dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Make interwiki transclusion use the WAN cache
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // phpcs:ignore Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102
103 # Regular expression for a non-newline space
104 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
105
106 # Flags for preprocessToDom
107 const PTD_FOR_INCLUSION = 1;
108
109 # Allowed values for $this->mOutputType
110 # Parameter to startExternalParse().
111 const OT_HTML = 1; # like parse()
112 const OT_WIKI = 2; # like preSaveTransform()
113 const OT_PREPROCESS = 3; # like preprocess()
114 const OT_MSG = 3;
115 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
116
117 /**
118 * @var string Prefix and suffix for temporary replacement strings
119 * for the multipass parser.
120 *
121 * \x7f should never appear in input as it's disallowed in XML.
122 * Using it at the front also gives us a little extra robustness
123 * since it shouldn't match when butted up against identifier-like
124 * string constructs.
125 *
126 * Must not consist of all title characters, or else it will change
127 * the behavior of <nowiki> in a link.
128 *
129 * Must have a character that needs escaping in attributes, otherwise
130 * someone could put a strip marker in an attribute, to get around
131 * escaping quote marks, and break out of the attribute. Thus we add
132 * `'".
133 */
134 const MARKER_SUFFIX = "-QINU`\"'\x7f";
135 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
136
137 # Markers used for wrapping the table of contents
138 const TOC_START = '<mw:toc>';
139 const TOC_END = '</mw:toc>';
140
141 /** @var int Assume that no output will later be saved this many seconds after parsing */
142 const MAX_TTS = 900;
143
144 # Persistent:
145 public $mTagHooks = [];
146 public $mTransparentTagHooks = [];
147 public $mFunctionHooks = [];
148 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
149 public $mFunctionTagHooks = [];
150 public $mStripList = [];
151 public $mDefaultStripList = [];
152 public $mVarCache = [];
153 public $mImageParams = [];
154 public $mImageParamsMagicArray = [];
155 public $mMarkerIndex = 0;
156 public $mFirstCall = true;
157
158 # Initialised by initialiseVariables()
159
160 /**
161 * @var MagicWordArray
162 */
163 public $mVariables;
164
165 /**
166 * @var MagicWordArray
167 */
168 public $mSubstWords;
169 # Initialised in constructor
170 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
171
172 # Initialized in getPreprocessor()
173 /** @var Preprocessor */
174 public $mPreprocessor;
175
176 # Cleared with clearState():
177 /**
178 * @var ParserOutput
179 */
180 public $mOutput;
181 public $mAutonumber;
182
183 /**
184 * @var StripState
185 */
186 public $mStripState;
187
188 public $mIncludeCount;
189 /**
190 * @var LinkHolderArray
191 */
192 public $mLinkHolders;
193
194 public $mLinkID;
195 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
196 public $mDefaultSort;
197 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
198 public $mExpensiveFunctionCount; # number of expensive parser function calls
199 public $mShowToc, $mForceTocPosition;
200
201 /**
202 * @var User
203 */
204 public $mUser; # User object; only used when doing pre-save transform
205
206 # Temporary
207 # These are variables reset at least once per parse regardless of $clearState
208
209 /**
210 * @var ParserOptions
211 */
212 public $mOptions;
213
214 /**
215 * @var Title
216 */
217 public $mTitle; # Title context, used for self-link rendering and similar things
218 public $mOutputType; # Output type, one of the OT_xxx constants
219 public $ot; # Shortcut alias, see setOutputType()
220 public $mRevisionObject; # The revision object of the specified revision ID
221 public $mRevisionId; # ID to display in {{REVISIONID}} tags
222 public $mRevisionTimestamp; # The timestamp of the specified revision ID
223 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
224 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
225 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
226 public $mInputSize = false; # For {{PAGESIZE}} on current page.
227
228 /**
229 * @var string Deprecated accessor for the strip marker prefix.
230 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
231 */
232 public $mUniqPrefix = self::MARKER_PREFIX;
233
234 /**
235 * @var array Array with the language name of each language link (i.e. the
236 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
237 * duplicate language links to the ParserOutput.
238 */
239 public $mLangLinkLanguages;
240
241 /**
242 * @var MapCacheLRU|null
243 * @since 1.24
244 *
245 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
246 */
247 public $currentRevisionCache;
248
249 /**
250 * @var bool|string Recursive call protection.
251 * This variable should be treated as if it were private.
252 */
253 public $mInParse = false;
254
255 /** @var SectionProfiler */
256 protected $mProfiler;
257
258 /**
259 * @var LinkRenderer
260 */
261 protected $mLinkRenderer;
262
263 /**
264 * @param array $conf
265 */
266 public function __construct( $conf = [] ) {
267 $this->mConf = $conf;
268 $this->mUrlProtocols = wfUrlProtocols();
269 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
270 self::EXT_LINK_ADDR .
271 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
272 if ( isset( $conf['preprocessorClass'] ) ) {
273 $this->mPreprocessorClass = $conf['preprocessorClass'];
274 } elseif ( defined( 'HPHP_VERSION' ) ) {
275 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
276 $this->mPreprocessorClass = Preprocessor_Hash::class;
277 } elseif ( extension_loaded( 'domxml' ) ) {
278 # PECL extension that conflicts with the core DOM extension (T15770)
279 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
280 $this->mPreprocessorClass = Preprocessor_Hash::class;
281 } elseif ( extension_loaded( 'dom' ) ) {
282 $this->mPreprocessorClass = Preprocessor_DOM::class;
283 } else {
284 $this->mPreprocessorClass = Preprocessor_Hash::class;
285 }
286 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
287 }
288
289 /**
290 * Reduce memory usage to reduce the impact of circular references
291 */
292 public function __destruct() {
293 if ( isset( $this->mLinkHolders ) ) {
294 unset( $this->mLinkHolders );
295 }
296 foreach ( $this as $name => $value ) {
297 unset( $this->$name );
298 }
299 }
300
301 /**
302 * Allow extensions to clean up when the parser is cloned
303 */
304 public function __clone() {
305 $this->mInParse = false;
306
307 // T58226: When you create a reference "to" an object field, that
308 // makes the object field itself be a reference too (until the other
309 // reference goes out of scope). When cloning, any field that's a
310 // reference is copied as a reference in the new object. Both of these
311 // are defined PHP5 behaviors, as inconvenient as it is for us when old
312 // hooks from PHP4 days are passing fields by reference.
313 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
314 // Make a non-reference copy of the field, then rebind the field to
315 // reference the new copy.
316 $tmp = $this->$k;
317 $this->$k =& $tmp;
318 unset( $tmp );
319 }
320
321 Hooks::run( 'ParserCloned', [ $this ] );
322 }
323
324 /**
325 * Do various kinds of initialisation on the first call of the parser
326 */
327 public function firstCallInit() {
328 if ( !$this->mFirstCall ) {
329 return;
330 }
331 $this->mFirstCall = false;
332
333 CoreParserFunctions::register( $this );
334 CoreTagHooks::register( $this );
335 $this->initialiseVariables();
336
337 // Avoid PHP 7.1 warning from passing $this by reference
338 $parser = $this;
339 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
340 }
341
342 /**
343 * Clear Parser state
344 *
345 * @private
346 */
347 public function clearState() {
348 if ( $this->mFirstCall ) {
349 $this->firstCallInit();
350 }
351 $this->mOutput = new ParserOutput;
352 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
353 $this->mAutonumber = 0;
354 $this->mIncludeCount = [];
355 $this->mLinkHolders = new LinkHolderArray( $this );
356 $this->mLinkID = 0;
357 $this->mRevisionObject = $this->mRevisionTimestamp =
358 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
359 $this->mVarCache = [];
360 $this->mUser = null;
361 $this->mLangLinkLanguages = [];
362 $this->currentRevisionCache = null;
363
364 $this->mStripState = new StripState( $this );
365
366 # Clear these on every parse, T6549
367 $this->mTplRedirCache = $this->mTplDomCache = [];
368
369 $this->mShowToc = true;
370 $this->mForceTocPosition = false;
371 $this->mIncludeSizes = [
372 'post-expand' => 0,
373 'arg' => 0,
374 ];
375 $this->mPPNodeCount = 0;
376 $this->mGeneratedPPNodeCount = 0;
377 $this->mHighestExpansionDepth = 0;
378 $this->mDefaultSort = false;
379 $this->mHeadings = [];
380 $this->mDoubleUnderscores = [];
381 $this->mExpensiveFunctionCount = 0;
382
383 # Fix cloning
384 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
385 $this->mPreprocessor = null;
386 }
387
388 $this->mProfiler = new SectionProfiler();
389
390 // Avoid PHP 7.1 warning from passing $this by reference
391 $parser = $this;
392 Hooks::run( 'ParserClearState', [ &$parser ] );
393 }
394
395 /**
396 * Convert wikitext to HTML
397 * Do not call this function recursively.
398 *
399 * @param string $text Text we want to parse
400 * @param Title $title
401 * @param ParserOptions $options
402 * @param bool $linestart
403 * @param bool $clearState
404 * @param int|null $revid Number to pass in {{REVISIONID}}
405 * @return ParserOutput A ParserOutput
406 */
407 public function parse(
408 $text, Title $title, ParserOptions $options,
409 $linestart = true, $clearState = true, $revid = null
410 ) {
411 if ( $clearState ) {
412 // We use U+007F DELETE to construct strip markers, so we have to make
413 // sure that this character does not occur in the input text.
414 $text = strtr( $text, "\x7f", "?" );
415 $magicScopeVariable = $this->lock();
416 }
417 // Strip U+0000 NULL (T159174)
418 $text = str_replace( "\000", '', $text );
419
420 $this->startParse( $title, $options, self::OT_HTML, $clearState );
421
422 $this->currentRevisionCache = null;
423 $this->mInputSize = strlen( $text );
424 if ( $this->mOptions->getEnableLimitReport() ) {
425 $this->mOutput->resetParseStartTime();
426 }
427
428 $oldRevisionId = $this->mRevisionId;
429 $oldRevisionObject = $this->mRevisionObject;
430 $oldRevisionTimestamp = $this->mRevisionTimestamp;
431 $oldRevisionUser = $this->mRevisionUser;
432 $oldRevisionSize = $this->mRevisionSize;
433 if ( $revid !== null ) {
434 $this->mRevisionId = $revid;
435 $this->mRevisionObject = null;
436 $this->mRevisionTimestamp = null;
437 $this->mRevisionUser = null;
438 $this->mRevisionSize = null;
439 }
440
441 // Avoid PHP 7.1 warning from passing $this by reference
442 $parser = $this;
443 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
444 # No more strip!
445 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
446 $text = $this->internalParse( $text );
447 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
448
449 $text = $this->internalParseHalfParsed( $text, true, $linestart );
450
451 /**
452 * A converted title will be provided in the output object if title and
453 * content conversion are enabled, the article text does not contain
454 * a conversion-suppressing double-underscore tag, and no
455 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
456 * automatic link conversion.
457 */
458 if ( !( $options->getDisableTitleConversion()
459 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
460 || isset( $this->mDoubleUnderscores['notitleconvert'] )
461 || $this->mOutput->getDisplayTitle() !== false )
462 ) {
463 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
464 if ( $convruletitle ) {
465 $this->mOutput->setTitleText( $convruletitle );
466 } else {
467 $titleText = $this->getTargetLanguage()->convertTitle( $title );
468 $this->mOutput->setTitleText( $titleText );
469 }
470 }
471
472 # Compute runtime adaptive expiry if set
473 $this->mOutput->finalizeAdaptiveCacheExpiry();
474
475 # Warn if too many heavyweight parser functions were used
476 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
477 $this->limitationWarn( 'expensive-parserfunction',
478 $this->mExpensiveFunctionCount,
479 $this->mOptions->getExpensiveParserFunctionLimit()
480 );
481 }
482
483 # Information on limits, for the benefit of users who try to skirt them
484 if ( $this->mOptions->getEnableLimitReport() ) {
485 $text .= $this->makeLimitReport();
486 }
487
488 # Wrap non-interface parser output in a <div> so it can be targeted
489 # with CSS (T37247)
490 $class = $this->mOptions->getWrapOutputClass();
491 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
492 $text = Html::rawElement( 'div', [ 'class' => $class ], $text );
493 }
494
495 $this->mOutput->setText( $text );
496
497 $this->mRevisionId = $oldRevisionId;
498 $this->mRevisionObject = $oldRevisionObject;
499 $this->mRevisionTimestamp = $oldRevisionTimestamp;
500 $this->mRevisionUser = $oldRevisionUser;
501 $this->mRevisionSize = $oldRevisionSize;
502 $this->mInputSize = false;
503 $this->currentRevisionCache = null;
504
505 return $this->mOutput;
506 }
507
508 /**
509 * Set the limit report data in the current ParserOutput, and return the
510 * limit report HTML comment.
511 *
512 * @return string
513 */
514 protected function makeLimitReport() {
515 global $wgShowHostnames;
516
517 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
518
519 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
520 if ( $cpuTime !== null ) {
521 $this->mOutput->setLimitReportData( 'limitreport-cputime',
522 sprintf( "%.3f", $cpuTime )
523 );
524 }
525
526 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
527 $this->mOutput->setLimitReportData( 'limitreport-walltime',
528 sprintf( "%.3f", $wallTime )
529 );
530
531 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
532 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
533 );
534 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
535 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
536 );
537 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
538 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
539 );
540 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
541 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
542 );
543 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
544 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
545 );
546 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
547 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
548 );
549
550 foreach ( $this->mStripState->getLimitReport() as list( $key, $value ) ) {
551 $this->mOutput->setLimitReportData( $key, $value );
552 }
553
554 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
555
556 $limitReport = "NewPP limit report\n";
557 if ( $wgShowHostnames ) {
558 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
559 }
560 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
561 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
562 $limitReport .= 'Dynamic content: ' .
563 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
564 "\n";
565
566 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
567 if ( Hooks::run( 'ParserLimitReportFormat',
568 [ $key, &$value, &$limitReport, false, false ]
569 ) ) {
570 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
571 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
572 ->inLanguage( 'en' )->useDatabase( false );
573 if ( !$valueMsg->exists() ) {
574 $valueMsg = new RawMessage( '$1' );
575 }
576 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
577 $valueMsg->params( $value );
578 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
579 }
580 }
581 }
582 // Since we're not really outputting HTML, decode the entities and
583 // then re-encode the things that need hiding inside HTML comments.
584 $limitReport = htmlspecialchars_decode( $limitReport );
585 // Run deprecated hook
586 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
587
588 // Sanitize for comment. Note '‐' in the replacement is U+2010,
589 // which looks much like the problematic '-'.
590 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
591 $text = "\n<!-- \n$limitReport-->\n";
592
593 // Add on template profiling data in human/machine readable way
594 $dataByFunc = $this->mProfiler->getFunctionStats();
595 uasort( $dataByFunc, function ( $a, $b ) {
596 return $b['real'] <=> $a['real']; // descending order
597 } );
598 $profileReport = [];
599 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
600 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
601 $item['%real'], $item['real'], $item['calls'],
602 htmlspecialchars( $item['name'] ) );
603 }
604 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
605 $text .= implode( "\n", $profileReport ) . "\n-->\n";
606
607 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
608
609 // Add other cache related metadata
610 if ( $wgShowHostnames ) {
611 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
612 }
613 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
614 $this->mOutput->getCacheTime() );
615 $this->mOutput->setLimitReportData( 'cachereport-ttl',
616 $this->mOutput->getCacheExpiry() );
617 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
618 $this->mOutput->hasDynamicContent() );
619
620 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
621 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
622 $this->mTitle->getPrefixedDBkey() );
623 }
624 return $text;
625 }
626
627 /**
628 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
629 * can be called from an extension tag hook.
630 *
631 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
632 * instead, which means that lists and links have not been fully parsed yet,
633 * and strip markers are still present.
634 *
635 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
636 *
637 * Use this function if you're a parser tag hook and you want to parse
638 * wikitext before or after applying additional transformations, and you
639 * intend to *return the result as hook output*, which will cause it to go
640 * through the rest of parsing process automatically.
641 *
642 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
643 * $text are not expanded
644 *
645 * @param string $text Text extension wants to have parsed
646 * @param bool|PPFrame $frame The frame to use for expanding any template variables
647 * @return string UNSAFE half-parsed HTML
648 */
649 public function recursiveTagParse( $text, $frame = false ) {
650 // Avoid PHP 7.1 warning from passing $this by reference
651 $parser = $this;
652 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
653 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
654 $text = $this->internalParse( $text, false, $frame );
655 return $text;
656 }
657
658 /**
659 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
660 * point can be called from an extension tag hook.
661 *
662 * The output of this function is fully-parsed HTML that is safe for output.
663 * If you're a parser tag hook, you might want to use recursiveTagParse()
664 * instead.
665 *
666 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
667 * $text are not expanded
668 *
669 * @since 1.25
670 *
671 * @param string $text Text extension wants to have parsed
672 * @param bool|PPFrame $frame The frame to use for expanding any template variables
673 * @return string Fully parsed HTML
674 */
675 public function recursiveTagParseFully( $text, $frame = false ) {
676 $text = $this->recursiveTagParse( $text, $frame );
677 $text = $this->internalParseHalfParsed( $text, false );
678 return $text;
679 }
680
681 /**
682 * Expand templates and variables in the text, producing valid, static wikitext.
683 * Also removes comments.
684 * Do not call this function recursively.
685 * @param string $text
686 * @param Title|null $title
687 * @param ParserOptions $options
688 * @param int|null $revid
689 * @param bool|PPFrame $frame
690 * @return mixed|string
691 */
692 public function preprocess( $text, Title $title = null,
693 ParserOptions $options, $revid = null, $frame = false
694 ) {
695 $magicScopeVariable = $this->lock();
696 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
697 if ( $revid !== null ) {
698 $this->mRevisionId = $revid;
699 }
700 // Avoid PHP 7.1 warning from passing $this by reference
701 $parser = $this;
702 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
703 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
704 $text = $this->replaceVariables( $text, $frame );
705 $text = $this->mStripState->unstripBoth( $text );
706 return $text;
707 }
708
709 /**
710 * Recursive parser entry point that can be called from an extension tag
711 * hook.
712 *
713 * @param string $text Text to be expanded
714 * @param bool|PPFrame $frame The frame to use for expanding any template variables
715 * @return string
716 * @since 1.19
717 */
718 public function recursivePreprocess( $text, $frame = false ) {
719 $text = $this->replaceVariables( $text, $frame );
720 $text = $this->mStripState->unstripBoth( $text );
721 return $text;
722 }
723
724 /**
725 * Process the wikitext for the "?preload=" feature. (T7210)
726 *
727 * "<noinclude>", "<includeonly>" etc. are parsed as for template
728 * transclusion, comments, templates, arguments, tags hooks and parser
729 * functions are untouched.
730 *
731 * @param string $text
732 * @param Title $title
733 * @param ParserOptions $options
734 * @param array $params
735 * @return string
736 */
737 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
738 $msg = new RawMessage( $text );
739 $text = $msg->params( $params )->plain();
740
741 # Parser (re)initialisation
742 $magicScopeVariable = $this->lock();
743 $this->startParse( $title, $options, self::OT_PLAIN, true );
744
745 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
746 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
747 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
748 $text = $this->mStripState->unstripBoth( $text );
749 return $text;
750 }
751
752 /**
753 * Set the current user.
754 * Should only be used when doing pre-save transform.
755 *
756 * @param User|null $user User object or null (to reset)
757 */
758 public function setUser( $user ) {
759 $this->mUser = $user;
760 }
761
762 /**
763 * Set the context title
764 *
765 * @param Title $t
766 */
767 public function setTitle( $t ) {
768 if ( !$t ) {
769 $t = Title::newFromText( 'NO TITLE' );
770 }
771
772 if ( $t->hasFragment() ) {
773 # Strip the fragment to avoid various odd effects
774 $this->mTitle = $t->createFragmentTarget( '' );
775 } else {
776 $this->mTitle = $t;
777 }
778 }
779
780 /**
781 * Accessor for the Title object
782 *
783 * @return Title
784 */
785 public function getTitle() {
786 return $this->mTitle;
787 }
788
789 /**
790 * Accessor/mutator for the Title object
791 *
792 * @param Title|null $x Title object or null to just get the current one
793 * @return Title
794 */
795 public function Title( $x = null ) {
796 return wfSetVar( $this->mTitle, $x );
797 }
798
799 /**
800 * Set the output type
801 *
802 * @param int $ot New value
803 */
804 public function setOutputType( $ot ) {
805 $this->mOutputType = $ot;
806 # Shortcut alias
807 $this->ot = [
808 'html' => $ot == self::OT_HTML,
809 'wiki' => $ot == self::OT_WIKI,
810 'pre' => $ot == self::OT_PREPROCESS,
811 'plain' => $ot == self::OT_PLAIN,
812 ];
813 }
814
815 /**
816 * Accessor/mutator for the output type
817 *
818 * @param int|null $x New value or null to just get the current one
819 * @return int
820 */
821 public function OutputType( $x = null ) {
822 return wfSetVar( $this->mOutputType, $x );
823 }
824
825 /**
826 * Get the ParserOutput object
827 *
828 * @return ParserOutput
829 */
830 public function getOutput() {
831 return $this->mOutput;
832 }
833
834 /**
835 * Get the ParserOptions object
836 *
837 * @return ParserOptions
838 */
839 public function getOptions() {
840 return $this->mOptions;
841 }
842
843 /**
844 * Accessor/mutator for the ParserOptions object
845 *
846 * @param ParserOptions|null $x New value or null to just get the current one
847 * @return ParserOptions Current ParserOptions object
848 */
849 public function Options( $x = null ) {
850 return wfSetVar( $this->mOptions, $x );
851 }
852
853 /**
854 * @return int
855 */
856 public function nextLinkID() {
857 return $this->mLinkID++;
858 }
859
860 /**
861 * @param int $id
862 */
863 public function setLinkID( $id ) {
864 $this->mLinkID = $id;
865 }
866
867 /**
868 * Get a language object for use in parser functions such as {{FORMATNUM:}}
869 * @return Language
870 */
871 public function getFunctionLang() {
872 return $this->getTargetLanguage();
873 }
874
875 /**
876 * Get the target language for the content being parsed. This is usually the
877 * language that the content is in.
