dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Create a custom log formatter that allows log entries to contain wikitext"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // phpcs:ignore Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102
103 # Regular expression for a non-newline space
104 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
105
106 # Flags for preprocessToDom
107 const PTD_FOR_INCLUSION = 1;
108
109 # Allowed values for $this->mOutputType
110 # Parameter to startExternalParse().
111 const OT_HTML = 1; # like parse()
112 const OT_WIKI = 2; # like preSaveTransform()
113 const OT_PREPROCESS = 3; # like preprocess()
114 const OT_MSG = 3;
115 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
116
117 /**
118 * @var string Prefix and suffix for temporary replacement strings
119 * for the multipass parser.
120 *
121 * \x7f should never appear in input as it's disallowed in XML.
122 * Using it at the front also gives us a little extra robustness
123 * since it shouldn't match when butted up against identifier-like
124 * string constructs.
125 *
126 * Must not consist of all title characters, or else it will change
127 * the behavior of <nowiki> in a link.
128 *
129 * Must have a character that needs escaping in attributes, otherwise
130 * someone could put a strip marker in an attribute, to get around
131 * escaping quote marks, and break out of the attribute. Thus we add
132 * `'".
133 */
134 const MARKER_SUFFIX = "-QINU`\"'\x7f";
135 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
136
137 # Markers used for wrapping the table of contents
138 const TOC_START = '<mw:toc>';
139 const TOC_END = '</mw:toc>';
140
141 # Persistent:
142 public $mTagHooks = [];
143 public $mTransparentTagHooks = [];
144 public $mFunctionHooks = [];
145 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
146 public $mFunctionTagHooks = [];
147 public $mStripList = [];
148 public $mDefaultStripList = [];
149 public $mVarCache = [];
150 public $mImageParams = [];
151 public $mImageParamsMagicArray = [];
152 public $mMarkerIndex = 0;
153 public $mFirstCall = true;
154
155 # Initialised by initialiseVariables()
156
157 /**
158 * @var MagicWordArray
159 */
160 public $mVariables;
161
162 /**
163 * @var MagicWordArray
164 */
165 public $mSubstWords;
166 # Initialised in constructor
167 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
168
169 # Initialized in getPreprocessor()
170 /** @var Preprocessor */
171 public $mPreprocessor;
172
173 # Cleared with clearState():
174 /**
175 * @var ParserOutput
176 */
177 public $mOutput;
178 public $mAutonumber;
179
180 /**
181 * @var StripState
182 */
183 public $mStripState;
184
185 public $mIncludeCount;
186 /**
187 * @var LinkHolderArray
188 */
189 public $mLinkHolders;
190
191 public $mLinkID;
192 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
193 public $mDefaultSort;
194 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
195 public $mExpensiveFunctionCount; # number of expensive parser function calls
196 public $mShowToc, $mForceTocPosition;
197
198 /**
199 * @var User
200 */
201 public $mUser; # User object; only used when doing pre-save transform
202
203 # Temporary
204 # These are variables reset at least once per parse regardless of $clearState
205
206 /**
207 * @var ParserOptions
208 */
209 public $mOptions;
210
211 /**
212 * @var Title
213 */
214 public $mTitle; # Title context, used for self-link rendering and similar things
215 public $mOutputType; # Output type, one of the OT_xxx constants
216 public $ot; # Shortcut alias, see setOutputType()
217 public $mRevisionObject; # The revision object of the specified revision ID
218 public $mRevisionId; # ID to display in {{REVISIONID}} tags
219 public $mRevisionTimestamp; # The timestamp of the specified revision ID
220 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
221 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
222 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
223 public $mInputSize = false; # For {{PAGESIZE}} on current page.
224
225 /**
226 * @var string Deprecated accessor for the strip marker prefix.
227 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
228 */
229 public $mUniqPrefix = self::MARKER_PREFIX;
230
231 /**
232 * @var array Array with the language name of each language link (i.e. the
233 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
234 * duplicate language links to the ParserOutput.
235 */
236 public $mLangLinkLanguages;
237
238 /**
239 * @var MapCacheLRU|null
240 * @since 1.24
241 *
242 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
243 */
244 public $currentRevisionCache;
245
246 /**
247 * @var bool|string Recursive call protection.
248 * This variable should be treated as if it were private.
249 */
250 public $mInParse = false;
251
252 /** @var SectionProfiler */
253 protected $mProfiler;
254
255 /**
256 * @var LinkRenderer
257 */
258 protected $mLinkRenderer;
259
260 /**
261 * @param array $conf
262 */
263 public function __construct( $conf = [] ) {
264 $this->mConf = $conf;
265 $this->mUrlProtocols = wfUrlProtocols();
266 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
267 self::EXT_LINK_ADDR .
268 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
269 if ( isset( $conf['preprocessorClass'] ) ) {
270 $this->mPreprocessorClass = $conf['preprocessorClass'];
271 } elseif ( defined( 'HPHP_VERSION' ) ) {
272 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
273 $this->mPreprocessorClass = Preprocessor_Hash::class;
274 } elseif ( extension_loaded( 'domxml' ) ) {
275 # PECL extension that conflicts with the core DOM extension (T15770)
276 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
277 $this->mPreprocessorClass = Preprocessor_Hash::class;
278 } elseif ( extension_loaded( 'dom' ) ) {
279 $this->mPreprocessorClass = Preprocessor_DOM::class;
280 } else {
281 $this->mPreprocessorClass = Preprocessor_Hash::class;
282 }
283 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
284 }
285
286 /**
287 * Reduce memory usage to reduce the impact of circular references
288 */
289 public function __destruct() {
290 if ( isset( $this->mLinkHolders ) ) {
291 unset( $this->mLinkHolders );
292 }
293 foreach ( $this as $name => $value ) {
294 unset( $this->$name );
295 }
296 }
297
298 /**
299 * Allow extensions to clean up when the parser is cloned
300 */
301 public function __clone() {
302 $this->mInParse = false;
303
304 // T58226: When you create a reference "to" an object field, that
305 // makes the object field itself be a reference too (until the other
306 // reference goes out of scope). When cloning, any field that's a
307 // reference is copied as a reference in the new object. Both of these
308 // are defined PHP5 behaviors, as inconvenient as it is for us when old
309 // hooks from PHP4 days are passing fields by reference.
310 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
311 // Make a non-reference copy of the field, then rebind the field to
312 // reference the new copy.
313 $tmp = $this->$k;
314 $this->$k =& $tmp;
315 unset( $tmp );
316 }
317
318 Hooks::run( 'ParserCloned', [ $this ] );
319 }
320
321 /**
322 * Do various kinds of initialisation on the first call of the parser
323 */
324 public function firstCallInit() {
325 if ( !$this->mFirstCall ) {
326 return;
327 }
328 $this->mFirstCall = false;
329
330 CoreParserFunctions::register( $this );
331 CoreTagHooks::register( $this );
332 $this->initialiseVariables();
333
334 // Avoid PHP 7.1 warning from passing $this by reference
335 $parser = $this;
336 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
337 }
338
339 /**
340 * Clear Parser state
341 *
342 * @private
343 */
344 public function clearState() {
345 if ( $this->mFirstCall ) {
346 $this->firstCallInit();
347 }
348 $this->mOutput = new ParserOutput;
349 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
350 $this->mAutonumber = 0;
351 $this->mIncludeCount = [];
352 $this->mLinkHolders = new LinkHolderArray( $this );
353 $this->mLinkID = 0;
354 $this->mRevisionObject = $this->mRevisionTimestamp =
355 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
356 $this->mVarCache = [];
357 $this->mUser = null;
358 $this->mLangLinkLanguages = [];
359 $this->currentRevisionCache = null;
360
361 $this->mStripState = new StripState;
362
363 # Clear these on every parse, T6549
364 $this->mTplRedirCache = $this->mTplDomCache = [];
365
366 $this->mShowToc = true;
367 $this->mForceTocPosition = false;
368 $this->mIncludeSizes = [
369 'post-expand' => 0,
370 'arg' => 0,
371 ];
372 $this->mPPNodeCount = 0;
373 $this->mGeneratedPPNodeCount = 0;
374 $this->mHighestExpansionDepth = 0;
375 $this->mDefaultSort = false;
376 $this->mHeadings = [];
377 $this->mDoubleUnderscores = [];
378 $this->mExpensiveFunctionCount = 0;
379
380 # Fix cloning
381 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
382 $this->mPreprocessor = null;
383 }
384
385 $this->mProfiler = new SectionProfiler();
386
387 // Avoid PHP 7.1 warning from passing $this by reference
388 $parser = $this;
389 Hooks::run( 'ParserClearState', [ &$parser ] );
390 }
391
392 /**
393 * Convert wikitext to HTML
394 * Do not call this function recursively.
395 *
396 * @param string $text Text we want to parse
397 * @param Title $title
398 * @param ParserOptions $options
399 * @param bool $linestart
400 * @param bool $clearState
401 * @param int $revid Number to pass in {{REVISIONID}}
402 * @return ParserOutput A ParserOutput
403 */
404 public function parse(
405 $text, Title $title, ParserOptions $options,
406 $linestart = true, $clearState = true, $revid = null
407 ) {
408 if ( $clearState ) {
409 // We use U+007F DELETE to construct strip markers, so we have to make
410 // sure that this character does not occur in the input text.
411 $text = strtr( $text, "\x7f", "?" );
412 $magicScopeVariable = $this->lock();
413 }
414 // Strip U+0000 NULL (T159174)
415 $text = str_replace( "\000", '', $text );
416
417 $this->startParse( $title, $options, self::OT_HTML, $clearState );
418
419 $this->currentRevisionCache = null;
420 $this->mInputSize = strlen( $text );
421 if ( $this->mOptions->getEnableLimitReport() ) {
422 $this->mOutput->resetParseStartTime();
423 }
424
425 $oldRevisionId = $this->mRevisionId;
426 $oldRevisionObject = $this->mRevisionObject;
427 $oldRevisionTimestamp = $this->mRevisionTimestamp;
428 $oldRevisionUser = $this->mRevisionUser;
429 $oldRevisionSize = $this->mRevisionSize;
430 if ( $revid !== null ) {
431 $this->mRevisionId = $revid;
432 $this->mRevisionObject = null;
433 $this->mRevisionTimestamp = null;
434 $this->mRevisionUser = null;
435 $this->mRevisionSize = null;
436 }
437
438 // Avoid PHP 7.1 warning from passing $this by reference
439 $parser = $this;
440 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
441 # No more strip!
442 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
443 $text = $this->internalParse( $text );
444 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
445
446 $text = $this->internalParseHalfParsed( $text, true, $linestart );
447
448 /**
449 * A converted title will be provided in the output object if title and
450 * content conversion are enabled, the article text does not contain
451 * a conversion-suppressing double-underscore tag, and no
452 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
453 * automatic link conversion.
454 */
455 if ( !( $options->getDisableTitleConversion()
456 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
457 || isset( $this->mDoubleUnderscores['notitleconvert'] )
458 || $this->mOutput->getDisplayTitle() !== false )
459 ) {
460 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
461 if ( $convruletitle ) {
462 $this->mOutput->setTitleText( $convruletitle );
463 } else {
464 $titleText = $this->getConverterLanguage()->convertTitle( $title );
465 $this->mOutput->setTitleText( $titleText );
466 }
467 }
468
469 # Compute runtime adaptive expiry if set
470 $this->mOutput->finalizeAdaptiveCacheExpiry();
471
472 # Warn if too many heavyweight parser functions were used
473 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
474 $this->limitationWarn( 'expensive-parserfunction',
475 $this->mExpensiveFunctionCount,
476 $this->mOptions->getExpensiveParserFunctionLimit()
477 );
478 }
479
480 # Information on limits, for the benefit of users who try to skirt them
481 if ( $this->mOptions->getEnableLimitReport() ) {
482 $text .= $this->makeLimitReport();
483 }
484
485 # Wrap non-interface parser output in a <div> so it can be targeted
486 # with CSS (T37247)
487 $class = $this->mOptions->getWrapOutputClass();
488 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
489 $text = Html::rawElement( 'div', [ 'class' => $class ], $text );
490 }
491
492 $this->mOutput->setText( $text );
493
494 $this->mRevisionId = $oldRevisionId;
495 $this->mRevisionObject = $oldRevisionObject;
496 $this->mRevisionTimestamp = $oldRevisionTimestamp;
497 $this->mRevisionUser = $oldRevisionUser;
498 $this->mRevisionSize = $oldRevisionSize;
499 $this->mInputSize = false;
500 $this->currentRevisionCache = null;
501
502 return $this->mOutput;
503 }
504
505 /**
506 * Set the limit report data in the current ParserOutput, and return the
507 * limit report HTML comment.
508 *
509 * @return string
510 */
511 protected function makeLimitReport() {
512 global $wgShowHostnames;
513
514 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
515
516 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
517 if ( $cpuTime !== null ) {
518 $this->mOutput->setLimitReportData( 'limitreport-cputime',
519 sprintf( "%.3f", $cpuTime )
520 );
521 }
522
523 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
524 $this->mOutput->setLimitReportData( 'limitreport-walltime',
525 sprintf( "%.3f", $wallTime )
526 );
527
528 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
529 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
530 );
531 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
532 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
533 );
534 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
535 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
536 );
537 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
538 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
539 );
540 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
541 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
542 );
543 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
544 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
545 );
546 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
547
548 $limitReport = "NewPP limit report\n";
549 if ( $wgShowHostnames ) {
550 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
551 }
552 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
553 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
554 $limitReport .= 'Dynamic content: ' .
555 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
556 "\n";
557
558 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
559 if ( Hooks::run( 'ParserLimitReportFormat',
560 [ $key, &$value, &$limitReport, false, false ]
561 ) ) {
562 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
563 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
564 ->inLanguage( 'en' )->useDatabase( false );
565 if ( !$valueMsg->exists() ) {
566 $valueMsg = new RawMessage( '$1' );
567 }
568 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
569 $valueMsg->params( $value );
570 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
571 }
572 }
573 }
574 // Since we're not really outputting HTML, decode the entities and
575 // then re-encode the things that need hiding inside HTML comments.
576 $limitReport = htmlspecialchars_decode( $limitReport );
577 // Run deprecated hook
578 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
579
580 // Sanitize for comment. Note '‐' in the replacement is U+2010,
581 // which looks much like the problematic '-'.
582 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
583 $text = "\n<!-- \n$limitReport-->\n";
584
585 // Add on template profiling data in human/machine readable way
586 $dataByFunc = $this->mProfiler->getFunctionStats();
587 uasort( $dataByFunc, function ( $a, $b ) {
588 return $a['real'] < $b['real']; // descending order
589 } );
590 $profileReport = [];
591 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
592 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
593 $item['%real'], $item['real'], $item['calls'],
594 htmlspecialchars( $item['name'] ) );
595 }
596 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
597 $text .= implode( "\n", $profileReport ) . "\n-->\n";
598
599 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
600
601 // Add other cache related metadata
602 if ( $wgShowHostnames ) {
603 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
604 }
605 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
606 $this->mOutput->getCacheTime() );
607 $this->mOutput->setLimitReportData( 'cachereport-ttl',
608 $this->mOutput->getCacheExpiry() );
609 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
610 $this->mOutput->hasDynamicContent() );
611
612 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
613 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
614 $this->mTitle->getPrefixedDBkey() );
615 }
616 return $text;
617 }
618
619 /**
620 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
621 * can be called from an extension tag hook.
622 *
623 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
624 * instead, which means that lists and links have not been fully parsed yet,
625 * and strip markers are still present.
626 *
627 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
628 *
629 * Use this function if you're a parser tag hook and you want to parse
630 * wikitext before or after applying additional transformations, and you
631 * intend to *return the result as hook output*, which will cause it to go
632 * through the rest of parsing process automatically.
633 *
634 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
635 * $text are not expanded
636 *
637 * @param string $text Text extension wants to have parsed
638 * @param bool|PPFrame $frame The frame to use for expanding any template variables
639 * @return string UNSAFE half-parsed HTML
640 */
641 public function recursiveTagParse( $text, $frame = false ) {
642 // Avoid PHP 7.1 warning from passing $this by reference
643 $parser = $this;
644 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
645 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
646 $text = $this->internalParse( $text, false, $frame );
647 return $text;
648 }
649
650 /**
651 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
652 * point can be called from an extension tag hook.
653 *
654 * The output of this function is fully-parsed HTML that is safe for output.
655 * If you're a parser tag hook, you might want to use recursiveTagParse()
656 * instead.
657 *
658 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
659 * $text are not expanded
660 *
661 * @since 1.25
662 *
663 * @param string $text Text extension wants to have parsed
664 * @param bool|PPFrame $frame The frame to use for expanding any template variables
665 * @return string Fully parsed HTML
666 */
667 public function recursiveTagParseFully( $text, $frame = false ) {
668 $text = $this->recursiveTagParse( $text, $frame );
669 $text = $this->internalParseHalfParsed( $text, false );
670 return $text;
671 }
672
673 /**
674 * Expand templates and variables in the text, producing valid, static wikitext.
675 * Also removes comments.
676 * Do not call this function recursively.
677 * @param string $text
678 * @param Title $title
679 * @param ParserOptions $options
680 * @param int|null $revid
681 * @param bool|PPFrame $frame
682 * @return mixed|string
683 */
684 public function preprocess( $text, Title $title = null,
685 ParserOptions $options, $revid = null, $frame = false
686 ) {
687 $magicScopeVariable = $this->lock();
688 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
689 if ( $revid !== null ) {
690 $this->mRevisionId = $revid;
691 }
692 // Avoid PHP 7.1 warning from passing $this by reference
693 $parser = $this;
694 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
695 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
696 $text = $this->replaceVariables( $text, $frame );
697 $text = $this->mStripState->unstripBoth( $text );
698 return $text;
699 }
700
701 /**
702 * Recursive parser entry point that can be called from an extension tag
703 * hook.
704 *
705 * @param string $text Text to be expanded
706 * @param bool|PPFrame $frame The frame to use for expanding any template variables
707 * @return string
708 * @since 1.19
709 */
710 public function recursivePreprocess( $text, $frame = false ) {
711 $text = $this->replaceVariables( $text, $frame );
712 $text = $this->mStripState->unstripBoth( $text );
713 return $text;
714 }
715
716 /**
717 * Process the wikitext for the "?preload=" feature. (T7210)
718 *
719 * "<noinclude>", "<includeonly>" etc. are parsed as for template
720 * transclusion, comments, templates, arguments, tags hooks and parser
721 * functions are untouched.
722 *
723 * @param string $text
724 * @param Title $title
725 * @param ParserOptions $options
726 * @param array $params
727 * @return string
728 */
729 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
730 $msg = new RawMessage( $text );
731 $text = $msg->params( $params )->plain();
732
733 # Parser (re)initialisation
734 $magicScopeVariable = $this->lock();
735 $this->startParse( $title, $options, self::OT_PLAIN, true );
736
737 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
738 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
739 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
740 $text = $this->mStripState->unstripBoth( $text );
741 return $text;
742 }
743
744 /**
745 * Set the current user.
746 * Should only be used when doing pre-save transform.
747 *
748 * @param User|null $user User object or null (to reset)
749 */
750 public function setUser( $user ) {
751 $this->mUser = $user;
752 }
753
754 /**
755 * Set the context title
756 *
757 * @param Title $t
758 */
759 public function setTitle( $t ) {
760 if ( !$t ) {
761 $t = Title::newFromText( 'NO TITLE' );
762 }
763
764 if ( $t->hasFragment() ) {
765 # Strip the fragment to avoid various odd effects
766 $this->mTitle = $t->createFragmentTarget( '' );
767 } else {
768 $this->mTitle = $t;
769 }
770 }
771
772 /**
773 * Accessor for the Title object
774 *
775 * @return Title
776 */
777 public function getTitle() {
778 return $this->mTitle;
779 }
780
781 /**
782 * Accessor/mutator for the Title object
783 *
784 * @param Title $x Title object or null to just get the current one
785 * @return Title
786 */
787 public function Title( $x = null ) {
788 return wfSetVar( $this->mTitle, $x );
789 }
790
791 /**
792 * Set the output type
793 *
794 * @param int $ot New value
795 */
796 public function setOutputType( $ot ) {
797 $this->mOutputType = $ot;
798 # Shortcut alias
799 $this->ot = [
800 'html' => $ot == self::OT_HTML,
801 'wiki' => $ot == self::OT_WIKI,
802 'pre' => $ot == self::OT_PREPROCESS,
803 'plain' => $ot == self::OT_PLAIN,
804 ];
805 }
806
807 /**
808 * Accessor/mutator for the output type
809 *
810 * @param int|null $x New value or null to just get the current one
811 * @return int
812 */
813 public function OutputType( $x = null ) {
814 return wfSetVar( $this->mOutputType, $x );
815 }
816
817 /**
818 * Get the ParserOutput object
819 *
820 * @return ParserOutput
821 */
822 public function getOutput() {
823 return $this->mOutput;
824 }
825
826 /**
827 * Get the ParserOptions object
828 *
829 * @return ParserOptions
830 */
831 public function getOptions() {
832 return $this->mOptions;
833 }
834
835 /**
836 * Accessor/mutator for the ParserOptions object
837 *
838 * @param ParserOptions $x New value or null to just get the current one
839 * @return ParserOptions Current ParserOptions object
840 */
841 public function Options( $x = null ) {
842 return wfSetVar( $this->mOptions, $x );
843 }
844
845 /**
846 * @return int
847 */
848 public function nextLinkID() {
849 return $this->mLinkID++;
850 }
851
852 /**
853 * @param int $id
854 */
855 public function setLinkID( $id ) {
856 $this->mLinkID = $id;
857 }
858
859 /**
860 * Get a language object for use in parser functions such as {{FORMATNUM:}}
861 * @return Language
862 */
863 public function getFunctionLang() {
864 return $this->getTargetLanguage();
865 }
866
867 /**
868 * Get the target language for the content being parsed. This is usually the
869 * language that the content is in.
