dépôts / lhc / web / wiklou.git / blob
? search:


078de3c601f0bb1f79ffe93fe27864e296a14c93
[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 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // @codingStandardsIgnoreStart Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102 // @codingStandardsIgnoreEnd
103
104 # Regular expression for a non-newline space
105 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
106
107 # Flags for preprocessToDom
108 const PTD_FOR_INCLUSION = 1;
109
110 # Allowed values for $this->mOutputType
111 # Parameter to startExternalParse().
112 const OT_HTML = 1; # like parse()
113 const OT_WIKI = 2; # like preSaveTransform()
114 const OT_PREPROCESS = 3; # like preprocess()
115 const OT_MSG = 3;
116 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
117
118 /**
119 * @var string Prefix and suffix for temporary replacement strings
120 * for the multipass parser.
121 *
122 * \x7f should never appear in input as it's disallowed in XML.
123 * Using it at the front also gives us a little extra robustness
124 * since it shouldn't match when butted up against identifier-like
125 * string constructs.
126 *
127 * Must not consist of all title characters, or else it will change
128 * the behavior of <nowiki> in a link.
129 *
130 * Must have a character that needs escaping in attributes, otherwise
131 * someone could put a strip marker in an attribute, to get around
132 * escaping quote marks, and break out of the attribute. Thus we add
133 * `'".
134 */
135 const MARKER_SUFFIX = "-QINU`\"'\x7f";
136 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
137
138 # Markers used for wrapping the table of contents
139 const TOC_START = '<mw:toc>';
140 const TOC_END = '</mw:toc>';
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 public $mFirstCall = true;
155
156 # Initialised by initialiseVariables()
157
158 /**
159 * @var MagicWordArray
160 */
161 public $mVariables;
162
163 /**
164 * @var MagicWordArray
165 */
166 public $mSubstWords;
167 # Initialised in constructor
168 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
169
170 # Initialized in getPreprocessor()
171 /** @var Preprocessor */
172 public $mPreprocessor;
173
174 # Cleared with clearState():
175 /**
176 * @var ParserOutput
177 */
178 public $mOutput;
179 public $mAutonumber;
180
181 /**
182 * @var StripState
183 */
184 public $mStripState;
185
186 public $mIncludeCount;
187 /**
188 * @var LinkHolderArray
189 */
190 public $mLinkHolders;
191
192 public $mLinkID;
193 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
194 public $mDefaultSort;
195 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
196 public $mExpensiveFunctionCount; # number of expensive parser function calls
197 public $mShowToc, $mForceTocPosition;
198
199 /**
200 * @var User
201 */
202 public $mUser; # User object; only used when doing pre-save transform
203
204 # Temporary
205 # These are variables reset at least once per parse regardless of $clearState
206
207 /**
208 * @var ParserOptions
209 */
210 public $mOptions;
211
212 /**
213 * @var Title
214 */
215 public $mTitle; # Title context, used for self-link rendering and similar things
216 public $mOutputType; # Output type, one of the OT_xxx constants
217 public $ot; # Shortcut alias, see setOutputType()
218 public $mRevisionObject; # The revision object of the specified revision ID
219 public $mRevisionId; # ID to display in {{REVISIONID}} tags
220 public $mRevisionTimestamp; # The timestamp of the specified revision ID
221 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
222 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
223 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
224 public $mInputSize = false; # For {{PAGESIZE}} on current page.
225
226 /**
227 * @var string Deprecated accessor for the strip marker prefix.
228 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
229 */
230 public $mUniqPrefix = self::MARKER_PREFIX;
231
232 /**
233 * @var array Array with the language name of each language link (i.e. the
234 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
235 * duplicate language links to the ParserOutput.
236 */
237 public $mLangLinkLanguages;
238
239 /**
240 * @var MapCacheLRU|null
241 * @since 1.24
242 *
243 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
244 */
245 public $currentRevisionCache;
246
247 /**
248 * @var bool|string Recursive call protection.
249 * This variable should be treated as if it were private.
250 */
251 public $mInParse = false;
252
253 /** @var SectionProfiler */
254 protected $mProfiler;
255
256 /**
257 * @var LinkRenderer
258 */
259 protected $mLinkRenderer;
260
261 /**
262 * @param array $conf
263 */
264 public function __construct( $conf = [] ) {
265 $this->mConf = $conf;
266 $this->mUrlProtocols = wfUrlProtocols();
267 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
268 self::EXT_LINK_ADDR .
269 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
270 if ( isset( $conf['preprocessorClass'] ) ) {
271 $this->mPreprocessorClass = $conf['preprocessorClass'];
272 } elseif ( defined( 'HPHP_VERSION' ) ) {
273 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
274 $this->mPreprocessorClass = 'Preprocessor_Hash';
275 } elseif ( extension_loaded( 'domxml' ) ) {
276 # PECL extension that conflicts with the core DOM extension (T15770)
277 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
278 $this->mPreprocessorClass = 'Preprocessor_Hash';
279 } elseif ( extension_loaded( 'dom' ) ) {
280 $this->mPreprocessorClass = 'Preprocessor_DOM';
281 } else {
282 $this->mPreprocessorClass = 'Preprocessor_Hash';
283 }
284 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
285 }
286
287 /**
288 * Reduce memory usage to reduce the impact of circular references
289 */
290 public function __destruct() {
291 if ( isset( $this->mLinkHolders ) ) {
292 unset( $this->mLinkHolders );
293 }
294 foreach ( $this as $name => $value ) {
295 unset( $this->$name );
296 }
297 }
298
299 /**
300 * Allow extensions to clean up when the parser is cloned
301 */
302 public function __clone() {
303 $this->mInParse = false;
304
305 // T58226: When you create a reference "to" an object field, that
306 // makes the object field itself be a reference too (until the other
307 // reference goes out of scope). When cloning, any field that's a
308 // reference is copied as a reference in the new object. Both of these
309 // are defined PHP5 behaviors, as inconvenient as it is for us when old
310 // hooks from PHP4 days are passing fields by reference.
311 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
312 // Make a non-reference copy of the field, then rebind the field to
313 // reference the new copy.
314 $tmp = $this->$k;
315 $this->$k =& $tmp;
316 unset( $tmp );
317 }
318
319 Hooks::run( 'ParserCloned', [ $this ] );
320 }
321
322 /**
323 * Do various kinds of initialisation on the first call of the parser
324 */
325 public function firstCallInit() {
326 if ( !$this->mFirstCall ) {
327 return;
328 }
329 $this->mFirstCall = false;
330
331 CoreParserFunctions::register( $this );
332 CoreTagHooks::register( $this );
333 $this->initialiseVariables();
334
335 // Avoid PHP 7.1 warning from passing $this by reference
336 $parser = $this;
337 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
338 }
339
340 /**
341 * Clear Parser state
342 *
343 * @private
344 */
345 public function clearState() {
346 if ( $this->mFirstCall ) {
347 $this->firstCallInit();
348 }
349 $this->mOutput = new ParserOutput;
350 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
351 $this->mAutonumber = 0;
352 $this->mIncludeCount = [];
353 $this->mLinkHolders = new LinkHolderArray( $this );
354 $this->mLinkID = 0;
355 $this->mRevisionObject = $this->mRevisionTimestamp =
356 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
357 $this->mVarCache = [];
358 $this->mUser = null;
359 $this->mLangLinkLanguages = [];
360 $this->currentRevisionCache = null;
361
362 $this->mStripState = new StripState;
363
364 # Clear these on every parse, T6549
365 $this->mTplRedirCache = $this->mTplDomCache = [];
366
367 $this->mShowToc = true;
368 $this->mForceTocPosition = false;
369 $this->mIncludeSizes = [
370 'post-expand' => 0,
371 'arg' => 0,
372 ];
373 $this->mPPNodeCount = 0;
374 $this->mGeneratedPPNodeCount = 0;
375 $this->mHighestExpansionDepth = 0;
376 $this->mDefaultSort = false;
377 $this->mHeadings = [];
378 $this->mDoubleUnderscores = [];
379 $this->mExpensiveFunctionCount = 0;
380
381 # Fix cloning
382 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
383 $this->mPreprocessor = null;
384 }
385
386 $this->mProfiler = new SectionProfiler();
387
388 // Avoid PHP 7.1 warning from passing $this by reference
389 $parser = $this;
390 Hooks::run( 'ParserClearState', [ &$parser ] );
391 }
392
393 /**
394 * Convert wikitext to HTML
395 * Do not call this function recursively.
396 *
397 * @param string $text Text we want to parse
398 * @param Title $title
399 * @param ParserOptions $options
400 * @param bool $linestart
401 * @param bool $clearState
402 * @param int $revid Number to pass in {{REVISIONID}}
403 * @return ParserOutput A ParserOutput
404 */
405 public function parse(
406 $text, Title $title, ParserOptions $options,
407 $linestart = true, $clearState = true, $revid = null
408 ) {
409 if ( $clearState ) {
410 // We use U+007F DELETE to construct strip markers, so we have to make
411 // sure that this character does not occur in the input text.
412 $text = strtr( $text, "\x7f", "?" );
413 $magicScopeVariable = $this->lock();
414 }
415 // Strip U+0000 NULL (T159174)
416 $text = str_replace( "\000", '', $text );
417
418 $this->startParse( $title, $options, self::OT_HTML, $clearState );
419
420 $this->currentRevisionCache = null;
421 $this->mInputSize = strlen( $text );
422 if ( $this->mOptions->getEnableLimitReport() ) {
423 $this->mOutput->resetParseStartTime();
424 }
425
426 $oldRevisionId = $this->mRevisionId;
427 $oldRevisionObject = $this->mRevisionObject;
428 $oldRevisionTimestamp = $this->mRevisionTimestamp;
429 $oldRevisionUser = $this->mRevisionUser;
430 $oldRevisionSize = $this->mRevisionSize;
431 if ( $revid !== null ) {
432 $this->mRevisionId = $revid;
433 $this->mRevisionObject = null;
434 $this->mRevisionTimestamp = null;
435 $this->mRevisionUser = null;
436 $this->mRevisionSize = null;
437 }
438
439 // Avoid PHP 7.1 warning from passing $this by reference
440 $parser = $this;
441 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
442 # No more strip!
443 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
444 $text = $this->internalParse( $text );
445 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
446
447 $text = $this->internalParseHalfParsed( $text, true, $linestart );
448
449 /**
450 * A converted title will be provided in the output object if title and
451 * content conversion are enabled, the article text does not contain
452 * a conversion-suppressing double-underscore tag, and no
453 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
454 * automatic link conversion.
455 */
456 if ( !( $options->getDisableTitleConversion()
457 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
458 || isset( $this->mDoubleUnderscores['notitleconvert'] )
459 || $this->mOutput->getDisplayTitle() !== false )
460 ) {
461 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
462 if ( $convruletitle ) {
463 $this->mOutput->setTitleText( $convruletitle );
464 } else {
465 $titleText = $this->getConverterLanguage()->convertTitle( $title );
466 $this->mOutput->setTitleText( $titleText );
467 }
468 }
469
470 # Compute runtime adaptive expiry if set
471 $this->mOutput->finalizeAdaptiveCacheExpiry();
472
473 # Warn if too many heavyweight parser functions were used
474 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
475 $this->limitationWarn( 'expensive-parserfunction',
476 $this->mExpensiveFunctionCount,
477 $this->mOptions->getExpensiveParserFunctionLimit()
478 );
479 }
480
481 # Information on limits, for the benefit of users who try to skirt them
482 if ( $this->mOptions->getEnableLimitReport() ) {
483 $text .= $this->makeLimitReport();
484 }
485
486 # Wrap non-interface parser output in a <div> so it can be targeted
487 # with CSS (T37247)
488 $class = $this->mOptions->getWrapOutputClass();
489 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
490 $text = Html::rawElement( 'div', [ 'class' => $class ], $text );
491 }
492
493 $this->mOutput->setText( $text );
494
495 $this->mRevisionId = $oldRevisionId;
496 $this->mRevisionObject = $oldRevisionObject;
497 $this->mRevisionTimestamp = $oldRevisionTimestamp;
498 $this->mRevisionUser = $oldRevisionUser;
499 $this->mRevisionSize = $oldRevisionSize;
500 $this->mInputSize = false;
501 $this->currentRevisionCache = null;
502
503 return $this->mOutput;
504 }
505
506 /**
507 * Set the limit report data in the current ParserOutput, and return the
508 * limit report HTML comment.
509 *
510 * @return string
511 */
512 protected function makeLimitReport() {
513 global $wgShowHostnames;
514
515 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
516
517 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
518 if ( $cpuTime !== null ) {
519 $this->mOutput->setLimitReportData( 'limitreport-cputime',
520 sprintf( "%.3f", $cpuTime )
521 );
522 }
523
524 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
525 $this->mOutput->setLimitReportData( 'limitreport-walltime',
526 sprintf( "%.3f", $wallTime )
527 );
528
529 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
530 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
531 );
532 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
533 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
534 );
535 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
536 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
537 );
538 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
539 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
540 );
541 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
542 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
543 );
544 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
545 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
546 );
547 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
548
549 $limitReport = "NewPP limit report\n";
550 if ( $wgShowHostnames ) {
551 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
552 }
553 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
554 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
555 $limitReport .= 'Dynamic content: ' .
556 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
557 "\n";
558
559 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
560 if ( Hooks::run( 'ParserLimitReportFormat',
561 [ $key, &$value, &$limitReport, false, false ]
562 ) ) {
563 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
564 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
565 ->inLanguage( 'en' )->useDatabase( false );
566 if ( !$valueMsg->exists() ) {
567 $valueMsg = new RawMessage( '$1' );
568 }
569 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
570 $valueMsg->params( $value );
571 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
572 }
573 }
574 }
575 // Since we're not really outputting HTML, decode the entities and
576 // then re-encode the things that need hiding inside HTML comments.
577 $limitReport = htmlspecialchars_decode( $limitReport );
578 // Run deprecated hook
579 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
580
581 // Sanitize for comment. Note '‐' in the replacement is U+2010,
582 // which looks much like the problematic '-'.
583 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
584 $text = "\n<!-- \n$limitReport-->\n";
585
586 // Add on template profiling data in human/machine readable way
587 $dataByFunc = $this->mProfiler->getFunctionStats();
588 uasort( $dataByFunc, function ( $a, $b ) {
589 return $a['real'] < $b['real']; // descending order
590 } );
591 $profileReport = [];
592 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
593 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
594 $item['%real'], $item['real'], $item['calls'],
595 htmlspecialchars( $item['name'] ) );
596 }
597 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
598 $text .= implode( "\n", $profileReport ) . "\n-->\n";
599
600 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
601
602 // Add other cache related metadata
603 if ( $wgShowHostnames ) {
604 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
605 }
606 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
607 $this->mOutput->getCacheTime() );
608 $this->mOutput->setLimitReportData( 'cachereport-ttl',
609 $this->mOutput->getCacheExpiry() );
610 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
611 $this->mOutput->hasDynamicContent() );
612
613 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
614 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
615 $this->mTitle->getPrefixedDBkey() );
616 }
617 return $text;
618 }
619
620 /**
621 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
622 * can be called from an extension tag hook.
623 *
624 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
625 * instead, which means that lists and links have not been fully parsed yet,
626 * and strip markers are still present.
627 *
628 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
629 *
630 * Use this function if you're a parser tag hook and you want to parse
631 * wikitext before or after applying additional transformations, and you
632 * intend to *return the result as hook output*, which will cause it to go
633 * through the rest of parsing process automatically.
634 *
635 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
636 * $text are not expanded
637 *
638 * @param string $text Text extension wants to have parsed
639 * @param bool|PPFrame $frame The frame to use for expanding any template variables
640 * @return string UNSAFE half-parsed HTML
641 */
642 public function recursiveTagParse( $text, $frame = false ) {
643 // Avoid PHP 7.1 warning from passing $this by reference
644 $parser = $this;
645 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
646 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
647 $text = $this->internalParse( $text, false, $frame );
648 return $text;
649 }
650
651 /**
652 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
653 * point can be called from an extension tag hook.
654 *
655 * The output of this function is fully-parsed HTML that is safe for output.
656 * If you're a parser tag hook, you might want to use recursiveTagParse()
657 * instead.
658 *
659 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
660 * $text are not expanded
661 *
662 * @since 1.25
663 *
664 * @param string $text Text extension wants to have parsed
665 * @param bool|PPFrame $frame The frame to use for expanding any template variables
666 * @return string Fully parsed HTML
667 */
668 public function recursiveTagParseFully( $text, $frame = false ) {
669 $text = $this->recursiveTagParse( $text, $frame );
670 $text = $this->internalParseHalfParsed( $text, false );
671 return $text;
672 }
673
674 /**
675 * Expand templates and variables in the text, producing valid, static wikitext.
676 * Also removes comments.
677 * Do not call this function recursively.
678 * @param string $text
679 * @param Title $title
680 * @param ParserOptions $options
681 * @param int|null $revid
682 * @param bool|PPFrame $frame
683 * @return mixed|string
684 */
685 public function preprocess( $text, Title $title = null,
686 ParserOptions $options, $revid = null, $frame = false
687 ) {
688 $magicScopeVariable = $this->lock();
689 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
690 if ( $revid !== null ) {
691 $this->mRevisionId = $revid;
692 }
693 // Avoid PHP 7.1 warning from passing $this by reference
694 $parser = $this;
695 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
696 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
697 $text = $this->replaceVariables( $text, $frame );
698 $text = $this->mStripState->unstripBoth( $text );
699 return $text;
700 }
701
702 /**
703 * Recursive parser entry point that can be called from an extension tag
704 * hook.
705 *
706 * @param string $text Text to be expanded
707 * @param bool|PPFrame $frame The frame to use for expanding any template variables
708 * @return string
709 * @since 1.19
710 */
711 public function recursivePreprocess( $text, $frame = false ) {
712 $text = $this->replaceVariables( $text, $frame );
713 $text = $this->mStripState->unstripBoth( $text );
714 return $text;
715 }
716
717 /**
718 * Process the wikitext for the "?preload=" feature. (T7210)
719 *
720 * "<noinclude>", "<includeonly>" etc. are parsed as for template
721 * transclusion, comments, templates, arguments, tags hooks and parser
722 * functions are untouched.
723 *
724 * @param string $text
725 * @param Title $title
726 * @param ParserOptions $options
727 * @param array $params
728 * @return string
729 */
730 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
731 $msg = new RawMessage( $text );
732 $text = $msg->params( $params )->plain();
733
734 # Parser (re)initialisation
735 $magicScopeVariable = $this->lock();
736 $this->startParse( $title, $options, self::OT_PLAIN, true );
737
738 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
739 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
740 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
741 $text = $this->mStripState->unstripBoth( $text );
742 return $text;
743 }
744
745 /**
746 * Set the current user.
747 * Should only be used when doing pre-save transform.
748 *
749 * @param User|null $user User object or null (to reset)
750 */
751 public function setUser( $user ) {
752 $this->mUser = $user;
753 }
754
755 /**
756 * Set the context title
757 *
758 * @param Title $t
759 */
760 public function setTitle( $t ) {
761 if ( !$t ) {
762 $t = Title::newFromText( 'NO TITLE' );
763 }
764
765 if ( $t->hasFragment() ) {
766 # Strip the fragment to avoid various odd effects
767 $this->mTitle = $t->createFragmentTarget( '' );
768 } else {
769 $this->mTitle = $t;
770 }
771 }
772
773 /**
774 * Accessor for the Title object
775 *
776 * @return Title
777 */
778 public function getTitle() {
779 return $this->mTitle;
780 }
781
782 /**
783 * Accessor/mutator for the Title object
784 *
785 * @param Title $x Title object or null to just get the current one
786 * @return Title
787 */
788 public function Title( $x = null ) {
789 return wfSetVar( $this->mTitle, $x );
790 }
791
792 /**
793 * Set the output type
794 *
795 * @param int $ot New value
796 */
797 public function setOutputType( $ot ) {
798 $this->mOutputType = $ot;
799 # Shortcut alias
800 $this->ot = [
801 'html' => $ot == self::OT_HTML,
802 'wiki' => $ot == self::OT_WIKI,
803 'pre' => $ot == self::OT_PREPROCESS,
804 'plain' => $ot == self::OT_PLAIN,
805 ];
806 }
807
808 /**
809 * Accessor/mutator for the output type
810 *
811 * @param int|null $x New value or null to just get the current one
812 * @return int
813 */
814 public function OutputType( $x = null ) {
815 return wfSetVar( $this->mOutputType, $x );
816 }
817
818 /**
819 * Get the ParserOutput object
820 *
821 * @return ParserOutput
822 */
823 public function getOutput() {
824 return $this->mOutput;
825 }
826
827 /**
828 * Get the ParserOptions object
829 *
830 * @return ParserOptions
831 */
832 public function getOptions() {
833 return $this->mOptions;
834 }
835
836 /**
837 * Accessor/mutator for the ParserOptions object
838 *
839 * @param ParserOptions $x New value or null to just get the current one
840 * @return ParserOptions Current ParserOptions object
841 */
842 public function Options( $x = null ) {
843 return wfSetVar( $this->mOptions, $x );
844 }
845
846 /**
847 * @return int
848 */
849 public function nextLinkID() {
850 return $this->mLinkID++;
851 }
852
853 /**
854 * @param int $id
855 */
856 public function setLinkID( $id ) {
857 $this->mLinkID = $id;
858 }
859
860 /**
861 * Get a language object for use in parser functions such as {{FORMATNUM:}}
862 * @return Language
863 */
864 public function getFunctionLang() {
865 return $this->getTargetLanguage();
866 }
867
868 /**
869 * Get the target language for the content being parsed. This is usually the
870 * language that the content is in.
871 *
872 * @since 1.19
873 *
874 * @throws MWException
875 * @return Language
876 */
877 public function getTargetLanguage() {
878 $target = $this->mOptions->getTargetLanguage();
879
880 if ( $target !== null ) {
881 return $target;
882 } elseif ( $this->mOptions->getInterfaceMessage() ) {
883 return $this->mOptions->getUserLangObj();
884 } elseif ( is_null( $this->mTitle ) ) {
885 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
886 }
887
888 return $this->mTitle->getPageLanguage();
889 }
890
891 /**
892 * Get the language object for language conversion
893 * @return Language|null
894 */
895 public function getConverterLanguage() {
896 return $this->getTargetLanguage();
897 }
898
899 /**
900 * Get a User object either from $this->mUser, if set, or from the
901 * ParserOptions object otherwise
902 *
903 * @return User
904 */
905 public function getUser() {
906 if ( !is_null( $this->mUser ) ) {
907 return $this->mUser;
908 }
909 return $this->mOptions->getUser();
910 }
911
912 /**
913 * Get a preprocessor object
914 *
915 * @return Preprocessor
916 */
917 public function getPreprocessor() {
918 if ( !isset( $this->mPreprocessor ) ) {
919 $class = $this->mPreprocessorClass;
920 $this->mPreprocessor = new $class( $this );
921 }
922 return $this->mPreprocessor;
923 }
924
925 /**
926 * Get a LinkRenderer instance to make links with
927 *
928 * @since 1.28
929 * @return LinkRenderer
930 */
931 public function getLinkRenderer() {
932 if ( !$this->mLinkRenderer ) {
933 $this->mLinkRenderer = MediaWikiServices::getInstance()
934 ->getLinkRendererFactory()->create();
935 $this->mLinkRenderer->setStubThreshold(
936 $this->getOptions()->getStubThreshold()
937 );
938 }
939
940 return $this->mLinkRenderer;
941 }
942
943 /**
944 * Replaces all occurrences of HTML-style comments and the given tags
945 * in the text with a random marker and returns the next text. The output
946 * parameter $matches will be an associative array filled with data in
947 * the form:
948 *
949 * @code
950 * 'UNIQ-xxxxx' => [
951 * 'element',
952 * 'tag content',
953 * [ 'param' => 'x' ],
954 * '<element param="x">tag content</element>' ]
955 * @endcode
956 *
957 * @param array $elements List of element names. Comments are always extracted.
