dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Revert "[MCR] Turn Revision into a proxy to new code.""
[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 # The interface for parser functions allows for extracting
3118 # flags into the local scope. Extract any forwarded flags
3119 # here.
3120 extract( $result );
3121 }
3122 }
3123
3124 # Finish mangling title and then check for loops.
3125 # Set $title to a Title object and $titleText to the PDBK
3126 if ( !$found ) {
3127 $ns = NS_TEMPLATE;
3128 # Split the title into page and subpage
3129 $subpage = '';
3130 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3131 if ( $part1 !== $relative ) {
3132 $part1 = $relative;
3133 $ns = $this->mTitle->getNamespace();
3134 }
3135 $title = Title::newFromText( $part1, $ns );
3136 if ( $title ) {
3137 $titleText = $title->getPrefixedText();
3138 # Check for language variants if the template is not found
3139 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3140 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3141 }
3142 # Do recursion depth check
3143 $limit = $this->mOptions->getMaxTemplateDepth();
3144 if ( $frame->depth >= $limit ) {
3145 $found = true;
3146 $text = '<span class="error">'
3147 . wfMessage( 'parser-template-recursion-depth-warning' )
3148 ->numParams( $limit )->inContentLanguage()->text()
3149 . '</span>';
3150 }
3151 }
3152 }
3153
3154 # Load from database
3155 if ( !$found && $title ) {
3156 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3157 if ( !$title->isExternal() ) {
3158 if ( $title->isSpecialPage()
3159 && $this->mOptions->getAllowSpecialInclusion()
3160 && $this->ot['html']
3161 ) {
3162 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3163 // Pass the template arguments as URL parameters.
3164 // "uselang" will have no effect since the Language object
3165 // is forced to the one defined in ParserOptions.
3166 $pageArgs = [];
3167 $argsLength = $args->getLength();
3168 for ( $i = 0; $i < $argsLength; $i++ ) {
3169 $bits = $args->item( $i )->splitArg();
3170 if ( strval( $bits['index'] ) === '' ) {
3171 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3172 $value = trim( $frame->expand( $bits['value'] ) );
3173 $pageArgs[$name] = $value;
3174 }
3175 }
3176
3177 // Create a new context to execute the special page
3178 $context = new RequestContext;
3179 $context->setTitle( $title );
3180 $context->setRequest( new FauxRequest( $pageArgs ) );
3181 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3182 $context->setUser( $this->getUser() );
3183 } else {
3184 // If this page is cached, then we better not be per user.
3185 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3186 }
3187 $context->setLanguage( $this->mOptions->getUserLangObj() );
3188 $ret = SpecialPageFactory::capturePath(
3189 $title, $context, $this->getLinkRenderer() );
3190 if ( $ret ) {
3191 $text = $context->getOutput()->getHTML();
3192 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3193 $found = true;
3194 $isHTML = true;
3195 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3196 $this->mOutput->updateRuntimeAdaptiveExpiry(
3197 $specialPage->maxIncludeCacheTime()
3198 );
3199 }
3200 }
3201 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3202 $found = false; # access denied
3203 wfDebug( __METHOD__ . ": template inclusion denied for " .
3204 $title->getPrefixedDBkey() . "\n" );
3205 } else {
3206 list( $text, $title ) = $this->getTemplateDom( $title );
3207 if ( $text !== false ) {
3208 $found = true;
3209 $isChildObj = true;
3210 }
3211 }
3212
3213 # If the title is valid but undisplayable, make a link to it
3214 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3215 $text = "[[:$titleText]]";
3216 $found = true;
3217 }
3218 } elseif ( $title->isTrans() ) {
3219 # Interwiki transclusion
3220 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3221 $text = $this->interwikiTransclude( $title, 'render' );
3222 $isHTML = true;
3223 } else {
3224 $text = $this->interwikiTransclude( $title, 'raw' );
3225 # Preprocess it like a template
3226 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3227 $isChildObj = true;
3228 }
3229 $found = true;
3230 }
3231
3232 # Do infinite loop check
3233 # This has to be done after redirect resolution to avoid infinite loops via redirects
3234 if ( !$frame->loopCheck( $title ) ) {
3235 $found = true;
3236 $text = '<span class="error">'
3237 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3238 . '</span>';
3239 $this->addTrackingCategory( 'template-loop-category' );
3240 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3241 wfEscapeWikiText( $titleText ) )->text() );
3242 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3243 }
3244 }
3245
3246 # If we haven't found text to substitute by now, we're done
3247 # Recover the source wikitext and return it
3248 if ( !$found ) {
3249 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3250 if ( $profileSection ) {
3251 $this->mProfiler->scopedProfileOut( $profileSection );
3252 }
3253 return [ 'object' => $text ];
3254 }
3255
3256 # Expand DOM-style return values in a child frame
3257 if ( $isChildObj ) {
3258 # Clean up argument array
3259 $newFrame = $frame->newChild( $args, $title );
3260
3261 if ( $nowiki ) {
3262 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3263 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3264 # Expansion is eligible for the empty-frame cache
3265 $text = $newFrame->cachedExpand( $titleText, $text );
3266 } else {
3267 # Uncached expansion
3268 $text = $newFrame->expand( $text );
3269 }
3270 }
3271 if ( $isLocalObj && $nowiki ) {
3272 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3273 $isLocalObj = false;
3274 }
3275
3276 if ( $profileSection ) {
3277 $this->mProfiler->scopedProfileOut( $profileSection );
3278 }
3279
3280 # Replace raw HTML by a placeholder
3281 if ( $isHTML ) {
3282 $text = $this->insertStripItem( $text );
3283 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3284 # Escape nowiki-style return values
3285 $text = wfEscapeWikiText( $text );
3286 } elseif ( is_string( $text )
3287 && !$piece['lineStart']
3288 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3289 ) {
3290 # T2529: if the template begins with a table or block-level
3291 # element, it should be treated as beginning a new line.
3292 # This behavior is somewhat controversial.
3293 $text = "\n" . $text;
3294 }
3295
3296 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3297 # Error, oversize inclusion
3298 if ( $titleText !== false ) {
3299 # Make a working, properly escaped link if possible (T25588)
3300 $text = "[[:$titleText]]";
3301 } else {
3302 # This will probably not be a working link, but at least it may
3303 # provide some hint of where the problem is
3304 preg_replace( '/^:/', '', $originalTitle );
3305 $text = "[[:$originalTitle]]";
3306 }
3307 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3308 . 'post-expand include size too large -->' );
3309 $this->limitationWarn( 'post-expand-template-inclusion' );
3310 }
3311
3312 if ( $isLocalObj ) {
3313 $ret = [ 'object' => $text ];
3314 } else {
3315 $ret = [ 'text' => $text ];
3316 }
3317
3318 return $ret;
3319 }
3320
3321 /**
3322 * Call a parser function and return an array with text and flags.
3323 *
3324 * The returned array will always contain a boolean 'found', indicating
3325 * whether the parser function was found or not. It may also contain the
3326 * following:
3327 * text: string|object, resulting wikitext or PP DOM object
3328 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3329 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3330 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3331 * nowiki: bool, wiki markup in $text should be escaped
3332 *
3333 * @since 1.21
3334 * @param PPFrame $frame The current frame, contains template arguments
3335 * @param string $function Function name
3336 * @param array $args Arguments to the function
3337 * @throws MWException
3338 * @return array
3339 */
3340 public function callParserFunction( $frame, $function, array $args = [] ) {
3341 global $wgContLang;
3342
3343 # Case sensitive functions
3344 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3345 $function = $this->mFunctionSynonyms[1][$function];
3346 } else {
3347 # Case insensitive functions
3348 $function = $wgContLang->lc( $function );
3349 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3350 $function = $this->mFunctionSynonyms[0][$function];
3351 } else {
3352 return [ 'found' => false ];
3353 }
3354 }
3355
3356 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3357
3358 // Avoid PHP 7.1 warning from passing $this by reference
3359 $parser = $this;
3360
3361 $allArgs = [ &$parser ];
3362 if ( $flags & self::SFH_OBJECT_ARGS ) {
3363 # Convert arguments to PPNodes and collect for appending to $allArgs
3364 $funcArgs = [];
3365 foreach ( $args as $k => $v ) {
3366 if ( $v instanceof PPNode || $k === 0 ) {
3367 $funcArgs[] = $v;
3368 } else {
3369 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3370 }
3371 }
3372
3373 # Add a frame parameter, and pass the arguments as an array
3374 $allArgs[] = $frame;
3375 $allArgs[] = $funcArgs;
3376 } else {
3377 # Convert arguments to plain text and append to $allArgs
3378 foreach ( $args as $k => $v ) {
3379 if ( $v instanceof PPNode ) {
3380 $allArgs[] = trim( $frame->expand( $v ) );
3381 } elseif ( is_int( $k ) && $k >= 0 ) {
3382 $allArgs[] = trim( $v );
3383 } else {
3384 $allArgs[] = trim( "$k=$v" );
3385 }
3386 }
3387 }
3388
3389 $result = call_user_func_array( $callback, $allArgs );
3390
3391 # The interface for function hooks allows them to return a wikitext
3392 # string or an array containing the string and any flags. This mungs
3393 # things around to match what this method should return.
3394 if ( !is_array( $result ) ) {
3395 $result = [
3396 'found' => true,
3397 'text' => $result,
3398 ];
3399 } else {
3400 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3401 $result['text'] = $result[0];
3402 }
3403 unset( $result[0] );
3404 $result += [
3405 'found' => true,
3406 ];
3407 }
3408
3409 $noparse = true;
3410 $preprocessFlags = 0;
3411 if ( isset( $result['noparse'] ) ) {
3412 $noparse = $result['noparse'];
3413 }
3414 if ( isset( $result['preprocessFlags'] ) ) {
3415 $preprocessFlags = $result['preprocessFlags'];
3416 }
3417
3418 if ( !$noparse ) {
3419 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3420 $result['isChildObj'] = true;
3421 }
3422
3423 return $result;
3424 }
3425
3426 /**
3427 * Get the semi-parsed DOM representation of a template with a given title,
3428 * and its redirect destination title. Cached.
3429 *
3430 * @param Title $title
3431 *
3432 * @return array
3433 */
3434 public function getTemplateDom( $title ) {
3435 $cacheTitle = $title;
3436 $titleText = $title->getPrefixedDBkey();
3437
3438 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3439 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3440 $title = Title::makeTitle( $ns, $dbk );
3441 $titleText = $title->getPrefixedDBkey();
3442 }
3443 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3444 return [ $this->mTplDomCache[$titleText], $title ];
3445 }
3446
3447 # Cache miss, go to the database
3448 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3449
3450 if ( $text === false ) {
3451 $this->mTplDomCache[$titleText] = false;
3452 return [ false, $title ];
3453 }
3454
3455 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3456 $this->mTplDomCache[$titleText] = $dom;
3457
3458 if ( !$title->equals( $cacheTitle ) ) {
3459 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3460 [ $title->getNamespace(), $title->getDBkey() ];
3461 }
3462
3463 return [ $dom, $title ];
3464 }
3465
3466 /**
3467 * Fetch the current revision of a given title. Note that the revision
3468 * (and even the title) may not exist in the database, so everything
3469 * contributing to the output of the parser should use this method
3470 * where possible, rather than getting the revisions themselves. This
3471 * method also caches its results, so using it benefits performance.
3472 *
3473 * @since 1.24
3474 * @param Title $title
3475 * @return Revision
3476 */
3477 public function fetchCurrentRevisionOfTitle( $title ) {
3478 $cacheKey = $title->getPrefixedDBkey();
3479 if ( !$this->currentRevisionCache ) {
3480 $this->currentRevisionCache = new MapCacheLRU( 100 );
3481 }
3482 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3483 $this->currentRevisionCache->set( $cacheKey,
3484 // Defaults to Parser::statelessFetchRevision()
3485 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3486 );
3487 }
3488 return $this->currentRevisionCache->get( $cacheKey );
3489 }
3490
3491 /**
3492 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3493 * without passing them on to it.