878 *
879 * @since 1.19
880 *
881 * @throws MWException
882 * @return Language
883 */
884 public function getTargetLanguage() {
885 $target = $this->mOptions->getTargetLanguage();
886
887 if ( $target !== null ) {
888 return $target;
889 } elseif ( $this->mOptions->getInterfaceMessage() ) {
890 return $this->mOptions->getUserLangObj();
891 } elseif ( is_null( $this->mTitle ) ) {
892 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
893 }
894
895 return $this->mTitle->getPageLanguage();
896 }
897
898 /**
899 * Get the language object for language conversion
900 * @deprecated since 1.32, just use getTargetLanguage()
901 * @return Language|null
902 */
903 public function getConverterLanguage() {
904 return $this->getTargetLanguage();
905 }
906
907 /**
908 * Get a User object either from $this->mUser, if set, or from the
909 * ParserOptions object otherwise
910 *
911 * @return User
912 */
913 public function getUser() {
914 if ( !is_null( $this->mUser ) ) {
915 return $this->mUser;
916 }
917 return $this->mOptions->getUser();
918 }
919
920 /**
921 * Get a preprocessor object
922 *
923 * @return Preprocessor
924 */
925 public function getPreprocessor() {
926 if ( !isset( $this->mPreprocessor ) ) {
927 $class = $this->mPreprocessorClass;
928 $this->mPreprocessor = new $class( $this );
929 }
930 return $this->mPreprocessor;
931 }
932
933 /**
934 * Get a LinkRenderer instance to make links with
935 *
936 * @since 1.28
937 * @return LinkRenderer
938 */
939 public function getLinkRenderer() {
940 if ( !$this->mLinkRenderer ) {
941 $this->mLinkRenderer = MediaWikiServices::getInstance()
942 ->getLinkRendererFactory()->create();
943 $this->mLinkRenderer->setStubThreshold(
944 $this->getOptions()->getStubThreshold()
945 );
946 }
947
948 return $this->mLinkRenderer;
949 }
950
951 /**
952 * Replaces all occurrences of HTML-style comments and the given tags
953 * in the text with a random marker and returns the next text. The output
954 * parameter $matches will be an associative array filled with data in
955 * the form:
956 *
957 * @code
958 * 'UNIQ-xxxxx' => [
959 * 'element',
960 * 'tag content',
961 * [ 'param' => 'x' ],
962 * '<element param="x">tag content</element>' ]
963 * @endcode
964 *
965 * @param array $elements List of element names. Comments are always extracted.
966 * @param string $text Source text string.
967 * @param array &$matches Out parameter, Array: extracted tags
968 * @return string Stripped text
969 */
970 public static function extractTagsAndParams( $elements, $text, &$matches ) {
971 static $n = 1;
972 $stripped = '';
973 $matches = [];
974
975 $taglist = implode( '|', $elements );
976 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
977
978 while ( $text != '' ) {
979 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
980 $stripped .= $p[0];
981 if ( count( $p ) < 5 ) {
982 break;
983 }
984 if ( count( $p ) > 5 ) {
985 # comment
986 $element = $p[4];
987 $attributes = '';
988 $close = '';
989 $inside = $p[5];
990 } else {
991 # tag
992 $element = $p[1];
993 $attributes = $p[2];
994 $close = $p[3];
995 $inside = $p[4];
996 }
997
998 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
999 $stripped .= $marker;
1000
1001 if ( $close === '/>' ) {
1002 # Empty element tag, <tag />
1003 $content = null;
1004 $text = $inside;
1005 $tail = null;
1006 } else {
1007 if ( $element === '!--' ) {
1008 $end = '/(-->)/';
1009 } else {
1010 $end = "/(<\\/$element\\s*>)/i";
1011 }
1012 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1013 $content = $q[0];
1014 if ( count( $q ) < 3 ) {
1015 # No end tag -- let it run out to the end of the text.
1016 $tail = '';
1017 $text = '';
1018 } else {
1019 $tail = $q[1];
1020 $text = $q[2];
1021 }
1022 }
1023
1024 $matches[$marker] = [ $element,
1025 $content,
1026 Sanitizer::decodeTagAttributes( $attributes ),
1027 "<$element$attributes$close$content$tail" ];
1028 }
1029 return $stripped;
1030 }
1031
1032 /**
1033 * Get a list of strippable XML-like elements
1034 *
1035 * @return array
1036 */
1037 public function getStripList() {
1038 return $this->mStripList;
1039 }
1040
1041 /**
1042 * Add an item to the strip state
1043 * Returns the unique tag which must be inserted into the stripped text
1044 * The tag will be replaced with the original text in unstrip()
1045 *
1046 * @param string $text
1047 *
1048 * @return string
1049 */
1050 public function insertStripItem( $text ) {
1051 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1052 $this->mMarkerIndex++;
1053 $this->mStripState->addGeneral( $marker, $text );
1054 return $marker;
1055 }
1056
1057 /**
1058 * parse the wiki syntax used to render tables
1059 *
1060 * @private
1061 * @param string $text
1062 * @return string
1063 */
1064 public function doTableStuff( $text ) {
1065 $lines = StringUtils::explode( "\n", $text );
1066 $out = '';
1067 $td_history = []; # Is currently a td tag open?
1068 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1069 $tr_history = []; # Is currently a tr tag open?
1070 $tr_attributes = []; # history of tr attributes
1071 $has_opened_tr = []; # Did this table open a <tr> element?
1072 $indent_level = 0; # indent level of the table
1073
1074 foreach ( $lines as $outLine ) {
1075 $line = trim( $outLine );
1076
1077 if ( $line === '' ) { # empty line, go to next line
1078 $out .= $outLine . "\n";
1079 continue;
1080 }
1081
1082 $first_character = $line[0];
1083 $first_two = substr( $line, 0, 2 );
1084 $matches = [];
1085
1086 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1087 # First check if we are starting a new table
1088 $indent_level = strlen( $matches[1] );
1089
1090 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1091 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1092
1093 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1094 array_push( $td_history, false );
1095 array_push( $last_tag_history, '' );
1096 array_push( $tr_history, false );
1097 array_push( $tr_attributes, '' );
1098 array_push( $has_opened_tr, false );
1099 } elseif ( count( $td_history ) == 0 ) {
1100 # Don't do any of the following
1101 $out .= $outLine . "\n";
1102 continue;
1103 } elseif ( $first_two === '|}' ) {
1104 # We are ending a table
1105 $line = '</table>' . substr( $line, 2 );
1106 $last_tag = array_pop( $last_tag_history );
1107
1108 if ( !array_pop( $has_opened_tr ) ) {
1109 $line = "<tr><td></td></tr>{$line}";
1110 }
1111
1112 if ( array_pop( $tr_history ) ) {
1113 $line = "</tr>{$line}";
1114 }
1115
1116 if ( array_pop( $td_history ) ) {
1117 $line = "</{$last_tag}>{$line}";
1118 }
1119 array_pop( $tr_attributes );
1120 if ( $indent_level > 0 ) {
1121 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1122 } else {
1123 $outLine = $line;
1124 }
1125 } elseif ( $first_two === '|-' ) {
1126 # Now we have a table row
1127 $line = preg_replace( '#^\|-+#', '', $line );
1128
1129 # Whats after the tag is now only attributes
1130 $attributes = $this->mStripState->unstripBoth( $line );
1131 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1132 array_pop( $tr_attributes );
1133 array_push( $tr_attributes, $attributes );
1134
1135 $line = '';
1136 $last_tag = array_pop( $last_tag_history );
1137 array_pop( $has_opened_tr );
1138 array_push( $has_opened_tr, true );
1139
1140 if ( array_pop( $tr_history ) ) {
1141 $line = '</tr>';
1142 }
1143
1144 if ( array_pop( $td_history ) ) {
1145 $line = "</{$last_tag}>{$line}";
1146 }
1147
1148 $outLine = $line;
1149 array_push( $tr_history, false );
1150 array_push( $td_history, false );
1151 array_push( $last_tag_history, '' );
1152 } elseif ( $first_character === '|'
1153 || $first_character === '!'
1154 || $first_two === '|+'
1155 ) {
1156 # This might be cell elements, td, th or captions
1157 if ( $first_two === '|+' ) {
1158 $first_character = '+';
1159 $line = substr( $line, 2 );
1160 } else {
1161 $line = substr( $line, 1 );
1162 }
1163
1164 // Implies both are valid for table headings.
1165 if ( $first_character === '!' ) {
1166 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1167 }
1168
1169 # Split up multiple cells on the same line.
1170 # FIXME : This can result in improper nesting of tags processed
1171 # by earlier parser steps.
1172 $cells = explode( '||', $line );
1173
1174 $outLine = '';
1175
1176 # Loop through each table cell
1177 foreach ( $cells as $cell ) {
1178 $previous = '';
1179 if ( $first_character !== '+' ) {
1180 $tr_after = array_pop( $tr_attributes );
1181 if ( !array_pop( $tr_history ) ) {
1182 $previous = "<tr{$tr_after}>\n";
1183 }
1184 array_push( $tr_history, true );
1185 array_push( $tr_attributes, '' );
1186 array_pop( $has_opened_tr );
1187 array_push( $has_opened_tr, true );
1188 }
1189
1190 $last_tag = array_pop( $last_tag_history );
1191
1192 if ( array_pop( $td_history ) ) {
1193 $previous = "</{$last_tag}>\n{$previous}";
1194 }
1195
1196 if ( $first_character === '|' ) {
1197 $last_tag = 'td';
1198 } elseif ( $first_character === '!' ) {
1199 $last_tag = 'th';
1200 } elseif ( $first_character === '+' ) {
1201 $last_tag = 'caption';
1202 } else {
1203 $last_tag = '';
1204 }
1205
1206 array_push( $last_tag_history, $last_tag );
1207
1208 # A cell could contain both parameters and data
1209 $cell_data = explode( '|', $cell, 2 );
1210
1211 # T2553: Note that a '|' inside an invalid link should not
1212 # be mistaken as delimiting cell parameters
1213 # Bug T153140: Neither should language converter markup.
1214 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1215 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1216 } elseif ( count( $cell_data ) == 1 ) {
1217 // Whitespace in cells is trimmed
1218 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1219 } else {
1220 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1221 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1222 // Whitespace in cells is trimmed
1223 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1224 }
1225
1226 $outLine .= $cell;
1227 array_push( $td_history, true );
1228 }
1229 }
1230 $out .= $outLine . "\n";
1231 }
1232
1233 # Closing open td, tr && table
1234 while ( count( $td_history ) > 0 ) {
1235 if ( array_pop( $td_history ) ) {
1236 $out .= "</td>\n";
1237 }
1238 if ( array_pop( $tr_history ) ) {
1239 $out .= "</tr>\n";
1240 }
1241 if ( !array_pop( $has_opened_tr ) ) {
1242 $out .= "<tr><td></td></tr>\n";
1243 }
1244
1245 $out .= "</table>\n";
1246 }
1247
1248 # Remove trailing line-ending (b/c)
1249 if ( substr( $out, -1 ) === "\n" ) {
1250 $out = substr( $out, 0, -1 );
1251 }
1252
1253 # special case: don't return empty table
1254 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1255 $out = '';
1256 }
1257
1258 return $out;
1259 }
1260
1261 /**
1262 * Helper function for parse() that transforms wiki markup into half-parsed
1263 * HTML. Only called for $mOutputType == self::OT_HTML.
1264 *
1265 * @private
1266 *
1267 * @param string $text The text to parse
1268 * @param bool $isMain Whether this is being called from the main parse() function
1269 * @param PPFrame|bool $frame A pre-processor frame
1270 *
1271 * @return string
1272 */
1273 public function internalParse( $text, $isMain = true, $frame = false ) {
1274 $origText = $text;
1275
1276 // Avoid PHP 7.1 warning from passing $this by reference
1277 $parser = $this;
1278
1279 # Hook to suspend the parser in this state
1280 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1281 return $text;
1282 }
1283
1284 # if $frame is provided, then use $frame for replacing any variables
1285 if ( $frame ) {
1286 # use frame depth to infer how include/noinclude tags should be handled
1287 # depth=0 means this is the top-level document; otherwise it's an included document
1288 if ( !$frame->depth ) {
1289 $flag = 0;
1290 } else {
1291 $flag = self::PTD_FOR_INCLUSION;
1292 }
1293 $dom = $this->preprocessToDom( $text, $flag );
1294 $text = $frame->expand( $dom );
1295 } else {
1296 # if $frame is not provided, then use old-style replaceVariables
1297 $text = $this->replaceVariables( $text );
1298 }
1299
1300 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1301 $text = Sanitizer::removeHTMLtags(
1302 $text,
1303 [ $this, 'attributeStripCallback' ],
1304 false,
1305 array_keys( $this->mTransparentTagHooks ),
1306 [],
1307 [ $this, 'addTrackingCategory' ]
1308 );
1309 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1310
1311 # Tables need to come after variable replacement for things to work
1312 # properly; putting them before other transformations should keep
1313 # exciting things like link expansions from showing up in surprising
1314 # places.
1315 $text = $this->doTableStuff( $text );
1316
1317 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1318
1319 $text = $this->doDoubleUnderscore( $text );
1320
1321 $text = $this->doHeadings( $text );
1322 $text = $this->replaceInternalLinks( $text );
1323 $text = $this->doAllQuotes( $text );
1324 $text = $this->replaceExternalLinks( $text );
1325
1326 # replaceInternalLinks may sometimes leave behind
1327 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1328 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1329
1330 $text = $this->doMagicLinks( $text );
1331 $text = $this->formatHeadings( $text, $origText, $isMain );
1332
1333 return $text;
1334 }
1335
1336 /**
1337 * Helper function for parse() that transforms half-parsed HTML into fully
1338 * parsed HTML.
1339 *
1340 * @param string $text
1341 * @param bool $isMain
1342 * @param bool $linestart
1343 * @return string
1344 */
1345 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1346 $text = $this->mStripState->unstripGeneral( $text );
1347
1348 // Avoid PHP 7.1 warning from passing $this by reference
1349 $parser = $this;
1350
1351 if ( $isMain ) {
1352 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1353 }
1354
1355 # Clean up special characters, only run once, next-to-last before doBlockLevels
1356 $text = Sanitizer::armorFrenchSpaces( $text );
1357
1358 $text = $this->doBlockLevels( $text, $linestart );
1359
1360 $this->replaceLinkHolders( $text );
1361
1362 /**
1363 * The input doesn't get language converted if
1364 * a) It's disabled
1365 * b) Content isn't converted
1366 * c) It's a conversion table
1367 * d) it is an interface message (which is in the user language)
1368 */
1369 if ( !( $this->mOptions->getDisableContentConversion()
1370 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1371 ) {
1372 if ( !$this->mOptions->getInterfaceMessage() ) {
1373 # The position of the convert() call should not be changed. it
1374 # assumes that the links are all replaced and the only thing left
1375 # is the <nowiki> mark.
1376 $text = $this->getTargetLanguage()->convert( $text );
1377 }
1378 }
1379
1380 $text = $this->mStripState->unstripNoWiki( $text );
1381
1382 if ( $isMain ) {
1383 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1384 }
1385
1386 $text = $this->replaceTransparentTags( $text );
1387 $text = $this->mStripState->unstripGeneral( $text );
1388
1389 $text = Sanitizer::normalizeCharReferences( $text );
1390
1391 if ( MWTidy::isEnabled() ) {
1392 if ( $this->mOptions->getTidy() ) {
1393 $text = MWTidy::tidy( $text );
1394 }
1395 } else {
1396 # attempt to sanitize at least some nesting problems
1397 # (T4702 and quite a few others)
1398 $tidyregs = [
1399 # ''Something [http://www.cool.com cool''] -->
1400 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1401 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1402 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1403 # fix up an anchor inside another anchor, only
1404 # at least for a single single nested link (T5695)
1405 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1406 '\\1\\2</a>\\3</a>\\1\\4</a>',
1407 # fix div inside inline elements- doBlockLevels won't wrap a line which
1408 # contains a div, so fix it up here; replace
1409 # div with escaped text
1410 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1411 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1412 # remove empty italic or bold tag pairs, some
1413 # introduced by rules above
1414 '/<([bi])><\/\\1>/' => '',
1415 ];
1416
1417 $text = preg_replace(
1418 array_keys( $tidyregs ),
1419 array_values( $tidyregs ),
1420 $text );
1421 }
1422
1423 if ( $isMain ) {
1424 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1425 }
1426
1427 return $text;
1428 }
1429
1430 /**
1431 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1432 * magic external links.
1433 *
1434 * DML
1435 * @private
1436 *
1437 * @param string $text
1438 *
1439 * @return string
1440 */
1441 public function doMagicLinks( $text ) {
1442 $prots = wfUrlProtocolsWithoutProtRel();
1443 $urlChar = self::EXT_LINK_URL_CLASS;
1444 $addr = self::EXT_LINK_ADDR;
1445 $space = self::SPACE_NOT_NL; # non-newline space
1446 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1447 $spaces = "$space++"; # possessive match of 1 or more spaces
1448 $text = preg_replace_callback(
1449 '!(?: # Start cases
1450 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1451 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1452 (\b # m[3]: Free external links
1453 (?i:$prots)
1454 ($addr$urlChar*) # m[4]: Post-protocol path
1455 ) |
1456 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1457 ([0-9]+)\b |
1458 \bISBN $spaces ( # m[6]: ISBN, capture number
1459 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1460 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1461 [0-9Xx] # check digit
1462 )\b
1463 )!xu", [ $this, 'magicLinkCallback' ], $text );
1464 return $text;
1465 }
1466
1467 /**
1468 * @throws MWException
1469 * @param array $m
1470 * @return string HTML
1471 */
1472 public function magicLinkCallback( $m ) {
1473 if ( isset( $m[1] ) && $m[1] !== '' ) {
1474 # Skip anchor
1475 return $m[0];
1476 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1477 # Skip HTML element
1478 return $m[0];
1479 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1480 # Free external link
1481 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1482 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1483 # RFC or PMID
1484 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1485 if ( !$this->mOptions->getMagicRFCLinks() ) {
1486 return $m[0];
1487 }
1488 $keyword = 'RFC';
1489 $urlmsg = 'rfcurl';
1490 $cssClass = 'mw-magiclink-rfc';
1491 $trackingCat = 'magiclink-tracking-rfc';
1492 $id = $m[5];
1493 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1494 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1495 return $m[0];
1496 }
1497 $keyword = 'PMID';
1498 $urlmsg = 'pubmedurl';
1499 $cssClass = 'mw-magiclink-pmid';
1500 $trackingCat = 'magiclink-tracking-pmid';
1501 $id = $m[5];
1502 } else {
1503 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1504 substr( $m[0], 0, 20 ) . '"' );
1505 }
1506 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1507 $this->addTrackingCategory( $trackingCat );
1508 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1509 } elseif ( isset( $m[6] ) && $m[6] !== ''
1510 && $this->mOptions->getMagicISBNLinks()
1511 ) {
1512 # ISBN
1513 $isbn = $m[6];
1514 $space = self::SPACE_NOT_NL; # non-newline space
1515 $isbn = preg_replace( "/$space/", ' ', $isbn );
1516 $num = strtr( $isbn, [
1517 '-' => '',
1518 ' ' => '',
1519 'x' => 'X',
1520 ] );
1521 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1522 return $this->getLinkRenderer()->makeKnownLink(
1523 SpecialPage::getTitleFor( 'Booksources', $num ),
1524 "ISBN $isbn",
1525 [
1526 'class' => 'internal mw-magiclink-isbn',
1527 'title' => false // suppress title attribute
1528 ]
1529 );
1530 } else {
1531 return $m[0];
1532 }
1533 }
1534
1535 /**
1536 * Make a free external link, given a user-supplied URL
1537 *
1538 * @param string $url
1539 * @param int $numPostProto
1540 * The number of characters after the protocol.
1541 * @return string HTML
1542 * @private
1543 */
1544 public function makeFreeExternalLink( $url, $numPostProto ) {
1545 $trail = '';
1546
1547 # The characters '<' and '>' (which were escaped by
1548 # removeHTMLtags()) should not be included in
1549 # URLs, per RFC 2396.
1550 # Make &nbsp; terminate a URL as well (bug T84937)
1551 $m2 = [];
1552 if ( preg_match(
1553 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1554 $url,
1555 $m2,
1556 PREG_OFFSET_CAPTURE
1557 ) ) {
1558 $trail = substr( $url, $m2[0][1] ) . $trail;
1559 $url = substr( $url, 0, $m2[0][1] );
1560 }
1561
1562 # Move trailing punctuation to $trail
1563 $sep = ',;\.:!?';
1564 # If there is no left bracket, then consider right brackets fair game too
1565 if ( strpos( $url, '(' ) === false ) {
1566 $sep .= ')';
1567 }
1568
1569 $urlRev = strrev( $url );
1570 $numSepChars = strspn( $urlRev, $sep );
1571 # Don't break a trailing HTML entity by moving the ; into $trail
1572 # This is in hot code, so use substr_compare to avoid having to
1573 # create a new string object for the comparison
1574 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1575 # more optimization: instead of running preg_match with a $
1576 # anchor, which can be slow, do the match on the reversed
1577 # string starting at the desired offset.
1578 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1579 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1580 $numSepChars--;
1581 }
1582 }
1583 if ( $numSepChars ) {
1584 $trail = substr( $url, -$numSepChars ) . $trail;
1585 $url = substr( $url, 0, -$numSepChars );
1586 }
1587
1588 # Verify that we still have a real URL after trail removal, and
1589 # not just lone protocol
1590 if ( strlen( $trail ) >= $numPostProto ) {
1591 return $url . $trail;
1592 }
1593
1594 $url = Sanitizer::cleanUrl( $url );
1595
1596 # Is this an external image?
1597 $text = $this->maybeMakeExternalImage( $url );
1598 if ( $text === false ) {
1599 # Not an image, make a link
1600 $text = Linker::makeExternalLink( $url,
1601 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1602 true, 'free',
1603 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1604 # Register it in the output object...
1605 $this->mOutput->addExternalLink( $url );
1606 }
1607 return $text . $trail;
1608 }
1609
1610 /**
1611 * Parse headers and return html
1612 *
1613 * @private
1614 *
1615 * @param string $text
1616 *
1617 * @return string
1618 */
1619 public function doHeadings( $text ) {
1620 for ( $i = 6; $i >= 1; --$i ) {
1621 $h = str_repeat( '=', $i );
1622 // Trim non-newline whitespace from headings
1623 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1624 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1625 }
1626 return $text;
1627 }
1628
1629 /**
1630 * Replace single quotes with HTML markup
1631 * @private
1632 *
1633 * @param string $text
1634 *
1635 * @return string The altered text
1636 */
1637 public function doAllQuotes( $text ) {
1638 $outtext = '';
1639 $lines = StringUtils::explode( "\n", $text );
1640 foreach ( $lines as $line ) {
1641 $outtext .= $this->doQuotes( $line ) . "\n";
1642 }
1643 $outtext = substr( $outtext, 0, -1 );
1644 return $outtext;
1645 }
1646
1647 /**
1648 * Helper function for doAllQuotes()
1649 *
1650 * @param string $text
1651 *
1652 * @return string
1653 */
1654 public function doQuotes( $text ) {
1655 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1656 $countarr = count( $arr );
1657 if ( $countarr == 1 ) {
1658 return $text;
1659 }
1660
1661 // First, do some preliminary work. This may shift some apostrophes from
1662 // being mark-up to being text. It also counts the number of occurrences
1663 // of bold and italics mark-ups.
1664 $numbold = 0;
1665 $numitalics = 0;
1666 for ( $i = 1; $i < $countarr; $i += 2 ) {
1667 $thislen = strlen( $arr[$i] );
1668 // If there are ever four apostrophes, assume the first is supposed to
1669 // be text, and the remaining three constitute mark-up for bold text.
