dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Deprecate Parser::replaceLinkHolders / replaceLinkHoldersText
[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\BadFileLookup;
24 use MediaWiki\Config\ServiceOptions;
25 use MediaWiki\Linker\LinkRenderer;
26 use MediaWiki\Linker\LinkRendererFactory;
27 use MediaWiki\Linker\LinkTarget;
28 use MediaWiki\MediaWikiServices;
29 use MediaWiki\Special\SpecialPageFactory;
30 use Psr\Log\NullLogger;
31 use Wikimedia\ScopedCallback;
32 use Psr\Log\LoggerInterface;
33
34 /**
35 * @defgroup Parser Parser
36 */
37
38 /**
39 * PHP Parser - Processes wiki markup (which uses a more user-friendly
40 * syntax, such as "[[link]]" for making links), and provides a one-way
41 * transformation of that wiki markup it into (X)HTML output / markup
42 * (which in turn the browser understands, and can display).
43 *
44 * There are seven main entry points into the Parser class:
45 *
46 * - Parser::parse()
47 * produces HTML output
48 * - Parser::preSaveTransform()
49 * produces altered wiki markup
50 * - Parser::preprocess()
51 * removes HTML comments and expands templates
52 * - Parser::cleanSig() and Parser::cleanSigInSig()
53 * cleans a signature before saving it to preferences
54 * - Parser::getSection()
55 * return the content of a section from an article for section editing
56 * - Parser::replaceSection()
57 * replaces a section by number inside an article
58 * - Parser::getPreloadText()
59 * removes <noinclude> sections and <includeonly> tags
60 *
61 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
62 *
63 * @par Settings:
64 * $wgNamespacesWithSubpages
65 *
66 * @par Settings only within ParserOptions:
67 * $wgAllowExternalImages
68 * $wgAllowSpecialInclusion
69 * $wgInterwikiMagic
70 * $wgMaxArticleSize
71 *
72 * @ingroup Parser
73 */
74 class Parser {
75 /**
76 * Update this version number when the ParserOutput format
77 * changes in an incompatible way, so the parser cache
78 * can automatically discard old data.
79 */
80 const VERSION = '1.6.4';
81
82 /**
83 * Update this version number when the output of serialiseHalfParsedText()
84 * changes in an incompatible way
85 */
86 const HALF_PARSED_VERSION = 2;
87
88 # Flags for Parser::setFunctionHook
89 const SFH_NO_HASH = 1;
90 const SFH_OBJECT_ARGS = 2;
91
92 # Constants needed for external link processing
93 # Everything except bracket, space, or control characters
94 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
95 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
96 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
97 # uses to replace invalid HTML characters.
98 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
99 # Simplified expression to match an IPv4 or IPv6 address, or
100 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
101 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
102 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
103 // phpcs:ignore Generic.Files.LineLength
104 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
105 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
106
107 # Regular expression for a non-newline space
108 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
109
110 # Flags for preprocessToDom
111 const PTD_FOR_INCLUSION = 1;
112
113 # Allowed values for $this->mOutputType
114 # Parameter to startExternalParse().
115 const OT_HTML = 1; # like parse()
116 const OT_WIKI = 2; # like preSaveTransform()
117 const OT_PREPROCESS = 3; # like preprocess()
118 const OT_MSG = 3;
119 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
120
121 /**
122 * @var string Prefix and suffix for temporary replacement strings
123 * for the multipass parser.
124 *
125 * \x7f should never appear in input as it's disallowed in XML.
126 * Using it at the front also gives us a little extra robustness
127 * since it shouldn't match when butted up against identifier-like
128 * string constructs.
129 *
130 * Must not consist of all title characters, or else it will change
131 * the behavior of <nowiki> in a link.
132 *
133 * Must have a character that needs escaping in attributes, otherwise
134 * someone could put a strip marker in an attribute, to get around
135 * escaping quote marks, and break out of the attribute. Thus we add
136 * `'".
137 */
138 const MARKER_SUFFIX = "-QINU`\"'\x7f";
139 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
140
141 # Markers used for wrapping the table of contents
142 const TOC_START = '<mw:toc>';
143 const TOC_END = '</mw:toc>';
144
145 /** @var int Assume that no output will later be saved this many seconds after parsing */
146 const MAX_TTS = 900;
147
148 # Persistent:
149 public $mTagHooks = [];
150 public $mTransparentTagHooks = [];
151 public $mFunctionHooks = [];
152 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
153 public $mFunctionTagHooks = [];
154 public $mStripList = [];
155 public $mDefaultStripList = [];
156 public $mVarCache = [];
157 public $mImageParams = [];
158 public $mImageParamsMagicArray = [];
159 public $mMarkerIndex = 0;
160 /**
161 * @var bool Whether firstCallInit still needs to be called
162 */
163 public $mFirstCall = true;
164
165 # Initialised by initializeVariables()
166
167 /**
168 * @var MagicWordArray
169 */
170 public $mVariables;
171
172 /**
173 * @var MagicWordArray
174 */
175 public $mSubstWords;
176
177 /**
178 * @deprecated since 1.34, there should be no need to use this
179 * @var array
180 */
181 public $mConf;
182
183 # Initialised in constructor
184 public $mExtLinkBracketedRegex, $mUrlProtocols;
185
186 # Initialized in getPreprocessor()
187 /** @var Preprocessor */
188 public $mPreprocessor;
189
190 # Cleared with clearState():
191 /**
192 * @var ParserOutput
193 */
194 public $mOutput;
195 public $mAutonumber;
196
197 /**
198 * @var StripState
199 */
200 public $mStripState;
201
202 public $mIncludeCount;
203 /**
204 * @var LinkHolderArray
205 */
206 public $mLinkHolders;
207
208 public $mLinkID;
209 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
210 public $mDefaultSort;
211 public $mTplRedirCache, $mHeadings, $mDoubleUnderscores;
212 public $mExpensiveFunctionCount; # number of expensive parser function calls
213 public $mShowToc, $mForceTocPosition;
214 /** @var array */
215 public $mTplDomCache;
216
217 /**
218 * @var User
219 */
220 public $mUser; # User object; only used when doing pre-save transform
221
222 # Temporary
223 # These are variables reset at least once per parse regardless of $clearState
224
225 /**
226 * @var ParserOptions
227 */
228 public $mOptions;
229
230 /**
231 * Since 1.34, leaving `mTitle` uninitialized or setting `mTitle` to
232 * `null` is deprecated.
233 *
234 * @internal
235 * @var Title|null
236 */
237 public $mTitle; # Title context, used for self-link rendering and similar things
238 public $mOutputType; # Output type, one of the OT_xxx constants
239 public $ot; # Shortcut alias, see setOutputType()
240 public $mRevisionObject; # The revision object of the specified revision ID
241 public $mRevisionId; # ID to display in {{REVISIONID}} tags
242 public $mRevisionTimestamp; # The timestamp of the specified revision ID
243 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
244 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
245 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
246 public $mInputSize = false; # For {{PAGESIZE}} on current page.
247
248 /**
249 * @var array Array with the language name of each language link (i.e. the
250 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
251 * duplicate language links to the ParserOutput.
252 */
253 public $mLangLinkLanguages;
254
255 /**
256 * @var MapCacheLRU|null
257 * @since 1.24
258 *
259 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
260 */
261 public $currentRevisionCache;
262
263 /**
264 * @var bool|string Recursive call protection.
265 * This variable should be treated as if it were private.
266 */
267 public $mInParse = false;
268
269 /** @var SectionProfiler */
270 protected $mProfiler;
271
272 /**
273 * @var LinkRenderer
274 */
275 protected $mLinkRenderer;
276
277 /** @var MagicWordFactory */
278 private $magicWordFactory;
279
280 /** @var Language */
281 private $contLang;
282
283 /** @var ParserFactory */
284 private $factory;
285
286 /** @var SpecialPageFactory */
287 private $specialPageFactory;
288
289 /**
290 * This is called $svcOptions instead of $options like elsewhere to avoid confusion with
291 * $mOptions, which is public and widely used, and also with the local variable $options used
292 * for ParserOptions throughout this file.
293 *
294 * @var ServiceOptions
295 */
296 private $svcOptions;
297
298 /** @var LinkRendererFactory */
299 private $linkRendererFactory;
300
301 /** @var NamespaceInfo */
302 private $nsInfo;
303
304 /** @var LoggerInterface */
305 private $logger;
306
307 /** @var BadFileLookup */
308 private $badFileLookup;
309
310 /**
311 * TODO Make this a const when HHVM support is dropped (T192166)
312 *
313 * @var array
314 * @since 1.33
315 */
316 public static $constructorOptions = [
317 // See $wgParserConf documentation
318 'class',
319 'preprocessorClass',
320 // See documentation for the corresponding config options
321 'ArticlePath',
322 'EnableScaryTranscluding',
323 'ExtraInterlanguageLinkPrefixes',
324 'FragmentMode',
325 'LanguageCode',
326 'MaxSigChars',
327 'MaxTocLevel',
328 'MiserMode',
329 'ScriptPath',
330 'Server',
331 'ServerName',
332 'ShowHostnames',
333 'Sitename',
334 'StylePath',
335 'TranscludeCacheExpiry',
336 ];
337
338 /**
339 * Constructing parsers directly is deprecated! Use a ParserFactory.
340 *
341 * @param ServiceOptions|null $svcOptions
342 * @param MagicWordFactory|null $magicWordFactory
343 * @param Language|null $contLang Content language
344 * @param ParserFactory|null $factory
345 * @param string|null $urlProtocols As returned from wfUrlProtocols()
346 * @param SpecialPageFactory|null $spFactory
347 * @param LinkRendererFactory|null $linkRendererFactory
348 * @param NamespaceInfo|null $nsInfo
349 * @param LoggerInterface|null $logger
350 * @param BadFileLookup|null $badFileLookup
351 */
352 public function __construct(
353 $svcOptions = null,
354 MagicWordFactory $magicWordFactory = null,
355 Language $contLang = null,
356 ParserFactory $factory = null,
357 $urlProtocols = null,
358 SpecialPageFactory $spFactory = null,
359 $linkRendererFactory = null,
360 $nsInfo = null,
361 $logger = null,
362 BadFileLookup $badFileLookup = null
363 ) {
364 if ( !$svcOptions || is_array( $svcOptions ) ) {
365 // Pre-1.34 calling convention is the first parameter is just ParserConf, the seventh is
366 // Config, and the eighth is LinkRendererFactory.
367 $this->mConf = (array)$svcOptions;
368 if ( empty( $this->mConf['class'] ) ) {
369 $this->mConf['class'] = self::class;
370 }
371 if ( empty( $this->mConf['preprocessorClass'] ) ) {
372 $this->mConf['preprocessorClass'] = self::getDefaultPreprocessorClass();
373 }
374 $this->svcOptions = new ServiceOptions( self::$constructorOptions,
375 $this->mConf, func_num_args() > 6
376 ? func_get_arg( 6 ) : MediaWikiServices::getInstance()->getMainConfig()
377 );
378 $linkRendererFactory = func_num_args() > 7 ? func_get_arg( 7 ) : null;
379 $nsInfo = func_num_args() > 8 ? func_get_arg( 8 ) : null;
380 } else {
381 // New calling convention
382 $svcOptions->assertRequiredOptions( self::$constructorOptions );
383 // $this->mConf is public, so we'll keep those two options there as well for
384 // compatibility until it's removed
385 $this->mConf = [
386 'class' => $svcOptions->get( 'class' ),
387 'preprocessorClass' => $svcOptions->get( 'preprocessorClass' ),
388 ];
389 $this->svcOptions = $svcOptions;
390 }
391
392 $this->mUrlProtocols = $urlProtocols ?? wfUrlProtocols();
393 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
394 self::EXT_LINK_ADDR .
395 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
396
397 $this->magicWordFactory = $magicWordFactory ??
398 MediaWikiServices::getInstance()->getMagicWordFactory();
399
400 $this->contLang = $contLang ?? MediaWikiServices::getInstance()->getContentLanguage();
401
402 $this->factory = $factory ?? MediaWikiServices::getInstance()->getParserFactory();
403 $this->specialPageFactory = $spFactory ??
404 MediaWikiServices::getInstance()->getSpecialPageFactory();
405 $this->linkRendererFactory = $linkRendererFactory ??
406 MediaWikiServices::getInstance()->getLinkRendererFactory();
407 $this->nsInfo = $nsInfo ?? MediaWikiServices::getInstance()->getNamespaceInfo();
408 $this->logger = $logger ?: new NullLogger();
409 $this->badFileLookup = $badFileLookup ??
410 MediaWikiServices::getInstance()->getBadFileLookup();
411 }
412
413 /**
414 * Reduce memory usage to reduce the impact of circular references
415 */
416 public function __destruct() {
417 if ( isset( $this->mLinkHolders ) ) {
418 // @phan-suppress-next-line PhanTypeObjectUnsetDeclaredProperty
419 unset( $this->mLinkHolders );
420 }
421 // @phan-suppress-next-line PhanTypeSuspiciousNonTraversableForeach
422 foreach ( $this as $name => $value ) {
423 unset( $this->$name );
424 }
425 }
426
427 /**
428 * Allow extensions to clean up when the parser is cloned
429 */
430 public function __clone() {
431 $this->mInParse = false;
432
433 // T58226: When you create a reference "to" an object field, that
434 // makes the object field itself be a reference too (until the other
435 // reference goes out of scope). When cloning, any field that's a
436 // reference is copied as a reference in the new object. Both of these
437 // are defined PHP5 behaviors, as inconvenient as it is for us when old
438 // hooks from PHP4 days are passing fields by reference.
439 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
440 // Make a non-reference copy of the field, then rebind the field to
441 // reference the new copy.
442 $tmp = $this->$k;
443 $this->$k =& $tmp;
444 unset( $tmp );
445 }
446
447 Hooks::run( 'ParserCloned', [ $this ] );
448 }
449
450 /**
451 * Which class should we use for the preprocessor if not otherwise specified?
452 *
453 * @since 1.34
454 * @deprecated since 1.34, removing configurability of preprocessor
455 * @return string
456 */
457 public static function getDefaultPreprocessorClass() {
458 return Preprocessor_Hash::class;
459 }
460
461 /**
462 * Do various kinds of initialisation on the first call of the parser
463 */
464 public function firstCallInit() {
465 if ( !$this->mFirstCall ) {
466 return;
467 }
468 $this->mFirstCall = false;
469
470 CoreParserFunctions::register( $this );
471 CoreTagHooks::register( $this );
472 $this->initializeVariables();
473
474 // Avoid PHP 7.1 warning from passing $this by reference
475 $parser = $this;
476 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
477 }
478
479 /**
480 * Clear Parser state
481 *
482 * @private
483 */
484 public function clearState() {
485 $this->firstCallInit();
486 $this->resetOutput();
487 $this->mAutonumber = 0;
488 $this->mIncludeCount = [];
489 $this->mLinkHolders = new LinkHolderArray( $this );
490 $this->mLinkID = 0;
491 $this->mRevisionObject = $this->mRevisionTimestamp =
492 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
493 $this->mVarCache = [];
494 $this->mUser = null;
495 $this->mLangLinkLanguages = [];
496 $this->currentRevisionCache = null;
497
498 $this->mStripState = new StripState( $this );
499
500 # Clear these on every parse, T6549
501 $this->mTplRedirCache = $this->mTplDomCache = [];
502
503 $this->mShowToc = true;
504 $this->mForceTocPosition = false;
505 $this->mIncludeSizes = [
506 'post-expand' => 0,
507 'arg' => 0,
508 ];
509 $this->mPPNodeCount = 0;
510 $this->mGeneratedPPNodeCount = 0;
511 $this->mHighestExpansionDepth = 0;
512 $this->mDefaultSort = false;
513 $this->mHeadings = [];
514 $this->mDoubleUnderscores = [];
515 $this->mExpensiveFunctionCount = 0;
516
517 # Fix cloning
518 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
519 $this->mPreprocessor = null;
520 }
521
522 $this->mProfiler = new SectionProfiler();
523
524 // Avoid PHP 7.1 warning from passing $this by reference
525 $parser = $this;
526 Hooks::run( 'ParserClearState', [ &$parser ] );
527 }
528
529 /**
530 * Reset the ParserOutput
531 */
532 public function resetOutput() {
533 $this->mOutput = new ParserOutput;
534 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
535 }
536
537 /**
538 * Convert wikitext to HTML
539 * Do not call this function recursively.
540 *
541 * @param string $text Text we want to parse
542 * @param-taint $text escapes_htmlnoent
543 * @param Title $title
544 * @param ParserOptions $options
545 * @param bool $linestart
546 * @param bool $clearState
547 * @param int|null $revid ID of the revision being rendered. This is used to render
548 * REVISION* magic words. 0 means that any current revision will be used. Null means
549 * that {{REVISIONID}}/{{REVISIONUSER}} will be empty and {{REVISIONTIMESTAMP}} will
550 * use the current timestamp.
551 * @return ParserOutput A ParserOutput
552 * @return-taint escaped
553 */
554 public function parse(
555 $text, Title $title, ParserOptions $options,
556 $linestart = true, $clearState = true, $revid = null
557 ) {
558 if ( $clearState ) {
559 // We use U+007F DELETE to construct strip markers, so we have to make
560 // sure that this character does not occur in the input text.
561 $text = strtr( $text, "\x7f", "?" );
562 $magicScopeVariable = $this->lock();
563 }
564 // Strip U+0000 NULL (T159174)
565 $text = str_replace( "\000", '', $text );
566
567 $this->startParse( $title, $options, self::OT_HTML, $clearState );
568
569 $this->currentRevisionCache = null;
570 $this->mInputSize = strlen( $text );
571 if ( $this->mOptions->getEnableLimitReport() ) {
572 $this->mOutput->resetParseStartTime();
573 }
574
575 $oldRevisionId = $this->mRevisionId;
576 $oldRevisionObject = $this->mRevisionObject;
577 $oldRevisionTimestamp = $this->mRevisionTimestamp;
578 $oldRevisionUser = $this->mRevisionUser;
579 $oldRevisionSize = $this->mRevisionSize;
580 if ( $revid !== null ) {
581 $this->mRevisionId = $revid;
582 $this->mRevisionObject = null;
583 $this->mRevisionTimestamp = null;
584 $this->mRevisionUser = null;
585 $this->mRevisionSize = null;
586 }
587
588 // Avoid PHP 7.1 warning from passing $this by reference
589 $parser = $this;
590 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
591 # No more strip!
592 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
593 $text = $this->internalParse( $text );
594 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
595
596 $text = $this->internalParseHalfParsed( $text, true, $linestart );
597
598 /**
599 * A converted title will be provided in the output object if title and
600 * content conversion are enabled, the article text does not contain
601 * a conversion-suppressing double-underscore tag, and no
602 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
603 * automatic link conversion.
604 */
605 if ( !( $options->getDisableTitleConversion()
606 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
607 || isset( $this->mDoubleUnderscores['notitleconvert'] )
608 || $this->mOutput->getDisplayTitle() !== false )
609 ) {
610 $convruletitle = $this->getTargetLanguage()->getConvRuleTitle();
611 if ( $convruletitle ) {
612 $this->mOutput->setTitleText( $convruletitle );
613 } else {
614 $titleText = $this->getTargetLanguage()->convertTitle( $title );
615 $this->mOutput->setTitleText( $titleText );
616 }
617 }
618
619 # Compute runtime adaptive expiry if set
620 $this->mOutput->finalizeAdaptiveCacheExpiry();
621
622 # Warn if too many heavyweight parser functions were used
623 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
624 $this->limitationWarn( 'expensive-parserfunction',
625 $this->mExpensiveFunctionCount,
626 $this->mOptions->getExpensiveParserFunctionLimit()
627 );
628 }
629
630 # Information on limits, for the benefit of users who try to skirt them
631 if ( $this->mOptions->getEnableLimitReport() ) {
632 $text .= $this->makeLimitReport();
633 }
634
635 # Wrap non-interface parser output in a <div> so it can be targeted
636 # with CSS (T37247)
637 $class = $this->mOptions->getWrapOutputClass();
638 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
639 $this->mOutput->addWrapperDivClass( $class );
640 }
641
642 $this->mOutput->setText( $text );
643
644 $this->mRevisionId = $oldRevisionId;
645 $this->mRevisionObject = $oldRevisionObject;
646 $this->mRevisionTimestamp = $oldRevisionTimestamp;
647 $this->mRevisionUser = $oldRevisionUser;
648 $this->mRevisionSize = $oldRevisionSize;
649 $this->mInputSize = false;
650 $this->currentRevisionCache = null;
651
652 return $this->mOutput;
653 }
654
655 /**
656 * Set the limit report data in the current ParserOutput, and return the
657 * limit report HTML comment.
658 *
659 * @return string
660 */
661 protected function makeLimitReport() {
662 $maxIncludeSize = $this->mOptions->getMaxIncludeSize();
663
664 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
665 if ( $cpuTime !== null ) {
666 $this->mOutput->setLimitReportData( 'limitreport-cputime',
667 sprintf( "%.3f", $cpuTime )
668 );
669 }
670
671 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
672 $this->mOutput->setLimitReportData( 'limitreport-walltime',
673 sprintf( "%.3f", $wallTime )
674 );
675
676 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
677 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
678 );
679 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
680 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
681 );
682 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
683 [ $this->mIncludeSizes['post-expand'], $maxIncludeSize ]
684 );
685 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
686 [ $this->mIncludeSizes['arg'], $maxIncludeSize ]
687 );
688 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
689 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
690 );
691 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
692 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
693 );
694
695 foreach ( $this->mStripState->getLimitReport() as list( $key, $value ) ) {
696 $this->mOutput->setLimitReportData( $key, $value );
697 }
698
699 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
700
701 $limitReport = "NewPP limit report\n";
702 if ( $this->svcOptions->get( 'ShowHostnames' ) ) {
703 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
704 }
705 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
706 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
707 $limitReport .= 'Dynamic content: ' .
708 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
709 "\n";
710 $limitReport .= 'Complications: [' . implode( ', ', $this->mOutput->getAllFlags() ) . "]\n";
711
712 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
713 if ( Hooks::run( 'ParserLimitReportFormat',
714 [ $key, &$value, &$limitReport, false, false ]
715 ) ) {
716 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
717 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
718 ->inLanguage( 'en' )->useDatabase( false );
719 if ( !$valueMsg->exists() ) {
720 $valueMsg = new RawMessage( '$1' );
721 }
722 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
723 $valueMsg->params( $value );
724 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
725 }
726 }
727 }
728 // Since we're not really outputting HTML, decode the entities and
729 // then re-encode the things that need hiding inside HTML comments.
730 $limitReport = htmlspecialchars_decode( $limitReport );
731
732 // Sanitize for comment. Note '‐' in the replacement is U+2010,
733 // which looks much like the problematic '-'.
734 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
735 $text = "\n<!-- \n$limitReport-->\n";
736
737 // Add on template profiling data in human/machine readable way
738 $dataByFunc = $this->mProfiler->getFunctionStats();
739 uasort( $dataByFunc, function ( $a, $b ) {
740 return $b['real'] <=> $a['real']; // descending order
741 } );
742 $profileReport = [];
743 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
744 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
745 $item['%real'], $item['real'], $item['calls'],
746 htmlspecialchars( $item['name'] ) );
747 }
748 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
749 $text .= implode( "\n", $profileReport ) . "\n-->\n";
750
751 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
752
753 // Add other cache related metadata
754 if ( $this->svcOptions->get( 'ShowHostnames' ) ) {
755 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
756 }
757 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
758 $this->mOutput->getCacheTime() );
759 $this->mOutput->setLimitReportData( 'cachereport-ttl',
760 $this->mOutput->getCacheExpiry() );
761 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
762 $this->mOutput->hasDynamicContent() );
763
764 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
765 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
766 $this->mTitle->getPrefixedDBkey() );
767 }
768 return $text;
769 }
770
771 /**
772 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
773 * can be called from an extension tag hook.
774 *
775 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
776 * instead, which means that lists and links have not been fully parsed yet,
777 * and strip markers are still present.
778 *
779 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
780 *
781 * Use this function if you're a parser tag hook and you want to parse
782 * wikitext before or after applying additional transformations, and you
783 * intend to *return the result as hook output*, which will cause it to go
784 * through the rest of parsing process automatically.
785 *
786 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
787 * $text are not expanded
788 *
789 * @param string $text Text extension wants to have parsed
790 * @param-taint $text escapes_htmlnoent
791 * @param bool|PPFrame $frame The frame to use for expanding any template variables
792 * @return string UNSAFE half-parsed HTML
793 * @return-taint escaped
794 */
795 public function recursiveTagParse( $text, $frame = false ) {
796 // Avoid PHP 7.1 warning from passing $this by reference
797 $parser = $this;
798 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
799 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
800 $text = $this->internalParse( $text, false, $frame );
801 return $text;
802 }
803
804 /**
805 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
806 * point can be called from an extension tag hook.
807 *
808 * The output of this function is fully-parsed HTML that is safe for output.
809 * If you're a parser tag hook, you might want to use recursiveTagParse()
810 * instead.
811 *
812 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
813 * $text are not expanded
814 *
815 * @since 1.25
816 *
817 * @param string $text Text extension wants to have parsed
818 * @param-taint $text escapes_htmlnoent
819 * @param bool|PPFrame $frame The frame to use for expanding any template variables
820 * @return string Fully parsed HTML
821 * @return-taint escaped
822 */
823 public function recursiveTagParseFully( $text, $frame = false ) {
824 $text = $this->recursiveTagParse( $text, $frame );
825 $text = $this->internalParseHalfParsed( $text, false );
826 return $text;
827 }
828
829 /**
830 * Expand templates and variables in the text, producing valid, static wikitext.
831 * Also removes comments.
832 * Do not call this function recursively.
833 * @param string $text
834 * @param Title|null $title
835 * @param ParserOptions $options
836 * @param int|null $revid
837 * @param bool|PPFrame $frame
838 * @return mixed|string
839 */
840 public function preprocess( $text, Title $title = null,
841 ParserOptions $options, $revid = null, $frame = false
842 ) {
843 $magicScopeVariable = $this->lock();
844 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
845 if ( $revid !== null ) {
846 $this->mRevisionId = $revid;
847 }
848 // Avoid PHP 7.1 warning from passing $this by reference
849 $parser = $this;
850 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
851 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
852 $text = $this->replaceVariables( $text, $frame );
853 $text = $this->mStripState->unstripBoth( $text );
854 return $text;
855 }
856
857 /**
858 * Recursive parser entry point that can be called from an extension tag
859 * hook.