3494 *
3495 * @since 1.24
3496 * @param Title $title
3497 * @param Parser|bool $parser
3498 * @return Revision|bool False if missing
3499 */
3500 public static function statelessFetchRevision( Title $title, $parser = false ) {
3501 $pageId = $title->getArticleID();
3502 $revId = $title->getLatestRevID();
3503
3504 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3505 if ( $rev ) {
3506 $rev->setTitle( $title );
3507 }
3508
3509 return $rev;
3510 }
3511
3512 /**
3513 * Fetch the unparsed text of a template and register a reference to it.
3514 * @param Title $title
3515 * @return array ( string or false, Title )
3516 */
3517 public function fetchTemplateAndTitle( $title ) {
3518 // Defaults to Parser::statelessFetchTemplate()
3519 $templateCb = $this->mOptions->getTemplateCallback();
3520 $stuff = call_user_func( $templateCb, $title, $this );
3521 // We use U+007F DELETE to distinguish strip markers from regular text.
3522 $text = $stuff['text'];
3523 if ( is_string( $stuff['text'] ) ) {
3524 $text = strtr( $text, "\x7f", "?" );
3525 }
3526 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3527 if ( isset( $stuff['deps'] ) ) {
3528 foreach ( $stuff['deps'] as $dep ) {
3529 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3530 if ( $dep['title']->equals( $this->getTitle() ) ) {
3531 // If we transclude ourselves, the final result
3532 // will change based on the new version of the page
3533 $this->mOutput->setFlag( 'vary-revision' );
3534 }
3535 }
3536 }
3537 return [ $text, $finalTitle ];
3538 }
3539
3540 /**
3541 * Fetch the unparsed text of a template and register a reference to it.
3542 * @param Title $title
3543 * @return string|bool
3544 */
3545 public function fetchTemplate( $title ) {
3546 return $this->fetchTemplateAndTitle( $title )[0];
3547 }
3548
3549 /**
3550 * Static function to get a template
3551 * Can be overridden via ParserOptions::setTemplateCallback().
3552 *
3553 * @param Title $title
3554 * @param bool|Parser $parser
3555 *
3556 * @return array
3557 */
3558 public static function statelessFetchTemplate( $title, $parser = false ) {
3559 $text = $skip = false;
3560 $finalTitle = $title;
3561 $deps = [];
3562
3563 # Loop to fetch the article, with up to 1 redirect
3564 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3565 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3566 // @codingStandardsIgnoreEnd
3567 # Give extensions a chance to select the revision instead
3568 $id = false; # Assume current
3569 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3570 [ $parser, $title, &$skip, &$id ] );
3571
3572 if ( $skip ) {
3573 $text = false;
3574 $deps[] = [
3575 'title' => $title,
3576 'page_id' => $title->getArticleID(),
3577 'rev_id' => null
3578 ];
3579 break;
3580 }
3581 # Get the revision
3582 if ( $id ) {
3583 $rev = Revision::newFromId( $id );
3584 } elseif ( $parser ) {
3585 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3586 } else {
3587 $rev = Revision::newFromTitle( $title );
3588 }
3589 $rev_id = $rev ? $rev->getId() : 0;
3590 # If there is no current revision, there is no page
3591 if ( $id === false && !$rev ) {
3592 $linkCache = LinkCache::singleton();
3593 $linkCache->addBadLinkObj( $title );
3594 }
3595
3596 $deps[] = [
3597 'title' => $title,
3598 'page_id' => $title->getArticleID(),
3599 'rev_id' => $rev_id ];
3600 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3601 # We fetched a rev from a different title; register it too...
3602 $deps[] = [
3603 'title' => $rev->getTitle(),
3604 'page_id' => $rev->getPage(),
3605 'rev_id' => $rev_id ];
3606 }
3607
3608 if ( $rev ) {
3609 $content = $rev->getContent();
3610 $text = $content ? $content->getWikitextForTransclusion() : null;
3611
3612 Hooks::run( 'ParserFetchTemplate',
3613 [ $parser, $title, $rev, &$text, &$deps ] );
3614
3615 if ( $text === false || $text === null ) {
3616 $text = false;
3617 break;
3618 }
3619 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3620 global $wgContLang;
3621 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3622 if ( !$message->exists() ) {
3623 $text = false;
3624 break;
3625 }
3626 $content = $message->content();
3627 $text = $message->plain();
3628 } else {
3629 break;
3630 }
3631 if ( !$content ) {
3632 break;
3633 }
3634 # Redirect?
3635 $finalTitle = $title;
3636 $title = $content->getRedirectTarget();
3637 }
3638 return [
3639 'text' => $text,
3640 'finalTitle' => $finalTitle,
3641 'deps' => $deps ];
3642 }
3643
3644 /**
3645 * Fetch a file and its title and register a reference to it.
3646 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3647 * @param Title $title
3648 * @param array $options Array of options to RepoGroup::findFile
3649 * @return File|bool
3650 */
3651 public function fetchFile( $title, $options = [] ) {
3652 return $this->fetchFileAndTitle( $title, $options )[0];
3653 }
3654
3655 /**
3656 * Fetch a file and its title and register a reference to it.
3657 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3658 * @param Title $title
3659 * @param array $options Array of options to RepoGroup::findFile
3660 * @return array ( File or false, Title of file )
3661 */
3662 public function fetchFileAndTitle( $title, $options = [] ) {
3663 $file = $this->fetchFileNoRegister( $title, $options );
3664
3665 $time = $file ? $file->getTimestamp() : false;
3666 $sha1 = $file ? $file->getSha1() : false;
3667 # Register the file as a dependency...
3668 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3669 if ( $file && !$title->equals( $file->getTitle() ) ) {
3670 # Update fetched file title
3671 $title = $file->getTitle();
3672 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3673 }
3674 return [ $file, $title ];
3675 }
3676
3677 /**
3678 * Helper function for fetchFileAndTitle.
3679 *
3680 * Also useful if you need to fetch a file but not use it yet,
3681 * for example to get the file's handler.
3682 *
3683 * @param Title $title
3684 * @param array $options Array of options to RepoGroup::findFile
3685 * @return File|bool
3686 */
3687 protected function fetchFileNoRegister( $title, $options = [] ) {
3688 if ( isset( $options['broken'] ) ) {
3689 $file = false; // broken thumbnail forced by hook
3690 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3691 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3692 } else { // get by (name,timestamp)
3693 $file = wfFindFile( $title, $options );
3694 }
3695 return $file;
3696 }
3697
3698 /**
3699 * Transclude an interwiki link.
3700 *
3701 * @param Title $title
3702 * @param string $action
3703 *
3704 * @return string
3705 */
3706 public function interwikiTransclude( $title, $action ) {
3707 global $wgEnableScaryTranscluding;
3708
3709 if ( !$wgEnableScaryTranscluding ) {
3710 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3711 }
3712
3713 $url = $title->getFullURL( [ 'action' => $action ] );
3714
3715 if ( strlen( $url ) > 255 ) {
3716 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3717 }
3718 return $this->fetchScaryTemplateMaybeFromCache( $url );
3719 }
3720
3721 /**
3722 * @param string $url
3723 * @return mixed|string
3724 */
3725 public function fetchScaryTemplateMaybeFromCache( $url ) {
3726 global $wgTranscludeCacheExpiry;
3727 $dbr = wfGetDB( DB_REPLICA );
3728 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3729 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3730 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3731 if ( $obj ) {
3732 return $obj->tc_contents;
3733 }
3734
3735 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3736 $status = $req->execute(); // Status object
3737 if ( $status->isOK() ) {
3738 $text = $req->getContent();
3739 } elseif ( $req->getStatus() != 200 ) {
3740 // Though we failed to fetch the content, this status is useless.
3741 return wfMessage( 'scarytranscludefailed-httpstatus' )
3742 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3743 } else {
3744 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3745 }
3746
3747 $dbw = wfGetDB( DB_MASTER );
3748 $dbw->replace( 'transcache', [ 'tc_url' ], [
3749 'tc_url' => $url,
3750 'tc_time' => $dbw->timestamp( time() ),
3751 'tc_contents' => $text
3752 ] );
3753 return $text;
3754 }
3755
3756 /**
3757 * Triple brace replacement -- used for template arguments
3758 * @private
3759 *
3760 * @param array $piece
3761 * @param PPFrame $frame
3762 *
3763 * @return array
3764 */
3765 public function argSubstitution( $piece, $frame ) {
3766 $error = false;
3767 $parts = $piece['parts'];
3768 $nameWithSpaces = $frame->expand( $piece['title'] );
3769 $argName = trim( $nameWithSpaces );
3770 $object = false;
3771 $text = $frame->getArgument( $argName );
3772 if ( $text === false && $parts->getLength() > 0
3773 && ( $this->ot['html']
3774 || $this->ot['pre']
3775 || ( $this->ot['wiki'] && $frame->isTemplate() )
3776 )
3777 ) {
3778 # No match in frame, use the supplied default
3779 $object = $parts->item( 0 )->getChildren();
3780 }
3781 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3782 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3783 $this->limitationWarn( 'post-expand-template-argument' );
3784 }
3785
3786 if ( $text === false && $object === false ) {
3787 # No match anywhere
3788 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3789 }
3790 if ( $error !== false ) {
3791 $text .= $error;
3792 }
3793 if ( $object !== false ) {
3794 $ret = [ 'object' => $object ];
3795 } else {
3796 $ret = [ 'text' => $text ];
3797 }
3798
3799 return $ret;
3800 }
3801
3802 /**
3803 * Return the text to be used for a given extension tag.
3804 * This is the ghost of strip().
3805 *
3806 * @param array $params Associative array of parameters:
3807 * name PPNode for the tag name
3808 * attr PPNode for unparsed text where tag attributes are thought to be
3809 * attributes Optional associative array of parsed attributes
3810 * inner Contents of extension element
3811 * noClose Original text did not have a close tag
3812 * @param PPFrame $frame
3813 *
3814 * @throws MWException
3815 * @return string
3816 */
3817 public function extensionSubstitution( $params, $frame ) {
3818 static $errorStr = '<span class="error">';
3819 static $errorLen = 20;
3820
3821 $name = $frame->expand( $params['name'] );
3822 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3823 // Probably expansion depth or node count exceeded. Just punt the
3824 // error up.
3825 return $name;
3826 }
3827
3828 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3829 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3830 // See above
3831 return $attrText;
3832 }
3833
3834 // We can't safely check if the expansion for $content resulted in an
3835 // error, because the content could happen to be the error string
3836 // (T149622).
3837 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3838
3839 $marker = self::MARKER_PREFIX . "-$name-"
3840 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3841
3842 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3843 ( $this->ot['html'] || $this->ot['pre'] );
3844 if ( $isFunctionTag ) {
3845 $markerType = 'none';
3846 } else {
3847 $markerType = 'general';
3848 }
3849 if ( $this->ot['html'] || $isFunctionTag ) {
3850 $name = strtolower( $name );
3851 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3852 if ( isset( $params['attributes'] ) ) {
3853 $attributes = $attributes + $params['attributes'];
3854 }
3855
3856 if ( isset( $this->mTagHooks[$name] ) ) {
3857 $output = call_user_func_array( $this->mTagHooks[$name],
3858 [ $content, $attributes, $this, $frame ] );
3859 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3860 list( $callback, ) = $this->mFunctionTagHooks[$name];
3861
3862 // Avoid PHP 7.1 warning from passing $this by reference
3863 $parser = $this;
3864 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3865 } else {
3866 $output = '<span class="error">Invalid tag extension name: ' .