1670 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1671 if ( $thislen == 4 ) {
1672 $arr[$i - 1] .= "'";
1673 $arr[$i] = "'''";
1674 $thislen = 3;
1675 } elseif ( $thislen > 5 ) {
1676 // If there are more than 5 apostrophes in a row, assume they're all
1677 // text except for the last 5.
1678 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1679 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1680 $arr[$i] = "'''''";
1681 $thislen = 5;
1682 }
1683 // Count the number of occurrences of bold and italics mark-ups.
1684 if ( $thislen == 2 ) {
1685 $numitalics++;
1686 } elseif ( $thislen == 3 ) {
1687 $numbold++;
1688 } elseif ( $thislen == 5 ) {
1689 $numitalics++;
1690 $numbold++;
1691 }
1692 }
1693
1694 // If there is an odd number of both bold and italics, it is likely
1695 // that one of the bold ones was meant to be an apostrophe followed
1696 // by italics. Which one we cannot know for certain, but it is more
1697 // likely to be one that has a single-letter word before it.
1698 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1699 $firstsingleletterword = -1;
1700 $firstmultiletterword = -1;
1701 $firstspace = -1;
1702 for ( $i = 1; $i < $countarr; $i += 2 ) {
1703 if ( strlen( $arr[$i] ) == 3 ) {
1704 $x1 = substr( $arr[$i - 1], -1 );
1705 $x2 = substr( $arr[$i - 1], -2, 1 );
1706 if ( $x1 === ' ' ) {
1707 if ( $firstspace == -1 ) {
1708 $firstspace = $i;
1709 }
1710 } elseif ( $x2 === ' ' ) {
1711 $firstsingleletterword = $i;
1712 // if $firstsingleletterword is set, we don't
1713 // look at the other options, so we can bail early.
1714 break;
1715 } else {
1716 if ( $firstmultiletterword == -1 ) {
1717 $firstmultiletterword = $i;
1718 }
1719 }
1720 }
1721 }
1722
1723 // If there is a single-letter word, use it!
1724 if ( $firstsingleletterword > -1 ) {
1725 $arr[$firstsingleletterword] = "''";
1726 $arr[$firstsingleletterword - 1] .= "'";
1727 } elseif ( $firstmultiletterword > -1 ) {
1728 // If not, but there's a multi-letter word, use that one.
1729 $arr[$firstmultiletterword] = "''";
1730 $arr[$firstmultiletterword - 1] .= "'";
1731 } elseif ( $firstspace > -1 ) {
1732 // ... otherwise use the first one that has neither.
1733 // (notice that it is possible for all three to be -1 if, for example,
1734 // there is only one pentuple-apostrophe in the line)
1735 $arr[$firstspace] = "''";
1736 $arr[$firstspace - 1] .= "'";
1737 }
1738 }
1739
1740 // Now let's actually convert our apostrophic mush to HTML!
1741 $output = '';
1742 $buffer = '';
1743 $state = '';
1744 $i = 0;
1745 foreach ( $arr as $r ) {
1746 if ( ( $i % 2 ) == 0 ) {
1747 if ( $state === 'both' ) {
1748 $buffer .= $r;
1749 } else {
1750 $output .= $r;
1751 }
1752 } else {
1753 $thislen = strlen( $r );
1754 if ( $thislen == 2 ) {
1755 if ( $state === 'i' ) {
1756 $output .= '</i>';
1757 $state = '';
1758 } elseif ( $state === 'bi' ) {
1759 $output .= '</i>';
1760 $state = 'b';
1761 } elseif ( $state === 'ib' ) {
1762 $output .= '</b></i><b>';
1763 $state = 'b';
1764 } elseif ( $state === 'both' ) {
1765 $output .= '<b><i>' . $buffer . '</i>';
1766 $state = 'b';
1767 } else { // $state can be 'b' or ''
1768 $output .= '<i>';
1769 $state .= 'i';
1770 }
1771 } elseif ( $thislen == 3 ) {
1772 if ( $state === 'b' ) {
1773 $output .= '</b>';
1774 $state = '';
1775 } elseif ( $state === 'bi' ) {
1776 $output .= '</i></b><i>';
1777 $state = 'i';
1778 } elseif ( $state === 'ib' ) {
1779 $output .= '</b>';
1780 $state = 'i';
1781 } elseif ( $state === 'both' ) {
1782 $output .= '<i><b>' . $buffer . '</b>';
1783 $state = 'i';
1784 } else { // $state can be 'i' or ''
1785 $output .= '<b>';
1786 $state .= 'b';
1787 }
1788 } elseif ( $thislen == 5 ) {
1789 if ( $state === 'b' ) {
1790 $output .= '</b><i>';
1791 $state = 'i';
1792 } elseif ( $state === 'i' ) {
1793 $output .= '</i><b>';
1794 $state = 'b';
1795 } elseif ( $state === 'bi' ) {
1796 $output .= '</i></b>';
1797 $state = '';
1798 } elseif ( $state === 'ib' ) {
1799 $output .= '</b></i>';
1800 $state = '';
1801 } elseif ( $state === 'both' ) {
1802 $output .= '<i><b>' . $buffer . '</b></i>';
1803 $state = '';
1804 } else { // ($state == '')
1805 $buffer = '';
1806 $state = 'both';
1807 }
1808 }
1809 }
1810 $i++;
1811 }
1812 // Now close all remaining tags. Notice that the order is important.
1813 if ( $state === 'b' || $state === 'ib' ) {
1814 $output .= '</b>';
1815 }
1816 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1817 $output .= '</i>';
1818 }
1819 if ( $state === 'bi' ) {
1820 $output .= '</b>';
1821 }
1822 // There might be lonely ''''', so make sure we have a buffer
1823 if ( $state === 'both' && $buffer ) {
1824 $output .= '<b><i>' . $buffer . '</i></b>';
1825 }
1826 return $output;
1827 }
1828
1829 /**
1830 * Replace external links (REL)
1831 *
1832 * Note: this is all very hackish and the order of execution matters a lot.
1833 * Make sure to run tests/parser/parserTests.php if you change this code.
1834 *
1835 * @private
1836 *
1837 * @param string $text
1838 *
1839 * @throws MWException
1840 * @return string
1841 */
1842 public function replaceExternalLinks( $text ) {
1843 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1844 if ( $bits === false ) {
1845 throw new MWException( "PCRE needs to be compiled with "
1846 . "--enable-unicode-properties in order for MediaWiki to function" );
1847 }
1848 $s = array_shift( $bits );
1849
1850 $i = 0;
1851 while ( $i < count( $bits ) ) {
1852 $url = $bits[$i++];
1853 $i++; // protocol
1854 $text = $bits[$i++];
1855 $trail = $bits[$i++];
1856
1857 # The characters '<' and '>' (which were escaped by
1858 # removeHTMLtags()) should not be included in
1859 # URLs, per RFC 2396.
1860 $m2 = [];
1861 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1862 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1863 $url = substr( $url, 0, $m2[0][1] );
1864 }
1865
1866 # If the link text is an image URL, replace it with an <img> tag
1867 # This happened by accident in the original parser, but some people used it extensively
1868 $img = $this->maybeMakeExternalImage( $text );
1869 if ( $img !== false ) {
1870 $text = $img;
1871 }
1872
1873 $dtrail = '';
1874
1875 # Set linktype for CSS
1876 $linktype = 'text';
1877
1878 # No link text, e.g. [http://domain.tld/some.link]
1879 if ( $text == '' ) {
1880 # Autonumber
1881 $langObj = $this->getTargetLanguage();
1882 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1883 $linktype = 'autonumber';
1884 } else {
1885 # Have link text, e.g. [http://domain.tld/some.link text]s
1886 # Check for trail
1887 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1888 }
1889
1890 // Excluding protocol-relative URLs may avoid many false positives.
1891 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
1892 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
1893 }
1894
1895 $url = Sanitizer::cleanUrl( $url );
1896
1897 # Use the encoded URL
1898 # This means that users can paste URLs directly into the text
1899 # Funny characters like ö aren't valid in URLs anyway
1900 # This was changed in August 2004
1901 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1902 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1903
1904 # Register link in the output object.
1905 $this->mOutput->addExternalLink( $url );
1906 }
1907
1908 return $s;
1909 }
1910
1911 /**
1912 * Get the rel attribute for a particular external link.
1913 *
1914 * @since 1.21
1915 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1916 * nofollow if appropriate
1917 * @param Title|null $title Optional Title, for wgNoFollowNsExceptions lookups
1918 * @return string|null Rel attribute for $url
1919 */
1920 public static function getExternalLinkRel( $url = false, $title = null ) {
1921 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1922 $ns = $title ? $title->getNamespace() : false;
1923 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1924 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1925 ) {
1926 return 'nofollow';
1927 }
1928 return null;
1929 }
1930
1931 /**
1932 * Get an associative array of additional HTML attributes appropriate for a
1933 * particular external link. This currently may include rel => nofollow
1934 * (depending on configuration, namespace, and the URL's domain) and/or a
1935 * target attribute (depending on configuration).
1936 *
1937 * @param string $url URL to extract the domain from for rel =>
1938 * nofollow if appropriate
1939 * @return array Associative array of HTML attributes
1940 */
1941 public function getExternalLinkAttribs( $url ) {
1942 $attribs = [];
1943 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1944
1945 $target = $this->mOptions->getExternalLinkTarget();
1946 if ( $target ) {
1947 $attribs['target'] = $target;
1948 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1949 // T133507. New windows can navigate parent cross-origin.
1950 // Including noreferrer due to lacking browser
1951 // support of noopener. Eventually noreferrer should be removed.
1952 if ( $rel !== '' ) {
1953 $rel .= ' ';
1954 }
1955 $rel .= 'noreferrer noopener';
1956 }
1957 }
1958 $attribs['rel'] = $rel;
1959 return $attribs;
1960 }
1961
1962 /**
1963 * Replace unusual escape codes in a URL with their equivalent characters
1964 *
1965 * This generally follows the syntax defined in RFC 3986, with special
1966 * consideration for HTTP query strings.
1967 *
1968 * @param string $url
1969 * @return string
1970 */
1971 public static function normalizeLinkUrl( $url ) {
1972 # First, make sure unsafe characters are encoded
1973 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1974 function ( $m ) {
1975 return rawurlencode( $m[0] );
1976 },
1977 $url
1978 );
1979
1980 $ret = '';
1981 $end = strlen( $url );
1982
1983 # Fragment part - 'fragment'
1984 $start = strpos( $url, '#' );
1985 if ( $start !== false && $start < $end ) {
1986 $ret = self::normalizeUrlComponent(
1987 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1988 $end = $start;
1989 }
1990
1991 # Query part - 'query' minus &=+;
1992 $start = strpos( $url, '?' );
1993 if ( $start !== false && $start < $end ) {
1994 $ret = self::normalizeUrlComponent(
1995 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1996 $end = $start;
1997 }
1998
1999 # Scheme and path part - 'pchar'
2000 # (we assume no userinfo or encoded colons in the host)
2001 $ret = self::normalizeUrlComponent(
2002 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2003
2004 return $ret;
2005 }
2006
2007 private static function normalizeUrlComponent( $component, $unsafe ) {
2008 $callback = function ( $matches ) use ( $unsafe ) {
2009 $char = urldecode( $matches[0] );
2010 $ord = ord( $char );
2011 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2012 # Unescape it
2013 return $char;
2014 } else {
2015 # Leave it escaped, but use uppercase for a-f
2016 return strtoupper( $matches[0] );
2017 }
2018 };
2019 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2020 }
2021
2022 /**
2023 * make an image if it's allowed, either through the global
2024 * option, through the exception, or through the on-wiki whitelist
2025 *
2026 * @param string $url
2027 *
2028 * @return string
2029 */
2030 private function maybeMakeExternalImage( $url ) {
2031 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2032 $imagesexception = !empty( $imagesfrom );
2033 $text = false;
2034 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2035 if ( $imagesexception && is_array( $imagesfrom ) ) {
2036 $imagematch = false;
2037 foreach ( $imagesfrom as $match ) {
2038 if ( strpos( $url, $match ) === 0 ) {
2039 $imagematch = true;
2040 break;
2041 }
2042 }
2043 } elseif ( $imagesexception ) {
2044 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2045 } else {
2046 $imagematch = false;
2047 }
2048
2049 if ( $this->mOptions->getAllowExternalImages()
2050 || ( $imagesexception && $imagematch )
2051 ) {
2052 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2053 # Image found
2054 $text = Linker::makeExternalImage( $url );
2055 }
2056 }
2057 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2058 && preg_match( self::EXT_IMAGE_REGEX, $url )
2059 ) {
2060 $whitelist = explode(
2061 "\n",
2062 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2063 );
2064
2065 foreach ( $whitelist as $entry ) {
2066 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2067 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2068 continue;
2069 }
2070 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2071 # Image matches a whitelist entry
2072 $text = Linker::makeExternalImage( $url );
2073 break;
2074 }
2075 }
2076 }
2077 return $text;
2078 }
2079
2080 /**
2081 * Process [[ ]] wikilinks
2082 *
2083 * @param string $s
2084 *
2085 * @return string Processed text
2086 *
2087 * @private
2088 */
2089 public function replaceInternalLinks( $s ) {
2090 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2091 return $s;
2092 }
2093
2094 /**
2095 * Process [[ ]] wikilinks (RIL)
2096 * @param string &$s
2097 * @throws MWException
2098 * @return LinkHolderArray
2099 *
2100 * @private
2101 */
2102 public function replaceInternalLinks2( &$s ) {
2103 global $wgExtraInterlanguageLinkPrefixes;
2104
2105 static $tc = false, $e1, $e1_img;
2106 # the % is needed to support urlencoded titles as well
2107 if ( !$tc ) {
2108 $tc = Title::legalChars() . '#%';
2109 # Match a link having the form [[namespace:link|alternate]]trail
2110 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2111 # Match cases where there is no "]]", which might still be images
2112 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2113 }
2114
2115 $holders = new LinkHolderArray( $this );
2116
2117 # split the entire text string on occurrences of [[
2118 $a = StringUtils::explode( '[[', ' ' . $s );
2119 # get the first element (all text up to first [[), and remove the space we added
2120 $s = $a->current();
2121 $a->next();
2122 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2123 $s = substr( $s, 1 );
2124
2125 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2126 $e2 = null;
2127 if ( $useLinkPrefixExtension ) {
2128 # Match the end of a line for a word that's not followed by whitespace,
2129 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2130 global $wgContLang;
2131 $charset = $wgContLang->linkPrefixCharset();
2132 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2133 }
2134
2135 if ( is_null( $this->mTitle ) ) {
2136 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2137 }
2138 $nottalk = !$this->mTitle->isTalkPage();
2139
2140 if ( $useLinkPrefixExtension ) {
2141 $m = [];
2142 if ( preg_match( $e2, $s, $m ) ) {
2143 $first_prefix = $m[2];
2144 } else {
2145 $first_prefix = false;
2146 }
2147 } else {
2148 $prefix = '';
2149 }
2150
2151 $useSubpages = $this->areSubpagesAllowed();
2152
2153 # Loop for each link
2154 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2155 # Check for excessive memory usage
2156 if ( $holders->isBig() ) {
2157 # Too big
2158 # Do the existence check, replace the link holders and clear the array
2159 $holders->replace( $s );
2160 $holders->clear();
2161 }
2162
2163 if ( $useLinkPrefixExtension ) {
2164 if ( preg_match( $e2, $s, $m ) ) {
2165 $prefix = $m[2];
2166 $s = $m[1];
2167 } else {
2168 $prefix = '';
2169 }
2170 # first link
2171 if ( $first_prefix ) {
2172 $prefix = $first_prefix;
2173 $first_prefix = false;
2174 }
2175 }
2176
2177 $might_be_img = false;
2178
2179 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2180 $text = $m[2];
2181 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2182 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2183 # the real problem is with the $e1 regex
2184 # See T1500.
2185 # Still some problems for cases where the ] is meant to be outside punctuation,
2186 # and no image is in sight. See T4095.
2187 if ( $text !== ''
2188 && substr( $m[3], 0, 1 ) === ']'
2189 && strpos( $text, '[' ) !== false
2190 ) {
2191 $text .= ']'; # so that replaceExternalLinks($text) works later
2192 $m[3] = substr( $m[3], 1 );
2193 }
2194 # fix up urlencoded title texts
2195 if ( strpos( $m[1], '%' ) !== false ) {
2196 # Should anchors '#' also be rejected?
2197 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2198 }
2199 $trail = $m[3];
2200 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2201 # Invalid, but might be an image with a link in its caption
2202 $might_be_img = true;
2203 $text = $m[2];
2204 if ( strpos( $m[1], '%' ) !== false ) {
2205 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2206 }
2207 $trail = "";
2208 } else { # Invalid form; output directly
2209 $s .= $prefix . '[[' . $line;
2210 continue;
2211 }
2212
2213 $origLink = ltrim( $m[1], ' ' );
2214
2215 # Don't allow internal links to pages containing
2216 # PROTO: where PROTO is a valid URL protocol; these
2217 # should be external links.
2218 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2219 $s .= $prefix . '[[' . $line;
2220 continue;
2221 }
2222
2223 # Make subpage if necessary
2224 if ( $useSubpages ) {
2225 $link = $this->maybeDoSubpageLink( $origLink, $text );
2226 } else {
2227 $link = $origLink;
2228 }
2229
2230 // \x7f isn't a default legal title char, so most likely strip
2231 // markers will force us into the "invalid form" path above. But,
2232 // just in case, let's assert that xmlish tags aren't valid in
2233 // the title position.
2234 $unstrip = $this->mStripState->killMarkers( $link );
2235 $noMarkers = ( $unstrip === $link );
2236
2237 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2238 if ( $nt === null ) {
2239 $s .= $prefix . '[[' . $line;
2240 continue;
2241 }
2242
2243 $ns = $nt->getNamespace();
2244 $iw = $nt->getInterwiki();
2245
2246 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2247
2248 if ( $might_be_img ) { # if this is actually an invalid link
2249 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2250 $found = false;
2251 while ( true ) {
2252 # look at the next 'line' to see if we can close it there
2253 $a->next();
2254 $next_line = $a->current();
2255 if ( $next_line === false || $next_line === null ) {
2256 break;
2257 }
2258 $m = explode( ']]', $next_line, 3 );
2259 if ( count( $m ) == 3 ) {
2260 # the first ]] closes the inner link, the second the image
2261 $found = true;
2262 $text .= "[[{$m[0]}]]{$m[1]}";
2263 $trail = $m[2];
2264 break;
2265 } elseif ( count( $m ) == 2 ) {
2266 # if there's exactly one ]] that's fine, we'll keep looking
2267 $text .= "[[{$m[0]}]]{$m[1]}";
2268 } else {
2269 # if $next_line is invalid too, we need look no further
2270 $text .= '[[' . $next_line;
2271 break;
2272 }
2273 }
2274 if ( !$found ) {
2275 # we couldn't find the end of this imageLink, so output it raw
2276 # but don't ignore what might be perfectly normal links in the text we've examined
2277 $holders->merge( $this->replaceInternalLinks2( $text ) );
2278 $s .= "{$prefix}[[$link|$text";
2279 # note: no $trail, because without an end, there *is* no trail
2280 continue;
2281 }
2282 } else { # it's not an image, so output it raw
2283 $s .= "{$prefix}[[$link|$text";
2284 # note: no $trail, because without an end, there *is* no trail
2285 continue;
2286 }
2287 }
2288
2289 $wasblank = ( $text == '' );
2290 if ( $wasblank ) {
2291 $text = $link;
2292 if ( !$noforce ) {
2293 # Strip off leading ':'
2294 $text = substr( $text, 1 );
2295 }
2296 } else {
2297 # T6598 madness. Handle the quotes only if they come from the alternate part
2298 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2299 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2300 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2301 $text = $this->doQuotes( $text );
2302 }
2303
2304 # Link not escaped by : , create the various objects
2305 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2306 # Interwikis
2307 if (
2308 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2309 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2310 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2311 )
2312 ) {
2313 # T26502: filter duplicates
2314 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2315 $this->mLangLinkLanguages[$iw] = true;
2316 $this->mOutput->addLanguageLink( $nt->getFullText() );
2317 }
2318
2319 /**
2320 * Strip the whitespace interwiki links produce, see T10897
2321 */
2322 $s = rtrim( $s . $prefix ) . $trail; # T175416
2323 continue;
2324 }
2325
2326 if ( $ns == NS_FILE ) {
2327 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2328 if ( $wasblank ) {
2329 # if no parameters were passed, $text
2330 # becomes something like "File:Foo.png",
2331 # which we don't want to pass on to the
2332 # image generator
2333 $text = '';
2334 } else {
2335 # recursively parse links inside the image caption
2336 # actually, this will parse them in any other parameters, too,
2337 # but it might be hard to fix that, and it doesn't matter ATM
2338 $text = $this->replaceExternalLinks( $text );
2339 $holders->merge( $this->replaceInternalLinks2( $text ) );
2340 }
2341 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2342 $s .= $prefix . $this->armorLinks(
2343 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2344 continue;
2345 }
2346 } elseif ( $ns == NS_CATEGORY ) {
2347 /**
2348 * Strip the whitespace Category links produce, see T2087
2349 */
2350 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2351
2352 if ( $wasblank ) {
2353 $sortkey = $this->getDefaultSort();
2354 } else {
2355 $sortkey = $text;
2356 }
2357 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2358 $sortkey = str_replace( "\n", '', $sortkey );
2359 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2360 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2361
2362 continue;
2363 }
2364 }
2365
2366 # Self-link checking. For some languages, variants of the title are checked in
2367 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2368 # for linking to a different variant.
2369 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2370 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2371 continue;
2372 }
2373
2374 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2375 # @todo FIXME: Should do batch file existence checks, see comment below
2376 if ( $ns == NS_MEDIA ) {
2377 # Give extensions a chance to select the file revision for us
2378 $options = [];
2379 $descQuery = false;
2380 Hooks::run( 'BeforeParserFetchFileAndTitle',
2381 [ $this, $nt, &$options, &$descQuery ] );
2382 # Fetch and register the file (file title may be different via hooks)
2383 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2384 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2385 $s .= $prefix . $this->armorLinks(
2386 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2387 continue;
2388 }
2389
2390 # Some titles, such as valid special pages or files in foreign repos, should
2391 # be shown as bluelinks even though they're not included in the page table
2392 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2393 # batch file existence checks for NS_FILE and NS_MEDIA
2394 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2395 $this->mOutput->addLink( $nt );
2396 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2397 } else {
2398 # Links will be added to the output link list after checking
2399 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2400 }
2401 }
2402 return $holders;
2403 }
2404
2405 /**
2406 * Render a forced-blue link inline; protect against double expansion of
2407 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2408 * Since this little disaster has to split off the trail text to avoid
2409 * breaking URLs in the following text without breaking trails on the
2410 * wiki links, it's been made into a horrible function.