870 *
871 * @since 1.19
872 *
873 * @throws MWException
874 * @return Language
875 */
876 public function getTargetLanguage() {
877 $target = $this->mOptions->getTargetLanguage();
878
879 if ( $target !== null ) {
880 return $target;
881 } elseif ( $this->mOptions->getInterfaceMessage() ) {
882 return $this->mOptions->getUserLangObj();
883 } elseif ( is_null( $this->mTitle ) ) {
884 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
885 }
886
887 return $this->mTitle->getPageLanguage();
888 }
889
890 /**
891 * Get the language object for language conversion
892 * @return Language|null
893 */
894 public function getConverterLanguage() {
895 return $this->getTargetLanguage();
896 }
897
898 /**
899 * Get a User object either from $this->mUser, if set, or from the
900 * ParserOptions object otherwise
901 *
902 * @return User
903 */
904 public function getUser() {
905 if ( !is_null( $this->mUser ) ) {
906 return $this->mUser;
907 }
908 return $this->mOptions->getUser();
909 }
910
911 /**
912 * Get a preprocessor object
913 *
914 * @return Preprocessor
915 */
916 public function getPreprocessor() {
917 if ( !isset( $this->mPreprocessor ) ) {
918 $class = $this->mPreprocessorClass;
919 $this->mPreprocessor = new $class( $this );
920 }
921 return $this->mPreprocessor;
922 }
923
924 /**
925 * Get a LinkRenderer instance to make links with
926 *
927 * @since 1.28
928 * @return LinkRenderer
929 */
930 public function getLinkRenderer() {
931 if ( !$this->mLinkRenderer ) {
932 $this->mLinkRenderer = MediaWikiServices::getInstance()
933 ->getLinkRendererFactory()->create();
934 $this->mLinkRenderer->setStubThreshold(
935 $this->getOptions()->getStubThreshold()
936 );
937 }
938
939 return $this->mLinkRenderer;
940 }
941
942 /**
943 * Replaces all occurrences of HTML-style comments and the given tags
944 * in the text with a random marker and returns the next text. The output
945 * parameter $matches will be an associative array filled with data in
946 * the form:
947 *
948 * @code
949 * 'UNIQ-xxxxx' => [
950 * 'element',
951 * 'tag content',
952 * [ 'param' => 'x' ],
953 * '<element param="x">tag content</element>' ]
954 * @endcode
955 *
956 * @param array $elements List of element names. Comments are always extracted.
957 * @param string $text Source text string.
958 * @param array &$matches Out parameter, Array: extracted tags
959 * @return string Stripped text
960 */
961 public static function extractTagsAndParams( $elements, $text, &$matches ) {
962 static $n = 1;
963 $stripped = '';
964 $matches = [];
965
966 $taglist = implode( '|', $elements );
967 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
968
969 while ( $text != '' ) {
970 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
971 $stripped .= $p[0];
972 if ( count( $p ) < 5 ) {
973 break;
974 }
975 if ( count( $p ) > 5 ) {
976 # comment
977 $element = $p[4];
978 $attributes = '';
979 $close = '';
980 $inside = $p[5];
981 } else {
982 # tag
983 $element = $p[1];
984 $attributes = $p[2];
985 $close = $p[3];
986 $inside = $p[4];
987 }
988
989 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
990 $stripped .= $marker;
991
992 if ( $close === '/>' ) {
993 # Empty element tag, <tag />
994 $content = null;
995 $text = $inside;
996 $tail = null;
997 } else {
998 if ( $element === '!--' ) {
999 $end = '/(-->)/';
1000 } else {
1001 $end = "/(<\\/$element\\s*>)/i";
1002 }
1003 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1004 $content = $q[0];
1005 if ( count( $q ) < 3 ) {
1006 # No end tag -- let it run out to the end of the text.
1007 $tail = '';
1008 $text = '';
1009 } else {
1010 $tail = $q[1];
1011 $text = $q[2];
1012 }
1013 }
1014
1015 $matches[$marker] = [ $element,
1016 $content,
1017 Sanitizer::decodeTagAttributes( $attributes ),
1018 "<$element$attributes$close$content$tail" ];
1019 }
1020 return $stripped;
1021 }
1022
1023 /**
1024 * Get a list of strippable XML-like elements
1025 *
1026 * @return array
1027 */
1028 public function getStripList() {
1029 return $this->mStripList;
1030 }
1031
1032 /**
1033 * Add an item to the strip state
1034 * Returns the unique tag which must be inserted into the stripped text
1035 * The tag will be replaced with the original text in unstrip()
1036 *
1037 * @param string $text
1038 *
1039 * @return string
1040 */
1041 public function insertStripItem( $text ) {
1042 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1043 $this->mMarkerIndex++;
1044 $this->mStripState->addGeneral( $marker, $text );
1045 return $marker;
1046 }
1047
1048 /**
1049 * parse the wiki syntax used to render tables
1050 *
1051 * @private
1052 * @param string $text
1053 * @return string
1054 */
1055 public function doTableStuff( $text ) {
1056 $lines = StringUtils::explode( "\n", $text );
1057 $out = '';
1058 $td_history = []; # Is currently a td tag open?
1059 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1060 $tr_history = []; # Is currently a tr tag open?
1061 $tr_attributes = []; # history of tr attributes
1062 $has_opened_tr = []; # Did this table open a <tr> element?
1063 $indent_level = 0; # indent level of the table
1064
1065 foreach ( $lines as $outLine ) {
1066 $line = trim( $outLine );
1067
1068 if ( $line === '' ) { # empty line, go to next line
1069 $out .= $outLine . "\n";
1070 continue;
1071 }
1072
1073 $first_character = $line[0];
1074 $first_two = substr( $line, 0, 2 );
1075 $matches = [];
1076
1077 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1078 # First check if we are starting a new table
1079 $indent_level = strlen( $matches[1] );
1080
1081 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1082 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1083
1084 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1085 array_push( $td_history, false );
1086 array_push( $last_tag_history, '' );
1087 array_push( $tr_history, false );
1088 array_push( $tr_attributes, '' );
1089 array_push( $has_opened_tr, false );
1090 } elseif ( count( $td_history ) == 0 ) {
1091 # Don't do any of the following
1092 $out .= $outLine . "\n";
1093 continue;
1094 } elseif ( $first_two === '|}' ) {
1095 # We are ending a table
1096 $line = '</table>' . substr( $line, 2 );
1097 $last_tag = array_pop( $last_tag_history );
1098
1099 if ( !array_pop( $has_opened_tr ) ) {
1100 $line = "<tr><td></td></tr>{$line}";
1101 }
1102
1103 if ( array_pop( $tr_history ) ) {
1104 $line = "</tr>{$line}";
1105 }
1106
1107 if ( array_pop( $td_history ) ) {
1108 $line = "</{$last_tag}>{$line}";
1109 }
1110 array_pop( $tr_attributes );
1111 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1112 } elseif ( $first_two === '|-' ) {
1113 # Now we have a table row
1114 $line = preg_replace( '#^\|-+#', '', $line );
1115
1116 # Whats after the tag is now only attributes
1117 $attributes = $this->mStripState->unstripBoth( $line );
1118 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1119 array_pop( $tr_attributes );
1120 array_push( $tr_attributes, $attributes );
1121
1122 $line = '';
1123 $last_tag = array_pop( $last_tag_history );
1124 array_pop( $has_opened_tr );
1125 array_push( $has_opened_tr, true );
1126
1127 if ( array_pop( $tr_history ) ) {
1128 $line = '</tr>';
1129 }
1130
1131 if ( array_pop( $td_history ) ) {
1132 $line = "</{$last_tag}>{$line}";
1133 }
1134
1135 $outLine = $line;
1136 array_push( $tr_history, false );
1137 array_push( $td_history, false );
1138 array_push( $last_tag_history, '' );
1139 } elseif ( $first_character === '|'
1140 || $first_character === '!'
1141 || $first_two === '|+'
1142 ) {
1143 # This might be cell elements, td, th or captions
1144 if ( $first_two === '|+' ) {
1145 $first_character = '+';
1146 $line = substr( $line, 2 );
1147 } else {
1148 $line = substr( $line, 1 );
1149 }
1150
1151 // Implies both are valid for table headings.
1152 if ( $first_character === '!' ) {
1153 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1154 }
1155
1156 # Split up multiple cells on the same line.
1157 # FIXME : This can result in improper nesting of tags processed
1158 # by earlier parser steps.
1159 $cells = explode( '||', $line );
1160
1161 $outLine = '';
1162
1163 # Loop through each table cell
1164 foreach ( $cells as $cell ) {
1165 $previous = '';
1166 if ( $first_character !== '+' ) {
1167 $tr_after = array_pop( $tr_attributes );
1168 if ( !array_pop( $tr_history ) ) {
1169 $previous = "<tr{$tr_after}>\n";
1170 }
1171 array_push( $tr_history, true );
1172 array_push( $tr_attributes, '' );
1173 array_pop( $has_opened_tr );
1174 array_push( $has_opened_tr, true );
1175 }
1176
1177 $last_tag = array_pop( $last_tag_history );
1178
1179 if ( array_pop( $td_history ) ) {
1180 $previous = "</{$last_tag}>\n{$previous}";
1181 }
1182
1183 if ( $first_character === '|' ) {
1184 $last_tag = 'td';
1185 } elseif ( $first_character === '!' ) {
1186 $last_tag = 'th';
1187 } elseif ( $first_character === '+' ) {
1188 $last_tag = 'caption';
1189 } else {
1190 $last_tag = '';
1191 }
1192
1193 array_push( $last_tag_history, $last_tag );
1194
1195 # A cell could contain both parameters and data
1196 $cell_data = explode( '|', $cell, 2 );
1197
1198 # T2553: Note that a '|' inside an invalid link should not
1199 # be mistaken as delimiting cell parameters
1200 # Bug T153140: Neither should language converter markup.
1201 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1202 $cell = "{$previous}<{$last_tag}>{$cell}";
1203 } elseif ( count( $cell_data ) == 1 ) {
1204 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1205 } else {
1206 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1207 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1208 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1209 }
1210
1211 $outLine .= $cell;
1212 array_push( $td_history, true );
1213 }
1214 }
1215 $out .= $outLine . "\n";
1216 }
1217
1218 # Closing open td, tr && table
1219 while ( count( $td_history ) > 0 ) {
1220 if ( array_pop( $td_history ) ) {
1221 $out .= "</td>\n";
1222 }
1223 if ( array_pop( $tr_history ) ) {
1224 $out .= "</tr>\n";
1225 }
1226 if ( !array_pop( $has_opened_tr ) ) {
1227 $out .= "<tr><td></td></tr>\n";
1228 }
1229
1230 $out .= "</table>\n";
1231 }
1232
1233 # Remove trailing line-ending (b/c)
1234 if ( substr( $out, -1 ) === "\n" ) {
1235 $out = substr( $out, 0, -1 );
1236 }
1237
1238 # special case: don't return empty table
1239 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1240 $out = '';
1241 }
1242
1243 return $out;
1244 }
1245
1246 /**
1247 * Helper function for parse() that transforms wiki markup into half-parsed
1248 * HTML. Only called for $mOutputType == self::OT_HTML.
1249 *
1250 * @private
1251 *
1252 * @param string $text The text to parse
1253 * @param bool $isMain Whether this is being called from the main parse() function
1254 * @param PPFrame|bool $frame A pre-processor frame
1255 *
1256 * @return string
1257 */
1258 public function internalParse( $text, $isMain = true, $frame = false ) {
1259 $origText = $text;
1260
1261 // Avoid PHP 7.1 warning from passing $this by reference
1262 $parser = $this;
1263
1264 # Hook to suspend the parser in this state
1265 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1266 return $text;
1267 }
1268
1269 # if $frame is provided, then use $frame for replacing any variables
1270 if ( $frame ) {
1271 # use frame depth to infer how include/noinclude tags should be handled
1272 # depth=0 means this is the top-level document; otherwise it's an included document
1273 if ( !$frame->depth ) {
1274 $flag = 0;
1275 } else {
1276 $flag = self::PTD_FOR_INCLUSION;
1277 }
1278 $dom = $this->preprocessToDom( $text, $flag );
1279 $text = $frame->expand( $dom );
1280 } else {
1281 # if $frame is not provided, then use old-style replaceVariables
1282 $text = $this->replaceVariables( $text );
1283 }
1284
1285 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1286 $text = Sanitizer::removeHTMLtags(
1287 $text,
1288 [ $this, 'attributeStripCallback' ],
1289 false,
1290 array_keys( $this->mTransparentTagHooks ),
1291 [],
1292 [ $this, 'addTrackingCategory' ]
1293 );
1294 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1295
1296 # Tables need to come after variable replacement for things to work
1297 # properly; putting them before other transformations should keep
1298 # exciting things like link expansions from showing up in surprising
1299 # places.
1300 $text = $this->doTableStuff( $text );
1301
1302 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1303
1304 $text = $this->doDoubleUnderscore( $text );
1305
1306 $text = $this->doHeadings( $text );
1307 $text = $this->replaceInternalLinks( $text );
1308 $text = $this->doAllQuotes( $text );
1309 $text = $this->replaceExternalLinks( $text );
1310
1311 # replaceInternalLinks may sometimes leave behind
1312 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1313 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1314
1315 $text = $this->doMagicLinks( $text );
1316 $text = $this->formatHeadings( $text, $origText, $isMain );
1317
1318 return $text;
1319 }
1320
1321 /**
1322 * Helper function for parse() that transforms half-parsed HTML into fully
1323 * parsed HTML.
1324 *
1325 * @param string $text
1326 * @param bool $isMain
1327 * @param bool $linestart
1328 * @return string
1329 */
1330 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1331 $text = $this->mStripState->unstripGeneral( $text );
1332
1333 // Avoid PHP 7.1 warning from passing $this by reference
1334 $parser = $this;
1335
1336 if ( $isMain ) {
1337 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1338 }
1339
1340 # Clean up special characters, only run once, next-to-last before doBlockLevels
1341 $fixtags = [
1342 # French spaces, last one Guillemet-left
1343 # only if there is something before the space
1344 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1345 # french spaces, Guillemet-right
1346 '/(\\302\\253) /' => '\\1&#160;',
1347 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1348 ];
1349 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1350
1351 $text = $this->doBlockLevels( $text, $linestart );
1352
1353 $this->replaceLinkHolders( $text );
1354
1355 /**
1356 * The input doesn't get language converted if
1357 * a) It's disabled
1358 * b) Content isn't converted
1359 * c) It's a conversion table
1360 * d) it is an interface message (which is in the user language)
1361 */
1362 if ( !( $this->mOptions->getDisableContentConversion()
1363 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1364 ) {
1365 if ( !$this->mOptions->getInterfaceMessage() ) {
1366 # The position of the convert() call should not be changed. it
1367 # assumes that the links are all replaced and the only thing left
1368 # is the <nowiki> mark.
1369 $text = $this->getConverterLanguage()->convert( $text );
1370 }
1371 }
1372
1373 $text = $this->mStripState->unstripNoWiki( $text );
1374
1375 if ( $isMain ) {
1376 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1377 }
1378
1379 $text = $this->replaceTransparentTags( $text );
1380 $text = $this->mStripState->unstripGeneral( $text );
1381
1382 $text = Sanitizer::normalizeCharReferences( $text );
1383
1384 if ( MWTidy::isEnabled() ) {
1385 if ( $this->mOptions->getTidy() ) {
1386 $text = MWTidy::tidy( $text );
1387 }
1388 } else {
1389 # attempt to sanitize at least some nesting problems
1390 # (T4702 and quite a few others)
1391 $tidyregs = [
1392 # ''Something [http://www.cool.com cool''] -->
1393 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1394 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1395 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1396 # fix up an anchor inside another anchor, only
1397 # at least for a single single nested link (T5695)
1398 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1399 '\\1\\2</a>\\3</a>\\1\\4</a>',
1400 # fix div inside inline elements- doBlockLevels won't wrap a line which
1401 # contains a div, so fix it up here; replace
1402 # div with escaped text
1403 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1404 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1405 # remove empty italic or bold tag pairs, some
1406 # introduced by rules above
1407 '/<([bi])><\/\\1>/' => '',
1408 ];
1409
1410 $text = preg_replace(
1411 array_keys( $tidyregs ),
1412 array_values( $tidyregs ),
1413 $text );
1414 }
1415
1416 if ( $isMain ) {
1417 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1418 }
1419
1420 return $text;
1421 }
1422
1423 /**
1424 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1425 * magic external links.
1426 *
1427 * DML
1428 * @private
1429 *
1430 * @param string $text
1431 *
1432 * @return string
1433 */
1434 public function doMagicLinks( $text ) {
1435 $prots = wfUrlProtocolsWithoutProtRel();
1436 $urlChar = self::EXT_LINK_URL_CLASS;
1437 $addr = self::EXT_LINK_ADDR;
1438 $space = self::SPACE_NOT_NL; # non-newline space
1439 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1440 $spaces = "$space++"; # possessive match of 1 or more spaces
1441 $text = preg_replace_callback(
1442 '!(?: # Start cases
1443 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1444 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1445 (\b # m[3]: Free external links
1446 (?i:$prots)
1447 ($addr$urlChar*) # m[4]: Post-protocol path
1448 ) |
1449 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1450 ([0-9]+)\b |
1451 \bISBN $spaces ( # m[6]: ISBN, capture number
1452 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1453 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1454 [0-9Xx] # check digit
1455 )\b
1456 )!xu", [ $this, 'magicLinkCallback' ], $text );
1457 return $text;
1458 }
1459
1460 /**
1461 * @throws MWException
1462 * @param array $m
1463 * @return HTML|string
1464 */
1465 public function magicLinkCallback( $m ) {
1466 if ( isset( $m[1] ) && $m[1] !== '' ) {
1467 # Skip anchor
1468 return $m[0];
1469 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1470 # Skip HTML element
1471 return $m[0];
1472 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1473 # Free external link
1474 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1475 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1476 # RFC or PMID
1477 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1478 if ( !$this->mOptions->getMagicRFCLinks() ) {
1479 return $m[0];
1480 }
1481 $keyword = 'RFC';
1482 $urlmsg = 'rfcurl';
1483 $cssClass = 'mw-magiclink-rfc';
1484 $trackingCat = 'magiclink-tracking-rfc';
1485 $id = $m[5];
1486 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1487 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1488 return $m[0];
1489 }
1490 $keyword = 'PMID';
1491 $urlmsg = 'pubmedurl';
1492 $cssClass = 'mw-magiclink-pmid';
1493 $trackingCat = 'magiclink-tracking-pmid';
1494 $id = $m[5];
1495 } else {
1496 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1497 substr( $m[0], 0, 20 ) . '"' );
1498 }
1499 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1500 $this->addTrackingCategory( $trackingCat );
1501 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1502 } elseif ( isset( $m[6] ) && $m[6] !== ''
1503 && $this->mOptions->getMagicISBNLinks()
1504 ) {
1505 # ISBN
1506 $isbn = $m[6];
1507 $space = self::SPACE_NOT_NL; # non-newline space
1508 $isbn = preg_replace( "/$space/", ' ', $isbn );
1509 $num = strtr( $isbn, [
1510 '-' => '',
1511 ' ' => '',
1512 'x' => 'X',
1513 ] );
1514 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1515 return $this->getLinkRenderer()->makeKnownLink(
1516 SpecialPage::getTitleFor( 'Booksources', $num ),
1517 "ISBN $isbn",
1518 [
1519 'class' => 'internal mw-magiclink-isbn',
1520 'title' => false // suppress title attribute
1521 ]
1522 );
1523 } else {
1524 return $m[0];
1525 }
1526 }
1527
1528 /**
1529 * Make a free external link, given a user-supplied URL
1530 *
1531 * @param string $url
1532 * @param int $numPostProto
1533 * The number of characters after the protocol.
1534 * @return string HTML
1535 * @private
1536 */
1537 public function makeFreeExternalLink( $url, $numPostProto ) {
1538 $trail = '';
1539
1540 # The characters '<' and '>' (which were escaped by
1541 # removeHTMLtags()) should not be included in
1542 # URLs, per RFC 2396.
1543 # Make &nbsp; terminate a URL as well (bug T84937)
1544 $m2 = [];
1545 if ( preg_match(
1546 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1547 $url,
1548 $m2,
1549 PREG_OFFSET_CAPTURE
1550 ) ) {
1551 $trail = substr( $url, $m2[0][1] ) . $trail;
1552 $url = substr( $url, 0, $m2[0][1] );
1553 }
1554
1555 # Move trailing punctuation to $trail
1556 $sep = ',;\.:!?';
1557 # If there is no left bracket, then consider right brackets fair game too
1558 if ( strpos( $url, '(' ) === false ) {
1559 $sep .= ')';
1560 }
1561
1562 $urlRev = strrev( $url );
1563 $numSepChars = strspn( $urlRev, $sep );
1564 # Don't break a trailing HTML entity by moving the ; into $trail
1565 # This is in hot code, so use substr_compare to avoid having to
1566 # create a new string object for the comparison
1567 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1568 # more optimization: instead of running preg_match with a $
1569 # anchor, which can be slow, do the match on the reversed
1570 # string starting at the desired offset.
1571 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1572 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1573 $numSepChars--;
1574 }
1575 }
1576 if ( $numSepChars ) {
1577 $trail = substr( $url, -$numSepChars ) . $trail;
1578 $url = substr( $url, 0, -$numSepChars );
1579 }
1580
1581 # Verify that we still have a real URL after trail removal, and
1582 # not just lone protocol
1583 if ( strlen( $trail ) >= $numPostProto ) {
1584 return $url . $trail;
1585 }
1586
1587 $url = Sanitizer::cleanUrl( $url );
1588
1589 # Is this an external image?
1590 $text = $this->maybeMakeExternalImage( $url );
1591 if ( $text === false ) {
1592 # Not an image, make a link
1593 $text = Linker::makeExternalLink( $url,
1594 $this->getConverterLanguage()->markNoConversion( $url, true ),
1595 true, 'free',
1596 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1597 # Register it in the output object...
1598 $this->mOutput->addExternalLink( $url );
1599 }
1600 return $text . $trail;
1601 }
1602
1603 /**
1604 * Parse headers and return html
1605 *
1606 * @private
1607 *
1608 * @param string $text
1609 *
1610 * @return string
1611 */
1612 public function doHeadings( $text ) {
1613 for ( $i = 6; $i >= 1; --$i ) {
1614 $h = str_repeat( '=', $i );
1615 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1616 }
1617 return $text;
1618 }
1619
1620 /**
1621 * Replace single quotes with HTML markup
1622 * @private
1623 *
1624 * @param string $text
1625 *
1626 * @return string The altered text
1627 */
1628 public function doAllQuotes( $text ) {
1629 $outtext = '';
1630 $lines = StringUtils::explode( "\n", $text );
1631 foreach ( $lines as $line ) {
1632 $outtext .= $this->doQuotes( $line ) . "\n";
1633 }
1634 $outtext = substr( $outtext, 0, -1 );
1635 return $outtext;
1636 }
1637
1638 /**
1639 * Helper function for doAllQuotes()
1640 *
1641 * @param string $text
1642 *
1643 * @return string
1644 */
1645 public function doQuotes( $text ) {
1646 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1647 $countarr = count( $arr );
1648 if ( $countarr == 1 ) {
1649 return $text;
1650 }
1651
1652 // First, do some preliminary work. This may shift some apostrophes from
1653 // being mark-up to being text. It also counts the number of occurrences
1654 // of bold and italics mark-ups.
1655 $numbold = 0;
1656 $numitalics = 0;
1657 for ( $i = 1; $i < $countarr; $i += 2 ) {
1658 $thislen = strlen( $arr[$i] );
1659 // If there are ever four apostrophes, assume the first is supposed to
1660 // be text, and the remaining three constitute mark-up for bold text.