3867 htmlspecialchars( $name ) . '</span>';
3868 }
3869
3870 if ( is_array( $output ) ) {
3871 # Extract flags to local scope (to override $markerType)
3872 $flags = $output;
3873 $output = $flags[0];
3874 unset( $flags[0] );
3875 extract( $flags );
3876 }
3877 } else {
3878 if ( is_null( $attrText ) ) {
3879 $attrText = '';
3880 }
3881 if ( isset( $params['attributes'] ) ) {
3882 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3883 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3884 htmlspecialchars( $attrValue ) . '"';
3885 }
3886 }
3887 if ( $content === null ) {
3888 $output = "<$name$attrText/>";
3889 } else {
3890 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3891 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3892 // See above
3893 return $close;
3894 }
3895 $output = "<$name$attrText>$content$close";
3896 }
3897 }
3898
3899 if ( $markerType === 'none' ) {
3900 return $output;
3901 } elseif ( $markerType === 'nowiki' ) {
3902 $this->mStripState->addNoWiki( $marker, $output );
3903 } elseif ( $markerType === 'general' ) {
3904 $this->mStripState->addGeneral( $marker, $output );
3905 } else {
3906 throw new MWException( __METHOD__ . ': invalid marker type' );
3907 }
3908 return $marker;
3909 }
3910
3911 /**
3912 * Increment an include size counter
3913 *
3914 * @param string $type The type of expansion
3915 * @param int $size The size of the text
3916 * @return bool False if this inclusion would take it over the maximum, true otherwise
3917 */
3918 public function incrementIncludeSize( $type, $size ) {
3919 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3920 return false;
3921 } else {
3922 $this->mIncludeSizes[$type] += $size;
3923 return true;
3924 }
3925 }
3926
3927 /**
3928 * Increment the expensive function count
3929 *
3930 * @return bool False if the limit has been exceeded
3931 */
3932 public function incrementExpensiveFunctionCount() {
3933 $this->mExpensiveFunctionCount++;
3934 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3935 }
3936
3937 /**
3938 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3939 * Fills $this->mDoubleUnderscores, returns the modified text
3940 *
3941 * @param string $text
3942 *
3943 * @return string
3944 */
3945 public function doDoubleUnderscore( $text ) {
3946 # The position of __TOC__ needs to be recorded
3947 $mw = MagicWord::get( 'toc' );
3948 if ( $mw->match( $text ) ) {
3949 $this->mShowToc = true;
3950 $this->mForceTocPosition = true;
3951
3952 # Set a placeholder. At the end we'll fill it in with the TOC.
3953 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
3954
3955 # Only keep the first one.
3956 $text = $mw->replace( '', $text );
3957 }
3958
3959 # Now match and remove the rest of them
3960 $mwa = MagicWord::getDoubleUnderscoreArray();
3961 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3962
3963 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3964 $this->mOutput->mNoGallery = true;
3965 }
3966 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3967 $this->mShowToc = false;
3968 }
3969 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3970 && $this->mTitle->getNamespace() == NS_CATEGORY
3971 ) {
3972 $this->addTrackingCategory( 'hidden-category-category' );
3973 }
3974 # (T10068) Allow control over whether robots index a page.
3975 # __INDEX__ always overrides __NOINDEX__, see T16899
3976 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3977 $this->mOutput->setIndexPolicy( 'noindex' );
3978 $this->addTrackingCategory( 'noindex-category' );
3979 }
3980 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3981 $this->mOutput->setIndexPolicy( 'index' );
3982 $this->addTrackingCategory( 'index-category' );
3983 }
3984
3985 # Cache all double underscores in the database
3986 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3987 $this->mOutput->setProperty( $key, '' );
3988 }
3989
3990 return $text;
3991 }
3992
3993 /**
3994 * @see ParserOutput::addTrackingCategory()
3995 * @param string $msg Message key
3996 * @return bool Whether the addition was successful
3997 */
3998 public function addTrackingCategory( $msg ) {
3999 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4000 }
4001
4002 /**
4003 * This function accomplishes several tasks:
4004 * 1) Auto-number headings if that option is enabled
4005 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4006 * 3) Add a Table of contents on the top for users who have enabled the option
4007 * 4) Auto-anchor headings
4008 *
4009 * It loops through all headlines, collects the necessary data, then splits up the
4010 * string and re-inserts the newly formatted headlines.
4011 *
4012 * @param string $text
4013 * @param string $origText Original, untouched wikitext
4014 * @param bool $isMain
4015 * @return mixed|string
4016 * @private
4017 */
4018 public function formatHeadings( $text, $origText, $isMain = true ) {
4019 global $wgMaxTocLevel;
4020
4021 # Inhibit editsection links if requested in the page
4022 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4023 $maybeShowEditLink = $showEditLink = false;
4024 } else {
4025 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4026 $showEditLink = $this->mOptions->getEditSection();
4027 }
4028 if ( $showEditLink ) {
4029 $this->mOutput->setEditSectionTokens( true );
4030 }
4031
4032 # Get all headlines for numbering them and adding funky stuff like [edit]
4033 # links - this is for later, but we need the number of headlines right now
4034 $matches = [];
4035 $numMatches = preg_match_all(
4036 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4037 $text,
4038 $matches
4039 );
4040
4041 # if there are fewer than 4 headlines in the article, do not show TOC
4042 # unless it's been explicitly enabled.
4043 $enoughToc = $this->mShowToc &&
4044 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4045
4046 # Allow user to stipulate that a page should have a "new section"
4047 # link added via __NEWSECTIONLINK__
4048 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4049 $this->mOutput->setNewSection( true );
4050 }
4051
4052 # Allow user to remove the "new section"
4053 # link via __NONEWSECTIONLINK__
4054 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4055 $this->mOutput->hideNewSection( true );
4056 }
4057
4058 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4059 # override above conditions and always show TOC above first header
4060 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4061 $this->mShowToc = true;
4062 $enoughToc = true;
4063 }
4064
4065 # headline counter
4066 $headlineCount = 0;
4067 $numVisible = 0;
4068
4069 # Ugh .. the TOC should have neat indentation levels which can be
4070 # passed to the skin functions. These are determined here
4071 $toc = '';
4072 $full = '';
4073 $head = [];
4074 $sublevelCount = [];
4075 $levelCount = [];
4076 $level = 0;
4077 $prevlevel = 0;
4078 $toclevel = 0;
4079 $prevtoclevel = 0;
4080 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4081 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4082 $oldType = $this->mOutputType;
4083 $this->setOutputType( self::OT_WIKI );
4084 $frame = $this->getPreprocessor()->newFrame();
4085 $root = $this->preprocessToDom( $origText );
4086 $node = $root->getFirstChild();
4087 $byteOffset = 0;
4088 $tocraw = [];
4089 $refers = [];
4090
4091 $headlines = $numMatches !== false ? $matches[3] : [];
4092
4093 foreach ( $headlines as $headline ) {
4094 $isTemplate = false;
4095 $titleText = false;
4096 $sectionIndex = false;
4097 $numbering = '';
4098 $markerMatches = [];
4099 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4100 $serial = $markerMatches[1];
4101 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4102 $isTemplate = ( $titleText != $baseTitleText );
4103 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4104 }
4105
4106 if ( $toclevel ) {
4107 $prevlevel = $level;
4108 }
4109 $level = $matches[1][$headlineCount];
4110
4111 if ( $level > $prevlevel ) {
4112 # Increase TOC level
4113 $toclevel++;
4114 $sublevelCount[$toclevel] = 0;
4115 if ( $toclevel < $wgMaxTocLevel ) {
4116 $prevtoclevel = $toclevel;
4117 $toc .= Linker::tocIndent();
4118 $numVisible++;
4119 }
4120 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4121 # Decrease TOC level, find level to jump to
4122
4123 for ( $i = $toclevel; $i > 0; $i-- ) {
4124 if ( $levelCount[$i] == $level ) {
4125 # Found last matching level
4126 $toclevel = $i;
4127 break;
4128 } elseif ( $levelCount[$i] < $level ) {
4129 # Found first matching level below current level
4130 $toclevel = $i + 1;
4131 break;
4132 }
4133 }
4134 if ( $i == 0 ) {
4135 $toclevel = 1;
4136 }
4137 if ( $toclevel < $wgMaxTocLevel ) {
4138 if ( $prevtoclevel < $wgMaxTocLevel ) {
4139 # Unindent only if the previous toc level was shown :p
4140 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4141 $prevtoclevel = $toclevel;
4142 } else {
4143 $toc .= Linker::tocLineEnd();
4144 }
4145 }
4146 } else {
4147 # No change in level, end TOC line
4148 if ( $toclevel < $wgMaxTocLevel ) {
4149 $toc .= Linker::tocLineEnd();
4150 }
4151 }
4152
4153 $levelCount[$toclevel] = $level;
4154
4155 # count number of headlines for each level
4156 $sublevelCount[$toclevel]++;
4157 $dot = 0;
4158 for ( $i = 1; $i <= $toclevel; $i++ ) {
4159 if ( !empty( $sublevelCount[$i] ) ) {
4160 if ( $dot ) {
4161 $numbering .= '.';
4162 }
4163 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4164 $dot = 1;
4165 }
4166 }
4167
4168 # The safe header is a version of the header text safe to use for links
4169
4170 # Remove link placeholders by the link text.
4171 # <!--LINK number-->
4172 # turns into
4173 # link text with suffix
4174 # Do this before unstrip since link text can contain strip markers
4175 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4176
4177 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4178 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4179
4180 # Strip out HTML (first regex removes any tag not allowed)
4181 # Allowed tags are:
4182 # * <sup> and <sub> (T10393)
4183 # * <i> (T28375)
4184 # * <b> (r105284)
4185 # * <bdi> (T74884)
4186 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4187 # * <s> and <strike> (T35715)
4188 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4189 # to allow setting directionality in toc items.
4190 $tocline = preg_replace(
4191 [
4192 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4193 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4194 ],
4195 [ '', '<$1>' ],
4196 $safeHeadline
4197 );
4198
4199 # Strip '<span></span>', which is the result from the above if
4200 # <span id="foo"></span> is used to produce an additional anchor
4201 # for a section.
4202 $tocline = str_replace( '<span></span>', '', $tocline );
4203
4204 $tocline = trim( $tocline );
4205
4206 # For the anchor, strip out HTML-y stuff period
4207 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4208 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4209
4210 # Save headline for section edit hint before it's escaped
4211 $headlineHint = $safeHeadline;
4212
4213 # Decode HTML entities
4214 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4215
4216 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4217
4218 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4219 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4220 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4221 if ( $fallbackHeadline === $safeHeadline ) {
4222 # No reason to have both (in fact, we can't)
4223 $fallbackHeadline = false;
4224 }
4225
4226 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4227 # @todo FIXME: We may be changing them depending on the current locale.