2411 *
2412 * @param Title $nt
2413 * @param string $text
2414 * @param string $trail
2415 * @param string $prefix
2416 * @return string HTML-wikitext mix oh yuck
2417 */
2418 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2419 list( $inside, $trail ) = Linker::splitTrail( $trail );
2420
2421 if ( $text == '' ) {
2422 $text = htmlspecialchars( $nt->getPrefixedText() );
2423 }
2424
2425 $link = $this->getLinkRenderer()->makeKnownLink(
2426 $nt, new HtmlArmor( "$prefix$text$inside" )
2427 );
2428
2429 return $this->armorLinks( $link ) . $trail;
2430 }
2431
2432 /**
2433 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2434 * going to go through further parsing steps before inline URL expansion.
2435 *
2436 * Not needed quite as much as it used to be since free links are a bit
2437 * more sensible these days. But bracketed links are still an issue.
2438 *
2439 * @param string $text More-or-less HTML
2440 * @return string Less-or-more HTML with NOPARSE bits
2441 */
2442 public function armorLinks( $text ) {
2443 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2444 self::MARKER_PREFIX . "NOPARSE$1", $text );
2445 }
2446
2447 /**
2448 * Return true if subpage links should be expanded on this page.
2449 * @return bool
2450 */
2451 public function areSubpagesAllowed() {
2452 # Some namespaces don't allow subpages
2453 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2454 }
2455
2456 /**
2457 * Handle link to subpage if necessary
2458 *
2459 * @param string $target The source of the link
2460 * @param string &$text The link text, modified as necessary
2461 * @return string The full name of the link
2462 * @private
2463 */
2464 public function maybeDoSubpageLink( $target, &$text ) {
2465 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2466 }
2467
2468 /**
2469 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2470 *
2471 * @param string $text
2472 * @param bool $linestart Whether or not this is at the start of a line.
2473 * @private
2474 * @return string The lists rendered as HTML
2475 */
2476 public function doBlockLevels( $text, $linestart ) {
2477 return BlockLevelPass::doBlockLevels( $text, $linestart );
2478 }
2479
2480 /**
2481 * Return value of a magic variable (like PAGENAME)
2482 *
2483 * @private
2484 *
2485 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2486 * @param bool|PPFrame $frame
2487 *
2488 * @throws MWException
2489 * @return string
2490 */
2491 public function getVariableValue( $index, $frame = false ) {
2492 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2493 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2494
2495 if ( is_null( $this->mTitle ) ) {
2496 // If no title set, bad things are going to happen
2497 // later. Title should always be set since this
2498 // should only be called in the middle of a parse
2499 // operation (but the unit-tests do funky stuff)
2500 throw new MWException( __METHOD__ . ' Should only be '
2501 . ' called while parsing (no title set)' );
2502 }
2503
2504 // Avoid PHP 7.1 warning from passing $this by reference
2505 $parser = $this;
2506
2507 /**
2508 * Some of these require message or data lookups and can be
2509 * expensive to check many times.
2510 */
2511 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2512 if ( isset( $this->mVarCache[$index] ) ) {
2513 return $this->mVarCache[$index];
2514 }
2515 }
2516
2517 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2518 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2519
2520 $pageLang = $this->getFunctionLang();
2521
2522 switch ( $index ) {
2523 case '!':
2524 $value = '|';
2525 break;
2526 case 'currentmonth':
2527 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2528 break;
2529 case 'currentmonth1':
2530 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2531 break;
2532 case 'currentmonthname':
2533 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2534 break;
2535 case 'currentmonthnamegen':
2536 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2537 break;
2538 case 'currentmonthabbrev':
2539 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2540 break;
2541 case 'currentday':
2542 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2543 break;
2544 case 'currentday2':
2545 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2546 break;
2547 case 'localmonth':
2548 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2549 break;
2550 case 'localmonth1':
2551 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2552 break;
2553 case 'localmonthname':
2554 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2555 break;
2556 case 'localmonthnamegen':
2557 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2558 break;
2559 case 'localmonthabbrev':
2560 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2561 break;
2562 case 'localday':
2563 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2564 break;
2565 case 'localday2':
2566 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2567 break;
2568 case 'pagename':
2569 $value = wfEscapeWikiText( $this->mTitle->getText() );
2570 break;
2571 case 'pagenamee':
2572 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2573 break;
2574 case 'fullpagename':
2575 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2576 break;
2577 case 'fullpagenamee':
2578 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2579 break;
2580 case 'subpagename':
2581 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2582 break;
2583 case 'subpagenamee':
2584 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2585 break;
2586 case 'rootpagename':
2587 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2588 break;
2589 case 'rootpagenamee':
2590 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2591 ' ',
2592 '_',
2593 $this->mTitle->getRootText()
2594 ) ) );
2595 break;
2596 case 'basepagename':
2597 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2598 break;
2599 case 'basepagenamee':
2600 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2601 ' ',
2602 '_',
2603 $this->mTitle->getBaseText()
2604 ) ) );
2605 break;
2606 case 'talkpagename':
2607 if ( $this->mTitle->canHaveTalkPage() ) {
2608 $talkPage = $this->mTitle->getTalkPage();
2609 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2610 } else {
2611 $value = '';
2612 }
2613 break;
2614 case 'talkpagenamee':
2615 if ( $this->mTitle->canHaveTalkPage() ) {
2616 $talkPage = $this->mTitle->getTalkPage();
2617 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2618 } else {
2619 $value = '';
2620 }
2621 break;
2622 case 'subjectpagename':
2623 $subjPage = $this->mTitle->getSubjectPage();
2624 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2625 break;
2626 case 'subjectpagenamee':
2627 $subjPage = $this->mTitle->getSubjectPage();
2628 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2629 break;
2630 case 'pageid': // requested in T25427
2631 $pageid = $this->getTitle()->getArticleID();
2632 if ( $pageid == 0 ) {
2633 # 0 means the page doesn't exist in the database,
2634 # which means the user is previewing a new page.
2635 # The vary-revision flag must be set, because the magic word
2636 # will have a different value once the page is saved.
2637 $this->mOutput->setFlag( 'vary-revision' );
2638 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2639 }
2640 $value = $pageid ?: null;
2641 break;
2642 case 'revisionid':
2643 # Let the edit saving system know we should parse the page
2644 # *after* a revision ID has been assigned.
2645 $this->mOutput->setFlag( 'vary-revision-id' );
2646 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2647 $value = $this->mRevisionId;
2648 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2649 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2650 $this->mOutput->setSpeculativeRevIdUsed( $value );
2651 }
2652 break;
2653 case 'revisionday':
2654 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2655 break;
2656 case 'revisionday2':
2657 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2658 break;
2659 case 'revisionmonth':
2660 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2661 break;
2662 case 'revisionmonth1':
2663 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2664 break;
2665 case 'revisionyear':
2666 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2667 break;
2668 case 'revisiontimestamp':
2669 # Let the edit saving system know we should parse the page
2670 # *after* a revision ID has been assigned. This is for null edits.
2671 $this->mOutput->setFlag( 'vary-revision' );
2672 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2673 $value = $this->getRevisionTimestamp();
2674 break;
2675 case 'revisionuser':
2676 # Let the edit saving system know we should parse the page
2677 # *after* a revision ID has been assigned for null edits.
2678 $this->mOutput->setFlag( 'vary-user' );
2679 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2680 $value = $this->getRevisionUser();
2681 break;
2682 case 'revisionsize':
2683 $value = $this->getRevisionSize();
2684 break;
2685 case 'namespace':
2686 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2687 break;
2688 case 'namespacee':
2689 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2690 break;
2691 case 'namespacenumber':
2692 $value = $this->mTitle->getNamespace();
2693 break;
2694 case 'talkspace':
2695 $value = $this->mTitle->canHaveTalkPage()
2696 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2697 : '';
2698 break;
2699 case 'talkspacee':
2700 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2701 break;
2702 case 'subjectspace':
2703 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2704 break;
2705 case 'subjectspacee':
2706 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2707 break;
2708 case 'currentdayname':
2709 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2710 break;
2711 case 'currentyear':
2712 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2713 break;
2714 case 'currenttime':
2715 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2716 break;
2717 case 'currenthour':
2718 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2719 break;
2720 case 'currentweek':
2721 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2722 # int to remove the padding
2723 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2724 break;
2725 case 'currentdow':
2726 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2727 break;
2728 case 'localdayname':
2729 $value = $pageLang->getWeekdayName(
2730 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2731 );
2732 break;
2733 case 'localyear':
2734 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2735 break;
2736 case 'localtime':
2737 $value = $pageLang->time(
2738 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2739 false,
2740 false
2741 );
2742 break;
2743 case 'localhour':
2744 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2745 break;
2746 case 'localweek':
2747 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2748 # int to remove the padding
2749 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2750 break;
2751 case 'localdow':
2752 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2753 break;
2754 case 'numberofarticles':
2755 $value = $pageLang->formatNum( SiteStats::articles() );
2756 break;
2757 case 'numberoffiles':
2758 $value = $pageLang->formatNum( SiteStats::images() );
2759 break;
2760 case 'numberofusers':
2761 $value = $pageLang->formatNum( SiteStats::users() );
2762 break;
2763 case 'numberofactiveusers':
2764 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2765 break;
2766 case 'numberofpages':
2767 $value = $pageLang->formatNum( SiteStats::pages() );
2768 break;
2769 case 'numberofadmins':
2770 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2771 break;
2772 case 'numberofedits':
2773 $value = $pageLang->formatNum( SiteStats::edits() );
2774 break;
2775 case 'currenttimestamp':
2776 $value = wfTimestamp( TS_MW, $ts );
2777 break;
2778 case 'localtimestamp':
2779 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2780 break;
2781 case 'currentversion':
2782 $value = SpecialVersion::getVersion();
2783 break;
2784 case 'articlepath':
2785 return $wgArticlePath;
2786 case 'sitename':
2787 return $wgSitename;
2788 case 'server':
2789 return $wgServer;
2790 case 'servername':
2791 return $wgServerName;
2792 case 'scriptpath':
2793 return $wgScriptPath;
2794 case 'stylepath':
2795 return $wgStylePath;
2796 case 'directionmark':
2797 return $pageLang->getDirMark();
2798 case 'contentlanguage':
2799 global $wgLanguageCode;
2800 return $wgLanguageCode;
2801 case 'pagelanguage':
2802 $value = $pageLang->getCode();
2803 break;
2804 case 'cascadingsources':
2805 $value = CoreParserFunctions::cascadingsources( $this );
2806 break;
2807 default:
2808 $ret = null;
2809 Hooks::run(
2810 'ParserGetVariableValueSwitch',
2811 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2812 );
2813
2814 return $ret;
2815 }
2816
2817 if ( $index ) {
2818 $this->mVarCache[$index] = $value;
2819 }
2820
2821 return $value;
2822 }
2823
2824 /**
2825 * @param int $start
2826 * @param int $len
2827 * @param int $mtts Max time-till-save; sets vary-revision if result might change by then
2828 * @param string $variable Parser variable name
2829 * @return string
2830 */
2831 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
2832 global $wgContLang;
2833
2834 # Get the timezone-adjusted timestamp to be used for this revision
2835 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
2836 # Possibly set vary-revision if there is not yet an associated revision
2837 if ( !$this->getRevisionObject() ) {
2838 # Get the timezone-adjusted timestamp $mtts seconds in the future
2839 $resThen = substr(
2840 $wgContLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
2841 $start,
2842 $len
2843 );
2844
2845 if ( $resNow !== $resThen ) {
2846 # Let the edit saving system know we should parse the page
2847 # *after* a revision ID has been assigned. This is for null edits.
2848 $this->mOutput->setFlag( 'vary-revision' );
2849 wfDebug( __METHOD__ . ": $variable used, setting vary-revision...\n" );
2850 }
2851 }
2852
2853 return $resNow;
2854 }
2855
2856 /**
2857 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2858 *
2859 * @private
2860 */
2861 public function initialiseVariables() {
2862 $variableIDs = MagicWord::getVariableIDs();
2863 $substIDs = MagicWord::getSubstIDs();
2864
2865 $this->mVariables = new MagicWordArray( $variableIDs );
2866 $this->mSubstWords = new MagicWordArray( $substIDs );
2867 }
2868
2869 /**
2870 * Preprocess some wikitext and return the document tree.
2871 * This is the ghost of replace_variables().
2872 *
2873 * @param string $text The text to parse
2874 * @param int $flags Bitwise combination of:
2875 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2876 * included. Default is to assume a direct page view.
2877 *
2878 * The generated DOM tree must depend only on the input text and the flags.
2879 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2880 *
2881 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2882 * change in the DOM tree for a given text, must be passed through the section identifier
2883 * in the section edit link and thus back to extractSections().
2884 *
2885 * The output of this function is currently only cached in process memory, but a persistent
2886 * cache may be implemented at a later date which takes further advantage of these strict
2887 * dependency requirements.
2888 *
2889 * @return PPNode
2890 */
2891 public function preprocessToDom( $text, $flags = 0 ) {
2892 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2893 return $dom;
2894 }
2895
2896 /**
2897 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2898 *
2899 * @param string $s
2900 *
2901 * @return array
2902 */
2903 public static function splitWhitespace( $s ) {
2904 $ltrimmed = ltrim( $s );
2905 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2906 $trimmed = rtrim( $ltrimmed );
2907 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2908 if ( $diff > 0 ) {
2909 $w2 = substr( $ltrimmed, -$diff );
2910 } else {
2911 $w2 = '';
2912 }
2913 return [ $w1, $trimmed, $w2 ];
2914 }
2915
2916 /**
2917 * Replace magic variables, templates, and template arguments
2918 * with the appropriate text. Templates are substituted recursively,
2919 * taking care to avoid infinite loops.
2920 *
2921 * Note that the substitution depends on value of $mOutputType:
2922 * self::OT_WIKI: only {{subst:}} templates
2923 * self::OT_PREPROCESS: templates but not extension tags
2924 * self::OT_HTML: all templates and extension tags
2925 *
2926 * @param string $text The text to transform
2927 * @param bool|PPFrame $frame Object describing the arguments passed to the
2928 * template. Arguments may also be provided as an associative array, as
2929 * was the usual case before MW1.12. Providing arguments this way may be
2930 * useful for extensions wishing to perform variable replacement
2931 * explicitly.
2932 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2933 * double-brace expansion.
2934 * @return string
2935 */
2936 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2937 # Is there any text? Also, Prevent too big inclusions!
2938 $textSize = strlen( $text );
2939 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2940 return $text;
2941 }
2942
2943 if ( $frame === false ) {
2944 $frame = $this->getPreprocessor()->newFrame();
2945 } elseif ( !( $frame instanceof PPFrame ) ) {
2946 wfDebug( __METHOD__ . " called using plain parameters instead of "
2947 . "a PPFrame instance. Creating custom frame.\n" );
2948 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2949 }
2950
2951 $dom = $this->preprocessToDom( $text );
2952 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2953 $text = $frame->expand( $dom, $flags );
2954
2955 return $text;
2956 }
2957
2958 /**
2959 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2960 *
2961 * @param array $args
2962 *
2963 * @return array
2964 */
2965 public static function createAssocArgs( $args ) {
2966 $assocArgs = [];
2967 $index = 1;
2968 foreach ( $args as $arg ) {
2969 $eqpos = strpos( $arg, '=' );
2970 if ( $eqpos === false ) {
2971 $assocArgs[$index++] = $arg;
2972 } else {
2973 $name = trim( substr( $arg, 0, $eqpos ) );
2974 $value = trim( substr( $arg, $eqpos + 1 ) );
2975 if ( $value === false ) {
2976 $value = '';
2977 }
2978 if ( $name !== false ) {
2979 $assocArgs[$name] = $value;
2980 }
2981 }
2982 }
2983
2984 return $assocArgs;
2985 }
2986
2987 /**
2988 * Warn the user when a parser limitation is reached
2989 * Will warn at most once the user per limitation type
2990 *
2991 * The results are shown during preview and run through the Parser (See EditPage.php)
2992 *
2993 * @param string $limitationType Should be one of:
2994 * 'expensive-parserfunction' (corresponding messages:
2995 * 'expensive-parserfunction-warning',
2996 * 'expensive-parserfunction-category')
2997 * 'post-expand-template-argument' (corresponding messages:
2998 * 'post-expand-template-argument-warning',
2999 * 'post-expand-template-argument-category')
3000 * 'post-expand-template-inclusion' (corresponding messages:
3001 * 'post-expand-template-inclusion-warning',
3002 * 'post-expand-template-inclusion-category')
3003 * 'node-count-exceeded' (corresponding messages:
3004 * 'node-count-exceeded-warning',
3005 * 'node-count-exceeded-category')
3006 * 'expansion-depth-exceeded' (corresponding messages:
3007 * 'expansion-depth-exceeded-warning',
3008 * 'expansion-depth-exceeded-category')
3009 * @param string|int|null $current Current value
3010 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3011 * exceeded, provide the values (optional)
3012 */
3013 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3014 # does no harm if $current and $max are present but are unnecessary for the message
3015 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3016 # only during preview, and that would split the parser cache unnecessarily.
3017 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3018 ->text();
3019 $this->mOutput->addWarning( $warning );
3020 $this->addTrackingCategory( "$limitationType-category" );
3021 }
3022
3023 /**
3024 * Return the text of a template, after recursively
3025 * replacing any variables or templates within the template.
3026 *
3027 * @param array $piece The parts of the template
3028 * $piece['title']: the title, i.e. the part before the |
3029 * $piece['parts']: the parameter array
3030 * $piece['lineStart']: whether the brace was at the start of a line
3031 * @param PPFrame $frame The current frame, contains template arguments
3032 * @throws Exception
3033 * @return string The text of the template
3034 */
3035 public function braceSubstitution( $piece, $frame ) {
3036 // Flags
3037
3038 // $text has been filled
3039 $found = false;
3040 // wiki markup in $text should be escaped
3041 $nowiki = false;
3042 // $text is HTML, armour it against wikitext transformation
3043 $isHTML = false;
3044 // Force interwiki transclusion to be done in raw mode not rendered
3045 $forceRawInterwiki = false;
3046 // $text is a DOM node needing expansion in a child frame
3047 $isChildObj = false;
3048 // $text is a DOM node needing expansion in the current frame
3049 $isLocalObj = false;
3050
3051 # Title object, where $text came from
3052 $title = false;
3053
3054 # $part1 is the bit before the first |, and must contain only title characters.
3055 # Various prefixes will be stripped from it later.
3056 $titleWithSpaces = $frame->expand( $piece['title'] );
3057 $part1 = trim( $titleWithSpaces );
3058 $titleText = false;
3059
3060 # Original title text preserved for various purposes
3061 $originalTitle = $part1;
3062
3063 # $args is a list of argument nodes, starting from index 0, not including $part1
3064 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3065 # below won't work b/c this $args isn't an object
3066 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3067
3068 $profileSection = null; // profile templates
3069
3070 # SUBST
3071 if ( !$found ) {
3072 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3073
3074 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3075 # Decide whether to expand template or keep wikitext as-is.
3076 if ( $this->ot['wiki'] ) {
3077 if ( $substMatch === false ) {
3078 $literal = true; # literal when in PST with no prefix
3079 } else {
3080 $literal = false; # expand when in PST with subst: or safesubst:
3081 }
3082 } else {
3083 if ( $substMatch == 'subst' ) {
3084 $literal = true; # literal when not in PST with plain subst:
3085 } else {
3086 $literal = false; # expand when not in PST with safesubst: or no prefix
3087 }
3088 }
3089 if ( $literal ) {
3090 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3091 $isLocalObj = true;
3092 $found = true;
3093 }
3094 }
3095
3096 # Variables
3097 if ( !$found && $args->getLength() == 0 ) {
3098 $id = $this->mVariables->matchStartToEnd( $part1 );
3099 if ( $id !== false ) {
3100 $text = $this->getVariableValue( $id, $frame );
3101 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3102 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3103 }
3104 $found = true;
3105 }
3106 }
3107
3108 # MSG, MSGNW and RAW
3109 if ( !$found ) {
3110 # Check for MSGNW:
3111 $mwMsgnw = MagicWord::get( 'msgnw' );
3112 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3113 $nowiki = true;
3114 } else {
3115 # Remove obsolete MSG:
3116 $mwMsg = MagicWord::get( 'msg' );
3117 $mwMsg->matchStartAndRemove( $part1 );
3118 }
3119
3120 # Check for RAW:
3121 $mwRaw = MagicWord::get( 'raw' );
3122 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3123 $forceRawInterwiki = true;
3124 }
3125 }
3126
3127 # Parser functions
3128 if ( !$found ) {
3129 $colonPos = strpos( $part1, ':' );
3130 if ( $colonPos !== false ) {
3131 $func = substr( $part1, 0, $colonPos );
3132 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3133 $argsLength = $args->getLength();
3134 for ( $i = 0; $i < $argsLength; $i++ ) {
3135 $funcArgs[] = $args->item( $i );
3136 }
3137
3138 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3139
3140 // Extract any forwarded flags
3141 if ( isset( $result['title'] ) ) {
3142 $title = $result['title'];
3143 }
3144 if ( isset( $result['found'] ) ) {
3145 $found = $result['found'];
3146 }
3147 if ( array_key_exists( 'text', $result ) ) {
3148 // a string or null
3149 $text = $result['text'];
3150 }
3151 if ( isset( $result['nowiki'] ) ) {
3152 $nowiki = $result['nowiki'];
3153 }
3154 if ( isset( $result['isHTML'] ) ) {
3155 $isHTML = $result['isHTML'];
3156 }
3157 if ( isset( $result['forceRawInterwiki'] ) ) {
3158 $forceRawInterwiki = $result['forceRawInterwiki'];
3159 }
3160 if ( isset( $result['isChildObj'] ) ) {
3161 $isChildObj = $result['isChildObj'];
3162 }
3163 if ( isset( $result['isLocalObj'] ) ) {
3164 $isLocalObj = $result['isLocalObj'];
3165 }
3166 }
3167 }
3168
3169 # Finish mangling title and then check for loops.
3170 # Set $title to a Title object and $titleText to the PDBK
3171 if ( !$found ) {
3172 $ns = NS_TEMPLATE;
3173 # Split the title into page and subpage
3174 $subpage = '';
3175 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3176 if ( $part1 !== $relative ) {
3177 $part1 = $relative;
3178 $ns = $this->mTitle->getNamespace();
3179 }
3180 $title = Title::newFromText( $part1, $ns );
3181 if ( $title ) {
3182 $titleText = $title->getPrefixedText();
3183 # Check for language variants if the template is not found
3184 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3185 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3186 }
3187 # Do recursion depth check
3188 $limit = $this->mOptions->getMaxTemplateDepth();
3189 if ( $frame->depth >= $limit ) {
3190 $found = true;
3191 $text = '<span class="error">'
3192 . wfMessage( 'parser-template-recursion-depth-warning' )
3193 ->numParams( $limit )->inContentLanguage()->text()
3194 . '</span>';
3195 }
3196 }
3197 }
3198
3199 # Load from database
3200 if ( !$found && $title ) {
3201 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3202 if ( !$title->isExternal() ) {
3203 if ( $title->isSpecialPage()
3204 && $this->mOptions->getAllowSpecialInclusion()
3205 && $this->ot['html']
3206 ) {
3207 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3208 // Pass the template arguments as URL parameters.