860 *
861 * @param string $text Text to be expanded
862 * @param bool|PPFrame $frame The frame to use for expanding any template variables
863 * @return string
864 * @since 1.19
865 */
866 public function recursivePreprocess( $text, $frame = false ) {
867 $text = $this->replaceVariables( $text, $frame );
868 $text = $this->mStripState->unstripBoth( $text );
869 return $text;
870 }
871
872 /**
873 * Process the wikitext for the "?preload=" feature. (T7210)
874 *
875 * "<noinclude>", "<includeonly>" etc. are parsed as for template
876 * transclusion, comments, templates, arguments, tags hooks and parser
877 * functions are untouched.
878 *
879 * @param string $text
880 * @param Title $title
881 * @param ParserOptions $options
882 * @param array $params
883 * @return string
884 */
885 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
886 $msg = new RawMessage( $text );
887 $text = $msg->params( $params )->plain();
888
889 # Parser (re)initialisation
890 $magicScopeVariable = $this->lock();
891 $this->startParse( $title, $options, self::OT_PLAIN, true );
892
893 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
894 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
895 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
896 $text = $this->mStripState->unstripBoth( $text );
897 return $text;
898 }
899
900 /**
901 * Set the current user.
902 * Should only be used when doing pre-save transform.
903 *
904 * @param User|null $user User object or null (to reset)
905 */
906 public function setUser( $user ) {
907 $this->mUser = $user;
908 }
909
910 /**
911 * Set the context title
912 *
913 * @param Title|null $t
914 */
915 public function setTitle( Title $t = null ) {
916 if ( !$t ) {
917 $t = Title::makeTitle( NS_SPECIAL, 'Badtitle/Parser' );
918 }
919
920 if ( $t->hasFragment() ) {
921 # Strip the fragment to avoid various odd effects
922 $this->mTitle = $t->createFragmentTarget( '' );
923 } else {
924 $this->mTitle = $t;
925 }
926 }
927
928 /**
929 * Accessor for the Title object
930 *
931 * Since 1.34, leaving `mTitle` uninitialized as `null` is deprecated.
932 *
933 * @return Title|null
934 */
935 public function getTitle() : ?Title {
936 if ( $this->mTitle === null ) {
937 wfDeprecated( 'Parser title should never be null', '1.34' );
938 }
939 return $this->mTitle;
940 }
941
942 /**
943 * Accessor/mutator for the Title object
944 *
945 * @param Title|null $x Title object or null to just get the current one
946 * @return Title|null
947 */
948 public function Title( Title $x = null ) : ?Title {
949 return wfSetVar( $this->mTitle, $x );
950 }
951
952 /**
953 * Set the output type
954 *
955 * @param int $ot New value
956 */
957 public function setOutputType( $ot ) {
958 $this->mOutputType = $ot;
959 # Shortcut alias
960 $this->ot = [
961 'html' => $ot == self::OT_HTML,
962 'wiki' => $ot == self::OT_WIKI,
963 'pre' => $ot == self::OT_PREPROCESS,
964 'plain' => $ot == self::OT_PLAIN,
965 ];
966 }
967
968 /**
969 * Accessor/mutator for the output type
970 *
971 * @param int|null $x New value or null to just get the current one
972 * @return int
973 */
974 public function OutputType( $x = null ) {
975 return wfSetVar( $this->mOutputType, $x );
976 }
977
978 /**
979 * Get the ParserOutput object
980 *
981 * @return ParserOutput
982 */
983 public function getOutput() {
984 return $this->mOutput;
985 }
986
987 /**
988 * Get the ParserOptions object
989 *
990 * @return ParserOptions
991 */
992 public function getOptions() {
993 return $this->mOptions;
994 }
995
996 /**
997 * Accessor/mutator for the ParserOptions object
998 *
999 * @param ParserOptions|null $x New value or null to just get the current one
1000 * @return ParserOptions Current ParserOptions object
1001 */
1002 public function Options( $x = null ) {
1003 return wfSetVar( $this->mOptions, $x );
1004 }
1005
1006 /**
1007 * @return int
1008 */
1009 public function nextLinkID() {
1010 return $this->mLinkID++;
1011 }
1012
1013 /**
1014 * @param int $id
1015 */
1016 public function setLinkID( $id ) {
1017 $this->mLinkID = $id;
1018 }
1019
1020 /**
1021 * Get a language object for use in parser functions such as {{FORMATNUM:}}
1022 * @return Language
1023 */
1024 public function getFunctionLang() {
1025 return $this->getTargetLanguage();
1026 }
1027
1028 /**
1029 * Get the target language for the content being parsed. This is usually the
1030 * language that the content is in.
1031 *
1032 * @since 1.19
1033 *
1034 * @throws MWException
1035 * @return Language
1036 */
1037 public function getTargetLanguage() {
1038 $target = $this->mOptions->getTargetLanguage();
1039
1040 if ( $target !== null ) {
1041 return $target;
1042 } elseif ( $this->mOptions->getInterfaceMessage() ) {
1043 return $this->mOptions->getUserLangObj();
1044 } elseif ( is_null( $this->mTitle ) ) {
1045 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
1046 }
1047
1048 return $this->mTitle->getPageLanguage();
1049 }
1050
1051 /**
1052 * Get the language object for language conversion
1053 * @deprecated since 1.32, just use getTargetLanguage()
1054 * @return Language|null
1055 */
1056 public function getConverterLanguage() {
1057 return $this->getTargetLanguage();
1058 }
1059
1060 /**
1061 * Get a User object either from $this->mUser, if set, or from the
1062 * ParserOptions object otherwise
1063 *
1064 * @return User
1065 */
1066 public function getUser() {
1067 if ( !is_null( $this->mUser ) ) {
1068 return $this->mUser;
1069 }
1070 return $this->mOptions->getUser();
1071 }
1072
1073 /**
1074 * Get a preprocessor object
1075 *
1076 * @return Preprocessor
1077 */
1078 public function getPreprocessor() {
1079 if ( !isset( $this->mPreprocessor ) ) {
1080 $class = $this->svcOptions->get( 'preprocessorClass' );
1081 $this->mPreprocessor = new $class( $this );
1082 }
1083 return $this->mPreprocessor;
1084 }
1085
1086 /**
1087 * Get a LinkRenderer instance to make links with
1088 *
1089 * @since 1.28
1090 * @return LinkRenderer
1091 */
1092 public function getLinkRenderer() {
1093 // XXX We make the LinkRenderer with current options and then cache it forever
1094 if ( !$this->mLinkRenderer ) {
1095 $this->mLinkRenderer = $this->linkRendererFactory->create();
1096 $this->mLinkRenderer->setStubThreshold(
1097 $this->getOptions()->getStubThreshold()
1098 );
1099 }
1100
1101 return $this->mLinkRenderer;
1102 }
1103
1104 /**
1105 * Get the MagicWordFactory that this Parser is using
1106 *
1107 * @since 1.32
1108 * @return MagicWordFactory
1109 */
1110 public function getMagicWordFactory() {
1111 return $this->magicWordFactory;
1112 }
1113
1114 /**
1115 * Get the content language that this Parser is using
1116 *
1117 * @since 1.32
1118 * @return Language
1119 */
1120 public function getContentLanguage() {
1121 return $this->contLang;
1122 }
1123
1124 /**
1125 * Replaces all occurrences of HTML-style comments and the given tags
1126 * in the text with a random marker and returns the next text. The output
1127 * parameter $matches will be an associative array filled with data in
1128 * the form:
1129 *
1130 * @code
1131 * 'UNIQ-xxxxx' => [
1132 * 'element',
1133 * 'tag content',
1134 * [ 'param' => 'x' ],
1135 * '<element param="x">tag content</element>' ]
1136 * @endcode
1137 *
1138 * @param array $elements List of element names. Comments are always extracted.
1139 * @param string $text Source text string.
1140 * @param array &$matches Out parameter, Array: extracted tags
1141 * @return string Stripped text
1142 */
1143 public static function extractTagsAndParams( $elements, $text, &$matches ) {
1144 static $n = 1;
1145 $stripped = '';
1146 $matches = [];
1147
1148 $taglist = implode( '|', $elements );
1149 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
1150
1151 while ( $text != '' ) {
1152 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
1153 $stripped .= $p[0];
1154 if ( count( $p ) < 5 ) {
1155 break;
1156 }
1157 if ( count( $p ) > 5 ) {
1158 # comment
1159 $element = $p[4];
1160 $attributes = '';
1161 $close = '';
1162 $inside = $p[5];
1163 } else {
1164 # tag
1165 list( , $element, $attributes, $close, $inside ) = $p;
1166 }
1167
1168 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1169 $stripped .= $marker;
1170
1171 if ( $close === '/>' ) {
1172 # Empty element tag, <tag />
1173 $content = null;
1174 $text = $inside;
1175 $tail = null;
1176 } else {
1177 if ( $element === '!--' ) {
1178 $end = '/(-->)/';
1179 } else {
1180 $end = "/(<\\/$element\\s*>)/i";
1181 }
1182 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1183 $content = $q[0];
1184 if ( count( $q ) < 3 ) {
1185 # No end tag -- let it run out to the end of the text.
1186 $tail = '';
1187 $text = '';
1188 } else {
1189 list( , $tail, $text ) = $q;
1190 }
1191 }
1192
1193 $matches[$marker] = [ $element,
1194 $content,
1195 Sanitizer::decodeTagAttributes( $attributes ),
1196 "<$element$attributes$close$content$tail" ];
1197 }
1198 return $stripped;
1199 }
1200
1201 /**
1202 * Get a list of strippable XML-like elements
1203 *
1204 * @return array
1205 */
1206 public function getStripList() {
1207 return $this->mStripList;
1208 }
1209
1210 /**
1211 * Get the StripState
1212 *
1213 * @return StripState
1214 */
1215 public function getStripState() {
1216 return $this->mStripState;
1217 }
1218
1219 /**
1220 * Add an item to the strip state
1221 * Returns the unique tag which must be inserted into the stripped text
1222 * The tag will be replaced with the original text in unstrip()
1223 *
1224 * @param string $text
1225 *
1226 * @return string
1227 */
1228 public function insertStripItem( $text ) {
1229 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1230 $this->mMarkerIndex++;
1231 $this->mStripState->addGeneral( $marker, $text );
1232 return $marker;
1233 }
1234
1235 /**
1236 * Parse the wiki syntax used to render tables.
1237 *
1238 * @private
1239 * @param string $text
1240 * @return string
1241 * @deprecated since 1.34; should not be used outside parser class.
1242 */
1243 public function doTableStuff( $text ) {
1244 wfDeprecated( __METHOD__, '1.34' );
1245 return $this->handleTables( $text );
1246 }
1247
1248 /**
1249 * Parse the wiki syntax used to render tables.
1250 *
1251 * @param string $text
1252 * @return string
1253 */
1254 private function handleTables( $text ) {
1255 $lines = StringUtils::explode( "\n", $text );
1256 $out = '';
1257 $td_history = []; # Is currently a td tag open?
1258 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1259 $tr_history = []; # Is currently a tr tag open?
1260 $tr_attributes = []; # history of tr attributes
1261 $has_opened_tr = []; # Did this table open a <tr> element?
1262 $indent_level = 0; # indent level of the table
1263
1264 foreach ( $lines as $outLine ) {
1265 $line = trim( $outLine );
1266
1267 if ( $line === '' ) { # empty line, go to next line
1268 $out .= $outLine . "\n";
1269 continue;
1270 }
1271
1272 $first_character = $line[0];
1273 $first_two = substr( $line, 0, 2 );
1274 $matches = [];
1275
1276 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1277 # First check if we are starting a new table
1278 $indent_level = strlen( $matches[1] );
1279
1280 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1281 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1282
1283 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1284 array_push( $td_history, false );
1285 array_push( $last_tag_history, '' );
1286 array_push( $tr_history, false );
1287 array_push( $tr_attributes, '' );
1288 array_push( $has_opened_tr, false );
1289 } elseif ( count( $td_history ) == 0 ) {
1290 # Don't do any of the following
1291 $out .= $outLine . "\n";
1292 continue;
1293 } elseif ( $first_two === '|}' ) {
1294 # We are ending a table
1295 $line = '</table>' . substr( $line, 2 );
1296 $last_tag = array_pop( $last_tag_history );
1297
1298 if ( !array_pop( $has_opened_tr ) ) {
1299 $line = "<tr><td></td></tr>{$line}";
1300 }
1301
1302 if ( array_pop( $tr_history ) ) {
1303 $line = "</tr>{$line}";
1304 }
1305
1306 if ( array_pop( $td_history ) ) {
1307 $line = "</{$last_tag}>{$line}";
1308 }
1309 array_pop( $tr_attributes );
1310 if ( $indent_level > 0 ) {
1311 $outLine = rtrim( $line ) . str_repeat( '</dd></dl>', $indent_level );
1312 } else {
1313 $outLine = $line;
1314 }
1315 } elseif ( $first_two === '|-' ) {
1316 # Now we have a table row
1317 $line = preg_replace( '#^\|-+#', '', $line );
1318
1319 # Whats after the tag is now only attributes
1320 $attributes = $this->mStripState->unstripBoth( $line );
1321 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1322 array_pop( $tr_attributes );
1323 array_push( $tr_attributes, $attributes );
1324
1325 $line = '';
1326 $last_tag = array_pop( $last_tag_history );
1327 array_pop( $has_opened_tr );
1328 array_push( $has_opened_tr, true );
1329
1330 if ( array_pop( $tr_history ) ) {
1331 $line = '</tr>';
1332 }
1333
1334 if ( array_pop( $td_history ) ) {
1335 $line = "</{$last_tag}>{$line}";
1336 }
1337
1338 $outLine = $line;
1339 array_push( $tr_history, false );
1340 array_push( $td_history, false );
1341 array_push( $last_tag_history, '' );
1342 } elseif ( $first_character === '|'
1343 || $first_character === '!'
1344 || $first_two === '|+'
1345 ) {
1346 # This might be cell elements, td, th or captions
1347 if ( $first_two === '|+' ) {
1348 $first_character = '+';
1349 $line = substr( $line, 2 );
1350 } else {
1351 $line = substr( $line, 1 );
1352 }
1353
1354 // Implies both are valid for table headings.
1355 if ( $first_character === '!' ) {
1356 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1357 }
1358
1359 # Split up multiple cells on the same line.
1360 # FIXME : This can result in improper nesting of tags processed
1361 # by earlier parser steps.
1362 $cells = explode( '||', $line );
1363
1364 $outLine = '';
1365
1366 # Loop through each table cell
1367 foreach ( $cells as $cell ) {
1368 $previous = '';
1369 if ( $first_character !== '+' ) {
1370 $tr_after = array_pop( $tr_attributes );
1371 if ( !array_pop( $tr_history ) ) {
1372 $previous = "<tr{$tr_after}>\n";
1373 }
1374 array_push( $tr_history, true );
1375 array_push( $tr_attributes, '' );
1376 array_pop( $has_opened_tr );
1377 array_push( $has_opened_tr, true );
1378 }
1379
1380 $last_tag = array_pop( $last_tag_history );
1381
1382 if ( array_pop( $td_history ) ) {
1383 $previous = "</{$last_tag}>\n{$previous}";
1384 }
1385
1386 if ( $first_character === '|' ) {
1387 $last_tag = 'td';
1388 } elseif ( $first_character === '!' ) {
1389 $last_tag = 'th';
1390 } elseif ( $first_character === '+' ) {
1391 $last_tag = 'caption';
1392 } else {
1393 $last_tag = '';
1394 }
1395
1396 array_push( $last_tag_history, $last_tag );
1397
1398 # A cell could contain both parameters and data
1399 $cell_data = explode( '|', $cell, 2 );
1400
1401 # T2553: Note that a '|' inside an invalid link should not
1402 # be mistaken as delimiting cell parameters
1403 # Bug T153140: Neither should language converter markup.
1404 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1405 $cell = "{$previous}<{$last_tag}>" . trim( $cell );
1406 } elseif ( count( $cell_data ) == 1 ) {
1407 // Whitespace in cells is trimmed
1408 $cell = "{$previous}<{$last_tag}>" . trim( $cell_data[0] );
1409 } else {
1410 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1411 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1412 // Whitespace in cells is trimmed
1413 $cell = "{$previous}<{$last_tag}{$attributes}>" . trim( $cell_data[1] );
1414 }
1415
1416 $outLine .= $cell;
1417 array_push( $td_history, true );
1418 }
1419 }
1420 $out .= $outLine . "\n";
1421 }
1422
1423 # Closing open td, tr && table
1424 while ( count( $td_history ) > 0 ) {
1425 if ( array_pop( $td_history ) ) {
1426 $out .= "</td>\n";
1427 }
1428 if ( array_pop( $tr_history ) ) {
1429 $out .= "</tr>\n";
1430 }
1431 if ( !array_pop( $has_opened_tr ) ) {
1432 $out .= "<tr><td></td></tr>\n";
1433 }
1434
1435 $out .= "</table>\n";
1436 }
1437
1438 # Remove trailing line-ending (b/c)
1439 if ( substr( $out, -1 ) === "\n" ) {
1440 $out = substr( $out, 0, -1 );
1441 }
1442
1443 # special case: don't return empty table
1444 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1445 $out = '';
1446 }
1447
1448 return $out;
1449 }
1450
1451 /**
1452 * Helper function for parse() that transforms wiki markup into half-parsed
1453 * HTML. Only called for $mOutputType == self::OT_HTML.
1454 *
1455 * @private
1456 *
1457 * @param string $text The text to parse
1458 * @param-taint $text escapes_html
1459 * @param bool $isMain Whether this is being called from the main parse() function
1460 * @param PPFrame|bool $frame A pre-processor frame
1461 *
1462 * @return string
1463 */
1464 public function internalParse( $text, $isMain = true, $frame = false ) {
1465 $origText = $text;
1466
1467 // Avoid PHP 7.1 warning from passing $this by reference
1468 $parser = $this;
1469
1470 # Hook to suspend the parser in this state
1471 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1472 return $text;
1473 }
1474
1475 # if $frame is provided, then use $frame for replacing any variables
1476 if ( $frame ) {
1477 # use frame depth to infer how include/noinclude tags should be handled
1478 # depth=0 means this is the top-level document; otherwise it's an included document
1479 if ( !$frame->depth ) {
1480 $flag = 0;
1481 } else {
1482 $flag = self::PTD_FOR_INCLUSION;
1483 }
1484 $dom = $this->preprocessToDom( $text, $flag );
1485 $text = $frame->expand( $dom );
1486 } else {
1487 # if $frame is not provided, then use old-style replaceVariables
1488 $text = $this->replaceVariables( $text );
1489 }
1490
1491 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1492 $text = Sanitizer::removeHTMLtags(
1493 $text,
1494 [ $this, 'attributeStripCallback' ],
1495 false,
1496 array_keys( $this->mTransparentTagHooks ),
1497 [],
1498 [ $this, 'addTrackingCategory' ]
1499 );
1500 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1501
1502 # Tables need to come after variable replacement for things to work
1503 # properly; putting them before other transformations should keep
1504 # exciting things like link expansions from showing up in surprising
1505 # places.
1506 $text = $this->handleTables( $text );
1507
1508 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1509
1510 $text = $this->handleDoubleUnderscore( $text );
1511
1512 $text = $this->handleHeadings( $text );
1513 $text = $this->handleInternalLinks( $text );
1514 $text = $this->handleAllQuotes( $text );
1515 $text = $this->handleExternalLinks( $text );
1516
1517 # handleInternalLinks may sometimes leave behind
1518 # absolute URLs, which have to be masked to hide them from handleExternalLinks
1519 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1520
1521 $text = $this->handleMagicLinks( $text );
1522 $text = $this->finalizeHeadings( $text, $origText, $isMain );
1523
1524 return $text;
1525 }
1526
1527 /**
1528 * Helper function for parse() that transforms half-parsed HTML into fully
1529 * parsed HTML.
1530 *
1531 * @param string $text
1532 * @param bool $isMain
1533 * @param bool $linestart
1534 * @return string
1535 */
1536 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1537 $text = $this->mStripState->unstripGeneral( $text );
1538
1539 // Avoid PHP 7.1 warning from passing $this by reference
1540 $parser = $this;
1541
1542 if ( $isMain ) {
1543 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1544 }
1545
1546 # Clean up special characters, only run once, next-to-last before doBlockLevels
1547 $text = Sanitizer::armorFrenchSpaces( $text );
1548
1549 $text = $this->doBlockLevels( $text, $linestart );
1550
1551 $this->replaceLinkHoldersPrivate( $text );
1552
1553 /**
1554 * The input doesn't get language converted if
1555 * a) It's disabled
1556 * b) Content isn't converted
1557 * c) It's a conversion table
1558 * d) it is an interface message (which is in the user language)
1559 */
1560 if ( !( $this->mOptions->getDisableContentConversion()
1561 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1562 && !$this->mOptions->getInterfaceMessage()
1563 ) {
1564 # The position of the convert() call should not be changed. it
1565 # assumes that the links are all replaced and the only thing left
1566 # is the <nowiki> mark.
1567 $text = $this->getTargetLanguage()->convert( $text );
1568 }
1569
1570 $text = $this->mStripState->unstripNoWiki( $text );
1571
1572 if ( $isMain ) {
1573 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1574 }
1575
1576 $text = $this->replaceTransparentTags( $text );
1577 $text = $this->mStripState->unstripGeneral( $text );
1578
1579 $text = Sanitizer::normalizeCharReferences( $text );
1580
1581 if ( MWTidy::isEnabled() ) {
1582 if ( $this->mOptions->getTidy() ) {
1583 $text = MWTidy::tidy( $text );
1584 }
1585 } else {
1586 # attempt to sanitize at least some nesting problems
1587 # (T4702 and quite a few others)
1588 # This code path is buggy and deprecated!
1589 wfDeprecated( 'disabling tidy', '1.33' );
1590 $tidyregs = [
1591 # ''Something [http://www.cool.com cool''] -->
1592 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1593 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1594 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1595 # fix up an anchor inside another anchor, only
1596 # at least for a single single nested link (T5695)
1597 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1598 '\\1\\2</a>\\3</a>\\1\\4</a>',
1599 # fix div inside inline elements- doBlockLevels won't wrap a line which
1600 # contains a div, so fix it up here; replace
1601 # div with escaped text
1602 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1603 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1604 # remove empty italic or bold tag pairs, some
1605 # introduced by rules above
1606 '/<([bi])><\/\\1>/' => '',
1607 ];
1608
1609 $text = preg_replace(
1610 array_keys( $tidyregs ),
1611 array_values( $tidyregs ),
1612 $text );
1613 }
1614
1615 if ( $isMain ) {
1616 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1617 }
1618
1619 return $text;
1620 }
1621
1622 /**
1623 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1624 * magic external links.
1625 *
1626 * DML
1627 * @private
1628 * @param string $text
1629 * @return string
1630 * @deprecated since 1.34; should not be used outside parser class.
1631 */
1632 public function doMagicLinks( $text ) {
1633 wfDeprecated( __METHOD__, '1.34' );
1634 return $this->handleMagicLinks( $text );
1635 }
1636
1637 /**
1638 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1639 * magic external links.
1640 *
1641 * DML
1642 *
1643 * @param string $text
1644 *
1645 * @return string
1646 */
1647 private function handleMagicLinks( $text ) {
1648 $prots = wfUrlProtocolsWithoutProtRel();
1649 $urlChar = self::EXT_LINK_URL_CLASS;
1650 $addr = self::EXT_LINK_ADDR;
1651 $space = self::SPACE_NOT_NL; # non-newline space
1652 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1653 $spaces = "$space++"; # possessive match of 1 or more spaces
1654 $text = preg_replace_callback(
1655 '!(?: # Start cases
1656 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1657 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1658 (\b # m[3]: Free external links
1659 (?i:$prots)
1660 ($addr$urlChar*) # m[4]: Post-protocol path
1661 ) |
1662 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1663 ([0-9]+)\b |
1664 \bISBN $spaces ( # m[6]: ISBN, capture number
1665 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1666 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1667 [0-9Xx] # check digit
1668 )\b
1669 )!xu", [ $this, 'magicLinkCallback' ], $text );
1670 return $text;
1671 }
1672
1673 /**
1674 * @throws MWException
1675 * @param array $m
1676 * @return string HTML
1677 */
1678 public function magicLinkCallback( $m ) {
1679 if ( isset( $m[1] ) && $m[1] !== '' ) {
1680 # Skip anchor
1681 return $m[0];
1682 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1683 # Skip HTML element
1684 return $m[0];
1685 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1686 # Free external link
1687 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1688 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1689 # RFC or PMID
1690 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1691 if ( !$this->mOptions->getMagicRFCLinks() ) {
1692 return $m[0];
1693 }
1694 $keyword = 'RFC';
1695 $urlmsg = 'rfcurl';
1696 $cssClass = 'mw-magiclink-rfc';
1697 $trackingCat = 'magiclink-tracking-rfc';
1698 $id = $m[5];
1699 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1700 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1701 return $m[0];
1702 }
1703 $keyword = 'PMID';
1704 $urlmsg = 'pubmedurl';
1705 $cssClass = 'mw-magiclink-pmid';
1706 $trackingCat = 'magiclink-tracking-pmid';
1707 $id = $m[5];
1708 } else {
1709 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1710 substr( $m[0], 0, 20 ) . '"' );
1711 }
1712 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1713 $this->addTrackingCategory( $trackingCat );
1714 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1715 } elseif ( isset( $m[6] ) && $m[6] !== ''
1716 && $this->mOptions->getMagicISBNLinks()
1717 ) {
1718 # ISBN
1719 $isbn = $m[6];
1720 $space = self::SPACE_NOT_NL; # non-newline space
1721 $isbn = preg_replace( "/$space/", ' ', $isbn );
1722 $num = strtr( $isbn, [
1723 '-' => '',
1724 ' ' => '',
1725 'x' => 'X',
1726 ] );
1727 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1728 return $this->getLinkRenderer()->makeKnownLink(
1729 SpecialPage::getTitleFor( 'Booksources', $num ),
1730 "ISBN $isbn",
1731 [
1732 'class' => 'internal mw-magiclink-isbn',
1733 'title' => false // suppress title attribute
1734 ]
1735 );
1736 } else {
1737 return $m[0];
1738 }
1739 }
1740
1741 /**
1742 * Make a free external link, given a user-supplied URL
1743 *
1744 * @param string $url
1745 * @param int $numPostProto
1746 * The number of characters after the protocol.