1661 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1662 if ( $thislen == 4 ) {
1663 $arr[$i - 1] .= "'";
1664 $arr[$i] = "'''";
1665 $thislen = 3;
1666 } elseif ( $thislen > 5 ) {
1667 // If there are more than 5 apostrophes in a row, assume they're all
1668 // text except for the last 5.
1669 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1670 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1671 $arr[$i] = "'''''";
1672 $thislen = 5;
1673 }
1674 // Count the number of occurrences of bold and italics mark-ups.
1675 if ( $thislen == 2 ) {
1676 $numitalics++;
1677 } elseif ( $thislen == 3 ) {
1678 $numbold++;
1679 } elseif ( $thislen == 5 ) {
1680 $numitalics++;
1681 $numbold++;
1682 }
1683 }
1684
1685 // If there is an odd number of both bold and italics, it is likely
1686 // that one of the bold ones was meant to be an apostrophe followed
1687 // by italics. Which one we cannot know for certain, but it is more
1688 // likely to be one that has a single-letter word before it.
1689 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1690 $firstsingleletterword = -1;
1691 $firstmultiletterword = -1;
1692 $firstspace = -1;
1693 for ( $i = 1; $i < $countarr; $i += 2 ) {
1694 if ( strlen( $arr[$i] ) == 3 ) {
1695 $x1 = substr( $arr[$i - 1], -1 );
1696 $x2 = substr( $arr[$i - 1], -2, 1 );
1697 if ( $x1 === ' ' ) {
1698 if ( $firstspace == -1 ) {
1699 $firstspace = $i;
1700 }
1701 } elseif ( $x2 === ' ' ) {
1702 $firstsingleletterword = $i;
1703 // if $firstsingleletterword is set, we don't
1704 // look at the other options, so we can bail early.
1705 break;
1706 } else {
1707 if ( $firstmultiletterword == -1 ) {
1708 $firstmultiletterword = $i;
1709 }
1710 }
1711 }
1712 }
1713
1714 // If there is a single-letter word, use it!
1715 if ( $firstsingleletterword > -1 ) {
1716 $arr[$firstsingleletterword] = "''";
1717 $arr[$firstsingleletterword - 1] .= "'";
1718 } elseif ( $firstmultiletterword > -1 ) {
1719 // If not, but there's a multi-letter word, use that one.
1720 $arr[$firstmultiletterword] = "''";
1721 $arr[$firstmultiletterword - 1] .= "'";
1722 } elseif ( $firstspace > -1 ) {
1723 // ... otherwise use the first one that has neither.
1724 // (notice that it is possible for all three to be -1 if, for example,
1725 // there is only one pentuple-apostrophe in the line)
1726 $arr[$firstspace] = "''";
1727 $arr[$firstspace - 1] .= "'";
1728 }
1729 }
1730
1731 // Now let's actually convert our apostrophic mush to HTML!
1732 $output = '';
1733 $buffer = '';
1734 $state = '';
1735 $i = 0;
1736 foreach ( $arr as $r ) {
1737 if ( ( $i % 2 ) == 0 ) {
1738 if ( $state === 'both' ) {
1739 $buffer .= $r;
1740 } else {
1741 $output .= $r;
1742 }
1743 } else {
1744 $thislen = strlen( $r );
1745 if ( $thislen == 2 ) {
1746 if ( $state === 'i' ) {
1747 $output .= '</i>';
1748 $state = '';
1749 } elseif ( $state === 'bi' ) {
1750 $output .= '</i>';
1751 $state = 'b';
1752 } elseif ( $state === 'ib' ) {
1753 $output .= '</b></i><b>';
1754 $state = 'b';
1755 } elseif ( $state === 'both' ) {
1756 $output .= '<b><i>' . $buffer . '</i>';
1757 $state = 'b';
1758 } else { // $state can be 'b' or ''
1759 $output .= '<i>';
1760 $state .= 'i';
1761 }
1762 } elseif ( $thislen == 3 ) {
1763 if ( $state === 'b' ) {
1764 $output .= '</b>';
1765 $state = '';
1766 } elseif ( $state === 'bi' ) {
1767 $output .= '</i></b><i>';
1768 $state = 'i';
1769 } elseif ( $state === 'ib' ) {
1770 $output .= '</b>';
1771 $state = 'i';
1772 } elseif ( $state === 'both' ) {
1773 $output .= '<i><b>' . $buffer . '</b>';
1774 $state = 'i';
1775 } else { // $state can be 'i' or ''
1776 $output .= '<b>';
1777 $state .= 'b';
1778 }
1779 } elseif ( $thislen == 5 ) {
1780 if ( $state === 'b' ) {
1781 $output .= '</b><i>';
1782 $state = 'i';
1783 } elseif ( $state === 'i' ) {
1784 $output .= '</i><b>';
1785 $state = 'b';
1786 } elseif ( $state === 'bi' ) {
1787 $output .= '</i></b>';
1788 $state = '';
1789 } elseif ( $state === 'ib' ) {
1790 $output .= '</b></i>';
1791 $state = '';
1792 } elseif ( $state === 'both' ) {
1793 $output .= '<i><b>' . $buffer . '</b></i>';
1794 $state = '';
1795 } else { // ($state == '')
1796 $buffer = '';
1797 $state = 'both';
1798 }
1799 }
1800 }
1801 $i++;
1802 }
1803 // Now close all remaining tags. Notice that the order is important.
1804 if ( $state === 'b' || $state === 'ib' ) {
1805 $output .= '</b>';
1806 }
1807 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1808 $output .= '</i>';
1809 }
1810 if ( $state === 'bi' ) {
1811 $output .= '</b>';
1812 }
1813 // There might be lonely ''''', so make sure we have a buffer
1814 if ( $state === 'both' && $buffer ) {
1815 $output .= '<b><i>' . $buffer . '</i></b>';
1816 }
1817 return $output;
1818 }
1819
1820 /**
1821 * Replace external links (REL)
1822 *
1823 * Note: this is all very hackish and the order of execution matters a lot.
1824 * Make sure to run tests/parser/parserTests.php if you change this code.
1825 *
1826 * @private
1827 *
1828 * @param string $text
1829 *
1830 * @throws MWException
1831 * @return string
1832 */
1833 public function replaceExternalLinks( $text ) {
1834 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1835 if ( $bits === false ) {
1836 throw new MWException( "PCRE needs to be compiled with "
1837 . "--enable-unicode-properties in order for MediaWiki to function" );
1838 }
1839 $s = array_shift( $bits );
1840
1841 $i = 0;
1842 while ( $i < count( $bits ) ) {
1843 $url = $bits[$i++];
1844 $i++; // protocol
1845 $text = $bits[$i++];
1846 $trail = $bits[$i++];
1847
1848 # The characters '<' and '>' (which were escaped by
1849 # removeHTMLtags()) should not be included in
1850 # URLs, per RFC 2396.
1851 $m2 = [];
1852 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1853 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1854 $url = substr( $url, 0, $m2[0][1] );
1855 }
1856
1857 # If the link text is an image URL, replace it with an <img> tag
1858 # This happened by accident in the original parser, but some people used it extensively
1859 $img = $this->maybeMakeExternalImage( $text );
1860 if ( $img !== false ) {
1861 $text = $img;
1862 }
1863
1864 $dtrail = '';
1865
1866 # Set linktype for CSS - if URL==text, link is essentially free
1867 $linktype = ( $text === $url ) ? 'free' : 'text';
1868
1869 # No link text, e.g. [http://domain.tld/some.link]
1870 if ( $text == '' ) {
1871 # Autonumber
1872 $langObj = $this->getTargetLanguage();
1873 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1874 $linktype = 'autonumber';
1875 } else {
1876 # Have link text, e.g. [http://domain.tld/some.link text]s
1877 # Check for trail
1878 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1879 }
1880
1881 $text = $this->getConverterLanguage()->markNoConversion( $text );
1882
1883 $url = Sanitizer::cleanUrl( $url );
1884
1885 # Use the encoded URL
1886 # This means that users can paste URLs directly into the text
1887 # Funny characters like ö aren't valid in URLs anyway
1888 # This was changed in August 2004
1889 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1890 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1891
1892 # Register link in the output object.
1893 $this->mOutput->addExternalLink( $url );
1894 }
1895
1896 return $s;
1897 }
1898
1899 /**
1900 * Get the rel attribute for a particular external link.
1901 *
1902 * @since 1.21
1903 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1904 * nofollow if appropriate
1905 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1906 * @return string|null Rel attribute for $url
1907 */
1908 public static function getExternalLinkRel( $url = false, $title = null ) {
1909 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1910 $ns = $title ? $title->getNamespace() : false;
1911 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1912 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1913 ) {
1914 return 'nofollow';
1915 }
1916 return null;
1917 }
1918
1919 /**
1920 * Get an associative array of additional HTML attributes appropriate for a
1921 * particular external link. This currently may include rel => nofollow
1922 * (depending on configuration, namespace, and the URL's domain) and/or a
1923 * target attribute (depending on configuration).
1924 *
1925 * @param string $url URL to extract the domain from for rel =>
1926 * nofollow if appropriate
1927 * @return array Associative array of HTML attributes
1928 */
1929 public function getExternalLinkAttribs( $url ) {
1930 $attribs = [];
1931 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1932
1933 $target = $this->mOptions->getExternalLinkTarget();
1934 if ( $target ) {
1935 $attribs['target'] = $target;
1936 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1937 // T133507. New windows can navigate parent cross-origin.
1938 // Including noreferrer due to lacking browser
1939 // support of noopener. Eventually noreferrer should be removed.
1940 if ( $rel !== '' ) {
1941 $rel .= ' ';
1942 }
1943 $rel .= 'noreferrer noopener';
1944 }
1945 }
1946 $attribs['rel'] = $rel;
1947 return $attribs;
1948 }
1949
1950 /**
1951 * Replace unusual escape codes in a URL with their equivalent characters
1952 *
1953 * This generally follows the syntax defined in RFC 3986, with special
1954 * consideration for HTTP query strings.
1955 *
1956 * @param string $url
1957 * @return string
1958 */
1959 public static function normalizeLinkUrl( $url ) {
1960 # First, make sure unsafe characters are encoded
1961 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1962 function ( $m ) {
1963 return rawurlencode( $m[0] );
1964 },
1965 $url
1966 );
1967
1968 $ret = '';
1969 $end = strlen( $url );
1970
1971 # Fragment part - 'fragment'
1972 $start = strpos( $url, '#' );
1973 if ( $start !== false && $start < $end ) {
1974 $ret = self::normalizeUrlComponent(
1975 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1976 $end = $start;
1977 }
1978
1979 # Query part - 'query' minus &=+;
1980 $start = strpos( $url, '?' );
1981 if ( $start !== false && $start < $end ) {
1982 $ret = self::normalizeUrlComponent(
1983 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1984 $end = $start;
1985 }
1986
1987 # Scheme and path part - 'pchar'
1988 # (we assume no userinfo or encoded colons in the host)
1989 $ret = self::normalizeUrlComponent(
1990 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1991
1992 return $ret;
1993 }
1994
1995 private static function normalizeUrlComponent( $component, $unsafe ) {
1996 $callback = function ( $matches ) use ( $unsafe ) {
1997 $char = urldecode( $matches[0] );
1998 $ord = ord( $char );
1999 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2000 # Unescape it
2001 return $char;
2002 } else {
2003 # Leave it escaped, but use uppercase for a-f
2004 return strtoupper( $matches[0] );
2005 }
2006 };
2007 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2008 }
2009
2010 /**
2011 * make an image if it's allowed, either through the global
2012 * option, through the exception, or through the on-wiki whitelist
2013 *
2014 * @param string $url
2015 *
2016 * @return string
2017 */
2018 private function maybeMakeExternalImage( $url ) {
2019 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2020 $imagesexception = !empty( $imagesfrom );
2021 $text = false;
2022 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2023 if ( $imagesexception && is_array( $imagesfrom ) ) {
2024 $imagematch = false;
2025 foreach ( $imagesfrom as $match ) {
2026 if ( strpos( $url, $match ) === 0 ) {
2027 $imagematch = true;
2028 break;
2029 }
2030 }
2031 } elseif ( $imagesexception ) {
2032 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2033 } else {
2034 $imagematch = false;
2035 }
2036
2037 if ( $this->mOptions->getAllowExternalImages()
2038 || ( $imagesexception && $imagematch )
2039 ) {
2040 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2041 # Image found
2042 $text = Linker::makeExternalImage( $url );
2043 }
2044 }
2045 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2046 && preg_match( self::EXT_IMAGE_REGEX, $url )
2047 ) {
2048 $whitelist = explode(
2049 "\n",
2050 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2051 );
2052
2053 foreach ( $whitelist as $entry ) {
2054 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2055 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2056 continue;
2057 }
2058 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2059 # Image matches a whitelist entry
2060 $text = Linker::makeExternalImage( $url );
2061 break;
2062 }
2063 }
2064 }
2065 return $text;
2066 }
2067
2068 /**
2069 * Process [[ ]] wikilinks
2070 *
2071 * @param string $s
2072 *
2073 * @return string Processed text
2074 *
2075 * @private
2076 */
2077 public function replaceInternalLinks( $s ) {
2078 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2079 return $s;
2080 }
2081
2082 /**
2083 * Process [[ ]] wikilinks (RIL)
2084 * @param string &$s
2085 * @throws MWException
2086 * @return LinkHolderArray
2087 *
2088 * @private
2089 */
2090 public function replaceInternalLinks2( &$s ) {
2091 global $wgExtraInterlanguageLinkPrefixes;
2092
2093 static $tc = false, $e1, $e1_img;
2094 # the % is needed to support urlencoded titles as well
2095 if ( !$tc ) {
2096 $tc = Title::legalChars() . '#%';
2097 # Match a link having the form [[namespace:link|alternate]]trail
2098 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2099 # Match cases where there is no "]]", which might still be images
2100 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2101 }
2102
2103 $holders = new LinkHolderArray( $this );
2104
2105 # split the entire text string on occurrences of [[
2106 $a = StringUtils::explode( '[[', ' ' . $s );
2107 # get the first element (all text up to first [[), and remove the space we added
2108 $s = $a->current();
2109 $a->next();
2110 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2111 $s = substr( $s, 1 );
2112
2113 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2114 $e2 = null;
2115 if ( $useLinkPrefixExtension ) {
2116 # Match the end of a line for a word that's not followed by whitespace,
2117 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2118 global $wgContLang;
2119 $charset = $wgContLang->linkPrefixCharset();
2120 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2121 }
2122
2123 if ( is_null( $this->mTitle ) ) {
2124 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2125 }
2126 $nottalk = !$this->mTitle->isTalkPage();
2127
2128 if ( $useLinkPrefixExtension ) {
2129 $m = [];
2130 if ( preg_match( $e2, $s, $m ) ) {
2131 $first_prefix = $m[2];
2132 } else {
2133 $first_prefix = false;
2134 }
2135 } else {
2136 $prefix = '';
2137 }
2138
2139 $useSubpages = $this->areSubpagesAllowed();
2140
2141 # Loop for each link
2142 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2143 # Check for excessive memory usage
2144 if ( $holders->isBig() ) {
2145 # Too big
2146 # Do the existence check, replace the link holders and clear the array
2147 $holders->replace( $s );
2148 $holders->clear();
2149 }
2150
2151 if ( $useLinkPrefixExtension ) {
2152 if ( preg_match( $e2, $s, $m ) ) {
2153 $prefix = $m[2];
2154 $s = $m[1];
2155 } else {
2156 $prefix = '';
2157 }
2158 # first link
2159 if ( $first_prefix ) {
2160 $prefix = $first_prefix;
2161 $first_prefix = false;
2162 }
2163 }
2164
2165 $might_be_img = false;
2166
2167 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2168 $text = $m[2];
2169 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2170 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2171 # the real problem is with the $e1 regex
2172 # See T1500.
2173 # Still some problems for cases where the ] is meant to be outside punctuation,
2174 # and no image is in sight. See T4095.
2175 if ( $text !== ''
2176 && substr( $m[3], 0, 1 ) === ']'
2177 && strpos( $text, '[' ) !== false
2178 ) {
2179 $text .= ']'; # so that replaceExternalLinks($text) works later
2180 $m[3] = substr( $m[3], 1 );
2181 }
2182 # fix up urlencoded title texts
2183 if ( strpos( $m[1], '%' ) !== false ) {
2184 # Should anchors '#' also be rejected?
2185 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2186 }
2187 $trail = $m[3];
2188 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2189 # Invalid, but might be an image with a link in its caption
2190 $might_be_img = true;
2191 $text = $m[2];
2192 if ( strpos( $m[1], '%' ) !== false ) {
2193 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2194 }
2195 $trail = "";
2196 } else { # Invalid form; output directly
2197 $s .= $prefix . '[[' . $line;
2198 continue;
2199 }
2200
2201 $origLink = ltrim( $m[1], ' ' );
2202
2203 # Don't allow internal links to pages containing
2204 # PROTO: where PROTO is a valid URL protocol; these
2205 # should be external links.
2206 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2207 $s .= $prefix . '[[' . $line;
2208 continue;
2209 }
2210
2211 # Make subpage if necessary
2212 if ( $useSubpages ) {
2213 $link = $this->maybeDoSubpageLink( $origLink, $text );
2214 } else {
2215 $link = $origLink;
2216 }
2217
2218 $unstrip = $this->mStripState->unstripNoWiki( $link );
2219 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2220 if ( $nt === null ) {
2221 $s .= $prefix . '[[' . $line;
2222 continue;
2223 }
2224
2225 $ns = $nt->getNamespace();
2226 $iw = $nt->getInterwiki();
2227
2228 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2229
2230 if ( $might_be_img ) { # if this is actually an invalid link
2231 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2232 $found = false;
2233 while ( true ) {
2234 # look at the next 'line' to see if we can close it there
2235 $a->next();
2236 $next_line = $a->current();
2237 if ( $next_line === false || $next_line === null ) {
2238 break;
2239 }
2240 $m = explode( ']]', $next_line, 3 );
2241 if ( count( $m ) == 3 ) {
2242 # the first ]] closes the inner link, the second the image
2243 $found = true;
2244 $text .= "[[{$m[0]}]]{$m[1]}";
2245 $trail = $m[2];
2246 break;
2247 } elseif ( count( $m ) == 2 ) {
2248 # if there's exactly one ]] that's fine, we'll keep looking
2249 $text .= "[[{$m[0]}]]{$m[1]}";
2250 } else {
2251 # if $next_line is invalid too, we need look no further
2252 $text .= '[[' . $next_line;
2253 break;
2254 }
2255 }
2256 if ( !$found ) {
2257 # we couldn't find the end of this imageLink, so output it raw
2258 # but don't ignore what might be perfectly normal links in the text we've examined
2259 $holders->merge( $this->replaceInternalLinks2( $text ) );
2260 $s .= "{$prefix}[[$link|$text";
2261 # note: no $trail, because without an end, there *is* no trail
2262 continue;
2263 }
2264 } else { # it's not an image, so output it raw
2265 $s .= "{$prefix}[[$link|$text";
2266 # note: no $trail, because without an end, there *is* no trail
2267 continue;
2268 }
2269 }
2270
2271 $wasblank = ( $text == '' );
2272 if ( $wasblank ) {
2273 $text = $link;
2274 if ( !$noforce ) {
2275 # Strip off leading ':'
2276 $text = substr( $text, 1 );
2277 }
2278 } else {
2279 # T6598 madness. Handle the quotes only if they come from the alternate part
2280 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2281 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2282 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2283 $text = $this->doQuotes( $text );
2284 }
2285
2286 # Link not escaped by : , create the various objects
2287 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2288 # Interwikis
2289 if (
2290 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2291 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2292 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2293 )
2294 ) {
2295 # T26502: filter duplicates
2296 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2297 $this->mLangLinkLanguages[$iw] = true;
2298 $this->mOutput->addLanguageLink( $nt->getFullText() );
2299 }
2300
2301 /**
2302 * Strip the whitespace interwiki links produce, see T10897
2303 */
2304 $s = rtrim( $s . $prefix ) . $trail; # T175416
2305 continue;
2306 }
2307
2308 if ( $ns == NS_FILE ) {
2309 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2310 if ( $wasblank ) {
2311 # if no parameters were passed, $text
2312 # becomes something like "File:Foo.png",
2313 # which we don't want to pass on to the
2314 # image generator
2315 $text = '';
2316 } else {
2317 # recursively parse links inside the image caption
2318 # actually, this will parse them in any other parameters, too,
2319 # but it might be hard to fix that, and it doesn't matter ATM
2320 $text = $this->replaceExternalLinks( $text );
2321 $holders->merge( $this->replaceInternalLinks2( $text ) );
2322 }
2323 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2324 $s .= $prefix . $this->armorLinks(
2325 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2326 continue;
2327 }
2328 } elseif ( $ns == NS_CATEGORY ) {
2329 /**
2330 * Strip the whitespace Category links produce, see T2087
2331 */
2332 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2333
2334 if ( $wasblank ) {
2335 $sortkey = $this->getDefaultSort();
2336 } else {
2337 $sortkey = $text;
2338 }
2339 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2340 $sortkey = str_replace( "\n", '', $sortkey );
2341 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2342 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2343
2344 continue;
2345 }
2346 }
2347
2348 # Self-link checking. For some languages, variants of the title are checked in
2349 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2350 # for linking to a different variant.
2351 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2352 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2353 continue;
2354 }
2355
2356 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2357 # @todo FIXME: Should do batch file existence checks, see comment below
2358 if ( $ns == NS_MEDIA ) {
2359 # Give extensions a chance to select the file revision for us
2360 $options = [];
2361 $descQuery = false;
2362 Hooks::run( 'BeforeParserFetchFileAndTitle',
2363 [ $this, $nt, &$options, &$descQuery ] );
2364 # Fetch and register the file (file title may be different via hooks)
2365 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2366 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2367 $s .= $prefix . $this->armorLinks(
2368 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2369 continue;
2370 }
2371
2372 # Some titles, such as valid special pages or files in foreign repos, should
2373 # be shown as bluelinks even though they're not included in the page table
2374 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2375 # batch file existence checks for NS_FILE and NS_MEDIA
2376 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2377 $this->mOutput->addLink( $nt );
2378 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2379 } else {
2380 # Links will be added to the output link list after checking
2381 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2382 }
2383 }
2384 return $holders;
2385 }
2386
2387 /**
2388 * Render a forced-blue link inline; protect against double expansion of
2389 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2390 * Since this little disaster has to split off the trail text to avoid
2391 * breaking URLs in the following text without breaking trails on the
2392 * wiki links, it's been made into a horrible function.