3209 // "uselang" will have no effect since the Language object
3210 // is forced to the one defined in ParserOptions.
3211 $pageArgs = [];
3212 $argsLength = $args->getLength();
3213 for ( $i = 0; $i < $argsLength; $i++ ) {
3214 $bits = $args->item( $i )->splitArg();
3215 if ( strval( $bits['index'] ) === '' ) {
3216 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3217 $value = trim( $frame->expand( $bits['value'] ) );
3218 $pageArgs[$name] = $value;
3219 }
3220 }
3221
3222 // Create a new context to execute the special page
3223 $context = new RequestContext;
3224 $context->setTitle( $title );
3225 $context->setRequest( new FauxRequest( $pageArgs ) );
3226 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3227 $context->setUser( $this->getUser() );
3228 } else {
3229 // If this page is cached, then we better not be per user.
3230 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3231 }
3232 $context->setLanguage( $this->mOptions->getUserLangObj() );
3233 $ret = SpecialPageFactory::capturePath(
3234 $title, $context, $this->getLinkRenderer() );
3235 if ( $ret ) {
3236 $text = $context->getOutput()->getHTML();
3237 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3238 $found = true;
3239 $isHTML = true;
3240 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3241 $this->mOutput->updateRuntimeAdaptiveExpiry(
3242 $specialPage->maxIncludeCacheTime()
3243 );
3244 }
3245 }
3246 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3247 $found = false; # access denied
3248 wfDebug( __METHOD__ . ": template inclusion denied for " .
3249 $title->getPrefixedDBkey() . "\n" );
3250 } else {
3251 list( $text, $title ) = $this->getTemplateDom( $title );
3252 if ( $text !== false ) {
3253 $found = true;
3254 $isChildObj = true;
3255 }
3256 }
3257
3258 # If the title is valid but undisplayable, make a link to it
3259 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3260 $text = "[[:$titleText]]";
3261 $found = true;
3262 }
3263 } elseif ( $title->isTrans() ) {
3264 # Interwiki transclusion
3265 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3266 $text = $this->interwikiTransclude( $title, 'render' );
3267 $isHTML = true;
3268 } else {
3269 $text = $this->interwikiTransclude( $title, 'raw' );
3270 # Preprocess it like a template
3271 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3272 $isChildObj = true;
3273 }
3274 $found = true;
3275 }
3276
3277 # Do infinite loop check
3278 # This has to be done after redirect resolution to avoid infinite loops via redirects
3279 if ( !$frame->loopCheck( $title ) ) {
3280 $found = true;
3281 $text = '<span class="error">'
3282 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3283 . '</span>';
3284 $this->addTrackingCategory( 'template-loop-category' );
3285 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3286 wfEscapeWikiText( $titleText ) )->text() );
3287 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3288 }
3289 }
3290
3291 # If we haven't found text to substitute by now, we're done
3292 # Recover the source wikitext and return it
3293 if ( !$found ) {
3294 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3295 if ( $profileSection ) {
3296 $this->mProfiler->scopedProfileOut( $profileSection );
3297 }
3298 return [ 'object' => $text ];
3299 }
3300
3301 # Expand DOM-style return values in a child frame
3302 if ( $isChildObj ) {
3303 # Clean up argument array
3304 $newFrame = $frame->newChild( $args, $title );
3305
3306 if ( $nowiki ) {
3307 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3308 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3309 # Expansion is eligible for the empty-frame cache
3310 $text = $newFrame->cachedExpand( $titleText, $text );
3311 } else {
3312 # Uncached expansion
3313 $text = $newFrame->expand( $text );
3314 }
3315 }
3316 if ( $isLocalObj && $nowiki ) {
3317 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3318 $isLocalObj = false;
3319 }
3320
3321 if ( $profileSection ) {
3322 $this->mProfiler->scopedProfileOut( $profileSection );
3323 }
3324
3325 # Replace raw HTML by a placeholder
3326 if ( $isHTML ) {
3327 $text = $this->insertStripItem( $text );
3328 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3329 # Escape nowiki-style return values
3330 $text = wfEscapeWikiText( $text );
3331 } elseif ( is_string( $text )
3332 && !$piece['lineStart']
3333 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3334 ) {
3335 # T2529: if the template begins with a table or block-level
3336 # element, it should be treated as beginning a new line.
3337 # This behavior is somewhat controversial.
3338 $text = "\n" . $text;
3339 }
3340
3341 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3342 # Error, oversize inclusion
3343 if ( $titleText !== false ) {
3344 # Make a working, properly escaped link if possible (T25588)
3345 $text = "[[:$titleText]]";
3346 } else {
3347 # This will probably not be a working link, but at least it may
3348 # provide some hint of where the problem is
3349 preg_replace( '/^:/', '', $originalTitle );
3350 $text = "[[:$originalTitle]]";
3351 }
3352 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3353 . 'post-expand include size too large -->' );
3354 $this->limitationWarn( 'post-expand-template-inclusion' );
3355 }
3356
3357 if ( $isLocalObj ) {
3358 $ret = [ 'object' => $text ];
3359 } else {
3360 $ret = [ 'text' => $text ];
3361 }
3362
3363 return $ret;
3364 }
3365
3366 /**
3367 * Call a parser function and return an array with text and flags.
3368 *
3369 * The returned array will always contain a boolean 'found', indicating
3370 * whether the parser function was found or not. It may also contain the
3371 * following:
3372 * text: string|object, resulting wikitext or PP DOM object
3373 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3374 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3375 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3376 * nowiki: bool, wiki markup in $text should be escaped
3377 *
3378 * @since 1.21
3379 * @param PPFrame $frame The current frame, contains template arguments
3380 * @param string $function Function name
3381 * @param array $args Arguments to the function
3382 * @throws MWException
3383 * @return array
3384 */
3385 public function callParserFunction( $frame, $function, array $args = [] ) {
3386 global $wgContLang;
3387
3388 # Case sensitive functions
3389 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3390 $function = $this->mFunctionSynonyms[1][$function];
3391 } else {
3392 # Case insensitive functions
3393 $function = $wgContLang->lc( $function );
3394 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3395 $function = $this->mFunctionSynonyms[0][$function];
3396 } else {
3397 return [ 'found' => false ];
3398 }
3399 }
3400
3401 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3402
3403 // Avoid PHP 7.1 warning from passing $this by reference
3404 $parser = $this;
3405
3406 $allArgs = [ &$parser ];
3407 if ( $flags & self::SFH_OBJECT_ARGS ) {
3408 # Convert arguments to PPNodes and collect for appending to $allArgs
3409 $funcArgs = [];
3410 foreach ( $args as $k => $v ) {
3411 if ( $v instanceof PPNode || $k === 0 ) {
3412 $funcArgs[] = $v;
3413 } else {
3414 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3415 }
3416 }
3417
3418 # Add a frame parameter, and pass the arguments as an array
3419 $allArgs[] = $frame;
3420 $allArgs[] = $funcArgs;
3421 } else {
3422 # Convert arguments to plain text and append to $allArgs
3423 foreach ( $args as $k => $v ) {
3424 if ( $v instanceof PPNode ) {
3425 $allArgs[] = trim( $frame->expand( $v ) );
3426 } elseif ( is_int( $k ) && $k >= 0 ) {
3427 $allArgs[] = trim( $v );
3428 } else {
3429 $allArgs[] = trim( "$k=$v" );
3430 }
3431 }
3432 }
3433
3434 $result = $callback( ...$allArgs );
3435
3436 # The interface for function hooks allows them to return a wikitext
3437 # string or an array containing the string and any flags. This mungs
3438 # things around to match what this method should return.
3439 if ( !is_array( $result ) ) {
3440 $result = [
3441 'found' => true,
3442 'text' => $result,
3443 ];
3444 } else {
3445 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3446 $result['text'] = $result[0];
3447 }
3448 unset( $result[0] );
3449 $result += [
3450 'found' => true,
3451 ];
3452 }
3453
3454 $noparse = true;
3455 $preprocessFlags = 0;
3456 if ( isset( $result['noparse'] ) ) {
3457 $noparse = $result['noparse'];
3458 }
3459 if ( isset( $result['preprocessFlags'] ) ) {
3460 $preprocessFlags = $result['preprocessFlags'];
3461 }
3462
3463 if ( !$noparse ) {
3464 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3465 $result['isChildObj'] = true;
3466 }
3467
3468 return $result;
3469 }
3470
3471 /**
3472 * Get the semi-parsed DOM representation of a template with a given title,
3473 * and its redirect destination title. Cached.
3474 *
3475 * @param Title $title
3476 *
3477 * @return array
3478 */
3479 public function getTemplateDom( $title ) {
3480 $cacheTitle = $title;
3481 $titleText = $title->getPrefixedDBkey();
3482
3483 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3484 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3485 $title = Title::makeTitle( $ns, $dbk );
3486 $titleText = $title->getPrefixedDBkey();
3487 }
3488 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3489 return [ $this->mTplDomCache[$titleText], $title ];
3490 }
3491
3492 # Cache miss, go to the database
3493 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3494
3495 if ( $text === false ) {
3496 $this->mTplDomCache[$titleText] = false;
3497 return [ false, $title ];
3498 }
3499
3500 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3501 $this->mTplDomCache[$titleText] = $dom;
3502
3503 if ( !$title->equals( $cacheTitle ) ) {
3504 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3505 [ $title->getNamespace(), $title->getDBkey() ];
3506 }
3507
3508 return [ $dom, $title ];
3509 }
3510
3511 /**
3512 * Fetch the current revision of a given title. Note that the revision
3513 * (and even the title) may not exist in the database, so everything
3514 * contributing to the output of the parser should use this method
3515 * where possible, rather than getting the revisions themselves. This
3516 * method also caches its results, so using it benefits performance.
3517 *
3518 * @since 1.24
3519 * @param Title $title
3520 * @return Revision
3521 */
3522 public function fetchCurrentRevisionOfTitle( $title ) {
3523 $cacheKey = $title->getPrefixedDBkey();
3524 if ( !$this->currentRevisionCache ) {
3525 $this->currentRevisionCache = new MapCacheLRU( 100 );
3526 }
3527 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3528 $this->currentRevisionCache->set( $cacheKey,
3529 // Defaults to Parser::statelessFetchRevision()
3530 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3531 );
3532 }
3533 return $this->currentRevisionCache->get( $cacheKey );
3534 }
3535
3536 /**
3537 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3538 * without passing them on to it.
3539 *
3540 * @since 1.24
3541 * @param Title $title
3542 * @param Parser|bool $parser
3543 * @return Revision|bool False if missing
3544 */
3545 public static function statelessFetchRevision( Title $title, $parser = false ) {
3546 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3547
3548 return $rev;
3549 }
3550
3551 /**
3552 * Fetch the unparsed text of a template and register a reference to it.
3553 * @param Title $title
3554 * @return array ( string or false, Title )
3555 */
3556 public function fetchTemplateAndTitle( $title ) {
3557 // Defaults to Parser::statelessFetchTemplate()
3558 $templateCb = $this->mOptions->getTemplateCallback();
3559 $stuff = call_user_func( $templateCb, $title, $this );
3560 // We use U+007F DELETE to distinguish strip markers from regular text.
3561 $text = $stuff['text'];
3562 if ( is_string( $stuff['text'] ) ) {
3563 $text = strtr( $text, "\x7f", "?" );
3564 }
3565 $finalTitle = $stuff['finalTitle'] ?? $title;
3566 if ( isset( $stuff['deps'] ) ) {
3567 foreach ( $stuff['deps'] as $dep ) {
3568 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3569 if ( $dep['title']->equals( $this->getTitle() ) ) {
3570 // If we transclude ourselves, the final result
3571 // will change based on the new version of the page
3572 $this->mOutput->setFlag( 'vary-revision' );
3573 }
3574 }
3575 }
3576 return [ $text, $finalTitle ];
3577 }
3578
3579 /**
3580 * Fetch the unparsed text of a template and register a reference to it.
3581 * @param Title $title
3582 * @return string|bool
3583 */
3584 public function fetchTemplate( $title ) {
3585 return $this->fetchTemplateAndTitle( $title )[0];
3586 }
3587
3588 /**
3589 * Static function to get a template
3590 * Can be overridden via ParserOptions::setTemplateCallback().
3591 *
3592 * @param Title $title
3593 * @param bool|Parser $parser
3594 *
3595 * @return array
3596 */
3597 public static function statelessFetchTemplate( $title, $parser = false ) {
3598 $text = $skip = false;
3599 $finalTitle = $title;
3600 $deps = [];
3601
3602 # Loop to fetch the article, with up to 1 redirect
3603 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall
3604 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3605 # Give extensions a chance to select the revision instead
3606 $id = false; # Assume current
3607 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3608 [ $parser, $title, &$skip, &$id ] );
3609
3610 if ( $skip ) {
3611 $text = false;
3612 $deps[] = [
3613 'title' => $title,
3614 'page_id' => $title->getArticleID(),
3615 'rev_id' => null
3616 ];
3617 break;
3618 }
3619 # Get the revision
3620 if ( $id ) {
3621 $rev = Revision::newFromId( $id );
3622 } elseif ( $parser ) {
3623 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3624 } else {
3625 $rev = Revision::newFromTitle( $title );
3626 }
3627 $rev_id = $rev ? $rev->getId() : 0;
3628 # If there is no current revision, there is no page
3629 if ( $id === false && !$rev ) {
3630 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3631 $linkCache->addBadLinkObj( $title );
3632 }
3633
3634 $deps[] = [
3635 'title' => $title,
3636 'page_id' => $title->getArticleID(),
3637 'rev_id' => $rev_id ];
3638 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3639 # We fetched a rev from a different title; register it too...
3640 $deps[] = [
3641 'title' => $rev->getTitle(),
3642 'page_id' => $rev->getPage(),
3643 'rev_id' => $rev_id ];
3644 }
3645
3646 if ( $rev ) {
3647 $content = $rev->getContent();
3648 $text = $content ? $content->getWikitextForTransclusion() : null;
3649
3650 Hooks::run( 'ParserFetchTemplate',
3651 [ $parser, $title, $rev, &$text, &$deps ] );
3652
3653 if ( $text === false || $text === null ) {
3654 $text = false;
3655 break;
3656 }
3657 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3658 global $wgContLang;
3659 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3660 if ( !$message->exists() ) {
3661 $text = false;
3662 break;
3663 }
3664 $content = $message->content();
3665 $text = $message->plain();
3666 } else {
3667 break;
3668 }
3669 if ( !$content ) {
3670 break;
3671 }
3672 # Redirect?
3673 $finalTitle = $title;
3674 $title = $content->getRedirectTarget();
3675 }
3676 return [
3677 'text' => $text,
3678 'finalTitle' => $finalTitle,
3679 'deps' => $deps ];
3680 }
3681
3682 /**
3683 * Fetch a file and its title and register a reference to it.
3684 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3685 * @param Title $title
3686 * @param array $options Array of options to RepoGroup::findFile
3687 * @return File|bool
3688 * @deprecated since 1.32, use fetchFileAndTitle instead
3689 */
3690 public function fetchFile( $title, $options = [] ) {
3691 wfDeprecated( __METHOD__, '1.32' );
3692 return $this->fetchFileAndTitle( $title, $options )[0];
3693 }
3694
3695 /**
3696 * Fetch a file and its title and register a reference to it.
3697 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3698 * @param Title $title
3699 * @param array $options Array of options to RepoGroup::findFile
3700 * @return array ( File or false, Title of file )
3701 */
3702 public function fetchFileAndTitle( $title, $options = [] ) {
3703 $file = $this->fetchFileNoRegister( $title, $options );
3704
3705 $time = $file ? $file->getTimestamp() : false;
3706 $sha1 = $file ? $file->getSha1() : false;
3707 # Register the file as a dependency...
3708 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3709 if ( $file && !$title->equals( $file->getTitle() ) ) {
3710 # Update fetched file title
3711 $title = $file->getTitle();
3712 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3713 }
3714 return [ $file, $title ];
3715 }
3716
3717 /**
3718 * Helper function for fetchFileAndTitle.
3719 *
3720 * Also useful if you need to fetch a file but not use it yet,
3721 * for example to get the file's handler.
3722 *
3723 * @param Title $title
3724 * @param array $options Array of options to RepoGroup::findFile
3725 * @return File|bool
3726 */
3727 protected function fetchFileNoRegister( $title, $options = [] ) {
3728 if ( isset( $options['broken'] ) ) {
3729 $file = false; // broken thumbnail forced by hook
3730 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3731 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3732 } else { // get by (name,timestamp)
3733 $file = wfFindFile( $title, $options );
3734 }
3735 return $file;
3736 }
3737
3738 /**
3739 * Transclude an interwiki link.
3740 *
3741 * @param Title $title
3742 * @param string $action Usually one of (raw, render)
3743 *
3744 * @return string
3745 */
3746 public function interwikiTransclude( $title, $action ) {
3747 global $wgEnableScaryTranscluding, $wgTranscludeCacheExpiry;
3748
3749 if ( !$wgEnableScaryTranscluding ) {
3750 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3751 }
3752
3753 $url = $title->getFullURL( [ 'action' => $action ] );
3754 if ( strlen( $url ) > 1024 ) {
3755 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3756 }
3757
3758 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
3759
3760 $fname = __METHOD__;
3761 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
3762
3763 $data = $cache->getWithSetCallback(
3764 $cache->makeGlobalKey(
3765 'interwiki-transclude',
3766 ( $wikiId !== false ) ? $wikiId : 'external',
3767 sha1( $url )
3768 ),
3769 $wgTranscludeCacheExpiry,
3770 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
3771 $req = MWHttpRequest::factory( $url, [], $fname );
3772
3773 $status = $req->execute(); // Status object
3774 if ( !$status->isOK() ) {
3775 $ttl = $cache::TTL_UNCACHEABLE;
3776 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
3777 $ttl = min( $cache::TTL_LAGGED, $ttl );
3778 }
3779
3780 return [
3781 'text' => $status->isOK() ? $req->getContent() : null,
3782 'code' => $req->getStatus()
3783 ];
3784 },
3785 [
3786 'checkKeys' => ( $wikiId !== false )
3787 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
3788 : [],
3789 'pcGroup' => 'interwiki-transclude:5',
3790 'pcTTL' => $cache::TTL_PROC_LONG
3791 ]
3792 );
3793
3794 if ( is_string( $data['text'] ) ) {
3795 $text = $data['text'];
3796 } elseif ( $data['code'] != 200 ) {
3797 // Though we failed to fetch the content, this status is useless.
3798 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
3799 ->params( $url, $data['code'] )->inContentLanguage()->text();
3800 } else {
3801 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3802 }
3803
3804 return $text;
3805 }
3806
3807 /**
3808 * Triple brace replacement -- used for template arguments
3809 * @private
3810 *
3811 * @param array $piece
3812 * @param PPFrame $frame
3813 *
3814 * @return array
3815 */
3816 public function argSubstitution( $piece, $frame ) {
3817 $error = false;
3818 $parts = $piece['parts'];
3819 $nameWithSpaces = $frame->expand( $piece['title'] );
3820 $argName = trim( $nameWithSpaces );
3821 $object = false;
3822 $text = $frame->getArgument( $argName );
3823 if ( $text === false && $parts->getLength() > 0
3824 && ( $this->ot['html']
3825 || $this->ot['pre']
3826 || ( $this->ot['wiki'] && $frame->isTemplate() )
3827 )
3828 ) {
3829 # No match in frame, use the supplied default
3830 $object = $parts->item( 0 )->getChildren();
3831 }
3832 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3833 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3834 $this->limitationWarn( 'post-expand-template-argument' );
3835 }
3836
3837 if ( $text === false && $object === false ) {
3838 # No match anywhere
3839 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3840 }
3841 if ( $error !== false ) {
3842 $text .= $error;
3843 }
3844 if ( $object !== false ) {
3845 $ret = [ 'object' => $object ];
3846 } else {
3847 $ret = [ 'text' => $text ];
3848 }
3849
3850 return $ret;
3851 }
3852
3853 /**
3854 * Return the text to be used for a given extension tag.
3855 * This is the ghost of strip().
3856 *
3857 * @param array $params Associative array of parameters:
3858 * name PPNode for the tag name
3859 * attr PPNode for unparsed text where tag attributes are thought to be
3860 * attributes Optional associative array of parsed attributes
3861 * inner Contents of extension element
3862 * noClose Original text did not have a close tag
3863 * @param PPFrame $frame
3864 *
3865 * @throws MWException
3866 * @return string
3867 */
3868 public function extensionSubstitution( $params, $frame ) {
3869 static $errorStr = '<span class="error">';
3870 static $errorLen = 20;
3871
3872 $name = $frame->expand( $params['name'] );
3873 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3874 // Probably expansion depth or node count exceeded. Just punt the
3875 // error up.
3876 return $name;
3877 }
3878
3879 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3880 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3881 // See above
3882 return $attrText;
3883 }
3884
3885 // We can't safely check if the expansion for $content resulted in an
3886 // error, because the content could happen to be the error string
3887 // (T149622).
3888 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3889
3890 $marker = self::MARKER_PREFIX . "-$name-"
3891 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3892
3893 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3894 ( $this->ot['html'] || $this->ot['pre'] );
3895 if ( $isFunctionTag ) {
3896 $markerType = 'none';
3897 } else {
3898 $markerType = 'general';
3899 }
3900 if ( $this->ot['html'] || $isFunctionTag ) {
3901 $name = strtolower( $name );
3902 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3903 if ( isset( $params['attributes'] ) ) {
3904 $attributes = $attributes + $params['attributes'];
3905 }
3906
3907 if ( isset( $this->mTagHooks[$name] ) ) {
3908 $output = call_user_func_array( $this->mTagHooks[$name],
3909 [ $content, $attributes, $this, $frame ] );
3910 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3911 list( $callback, ) = $this->mFunctionTagHooks[$name];
3912
3913 // Avoid PHP 7.1 warning from passing $this by reference
3914 $parser = $this;
3915 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3916 } else {
3917 $output = '<span class="error">Invalid tag extension name: ' .