4228 $arrayKey = strtolower( $safeHeadline );
4229 if ( $fallbackHeadline === false ) {
4230 $fallbackArrayKey = false;
4231 } else {
4232 $fallbackArrayKey = strtolower( $fallbackHeadline );
4233 }
4234
4235 # Create the anchor for linking from the TOC to the section
4236 $anchor = $safeHeadline;
4237 $fallbackAnchor = $fallbackHeadline;
4238 if ( isset( $refers[$arrayKey] ) ) {
4239 // @codingStandardsIgnoreStart
4240 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4241 // @codingStandardsIgnoreEnd
4242 $anchor .= "_$i";
4243 $linkAnchor .= "_$i";
4244 $refers["${arrayKey}_$i"] = true;
4245 } else {
4246 $refers[$arrayKey] = true;
4247 }
4248 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4249 // @codingStandardsIgnoreStart
4250 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4251 // @codingStandardsIgnoreEnd
4252 $fallbackAnchor .= "_$i";
4253 $refers["${fallbackArrayKey}_$i"] = true;
4254 } else {
4255 $refers[$fallbackArrayKey] = true;
4256 }
4257
4258 # Don't number the heading if it is the only one (looks silly)
4259 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4260 # the two are different if the line contains a link
4261 $headline = Html::element(
4262 'span',
4263 [ 'class' => 'mw-headline-number' ],
4264 $numbering
4265 ) . ' ' . $headline;
4266 }
4267
4268 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4269 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4270 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4271 }
4272
4273 # Add the section to the section tree
4274 # Find the DOM node for this header
4275 $noOffset = ( $isTemplate || $sectionIndex === false );
4276 while ( $node && !$noOffset ) {
4277 if ( $node->getName() === 'h' ) {
4278 $bits = $node->splitHeading();
4279 if ( $bits['i'] == $sectionIndex ) {
4280 break;
4281 }
4282 }
4283 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4284 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4285 $node = $node->getNextSibling();
4286 }
4287 $tocraw[] = [
4288 'toclevel' => $toclevel,
4289 'level' => $level,
4290 'line' => $tocline,
4291 'number' => $numbering,
4292 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4293 'fromtitle' => $titleText,
4294 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4295 'anchor' => $anchor,
4296 ];
4297
4298 # give headline the correct <h#> tag
4299 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4300 // Output edit section links as markers with styles that can be customized by skins
4301 if ( $isTemplate ) {
4302 # Put a T flag in the section identifier, to indicate to extractSections()
4303 # that sections inside <includeonly> should be counted.
4304 $editsectionPage = $titleText;
4305 $editsectionSection = "T-$sectionIndex";
4306 $editsectionContent = null;
4307 } else {
4308 $editsectionPage = $this->mTitle->getPrefixedText();
4309 $editsectionSection = $sectionIndex;
4310 $editsectionContent = $headlineHint;
4311 }
4312 // We use a bit of pesudo-xml for editsection markers. The
4313 // language converter is run later on. Using a UNIQ style marker
4314 // leads to the converter screwing up the tokens when it
4315 // converts stuff. And trying to insert strip tags fails too. At
4316 // this point all real inputted tags have already been escaped,
4317 // so we don't have to worry about a user trying to input one of
4318 // these markers directly. We use a page and section attribute
4319 // to stop the language converter from converting these
4320 // important bits of data, but put the headline hint inside a
4321 // content block because the language converter is supposed to
4322 // be able to convert that piece of data.
4323 // Gets replaced with html in ParserOutput::getText
4324 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4325 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4326 if ( $editsectionContent !== null ) {
4327 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4328 } else {
4329 $editlink .= '/>';
4330 }
4331 } else {
4332 $editlink = '';
4333 }
4334 $head[$headlineCount] = Linker::makeHeadline( $level,
4335 $matches['attrib'][$headlineCount], $anchor, $headline,
4336 $editlink, $fallbackAnchor );
4337
4338 $headlineCount++;
4339 }
4340
4341 $this->setOutputType( $oldType );
4342
4343 # Never ever show TOC if no headers
4344 if ( $numVisible < 1 ) {
4345 $enoughToc = false;
4346 }
4347
4348 if ( $enoughToc ) {
4349 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4350 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4351 }
4352 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4353 $this->mOutput->setTOCHTML( $toc );
4354 $toc = self::TOC_START . $toc . self::TOC_END;
4355 }
4356
4357 if ( $isMain ) {
4358 $this->mOutput->setSections( $tocraw );
4359 }
4360
4361 # split up and insert constructed headlines
4362 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4363 $i = 0;
4364
4365 // build an array of document sections
4366 $sections = [];
4367 foreach ( $blocks as $block ) {
4368 // $head is zero-based, sections aren't.
4369 if ( empty( $head[$i - 1] ) ) {
4370 $sections[$i] = $block;
4371 } else {
4372 $sections[$i] = $head[$i - 1] . $block;
4373 }
4374
4375 /**
4376 * Send a hook, one per section.
4377 * The idea here is to be able to make section-level DIVs, but to do so in a
4378 * lower-impact, more correct way than r50769
4379 *
4380 * $this : caller
4381 * $section : the section number
4382 * &$sectionContent : ref to the content of the section
4383 * $showEditLinks : boolean describing whether this section has an edit link
4384 */
4385 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4386
4387 $i++;
4388 }
4389
4390 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4391 // append the TOC at the beginning
4392 // Top anchor now in skin
4393 $sections[0] = $sections[0] . $toc . "\n";
4394 }
4395
4396 $full .= implode( '', $sections );
4397
4398 if ( $this->mForceTocPosition ) {
4399 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4400 } else {
4401 return $full;
4402 }
4403 }
4404
4405 /**
4406 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4407 * conversion, substituting signatures, {{subst:}} templates, etc.
4408 *
4409 * @param string $text The text to transform
4410 * @param Title $title The Title object for the current article
4411 * @param User $user The User object describing the current user
4412 * @param ParserOptions $options Parsing options
4413 * @param bool $clearState Whether to clear the parser state first
4414 * @return string The altered wiki markup
4415 */
4416 public function preSaveTransform( $text, Title $title, User $user,
4417 ParserOptions $options, $clearState = true
4418 ) {
4419 if ( $clearState ) {
4420 $magicScopeVariable = $this->lock();
4421 }
4422 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4423 $this->setUser( $user );
4424
4425 // Strip U+0000 NULL (T159174)
4426 $text = str_replace( "\000", '', $text );
4427
4428 // We still normalize line endings for backwards-compatibility
4429 // with other code that just calls PST, but this should already
4430 // be handled in TextContent subclasses
4431 $text = TextContent::normalizeLineEndings( $text );
4432
4433 if ( $options->getPreSaveTransform() ) {
4434 $text = $this->pstPass2( $text, $user );
4435 }
4436 $text = $this->mStripState->unstripBoth( $text );
4437
4438 $this->setUser( null ); # Reset
4439
4440 return $text;
4441 }
4442
4443 /**
4444 * Pre-save transform helper function
4445 *
4446 * @param string $text
4447 * @param User $user
4448 *
4449 * @return string
4450 */
4451 private function pstPass2( $text, $user ) {
4452 global $wgContLang;
4453
4454 # Note: This is the timestamp saved as hardcoded wikitext to
4455 # the database, we use $wgContLang here in order to give
4456 # everyone the same signature and use the default one rather
4457 # than the one selected in each user's preferences.
4458 # (see also T14815)
4459 $ts = $this->mOptions->getTimestamp();
4460 $timestamp = MWTimestamp::getLocalInstance( $ts );
4461 $ts = $timestamp->format( 'YmdHis' );
4462 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4463
4464 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4465
4466 # Variable replacement
4467 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4468 $text = $this->replaceVariables( $text );
4469
4470 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4471 # which may corrupt this parser instance via its wfMessage()->text() call-
4472
4473 # Signatures
4474 if ( strpos( $text, '~~~' ) !== false ) {
4475 $sigText = $this->getUserSig( $user );
4476 $text = strtr( $text, [
4477 '~~~~~' => $d,
4478 '~~~~' => "$sigText $d",
4479 '~~~' => $sigText
4480 ] );
4481 # The main two signature forms used above are time-sensitive
4482 $this->mOutput->setFlag( 'user-signature' );
4483 }
4484
4485 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4486 $tc = '[' . Title::legalChars() . ']';
4487 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4488
4489 // [[ns:page (context)|]]
4490 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4491 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4492 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4493 // [[ns:page (context), context|]] (using either single or double-width comma)
4494 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4495 // [[|page]] (reverse pipe trick: add context from page title)
4496 $p2 = "/\[\[\\|($tc+)]]/";
4497
4498 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4499 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4500 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4501 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4502
4503 $t = $this->mTitle->getText();
4504 $m = [];
4505 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4506 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4507 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4508 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4509 } else {
4510 # if there's no context, don't bother duplicating the title
4511 $text = preg_replace( $p2, '[[\\1]]', $text );
4512 }
4513
4514 return $text;
4515 }
4516
4517 /**
4518 * Fetch the user's signature text, if any, and normalize to
4519 * validated, ready-to-insert wikitext.
4520 * If you have pre-fetched the nickname or the fancySig option, you can
4521 * specify them here to save a database query.
4522 * Do not reuse this parser instance after calling getUserSig(),
4523 * as it may have changed if it's the $wgParser.
4524 *
4525 * @param User &$user
4526 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4527 * @param bool|null $fancySig whether the nicknname is the complete signature
4528 * or null to use default value
4529 * @return string
4530 */
4531 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4532 global $wgMaxSigChars;
4533
4534 $username = $user->getName();
4535
4536 # If not given, retrieve from the user object.
4537 if ( $nickname === false ) {
4538 $nickname = $user->getOption( 'nickname' );
4539 }
4540
4541 if ( is_null( $fancySig ) ) {
4542 $fancySig = $user->getBoolOption( 'fancysig' );
4543 }
4544
4545 $nickname = $nickname == null ? $username : $nickname;
4546
4547 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4548 $nickname = $username;
4549 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4550 } elseif ( $fancySig !== false ) {
4551 # Sig. might contain markup; validate this
4552 if ( $this->validateSig( $nickname ) !== false ) {
4553 # Validated; clean up (if needed) and return it
4554 return $this->cleanSig( $nickname, true );
4555 } else {
4556 # Failed to validate; fall back to the default
4557 $nickname = $username;
4558 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4559 }
4560 }
4561
4562 # Make sure nickname doesnt get a sig in a sig
4563 $nickname = self::cleanSigInSig( $nickname );
4564
4565 # If we're still here, make it a link to the user page
4566 $userText = wfEscapeWikiText( $username );
4567 $nickText = wfEscapeWikiText( $nickname );
4568 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4569
4570 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4571 ->title( $this->getTitle() )->text();
4572 }
4573
4574 /**
4575 * Check that the user's signature contains no bad XML
4576 *
4577 * @param string $text
4578 * @return string|bool An expanded string, or false if invalid.
4579 */
4580 public function validateSig( $text ) {
4581 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4582 }
4583
4584 /**
4585 * Clean up signature text
4586 *
4587 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4588 * 2) Substitute all transclusions
4589 *
4590 * @param string $text
4591 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4592 * @return string Signature text
4593 */
4594 public function cleanSig( $text, $parsing = false ) {
4595 if ( !$parsing ) {
4596 global $wgTitle;
4597 $magicScopeVariable = $this->lock();
4598 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4599 }
4600
4601 # Option to disable this feature
4602 if ( !$this->mOptions->getCleanSignatures() ) {
4603 return $text;
4604 }
4605
4606 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4607 # => Move this logic to braceSubstitution()
4608 $substWord = MagicWord::get( 'subst' );
4609 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4610 $substText = '{{' . $substWord->getSynonym( 0 );
4611
4612 $text = preg_replace( $substRegex, $substText, $text );
4613 $text = self::cleanSigInSig( $text );
4614 $dom = $this->preprocessToDom( $text );
4615 $frame = $this->getPreprocessor()->newFrame();
4616 $text = $frame->expand( $dom );
4617
4618 if ( !$parsing ) {
4619 $text = $this->mStripState->unstripBoth( $text );
4620 }
4621
4622 return $text;
4623 }
4624
4625 /**
4626 * Strip 3, 4 or 5 tildes out of signatures.