2393 *
2394 * @param Title $nt
2395 * @param string $text
2396 * @param string $trail
2397 * @param string $prefix
2398 * @return string HTML-wikitext mix oh yuck
2399 */
2400 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2401 list( $inside, $trail ) = Linker::splitTrail( $trail );
2402
2403 if ( $text == '' ) {
2404 $text = htmlspecialchars( $nt->getPrefixedText() );
2405 }
2406
2407 $link = $this->getLinkRenderer()->makeKnownLink(
2408 $nt, new HtmlArmor( "$prefix$text$inside" )
2409 );
2410
2411 return $this->armorLinks( $link ) . $trail;
2412 }
2413
2414 /**
2415 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2416 * going to go through further parsing steps before inline URL expansion.
2417 *
2418 * Not needed quite as much as it used to be since free links are a bit
2419 * more sensible these days. But bracketed links are still an issue.
2420 *
2421 * @param string $text More-or-less HTML
2422 * @return string Less-or-more HTML with NOPARSE bits
2423 */
2424 public function armorLinks( $text ) {
2425 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2426 self::MARKER_PREFIX . "NOPARSE$1", $text );
2427 }
2428
2429 /**
2430 * Return true if subpage links should be expanded on this page.
2431 * @return bool
2432 */
2433 public function areSubpagesAllowed() {
2434 # Some namespaces don't allow subpages
2435 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2436 }
2437
2438 /**
2439 * Handle link to subpage if necessary
2440 *
2441 * @param string $target The source of the link
2442 * @param string &$text The link text, modified as necessary
2443 * @return string The full name of the link
2444 * @private
2445 */
2446 public function maybeDoSubpageLink( $target, &$text ) {
2447 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2448 }
2449
2450 /**
2451 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2452 *
2453 * @param string $text
2454 * @param bool $linestart Whether or not this is at the start of a line.
2455 * @private
2456 * @return string The lists rendered as HTML
2457 */
2458 public function doBlockLevels( $text, $linestart ) {
2459 return BlockLevelPass::doBlockLevels( $text, $linestart );
2460 }
2461
2462 /**
2463 * Return value of a magic variable (like PAGENAME)
2464 *
2465 * @private
2466 *
2467 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2468 * @param bool|PPFrame $frame
2469 *
2470 * @throws MWException
2471 * @return string
2472 */
2473 public function getVariableValue( $index, $frame = false ) {
2474 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2475 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2476
2477 if ( is_null( $this->mTitle ) ) {
2478 // If no title set, bad things are going to happen
2479 // later. Title should always be set since this
2480 // should only be called in the middle of a parse
2481 // operation (but the unit-tests do funky stuff)
2482 throw new MWException( __METHOD__ . ' Should only be '
2483 . ' called while parsing (no title set)' );
2484 }
2485
2486 // Avoid PHP 7.1 warning from passing $this by reference
2487 $parser = $this;
2488
2489 /**
2490 * Some of these require message or data lookups and can be
2491 * expensive to check many times.
2492 */
2493 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2494 if ( isset( $this->mVarCache[$index] ) ) {
2495 return $this->mVarCache[$index];
2496 }
2497 }
2498
2499 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2500 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2501
2502 $pageLang = $this->getFunctionLang();
2503
2504 switch ( $index ) {
2505 case '!':
2506 $value = '|';
2507 break;
2508 case 'currentmonth':
2509 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2510 break;
2511 case 'currentmonth1':
2512 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2513 break;
2514 case 'currentmonthname':
2515 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2516 break;
2517 case 'currentmonthnamegen':
2518 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2519 break;
2520 case 'currentmonthabbrev':
2521 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2522 break;
2523 case 'currentday':
2524 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2525 break;
2526 case 'currentday2':
2527 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2528 break;
2529 case 'localmonth':
2530 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2531 break;
2532 case 'localmonth1':
2533 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2534 break;
2535 case 'localmonthname':
2536 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2537 break;
2538 case 'localmonthnamegen':
2539 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2540 break;
2541 case 'localmonthabbrev':
2542 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2543 break;
2544 case 'localday':
2545 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2546 break;
2547 case 'localday2':
2548 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2549 break;
2550 case 'pagename':
2551 $value = wfEscapeWikiText( $this->mTitle->getText() );
2552 break;
2553 case 'pagenamee':
2554 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2555 break;
2556 case 'fullpagename':
2557 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2558 break;
2559 case 'fullpagenamee':
2560 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2561 break;
2562 case 'subpagename':
2563 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2564 break;
2565 case 'subpagenamee':
2566 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2567 break;
2568 case 'rootpagename':
2569 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2570 break;
2571 case 'rootpagenamee':
2572 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2573 ' ',
2574 '_',
2575 $this->mTitle->getRootText()
2576 ) ) );
2577 break;
2578 case 'basepagename':
2579 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2580 break;
2581 case 'basepagenamee':
2582 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2583 ' ',
2584 '_',
2585 $this->mTitle->getBaseText()
2586 ) ) );
2587 break;
2588 case 'talkpagename':
2589 if ( $this->mTitle->canHaveTalkPage() ) {
2590 $talkPage = $this->mTitle->getTalkPage();
2591 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2592 } else {
2593 $value = '';
2594 }
2595 break;
2596 case 'talkpagenamee':
2597 if ( $this->mTitle->canHaveTalkPage() ) {
2598 $talkPage = $this->mTitle->getTalkPage();
2599 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2600 } else {
2601 $value = '';
2602 }
2603 break;
2604 case 'subjectpagename':
2605 $subjPage = $this->mTitle->getSubjectPage();
2606 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2607 break;
2608 case 'subjectpagenamee':
2609 $subjPage = $this->mTitle->getSubjectPage();
2610 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2611 break;
2612 case 'pageid': // requested in T25427
2613 $pageid = $this->getTitle()->getArticleID();
2614 if ( $pageid == 0 ) {
2615 # 0 means the page doesn't exist in the database,
2616 # which means the user is previewing a new page.
2617 # The vary-revision flag must be set, because the magic word
2618 # will have a different value once the page is saved.
2619 $this->mOutput->setFlag( 'vary-revision' );
2620 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2621 }
2622 $value = $pageid ? $pageid : null;
2623 break;
2624 case 'revisionid':
2625 # Let the edit saving system know we should parse the page
2626 # *after* a revision ID has been assigned.
2627 $this->mOutput->setFlag( 'vary-revision-id' );
2628 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2629 $value = $this->mRevisionId;
2630 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2631 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2632 $this->mOutput->setSpeculativeRevIdUsed( $value );
2633 }
2634 break;
2635 case 'revisionday':
2636 # Let the edit saving system know we should parse the page
2637 # *after* a revision ID has been assigned. This is for null edits.
2638 $this->mOutput->setFlag( 'vary-revision' );
2639 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2640 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2641 break;
2642 case 'revisionday2':
2643 # Let the edit saving system know we should parse the page
2644 # *after* a revision ID has been assigned. This is for null edits.
2645 $this->mOutput->setFlag( 'vary-revision' );
2646 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2647 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2648 break;
2649 case 'revisionmonth':
2650 # Let the edit saving system know we should parse the page
2651 # *after* a revision ID has been assigned. This is for null edits.
2652 $this->mOutput->setFlag( 'vary-revision' );
2653 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2654 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2655 break;
2656 case 'revisionmonth1':
2657 # Let the edit saving system know we should parse the page
2658 # *after* a revision ID has been assigned. This is for null edits.
2659 $this->mOutput->setFlag( 'vary-revision' );
2660 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2661 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2662 break;
2663 case 'revisionyear':
2664 # Let the edit saving system know we should parse the page
2665 # *after* a revision ID has been assigned. This is for null edits.
2666 $this->mOutput->setFlag( 'vary-revision' );
2667 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2668 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2669 break;
2670 case 'revisiontimestamp':
2671 # Let the edit saving system know we should parse the page
2672 # *after* a revision ID has been assigned. This is for null edits.
2673 $this->mOutput->setFlag( 'vary-revision' );
2674 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2675 $value = $this->getRevisionTimestamp();
2676 break;
2677 case 'revisionuser':
2678 # Let the edit saving system know we should parse the page
2679 # *after* a revision ID has been assigned for null edits.
2680 $this->mOutput->setFlag( 'vary-user' );
2681 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2682 $value = $this->getRevisionUser();
2683 break;
2684 case 'revisionsize':
2685 $value = $this->getRevisionSize();
2686 break;
2687 case 'namespace':
2688 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2689 break;
2690 case 'namespacee':
2691 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2692 break;
2693 case 'namespacenumber':
2694 $value = $this->mTitle->getNamespace();
2695 break;
2696 case 'talkspace':
2697 $value = $this->mTitle->canHaveTalkPage()
2698 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2699 : '';
2700 break;
2701 case 'talkspacee':
2702 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2703 break;
2704 case 'subjectspace':
2705 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2706 break;
2707 case 'subjectspacee':
2708 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2709 break;
2710 case 'currentdayname':
2711 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2712 break;
2713 case 'currentyear':
2714 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2715 break;
2716 case 'currenttime':
2717 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2718 break;
2719 case 'currenthour':
2720 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2721 break;
2722 case 'currentweek':
2723 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2724 # int to remove the padding
2725 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2726 break;
2727 case 'currentdow':
2728 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2729 break;
2730 case 'localdayname':
2731 $value = $pageLang->getWeekdayName(
2732 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2733 );
2734 break;
2735 case 'localyear':
2736 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2737 break;
2738 case 'localtime':
2739 $value = $pageLang->time(
2740 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2741 false,
2742 false
2743 );
2744 break;
2745 case 'localhour':
2746 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2747 break;
2748 case 'localweek':
2749 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2750 # int to remove the padding
2751 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2752 break;
2753 case 'localdow':
2754 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2755 break;
2756 case 'numberofarticles':
2757 $value = $pageLang->formatNum( SiteStats::articles() );
2758 break;
2759 case 'numberoffiles':
2760 $value = $pageLang->formatNum( SiteStats::images() );
2761 break;
2762 case 'numberofusers':
2763 $value = $pageLang->formatNum( SiteStats::users() );
2764 break;
2765 case 'numberofactiveusers':
2766 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2767 break;
2768 case 'numberofpages':
2769 $value = $pageLang->formatNum( SiteStats::pages() );
2770 break;
2771 case 'numberofadmins':
2772 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2773 break;
2774 case 'numberofedits':
2775 $value = $pageLang->formatNum( SiteStats::edits() );
2776 break;
2777 case 'currenttimestamp':
2778 $value = wfTimestamp( TS_MW, $ts );
2779 break;
2780 case 'localtimestamp':
2781 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2782 break;
2783 case 'currentversion':
2784 $value = SpecialVersion::getVersion();
2785 break;
2786 case 'articlepath':
2787 return $wgArticlePath;
2788 case 'sitename':
2789 return $wgSitename;
2790 case 'server':
2791 return $wgServer;
2792 case 'servername':
2793 return $wgServerName;
2794 case 'scriptpath':
2795 return $wgScriptPath;
2796 case 'stylepath':
2797 return $wgStylePath;
2798 case 'directionmark':
2799 return $pageLang->getDirMark();
2800 case 'contentlanguage':
2801 global $wgLanguageCode;
2802 return $wgLanguageCode;
2803 case 'pagelanguage':
2804 $value = $pageLang->getCode();
2805 break;
2806 case 'cascadingsources':
2807 $value = CoreParserFunctions::cascadingsources( $this );
2808 break;
2809 default:
2810 $ret = null;
2811 Hooks::run(
2812 'ParserGetVariableValueSwitch',
2813 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2814 );
2815
2816 return $ret;
2817 }
2818
2819 if ( $index ) {
2820 $this->mVarCache[$index] = $value;
2821 }
2822
2823 return $value;
2824 }
2825
2826 /**
2827 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2828 *
2829 * @private
2830 */
2831 public function initialiseVariables() {
2832 $variableIDs = MagicWord::getVariableIDs();
2833 $substIDs = MagicWord::getSubstIDs();
2834
2835 $this->mVariables = new MagicWordArray( $variableIDs );
2836 $this->mSubstWords = new MagicWordArray( $substIDs );
2837 }
2838
2839 /**
2840 * Preprocess some wikitext and return the document tree.
2841 * This is the ghost of replace_variables().
2842 *
2843 * @param string $text The text to parse
2844 * @param int $flags Bitwise combination of:
2845 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2846 * included. Default is to assume a direct page view.
2847 *
2848 * The generated DOM tree must depend only on the input text and the flags.
2849 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2850 *
2851 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2852 * change in the DOM tree for a given text, must be passed through the section identifier
2853 * in the section edit link and thus back to extractSections().
2854 *
2855 * The output of this function is currently only cached in process memory, but a persistent
2856 * cache may be implemented at a later date which takes further advantage of these strict
2857 * dependency requirements.
2858 *
2859 * @return PPNode
2860 */
2861 public function preprocessToDom( $text, $flags = 0 ) {
2862 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2863 return $dom;
2864 }
2865
2866 /**
2867 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2868 *
2869 * @param string $s
2870 *
2871 * @return array
2872 */
2873 public static function splitWhitespace( $s ) {
2874 $ltrimmed = ltrim( $s );
2875 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2876 $trimmed = rtrim( $ltrimmed );
2877 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2878 if ( $diff > 0 ) {
2879 $w2 = substr( $ltrimmed, -$diff );
2880 } else {
2881 $w2 = '';
2882 }
2883 return [ $w1, $trimmed, $w2 ];
2884 }
2885
2886 /**
2887 * Replace magic variables, templates, and template arguments
2888 * with the appropriate text. Templates are substituted recursively,
2889 * taking care to avoid infinite loops.
2890 *
2891 * Note that the substitution depends on value of $mOutputType:
2892 * self::OT_WIKI: only {{subst:}} templates
2893 * self::OT_PREPROCESS: templates but not extension tags
2894 * self::OT_HTML: all templates and extension tags
2895 *
2896 * @param string $text The text to transform
2897 * @param bool|PPFrame $frame Object describing the arguments passed to the
2898 * template. Arguments may also be provided as an associative array, as
2899 * was the usual case before MW1.12. Providing arguments this way may be
2900 * useful for extensions wishing to perform variable replacement
2901 * explicitly.
2902 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2903 * double-brace expansion.
2904 * @return string
2905 */
2906 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2907 # Is there any text? Also, Prevent too big inclusions!
2908 $textSize = strlen( $text );
2909 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2910 return $text;
2911 }
2912
2913 if ( $frame === false ) {
2914 $frame = $this->getPreprocessor()->newFrame();
2915 } elseif ( !( $frame instanceof PPFrame ) ) {
2916 wfDebug( __METHOD__ . " called using plain parameters instead of "
2917 . "a PPFrame instance. Creating custom frame.\n" );
2918 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2919 }
2920
2921 $dom = $this->preprocessToDom( $text );
2922 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2923 $text = $frame->expand( $dom, $flags );
2924
2925 return $text;
2926 }
2927
2928 /**
2929 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2930 *
2931 * @param array $args
2932 *
2933 * @return array
2934 */
2935 public static function createAssocArgs( $args ) {
2936 $assocArgs = [];
2937 $index = 1;
2938 foreach ( $args as $arg ) {
2939 $eqpos = strpos( $arg, '=' );
2940 if ( $eqpos === false ) {
2941 $assocArgs[$index++] = $arg;
2942 } else {
2943 $name = trim( substr( $arg, 0, $eqpos ) );
2944 $value = trim( substr( $arg, $eqpos + 1 ) );
2945 if ( $value === false ) {
2946 $value = '';
2947 }
2948 if ( $name !== false ) {
2949 $assocArgs[$name] = $value;
2950 }
2951 }
2952 }
2953
2954 return $assocArgs;
2955 }
2956
2957 /**
2958 * Warn the user when a parser limitation is reached
2959 * Will warn at most once the user per limitation type
2960 *
2961 * The results are shown during preview and run through the Parser (See EditPage.php)
2962 *
2963 * @param string $limitationType Should be one of:
2964 * 'expensive-parserfunction' (corresponding messages:
2965 * 'expensive-parserfunction-warning',
2966 * 'expensive-parserfunction-category')
2967 * 'post-expand-template-argument' (corresponding messages:
2968 * 'post-expand-template-argument-warning',
2969 * 'post-expand-template-argument-category')
2970 * 'post-expand-template-inclusion' (corresponding messages:
2971 * 'post-expand-template-inclusion-warning',
2972 * 'post-expand-template-inclusion-category')
2973 * 'node-count-exceeded' (corresponding messages:
2974 * 'node-count-exceeded-warning',
2975 * 'node-count-exceeded-category')
2976 * 'expansion-depth-exceeded' (corresponding messages:
2977 * 'expansion-depth-exceeded-warning',
2978 * 'expansion-depth-exceeded-category')
2979 * @param string|int|null $current Current value
2980 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2981 * exceeded, provide the values (optional)
2982 */
2983 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2984 # does no harm if $current and $max are present but are unnecessary for the message
2985 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2986 # only during preview, and that would split the parser cache unnecessarily.
2987 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2988 ->text();
2989 $this->mOutput->addWarning( $warning );
2990 $this->addTrackingCategory( "$limitationType-category" );
2991 }
2992
2993 /**
2994 * Return the text of a template, after recursively
2995 * replacing any variables or templates within the template.
2996 *
2997 * @param array $piece The parts of the template
2998 * $piece['title']: the title, i.e. the part before the |
2999 * $piece['parts']: the parameter array
3000 * $piece['lineStart']: whether the brace was at the start of a line
3001 * @param PPFrame $frame The current frame, contains template arguments
3002 * @throws Exception
3003 * @return string The text of the template
3004 */
3005 public function braceSubstitution( $piece, $frame ) {
3006 // Flags
3007
3008 // $text has been filled
3009 $found = false;
3010 // wiki markup in $text should be escaped
3011 $nowiki = false;
3012 // $text is HTML, armour it against wikitext transformation
3013 $isHTML = false;
3014 // Force interwiki transclusion to be done in raw mode not rendered
3015 $forceRawInterwiki = false;
3016 // $text is a DOM node needing expansion in a child frame
3017 $isChildObj = false;
3018 // $text is a DOM node needing expansion in the current frame
3019 $isLocalObj = false;
3020
3021 # Title object, where $text came from
3022 $title = false;
3023
3024 # $part1 is the bit before the first |, and must contain only title characters.
3025 # Various prefixes will be stripped from it later.
3026 $titleWithSpaces = $frame->expand( $piece['title'] );
3027 $part1 = trim( $titleWithSpaces );
3028 $titleText = false;
3029
3030 # Original title text preserved for various purposes
3031 $originalTitle = $part1;
3032
3033 # $args is a list of argument nodes, starting from index 0, not including $part1
3034 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3035 # below won't work b/c this $args isn't an object
3036 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3037
3038 $profileSection = null; // profile templates
3039
3040 # SUBST
3041 if ( !$found ) {
3042 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3043
3044 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3045 # Decide whether to expand template or keep wikitext as-is.
3046 if ( $this->ot['wiki'] ) {
3047 if ( $substMatch === false ) {
3048 $literal = true; # literal when in PST with no prefix
3049 } else {
3050 $literal = false; # expand when in PST with subst: or safesubst:
3051 }
3052 } else {
3053 if ( $substMatch == 'subst' ) {
3054 $literal = true; # literal when not in PST with plain subst:
3055 } else {
3056 $literal = false; # expand when not in PST with safesubst: or no prefix
3057 }
3058 }
3059 if ( $literal ) {
3060 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3061 $isLocalObj = true;
3062 $found = true;
3063 }
3064 }
3065
3066 # Variables
3067 if ( !$found && $args->getLength() == 0 ) {
3068 $id = $this->mVariables->matchStartToEnd( $part1 );
3069 if ( $id !== false ) {
3070 $text = $this->getVariableValue( $id, $frame );
3071 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3072 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3073 }
3074 $found = true;
3075 }
3076 }
3077
3078 # MSG, MSGNW and RAW
3079 if ( !$found ) {
3080 # Check for MSGNW:
3081 $mwMsgnw = MagicWord::get( 'msgnw' );
3082 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3083 $nowiki = true;
3084 } else {
3085 # Remove obsolete MSG:
3086 $mwMsg = MagicWord::get( 'msg' );
3087 $mwMsg->matchStartAndRemove( $part1 );
3088 }
3089
3090 # Check for RAW:
3091 $mwRaw = MagicWord::get( 'raw' );
3092 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3093 $forceRawInterwiki = true;
3094 }
3095 }
3096
3097 # Parser functions
3098 if ( !$found ) {
3099 $colonPos = strpos( $part1, ':' );
3100 if ( $colonPos !== false ) {
3101 $func = substr( $part1, 0, $colonPos );
3102 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3103 $argsLength = $args->getLength();
3104 for ( $i = 0; $i < $argsLength; $i++ ) {
3105 $funcArgs[] = $args->item( $i );
3106 }
3107 try {
3108 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3109 } catch ( Exception $ex ) {
3110 throw $ex;
3111 }
3112
3113 // Extract any forwarded flags
3114 if ( isset( $result['title'] ) ) {
3115 $title = $result['title'];
3116 }
3117 if ( isset( $result['found'] ) ) {
3118 $found = $result['found'];
3119 }
3120 if ( array_key_exists( 'text', $result ) ) {
3121 // a string or null
3122 $text = $result['text'];
3123 }
3124 if ( isset( $result['nowiki'] ) ) {
3125 $nowiki = $result['nowiki'];
3126 }
3127 if ( isset( $result['isHTML'] ) ) {
3128 $isHTML = $result['isHTML'];
3129 }
3130 if ( isset( $result['forceRawInterwiki'] ) ) {
3131 $forceRawInterwiki = $result['forceRawInterwiki'];
3132 }
3133 if ( isset( $result['isChildObj'] ) ) {
3134 $isChildObj = $result['isChildObj'];
3135 }
3136 if ( isset( $result['isLocalObj'] ) ) {
3137 $isLocalObj = $result['isLocalObj'];
3138 }
3139 }
3140 }
3141
3142 # Finish mangling title and then check for loops.
3143 # Set $title to a Title object and $titleText to the PDBK
3144 if ( !$found ) {
3145 $ns = NS_TEMPLATE;
3146 # Split the title into page and subpage
3147 $subpage = '';
3148 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3149 if ( $part1 !== $relative ) {
3150 $part1 = $relative;
3151 $ns = $this->mTitle->getNamespace();
3152 }
3153 $title = Title::newFromText( $part1, $ns );
3154 if ( $title ) {
3155 $titleText = $title->getPrefixedText();
3156 # Check for language variants if the template is not found
3157 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3158 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3159 }
3160 # Do recursion depth check
3161 $limit = $this->mOptions->getMaxTemplateDepth();
3162 if ( $frame->depth >= $limit ) {
3163 $found = true;
3164 $text = '<span class="error">'
3165 . wfMessage( 'parser-template-recursion-depth-warning' )
3166 ->numParams( $limit )->inContentLanguage()->text()
3167 . '</span>';
3168 }
3169 }
3170 }
3171
3172 # Load from database
3173 if ( !$found && $title ) {
3174 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3175 if ( !$title->isExternal() ) {
3176 if ( $title->isSpecialPage()
3177 && $this->mOptions->getAllowSpecialInclusion()
3178 && $this->ot['html']
3179 ) {
3180 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3181 // Pass the template arguments as URL parameters.