1747 * @return string HTML
1748 * @private
1749 */
1750 public function makeFreeExternalLink( $url, $numPostProto ) {
1751 $trail = '';
1752
1753 # The characters '<' and '>' (which were escaped by
1754 # removeHTMLtags()) should not be included in
1755 # URLs, per RFC 2396.
1756 # Make &nbsp; terminate a URL as well (bug T84937)
1757 $m2 = [];
1758 if ( preg_match(
1759 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1760 $url,
1761 $m2,
1762 PREG_OFFSET_CAPTURE
1763 ) ) {
1764 $trail = substr( $url, $m2[0][1] ) . $trail;
1765 $url = substr( $url, 0, $m2[0][1] );
1766 }
1767
1768 # Move trailing punctuation to $trail
1769 $sep = ',;\.:!?';
1770 # If there is no left bracket, then consider right brackets fair game too
1771 if ( strpos( $url, '(' ) === false ) {
1772 $sep .= ')';
1773 }
1774
1775 $urlRev = strrev( $url );
1776 $numSepChars = strspn( $urlRev, $sep );
1777 # Don't break a trailing HTML entity by moving the ; into $trail
1778 # This is in hot code, so use substr_compare to avoid having to
1779 # create a new string object for the comparison
1780 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1781 # more optimization: instead of running preg_match with a $
1782 # anchor, which can be slow, do the match on the reversed
1783 # string starting at the desired offset.
1784 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1785 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1786 $numSepChars--;
1787 }
1788 }
1789 if ( $numSepChars ) {
1790 $trail = substr( $url, -$numSepChars ) . $trail;
1791 $url = substr( $url, 0, -$numSepChars );
1792 }
1793
1794 # Verify that we still have a real URL after trail removal, and
1795 # not just lone protocol
1796 if ( strlen( $trail ) >= $numPostProto ) {
1797 return $url . $trail;
1798 }
1799
1800 $url = Sanitizer::cleanUrl( $url );
1801
1802 # Is this an external image?
1803 $text = $this->maybeMakeExternalImage( $url );
1804 if ( $text === false ) {
1805 # Not an image, make a link
1806 $text = Linker::makeExternalLink( $url,
1807 $this->getTargetLanguage()->getConverter()->markNoConversion( $url ),
1808 true, 'free',
1809 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1810 # Register it in the output object...
1811 $this->mOutput->addExternalLink( $url );
1812 }
1813 return $text . $trail;
1814 }
1815
1816 /**
1817 * Parse headers and return html.
1818 *
1819 * @private
1820 * @param string $text
1821 * @return string
1822 * @deprecated since 1.34; should not be used outside parser class.
1823 */
1824 public function doHeadings( $text ) {
1825 wfDeprecated( __METHOD__, '1.34' );
1826 return $this->handleHeadings( $text );
1827 }
1828
1829 /**
1830 * Parse headers and return html
1831 *
1832 * @param string $text
1833 * @return string
1834 */
1835 private function handleHeadings( $text ) {
1836 for ( $i = 6; $i >= 1; --$i ) {
1837 $h = str_repeat( '=', $i );
1838 // Trim non-newline whitespace from headings
1839 // Using \s* will break for: "==\n===\n" and parse as <h2>=</h2>
1840 $text = preg_replace( "/^(?:$h)[ \\t]*(.+?)[ \\t]*(?:$h)\\s*$/m", "<h$i>\\1</h$i>", $text );
1841 }
1842 return $text;
1843 }
1844
1845 /**
1846 * Replace single quotes with HTML markup
1847 * @private
1848 *
1849 * @param string $text
1850 *
1851 * @return string The altered text
1852 * @deprecated since 1.34; should not be used outside parser class.
1853 */
1854 public function doAllQuotes( $text ) {
1855 wfDeprecated( __METHOD__, '1.34' );
1856 return $this->handleAllQuotes( $text );
1857 }
1858
1859 /**
1860 * Replace single quotes with HTML markup
1861 *
1862 * @param string $text
1863 *
1864 * @return string The altered text
1865 */
1866 private function handleAllQuotes( $text ) {
1867 $outtext = '';
1868 $lines = StringUtils::explode( "\n", $text );
1869 foreach ( $lines as $line ) {
1870 $outtext .= $this->doQuotes( $line ) . "\n";
1871 }
1872 $outtext = substr( $outtext, 0, -1 );
1873 return $outtext;
1874 }
1875
1876 /**
1877 * Helper function for doAllQuotes()
1878 *
1879 * @param string $text
1880 *
1881 * @return string
1882 * @internal
1883 */
1884 public function doQuotes( $text ) {
1885 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1886 $countarr = count( $arr );
1887 if ( $countarr == 1 ) {
1888 return $text;
1889 }
1890
1891 // First, do some preliminary work. This may shift some apostrophes from
1892 // being mark-up to being text. It also counts the number of occurrences
1893 // of bold and italics mark-ups.
1894 $numbold = 0;
1895 $numitalics = 0;
1896 for ( $i = 1; $i < $countarr; $i += 2 ) {
1897 $thislen = strlen( $arr[$i] );
1898 // If there are ever four apostrophes, assume the first is supposed to
1899 // be text, and the remaining three constitute mark-up for bold text.
1900 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1901 if ( $thislen == 4 ) {
1902 $arr[$i - 1] .= "'";
1903 $arr[$i] = "'''";
1904 $thislen = 3;
1905 } elseif ( $thislen > 5 ) {
1906 // If there are more than 5 apostrophes in a row, assume they're all
1907 // text except for the last 5.
1908 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1909 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1910 $arr[$i] = "'''''";
1911 $thislen = 5;
1912 }
1913 // Count the number of occurrences of bold and italics mark-ups.
1914 if ( $thislen == 2 ) {
1915 $numitalics++;
1916 } elseif ( $thislen == 3 ) {
1917 $numbold++;
1918 } elseif ( $thislen == 5 ) {
1919 $numitalics++;
1920 $numbold++;
1921 }
1922 }
1923
1924 // If there is an odd number of both bold and italics, it is likely
1925 // that one of the bold ones was meant to be an apostrophe followed
1926 // by italics. Which one we cannot know for certain, but it is more
1927 // likely to be one that has a single-letter word before it.
1928 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1929 $firstsingleletterword = -1;
1930 $firstmultiletterword = -1;
1931 $firstspace = -1;
1932 for ( $i = 1; $i < $countarr; $i += 2 ) {
1933 if ( strlen( $arr[$i] ) == 3 ) {
1934 $x1 = substr( $arr[$i - 1], -1 );
1935 $x2 = substr( $arr[$i - 1], -2, 1 );
1936 if ( $x1 === ' ' ) {
1937 if ( $firstspace == -1 ) {
1938 $firstspace = $i;
1939 }
1940 } elseif ( $x2 === ' ' ) {
1941 $firstsingleletterword = $i;
1942 // if $firstsingleletterword is set, we don't
1943 // look at the other options, so we can bail early.
1944 break;
1945 } elseif ( $firstmultiletterword == -1 ) {
1946 $firstmultiletterword = $i;
1947 }
1948 }
1949 }
1950
1951 // If there is a single-letter word, use it!
1952 if ( $firstsingleletterword > -1 ) {
1953 $arr[$firstsingleletterword] = "''";
1954 $arr[$firstsingleletterword - 1] .= "'";
1955 } elseif ( $firstmultiletterword > -1 ) {
1956 // If not, but there's a multi-letter word, use that one.
1957 $arr[$firstmultiletterword] = "''";
1958 $arr[$firstmultiletterword - 1] .= "'";
1959 } elseif ( $firstspace > -1 ) {
1960 // ... otherwise use the first one that has neither.
1961 // (notice that it is possible for all three to be -1 if, for example,
1962 // there is only one pentuple-apostrophe in the line)
1963 $arr[$firstspace] = "''";
1964 $arr[$firstspace - 1] .= "'";
1965 }
1966 }
1967
1968 // Now let's actually convert our apostrophic mush to HTML!
1969 $output = '';
1970 $buffer = '';
1971 $state = '';
1972 $i = 0;
1973 foreach ( $arr as $r ) {
1974 if ( ( $i % 2 ) == 0 ) {
1975 if ( $state === 'both' ) {
1976 $buffer .= $r;
1977 } else {
1978 $output .= $r;
1979 }
1980 } else {
1981 $thislen = strlen( $r );
1982 if ( $thislen == 2 ) {
1983 if ( $state === 'i' ) {
1984 $output .= '</i>';
1985 $state = '';
1986 } elseif ( $state === 'bi' ) {
1987 $output .= '</i>';
1988 $state = 'b';
1989 } elseif ( $state === 'ib' ) {
1990 $output .= '</b></i><b>';
1991 $state = 'b';
1992 } elseif ( $state === 'both' ) {
1993 $output .= '<b><i>' . $buffer . '</i>';
1994 $state = 'b';
1995 } else { // $state can be 'b' or ''
1996 $output .= '<i>';
1997 $state .= 'i';
1998 }
1999 } elseif ( $thislen == 3 ) {
2000 if ( $state === 'b' ) {
2001 $output .= '</b>';
2002 $state = '';
2003 } elseif ( $state === 'bi' ) {
2004 $output .= '</i></b><i>';
2005 $state = 'i';
2006 } elseif ( $state === 'ib' ) {
2007 $output .= '</b>';
2008 $state = 'i';
2009 } elseif ( $state === 'both' ) {
2010 $output .= '<i><b>' . $buffer . '</b>';
2011 $state = 'i';
2012 } else { // $state can be 'i' or ''
2013 $output .= '<b>';
2014 $state .= 'b';
2015 }
2016 } elseif ( $thislen == 5 ) {
2017 if ( $state === 'b' ) {
2018 $output .= '</b><i>';
2019 $state = 'i';
2020 } elseif ( $state === 'i' ) {
2021 $output .= '</i><b>';
2022 $state = 'b';
2023 } elseif ( $state === 'bi' ) {
2024 $output .= '</i></b>';
2025 $state = '';
2026 } elseif ( $state === 'ib' ) {
2027 $output .= '</b></i>';
2028 $state = '';
2029 } elseif ( $state === 'both' ) {
2030 $output .= '<i><b>' . $buffer . '</b></i>';
2031 $state = '';
2032 } else { // ($state == '')
2033 $buffer = '';
2034 $state = 'both';
2035 }
2036 }
2037 }
2038 $i++;
2039 }
2040 // Now close all remaining tags. Notice that the order is important.
2041 if ( $state === 'b' || $state === 'ib' ) {
2042 $output .= '</b>';
2043 }
2044 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
2045 $output .= '</i>';
2046 }
2047 if ( $state === 'bi' ) {
2048 $output .= '</b>';
2049 }
2050 // There might be lonely ''''', so make sure we have a buffer
2051 if ( $state === 'both' && $buffer ) {
2052 $output .= '<b><i>' . $buffer . '</i></b>';
2053 }
2054 return $output;
2055 }
2056
2057 /**
2058 * Replace external links (REL)
2059 *
2060 * Note: this is all very hackish and the order of execution matters a lot.
2061 * Make sure to run tests/parser/parserTests.php if you change this code.
2062 *
2063 * @private
2064 *
2065 * @param string $text
2066 *
2067 * @throws MWException
2068 * @return string
2069 */
2070 public function replaceExternalLinks( $text ) {
2071 wfDeprecated( __METHOD__, '1.34' );
2072 return $this->handleExternalLinks( $text );
2073 }
2074
2075 /**
2076 * Replace external links (REL)
2077 *
2078 * Note: this is all very hackish and the order of execution matters a lot.
2079 * Make sure to run tests/parser/parserTests.php if you change this code.
2080 *
2081 * @param string $text
2082 * @throws MWException
2083 * @return string
2084 */
2085 private function handleExternalLinks( $text ) {
2086 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
2087 // @phan-suppress-next-line PhanTypeComparisonFromArray See phan issue #3161
2088 if ( $bits === false ) {
2089 throw new MWException( "PCRE needs to be compiled with "
2090 . "--enable-unicode-properties in order for MediaWiki to function" );
2091 }
2092 $s = array_shift( $bits );
2093
2094 $i = 0;
2095 while ( $i < count( $bits ) ) {
2096 $url = $bits[$i++];
2097 $i++; // protocol
2098 $text = $bits[$i++];
2099 $trail = $bits[$i++];
2100
2101 # The characters '<' and '>' (which were escaped by
2102 # removeHTMLtags()) should not be included in
2103 # URLs, per RFC 2396.
2104 $m2 = [];
2105 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
2106 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
2107 $url = substr( $url, 0, $m2[0][1] );
2108 }
2109
2110 # If the link text is an image URL, replace it with an <img> tag
2111 # This happened by accident in the original parser, but some people used it extensively
2112 $img = $this->maybeMakeExternalImage( $text );
2113 if ( $img !== false ) {
2114 $text = $img;
2115 }
2116
2117 $dtrail = '';
2118
2119 # Set linktype for CSS
2120 $linktype = 'text';
2121
2122 # No link text, e.g. [http://domain.tld/some.link]
2123 if ( $text == '' ) {
2124 # Autonumber
2125 $langObj = $this->getTargetLanguage();
2126 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
2127 $linktype = 'autonumber';
2128 } else {
2129 # Have link text, e.g. [http://domain.tld/some.link text]s
2130 # Check for trail
2131 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
2132 }
2133
2134 // Excluding protocol-relative URLs may avoid many false positives.
2135 if ( preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
2136 $text = $this->getTargetLanguage()->getConverter()->markNoConversion( $text );
2137 }
2138
2139 $url = Sanitizer::cleanUrl( $url );
2140
2141 # Use the encoded URL
2142 # This means that users can paste URLs directly into the text
2143 # Funny characters like ö aren't valid in URLs anyway
2144 # This was changed in August 2004
2145 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
2146 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
2147
2148 # Register link in the output object.
2149 $this->mOutput->addExternalLink( $url );
2150 }
2151
2152 return $s;
2153 }
2154
2155 /**
2156 * Get the rel attribute for a particular external link.
2157 *
2158 * @since 1.21
2159 * @internal
2160 * @param string|bool $url Optional URL, to extract the domain from for rel =>
2161 * nofollow if appropriate
2162 * @param LinkTarget|null $title Optional LinkTarget, for wgNoFollowNsExceptions lookups
2163 * @return string|null Rel attribute for $url
2164 */
2165 public static function getExternalLinkRel( $url = false, $title = null ) {
2166 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
2167 $ns = $title ? $title->getNamespace() : false;
2168 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
2169 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
2170 ) {
2171 return 'nofollow';
2172 }
2173 return null;
2174 }
2175
2176 /**
2177 * Get an associative array of additional HTML attributes appropriate for a
2178 * particular external link. This currently may include rel => nofollow
2179 * (depending on configuration, namespace, and the URL's domain) and/or a
2180 * target attribute (depending on configuration).
2181 *
2182 * @internal
2183 * @param string $url URL to extract the domain from for rel =>
2184 * nofollow if appropriate
2185 * @return array Associative array of HTML attributes
2186 */
2187 public function getExternalLinkAttribs( $url ) {
2188 $attribs = [];
2189 $rel = self::getExternalLinkRel( $url, $this->mTitle );
2190
2191 $target = $this->mOptions->getExternalLinkTarget();
2192 if ( $target ) {
2193 $attribs['target'] = $target;
2194 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
2195 // T133507. New windows can navigate parent cross-origin.
2196 // Including noreferrer due to lacking browser
2197 // support of noopener. Eventually noreferrer should be removed.
2198 if ( $rel !== '' ) {
2199 $rel .= ' ';
2200 }
2201 $rel .= 'noreferrer noopener';
2202 }
2203 }
2204 $attribs['rel'] = $rel;
2205 return $attribs;
2206 }
2207
2208 /**
2209 * Replace unusual escape codes in a URL with their equivalent characters
2210 *
2211 * This generally follows the syntax defined in RFC 3986, with special
2212 * consideration for HTTP query strings.
2213 *
2214 * @internal
2215 * @param string $url
2216 * @return string
2217 */
2218 public static function normalizeLinkUrl( $url ) {
2219 # Test for RFC 3986 IPv6 syntax
2220 $scheme = '[a-z][a-z0-9+.-]*:';
2221 $userinfo = '(?:[a-z0-9\-._~!$&\'()*+,;=:]|%[0-9a-f]{2})*';
2222 $ipv6Host = '\\[((?:[0-9a-f:]|%3[0-A]|%[46][1-6])+)\\]';
2223 if ( preg_match( "<^(?:{$scheme})?//(?:{$userinfo}@)?{$ipv6Host}(?:[:/?#].*|)$>i", $url, $m ) &&
2224 IP::isValid( rawurldecode( $m[1] ) )
2225 ) {
2226 $isIPv6 = rawurldecode( $m[1] );
2227 } else {
2228 $isIPv6 = false;
2229 }
2230
2231 # Make sure unsafe characters are encoded
2232 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
2233 function ( $m ) {
2234 return rawurlencode( $m[0] );
2235 },
2236 $url
2237 );
2238
2239 $ret = '';
2240 $end = strlen( $url );
2241
2242 # Fragment part - 'fragment'
2243 $start = strpos( $url, '#' );
2244 if ( $start !== false && $start < $end ) {
2245 $ret = self::normalizeUrlComponent(
2246 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
2247 $end = $start;
2248 }
2249
2250 # Query part - 'query' minus &=+;
2251 $start = strpos( $url, '?' );
2252 if ( $start !== false && $start < $end ) {
2253 $ret = self::normalizeUrlComponent(
2254 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2255 $end = $start;
2256 }
2257
2258 # Scheme and path part - 'pchar'
2259 # (we assume no userinfo or encoded colons in the host)
2260 $ret = self::normalizeUrlComponent(
2261 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2262
2263 # Fix IPv6 syntax
2264 if ( $isIPv6 !== false ) {
2265 $ipv6Host = "%5B({$isIPv6})%5D";
2266 $ret = preg_replace(
2267 "<^((?:{$scheme})?//(?:{$userinfo}@)?){$ipv6Host}(?=[:/?#]|$)>i",
2268 "$1[$2]",
2269 $ret
2270 );
2271 }
2272
2273 return $ret;
2274 }
2275
2276 private static function normalizeUrlComponent( $component, $unsafe ) {
2277 $callback = function ( $matches ) use ( $unsafe ) {
2278 $char = urldecode( $matches[0] );
2279 $ord = ord( $char );
2280 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2281 # Unescape it
2282 return $char;
2283 } else {
2284 # Leave it escaped, but use uppercase for a-f
2285 return strtoupper( $matches[0] );
2286 }
2287 };
2288 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2289 }
2290
2291 /**
2292 * make an image if it's allowed, either through the global
2293 * option, through the exception, or through the on-wiki whitelist
2294 *
2295 * @param string $url
2296 *
2297 * @return string
2298 */
2299 private function maybeMakeExternalImage( $url ) {
2300 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2301 $imagesexception = !empty( $imagesfrom );
2302 $text = false;
2303 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2304 if ( $imagesexception && is_array( $imagesfrom ) ) {
2305 $imagematch = false;
2306 foreach ( $imagesfrom as $match ) {
2307 if ( strpos( $url, $match ) === 0 ) {
2308 $imagematch = true;
2309 break;
2310 }
2311 }
2312 } elseif ( $imagesexception ) {
2313 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2314 } else {
2315 $imagematch = false;
2316 }
2317
2318 if ( $this->mOptions->getAllowExternalImages()
2319 || ( $imagesexception && $imagematch )
2320 ) {
2321 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2322 # Image found
2323 $text = Linker::makeExternalImage( $url );
2324 }
2325 }
2326 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2327 && preg_match( self::EXT_IMAGE_REGEX, $url )
2328 ) {
2329 $whitelist = explode(
2330 "\n",
2331 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2332 );
2333
2334 foreach ( $whitelist as $entry ) {
2335 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2336 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2337 continue;
2338 }
2339 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2340 # Image matches a whitelist entry
2341 $text = Linker::makeExternalImage( $url );
2342 break;
2343 }
2344 }
2345 }
2346 return $text;
2347 }
2348
2349 /**
2350 * Process [[ ]] wikilinks
2351 *
2352 * @param string $text
2353 *
2354 * @return string Processed text
2355 *
2356 * @private
2357 * @deprecated since 1.34; should not be used outside parser class.
2358 */
2359 public function replaceInternalLinks( $text ) {
2360 wfDeprecated( __METHOD__, '1.34' );
2361 return $this->handleInternalLinks( $text );
2362 }
2363
2364 /**
2365 * Process [[ ]] wikilinks
2366 *
2367 * @param string $text
2368 *
2369 * @return string Processed text
2370 */
2371 private function handleInternalLinks( $text ) {
2372 $this->mLinkHolders->merge( $this->handleInternalLinks2( $text ) );
2373 return $text;
2374 }
2375
2376 /**
2377 * Process [[ ]] wikilinks (RIL)
2378 * @param string &$text
2379 * @throws MWException
2380 * @return LinkHolderArray
2381 *
2382 * @private
2383 * @deprecated since 1.34; should not be used outside parser class.
2384 */
2385 public function replaceInternalLinks2( &$text ) {
2386 wfDeprecated( __METHOD__, '1.34' );
2387 return $this->handleInternalLinks2( $text );
2388 }
2389
2390 /**
2391 * Process [[ ]] wikilinks (RIL)
2392 * @param string &$s
2393 * @throws MWException
2394 * @return LinkHolderArray
2395 */
2396 private function handleInternalLinks2( &$s ) {
2397 static $tc = false, $e1, $e1_img;
2398 # the % is needed to support urlencoded titles as well
2399 if ( !$tc ) {
2400 $tc = Title::legalChars() . '#%';
2401 # Match a link having the form [[namespace:link|alternate]]trail
2402 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2403 # Match cases where there is no "]]", which might still be images
2404 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2405 }
2406
2407 $holders = new LinkHolderArray( $this );
2408
2409 # split the entire text string on occurrences of [[
2410 $a = StringUtils::explode( '[[', ' ' . $s );
2411 # get the first element (all text up to first [[), and remove the space we added
2412 $s = $a->current();
2413 $a->next();
2414 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2415 $s = substr( $s, 1 );
2416
2417 if ( is_null( $this->mTitle ) ) {
2418 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2419 }
2420 $nottalk = !$this->mTitle->isTalkPage();
2421
2422 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2423 $e2 = null;
2424 if ( $useLinkPrefixExtension ) {
2425 # Match the end of a line for a word that's not followed by whitespace,
2426 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2427 $charset = $this->contLang->linkPrefixCharset();
2428 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2429 $m = [];
2430 if ( preg_match( $e2, $s, $m ) ) {
2431 $first_prefix = $m[2];
2432 } else {
2433 $first_prefix = false;
2434 }
2435 } else {
2436 $prefix = '';
2437 }
2438
2439 # Some namespaces don't allow subpages
2440 $useSubpages = $this->nsInfo->hasSubpages(
2441 $this->mTitle->getNamespace()
2442 );
2443
2444 # Loop for each link
2445 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2446 # Check for excessive memory usage
2447 if ( $holders->isBig() ) {
2448 # Too big
2449 # Do the existence check, replace the link holders and clear the array
2450 $holders->replace( $s );
2451 $holders->clear();
2452 }
2453
2454 if ( $useLinkPrefixExtension ) {
2455 if ( preg_match( $e2, $s, $m ) ) {
2456 list( , $s, $prefix ) = $m;
2457 } else {
2458 $prefix = '';
2459 }
2460 # first link
2461 if ( $first_prefix ) {
2462 $prefix = $first_prefix;
2463 $first_prefix = false;
2464 }
2465 }
2466
2467 $might_be_img = false;
2468
2469 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2470 $text = $m[2];
2471 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2472 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2473 # the real problem is with the $e1 regex
2474 # See T1500.
2475 # Still some problems for cases where the ] is meant to be outside punctuation,
2476 # and no image is in sight. See T4095.
2477 if ( $text !== ''
2478 && substr( $m[3], 0, 1 ) === ']'
2479 && strpos( $text, '[' ) !== false
2480 ) {
2481 $text .= ']'; # so that handleExternalLinks($text) works later
2482 $m[3] = substr( $m[3], 1 );
2483 }
2484 # fix up urlencoded title texts
2485 if ( strpos( $m[1], '%' ) !== false ) {
2486 # Should anchors '#' also be rejected?
2487 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2488 }
2489 $trail = $m[3];
2490 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2491 # Invalid, but might be an image with a link in its caption
2492 $might_be_img = true;
2493 $text = $m[2];
2494 if ( strpos( $m[1], '%' ) !== false ) {
2495 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2496 }
2497 $trail = "";
2498 } else { # Invalid form; output directly
2499 $s .= $prefix . '[[' . $line;
2500 continue;
2501 }
2502
2503 $origLink = ltrim( $m[1], ' ' );
2504
2505 # Don't allow internal links to pages containing
2506 # PROTO: where PROTO is a valid URL protocol; these
2507 # should be external links.
2508 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2509 $s .= $prefix . '[[' . $line;
2510 continue;
2511 }
2512
2513 # Make subpage if necessary
2514 if ( $useSubpages ) {
2515 $link = Linker::normalizeSubpageLink(
2516 $this->mTitle, $origLink, $text
2517 );
2518 } else {
2519 $link = $origLink;
2520 }
2521
2522 // \x7f isn't a default legal title char, so most likely strip
2523 // markers will force us into the "invalid form" path above. But,
2524 // just in case, let's assert that xmlish tags aren't valid in
2525 // the title position.