4627 *
4628 * @param string $text
4629 * @return string Signature text with /~{3,5}/ removed
4630 */
4631 public static function cleanSigInSig( $text ) {
4632 $text = preg_replace( '/~{3,5}/', '', $text );
4633 return $text;
4634 }
4635
4636 /**
4637 * Set up some variables which are usually set up in parse()
4638 * so that an external function can call some class members with confidence
4639 *
4640 * @param Title|null $title
4641 * @param ParserOptions $options
4642 * @param int $outputType
4643 * @param bool $clearState
4644 */
4645 public function startExternalParse( Title $title = null, ParserOptions $options,
4646 $outputType, $clearState = true
4647 ) {
4648 $this->startParse( $title, $options, $outputType, $clearState );
4649 }
4650
4651 /**
4652 * @param Title|null $title
4653 * @param ParserOptions $options
4654 * @param int $outputType
4655 * @param bool $clearState
4656 */
4657 private function startParse( Title $title = null, ParserOptions $options,
4658 $outputType, $clearState = true
4659 ) {
4660 $this->setTitle( $title );
4661 $this->mOptions = $options;
4662 $this->setOutputType( $outputType );
4663 if ( $clearState ) {
4664 $this->clearState();
4665 }
4666 }
4667
4668 /**
4669 * Wrapper for preprocess()
4670 *
4671 * @param string $text The text to preprocess
4672 * @param ParserOptions $options Options
4673 * @param Title|null $title Title object or null to use $wgTitle
4674 * @return string
4675 */
4676 public function transformMsg( $text, $options, $title = null ) {
4677 static $executing = false;
4678
4679 # Guard against infinite recursion
4680 if ( $executing ) {
4681 return $text;
4682 }
4683 $executing = true;
4684
4685 if ( !$title ) {
4686 global $wgTitle;
4687 $title = $wgTitle;
4688 }
4689
4690 $text = $this->preprocess( $text, $title, $options );
4691
4692 $executing = false;
4693 return $text;
4694 }
4695
4696 /**
4697 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4698 * The callback should have the following form:
4699 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4700 *
4701 * Transform and return $text. Use $parser for any required context, e.g. use
4702 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4703 *
4704 * Hooks may return extended information by returning an array, of which the
4705 * first numbered element (index 0) must be the return string, and all other
4706 * entries are extracted into local variables within an internal function
4707 * in the Parser class.
4708 *
4709 * This interface (introduced r61913) appears to be undocumented, but
4710 * 'markerType' is used by some core tag hooks to override which strip
4711 * array their results are placed in. **Use great caution if attempting
4712 * this interface, as it is not documented and injudicious use could smash
4713 * private variables.**
4714 *
4715 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4716 * @param callable $callback The callback function (and object) to use for the tag
4717 * @throws MWException
4718 * @return callable|null The old value of the mTagHooks array associated with the hook
4719 */
4720 public function setHook( $tag, callable $callback ) {
4721 $tag = strtolower( $tag );
4722 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4723 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4724 }
4725 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4726 $this->mTagHooks[$tag] = $callback;
4727 if ( !in_array( $tag, $this->mStripList ) ) {
4728 $this->mStripList[] = $tag;
4729 }
4730
4731 return $oldVal;
4732 }
4733
4734 /**
4735 * As setHook(), but letting the contents be parsed.
4736 *
4737 * Transparent tag hooks are like regular XML-style tag hooks, except they
4738 * operate late in the transformation sequence, on HTML instead of wikitext.
4739 *
4740 * This is probably obsoleted by things dealing with parser frames?
4741 * The only extension currently using it is geoserver.
4742 *
4743 * @since 1.10
4744 * @todo better document or deprecate this
4745 *
4746 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4747 * @param callable $callback The callback function (and object) to use for the tag
4748 * @throws MWException
4749 * @return callable|null The old value of the mTagHooks array associated with the hook
4750 */
4751 public function setTransparentTagHook( $tag, callable $callback ) {
4752 $tag = strtolower( $tag );
4753 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4754 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4755 }
4756 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4757 $this->mTransparentTagHooks[$tag] = $callback;
4758
4759 return $oldVal;
4760 }
4761
4762 /**
4763 * Remove all tag hooks
4764 */
4765 public function clearTagHooks() {
4766 $this->mTagHooks = [];
4767 $this->mFunctionTagHooks = [];
4768 $this->mStripList = $this->mDefaultStripList;
4769 }
4770
4771 /**
4772 * Create a function, e.g. {{sum:1|2|3}}
4773 * The callback function should have the form:
4774 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4775 *
4776 * Or with Parser::SFH_OBJECT_ARGS:
4777 * function myParserFunction( $parser, $frame, $args ) { ... }
4778 *
4779 * The callback may either return the text result of the function, or an array with the text
4780 * in element 0, and a number of flags in the other elements. The names of the flags are
4781 * specified in the keys. Valid flags are:
4782 * found The text returned is valid, stop processing the template. This
4783 * is on by default.
4784 * nowiki Wiki markup in the return value should be escaped
4785 * isHTML The returned text is HTML, armour it against wikitext transformation
4786 *
4787 * @param string $id The magic word ID
4788 * @param callable $callback The callback function (and object) to use
4789 * @param int $flags A combination of the following flags:
4790 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4791 *
4792 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4793 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4794 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4795 * the arguments, and to control the way they are expanded.
4796 *
4797 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4798 * arguments, for instance:
4799 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4800 *
4801 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4802 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4803 * working if/when this is changed.
4804 *
4805 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4806 * expansion.
4807 *
4808 * Please read the documentation in includes/parser/Preprocessor.php for more information
4809 * about the methods available in PPFrame and PPNode.
4810 *
4811 * @throws MWException
4812 * @return string|callable The old callback function for this name, if any
4813 */
4814 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4815 global $wgContLang;
4816
4817 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4818 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4819
4820 # Add to function cache
4821 $mw = MagicWord::get( $id );
4822 if ( !$mw ) {
4823 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4824 }
4825
4826 $synonyms = $mw->getSynonyms();
4827 $sensitive = intval( $mw->isCaseSensitive() );
4828
4829 foreach ( $synonyms as $syn ) {
4830 # Case
4831 if ( !$sensitive ) {
4832 $syn = $wgContLang->lc( $syn );
4833 }
4834 # Add leading hash
4835 if ( !( $flags & self::SFH_NO_HASH ) ) {
4836 $syn = '#' . $syn;
4837 }
4838 # Remove trailing colon
4839 if ( substr( $syn, -1, 1 ) === ':' ) {
4840 $syn = substr( $syn, 0, -1 );
4841 }
4842 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4843 }
4844 return $oldVal;
4845 }
4846
4847 /**
4848 * Get all registered function hook identifiers
4849 *
4850 * @return array
4851 */
4852 public function getFunctionHooks() {
4853 return array_keys( $this->mFunctionHooks );
4854 }
4855
4856 /**
4857 * Create a tag function, e.g. "<test>some stuff</test>".
4858 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4859 * Unlike parser functions, their content is not preprocessed.
4860 * @param string $tag
4861 * @param callable $callback
4862 * @param int $flags
4863 * @throws MWException
4864 * @return null
4865 */
4866 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4867 $tag = strtolower( $tag );
4868 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4869 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4870 }
4871 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4872 $this->mFunctionTagHooks[$tag] : null;
4873 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4874
4875 if ( !in_array( $tag, $this->mStripList ) ) {
4876 $this->mStripList[] = $tag;
4877 }
4878
4879 return $old;
4880 }
4881
4882 /**
4883 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4884 * Placeholders created in Linker::link()
4885 *
4886 * @param string &$text
4887 * @param int $options
4888 */
4889 public function replaceLinkHolders( &$text, $options = 0 ) {
4890 $this->mLinkHolders->replace( $text );
4891 }
4892
4893 /**
4894 * Replace "<!--LINK-->" link placeholders with plain text of links
4895 * (not HTML-formatted).
4896 *
4897 * @param string $text
4898 * @return string
4899 */
4900 public function replaceLinkHoldersText( $text ) {
4901 return $this->mLinkHolders->replaceText( $text );
4902 }
4903
4904 /**
4905 * Renders an image gallery from a text with one line per image.
4906 * text labels may be given by using |-style alternative text. E.g.
4907 * Image:one.jpg|The number "1"
4908 * Image:tree.jpg|A tree
4909 * given as text will return the HTML of a gallery with two images,
4910 * labeled 'The number "1"' and
4911 * 'A tree'.
4912 *
4913 * @param string $text
4914 * @param array $params
4915 * @return string HTML
4916 */
4917 public function renderImageGallery( $text, $params ) {
4918 $mode = false;
4919 if ( isset( $params['mode'] ) ) {
4920 $mode = $params['mode'];
4921 }
4922
4923 try {
4924 $ig = ImageGalleryBase::factory( $mode );
4925 } catch ( Exception $e ) {
4926 // If invalid type set, fallback to default.
4927 $ig = ImageGalleryBase::factory( false );
4928 }
4929
4930 $ig->setContextTitle( $this->mTitle );
4931 $ig->setShowBytes( false );
4932 $ig->setShowDimensions( false );
4933 $ig->setShowFilename( false );
4934 $ig->setParser( $this );
4935 $ig->setHideBadImages();
4936 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4937
4938 if ( isset( $params['showfilename'] ) ) {
4939 $ig->setShowFilename( true );
4940 } else {
4941 $ig->setShowFilename( false );
4942 }
4943 if ( isset( $params['caption'] ) ) {
4944 $caption = $params['caption'];
4945 $caption = htmlspecialchars( $caption );
4946 $caption = $this->replaceInternalLinks( $caption );
4947 $ig->setCaptionHtml( $caption );
4948 }
4949 if ( isset( $params['perrow'] ) ) {
4950 $ig->setPerRow( $params['perrow'] );
4951 }
4952 if ( isset( $params['widths'] ) ) {
4953 $ig->setWidths( $params['widths'] );
4954 }
4955 if ( isset( $params['heights'] ) ) {
4956 $ig->setHeights( $params['heights'] );
4957 }
4958 $ig->setAdditionalOptions( $params );
4959
4960 // Avoid PHP 7.1 warning from passing $this by reference
4961 $parser = $this;
4962 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
4963
4964 $lines = StringUtils::explode( "\n", $text );
4965 foreach ( $lines as $line ) {
4966 # match lines like these:
4967 # Image:someimage.jpg|This is some image
4968 $matches = [];
4969 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4970 # Skip empty lines
4971 if ( count( $matches ) == 0 ) {
4972 continue;
4973 }
4974
4975 if ( strpos( $matches[0], '%' ) !== false ) {
4976 $matches[1] = rawurldecode( $matches[1] );
4977 }
4978 $title = Title::newFromText( $matches[1], NS_FILE );
4979 if ( is_null( $title ) ) {
4980 # Bogus title. Ignore these so we don't bomb out later.
4981 continue;
4982 }
4983
4984 # We need to get what handler the file uses, to figure out parameters.
4985 # Note, a hook can overide the file name, and chose an entirely different
4986 # file (which potentially could be of a different type and have different handler).
4987 $options = [];
4988 $descQuery = false;
4989 Hooks::run( 'BeforeParserFetchFileAndTitle',
4990 [ $this, $title, &$options, &$descQuery ] );
4991 # Don't register it now, as TraditionalImageGallery does that later.
4992 $file = $this->fetchFileNoRegister( $title, $options );
4993 $handler = $file ? $file->getHandler() : false;
4994
4995 $paramMap = [
4996 'img_alt' => 'gallery-internal-alt',
4997 'img_link' => 'gallery-internal-link',
4998 ];
4999 if ( $handler ) {
5000 $paramMap = $paramMap + $handler->getParamMap();
5001 // We don't want people to specify per-image widths.
5002 // Additionally the width parameter would need special casing anyhow.
5003 unset( $paramMap['img_width'] );
5004 }
5005
5006 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5007
5008 $label = '';
5009 $alt = '';
5010 $link = '';
5011 $handlerOptions = [];
5012 if ( isset( $matches[3] ) ) {
5013 // look for an |alt= definition while trying not to break existing
5014 // captions with multiple pipes (|) in it, until a more sensible grammar
5015 // is defined for images in galleries
5016
5017 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5018 // splitting on '|' is a bit odd, and different from makeImage.