3182 // "uselang" will have no effect since the Language object
3183 // is forced to the one defined in ParserOptions.
3184 $pageArgs = [];
3185 $argsLength = $args->getLength();
3186 for ( $i = 0; $i < $argsLength; $i++ ) {
3187 $bits = $args->item( $i )->splitArg();
3188 if ( strval( $bits['index'] ) === '' ) {
3189 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3190 $value = trim( $frame->expand( $bits['value'] ) );
3191 $pageArgs[$name] = $value;
3192 }
3193 }
3194
3195 // Create a new context to execute the special page
3196 $context = new RequestContext;
3197 $context->setTitle( $title );
3198 $context->setRequest( new FauxRequest( $pageArgs ) );
3199 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3200 $context->setUser( $this->getUser() );
3201 } else {
3202 // If this page is cached, then we better not be per user.
3203 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3204 }
3205 $context->setLanguage( $this->mOptions->getUserLangObj() );
3206 $ret = SpecialPageFactory::capturePath(
3207 $title, $context, $this->getLinkRenderer() );
3208 if ( $ret ) {
3209 $text = $context->getOutput()->getHTML();
3210 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3211 $found = true;
3212 $isHTML = true;
3213 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3214 $this->mOutput->updateRuntimeAdaptiveExpiry(
3215 $specialPage->maxIncludeCacheTime()
3216 );
3217 }
3218 }
3219 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3220 $found = false; # access denied
3221 wfDebug( __METHOD__ . ": template inclusion denied for " .
3222 $title->getPrefixedDBkey() . "\n" );
3223 } else {
3224 list( $text, $title ) = $this->getTemplateDom( $title );
3225 if ( $text !== false ) {
3226 $found = true;
3227 $isChildObj = true;
3228 }
3229 }
3230
3231 # If the title is valid but undisplayable, make a link to it
3232 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3233 $text = "[[:$titleText]]";
3234 $found = true;
3235 }
3236 } elseif ( $title->isTrans() ) {
3237 # Interwiki transclusion
3238 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3239 $text = $this->interwikiTransclude( $title, 'render' );
3240 $isHTML = true;
3241 } else {
3242 $text = $this->interwikiTransclude( $title, 'raw' );
3243 # Preprocess it like a template
3244 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3245 $isChildObj = true;
3246 }
3247 $found = true;
3248 }
3249
3250 # Do infinite loop check
3251 # This has to be done after redirect resolution to avoid infinite loops via redirects
3252 if ( !$frame->loopCheck( $title ) ) {
3253 $found = true;
3254 $text = '<span class="error">'
3255 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3256 . '</span>';
3257 $this->addTrackingCategory( 'template-loop-category' );
3258 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3259 wfEscapeWikiText( $titleText ) )->text() );
3260 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3261 }
3262 }
3263
3264 # If we haven't found text to substitute by now, we're done
3265 # Recover the source wikitext and return it
3266 if ( !$found ) {
3267 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3268 if ( $profileSection ) {
3269 $this->mProfiler->scopedProfileOut( $profileSection );
3270 }
3271 return [ 'object' => $text ];
3272 }
3273
3274 # Expand DOM-style return values in a child frame
3275 if ( $isChildObj ) {
3276 # Clean up argument array
3277 $newFrame = $frame->newChild( $args, $title );
3278
3279 if ( $nowiki ) {
3280 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3281 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3282 # Expansion is eligible for the empty-frame cache
3283 $text = $newFrame->cachedExpand( $titleText, $text );
3284 } else {
3285 # Uncached expansion
3286 $text = $newFrame->expand( $text );
3287 }
3288 }
3289 if ( $isLocalObj && $nowiki ) {
3290 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3291 $isLocalObj = false;
3292 }
3293
3294 if ( $profileSection ) {
3295 $this->mProfiler->scopedProfileOut( $profileSection );
3296 }
3297
3298 # Replace raw HTML by a placeholder
3299 if ( $isHTML ) {
3300 $text = $this->insertStripItem( $text );
3301 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3302 # Escape nowiki-style return values
3303 $text = wfEscapeWikiText( $text );
3304 } elseif ( is_string( $text )
3305 && !$piece['lineStart']
3306 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3307 ) {
3308 # T2529: if the template begins with a table or block-level
3309 # element, it should be treated as beginning a new line.
3310 # This behavior is somewhat controversial.
3311 $text = "\n" . $text;
3312 }
3313
3314 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3315 # Error, oversize inclusion
3316 if ( $titleText !== false ) {
3317 # Make a working, properly escaped link if possible (T25588)
3318 $text = "[[:$titleText]]";
3319 } else {
3320 # This will probably not be a working link, but at least it may
3321 # provide some hint of where the problem is
3322 preg_replace( '/^:/', '', $originalTitle );
3323 $text = "[[:$originalTitle]]";
3324 }
3325 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3326 . 'post-expand include size too large -->' );
3327 $this->limitationWarn( 'post-expand-template-inclusion' );
3328 }
3329
3330 if ( $isLocalObj ) {
3331 $ret = [ 'object' => $text ];
3332 } else {
3333 $ret = [ 'text' => $text ];
3334 }
3335
3336 return $ret;
3337 }
3338
3339 /**
3340 * Call a parser function and return an array with text and flags.
3341 *
3342 * The returned array will always contain a boolean 'found', indicating
3343 * whether the parser function was found or not. It may also contain the
3344 * following:
3345 * text: string|object, resulting wikitext or PP DOM object
3346 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3347 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3348 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3349 * nowiki: bool, wiki markup in $text should be escaped
3350 *
3351 * @since 1.21
3352 * @param PPFrame $frame The current frame, contains template arguments
3353 * @param string $function Function name
3354 * @param array $args Arguments to the function
3355 * @throws MWException
3356 * @return array
3357 */
3358 public function callParserFunction( $frame, $function, array $args = [] ) {
3359 global $wgContLang;
3360
3361 # Case sensitive functions
3362 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3363 $function = $this->mFunctionSynonyms[1][$function];
3364 } else {
3365 # Case insensitive functions
3366 $function = $wgContLang->lc( $function );
3367 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3368 $function = $this->mFunctionSynonyms[0][$function];
3369 } else {
3370 return [ 'found' => false ];
3371 }
3372 }
3373
3374 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3375
3376 // Avoid PHP 7.1 warning from passing $this by reference
3377 $parser = $this;
3378
3379 $allArgs = [ &$parser ];
3380 if ( $flags & self::SFH_OBJECT_ARGS ) {
3381 # Convert arguments to PPNodes and collect for appending to $allArgs
3382 $funcArgs = [];
3383 foreach ( $args as $k => $v ) {
3384 if ( $v instanceof PPNode || $k === 0 ) {
3385 $funcArgs[] = $v;
3386 } else {
3387 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3388 }
3389 }
3390
3391 # Add a frame parameter, and pass the arguments as an array
3392 $allArgs[] = $frame;
3393 $allArgs[] = $funcArgs;
3394 } else {
3395 # Convert arguments to plain text and append to $allArgs
3396 foreach ( $args as $k => $v ) {
3397 if ( $v instanceof PPNode ) {
3398 $allArgs[] = trim( $frame->expand( $v ) );
3399 } elseif ( is_int( $k ) && $k >= 0 ) {
3400 $allArgs[] = trim( $v );
3401 } else {
3402 $allArgs[] = trim( "$k=$v" );
3403 }
3404 }
3405 }
3406
3407 $result = call_user_func_array( $callback, $allArgs );
3408
3409 # The interface for function hooks allows them to return a wikitext
3410 # string or an array containing the string and any flags. This mungs
3411 # things around to match what this method should return.
3412 if ( !is_array( $result ) ) {
3413 $result = [
3414 'found' => true,
3415 'text' => $result,
3416 ];
3417 } else {
3418 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3419 $result['text'] = $result[0];
3420 }
3421 unset( $result[0] );
3422 $result += [
3423 'found' => true,
3424 ];
3425 }
3426
3427 $noparse = true;
3428 $preprocessFlags = 0;
3429 if ( isset( $result['noparse'] ) ) {
3430 $noparse = $result['noparse'];
3431 }
3432 if ( isset( $result['preprocessFlags'] ) ) {
3433 $preprocessFlags = $result['preprocessFlags'];
3434 }
3435
3436 if ( !$noparse ) {
3437 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3438 $result['isChildObj'] = true;
3439 }
3440
3441 return $result;
3442 }
3443
3444 /**
3445 * Get the semi-parsed DOM representation of a template with a given title,
3446 * and its redirect destination title. Cached.
3447 *
3448 * @param Title $title
3449 *
3450 * @return array
3451 */
3452 public function getTemplateDom( $title ) {
3453 $cacheTitle = $title;
3454 $titleText = $title->getPrefixedDBkey();
3455
3456 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3457 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3458 $title = Title::makeTitle( $ns, $dbk );
3459 $titleText = $title->getPrefixedDBkey();
3460 }
3461 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3462 return [ $this->mTplDomCache[$titleText], $title ];
3463 }
3464
3465 # Cache miss, go to the database
3466 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3467
3468 if ( $text === false ) {
3469 $this->mTplDomCache[$titleText] = false;
3470 return [ false, $title ];
3471 }
3472
3473 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3474 $this->mTplDomCache[$titleText] = $dom;
3475
3476 if ( !$title->equals( $cacheTitle ) ) {
3477 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3478 [ $title->getNamespace(), $title->getDBkey() ];
3479 }
3480
3481 return [ $dom, $title ];
3482 }
3483
3484 /**
3485 * Fetch the current revision of a given title. Note that the revision
3486 * (and even the title) may not exist in the database, so everything
3487 * contributing to the output of the parser should use this method
3488 * where possible, rather than getting the revisions themselves. This
3489 * method also caches its results, so using it benefits performance.
3490 *
3491 * @since 1.24
3492 * @param Title $title
3493 * @return Revision
3494 */
3495 public function fetchCurrentRevisionOfTitle( $title ) {
3496 $cacheKey = $title->getPrefixedDBkey();
3497 if ( !$this->currentRevisionCache ) {
3498 $this->currentRevisionCache = new MapCacheLRU( 100 );
3499 }
3500 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3501 $this->currentRevisionCache->set( $cacheKey,
3502 // Defaults to Parser::statelessFetchRevision()
3503 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3504 );
3505 }
3506 return $this->currentRevisionCache->get( $cacheKey );
3507 }
3508
3509 /**
3510 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3511 * without passing them on to it.
3512 *
3513 * @since 1.24
3514 * @param Title $title
3515 * @param Parser|bool $parser
3516 * @return Revision|bool False if missing
3517 */
3518 public static function statelessFetchRevision( Title $title, $parser = false ) {
3519 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3520
3521 return $rev;
3522 }
3523
3524 /**
3525 * Fetch the unparsed text of a template and register a reference to it.
3526 * @param Title $title
3527 * @return array ( string or false, Title )
3528 */
3529 public function fetchTemplateAndTitle( $title ) {
3530 // Defaults to Parser::statelessFetchTemplate()
3531 $templateCb = $this->mOptions->getTemplateCallback();
3532 $stuff = call_user_func( $templateCb, $title, $this );
3533 // We use U+007F DELETE to distinguish strip markers from regular text.
3534 $text = $stuff['text'];
3535 if ( is_string( $stuff['text'] ) ) {
3536 $text = strtr( $text, "\x7f", "?" );
3537 }
3538 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3539 if ( isset( $stuff['deps'] ) ) {
3540 foreach ( $stuff['deps'] as $dep ) {
3541 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3542 if ( $dep['title']->equals( $this->getTitle() ) ) {
3543 // If we transclude ourselves, the final result
3544 // will change based on the new version of the page
3545 $this->mOutput->setFlag( 'vary-revision' );
3546 }
3547 }
3548 }
3549 return [ $text, $finalTitle ];
3550 }
3551
3552 /**
3553 * Fetch the unparsed text of a template and register a reference to it.
3554 * @param Title $title
3555 * @return string|bool
3556 */
3557 public function fetchTemplate( $title ) {
3558 return $this->fetchTemplateAndTitle( $title )[0];
3559 }
3560
3561 /**
3562 * Static function to get a template
3563 * Can be overridden via ParserOptions::setTemplateCallback().
3564 *
3565 * @param Title $title
3566 * @param bool|Parser $parser
3567 *
3568 * @return array
3569 */
3570 public static function statelessFetchTemplate( $title, $parser = false ) {
3571 $text = $skip = false;
3572 $finalTitle = $title;
3573 $deps = [];
3574
3575 # Loop to fetch the article, with up to 1 redirect
3576 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall
3577 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3578 # Give extensions a chance to select the revision instead
3579 $id = false; # Assume current
3580 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3581 [ $parser, $title, &$skip, &$id ] );
3582
3583 if ( $skip ) {
3584 $text = false;
3585 $deps[] = [
3586 'title' => $title,
3587 'page_id' => $title->getArticleID(),
3588 'rev_id' => null
3589 ];
3590 break;
3591 }
3592 # Get the revision
3593 if ( $id ) {
3594 $rev = Revision::newFromId( $id );
3595 } elseif ( $parser ) {
3596 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3597 } else {
3598 $rev = Revision::newFromTitle( $title );
3599 }
3600 $rev_id = $rev ? $rev->getId() : 0;
3601 # If there is no current revision, there is no page
3602 if ( $id === false && !$rev ) {
3603 $linkCache = LinkCache::singleton();
3604 $linkCache->addBadLinkObj( $title );
3605 }
3606
3607 $deps[] = [
3608 'title' => $title,
3609 'page_id' => $title->getArticleID(),
3610 'rev_id' => $rev_id ];
3611 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3612 # We fetched a rev from a different title; register it too...
3613 $deps[] = [
3614 'title' => $rev->getTitle(),
3615 'page_id' => $rev->getPage(),
3616 'rev_id' => $rev_id ];
3617 }
3618
3619 if ( $rev ) {
3620 $content = $rev->getContent();
3621 $text = $content ? $content->getWikitextForTransclusion() : null;
3622
3623 Hooks::run( 'ParserFetchTemplate',
3624 [ $parser, $title, $rev, &$text, &$deps ] );
3625
3626 if ( $text === false || $text === null ) {
3627 $text = false;
3628 break;
3629 }
3630 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3631 global $wgContLang;
3632 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3633 if ( !$message->exists() ) {
3634 $text = false;
3635 break;
3636 }
3637 $content = $message->content();
3638 $text = $message->plain();
3639 } else {
3640 break;
3641 }
3642 if ( !$content ) {
3643 break;
3644 }
3645 # Redirect?
3646 $finalTitle = $title;
3647 $title = $content->getRedirectTarget();
3648 }
3649 return [
3650 'text' => $text,
3651 'finalTitle' => $finalTitle,
3652 'deps' => $deps ];
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 File|bool
3661 */
3662 public function fetchFile( $title, $options = [] ) {
3663 return $this->fetchFileAndTitle( $title, $options )[0];
3664 }
3665
3666 /**
3667 * Fetch a file and its title and register a reference to it.
3668 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3669 * @param Title $title
3670 * @param array $options Array of options to RepoGroup::findFile
3671 * @return array ( File or false, Title of file )
3672 */
3673 public function fetchFileAndTitle( $title, $options = [] ) {
3674 $file = $this->fetchFileNoRegister( $title, $options );
3675
3676 $time = $file ? $file->getTimestamp() : false;
3677 $sha1 = $file ? $file->getSha1() : false;
3678 # Register the file as a dependency...
3679 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3680 if ( $file && !$title->equals( $file->getTitle() ) ) {
3681 # Update fetched file title
3682 $title = $file->getTitle();
3683 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3684 }
3685 return [ $file, $title ];
3686 }
3687
3688 /**
3689 * Helper function for fetchFileAndTitle.
3690 *
3691 * Also useful if you need to fetch a file but not use it yet,
3692 * for example to get the file's handler.
3693 *
3694 * @param Title $title
3695 * @param array $options Array of options to RepoGroup::findFile
3696 * @return File|bool
3697 */
3698 protected function fetchFileNoRegister( $title, $options = [] ) {
3699 if ( isset( $options['broken'] ) ) {
3700 $file = false; // broken thumbnail forced by hook
3701 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3702 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3703 } else { // get by (name,timestamp)
3704 $file = wfFindFile( $title, $options );
3705 }
3706 return $file;
3707 }
3708
3709 /**
3710 * Transclude an interwiki link.
3711 *
3712 * @param Title $title
3713 * @param string $action
3714 *
3715 * @return string
3716 */
3717 public function interwikiTransclude( $title, $action ) {
3718 global $wgEnableScaryTranscluding;
3719
3720 if ( !$wgEnableScaryTranscluding ) {
3721 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3722 }
3723
3724 $url = $title->getFullURL( [ 'action' => $action ] );
3725
3726 if ( strlen( $url ) > 255 ) {
3727 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3728 }
3729 return $this->fetchScaryTemplateMaybeFromCache( $url );
3730 }
3731
3732 /**
3733 * @param string $url
3734 * @return mixed|string
3735 */
3736 public function fetchScaryTemplateMaybeFromCache( $url ) {
3737 global $wgTranscludeCacheExpiry;
3738 $dbr = wfGetDB( DB_REPLICA );
3739 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3740 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3741 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3742 if ( $obj ) {
3743 return $obj->tc_contents;
3744 }
3745
3746 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3747 $status = $req->execute(); // Status object
3748 if ( $status->isOK() ) {
3749 $text = $req->getContent();
3750 } elseif ( $req->getStatus() != 200 ) {
3751 // Though we failed to fetch the content, this status is useless.
3752 return wfMessage( 'scarytranscludefailed-httpstatus' )
3753 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3754 } else {
3755 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3756 }
3757
3758 $dbw = wfGetDB( DB_MASTER );
3759 $dbw->replace( 'transcache', [ 'tc_url' ], [
3760 'tc_url' => $url,
3761 'tc_time' => $dbw->timestamp( time() ),
3762 'tc_contents' => $text
3763 ] );
3764 return $text;
3765 }
3766
3767 /**
3768 * Triple brace replacement -- used for template arguments
3769 * @private
3770 *
3771 * @param array $piece
3772 * @param PPFrame $frame
3773 *
3774 * @return array
3775 */
3776 public function argSubstitution( $piece, $frame ) {
3777 $error = false;
3778 $parts = $piece['parts'];
3779 $nameWithSpaces = $frame->expand( $piece['title'] );
3780 $argName = trim( $nameWithSpaces );
3781 $object = false;
3782 $text = $frame->getArgument( $argName );
3783 if ( $text === false && $parts->getLength() > 0
3784 && ( $this->ot['html']
3785 || $this->ot['pre']
3786 || ( $this->ot['wiki'] && $frame->isTemplate() )
3787 )
3788 ) {
3789 # No match in frame, use the supplied default
3790 $object = $parts->item( 0 )->getChildren();
3791 }
3792 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3793 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3794 $this->limitationWarn( 'post-expand-template-argument' );
3795 }
3796
3797 if ( $text === false && $object === false ) {
3798 # No match anywhere
3799 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3800 }
3801 if ( $error !== false ) {
3802 $text .= $error;
3803 }
3804 if ( $object !== false ) {
3805 $ret = [ 'object' => $object ];
3806 } else {
3807 $ret = [ 'text' => $text ];
3808 }
3809
3810 return $ret;
3811 }
3812
3813 /**
3814 * Return the text to be used for a given extension tag.
3815 * This is the ghost of strip().
3816 *
3817 * @param array $params Associative array of parameters:
3818 * name PPNode for the tag name
3819 * attr PPNode for unparsed text where tag attributes are thought to be
3820 * attributes Optional associative array of parsed attributes
3821 * inner Contents of extension element
3822 * noClose Original text did not have a close tag
3823 * @param PPFrame $frame
3824 *
3825 * @throws MWException
3826 * @return string
3827 */
3828 public function extensionSubstitution( $params, $frame ) {
3829 static $errorStr = '<span class="error">';
3830 static $errorLen = 20;
3831
3832 $name = $frame->expand( $params['name'] );
3833 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3834 // Probably expansion depth or node count exceeded. Just punt the
3835 // error up.
3836 return $name;
3837 }
3838
3839 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3840 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3841 // See above
3842 return $attrText;
3843 }
3844
3845 // We can't safely check if the expansion for $content resulted in an
3846 // error, because the content could happen to be the error string
3847 // (T149622).
3848 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3849
3850 $marker = self::MARKER_PREFIX . "-$name-"
3851 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3852
3853 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3854 ( $this->ot['html'] || $this->ot['pre'] );
3855 if ( $isFunctionTag ) {
3856 $markerType = 'none';
3857 } else {
3858 $markerType = 'general';
3859 }
3860 if ( $this->ot['html'] || $isFunctionTag ) {
3861 $name = strtolower( $name );
3862 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3863 if ( isset( $params['attributes'] ) ) {
3864 $attributes = $attributes + $params['attributes'];
3865 }
3866
3867 if ( isset( $this->mTagHooks[$name] ) ) {
3868 $output = call_user_func_array( $this->mTagHooks[$name],
3869 [ $content, $attributes, $this, $frame ] );
3870 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3871 list( $callback, ) = $this->mFunctionTagHooks[$name];
3872
3873 // Avoid PHP 7.1 warning from passing $this by reference
3874 $parser = $this;
3875 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3876 } else {
3877 $output = '<span class="error">Invalid tag extension name: ' .