3918 htmlspecialchars( $name ) . '</span>';
3919 }
3920
3921 if ( is_array( $output ) ) {
3922 // Extract flags
3923 $flags = $output;
3924 $output = $flags[0];
3925 if ( isset( $flags['markerType'] ) ) {
3926 $markerType = $flags['markerType'];
3927 }
3928 }
3929 } else {
3930 if ( is_null( $attrText ) ) {
3931 $attrText = '';
3932 }
3933 if ( isset( $params['attributes'] ) ) {
3934 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3935 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3936 htmlspecialchars( $attrValue ) . '"';
3937 }
3938 }
3939 if ( $content === null ) {
3940 $output = "<$name$attrText/>";
3941 } else {
3942 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3943 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3944 // See above
3945 return $close;
3946 }
3947 $output = "<$name$attrText>$content$close";
3948 }
3949 }
3950
3951 if ( $markerType === 'none' ) {
3952 return $output;
3953 } elseif ( $markerType === 'nowiki' ) {
3954 $this->mStripState->addNoWiki( $marker, $output );
3955 } elseif ( $markerType === 'general' ) {
3956 $this->mStripState->addGeneral( $marker, $output );
3957 } else {
3958 throw new MWException( __METHOD__ . ': invalid marker type' );
3959 }
3960 return $marker;
3961 }
3962
3963 /**
3964 * Increment an include size counter
3965 *
3966 * @param string $type The type of expansion
3967 * @param int $size The size of the text
3968 * @return bool False if this inclusion would take it over the maximum, true otherwise
3969 */
3970 public function incrementIncludeSize( $type, $size ) {
3971 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3972 return false;
3973 } else {
3974 $this->mIncludeSizes[$type] += $size;
3975 return true;
3976 }
3977 }
3978
3979 /**
3980 * Increment the expensive function count
3981 *
3982 * @return bool False if the limit has been exceeded
3983 */
3984 public function incrementExpensiveFunctionCount() {
3985 $this->mExpensiveFunctionCount++;
3986 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3987 }
3988
3989 /**
3990 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3991 * Fills $this->mDoubleUnderscores, returns the modified text
3992 *
3993 * @param string $text
3994 *
3995 * @return string
3996 */
3997 public function doDoubleUnderscore( $text ) {
3998 # The position of __TOC__ needs to be recorded
3999 $mw = MagicWord::get( 'toc' );
4000 if ( $mw->match( $text ) ) {
4001 $this->mShowToc = true;
4002 $this->mForceTocPosition = true;
4003
4004 # Set a placeholder. At the end we'll fill it in with the TOC.
4005 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4006
4007 # Only keep the first one.
4008 $text = $mw->replace( '', $text );
4009 }
4010
4011 # Now match and remove the rest of them
4012 $mwa = MagicWord::getDoubleUnderscoreArray();
4013 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4014
4015 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4016 $this->mOutput->mNoGallery = true;
4017 }
4018 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4019 $this->mShowToc = false;
4020 }
4021 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4022 && $this->mTitle->getNamespace() == NS_CATEGORY
4023 ) {
4024 $this->addTrackingCategory( 'hidden-category-category' );
4025 }
4026 # (T10068) Allow control over whether robots index a page.
4027 # __INDEX__ always overrides __NOINDEX__, see T16899
4028 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4029 $this->mOutput->setIndexPolicy( 'noindex' );
4030 $this->addTrackingCategory( 'noindex-category' );
4031 }
4032 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4033 $this->mOutput->setIndexPolicy( 'index' );
4034 $this->addTrackingCategory( 'index-category' );
4035 }
4036
4037 # Cache all double underscores in the database
4038 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4039 $this->mOutput->setProperty( $key, '' );
4040 }
4041
4042 return $text;
4043 }
4044
4045 /**
4046 * @see ParserOutput::addTrackingCategory()
4047 * @param string $msg Message key
4048 * @return bool Whether the addition was successful
4049 */
4050 public function addTrackingCategory( $msg ) {
4051 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4052 }
4053
4054 /**
4055 * This function accomplishes several tasks:
4056 * 1) Auto-number headings if that option is enabled
4057 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4058 * 3) Add a Table of contents on the top for users who have enabled the option
4059 * 4) Auto-anchor headings
4060 *
4061 * It loops through all headlines, collects the necessary data, then splits up the
4062 * string and re-inserts the newly formatted headlines.
4063 *
4064 * @param string $text
4065 * @param string $origText Original, untouched wikitext
4066 * @param bool $isMain
4067 * @return mixed|string
4068 * @private
4069 */
4070 public function formatHeadings( $text, $origText, $isMain = true ) {
4071 global $wgMaxTocLevel;
4072
4073 # Inhibit editsection links if requested in the page
4074 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4075 $maybeShowEditLink = false;
4076 } else {
4077 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4078 }
4079
4080 # Get all headlines for numbering them and adding funky stuff like [edit]
4081 # links - this is for later, but we need the number of headlines right now
4082 # NOTE: white space in headings have been trimmed in doHeadings. They shouldn't
4083 # be trimmed here since whitespace in HTML headings is significant.
4084 $matches = [];
4085 $numMatches = preg_match_all(
4086 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4087 $text,
4088 $matches
4089 );
4090
4091 # if there are fewer than 4 headlines in the article, do not show TOC
4092 # unless it's been explicitly enabled.
4093 $enoughToc = $this->mShowToc &&
4094 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4095
4096 # Allow user to stipulate that a page should have a "new section"
4097 # link added via __NEWSECTIONLINK__
4098 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4099 $this->mOutput->setNewSection( true );
4100 }
4101
4102 # Allow user to remove the "new section"
4103 # link via __NONEWSECTIONLINK__
4104 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4105 $this->mOutput->hideNewSection( true );
4106 }
4107
4108 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4109 # override above conditions and always show TOC above first header
4110 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4111 $this->mShowToc = true;
4112 $enoughToc = true;
4113 }
4114
4115 # headline counter
4116 $headlineCount = 0;
4117 $numVisible = 0;
4118
4119 # Ugh .. the TOC should have neat indentation levels which can be
4120 # passed to the skin functions. These are determined here
4121 $toc = '';
4122 $full = '';
4123 $head = [];
4124 $sublevelCount = [];
4125 $levelCount = [];
4126 $level = 0;
4127 $prevlevel = 0;
4128 $toclevel = 0;
4129 $prevtoclevel = 0;
4130 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4131 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4132 $oldType = $this->mOutputType;
4133 $this->setOutputType( self::OT_WIKI );
4134 $frame = $this->getPreprocessor()->newFrame();
4135 $root = $this->preprocessToDom( $origText );
4136 $node = $root->getFirstChild();
4137 $byteOffset = 0;
4138 $tocraw = [];
4139 $refers = [];
4140
4141 $headlines = $numMatches !== false ? $matches[3] : [];
4142
4143 foreach ( $headlines as $headline ) {
4144 $isTemplate = false;
4145 $titleText = false;
4146 $sectionIndex = false;
4147 $numbering = '';
4148 $markerMatches = [];
4149 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4150 $serial = $markerMatches[1];
4151 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4152 $isTemplate = ( $titleText != $baseTitleText );
4153 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4154 }
4155
4156 if ( $toclevel ) {
4157 $prevlevel = $level;
4158 }
4159 $level = $matches[1][$headlineCount];
4160
4161 if ( $level > $prevlevel ) {
4162 # Increase TOC level
4163 $toclevel++;
4164 $sublevelCount[$toclevel] = 0;
4165 if ( $toclevel < $wgMaxTocLevel ) {
4166 $prevtoclevel = $toclevel;
4167 $toc .= Linker::tocIndent();
4168 $numVisible++;
4169 }
4170 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4171 # Decrease TOC level, find level to jump to
4172
4173 for ( $i = $toclevel; $i > 0; $i-- ) {
4174 if ( $levelCount[$i] == $level ) {
4175 # Found last matching level
4176 $toclevel = $i;
4177 break;
4178 } elseif ( $levelCount[$i] < $level ) {
4179 # Found first matching level below current level
4180 $toclevel = $i + 1;
4181 break;
4182 }
4183 }
4184 if ( $i == 0 ) {
4185 $toclevel = 1;
4186 }
4187 if ( $toclevel < $wgMaxTocLevel ) {
4188 if ( $prevtoclevel < $wgMaxTocLevel ) {
4189 # Unindent only if the previous toc level was shown :p
4190 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4191 $prevtoclevel = $toclevel;
4192 } else {
4193 $toc .= Linker::tocLineEnd();
4194 }
4195 }
4196 } else {
4197 # No change in level, end TOC line
4198 if ( $toclevel < $wgMaxTocLevel ) {
4199 $toc .= Linker::tocLineEnd();
4200 }
4201 }
4202
4203 $levelCount[$toclevel] = $level;
4204
4205 # count number of headlines for each level
4206 $sublevelCount[$toclevel]++;
4207 $dot = 0;
4208 for ( $i = 1; $i <= $toclevel; $i++ ) {
4209 if ( !empty( $sublevelCount[$i] ) ) {
4210 if ( $dot ) {
4211 $numbering .= '.';
4212 }
4213 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4214 $dot = 1;
4215 }
4216 }
4217
4218 # The safe header is a version of the header text safe to use for links
4219
4220 # Remove link placeholders by the link text.
4221 # <!--LINK number-->
4222 # turns into
4223 # link text with suffix
4224 # Do this before unstrip since link text can contain strip markers
4225 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4226
4227 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4228 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4229
4230 # Remove any <style> or <script> tags (T198618)
4231 $safeHeadline = preg_replace(
4232 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4233 '',
4234 $safeHeadline
4235 );
4236
4237 # Strip out HTML (first regex removes any tag not allowed)
4238 # Allowed tags are:
4239 # * <sup> and <sub> (T10393)
4240 # * <i> (T28375)
4241 # * <b> (r105284)
4242 # * <bdi> (T74884)
4243 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4244 # * <s> and <strike> (T35715)
4245 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4246 # to allow setting directionality in toc items.
4247 $tocline = preg_replace(
4248 [
4249 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4250 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4251 ],
4252 [ '', '<$1>' ],
4253 $safeHeadline
4254 );
4255
4256 # Strip '<span></span>', which is the result from the above if
4257 # <span id="foo"></span> is used to produce an additional anchor
4258 # for a section.
4259 $tocline = str_replace( '<span></span>', '', $tocline );
4260
4261 $tocline = trim( $tocline );
4262
4263 # For the anchor, strip out HTML-y stuff period
4264 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4265 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4266
4267 # Save headline for section edit hint before it's escaped
4268 $headlineHint = $safeHeadline;
4269
4270 # Decode HTML entities
4271 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4272
4273 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4274
4275 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4276 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4277 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4278 if ( $fallbackHeadline === $safeHeadline ) {
4279 # No reason to have both (in fact, we can't)
4280 $fallbackHeadline = false;
4281 }
4282
4283 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4284 # @todo FIXME: We may be changing them depending on the current locale.
4285 $arrayKey = strtolower( $safeHeadline );
4286 if ( $fallbackHeadline === false ) {
4287 $fallbackArrayKey = false;
4288 } else {
4289 $fallbackArrayKey = strtolower( $fallbackHeadline );
4290 }
4291
4292 # Create the anchor for linking from the TOC to the section
4293 $anchor = $safeHeadline;
4294 $fallbackAnchor = $fallbackHeadline;
4295 if ( isset( $refers[$arrayKey] ) ) {
4296 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4297 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4298 $anchor .= "_$i";
4299 $linkAnchor .= "_$i";
4300 $refers["${arrayKey}_$i"] = true;
4301 } else {
4302 $refers[$arrayKey] = true;
4303 }
4304 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4305 // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall,Generic.Formatting.DisallowMultipleStatements
4306 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4307 $fallbackAnchor .= "_$i";
4308 $refers["${fallbackArrayKey}_$i"] = true;
4309 } else {
4310 $refers[$fallbackArrayKey] = true;
4311 }
4312
4313 # Don't number the heading if it is the only one (looks silly)
4314 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4315 # the two are different if the line contains a link
4316 $headline = Html::element(
4317 'span',
4318 [ 'class' => 'mw-headline-number' ],
4319 $numbering
4320 ) . ' ' . $headline;
4321 }
4322
4323 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4324 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4325 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4326 }
4327
4328 # Add the section to the section tree
4329 # Find the DOM node for this header
4330 $noOffset = ( $isTemplate || $sectionIndex === false );
4331 while ( $node && !$noOffset ) {
4332 if ( $node->getName() === 'h' ) {
4333 $bits = $node->splitHeading();
4334 if ( $bits['i'] == $sectionIndex ) {
4335 break;
4336 }
4337 }
4338 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4339 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4340 $node = $node->getNextSibling();
4341 }
4342 $tocraw[] = [
4343 'toclevel' => $toclevel,
4344 'level' => $level,
4345 'line' => $tocline,
4346 'number' => $numbering,
4347 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4348 'fromtitle' => $titleText,
4349 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4350 'anchor' => $anchor,
4351 ];
4352
4353 # give headline the correct <h#> tag
4354 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4355 // Output edit section links as markers with styles that can be customized by skins
4356 if ( $isTemplate ) {
4357 # Put a T flag in the section identifier, to indicate to extractSections()
4358 # that sections inside <includeonly> should be counted.
4359 $editsectionPage = $titleText;
4360 $editsectionSection = "T-$sectionIndex";
4361 $editsectionContent = null;
4362 } else {
4363 $editsectionPage = $this->mTitle->getPrefixedText();
4364 $editsectionSection = $sectionIndex;
4365 $editsectionContent = $headlineHint;
4366 }
4367 // We use a bit of pesudo-xml for editsection markers. The
4368 // language converter is run later on. Using a UNIQ style marker
4369 // leads to the converter screwing up the tokens when it
4370 // converts stuff. And trying to insert strip tags fails too. At
4371 // this point all real inputted tags have already been escaped,
4372 // so we don't have to worry about a user trying to input one of
4373 // these markers directly. We use a page and section attribute
4374 // to stop the language converter from converting these
4375 // important bits of data, but put the headline hint inside a
4376 // content block because the language converter is supposed to
4377 // be able to convert that piece of data.
4378 // Gets replaced with html in ParserOutput::getText
4379 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4380 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4381 if ( $editsectionContent !== null ) {
4382 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4383 } else {
4384 $editlink .= '/>';
4385 }
4386 } else {
4387 $editlink = '';
4388 }
4389 $head[$headlineCount] = Linker::makeHeadline( $level,
4390 $matches['attrib'][$headlineCount], $anchor, $headline,
4391 $editlink, $fallbackAnchor );
4392
4393 $headlineCount++;
4394 }
4395
4396 $this->setOutputType( $oldType );
4397
4398 # Never ever show TOC if no headers
4399 if ( $numVisible < 1 ) {
4400 $enoughToc = false;
4401 }
4402
4403 if ( $enoughToc ) {
4404 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4405 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4406 }
4407 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4408 $this->mOutput->setTOCHTML( $toc );
4409 $toc = self::TOC_START . $toc . self::TOC_END;
4410 }
4411
4412 if ( $isMain ) {
4413 $this->mOutput->setSections( $tocraw );
4414 }
4415
4416 # split up and insert constructed headlines
4417 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4418 $i = 0;
4419
4420 // build an array of document sections
4421 $sections = [];
4422 foreach ( $blocks as $block ) {
4423 // $head is zero-based, sections aren't.
4424 if ( empty( $head[$i - 1] ) ) {
4425 $sections[$i] = $block;
4426 } else {
4427 $sections[$i] = $head[$i - 1] . $block;
4428 }
4429
4430 /**
4431 * Send a hook, one per section.
4432 * The idea here is to be able to make section-level DIVs, but to do so in a
4433 * lower-impact, more correct way than r50769
4434 *
4435 * $this : caller
4436 * $section : the section number
4437 * &$sectionContent : ref to the content of the section
4438 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4439 */
4440 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4441
4442 $i++;
4443 }
4444
4445 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4446 // append the TOC at the beginning
4447 // Top anchor now in skin
4448 $sections[0] = $sections[0] . $toc . "\n";
4449 }
4450
4451 $full .= implode( '', $sections );
4452
4453 if ( $this->mForceTocPosition ) {
4454 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4455 } else {
4456 return $full;
4457 }
4458 }
4459
4460 /**
4461 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4462 * conversion, substituting signatures, {{subst:}} templates, etc.
4463 *
4464 * @param string $text The text to transform
4465 * @param Title $title The Title object for the current article
4466 * @param User $user The User object describing the current user
4467 * @param ParserOptions $options Parsing options
4468 * @param bool $clearState Whether to clear the parser state first
4469 * @return string The altered wiki markup
4470 */
4471 public function preSaveTransform( $text, Title $title, User $user,
4472 ParserOptions $options, $clearState = true
4473 ) {
4474 if ( $clearState ) {
4475 $magicScopeVariable = $this->lock();
4476 }
4477 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4478 $this->setUser( $user );
4479
4480 // Strip U+0000 NULL (T159174)
4481 $text = str_replace( "\000", '', $text );
4482
4483 // We still normalize line endings for backwards-compatibility
4484 // with other code that just calls PST, but this should already
4485 // be handled in TextContent subclasses
4486 $text = TextContent::normalizeLineEndings( $text );
4487
4488 if ( $options->getPreSaveTransform() ) {
4489 $text = $this->pstPass2( $text, $user );
4490 }
4491 $text = $this->mStripState->unstripBoth( $text );
4492
4493 $this->setUser( null ); # Reset
4494
4495 return $text;
4496 }
4497
4498 /**
4499 * Pre-save transform helper function
4500 *
4501 * @param string $text
4502 * @param User $user
4503 *
4504 * @return string
4505 */
4506 private function pstPass2( $text, $user ) {
4507 global $wgContLang;
4508
4509 # Note: This is the timestamp saved as hardcoded wikitext to
4510 # the database, we use $wgContLang here in order to give
4511 # everyone the same signature and use the default one rather
4512 # than the one selected in each user's preferences.
4513 # (see also T14815)
4514 $ts = $this->mOptions->getTimestamp();
4515 $timestamp = MWTimestamp::getLocalInstance( $ts );
4516 $ts = $timestamp->format( 'YmdHis' );
4517 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4518
4519 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4520
4521 # Variable replacement
4522 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4523 $text = $this->replaceVariables( $text );
4524
4525 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4526 # which may corrupt this parser instance via its wfMessage()->text() call-
4527
4528 # Signatures
4529 if ( strpos( $text, '~~~' ) !== false ) {
4530 $sigText = $this->getUserSig( $user );
4531 $text = strtr( $text, [
4532 '~~~~~' => $d,
4533 '~~~~' => "$sigText $d",
4534 '~~~' => $sigText
4535 ] );
4536 # The main two signature forms used above are time-sensitive
4537 $this->mOutput->setFlag( 'user-signature' );
4538 }
4539
4540 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4541 $tc = '[' . Title::legalChars() . ']';
4542 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4543
4544 // [[ns:page (context)|]]
4545 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4546 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4547 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4548 // [[ns:page (context), context|]] (using either single or double-width comma)
4549 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4550 // [[|page]] (reverse pipe trick: add context from page title)
4551 $p2 = "/\[\[\\|($tc+)]]/";
4552
4553 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4554 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4555 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4556 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4557
4558 $t = $this->mTitle->getText();
4559 $m = [];
4560 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4561 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4562 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4563 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4564 } else {
4565 # if there's no context, don't bother duplicating the title
4566 $text = preg_replace( $p2, '[[\\1]]', $text );
4567 }
4568
4569 return $text;
4570 }
4571
4572 /**
4573 * Fetch the user's signature text, if any, and normalize to
4574 * validated, ready-to-insert wikitext.
4575 * If you have pre-fetched the nickname or the fancySig option, you can
4576 * specify them here to save a database query.
4577 * Do not reuse this parser instance after calling getUserSig(),
4578 * as it may have changed if it's the $wgParser.
4579 *
4580 * @param User &$user
4581 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4582 * @param bool|null $fancySig whether the nicknname is the complete signature
4583 * or null to use default value
4584 * @return string
4585 */
4586 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4587 global $wgMaxSigChars;
4588
4589 $username = $user->getName();
4590
4591 # If not given, retrieve from the user object.
4592 if ( $nickname === false ) {
4593 $nickname = $user->getOption( 'nickname' );
4594 }
4595
4596 if ( is_null( $fancySig ) ) {
4597 $fancySig = $user->getBoolOption( 'fancysig' );
4598 }
4599
4600 $nickname = $nickname == null ? $username : $nickname;
4601
4602 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4603 $nickname = $username;
4604 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4605 } elseif ( $fancySig !== false ) {
4606 # Sig. might contain markup; validate this
4607 if ( $this->validateSig( $nickname ) !== false ) {
4608 # Validated; clean up (if needed) and return it
4609 return $this->cleanSig( $nickname, true );
4610 } else {
4611 # Failed to validate; fall back to the default
4612 $nickname = $username;
4613 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4614 }
4615 }
4616
4617 # Make sure nickname doesnt get a sig in a sig
4618 $nickname = self::cleanSigInSig( $nickname );
4619
4620 # If we're still here, make it a link to the user page
4621 $userText = wfEscapeWikiText( $username );
4622 $nickText = wfEscapeWikiText( $nickname );
4623 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4624
4625 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4626 ->title( $this->getTitle() )->text();
4627 }
4628
4629 /**
4630 * Check that the user's signature contains no bad XML
4631 *
4632 * @param string $text
4633 * @return string|bool An expanded string, or false if invalid.
4634 */
4635 public function validateSig( $text ) {
4636 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4637 }
4638
4639 /**
4640 * Clean up signature text
4641 *
4642 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4643 * 2) Substitute all transclusions
4644 *
4645 * @param string $text
4646 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4647 * @return string Signature text
4648 */
4649 public function cleanSig( $text, $parsing = false ) {
4650 if ( !$parsing ) {
4651 global $wgTitle;
4652 $magicScopeVariable = $this->lock();
4653 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4654 }
4655
4656 # Option to disable this feature
4657 if ( !$this->mOptions->getCleanSignatures() ) {
4658 return $text;
4659 }
4660
4661 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4662 # => Move this logic to braceSubstitution()
4663 $substWord = MagicWord::get( 'subst' );
4664 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4665 $substText = '{{' . $substWord->getSynonym( 0 );
4666
4667 $text = preg_replace( $substRegex, $substText, $text );
4668 $text = self::cleanSigInSig( $text );
4669 $dom = $this->preprocessToDom( $text );
4670 $frame = $this->getPreprocessor()->newFrame();
4671 $text = $frame->expand( $dom );
4672
4673 if ( !$parsing ) {
4674 $text = $this->mStripState->unstripBoth( $text );
4675 }
4676
4677 return $text;
4678 }
4679
4680 /**
4681 * Strip 3, 4 or 5 tildes out of signatures.
4682 *
4683 * @param string $text
4684 * @return string Signature text with /~{3,5}/ removed
4685 */
4686 public static function cleanSigInSig( $text ) {
4687 $text = preg_replace( '/~{3,5}/', '', $text );
4688 return $text;
4689 }
4690
4691 /**
4692 * Set up some variables which are usually set up in parse()
4693 * so that an external function can call some class members with confidence
4694 *
4695 * @param Title|null $title
4696 * @param ParserOptions $options
4697 * @param int $outputType
4698 * @param bool $clearState
4699 */
4700 public function startExternalParse( Title $title = null, ParserOptions $options,
4701 $outputType, $clearState = true
4702 ) {
4703 $this->startParse( $title, $options, $outputType, $clearState );
4704 }
4705
4706 /**
4707 * @param Title|null $title
4708 * @param ParserOptions $options
4709 * @param int $outputType
4710 * @param bool $clearState
4711 */
4712 private function startParse( Title $title = null, ParserOptions $options,
4713 $outputType, $clearState = true
4714 ) {
4715 $this->setTitle( $title );
4716 $this->mOptions = $options;
4717 $this->setOutputType( $outputType );
4718 if ( $clearState ) {
4719 $this->clearState();
4720 }
4721 }
4722
4723 /**
4724 * Wrapper for preprocess()
4725 *
4726 * @param string $text The text to preprocess
4727 * @param ParserOptions $options
4728 * @param Title|null $title Title object or null to use $wgTitle
4729 * @return string
4730 */
4731 public function transformMsg( $text, $options, $title = null ) {
4732 static $executing = false;
4733
4734 # Guard against infinite recursion
4735 if ( $executing ) {
4736 return $text;
4737 }
4738 $executing = true;
4739
4740 if ( !$title ) {
4741 global $wgTitle;
4742 $title = $wgTitle;
4743 }
4744
4745 $text = $this->preprocess( $text, $title, $options );
4746
4747 $executing = false;
4748 return $text;
4749 }
4750
4751 /**
4752 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4753 * The callback should have the following form:
4754 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4755 *
4756 * Transform and return $text. Use $parser for any required context, e.g. use
4757 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4758 *
4759 * Hooks may return extended information by returning an array, of which the
4760 * first numbered element (index 0) must be the return string, and all other
4761 * entries are extracted into local variables within an internal function
4762 * in the Parser class.