5019 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5020 // Protect LanguageConverter markup
5021 $parameterMatches = StringUtils::delimiterExplode(
5022 '-{', '}-', '|', $matches[3], true /* nested */
5023 );
5024
5025 foreach ( $parameterMatches as $parameterMatch ) {
5026 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5027 if ( $magicName ) {
5028 $paramName = $paramMap[$magicName];
5029
5030 switch ( $paramName ) {
5031 case 'gallery-internal-alt':
5032 $alt = $this->stripAltText( $match, false );
5033 break;
5034 case 'gallery-internal-link':
5035 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5036 $chars = self::EXT_LINK_URL_CLASS;
5037 $addr = self::EXT_LINK_ADDR;
5038 $prots = $this->mUrlProtocols;
5039 // check to see if link matches an absolute url, if not then it must be a wiki link.
5040 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5041 // Result of LanguageConverter::markNoConversion
5042 // invoked on an external link.
5043 $linkValue = substr( $linkValue, 4, -2 );
5044 }
5045 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5046 $link = $linkValue;
5047 $this->mOutput->addExternalLink( $link );
5048 } else {
5049 $localLinkTitle = Title::newFromText( $linkValue );
5050 if ( $localLinkTitle !== null ) {
5051 $this->mOutput->addLink( $localLinkTitle );
5052 $link = $localLinkTitle->getLinkURL();
5053 }
5054 }
5055 break;
5056 default:
5057 // Must be a handler specific parameter.
5058 if ( $handler->validateParam( $paramName, $match ) ) {
5059 $handlerOptions[$paramName] = $match;
5060 } else {
5061 // Guess not, consider it as caption.
5062 wfDebug( "$parameterMatch failed parameter validation\n" );
5063 $label = '|' . $parameterMatch;
5064 }
5065 }
5066
5067 } else {
5068 // Last pipe wins.
5069 $label = '|' . $parameterMatch;
5070 }
5071 }
5072 // Remove the pipe.
5073 $label = substr( $label, 1 );
5074 }
5075
5076 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5077 }
5078 $html = $ig->toHTML();
5079 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5080 return $html;
5081 }
5082
5083 /**
5084 * @param MediaHandler $handler
5085 * @return array
5086 */
5087 public function getImageParams( $handler ) {
5088 if ( $handler ) {
5089 $handlerClass = get_class( $handler );
5090 } else {
5091 $handlerClass = '';
5092 }
5093 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5094 # Initialise static lists
5095 static $internalParamNames = [
5096 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5097 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5098 'bottom', 'text-bottom' ],
5099 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5100 'upright', 'border', 'link', 'alt', 'class' ],
5101 ];
5102 static $internalParamMap;
5103 if ( !$internalParamMap ) {
5104 $internalParamMap = [];
5105 foreach ( $internalParamNames as $type => $names ) {
5106 foreach ( $names as $name ) {
5107 // For grep: img_left, img_right, img_center, img_none,
5108 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5109 // img_bottom, img_text_bottom,
5110 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5111 // img_border, img_link, img_alt, img_class
5112 $magicName = str_replace( '-', '_', "img_$name" );
5113 $internalParamMap[$magicName] = [ $type, $name ];
5114 }
5115 }
5116 }
5117
5118 # Add handler params
5119 $paramMap = $internalParamMap;
5120 if ( $handler ) {
5121 $handlerParamMap = $handler->getParamMap();
5122 foreach ( $handlerParamMap as $magic => $paramName ) {
5123 $paramMap[$magic] = [ 'handler', $paramName ];
5124 }
5125 }
5126 $this->mImageParams[$handlerClass] = $paramMap;
5127 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5128 }
5129 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5130 }
5131
5132 /**
5133 * Parse image options text and use it to make an image
5134 *
5135 * @param Title $title
5136 * @param string $options
5137 * @param LinkHolderArray|bool $holders
5138 * @return string HTML
5139 */
5140 public function makeImage( $title, $options, $holders = false ) {
5141 # Check if the options text is of the form "options|alt text"
5142 # Options are:
5143 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5144 # * left no resizing, just left align. label is used for alt= only
5145 # * right same, but right aligned
5146 # * none same, but not aligned
5147 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5148 # * center center the image
5149 # * frame Keep original image size, no magnify-button.
5150 # * framed Same as "frame"
5151 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5152 # * upright reduce width for upright images, rounded to full __0 px
5153 # * border draw a 1px border around the image
5154 # * alt Text for HTML alt attribute (defaults to empty)
5155 # * class Set a class for img node
5156 # * link Set the target of the image link. Can be external, interwiki, or local
5157 # vertical-align values (no % or length right now):
5158 # * baseline
5159 # * sub
5160 # * super
5161 # * top
5162 # * text-top
5163 # * middle
5164 # * bottom
5165 # * text-bottom
5166
5167 # Protect LanguageConverter markup when splitting into parts
5168 $parts = StringUtils::delimiterExplode(
5169 '-{', '}-', '|', $options, true /* allow nesting */
5170 );
5171
5172 # Give extensions a chance to select the file revision for us
5173 $options = [];
5174 $descQuery = false;
5175 Hooks::run( 'BeforeParserFetchFileAndTitle',
5176 [ $this, $title, &$options, &$descQuery ] );
5177 # Fetch and register the file (file title may be different via hooks)
5178 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5179
5180 # Get parameter map
5181 $handler = $file ? $file->getHandler() : false;
5182
5183 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5184
5185 if ( !$file ) {
5186 $this->addTrackingCategory( 'broken-file-category' );
5187 }
5188
5189 # Process the input parameters
5190 $caption = '';
5191 $params = [ 'frame' => [], 'handler' => [],
5192 'horizAlign' => [], 'vertAlign' => [] ];
5193 $seenformat = false;
5194 foreach ( $parts as $part ) {
5195 $part = trim( $part );
5196 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5197 $validated = false;
5198 if ( isset( $paramMap[$magicName] ) ) {
5199 list( $type, $paramName ) = $paramMap[$magicName];
5200
5201 # Special case; width and height come in one variable together
5202 if ( $type === 'handler' && $paramName === 'width' ) {
5203 $parsedWidthParam = $this->parseWidthParam( $value );
5204 if ( isset( $parsedWidthParam['width'] ) ) {
5205 $width = $parsedWidthParam['width'];
5206 if ( $handler->validateParam( 'width', $width ) ) {
5207 $params[$type]['width'] = $width;
5208 $validated = true;
5209 }
5210 }
5211 if ( isset( $parsedWidthParam['height'] ) ) {
5212 $height = $parsedWidthParam['height'];
5213 if ( $handler->validateParam( 'height', $height ) ) {
5214 $params[$type]['height'] = $height;
5215 $validated = true;
5216 }
5217 }
5218 # else no validation -- T15436
5219 } else {
5220 if ( $type === 'handler' ) {
5221 # Validate handler parameter
5222 $validated = $handler->validateParam( $paramName, $value );
5223 } else {
5224 # Validate internal parameters
5225 switch ( $paramName ) {
5226 case 'manualthumb':
5227 case 'alt':
5228 case 'class':
5229 # @todo FIXME: Possibly check validity here for
5230 # manualthumb? downstream behavior seems odd with
5231 # missing manual thumbs.
5232 $validated = true;
5233 $value = $this->stripAltText( $value, $holders );
5234 break;
5235 case 'link':
5236 $chars = self::EXT_LINK_URL_CLASS;
5237 $addr = self::EXT_LINK_ADDR;
5238 $prots = $this->mUrlProtocols;
5239 if ( $value === '' ) {
5240 $paramName = 'no-link';
5241 $value = true;
5242 $validated = true;
5243 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5244 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5245 $paramName = 'link-url';
5246 $this->mOutput->addExternalLink( $value );
5247 if ( $this->mOptions->getExternalLinkTarget() ) {
5248 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5249 }
5250 $validated = true;
5251 }
5252 } else {
5253 $linkTitle = Title::newFromText( $value );
5254 if ( $linkTitle ) {
5255 $paramName = 'link-title';
5256 $value = $linkTitle;
5257 $this->mOutput->addLink( $linkTitle );
5258 $validated = true;
5259 }
5260 }
5261 break;
5262 case 'frameless':
5263 case 'framed':
5264 case 'thumbnail':
5265 // use first appearing option, discard others.
5266 $validated = !$seenformat;
5267 $seenformat = true;
5268 break;
5269 default:
5270 # Most other things appear to be empty or numeric...
5271 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5272 }
5273 }
5274
5275 if ( $validated ) {
5276 $params[$type][$paramName] = $value;
5277 }
5278 }
5279 }
5280 if ( !$validated ) {
5281 $caption = $part;
5282 }
5283 }
5284
5285 # Process alignment parameters
5286 if ( $params['horizAlign'] ) {
5287 $params['frame']['align'] = key( $params['horizAlign'] );
5288 }
5289 if ( $params['vertAlign'] ) {
5290 $params['frame']['valign'] = key( $params['vertAlign'] );
5291 }
5292
5293 $params['frame']['caption'] = $caption;
5294
5295 # Will the image be presented in a frame, with the caption below?
5296 $imageIsFramed = isset( $params['frame']['frame'] )
5297 || isset( $params['frame']['framed'] )
5298 || isset( $params['frame']['thumbnail'] )
5299 || isset( $params['frame']['manualthumb'] );
5300
5301 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5302 # came to also set the caption, ordinary text after the image -- which
5303 # makes no sense, because that just repeats the text multiple times in
5304 # screen readers. It *also* came to set the title attribute.
5305 # Now that we have an alt attribute, we should not set the alt text to
5306 # equal the caption: that's worse than useless, it just repeats the
5307 # text. This is the framed/thumbnail case. If there's no caption, we
5308 # use the unnamed parameter for alt text as well, just for the time be-
5309 # ing, if the unnamed param is set and the alt param is not.
5310 # For the future, we need to figure out if we want to tweak this more,
5311 # e.g., introducing a title= parameter for the title; ignoring the un-
5312 # named parameter entirely for images without a caption; adding an ex-
5313 # plicit caption= parameter and preserving the old magic unnamed para-
5314 # meter for BC; ...
5315 if ( $imageIsFramed ) { # Framed image
5316 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5317 # No caption or alt text, add the filename as the alt text so
5318 # that screen readers at least get some description of the image
5319 $params['frame']['alt'] = $title->getText();
5320 }
5321 # Do not set $params['frame']['title'] because tooltips don't make sense
5322 # for framed images
5323 } else { # Inline image
5324 if ( !isset( $params['frame']['alt'] ) ) {
5325 # No alt text, use the "caption" for the alt text
5326 if ( $caption !== '' ) {
5327 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5328 } else {
5329 # No caption, fall back to using the filename for the
5330 # alt text
5331 $params['frame']['alt'] = $title->getText();
5332 }
5333 }
5334 # Use the "caption" for the tooltip text
5335 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5336 }
5337
5338 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5339
5340 # Linker does the rest
5341 $time = isset( $options['time'] ) ? $options['time'] : false;
5342 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5343 $time, $descQuery, $this->mOptions->getThumbSize() );
5344
5345 # Give the handler a chance to modify the parser object
5346 if ( $handler ) {
5347 $handler->parserTransformHook( $this, $file );
5348 }
5349
5350 return $ret;
5351 }
5352
5353 /**
5354 * @param string $caption
5355 * @param LinkHolderArray|bool $holders
5356 * @return mixed|string
5357 */
5358 protected function stripAltText( $caption, $holders ) {
5359 # Strip bad stuff out of the title (tooltip). We can't just use
5360 # replaceLinkHoldersText() here, because if this function is called
5361 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5362 if ( $holders ) {
5363 $tooltip = $holders->replaceText( $caption );
5364 } else {
5365 $tooltip = $this->replaceLinkHoldersText( $caption );
5366 }
5367
5368 # make sure there are no placeholders in thumbnail attributes
5369 # that are later expanded to html- so expand them now and
5370 # remove the tags
5371 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5372 $tooltip = Sanitizer::stripAllTags( $tooltip );
5373
5374 return $tooltip;
5375 }
5376
5377 /**
5378 * Set a flag in the output object indicating that the content is dynamic and
5379 * shouldn't be cached.