3878 htmlspecialchars( $name ) . '</span>';
3879 }
3880
3881 if ( is_array( $output ) ) {
3882 // Extract flags
3883 $flags = $output;
3884 $output = $flags[0];
3885 if ( isset( $flags['markerType'] ) ) {
3886 $markerType = $flags['markerType'];
3887 }
3888 }
3889 } else {
3890 if ( is_null( $attrText ) ) {
3891 $attrText = '';
3892 }
3893 if ( isset( $params['attributes'] ) ) {
3894 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3895 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3896 htmlspecialchars( $attrValue ) . '"';
3897 }
3898 }
3899 if ( $content === null ) {
3900 $output = "<$name$attrText/>";
3901 } else {
3902 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3903 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3904 // See above
3905 return $close;
3906 }
3907 $output = "<$name$attrText>$content$close";
3908 }
3909 }
3910
3911 if ( $markerType === 'none' ) {
3912 return $output;
3913 } elseif ( $markerType === 'nowiki' ) {
3914 $this->mStripState->addNoWiki( $marker, $output );
3915 } elseif ( $markerType === 'general' ) {
3916 $this->mStripState->addGeneral( $marker, $output );
3917 } else {
3918 throw new MWException( __METHOD__ . ': invalid marker type' );
3919 }
3920 return $marker;
3921 }
3922
3923 /**
3924 * Increment an include size counter
3925 *
3926 * @param string $type The type of expansion
3927 * @param int $size The size of the text
3928 * @return bool False if this inclusion would take it over the maximum, true otherwise
3929 */
3930 public function incrementIncludeSize( $type, $size ) {
3931 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3932 return false;
3933 } else {
3934 $this->mIncludeSizes[$type] += $size;
3935 return true;
3936 }
3937 }
3938
3939 /**
3940 * Increment the expensive function count
3941 *
3942 * @return bool False if the limit has been exceeded
3943 */
3944 public function incrementExpensiveFunctionCount() {
3945 $this->mExpensiveFunctionCount++;
3946 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3947 }
3948
3949 /**
3950 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3951 * Fills $this->mDoubleUnderscores, returns the modified text
3952 *
3953 * @param string $text
3954 *
3955 * @return string
3956 */
3957 public function doDoubleUnderscore( $text ) {
3958 # The position of __TOC__ needs to be recorded
3959 $mw = MagicWord::get( 'toc' );
3960 if ( $mw->match( $text ) ) {
3961 $this->mShowToc = true;
3962 $this->mForceTocPosition = true;
3963
3964 # Set a placeholder. At the end we'll fill it in with the TOC.
3965 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
3966
3967 # Only keep the first one.
3968 $text = $mw->replace( '', $text );
3969 }
3970
3971 # Now match and remove the rest of them
3972 $mwa = MagicWord::getDoubleUnderscoreArray();
3973 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3974
3975 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3976 $this->mOutput->mNoGallery = true;
3977 }
3978 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3979 $this->mShowToc = false;
3980 }
3981 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3982 && $this->mTitle->getNamespace() == NS_CATEGORY
3983 ) {
3984 $this->addTrackingCategory( 'hidden-category-category' );
3985 }
3986 # (T10068) Allow control over whether robots index a page.
3987 # __INDEX__ always overrides __NOINDEX__, see T16899
3988 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3989 $this->mOutput->setIndexPolicy( 'noindex' );
3990 $this->addTrackingCategory( 'noindex-category' );
3991 }
3992 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3993 $this->mOutput->setIndexPolicy( 'index' );
3994 $this->addTrackingCategory( 'index-category' );
3995 }
3996
3997 # Cache all double underscores in the database
3998 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3999 $this->mOutput->setProperty( $key, '' );
4000 }
4001
4002 return $text;
4003 }
4004
4005 /**
4006 * @see ParserOutput::addTrackingCategory()
4007 * @param string $msg Message key
4008 * @return bool Whether the addition was successful
4009 */
4010 public function addTrackingCategory( $msg ) {
4011 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4012 }
4013
4014 /**
4015 * This function accomplishes several tasks:
4016 * 1) Auto-number headings if that option is enabled
4017 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4018 * 3) Add a Table of contents on the top for users who have enabled the option
4019 * 4) Auto-anchor headings
4020 *
4021 * It loops through all headlines, collects the necessary data, then splits up the
4022 * string and re-inserts the newly formatted headlines.
4023 *
4024 * @param string $text
4025 * @param string $origText Original, untouched wikitext
4026 * @param bool $isMain
4027 * @return mixed|string
4028 * @private
4029 */
4030 public function formatHeadings( $text, $origText, $isMain = true ) {
4031 global $wgMaxTocLevel;
4032
4033 # Inhibit editsection links if requested in the page
4034 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4035 $maybeShowEditLink = false;
4036 } else {
4037 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4038 }
4039
4040 # Get all headlines for numbering them and adding funky stuff like [edit]
4041 # links - this is for later, but we need the number of headlines right now
4042 $matches = [];
4043 $numMatches = preg_match_all(
4044 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4045 $text,
4046 $matches
4047 );
4048
4049 # if there are fewer than 4 headlines in the article, do not show TOC
4050 # unless it's been explicitly enabled.
4051 $enoughToc = $this->mShowToc &&
4052 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4053
4054 # Allow user to stipulate that a page should have a "new section"
4055 # link added via __NEWSECTIONLINK__
4056 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4057 $this->mOutput->setNewSection( true );
4058 }
4059
4060 # Allow user to remove the "new section"
4061 # link via __NONEWSECTIONLINK__
4062 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4063 $this->mOutput->hideNewSection( true );
4064 }
4065
4066 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4067 # override above conditions and always show TOC above first header
4068 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4069 $this->mShowToc = true;
4070 $enoughToc = true;
4071 }
4072
4073 # headline counter
4074 $headlineCount = 0;
4075 $numVisible = 0;
4076
4077 # Ugh .. the TOC should have neat indentation levels which can be
4078 # passed to the skin functions. These are determined here
4079 $toc = '';
4080 $full = '';
4081 $head = [];
4082 $sublevelCount = [];
4083 $levelCount = [];
4084 $level = 0;
4085 $prevlevel = 0;
4086 $toclevel = 0;
4087 $prevtoclevel = 0;
4088 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4089 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4090 $oldType = $this->mOutputType;
4091 $this->setOutputType( self::OT_WIKI );
4092 $frame = $this->getPreprocessor()->newFrame();
4093 $root = $this->preprocessToDom( $origText );
4094 $node = $root->getFirstChild();
4095 $byteOffset = 0;
4096 $tocraw = [];
4097 $refers = [];
4098
4099 $headlines = $numMatches !== false ? $matches[3] : [];
4100
4101 foreach ( $headlines as $headline ) {
4102 $isTemplate = false;
4103 $titleText = false;
4104 $sectionIndex = false;
4105 $numbering = '';
4106 $markerMatches = [];
4107 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4108 $serial = $markerMatches[1];
4109 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4110 $isTemplate = ( $titleText != $baseTitleText );
4111 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4112 }
4113
4114 if ( $toclevel ) {
4115 $prevlevel = $level;
4116 }
4117 $level = $matches[1][$headlineCount];
4118
4119 if ( $level > $prevlevel ) {
4120 # Increase TOC level
4121 $toclevel++;
4122 $sublevelCount[$toclevel] = 0;
4123 if ( $toclevel < $wgMaxTocLevel ) {
4124 $prevtoclevel = $toclevel;
4125 $toc .= Linker::tocIndent();
4126 $numVisible++;
4127 }
4128 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4129 # Decrease TOC level, find level to jump to
4130
4131 for ( $i = $toclevel; $i > 0; $i-- ) {
4132 if ( $levelCount[$i] == $level ) {
4133 # Found last matching level
4134 $toclevel = $i;
4135 break;
4136 } elseif ( $levelCount[$i] < $level ) {
4137 # Found first matching level below current level
4138 $toclevel = $i + 1;
4139 break;
4140 }
4141 }
4142 if ( $i == 0 ) {
4143 $toclevel = 1;
4144 }
4145 if ( $toclevel < $wgMaxTocLevel ) {
4146 if ( $prevtoclevel < $wgMaxTocLevel ) {
4147 # Unindent only if the previous toc level was shown :p
4148 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4149 $prevtoclevel = $toclevel;
4150 } else {
4151 $toc .= Linker::tocLineEnd();
4152 }
4153 }
4154 } else {
4155 # No change in level, end TOC line
4156 if ( $toclevel < $wgMaxTocLevel ) {
4157 $toc .= Linker::tocLineEnd();
4158 }
4159 }
4160
4161 $levelCount[$toclevel] = $level;
4162
4163 # count number of headlines for each level
4164 $sublevelCount[$toclevel]++;
4165 $dot = 0;
4166 for ( $i = 1; $i <= $toclevel; $i++ ) {
4167 if ( !empty( $sublevelCount[$i] ) ) {
4168 if ( $dot ) {
4169 $numbering .= '.';
4170 }
4171 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4172 $dot = 1;
4173 }
4174 }
4175
4176 # The safe header is a version of the header text safe to use for links
4177
4178 # Remove link placeholders by the link text.
4179 # <!--LINK number-->
4180 # turns into
4181 # link text with suffix
4182 # Do this before unstrip since link text can contain strip markers
4183 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4184
4185 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4186 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4187
4188 # Strip out HTML (first regex removes any tag not allowed)
4189 # Allowed tags are:
4190 # * <sup> and <sub> (T10393)
4191 # * <i> (T28375)
4192 # * <b> (r105284)
4193 # * <bdi> (T74884)
4194 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4195 # * <s> and <strike> (T35715)
4196 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4197 # to allow setting directionality in toc items.
4198 $tocline = preg_replace(
4199 [
4200 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4201 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4202 ],
4203 [ '', '<$1>' ],
4204 $safeHeadline
4205 );
4206
4207 # Strip '<span></span>', which is the result from the above if
4208 # <span id="foo"></span> is used to produce an additional anchor
4209 # for a section.
4210 $tocline = str_replace( '<span></span>', '', $tocline );
4211
4212 $tocline = trim( $tocline );
4213
4214 # For the anchor, strip out HTML-y stuff period
4215 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4216 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4217
4218 # Save headline for section edit hint before it's escaped
4219 $headlineHint = $safeHeadline;
4220
4221 # Decode HTML entities
4222 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4223
4224 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4225
4226 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4227 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4228 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4229 if ( $fallbackHeadline === $safeHeadline ) {
4230 # No reason to have both (in fact, we can't)
4231 $fallbackHeadline = false;
4232 }
4233
4234 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4235 # @todo FIXME: We may be changing them depending on the current locale.
4236 $arrayKey = strtolower( $safeHeadline );
4237 if ( $fallbackHeadline === false ) {
4238 $fallbackArrayKey = false;
4239 } else {
4240 $fallbackArrayKey = strtolower( $fallbackHeadline );
4241 }
4242
4243 # Create the anchor for linking from the TOC to the section
4244 $anchor = $safeHeadline;
4245 $fallbackAnchor = $fallbackHeadline;
4246 if ( isset( $refers[$arrayKey] ) ) {
4247 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4248 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4249 $anchor .= "_$i";
4250 $linkAnchor .= "_$i";
4251 $refers["${arrayKey}_$i"] = true;
4252 } else {
4253 $refers[$arrayKey] = true;
4254 }
4255 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4256 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4257 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4258 $fallbackAnchor .= "_$i";
4259 $refers["${fallbackArrayKey}_$i"] = true;
4260 } else {
4261 $refers[$fallbackArrayKey] = true;
4262 }
4263
4264 # Don't number the heading if it is the only one (looks silly)
4265 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4266 # the two are different if the line contains a link
4267 $headline = Html::element(
4268 'span',
4269 [ 'class' => 'mw-headline-number' ],
4270 $numbering
4271 ) . ' ' . $headline;
4272 }
4273
4274 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4275 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4276 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4277 }
4278
4279 # Add the section to the section tree
4280 # Find the DOM node for this header
4281 $noOffset = ( $isTemplate || $sectionIndex === false );
4282 while ( $node && !$noOffset ) {
4283 if ( $node->getName() === 'h' ) {
4284 $bits = $node->splitHeading();
4285 if ( $bits['i'] == $sectionIndex ) {
4286 break;
4287 }
4288 }
4289 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4290 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4291 $node = $node->getNextSibling();
4292 }
4293 $tocraw[] = [
4294 'toclevel' => $toclevel,
4295 'level' => $level,
4296 'line' => $tocline,
4297 'number' => $numbering,
4298 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4299 'fromtitle' => $titleText,
4300 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4301 'anchor' => $anchor,
4302 ];
4303
4304 # give headline the correct <h#> tag
4305 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4306 // Output edit section links as markers with styles that can be customized by skins
4307 if ( $isTemplate ) {
4308 # Put a T flag in the section identifier, to indicate to extractSections()
4309 # that sections inside <includeonly> should be counted.
4310 $editsectionPage = $titleText;
4311 $editsectionSection = "T-$sectionIndex";
4312 $editsectionContent = null;
4313 } else {
4314 $editsectionPage = $this->mTitle->getPrefixedText();
4315 $editsectionSection = $sectionIndex;
4316 $editsectionContent = $headlineHint;
4317 }
4318 // We use a bit of pesudo-xml for editsection markers. The
4319 // language converter is run later on. Using a UNIQ style marker
4320 // leads to the converter screwing up the tokens when it
4321 // converts stuff. And trying to insert strip tags fails too. At
4322 // this point all real inputted tags have already been escaped,
4323 // so we don't have to worry about a user trying to input one of
4324 // these markers directly. We use a page and section attribute
4325 // to stop the language converter from converting these
4326 // important bits of data, but put the headline hint inside a
4327 // content block because the language converter is supposed to
4328 // be able to convert that piece of data.
4329 // Gets replaced with html in ParserOutput::getText
4330 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4331 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4332 if ( $editsectionContent !== null ) {
4333 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4334 } else {
4335 $editlink .= '/>';
4336 }
4337 } else {
4338 $editlink = '';
4339 }
4340 $head[$headlineCount] = Linker::makeHeadline( $level,
4341 $matches['attrib'][$headlineCount], $anchor, $headline,
4342 $editlink, $fallbackAnchor );
4343
4344 $headlineCount++;
4345 }
4346
4347 $this->setOutputType( $oldType );
4348
4349 # Never ever show TOC if no headers
4350 if ( $numVisible < 1 ) {
4351 $enoughToc = false;
4352 }
4353
4354 if ( $enoughToc ) {
4355 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4356 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4357 }
4358 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4359 $this->mOutput->setTOCHTML( $toc );
4360 $toc = self::TOC_START . $toc . self::TOC_END;
4361 }
4362
4363 if ( $isMain ) {
4364 $this->mOutput->setSections( $tocraw );
4365 }
4366
4367 # split up and insert constructed headlines
4368 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4369 $i = 0;
4370
4371 // build an array of document sections
4372 $sections = [];
4373 foreach ( $blocks as $block ) {
4374 // $head is zero-based, sections aren't.
4375 if ( empty( $head[$i - 1] ) ) {
4376 $sections[$i] = $block;
4377 } else {
4378 $sections[$i] = $head[$i - 1] . $block;
4379 }
4380
4381 /**
4382 * Send a hook, one per section.
4383 * The idea here is to be able to make section-level DIVs, but to do so in a
4384 * lower-impact, more correct way than r50769
4385 *
4386 * $this : caller
4387 * $section : the section number
4388 * &$sectionContent : ref to the content of the section
4389 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4390 */
4391 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4392
4393 $i++;
4394 }
4395
4396 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4397 // append the TOC at the beginning
4398 // Top anchor now in skin
4399 $sections[0] = $sections[0] . $toc . "\n";
4400 }
4401
4402 $full .= implode( '', $sections );
4403
4404 if ( $this->mForceTocPosition ) {
4405 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4406 } else {
4407 return $full;
4408 }
4409 }
4410
4411 /**
4412 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4413 * conversion, substituting signatures, {{subst:}} templates, etc.
4414 *
4415 * @param string $text The text to transform
4416 * @param Title $title The Title object for the current article
4417 * @param User $user The User object describing the current user
4418 * @param ParserOptions $options Parsing options
4419 * @param bool $clearState Whether to clear the parser state first
4420 * @return string The altered wiki markup
4421 */
4422 public function preSaveTransform( $text, Title $title, User $user,
4423 ParserOptions $options, $clearState = true
4424 ) {
4425 if ( $clearState ) {
4426 $magicScopeVariable = $this->lock();
4427 }
4428 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4429 $this->setUser( $user );
4430
4431 // Strip U+0000 NULL (T159174)
4432 $text = str_replace( "\000", '', $text );
4433
4434 // We still normalize line endings for backwards-compatibility
4435 // with other code that just calls PST, but this should already
4436 // be handled in TextContent subclasses
4437 $text = TextContent::normalizeLineEndings( $text );
4438
4439 if ( $options->getPreSaveTransform() ) {
4440 $text = $this->pstPass2( $text, $user );
4441 }
4442 $text = $this->mStripState->unstripBoth( $text );
4443
4444 $this->setUser( null ); # Reset
4445
4446 return $text;
4447 }
4448
4449 /**
4450 * Pre-save transform helper function
4451 *
4452 * @param string $text
4453 * @param User $user
4454 *
4455 * @return string
4456 */
4457 private function pstPass2( $text, $user ) {
4458 global $wgContLang;
4459
4460 # Note: This is the timestamp saved as hardcoded wikitext to
4461 # the database, we use $wgContLang here in order to give
4462 # everyone the same signature and use the default one rather
4463 # than the one selected in each user's preferences.
4464 # (see also T14815)
4465 $ts = $this->mOptions->getTimestamp();
4466 $timestamp = MWTimestamp::getLocalInstance( $ts );
4467 $ts = $timestamp->format( 'YmdHis' );
4468 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4469
4470 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4471
4472 # Variable replacement
4473 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4474 $text = $this->replaceVariables( $text );
4475
4476 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4477 # which may corrupt this parser instance via its wfMessage()->text() call-
4478
4479 # Signatures
4480 if ( strpos( $text, '~~~' ) !== false ) {
4481 $sigText = $this->getUserSig( $user );
4482 $text = strtr( $text, [
4483 '~~~~~' => $d,
4484 '~~~~' => "$sigText $d",
4485 '~~~' => $sigText
4486 ] );
4487 # The main two signature forms used above are time-sensitive
4488 $this->mOutput->setFlag( 'user-signature' );
4489 }
4490
4491 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4492 $tc = '[' . Title::legalChars() . ']';
4493 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4494
4495 // [[ns:page (context)|]]
4496 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4497 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4498 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4499 // [[ns:page (context), context|]] (using either single or double-width comma)
4500 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4501 // [[|page]] (reverse pipe trick: add context from page title)
4502 $p2 = "/\[\[\\|($tc+)]]/";
4503
4504 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4505 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4506 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4507 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4508
4509 $t = $this->mTitle->getText();
4510 $m = [];
4511 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4512 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4513 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4514 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4515 } else {
4516 # if there's no context, don't bother duplicating the title
4517 $text = preg_replace( $p2, '[[\\1]]', $text );
4518 }
4519
4520 return $text;
4521 }
4522
4523 /**
4524 * Fetch the user's signature text, if any, and normalize to
4525 * validated, ready-to-insert wikitext.
4526 * If you have pre-fetched the nickname or the fancySig option, you can
4527 * specify them here to save a database query.
4528 * Do not reuse this parser instance after calling getUserSig(),
4529 * as it may have changed if it's the $wgParser.
4530 *
4531 * @param User &$user
4532 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4533 * @param bool|null $fancySig whether the nicknname is the complete signature
4534 * or null to use default value
4535 * @return string
4536 */
4537 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4538 global $wgMaxSigChars;
4539
4540 $username = $user->getName();
4541
4542 # If not given, retrieve from the user object.
4543 if ( $nickname === false ) {
4544 $nickname = $user->getOption( 'nickname' );
4545 }
4546
4547 if ( is_null( $fancySig ) ) {
4548 $fancySig = $user->getBoolOption( 'fancysig' );
4549 }
4550
4551 $nickname = $nickname == null ? $username : $nickname;
4552
4553 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4554 $nickname = $username;
4555 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4556 } elseif ( $fancySig !== false ) {
4557 # Sig. might contain markup; validate this
4558 if ( $this->validateSig( $nickname ) !== false ) {
4559 # Validated; clean up (if needed) and return it
4560 return $this->cleanSig( $nickname, true );
4561 } else {
4562 # Failed to validate; fall back to the default
4563 $nickname = $username;
4564 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4565 }
4566 }
4567
4568 # Make sure nickname doesnt get a sig in a sig
4569 $nickname = self::cleanSigInSig( $nickname );
4570
4571 # If we're still here, make it a link to the user page
4572 $userText = wfEscapeWikiText( $username );
4573 $nickText = wfEscapeWikiText( $nickname );
4574 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4575
4576 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4577 ->title( $this->getTitle() )->text();
4578 }
4579
4580 /**
4581 * Check that the user's signature contains no bad XML
4582 *
4583 * @param string $text
4584 * @return string|bool An expanded string, or false if invalid.
4585 */
4586 public function validateSig( $text ) {
4587 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4588 }
4589
4590 /**
4591 * Clean up signature text
4592 *
4593 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4594 * 2) Substitute all transclusions
4595 *
4596 * @param string $text
4597 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4598 * @return string Signature text
4599 */
4600 public function cleanSig( $text, $parsing = false ) {
4601 if ( !$parsing ) {
4602 global $wgTitle;
4603 $magicScopeVariable = $this->lock();
4604 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4605 }
4606
4607 # Option to disable this feature
4608 if ( !$this->mOptions->getCleanSignatures() ) {
4609 return $text;
4610 }
4611
4612 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4613 # => Move this logic to braceSubstitution()
4614 $substWord = MagicWord::get( 'subst' );
4615 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4616 $substText = '{{' . $substWord->getSynonym( 0 );
4617
4618 $text = preg_replace( $substRegex, $substText, $text );
4619 $text = self::cleanSigInSig( $text );
4620 $dom = $this->preprocessToDom( $text );
4621 $frame = $this->getPreprocessor()->newFrame();
4622 $text = $frame->expand( $dom );
4623
4624 if ( !$parsing ) {
4625 $text = $this->mStripState->unstripBoth( $text );
4626 }
4627
4628 return $text;
4629 }
4630
4631 /**
4632 * Strip 3, 4 or 5 tildes out of signatures.
4633 *
4634 * @param string $text
4635 * @return string Signature text with /~{3,5}/ removed
4636 */
4637 public static function cleanSigInSig( $text ) {
4638 $text = preg_replace( '/~{3,5}/', '', $text );
4639 return $text;
4640 }
4641
4642 /**
4643 * Set up some variables which are usually set up in parse()
4644 * so that an external function can call some class members with confidence
4645 *
4646 * @param Title|null $title
4647 * @param ParserOptions $options
4648 * @param int $outputType
4649 * @param bool $clearState
4650 */
4651 public function startExternalParse( Title $title = null, ParserOptions $options,
4652 $outputType, $clearState = true
4653 ) {
4654 $this->startParse( $title, $options, $outputType, $clearState );
4655 }
4656
4657 /**
4658 * @param Title|null $title
4659 * @param ParserOptions $options
4660 * @param int $outputType
4661 * @param bool $clearState
4662 */
4663 private function startParse( Title $title = null, ParserOptions $options,
4664 $outputType, $clearState = true
4665 ) {
4666 $this->setTitle( $title );
4667 $this->mOptions = $options;
4668 $this->setOutputType( $outputType );
4669 if ( $clearState ) {
4670 $this->clearState();
4671 }
4672 }
4673
4674 /**
4675 * Wrapper for preprocess()
4676 *
4677 * @param string $text The text to preprocess
4678 * @param ParserOptions $options
4679 * @param Title|null $title Title object or null to use $wgTitle
4680 * @return string
4681 */
4682 public function transformMsg( $text, $options, $title = null ) {
4683 static $executing = false;
4684
4685 # Guard against infinite recursion
4686 if ( $executing ) {
4687 return $text;
4688 }
4689 $executing = true;
4690
4691 if ( !$title ) {
4692 global $wgTitle;
4693 $title = $wgTitle;
4694 }
4695
4696 $text = $this->preprocess( $text, $title, $options );
4697
4698 $executing = false;
4699 return $text;
4700 }
4701
4702 /**
4703 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4704 * The callback should have the following form:
4705 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4706 *
4707 * Transform and return $text. Use $parser for any required context, e.g. use
4708 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4709 *
4710 * Hooks may return extended information by returning an array, of which the
4711 * first numbered element (index 0) must be the return string, and all other
4712 * entries are extracted into local variables within an internal function
4713 * in the Parser class.