2526 $unstrip = $this->mStripState->killMarkers( $link );
2527 $noMarkers = ( $unstrip === $link );
2528
2529 $nt = $noMarkers ? Title::newFromText( $link ) : null;
2530 if ( $nt === null ) {
2531 $s .= $prefix . '[[' . $line;
2532 continue;
2533 }
2534
2535 $ns = $nt->getNamespace();
2536 $iw = $nt->getInterwiki();
2537
2538 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2539
2540 if ( $might_be_img ) { # if this is actually an invalid link
2541 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2542 $found = false;
2543 while ( true ) {
2544 # look at the next 'line' to see if we can close it there
2545 $a->next();
2546 $next_line = $a->current();
2547 if ( $next_line === false || $next_line === null ) {
2548 break;
2549 }
2550 $m = explode( ']]', $next_line, 3 );
2551 if ( count( $m ) == 3 ) {
2552 # the first ]] closes the inner link, the second the image
2553 $found = true;
2554 $text .= "[[{$m[0]}]]{$m[1]}";
2555 $trail = $m[2];
2556 break;
2557 } elseif ( count( $m ) == 2 ) {
2558 # if there's exactly one ]] that's fine, we'll keep looking
2559 $text .= "[[{$m[0]}]]{$m[1]}";
2560 } else {
2561 # if $next_line is invalid too, we need look no further
2562 $text .= '[[' . $next_line;
2563 break;
2564 }
2565 }
2566 if ( !$found ) {
2567 # we couldn't find the end of this imageLink, so output it raw
2568 # but don't ignore what might be perfectly normal links in the text we've examined
2569 $holders->merge( $this->handleInternalLinks2( $text ) );
2570 $s .= "{$prefix}[[$link|$text";
2571 # note: no $trail, because without an end, there *is* no trail
2572 continue;
2573 }
2574 } else { # it's not an image, so output it raw
2575 $s .= "{$prefix}[[$link|$text";
2576 # note: no $trail, because without an end, there *is* no trail
2577 continue;
2578 }
2579 }
2580
2581 $wasblank = ( $text == '' );
2582 if ( $wasblank ) {
2583 $text = $link;
2584 if ( !$noforce ) {
2585 # Strip off leading ':'
2586 $text = substr( $text, 1 );
2587 }
2588 } else {
2589 # T6598 madness. Handle the quotes only if they come from the alternate part
2590 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2591 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2592 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2593 $text = $this->doQuotes( $text );
2594 }
2595
2596 # Link not escaped by : , create the various objects
2597 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2598 # Interwikis
2599 if (
2600 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2601 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2602 in_array( $iw, $this->svcOptions->get( 'ExtraInterlanguageLinkPrefixes' ) )
2603 )
2604 ) {
2605 # T26502: filter duplicates
2606 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2607 $this->mLangLinkLanguages[$iw] = true;
2608 $this->mOutput->addLanguageLink( $nt->getFullText() );
2609 }
2610
2611 /**
2612 * Strip the whitespace interwiki links produce, see T10897
2613 */
2614 $s = rtrim( $s . $prefix ) . $trail; # T175416
2615 continue;
2616 }
2617
2618 if ( $ns == NS_FILE ) {
2619 if ( !$this->badFileLookup->isBadFile( $nt->getDBkey(), $this->mTitle ) ) {
2620 if ( $wasblank ) {
2621 # if no parameters were passed, $text
2622 # becomes something like "File:Foo.png",
2623 # which we don't want to pass on to the
2624 # image generator
2625 $text = '';
2626 } else {
2627 # recursively parse links inside the image caption
2628 # actually, this will parse them in any other parameters, too,
2629 # but it might be hard to fix that, and it doesn't matter ATM
2630 $text = $this->handleExternalLinks( $text );
2631 $holders->merge( $this->handleInternalLinks2( $text ) );
2632 }
2633 # cloak any absolute URLs inside the image markup, so handleExternalLinks() won't touch them
2634 $s .= $prefix . $this->armorLinks(
2635 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2636 continue;
2637 }
2638 } elseif ( $ns == NS_CATEGORY ) {
2639 /**
2640 * Strip the whitespace Category links produce, see T2087
2641 */
2642 $s = rtrim( $s . $prefix ) . $trail; # T2087, T87753
2643
2644 if ( $wasblank ) {
2645 $sortkey = $this->getDefaultSort();
2646 } else {
2647 $sortkey = $text;
2648 }
2649 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2650 $sortkey = str_replace( "\n", '', $sortkey );
2651 $sortkey = $this->getTargetLanguage()->convertCategoryKey( $sortkey );
2652 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2653
2654 continue;
2655 }
2656 }
2657
2658 # Self-link checking. For some languages, variants of the title are checked in
2659 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2660 # for linking to a different variant.
2661 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2662 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2663 continue;
2664 }
2665
2666 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2667 # @todo FIXME: Should do batch file existence checks, see comment below
2668 if ( $ns == NS_MEDIA ) {
2669 # Give extensions a chance to select the file revision for us
2670 $options = [];
2671 $descQuery = false;
2672 Hooks::run( 'BeforeParserFetchFileAndTitle',
2673 [ $this, $nt, &$options, &$descQuery ] );
2674 # Fetch and register the file (file title may be different via hooks)
2675 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2676 # Cloak with NOPARSE to avoid replacement in handleExternalLinks
2677 $s .= $prefix . $this->armorLinks(
2678 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2679 continue;
2680 }
2681
2682 # Some titles, such as valid special pages or files in foreign repos, should
2683 # be shown as bluelinks even though they're not included in the page table
2684 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2685 # batch file existence checks for NS_FILE and NS_MEDIA
2686 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2687 $this->mOutput->addLink( $nt );
2688 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2689 } else {
2690 # Links will be added to the output link list after checking
2691 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2692 }
2693 }
2694 return $holders;
2695 }
2696
2697 /**
2698 * Render a forced-blue link inline; protect against double expansion of
2699 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2700 * Since this little disaster has to split off the trail text to avoid
2701 * breaking URLs in the following text without breaking trails on the
2702 * wiki links, it's been made into a horrible function.
2703 *
2704 * @param Title $nt
2705 * @param string $text
2706 * @param string $trail
2707 * @param string $prefix
2708 * @return string HTML-wikitext mix oh yuck
2709 */
2710 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2711 list( $inside, $trail ) = Linker::splitTrail( $trail );
2712
2713 if ( $text == '' ) {
2714 $text = htmlspecialchars( $nt->getPrefixedText() );
2715 }
2716
2717 $link = $this->getLinkRenderer()->makeKnownLink(
2718 $nt, new HtmlArmor( "$prefix$text$inside" )
2719 );
2720
2721 return $this->armorLinks( $link ) . $trail;
2722 }
2723
2724 /**
2725 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2726 * going to go through further parsing steps before inline URL expansion.
2727 *
2728 * Not needed quite as much as it used to be since free links are a bit
2729 * more sensible these days. But bracketed links are still an issue.
2730 *
2731 * @param string $text More-or-less HTML
2732 * @return string Less-or-more HTML with NOPARSE bits
2733 */
2734 public function armorLinks( $text ) {
2735 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2736 self::MARKER_PREFIX . "NOPARSE$1", $text );
2737 }
2738
2739 /**
2740 * Return true if subpage links should be expanded on this page.
2741 * @return bool
2742 * @deprecated since 1.34; should not be used outside parser class.
2743 */
2744 public function areSubpagesAllowed() {
2745 # Some namespaces don't allow subpages
2746 wfDeprecated( __METHOD__, '1.34' );
2747 return $this->nsInfo->hasSubpages( $this->mTitle->getNamespace() );
2748 }
2749
2750 /**
2751 * Handle link to subpage if necessary
2752 *
2753 * @param string $target The source of the link
2754 * @param string &$text The link text, modified as necessary
2755 * @return string The full name of the link
2756 * @private
2757 * @deprecated since 1.34; should not be used outside parser class.
2758 */
2759 public function maybeDoSubpageLink( $target, &$text ) {
2760 wfDeprecated( __METHOD__, '1.34' );
2761 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2762 }
2763
2764 /**
2765 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2766 *
2767 * @param string $text
2768 * @param bool $linestart Whether or not this is at the start of a line.
2769 * @private
2770 * @return string The lists rendered as HTML
2771 */
2772 public function doBlockLevels( $text, $linestart ) {
2773 return BlockLevelPass::doBlockLevels( $text, $linestart );
2774 }
2775
2776 /**
2777 * Return value of a magic variable (like PAGENAME)
2778 *
2779 * @private
2780 *
2781 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2782 * @param bool|PPFrame $frame
2783 *
2784 * @throws MWException
2785 * @return string
2786 * @deprecated since 1.34; should not be used outside parser class.
2787 */
2788 public function getVariableValue( $index, $frame = false ) {
2789 wfDeprecated( __METHOD__, '1.34' );
2790 return $this->expandMagicVariable( $index, $frame );
2791 }
2792
2793 /**
2794 * Return value of a magic variable (like PAGENAME)
2795 *
2796 * @param string $index Magic variable identifier as mapped in MagicWordFactory::$mVariableIDs
2797 * @param bool|PPFrame $frame
2798 *
2799 * @throws MWException
2800 * @return string
2801 */
2802 private function expandMagicVariable( $index, $frame = false ) {
2803 // XXX This function should be moved out of Parser class for
2804 // reuse by Parsoid/etc.
2805 if ( is_null( $this->mTitle ) ) {
2806 // If no title set, bad things are going to happen
2807 // later. Title should always be set since this
2808 // should only be called in the middle of a parse
2809 // operation (but the unit-tests do funky stuff)
2810 throw new MWException( __METHOD__ . ' Should only be '
2811 . ' called while parsing (no title set)' );
2812 }
2813
2814 // Avoid PHP 7.1 warning from passing $this by reference
2815 $parser = $this;
2816
2817 /**
2818 * Some of these require message or data lookups and can be
2819 * expensive to check many times.
2820 */
2821 if (
2822 Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) &&
2823 isset( $this->mVarCache[$index] )
2824 ) {
2825 return $this->mVarCache[$index];
2826 }
2827
2828 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2829 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2830
2831 $pageLang = $this->getFunctionLang();
2832
2833 switch ( $index ) {
2834 case '!':
2835 $value = '|';
2836 break;
2837 case 'currentmonth':
2838 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ), true );
2839 break;
2840 case 'currentmonth1':
2841 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ), true );
2842 break;
2843 case 'currentmonthname':
2844 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2845 break;
2846 case 'currentmonthnamegen':
2847 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2848 break;
2849 case 'currentmonthabbrev':
2850 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2851 break;
2852 case 'currentday':
2853 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ), true );
2854 break;
2855 case 'currentday2':
2856 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ), true );
2857 break;
2858 case 'localmonth':
2859 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ), true );
2860 break;
2861 case 'localmonth1':
2862 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ), true );
2863 break;
2864 case 'localmonthname':
2865 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2866 break;
2867 case 'localmonthnamegen':
2868 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2869 break;
2870 case 'localmonthabbrev':
2871 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2872 break;
2873 case 'localday':
2874 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ), true );
2875 break;
2876 case 'localday2':
2877 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ), true );
2878 break;
2879 case 'pagename':
2880 $value = wfEscapeWikiText( $this->mTitle->getText() );
2881 break;
2882 case 'pagenamee':
2883 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2884 break;
2885 case 'fullpagename':
2886 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2887 break;
2888 case 'fullpagenamee':
2889 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2890 break;
2891 case 'subpagename':
2892 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2893 break;
2894 case 'subpagenamee':
2895 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2896 break;
2897 case 'rootpagename':
2898 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2899 break;
2900 case 'rootpagenamee':
2901 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2902 ' ',
2903 '_',
2904 $this->mTitle->getRootText()
2905 ) ) );
2906 break;
2907 case 'basepagename':
2908 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2909 break;
2910 case 'basepagenamee':
2911 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2912 ' ',
2913 '_',
2914 $this->mTitle->getBaseText()
2915 ) ) );
2916 break;
2917 case 'talkpagename':
2918 if ( $this->mTitle->canHaveTalkPage() ) {
2919 $talkPage = $this->mTitle->getTalkPage();
2920 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2921 } else {
2922 $value = '';
2923 }
2924 break;
2925 case 'talkpagenamee':
2926 if ( $this->mTitle->canHaveTalkPage() ) {
2927 $talkPage = $this->mTitle->getTalkPage();
2928 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2929 } else {
2930 $value = '';
2931 }
2932 break;
2933 case 'subjectpagename':
2934 $subjPage = $this->mTitle->getSubjectPage();
2935 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2936 break;
2937 case 'subjectpagenamee':
2938 $subjPage = $this->mTitle->getSubjectPage();
2939 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2940 break;
2941 case 'pageid': // requested in T25427
2942 # Inform the edit saving system that getting the canonical output
2943 # after page insertion requires a parse that used that exact page ID
2944 $this->setOutputFlag( 'vary-page-id', '{{PAGEID}} used' );
2945 $value = $this->mTitle->getArticleID();
2946 if ( !$value ) {
2947 $value = $this->mOptions->getSpeculativePageId();
2948 if ( $value ) {
2949 $this->mOutput->setSpeculativePageIdUsed( $value );
2950 }
2951 }
2952 break;
2953 case 'revisionid':
2954 if (
2955 $this->svcOptions->get( 'MiserMode' ) &&
2956 !$this->mOptions->getInterfaceMessage() &&
2957 // @TODO: disallow this word on all namespaces
2958 $this->nsInfo->isContent( $this->mTitle->getNamespace() )
2959 ) {
2960 // Use a stub result instead of the actual revision ID in order to avoid
2961 // double parses on page save but still allow preview detection (T137900)
2962 if ( $this->getRevisionId() || $this->mOptions->getSpeculativeRevId() ) {
2963 $value = '-';
2964 } else {
2965 $this->setOutputFlag( 'vary-revision-exists', '{{REVISIONID}} used' );
2966 $value = '';
2967 }
2968 } else {
2969 # Inform the edit saving system that getting the canonical output after
2970 # revision insertion requires a parse that used that exact revision ID
2971 $this->setOutputFlag( 'vary-revision-id', '{{REVISIONID}} used' );
2972 $value = $this->getRevisionId();
2973 if ( $value === 0 ) {
2974 $rev = $this->getRevisionObject();
2975 $value = $rev ? $rev->getId() : $value;
2976 }
2977 if ( !$value ) {
2978 $value = $this->mOptions->getSpeculativeRevId();
2979 if ( $value ) {
2980 $this->mOutput->setSpeculativeRevIdUsed( $value );
2981 }
2982 }
2983 }
2984 break;
2985 case 'revisionday':
2986 $value = (int)$this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2987 break;
2988 case 'revisionday2':
2989 $value = $this->getRevisionTimestampSubstring( 6, 2, self::MAX_TTS, $index );
2990 break;
2991 case 'revisionmonth':
2992 $value = $this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2993 break;
2994 case 'revisionmonth1':
2995 $value = (int)$this->getRevisionTimestampSubstring( 4, 2, self::MAX_TTS, $index );
2996 break;
2997 case 'revisionyear':
2998 $value = $this->getRevisionTimestampSubstring( 0, 4, self::MAX_TTS, $index );
2999 break;
3000 case 'revisiontimestamp':
3001 $value = $this->getRevisionTimestampSubstring( 0, 14, self::MAX_TTS, $index );
3002 break;
3003 case 'revisionuser':
3004 # Inform the edit saving system that getting the canonical output after
3005 # revision insertion requires a parse that used the actual user ID
3006 $this->setOutputFlag( 'vary-user', '{{REVISIONUSER}} used' );
3007 $value = $this->getRevisionUser();
3008 break;
3009 case 'revisionsize':
3010 $value = $this->getRevisionSize();
3011 break;
3012 case 'namespace':
3013 $value = str_replace( '_', ' ',
3014 $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
3015 break;
3016 case 'namespacee':
3017 $value = wfUrlencode( $this->contLang->getNsText( $this->mTitle->getNamespace() ) );
3018 break;
3019 case 'namespacenumber':
3020 $value = $this->mTitle->getNamespace();
3021 break;
3022 case 'talkspace':
3023 $value = $this->mTitle->canHaveTalkPage()
3024 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
3025 : '';
3026 break;
3027 case 'talkspacee':
3028 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
3029 break;
3030 case 'subjectspace':
3031 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
3032 break;
3033 case 'subjectspacee':
3034 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
3035 break;
3036 case 'currentdayname':
3037 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
3038 break;
3039 case 'currentyear':
3040 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
3041 break;
3042 case 'currenttime':
3043 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
3044 break;
3045 case 'currenthour':
3046 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
3047 break;
3048 case 'currentweek':
3049 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
3050 # int to remove the padding
3051 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
3052 break;
3053 case 'currentdow':
3054 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
3055 break;
3056 case 'localdayname':
3057 $value = $pageLang->getWeekdayName(
3058 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
3059 );
3060 break;
3061 case 'localyear':
3062 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
3063 break;
3064 case 'localtime':
3065 $value = $pageLang->time(
3066 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
3067 false,
3068 false
3069 );
3070 break;
3071 case 'localhour':
3072 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
3073 break;
3074 case 'localweek':
3075 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
3076 # int to remove the padding
3077 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
3078 break;
3079 case 'localdow':
3080 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
3081 break;
3082 case 'numberofarticles':
3083 $value = $pageLang->formatNum( SiteStats::articles() );
3084 break;
3085 case 'numberoffiles':
3086 $value = $pageLang->formatNum( SiteStats::images() );
3087 break;
3088 case 'numberofusers':
3089 $value = $pageLang->formatNum( SiteStats::users() );
3090 break;
3091 case 'numberofactiveusers':
3092 $value = $pageLang->formatNum( SiteStats::activeUsers() );
3093 break;
3094 case 'numberofpages':
3095 $value = $pageLang->formatNum( SiteStats::pages() );
3096 break;
3097 case 'numberofadmins':
3098 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
3099 break;
3100 case 'numberofedits':
3101 $value = $pageLang->formatNum( SiteStats::edits() );
3102 break;
3103 case 'currenttimestamp':
3104 $value = wfTimestamp( TS_MW, $ts );
3105 break;
3106 case 'localtimestamp':
3107 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
3108 break;
3109 case 'currentversion':
3110 $value = SpecialVersion::getVersion();
3111 break;
3112 case 'articlepath':
3113 return $this->svcOptions->get( 'ArticlePath' );
3114 case 'sitename':
3115 return $this->svcOptions->get( 'Sitename' );
3116 case 'server':
3117 return $this->svcOptions->get( 'Server' );
3118 case 'servername':
3119 return $this->svcOptions->get( 'ServerName' );
3120 case 'scriptpath':
3121 return $this->svcOptions->get( 'ScriptPath' );
3122 case 'stylepath':
3123 return $this->svcOptions->get( 'StylePath' );
3124 case 'directionmark':
3125 return $pageLang->getDirMark();
3126 case 'contentlanguage':
3127 return $this->svcOptions->get( 'LanguageCode' );
3128 case 'pagelanguage':
3129 $value = $pageLang->getCode();
3130 break;
3131 case 'cascadingsources':
3132 $value = CoreParserFunctions::cascadingsources( $this );
3133 break;
3134 default:
3135 $ret = null;
3136 Hooks::run(
3137 'ParserGetVariableValueSwitch',
3138 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
3139 );
3140
3141 return $ret;
3142 }
3143
3144 if ( $index ) {
3145 $this->mVarCache[$index] = $value;
3146 }
3147
3148 return $value;
3149 }
3150
3151 /**
3152 * @param int $start
3153 * @param int $len
3154 * @param int $mtts Max time-till-save; sets vary-revision-timestamp if result changes by then
3155 * @param string $variable Parser variable name
3156 * @return string
3157 */
3158 private function getRevisionTimestampSubstring( $start, $len, $mtts, $variable ) {
3159 # Get the timezone-adjusted timestamp to be used for this revision
3160 $resNow = substr( $this->getRevisionTimestamp(), $start, $len );
3161 # Possibly set vary-revision if there is not yet an associated revision
3162 if ( !$this->getRevisionObject() ) {
3163 # Get the timezone-adjusted timestamp $mtts seconds in the future.
3164 # This future is relative to the current time and not that of the
3165 # parser options. The rendered timestamp can be compared to that
3166 # of the timestamp specified by the parser options.
3167 $resThen = substr(
3168 $this->contLang->userAdjust( wfTimestamp( TS_MW, time() + $mtts ), '' ),
3169 $start,
3170 $len
3171 );
3172
3173 if ( $resNow !== $resThen ) {
3174 # Inform the edit saving system that getting the canonical output after
3175 # revision insertion requires a parse that used an actual revision timestamp
3176 $this->setOutputFlag( 'vary-revision-timestamp', "$variable used" );
3177 }
3178 }
3179
3180 return $resNow;
3181 }
3182
3183 /**
3184 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3185 *
3186 * @private
3187 * @deprecated since 1.34; should not be used outside parser class.
3188 */
3189 public function initialiseVariables() {
3190 wfDeprecated( __METHOD__, '1.34' );
3191 $this->initializeVariables();
3192 }
3193
3194 /**
3195 * Initialize the magic variables (like CURRENTMONTHNAME) and
3196 * substitution modifiers.
3197 */
3198 private function initializeVariables() {
3199 $variableIDs = $this->magicWordFactory->getVariableIDs();
3200 $substIDs = $this->magicWordFactory->getSubstIDs();
3201
3202 $this->mVariables = $this->magicWordFactory->newArray( $variableIDs );
3203 $this->mSubstWords = $this->magicWordFactory->newArray( $substIDs );
3204 }
3205
3206 /**
3207 * Preprocess some wikitext and return the document tree.
3208 * This is the ghost of replace_variables().
3209 *
3210 * @param string $text The text to parse
3211 * @param int $flags Bitwise combination of:
3212 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3213 * included. Default is to assume a direct page view.
3214 *
3215 * The generated DOM tree must depend only on the input text and the flags.
3216 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
3217 *
3218 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3219 * change in the DOM tree for a given text, must be passed through the section identifier
3220 * in the section edit link and thus back to extractSections().
3221 *
3222 * The output of this function is currently only cached in process memory, but a persistent
3223 * cache may be implemented at a later date which takes further advantage of these strict
3224 * dependency requirements.
3225 *
3226 * @return PPNode
3227 */
3228 public function preprocessToDom( $text, $flags = 0 ) {
3229 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3230 return $dom;
3231 }
3232
3233 /**
3234 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3235 *
3236 * @param string $s
3237 *
3238 * @return array
3239 * @deprecated since 1.34; appears to be unused.
3240 */
3241 public static function splitWhitespace( $s ) {
3242 wfDeprecated( __METHOD__, '1.34' );
3243 $ltrimmed = ltrim( $s );
3244 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3245 $trimmed = rtrim( $ltrimmed );
3246 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3247 if ( $diff > 0 ) {
3248 $w2 = substr( $ltrimmed, -$diff );
3249 } else {
3250 $w2 = '';
3251 }
3252 return [ $w1, $trimmed, $w2 ];
3253 }
3254
3255 /**
3256 * Replace magic variables, templates, and template arguments
3257 * with the appropriate text. Templates are substituted recursively,
3258 * taking care to avoid infinite loops.
3259 *
3260 * Note that the substitution depends on value of $mOutputType:
3261 * self::OT_WIKI: only {{subst:}} templates
3262 * self::OT_PREPROCESS: templates but not extension tags
3263 * self::OT_HTML: all templates and extension tags
3264 *
3265 * @param string $text The text to transform
3266 * @param false|PPFrame|array $frame Object describing the arguments passed to the
3267 * template. Arguments may also be provided as an associative array, as
3268 * was the usual case before MW1.12. Providing arguments this way may be
3269 * useful for extensions wishing to perform variable replacement
3270 * explicitly.
3271 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3272 * double-brace expansion.
3273 * @return string
3274 */
3275 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3276 # Is there any text? Also, Prevent too big inclusions!
3277 $textSize = strlen( $text );
3278 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3279 return $text;
3280 }
3281
3282 if ( $frame === false ) {
3283 $frame = $this->getPreprocessor()->newFrame();
3284 } elseif ( !( $frame instanceof PPFrame ) ) {
3285 $this->logger->debug(
3286 __METHOD__ . " called using plain parameters instead of " .
3287 "a PPFrame instance. Creating custom frame."
3288 );
3289 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3290 }
3291
3292 $dom = $this->preprocessToDom( $text );
3293 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3294 $text = $frame->expand( $dom, $flags );
3295
3296 return $text;
3297 }
3298
3299 /**
3300 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3301 *
3302 * @param array $args
3303 *
3304 * @return array
3305 * @deprecated since 1.34; appears to be unused in core.
3306 */
3307 public static function createAssocArgs( $args ) {
3308 wfDeprecated( __METHOD__, '1.34' );
3309 $assocArgs = [];
3310 $index = 1;
3311 foreach ( $args as $arg ) {
3312 $eqpos = strpos( $arg, '=' );
3313 if ( $eqpos === false ) {
3314 $assocArgs[$index++] = $arg;
3315 } else {
3316 $name = trim( substr( $arg, 0, $eqpos ) );
3317 $value = trim( substr( $arg, $eqpos + 1 ) );
3318 if ( $value === false ) {
3319 $value = '';
3320 }
3321 if ( $name !== false ) {
3322 $assocArgs[$name] = $value;
3323 }
3324 }
3325 }
3326
3327 return $assocArgs;
3328 }
3329
3330 /**
3331 * Warn the user when a parser limitation is reached
3332 * Will warn at most once the user per limitation type
3333 *
3334 * The results are shown during preview and run through the Parser (See EditPage.php)
3335 *
3336 * @param string $limitationType Should be one of:
3337 * 'expensive-parserfunction' (corresponding messages:
3338 * 'expensive-parserfunction-warning',
3339 * 'expensive-parserfunction-category')
3340 * 'post-expand-template-argument' (corresponding messages:
3341 * 'post-expand-template-argument-warning',
3342 * 'post-expand-template-argument-category')
3343 * 'post-expand-template-inclusion' (corresponding messages:
3344 * 'post-expand-template-inclusion-warning',
3345 * 'post-expand-template-inclusion-category')
3346 * 'node-count-exceeded' (corresponding messages:
3347 * 'node-count-exceeded-warning',
3348 * 'node-count-exceeded-category')
3349 * 'expansion-depth-exceeded' (corresponding messages:
3350 * 'expansion-depth-exceeded-warning',
3351 * 'expansion-depth-exceeded-category')
3352 * @param string|int|null $current Current value
3353 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3354 * exceeded, provide the values (optional)
3355 */
3356 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3357 # does no harm if $current and $max are present but are unnecessary for the message
3358 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3359 # only during preview, and that would split the parser cache unnecessarily.
3360 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3361 ->text();
3362 $this->mOutput->addWarning( $warning );
3363 $this->addTrackingCategory( "$limitationType-category" );
3364 }
3365
3366 /**
3367 * Return the text of a template, after recursively
3368 * replacing any variables or templates within the template.
3369 *
3370 * @param array $piece The parts of the template
3371 * $piece['title']: the title, i.e. the part before the |
3372 * $piece['parts']: the parameter array
3373 * $piece['lineStart']: whether the brace was at the start of a line
3374 * @param PPFrame $frame The current frame, contains template arguments
3375 * @throws Exception
3376 * @return string|array The text of the template
3377 * @internal
3378 */
3379 public function braceSubstitution( $piece, $frame ) {
3380 // Flags
3381
3382 // $text has been filled
3383 $found = false;
3384 // wiki markup in $text should be escaped
3385 $nowiki = false;
3386 // $text is HTML, armour it against wikitext transformation
3387 $isHTML = false;
3388 // Force interwiki transclusion to be done in raw mode not rendered
3389 $forceRawInterwiki = false;
3390 // $text is a DOM node needing expansion in a child frame
3391 $isChildObj = false;
3392 // $text is a DOM node needing expansion in the current frame
3393 $isLocalObj = false;
3394
3395 # Title object, where $text came from
3396 $title = false;
3397
3398 # $part1 is the bit before the first |, and must contain only title characters.