4763 *
4764 * This interface (introduced r61913) appears to be undocumented, but
4765 * 'markerType' is used by some core tag hooks to override which strip
4766 * array their results are placed in. **Use great caution if attempting
4767 * this interface, as it is not documented and injudicious use could smash
4768 * private variables.**
4769 *
4770 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4771 * @param callable $callback The callback function (and object) to use for the tag
4772 * @throws MWException
4773 * @return callable|null The old value of the mTagHooks array associated with the hook
4774 */
4775 public function setHook( $tag, callable $callback ) {
4776 $tag = strtolower( $tag );
4777 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4778 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4779 }
4780 $oldVal = $this->mTagHooks[$tag] ?? null;
4781 $this->mTagHooks[$tag] = $callback;
4782 if ( !in_array( $tag, $this->mStripList ) ) {
4783 $this->mStripList[] = $tag;
4784 }
4785
4786 return $oldVal;
4787 }
4788
4789 /**
4790 * As setHook(), but letting the contents be parsed.
4791 *
4792 * Transparent tag hooks are like regular XML-style tag hooks, except they
4793 * operate late in the transformation sequence, on HTML instead of wikitext.
4794 *
4795 * This is probably obsoleted by things dealing with parser frames?
4796 * The only extension currently using it is geoserver.
4797 *
4798 * @since 1.10
4799 * @todo better document or deprecate this
4800 *
4801 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4802 * @param callable $callback The callback function (and object) to use for the tag
4803 * @throws MWException
4804 * @return callable|null The old value of the mTagHooks array associated with the hook
4805 */
4806 public function setTransparentTagHook( $tag, callable $callback ) {
4807 $tag = strtolower( $tag );
4808 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4809 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4810 }
4811 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
4812 $this->mTransparentTagHooks[$tag] = $callback;
4813
4814 return $oldVal;
4815 }
4816
4817 /**
4818 * Remove all tag hooks
4819 */
4820 public function clearTagHooks() {
4821 $this->mTagHooks = [];
4822 $this->mFunctionTagHooks = [];
4823 $this->mStripList = $this->mDefaultStripList;
4824 }
4825
4826 /**
4827 * Create a function, e.g. {{sum:1|2|3}}
4828 * The callback function should have the form:
4829 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4830 *
4831 * Or with Parser::SFH_OBJECT_ARGS:
4832 * function myParserFunction( $parser, $frame, $args ) { ... }
4833 *
4834 * The callback may either return the text result of the function, or an array with the text
4835 * in element 0, and a number of flags in the other elements. The names of the flags are
4836 * specified in the keys. Valid flags are:
4837 * found The text returned is valid, stop processing the template. This
4838 * is on by default.
4839 * nowiki Wiki markup in the return value should be escaped
4840 * isHTML The returned text is HTML, armour it against wikitext transformation
4841 *
4842 * @param string $id The magic word ID
4843 * @param callable $callback The callback function (and object) to use
4844 * @param int $flags A combination of the following flags:
4845 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4846 *
4847 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4848 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4849 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4850 * the arguments, and to control the way they are expanded.
4851 *
4852 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4853 * arguments, for instance:
4854 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4855 *
4856 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4857 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4858 * working if/when this is changed.
4859 *
4860 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4861 * expansion.
4862 *
4863 * Please read the documentation in includes/parser/Preprocessor.php for more information
4864 * about the methods available in PPFrame and PPNode.
4865 *
4866 * @throws MWException
4867 * @return string|callable The old callback function for this name, if any
4868 */
4869 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4870 global $wgContLang;
4871
4872 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4873 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4874
4875 # Add to function cache
4876 $mw = MagicWord::get( $id );
4877 if ( !$mw ) {
4878 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4879 }
4880
4881 $synonyms = $mw->getSynonyms();
4882 $sensitive = intval( $mw->isCaseSensitive() );
4883
4884 foreach ( $synonyms as $syn ) {
4885 # Case
4886 if ( !$sensitive ) {
4887 $syn = $wgContLang->lc( $syn );
4888 }
4889 # Add leading hash
4890 if ( !( $flags & self::SFH_NO_HASH ) ) {
4891 $syn = '#' . $syn;
4892 }
4893 # Remove trailing colon
4894 if ( substr( $syn, -1, 1 ) === ':' ) {
4895 $syn = substr( $syn, 0, -1 );
4896 }
4897 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4898 }
4899 return $oldVal;
4900 }
4901
4902 /**
4903 * Get all registered function hook identifiers
4904 *
4905 * @return array
4906 */
4907 public function getFunctionHooks() {
4908 return array_keys( $this->mFunctionHooks );
4909 }
4910
4911 /**
4912 * Create a tag function, e.g. "<test>some stuff</test>".
4913 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4914 * Unlike parser functions, their content is not preprocessed.
4915 * @param string $tag
4916 * @param callable $callback
4917 * @param int $flags
4918 * @throws MWException
4919 * @return null
4920 */
4921 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4922 $tag = strtolower( $tag );
4923 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4924 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4925 }
4926 $old = $this->mFunctionTagHooks[$tag] ?? null;
4927 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4928
4929 if ( !in_array( $tag, $this->mStripList ) ) {
4930 $this->mStripList[] = $tag;
4931 }
4932
4933 return $old;
4934 }
4935
4936 /**
4937 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4938 * Placeholders created in Linker::link()
4939 *
4940 * @param string &$text
4941 * @param int $options
4942 */
4943 public function replaceLinkHolders( &$text, $options = 0 ) {
4944 $this->mLinkHolders->replace( $text );
4945 }
4946
4947 /**
4948 * Replace "<!--LINK-->" link placeholders with plain text of links
4949 * (not HTML-formatted).
4950 *
4951 * @param string $text
4952 * @return string
4953 */
4954 public function replaceLinkHoldersText( $text ) {
4955 return $this->mLinkHolders->replaceText( $text );
4956 }
4957
4958 /**
4959 * Renders an image gallery from a text with one line per image.
4960 * text labels may be given by using |-style alternative text. E.g.
4961 * Image:one.jpg|The number "1"
4962 * Image:tree.jpg|A tree
4963 * given as text will return the HTML of a gallery with two images,
4964 * labeled 'The number "1"' and
4965 * 'A tree'.
4966 *
4967 * @param string $text
4968 * @param array $params
4969 * @return string HTML
4970 */
4971 public function renderImageGallery( $text, $params ) {
4972 $mode = false;
4973 if ( isset( $params['mode'] ) ) {
4974 $mode = $params['mode'];
4975 }
4976
4977 try {
4978 $ig = ImageGalleryBase::factory( $mode );
4979 } catch ( Exception $e ) {
4980 // If invalid type set, fallback to default.
4981 $ig = ImageGalleryBase::factory( false );
4982 }
4983
4984 $ig->setContextTitle( $this->mTitle );
4985 $ig->setShowBytes( false );
4986 $ig->setShowDimensions( false );
4987 $ig->setShowFilename( false );
4988 $ig->setParser( $this );
4989 $ig->setHideBadImages();
4990 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4991
4992 if ( isset( $params['showfilename'] ) ) {
4993 $ig->setShowFilename( true );
4994 } else {
4995 $ig->setShowFilename( false );
4996 }
4997 if ( isset( $params['caption'] ) ) {
4998 $caption = $params['caption'];
4999 $caption = htmlspecialchars( $caption );
5000 $caption = $this->replaceInternalLinks( $caption );
5001 $ig->setCaptionHtml( $caption );
5002 }
5003 if ( isset( $params['perrow'] ) ) {
5004 $ig->setPerRow( $params['perrow'] );
5005 }
5006 if ( isset( $params['widths'] ) ) {
5007 $ig->setWidths( $params['widths'] );
5008 }
5009 if ( isset( $params['heights'] ) ) {
5010 $ig->setHeights( $params['heights'] );
5011 }
5012 $ig->setAdditionalOptions( $params );
5013
5014 // Avoid PHP 7.1 warning from passing $this by reference
5015 $parser = $this;
5016 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5017
5018 $lines = StringUtils::explode( "\n", $text );
5019 foreach ( $lines as $line ) {
5020 # match lines like these:
5021 # Image:someimage.jpg|This is some image
5022 $matches = [];
5023 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5024 # Skip empty lines
5025 if ( count( $matches ) == 0 ) {
5026 continue;
5027 }
5028
5029 if ( strpos( $matches[0], '%' ) !== false ) {
5030 $matches[1] = rawurldecode( $matches[1] );
5031 }
5032 $title = Title::newFromText( $matches[1], NS_FILE );
5033 if ( is_null( $title ) ) {
5034 # Bogus title. Ignore these so we don't bomb out later.
5035 continue;
5036 }
5037
5038 # We need to get what handler the file uses, to figure out parameters.
5039 # Note, a hook can overide the file name, and chose an entirely different
5040 # file (which potentially could be of a different type and have different handler).
5041 $options = [];
5042 $descQuery = false;
5043 Hooks::run( 'BeforeParserFetchFileAndTitle',
5044 [ $this, $title, &$options, &$descQuery ] );
5045 # Don't register it now, as TraditionalImageGallery does that later.
5046 $file = $this->fetchFileNoRegister( $title, $options );
5047 $handler = $file ? $file->getHandler() : false;
5048
5049 $paramMap = [
5050 'img_alt' => 'gallery-internal-alt',
5051 'img_link' => 'gallery-internal-link',
5052 ];
5053 if ( $handler ) {
5054 $paramMap = $paramMap + $handler->getParamMap();
5055 // We don't want people to specify per-image widths.
5056 // Additionally the width parameter would need special casing anyhow.
5057 unset( $paramMap['img_width'] );
5058 }
5059
5060 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5061
5062 $label = '';
5063 $alt = '';
5064 $link = '';
5065 $handlerOptions = [];
5066 if ( isset( $matches[3] ) ) {
5067 // look for an |alt= definition while trying not to break existing
5068 // captions with multiple pipes (|) in it, until a more sensible grammar
5069 // is defined for images in galleries
5070
5071 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5072 // splitting on '|' is a bit odd, and different from makeImage.
5073 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5074 // Protect LanguageConverter markup
5075 $parameterMatches = StringUtils::delimiterExplode(
5076 '-{', '}-', '|', $matches[3], true /* nested */
5077 );
5078
5079 foreach ( $parameterMatches as $parameterMatch ) {
5080 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5081 if ( $magicName ) {
5082 $paramName = $paramMap[$magicName];
5083
5084 switch ( $paramName ) {
5085 case 'gallery-internal-alt':
5086 $alt = $this->stripAltText( $match, false );
5087 break;
5088 case 'gallery-internal-link':
5089 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5090 $chars = self::EXT_LINK_URL_CLASS;
5091 $addr = self::EXT_LINK_ADDR;
5092 $prots = $this->mUrlProtocols;
5093 // check to see if link matches an absolute url, if not then it must be a wiki link.
5094 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5095 // Result of LanguageConverter::markNoConversion
5096 // invoked on an external link.
5097 $linkValue = substr( $linkValue, 4, -2 );
5098 }
5099 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5100 $link = $linkValue;
5101 $this->mOutput->addExternalLink( $link );
5102 } else {
5103 $localLinkTitle = Title::newFromText( $linkValue );
5104 if ( $localLinkTitle !== null ) {
5105 $this->mOutput->addLink( $localLinkTitle );
5106 $link = $localLinkTitle->getLinkURL();
5107 }
5108 }
5109 break;
5110 default:
5111 // Must be a handler specific parameter.
5112 if ( $handler->validateParam( $paramName, $match ) ) {
5113 $handlerOptions[$paramName] = $match;
5114 } else {
5115 // Guess not, consider it as caption.
5116 wfDebug( "$parameterMatch failed parameter validation\n" );
5117 $label = $parameterMatch;
5118 }
5119 }
5120
5121 } else {
5122 // Last pipe wins.
5123 $label = $parameterMatch;
5124 }
5125 }
5126 }
5127
5128 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5129 }
5130 $html = $ig->toHTML();
5131 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5132 return $html;
5133 }
5134
5135 /**
5136 * @param MediaHandler $handler
5137 * @return array
5138 */
5139 public function getImageParams( $handler ) {
5140 if ( $handler ) {
5141 $handlerClass = get_class( $handler );
5142 } else {
5143 $handlerClass = '';
5144 }
5145 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5146 # Initialise static lists
5147 static $internalParamNames = [
5148 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5149 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5150 'bottom', 'text-bottom' ],
5151 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5152 'upright', 'border', 'link', 'alt', 'class' ],
5153 ];
5154 static $internalParamMap;
5155 if ( !$internalParamMap ) {
5156 $internalParamMap = [];
5157 foreach ( $internalParamNames as $type => $names ) {
5158 foreach ( $names as $name ) {
5159 // For grep: img_left, img_right, img_center, img_none,
5160 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5161 // img_bottom, img_text_bottom,
5162 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5163 // img_border, img_link, img_alt, img_class
5164 $magicName = str_replace( '-', '_', "img_$name" );
5165 $internalParamMap[$magicName] = [ $type, $name ];
5166 }
5167 }
5168 }
5169
5170 # Add handler params
5171 $paramMap = $internalParamMap;
5172 if ( $handler ) {
5173 $handlerParamMap = $handler->getParamMap();
5174 foreach ( $handlerParamMap as $magic => $paramName ) {
5175 $paramMap[$magic] = [ 'handler', $paramName ];
5176 }
5177 }
5178 $this->mImageParams[$handlerClass] = $paramMap;
5179 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5180 }
5181 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5182 }
5183
5184 /**
5185 * Parse image options text and use it to make an image
5186 *
5187 * @param Title $title
5188 * @param string $options
5189 * @param LinkHolderArray|bool $holders
5190 * @return string HTML
5191 */
5192 public function makeImage( $title, $options, $holders = false ) {
5193 # Check if the options text is of the form "options|alt text"
5194 # Options are:
5195 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5196 # * left no resizing, just left align. label is used for alt= only
5197 # * right same, but right aligned
5198 # * none same, but not aligned
5199 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5200 # * center center the image
5201 # * frame Keep original image size, no magnify-button.
5202 # * framed Same as "frame"
5203 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5204 # * upright reduce width for upright images, rounded to full __0 px
5205 # * border draw a 1px border around the image
5206 # * alt Text for HTML alt attribute (defaults to empty)
5207 # * class Set a class for img node
5208 # * link Set the target of the image link. Can be external, interwiki, or local
5209 # vertical-align values (no % or length right now):
5210 # * baseline
5211 # * sub
5212 # * super
5213 # * top
5214 # * text-top
5215 # * middle
5216 # * bottom
5217 # * text-bottom
5218
5219 # Protect LanguageConverter markup when splitting into parts
5220 $parts = StringUtils::delimiterExplode(
5221 '-{', '}-', '|', $options, true /* allow nesting */
5222 );
5223
5224 # Give extensions a chance to select the file revision for us
5225 $options = [];
5226 $descQuery = false;
5227 Hooks::run( 'BeforeParserFetchFileAndTitle',
5228 [ $this, $title, &$options, &$descQuery ] );
5229 # Fetch and register the file (file title may be different via hooks)
5230 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5231
5232 # Get parameter map
5233 $handler = $file ? $file->getHandler() : false;
5234
5235 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5236
5237 if ( !$file ) {
5238 $this->addTrackingCategory( 'broken-file-category' );
5239 }
5240
5241 # Process the input parameters
5242 $caption = '';
5243 $params = [ 'frame' => [], 'handler' => [],
5244 'horizAlign' => [], 'vertAlign' => [] ];
5245 $seenformat = false;
5246 foreach ( $parts as $part ) {
5247 $part = trim( $part );
5248 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5249 $validated = false;
5250 if ( isset( $paramMap[$magicName] ) ) {
5251 list( $type, $paramName ) = $paramMap[$magicName];
5252
5253 # Special case; width and height come in one variable together
5254 if ( $type === 'handler' && $paramName === 'width' ) {
5255 $parsedWidthParam = self::parseWidthParam( $value );
5256 if ( isset( $parsedWidthParam['width'] ) ) {
5257 $width = $parsedWidthParam['width'];
5258 if ( $handler->validateParam( 'width', $width ) ) {
5259 $params[$type]['width'] = $width;
5260 $validated = true;
5261 }
5262 }
5263 if ( isset( $parsedWidthParam['height'] ) ) {
5264 $height = $parsedWidthParam['height'];
5265 if ( $handler->validateParam( 'height', $height ) ) {
5266 $params[$type]['height'] = $height;
5267 $validated = true;
5268 }
5269 }
5270 # else no validation -- T15436
5271 } else {
5272 if ( $type === 'handler' ) {
5273 # Validate handler parameter
5274 $validated = $handler->validateParam( $paramName, $value );
5275 } else {
5276 # Validate internal parameters
5277 switch ( $paramName ) {
5278 case 'manualthumb':
5279 case 'alt':
5280 case 'class':
5281 # @todo FIXME: Possibly check validity here for
5282 # manualthumb? downstream behavior seems odd with
5283 # missing manual thumbs.
5284 $validated = true;
5285 $value = $this->stripAltText( $value, $holders );
5286 break;
5287 case 'link':
5288 $chars = self::EXT_LINK_URL_CLASS;
5289 $addr = self::EXT_LINK_ADDR;
5290 $prots = $this->mUrlProtocols;
5291 if ( $value === '' ) {
5292 $paramName = 'no-link';
5293 $value = true;
5294 $validated = true;
5295 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5296 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5297 $paramName = 'link-url';
5298 $this->mOutput->addExternalLink( $value );
5299 if ( $this->mOptions->getExternalLinkTarget() ) {
5300 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5301 }
5302 $validated = true;
5303 }
5304 } else {
5305 $linkTitle = Title::newFromText( $value );
5306 if ( $linkTitle ) {
5307 $paramName = 'link-title';
5308 $value = $linkTitle;
5309 $this->mOutput->addLink( $linkTitle );
5310 $validated = true;
5311 }
5312 }
5313 break;
5314 case 'frameless':
5315 case 'framed':
5316 case 'thumbnail':
5317 // use first appearing option, discard others.
5318 $validated = !$seenformat;
5319 $seenformat = true;
5320 break;
5321 default:
5322 # Most other things appear to be empty or numeric...
5323 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5324 }
5325 }
5326
5327 if ( $validated ) {
5328 $params[$type][$paramName] = $value;
5329 }
5330 }
5331 }
5332 if ( !$validated ) {
5333 $caption = $part;
5334 }
5335 }
5336
5337 # Process alignment parameters
5338 if ( $params['horizAlign'] ) {
5339 $params['frame']['align'] = key( $params['horizAlign'] );
5340 }
5341 if ( $params['vertAlign'] ) {
5342 $params['frame']['valign'] = key( $params['vertAlign'] );
5343 }
5344
5345 $params['frame']['caption'] = $caption;
5346
5347 # Will the image be presented in a frame, with the caption below?
5348 $imageIsFramed = isset( $params['frame']['frame'] )
5349 || isset( $params['frame']['framed'] )
5350 || isset( $params['frame']['thumbnail'] )
5351 || isset( $params['frame']['manualthumb'] );
5352
5353 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5354 # came to also set the caption, ordinary text after the image -- which
5355 # makes no sense, because that just repeats the text multiple times in
5356 # screen readers. It *also* came to set the title attribute.
5357 # Now that we have an alt attribute, we should not set the alt text to
5358 # equal the caption: that's worse than useless, it just repeats the
5359 # text. This is the framed/thumbnail case. If there's no caption, we
5360 # use the unnamed parameter for alt text as well, just for the time be-
5361 # ing, if the unnamed param is set and the alt param is not.
5362 # For the future, we need to figure out if we want to tweak this more,
5363 # e.g., introducing a title= parameter for the title; ignoring the un-
5364 # named parameter entirely for images without a caption; adding an ex-
5365 # plicit caption= parameter and preserving the old magic unnamed para-
5366 # meter for BC; ...
5367 if ( $imageIsFramed ) { # Framed image
5368 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5369 # No caption or alt text, add the filename as the alt text so
5370 # that screen readers at least get some description of the image
5371 $params['frame']['alt'] = $title->getText();
5372 }
5373 # Do not set $params['frame']['title'] because tooltips don't make sense
5374 # for framed images
5375 } else { # Inline image
5376 if ( !isset( $params['frame']['alt'] ) ) {
5377 # No alt text, use the "caption" for the alt text
5378 if ( $caption !== '' ) {
5379 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5380 } else {
5381 # No caption, fall back to using the filename for the
5382 # alt text
5383 $params['frame']['alt'] = $title->getText();
5384 }
5385 }
5386 # Use the "caption" for the tooltip text
5387 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5388 }
5389
5390 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5391
5392 # Linker does the rest
5393 $time = $options['time'] ?? false;
5394 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5395 $time, $descQuery, $this->mOptions->getThumbSize() );
5396
5397 # Give the handler a chance to modify the parser object
5398 if ( $handler ) {
5399 $handler->parserTransformHook( $this, $file );
5400 }
5401
5402 return $ret;
5403 }
5404
5405 /**
5406 * @param string $caption
5407 * @param LinkHolderArray|bool $holders
5408 * @return mixed|string
5409 */
5410 protected function stripAltText( $caption, $holders ) {
5411 # Strip bad stuff out of the title (tooltip). We can't just use
5412 # replaceLinkHoldersText() here, because if this function is called
5413 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5414 if ( $holders ) {
5415 $tooltip = $holders->replaceText( $caption );
5416 } else {
5417 $tooltip = $this->replaceLinkHoldersText( $caption );
5418 }
5419
5420 # make sure there are no placeholders in thumbnail attributes
5421 # that are later expanded to html- so expand them now and
5422 # remove the tags
5423 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5424 $tooltip = Sanitizer::stripAllTags( $tooltip );
5425
5426 return $tooltip;
5427 }
5428
5429 /**
5430 * Set a flag in the output object indicating that the content is dynamic and
5431 * shouldn't be cached.
5432 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5433 */
5434 public function disableCache() {
5435 wfDebug( "Parser output marked as uncacheable.\n" );
5436 if ( !$this->mOutput ) {
5437 throw new MWException( __METHOD__ .