4714 *
4715 * This interface (introduced r61913) appears to be undocumented, but
4716 * 'markerType' is used by some core tag hooks to override which strip
4717 * array their results are placed in. **Use great caution if attempting
4718 * this interface, as it is not documented and injudicious use could smash
4719 * private variables.**
4720 *
4721 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4722 * @param callable $callback The callback function (and object) to use for the tag
4723 * @throws MWException
4724 * @return callable|null The old value of the mTagHooks array associated with the hook
4725 */
4726 public function setHook( $tag, callable $callback ) {
4727 $tag = strtolower( $tag );
4728 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4729 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4730 }
4731 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4732 $this->mTagHooks[$tag] = $callback;
4733 if ( !in_array( $tag, $this->mStripList ) ) {
4734 $this->mStripList[] = $tag;
4735 }
4736
4737 return $oldVal;
4738 }
4739
4740 /**
4741 * As setHook(), but letting the contents be parsed.
4742 *
4743 * Transparent tag hooks are like regular XML-style tag hooks, except they
4744 * operate late in the transformation sequence, on HTML instead of wikitext.
4745 *
4746 * This is probably obsoleted by things dealing with parser frames?
4747 * The only extension currently using it is geoserver.
4748 *
4749 * @since 1.10
4750 * @todo better document or deprecate this
4751 *
4752 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4753 * @param callable $callback The callback function (and object) to use for the tag
4754 * @throws MWException
4755 * @return callable|null The old value of the mTagHooks array associated with the hook
4756 */
4757 public function setTransparentTagHook( $tag, callable $callback ) {
4758 $tag = strtolower( $tag );
4759 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4760 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4761 }
4762 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4763 $this->mTransparentTagHooks[$tag] = $callback;
4764
4765 return $oldVal;
4766 }
4767
4768 /**
4769 * Remove all tag hooks
4770 */
4771 public function clearTagHooks() {
4772 $this->mTagHooks = [];
4773 $this->mFunctionTagHooks = [];
4774 $this->mStripList = $this->mDefaultStripList;
4775 }
4776
4777 /**
4778 * Create a function, e.g. {{sum:1|2|3}}
4779 * The callback function should have the form:
4780 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4781 *
4782 * Or with Parser::SFH_OBJECT_ARGS:
4783 * function myParserFunction( $parser, $frame, $args ) { ... }
4784 *
4785 * The callback may either return the text result of the function, or an array with the text
4786 * in element 0, and a number of flags in the other elements. The names of the flags are
4787 * specified in the keys. Valid flags are:
4788 * found The text returned is valid, stop processing the template. This
4789 * is on by default.
4790 * nowiki Wiki markup in the return value should be escaped
4791 * isHTML The returned text is HTML, armour it against wikitext transformation
4792 *
4793 * @param string $id The magic word ID
4794 * @param callable $callback The callback function (and object) to use
4795 * @param int $flags A combination of the following flags:
4796 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4797 *
4798 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4799 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4800 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4801 * the arguments, and to control the way they are expanded.
4802 *
4803 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4804 * arguments, for instance:
4805 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4806 *
4807 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4808 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4809 * working if/when this is changed.
4810 *
4811 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4812 * expansion.
4813 *
4814 * Please read the documentation in includes/parser/Preprocessor.php for more information
4815 * about the methods available in PPFrame and PPNode.
4816 *
4817 * @throws MWException
4818 * @return string|callable The old callback function for this name, if any
4819 */
4820 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4821 global $wgContLang;
4822
4823 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4824 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4825
4826 # Add to function cache
4827 $mw = MagicWord::get( $id );
4828 if ( !$mw ) {
4829 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4830 }
4831
4832 $synonyms = $mw->getSynonyms();
4833 $sensitive = intval( $mw->isCaseSensitive() );
4834
4835 foreach ( $synonyms as $syn ) {
4836 # Case
4837 if ( !$sensitive ) {
4838 $syn = $wgContLang->lc( $syn );
4839 }
4840 # Add leading hash
4841 if ( !( $flags & self::SFH_NO_HASH ) ) {
4842 $syn = '#' . $syn;
4843 }
4844 # Remove trailing colon
4845 if ( substr( $syn, -1, 1 ) === ':' ) {
4846 $syn = substr( $syn, 0, -1 );
4847 }
4848 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4849 }
4850 return $oldVal;
4851 }
4852
4853 /**
4854 * Get all registered function hook identifiers
4855 *
4856 * @return array
4857 */
4858 public function getFunctionHooks() {
4859 return array_keys( $this->mFunctionHooks );
4860 }
4861
4862 /**
4863 * Create a tag function, e.g. "<test>some stuff</test>".
4864 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4865 * Unlike parser functions, their content is not preprocessed.
4866 * @param string $tag
4867 * @param callable $callback
4868 * @param int $flags
4869 * @throws MWException
4870 * @return null
4871 */
4872 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4873 $tag = strtolower( $tag );
4874 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4875 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4876 }
4877 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4878 $this->mFunctionTagHooks[$tag] : null;
4879 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4880
4881 if ( !in_array( $tag, $this->mStripList ) ) {
4882 $this->mStripList[] = $tag;
4883 }
4884
4885 return $old;
4886 }
4887
4888 /**
4889 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4890 * Placeholders created in Linker::link()
4891 *
4892 * @param string &$text
4893 * @param int $options
4894 */
4895 public function replaceLinkHolders( &$text, $options = 0 ) {
4896 $this->mLinkHolders->replace( $text );
4897 }
4898
4899 /**
4900 * Replace "<!--LINK-->" link placeholders with plain text of links
4901 * (not HTML-formatted).
4902 *
4903 * @param string $text
4904 * @return string
4905 */
4906 public function replaceLinkHoldersText( $text ) {
4907 return $this->mLinkHolders->replaceText( $text );
4908 }
4909
4910 /**
4911 * Renders an image gallery from a text with one line per image.
4912 * text labels may be given by using |-style alternative text. E.g.
4913 * Image:one.jpg|The number "1"
4914 * Image:tree.jpg|A tree
4915 * given as text will return the HTML of a gallery with two images,
4916 * labeled 'The number "1"' and
4917 * 'A tree'.
4918 *
4919 * @param string $text
4920 * @param array $params
4921 * @return string HTML
4922 */
4923 public function renderImageGallery( $text, $params ) {
4924 $mode = false;
4925 if ( isset( $params['mode'] ) ) {
4926 $mode = $params['mode'];
4927 }
4928
4929 try {
4930 $ig = ImageGalleryBase::factory( $mode );
4931 } catch ( Exception $e ) {
4932 // If invalid type set, fallback to default.
4933 $ig = ImageGalleryBase::factory( false );
4934 }
4935
4936 $ig->setContextTitle( $this->mTitle );
4937 $ig->setShowBytes( false );
4938 $ig->setShowDimensions( false );
4939 $ig->setShowFilename( false );
4940 $ig->setParser( $this );
4941 $ig->setHideBadImages();
4942 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4943
4944 if ( isset( $params['showfilename'] ) ) {
4945 $ig->setShowFilename( true );
4946 } else {
4947 $ig->setShowFilename( false );
4948 }
4949 if ( isset( $params['caption'] ) ) {
4950 $caption = $params['caption'];
4951 $caption = htmlspecialchars( $caption );
4952 $caption = $this->replaceInternalLinks( $caption );
4953 $ig->setCaptionHtml( $caption );
4954 }
4955 if ( isset( $params['perrow'] ) ) {
4956 $ig->setPerRow( $params['perrow'] );
4957 }
4958 if ( isset( $params['widths'] ) ) {
4959 $ig->setWidths( $params['widths'] );
4960 }
4961 if ( isset( $params['heights'] ) ) {
4962 $ig->setHeights( $params['heights'] );
4963 }
4964 $ig->setAdditionalOptions( $params );
4965
4966 // Avoid PHP 7.1 warning from passing $this by reference
4967 $parser = $this;
4968 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
4969
4970 $lines = StringUtils::explode( "\n", $text );
4971 foreach ( $lines as $line ) {
4972 # match lines like these:
4973 # Image:someimage.jpg|This is some image
4974 $matches = [];
4975 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4976 # Skip empty lines
4977 if ( count( $matches ) == 0 ) {
4978 continue;
4979 }
4980
4981 if ( strpos( $matches[0], '%' ) !== false ) {
4982 $matches[1] = rawurldecode( $matches[1] );
4983 }
4984 $title = Title::newFromText( $matches[1], NS_FILE );
4985 if ( is_null( $title ) ) {
4986 # Bogus title. Ignore these so we don't bomb out later.
4987 continue;
4988 }
4989
4990 # We need to get what handler the file uses, to figure out parameters.
4991 # Note, a hook can overide the file name, and chose an entirely different
4992 # file (which potentially could be of a different type and have different handler).
4993 $options = [];
4994 $descQuery = false;
4995 Hooks::run( 'BeforeParserFetchFileAndTitle',
4996 [ $this, $title, &$options, &$descQuery ] );
4997 # Don't register it now, as TraditionalImageGallery does that later.
4998 $file = $this->fetchFileNoRegister( $title, $options );
4999 $handler = $file ? $file->getHandler() : false;
5000
5001 $paramMap = [
5002 'img_alt' => 'gallery-internal-alt',
5003 'img_link' => 'gallery-internal-link',
5004 ];
5005 if ( $handler ) {
5006 $paramMap = $paramMap + $handler->getParamMap();
5007 // We don't want people to specify per-image widths.
5008 // Additionally the width parameter would need special casing anyhow.
5009 unset( $paramMap['img_width'] );
5010 }
5011
5012 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5013
5014 $label = '';
5015 $alt = '';
5016 $link = '';
5017 $handlerOptions = [];
5018 if ( isset( $matches[3] ) ) {
5019 // look for an |alt= definition while trying not to break existing
5020 // captions with multiple pipes (|) in it, until a more sensible grammar
5021 // is defined for images in galleries
5022
5023 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5024 // splitting on '|' is a bit odd, and different from makeImage.
5025 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5026 // Protect LanguageConverter markup
5027 $parameterMatches = StringUtils::delimiterExplode(
5028 '-{', '}-', '|', $matches[3], true /* nested */
5029 );
5030
5031 foreach ( $parameterMatches as $parameterMatch ) {
5032 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5033 if ( $magicName ) {
5034 $paramName = $paramMap[$magicName];
5035
5036 switch ( $paramName ) {
5037 case 'gallery-internal-alt':
5038 $alt = $this->stripAltText( $match, false );
5039 break;
5040 case 'gallery-internal-link':
5041 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5042 $chars = self::EXT_LINK_URL_CLASS;
5043 $addr = self::EXT_LINK_ADDR;
5044 $prots = $this->mUrlProtocols;
5045 // check to see if link matches an absolute url, if not then it must be a wiki link.
5046 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5047 // Result of LanguageConverter::markNoConversion
5048 // invoked on an external link.
5049 $linkValue = substr( $linkValue, 4, -2 );
5050 }
5051 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5052 $link = $linkValue;
5053 $this->mOutput->addExternalLink( $link );
5054 } else {
5055 $localLinkTitle = Title::newFromText( $linkValue );
5056 if ( $localLinkTitle !== null ) {
5057 $this->mOutput->addLink( $localLinkTitle );
5058 $link = $localLinkTitle->getLinkURL();
5059 }
5060 }
5061 break;
5062 default:
5063 // Must be a handler specific parameter.
5064 if ( $handler->validateParam( $paramName, $match ) ) {
5065 $handlerOptions[$paramName] = $match;
5066 } else {
5067 // Guess not, consider it as caption.
5068 wfDebug( "$parameterMatch failed parameter validation\n" );
5069 $label = '|' . $parameterMatch;
5070 }
5071 }
5072
5073 } else {
5074 // Last pipe wins.
5075 $label = '|' . $parameterMatch;
5076 }
5077 }
5078 // Remove the pipe.
5079 $label = substr( $label, 1 );
5080 }
5081
5082 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5083 }
5084 $html = $ig->toHTML();
5085 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5086 return $html;
5087 }
5088
5089 /**
5090 * @param MediaHandler $handler
5091 * @return array
5092 */
5093 public function getImageParams( $handler ) {
5094 if ( $handler ) {
5095 $handlerClass = get_class( $handler );
5096 } else {
5097 $handlerClass = '';
5098 }
5099 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5100 # Initialise static lists
5101 static $internalParamNames = [
5102 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5103 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5104 'bottom', 'text-bottom' ],
5105 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5106 'upright', 'border', 'link', 'alt', 'class' ],
5107 ];
5108 static $internalParamMap;
5109 if ( !$internalParamMap ) {
5110 $internalParamMap = [];
5111 foreach ( $internalParamNames as $type => $names ) {
5112 foreach ( $names as $name ) {
5113 // For grep: img_left, img_right, img_center, img_none,
5114 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5115 // img_bottom, img_text_bottom,
5116 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5117 // img_border, img_link, img_alt, img_class
5118 $magicName = str_replace( '-', '_', "img_$name" );
5119 $internalParamMap[$magicName] = [ $type, $name ];
5120 }
5121 }
5122 }
5123
5124 # Add handler params
5125 $paramMap = $internalParamMap;
5126 if ( $handler ) {
5127 $handlerParamMap = $handler->getParamMap();
5128 foreach ( $handlerParamMap as $magic => $paramName ) {
5129 $paramMap[$magic] = [ 'handler', $paramName ];
5130 }
5131 }
5132 $this->mImageParams[$handlerClass] = $paramMap;
5133 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5134 }
5135 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5136 }
5137
5138 /**
5139 * Parse image options text and use it to make an image
5140 *
5141 * @param Title $title
5142 * @param string $options
5143 * @param LinkHolderArray|bool $holders
5144 * @return string HTML
5145 */
5146 public function makeImage( $title, $options, $holders = false ) {
5147 # Check if the options text is of the form "options|alt text"
5148 # Options are:
5149 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5150 # * left no resizing, just left align. label is used for alt= only
5151 # * right same, but right aligned
5152 # * none same, but not aligned
5153 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5154 # * center center the image
5155 # * frame Keep original image size, no magnify-button.
5156 # * framed Same as "frame"
5157 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5158 # * upright reduce width for upright images, rounded to full __0 px
5159 # * border draw a 1px border around the image
5160 # * alt Text for HTML alt attribute (defaults to empty)
5161 # * class Set a class for img node
5162 # * link Set the target of the image link. Can be external, interwiki, or local
5163 # vertical-align values (no % or length right now):
5164 # * baseline
5165 # * sub
5166 # * super
5167 # * top
5168 # * text-top
5169 # * middle
5170 # * bottom
5171 # * text-bottom
5172
5173 # Protect LanguageConverter markup when splitting into parts
5174 $parts = StringUtils::delimiterExplode(
5175 '-{', '}-', '|', $options, true /* allow nesting */
5176 );
5177
5178 # Give extensions a chance to select the file revision for us
5179 $options = [];
5180 $descQuery = false;
5181 Hooks::run( 'BeforeParserFetchFileAndTitle',
5182 [ $this, $title, &$options, &$descQuery ] );
5183 # Fetch and register the file (file title may be different via hooks)
5184 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5185
5186 # Get parameter map
5187 $handler = $file ? $file->getHandler() : false;
5188
5189 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5190
5191 if ( !$file ) {
5192 $this->addTrackingCategory( 'broken-file-category' );
5193 }
5194
5195 # Process the input parameters
5196 $caption = '';
5197 $params = [ 'frame' => [], 'handler' => [],
5198 'horizAlign' => [], 'vertAlign' => [] ];
5199 $seenformat = false;
5200 foreach ( $parts as $part ) {
5201 $part = trim( $part );
5202 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5203 $validated = false;
5204 if ( isset( $paramMap[$magicName] ) ) {
5205 list( $type, $paramName ) = $paramMap[$magicName];
5206
5207 # Special case; width and height come in one variable together
5208 if ( $type === 'handler' && $paramName === 'width' ) {
5209 $parsedWidthParam = self::parseWidthParam( $value );
5210 if ( isset( $parsedWidthParam['width'] ) ) {
5211 $width = $parsedWidthParam['width'];
5212 if ( $handler->validateParam( 'width', $width ) ) {
5213 $params[$type]['width'] = $width;
5214 $validated = true;
5215 }
5216 }
5217 if ( isset( $parsedWidthParam['height'] ) ) {
5218 $height = $parsedWidthParam['height'];
5219 if ( $handler->validateParam( 'height', $height ) ) {
5220 $params[$type]['height'] = $height;
5221 $validated = true;
5222 }
5223 }
5224 # else no validation -- T15436
5225 } else {
5226 if ( $type === 'handler' ) {
5227 # Validate handler parameter
5228 $validated = $handler->validateParam( $paramName, $value );
5229 } else {
5230 # Validate internal parameters
5231 switch ( $paramName ) {
5232 case 'manualthumb':
5233 case 'alt':
5234 case 'class':
5235 # @todo FIXME: Possibly check validity here for
5236 # manualthumb? downstream behavior seems odd with
5237 # missing manual thumbs.
5238 $validated = true;
5239 $value = $this->stripAltText( $value, $holders );
5240 break;
5241 case 'link':
5242 $chars = self::EXT_LINK_URL_CLASS;
5243 $addr = self::EXT_LINK_ADDR;
5244 $prots = $this->mUrlProtocols;
5245 if ( $value === '' ) {
5246 $paramName = 'no-link';
5247 $value = true;
5248 $validated = true;
5249 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5250 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5251 $paramName = 'link-url';
5252 $this->mOutput->addExternalLink( $value );
5253 if ( $this->mOptions->getExternalLinkTarget() ) {
5254 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5255 }
5256 $validated = true;
5257 }
5258 } else {
5259 $linkTitle = Title::newFromText( $value );
5260 if ( $linkTitle ) {
5261 $paramName = 'link-title';
5262 $value = $linkTitle;
5263 $this->mOutput->addLink( $linkTitle );
5264 $validated = true;
5265 }
5266 }
5267 break;
5268 case 'frameless':
5269 case 'framed':
5270 case 'thumbnail':
5271 // use first appearing option, discard others.
5272 $validated = !$seenformat;
5273 $seenformat = true;
5274 break;
5275 default:
5276 # Most other things appear to be empty or numeric...
5277 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5278 }
5279 }
5280
5281 if ( $validated ) {
5282 $params[$type][$paramName] = $value;
5283 }
5284 }
5285 }
5286 if ( !$validated ) {
5287 $caption = $part;
5288 }
5289 }
5290
5291 # Process alignment parameters
5292 if ( $params['horizAlign'] ) {
5293 $params['frame']['align'] = key( $params['horizAlign'] );
5294 }
5295 if ( $params['vertAlign'] ) {
5296 $params['frame']['valign'] = key( $params['vertAlign'] );
5297 }
5298
5299 $params['frame']['caption'] = $caption;
5300
5301 # Will the image be presented in a frame, with the caption below?
5302 $imageIsFramed = isset( $params['frame']['frame'] )
5303 || isset( $params['frame']['framed'] )
5304 || isset( $params['frame']['thumbnail'] )
5305 || isset( $params['frame']['manualthumb'] );
5306
5307 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5308 # came to also set the caption, ordinary text after the image -- which
5309 # makes no sense, because that just repeats the text multiple times in
5310 # screen readers. It *also* came to set the title attribute.
5311 # Now that we have an alt attribute, we should not set the alt text to
5312 # equal the caption: that's worse than useless, it just repeats the
5313 # text. This is the framed/thumbnail case. If there's no caption, we
5314 # use the unnamed parameter for alt text as well, just for the time be-
5315 # ing, if the unnamed param is set and the alt param is not.
5316 # For the future, we need to figure out if we want to tweak this more,
5317 # e.g., introducing a title= parameter for the title; ignoring the un-
5318 # named parameter entirely for images without a caption; adding an ex-
5319 # plicit caption= parameter and preserving the old magic unnamed para-
5320 # meter for BC; ...
5321 if ( $imageIsFramed ) { # Framed image
5322 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5323 # No caption or alt text, add the filename as the alt text so
5324 # that screen readers at least get some description of the image
5325 $params['frame']['alt'] = $title->getText();
5326 }
5327 # Do not set $params['frame']['title'] because tooltips don't make sense
5328 # for framed images
5329 } else { # Inline image
5330 if ( !isset( $params['frame']['alt'] ) ) {
5331 # No alt text, use the "caption" for the alt text
5332 if ( $caption !== '' ) {
5333 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5334 } else {
5335 # No caption, fall back to using the filename for the
5336 # alt text
5337 $params['frame']['alt'] = $title->getText();
5338 }
5339 }
5340 # Use the "caption" for the tooltip text
5341 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5342 }
5343
5344 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5345
5346 # Linker does the rest
5347 $time = isset( $options['time'] ) ? $options['time'] : false;
5348 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5349 $time, $descQuery, $this->mOptions->getThumbSize() );
5350
5351 # Give the handler a chance to modify the parser object
5352 if ( $handler ) {
5353 $handler->parserTransformHook( $this, $file );
5354 }
5355
5356 return $ret;
5357 }
5358
5359 /**
5360 * @param string $caption
5361 * @param LinkHolderArray|bool $holders
5362 * @return mixed|string
5363 */
5364 protected function stripAltText( $caption, $holders ) {
5365 # Strip bad stuff out of the title (tooltip). We can't just use
5366 # replaceLinkHoldersText() here, because if this function is called
5367 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5368 if ( $holders ) {
5369 $tooltip = $holders->replaceText( $caption );
5370 } else {
5371 $tooltip = $this->replaceLinkHoldersText( $caption );
5372 }
5373
5374 # make sure there are no placeholders in thumbnail attributes
5375 # that are later expanded to html- so expand them now and
5376 # remove the tags
5377 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5378 $tooltip = Sanitizer::stripAllTags( $tooltip );
5379
5380 return $tooltip;
5381 }
5382
5383 /**
5384 * Set a flag in the output object indicating that the content is dynamic and
5385 * shouldn't be cached.
5386 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5387 */
5388 public function disableCache() {
5389 wfDebug( "Parser output marked as uncacheable.\n" );
5390 if ( !$this->mOutput ) {
5391 throw new MWException( __METHOD__ .