3399 # Various prefixes will be stripped from it later.
3400 $titleWithSpaces = $frame->expand( $piece['title'] );
3401 $part1 = trim( $titleWithSpaces );
3402 $titleText = false;
3403
3404 # Original title text preserved for various purposes
3405 $originalTitle = $part1;
3406
3407 # $args is a list of argument nodes, starting from index 0, not including $part1
3408 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3409 # below won't work b/c this $args isn't an object
3410 $args = ( $piece['parts'] == null ) ? [] : $piece['parts'];
3411
3412 $profileSection = null; // profile templates
3413
3414 # SUBST
3415 if ( !$found ) {
3416 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3417
3418 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3419 # Decide whether to expand template or keep wikitext as-is.
3420 if ( $this->ot['wiki'] ) {
3421 if ( $substMatch === false ) {
3422 $literal = true; # literal when in PST with no prefix
3423 } else {
3424 $literal = false; # expand when in PST with subst: or safesubst:
3425 }
3426 } else {
3427 if ( $substMatch == 'subst' ) {
3428 $literal = true; # literal when not in PST with plain subst:
3429 } else {
3430 $literal = false; # expand when not in PST with safesubst: or no prefix
3431 }
3432 }
3433 if ( $literal ) {
3434 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3435 $isLocalObj = true;
3436 $found = true;
3437 }
3438 }
3439
3440 # Variables
3441 if ( !$found && $args->getLength() == 0 ) {
3442 $id = $this->mVariables->matchStartToEnd( $part1 );
3443 if ( $id !== false ) {
3444 $text = $this->expandMagicVariable( $id, $frame );
3445 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3446 $this->mOutput->updateCacheExpiry(
3447 $this->magicWordFactory->getCacheTTL( $id ) );
3448 }
3449 $found = true;
3450 }
3451 }
3452
3453 # MSG, MSGNW and RAW
3454 if ( !$found ) {
3455 # Check for MSGNW:
3456 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3457 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3458 $nowiki = true;
3459 } else {
3460 # Remove obsolete MSG:
3461 $mwMsg = $this->magicWordFactory->get( 'msg' );
3462 $mwMsg->matchStartAndRemove( $part1 );
3463 }
3464
3465 # Check for RAW:
3466 $mwRaw = $this->magicWordFactory->get( 'raw' );
3467 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3468 $forceRawInterwiki = true;
3469 }
3470 }
3471
3472 # Parser functions
3473 if ( !$found ) {
3474 $colonPos = strpos( $part1, ':' );
3475 if ( $colonPos !== false ) {
3476 $func = substr( $part1, 0, $colonPos );
3477 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3478 $argsLength = $args->getLength();
3479 for ( $i = 0; $i < $argsLength; $i++ ) {
3480 $funcArgs[] = $args->item( $i );
3481 }
3482
3483 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3484
3485 // Extract any forwarded flags
3486 if ( isset( $result['title'] ) ) {
3487 $title = $result['title'];
3488 }
3489 if ( isset( $result['found'] ) ) {
3490 $found = $result['found'];
3491 }
3492 if ( array_key_exists( 'text', $result ) ) {
3493 // a string or null
3494 $text = $result['text'];
3495 }
3496 if ( isset( $result['nowiki'] ) ) {
3497 $nowiki = $result['nowiki'];
3498 }
3499 if ( isset( $result['isHTML'] ) ) {
3500 $isHTML = $result['isHTML'];
3501 }
3502 if ( isset( $result['forceRawInterwiki'] ) ) {
3503 $forceRawInterwiki = $result['forceRawInterwiki'];
3504 }
3505 if ( isset( $result['isChildObj'] ) ) {
3506 $isChildObj = $result['isChildObj'];
3507 }
3508 if ( isset( $result['isLocalObj'] ) ) {
3509 $isLocalObj = $result['isLocalObj'];
3510 }
3511 }
3512 }
3513
3514 # Finish mangling title and then check for loops.
3515 # Set $title to a Title object and $titleText to the PDBK
3516 if ( !$found ) {
3517 $ns = NS_TEMPLATE;
3518 # Split the title into page and subpage
3519 $subpage = '';
3520 $relative = Linker::normalizeSubpageLink(
3521 $this->mTitle, $part1, $subpage
3522 );
3523 if ( $part1 !== $relative ) {
3524 $part1 = $relative;
3525 $ns = $this->mTitle->getNamespace();
3526 }
3527 $title = Title::newFromText( $part1, $ns );
3528 if ( $title ) {
3529 $titleText = $title->getPrefixedText();
3530 # Check for language variants if the template is not found
3531 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3532 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3533 }
3534 # Do recursion depth check
3535 $limit = $this->mOptions->getMaxTemplateDepth();
3536 if ( $frame->depth >= $limit ) {
3537 $found = true;
3538 $text = '<span class="error">'
3539 . wfMessage( 'parser-template-recursion-depth-warning' )
3540 ->numParams( $limit )->inContentLanguage()->text()
3541 . '</span>';
3542 }
3543 }
3544 }
3545
3546 # Load from database
3547 if ( !$found && $title ) {
3548 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3549 if ( !$title->isExternal() ) {
3550 if ( $title->isSpecialPage()
3551 && $this->mOptions->getAllowSpecialInclusion()
3552 && $this->ot['html']
3553 ) {
3554 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3555 // Pass the template arguments as URL parameters.
3556 // "uselang" will have no effect since the Language object
3557 // is forced to the one defined in ParserOptions.
3558 $pageArgs = [];
3559 $argsLength = $args->getLength();
3560 for ( $i = 0; $i < $argsLength; $i++ ) {
3561 $bits = $args->item( $i )->splitArg();
3562 if ( strval( $bits['index'] ) === '' ) {
3563 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3564 $value = trim( $frame->expand( $bits['value'] ) );
3565 $pageArgs[$name] = $value;
3566 }
3567 }
3568
3569 // Create a new context to execute the special page
3570 $context = new RequestContext;
3571 $context->setTitle( $title );
3572 $context->setRequest( new FauxRequest( $pageArgs ) );
3573 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3574 $context->setUser( $this->getUser() );
3575 } else {
3576 // If this page is cached, then we better not be per user.
3577 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3578 }
3579 $context->setLanguage( $this->mOptions->getUserLangObj() );
3580 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3581 if ( $ret ) {
3582 $text = $context->getOutput()->getHTML();
3583 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3584 $found = true;
3585 $isHTML = true;
3586 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3587 $this->mOutput->updateRuntimeAdaptiveExpiry(
3588 $specialPage->maxIncludeCacheTime()
3589 );
3590 }
3591 }
3592 } elseif ( $this->nsInfo->isNonincludable( $title->getNamespace() ) ) {
3593 $found = false; # access denied
3594 $this->logger->debug(
3595 __METHOD__ .
3596 ": template inclusion denied for " . $title->getPrefixedDBkey()
3597 );
3598 } else {
3599 list( $text, $title ) = $this->getTemplateDom( $title );
3600 if ( $text !== false ) {
3601 $found = true;
3602 $isChildObj = true;
3603 }
3604 }
3605
3606 # If the title is valid but undisplayable, make a link to it
3607 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3608 $text = "[[:$titleText]]";
3609 $found = true;
3610 }
3611 } elseif ( $title->isTrans() ) {
3612 # Interwiki transclusion
3613 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3614 $text = $this->interwikiTransclude( $title, 'render' );
3615 $isHTML = true;
3616 } else {
3617 $text = $this->interwikiTransclude( $title, 'raw' );
3618 # Preprocess it like a template
3619 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3620 $isChildObj = true;
3621 }
3622 $found = true;
3623 }
3624
3625 # Do infinite loop check
3626 # This has to be done after redirect resolution to avoid infinite loops via redirects
3627 if ( !$frame->loopCheck( $title ) ) {
3628 $found = true;
3629 $text = '<span class="error">'
3630 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3631 . '</span>';
3632 $this->addTrackingCategory( 'template-loop-category' );
3633 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3634 wfEscapeWikiText( $titleText ) )->text() );
3635 $this->logger->debug( __METHOD__ . ": template loop broken at '$titleText'" );
3636 }
3637 }
3638
3639 # If we haven't found text to substitute by now, we're done
3640 # Recover the source wikitext and return it
3641 if ( !$found ) {
3642 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3643 if ( $profileSection ) {
3644 $this->mProfiler->scopedProfileOut( $profileSection );
3645 }
3646 return [ 'object' => $text ];
3647 }
3648
3649 # Expand DOM-style return values in a child frame
3650 if ( $isChildObj ) {
3651 # Clean up argument array
3652 $newFrame = $frame->newChild( $args, $title );
3653
3654 if ( $nowiki ) {
3655 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3656 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3657 # Expansion is eligible for the empty-frame cache
3658 $text = $newFrame->cachedExpand( $titleText, $text );
3659 } else {
3660 # Uncached expansion
3661 $text = $newFrame->expand( $text );
3662 }
3663 }
3664 if ( $isLocalObj && $nowiki ) {
3665 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3666 $isLocalObj = false;
3667 }
3668
3669 if ( $profileSection ) {
3670 $this->mProfiler->scopedProfileOut( $profileSection );
3671 }
3672
3673 # Replace raw HTML by a placeholder
3674 if ( $isHTML ) {
3675 $text = $this->insertStripItem( $text );
3676 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3677 # Escape nowiki-style return values
3678 $text = wfEscapeWikiText( $text );
3679 } elseif ( is_string( $text )
3680 && !$piece['lineStart']
3681 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3682 ) {
3683 # T2529: if the template begins with a table or block-level
3684 # element, it should be treated as beginning a new line.
3685 # This behavior is somewhat controversial.
3686 $text = "\n" . $text;
3687 }
3688
3689 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3690 # Error, oversize inclusion
3691 if ( $titleText !== false ) {
3692 # Make a working, properly escaped link if possible (T25588)
3693 $text = "[[:$titleText]]";
3694 } else {
3695 # This will probably not be a working link, but at least it may
3696 # provide some hint of where the problem is
3697 preg_replace( '/^:/', '', $originalTitle );
3698 $text = "[[:$originalTitle]]";
3699 }
3700 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3701 . 'post-expand include size too large -->' );
3702 $this->limitationWarn( 'post-expand-template-inclusion' );
3703 }
3704
3705 if ( $isLocalObj ) {
3706 $ret = [ 'object' => $text ];
3707 } else {
3708 $ret = [ 'text' => $text ];
3709 }
3710
3711 return $ret;
3712 }
3713
3714 /**
3715 * Call a parser function and return an array with text and flags.
3716 *
3717 * The returned array will always contain a boolean 'found', indicating
3718 * whether the parser function was found or not. It may also contain the
3719 * following:
3720 * text: string|object, resulting wikitext or PP DOM object
3721 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3722 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3723 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3724 * nowiki: bool, wiki markup in $text should be escaped
3725 *
3726 * @since 1.21
3727 * @param PPFrame $frame The current frame, contains template arguments
3728 * @param string $function Function name
3729 * @param array $args Arguments to the function
3730 * @throws MWException
3731 * @return array
3732 */
3733 public function callParserFunction( $frame, $function, array $args = [] ) {
3734 # Case sensitive functions
3735 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3736 $function = $this->mFunctionSynonyms[1][$function];
3737 } else {
3738 # Case insensitive functions
3739 $function = $this->contLang->lc( $function );
3740 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3741 $function = $this->mFunctionSynonyms[0][$function];
3742 } else {
3743 return [ 'found' => false ];
3744 }
3745 }
3746
3747 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3748
3749 // Avoid PHP 7.1 warning from passing $this by reference
3750 $parser = $this;
3751
3752 $allArgs = [ &$parser ];
3753 if ( $flags & self::SFH_OBJECT_ARGS ) {
3754 # Convert arguments to PPNodes and collect for appending to $allArgs
3755 $funcArgs = [];
3756 foreach ( $args as $k => $v ) {
3757 if ( $v instanceof PPNode || $k === 0 ) {
3758 $funcArgs[] = $v;
3759 } else {
3760 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3761 }
3762 }
3763
3764 # Add a frame parameter, and pass the arguments as an array
3765 $allArgs[] = $frame;
3766 $allArgs[] = $funcArgs;
3767 } else {
3768 # Convert arguments to plain text and append to $allArgs
3769 foreach ( $args as $k => $v ) {
3770 if ( $v instanceof PPNode ) {
3771 $allArgs[] = trim( $frame->expand( $v ) );
3772 } elseif ( is_int( $k ) && $k >= 0 ) {
3773 $allArgs[] = trim( $v );
3774 } else {
3775 $allArgs[] = trim( "$k=$v" );
3776 }
3777 }
3778 }
3779
3780 $result = $callback( ...$allArgs );
3781
3782 # The interface for function hooks allows them to return a wikitext
3783 # string or an array containing the string and any flags. This mungs
3784 # things around to match what this method should return.
3785 if ( !is_array( $result ) ) {
3786 $result = [
3787 'found' => true,
3788 'text' => $result,
3789 ];
3790 } else {
3791 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3792 $result['text'] = $result[0];
3793 }
3794 unset( $result[0] );
3795 $result += [
3796 'found' => true,
3797 ];
3798 }
3799
3800 $noparse = true;
3801 $preprocessFlags = 0;
3802 if ( isset( $result['noparse'] ) ) {
3803 $noparse = $result['noparse'];
3804 }
3805 if ( isset( $result['preprocessFlags'] ) ) {
3806 $preprocessFlags = $result['preprocessFlags'];
3807 }
3808
3809 if ( !$noparse ) {
3810 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3811 $result['isChildObj'] = true;
3812 }
3813
3814 return $result;
3815 }
3816
3817 /**
3818 * Get the semi-parsed DOM representation of a template with a given title,
3819 * and its redirect destination title. Cached.
3820 *
3821 * @param Title $title
3822 *
3823 * @return array
3824 */
3825 public function getTemplateDom( $title ) {
3826 $cacheTitle = $title;
3827 $titleText = $title->getPrefixedDBkey();
3828
3829 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3830 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3831 $title = Title::makeTitle( $ns, $dbk );
3832 $titleText = $title->getPrefixedDBkey();
3833 }
3834 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3835 return [ $this->mTplDomCache[$titleText], $title ];
3836 }
3837
3838 # Cache miss, go to the database
3839 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3840
3841 if ( $text === false ) {
3842 $this->mTplDomCache[$titleText] = false;
3843 return [ false, $title ];
3844 }
3845
3846 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3847 $this->mTplDomCache[$titleText] = $dom;
3848
3849 if ( !$title->equals( $cacheTitle ) ) {
3850 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3851 [ $title->getNamespace(), $title->getDBkey() ];
3852 }
3853
3854 return [ $dom, $title ];
3855 }
3856
3857 /**
3858 * Fetch the current revision of a given title. Note that the revision
3859 * (and even the title) may not exist in the database, so everything
3860 * contributing to the output of the parser should use this method
3861 * where possible, rather than getting the revisions themselves. This
3862 * method also caches its results, so using it benefits performance.
3863 *
3864 * @since 1.24
3865 * @param Title $title
3866 * @return Revision
3867 */
3868 public function fetchCurrentRevisionOfTitle( $title ) {
3869 $cacheKey = $title->getPrefixedDBkey();
3870 if ( !$this->currentRevisionCache ) {
3871 $this->currentRevisionCache = new MapCacheLRU( 100 );
3872 }
3873 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3874 $this->currentRevisionCache->set( $cacheKey,
3875 // Defaults to Parser::statelessFetchRevision()
3876 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3877 );
3878 }
3879 return $this->currentRevisionCache->get( $cacheKey );
3880 }
3881
3882 /**
3883 * @param Title $title
3884 * @return bool
3885 * @since 1.34
3886 */
3887 public function isCurrentRevisionOfTitleCached( $title ) {
3888 return (
3889 $this->currentRevisionCache &&
3890 $this->currentRevisionCache->has( $title->getPrefixedText() )
3891 );
3892 }
3893
3894 /**
3895 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3896 * without passing them on to it.
3897 *
3898 * @since 1.24
3899 * @param Title $title
3900 * @param Parser|bool $parser
3901 * @return Revision|bool False if missing
3902 */
3903 public static function statelessFetchRevision( Title $title, $parser = false ) {
3904 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3905
3906 return $rev;
3907 }
3908
3909 /**
3910 * Fetch the unparsed text of a template and register a reference to it.
3911 * @param Title $title
3912 * @return array ( string or false, Title )
3913 */
3914 public function fetchTemplateAndTitle( $title ) {
3915 // Defaults to Parser::statelessFetchTemplate()
3916 $templateCb = $this->mOptions->getTemplateCallback();
3917 $stuff = call_user_func( $templateCb, $title, $this );
3918 $rev = $stuff['revision'] ?? null;
3919 $text = $stuff['text'];
3920 if ( is_string( $stuff['text'] ) ) {
3921 // We use U+007F DELETE to distinguish strip markers from regular text
3922 $text = strtr( $text, "\x7f", "?" );
3923 }
3924 $finalTitle = $stuff['finalTitle'] ?? $title;
3925 foreach ( ( $stuff['deps'] ?? [] ) as $dep ) {
3926 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3927 if ( $dep['title']->equals( $this->getTitle() ) && $rev instanceof Revision ) {
3928 // Self-transclusion; final result may change based on the new page version
3929 $this->setOutputFlag( 'vary-revision-sha1', 'Self transclusion' );
3930 $this->getOutput()->setRevisionUsedSha1Base36( $rev->getSha1() );
3931 }
3932 }
3933
3934 return [ $text, $finalTitle ];
3935 }
3936
3937 /**
3938 * Fetch the unparsed text of a template and register a reference to it.
3939 * @param Title $title
3940 * @return string|bool
3941 */
3942 public function fetchTemplate( $title ) {
3943 return $this->fetchTemplateAndTitle( $title )[0];
3944 }
3945
3946 /**
3947 * Static function to get a template
3948 * Can be overridden via ParserOptions::setTemplateCallback().
3949 *
3950 * @param Title $title
3951 * @param bool|Parser $parser
3952 *
3953 * @return array
3954 */
3955 public static function statelessFetchTemplate( $title, $parser = false ) {
3956 $text = $skip = false;
3957 $finalTitle = $title;
3958 $deps = [];
3959 $rev = null;
3960
3961 # Loop to fetch the article, with up to 1 redirect
3962 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3963 # Give extensions a chance to select the revision instead
3964 $id = false; # Assume current
3965 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3966 [ $parser, $title, &$skip, &$id ] );
3967
3968 if ( $skip ) {
3969 $text = false;
3970 $deps[] = [
3971 'title' => $title,
3972 'page_id' => $title->getArticleID(),
3973 'rev_id' => null
3974 ];
3975 break;
3976 }
3977 # Get the revision
3978 if ( $id ) {
3979 $rev = Revision::newFromId( $id );
3980 } elseif ( $parser ) {
3981 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3982 } else {
3983 $rev = Revision::newFromTitle( $title );
3984 }
3985 $rev_id = $rev ? $rev->getId() : 0;
3986 # If there is no current revision, there is no page
3987 if ( $id === false && !$rev ) {
3988 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3989 $linkCache->addBadLinkObj( $title );
3990 }
3991
3992 $deps[] = [
3993 'title' => $title,
3994 'page_id' => $title->getArticleID(),
3995 'rev_id' => $rev_id
3996 ];
3997 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3998 # We fetched a rev from a different title; register it too...
3999 $deps[] = [
4000 'title' => $rev->getTitle(),
4001 'page_id' => $rev->getPage(),
4002 'rev_id' => $rev_id
4003 ];
4004 }
4005
4006 if ( $rev ) {
4007 $content = $rev->getContent();
4008 $text = $content ? $content->getWikitextForTransclusion() : null;
4009
4010 Hooks::run( 'ParserFetchTemplate',
4011 [ $parser, $title, $rev, &$text, &$deps ] );
4012
4013 if ( $text === false || $text === null ) {
4014 $text = false;
4015 break;
4016 }
4017 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
4018 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
4019 lcfirst( $title->getText() ) )->inContentLanguage();
4020 if ( !$message->exists() ) {
4021 $text = false;
4022 break;
4023 }
4024 $content = $message->content();
4025 $text = $message->plain();
4026 } else {
4027 break;
4028 }
4029 if ( !$content ) {
4030 break;
4031 }
4032 # Redirect?
4033 $finalTitle = $title;
4034 $title = $content->getRedirectTarget();
4035 }
4036 return [
4037 'revision' => $rev,
4038 'text' => $text,
4039 'finalTitle' => $finalTitle,
4040 'deps' => $deps
4041 ];
4042 }
4043
4044 /**
4045 * Fetch a file and its title and register a reference to it.
4046 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
4047 * @param Title $title
4048 * @param array $options Array of options to RepoGroup::findFile
4049 * @return array ( File or false, Title of file )
4050 */
4051 public function fetchFileAndTitle( $title, $options = [] ) {
4052 $file = $this->fetchFileNoRegister( $title, $options );
4053
4054 $time = $file ? $file->getTimestamp() : false;
4055 $sha1 = $file ? $file->getSha1() : false;
4056 # Register the file as a dependency...
4057 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4058 if ( $file && !$title->equals( $file->getTitle() ) ) {
4059 # Update fetched file title
4060 $title = $file->getTitle();
4061 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4062 }
4063 return [ $file, $title ];
4064 }
4065
4066 /**
4067 * Helper function for fetchFileAndTitle.
4068 *
4069 * Also useful if you need to fetch a file but not use it yet,
4070 * for example to get the file's handler.
4071 *
4072 * @param Title $title
4073 * @param array $options Array of options to RepoGroup::findFile
4074 * @return File|bool
4075 */
4076 protected function fetchFileNoRegister( $title, $options = [] ) {
4077 if ( isset( $options['broken'] ) ) {
4078 $file = false; // broken thumbnail forced by hook
4079 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
4080 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
4081 } else { // get by (name,timestamp)
4082 $file = MediaWikiServices::getInstance()->getRepoGroup()->findFile( $title, $options );
4083 }
4084 return $file;
4085 }
4086
4087 /**
4088 * Transclude an interwiki link.
4089 *
4090 * @param Title $title
4091 * @param string $action Usually one of (raw, render)
4092 *
4093 * @return string
4094 */
4095 public function interwikiTransclude( $title, $action ) {
4096 if ( !$this->svcOptions->get( 'EnableScaryTranscluding' ) ) {
4097 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
4098 }
4099
4100 $url = $title->getFullURL( [ 'action' => $action ] );
4101 if ( strlen( $url ) > 1024 ) {
4102 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
4103 }
4104
4105 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
4106
4107 $fname = __METHOD__;
4108 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
4109
4110 $data = $cache->getWithSetCallback(
4111 $cache->makeGlobalKey(
4112 'interwiki-transclude',
4113 ( $wikiId !== false ) ? $wikiId : 'external',
4114 sha1( $url )
4115 ),
4116 $this->svcOptions->get( 'TranscludeCacheExpiry' ),
4117 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
4118 $req = MWHttpRequest::factory( $url, [], $fname );
4119
4120 $status = $req->execute(); // Status object
4121 if ( !$status->isOK() ) {
4122 $ttl = $cache::TTL_UNCACHEABLE;
4123 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
4124 $ttl = min( $cache::TTL_LAGGED, $ttl );
4125 }
4126
4127 return [
4128 'text' => $status->isOK() ? $req->getContent() : null,
4129 'code' => $req->getStatus()
4130 ];
4131 },
4132 [
4133 'checkKeys' => ( $wikiId !== false )
4134 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
4135 : [],
4136 'pcGroup' => 'interwiki-transclude:5',
4137 'pcTTL' => $cache::TTL_PROC_LONG
4138 ]
4139 );
4140
4141 if ( is_string( $data['text'] ) ) {
4142 $text = $data['text'];
4143 } elseif ( $data['code'] != 200 ) {
4144 // Though we failed to fetch the content, this status is useless.
4145 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
4146 ->params( $url, $data['code'] )->inContentLanguage()->text();
4147 } else {
4148 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4149 }
4150
4151 return $text;
4152 }
4153
4154 /**
4155 * Triple brace replacement -- used for template arguments
4156 * @internal
4157 *
4158 * @param array $piece
4159 * @param PPFrame $frame
4160 *
4161 * @return array
4162 */
4163 public function argSubstitution( $piece, $frame ) {
4164 $error = false;
4165 $parts = $piece['parts'];
4166 $nameWithSpaces = $frame->expand( $piece['title'] );
4167 $argName = trim( $nameWithSpaces );
4168 $object = false;
4169 $text = $frame->getArgument( $argName );
4170 if ( $text === false && $parts->getLength() > 0
4171 && ( $this->ot['html']
4172 || $this->ot['pre']
4173 || ( $this->ot['wiki'] && $frame->isTemplate() )
4174 )
4175 ) {
4176 # No match in frame, use the supplied default
4177 $object = $parts->item( 0 )->getChildren();
4178 }
4179 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4180 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4181 $this->limitationWarn( 'post-expand-template-argument' );
4182 }
4183
4184 if ( $text === false && $object === false ) {
4185 # No match anywhere
4186 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4187 }
4188 if ( $error !== false ) {
4189 $text .= $error;
4190 }
4191 if ( $object !== false ) {
4192 $ret = [ 'object' => $object ];
4193 } else {
4194 $ret = [ 'text' => $text ];
4195 }
4196
4197 return $ret;
4198 }
4199
4200 /**
4201 * Return the text to be used for a given extension tag.
4202 * This is the ghost of strip().
4203 *
4204 * @param array $params Associative array of parameters:
4205 * name PPNode for the tag name
4206 * attr PPNode for unparsed text where tag attributes are thought to be
4207 * attributes Optional associative array of parsed attributes
4208 * inner Contents of extension element
4209 * noClose Original text did not have a close tag
4210 * @param PPFrame $frame
4211 *
4212 * @throws MWException
4213 * @return string
4214 * @internal
4215 */
4216 public function extensionSubstitution( $params, $frame ) {
4217 static $errorStr = '<span class="error">';
4218 static $errorLen = 20;
4219
4220 $name = $frame->expand( $params['name'] );
4221 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4222 // Probably expansion depth or node count exceeded. Just punt the
4223 // error up.
4224 return $name;
4225 }
4226
4227 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
4228 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4229 // See above
4230 return $attrText;
4231 }
4232
4233 // We can't safely check if the expansion for $content resulted in an
4234 // error, because the content could happen to be the error string
4235 // (T149622).