958 * @param string $text Source text string.
959 * @param array &$matches Out parameter, Array: extracted tags
960 * @return string Stripped text
961 */
962 public static function extractTagsAndParams( $elements, $text, &$matches ) {
963 static $n = 1;
964 $stripped = '';
965 $matches = [];
966
967 $taglist = implode( '|', $elements );
968 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
969
970 while ( $text != '' ) {
971 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
972 $stripped .= $p[0];
973 if ( count( $p ) < 5 ) {
974 break;
975 }
976 if ( count( $p ) > 5 ) {
977 # comment
978 $element = $p[4];
979 $attributes = '';
980 $close = '';
981 $inside = $p[5];
982 } else {
983 # tag
984 $element = $p[1];
985 $attributes = $p[2];
986 $close = $p[3];
987 $inside = $p[4];
988 }
989
990 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
991 $stripped .= $marker;
992
993 if ( $close === '/>' ) {
994 # Empty element tag, <tag />
995 $content = null;
996 $text = $inside;
997 $tail = null;
998 } else {
999 if ( $element === '!--' ) {
1000 $end = '/(-->)/';
1001 } else {
1002 $end = "/(<\\/$element\\s*>)/i";
1003 }
1004 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1005 $content = $q[0];
1006 if ( count( $q ) < 3 ) {
1007 # No end tag -- let it run out to the end of the text.
1008 $tail = '';
1009 $text = '';
1010 } else {
1011 $tail = $q[1];
1012 $text = $q[2];
1013 }
1014 }
1015
1016 $matches[$marker] = [ $element,
1017 $content,
1018 Sanitizer::decodeTagAttributes( $attributes ),
1019 "<$element$attributes$close$content$tail" ];
1020 }
1021 return $stripped;
1022 }
1023
1024 /**
1025 * Get a list of strippable XML-like elements
1026 *
1027 * @return array
1028 */
1029 public function getStripList() {
1030 return $this->mStripList;
1031 }
1032
1033 /**
1034 * Add an item to the strip state
1035 * Returns the unique tag which must be inserted into the stripped text
1036 * The tag will be replaced with the original text in unstrip()
1037 *
1038 * @param string $text
1039 *
1040 * @return string
1041 */
1042 public function insertStripItem( $text ) {
1043 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1044 $this->mMarkerIndex++;
1045 $this->mStripState->addGeneral( $marker, $text );
1046 return $marker;
1047 }
1048
1049 /**
1050 * parse the wiki syntax used to render tables
1051 *
1052 * @private
1053 * @param string $text
1054 * @return string
1055 */
1056 public function doTableStuff( $text ) {
1057 $lines = StringUtils::explode( "\n", $text );
1058 $out = '';
1059 $td_history = []; # Is currently a td tag open?
1060 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1061 $tr_history = []; # Is currently a tr tag open?
1062 $tr_attributes = []; # history of tr attributes
1063 $has_opened_tr = []; # Did this table open a <tr> element?
1064 $indent_level = 0; # indent level of the table
1065
1066 foreach ( $lines as $outLine ) {
1067 $line = trim( $outLine );
1068
1069 if ( $line === '' ) { # empty line, go to next line
1070 $out .= $outLine . "\n";
1071 continue;
1072 }
1073
1074 $first_character = $line[0];
1075 $first_two = substr( $line, 0, 2 );
1076 $matches = [];
1077
1078 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1079 # First check if we are starting a new table
1080 $indent_level = strlen( $matches[1] );
1081
1082 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1083 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1084
1085 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1086 array_push( $td_history, false );
1087 array_push( $last_tag_history, '' );
1088 array_push( $tr_history, false );
1089 array_push( $tr_attributes, '' );
1090 array_push( $has_opened_tr, false );
1091 } elseif ( count( $td_history ) == 0 ) {
1092 # Don't do any of the following
1093 $out .= $outLine . "\n";
1094 continue;
1095 } elseif ( $first_two === '|}' ) {
1096 # We are ending a table
1097 $line = '</table>' . substr( $line, 2 );
1098 $last_tag = array_pop( $last_tag_history );
1099
1100 if ( !array_pop( $has_opened_tr ) ) {
1101 $line = "<tr><td></td></tr>{$line}";
1102 }
1103
1104 if ( array_pop( $tr_history ) ) {
1105 $line = "</tr>{$line}";
1106 }
1107
1108 if ( array_pop( $td_history ) ) {
1109 $line = "</{$last_tag}>{$line}";
1110 }
1111 array_pop( $tr_attributes );
1112 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1113 } elseif ( $first_two === '|-' ) {
1114 # Now we have a table row
1115 $line = preg_replace( '#^\|-+#', '', $line );
1116
1117 # Whats after the tag is now only attributes
1118 $attributes = $this->mStripState->unstripBoth( $line );
1119 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1120 array_pop( $tr_attributes );
1121 array_push( $tr_attributes, $attributes );
1122
1123 $line = '';
1124 $last_tag = array_pop( $last_tag_history );
1125 array_pop( $has_opened_tr );
1126 array_push( $has_opened_tr, true );
1127
1128 if ( array_pop( $tr_history ) ) {
1129 $line = '</tr>';
1130 }
1131
1132 if ( array_pop( $td_history ) ) {
1133 $line = "</{$last_tag}>{$line}";
1134 }
1135
1136 $outLine = $line;
1137 array_push( $tr_history, false );
1138 array_push( $td_history, false );
1139 array_push( $last_tag_history, '' );
1140 } elseif ( $first_character === '|'
1141 || $first_character === '!'
1142 || $first_two === '|+'
1143 ) {
1144 # This might be cell elements, td, th or captions
1145 if ( $first_two === '|+' ) {
1146 $first_character = '+';
1147 $line = substr( $line, 2 );
1148 } else {
1149 $line = substr( $line, 1 );
1150 }
1151
1152 // Implies both are valid for table headings.
1153 if ( $first_character === '!' ) {
1154 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1155 }
1156
1157 # Split up multiple cells on the same line.
1158 # FIXME : This can result in improper nesting of tags processed
1159 # by earlier parser steps.
1160 $cells = explode( '||', $line );
1161
1162 $outLine = '';
1163
1164 # Loop through each table cell
1165 foreach ( $cells as $cell ) {
1166 $previous = '';
1167 if ( $first_character !== '+' ) {
1168 $tr_after = array_pop( $tr_attributes );
1169 if ( !array_pop( $tr_history ) ) {
1170 $previous = "<tr{$tr_after}>\n";
1171 }
1172 array_push( $tr_history, true );
1173 array_push( $tr_attributes, '' );
1174 array_pop( $has_opened_tr );
1175 array_push( $has_opened_tr, true );
1176 }
1177
1178 $last_tag = array_pop( $last_tag_history );
1179
1180 if ( array_pop( $td_history ) ) {
1181 $previous = "</{$last_tag}>\n{$previous}";
1182 }
1183
1184 if ( $first_character === '|' ) {
1185 $last_tag = 'td';
1186 } elseif ( $first_character === '!' ) {
1187 $last_tag = 'th';
1188 } elseif ( $first_character === '+' ) {
1189 $last_tag = 'caption';
1190 } else {
1191 $last_tag = '';
1192 }
1193
1194 array_push( $last_tag_history, $last_tag );
1195
1196 # A cell could contain both parameters and data
1197 $cell_data = explode( '|', $cell, 2 );
1198
1199 # T2553: Note that a '|' inside an invalid link should not
1200 # be mistaken as delimiting cell parameters
1201 # Bug T153140: Neither should language converter markup.
1202 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1203 $cell = "{$previous}<{$last_tag}>{$cell}";
1204 } elseif ( count( $cell_data ) == 1 ) {
1205 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1206 } else {
1207 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1208 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1209 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1210 }
1211
1212 $outLine .= $cell;
1213 array_push( $td_history, true );
1214 }
1215 }
1216 $out .= $outLine . "\n";
1217 }
1218
1219 # Closing open td, tr && table
1220 while ( count( $td_history ) > 0 ) {
1221 if ( array_pop( $td_history ) ) {
1222 $out .= "</td>\n";
1223 }
1224 if ( array_pop( $tr_history ) ) {
1225 $out .= "</tr>\n";
1226 }
1227 if ( !array_pop( $has_opened_tr ) ) {
1228 $out .= "<tr><td></td></tr>\n";
1229 }
1230
1231 $out .= "</table>\n";
1232 }
1233
1234 # Remove trailing line-ending (b/c)
1235 if ( substr( $out, -1 ) === "\n" ) {
1236 $out = substr( $out, 0, -1 );
1237 }
1238
1239 # special case: don't return empty table
1240 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1241 $out = '';
1242 }
1243
1244 return $out;
1245 }
1246
1247 /**
1248 * Helper function for parse() that transforms wiki markup into half-parsed
1249 * HTML. Only called for $mOutputType == self::OT_HTML.
1250 *
1251 * @private
1252 *
1253 * @param string $text The text to parse
1254 * @param bool $isMain Whether this is being called from the main parse() function
1255 * @param PPFrame|bool $frame A pre-processor frame
1256 *
1257 * @return string
1258 */
1259 public function internalParse( $text, $isMain = true, $frame = false ) {
1260 $origText = $text;
1261
1262 // Avoid PHP 7.1 warning from passing $this by reference
1263 $parser = $this;
1264
1265 # Hook to suspend the parser in this state
1266 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1267 return $text;
1268 }
1269
1270 # if $frame is provided, then use $frame for replacing any variables
1271 if ( $frame ) {
1272 # use frame depth to infer how include/noinclude tags should be handled
1273 # depth=0 means this is the top-level document; otherwise it's an included document
1274 if ( !$frame->depth ) {
1275 $flag = 0;
1276 } else {
1277 $flag = self::PTD_FOR_INCLUSION;
1278 }
1279 $dom = $this->preprocessToDom( $text, $flag );
1280 $text = $frame->expand( $dom );
1281 } else {
1282 # if $frame is not provided, then use old-style replaceVariables
1283 $text = $this->replaceVariables( $text );
1284 }
1285
1286 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1287 $text = Sanitizer::removeHTMLtags(
1288 $text,
1289 [ $this, 'attributeStripCallback' ],
1290 false,
1291 array_keys( $this->mTransparentTagHooks ),
1292 [],
1293 [ $this, 'addTrackingCategory' ]
1294 );
1295 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1296
1297 # Tables need to come after variable replacement for things to work
1298 # properly; putting them before other transformations should keep
1299 # exciting things like link expansions from showing up in surprising
1300 # places.
1301 $text = $this->doTableStuff( $text );
1302
1303 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1304
1305 $text = $this->doDoubleUnderscore( $text );
1306
1307 $text = $this->doHeadings( $text );
1308 $text = $this->replaceInternalLinks( $text );
1309 $text = $this->doAllQuotes( $text );
1310 $text = $this->replaceExternalLinks( $text );
1311
1312 # replaceInternalLinks may sometimes leave behind
1313 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1314 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1315
1316 $text = $this->doMagicLinks( $text );
1317 $text = $this->formatHeadings( $text, $origText, $isMain );
1318
1319 return $text;
1320 }
1321
1322 /**
1323 * Helper function for parse() that transforms half-parsed HTML into fully
1324 * parsed HTML.
1325 *
1326 * @param string $text
1327 * @param bool $isMain
1328 * @param bool $linestart
1329 * @return string
1330 */
1331 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1332 $text = $this->mStripState->unstripGeneral( $text );
1333
1334 // Avoid PHP 7.1 warning from passing $this by reference
1335 $parser = $this;
1336
1337 if ( $isMain ) {
1338 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1339 }
1340
1341 # Clean up special characters, only run once, next-to-last before doBlockLevels
1342 $fixtags = [
1343 # French spaces, last one Guillemet-left
1344 # only if there is something before the space
1345 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1346 # french spaces, Guillemet-right
1347 '/(\\302\\253) /' => '\\1&#160;',
1348 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1349 ];
1350 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1351
1352 $text = $this->doBlockLevels( $text, $linestart );
1353
1354 $this->replaceLinkHolders( $text );
1355
1356 /**
1357 * The input doesn't get language converted if
1358 * a) It's disabled
1359 * b) Content isn't converted
1360 * c) It's a conversion table
1361 * d) it is an interface message (which is in the user language)
1362 */
1363 if ( !( $this->mOptions->getDisableContentConversion()
1364 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1365 ) {
1366 if ( !$this->mOptions->getInterfaceMessage() ) {
1367 # The position of the convert() call should not be changed. it
1368 # assumes that the links are all replaced and the only thing left
1369 # is the <nowiki> mark.
1370 $text = $this->getConverterLanguage()->convert( $text );
1371 }
1372 }
1373
1374 $text = $this->mStripState->unstripNoWiki( $text );
1375
1376 if ( $isMain ) {
1377 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1378 }
1379
1380 $text = $this->replaceTransparentTags( $text );
1381 $text = $this->mStripState->unstripGeneral( $text );
1382
1383 $text = Sanitizer::normalizeCharReferences( $text );
1384
1385 if ( MWTidy::isEnabled() ) {
1386 if ( $this->mOptions->getTidy() ) {
1387 $text = MWTidy::tidy( $text );
1388 }
1389 } else {
1390 # attempt to sanitize at least some nesting problems
1391 # (T4702 and quite a few others)
1392 $tidyregs = [
1393 # ''Something [http://www.cool.com cool''] -->
1394 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1395 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1396 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1397 # fix up an anchor inside another anchor, only
1398 # at least for a single single nested link (T5695)
1399 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1400 '\\1\\2</a>\\3</a>\\1\\4</a>',
1401 # fix div inside inline elements- doBlockLevels won't wrap a line which
1402 # contains a div, so fix it up here; replace
1403 # div with escaped text
1404 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1405 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1406 # remove empty italic or bold tag pairs, some
1407 # introduced by rules above
1408 '/<([bi])><\/\\1>/' => '',
1409 ];
1410
1411 $text = preg_replace(
1412 array_keys( $tidyregs ),
1413 array_values( $tidyregs ),
1414 $text );
1415 }
1416
1417 if ( $isMain ) {
1418 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1419 }
1420
1421 return $text;
1422 }
1423
1424 /**
1425 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1426 * magic external links.
1427 *
1428 * DML
1429 * @private
1430 *
1431 * @param string $text
1432 *
1433 * @return string
1434 */
1435 public function doMagicLinks( $text ) {
1436 $prots = wfUrlProtocolsWithoutProtRel();
1437 $urlChar = self::EXT_LINK_URL_CLASS;
1438 $addr = self::EXT_LINK_ADDR;
1439 $space = self::SPACE_NOT_NL; # non-newline space
1440 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1441 $spaces = "$space++"; # possessive match of 1 or more spaces
1442 $text = preg_replace_callback(
1443 '!(?: # Start cases
1444 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1445 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1446 (\b # m[3]: Free external links
1447 (?i:$prots)
1448 ($addr$urlChar*) # m[4]: Post-protocol path
1449 ) |
1450 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1451 ([0-9]+)\b |
1452 \bISBN $spaces ( # m[6]: ISBN, capture number
1453 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1454 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1455 [0-9Xx] # check digit
1456 )\b
1457 )!xu", [ $this, 'magicLinkCallback' ], $text );
1458 return $text;
1459 }
1460
1461 /**
1462 * @throws MWException
1463 * @param array $m
1464 * @return HTML|string
1465 */
1466 public function magicLinkCallback( $m ) {
1467 if ( isset( $m[1] ) && $m[1] !== '' ) {
1468 # Skip anchor
1469 return $m[0];
1470 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1471 # Skip HTML element
1472 return $m[0];
1473 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1474 # Free external link
1475 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1476 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1477 # RFC or PMID
1478 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1479 if ( !$this->mOptions->getMagicRFCLinks() ) {
1480 return $m[0];
1481 }
1482 $keyword = 'RFC';
1483 $urlmsg = 'rfcurl';
1484 $cssClass = 'mw-magiclink-rfc';
1485 $trackingCat = 'magiclink-tracking-rfc';
1486 $id = $m[5];
1487 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1488 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1489 return $m[0];
1490 }
1491 $keyword = 'PMID';
1492 $urlmsg = 'pubmedurl';
1493 $cssClass = 'mw-magiclink-pmid';
1494 $trackingCat = 'magiclink-tracking-pmid';
1495 $id = $m[5];
1496 } else {
1497 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1498 substr( $m[0], 0, 20 ) . '"' );
1499 }
1500 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1501 $this->addTrackingCategory( $trackingCat );
1502 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1503 } elseif ( isset( $m[6] ) && $m[6] !== ''
1504 && $this->mOptions->getMagicISBNLinks()
1505 ) {
1506 # ISBN
1507 $isbn = $m[6];
1508 $space = self::SPACE_NOT_NL; # non-newline space
1509 $isbn = preg_replace( "/$space/", ' ', $isbn );
1510 $num = strtr( $isbn, [
1511 '-' => '',
1512 ' ' => '',
1513 'x' => 'X',
1514 ] );
1515 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1516 return $this->getLinkRenderer()->makeKnownLink(
1517 SpecialPage::getTitleFor( 'Booksources', $num ),
1518 "ISBN $isbn",
1519 [
1520 'class' => 'internal mw-magiclink-isbn',
1521 'title' => false // suppress title attribute
1522 ]
1523 );
1524 } else {
1525 return $m[0];
1526 }
1527 }
1528
1529 /**
1530 * Make a free external link, given a user-supplied URL
1531 *
1532 * @param string $url
1533 * @param int $numPostProto
1534 * The number of characters after the protocol.
1535 * @return string HTML
1536 * @private
1537 */
1538 public function makeFreeExternalLink( $url, $numPostProto ) {
1539 $trail = '';
1540
1541 # The characters '<' and '>' (which were escaped by
1542 # removeHTMLtags()) should not be included in
1543 # URLs, per RFC 2396.
1544 # Make &nbsp; terminate a URL as well (bug T84937)
1545 $m2 = [];
1546 if ( preg_match(
1547 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1548 $url,
1549 $m2,
1550 PREG_OFFSET_CAPTURE
1551 ) ) {
1552 $trail = substr( $url, $m2[0][1] ) . $trail;
1553 $url = substr( $url, 0, $m2[0][1] );
1554 }
1555
1556 # Move trailing punctuation to $trail
1557 $sep = ',;\.:!?';
1558 # If there is no left bracket, then consider right brackets fair game too
1559 if ( strpos( $url, '(' ) === false ) {
1560 $sep .= ')';
1561 }
1562
1563 $urlRev = strrev( $url );
1564 $numSepChars = strspn( $urlRev, $sep );
1565 # Don't break a trailing HTML entity by moving the ; into $trail
1566 # This is in hot code, so use substr_compare to avoid having to
1567 # create a new string object for the comparison
1568 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1569 # more optimization: instead of running preg_match with a $
1570 # anchor, which can be slow, do the match on the reversed
1571 # string starting at the desired offset.
1572 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1573 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1574 $numSepChars--;
1575 }
1576 }
1577 if ( $numSepChars ) {
1578 $trail = substr( $url, -$numSepChars ) . $trail;
1579 $url = substr( $url, 0, -$numSepChars );
1580 }
1581
1582 # Verify that we still have a real URL after trail removal, and
1583 # not just lone protocol
1584 if ( strlen( $trail ) >= $numPostProto ) {
1585 return $url . $trail;
1586 }
1587
1588 $url = Sanitizer::cleanUrl( $url );
1589
1590 # Is this an external image?
1591 $text = $this->maybeMakeExternalImage( $url );
1592 if ( $text === false ) {
1593 # Not an image, make a link
1594 $text = Linker::makeExternalLink( $url,
1595 $this->getConverterLanguage()->markNoConversion( $url, true ),
1596 true, 'free',
1597 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1598 # Register it in the output object...
1599 $this->mOutput->addExternalLink( $url );
1600 }
1601 return $text . $trail;
1602 }
1603
1604 /**
1605 * Parse headers and return html
1606 *
1607 * @private
1608 *
1609 * @param string $text
1610 *
1611 * @return string
1612 */
1613 public function doHeadings( $text ) {
1614 for ( $i = 6; $i >= 1; --$i ) {
1615 $h = str_repeat( '=', $i );
1616 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1617 }
1618 return $text;
1619 }
1620
1621 /**
1622 * Replace single quotes with HTML markup
1623 * @private
1624 *
1625 * @param string $text
1626 *
1627 * @return string The altered text
1628 */
1629 public function doAllQuotes( $text ) {
1630 $outtext = '';
1631 $lines = StringUtils::explode( "\n", $text );
1632 foreach ( $lines as $line ) {
1633 $outtext .= $this->doQuotes( $line ) . "\n";
1634 }
1635 $outtext = substr( $outtext, 0, -1 );
1636 return $outtext;
1637 }
1638
1639 /**
1640 * Helper function for doAllQuotes()
1641 *
1642 * @param string $text
1643 *
1644 * @return string
1645 */
1646 public function doQuotes( $text ) {
1647 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1648 $countarr = count( $arr );
1649 if ( $countarr == 1 ) {
1650 return $text;
1651 }
1652
1653 // First, do some preliminary work. This may shift some apostrophes from
1654 // being mark-up to being text. It also counts the number of occurrences
1655 // of bold and italics mark-ups.
1656 $numbold = 0;
1657 $numitalics = 0;
1658 for ( $i = 1; $i < $countarr; $i += 2 ) {
1659 $thislen = strlen( $arr[$i] );
1660 // If there are ever four apostrophes, assume the first is supposed to
1661 // be text, and the remaining three constitute mark-up for bold text.
1662 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1663 if ( $thislen == 4 ) {
1664 $arr[$i - 1] .= "'";
1665 $arr[$i] = "'''";
1666 $thislen = 3;
1667 } elseif ( $thislen > 5 ) {
1668 // If there are more than 5 apostrophes in a row, assume they're all
1669 // text except for the last 5.