5392 " can only be called when actually parsing something" );
5393 }
5394 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5395 }
5396
5397 /**
5398 * Callback from the Sanitizer for expanding items found in HTML attribute
5399 * values, so they can be safely tested and escaped.
5400 *
5401 * @param string &$text
5402 * @param bool|PPFrame $frame
5403 * @return string
5404 */
5405 public function attributeStripCallback( &$text, $frame = false ) {
5406 $text = $this->replaceVariables( $text, $frame );
5407 $text = $this->mStripState->unstripBoth( $text );
5408 return $text;
5409 }
5410
5411 /**
5412 * Accessor
5413 *
5414 * @return array
5415 */
5416 public function getTags() {
5417 return array_merge(
5418 array_keys( $this->mTransparentTagHooks ),
5419 array_keys( $this->mTagHooks ),
5420 array_keys( $this->mFunctionTagHooks )
5421 );
5422 }
5423
5424 /**
5425 * Replace transparent tags in $text with the values given by the callbacks.
5426 *
5427 * Transparent tag hooks are like regular XML-style tag hooks, except they
5428 * operate late in the transformation sequence, on HTML instead of wikitext.
5429 *
5430 * @param string $text
5431 *
5432 * @return string
5433 */
5434 public function replaceTransparentTags( $text ) {
5435 $matches = [];
5436 $elements = array_keys( $this->mTransparentTagHooks );
5437 $text = self::extractTagsAndParams( $elements, $text, $matches );
5438 $replacements = [];
5439
5440 foreach ( $matches as $marker => $data ) {
5441 list( $element, $content, $params, $tag ) = $data;
5442 $tagName = strtolower( $element );
5443 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5444 $output = call_user_func_array(
5445 $this->mTransparentTagHooks[$tagName],
5446 [ $content, $params, $this ]
5447 );
5448 } else {
5449 $output = $tag;
5450 }
5451 $replacements[$marker] = $output;
5452 }
5453 return strtr( $text, $replacements );
5454 }
5455
5456 /**
5457 * Break wikitext input into sections, and either pull or replace
5458 * some particular section's text.
5459 *
5460 * External callers should use the getSection and replaceSection methods.
5461 *
5462 * @param string $text Page wikitext
5463 * @param string|int $sectionId A section identifier string of the form:
5464 * "<flag1> - <flag2> - ... - <section number>"
5465 *
5466 * Currently the only recognised flag is "T", which means the target section number
5467 * was derived during a template inclusion parse, in other words this is a template
5468 * section edit link. If no flags are given, it was an ordinary section edit link.
5469 * This flag is required to avoid a section numbering mismatch when a section is
5470 * enclosed by "<includeonly>" (T8563).
5471 *
5472 * The section number 0 pulls the text before the first heading; other numbers will
5473 * pull the given section along with its lower-level subsections. If the section is
5474 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5475 *
5476 * Section 0 is always considered to exist, even if it only contains the empty
5477 * string. If $text is the empty string and section 0 is replaced, $newText is
5478 * returned.
5479 *
5480 * @param string $mode One of "get" or "replace"
5481 * @param string $newText Replacement text for section data.
5482 * @return string For "get", the extracted section text.
5483 * for "replace", the whole page with the section replaced.
5484 */
5485 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5486 global $wgTitle; # not generally used but removes an ugly failure mode
5487
5488 $magicScopeVariable = $this->lock();
5489 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5490 $outText = '';
5491 $frame = $this->getPreprocessor()->newFrame();
5492
5493 # Process section extraction flags
5494 $flags = 0;
5495 $sectionParts = explode( '-', $sectionId );
5496 $sectionIndex = array_pop( $sectionParts );
5497 foreach ( $sectionParts as $part ) {
5498 if ( $part === 'T' ) {
5499 $flags |= self::PTD_FOR_INCLUSION;
5500 }
5501 }
5502
5503 # Check for empty input
5504 if ( strval( $text ) === '' ) {
5505 # Only sections 0 and T-0 exist in an empty document
5506 if ( $sectionIndex == 0 ) {
5507 if ( $mode === 'get' ) {
5508 return '';
5509 } else {
5510 return $newText;
5511 }
5512 } else {
5513 if ( $mode === 'get' ) {
5514 return $newText;
5515 } else {
5516 return $text;
5517 }
5518 }
5519 }
5520
5521 # Preprocess the text
5522 $root = $this->preprocessToDom( $text, $flags );
5523
5524 # <h> nodes indicate section breaks
5525 # They can only occur at the top level, so we can find them by iterating the root's children
5526 $node = $root->getFirstChild();
5527
5528 # Find the target section
5529 if ( $sectionIndex == 0 ) {
5530 # Section zero doesn't nest, level=big
5531 $targetLevel = 1000;
5532 } else {
5533 while ( $node ) {
5534 if ( $node->getName() === 'h' ) {
5535 $bits = $node->splitHeading();
5536 if ( $bits['i'] == $sectionIndex ) {
5537 $targetLevel = $bits['level'];
5538 break;
5539 }
5540 }
5541 if ( $mode === 'replace' ) {
5542 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5543 }
5544 $node = $node->getNextSibling();
5545 }
5546 }
5547
5548 if ( !$node ) {
5549 # Not found
5550 if ( $mode === 'get' ) {
5551 return $newText;
5552 } else {
5553 return $text;
5554 }
5555 }
5556
5557 # Find the end of the section, including nested sections
5558 do {
5559 if ( $node->getName() === 'h' ) {
5560 $bits = $node->splitHeading();
5561 $curLevel = $bits['level'];
5562 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5563 break;
5564 }
5565 }
5566 if ( $mode === 'get' ) {
5567 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5568 }
5569 $node = $node->getNextSibling();
5570 } while ( $node );
5571
5572 # Write out the remainder (in replace mode only)
5573 if ( $mode === 'replace' ) {
5574 # Output the replacement text
5575 # Add two newlines on -- trailing whitespace in $newText is conventionally
5576 # stripped by the editor, so we need both newlines to restore the paragraph gap
5577 # Only add trailing whitespace if there is newText
5578 if ( $newText != "" ) {
5579 $outText .= $newText . "\n\n";
5580 }
5581
5582 while ( $node ) {
5583 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5584 $node = $node->getNextSibling();
5585 }
5586 }
5587
5588 if ( is_string( $outText ) ) {
5589 # Re-insert stripped tags
5590 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5591 }
5592
5593 return $outText;
5594 }
5595
5596 /**
5597 * This function returns the text of a section, specified by a number ($section).
5598 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5599 * the first section before any such heading (section 0).
5600 *
5601 * If a section contains subsections, these are also returned.
5602 *
5603 * @param string $text Text to look in
5604 * @param string|int $sectionId Section identifier as a number or string
5605 * (e.g. 0, 1 or 'T-1').
5606 * @param string $defaultText Default to return if section is not found
5607 *
5608 * @return string Text of the requested section
5609 */
5610 public function getSection( $text, $sectionId, $defaultText = '' ) {
5611 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5612 }
5613
5614 /**
5615 * This function returns $oldtext after the content of the section
5616 * specified by $section has been replaced with $text. If the target
5617 * section does not exist, $oldtext is returned unchanged.
5618 *
5619 * @param string $oldText Former text of the article
5620 * @param string|int $sectionId Section identifier as a number or string
5621 * (e.g. 0, 1 or 'T-1').
5622 * @param string $newText Replacing text
5623 *
5624 * @return string Modified text
5625 */
5626 public function replaceSection( $oldText, $sectionId, $newText ) {
5627 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5628 }
5629
5630 /**
5631 * Get the ID of the revision we are parsing
5632 *
5633 * @return int|null
5634 */
5635 public function getRevisionId() {
5636 return $this->mRevisionId;
5637 }
5638
5639 /**
5640 * Get the revision object for $this->mRevisionId
5641 *
5642 * @return Revision|null Either a Revision object or null
5643 * @since 1.23 (public since 1.23)
5644 */
5645 public function getRevisionObject() {
5646 if ( !is_null( $this->mRevisionObject ) ) {
5647 return $this->mRevisionObject;
5648 }
5649 if ( is_null( $this->mRevisionId ) ) {
5650 return null;
5651 }
5652
5653 $rev = call_user_func(
5654 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5655 );
5656
5657 # If the parse is for a new revision, then the callback should have
5658 # already been set to force the object and should match mRevisionId.
5659 # If not, try to fetch by mRevisionId for sanity.
5660 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5661 $rev = Revision::newFromId( $this->mRevisionId );
5662 }
5663
5664 $this->mRevisionObject = $rev;
5665
5666 return $this->mRevisionObject;
5667 }
5668
5669 /**
5670 * Get the timestamp associated with the current revision, adjusted for
5671 * the default server-local timestamp
5672 * @return string
5673 */
5674 public function getRevisionTimestamp() {
5675 if ( is_null( $this->mRevisionTimestamp ) ) {
5676 global $wgContLang;
5677
5678 $revObject = $this->getRevisionObject();
5679 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5680
5681 # The cryptic '' timezone parameter tells to use the site-default
5682 # timezone offset instead of the user settings.
5683 # Since this value will be saved into the parser cache, served
5684 # to other users, and potentially even used inside links and such,
5685 # it needs to be consistent for all visitors.
5686 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5687
5688 }
5689 return $this->mRevisionTimestamp;
5690 }
5691
5692 /**
5693 * Get the name of the user that edited the last revision
5694 *
5695 * @return string User name
5696 */
5697 public function getRevisionUser() {
5698 if ( is_null( $this->mRevisionUser ) ) {
5699 $revObject = $this->getRevisionObject();
5700
5701 # if this template is subst: the revision id will be blank,
5702 # so just use the current user's name
5703 if ( $revObject ) {
5704 $this->mRevisionUser = $revObject->getUserText();
5705 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5706 $this->mRevisionUser = $this->getUser()->getName();
5707 }
5708 }
5709 return $this->mRevisionUser;
5710 }
5711
5712 /**
5713 * Get the size of the revision
5714 *
5715 * @return int|null Revision size
5716 */
5717 public function getRevisionSize() {
5718 if ( is_null( $this->mRevisionSize ) ) {
5719 $revObject = $this->getRevisionObject();
5720
5721 # if this variable is subst: the revision id will be blank,
5722 # so just use the parser input size, because the own substituation
5723 # will change the size.
5724 if ( $revObject ) {
5725 $this->mRevisionSize = $revObject->getSize();
5726 } else {
5727 $this->mRevisionSize = $this->mInputSize;
5728 }
5729 }
5730 return $this->mRevisionSize;
5731 }
5732
5733 /**
5734 * Mutator for $mDefaultSort
5735 *
5736 * @param string $sort New value
5737 */
5738 public function setDefaultSort( $sort ) {
5739 $this->mDefaultSort = $sort;
5740 $this->mOutput->setProperty( 'defaultsort', $sort );
5741 }
5742
5743 /**
5744 * Accessor for $mDefaultSort
5745 * Will use the empty string if none is set.
5746 *
5747 * This value is treated as a prefix, so the
5748 * empty string is equivalent to sorting by
5749 * page name.
5750 *
5751 * @return string
5752 */
5753 public function getDefaultSort() {
5754 if ( $this->mDefaultSort !== false ) {
5755 return $this->mDefaultSort;
5756 } else {
5757 return '';
5758 }
5759 }
5760
5761 /**
5762 * Accessor for $mDefaultSort
5763 * Unlike getDefaultSort(), will return false if none is set
5764 *
5765 * @return string|bool
5766 */
5767 public function getCustomDefaultSort() {
5768 return $this->mDefaultSort;
5769 }
5770
5771 private static function getSectionNameFromStrippedText( $text ) {
5772 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5773 $text = Sanitizer::decodeCharReferences( $text );
5774 $text = self::normalizeSectionName( $text );
5775 return $text;
5776 }
5777
5778 private static function makeAnchor( $sectionName ) {
5779 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5780 }
5781
5782 private static function makeLegacyAnchor( $sectionName ) {
5783 global $wgFragmentMode;
5784 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5785 // ForAttribute() and ForLink() are the same for legacy encoding
5786 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5787 } else {
5788 $id = Sanitizer::escapeIdForLink( $sectionName );
5789 }
5790
5791 return "#$id";
5792 }
5793
5794 /**
5795 * Try to guess the section anchor name based on a wikitext fragment
5796 * presumably extracted from a heading, for example "Header" from
5797 * "== Header ==".
5798 *
5799 * @param string $text
5800 * @return string Anchor (starting with '#')
5801 */
5802 public function guessSectionNameFromWikiText( $text ) {
5803 # Strip out wikitext links(they break the anchor)
5804 $text = $this->stripSectionName( $text );
5805 $sectionName = self::getSectionNameFromStrippedText( $text );
5806 return self::makeAnchor( $sectionName );
5807 }
5808
5809 /**
5810 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5811 * instead, if possible. For use in redirects, since various versions
5812 * of Microsoft browsers interpret Location: headers as something other
5813 * than UTF-8, resulting in breakage.
5814 *
5815 * @param string $text The section name
5816 * @return string Anchor (starting with '#')
5817 */
5818 public function guessLegacySectionNameFromWikiText( $text ) {
5819 # Strip out wikitext links(they break the anchor)
5820 $text = $this->stripSectionName( $text );
5821 $sectionName = self::getSectionNameFromStrippedText( $text );
5822 return self::makeLegacyAnchor( $sectionName );
5823 }
5824
5825 /**
5826 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5827 * @param string $text Section name (plain text)
5828 * @return string Anchor (starting with '#')
5829 */
5830 public static function guessSectionNameFromStrippedText( $text ) {
5831 $sectionName = self::getSectionNameFromStrippedText( $text );
5832 return self::makeAnchor( $sectionName );
5833 }
5834
5835 /**
5836 * Apply the same normalization as code making links to this section would
5837 *
5838 * @param string $text
5839 * @return string
5840 */
5841 private static function normalizeSectionName( $text ) {
5842 # T90902: ensure the same normalization is applied for IDs as to links
5843 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5844 try {
5845
5846 $parts = $titleParser->splitTitleString( "#$text" );
5847 } catch ( MalformedTitleException $ex ) {
5848 return $text;
5849 }
5850 return $parts['fragment'];
5851 }
5852
5853 /**
5854 * Strips a text string of wikitext for use in a section anchor
5855 *
5856 * Accepts a text string and then removes all wikitext from the
5857 * string and leaves only the resultant text (i.e. the result of
5858 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5859 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5860 * to create valid section anchors by mimicing the output of the
5861 * parser when headings are parsed.
5862 *
5863 * @param string $text Text string to be stripped of wikitext
5864 * for use in a Section anchor
5865 * @return string Filtered text string
5866 */
5867 public function stripSectionName( $text ) {
5868 # Strip internal link markup
5869 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5870 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5871
5872 # Strip external link markup
5873 # @todo FIXME: Not tolerant to blank link text
5874 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5875 # on how many empty links there are on the page - need to figure that out.
5876 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5877
5878 # Parse wikitext quotes (italics & bold)
5879 $text = $this->doQuotes( $text );
5880
5881 # Strip HTML tags
5882 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5883 return $text;
5884 }
5885
5886 /**
5887 * strip/replaceVariables/unstrip for preprocessor regression testing
5888 *
5889 * @param string $text
5890 * @param Title $title
5891 * @param ParserOptions $options
5892 * @param int $outputType
5893 *
5894 * @return string
5895 */
5896 public function testSrvus( $text, Title $title, ParserOptions $options,
5897 $outputType = self::OT_HTML
5898 ) {
5899 $magicScopeVariable = $this->lock();
5900 $this->startParse( $title, $options, $outputType, true );
5901
5902 $text = $this->replaceVariables( $text );
5903 $text = $this->mStripState->unstripBoth( $text );
5904 $text = Sanitizer::removeHTMLtags( $text );
5905 return $text;
5906 }
5907
5908 /**
5909 * @param string $text
5910 * @param Title $title
5911 * @param ParserOptions $options
5912 * @return string
5913 */
5914 public function testPst( $text, Title $title, ParserOptions $options ) {
5915 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5916 }
5917
5918 /**
5919 * @param string $text
5920 * @param Title $title
5921 * @param ParserOptions $options
5922 * @return string
5923 */
5924 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5925 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5926 }
5927
5928 /**
5929 * Call a callback function on all regions of the given text that are not
5930 * inside strip markers, and replace those regions with the return value
5931 * of the callback. For example, with input:
5932 *
5933 * aaa<MARKER>bbb
5934 *
5935 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5936 * two strings will be replaced with the value returned by the callback in
5937 * each case.
5938 *
5939 * @param string $s
5940 * @param callable $callback
5941 *
5942 * @return string
5943 */
5944 public function markerSkipCallback( $s, $callback ) {
5945 $i = 0;
5946 $out = '';
5947 while ( $i < strlen( $s ) ) {
5948 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5949 if ( $markerStart === false ) {
5950 $out .= call_user_func( $callback, substr( $s, $i ) );
5951 break;
5952 } else {
5953 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5954 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5955 if ( $markerEnd === false ) {
5956 $out .= substr( $s, $markerStart );
5957 break;
5958 } else {
5959 $markerEnd += strlen( self::MARKER_SUFFIX );
5960 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5961 $i = $markerEnd;
5962 }
5963 }
5964 }
5965 return $out;
5966 }
5967
5968 /**
5969 * Remove any strip markers found in the given text.
5970 *
5971 * @param string $text
5972 * @return string
5973 */
5974 public function killMarkers( $text ) {
5975 return $this->mStripState->killMarkers( $text );
5976 }
5977
5978 /**
5979 * Save the parser state required to convert the given half-parsed text to
5980 * HTML. "Half-parsed" in this context means the output of
5981 * recursiveTagParse() or internalParse(). This output has strip markers
5982 * from replaceVariables (extensionSubstitution() etc.), and link
5983 * placeholders from replaceLinkHolders().
5984 *
5985 * Returns an array which can be serialized and stored persistently. This
5986 * array can later be loaded into another parser instance with
5987 * unserializeHalfParsedText(). The text can then be safely incorporated into
5988 * the return value of a parser hook.
5989 *
5990 * @param string $text
5991 *
5992 * @return array
5993 */
5994 public function serializeHalfParsedText( $text ) {
5995 $data = [
5996 'text' => $text,
5997 'version' => self::HALF_PARSED_VERSION,
5998 'stripState' => $this->mStripState->getSubState( $text ),
5999 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6000 ];
6001 return $data;
6002 }
6003
6004 /**
6005 * Load the parser state given in the $data array, which is assumed to
6006 * have been generated by serializeHalfParsedText(). The text contents is
6007 * extracted from the array, and its markers are transformed into markers
6008 * appropriate for the current Parser instance. This transformed text is
6009 * returned, and can be safely included in the return value of a parser
6010 * hook.
6011 *
6012 * If the $data array has been stored persistently, the caller should first
6013 * check whether it is still valid, by calling isValidHalfParsedText().
6014 *
6015 * @param array $data Serialized data
6016 * @throws MWException
6017 * @return string
6018 */
6019 public function unserializeHalfParsedText( $data ) {
6020 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6021 throw new MWException( __METHOD__ . ': invalid version' );
6022 }
6023
6024 # First, extract the strip state.
6025 $texts = [ $data['text'] ];
6026 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6027
6028 # Now renumber links
6029 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6030
6031 # Should be good to go.
6032 return $texts[0];
6033 }
6034
6035 /**
6036 * Returns true if the given array, presumed to be generated by
6037 * serializeHalfParsedText(), is compatible with the current version of the
6038 * parser.
6039 *
6040 * @param array $data
6041 *
6042 * @return bool
6043 */
6044 public function isValidHalfParsedText( $data ) {
6045 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6046 }
6047
6048 /**
6049 * Parsed a width param of imagelink like 300px or 200x300px
6050 *
6051 * @param string $value
6052 * @param bool $parseHeight
6053 *
6054 * @return array
6055 * @since 1.20
6056 */
6057 public static function parseWidthParam( $value, $parseHeight = true ) {
6058 $parsedWidthParam = [];
6059 if ( $value === '' ) {
6060 return $parsedWidthParam;
6061 }
6062 $m = [];
6063 # (T15500) In both cases (width/height and width only),
6064 # permit trailing "px" for backward compatibility.
6065 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6066 $width = intval( $m[1] );
6067 $height = intval( $m[2] );
6068 $parsedWidthParam['width'] = $width;
6069 $parsedWidthParam['height'] = $height;
6070 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6071 $width = intval( $value );
6072 $parsedWidthParam['width'] = $width;
6073 }
6074 return $parsedWidthParam;
6075 }
6076
6077 /**
6078 * Lock the current instance of the parser.
6079 *
6080 * This is meant to stop someone from calling the parser
6081 * recursively and messing up all the strip state.
6082 *
6083 * @throws MWException If parser is in a parse
6084 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6085 */
6086 protected function lock() {
6087 if ( $this->mInParse ) {
6088 throw new MWException( "Parser state cleared while parsing. "
6089 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6090 }
6091
6092 // Save the backtrace when locking, so that if some code tries locking again,
6093 // we can print the lock owner's backtrace for easier debugging
6094 $e = new Exception;
6095 $this->mInParse = $e->getTraceAsString();
6096
6097 $recursiveCheck = new ScopedCallback( function () {
6098 $this->mInParse = false;
6099 } );
6100
6101 return $recursiveCheck;
6102 }
6103
6104 /**
6105 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6106 *
6107 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6108 * or if there is more than one <p/> tag in the input HTML.
6109 *
6110 * @param string $html
6111 * @return string
6112 * @since 1.24
6113 */
6114 public static function stripOuterParagraph( $html ) {
6115 $m = [];
6116 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6117 if ( strpos( $m[1], '</p>' ) === false ) {
6118 $html = $m[1];
6119 }
6120 }
6121
6122 return $html;
6123 }
6124
6125 /**
6126 * Return this parser if it is not doing anything, otherwise
6127 * get a fresh parser. You can use this method by doing
6128 * $myParser = $wgParser->getFreshParser(), or more simply
6129 * $wgParser->getFreshParser()->parse( ... );
6130 * if you're unsure if $wgParser is safe to use.
6131 *
6132 * @since 1.24
6133 * @return Parser A parser object that is not parsing anything
6134 */
6135 public function getFreshParser() {
6136 global $wgParserConf;
6137 if ( $this->mInParse ) {
6138 return new $wgParserConf['class']( $wgParserConf );
6139 } else {
6140 return $this;
6141 }
6142 }
6143
6144 /**
6145 * Set's up the PHP implementation of OOUI for use in this request
6146 * and instructs OutputPage to enable OOUI for itself.
6147 *
6148 * @since 1.26
6149 */
6150 public function enableOOUI() {
6151 OutputPage::setupOOUI();
6152 $this->mOutput->setEnableOOUI( true );
6153 }
6154 }
Wiklou, le Wiki du Wiklou