4236 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
4237
4238 $marker = self::MARKER_PREFIX . "-$name-"
4239 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
4240
4241 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4242 ( $this->ot['html'] || $this->ot['pre'] );
4243 if ( $isFunctionTag ) {
4244 $markerType = 'none';
4245 } else {
4246 $markerType = 'general';
4247 }
4248 if ( $this->ot['html'] || $isFunctionTag ) {
4249 $name = strtolower( $name );
4250 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4251 if ( isset( $params['attributes'] ) ) {
4252 $attributes += $params['attributes'];
4253 }
4254
4255 if ( isset( $this->mTagHooks[$name] ) ) {
4256 $output = call_user_func_array( $this->mTagHooks[$name],
4257 [ $content, $attributes, $this, $frame ] );
4258 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4259 list( $callback, ) = $this->mFunctionTagHooks[$name];
4260
4261 // Avoid PHP 7.1 warning from passing $this by reference
4262 $parser = $this;
4263 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4264 } else {
4265 $output = '<span class="error">Invalid tag extension name: ' .
4266 htmlspecialchars( $name ) . '</span>';
4267 }
4268
4269 if ( is_array( $output ) ) {
4270 // Extract flags
4271 $flags = $output;
4272 $output = $flags[0];
4273 if ( isset( $flags['markerType'] ) ) {
4274 $markerType = $flags['markerType'];
4275 }
4276 }
4277 } else {
4278 if ( is_null( $attrText ) ) {
4279 $attrText = '';
4280 }
4281 if ( isset( $params['attributes'] ) ) {
4282 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4283 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4284 htmlspecialchars( $attrValue ) . '"';
4285 }
4286 }
4287 if ( $content === null ) {
4288 $output = "<$name$attrText/>";
4289 } else {
4290 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4291 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4292 // See above
4293 return $close;
4294 }
4295 $output = "<$name$attrText>$content$close";
4296 }
4297 }
4298
4299 if ( $markerType === 'none' ) {
4300 return $output;
4301 } elseif ( $markerType === 'nowiki' ) {
4302 $this->mStripState->addNoWiki( $marker, $output );
4303 } elseif ( $markerType === 'general' ) {
4304 $this->mStripState->addGeneral( $marker, $output );
4305 } else {
4306 throw new MWException( __METHOD__ . ': invalid marker type' );
4307 }
4308 return $marker;
4309 }
4310
4311 /**
4312 * Increment an include size counter
4313 *
4314 * @param string $type The type of expansion
4315 * @param int $size The size of the text
4316 * @return bool False if this inclusion would take it over the maximum, true otherwise
4317 */
4318 public function incrementIncludeSize( $type, $size ) {
4319 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4320 return false;
4321 } else {
4322 $this->mIncludeSizes[$type] += $size;
4323 return true;
4324 }
4325 }
4326
4327 /**
4328 * Increment the expensive function count
4329 *
4330 * @return bool False if the limit has been exceeded
4331 */
4332 public function incrementExpensiveFunctionCount() {
4333 $this->mExpensiveFunctionCount++;
4334 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4335 }
4336
4337 /**
4338 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4339 * Fills $this->mDoubleUnderscores, returns the modified text
4340 *
4341 * @param string $text
4342 * @return string
4343 * @deprecated since 1.34; should not be used outside parser class.
4344 */
4345 public function doDoubleUnderscore( $text ) {
4346 wfDeprecated( __METHOD__, '1.34' );
4347 return $this->handleDoubleUnderscore( $text );
4348 }
4349
4350 /**
4351 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4352 * Fills $this->mDoubleUnderscores, returns the modified text
4353 *
4354 * @param string $text
4355 * @return string
4356 */
4357 private function handleDoubleUnderscore( $text ) {
4358 # The position of __TOC__ needs to be recorded
4359 $mw = $this->magicWordFactory->get( 'toc' );
4360 if ( $mw->match( $text ) ) {
4361 $this->mShowToc = true;
4362 $this->mForceTocPosition = true;
4363
4364 # Set a placeholder. At the end we'll fill it in with the TOC.
4365 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4366
4367 # Only keep the first one.
4368 $text = $mw->replace( '', $text );
4369 }
4370
4371 # Now match and remove the rest of them
4372 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4373 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4374
4375 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4376 $this->mOutput->mNoGallery = true;
4377 }
4378 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4379 $this->mShowToc = false;
4380 }
4381 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4382 && $this->mTitle->getNamespace() == NS_CATEGORY
4383 ) {
4384 $this->addTrackingCategory( 'hidden-category-category' );
4385 }
4386 # (T10068) Allow control over whether robots index a page.
4387 # __INDEX__ always overrides __NOINDEX__, see T16899
4388 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4389 $this->mOutput->setIndexPolicy( 'noindex' );
4390 $this->addTrackingCategory( 'noindex-category' );
4391 }
4392 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4393 $this->mOutput->setIndexPolicy( 'index' );
4394 $this->addTrackingCategory( 'index-category' );
4395 }
4396
4397 # Cache all double underscores in the database
4398 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4399 $this->mOutput->setProperty( $key, '' );
4400 }
4401
4402 return $text;
4403 }
4404
4405 /**
4406 * @see ParserOutput::addTrackingCategory()
4407 * @param string $msg Message key
4408 * @return bool Whether the addition was successful
4409 */
4410 public function addTrackingCategory( $msg ) {
4411 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4412 }
4413
4414 /**
4415 * This function accomplishes several tasks:
4416 * 1) Auto-number headings if that option is enabled
4417 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4418 * 3) Add a Table of contents on the top for users who have enabled the option
4419 * 4) Auto-anchor headings
4420 *
4421 * It loops through all headlines, collects the necessary data, then splits up the
4422 * string and re-inserts the newly formatted headlines.
4423 *
4424 * @param string $text
4425 * @param string $origText Original, untouched wikitext
4426 * @param bool $isMain
4427 * @return mixed|string
4428 * @private
4429 * @deprecated since 1.34; should not be used outside parser class.
4430 */
4431 public function formatHeadings( $text, $origText, $isMain = true ) {
4432 wfDeprecated( __METHOD__, '1.34' );
4433 return $this->finalizeHeadings( $text, $origText, $isMain );
4434 }
4435
4436 /**
4437 * This function accomplishes several tasks:
4438 * 1) Auto-number headings if that option is enabled
4439 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4440 * 3) Add a Table of contents on the top for users who have enabled the option
4441 * 4) Auto-anchor headings
4442 *
4443 * It loops through all headlines, collects the necessary data, then splits up the
4444 * string and re-inserts the newly formatted headlines.
4445 *
4446 * @param string $text
4447 * @param string $origText Original, untouched wikitext
4448 * @param bool $isMain
4449 * @return mixed|string
4450 */
4451 private function finalizeHeadings( $text, $origText, $isMain = true ) {
4452 # Inhibit editsection links if requested in the page
4453 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4454 $maybeShowEditLink = false;
4455 } else {
4456 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4457 }
4458
4459 # Get all headlines for numbering them and adding funky stuff like [edit]
4460 # links - this is for later, but we need the number of headlines right now
4461 # NOTE: white space in headings have been trimmed in handleHeadings. They shouldn't
4462 # be trimmed here since whitespace in HTML headings is significant.
4463 $matches = [];
4464 $numMatches = preg_match_all(
4465 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4466 $text,
4467 $matches
4468 );
4469
4470 # if there are fewer than 4 headlines in the article, do not show TOC
4471 # unless it's been explicitly enabled.
4472 $enoughToc = $this->mShowToc &&
4473 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4474
4475 # Allow user to stipulate that a page should have a "new section"
4476 # link added via __NEWSECTIONLINK__
4477 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4478 $this->mOutput->setNewSection( true );
4479 }
4480
4481 # Allow user to remove the "new section"
4482 # link via __NONEWSECTIONLINK__
4483 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4484 $this->mOutput->hideNewSection( true );
4485 }
4486
4487 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4488 # override above conditions and always show TOC above first header
4489 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4490 $this->mShowToc = true;
4491 $enoughToc = true;
4492 }
4493
4494 # headline counter
4495 $headlineCount = 0;
4496 $numVisible = 0;
4497
4498 # Ugh .. the TOC should have neat indentation levels which can be
4499 # passed to the skin functions. These are determined here
4500 $toc = '';
4501 $full = '';
4502 $head = [];
4503 $sublevelCount = [];
4504 $levelCount = [];
4505 $level = 0;
4506 $prevlevel = 0;
4507 $toclevel = 0;
4508 $prevtoclevel = 0;
4509 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4510 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4511 $oldType = $this->mOutputType;
4512 $this->setOutputType( self::OT_WIKI );
4513 $frame = $this->getPreprocessor()->newFrame();
4514 $root = $this->preprocessToDom( $origText );
4515 $node = $root->getFirstChild();
4516 $byteOffset = 0;
4517 $tocraw = [];
4518 $refers = [];
4519
4520 $headlines = $numMatches !== false ? $matches[3] : [];
4521
4522 $maxTocLevel = $this->svcOptions->get( 'MaxTocLevel' );
4523 foreach ( $headlines as $headline ) {
4524 $isTemplate = false;
4525 $titleText = false;
4526 $sectionIndex = false;
4527 $numbering = '';
4528 $markerMatches = [];
4529 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4530 $serial = $markerMatches[1];
4531 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4532 $isTemplate = ( $titleText != $baseTitleText );
4533 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4534 }
4535
4536 if ( $toclevel ) {
4537 $prevlevel = $level;
4538 }
4539 $level = $matches[1][$headlineCount];
4540
4541 if ( $level > $prevlevel ) {
4542 # Increase TOC level
4543 $toclevel++;
4544 $sublevelCount[$toclevel] = 0;
4545 if ( $toclevel < $maxTocLevel ) {
4546 $prevtoclevel = $toclevel;
4547 $toc .= Linker::tocIndent();
4548 $numVisible++;
4549 }
4550 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4551 # Decrease TOC level, find level to jump to
4552
4553 for ( $i = $toclevel; $i > 0; $i-- ) {
4554 if ( $levelCount[$i] == $level ) {
4555 # Found last matching level
4556 $toclevel = $i;
4557 break;
4558 } elseif ( $levelCount[$i] < $level ) {
4559 # Found first matching level below current level
4560 $toclevel = $i + 1;
4561 break;
4562 }
4563 }
4564 if ( $i == 0 ) {
4565 $toclevel = 1;
4566 }
4567 if ( $toclevel < $maxTocLevel ) {
4568 if ( $prevtoclevel < $maxTocLevel ) {
4569 # Unindent only if the previous toc level was shown :p
4570 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4571 $prevtoclevel = $toclevel;
4572 } else {
4573 $toc .= Linker::tocLineEnd();
4574 }
4575 }
4576 } else {
4577 # No change in level, end TOC line
4578 if ( $toclevel < $maxTocLevel ) {
4579 $toc .= Linker::tocLineEnd();
4580 }
4581 }
4582
4583 $levelCount[$toclevel] = $level;
4584
4585 # count number of headlines for each level
4586 $sublevelCount[$toclevel]++;
4587 $dot = 0;
4588 for ( $i = 1; $i <= $toclevel; $i++ ) {
4589 if ( !empty( $sublevelCount[$i] ) ) {
4590 if ( $dot ) {
4591 $numbering .= '.';
4592 }
4593 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4594 $dot = 1;
4595 }
4596 }
4597
4598 # The safe header is a version of the header text safe to use for links
4599
4600 # Remove link placeholders by the link text.
4601 # <!--LINK number-->
4602 # turns into
4603 # link text with suffix
4604 # Do this before unstrip since link text can contain strip markers
4605 $safeHeadline = $this->replaceLinkHoldersTextPrivate( $headline );
4606
4607 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4608 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4609
4610 # Remove any <style> or <script> tags (T198618)
4611 $safeHeadline = preg_replace(
4612 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4613 '',
4614 $safeHeadline
4615 );
4616
4617 # Strip out HTML (first regex removes any tag not allowed)
4618 # Allowed tags are:
4619 # * <sup> and <sub> (T10393)
4620 # * <i> (T28375)
4621 # * <b> (r105284)
4622 # * <bdi> (T74884)
4623 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4624 # * <s> and <strike> (T35715)
4625 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4626 # to allow setting directionality in toc items.
4627 $tocline = preg_replace(
4628 [
4629 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4630 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4631 ],
4632 [ '', '<$1>' ],
4633 $safeHeadline
4634 );
4635
4636 # Strip '<span></span>', which is the result from the above if
4637 # <span id="foo"></span> is used to produce an additional anchor
4638 # for a section.
4639 $tocline = str_replace( '<span></span>', '', $tocline );
4640
4641 $tocline = trim( $tocline );
4642
4643 # For the anchor, strip out HTML-y stuff period
4644 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4645 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4646
4647 # Save headline for section edit hint before it's escaped
4648 $headlineHint = $safeHeadline;
4649
4650 # Decode HTML entities
4651 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4652
4653 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4654
4655 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4656 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4657 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4658 if ( $fallbackHeadline === $safeHeadline ) {
4659 # No reason to have both (in fact, we can't)
4660 $fallbackHeadline = false;
4661 }
4662
4663 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4664 # @todo FIXME: We may be changing them depending on the current locale.
4665 $arrayKey = strtolower( $safeHeadline );
4666 if ( $fallbackHeadline === false ) {
4667 $fallbackArrayKey = false;
4668 } else {
4669 $fallbackArrayKey = strtolower( $fallbackHeadline );
4670 }
4671
4672 # Create the anchor for linking from the TOC to the section
4673 $anchor = $safeHeadline;
4674 $fallbackAnchor = $fallbackHeadline;
4675 if ( isset( $refers[$arrayKey] ) ) {
4676 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4677 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4678 $anchor .= "_$i";
4679 $linkAnchor .= "_$i";
4680 $refers["${arrayKey}_$i"] = true;
4681 } else {
4682 $refers[$arrayKey] = true;
4683 }
4684 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4685 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4686 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4687 $fallbackAnchor .= "_$i";
4688 $refers["${fallbackArrayKey}_$i"] = true;
4689 } else {
4690 $refers[$fallbackArrayKey] = true;
4691 }
4692
4693 # Don't number the heading if it is the only one (looks silly)
4694 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4695 # the two are different if the line contains a link
4696 $headline = Html::element(
4697 'span',
4698 [ 'class' => 'mw-headline-number' ],
4699 $numbering
4700 ) . ' ' . $headline;
4701 }
4702
4703 if ( $enoughToc && ( !isset( $maxTocLevel ) || $toclevel < $maxTocLevel ) ) {
4704 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4705 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4706 }
4707
4708 # Add the section to the section tree
4709 # Find the DOM node for this header
4710 $noOffset = ( $isTemplate || $sectionIndex === false );
4711 while ( $node && !$noOffset ) {
4712 if ( $node->getName() === 'h' ) {
4713 $bits = $node->splitHeading();
4714 if ( $bits['i'] == $sectionIndex ) {
4715 break;
4716 }
4717 }
4718 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4719 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4720 $node = $node->getNextSibling();
4721 }
4722 $tocraw[] = [
4723 'toclevel' => $toclevel,
4724 'level' => $level,
4725 'line' => $tocline,
4726 'number' => $numbering,
4727 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4728 'fromtitle' => $titleText,
4729 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4730 'anchor' => $anchor,
4731 ];
4732
4733 # give headline the correct <h#> tag
4734 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4735 // Output edit section links as markers with styles that can be customized by skins
4736 if ( $isTemplate ) {
4737 # Put a T flag in the section identifier, to indicate to extractSections()
4738 # that sections inside <includeonly> should be counted.
4739 $editsectionPage = $titleText;
4740 $editsectionSection = "T-$sectionIndex";
4741 $editsectionContent = null;
4742 } else {
4743 $editsectionPage = $this->mTitle->getPrefixedText();
4744 $editsectionSection = $sectionIndex;
4745 $editsectionContent = $headlineHint;
4746 }
4747 // We use a bit of pesudo-xml for editsection markers. The
4748 // language converter is run later on. Using a UNIQ style marker
4749 // leads to the converter screwing up the tokens when it
4750 // converts stuff. And trying to insert strip tags fails too. At
4751 // this point all real inputted tags have already been escaped,
4752 // so we don't have to worry about a user trying to input one of
4753 // these markers directly. We use a page and section attribute
4754 // to stop the language converter from converting these
4755 // important bits of data, but put the headline hint inside a
4756 // content block because the language converter is supposed to
4757 // be able to convert that piece of data.
4758 // Gets replaced with html in ParserOutput::getText
4759 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4760 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4761 if ( $editsectionContent !== null ) {
4762 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4763 } else {
4764 $editlink .= '/>';
4765 }
4766 } else {
4767 $editlink = '';
4768 }
4769 $head[$headlineCount] = Linker::makeHeadline( $level,
4770 $matches['attrib'][$headlineCount], $anchor, $headline,
4771 $editlink, $fallbackAnchor );
4772
4773 $headlineCount++;
4774 }
4775
4776 $this->setOutputType( $oldType );
4777
4778 # Never ever show TOC if no headers
4779 if ( $numVisible < 1 ) {
4780 $enoughToc = false;
4781 }
4782
4783 if ( $enoughToc ) {
4784 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4785 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4786 }
4787 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4788 $this->mOutput->setTOCHTML( $toc );
4789 $toc = self::TOC_START . $toc . self::TOC_END;
4790 }
4791
4792 if ( $isMain ) {
4793 $this->mOutput->setSections( $tocraw );
4794 }
4795
4796 # split up and insert constructed headlines
4797 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4798 $i = 0;
4799
4800 // build an array of document sections
4801 $sections = [];
4802 foreach ( $blocks as $block ) {
4803 // $head is zero-based, sections aren't.
4804 if ( empty( $head[$i - 1] ) ) {
4805 $sections[$i] = $block;
4806 } else {
4807 $sections[$i] = $head[$i - 1] . $block;
4808 }
4809
4810 /**
4811 * Send a hook, one per section.
4812 * The idea here is to be able to make section-level DIVs, but to do so in a
4813 * lower-impact, more correct way than r50769
4814 *
4815 * $this : caller
4816 * $section : the section number
4817 * &$sectionContent : ref to the content of the section
4818 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4819 */
4820 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4821
4822 $i++;
4823 }
4824
4825 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4826 // append the TOC at the beginning
4827 // Top anchor now in skin
4828 $sections[0] .= $toc . "\n";
4829 }
4830
4831 $full .= implode( '', $sections );
4832
4833 if ( $this->mForceTocPosition ) {
4834 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4835 } else {
4836 return $full;
4837 }
4838 }
4839
4840 /**
4841 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4842 * conversion, substituting signatures, {{subst:}} templates, etc.
4843 *
4844 * @param string $text The text to transform
4845 * @param Title $title The Title object for the current article
4846 * @param User $user The User object describing the current user
4847 * @param ParserOptions $options Parsing options
4848 * @param bool $clearState Whether to clear the parser state first
4849 * @return string The altered wiki markup
4850 */
4851 public function preSaveTransform( $text, Title $title, User $user,
4852 ParserOptions $options, $clearState = true
4853 ) {
4854 if ( $clearState ) {
4855 $magicScopeVariable = $this->lock();
4856 }
4857 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4858 $this->setUser( $user );
4859
4860 // Strip U+0000 NULL (T159174)
4861 $text = str_replace( "\000", '', $text );
4862
4863 // We still normalize line endings for backwards-compatibility
4864 // with other code that just calls PST, but this should already
4865 // be handled in TextContent subclasses
4866 $text = TextContent::normalizeLineEndings( $text );
4867
4868 if ( $options->getPreSaveTransform() ) {
4869 $text = $this->pstPass2( $text, $user );
4870 }
4871 $text = $this->mStripState->unstripBoth( $text );
4872
4873 $this->setUser( null ); # Reset
4874
4875 return $text;
4876 }
4877
4878 /**
4879 * Pre-save transform helper function
4880 *
4881 * @param string $text
4882 * @param User $user
4883 *
4884 * @return string
4885 */
4886 private function pstPass2( $text, $user ) {
4887 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4888 # $this->contLang here in order to give everyone the same signature and use the default one
4889 # rather than the one selected in each user's preferences. (see also T14815)
4890 $ts = $this->mOptions->getTimestamp();
4891 $timestamp = MWTimestamp::getLocalInstance( $ts );
4892 $ts = $timestamp->format( 'YmdHis' );
4893 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4894
4895 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4896
4897 # Variable replacement
4898 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4899 $text = $this->replaceVariables( $text );
4900
4901 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4902 # which may corrupt this parser instance via its wfMessage()->text() call-
4903
4904 # Signatures
4905 if ( strpos( $text, '~~~' ) !== false ) {
4906 $sigText = $this->getUserSig( $user );
4907 $text = strtr( $text, [
4908 '~~~~~' => $d,
4909 '~~~~' => "$sigText $d",
4910 '~~~' => $sigText
4911 ] );
4912 # The main two signature forms used above are time-sensitive
4913 $this->setOutputFlag( 'user-signature', 'User signature detected' );
4914 }
4915
4916 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4917 $tc = '[' . Title::legalChars() . ']';
4918 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4919
4920 // [[ns:page (context)|]]
4921 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4922 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4923 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4924 // [[ns:page (context), context|]] (using either single or double-width comma)
4925 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4926 // [[|page]] (reverse pipe trick: add context from page title)
4927 $p2 = "/\[\[\\|($tc+)]]/";
4928
4929 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4930 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4931 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4932 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4933
4934 $t = $this->mTitle->getText();
4935 $m = [];
4936 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4937 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4938 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4939 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4940 } else {
4941 # if there's no context, don't bother duplicating the title
4942 $text = preg_replace( $p2, '[[\\1]]', $text );
4943 }
4944
4945 return $text;
4946 }
4947
4948 /**
4949 * Fetch the user's signature text, if any, and normalize to
4950 * validated, ready-to-insert wikitext.
4951 * If you have pre-fetched the nickname or the fancySig option, you can
4952 * specify them here to save a database query.
4953 * Do not reuse this parser instance after calling getUserSig(),
4954 * as it may have changed.
4955 *
4956 * @param User &$user
4957 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4958 * @param bool|null $fancySig whether the nicknname is the complete signature
4959 * or null to use default value
4960 * @return string
4961 */
4962 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4963 $username = $user->getName();
4964
4965 # If not given, retrieve from the user object.
4966 if ( $nickname === false ) {
4967 $nickname = $user->getOption( 'nickname' );
4968 }
4969
4970 if ( is_null( $fancySig ) ) {
4971 $fancySig = $user->getBoolOption( 'fancysig' );
4972 }
4973
4974 $nickname = $nickname == null ? $username : $nickname;
4975
4976 if ( mb_strlen( $nickname ) > $this->svcOptions->get( 'MaxSigChars' ) ) {
4977 $nickname = $username;
4978 $this->logger->debug( __METHOD__ . ": $username has overlong signature." );
4979 } elseif ( $fancySig !== false ) {
4980 # Sig. might contain markup; validate this
4981 if ( $this->validateSig( $nickname ) !== false ) {
4982 # Validated; clean up (if needed) and return it
4983 return $this->cleanSig( $nickname, true );
4984 } else {
4985 # Failed to validate; fall back to the default
4986 $nickname = $username;
4987 $this->logger->debug( __METHOD__ . ": $username has bad XML tags in signature." );
4988 }
4989 }
4990
4991 # Make sure nickname doesnt get a sig in a sig
4992 $nickname = self::cleanSigInSig( $nickname );
4993
4994 # If we're still here, make it a link to the user page
4995 $userText = wfEscapeWikiText( $username );
4996 $nickText = wfEscapeWikiText( $nickname );
4997 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4998
4999 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
5000 ->title( $this->getTitle() )->text();
5001 }
5002
5003 /**
5004 * Check that the user's signature contains no bad XML
5005 *
5006 * @param string $text
5007 * @return string|bool An expanded string, or false if invalid.
5008 */
5009 public function validateSig( $text ) {
5010 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
5011 }
5012
5013 /**
5014 * Clean up signature text
5015 *
5016 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
5017 * 2) Substitute all transclusions
5018 *
5019 * @param string $text
5020 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
5021 * @return string Signature text
5022 */
5023 public function cleanSig( $text, $parsing = false ) {
5024 if ( !$parsing ) {
5025 global $wgTitle;
5026 $magicScopeVariable = $this->lock();
5027 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
5028 }
5029
5030 # Option to disable this feature
5031 if ( !$this->mOptions->getCleanSignatures() ) {
5032 return $text;
5033 }
5034
5035 # @todo FIXME: Regex doesn't respect extension tags or nowiki
5036 # => Move this logic to braceSubstitution()
5037 $substWord = $this->magicWordFactory->get( 'subst' );
5038 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
5039 $substText = '{{' . $substWord->getSynonym( 0 );
5040
5041 $text = preg_replace( $substRegex, $substText, $text );
5042 $text = self::cleanSigInSig( $text );
5043 $dom = $this->preprocessToDom( $text );
5044 $frame = $this->getPreprocessor()->newFrame();
5045 $text = $frame->expand( $dom );
5046
5047 if ( !$parsing ) {
5048 $text = $this->mStripState->unstripBoth( $text );
5049 }
5050
5051 return $text;
5052 }
5053
5054 /**
5055 * Strip 3, 4 or 5 tildes out of signatures.