1670 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1671 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1672 $arr[$i] = "'''''";
1673 $thislen = 5;
1674 }
1675 // Count the number of occurrences of bold and italics mark-ups.
1676 if ( $thislen == 2 ) {
1677 $numitalics++;
1678 } elseif ( $thislen == 3 ) {
1679 $numbold++;
1680 } elseif ( $thislen == 5 ) {
1681 $numitalics++;
1682 $numbold++;
1683 }
1684 }
1685
1686 // If there is an odd number of both bold and italics, it is likely
1687 // that one of the bold ones was meant to be an apostrophe followed
1688 // by italics. Which one we cannot know for certain, but it is more
1689 // likely to be one that has a single-letter word before it.
1690 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1691 $firstsingleletterword = -1;
1692 $firstmultiletterword = -1;
1693 $firstspace = -1;
1694 for ( $i = 1; $i < $countarr; $i += 2 ) {
1695 if ( strlen( $arr[$i] ) == 3 ) {
1696 $x1 = substr( $arr[$i - 1], -1 );
1697 $x2 = substr( $arr[$i - 1], -2, 1 );
1698 if ( $x1 === ' ' ) {
1699 if ( $firstspace == -1 ) {
1700 $firstspace = $i;
1701 }
1702 } elseif ( $x2 === ' ' ) {
1703 $firstsingleletterword = $i;
1704 // if $firstsingleletterword is set, we don't
1705 // look at the other options, so we can bail early.
1706 break;
1707 } else {
1708 if ( $firstmultiletterword == -1 ) {
1709 $firstmultiletterword = $i;
1710 }
1711 }
1712 }
1713 }
1714
1715 // If there is a single-letter word, use it!
1716 if ( $firstsingleletterword > -1 ) {
1717 $arr[$firstsingleletterword] = "''";
1718 $arr[$firstsingleletterword - 1] .= "'";
1719 } elseif ( $firstmultiletterword > -1 ) {
1720 // If not, but there's a multi-letter word, use that one.
1721 $arr[$firstmultiletterword] = "''";
1722 $arr[$firstmultiletterword - 1] .= "'";
1723 } elseif ( $firstspace > -1 ) {
1724 // ... otherwise use the first one that has neither.
1725 // (notice that it is possible for all three to be -1 if, for example,
1726 // there is only one pentuple-apostrophe in the line)
1727 $arr[$firstspace] = "''";
1728 $arr[$firstspace - 1] .= "'";
1729 }
1730 }
1731
1732 // Now let's actually convert our apostrophic mush to HTML!
1733 $output = '';
1734 $buffer = '';
1735 $state = '';
1736 $i = 0;
1737 foreach ( $arr as $r ) {
1738 if ( ( $i % 2 ) == 0 ) {
1739 if ( $state === 'both' ) {
1740 $buffer .= $r;
1741 } else {
1742 $output .= $r;
1743 }
1744 } else {
1745 $thislen = strlen( $r );
1746 if ( $thislen == 2 ) {
1747 if ( $state === 'i' ) {
1748 $output .= '</i>';
1749 $state = '';
1750 } elseif ( $state === 'bi' ) {
1751 $output .= '</i>';
1752 $state = 'b';
1753 } elseif ( $state === 'ib' ) {
1754 $output .= '</b></i><b>';
1755 $state = 'b';
1756 } elseif ( $state === 'both' ) {
1757 $output .= '<b><i>' . $buffer . '</i>';
1758 $state = 'b';
1759 } else { // $state can be 'b' or ''
1760 $output .= '<i>';
1761 $state .= 'i';
1762 }
1763 } elseif ( $thislen == 3 ) {
1764 if ( $state === 'b' ) {
1765 $output .= '</b>';
1766 $state = '';
1767 } elseif ( $state === 'bi' ) {
1768 $output .= '</i></b><i>';
1769 $state = 'i';
1770 } elseif ( $state === 'ib' ) {
1771 $output .= '</b>';
1772 $state = 'i';
1773 } elseif ( $state === 'both' ) {
1774 $output .= '<i><b>' . $buffer . '</b>';
1775 $state = 'i';
1776 } else { // $state can be 'i' or ''
1777 $output .= '<b>';
1778 $state .= 'b';
1779 }
1780 } elseif ( $thislen == 5 ) {
1781 if ( $state === 'b' ) {
1782 $output .= '</b><i>';
1783 $state = 'i';
1784 } elseif ( $state === 'i' ) {
1785 $output .= '</i><b>';
1786 $state = 'b';
1787 } elseif ( $state === 'bi' ) {
1788 $output .= '</i></b>';
1789 $state = '';
1790 } elseif ( $state === 'ib' ) {
1791 $output .= '</b></i>';
1792 $state = '';
1793 } elseif ( $state === 'both' ) {
1794 $output .= '<i><b>' . $buffer . '</b></i>';
1795 $state = '';
1796 } else { // ($state == '')
1797 $buffer = '';
1798 $state = 'both';
1799 }
1800 }
1801 }
1802 $i++;
1803 }
1804 // Now close all remaining tags. Notice that the order is important.
1805 if ( $state === 'b' || $state === 'ib' ) {
1806 $output .= '</b>';
1807 }
1808 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1809 $output .= '</i>';
1810 }
1811 if ( $state === 'bi' ) {
1812 $output .= '</b>';
1813 }
1814 // There might be lonely ''''', so make sure we have a buffer
1815 if ( $state === 'both' && $buffer ) {
1816 $output .= '<b><i>' . $buffer . '</i></b>';
1817 }
1818 return $output;
1819 }
1820
1821 /**
1822 * Replace external links (REL)
1823 *
1824 * Note: this is all very hackish and the order of execution matters a lot.
1825 * Make sure to run tests/parser/parserTests.php if you change this code.
1826 *
1827 * @private
1828 *
1829 * @param string $text
1830 *
1831 * @throws MWException
1832 * @return string
1833 */
1834 public function replaceExternalLinks( $text ) {
1835 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1836 if ( $bits === false ) {
1837 throw new MWException( "PCRE needs to be compiled with "
1838 . "--enable-unicode-properties in order for MediaWiki to function" );
1839 }
1840 $s = array_shift( $bits );
1841
1842 $i = 0;
1843 while ( $i < count( $bits ) ) {
1844 $url = $bits[$i++];
1845 $i++; // protocol
1846 $text = $bits[$i++];
1847 $trail = $bits[$i++];
1848
1849 # The characters '<' and '>' (which were escaped by
1850 # removeHTMLtags()) should not be included in
1851 # URLs, per RFC 2396.
1852 $m2 = [];
1853 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1854 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1855 $url = substr( $url, 0, $m2[0][1] );
1856 }
1857
1858 # If the link text is an image URL, replace it with an <img> tag
1859 # This happened by accident in the original parser, but some people used it extensively
1860 $img = $this->maybeMakeExternalImage( $text );
1861 if ( $img !== false ) {
1862 $text = $img;
1863 }
1864
1865 $dtrail = '';
1866
1867 # Set linktype for CSS - if URL==text, link is essentially free
1868 $linktype = ( $text === $url ) ? 'free' : 'text';
1869
1870 # No link text, e.g. [http://domain.tld/some.link]
1871 if ( $text == '' ) {
1872 # Autonumber
1873 $langObj = $this->getTargetLanguage();
1874 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1875 $linktype = 'autonumber';
1876 } else {
1877 # Have link text, e.g. [http://domain.tld/some.link text]s
1878 # Check for trail
1879 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1880 }
1881
1882 $text = $this->getConverterLanguage()->markNoConversion( $text );
1883
1884 $url = Sanitizer::cleanUrl( $url );
1885
1886 # Use the encoded URL
1887 # This means that users can paste URLs directly into the text
1888 # Funny characters like ö aren't valid in URLs anyway
1889 # This was changed in August 2004
1890 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1891 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1892
1893 # Register link in the output object.
1894 $this->mOutput->addExternalLink( $url );
1895 }
1896
1897 return $s;
1898 }
1899
1900 /**
1901 * Get the rel attribute for a particular external link.
1902 *
1903 * @since 1.21
1904 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1905 * nofollow if appropriate
1906 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1907 * @return string|null Rel attribute for $url
1908 */
1909 public static function getExternalLinkRel( $url = false, $title = null ) {
1910 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1911 $ns = $title ? $title->getNamespace() : false;
1912 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1913 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1914 ) {
1915 return 'nofollow';
1916 }
1917 return null;
1918 }
1919
1920 /**
1921 * Get an associative array of additional HTML attributes appropriate for a
1922 * particular external link. This currently may include rel => nofollow
1923 * (depending on configuration, namespace, and the URL's domain) and/or a
1924 * target attribute (depending on configuration).
1925 *
1926 * @param string $url URL to extract the domain from for rel =>
1927 * nofollow if appropriate
1928 * @return array Associative array of HTML attributes
1929 */
1930 public function getExternalLinkAttribs( $url ) {
1931 $attribs = [];
1932 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1933
1934 $target = $this->mOptions->getExternalLinkTarget();
1935 if ( $target ) {
1936 $attribs['target'] = $target;
1937 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1938 // T133507. New windows can navigate parent cross-origin.
1939 // Including noreferrer due to lacking browser
1940 // support of noopener. Eventually noreferrer should be removed.
1941 if ( $rel !== '' ) {
1942 $rel .= ' ';
1943 }
1944 $rel .= 'noreferrer noopener';
1945 }
1946 }
1947 $attribs['rel'] = $rel;
1948 return $attribs;
1949 }
1950
1951 /**
1952 * Replace unusual escape codes in a URL with their equivalent characters
1953 *
1954 * This generally follows the syntax defined in RFC 3986, with special
1955 * consideration for HTTP query strings.
1956 *
1957 * @param string $url
1958 * @return string
1959 */
1960 public static function normalizeLinkUrl( $url ) {
1961 # First, make sure unsafe characters are encoded
1962 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1963 function ( $m ) {
1964 return rawurlencode( $m[0] );
1965 },
1966 $url
1967 );
1968
1969 $ret = '';
1970 $end = strlen( $url );
1971
1972 # Fragment part - 'fragment'
1973 $start = strpos( $url, '#' );
1974 if ( $start !== false && $start < $end ) {
1975 $ret = self::normalizeUrlComponent(
1976 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1977 $end = $start;
1978 }
1979
1980 # Query part - 'query' minus &=+;
1981 $start = strpos( $url, '?' );
1982 if ( $start !== false && $start < $end ) {
1983 $ret = self::normalizeUrlComponent(
1984 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1985 $end = $start;
1986 }
1987
1988 # Scheme and path part - 'pchar'
1989 # (we assume no userinfo or encoded colons in the host)
1990 $ret = self::normalizeUrlComponent(
1991 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1992
1993 return $ret;
1994 }
1995
1996 private static function normalizeUrlComponent( $component, $unsafe ) {
1997 $callback = function ( $matches ) use ( $unsafe ) {
1998 $char = urldecode( $matches[0] );
1999 $ord = ord( $char );
2000 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2001 # Unescape it
2002 return $char;
2003 } else {
2004 # Leave it escaped, but use uppercase for a-f
2005 return strtoupper( $matches[0] );
2006 }
2007 };
2008 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2009 }
2010
2011 /**
2012 * make an image if it's allowed, either through the global
2013 * option, through the exception, or through the on-wiki whitelist
2014 *
2015 * @param string $url
2016 *
2017 * @return string
2018 */
2019 private function maybeMakeExternalImage( $url ) {
2020 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2021 $imagesexception = !empty( $imagesfrom );
2022 $text = false;
2023 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2024 if ( $imagesexception && is_array( $imagesfrom ) ) {
2025 $imagematch = false;
2026 foreach ( $imagesfrom as $match ) {
2027 if ( strpos( $url, $match ) === 0 ) {
2028 $imagematch = true;
2029 break;
2030 }
2031 }
2032 } elseif ( $imagesexception ) {
2033 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2034 } else {
2035 $imagematch = false;
2036 }
2037
2038 if ( $this->mOptions->getAllowExternalImages()
2039 || ( $imagesexception && $imagematch )
2040 ) {
2041 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2042 # Image found
2043 $text = Linker::makeExternalImage( $url );
2044 }
2045 }
2046 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2047 && preg_match( self::EXT_IMAGE_REGEX, $url )
2048 ) {
2049 $whitelist = explode(
2050 "\n",
2051 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2052 );
2053
2054 foreach ( $whitelist as $entry ) {
2055 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2056 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2057 continue;
2058 }
2059 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2060 # Image matches a whitelist entry
2061 $text = Linker::makeExternalImage( $url );
2062 break;
2063 }
2064 }
2065 }
2066 return $text;
2067 }
2068
2069 /**
2070 * Process [[ ]] wikilinks
2071 *
2072 * @param string $s
2073 *
2074 * @return string Processed text
2075 *
2076 * @private
2077 */
2078 public function replaceInternalLinks( $s ) {
2079 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2080 return $s;
2081 }
2082
2083 /**
2084 * Process [[ ]] wikilinks (RIL)
2085 * @param string &$s
2086 * @throws MWException
2087 * @return LinkHolderArray
2088 *
2089 * @private
2090 */
2091 public function replaceInternalLinks2( &$s ) {
2092 global $wgExtraInterlanguageLinkPrefixes;
2093
2094 static $tc = false, $e1, $e1_img;
2095 # the % is needed to support urlencoded titles as well
2096 if ( !$tc ) {
2097 $tc = Title::legalChars() . '#%';
2098 # Match a link having the form [[namespace:link|alternate]]trail
2099 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2100 # Match cases where there is no "]]", which might still be images
2101 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2102 }
2103
2104 $holders = new LinkHolderArray( $this );
2105
2106 # split the entire text string on occurrences of [[
2107 $a = StringUtils::explode( '[[', ' ' . $s );
2108 # get the first element (all text up to first [[), and remove the space we added
2109 $s = $a->current();
2110 $a->next();
2111 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2112 $s = substr( $s, 1 );
2113
2114 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2115 $e2 = null;
2116 if ( $useLinkPrefixExtension ) {
2117 # Match the end of a line for a word that's not followed by whitespace,
2118 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2119 global $wgContLang;
2120 $charset = $wgContLang->linkPrefixCharset();
2121 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2122 }
2123
2124 if ( is_null( $this->mTitle ) ) {
2125 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2126 }
2127 $nottalk = !$this->mTitle->isTalkPage();
2128
2129 if ( $useLinkPrefixExtension ) {
2130 $m = [];
2131 if ( preg_match( $e2, $s, $m ) ) {
2132 $first_prefix = $m[2];
2133 } else {
2134 $first_prefix = false;
2135 }
2136 } else {
2137 $prefix = '';
2138 }
2139
2140 $useSubpages = $this->areSubpagesAllowed();
2141
2142 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2143 # Loop for each link
2144 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2145 // @codingStandardsIgnoreEnd
2146
2147 # Check for excessive memory usage
2148 if ( $holders->isBig() ) {
2149 # Too big
2150 # Do the existence check, replace the link holders and clear the array
2151 $holders->replace( $s );
2152 $holders->clear();
2153 }
2154
2155 if ( $useLinkPrefixExtension ) {
2156 if ( preg_match( $e2, $s, $m ) ) {
2157 $prefix = $m[2];
2158 $s = $m[1];
2159 } else {
2160 $prefix = '';
2161 }
2162 # first link
2163 if ( $first_prefix ) {
2164 $prefix = $first_prefix;
2165 $first_prefix = false;
2166 }
2167 }
2168
2169 $might_be_img = false;
2170
2171 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2172 $text = $m[2];
2173 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2174 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2175 # the real problem is with the $e1 regex
2176 # See T1500.
2177 # Still some problems for cases where the ] is meant to be outside punctuation,
2178 # and no image is in sight. See T4095.
2179 if ( $text !== ''
2180 && substr( $m[3], 0, 1 ) === ']'
2181 && strpos( $text, '[' ) !== false
2182 ) {
2183 $text .= ']'; # so that replaceExternalLinks($text) works later
2184 $m[3] = substr( $m[3], 1 );
2185 }
2186 # fix up urlencoded title texts
2187 if ( strpos( $m[1], '%' ) !== false ) {
2188 # Should anchors '#' also be rejected?
2189 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2190 }
2191 $trail = $m[3];
2192 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2193 # Invalid, but might be an image with a link in its caption
2194 $might_be_img = true;
2195 $text = $m[2];
2196 if ( strpos( $m[1], '%' ) !== false ) {
2197 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2198 }
2199 $trail = "";
2200 } else { # Invalid form; output directly
2201 $s .= $prefix . '[[' . $line;
2202 continue;
2203 }
2204
2205 $origLink = ltrim( $m[1], ' ' );
2206
2207 # Don't allow internal links to pages containing
2208 # PROTO: where PROTO is a valid URL protocol; these
2209 # should be external links.
2210 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2211 $s .= $prefix . '[[' . $line;
2212 continue;
2213 }
2214
2215 # Make subpage if necessary
2216 if ( $useSubpages ) {
2217 $link = $this->maybeDoSubpageLink( $origLink, $text );
2218 } else {
2219 $link = $origLink;
2220 }
2221
2222 $unstrip = $this->mStripState->unstripNoWiki( $link );
2223 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2224 if ( $nt === null ) {
2225 $s .= $prefix . '[[' . $line;
2226 continue;
2227 }
2228
2229 $ns = $nt->getNamespace();
2230 $iw = $nt->getInterwiki();
2231
2232 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2233
2234 if ( $might_be_img ) { # if this is actually an invalid link
2235 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2236 $found = false;
2237 while ( true ) {
2238 # look at the next 'line' to see if we can close it there
2239 $a->next();
2240 $next_line = $a->current();
2241 if ( $next_line === false || $next_line === null ) {
2242 break;
2243 }
2244 $m = explode( ']]', $next_line, 3 );
2245 if ( count( $m ) == 3 ) {
2246 # the first ]] closes the inner link, the second the image
2247 $found = true;
2248 $text .= "[[{$m[0]}]]{$m[1]}";
2249 $trail = $m[2];
2250 break;
2251 } elseif ( count( $m ) == 2 ) {
2252 # if there's exactly one ]] that's fine, we'll keep looking
2253 $text .= "[[{$m[0]}]]{$m[1]}";
2254 } else {
2255 # if $next_line is invalid too, we need look no further
2256 $text .= '[[' . $next_line;
2257 break;
2258 }
2259 }
2260 if ( !$found ) {
2261 # we couldn't find the end of this imageLink, so output it raw
2262 # but don't ignore what might be perfectly normal links in the text we've examined
2263 $holders->merge( $this->replaceInternalLinks2( $text ) );
2264 $s .= "{$prefix}[[$link|$text";
2265 # note: no $trail, because without an end, there *is* no trail
2266 continue;
2267 }
2268 } else { # it's not an image, so output it raw
2269 $s .= "{$prefix}[[$link|$text";
2270 # note: no $trail, because without an end, there *is* no trail
2271 continue;
2272 }
2273 }
2274
2275 $wasblank = ( $text == '' );
2276 if ( $wasblank ) {
2277 $text = $link;
2278 if ( !$noforce ) {
2279 # Strip off leading ':'
2280 $text = substr( $text, 1 );
2281 }
2282 } else {
2283 # T6598 madness. Handle the quotes only if they come from the alternate part
2284 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2285 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2286 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2287 $text = $this->doQuotes( $text );
2288 }
2289
2290 # Link not escaped by : , create the various objects
2291 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2292 # Interwikis
2293 if (
2294 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2295 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2296 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2297 )
2298 ) {
2299 # T26502: filter duplicates
2300 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2301 $this->mLangLinkLanguages[$iw] = true;
2302 $this->mOutput->addLanguageLink( $nt->getFullText() );
2303 }
2304
2305 /**
2306 * Strip the whitespace interwiki links produce, see T10897
2307 */
2308 $s = rtrim( $s . $prefix ) . $trail; # T175416
2309 continue;
2310 }
2311
2312 if ( $ns == NS_FILE ) {
2313 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2314 if ( $wasblank ) {
2315 # if no parameters were passed, $text
2316 # becomes something like "File:Foo.png",
2317 # which we don't want to pass on to the
2318 # image generator
2319 $text = '';
2320 } else {
2321 # recursively parse links inside the image caption
2322 # actually, this will parse them in any other parameters, too,
2323 # but it might be hard to fix that, and it doesn't matter ATM
2324 $text = $this->replaceExternalLinks( $text );
2325 $holders->merge( $this->replaceInternalLinks2( $text ) );
2326 }
2327 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2328 $s .= $prefix . $this->armorLinks(
2329 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2330 continue;
2331 }
2332 } elseif ( $ns == NS_CATEGORY ) {
2333 /**
2334 * Strip the whitespace Category links produce, see T2087
2335 */
2336 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2337
2338 if ( $wasblank ) {
2339 $sortkey = $this->getDefaultSort();
2340 } else {
2341 $sortkey = $text;
2342 }
2343 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2344 $sortkey = str_replace( "\n", '', $sortkey );
2345 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2346 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2347
2348 continue;
2349 }
2350 }
2351
2352 # Self-link checking. For some languages, variants of the title are checked in
2353 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2354 # for linking to a different variant.
2355 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2356 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2357 continue;
2358 }
2359
2360 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2361 # @todo FIXME: Should do batch file existence checks, see comment below
2362 if ( $ns == NS_MEDIA ) {
2363 # Give extensions a chance to select the file revision for us
2364 $options = [];
2365 $descQuery = false;
2366 Hooks::run( 'BeforeParserFetchFileAndTitle',
2367 [ $this, $nt, &$options, &$descQuery ] );
2368 # Fetch and register the file (file title may be different via hooks)
2369 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2370 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2371 $s .= $prefix . $this->armorLinks(
2372 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2373 continue;
2374 }
2375
2376 # Some titles, such as valid special pages or files in foreign repos, should
2377 # be shown as bluelinks even though they're not included in the page table
2378 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2379 # batch file existence checks for NS_FILE and NS_MEDIA
2380 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2381 $this->mOutput->addLink( $nt );
2382 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2383 } else {
2384 # Links will be added to the output link list after checking
2385 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2386 }
2387 }
2388 return $holders;
2389 }
2390
2391 /**
2392 * Render a forced-blue link inline; protect against double expansion of
2393 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2394 * Since this little disaster has to split off the trail text to avoid
2395 * breaking URLs in the following text without breaking trails on the
2396 * wiki links, it's been made into a horrible function.
2397 *
2398 * @param Title $nt
2399 * @param string $text
2400 * @param string $trail
2401 * @param string $prefix
2402 * @return string HTML-wikitext mix oh yuck
2403 */
2404 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2405 list( $inside, $trail ) = Linker::splitTrail( $trail );
2406
2407 if ( $text == '' ) {
2408 $text = htmlspecialchars( $nt->getPrefixedText() );
2409 }
2410
2411 $link = $this->getLinkRenderer()->makeKnownLink(
2412 $nt, new HtmlArmor( "$prefix$text$inside" )
2413 );
2414
2415 return $this->armorLinks( $link ) . $trail;
2416 }
2417
2418 /**
2419 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2420 * going to go through further parsing steps before inline URL expansion.