5380 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5381 */
5382 public function disableCache() {
5383 wfDebug( "Parser output marked as uncacheable.\n" );
5384 if ( !$this->mOutput ) {
5385 throw new MWException( __METHOD__ .
5386 " can only be called when actually parsing something" );
5387 }
5388 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5389 }
5390
5391 /**
5392 * Callback from the Sanitizer for expanding items found in HTML attribute
5393 * values, so they can be safely tested and escaped.
5394 *
5395 * @param string &$text
5396 * @param bool|PPFrame $frame
5397 * @return string
5398 */
5399 public function attributeStripCallback( &$text, $frame = false ) {
5400 $text = $this->replaceVariables( $text, $frame );
5401 $text = $this->mStripState->unstripBoth( $text );
5402 return $text;
5403 }
5404
5405 /**
5406 * Accessor
5407 *
5408 * @return array
5409 */
5410 public function getTags() {
5411 return array_merge(
5412 array_keys( $this->mTransparentTagHooks ),
5413 array_keys( $this->mTagHooks ),
5414 array_keys( $this->mFunctionTagHooks )
5415 );
5416 }
5417
5418 /**
5419 * Replace transparent tags in $text with the values given by the callbacks.
5420 *
5421 * Transparent tag hooks are like regular XML-style tag hooks, except they
5422 * operate late in the transformation sequence, on HTML instead of wikitext.
5423 *
5424 * @param string $text
5425 *
5426 * @return string
5427 */
5428 public function replaceTransparentTags( $text ) {
5429 $matches = [];
5430 $elements = array_keys( $this->mTransparentTagHooks );
5431 $text = self::extractTagsAndParams( $elements, $text, $matches );
5432 $replacements = [];
5433
5434 foreach ( $matches as $marker => $data ) {
5435 list( $element, $content, $params, $tag ) = $data;
5436 $tagName = strtolower( $element );
5437 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5438 $output = call_user_func_array(
5439 $this->mTransparentTagHooks[$tagName],
5440 [ $content, $params, $this ]
5441 );
5442 } else {
5443 $output = $tag;
5444 }
5445 $replacements[$marker] = $output;
5446 }
5447 return strtr( $text, $replacements );
5448 }
5449
5450 /**
5451 * Break wikitext input into sections, and either pull or replace
5452 * some particular section's text.
5453 *
5454 * External callers should use the getSection and replaceSection methods.
5455 *
5456 * @param string $text Page wikitext
5457 * @param string|int $sectionId A section identifier string of the form:
5458 * "<flag1> - <flag2> - ... - <section number>"
5459 *
5460 * Currently the only recognised flag is "T", which means the target section number
5461 * was derived during a template inclusion parse, in other words this is a template
5462 * section edit link. If no flags are given, it was an ordinary section edit link.
5463 * This flag is required to avoid a section numbering mismatch when a section is
5464 * enclosed by "<includeonly>" (T8563).
5465 *
5466 * The section number 0 pulls the text before the first heading; other numbers will
5467 * pull the given section along with its lower-level subsections. If the section is
5468 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5469 *
5470 * Section 0 is always considered to exist, even if it only contains the empty
5471 * string. If $text is the empty string and section 0 is replaced, $newText is
5472 * returned.
5473 *
5474 * @param string $mode One of "get" or "replace"
5475 * @param string $newText Replacement text for section data.
5476 * @return string For "get", the extracted section text.
5477 * for "replace", the whole page with the section replaced.
5478 */
5479 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5480 global $wgTitle; # not generally used but removes an ugly failure mode
5481
5482 $magicScopeVariable = $this->lock();
5483 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5484 $outText = '';
5485 $frame = $this->getPreprocessor()->newFrame();
5486
5487 # Process section extraction flags
5488 $flags = 0;
5489 $sectionParts = explode( '-', $sectionId );
5490 $sectionIndex = array_pop( $sectionParts );
5491 foreach ( $sectionParts as $part ) {
5492 if ( $part === 'T' ) {
5493 $flags |= self::PTD_FOR_INCLUSION;
5494 }
5495 }
5496
5497 # Check for empty input
5498 if ( strval( $text ) === '' ) {
5499 # Only sections 0 and T-0 exist in an empty document
5500 if ( $sectionIndex == 0 ) {
5501 if ( $mode === 'get' ) {
5502 return '';
5503 } else {
5504 return $newText;
5505 }
5506 } else {
5507 if ( $mode === 'get' ) {
5508 return $newText;
5509 } else {
5510 return $text;
5511 }
5512 }
5513 }
5514
5515 # Preprocess the text
5516 $root = $this->preprocessToDom( $text, $flags );
5517
5518 # <h> nodes indicate section breaks
5519 # They can only occur at the top level, so we can find them by iterating the root's children
5520 $node = $root->getFirstChild();
5521
5522 # Find the target section
5523 if ( $sectionIndex == 0 ) {
5524 # Section zero doesn't nest, level=big
5525 $targetLevel = 1000;
5526 } else {
5527 while ( $node ) {
5528 if ( $node->getName() === 'h' ) {
5529 $bits = $node->splitHeading();
5530 if ( $bits['i'] == $sectionIndex ) {
5531 $targetLevel = $bits['level'];
5532 break;
5533 }
5534 }
5535 if ( $mode === 'replace' ) {
5536 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5537 }
5538 $node = $node->getNextSibling();
5539 }
5540 }
5541
5542 if ( !$node ) {
5543 # Not found
5544 if ( $mode === 'get' ) {
5545 return $newText;
5546 } else {
5547 return $text;
5548 }
5549 }
5550
5551 # Find the end of the section, including nested sections
5552 do {
5553 if ( $node->getName() === 'h' ) {
5554 $bits = $node->splitHeading();
5555 $curLevel = $bits['level'];
5556 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5557 break;
5558 }
5559 }
5560 if ( $mode === 'get' ) {
5561 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5562 }
5563 $node = $node->getNextSibling();
5564 } while ( $node );
5565
5566 # Write out the remainder (in replace mode only)
5567 if ( $mode === 'replace' ) {
5568 # Output the replacement text
5569 # Add two newlines on -- trailing whitespace in $newText is conventionally
5570 # stripped by the editor, so we need both newlines to restore the paragraph gap
5571 # Only add trailing whitespace if there is newText
5572 if ( $newText != "" ) {
5573 $outText .= $newText . "\n\n";
5574 }
5575
5576 while ( $node ) {
5577 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5578 $node = $node->getNextSibling();
5579 }
5580 }
5581
5582 if ( is_string( $outText ) ) {
5583 # Re-insert stripped tags
5584 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5585 }
5586
5587 return $outText;
5588 }
5589
5590 /**
5591 * This function returns the text of a section, specified by a number ($section).
5592 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5593 * the first section before any such heading (section 0).
5594 *
5595 * If a section contains subsections, these are also returned.
5596 *
5597 * @param string $text Text to look in
5598 * @param string|int $sectionId Section identifier as a number or string
5599 * (e.g. 0, 1 or 'T-1').
5600 * @param string $defaultText Default to return if section is not found
5601 *
5602 * @return string Text of the requested section
5603 */
5604 public function getSection( $text, $sectionId, $defaultText = '' ) {
5605 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5606 }
5607
5608 /**
5609 * This function returns $oldtext after the content of the section
5610 * specified by $section has been replaced with $text. If the target
5611 * section does not exist, $oldtext is returned unchanged.
5612 *
5613 * @param string $oldText Former text of the article
5614 * @param string|int $sectionId Section identifier as a number or string
5615 * (e.g. 0, 1 or 'T-1').
5616 * @param string $newText Replacing text
5617 *
5618 * @return string Modified text
5619 */
5620 public function replaceSection( $oldText, $sectionId, $newText ) {
5621 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5622 }
5623
5624 /**
5625 * Get the ID of the revision we are parsing
5626 *
5627 * @return int|null
5628 */
5629 public function getRevisionId() {
5630 return $this->mRevisionId;
5631 }
5632
5633 /**
5634 * Get the revision object for $this->mRevisionId
5635 *
5636 * @return Revision|null Either a Revision object or null
5637 * @since 1.23 (public since 1.23)
5638 */
5639 public function getRevisionObject() {
5640 if ( !is_null( $this->mRevisionObject ) ) {
5641 return $this->mRevisionObject;
5642 }
5643 if ( is_null( $this->mRevisionId ) ) {
5644 return null;
5645 }
5646
5647 $rev = call_user_func(
5648 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5649 );
5650
5651 # If the parse is for a new revision, then the callback should have
5652 # already been set to force the object and should match mRevisionId.
5653 # If not, try to fetch by mRevisionId for sanity.
5654 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5655 $rev = Revision::newFromId( $this->mRevisionId );
5656 }
5657
5658 $this->mRevisionObject = $rev;
5659
5660 return $this->mRevisionObject;
5661 }
5662
5663 /**
5664 * Get the timestamp associated with the current revision, adjusted for
5665 * the default server-local timestamp
5666 * @return string
5667 */
5668 public function getRevisionTimestamp() {
5669 if ( is_null( $this->mRevisionTimestamp ) ) {
5670 global $wgContLang;
5671
5672 $revObject = $this->getRevisionObject();
5673 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5674
5675 # The cryptic '' timezone parameter tells to use the site-default
5676 # timezone offset instead of the user settings.
5677 # Since this value will be saved into the parser cache, served
5678 # to other users, and potentially even used inside links and such,
5679 # it needs to be consistent for all visitors.
5680 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5681
5682 }
5683 return $this->mRevisionTimestamp;
5684 }
5685
5686 /**
5687 * Get the name of the user that edited the last revision
5688 *
5689 * @return string User name
5690 */
5691 public function getRevisionUser() {
5692 if ( is_null( $this->mRevisionUser ) ) {
5693 $revObject = $this->getRevisionObject();
5694
5695 # if this template is subst: the revision id will be blank,
5696 # so just use the current user's name
5697 if ( $revObject ) {
5698 $this->mRevisionUser = $revObject->getUserText();
5699 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5700 $this->mRevisionUser = $this->getUser()->getName();
5701 }
5702 }
5703 return $this->mRevisionUser;
5704 }
5705
5706 /**
5707 * Get the size of the revision
5708 *
5709 * @return int|null Revision size
5710 */
5711 public function getRevisionSize() {
5712 if ( is_null( $this->mRevisionSize ) ) {
5713 $revObject = $this->getRevisionObject();
5714
5715 # if this variable is subst: the revision id will be blank,
5716 # so just use the parser input size, because the own substituation
5717 # will change the size.
5718 if ( $revObject ) {
5719 $this->mRevisionSize = $revObject->getSize();
5720 } else {
5721 $this->mRevisionSize = $this->mInputSize;
5722 }
5723 }
5724 return $this->mRevisionSize;
5725 }
5726
5727 /**
5728 * Mutator for $mDefaultSort
5729 *
5730 * @param string $sort New value
5731 */
5732 public function setDefaultSort( $sort ) {
5733 $this->mDefaultSort = $sort;
5734 $this->mOutput->setProperty( 'defaultsort', $sort );
5735 }
5736
5737 /**
5738 * Accessor for $mDefaultSort
5739 * Will use the empty string if none is set.
5740 *
5741 * This value is treated as a prefix, so the
5742 * empty string is equivalent to sorting by
5743 * page name.