5056 *
5057 * @param string $text
5058 * @return string Signature text with /~{3,5}/ removed
5059 */
5060 public static function cleanSigInSig( $text ) {
5061 $text = preg_replace( '/~{3,5}/', '', $text );
5062 return $text;
5063 }
5064
5065 /**
5066 * Set up some variables which are usually set up in parse()
5067 * so that an external function can call some class members with confidence
5068 *
5069 * @param Title|null $title
5070 * @param ParserOptions $options
5071 * @param int $outputType
5072 * @param bool $clearState
5073 * @param int|null $revId
5074 */
5075 public function startExternalParse( Title $title = null, ParserOptions $options,
5076 $outputType, $clearState = true, $revId = null
5077 ) {
5078 $this->startParse( $title, $options, $outputType, $clearState );
5079 if ( $revId !== null ) {
5080 $this->mRevisionId = $revId;
5081 }
5082 }
5083
5084 /**
5085 * @param Title|null $title
5086 * @param ParserOptions $options
5087 * @param int $outputType
5088 * @param bool $clearState
5089 */
5090 private function startParse( Title $title = null, ParserOptions $options,
5091 $outputType, $clearState = true
5092 ) {
5093 $this->setTitle( $title );
5094 $this->mOptions = $options;
5095 $this->setOutputType( $outputType );
5096 if ( $clearState ) {
5097 $this->clearState();
5098 }
5099 }
5100
5101 /**
5102 * Wrapper for preprocess()
5103 *
5104 * @param string $text The text to preprocess
5105 * @param ParserOptions $options
5106 * @param Title|null $title Title object or null to use $wgTitle
5107 * @return string
5108 */
5109 public function transformMsg( $text, $options, $title = null ) {
5110 static $executing = false;
5111
5112 # Guard against infinite recursion
5113 if ( $executing ) {
5114 return $text;
5115 }
5116 $executing = true;
5117
5118 if ( !$title ) {
5119 global $wgTitle;
5120 $title = $wgTitle;
5121 }
5122
5123 $text = $this->preprocess( $text, $title, $options );
5124
5125 $executing = false;
5126 return $text;
5127 }
5128
5129 /**
5130 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
5131 * The callback should have the following form:
5132 * function myParserHook( $text, $params, $parser, $frame ) { ... }
5133 *
5134 * Transform and return $text. Use $parser for any required context, e.g. use
5135 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
5136 *
5137 * Hooks may return extended information by returning an array, of which the
5138 * first numbered element (index 0) must be the return string, and all other
5139 * entries are extracted into local variables within an internal function
5140 * in the Parser class.
5141 *
5142 * This interface (introduced r61913) appears to be undocumented, but
5143 * 'markerType' is used by some core tag hooks to override which strip
5144 * array their results are placed in. **Use great caution if attempting
5145 * this interface, as it is not documented and injudicious use could smash
5146 * private variables.**
5147 *
5148 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5149 * @param callable $callback The callback function (and object) to use for the tag
5150 * @throws MWException
5151 * @return callable|null The old value of the mTagHooks array associated with the hook
5152 */
5153 public function setHook( $tag, callable $callback ) {
5154 $tag = strtolower( $tag );
5155 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5156 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
5157 }
5158 $oldVal = $this->mTagHooks[$tag] ?? null;
5159 $this->mTagHooks[$tag] = $callback;
5160 if ( !in_array( $tag, $this->mStripList ) ) {
5161 $this->mStripList[] = $tag;
5162 }
5163
5164 return $oldVal;
5165 }
5166
5167 /**
5168 * As setHook(), but letting the contents be parsed.
5169 *
5170 * Transparent tag hooks are like regular XML-style tag hooks, except they
5171 * operate late in the transformation sequence, on HTML instead of wikitext.
5172 *
5173 * This is probably obsoleted by things dealing with parser frames?
5174 * The only extension currently using it is geoserver.
5175 *
5176 * @since 1.10
5177 * @todo better document or deprecate this
5178 *
5179 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5180 * @param callable $callback The callback function (and object) to use for the tag
5181 * @throws MWException
5182 * @return callable|null The old value of the mTagHooks array associated with the hook
5183 */
5184 public function setTransparentTagHook( $tag, callable $callback ) {
5185 $tag = strtolower( $tag );
5186 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5187 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5188 }
5189 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
5190 $this->mTransparentTagHooks[$tag] = $callback;
5191
5192 return $oldVal;
5193 }
5194
5195 /**
5196 * Remove all tag hooks
5197 */
5198 public function clearTagHooks() {
5199 $this->mTagHooks = [];
5200 $this->mFunctionTagHooks = [];
5201 $this->mStripList = $this->mDefaultStripList;
5202 }
5203
5204 /**
5205 * Create a function, e.g. {{sum:1|2|3}}
5206 * The callback function should have the form:
5207 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5208 *
5209 * Or with Parser::SFH_OBJECT_ARGS:
5210 * function myParserFunction( $parser, $frame, $args ) { ... }
5211 *
5212 * The callback may either return the text result of the function, or an array with the text
5213 * in element 0, and a number of flags in the other elements. The names of the flags are
5214 * specified in the keys. Valid flags are:
5215 * found The text returned is valid, stop processing the template. This
5216 * is on by default.
5217 * nowiki Wiki markup in the return value should be escaped
5218 * isHTML The returned text is HTML, armour it against wikitext transformation
5219 *
5220 * @param string $id The magic word ID
5221 * @param callable $callback The callback function (and object) to use
5222 * @param int $flags A combination of the following flags:
5223 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5224 *
5225 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5226 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5227 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5228 * the arguments, and to control the way they are expanded.
5229 *
5230 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5231 * arguments, for instance:
5232 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5233 *
5234 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5235 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5236 * working if/when this is changed.
5237 *
5238 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5239 * expansion.
5240 *
5241 * Please read the documentation in includes/parser/Preprocessor.php for more information
5242 * about the methods available in PPFrame and PPNode.
5243 *
5244 * @throws MWException
5245 * @return string|callable The old callback function for this name, if any
5246 */
5247 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
5248 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
5249 $this->mFunctionHooks[$id] = [ $callback, $flags ];
5250
5251 # Add to function cache
5252 $mw = $this->magicWordFactory->get( $id );
5253 if ( !$mw ) {
5254 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
5255 }
5256
5257 $synonyms = $mw->getSynonyms();
5258 $sensitive = intval( $mw->isCaseSensitive() );
5259
5260 foreach ( $synonyms as $syn ) {
5261 # Case
5262 if ( !$sensitive ) {
5263 $syn = $this->contLang->lc( $syn );
5264 }
5265 # Add leading hash
5266 if ( !( $flags & self::SFH_NO_HASH ) ) {
5267 $syn = '#' . $syn;
5268 }
5269 # Remove trailing colon
5270 if ( substr( $syn, -1, 1 ) === ':' ) {
5271 $syn = substr( $syn, 0, -1 );
5272 }
5273 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5274 }
5275 return $oldVal;
5276 }
5277
5278 /**
5279 * Get all registered function hook identifiers
5280 *
5281 * @return array
5282 */
5283 public function getFunctionHooks() {
5284 $this->firstCallInit();
5285 return array_keys( $this->mFunctionHooks );
5286 }
5287
5288 /**
5289 * Create a tag function, e.g. "<test>some stuff</test>".
5290 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5291 * Unlike parser functions, their content is not preprocessed.
5292 * @param string $tag
5293 * @param callable $callback
5294 * @param int $flags
5295 * @throws MWException
5296 * @return null
5297 */
5298 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
5299 $tag = strtolower( $tag );
5300 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5301 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5302 }
5303 $old = $this->mFunctionTagHooks[$tag] ?? null;
5304 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
5305
5306 if ( !in_array( $tag, $this->mStripList ) ) {
5307 $this->mStripList[] = $tag;
5308 }
5309
5310 return $old;
5311 }
5312
5313 /**
5314 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5315 * Placeholders created in Linker::link()
5316 *
5317 * @param string &$text
5318 * @param int $options
5319 * @deprecated since 1.34; should not be used outside parser class.
5320 */
5321 public function replaceLinkHolders( &$text, $options = 0 ) {
5322 $this->replaceLinkHoldersPrivate( $text, $options );
5323 }
5324
5325 /**
5326 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5327 * Placeholders created in Linker::link()
5328 *
5329 * @param string &$text
5330 * @param int $options
5331 */
5332 private function replaceLinkHoldersPrivate( &$text, $options = 0 ) {
5333 $this->mLinkHolders->replace( $text );
5334 }
5335
5336 /**
5337 * Replace "<!--LINK-->" link placeholders with plain text of links
5338 * (not HTML-formatted).
5339 *
5340 * @param string $text
5341 * @return string
5342 * @deprecated since 1.34; should not be used outside parser class.
5343 */
5344 public function replaceLinkHoldersText( $text ) {
5345 wfDeprecated( __METHOD__, '1.34' );
5346 return $this->replaceLinkHoldersTextPrivate( $text );
5347 }
5348
5349 /**
5350 * Replace "<!--LINK-->" link placeholders with plain text of links
5351 * (not HTML-formatted).
5352 *
5353 * @param string $text
5354 * @return string
5355 */
5356 private function replaceLinkHoldersTextPrivate( $text ) {
5357 return $this->mLinkHolders->replaceText( $text );
5358 }
5359
5360 /**
5361 * Renders an image gallery from a text with one line per image.
5362 * text labels may be given by using |-style alternative text. E.g.
5363 * Image:one.jpg|The number "1"
5364 * Image:tree.jpg|A tree
5365 * given as text will return the HTML of a gallery with two images,
5366 * labeled 'The number "1"' and
5367 * 'A tree'.
5368 *
5369 * @param string $text
5370 * @param array $params
5371 * @return string HTML
5372 */
5373 public function renderImageGallery( $text, $params ) {
5374 $mode = false;
5375 if ( isset( $params['mode'] ) ) {
5376 $mode = $params['mode'];
5377 }
5378
5379 try {
5380 $ig = ImageGalleryBase::factory( $mode );
5381 } catch ( Exception $e ) {
5382 // If invalid type set, fallback to default.
5383 $ig = ImageGalleryBase::factory( false );
5384 }
5385
5386 $ig->setContextTitle( $this->mTitle );
5387 $ig->setShowBytes( false );
5388 $ig->setShowDimensions( false );
5389 $ig->setShowFilename( false );
5390 $ig->setParser( $this );
5391 $ig->setHideBadImages();
5392 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5393
5394 if ( isset( $params['showfilename'] ) ) {
5395 $ig->setShowFilename( true );
5396 } else {
5397 $ig->setShowFilename( false );
5398 }
5399 if ( isset( $params['caption'] ) ) {
5400 // NOTE: We aren't passing a frame here or below. Frame info
5401 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5402 // See T107332#4030581
5403 $caption = $this->recursiveTagParse( $params['caption'] );
5404 $ig->setCaptionHtml( $caption );
5405 }
5406 if ( isset( $params['perrow'] ) ) {
5407 $ig->setPerRow( $params['perrow'] );
5408 }
5409 if ( isset( $params['widths'] ) ) {
5410 $ig->setWidths( $params['widths'] );
5411 }
5412 if ( isset( $params['heights'] ) ) {
5413 $ig->setHeights( $params['heights'] );
5414 }
5415 $ig->setAdditionalOptions( $params );
5416
5417 // Avoid PHP 7.1 warning from passing $this by reference
5418 $parser = $this;
5419 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5420
5421 $lines = StringUtils::explode( "\n", $text );
5422 foreach ( $lines as $line ) {
5423 # match lines like these:
5424 # Image:someimage.jpg|This is some image
5425 $matches = [];
5426 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5427 # Skip empty lines
5428 if ( count( $matches ) == 0 ) {
5429 continue;
5430 }
5431
5432 if ( strpos( $matches[0], '%' ) !== false ) {
5433 $matches[1] = rawurldecode( $matches[1] );
5434 }
5435 $title = Title::newFromText( $matches[1], NS_FILE );
5436 if ( is_null( $title ) ) {
5437 # Bogus title. Ignore these so we don't bomb out later.
5438 continue;
5439 }
5440
5441 # We need to get what handler the file uses, to figure out parameters.
5442 # Note, a hook can overide the file name, and chose an entirely different
5443 # file (which potentially could be of a different type and have different handler).
5444 $options = [];
5445 $descQuery = false;
5446 Hooks::run( 'BeforeParserFetchFileAndTitle',
5447 [ $this, $title, &$options, &$descQuery ] );
5448 # Don't register it now, as TraditionalImageGallery does that later.
5449 $file = $this->fetchFileNoRegister( $title, $options );
5450 $handler = $file ? $file->getHandler() : false;
5451
5452 $paramMap = [
5453 'img_alt' => 'gallery-internal-alt',
5454 'img_link' => 'gallery-internal-link',
5455 ];
5456 if ( $handler ) {
5457 $paramMap += $handler->getParamMap();
5458 // We don't want people to specify per-image widths.
5459 // Additionally the width parameter would need special casing anyhow.
5460 unset( $paramMap['img_width'] );
5461 }
5462
5463 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5464
5465 $label = '';
5466 $alt = '';
5467 $link = '';
5468 $handlerOptions = [];
5469 if ( isset( $matches[3] ) ) {
5470 // look for an |alt= definition while trying not to break existing
5471 // captions with multiple pipes (|) in it, until a more sensible grammar
5472 // is defined for images in galleries
5473
5474 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5475 // splitting on '|' is a bit odd, and different from makeImage.
5476 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5477 // Protect LanguageConverter markup
5478 $parameterMatches = StringUtils::delimiterExplode(
5479 '-{', '}-', '|', $matches[3], true /* nested */
5480 );
5481
5482 foreach ( $parameterMatches as $parameterMatch ) {
5483 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5484 if ( $magicName ) {
5485 $paramName = $paramMap[$magicName];
5486
5487 switch ( $paramName ) {
5488 case 'gallery-internal-alt':
5489 $alt = $this->stripAltText( $match, false );
5490 break;
5491 case 'gallery-internal-link':
5492 $linkValue = $this->stripAltText( $match, false );
5493 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5494 // Result of LanguageConverter::markNoConversion
5495 // invoked on an external link.
5496 $linkValue = substr( $linkValue, 4, -2 );
5497 }
5498 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5499 if ( $type === 'link-url' ) {
5500 $link = $target;
5501 $this->mOutput->addExternalLink( $target );
5502 } elseif ( $type === 'link-title' ) {
5503 $link = $target->getLinkURL();
5504 $this->mOutput->addLink( $target );
5505 }
5506 break;
5507 default:
5508 // Must be a handler specific parameter.
5509 if ( $handler->validateParam( $paramName, $match ) ) {
5510 $handlerOptions[$paramName] = $match;
5511 } else {
5512 // Guess not, consider it as caption.
5513 $this->logger->debug(
5514 "$parameterMatch failed parameter validation" );
5515 $label = $parameterMatch;
5516 }
5517 }
5518
5519 } else {
5520 // Last pipe wins.
5521 $label = $parameterMatch;
5522 }
5523 }
5524 }
5525
5526 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5527 }
5528 $html = $ig->toHTML();
5529 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5530 return $html;
5531 }
5532
5533 /**
5534 * @param MediaHandler $handler
5535 * @return array
5536 */
5537 public function getImageParams( $handler ) {
5538 if ( $handler ) {
5539 $handlerClass = get_class( $handler );
5540 } else {
5541 $handlerClass = '';
5542 }
5543 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5544 # Initialise static lists
5545 static $internalParamNames = [
5546 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5547 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5548 'bottom', 'text-bottom' ],
5549 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5550 'upright', 'border', 'link', 'alt', 'class' ],
5551 ];
5552 static $internalParamMap;
5553 if ( !$internalParamMap ) {
5554 $internalParamMap = [];
5555 foreach ( $internalParamNames as $type => $names ) {
5556 foreach ( $names as $name ) {
5557 // For grep: img_left, img_right, img_center, img_none,
5558 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5559 // img_bottom, img_text_bottom,
5560 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5561 // img_border, img_link, img_alt, img_class
5562 $magicName = str_replace( '-', '_', "img_$name" );
5563 $internalParamMap[$magicName] = [ $type, $name ];
5564 }
5565 }
5566 }
5567
5568 # Add handler params
5569 $paramMap = $internalParamMap;
5570 if ( $handler ) {
5571 $handlerParamMap = $handler->getParamMap();
5572 foreach ( $handlerParamMap as $magic => $paramName ) {
5573 $paramMap[$magic] = [ 'handler', $paramName ];
5574 }
5575 }
5576 $this->mImageParams[$handlerClass] = $paramMap;
5577 $this->mImageParamsMagicArray[$handlerClass] =
5578 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5579 }
5580 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5581 }
5582
5583 /**
5584 * Parse image options text and use it to make an image
5585 *
5586 * @param Title $title
5587 * @param string $options
5588 * @param LinkHolderArray|bool $holders
5589 * @return string HTML
5590 */
5591 public function makeImage( $title, $options, $holders = false ) {
5592 # Check if the options text is of the form "options|alt text"
5593 # Options are:
5594 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5595 # * left no resizing, just left align. label is used for alt= only
5596 # * right same, but right aligned
5597 # * none same, but not aligned
5598 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5599 # * center center the image
5600 # * frame Keep original image size, no magnify-button.
5601 # * framed Same as "frame"
5602 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5603 # * upright reduce width for upright images, rounded to full __0 px
5604 # * border draw a 1px border around the image
5605 # * alt Text for HTML alt attribute (defaults to empty)
5606 # * class Set a class for img node
5607 # * link Set the target of the image link. Can be external, interwiki, or local
5608 # vertical-align values (no % or length right now):
5609 # * baseline
5610 # * sub
5611 # * super
5612 # * top
5613 # * text-top
5614 # * middle
5615 # * bottom
5616 # * text-bottom
5617
5618 # Protect LanguageConverter markup when splitting into parts
5619 $parts = StringUtils::delimiterExplode(
5620 '-{', '}-', '|', $options, true /* allow nesting */
5621 );
5622
5623 # Give extensions a chance to select the file revision for us
5624 $options = [];
5625 $descQuery = false;
5626 Hooks::run( 'BeforeParserFetchFileAndTitle',
5627 [ $this, $title, &$options, &$descQuery ] );
5628 # Fetch and register the file (file title may be different via hooks)
5629 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5630
5631 # Get parameter map
5632 $handler = $file ? $file->getHandler() : false;
5633
5634 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5635
5636 if ( !$file ) {
5637 $this->addTrackingCategory( 'broken-file-category' );
5638 }
5639
5640 # Process the input parameters
5641 $caption = '';
5642 $params = [ 'frame' => [], 'handler' => [],
5643 'horizAlign' => [], 'vertAlign' => [] ];
5644 $seenformat = false;
5645 foreach ( $parts as $part ) {
5646 $part = trim( $part );
5647 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5648 $validated = false;
5649 if ( isset( $paramMap[$magicName] ) ) {
5650 list( $type, $paramName ) = $paramMap[$magicName];
5651
5652 # Special case; width and height come in one variable together
5653 if ( $type === 'handler' && $paramName === 'width' ) {
5654 $parsedWidthParam = self::parseWidthParam( $value );
5655 if ( isset( $parsedWidthParam['width'] ) ) {
5656 $width = $parsedWidthParam['width'];
5657 if ( $handler->validateParam( 'width', $width ) ) {
5658 $params[$type]['width'] = $width;
5659 $validated = true;
5660 }
5661 }
5662 if ( isset( $parsedWidthParam['height'] ) ) {
5663 $height = $parsedWidthParam['height'];
5664 if ( $handler->validateParam( 'height', $height ) ) {
5665 $params[$type]['height'] = $height;
5666 $validated = true;
5667 }
5668 }
5669 # else no validation -- T15436
5670 } else {
5671 if ( $type === 'handler' ) {
5672 # Validate handler parameter
5673 $validated = $handler->validateParam( $paramName, $value );
5674 } else {
5675 # Validate internal parameters
5676 switch ( $paramName ) {
5677 case 'manualthumb':
5678 case 'alt':
5679 case 'class':
5680 # @todo FIXME: Possibly check validity here for
5681 # manualthumb? downstream behavior seems odd with
5682 # missing manual thumbs.
5683 $validated = true;
5684 $value = $this->stripAltText( $value, $holders );
5685 break;
5686 case 'link':
5687 list( $paramName, $value ) =
5688 $this->parseLinkParameter(
5689 $this->stripAltText( $value, $holders )
5690 );
5691 if ( $paramName ) {
5692 $validated = true;
5693 if ( $paramName === 'no-link' ) {
5694 $value = true;
5695 }
5696 if ( ( $paramName === 'link-url' ) && $this->mOptions->getExternalLinkTarget() ) {
5697 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5698 }
5699 }
5700 break;
5701 case 'frameless':
5702 case 'framed':
5703 case 'thumbnail':
5704 // use first appearing option, discard others.
5705 $validated = !$seenformat;
5706 $seenformat = true;
5707 break;
5708 default:
5709 # Most other things appear to be empty or numeric...
5710 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5711 }
5712 }
5713
5714 if ( $validated ) {
5715 $params[$type][$paramName] = $value;
5716 }
5717 }
5718 }
5719 if ( !$validated ) {
5720 $caption = $part;
5721 }
5722 }
5723
5724 # Process alignment parameters
5725 if ( $params['horizAlign'] ) {
5726 $params['frame']['align'] = key( $params['horizAlign'] );
5727 }
5728 if ( $params['vertAlign'] ) {
5729 $params['frame']['valign'] = key( $params['vertAlign'] );
5730 }
5731
5732 $params['frame']['caption'] = $caption;
5733
5734 # Will the image be presented in a frame, with the caption below?
5735 $imageIsFramed = isset( $params['frame']['frame'] )
5736 || isset( $params['frame']['framed'] )
5737 || isset( $params['frame']['thumbnail'] )
5738 || isset( $params['frame']['manualthumb'] );
5739
5740 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5741 # came to also set the caption, ordinary text after the image -- which
5742 # makes no sense, because that just repeats the text multiple times in
5743 # screen readers. It *also* came to set the title attribute.
5744 # Now that we have an alt attribute, we should not set the alt text to
5745 # equal the caption: that's worse than useless, it just repeats the
5746 # text. This is the framed/thumbnail case. If there's no caption, we
5747 # use the unnamed parameter for alt text as well, just for the time be-
5748 # ing, if the unnamed param is set and the alt param is not.
5749 # For the future, we need to figure out if we want to tweak this more,
5750 # e.g., introducing a title= parameter for the title; ignoring the un-
5751 # named parameter entirely for images without a caption; adding an ex-
5752 # plicit caption= parameter and preserving the old magic unnamed para-
5753 # meter for BC; ...
5754 if ( $imageIsFramed ) { # Framed image
5755 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5756 # No caption or alt text, add the filename as the alt text so
5757 # that screen readers at least get some description of the image
5758 $params['frame']['alt'] = $title->getText();
5759 }
5760 # Do not set $params['frame']['title'] because tooltips don't make sense
5761 # for framed images
5762 } else { # Inline image
5763 if ( !isset( $params['frame']['alt'] ) ) {
5764 # No alt text, use the "caption" for the alt text
5765 if ( $caption !== '' ) {
5766 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5767 } else {
5768 # No caption, fall back to using the filename for the
5769 # alt text
5770 $params['frame']['alt'] = $title->getText();
5771 }
5772 }
5773 # Use the "caption" for the tooltip text
5774 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5775 }
5776 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5777
5778 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5779
5780 # Linker does the rest
5781 $time = $options['time'] ?? false;
5782 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5783 $time, $descQuery, $this->mOptions->getThumbSize() );
5784
5785 # Give the handler a chance to modify the parser object
5786 if ( $handler ) {
5787 $handler->parserTransformHook( $this, $file );
5788 }
5789
5790 return $ret;
5791 }
5792
5793 /**
5794 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5795 *
5796 * Adds an entry to appropriate link tables.
5797 *
5798 * @since 1.32
5799 * @param string $value
5800 * @return array of `[ type, target ]`, where:
5801 * - `type` is one of:
5802 * - `null`: Given value is not a valid link target, use default
5803 * - `'no-link'`: Given value is empty, do not generate a link
5804 * - `'link-url'`: Given value is a valid external link
5805 * - `'link-title'`: Given value is a valid internal link
5806 * - `target` is:
5807 * - When `type` is `null` or `'no-link'`: `false`
5808 * - When `type` is `'link-url'`: URL string corresponding to given value
5809 * - When `type` is `'link-title'`: Title object corresponding to given value
5810 */
5811 public function parseLinkParameter( $value ) {
5812 $chars = self::EXT_LINK_URL_CLASS;
5813 $addr = self::EXT_LINK_ADDR;
5814 $prots = $this->mUrlProtocols;
5815 $type = null;
5816 $target = false;
5817 if ( $value === '' ) {
5818 $type = 'no-link';
5819 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5820 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5821 $this->mOutput->addExternalLink( $value );
5822 $type = 'link-url';
5823 $target = $value;
5824 }
5825 } else {
5826 $linkTitle = Title::newFromText( $value );
5827 if ( $linkTitle ) {
5828 $this->mOutput->addLink( $linkTitle );
5829 $type = 'link-title';
5830 $target = $linkTitle;
5831 }
5832 }
5833 return [ $type, $target ];
5834 }
5835
5836 /**
5837 * @param string $caption
5838 * @param LinkHolderArray|bool $holders
5839 * @return mixed|string
5840 */
5841 protected function stripAltText( $caption, $holders ) {
5842 # Strip bad stuff out of the title (tooltip). We can't just use
5843 # replaceLinkHoldersText() here, because if this function is called
5844 # from handleInternalLinks2(), mLinkHolders won't be up-to-date.
5845 if ( $holders ) {
5846 $tooltip = $holders->replaceText( $caption );
5847 } else {
5848 $tooltip = $this->replaceLinkHoldersTextPrivate( $caption );
5849 }
5850
5851 # make sure there are no placeholders in thumbnail attributes
5852 # that are later expanded to html- so expand them now and
5853 # remove the tags
5854 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5855 # Compatibility hack! In HTML certain entity references not terminated
5856 # by a semicolon are decoded (but not if we're in an attribute; that's
5857 # how link URLs get away without properly escaping & in queries).
5858 # But wikitext has always required semicolon-termination of entities,
5859 # so encode & where needed to avoid decode of semicolon-less entities.
5860 # See T209236 and
5861 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5862 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5863 $tooltip = preg_replace( "/
5864 & # 1. entity prefix
5865 (?= # 2. followed by:
5866 (?: # a. one of the legacy semicolon-less named entities
5867 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5868 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5869 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5870 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5871 U(?:acute|circ|grave|uml)|Yacute|
5872 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5873 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5874 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5875 frac(?:1(?:2|4)|34)|
5876 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5877 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5878 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5879 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5880 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5881 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5882 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5883 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5884 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5885 )
5886 (?:[^;]|$)) # b. and not followed by a semicolon
5887 # S = study, for efficiency
5888 /Sx", '&amp;', $tooltip );
5889 $tooltip = Sanitizer::stripAllTags( $tooltip );
5890
5891 return $tooltip;
5892 }
5893
5894 /**
5895 * Set a flag in the output object indicating that the content is dynamic and
5896 * shouldn't be cached.
5897 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5898 */
5899 public function disableCache() {
5900 $this->logger->debug( "Parser output marked as uncacheable." );
5901 if ( !$this->mOutput ) {
5902 throw new MWException( __METHOD__ .
5903 " can only be called when actually parsing something" );
5904 }
5905 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5906 }
5907
5908 /**
5909 * Callback from the Sanitizer for expanding items found in HTML attribute
5910 * values, so they can be safely tested and escaped.