2421 *
2422 * Not needed quite as much as it used to be since free links are a bit
2423 * more sensible these days. But bracketed links are still an issue.
2424 *
2425 * @param string $text More-or-less HTML
2426 * @return string Less-or-more HTML with NOPARSE bits
2427 */
2428 public function armorLinks( $text ) {
2429 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2430 self::MARKER_PREFIX . "NOPARSE$1", $text );
2431 }
2432
2433 /**
2434 * Return true if subpage links should be expanded on this page.
2435 * @return bool
2436 */
2437 public function areSubpagesAllowed() {
2438 # Some namespaces don't allow subpages
2439 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2440 }
2441
2442 /**
2443 * Handle link to subpage if necessary
2444 *
2445 * @param string $target The source of the link
2446 * @param string &$text The link text, modified as necessary
2447 * @return string The full name of the link
2448 * @private
2449 */
2450 public function maybeDoSubpageLink( $target, &$text ) {
2451 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2452 }
2453
2454 /**
2455 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2456 *
2457 * @param string $text
2458 * @param bool $linestart Whether or not this is at the start of a line.
2459 * @private
2460 * @return string The lists rendered as HTML
2461 */
2462 public function doBlockLevels( $text, $linestart ) {
2463 return BlockLevelPass::doBlockLevels( $text, $linestart );
2464 }
2465
2466 /**
2467 * Return value of a magic variable (like PAGENAME)
2468 *
2469 * @private
2470 *
2471 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2472 * @param bool|PPFrame $frame
2473 *
2474 * @throws MWException
2475 * @return string
2476 */
2477 public function getVariableValue( $index, $frame = false ) {
2478 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2479 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2480
2481 if ( is_null( $this->mTitle ) ) {
2482 // If no title set, bad things are going to happen
2483 // later. Title should always be set since this
2484 // should only be called in the middle of a parse
2485 // operation (but the unit-tests do funky stuff)
2486 throw new MWException( __METHOD__ . ' Should only be '
2487 . ' called while parsing (no title set)' );
2488 }
2489
2490 // Avoid PHP 7.1 warning from passing $this by reference
2491 $parser = $this;
2492
2493 /**
2494 * Some of these require message or data lookups and can be
2495 * expensive to check many times.
2496 */
2497 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2498 if ( isset( $this->mVarCache[$index] ) ) {
2499 return $this->mVarCache[$index];
2500 }
2501 }
2502
2503 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2504 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2505
2506 $pageLang = $this->getFunctionLang();
2507
2508 switch ( $index ) {
2509 case '!':
2510 $value = '|';
2511 break;
2512 case 'currentmonth':
2513 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2514 break;
2515 case 'currentmonth1':
2516 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2517 break;
2518 case 'currentmonthname':
2519 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2520 break;
2521 case 'currentmonthnamegen':
2522 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2523 break;
2524 case 'currentmonthabbrev':
2525 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2526 break;
2527 case 'currentday':
2528 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2529 break;
2530 case 'currentday2':
2531 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2532 break;
2533 case 'localmonth':
2534 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2535 break;
2536 case 'localmonth1':
2537 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2538 break;
2539 case 'localmonthname':
2540 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2541 break;
2542 case 'localmonthnamegen':
2543 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2544 break;
2545 case 'localmonthabbrev':
2546 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2547 break;
2548 case 'localday':
2549 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2550 break;
2551 case 'localday2':
2552 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2553 break;
2554 case 'pagename':
2555 $value = wfEscapeWikiText( $this->mTitle->getText() );
2556 break;
2557 case 'pagenamee':
2558 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2559 break;
2560 case 'fullpagename':
2561 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2562 break;
2563 case 'fullpagenamee':
2564 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2565 break;
2566 case 'subpagename':
2567 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2568 break;
2569 case 'subpagenamee':
2570 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2571 break;
2572 case 'rootpagename':
2573 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2574 break;
2575 case 'rootpagenamee':
2576 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2577 ' ',
2578 '_',
2579 $this->mTitle->getRootText()
2580 ) ) );
2581 break;
2582 case 'basepagename':
2583 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2584 break;
2585 case 'basepagenamee':
2586 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2587 ' ',
2588 '_',
2589 $this->mTitle->getBaseText()
2590 ) ) );
2591 break;
2592 case 'talkpagename':
2593 if ( $this->mTitle->canHaveTalkPage() ) {
2594 $talkPage = $this->mTitle->getTalkPage();
2595 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2596 } else {
2597 $value = '';
2598 }
2599 break;
2600 case 'talkpagenamee':
2601 if ( $this->mTitle->canHaveTalkPage() ) {
2602 $talkPage = $this->mTitle->getTalkPage();
2603 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2604 } else {
2605 $value = '';
2606 }
2607 break;
2608 case 'subjectpagename':
2609 $subjPage = $this->mTitle->getSubjectPage();
2610 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2611 break;
2612 case 'subjectpagenamee':
2613 $subjPage = $this->mTitle->getSubjectPage();
2614 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2615 break;
2616 case 'pageid': // requested in T25427
2617 $pageid = $this->getTitle()->getArticleID();
2618 if ( $pageid == 0 ) {
2619 # 0 means the page doesn't exist in the database,
2620 # which means the user is previewing a new page.
2621 # The vary-revision flag must be set, because the magic word
2622 # will have a different value once the page is saved.
2623 $this->mOutput->setFlag( 'vary-revision' );
2624 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2625 }
2626 $value = $pageid ? $pageid : null;
2627 break;
2628 case 'revisionid':
2629 # Let the edit saving system know we should parse the page
2630 # *after* a revision ID has been assigned.
2631 $this->mOutput->setFlag( 'vary-revision-id' );
2632 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2633 $value = $this->mRevisionId;
2634 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2635 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2636 $this->mOutput->setSpeculativeRevIdUsed( $value );
2637 }
2638 break;
2639 case 'revisionday':
2640 # Let the edit saving system know we should parse the page
2641 # *after* a revision ID has been assigned. This is for null edits.
2642 $this->mOutput->setFlag( 'vary-revision' );
2643 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2644 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2645 break;
2646 case 'revisionday2':
2647 # Let the edit saving system know we should parse the page
2648 # *after* a revision ID has been assigned. This is for null edits.
2649 $this->mOutput->setFlag( 'vary-revision' );
2650 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2651 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2652 break;
2653 case 'revisionmonth':
2654 # Let the edit saving system know we should parse the page
2655 # *after* a revision ID has been assigned. This is for null edits.
2656 $this->mOutput->setFlag( 'vary-revision' );
2657 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2658 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2659 break;
2660 case 'revisionmonth1':
2661 # Let the edit saving system know we should parse the page
2662 # *after* a revision ID has been assigned. This is for null edits.
2663 $this->mOutput->setFlag( 'vary-revision' );
2664 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2665 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2666 break;
2667 case 'revisionyear':
2668 # Let the edit saving system know we should parse the page
2669 # *after* a revision ID has been assigned. This is for null edits.
2670 $this->mOutput->setFlag( 'vary-revision' );
2671 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2672 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2673 break;
2674 case 'revisiontimestamp':
2675 # Let the edit saving system know we should parse the page
2676 # *after* a revision ID has been assigned. This is for null edits.
2677 $this->mOutput->setFlag( 'vary-revision' );
2678 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2679 $value = $this->getRevisionTimestamp();
2680 break;
2681 case 'revisionuser':
2682 # Let the edit saving system know we should parse the page
2683 # *after* a revision ID has been assigned for null edits.
2684 $this->mOutput->setFlag( 'vary-user' );
2685 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2686 $value = $this->getRevisionUser();
2687 break;
2688 case 'revisionsize':
2689 $value = $this->getRevisionSize();
2690 break;
2691 case 'namespace':
2692 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2693 break;
2694 case 'namespacee':
2695 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2696 break;
2697 case 'namespacenumber':
2698 $value = $this->mTitle->getNamespace();
2699 break;
2700 case 'talkspace':
2701 $value = $this->mTitle->canHaveTalkPage()
2702 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2703 : '';
2704 break;
2705 case 'talkspacee':
2706 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2707 break;
2708 case 'subjectspace':
2709 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2710 break;
2711 case 'subjectspacee':
2712 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2713 break;
2714 case 'currentdayname':
2715 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2716 break;
2717 case 'currentyear':
2718 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2719 break;
2720 case 'currenttime':
2721 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2722 break;
2723 case 'currenthour':
2724 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2725 break;
2726 case 'currentweek':
2727 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2728 # int to remove the padding
2729 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2730 break;
2731 case 'currentdow':
2732 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2733 break;
2734 case 'localdayname':
2735 $value = $pageLang->getWeekdayName(
2736 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2737 );
2738 break;
2739 case 'localyear':
2740 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2741 break;
2742 case 'localtime':
2743 $value = $pageLang->time(
2744 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2745 false,
2746 false
2747 );
2748 break;
2749 case 'localhour':
2750 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2751 break;
2752 case 'localweek':
2753 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2754 # int to remove the padding
2755 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2756 break;
2757 case 'localdow':
2758 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2759 break;
2760 case 'numberofarticles':
2761 $value = $pageLang->formatNum( SiteStats::articles() );
2762 break;
2763 case 'numberoffiles':
2764 $value = $pageLang->formatNum( SiteStats::images() );
2765 break;
2766 case 'numberofusers':
2767 $value = $pageLang->formatNum( SiteStats::users() );
2768 break;
2769 case 'numberofactiveusers':
2770 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2771 break;
2772 case 'numberofpages':
2773 $value = $pageLang->formatNum( SiteStats::pages() );
2774 break;
2775 case 'numberofadmins':
2776 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2777 break;
2778 case 'numberofedits':
2779 $value = $pageLang->formatNum( SiteStats::edits() );
2780 break;
2781 case 'currenttimestamp':
2782 $value = wfTimestamp( TS_MW, $ts );
2783 break;
2784 case 'localtimestamp':
2785 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2786 break;
2787 case 'currentversion':
2788 $value = SpecialVersion::getVersion();
2789 break;
2790 case 'articlepath':
2791 return $wgArticlePath;
2792 case 'sitename':
2793 return $wgSitename;
2794 case 'server':
2795 return $wgServer;
2796 case 'servername':
2797 return $wgServerName;
2798 case 'scriptpath':
2799 return $wgScriptPath;
2800 case 'stylepath':
2801 return $wgStylePath;
2802 case 'directionmark':
2803 return $pageLang->getDirMark();
2804 case 'contentlanguage':
2805 global $wgLanguageCode;
2806 return $wgLanguageCode;
2807 case 'pagelanguage':
2808 $value = $pageLang->getCode();
2809 break;
2810 case 'cascadingsources':
2811 $value = CoreParserFunctions::cascadingsources( $this );
2812 break;
2813 default:
2814 $ret = null;
2815 Hooks::run(
2816 'ParserGetVariableValueSwitch',
2817 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2818 );
2819
2820 return $ret;
2821 }
2822
2823 if ( $index ) {
2824 $this->mVarCache[$index] = $value;
2825 }
2826
2827 return $value;
2828 }
2829
2830 /**
2831 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2832 *
2833 * @private
2834 */
2835 public function initialiseVariables() {
2836 $variableIDs = MagicWord::getVariableIDs();
2837 $substIDs = MagicWord::getSubstIDs();
2838
2839 $this->mVariables = new MagicWordArray( $variableIDs );
2840 $this->mSubstWords = new MagicWordArray( $substIDs );
2841 }
2842
2843 /**
2844 * Preprocess some wikitext and return the document tree.
2845 * This is the ghost of replace_variables().
2846 *
2847 * @param string $text The text to parse
2848 * @param int $flags Bitwise combination of:
2849 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2850 * included. Default is to assume a direct page view.
2851 *
2852 * The generated DOM tree must depend only on the input text and the flags.
2853 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2854 *
2855 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2856 * change in the DOM tree for a given text, must be passed through the section identifier
2857 * in the section edit link and thus back to extractSections().
2858 *
2859 * The output of this function is currently only cached in process memory, but a persistent
2860 * cache may be implemented at a later date which takes further advantage of these strict
2861 * dependency requirements.
2862 *
2863 * @return PPNode
2864 */
2865 public function preprocessToDom( $text, $flags = 0 ) {
2866 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2867 return $dom;
2868 }
2869
2870 /**
2871 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2872 *
2873 * @param string $s
2874 *
2875 * @return array
2876 */
2877 public static function splitWhitespace( $s ) {
2878 $ltrimmed = ltrim( $s );
2879 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2880 $trimmed = rtrim( $ltrimmed );
2881 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2882 if ( $diff > 0 ) {
2883 $w2 = substr( $ltrimmed, -$diff );
2884 } else {
2885 $w2 = '';
2886 }
2887 return [ $w1, $trimmed, $w2 ];
2888 }
2889
2890 /**
2891 * Replace magic variables, templates, and template arguments
2892 * with the appropriate text. Templates are substituted recursively,
2893 * taking care to avoid infinite loops.
2894 *
2895 * Note that the substitution depends on value of $mOutputType:
2896 * self::OT_WIKI: only {{subst:}} templates
2897 * self::OT_PREPROCESS: templates but not extension tags
2898 * self::OT_HTML: all templates and extension tags
2899 *
2900 * @param string $text The text to transform
2901 * @param bool|PPFrame $frame Object describing the arguments passed to the
2902 * template. Arguments may also be provided as an associative array, as
2903 * was the usual case before MW1.12. Providing arguments this way may be
2904 * useful for extensions wishing to perform variable replacement
2905 * explicitly.
2906 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2907 * double-brace expansion.
2908 * @return string
2909 */
2910 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2911 # Is there any text? Also, Prevent too big inclusions!
2912 $textSize = strlen( $text );
2913 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2914 return $text;
2915 }
2916
2917 if ( $frame === false ) {
2918 $frame = $this->getPreprocessor()->newFrame();
2919 } elseif ( !( $frame instanceof PPFrame ) ) {
2920 wfDebug( __METHOD__ . " called using plain parameters instead of "
2921 . "a PPFrame instance. Creating custom frame.\n" );
2922 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2923 }
2924
2925 $dom = $this->preprocessToDom( $text );
2926 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2927 $text = $frame->expand( $dom, $flags );
2928
2929 return $text;
2930 }
2931
2932 /**
2933 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2934 *
2935 * @param array $args
2936 *
2937 * @return array
2938 */
2939 public static function createAssocArgs( $args ) {
2940 $assocArgs = [];
2941 $index = 1;
2942 foreach ( $args as $arg ) {
2943 $eqpos = strpos( $arg, '=' );
2944 if ( $eqpos === false ) {
2945 $assocArgs[$index++] = $arg;
2946 } else {
2947 $name = trim( substr( $arg, 0, $eqpos ) );
2948 $value = trim( substr( $arg, $eqpos + 1 ) );
2949 if ( $value === false ) {
2950 $value = '';
2951 }
2952 if ( $name !== false ) {
2953 $assocArgs[$name] = $value;
2954 }
2955 }
2956 }
2957
2958 return $assocArgs;
2959 }
2960
2961 /**
2962 * Warn the user when a parser limitation is reached
2963 * Will warn at most once the user per limitation type
2964 *
2965 * The results are shown during preview and run through the Parser (See EditPage.php)
2966 *
2967 * @param string $limitationType Should be one of:
2968 * 'expensive-parserfunction' (corresponding messages:
2969 * 'expensive-parserfunction-warning',
2970 * 'expensive-parserfunction-category')
2971 * 'post-expand-template-argument' (corresponding messages:
2972 * 'post-expand-template-argument-warning',
2973 * 'post-expand-template-argument-category')
2974 * 'post-expand-template-inclusion' (corresponding messages:
2975 * 'post-expand-template-inclusion-warning',
2976 * 'post-expand-template-inclusion-category')
2977 * 'node-count-exceeded' (corresponding messages:
2978 * 'node-count-exceeded-warning',
2979 * 'node-count-exceeded-category')
2980 * 'expansion-depth-exceeded' (corresponding messages:
2981 * 'expansion-depth-exceeded-warning',
2982 * 'expansion-depth-exceeded-category')
2983 * @param string|int|null $current Current value
2984 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2985 * exceeded, provide the values (optional)
2986 */
2987 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2988 # does no harm if $current and $max are present but are unnecessary for the message
2989 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2990 # only during preview, and that would split the parser cache unnecessarily.
2991 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2992 ->text();
2993 $this->mOutput->addWarning( $warning );
2994 $this->addTrackingCategory( "$limitationType-category" );
2995 }
2996
2997 /**
2998 * Return the text of a template, after recursively
2999 * replacing any variables or templates within the template.
3000 *
3001 * @param array $piece The parts of the template
3002 * $piece['title']: the title, i.e. the part before the |
3003 * $piece['parts']: the parameter array
3004 * $piece['lineStart']: whether the brace was at the start of a line
3005 * @param PPFrame $frame The current frame, contains template arguments
3006 * @throws Exception
3007 * @return string The text of the template
3008 */
3009 public function braceSubstitution( $piece, $frame ) {
3010 // Flags
3011
3012 // $text has been filled
3013 $found = false;
3014 // wiki markup in $text should be escaped
3015 $nowiki = false;
3016 // $text is HTML, armour it against wikitext transformation
3017 $isHTML = false;
3018 // Force interwiki transclusion to be done in raw mode not rendered
3019 $forceRawInterwiki = false;
3020 // $text is a DOM node needing expansion in a child frame
3021 $isChildObj = false;
3022 // $text is a DOM node needing expansion in the current frame
3023 $isLocalObj = false;
3024
3025 # Title object, where $text came from
3026 $title = false;
3027
3028 # $part1 is the bit before the first |, and must contain only title characters.
3029 # Various prefixes will be stripped from it later.
3030 $titleWithSpaces = $frame->expand( $piece['title'] );
3031 $part1 = trim( $titleWithSpaces );
3032 $titleText = false;
3033
3034 # Original title text preserved for various purposes
3035 $originalTitle = $part1;
3036
3037 # $args is a list of argument nodes, starting from index 0, not including $part1
3038 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3039 # below won't work b/c this $args isn't an object
3040 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3041
3042 $profileSection = null; // profile templates
3043
3044 # SUBST
3045 if ( !$found ) {
3046 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3047
3048 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3049 # Decide whether to expand template or keep wikitext as-is.
3050 if ( $this->ot['wiki'] ) {
3051 if ( $substMatch === false ) {
3052 $literal = true; # literal when in PST with no prefix
3053 } else {
3054 $literal = false; # expand when in PST with subst: or safesubst:
3055 }
3056 } else {
3057 if ( $substMatch == 'subst' ) {
3058 $literal = true; # literal when not in PST with plain subst:
3059 } else {
3060 $literal = false; # expand when not in PST with safesubst: or no prefix
3061 }
3062 }
3063 if ( $literal ) {
3064 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3065 $isLocalObj = true;
3066 $found = true;
3067 }
3068 }
3069
3070 # Variables
3071 if ( !$found && $args->getLength() == 0 ) {
3072 $id = $this->mVariables->matchStartToEnd( $part1 );
3073 if ( $id !== false ) {
3074 $text = $this->getVariableValue( $id, $frame );
3075 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3076 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3077 }
3078 $found = true;
3079 }
3080 }
3081
3082 # MSG, MSGNW and RAW
3083 if ( !$found ) {
3084 # Check for MSGNW:
3085 $mwMsgnw = MagicWord::get( 'msgnw' );
3086 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3087 $nowiki = true;
3088 } else {
3089 # Remove obsolete MSG:
3090 $mwMsg = MagicWord::get( 'msg' );
3091 $mwMsg->matchStartAndRemove( $part1 );
3092 }
3093
3094 # Check for RAW:
3095 $mwRaw = MagicWord::get( 'raw' );
3096 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3097 $forceRawInterwiki = true;
3098 }
3099 }
3100
3101 # Parser functions
3102 if ( !$found ) {
3103 $colonPos = strpos( $part1, ':' );
3104 if ( $colonPos !== false ) {
3105 $func = substr( $part1, 0, $colonPos );
3106 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3107 $argsLength = $args->getLength();
3108 for ( $i = 0; $i < $argsLength; $i++ ) {
3109 $funcArgs[] = $args->item( $i );
3110 }
3111 try {
3112 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3113 } catch ( Exception $ex ) {
3114 throw $ex;
3115 }
3116
3117 // Extract any forwarded flags
3118 if ( isset( $result['found'] ) ) {
3119 $found = $result['found'];
3120 }
3121 if ( array_key_exists( 'text', $result ) ) {
3122 // a string or null
3123 $text = $result['text'];
3124 }
3125 if ( isset( $result['nowiki'] ) ) {
3126 $nowiki = $result['nowiki'];
3127 }
3128 if ( isset( $result['isHTML'] ) ) {
3129 $isHTML = $result['isHTML'];
3130 }
3131 if ( isset( $result['forceRawInterwiki'] ) ) {
3132 $forceRawInterwiki = $result['forceRawInterwiki'];
3133 }
3134 if ( isset( $result['isChildObj'] ) ) {
3135 $isChildObj = $result['isChildObj'];
3136 }
3137 if ( isset( $result['isLocalObj'] ) ) {
3138 $isLocalObj = $result['isLocalObj'];
3139 }
3140 }
3141 }
3142
3143 # Finish mangling title and then check for loops.
3144 # Set $title to a Title object and $titleText to the PDBK
3145 if ( !$found ) {
3146 $ns = NS_TEMPLATE;
3147 # Split the title into page and subpage
3148 $subpage = '';
3149 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3150 if ( $part1 !== $relative ) {
3151 $part1 = $relative;
3152 $ns = $this->mTitle->getNamespace();
3153 }
3154 $title = Title::newFromText( $part1, $ns );
3155 if ( $title ) {
3156 $titleText = $title->getPrefixedText();
3157 # Check for language variants if the template is not found
3158 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3159 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3160 }
3161 # Do recursion depth check
3162 $limit = $this->mOptions->getMaxTemplateDepth();
3163 if ( $frame->depth >= $limit ) {
3164 $found = true;
3165 $text = '<span class="error">'
3166 . wfMessage( 'parser-template-recursion-depth-warning' )
3167 ->numParams( $limit )->inContentLanguage()->text()
3168 . '</span>';
3169 }
3170 }
3171 }
3172
3173 # Load from database
3174 if ( !$found && $title ) {
3175 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3176 if ( !$title->isExternal() ) {
3177 if ( $title->isSpecialPage()
3178 && $this->mOptions->getAllowSpecialInclusion()
3179 && $this->ot['html']
3180 ) {
3181 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3182 // Pass the template arguments as URL parameters.
3183 // "uselang" will have no effect since the Language object
3184 // is forced to the one defined in ParserOptions.