5438 " can only be called when actually parsing something" );
5439 }
5440 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5441 }
5442
5443 /**
5444 * Callback from the Sanitizer for expanding items found in HTML attribute
5445 * values, so they can be safely tested and escaped.
5446 *
5447 * @param string &$text
5448 * @param bool|PPFrame $frame
5449 * @return string
5450 */
5451 public function attributeStripCallback( &$text, $frame = false ) {
5452 $text = $this->replaceVariables( $text, $frame );
5453 $text = $this->mStripState->unstripBoth( $text );
5454 return $text;
5455 }
5456
5457 /**
5458 * Accessor
5459 *
5460 * @return array
5461 */
5462 public function getTags() {
5463 return array_merge(
5464 array_keys( $this->mTransparentTagHooks ),
5465 array_keys( $this->mTagHooks ),
5466 array_keys( $this->mFunctionTagHooks )
5467 );
5468 }
5469
5470 /**
5471 * Replace transparent tags in $text with the values given by the callbacks.
5472 *
5473 * Transparent tag hooks are like regular XML-style tag hooks, except they
5474 * operate late in the transformation sequence, on HTML instead of wikitext.
5475 *
5476 * @param string $text
5477 *
5478 * @return string
5479 */
5480 public function replaceTransparentTags( $text ) {
5481 $matches = [];
5482 $elements = array_keys( $this->mTransparentTagHooks );
5483 $text = self::extractTagsAndParams( $elements, $text, $matches );
5484 $replacements = [];
5485
5486 foreach ( $matches as $marker => $data ) {
5487 list( $element, $content, $params, $tag ) = $data;
5488 $tagName = strtolower( $element );
5489 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5490 $output = call_user_func_array(
5491 $this->mTransparentTagHooks[$tagName],
5492 [ $content, $params, $this ]
5493 );
5494 } else {
5495 $output = $tag;
5496 }
5497 $replacements[$marker] = $output;
5498 }
5499 return strtr( $text, $replacements );
5500 }
5501
5502 /**
5503 * Break wikitext input into sections, and either pull or replace
5504 * some particular section's text.
5505 *
5506 * External callers should use the getSection and replaceSection methods.
5507 *
5508 * @param string $text Page wikitext
5509 * @param string|int $sectionId A section identifier string of the form:
5510 * "<flag1> - <flag2> - ... - <section number>"
5511 *
5512 * Currently the only recognised flag is "T", which means the target section number
5513 * was derived during a template inclusion parse, in other words this is a template
5514 * section edit link. If no flags are given, it was an ordinary section edit link.
5515 * This flag is required to avoid a section numbering mismatch when a section is
5516 * enclosed by "<includeonly>" (T8563).
5517 *
5518 * The section number 0 pulls the text before the first heading; other numbers will
5519 * pull the given section along with its lower-level subsections. If the section is
5520 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5521 *
5522 * Section 0 is always considered to exist, even if it only contains the empty
5523 * string. If $text is the empty string and section 0 is replaced, $newText is
5524 * returned.
5525 *
5526 * @param string $mode One of "get" or "replace"
5527 * @param string $newText Replacement text for section data.
5528 * @return string For "get", the extracted section text.
5529 * for "replace", the whole page with the section replaced.
5530 */
5531 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5532 global $wgTitle; # not generally used but removes an ugly failure mode
5533
5534 $magicScopeVariable = $this->lock();
5535 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5536 $outText = '';
5537 $frame = $this->getPreprocessor()->newFrame();
5538
5539 # Process section extraction flags
5540 $flags = 0;
5541 $sectionParts = explode( '-', $sectionId );
5542 $sectionIndex = array_pop( $sectionParts );
5543 foreach ( $sectionParts as $part ) {
5544 if ( $part === 'T' ) {
5545 $flags |= self::PTD_FOR_INCLUSION;
5546 }
5547 }
5548
5549 # Check for empty input
5550 if ( strval( $text ) === '' ) {
5551 # Only sections 0 and T-0 exist in an empty document
5552 if ( $sectionIndex == 0 ) {
5553 if ( $mode === 'get' ) {
5554 return '';
5555 } else {
5556 return $newText;
5557 }
5558 } else {
5559 if ( $mode === 'get' ) {
5560 return $newText;
5561 } else {
5562 return $text;
5563 }
5564 }
5565 }
5566
5567 # Preprocess the text
5568 $root = $this->preprocessToDom( $text, $flags );
5569
5570 # <h> nodes indicate section breaks
5571 # They can only occur at the top level, so we can find them by iterating the root's children
5572 $node = $root->getFirstChild();
5573
5574 # Find the target section
5575 if ( $sectionIndex == 0 ) {
5576 # Section zero doesn't nest, level=big
5577 $targetLevel = 1000;
5578 } else {
5579 while ( $node ) {
5580 if ( $node->getName() === 'h' ) {
5581 $bits = $node->splitHeading();
5582 if ( $bits['i'] == $sectionIndex ) {
5583 $targetLevel = $bits['level'];
5584 break;
5585 }
5586 }
5587 if ( $mode === 'replace' ) {
5588 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5589 }
5590 $node = $node->getNextSibling();
5591 }
5592 }
5593
5594 if ( !$node ) {
5595 # Not found
5596 if ( $mode === 'get' ) {
5597 return $newText;
5598 } else {
5599 return $text;
5600 }
5601 }
5602
5603 # Find the end of the section, including nested sections
5604 do {
5605 if ( $node->getName() === 'h' ) {
5606 $bits = $node->splitHeading();
5607 $curLevel = $bits['level'];
5608 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5609 break;
5610 }
5611 }
5612 if ( $mode === 'get' ) {
5613 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5614 }
5615 $node = $node->getNextSibling();
5616 } while ( $node );
5617
5618 # Write out the remainder (in replace mode only)
5619 if ( $mode === 'replace' ) {
5620 # Output the replacement text
5621 # Add two newlines on -- trailing whitespace in $newText is conventionally
5622 # stripped by the editor, so we need both newlines to restore the paragraph gap
5623 # Only add trailing whitespace if there is newText
5624 if ( $newText != "" ) {
5625 $outText .= $newText . "\n\n";
5626 }
5627
5628 while ( $node ) {
5629 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5630 $node = $node->getNextSibling();
5631 }
5632 }
5633
5634 if ( is_string( $outText ) ) {
5635 # Re-insert stripped tags
5636 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5637 }
5638
5639 return $outText;
5640 }
5641
5642 /**
5643 * This function returns the text of a section, specified by a number ($section).
5644 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5645 * the first section before any such heading (section 0).
5646 *
5647 * If a section contains subsections, these are also returned.
5648 *
5649 * @param string $text Text to look in
5650 * @param string|int $sectionId Section identifier as a number or string
5651 * (e.g. 0, 1 or 'T-1').
5652 * @param string $defaultText Default to return if section is not found
5653 *
5654 * @return string Text of the requested section
5655 */
5656 public function getSection( $text, $sectionId, $defaultText = '' ) {
5657 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5658 }
5659
5660 /**
5661 * This function returns $oldtext after the content of the section
5662 * specified by $section has been replaced with $text. If the target
5663 * section does not exist, $oldtext is returned unchanged.
5664 *
5665 * @param string $oldText Former text of the article
5666 * @param string|int $sectionId Section identifier as a number or string
5667 * (e.g. 0, 1 or 'T-1').
5668 * @param string $newText Replacing text
5669 *
5670 * @return string Modified text
5671 */
5672 public function replaceSection( $oldText, $sectionId, $newText ) {
5673 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5674 }
5675
5676 /**
5677 * Get the ID of the revision we are parsing
5678 *
5679 * @return int|null
5680 */
5681 public function getRevisionId() {
5682 return $this->mRevisionId;
5683 }
5684
5685 /**
5686 * Get the revision object for $this->mRevisionId
5687 *
5688 * @return Revision|null Either a Revision object or null
5689 * @since 1.23 (public since 1.23)
5690 */
5691 public function getRevisionObject() {
5692 if ( !is_null( $this->mRevisionObject ) ) {
5693 return $this->mRevisionObject;
5694 }
5695 if ( is_null( $this->mRevisionId ) ) {
5696 return null;
5697 }
5698
5699 $rev = call_user_func(
5700 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5701 );
5702
5703 # If the parse is for a new revision, then the callback should have
5704 # already been set to force the object and should match mRevisionId.
5705 # If not, try to fetch by mRevisionId for sanity.
5706 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5707 $rev = Revision::newFromId( $this->mRevisionId );
5708 }
5709
5710 $this->mRevisionObject = $rev;
5711
5712 return $this->mRevisionObject;
5713 }
5714
5715 /**
5716 * Get the timestamp associated with the current revision, adjusted for
5717 * the default server-local timestamp
5718 * @return string
5719 */
5720 public function getRevisionTimestamp() {
5721 if ( is_null( $this->mRevisionTimestamp ) ) {
5722 global $wgContLang;
5723
5724 $revObject = $this->getRevisionObject();
5725 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5726
5727 # The cryptic '' timezone parameter tells to use the site-default
5728 # timezone offset instead of the user settings.
5729 # Since this value will be saved into the parser cache, served
5730 # to other users, and potentially even used inside links and such,
5731 # it needs to be consistent for all visitors.
5732 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5733
5734 }
5735 return $this->mRevisionTimestamp;
5736 }
5737
5738 /**
5739 * Get the name of the user that edited the last revision
5740 *
5741 * @return string User name
5742 */
5743 public function getRevisionUser() {
5744 if ( is_null( $this->mRevisionUser ) ) {
5745 $revObject = $this->getRevisionObject();
5746
5747 # if this template is subst: the revision id will be blank,
5748 # so just use the current user's name
5749 if ( $revObject ) {
5750 $this->mRevisionUser = $revObject->getUserText();
5751 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5752 $this->mRevisionUser = $this->getUser()->getName();
5753 }
5754 }
5755 return $this->mRevisionUser;
5756 }
5757
5758 /**
5759 * Get the size of the revision
5760 *
5761 * @return int|null Revision size
5762 */
5763 public function getRevisionSize() {
5764 if ( is_null( $this->mRevisionSize ) ) {
5765 $revObject = $this->getRevisionObject();
5766
5767 # if this variable is subst: the revision id will be blank,
5768 # so just use the parser input size, because the own substituation
5769 # will change the size.
5770 if ( $revObject ) {
5771 $this->mRevisionSize = $revObject->getSize();
5772 } else {
5773 $this->mRevisionSize = $this->mInputSize;
5774 }
5775 }
5776 return $this->mRevisionSize;
5777 }
5778
5779 /**
5780 * Mutator for $mDefaultSort
5781 *
5782 * @param string $sort New value
5783 */
5784 public function setDefaultSort( $sort ) {
5785 $this->mDefaultSort = $sort;
5786 $this->mOutput->setProperty( 'defaultsort', $sort );
5787 }
5788
5789 /**
5790 * Accessor for $mDefaultSort
5791 * Will use the empty string if none is set.
5792 *
5793 * This value is treated as a prefix, so the
5794 * empty string is equivalent to sorting by
5795 * page name.
5796 *
5797 * @return string
5798 */
5799 public function getDefaultSort() {
5800 if ( $this->mDefaultSort !== false ) {
5801 return $this->mDefaultSort;
5802 } else {
5803 return '';
5804 }
5805 }
5806
5807 /**
5808 * Accessor for $mDefaultSort
5809 * Unlike getDefaultSort(), will return false if none is set
5810 *
5811 * @return string|bool
5812 */
5813 public function getCustomDefaultSort() {
5814 return $this->mDefaultSort;
5815 }
5816
5817 private static function getSectionNameFromStrippedText( $text ) {
5818 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5819 $text = Sanitizer::decodeCharReferences( $text );
5820 $text = self::normalizeSectionName( $text );
5821 return $text;
5822 }
5823
5824 private static function makeAnchor( $sectionName ) {
5825 return '#' . Sanitizer::escapeIdForLink( $sectionName );
5826 }
5827
5828 private static function makeLegacyAnchor( $sectionName ) {
5829 global $wgFragmentMode;
5830 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5831 // ForAttribute() and ForLink() are the same for legacy encoding
5832 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
5833 } else {
5834 $id = Sanitizer::escapeIdForLink( $sectionName );
5835 }
5836
5837 return "#$id";
5838 }
5839
5840 /**
5841 * Try to guess the section anchor name based on a wikitext fragment
5842 * presumably extracted from a heading, for example "Header" from
5843 * "== Header ==".
5844 *
5845 * @param string $text
5846 * @return string Anchor (starting with '#')
5847 */
5848 public function guessSectionNameFromWikiText( $text ) {
5849 # Strip out wikitext links(they break the anchor)
5850 $text = $this->stripSectionName( $text );
5851 $sectionName = self::getSectionNameFromStrippedText( $text );
5852 return self::makeAnchor( $sectionName );
5853 }
5854
5855 /**
5856 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5857 * instead, if possible. For use in redirects, since various versions
5858 * of Microsoft browsers interpret Location: headers as something other
5859 * than UTF-8, resulting in breakage.
5860 *
5861 * @param string $text The section name
5862 * @return string Anchor (starting with '#')
5863 */
5864 public function guessLegacySectionNameFromWikiText( $text ) {
5865 # Strip out wikitext links(they break the anchor)
5866 $text = $this->stripSectionName( $text );
5867 $sectionName = self::getSectionNameFromStrippedText( $text );
5868 return self::makeLegacyAnchor( $sectionName );
5869 }
5870
5871 /**
5872 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
5873 * @param string $text Section name (plain text)
5874 * @return string Anchor (starting with '#')
5875 */
5876 public static function guessSectionNameFromStrippedText( $text ) {
5877 $sectionName = self::getSectionNameFromStrippedText( $text );
5878 return self::makeAnchor( $sectionName );
5879 }
5880
5881 /**
5882 * Apply the same normalization as code making links to this section would
5883 *
5884 * @param string $text
5885 * @return string
5886 */
5887 private static function normalizeSectionName( $text ) {
5888 # T90902: ensure the same normalization is applied for IDs as to links
5889 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
5890 try {
5891
5892 $parts = $titleParser->splitTitleString( "#$text" );
5893 } catch ( MalformedTitleException $ex ) {
5894 return $text;
5895 }
5896 return $parts['fragment'];
5897 }
5898
5899 /**
5900 * Strips a text string of wikitext for use in a section anchor
5901 *
5902 * Accepts a text string and then removes all wikitext from the
5903 * string and leaves only the resultant text (i.e. the result of
5904 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5905 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5906 * to create valid section anchors by mimicing the output of the
5907 * parser when headings are parsed.
5908 *
5909 * @param string $text Text string to be stripped of wikitext
5910 * for use in a Section anchor
5911 * @return string Filtered text string
5912 */
5913 public function stripSectionName( $text ) {
5914 # Strip internal link markup
5915 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5916 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5917
5918 # Strip external link markup
5919 # @todo FIXME: Not tolerant to blank link text
5920 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5921 # on how many empty links there are on the page - need to figure that out.
5922 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5923
5924 # Parse wikitext quotes (italics & bold)
5925 $text = $this->doQuotes( $text );
5926
5927 # Strip HTML tags
5928 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5929 return $text;
5930 }
5931
5932 /**
5933 * strip/replaceVariables/unstrip for preprocessor regression testing
5934 *
5935 * @param string $text
5936 * @param Title $title
5937 * @param ParserOptions $options
5938 * @param int $outputType
5939 *
5940 * @return string
5941 */
5942 public function testSrvus( $text, Title $title, ParserOptions $options,
5943 $outputType = self::OT_HTML
5944 ) {
5945 $magicScopeVariable = $this->lock();
5946 $this->startParse( $title, $options, $outputType, true );
5947
5948 $text = $this->replaceVariables( $text );
5949 $text = $this->mStripState->unstripBoth( $text );
5950 $text = Sanitizer::removeHTMLtags( $text );
5951 return $text;
5952 }
5953
5954 /**
5955 * @param string $text
5956 * @param Title $title
5957 * @param ParserOptions $options
5958 * @return string
5959 */
5960 public function testPst( $text, Title $title, ParserOptions $options ) {
5961 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5962 }
5963
5964 /**
5965 * @param string $text
5966 * @param Title $title
5967 * @param ParserOptions $options
5968 * @return string
5969 */
5970 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5971 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5972 }
5973
5974 /**
5975 * Call a callback function on all regions of the given text that are not
5976 * inside strip markers, and replace those regions with the return value
5977 * of the callback. For example, with input:
5978 *
5979 * aaa<MARKER>bbb
5980 *
5981 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5982 * two strings will be replaced with the value returned by the callback in
5983 * each case.
5984 *
5985 * @param string $s
5986 * @param callable $callback
5987 *
5988 * @return string
5989 */
5990 public function markerSkipCallback( $s, $callback ) {
5991 $i = 0;
5992 $out = '';
5993 while ( $i < strlen( $s ) ) {
5994 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5995 if ( $markerStart === false ) {
5996 $out .= call_user_func( $callback, substr( $s, $i ) );
5997 break;
5998 } else {
5999 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6000 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6001 if ( $markerEnd === false ) {
6002 $out .= substr( $s, $markerStart );
6003 break;
6004 } else {
6005 $markerEnd += strlen( self::MARKER_SUFFIX );
6006 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6007 $i = $markerEnd;
6008 }
6009 }
6010 }
6011 return $out;
6012 }
6013
6014 /**
6015 * Remove any strip markers found in the given text.
6016 *
6017 * @param string $text
6018 * @return string
6019 */
6020 public function killMarkers( $text ) {
6021 return $this->mStripState->killMarkers( $text );
6022 }
6023
6024 /**
6025 * Save the parser state required to convert the given half-parsed text to
6026 * HTML. "Half-parsed" in this context means the output of
6027 * recursiveTagParse() or internalParse(). This output has strip markers
6028 * from replaceVariables (extensionSubstitution() etc.), and link
6029 * placeholders from replaceLinkHolders().
6030 *
6031 * Returns an array which can be serialized and stored persistently. This
6032 * array can later be loaded into another parser instance with
6033 * unserializeHalfParsedText(). The text can then be safely incorporated into
6034 * the return value of a parser hook.
6035 *
6036 * @deprecated since 1.31
6037 * @param string $text
6038 *
6039 * @return array
6040 */
6041 public function serializeHalfParsedText( $text ) {
6042 wfDeprecated( __METHOD__, '1.31' );
6043 $data = [
6044 'text' => $text,
6045 'version' => self::HALF_PARSED_VERSION,
6046 'stripState' => $this->mStripState->getSubState( $text ),
6047 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6048 ];
6049 return $data;
6050 }
6051
6052 /**
6053 * Load the parser state given in the $data array, which is assumed to
6054 * have been generated by serializeHalfParsedText(). The text contents is
6055 * extracted from the array, and its markers are transformed into markers
6056 * appropriate for the current Parser instance. This transformed text is
6057 * returned, and can be safely included in the return value of a parser
6058 * hook.
6059 *
6060 * If the $data array has been stored persistently, the caller should first
6061 * check whether it is still valid, by calling isValidHalfParsedText().
6062 *
6063 * @deprecated since 1.31
6064 * @param array $data Serialized data
6065 * @throws MWException
6066 * @return string
6067 */
6068 public function unserializeHalfParsedText( $data ) {
6069 wfDeprecated( __METHOD__, '1.31' );
6070 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6071 throw new MWException( __METHOD__ . ': invalid version' );
6072 }
6073
6074 # First, extract the strip state.
6075 $texts = [ $data['text'] ];
6076 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6077
6078 # Now renumber links
6079 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6080
6081 # Should be good to go.
6082 return $texts[0];
6083 }
6084
6085 /**
6086 * Returns true if the given array, presumed to be generated by
6087 * serializeHalfParsedText(), is compatible with the current version of the
6088 * parser.
6089 *
6090 * @deprecated since 1.31
6091 * @param array $data
6092 *
6093 * @return bool
6094 */
6095 public function isValidHalfParsedText( $data ) {
6096 wfDeprecated( __METHOD__, '1.31' );
6097 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6098 }
6099
6100 /**
6101 * Parsed a width param of imagelink like 300px or 200x300px
6102 *
6103 * @param string $value
6104 * @param bool $parseHeight
6105 *
6106 * @return array
6107 * @since 1.20
6108 */
6109 public static function parseWidthParam( $value, $parseHeight = true ) {
6110 $parsedWidthParam = [];
6111 if ( $value === '' ) {
6112 return $parsedWidthParam;
6113 }
6114 $m = [];
6115 # (T15500) In both cases (width/height and width only),
6116 # permit trailing "px" for backward compatibility.
6117 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6118 $width = intval( $m[1] );
6119 $height = intval( $m[2] );
6120 $parsedWidthParam['width'] = $width;
6121 $parsedWidthParam['height'] = $height;
6122 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6123 $width = intval( $value );
6124 $parsedWidthParam['width'] = $width;
6125 }
6126 return $parsedWidthParam;
6127 }
6128
6129 /**
6130 * Lock the current instance of the parser.
6131 *
6132 * This is meant to stop someone from calling the parser
6133 * recursively and messing up all the strip state.
6134 *
6135 * @throws MWException If parser is in a parse
6136 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6137 */
6138 protected function lock() {
6139 if ( $this->mInParse ) {
6140 throw new MWException( "Parser state cleared while parsing. "
6141 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6142 }
6143
6144 // Save the backtrace when locking, so that if some code tries locking again,
6145 // we can print the lock owner's backtrace for easier debugging
6146 $e = new Exception;
6147 $this->mInParse = $e->getTraceAsString();
6148
6149 $recursiveCheck = new ScopedCallback( function () {
6150 $this->mInParse = false;
6151 } );
6152
6153 return $recursiveCheck;
6154 }
6155
6156 /**
6157 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6158 *
6159 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6160 * or if there is more than one <p/> tag in the input HTML.
6161 *
6162 * @param string $html
6163 * @return string
6164 * @since 1.24
6165 */
6166 public static function stripOuterParagraph( $html ) {
6167 $m = [];
6168 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6169 if ( strpos( $m[1], '</p>' ) === false ) {
6170 $html = $m[1];
6171 }
6172 }
6173
6174 return $html;
6175 }
6176
6177 /**
6178 * Return this parser if it is not doing anything, otherwise
6179 * get a fresh parser. You can use this method by doing
6180 * $myParser = $wgParser->getFreshParser(), or more simply
6181 * $wgParser->getFreshParser()->parse( ... );
6182 * if you're unsure if $wgParser is safe to use.
6183 *
6184 * @since 1.24
6185 * @return Parser A parser object that is not parsing anything
6186 */
6187 public function getFreshParser() {
6188 global $wgParserConf;
6189 if ( $this->mInParse ) {
6190 return new $wgParserConf['class']( $wgParserConf );
6191 } else {
6192 return $this;
6193 }
6194 }
6195
6196 /**
6197 * Set's up the PHP implementation of OOUI for use in this request
6198 * and instructs OutputPage to enable OOUI for itself.
6199 *
6200 * @since 1.26
6201 */
6202 public function enableOOUI() {
6203 OutputPage::setupOOUI();
6204 $this->mOutput->setEnableOOUI( true );
6205 }
6206 }
Wiklou, le Wiki du Wiklou