5911 *
5912 * @param string &$text
5913 * @param bool|PPFrame $frame
5914 * @return string
5915 */
5916 public function attributeStripCallback( &$text, $frame = false ) {
5917 $text = $this->replaceVariables( $text, $frame );
5918 $text = $this->mStripState->unstripBoth( $text );
5919 return $text;
5920 }
5921
5922 /**
5923 * Accessor
5924 *
5925 * @return array
5926 */
5927 public function getTags() {
5928 $this->firstCallInit();
5929 return array_merge(
5930 array_keys( $this->mTransparentTagHooks ),
5931 array_keys( $this->mTagHooks ),
5932 array_keys( $this->mFunctionTagHooks )
5933 );
5934 }
5935
5936 /**
5937 * @since 1.32
5938 * @return array
5939 */
5940 public function getFunctionSynonyms() {
5941 $this->firstCallInit();
5942 return $this->mFunctionSynonyms;
5943 }
5944
5945 /**
5946 * @since 1.32
5947 * @return string
5948 */
5949 public function getUrlProtocols() {
5950 return $this->mUrlProtocols;
5951 }
5952
5953 /**
5954 * Replace transparent tags in $text with the values given by the callbacks.
5955 *
5956 * Transparent tag hooks are like regular XML-style tag hooks, except they
5957 * operate late in the transformation sequence, on HTML instead of wikitext.
5958 *
5959 * @param string $text
5960 *
5961 * @return string
5962 */
5963 public function replaceTransparentTags( $text ) {
5964 $matches = [];
5965 $elements = array_keys( $this->mTransparentTagHooks );
5966 $text = self::extractTagsAndParams( $elements, $text, $matches );
5967 $replacements = [];
5968
5969 foreach ( $matches as $marker => $data ) {
5970 list( $element, $content, $params, $tag ) = $data;
5971 $tagName = strtolower( $element );
5972 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5973 $output = call_user_func_array(
5974 $this->mTransparentTagHooks[$tagName],
5975 [ $content, $params, $this ]
5976 );
5977 } else {
5978 $output = $tag;
5979 }
5980 $replacements[$marker] = $output;
5981 }
5982 return strtr( $text, $replacements );
5983 }
5984
5985 /**
5986 * Break wikitext input into sections, and either pull or replace
5987 * some particular section's text.
5988 *
5989 * External callers should use the getSection and replaceSection methods.
5990 *
5991 * @param string $text Page wikitext
5992 * @param string|int $sectionId A section identifier string of the form:
5993 * "<flag1> - <flag2> - ... - <section number>"
5994 *
5995 * Currently the only recognised flag is "T", which means the target section number
5996 * was derived during a template inclusion parse, in other words this is a template
5997 * section edit link. If no flags are given, it was an ordinary section edit link.
5998 * This flag is required to avoid a section numbering mismatch when a section is
5999 * enclosed by "<includeonly>" (T8563).
6000 *
6001 * The section number 0 pulls the text before the first heading; other numbers will
6002 * pull the given section along with its lower-level subsections. If the section is
6003 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
6004 *
6005 * Section 0 is always considered to exist, even if it only contains the empty
6006 * string. If $text is the empty string and section 0 is replaced, $newText is
6007 * returned.
6008 *
6009 * @param string $mode One of "get" or "replace"
6010 * @param string $newText Replacement text for section data.
6011 * @return string For "get", the extracted section text.
6012 * for "replace", the whole page with the section replaced.
6013 */
6014 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
6015 global $wgTitle; # not generally used but removes an ugly failure mode
6016
6017 $magicScopeVariable = $this->lock();
6018 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
6019 $outText = '';
6020 $frame = $this->getPreprocessor()->newFrame();
6021
6022 # Process section extraction flags
6023 $flags = 0;
6024 $sectionParts = explode( '-', $sectionId );
6025 $sectionIndex = array_pop( $sectionParts );
6026 foreach ( $sectionParts as $part ) {
6027 if ( $part === 'T' ) {
6028 $flags |= self::PTD_FOR_INCLUSION;
6029 }
6030 }
6031
6032 # Check for empty input
6033 if ( strval( $text ) === '' ) {
6034 # Only sections 0 and T-0 exist in an empty document
6035 if ( $sectionIndex == 0 ) {
6036 if ( $mode === 'get' ) {
6037 return '';
6038 }
6039
6040 return $newText;
6041 } else {
6042 if ( $mode === 'get' ) {
6043 return $newText;
6044 }
6045
6046 return $text;
6047 }
6048 }
6049
6050 # Preprocess the text
6051 $root = $this->preprocessToDom( $text, $flags );
6052
6053 # <h> nodes indicate section breaks
6054 # They can only occur at the top level, so we can find them by iterating the root's children
6055 $node = $root->getFirstChild();
6056
6057 # Find the target section
6058 if ( $sectionIndex == 0 ) {
6059 # Section zero doesn't nest, level=big
6060 $targetLevel = 1000;
6061 } else {
6062 while ( $node ) {
6063 if ( $node->getName() === 'h' ) {
6064 $bits = $node->splitHeading();
6065 if ( $bits['i'] == $sectionIndex ) {
6066 $targetLevel = $bits['level'];
6067 break;
6068 }
6069 }
6070 if ( $mode === 'replace' ) {
6071 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
6072 }
6073 $node = $node->getNextSibling();
6074 }
6075 }
6076
6077 if ( !$node ) {
6078 # Not found
6079 if ( $mode === 'get' ) {
6080 return $newText;
6081 } else {
6082 return $text;
6083 }
6084 }
6085
6086 # Find the end of the section, including nested sections
6087 do {
6088 if ( $node->getName() === 'h' ) {
6089 $bits = $node->splitHeading();
6090 $curLevel = $bits['level'];
6091 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
6092 break;
6093 }
6094 }
6095 if ( $mode === 'get' ) {
6096 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
6097 }
6098 $node = $node->getNextSibling();
6099 } while ( $node );
6100
6101 # Write out the remainder (in replace mode only)
6102 if ( $mode === 'replace' ) {
6103 # Output the replacement text
6104 # Add two newlines on -- trailing whitespace in $newText is conventionally
6105 # stripped by the editor, so we need both newlines to restore the paragraph gap
6106 # Only add trailing whitespace if there is newText
6107 if ( $newText != "" ) {
6108 $outText .= $newText . "\n\n";
6109 }
6110
6111 while ( $node ) {
6112 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
6113 $node = $node->getNextSibling();
6114 }
6115 }
6116
6117 if ( is_string( $outText ) ) {
6118 # Re-insert stripped tags
6119 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
6120 }
6121
6122 return $outText;
6123 }
6124
6125 /**
6126 * This function returns the text of a section, specified by a number ($section).
6127 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
6128 * the first section before any such heading (section 0).
6129 *
6130 * If a section contains subsections, these are also returned.
6131 *
6132 * @param string $text Text to look in
6133 * @param string|int $sectionId Section identifier as a number or string
6134 * (e.g. 0, 1 or 'T-1').
6135 * @param string $defaultText Default to return if section is not found
6136 *
6137 * @return string Text of the requested section
6138 */
6139 public function getSection( $text, $sectionId, $defaultText = '' ) {
6140 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
6141 }
6142
6143 /**
6144 * This function returns $oldtext after the content of the section
6145 * specified by $section has been replaced with $text. If the target
6146 * section does not exist, $oldtext is returned unchanged.
6147 *
6148 * @param string $oldText Former text of the article
6149 * @param string|int $sectionId Section identifier as a number or string
6150 * (e.g. 0, 1 or 'T-1').
6151 * @param string $newText Replacing text
6152 *
6153 * @return string Modified text
6154 */
6155 public function replaceSection( $oldText, $sectionId, $newText ) {
6156 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
6157 }
6158
6159 /**
6160 * Get the ID of the revision we are parsing
6161 *
6162 * The return value will be either:
6163 * - a) Positive, indicating a specific revision ID (current or old)
6164 * - b) Zero, meaning the revision ID is specified by getCurrentRevisionCallback()
6165 * - c) Null, meaning the parse is for preview mode and there is no revision
6166 *
6167 * @return int|null
6168 */
6169 public function getRevisionId() {
6170 return $this->mRevisionId;
6171 }
6172
6173 /**
6174 * Get the revision object for $this->mRevisionId
6175 *
6176 * @return Revision|null Either a Revision object or null
6177 * @since 1.23 (public since 1.23)
6178 */
6179 public function getRevisionObject() {
6180 if ( $this->mRevisionObject ) {
6181 return $this->mRevisionObject;
6182 }
6183
6184 // NOTE: try to get the RevisionObject even if mRevisionId is null.
6185 // This is useful when parsing a revision that has not yet been saved.
6186 // However, if we get back a saved revision even though we are in
6187 // preview mode, we'll have to ignore it, see below.
6188 // NOTE: This callback may be used to inject an OLD revision that was
6189 // already loaded, so "current" is a bit of a misnomer. We can't just
6190 // skip it if mRevisionId is set.
6191 $rev = call_user_func(
6192 $this->mOptions->getCurrentRevisionCallback(),
6193 $this->getTitle(),
6194 $this
6195 );
6196
6197 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
6198 // We are in preview mode (mRevisionId is null), and the current revision callback
6199 // returned an existing revision. Ignore it and return null, it's probably the page's
6200 // current revision, which is not what we want here. Note that we do want to call the
6201 // callback to allow the unsaved revision to be injected here, e.g. for
6202 // self-transclusion previews.
6203 return null;
6204 }
6205
6206 // If the parse is for a new revision, then the callback should have
6207 // already been set to force the object and should match mRevisionId.
6208 // If not, try to fetch by mRevisionId for sanity.
6209 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
6210 $rev = Revision::newFromId( $this->mRevisionId );
6211 }
6212
6213 $this->mRevisionObject = $rev;
6214
6215 return $this->mRevisionObject;
6216 }
6217
6218 /**
6219 * Get the timestamp associated with the current revision, adjusted for
6220 * the default server-local timestamp
6221 * @return string TS_MW timestamp
6222 */
6223 public function getRevisionTimestamp() {
6224 if ( $this->mRevisionTimestamp !== null ) {
6225 return $this->mRevisionTimestamp;
6226 }
6227
6228 # Use specified revision timestamp, falling back to the current timestamp
6229 $revObject = $this->getRevisionObject();
6230 $timestamp = $revObject ? $revObject->getTimestamp() : $this->mOptions->getTimestamp();
6231 $this->mOutput->setRevisionTimestampUsed( $timestamp ); // unadjusted time zone
6232
6233 # The cryptic '' timezone parameter tells to use the site-default
6234 # timezone offset instead of the user settings.
6235 # Since this value will be saved into the parser cache, served
6236 # to other users, and potentially even used inside links and such,
6237 # it needs to be consistent for all visitors.
6238 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
6239
6240 return $this->mRevisionTimestamp;
6241 }
6242
6243 /**
6244 * Get the name of the user that edited the last revision
6245 *
6246 * @return string User name
6247 */
6248 public function getRevisionUser() {
6249 if ( is_null( $this->mRevisionUser ) ) {
6250 $revObject = $this->getRevisionObject();
6251
6252 # if this template is subst: the revision id will be blank,
6253 # so just use the current user's name
6254 if ( $revObject ) {
6255 $this->mRevisionUser = $revObject->getUserText();
6256 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6257 $this->mRevisionUser = $this->getUser()->getName();
6258 }
6259 }
6260 return $this->mRevisionUser;
6261 }
6262
6263 /**
6264 * Get the size of the revision
6265 *
6266 * @return int|null Revision size
6267 */
6268 public function getRevisionSize() {
6269 if ( is_null( $this->mRevisionSize ) ) {
6270 $revObject = $this->getRevisionObject();
6271
6272 # if this variable is subst: the revision id will be blank,
6273 # so just use the parser input size, because the own substituation
6274 # will change the size.
6275 if ( $revObject ) {
6276 $this->mRevisionSize = $revObject->getSize();
6277 } else {
6278 $this->mRevisionSize = $this->mInputSize;
6279 }
6280 }
6281 return $this->mRevisionSize;
6282 }
6283
6284 /**
6285 * Mutator for $mDefaultSort
6286 *
6287 * @param string $sort New value
6288 */
6289 public function setDefaultSort( $sort ) {
6290 $this->mDefaultSort = $sort;
6291 $this->mOutput->setProperty( 'defaultsort', $sort );
6292 }
6293
6294 /**
6295 * Accessor for $mDefaultSort
6296 * Will use the empty string if none is set.
6297 *
6298 * This value is treated as a prefix, so the
6299 * empty string is equivalent to sorting by
6300 * page name.
6301 *
6302 * @return string
6303 */
6304 public function getDefaultSort() {
6305 if ( $this->mDefaultSort !== false ) {
6306 return $this->mDefaultSort;
6307 } else {
6308 return '';
6309 }
6310 }
6311
6312 /**
6313 * Accessor for $mDefaultSort
6314 * Unlike getDefaultSort(), will return false if none is set
6315 *
6316 * @return string|bool
6317 */
6318 public function getCustomDefaultSort() {
6319 return $this->mDefaultSort;
6320 }
6321
6322 private static function getSectionNameFromStrippedText( $text ) {
6323 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6324 $text = Sanitizer::decodeCharReferences( $text );
6325 $text = self::normalizeSectionName( $text );
6326 return $text;
6327 }
6328
6329 private static function makeAnchor( $sectionName ) {
6330 return '#' . Sanitizer::escapeIdForLink( $sectionName );
6331 }
6332
6333 private function makeLegacyAnchor( $sectionName ) {
6334 $fragmentMode = $this->svcOptions->get( 'FragmentMode' );
6335 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6336 // ForAttribute() and ForLink() are the same for legacy encoding
6337 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
6338 } else {
6339 $id = Sanitizer::escapeIdForLink( $sectionName );
6340 }
6341
6342 return "#$id";
6343 }
6344
6345 /**
6346 * Try to guess the section anchor name based on a wikitext fragment
6347 * presumably extracted from a heading, for example "Header" from
6348 * "== Header ==".
6349 *
6350 * @param string $text
6351 * @return string Anchor (starting with '#')
6352 */
6353 public function guessSectionNameFromWikiText( $text ) {
6354 # Strip out wikitext links(they break the anchor)
6355 $text = $this->stripSectionName( $text );
6356 $sectionName = self::getSectionNameFromStrippedText( $text );
6357 return self::makeAnchor( $sectionName );
6358 }
6359
6360 /**
6361 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6362 * instead, if possible. For use in redirects, since various versions
6363 * of Microsoft browsers interpret Location: headers as something other
6364 * than UTF-8, resulting in breakage.
6365 *
6366 * @param string $text The section name
6367 * @return string Anchor (starting with '#')
6368 */
6369 public function guessLegacySectionNameFromWikiText( $text ) {
6370 # Strip out wikitext links(they break the anchor)
6371 $text = $this->stripSectionName( $text );
6372 $sectionName = self::getSectionNameFromStrippedText( $text );
6373 return $this->makeLegacyAnchor( $sectionName );
6374 }
6375
6376 /**
6377 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6378 * @param string $text Section name (plain text)
6379 * @return string Anchor (starting with '#')
6380 */
6381 public static function guessSectionNameFromStrippedText( $text ) {
6382 $sectionName = self::getSectionNameFromStrippedText( $text );
6383 return self::makeAnchor( $sectionName );
6384 }
6385
6386 /**
6387 * Apply the same normalization as code making links to this section would
6388 *
6389 * @param string $text
6390 * @return string
6391 */
6392 private static function normalizeSectionName( $text ) {
6393 # T90902: ensure the same normalization is applied for IDs as to links
6394 /** @var MediaWikiTitleCodec $titleParser */
6395 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
6396 '@phan-var MediaWikiTitleCodec $titleParser';
6397 try {
6398
6399 $parts = $titleParser->splitTitleString( "#$text" );
6400 } catch ( MalformedTitleException $ex ) {
6401 return $text;
6402 }
6403 return $parts['fragment'];
6404 }
6405
6406 /**
6407 * Strips a text string of wikitext for use in a section anchor
6408 *
6409 * Accepts a text string and then removes all wikitext from the
6410 * string and leaves only the resultant text (i.e. the result of
6411 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6412 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6413 * to create valid section anchors by mimicing the output of the
6414 * parser when headings are parsed.
6415 *
6416 * @param string $text Text string to be stripped of wikitext
6417 * for use in a Section anchor
6418 * @return string Filtered text string
6419 */
6420 public function stripSectionName( $text ) {
6421 # Strip internal link markup
6422 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6423 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6424
6425 # Strip external link markup
6426 # @todo FIXME: Not tolerant to blank link text
6427 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6428 # on how many empty links there are on the page - need to figure that out.
6429 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6430
6431 # Parse wikitext quotes (italics & bold)
6432 $text = $this->doQuotes( $text );
6433
6434 # Strip HTML tags
6435 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6436 return $text;
6437 }
6438
6439 /**
6440 * strip/replaceVariables/unstrip for preprocessor regression testing
6441 *
6442 * @param string $text
6443 * @param Title $title
6444 * @param ParserOptions $options
6445 * @param int $outputType
6446 *
6447 * @return string
6448 * @deprecated since 1.34; should not be used outside parser class.
6449 */
6450 public function testSrvus( $text, Title $title, ParserOptions $options,
6451 $outputType = self::OT_HTML
6452 ) {
6453 wfDeprecated( __METHOD__, '1.34' );
6454 return $this->fuzzTestSrvus( $text, $title, $options, $outputType );
6455 }
6456
6457 /**
6458 * Strip/replaceVariables/unstrip for preprocessor regression testing
6459 *
6460 * @param string $text
6461 * @param Title $title
6462 * @param ParserOptions $options
6463 * @param int $outputType
6464 *
6465 * @return string
6466 */
6467 private function fuzzTestSrvus( $text, Title $title, ParserOptions $options,
6468 $outputType = self::OT_HTML
6469 ) {
6470 $magicScopeVariable = $this->lock();
6471 $this->startParse( $title, $options, $outputType, true );
6472
6473 $text = $this->replaceVariables( $text );
6474 $text = $this->mStripState->unstripBoth( $text );
6475 $text = Sanitizer::removeHTMLtags( $text );
6476 return $text;
6477 }
6478
6479 /**
6480 * @param string $text
6481 * @param Title $title
6482 * @param ParserOptions $options
6483 * @return string
6484 * @deprecated since 1.34; should not be used outside parser class.
6485 */
6486 public function testPst( $text, Title $title, ParserOptions $options ) {
6487 wfDeprecated( __METHOD__, '1.34' );
6488 return $this->fuzzTestPst( $text, $title, $options );
6489 }
6490
6491 /**
6492 * @param string $text
6493 * @param Title $title
6494 * @param ParserOptions $options
6495 * @return string
6496 */
6497 private function fuzzTestPst( $text, Title $title, ParserOptions $options ) {
6498 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6499 }
6500
6501 /**
6502 * @param string $text
6503 * @param Title $title
6504 * @param ParserOptions $options
6505 * @return string
6506 * @deprecated since 1.34; should not be used outside parser class.
6507 */
6508 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6509 wfDeprecated( __METHOD__, '1.34' );
6510 return $this->fuzzTestPreprocess( $text, $title, $options );
6511 }
6512
6513 /**
6514 * @param string $text
6515 * @param Title $title
6516 * @param ParserOptions $options
6517 * @return string
6518 */
6519 private function fuzzTestPreprocess( $text, Title $title, ParserOptions $options ) {
6520 return $this->fuzzTestSrvus( $text, $title, $options, self::OT_PREPROCESS );
6521 }
6522
6523 /**
6524 * Call a callback function on all regions of the given text that are not
6525 * inside strip markers, and replace those regions with the return value
6526 * of the callback. For example, with input:
6527 *
6528 * aaa<MARKER>bbb
6529 *
6530 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6531 * two strings will be replaced with the value returned by the callback in
6532 * each case.
6533 *
6534 * @param string $s
6535 * @param callable $callback
6536 *
6537 * @return string
6538 */
6539 public function markerSkipCallback( $s, $callback ) {
6540 $i = 0;
6541 $out = '';
6542 while ( $i < strlen( $s ) ) {
6543 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6544 if ( $markerStart === false ) {
6545 $out .= call_user_func( $callback, substr( $s, $i ) );
6546 break;
6547 } else {
6548 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6549 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6550 if ( $markerEnd === false ) {
6551 $out .= substr( $s, $markerStart );
6552 break;
6553 } else {
6554 $markerEnd += strlen( self::MARKER_SUFFIX );
6555 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6556 $i = $markerEnd;
6557 }
6558 }
6559 }
6560 return $out;
6561 }
6562
6563 /**
6564 * Remove any strip markers found in the given text.
6565 *
6566 * @param string $text
6567 * @return string
6568 */
6569 public function killMarkers( $text ) {
6570 return $this->mStripState->killMarkers( $text );
6571 }
6572
6573 /**
6574 * Save the parser state required to convert the given half-parsed text to
6575 * HTML. "Half-parsed" in this context means the output of
6576 * recursiveTagParse() or internalParse(). This output has strip markers
6577 * from replaceVariables (extensionSubstitution() etc.), and link
6578 * placeholders from replaceLinkHolders().
6579 *
6580 * Returns an array which can be serialized and stored persistently. This
6581 * array can later be loaded into another parser instance with
6582 * unserializeHalfParsedText(). The text can then be safely incorporated into
6583 * the return value of a parser hook.
6584 *
6585 * @deprecated since 1.31
6586 * @param string $text
6587 *
6588 * @return array
6589 */
6590 public function serializeHalfParsedText( $text ) {
6591 wfDeprecated( __METHOD__, '1.31' );
6592 $data = [
6593 'text' => $text,
6594 'version' => self::HALF_PARSED_VERSION,
6595 'stripState' => $this->mStripState->getSubState( $text ),
6596 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6597 ];
6598 return $data;
6599 }
6600
6601 /**
6602 * Load the parser state given in the $data array, which is assumed to
6603 * have been generated by serializeHalfParsedText(). The text contents is
6604 * extracted from the array, and its markers are transformed into markers
6605 * appropriate for the current Parser instance. This transformed text is
6606 * returned, and can be safely included in the return value of a parser
6607 * hook.
6608 *
6609 * If the $data array has been stored persistently, the caller should first
6610 * check whether it is still valid, by calling isValidHalfParsedText().
6611 *
6612 * @deprecated since 1.31
6613 * @param array $data Serialized data
6614 * @throws MWException
6615 * @return string
6616 */
6617 public function unserializeHalfParsedText( $data ) {
6618 wfDeprecated( __METHOD__, '1.31' );
6619 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6620 throw new MWException( __METHOD__ . ': invalid version' );
6621 }
6622
6623 # First, extract the strip state.
6624 $texts = [ $data['text'] ];
6625 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6626
6627 # Now renumber links
6628 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6629
6630 # Should be good to go.
6631 return $texts[0];
6632 }
6633
6634 /**
6635 * Returns true if the given array, presumed to be generated by
6636 * serializeHalfParsedText(), is compatible with the current version of the
6637 * parser.
6638 *
6639 * @deprecated since 1.31
6640 * @param array $data
6641 *
6642 * @return bool
6643 */
6644 public function isValidHalfParsedText( $data ) {
6645 wfDeprecated( __METHOD__, '1.31' );
6646 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6647 }
6648
6649 /**
6650 * Parsed a width param of imagelink like 300px or 200x300px
6651 *
6652 * @param string $value
6653 * @param bool $parseHeight
6654 *
6655 * @return array
6656 * @since 1.20
6657 */
6658 public static function parseWidthParam( $value, $parseHeight = true ) {
6659 $parsedWidthParam = [];
6660 if ( $value === '' ) {
6661 return $parsedWidthParam;
6662 }
6663 $m = [];
6664 # (T15500) In both cases (width/height and width only),
6665 # permit trailing "px" for backward compatibility.
6666 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6667 $width = intval( $m[1] );
6668 $height = intval( $m[2] );
6669 $parsedWidthParam['width'] = $width;
6670 $parsedWidthParam['height'] = $height;
6671 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6672 $width = intval( $value );
6673 $parsedWidthParam['width'] = $width;
6674 }
6675 return $parsedWidthParam;
6676 }
6677
6678 /**
6679 * Lock the current instance of the parser.
6680 *
6681 * This is meant to stop someone from calling the parser
6682 * recursively and messing up all the strip state.
6683 *
6684 * @throws MWException If parser is in a parse
6685 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6686 */
6687 protected function lock() {
6688 if ( $this->mInParse ) {
6689 throw new MWException( "Parser state cleared while parsing. "
6690 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6691 }
6692
6693 // Save the backtrace when locking, so that if some code tries locking again,
6694 // we can print the lock owner's backtrace for easier debugging
6695 $e = new Exception;
6696 $this->mInParse = $e->getTraceAsString();
6697
6698 $recursiveCheck = new ScopedCallback( function () {
6699 $this->mInParse = false;
6700 } );
6701
6702 return $recursiveCheck;
6703 }
6704
6705 /**
6706 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6707 *
6708 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6709 * or if there is more than one <p/> tag in the input HTML.
6710 *
6711 * @param string $html
6712 * @return string
6713 * @since 1.24
6714 */
6715 public static function stripOuterParagraph( $html ) {
6716 $m = [];
6717 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6718 $html = $m[1];
6719 }
6720
6721 return $html;
6722 }
6723
6724 /**
6725 * Return this parser if it is not doing anything, otherwise
6726 * get a fresh parser. You can use this method by doing
6727 * $newParser = $oldParser->getFreshParser(), or more simply
6728 * $oldParser->getFreshParser()->parse( ... );
6729 * if you're unsure if $oldParser is safe to use.
6730 *
6731 * @since 1.24
6732 * @return Parser A parser object that is not parsing anything
6733 */
6734 public function getFreshParser() {
6735 if ( $this->mInParse ) {
6736 return $this->factory->create();
6737 } else {
6738 return $this;
6739 }
6740 }
6741
6742 /**
6743 * Set's up the PHP implementation of OOUI for use in this request
6744 * and instructs OutputPage to enable OOUI for itself.
6745 *
6746 * @since 1.26
6747 */
6748 public function enableOOUI() {
6749 OutputPage::setupOOUI();
6750 $this->mOutput->setEnableOOUI( true );
6751 }
6752
6753 /**
6754 * @param string $flag
6755 * @param string $reason
6756 */
6757 protected function setOutputFlag( $flag, $reason ) {
6758 $this->mOutput->setFlag( $flag );
6759 $name = $this->mTitle->getPrefixedText();
6760 $this->logger->debug( __METHOD__ . ": set $flag flag on '$name'; $reason" );
6761 }
6762 }
Wiklou, le Wiki du Wiklou