3185 $pageArgs = [];
3186 $argsLength = $args->getLength();
3187 for ( $i = 0; $i < $argsLength; $i++ ) {
3188 $bits = $args->item( $i )->splitArg();
3189 if ( strval( $bits['index'] ) === '' ) {
3190 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3191 $value = trim( $frame->expand( $bits['value'] ) );
3192 $pageArgs[$name] = $value;
3193 }
3194 }
3195
3196 // Create a new context to execute the special page
3197 $context = new RequestContext;
3198 $context->setTitle( $title );
3199 $context->setRequest( new FauxRequest( $pageArgs ) );
3200 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3201 $context->setUser( $this->getUser() );
3202 } else {
3203 // If this page is cached, then we better not be per user.
3204 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3205 }
3206 $context->setLanguage( $this->mOptions->getUserLangObj() );
3207 $ret = SpecialPageFactory::capturePath(
3208 $title, $context, $this->getLinkRenderer() );
3209 if ( $ret ) {
3210 $text = $context->getOutput()->getHTML();
3211 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3212 $found = true;
3213 $isHTML = true;
3214 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3215 $this->mOutput->updateRuntimeAdaptiveExpiry(
3216 $specialPage->maxIncludeCacheTime()
3217 );
3218 }
3219 }
3220 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3221 $found = false; # access denied
3222 wfDebug( __METHOD__ . ": template inclusion denied for " .
3223 $title->getPrefixedDBkey() . "\n" );
3224 } else {
3225 list( $text, $title ) = $this->getTemplateDom( $title );
3226 if ( $text !== false ) {
3227 $found = true;
3228 $isChildObj = true;
3229 }
3230 }
3231
3232 # If the title is valid but undisplayable, make a link to it
3233 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3234 $text = "[[:$titleText]]";
3235 $found = true;
3236 }
3237 } elseif ( $title->isTrans() ) {
3238 # Interwiki transclusion
3239 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3240 $text = $this->interwikiTransclude( $title, 'render' );
3241 $isHTML = true;
3242 } else {
3243 $text = $this->interwikiTransclude( $title, 'raw' );
3244 # Preprocess it like a template
3245 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3246 $isChildObj = true;
3247 }
3248 $found = true;
3249 }
3250
3251 # Do infinite loop check
3252 # This has to be done after redirect resolution to avoid infinite loops via redirects
3253 if ( !$frame->loopCheck( $title ) ) {
3254 $found = true;
3255 $text = '<span class="error">'
3256 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3257 . '</span>';
3258 $this->addTrackingCategory( 'template-loop-category' );
3259 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3260 wfEscapeWikiText( $titleText ) )->text() );
3261 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3262 }
3263 }
3264
3265 # If we haven't found text to substitute by now, we're done
3266 # Recover the source wikitext and return it
3267 if ( !$found ) {
3268 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3269 if ( $profileSection ) {
3270 $this->mProfiler->scopedProfileOut( $profileSection );
3271 }
3272 return [ 'object' => $text ];
3273 }
3274
3275 # Expand DOM-style return values in a child frame
3276 if ( $isChildObj ) {
3277 # Clean up argument array
3278 $newFrame = $frame->newChild( $args, $title );
3279
3280 if ( $nowiki ) {
3281 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3282 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3283 # Expansion is eligible for the empty-frame cache
3284 $text = $newFrame->cachedExpand( $titleText, $text );
3285 } else {
3286 # Uncached expansion
3287 $text = $newFrame->expand( $text );
3288 }
3289 }
3290 if ( $isLocalObj && $nowiki ) {
3291 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3292 $isLocalObj = false;
3293 }
3294
3295 if ( $profileSection ) {
3296 $this->mProfiler->scopedProfileOut( $profileSection );
3297 }
3298
3299 # Replace raw HTML by a placeholder
3300 if ( $isHTML ) {
3301 $text = $this->insertStripItem( $text );
3302 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3303 # Escape nowiki-style return values
3304 $text = wfEscapeWikiText( $text );
3305 } elseif ( is_string( $text )
3306 && !$piece['lineStart']
3307 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3308 ) {
3309 # T2529: if the template begins with a table or block-level
3310 # element, it should be treated as beginning a new line.
3311 # This behavior is somewhat controversial.
3312 $text = "\n" . $text;
3313 }
3314
3315 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3316 # Error, oversize inclusion
3317 if ( $titleText !== false ) {
3318 # Make a working, properly escaped link if possible (T25588)
3319 $text = "[[:$titleText]]";
3320 } else {
3321 # This will probably not be a working link, but at least it may
3322 # provide some hint of where the problem is
3323 preg_replace( '/^:/', '', $originalTitle );
3324 $text = "[[:$originalTitle]]";
3325 }
3326 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3327 . 'post-expand include size too large -->' );
3328 $this->limitationWarn( 'post-expand-template-inclusion' );
3329 }
3330
3331 if ( $isLocalObj ) {
3332 $ret = [ 'object' => $text ];
3333 } else {
3334 $ret = [ 'text' => $text ];
3335 }
3336
3337 return $ret;
3338 }
3339
3340 /**
3341 * Call a parser function and return an array with text and flags.
3342 *
3343 * The returned array will always contain a boolean 'found', indicating
3344 * whether the parser function was found or not. It may also contain the
3345 * following:
3346 * text: string|object, resulting wikitext or PP DOM object
3347 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3348 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3349 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3350 * nowiki: bool, wiki markup in $text should be escaped
3351 *
3352 * @since 1.21
3353 * @param PPFrame $frame The current frame, contains template arguments
3354 * @param string $function Function name
3355 * @param array $args Arguments to the function
3356 * @throws MWException
3357 * @return array
3358 */
3359 public function callParserFunction( $frame, $function, array $args = [] ) {
3360 global $wgContLang;
3361
3362 # Case sensitive functions
3363 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3364 $function = $this->mFunctionSynonyms[1][$function];
3365 } else {
3366 # Case insensitive functions
3367 $function = $wgContLang->lc( $function );
3368 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3369 $function = $this->mFunctionSynonyms[0][$function];
3370 } else {
3371 return [ 'found' => false ];
3372 }
3373 }
3374
3375 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3376
3377 // Avoid PHP 7.1 warning from passing $this by reference
3378 $parser = $this;
3379
3380 $allArgs = [ &$parser ];
3381 if ( $flags & self::SFH_OBJECT_ARGS ) {
3382 # Convert arguments to PPNodes and collect for appending to $allArgs
3383 $funcArgs = [];
3384 foreach ( $args as $k => $v ) {
3385 if ( $v instanceof PPNode || $k === 0 ) {
3386 $funcArgs[] = $v;
3387 } else {
3388 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3389 }
3390 }
3391
3392 # Add a frame parameter, and pass the arguments as an array
3393 $allArgs[] = $frame;
3394 $allArgs[] = $funcArgs;
3395 } else {
3396 # Convert arguments to plain text and append to $allArgs
3397 foreach ( $args as $k => $v ) {
3398 if ( $v instanceof PPNode ) {
3399 $allArgs[] = trim( $frame->expand( $v ) );
3400 } elseif ( is_int( $k ) && $k >= 0 ) {
3401 $allArgs[] = trim( $v );
3402 } else {
3403 $allArgs[] = trim( "$k=$v" );
3404 }
3405 }
3406 }
3407
3408 $result = call_user_func_array( $callback, $allArgs );
3409
3410 # The interface for function hooks allows them to return a wikitext
3411 # string or an array containing the string and any flags. This mungs
3412 # things around to match what this method should return.
3413 if ( !is_array( $result ) ) {
3414 $result = [
3415 'found' => true,
3416 'text' => $result,
3417 ];
3418 } else {
3419 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3420 $result['text'] = $result[0];
3421 }
3422 unset( $result[0] );
3423 $result += [
3424 'found' => true,
3425 ];
3426 }
3427
3428 $noparse = true;
3429 $preprocessFlags = 0;
3430 if ( isset( $result['noparse'] ) ) {
3431 $noparse = $result['noparse'];
3432 }
3433 if ( isset( $result['preprocessFlags'] ) ) {
3434 $preprocessFlags = $result['preprocessFlags'];
3435 }
3436
3437 if ( !$noparse ) {
3438 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3439 $result['isChildObj'] = true;
3440 }
3441
3442 return $result;
3443 }
3444
3445 /**
3446 * Get the semi-parsed DOM representation of a template with a given title,
3447 * and its redirect destination title. Cached.
3448 *
3449 * @param Title $title
3450 *
3451 * @return array
3452 */
3453 public function getTemplateDom( $title ) {
3454 $cacheTitle = $title;
3455 $titleText = $title->getPrefixedDBkey();
3456
3457 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3458 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3459 $title = Title::makeTitle( $ns, $dbk );
3460 $titleText = $title->getPrefixedDBkey();
3461 }
3462 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3463 return [ $this->mTplDomCache[$titleText], $title ];
3464 }
3465
3466 # Cache miss, go to the database
3467 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3468
3469 if ( $text === false ) {
3470 $this->mTplDomCache[$titleText] = false;
3471 return [ false, $title ];
3472 }
3473
3474 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3475 $this->mTplDomCache[$titleText] = $dom;
3476
3477 if ( !$title->equals( $cacheTitle ) ) {
3478 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3479 [ $title->getNamespace(), $title->getDBkey() ];
3480 }
3481
3482 return [ $dom, $title ];
3483 }
3484
3485 /**
3486 * Fetch the current revision of a given title. Note that the revision
3487 * (and even the title) may not exist in the database, so everything
3488 * contributing to the output of the parser should use this method
3489 * where possible, rather than getting the revisions themselves. This
3490 * method also caches its results, so using it benefits performance.
3491 *
3492 * @since 1.24
3493 * @param Title $title
3494 * @return Revision
3495 */
3496 public function fetchCurrentRevisionOfTitle( $title ) {
3497 $cacheKey = $title->getPrefixedDBkey();
3498 if ( !$this->currentRevisionCache ) {
3499 $this->currentRevisionCache = new MapCacheLRU( 100 );
3500 }
3501 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3502 $this->currentRevisionCache->set( $cacheKey,
3503 // Defaults to Parser::statelessFetchRevision()
3504 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3505 );
3506 }
3507 return $this->currentRevisionCache->get( $cacheKey );
3508 }
3509
3510 /**
3511 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3512 * without passing them on to it.
3513 *
3514 * @since 1.24
3515 * @param Title $title
3516 * @param Parser|bool $parser
3517 * @return Revision|bool False if missing
3518 */
3519 public static function statelessFetchRevision( Title $title, $parser = false ) {
3520 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3521
3522 return $rev;
3523 }
3524
3525 /**
3526 * Fetch the unparsed text of a template and register a reference to it.
3527 * @param Title $title
3528 * @return array ( string or false, Title )
3529 */
3530 public function fetchTemplateAndTitle( $title ) {
3531 // Defaults to Parser::statelessFetchTemplate()
3532 $templateCb = $this->mOptions->getTemplateCallback();
3533 $stuff = call_user_func( $templateCb, $title, $this );
3534 // We use U+007F DELETE to distinguish strip markers from regular text.
3535 $text = $stuff['text'];
3536 if ( is_string( $stuff['text'] ) ) {
3537 $text = strtr( $text, "\x7f", "?" );
3538 }
3539 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3540 if ( isset( $stuff['deps'] ) ) {
3541 foreach ( $stuff['deps'] as $dep ) {
3542 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3543 if ( $dep['title']->equals( $this->getTitle() ) ) {
3544 // If we transclude ourselves, the final result
3545 // will change based on the new version of the page
3546 $this->mOutput->setFlag( 'vary-revision' );
3547 }
3548 }
3549 }
3550 return [ $text, $finalTitle ];
3551 }
3552
3553 /**
3554 * Fetch the unparsed text of a template and register a reference to it.
3555 * @param Title $title
3556 * @return string|bool
3557 */
3558 public function fetchTemplate( $title ) {
3559 return $this->fetchTemplateAndTitle( $title )[0];
3560 }
3561
3562 /**
3563 * Static function to get a template
3564 * Can be overridden via ParserOptions::setTemplateCallback().
3565 *
3566 * @param Title $title
3567 * @param bool|Parser $parser
3568 *
3569 * @return array
3570 */
3571 public static function statelessFetchTemplate( $title, $parser = false ) {
3572 $text = $skip = false;
3573 $finalTitle = $title;
3574 $deps = [];
3575
3576 # Loop to fetch the article, with up to 1 redirect
3577 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3578 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3579 // @codingStandardsIgnoreEnd
3580 # Give extensions a chance to select the revision instead
3581 $id = false; # Assume current
3582 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3583 [ $parser, $title, &$skip, &$id ] );
3584
3585 if ( $skip ) {
3586 $text = false;
3587 $deps[] = [
3588 'title' => $title,
3589 'page_id' => $title->getArticleID(),
3590 'rev_id' => null
3591 ];
3592 break;
3593 }
3594 # Get the revision
3595 if ( $id ) {
3596 $rev = Revision::newFromId( $id );
3597 } elseif ( $parser ) {
3598 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3599 } else {
3600 $rev = Revision::newFromTitle( $title );
3601 }
3602 $rev_id = $rev ? $rev->getId() : 0;
3603 # If there is no current revision, there is no page
3604 if ( $id === false && !$rev ) {
3605 $linkCache = LinkCache::singleton();
3606 $linkCache->addBadLinkObj( $title );
3607 }
3608
3609 $deps[] = [
3610 'title' => $title,
3611 'page_id' => $title->getArticleID(),
3612 'rev_id' => $rev_id ];
3613 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3614 # We fetched a rev from a different title; register it too...
3615 $deps[] = [
3616 'title' => $rev->getTitle(),
3617 'page_id' => $rev->getPage(),
3618 'rev_id' => $rev_id ];
3619 }
3620
3621 if ( $rev ) {
3622 $content = $rev->getContent();
3623 $text = $content ? $content->getWikitextForTransclusion() : null;
3624
3625 Hooks::run( 'ParserFetchTemplate',
3626 [ $parser, $title, $rev, &$text, &$deps ] );
3627
3628 if ( $text === false || $text === null ) {
3629 $text = false;
3630 break;
3631 }
3632 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3633 global $wgContLang;
3634 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3635 if ( !$message->exists() ) {
3636 $text = false;
3637 break;
3638 }
3639 $content = $message->content();
3640 $text = $message->plain();
3641 } else {
3642 break;
3643 }
3644 if ( !$content ) {
3645 break;
3646 }
3647 # Redirect?
3648 $finalTitle = $title;
3649 $title = $content->getRedirectTarget();
3650 }
3651 return [
3652 'text' => $text,
3653 'finalTitle' => $finalTitle,
3654 'deps' => $deps ];
3655 }
3656
3657 /**
3658 * Fetch a file and its title and register a reference to it.
3659 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3660 * @param Title $title
3661 * @param array $options Array of options to RepoGroup::findFile
3662 * @return File|bool
3663 */
3664 public function fetchFile( $title, $options = [] ) {
3665 return $this->fetchFileAndTitle( $title, $options )[0];
3666 }
3667
3668 /**
3669 * Fetch a file and its title and register a reference to it.
3670 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3671 * @param Title $title
3672 * @param array $options Array of options to RepoGroup::findFile
3673 * @return array ( File or false, Title of file )
3674 */
3675 public function fetchFileAndTitle( $title, $options = [] ) {
3676 $file = $this->fetchFileNoRegister( $title, $options );
3677
3678 $time = $file ? $file->getTimestamp() : false;
3679 $sha1 = $file ? $file->getSha1() : false;
3680 # Register the file as a dependency...
3681 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3682 if ( $file && !$title->equals( $file->getTitle() ) ) {
3683 # Update fetched file title
3684 $title = $file->getTitle();
3685 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3686 }
3687 return [ $file, $title ];
3688 }
3689
3690 /**
3691 * Helper function for fetchFileAndTitle.
3692 *
3693 * Also useful if you need to fetch a file but not use it yet,
3694 * for example to get the file's handler.
3695 *
3696 * @param Title $title
3697 * @param array $options Array of options to RepoGroup::findFile
3698 * @return File|bool
3699 */
3700 protected function fetchFileNoRegister( $title, $options = [] ) {
3701 if ( isset( $options['broken'] ) ) {
3702 $file = false; // broken thumbnail forced by hook
3703 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3704 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3705 } else { // get by (name,timestamp)
3706 $file = wfFindFile( $title, $options );
3707 }
3708 return $file;
3709 }
3710
3711 /**
3712 * Transclude an interwiki link.
3713 *
3714 * @param Title $title
3715 * @param string $action
3716 *
3717 * @return string
3718 */
3719 public function interwikiTransclude( $title, $action ) {
3720 global $wgEnableScaryTranscluding;
3721
3722 if ( !$wgEnableScaryTranscluding ) {
3723 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3724 }
3725
3726 $url = $title->getFullURL( [ 'action' => $action ] );
3727
3728 if ( strlen( $url ) > 255 ) {
3729 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3730 }
3731 return $this->fetchScaryTemplateMaybeFromCache( $url );
3732 }
3733
3734 /**
3735 * @param string $url
3736 * @return mixed|string
3737 */
3738 public function fetchScaryTemplateMaybeFromCache( $url ) {
3739 global $wgTranscludeCacheExpiry;
3740 $dbr = wfGetDB( DB_REPLICA );
3741 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3742 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3743 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3744 if ( $obj ) {
3745 return $obj->tc_contents;
3746 }
3747
3748 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3749 $status = $req->execute(); // Status object
3750 if ( $status->isOK() ) {
3751 $text = $req->getContent();
3752 } elseif ( $req->getStatus() != 200 ) {
3753 // Though we failed to fetch the content, this status is useless.
3754 return wfMessage( 'scarytranscludefailed-httpstatus' )
3755 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3756 } else {
3757 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3758 }
3759
3760 $dbw = wfGetDB( DB_MASTER );
3761 $dbw->replace( 'transcache', [ 'tc_url' ], [
3762 'tc_url' => $url,
3763 'tc_time' => $dbw->timestamp( time() ),
3764 'tc_contents' => $text
3765 ] );
3766 return $text;
3767 }
3768
3769 /**
3770 * Triple brace replacement -- used for template arguments
3771 * @private
3772 *
3773 * @param array $piece
3774 * @param PPFrame $frame
3775 *
3776 * @return array
3777 */
3778 public function argSubstitution( $piece, $frame ) {
3779 $error = false;
3780 $parts = $piece['parts'];
3781 $nameWithSpaces = $frame->expand( $piece['title'] );
3782 $argName = trim( $nameWithSpaces );
3783 $object = false;
3784 $text = $frame->getArgument( $argName );
3785 if ( $text === false && $parts->getLength() > 0
3786 && ( $this->ot['html']
3787 || $this->ot['pre']
3788 || ( $this->ot['wiki'] && $frame->isTemplate() )
3789 )
3790 ) {
3791 # No match in frame, use the supplied default
3792 $object = $parts->item( 0 )->getChildren();
3793 }
3794 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3795 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3796 $this->limitationWarn( 'post-expand-template-argument' );
3797 }
3798
3799 if ( $text === false && $object === false ) {
3800 # No match anywhere
3801 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3802 }
3803 if ( $error !== false ) {
3804 $text .= $error;
3805 }
3806 if ( $object !== false ) {
3807 $ret = [ 'object' => $object ];
3808 } else {
3809 $ret = [ 'text' => $text ];
3810 }
3811
3812 return $ret;
3813 }
3814
3815 /**
3816 * Return the text to be used for a given extension tag.
3817 * This is the ghost of strip().
3818 *
3819 * @param array $params Associative array of parameters:
3820 * name PPNode for the tag name
3821 * attr PPNode for unparsed text where tag attributes are thought to be
3822 * attributes Optional associative array of parsed attributes
3823 * inner Contents of extension element
3824 * noClose Original text did not have a close tag
3825 * @param PPFrame $frame
3826 *
3827 * @throws MWException
3828 * @return string
3829 */
3830 public function extensionSubstitution( $params, $frame ) {
3831 static $errorStr = '<span class="error">';
3832 static $errorLen = 20;
3833
3834 $name = $frame->expand( $params['name'] );
3835 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3836 // Probably expansion depth or node count exceeded. Just punt the
3837 // error up.
3838 return $name;
3839 }
3840
3841 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3842 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3843 // See above
3844 return $attrText;
3845 }
3846
3847 // We can't safely check if the expansion for $content resulted in an
3848 // error, because the content could happen to be the error string
3849 // (T149622).
3850 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3851
3852 $marker = self::MARKER_PREFIX . "-$name-"
3853 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3854
3855 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3856 ( $this->ot['html'] || $this->ot['pre'] );
3857 if ( $isFunctionTag ) {
3858 $markerType = 'none';
3859 } else {
3860 $markerType = 'general';
3861 }
3862 if ( $this->ot['html'] || $isFunctionTag ) {
3863 $name = strtolower( $name );
3864 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3865 if ( isset( $params['attributes'] ) ) {
3866 $attributes = $attributes + $params['attributes'];
3867 }
3868
3869 if ( isset( $this->mTagHooks[$name] ) ) {
3870 $output = call_user_func_array( $this->mTagHooks[$name],
3871 [ $content, $attributes, $this, $frame ] );
3872 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3873 list( $callback, ) = $this->mFunctionTagHooks[$name];
3874
3875 // Avoid PHP 7.1 warning from passing $this by reference
3876 $parser = $this;
3877 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3878 } else {
3879 $output = '<span class="error">Invalid tag extension name: ' .
3880 htmlspecialchars( $name ) . '</span>';
3881 }
3882
3883 if ( is_array( $output ) ) {
3884 // Extract flags
3885 $flags = $output;
3886 $output = $flags[0];
3887 if ( isset( $flags['markerType'] ) ) {
3888 $markerType = $flags['markerType'];
3889 }
3890 }
3891 } else {
3892 if ( is_null( $attrText ) ) {
3893 $attrText = '';
3894 }
3895 if ( isset( $params['attributes'] ) ) {
3896 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3897 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3898 htmlspecialchars( $attrValue ) . '"';
3899 }
3900 }
3901 if ( $content === null ) {
3902 $output = "<$name$attrText/>";
3903 } else {
3904 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3905 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3906 // See above
3907 return $close;
3908 }
3909 $output = "<$name$attrText>$content$close";
3910 }
3911 }
3912
3913 if ( $markerType === 'none' ) {
3914 return $output;
3915 } elseif ( $markerType === 'nowiki' ) {
3916 $this->mStripState->addNoWiki( $marker, $output );
3917 } elseif ( $markerType === 'general' ) {
3918 $this->mStripState->addGeneral( $marker, $output );
3919 } else {
3920 throw new MWException( __METHOD__ . ': invalid marker type' );
3921 }
3922 return $marker;
3923 }
3924
3925 /**
3926 * Increment an include size counter
3927 *
3928 * @param string $type The type of expansion
3929 * @param int $size The size of the text
3930 * @return bool False if this inclusion would take it over the maximum, true otherwise
3931 */
3932 public function incrementIncludeSize( $type, $size ) {
3933 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3934 return false;
3935 } else {
3936 $this->mIncludeSizes[$type] += $size;
3937 return true;
3938 }
3939 }
3940
3941 /**
3942 * Increment the expensive function count
3943 *
3944 * @return bool False if the limit has been exceeded
3945 */
3946 public function incrementExpensiveFunctionCount() {
3947 $this->mExpensiveFunctionCount++;
3948 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3949 }
3950
3951 /**
3952 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3953 * Fills $this->mDoubleUnderscores, returns the modified text
3954 *
3955 * @param string $text
3956 *
3957 * @return string
3958 */
3959 public function doDoubleUnderscore( $text ) {
3960 # The position of __TOC__ needs to be recorded
3961 $mw = MagicWord::get( 'toc' );
3962 if ( $mw->match( $text ) ) {
3963 $this->mShowToc = true;
3964 $this->mForceTocPosition = true;
3965
3966 # Set a placeholder. At the end we'll fill it in with the TOC.