5744 *
5745 * @return string
5746 */
5747 public function getDefaultSort() {
5748 if ( $this->mDefaultSort !== false ) {
5749 return $this->mDefaultSort;
5750 } else {
5751 return '';
5752 }
5753 }
5754
5755 /**
5756 * Accessor for $mDefaultSort
5757 * Unlike getDefaultSort(), will return false if none is set
5758 *
5759 * @return string|bool
5760 */
5761 public function getCustomDefaultSort() {
5762 return $this->mDefaultSort;
5763 }
5764
5765 private static function getSectionNameFromStrippedText( $text ) {
5766 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5767 $text = Sanitizer::decodeCharReferences( $text );
5768 $text = self::normalizeSectionName( $text );
5769 return $text;
5770 }
5771
5772 private static function makeAnchor( $sectionName ) {
5773 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5774 }
5775
5776 private static function makeLegacyAnchor( $sectionName ) {
5777 global $wgFragmentMode;
5778 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5779 // ForAttribute() and ForLink() are the same for legacy encoding
5780 $id = Sanitizer::escapeIdForAttribute( $text, Sanitizer::ID_FALLBACK );
5781 } else {
5782 $id = Sanitizer::escapeIdForLink( $text );
5783 }
5784
5785 return "#$id";
5786 }
5787
5788 /**
5789 * Try to guess the section anchor name based on a wikitext fragment
5790 * presumably extracted from a heading, for example "Header" from
5791 * "== Header ==".
5792 *
5793 * @param string $text
5794 * @return string Anchor (starting with '#')
5795 */
5796 public function guessSectionNameFromWikiText( $text ) {
5797 # Strip out wikitext links(they break the anchor)
5798 $text = $this->stripSectionName( $text );
5799 $sectionName = self::getSectionNameFromStrippedText( $text );
5800 return self::makeAnchor( $sectionName );
5801 }
5802
5803 /**
5804 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5805 * instead, if possible. For use in redirects, since various versions
5806 * of Microsoft browsers interpret Location: headers as something other
5807 * than UTF-8, resulting in breakage.
5808 *
5809 * @param string $text The section name
5810 * @return string Anchor (starting with '#')
5811 */
5812 public function guessLegacySectionNameFromWikiText( $text ) {
5813 # Strip out wikitext links(they break the anchor)
5814 $text = $this->stripSectionName( $text );
5815 $sectionName = self::getSectionNameFromStrippedText( $text );
5816 return self::makeLegacyAnchor( $sectionName );
5817 }
5818
5819 /**
5820 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5821 * @param string $text Section name (plain text)
5822 * @return string Anchor (starting with '#')
5823 */
5824 public static function guessSectionNameFromStrippedText( $text ) {
5825 $sectionName = self::getSectionNameFromStrippedText( $text );
5826 return self::makeAnchor( $sectionName );
5827 }
5828
5829 /**
5830 * Apply the same normalization as code making links to this section would
5831 *
5832 * @param string $text
5833 * @return string
5834 */
5835 private static function normalizeSectionName( $text ) {
5836 # T90902: ensure the same normalization is applied for IDs as to links
5837 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5838 try {
5839
5840 $parts = $titleParser->splitTitleString( "#$text" );
5841 } catch ( MalformedTitleException $ex ) {
5842 return $text;
5843 }
5844 return $parts['fragment'];
5845 }
5846
5847 /**
5848 * Strips a text string of wikitext for use in a section anchor
5849 *
5850 * Accepts a text string and then removes all wikitext from the
5851 * string and leaves only the resultant text (i.e. the result of
5852 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5853 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5854 * to create valid section anchors by mimicing the output of the
5855 * parser when headings are parsed.
5856 *
5857 * @param string $text Text string to be stripped of wikitext
5858 * for use in a Section anchor
5859 * @return string Filtered text string
5860 */
5861 public function stripSectionName( $text ) {
5862 # Strip internal link markup
5863 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5864 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5865
5866 # Strip external link markup
5867 # @todo FIXME: Not tolerant to blank link text
5868 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5869 # on how many empty links there are on the page - need to figure that out.
5870 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5871
5872 # Parse wikitext quotes (italics & bold)
5873 $text = $this->doQuotes( $text );
5874
5875 # Strip HTML tags
5876 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5877 return $text;
5878 }
5879
5880 /**
5881 * strip/replaceVariables/unstrip for preprocessor regression testing
5882 *
5883 * @param string $text
5884 * @param Title $title
5885 * @param ParserOptions $options
5886 * @param int $outputType
5887 *
5888 * @return string
5889 */
5890 public function testSrvus( $text, Title $title, ParserOptions $options,
5891 $outputType = self::OT_HTML
5892 ) {
5893 $magicScopeVariable = $this->lock();
5894 $this->startParse( $title, $options, $outputType, true );
5895
5896 $text = $this->replaceVariables( $text );
5897 $text = $this->mStripState->unstripBoth( $text );
5898 $text = Sanitizer::removeHTMLtags( $text );
5899 return $text;
5900 }
5901
5902 /**
5903 * @param string $text
5904 * @param Title $title
5905 * @param ParserOptions $options
5906 * @return string
5907 */
5908 public function testPst( $text, Title $title, ParserOptions $options ) {
5909 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5910 }
5911
5912 /**
5913 * @param string $text
5914 * @param Title $title
5915 * @param ParserOptions $options
5916 * @return string
5917 */
5918 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5919 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5920 }
5921
5922 /**
5923 * Call a callback function on all regions of the given text that are not
5924 * inside strip markers, and replace those regions with the return value
5925 * of the callback. For example, with input:
5926 *
5927 * aaa<MARKER>bbb
5928 *
5929 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5930 * two strings will be replaced with the value returned by the callback in
5931 * each case.
5932 *
5933 * @param string $s
5934 * @param callable $callback
5935 *
5936 * @return string
5937 */
5938 public function markerSkipCallback( $s, $callback ) {
5939 $i = 0;
5940 $out = '';
5941 while ( $i < strlen( $s ) ) {
5942 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5943 if ( $markerStart === false ) {
5944 $out .= call_user_func( $callback, substr( $s, $i ) );
5945 break;
5946 } else {
5947 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5948 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5949 if ( $markerEnd === false ) {
5950 $out .= substr( $s, $markerStart );
5951 break;
5952 } else {
5953 $markerEnd += strlen( self::MARKER_SUFFIX );
5954 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5955 $i = $markerEnd;
5956 }
5957 }
5958 }
5959 return $out;
5960 }
5961
5962 /**
5963 * Remove any strip markers found in the given text.
5964 *
5965 * @param string $text Input string
5966 * @return string
5967 */
5968 public function killMarkers( $text ) {
5969 return $this->mStripState->killMarkers( $text );
5970 }
5971
5972 /**
5973 * Save the parser state required to convert the given half-parsed text to
5974 * HTML. "Half-parsed" in this context means the output of
5975 * recursiveTagParse() or internalParse(). This output has strip markers
5976 * from replaceVariables (extensionSubstitution() etc.), and link
5977 * placeholders from replaceLinkHolders().
5978 *
5979 * Returns an array which can be serialized and stored persistently. This
5980 * array can later be loaded into another parser instance with
5981 * unserializeHalfParsedText(). The text can then be safely incorporated into
5982 * the return value of a parser hook.
5983 *
5984 * @param string $text
5985 *
5986 * @return array
5987 */
5988 public function serializeHalfParsedText( $text ) {
5989 $data = [
5990 'text' => $text,
5991 'version' => self::HALF_PARSED_VERSION,
5992 'stripState' => $this->mStripState->getSubState( $text ),
5993 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5994 ];
5995 return $data;
5996 }
5997
5998 /**
5999 * Load the parser state given in the $data array, which is assumed to
6000 * have been generated by serializeHalfParsedText(). The text contents is
6001 * extracted from the array, and its markers are transformed into markers
6002 * appropriate for the current Parser instance. This transformed text is
6003 * returned, and can be safely included in the return value of a parser
6004 * hook.
6005 *
6006 * If the $data array has been stored persistently, the caller should first
6007 * check whether it is still valid, by calling isValidHalfParsedText().
6008 *
6009 * @param array $data Serialized data
6010 * @throws MWException
6011 * @return string
6012 */
6013 public function unserializeHalfParsedText( $data ) {
6014 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6015 throw new MWException( __METHOD__ . ': invalid version' );
6016 }
6017
6018 # First, extract the strip state.
6019 $texts = [ $data['text'] ];
6020 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6021
6022 # Now renumber links
6023 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6024
6025 # Should be good to go.
6026 return $texts[0];
6027 }
6028
6029 /**
6030 * Returns true if the given array, presumed to be generated by
6031 * serializeHalfParsedText(), is compatible with the current version of the
6032 * parser.
6033 *
6034 * @param array $data
6035 *
6036 * @return bool
6037 */
6038 public function isValidHalfParsedText( $data ) {
6039 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6040 }
6041
6042 /**
6043 * Parsed a width param of imagelink like 300px or 200x300px
6044 *
6045 * @param string $value
6046 *
6047 * @return array
6048 * @since 1.20
6049 */
6050 public function parseWidthParam( $value ) {
6051 $parsedWidthParam = [];
6052 if ( $value === '' ) {
6053 return $parsedWidthParam;
6054 }
6055 $m = [];
6056 # (T15500) In both cases (width/height and width only),
6057 # permit trailing "px" for backward compatibility.
6058 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6059 $width = intval( $m[1] );
6060 $height = intval( $m[2] );
6061 $parsedWidthParam['width'] = $width;
6062 $parsedWidthParam['height'] = $height;
6063 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6064 $width = intval( $value );
6065 $parsedWidthParam['width'] = $width;
6066 }
6067 return $parsedWidthParam;
6068 }
6069
6070 /**
6071 * Lock the current instance of the parser.
6072 *
6073 * This is meant to stop someone from calling the parser
6074 * recursively and messing up all the strip state.
6075 *
6076 * @throws MWException If parser is in a parse
6077 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6078 */
6079 protected function lock() {
6080 if ( $this->mInParse ) {
6081 throw new MWException( "Parser state cleared while parsing. "
6082 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6083 }
6084
6085 // Save the backtrace when locking, so that if some code tries locking again,
6086 // we can print the lock owner's backtrace for easier debugging
6087 $e = new Exception;
6088 $this->mInParse = $e->getTraceAsString();
6089
6090 $recursiveCheck = new ScopedCallback( function () {
6091 $this->mInParse = false;
6092 } );
6093
6094 return $recursiveCheck;
6095 }
6096
6097 /**
6098 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6099 *
6100 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6101 * or if there is more than one <p/> tag in the input HTML.
6102 *
6103 * @param string $html
6104 * @return string
6105 * @since 1.24
6106 */
6107 public static function stripOuterParagraph( $html ) {
6108 $m = [];
6109 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6110 if ( strpos( $m[1], '</p>' ) === false ) {
6111 $html = $m[1];
6112 }
6113 }
6114
6115 return $html;
6116 }
6117
6118 /**
6119 * Return this parser if it is not doing anything, otherwise
6120 * get a fresh parser. You can use this method by doing
6121 * $myParser = $wgParser->getFreshParser(), or more simply
6122 * $wgParser->getFreshParser()->parse( ... );
6123 * if you're unsure if $wgParser is safe to use.
6124 *
6125 * @since 1.24
6126 * @return Parser A parser object that is not parsing anything
6127 */
6128 public function getFreshParser() {
6129 global $wgParserConf;
6130 if ( $this->mInParse ) {
6131 return new $wgParserConf['class']( $wgParserConf );
6132 } else {
6133 return $this;
6134 }
6135 }
6136
6137 /**
6138 * Set's up the PHP implementation of OOUI for use in this request
6139 * and instructs OutputPage to enable OOUI for itself.
6140 *
6141 * @since 1.26
6142 */
6143 public function enableOOUI() {
6144 OutputPage::setupOOUI();
6145 $this->mOutput->setEnableOOUI( true );
6146 }
6147 }
Wiklou, le Wiki du Wiklou