3967 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
3968
3969 # Only keep the first one.
3970 $text = $mw->replace( '', $text );
3971 }
3972
3973 # Now match and remove the rest of them
3974 $mwa = MagicWord::getDoubleUnderscoreArray();
3975 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3976
3977 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3978 $this->mOutput->mNoGallery = true;
3979 }
3980 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3981 $this->mShowToc = false;
3982 }
3983 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3984 && $this->mTitle->getNamespace() == NS_CATEGORY
3985 ) {
3986 $this->addTrackingCategory( 'hidden-category-category' );
3987 }
3988 # (T10068) Allow control over whether robots index a page.
3989 # __INDEX__ always overrides __NOINDEX__, see T16899
3990 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3991 $this->mOutput->setIndexPolicy( 'noindex' );
3992 $this->addTrackingCategory( 'noindex-category' );
3993 }
3994 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3995 $this->mOutput->setIndexPolicy( 'index' );
3996 $this->addTrackingCategory( 'index-category' );
3997 }
3998
3999 # Cache all double underscores in the database
4000 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4001 $this->mOutput->setProperty( $key, '' );
4002 }
4003
4004 return $text;
4005 }
4006
4007 /**
4008 * @see ParserOutput::addTrackingCategory()
4009 * @param string $msg Message key
4010 * @return bool Whether the addition was successful
4011 */
4012 public function addTrackingCategory( $msg ) {
4013 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4014 }
4015
4016 /**
4017 * This function accomplishes several tasks:
4018 * 1) Auto-number headings if that option is enabled
4019 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4020 * 3) Add a Table of contents on the top for users who have enabled the option
4021 * 4) Auto-anchor headings
4022 *
4023 * It loops through all headlines, collects the necessary data, then splits up the
4024 * string and re-inserts the newly formatted headlines.
4025 *
4026 * @param string $text
4027 * @param string $origText Original, untouched wikitext
4028 * @param bool $isMain
4029 * @return mixed|string
4030 * @private
4031 */
4032 public function formatHeadings( $text, $origText, $isMain = true ) {
4033 global $wgMaxTocLevel;
4034
4035 # Inhibit editsection links if requested in the page
4036 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4037 $maybeShowEditLink = $showEditLink = false;
4038 } else {
4039 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4040 $showEditLink = $this->mOptions->getEditSection();
4041 }
4042 if ( $showEditLink ) {
4043 $this->mOutput->setEditSectionTokens( true );
4044 }
4045
4046 # Get all headlines for numbering them and adding funky stuff like [edit]
4047 # links - this is for later, but we need the number of headlines right now
4048 $matches = [];
4049 $numMatches = preg_match_all(
4050 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4051 $text,
4052 $matches
4053 );
4054
4055 # if there are fewer than 4 headlines in the article, do not show TOC
4056 # unless it's been explicitly enabled.
4057 $enoughToc = $this->mShowToc &&
4058 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4059
4060 # Allow user to stipulate that a page should have a "new section"
4061 # link added via __NEWSECTIONLINK__
4062 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4063 $this->mOutput->setNewSection( true );
4064 }
4065
4066 # Allow user to remove the "new section"
4067 # link via __NONEWSECTIONLINK__
4068 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4069 $this->mOutput->hideNewSection( true );
4070 }
4071
4072 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4073 # override above conditions and always show TOC above first header
4074 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4075 $this->mShowToc = true;
4076 $enoughToc = true;
4077 }
4078
4079 # headline counter
4080 $headlineCount = 0;
4081 $numVisible = 0;
4082
4083 # Ugh .. the TOC should have neat indentation levels which can be
4084 # passed to the skin functions. These are determined here
4085 $toc = '';
4086 $full = '';
4087 $head = [];
4088 $sublevelCount = [];
4089 $levelCount = [];
4090 $level = 0;
4091 $prevlevel = 0;
4092 $toclevel = 0;
4093 $prevtoclevel = 0;
4094 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4095 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4096 $oldType = $this->mOutputType;
4097 $this->setOutputType( self::OT_WIKI );
4098 $frame = $this->getPreprocessor()->newFrame();
4099 $root = $this->preprocessToDom( $origText );
4100 $node = $root->getFirstChild();
4101 $byteOffset = 0;
4102 $tocraw = [];
4103 $refers = [];
4104
4105 $headlines = $numMatches !== false ? $matches[3] : [];
4106
4107 foreach ( $headlines as $headline ) {
4108 $isTemplate = false;
4109 $titleText = false;
4110 $sectionIndex = false;
4111 $numbering = '';
4112 $markerMatches = [];
4113 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4114 $serial = $markerMatches[1];
4115 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4116 $isTemplate = ( $titleText != $baseTitleText );
4117 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4118 }
4119
4120 if ( $toclevel ) {
4121 $prevlevel = $level;
4122 }
4123 $level = $matches[1][$headlineCount];
4124
4125 if ( $level > $prevlevel ) {
4126 # Increase TOC level
4127 $toclevel++;
4128 $sublevelCount[$toclevel] = 0;
4129 if ( $toclevel < $wgMaxTocLevel ) {
4130 $prevtoclevel = $toclevel;
4131 $toc .= Linker::tocIndent();
4132 $numVisible++;
4133 }
4134 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4135 # Decrease TOC level, find level to jump to
4136
4137 for ( $i = $toclevel; $i > 0; $i-- ) {
4138 if ( $levelCount[$i] == $level ) {
4139 # Found last matching level
4140 $toclevel = $i;
4141 break;
4142 } elseif ( $levelCount[$i] < $level ) {
4143 # Found first matching level below current level
4144 $toclevel = $i + 1;
4145 break;
4146 }
4147 }
4148 if ( $i == 0 ) {
4149 $toclevel = 1;
4150 }
4151 if ( $toclevel < $wgMaxTocLevel ) {
4152 if ( $prevtoclevel < $wgMaxTocLevel ) {
4153 # Unindent only if the previous toc level was shown :p
4154 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4155 $prevtoclevel = $toclevel;
4156 } else {
4157 $toc .= Linker::tocLineEnd();
4158 }
4159 }
4160 } else {
4161 # No change in level, end TOC line
4162 if ( $toclevel < $wgMaxTocLevel ) {
4163 $toc .= Linker::tocLineEnd();
4164 }
4165 }
4166
4167 $levelCount[$toclevel] = $level;
4168
4169 # count number of headlines for each level
4170 $sublevelCount[$toclevel]++;
4171 $dot = 0;
4172 for ( $i = 1; $i <= $toclevel; $i++ ) {
4173 if ( !empty( $sublevelCount[$i] ) ) {
4174 if ( $dot ) {
4175 $numbering .= '.';
4176 }
4177 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4178 $dot = 1;
4179 }
4180 }
4181
4182 # The safe header is a version of the header text safe to use for links
4183
4184 # Remove link placeholders by the link text.
4185 # <!--LINK number-->
4186 # turns into
4187 # link text with suffix
4188 # Do this before unstrip since link text can contain strip markers
4189 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4190
4191 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4192 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4193
4194 # Strip out HTML (first regex removes any tag not allowed)
4195 # Allowed tags are:
4196 # * <sup> and <sub> (T10393)
4197 # * <i> (T28375)
4198 # * <b> (r105284)
4199 # * <bdi> (T74884)
4200 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4201 # * <s> and <strike> (T35715)
4202 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4203 # to allow setting directionality in toc items.
4204 $tocline = preg_replace(
4205 [
4206 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4207 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4208 ],
4209 [ '', '<$1>' ],
4210 $safeHeadline
4211 );
4212
4213 # Strip '<span></span>', which is the result from the above if
4214 # <span id="foo"></span> is used to produce an additional anchor
4215 # for a section.
4216 $tocline = str_replace( '<span></span>', '', $tocline );
4217
4218 $tocline = trim( $tocline );
4219
4220 # For the anchor, strip out HTML-y stuff period
4221 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4222 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4223
4224 # Save headline for section edit hint before it's escaped
4225 $headlineHint = $safeHeadline;
4226
4227 # Decode HTML entities
4228 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4229
4230 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4231
4232 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4233 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4234 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4235 if ( $fallbackHeadline === $safeHeadline ) {
4236 # No reason to have both (in fact, we can't)
4237 $fallbackHeadline = false;
4238 }
4239
4240 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4241 # @todo FIXME: We may be changing them depending on the current locale.
4242 $arrayKey = strtolower( $safeHeadline );
4243 if ( $fallbackHeadline === false ) {
4244 $fallbackArrayKey = false;
4245 } else {
4246 $fallbackArrayKey = strtolower( $fallbackHeadline );
4247 }
4248
4249 # Create the anchor for linking from the TOC to the section
4250 $anchor = $safeHeadline;
4251 $fallbackAnchor = $fallbackHeadline;
4252 if ( isset( $refers[$arrayKey] ) ) {
4253 // @codingStandardsIgnoreStart
4254 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4255 // @codingStandardsIgnoreEnd
4256 $anchor .= "_$i";
4257 $linkAnchor .= "_$i";
4258 $refers["${arrayKey}_$i"] = true;
4259 } else {
4260 $refers[$arrayKey] = true;
4261 }
4262 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4263 // @codingStandardsIgnoreStart
4264 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4265 // @codingStandardsIgnoreEnd
4266 $fallbackAnchor .= "_$i";
4267 $refers["${fallbackArrayKey}_$i"] = true;
4268 } else {
4269 $refers[$fallbackArrayKey] = true;
4270 }
4271
4272 # Don't number the heading if it is the only one (looks silly)
4273 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4274 # the two are different if the line contains a link
4275 $headline = Html::element(
4276 'span',
4277 [ 'class' => 'mw-headline-number' ],
4278 $numbering
4279 ) . ' ' . $headline;
4280 }
4281
4282 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4283 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4284 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4285 }
4286
4287 # Add the section to the section tree
4288 # Find the DOM node for this header
4289 $noOffset = ( $isTemplate || $sectionIndex === false );
4290 while ( $node && !$noOffset ) {
4291 if ( $node->getName() === 'h' ) {
4292 $bits = $node->splitHeading();
4293 if ( $bits['i'] == $sectionIndex ) {
4294 break;
4295 }
4296 }
4297 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4298 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4299 $node = $node->getNextSibling();
4300 }
4301 $tocraw[] = [
4302 'toclevel' => $toclevel,
4303 'level' => $level,
4304 'line' => $tocline,
4305 'number' => $numbering,
4306 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4307 'fromtitle' => $titleText,
4308 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4309 'anchor' => $anchor,
4310 ];
4311
4312 # give headline the correct <h#> tag
4313 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4314 // Output edit section links as markers with styles that can be customized by skins
4315 if ( $isTemplate ) {
4316 # Put a T flag in the section identifier, to indicate to extractSections()
4317 # that sections inside <includeonly> should be counted.
4318 $editsectionPage = $titleText;
4319 $editsectionSection = "T-$sectionIndex";
4320 $editsectionContent = null;
4321 } else {
4322 $editsectionPage = $this->mTitle->getPrefixedText();
4323 $editsectionSection = $sectionIndex;
4324 $editsectionContent = $headlineHint;
4325 }
4326 // We use a bit of pesudo-xml for editsection markers. The
4327 // language converter is run later on. Using a UNIQ style marker
4328 // leads to the converter screwing up the tokens when it
4329 // converts stuff. And trying to insert strip tags fails too. At
4330 // this point all real inputted tags have already been escaped,
4331 // so we don't have to worry about a user trying to input one of
4332 // these markers directly. We use a page and section attribute
4333 // to stop the language converter from converting these
4334 // important bits of data, but put the headline hint inside a
4335 // content block because the language converter is supposed to
4336 // be able to convert that piece of data.
4337 // Gets replaced with html in ParserOutput::getText
4338 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4339 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4340 if ( $editsectionContent !== null ) {
4341 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4342 } else {
4343 $editlink .= '/>';
4344 }
4345 } else {
4346 $editlink = '';
4347 }
4348 $head[$headlineCount] = Linker::makeHeadline( $level,
4349 $matches['attrib'][$headlineCount], $anchor, $headline,
4350 $editlink, $fallbackAnchor );
4351
4352 $headlineCount++;
4353 }
4354
4355 $this->setOutputType( $oldType );
4356
4357 # Never ever show TOC if no headers
4358 if ( $numVisible < 1 ) {
4359 $enoughToc = false;
4360 }
4361
4362 if ( $enoughToc ) {
4363 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4364 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4365 }
4366 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4367 $this->mOutput->setTOCHTML( $toc );
4368 $toc = self::TOC_START . $toc . self::TOC_END;
4369 }
4370
4371 if ( $isMain ) {
4372 $this->mOutput->setSections( $tocraw );
4373 }
4374
4375 # split up and insert constructed headlines
4376 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4377 $i = 0;
4378
4379 // build an array of document sections
4380 $sections = [];
4381 foreach ( $blocks as $block ) {
4382 // $head is zero-based, sections aren't.
4383 if ( empty( $head[$i - 1] ) ) {
4384 $sections[$i] = $block;
4385 } else {
4386 $sections[$i] = $head[$i - 1] . $block;
4387 }
4388
4389 /**
4390 * Send a hook, one per section.
4391 * The idea here is to be able to make section-level DIVs, but to do so in a
4392 * lower-impact, more correct way than r50769
4393 *
4394 * $this : caller
4395 * $section : the section number
4396 * &$sectionContent : ref to the content of the section
4397 * $showEditLinks : boolean describing whether this section has an edit link
4398 */
4399 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4400
4401 $i++;
4402 }
4403
4404 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4405 // append the TOC at the beginning
4406 // Top anchor now in skin
4407 $sections[0] = $sections[0] . $toc . "\n";
4408 }
4409
4410 $full .= implode( '', $sections );
4411
4412 if ( $this->mForceTocPosition ) {
4413 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4414 } else {
4415 return $full;
4416 }
4417 }
4418
4419 /**
4420 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4421 * conversion, substituting signatures, {{subst:}} templates, etc.
4422 *
4423 * @param string $text The text to transform
4424 * @param Title $title The Title object for the current article
4425 * @param User $user The User object describing the current user
4426 * @param ParserOptions $options Parsing options
4427 * @param bool $clearState Whether to clear the parser state first
4428 * @return string The altered wiki markup
4429 */
4430 public function preSaveTransform( $text, Title $title, User $user,
4431 ParserOptions $options, $clearState = true
4432 ) {
4433 if ( $clearState ) {
4434 $magicScopeVariable = $this->lock();
4435 }
4436 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4437 $this->setUser( $user );
4438
4439 // Strip U+0000 NULL (T159174)
4440 $text = str_replace( "\000", '', $text );
4441
4442 // We still normalize line endings for backwards-compatibility
4443 // with other code that just calls PST, but this should already
4444 // be handled in TextContent subclasses
4445 $text = TextContent::normalizeLineEndings( $text );
4446
4447 if ( $options->getPreSaveTransform() ) {
4448 $text = $this->pstPass2( $text, $user );
4449 }
4450 $text = $this->mStripState->unstripBoth( $text );
4451
4452 $this->setUser( null ); # Reset
4453
4454 return $text;
4455 }
4456
4457 /**
4458 * Pre-save transform helper function
4459 *
4460 * @param string $text
4461 * @param User $user
4462 *
4463 * @return string
4464 */
4465 private function pstPass2( $text, $user ) {
4466 global $wgContLang;
4467
4468 # Note: This is the timestamp saved as hardcoded wikitext to
4469 # the database, we use $wgContLang here in order to give
4470 # everyone the same signature and use the default one rather
4471 # than the one selected in each user's preferences.
4472 # (see also T14815)
4473 $ts = $this->mOptions->getTimestamp();
4474 $timestamp = MWTimestamp::getLocalInstance( $ts );
4475 $ts = $timestamp->format( 'YmdHis' );
4476 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4477
4478 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4479
4480 # Variable replacement
4481 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4482 $text = $this->replaceVariables( $text );
4483
4484 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4485 # which may corrupt this parser instance via its wfMessage()->text() call-
4486
4487 # Signatures
4488 if ( strpos( $text, '~~~' ) !== false ) {
4489 $sigText = $this->getUserSig( $user );
4490 $text = strtr( $text, [
4491 '~~~~~' => $d,
4492 '~~~~' => "$sigText $d",
4493 '~~~' => $sigText
4494 ] );
4495 # The main two signature forms used above are time-sensitive
4496 $this->mOutput->setFlag( 'user-signature' );
4497 }
4498
4499 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4500 $tc = '[' . Title::legalChars() . ']';
4501 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4502
4503 // [[ns:page (context)|]]
4504 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4505 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4506 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4507 // [[ns:page (context), context|]] (using either single or double-width comma)
4508 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4509 // [[|page]] (reverse pipe trick: add context from page title)
4510 $p2 = "/\[\[\\|($tc+)]]/";
4511
4512 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4513 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4514 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4515 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4516
4517 $t = $this->mTitle->getText();
4518 $m = [];
4519 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4520 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4521 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4522 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4523 } else {
4524 # if there's no context, don't bother duplicating the title
4525 $text = preg_replace( $p2, '[[\\1]]', $text );
4526 }
4527
4528 return $text;
4529 }
4530
4531 /**
4532 * Fetch the user's signature text, if any, and normalize to
4533 * validated, ready-to-insert wikitext.
4534 * If you have pre-fetched the nickname or the fancySig option, you can
4535 * specify them here to save a database query.
4536 * Do not reuse this parser instance after calling getUserSig(),
4537 * as it may have changed if it's the $wgParser.
4538 *
4539 * @param User &$user
4540 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4541 * @param bool|null $fancySig whether the nicknname is the complete signature
4542 * or null to use default value
4543 * @return string
4544 */
4545 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4546 global $wgMaxSigChars;
4547
4548 $username = $user->getName();
4549
4550 # If not given, retrieve from the user object.
4551 if ( $nickname === false ) {
4552 $nickname = $user->getOption( 'nickname' );
4553 }
4554
4555 if ( is_null( $fancySig ) ) {
4556 $fancySig = $user->getBoolOption( 'fancysig' );
4557 }
4558
4559 $nickname = $nickname == null ? $username : $nickname;
4560
4561 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4562 $nickname = $username;
4563 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4564 } elseif ( $fancySig !== false ) {
4565 # Sig. might contain markup; validate this
4566 if ( $this->validateSig( $nickname ) !== false ) {
4567 # Validated; clean up (if needed) and return it
4568 return $this->cleanSig( $nickname, true );
4569 } else {
4570 # Failed to validate; fall back to the default
4571 $nickname = $username;
4572 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4573 }
4574 }
4575
4576 # Make sure nickname doesnt get a sig in a sig
4577 $nickname = self::cleanSigInSig( $nickname );
4578
4579 # If we're still here, make it a link to the user page
4580 $userText = wfEscapeWikiText( $username );
4581 $nickText = wfEscapeWikiText( $nickname );
4582 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4583
4584 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4585 ->title( $this->getTitle() )->text();
4586 }
4587
4588 /**
4589 * Check that the user's signature contains no bad XML
4590 *
4591 * @param string $text
4592 * @return string|bool An expanded string, or false if invalid.
4593 */
4594 public function validateSig( $text ) {
4595 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4596 }
4597
4598 /**
4599 * Clean up signature text
4600 *
4601 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4602 * 2) Substitute all transclusions
4603 *
4604 * @param string $text
4605 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4606 * @return string Signature text
4607 */
4608 public function cleanSig( $text, $parsing = false ) {
4609 if ( !$parsing ) {
4610 global $wgTitle;
4611 $magicScopeVariable = $this->lock();
4612 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4613 }
4614
4615 # Option to disable this feature
4616 if ( !$this->mOptions->getCleanSignatures() ) {
4617 return $text;
4618 }
4619
4620 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4621 # => Move this logic to braceSubstitution()
4622 $substWord = MagicWord::get( 'subst' );
4623 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4624 $substText = '{{' . $substWord->getSynonym( 0 );
4625
4626 $text = preg_replace( $substRegex, $substText, $text );
4627 $text = self::cleanSigInSig( $text );
4628 $dom = $this->preprocessToDom( $text );
4629 $frame = $this->getPreprocessor()->newFrame();
4630 $text = $frame->expand( $dom );
4631
4632 if ( !$parsing ) {
4633 $text = $this->mStripState->unstripBoth( $text );
4634 }
4635
4636 return $text;
4637 }
4638
4639 /**
4640 * Strip 3, 4 or 5 tildes out of signatures.
4641 *
4642 * @param string $text
4643 * @return string Signature text with /~{3,5}/ removed
4644 */
4645 public static function cleanSigInSig( $text ) {
4646 $text = preg_replace( '/~{3,5}/', '', $text );
4647 return $text;
4648 }
4649
4650 /**
4651 * Set up some variables which are usually set up in parse()
4652 * so that an external function can call some class members with confidence
4653 *
4654 * @param Title|null $title
4655 * @param ParserOptions $options
4656 * @param int $outputType
4657 * @param bool $clearState
4658 */
4659 public function startExternalParse( Title $title = null, ParserOptions $options,
4660 $outputType, $clearState = true
4661 ) {
4662 $this->startParse( $title, $options, $outputType, $clearState );
4663 }
4664
4665 /**
4666 * @param Title|null $title
4667 * @param ParserOptions $options
4668 * @param int $outputType
4669 * @param bool $clearState
4670 */
4671 private function startParse( Title $title = null, ParserOptions $options,
4672 $outputType, $clearState = true
4673 ) {
4674 $this->setTitle( $title );
4675 $this->mOptions = $options;
4676 $this->setOutputType( $outputType );
4677 if ( $clearState ) {
4678 $this->clearState();
4679 }
4680 }
4681
4682 /**
4683 * Wrapper for preprocess()
4684 *
4685 * @param string $text The text to preprocess
4686 * @param ParserOptions $options Options
4687 * @param Title|null $title Title object or null to use $wgTitle
4688 * @return string
4689 */
4690 public function transformMsg( $text, $options, $title = null ) {
4691 static $executing = false;
4692
4693 # Guard against infinite recursion
4694 if ( $executing ) {
4695 return $text;
4696 }
4697 $executing = true;
4698
4699 if ( !$title ) {
4700 global $wgTitle;
4701 $title = $wgTitle;
4702 }
4703
4704 $text = $this->preprocess( $text, $title, $options );
4705
4706 $executing = false;
4707 return $text;
4708 }
4709
4710 /**
4711 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4712 * The callback should have the following form:
4713 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4714 *
4715 * Transform and return $text. Use $parser for any required context, e.g. use
4716 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4717 *
4718 * Hooks may return extended information by returning an array, of which the
4719 * first numbered element (index 0) must be the return string, and all other
4720 * entries are extracted into local variables within an internal function
4721 * in the Parser class.
4722 *
4723 * This interface (introduced r61913) appears to be undocumented, but
4724 * 'markerType' is used by some core tag hooks to override which strip
4725 * array their results are placed in. **Use great caution if attempting
4726 * this interface, as it is not documented and injudicious use could smash
4727 * private variables.**
4728 *
4729 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4730 * @param callable $callback The callback function (and object) to use for the tag
4731 * @throws MWException
4732 * @return callable|null The old value of the mTagHooks array associated with the hook
4733 */
4734 public function setHook( $tag, callable $callback ) {
4735 $tag = strtolower( $tag );
4736 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4737 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4738 }
4739 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4740 $this->mTagHooks[$tag] = $callback;
4741 if ( !in_array( $tag, $this->mStripList ) ) {
4742 $this->mStripList[] = $tag;
4743 }
4744
4745 return $oldVal;
4746 }
4747
4748 /**
4749 * As setHook(), but letting the contents be parsed.
4750 *
4751 * Transparent tag hooks are like regular XML-style tag hooks, except they
4752 * operate late in the transformation sequence, on HTML instead of wikitext.
4753 *
4754 * This is probably obsoleted by things dealing with parser frames?
4755 * The only extension currently using it is geoserver.
4756 *
4757 * @since 1.10
4758 * @todo better document or deprecate this
4759 *
4760 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4761 * @param callable $callback The callback function (and object) to use for the tag
4762 * @throws MWException
4763 * @return callable|null The old value of the mTagHooks array associated with the hook
4764 */
4765 public function setTransparentTagHook( $tag, callable $callback ) {
4766 $tag = strtolower( $tag );
4767 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4768 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4769 }
4770 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4771 $this->mTransparentTagHooks[$tag] = $callback;
4772
4773 return $oldVal;
4774 }
4775
4776 /**
4777 * Remove all tag hooks
4778 */
4779 public function clearTagHooks() {
4780 $this->mTagHooks = [];
4781 $this->mFunctionTagHooks = [];
4782 $this->mStripList = $this->mDefaultStripList;
4783 }
4784
4785 /**
4786 * Create a function, e.g. {{sum:1|2|3}}
4787 * The callback function should have the form:
4788 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4789 *
4790 * Or with Parser::SFH_OBJECT_ARGS:
4791 * function myParserFunction( $parser, $frame, $args ) { ... }
4792 *
4793 * The callback may either return the text result of the function, or an array with the text
4794 * in element 0, and a number of flags in the other elements. The names of the flags are
4795 * specified in the keys. Valid flags are:
4796 * found The text returned is valid, stop processing the template. This
4797 * is on by default.
4798 * nowiki Wiki markup in the return value should be escaped
4799 * isHTML The returned text is HTML, armour it against wikitext transformation
4800 *
4801 * @param string $id The magic word ID
4802 * @param callable $callback The callback function (and object) to use
4803 * @param int $flags A combination of the following flags:
4804 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4805 *
4806 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4807 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4808 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4809 * the arguments, and to control the way they are expanded.
4810 *
4811 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4812 * arguments, for instance:
4813 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4814 *
4815 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4816 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4817 * working if/when this is changed.
4818 *
4819 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4820 * expansion.
4821 *
4822 * Please read the documentation in includes/parser/Preprocessor.php for more information
4823 * about the methods available in PPFrame and PPNode.
4824 *
4825 * @throws MWException
4826 * @return string|callable The old callback function for this name, if any
4827 */
4828 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4829 global $wgContLang;
4830
4831 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4832 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4833
4834 # Add to function cache
4835 $mw = MagicWord::get( $id );
4836 if ( !$mw ) {
4837 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4838 }
4839
4840 $synonyms = $mw->getSynonyms();
4841 $sensitive = intval( $mw->isCaseSensitive() );
4842
4843 foreach ( $synonyms as $syn ) {
4844 # Case
4845 if ( !$sensitive ) {
4846 $syn = $wgContLang->lc( $syn );
4847 }
4848 # Add leading hash
4849 if ( !( $flags & self::SFH_NO_HASH ) ) {
4850 $syn = '#' . $syn;
4851 }
4852 # Remove trailing colon
4853 if ( substr( $syn, -1, 1 ) === ':' ) {
4854 $syn = substr( $syn, 0, -1 );
4855 }
4856 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4857 }
4858 return $oldVal;
4859 }
4860
4861 /**
4862 * Get all registered function hook identifiers
4863 *
4864 * @return array
4865 */
4866 public function getFunctionHooks() {
4867 return array_keys( $this->mFunctionHooks );
4868 }
4869
4870 /**
4871 * Create a tag function, e.g. "<test>some stuff</test>".
4872 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4873 * Unlike parser functions, their content is not preprocessed.
4874 * @param string $tag
4875 * @param callable $callback
4876 * @param int $flags
4877 * @throws MWException
4878 * @return null
4879 */
4880 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4881 $tag = strtolower( $tag );
4882 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4883 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4884 }
4885 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4886 $this->mFunctionTagHooks[$tag] : null;
4887 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4888
4889 if ( !in_array( $tag, $this->mStripList ) ) {
4890 $this->mStripList[] = $tag;
4891 }
4892
4893 return $old;
4894 }
4895
4896 /**
4897 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4898 * Placeholders created in Linker::link()
4899 *
4900 * @param string &$text
4901 * @param int $options
4902 */
4903 public function replaceLinkHolders( &$text, $options = 0 ) {
4904 $this->mLinkHolders->replace( $text );
4905 }
4906
4907 /**
4908 * Replace "<!--LINK-->" link placeholders with plain text of links
4909 * (not HTML-formatted).
4910 *
4911 * @param string $text
4912 * @return string
4913 */
4914 public function replaceLinkHoldersText( $text ) {
4915 return $this->mLinkHolders->replaceText( $text );
4916 }
4917
4918 /**
4919 * Renders an image gallery from a text with one line per image.
4920 * text labels may be given by using |-style alternative text. E.g.
4921 * Image:one.jpg|The number "1"
4922 * Image:tree.jpg|A tree
4923 * given as text will return the HTML of a gallery with two images,
4924 * labeled 'The number "1"' and
4925 * 'A tree'.
4926 *
4927 * @param string $text
4928 * @param array $params
4929 * @return string HTML
4930 */
4931 public function renderImageGallery( $text, $params ) {
4932 $mode = false;
4933 if ( isset( $params['mode'] ) ) {
4934 $mode = $params['mode'];
4935 }
4936
4937 try {
4938 $ig = ImageGalleryBase::factory( $mode );
4939 } catch ( Exception $e ) {
4940 // If invalid type set, fallback to default.
4941 $ig = ImageGalleryBase::factory( false );
4942 }
4943
4944 $ig->setContextTitle( $this->mTitle );
4945 $ig->setShowBytes( false );
4946 $ig->setShowDimensions( false );
4947 $ig->setShowFilename( false );
4948 $ig->setParser( $this );
4949 $ig->setHideBadImages();
4950 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4951
4952 if ( isset( $params['showfilename'] ) ) {
4953 $ig->setShowFilename( true );
4954 } else {
4955 $ig->setShowFilename( false );
4956 }
4957 if ( isset( $params['caption'] ) ) {
4958 $caption = $params['caption'];
4959 $caption = htmlspecialchars( $caption );
4960 $caption = $this->replaceInternalLinks( $caption );
4961 $ig->setCaptionHtml( $caption );
4962 }
4963 if ( isset( $params['perrow'] ) ) {
4964 $ig->setPerRow( $params['perrow'] );
4965 }
4966 if ( isset( $params['widths'] ) ) {
4967 $ig->setWidths( $params['widths'] );
4968 }
4969 if ( isset( $params['heights'] ) ) {
4970 $ig->setHeights( $params['heights'] );
4971 }
4972 $ig->setAdditionalOptions( $params );
4973
4974 // Avoid PHP 7.1 warning from passing $this by reference
4975 $parser = $this;
4976 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
4977
4978 $lines = StringUtils::explode( "\n", $text );
4979 foreach ( $lines as $line ) {
4980 # match lines like these:
4981 # Image:someimage.jpg|This is some image
4982 $matches = [];
4983 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4984 # Skip empty lines
4985 if ( count( $matches ) == 0 ) {
4986 continue;
4987 }
4988
4989 if ( strpos( $matches[0], '%' ) !== false ) {
4990 $matches[1] = rawurldecode( $matches[1] );
4991 }
4992 $title = Title::newFromText( $matches[1], NS_FILE );
4993 if ( is_null( $title ) ) {
4994 # Bogus title. Ignore these so we don't bomb out later.
4995 continue;
4996 }
4997
4998 # We need to get what handler the file uses, to figure out parameters.
4999 # Note, a hook can overide the file name, and chose an entirely different
5000 # file (which potentially could be of a different type and have different handler).
5001 $options = [];
5002 $descQuery = false;
5003 Hooks::run( 'BeforeParserFetchFileAndTitle',
5004 [ $this, $title, &$options, &$descQuery ] );
5005 # Don't register it now, as TraditionalImageGallery does that later.
5006 $file = $this->fetchFileNoRegister( $title, $options );
5007 $handler = $file ? $file->getHandler() : false;
5008
5009 $paramMap = [
5010 'img_alt' => 'gallery-internal-alt',
5011 'img_link' => 'gallery-internal-link',
5012 ];
5013 if ( $handler ) {
5014 $paramMap = $paramMap + $handler->getParamMap();
5015 // We don't want people to specify per-image widths.
5016 // Additionally the width parameter would need special casing anyhow.
5017 unset( $paramMap['img_width'] );
5018 }
5019
5020 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5021
5022 $label = '';
5023 $alt = '';
5024 $link = '';
5025 $handlerOptions = [];
5026 if ( isset( $matches[3] ) ) {
5027 // look for an |alt= definition while trying not to break existing
5028 // captions with multiple pipes (|) in it, until a more sensible grammar
5029 // is defined for images in galleries
5030
5031 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5032 // splitting on '|' is a bit odd, and different from makeImage.
5033 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5034 // Protect LanguageConverter markup
5035 $parameterMatches = StringUtils::delimiterExplode(
5036 '-{', '}-', '|', $matches[3], true /* nested */
5037 );
5038
5039 foreach ( $parameterMatches as $parameterMatch ) {
5040 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5041 if ( $magicName ) {
5042 $paramName = $paramMap[$magicName];
5043
5044 switch ( $paramName ) {
5045 case 'gallery-internal-alt':
5046 $alt = $this->stripAltText( $match, false );
5047 break;
5048 case 'gallery-internal-link':
5049 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5050 $chars = self::EXT_LINK_URL_CLASS;
5051 $addr = self::EXT_LINK_ADDR;
5052 $prots = $this->mUrlProtocols;
5053 // check to see if link matches an absolute url, if not then it must be a wiki link.
5054 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5055 // Result of LanguageConverter::markNoConversion
5056 // invoked on an external link.
5057 $linkValue = substr( $linkValue, 4, -2 );
5058 }
5059 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5060 $link = $linkValue;
5061 $this->mOutput->addExternalLink( $link );
5062 } else {
5063 $localLinkTitle = Title::newFromText( $linkValue );
5064 if ( $localLinkTitle !== null ) {
5065 $this->mOutput->addLink( $localLinkTitle );
5066 $link = $localLinkTitle->getLinkURL();
5067 }
5068 }
5069 break;
5070 default:
5071 // Must be a handler specific parameter.
5072 if ( $handler->validateParam( $paramName, $match ) ) {
5073 $handlerOptions[$paramName] = $match;
5074 } else {
5075 // Guess not, consider it as caption.
5076 wfDebug( "$parameterMatch failed parameter validation\n" );
5077 $label = '|' . $parameterMatch;
5078 }
5079 }
5080
5081 } else {
5082 // Last pipe wins.
5083 $label = '|' . $parameterMatch;
5084 }
5085 }
5086 // Remove the pipe.
5087 $label = substr( $label, 1 );
5088 }
5089
5090 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5091 }
5092 $html = $ig->toHTML();
5093 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5094 return $html;
5095 }
5096
5097 /**
5098 * @param MediaHandler $handler
5099 * @return array
5100 */
5101 public function getImageParams( $handler ) {
5102 if ( $handler ) {
5103 $handlerClass = get_class( $handler );
5104 } else {
5105 $handlerClass = '';
5106 }
5107 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5108 # Initialise static lists
5109 static $internalParamNames = [
5110 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5111 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5112 'bottom', 'text-bottom' ],
5113 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5114 'upright', 'border', 'link', 'alt', 'class' ],
5115 ];
5116 static $internalParamMap;
5117 if ( !$internalParamMap ) {
5118 $internalParamMap = [];
5119 foreach ( $internalParamNames as $type => $names ) {
5120 foreach ( $names as $name ) {
5121 // For grep: img_left, img_right, img_center, img_none,
5122 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5123 // img_bottom, img_text_bottom,
5124 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5125 // img_border, img_link, img_alt, img_class
5126 $magicName = str_replace( '-', '_', "img_$name" );
5127 $internalParamMap[$magicName] = [ $type, $name ];
5128 }
5129 }
5130 }
5131
5132 # Add handler params
5133 $paramMap = $internalParamMap;
5134 if ( $handler ) {
5135 $handlerParamMap = $handler->getParamMap();
5136 foreach ( $handlerParamMap as $magic => $paramName ) {
5137 $paramMap[$magic] = [ 'handler', $paramName ];
5138 }
5139 }
5140 $this->mImageParams[$handlerClass] = $paramMap;
5141 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5142 }
5143 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5144 }
5145
5146 /**
5147 * Parse image options text and use it to make an image
5148 *
5149 * @param Title $title
5150 * @param string $options
5151 * @param LinkHolderArray|bool $holders
5152 * @return string HTML
5153 */
5154 public function makeImage( $title, $options, $holders = false ) {
5155 # Check if the options text is of the form "options|alt text"
5156 # Options are:
5157 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5158 # * left no resizing, just left align. label is used for alt= only
5159 # * right same, but right aligned
5160 # * none same, but not aligned
5161 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5162 # * center center the image
5163 # * frame Keep original image size, no magnify-button.
5164 # * framed Same as "frame"
5165 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5166 # * upright reduce width for upright images, rounded to full __0 px
5167 # * border draw a 1px border around the image
5168 # * alt Text for HTML alt attribute (defaults to empty)
5169 # * class Set a class for img node
5170 # * link Set the target of the image link. Can be external, interwiki, or local
5171 # vertical-align values (no % or length right now):
5172 # * baseline
5173 # * sub
5174 # * super
5175 # * top
5176 # * text-top
5177 # * middle
5178 # * bottom
5179 # * text-bottom
5180
5181 # Protect LanguageConverter markup when splitting into parts
5182 $parts = StringUtils::delimiterExplode(
5183 '-{', '}-', '|', $options, true /* allow nesting */
5184 );
5185
5186 # Give extensions a chance to select the file revision for us
5187 $options = [];
5188 $descQuery = false;
5189 Hooks::run( 'BeforeParserFetchFileAndTitle',
5190 [ $this, $title, &$options, &$descQuery ] );
5191 # Fetch and register the file (file title may be different via hooks)
5192 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5193
5194 # Get parameter map
5195 $handler = $file ? $file->getHandler() : false;
5196
5197 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5198
5199 if ( !$file ) {
5200 $this->addTrackingCategory( 'broken-file-category' );
5201 }
5202
5203 # Process the input parameters
5204 $caption = '';
5205 $params = [ 'frame' => [], 'handler' => [],
5206 'horizAlign' => [], 'vertAlign' => [] ];
5207 $seenformat = false;
5208 foreach ( $parts as $part ) {
5209 $part = trim( $part );
5210 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5211 $validated = false;
5212 if ( isset( $paramMap[$magicName] ) ) {
5213 list( $type, $paramName ) = $paramMap[$magicName];
5214
5215 # Special case; width and height come in one variable together
5216 if ( $type === 'handler' && $paramName === 'width' ) {
5217 $parsedWidthParam = $this->parseWidthParam( $value );
5218 if ( isset( $parsedWidthParam['width'] ) ) {
5219 $width = $parsedWidthParam['width'];
5220 if ( $handler->validateParam( 'width', $width ) ) {
5221 $params[$type]['width'] = $width;
5222 $validated = true;
5223 }
5224 }
5225 if ( isset( $parsedWidthParam['height'] ) ) {
5226 $height = $parsedWidthParam['height'];
5227 if ( $handler->validateParam( 'height', $height ) ) {
5228 $params[$type]['height'] = $height;
5229 $validated = true;
5230 }
5231 }
5232 # else no validation -- T15436
5233 } else {
5234 if ( $type === 'handler' ) {
5235 # Validate handler parameter
5236 $validated = $handler->validateParam( $paramName, $value );
5237 } else {
5238 # Validate internal parameters
5239 switch ( $paramName ) {
5240 case 'manualthumb':
5241 case 'alt':
5242 case 'class':
5243 # @todo FIXME: Possibly check validity here for
5244 # manualthumb? downstream behavior seems odd with
5245 # missing manual thumbs.
5246 $validated = true;
5247 $value = $this->stripAltText( $value, $holders );
5248 break;
5249 case 'link':
5250 $chars = self::EXT_LINK_URL_CLASS;
5251 $addr = self::EXT_LINK_ADDR;
5252 $prots = $this->mUrlProtocols;
5253 if ( $value === '' ) {
5254 $paramName = 'no-link';
5255 $value = true;
5256 $validated = true;
5257 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5258 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5259 $paramName = 'link-url';
5260 $this->mOutput->addExternalLink( $value );
5261 if ( $this->mOptions->getExternalLinkTarget() ) {
5262 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5263 }
5264 $validated = true;
5265 }
5266 } else {
5267 $linkTitle = Title::newFromText( $value );
5268 if ( $linkTitle ) {
5269 $paramName = 'link-title';
5270 $value = $linkTitle;
5271 $this->mOutput->addLink( $linkTitle );
5272 $validated = true;
5273 }
5274 }
5275 break;
5276 case 'frameless':
5277 case 'framed':
5278 case 'thumbnail':
5279 // use first appearing option, discard others.
5280 $validated = !$seenformat;
5281 $seenformat = true;
5282 break;
5283 default:
5284 # Most other things appear to be empty or numeric...
5285 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5286 }
5287 }
5288
5289 if ( $validated ) {
5290 $params[$type][$paramName] = $value;
5291 }
5292 }
5293 }
5294 if ( !$validated ) {
5295 $caption = $part;
5296 }
5297 }
5298
5299 # Process alignment parameters
5300 if ( $params['horizAlign'] ) {
5301 $params['frame']['align'] = key( $params['horizAlign'] );
5302 }
5303 if ( $params['vertAlign'] ) {
5304 $params['frame']['valign'] = key( $params['vertAlign'] );
5305 }
5306
5307 $params['frame']['caption'] = $caption;
5308
5309 # Will the image be presented in a frame, with the caption below?
5310 $imageIsFramed = isset( $params['frame']['frame'] )
5311 || isset( $params['frame']['framed'] )
5312 || isset( $params['frame']['thumbnail'] )
5313 || isset( $params['frame']['manualthumb'] );
5314
5315 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5316 # came to also set the caption, ordinary text after the image -- which
5317 # makes no sense, because that just repeats the text multiple times in
5318 # screen readers. It *also* came to set the title attribute.
5319 # Now that we have an alt attribute, we should not set the alt text to
5320 # equal the caption: that's worse than useless, it just repeats the
5321 # text. This is the framed/thumbnail case. If there's no caption, we
5322 # use the unnamed parameter for alt text as well, just for the time be-
5323 # ing, if the unnamed param is set and the alt param is not.
5324 # For the future, we need to figure out if we want to tweak this more,
5325 # e.g., introducing a title= parameter for the title; ignoring the un-
5326 # named parameter entirely for images without a caption; adding an ex-
5327 # plicit caption= parameter and preserving the old magic unnamed para-
5328 # meter for BC; ...
5329 if ( $imageIsFramed ) { # Framed image
5330 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5331 # No caption or alt text, add the filename as the alt text so
5332 # that screen readers at least get some description of the image
5333 $params['frame']['alt'] = $title->getText();
5334 }
5335 # Do not set $params['frame']['title'] because tooltips don't make sense
5336 # for framed images
5337 } else { # Inline image
5338 if ( !isset( $params['frame']['alt'] ) ) {
5339 # No alt text, use the "caption" for the alt text
5340 if ( $caption !== '' ) {
5341 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5342 } else {
5343 # No caption, fall back to using the filename for the
5344 # alt text
5345 $params['frame']['alt'] = $title->getText();
5346 }
5347 }
5348 # Use the "caption" for the tooltip text
5349 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5350 }
5351
5352 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5353
5354 # Linker does the rest
5355 $time = isset( $options['time'] ) ? $options['time'] : false;
5356 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5357 $time, $descQuery, $this->mOptions->getThumbSize() );
5358
5359 # Give the handler a chance to modify the parser object
5360 if ( $handler ) {
5361 $handler->parserTransformHook( $this, $file );
5362 }
5363
5364 return $ret;
5365 }
5366
5367 /**
5368 * @param string $caption
5369 * @param LinkHolderArray|bool $holders
5370 * @return mixed|string
5371 */
5372 protected function stripAltText( $caption, $holders ) {
5373 # Strip bad stuff out of the title (tooltip). We can't just use
5374 # replaceLinkHoldersText() here, because if this function is called
5375 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5376 if ( $holders ) {
5377 $tooltip = $holders->replaceText( $caption );
5378 } else {
5379 $tooltip = $this->replaceLinkHoldersText( $caption );
5380 }
5381
5382 # make sure there are no placeholders in thumbnail attributes
5383 # that are later expanded to html- so expand them now and
5384 # remove the tags
5385 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5386 $tooltip = Sanitizer::stripAllTags( $tooltip );
5387
5388 return $tooltip;
5389 }
5390
5391 /**
5392 * Set a flag in the output object indicating that the content is dynamic and
5393 * shouldn't be cached.
5394 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5395 */
5396 public function disableCache() {
5397 wfDebug( "Parser output marked as uncacheable.\n" );
5398 if ( !$this->mOutput ) {
5399 throw new MWException( __METHOD__ .
5400 " can only be called when actually parsing something" );
5401 }
5402 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5403 }
5404
5405 /**
5406 * Callback from the Sanitizer for expanding items found in HTML attribute
5407 * values, so they can be safely tested and escaped.
5408 *
5409 * @param string &$text
5410 * @param bool|PPFrame $frame
5411 * @return string
5412 */
5413 public function attributeStripCallback( &$text, $frame = false ) {
5414 $text = $this->replaceVariables( $text, $frame );
5415 $text = $this->mStripState->unstripBoth( $text );
5416 return $text;
5417 }
5418
5419 /**
5420 * Accessor
5421 *
5422 * @return array
5423 */
5424 public function getTags() {
5425 return array_merge(
5426 array_keys( $this->mTransparentTagHooks ),
5427 array_keys( $this->mTagHooks ),
5428 array_keys( $this->mFunctionTagHooks )
5429 );
5430 }
5431
5432 /**
5433 * Replace transparent tags in $text with the values given by the callbacks.
5434 *
5435 * Transparent tag hooks are like regular XML-style tag hooks, except they
5436 * operate late in the transformation sequence, on HTML instead of wikitext.
5437 *
5438 * @param string $text
5439 *
5440 * @return string
5441 */
5442 public function replaceTransparentTags( $text ) {
5443 $matches = [];
5444 $elements = array_keys( $this->mTransparentTagHooks );
5445 $text = self::extractTagsAndParams( $elements, $text, $matches );
5446 $replacements = [];
5447
5448 foreach ( $matches as $marker => $data ) {
5449 list( $element, $content, $params, $tag ) = $data;
5450 $tagName = strtolower( $element );
5451 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5452 $output = call_user_func_array(
5453 $this->mTransparentTagHooks[$tagName],
5454 [ $content, $params, $this ]
5455 );
5456 } else {
5457 $output = $tag;
5458 }
5459 $replacements[$marker] = $output;
5460 }
5461 return strtr( $text, $replacements );
5462 }
5463
5464 /**
5465 * Break wikitext input into sections, and either pull or replace
5466 * some particular section's text.
5467 *
5468 * External callers should use the getSection and replaceSection methods.
5469 *
5470 * @param string $text Page wikitext
5471 * @param string|int $sectionId A section identifier string of the form:
5472 * "<flag1> - <flag2> - ... - <section number>"
5473 *
5474 * Currently the only recognised flag is "T", which means the target section number
5475 * was derived during a template inclusion parse, in other words this is a template
5476 * section edit link. If no flags are given, it was an ordinary section edit link.
5477 * This flag is required to avoid a section numbering mismatch when a section is
5478 * enclosed by "<includeonly>" (T8563).
5479 *
5480 * The section number 0 pulls the text before the first heading; other numbers will
5481 * pull the given section along with its lower-level subsections. If the section is
5482 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5483 *
5484 * Section 0 is always considered to exist, even if it only contains the empty
5485 * string. If $text is the empty string and section 0 is replaced, $newText is
5486 * returned.
5487 *
5488 * @param string $mode One of "get" or "replace"
5489 * @param string $newText Replacement text for section data.
5490 * @return string For "get", the extracted section text.
5491 * for "replace", the whole page with the section replaced.
5492 */
5493 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5494 global $wgTitle; # not generally used but removes an ugly failure mode
5495
5496 $magicScopeVariable = $this->lock();
5497 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5498 $outText = '';
5499 $frame = $this->getPreprocessor()->newFrame();
5500
5501 # Process section extraction flags
5502 $flags = 0;
5503 $sectionParts = explode( '-', $sectionId );
5504 $sectionIndex = array_pop( $sectionParts );
5505 foreach ( $sectionParts as $part ) {
5506 if ( $part === 'T' ) {
5507 $flags |= self::PTD_FOR_INCLUSION;
5508 }
5509 }
5510
5511 # Check for empty input
5512 if ( strval( $text ) === '' ) {
5513 # Only sections 0 and T-0 exist in an empty document
5514 if ( $sectionIndex == 0 ) {
5515 if ( $mode === 'get' ) {
5516 return '';
5517 } else {
5518 return $newText;
5519 }
5520 } else {
5521 if ( $mode === 'get' ) {
5522 return $newText;
5523 } else {
5524 return $text;
5525 }
5526 }
5527 }
5528
5529 # Preprocess the text
5530 $root = $this->preprocessToDom( $text, $flags );
5531
5532 # <h> nodes indicate section breaks
5533 # They can only occur at the top level, so we can find them by iterating the root's children
5534 $node = $root->getFirstChild();
5535
5536 # Find the target section
5537 if ( $sectionIndex == 0 ) {
5538 # Section zero doesn't nest, level=big
5539 $targetLevel = 1000;
5540 } else {
5541 while ( $node ) {
5542 if ( $node->getName() === 'h' ) {
5543 $bits = $node->splitHeading();
5544 if ( $bits['i'] == $sectionIndex ) {
5545 $targetLevel = $bits['level'];
5546 break;
5547 }
5548 }
5549 if ( $mode === 'replace' ) {
5550 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5551 }
5552 $node = $node->getNextSibling();
5553 }
5554 }
5555
5556 if ( !$node ) {
5557 # Not found
5558 if ( $mode === 'get' ) {
5559 return $newText;
5560 } else {
5561 return $text;
5562 }
5563 }
5564
5565 # Find the end of the section, including nested sections
5566 do {
5567 if ( $node->getName() === 'h' ) {
5568 $bits = $node->splitHeading();
5569 $curLevel = $bits['level'];
5570 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5571 break;
5572 }
5573 }
5574 if ( $mode === 'get' ) {
5575 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5576 }
5577 $node = $node->getNextSibling();
5578 } while ( $node );
5579
5580 # Write out the remainder (in replace mode only)
5581 if ( $mode === 'replace' ) {
5582 # Output the replacement text
5583 # Add two newlines on -- trailing whitespace in $newText is conventionally
5584 # stripped by the editor, so we need both newlines to restore the paragraph gap
5585 # Only add trailing whitespace if there is newText
5586 if ( $newText != "" ) {
5587 $outText .= $newText . "\n\n";
5588 }
5589
5590 while ( $node ) {
5591 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5592 $node = $node->getNextSibling();
5593 }
5594 }
5595
5596 if ( is_string( $outText ) ) {
5597 # Re-insert stripped tags
5598 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5599 }
5600
5601 return $outText;
5602 }
5603
5604 /**
5605 * This function returns the text of a section, specified by a number ($section).
5606 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5607 * the first section before any such heading (section 0).
5608 *
5609 * If a section contains subsections, these are also returned.
5610 *
5611 * @param string $text Text to look in
5612 * @param string|int $sectionId Section identifier as a number or string
5613 * (e.g. 0, 1 or 'T-1').
5614 * @param string $defaultText Default to return if section is not found
5615 *
5616 * @return string Text of the requested section
5617 */
5618 public function getSection( $text, $sectionId, $defaultText = '' ) {
5619 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5620 }
5621
5622 /**
5623 * This function returns $oldtext after the content of the section
5624 * specified by $section has been replaced with $text. If the target
5625 * section does not exist, $oldtext is returned unchanged.
5626 *
5627 * @param string $oldText Former text of the article
5628 * @param string|int $sectionId Section identifier as a number or string
5629 * (e.g. 0, 1 or 'T-1').
5630 * @param string $newText Replacing text
5631 *
5632 * @return string Modified text
5633 */
5634 public function replaceSection( $oldText, $sectionId, $newText ) {
5635 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5636 }
5637
5638 /**
5639 * Get the ID of the revision we are parsing
5640 *
5641 * @return int|null
5642 */
5643 public function getRevisionId() {
5644 return $this->mRevisionId;
5645 }
5646
5647 /**
5648 * Get the revision object for $this->mRevisionId
5649 *
5650 * @return Revision|null Either a Revision object or null
5651 * @since 1.23 (public since 1.23)
5652 */
5653 public function getRevisionObject() {
5654 if ( !is_null( $this->mRevisionObject ) ) {
5655 return $this->mRevisionObject;
5656 }
5657 if ( is_null( $this->mRevisionId ) ) {
5658 return null;
5659 }
5660
5661 $rev = call_user_func(
5662 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5663 );
5664
5665 # If the parse is for a new revision, then the callback should have
5666 # already been set to force the object and should match mRevisionId.
5667 # If not, try to fetch by mRevisionId for sanity.
5668 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5669 $rev = Revision::newFromId( $this->mRevisionId );
5670 }
5671
5672 $this->mRevisionObject = $rev;
5673
5674 return $this->mRevisionObject;
5675 }
5676
5677 /**
5678 * Get the timestamp associated with the current revision, adjusted for
5679 * the default server-local timestamp
5680 * @return string
5681 */
5682 public function getRevisionTimestamp() {
5683 if ( is_null( $this->mRevisionTimestamp ) ) {
5684 global $wgContLang;
5685
5686 $revObject = $this->getRevisionObject();
5687 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5688
5689 # The cryptic '' timezone parameter tells to use the site-default
5690 # timezone offset instead of the user settings.
5691 # Since this value will be saved into the parser cache, served
5692 # to other users, and potentially even used inside links and such,
5693 # it needs to be consistent for all visitors.
5694 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5695
5696 }
5697 return $this->mRevisionTimestamp;
5698 }
5699
5700 /**
5701 * Get the name of the user that edited the last revision
5702 *
5703 * @return string User name
5704 */
5705 public function getRevisionUser() {
5706 if ( is_null( $this->mRevisionUser ) ) {
5707 $revObject = $this->getRevisionObject();
5708
5709 # if this template is subst: the revision id will be blank,
5710 # so just use the current user's name
5711 if ( $revObject ) {
5712 $this->mRevisionUser = $revObject->getUserText();
5713 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5714 $this->mRevisionUser = $this->getUser()->getName();
5715 }
5716 }
5717 return $this->mRevisionUser;
5718 }
5719
5720 /**
5721 * Get the size of the revision
5722 *
5723 * @return int|null Revision size
5724 */
5725 public function getRevisionSize() {
5726 if ( is_null( $this->mRevisionSize ) ) {
5727 $revObject = $this->getRevisionObject();
5728
5729 # if this variable is subst: the revision id will be blank,
5730 # so just use the parser input size, because the own substituation
5731 # will change the size.
5732 if ( $revObject ) {
5733 $this->mRevisionSize = $revObject->getSize();
5734 } else {
5735 $this->mRevisionSize = $this->mInputSize;
5736 }
5737 }
5738 return $this->mRevisionSize;
5739 }
5740
5741 /**
5742 * Mutator for $mDefaultSort
5743 *
5744 * @param string $sort New value
5745 */
5746 public function setDefaultSort( $sort ) {
5747 $this->mDefaultSort = $sort;
5748 $this->mOutput->setProperty( 'defaultsort', $sort );
5749 }
5750
5751 /**
5752 * Accessor for $mDefaultSort
5753 * Will use the empty string if none is set.
5754 *
5755 * This value is treated as a prefix, so the
5756 * empty string is equivalent to sorting by
5757 * page name.
5758 *
5759 * @return string
5760 */
5761 public function getDefaultSort() {
5762 if ( $this->mDefaultSort !== false ) {
5763 return $this->mDefaultSort;
5764 } else {
5765 return '';
5766 }
5767 }
5768
5769 /**
5770 * Accessor for $mDefaultSort
5771 * Unlike getDefaultSort(), will return false if none is set
5772 *
5773 * @return string|bool
5774 */
5775 public function getCustomDefaultSort() {
5776 return $this->mDefaultSort;
5777 }
5778
5779 private static function getSectionNameFromStrippedText( $text ) {
5780 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5781 $text = Sanitizer::decodeCharReferences( $text );
5782 $text = self::normalizeSectionName( $text );
5783 return $text;
5784 }
5785
5786 private static function makeAnchor( $sectionName ) {
5787 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5788 }
5789
5790 private static function makeLegacyAnchor( $sectionName ) {
5791 global $wgFragmentMode;
5792 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5793 // ForAttribute() and ForLink() are the same for legacy encoding
5794 $id = Sanitizer::escapeIdForAttribute( $text, Sanitizer::ID_FALLBACK );
5795 } else {
5796 $id = Sanitizer::escapeIdForLink( $text );
5797 }
5798
5799 return "#$id";
5800 }
5801
5802 /**
5803 * Try to guess the section anchor name based on a wikitext fragment
5804 * presumably extracted from a heading, for example "Header" from
5805 * "== Header ==".
5806 *
5807 * @param string $text
5808 * @return string Anchor (starting with '#')
5809 */
5810 public function guessSectionNameFromWikiText( $text ) {
5811 # Strip out wikitext links(they break the anchor)
5812 $text = $this->stripSectionName( $text );
5813 $sectionName = self::getSectionNameFromStrippedText( $text );
5814 return self::makeAnchor( $sectionName );
5815 }
5816
5817 /**
5818 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5819 * instead, if possible. For use in redirects, since various versions
5820 * of Microsoft browsers interpret Location: headers as something other
5821 * than UTF-8, resulting in breakage.
5822 *
5823 * @param string $text The section name
5824 * @return string Anchor (starting with '#')
5825 */
5826 public function guessLegacySectionNameFromWikiText( $text ) {
5827 # Strip out wikitext links(they break the anchor)
5828 $text = $this->stripSectionName( $text );
5829 $sectionName = self::getSectionNameFromStrippedText( $text );
5830 return self::makeLegacyAnchor( $sectionName );
5831 }
5832
5833 /**
5834 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5835 * @param string $text Section name (plain text)
5836 * @return string Anchor (starting with '#')
5837 */
5838 public static function guessSectionNameFromStrippedText( $text ) {
5839 $sectionName = self::getSectionNameFromStrippedText( $text );
5840 return self::makeAnchor( $sectionName );
5841 }
5842
5843 /**
5844 * Apply the same normalization as code making links to this section would
5845 *
5846 * @param string $text
5847 * @return string
5848 */
5849 private static function normalizeSectionName( $text ) {
5850 # T90902: ensure the same normalization is applied for IDs as to links
5851 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5852 try {
5853
5854 $parts = $titleParser->splitTitleString( "#$text" );
5855 } catch ( MalformedTitleException $ex ) {
5856 return $text;
5857 }
5858 return $parts['fragment'];
5859 }
5860
5861 /**
5862 * Strips a text string of wikitext for use in a section anchor
5863 *
5864 * Accepts a text string and then removes all wikitext from the
5865 * string and leaves only the resultant text (i.e. the result of
5866 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5867 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5868 * to create valid section anchors by mimicing the output of the
5869 * parser when headings are parsed.
5870 *
5871 * @param string $text Text string to be stripped of wikitext
5872 * for use in a Section anchor
5873 * @return string Filtered text string
5874 */
5875 public function stripSectionName( $text ) {
5876 # Strip internal link markup
5877 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5878 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5879
5880 # Strip external link markup
5881 # @todo FIXME: Not tolerant to blank link text
5882 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5883 # on how many empty links there are on the page - need to figure that out.
5884 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5885
5886 # Parse wikitext quotes (italics & bold)
5887 $text = $this->doQuotes( $text );
5888
5889 # Strip HTML tags
5890 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5891 return $text;
5892 }
5893
5894 /**
5895 * strip/replaceVariables/unstrip for preprocessor regression testing
5896 *
5897 * @param string $text
5898 * @param Title $title
5899 * @param ParserOptions $options
5900 * @param int $outputType
5901 *
5902 * @return string
5903 */
5904 public function testSrvus( $text, Title $title, ParserOptions $options,
5905 $outputType = self::OT_HTML
5906 ) {
5907 $magicScopeVariable = $this->lock();
5908 $this->startParse( $title, $options, $outputType, true );
5909
5910 $text = $this->replaceVariables( $text );
5911 $text = $this->mStripState->unstripBoth( $text );
5912 $text = Sanitizer::removeHTMLtags( $text );
5913 return $text;
5914 }
5915
5916 /**
5917 * @param string $text
5918 * @param Title $title
5919 * @param ParserOptions $options
5920 * @return string
5921 */
5922 public function testPst( $text, Title $title, ParserOptions $options ) {
5923 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5924 }
5925
5926 /**
5927 * @param string $text
5928 * @param Title $title
5929 * @param ParserOptions $options
5930 * @return string
5931 */
5932 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5933 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5934 }
5935
5936 /**
5937 * Call a callback function on all regions of the given text that are not
5938 * inside strip markers, and replace those regions with the return value
5939 * of the callback. For example, with input:
5940 *
5941 * aaa<MARKER>bbb
5942 *
5943 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5944 * two strings will be replaced with the value returned by the callback in
5945 * each case.
5946 *
5947 * @param string $s
5948 * @param callable $callback
5949 *
5950 * @return string
5951 */
5952 public function markerSkipCallback( $s, $callback ) {
5953 $i = 0;
5954 $out = '';
5955 while ( $i < strlen( $s ) ) {
5956 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5957 if ( $markerStart === false ) {
5958 $out .= call_user_func( $callback, substr( $s, $i ) );
5959 break;
5960 } else {
5961 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5962 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5963 if ( $markerEnd === false ) {
5964 $out .= substr( $s, $markerStart );
5965 break;
5966 } else {
5967 $markerEnd += strlen( self::MARKER_SUFFIX );
5968 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5969 $i = $markerEnd;
5970 }
5971 }
5972 }
5973 return $out;
5974 }
5975
5976 /**
5977 * Remove any strip markers found in the given text.
5978 *
5979 * @param string $text Input string
5980 * @return string
5981 */
5982 public function killMarkers( $text ) {
5983 return $this->mStripState->killMarkers( $text );
5984 }
5985
5986 /**
5987 * Save the parser state required to convert the given half-parsed text to
5988 * HTML. "Half-parsed" in this context means the output of
5989 * recursiveTagParse() or internalParse(). This output has strip markers
5990 * from replaceVariables (extensionSubstitution() etc.), and link
5991 * placeholders from replaceLinkHolders().
5992 *
5993 * Returns an array which can be serialized and stored persistently. This
5994 * array can later be loaded into another parser instance with
5995 * unserializeHalfParsedText(). The text can then be safely incorporated into
5996 * the return value of a parser hook.
5997 *
5998 * @param string $text
5999 *
6000 * @return array
6001 */
6002 public function serializeHalfParsedText( $text ) {
6003 $data = [
6004 'text' => $text,
6005 'version' => self::HALF_PARSED_VERSION,
6006 'stripState' => $this->mStripState->getSubState( $text ),
6007 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6008 ];
6009 return $data;
6010 }
6011
6012 /**
6013 * Load the parser state given in the $data array, which is assumed to
6014 * have been generated by serializeHalfParsedText(). The text contents is
6015 * extracted from the array, and its markers are transformed into markers
6016 * appropriate for the current Parser instance. This transformed text is
6017 * returned, and can be safely included in the return value of a parser
6018 * hook.
6019 *
6020 * If the $data array has been stored persistently, the caller should first
6021 * check whether it is still valid, by calling isValidHalfParsedText().
6022 *
6023 * @param array $data Serialized data
6024 * @throws MWException
6025 * @return string
6026 */
6027 public function unserializeHalfParsedText( $data ) {
6028 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6029 throw new MWException( __METHOD__ . ': invalid version' );
6030 }
6031
6032 # First, extract the strip state.
6033 $texts = [ $data['text'] ];
6034 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6035
6036 # Now renumber links
6037 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6038
6039 # Should be good to go.
6040 return $texts[0];
6041 }
6042
6043 /**
6044 * Returns true if the given array, presumed to be generated by
6045 * serializeHalfParsedText(), is compatible with the current version of the
6046 * parser.
6047 *
6048 * @param array $data
6049 *
6050 * @return bool
6051 */
6052 public function isValidHalfParsedText( $data ) {
6053 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6054 }
6055
6056 /**
6057 * Parsed a width param of imagelink like 300px or 200x300px
6058 *
6059 * @param string $value
6060 *
6061 * @return array
6062 * @since 1.20
6063 */
6064 public function parseWidthParam( $value ) {
6065 $parsedWidthParam = [];
6066 if ( $value === '' ) {
6067 return $parsedWidthParam;
6068 }
6069 $m = [];
6070 # (T15500) In both cases (width/height and width only),
6071 # permit trailing "px" for backward compatibility.
6072 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6073 $width = intval( $m[1] );
6074 $height = intval( $m[2] );
6075 $parsedWidthParam['width'] = $width;
6076 $parsedWidthParam['height'] = $height;
6077 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6078 $width = intval( $value );
6079 $parsedWidthParam['width'] = $width;
6080 }
6081 return $parsedWidthParam;
6082 }
6083
6084 /**
6085 * Lock the current instance of the parser.
6086 *
6087 * This is meant to stop someone from calling the parser
6088 * recursively and messing up all the strip state.
6089 *
6090 * @throws MWException If parser is in a parse
6091 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6092 */
6093 protected function lock() {
6094 if ( $this->mInParse ) {
6095 throw new MWException( "Parser state cleared while parsing. "
6096 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6097 }
6098
6099 // Save the backtrace when locking, so that if some code tries locking again,
6100 // we can print the lock owner's backtrace for easier debugging
6101 $e = new Exception;
6102 $this->mInParse = $e->getTraceAsString();
6103
6104 $recursiveCheck = new ScopedCallback( function () {
6105 $this->mInParse = false;
6106 } );
6107
6108 return $recursiveCheck;
6109 }
6110
6111 /**
6112 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6113 *
6114 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6115 * or if there is more than one <p/> tag in the input HTML.
6116 *
6117 * @param string $html
6118 * @return string
6119 * @since 1.24
6120 */
6121 public static function stripOuterParagraph( $html ) {
6122 $m = [];
6123 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6124 if ( strpos( $m[1], '</p>' ) === false ) {
6125 $html = $m[1];
6126 }
6127 }
6128
6129 return $html;
6130 }
6131
6132 /**
6133 * Return this parser if it is not doing anything, otherwise
6134 * get a fresh parser. You can use this method by doing
6135 * $myParser = $wgParser->getFreshParser(), or more simply
6136 * $wgParser->getFreshParser()->parse( ... );
6137 * if you're unsure if $wgParser is safe to use.
6138 *
6139 * @since 1.24
6140 * @return Parser A parser object that is not parsing anything
6141 */
6142 public function getFreshParser() {
6143 global $wgParserConf;
6144 if ( $this->mInParse ) {
6145 return new $wgParserConf['class']( $wgParserConf );
6146 } else {
6147 return $this;
6148 }
6149 }
6150
6151 /**
6152 * Set's up the PHP implementation of OOUI for use in this request
6153 * and instructs OutputPage to enable OOUI for itself.
6154 *
6155 * @since 1.26
6156 */
6157 public function enableOOUI() {
6158 OutputPage::setupOOUI();
6159 $this->mOutput->setEnableOOUI( true );
6160 }
6161 }
Wiklou, le Wiki du Wiklou