dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Deprecate Parser::areSubpagesAllowed() / Parser::maybeDoSubpageLink()
[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->replaceLinkHolders( $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 */
3240 public static function splitWhitespace( $s ) {
3241 $ltrimmed = ltrim( $s );
3242 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3243 $trimmed = rtrim( $ltrimmed );
3244 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3245 if ( $diff > 0 ) {
3246 $w2 = substr( $ltrimmed, -$diff );
3247 } else {
3248 $w2 = '';
3249 }
3250 return [ $w1, $trimmed, $w2 ];
3251 }
3252
3253 /**
3254 * Replace magic variables, templates, and template arguments
3255 * with the appropriate text. Templates are substituted recursively,
3256 * taking care to avoid infinite loops.
3257 *
3258 * Note that the substitution depends on value of $mOutputType:
3259 * self::OT_WIKI: only {{subst:}} templates
3260 * self::OT_PREPROCESS: templates but not extension tags
3261 * self::OT_HTML: all templates and extension tags
3262 *
3263 * @param string $text The text to transform
3264 * @param false|PPFrame|array $frame Object describing the arguments passed to the
3265 * template. Arguments may also be provided as an associative array, as
3266 * was the usual case before MW1.12. Providing arguments this way may be
3267 * useful for extensions wishing to perform variable replacement
3268 * explicitly.
3269 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3270 * double-brace expansion.
3271 * @return string
3272 */
3273 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3274 # Is there any text? Also, Prevent too big inclusions!
3275 $textSize = strlen( $text );
3276 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3277 return $text;
3278 }
3279
3280 if ( $frame === false ) {
3281 $frame = $this->getPreprocessor()->newFrame();
3282 } elseif ( !( $frame instanceof PPFrame ) ) {
3283 $this->logger->debug(
3284 __METHOD__ . " called using plain parameters instead of " .
3285 "a PPFrame instance. Creating custom frame."
3286 );
3287 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3288 }
3289
3290 $dom = $this->preprocessToDom( $text );
3291 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3292 $text = $frame->expand( $dom, $flags );
3293
3294 return $text;
3295 }
3296
3297 /**
3298 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3299 *
3300 * @param array $args
3301 *
3302 * @return array
3303 */
3304 public static function createAssocArgs( $args ) {
3305 $assocArgs = [];
3306 $index = 1;
3307 foreach ( $args as $arg ) {
3308 $eqpos = strpos( $arg, '=' );
3309 if ( $eqpos === false ) {
3310 $assocArgs[$index++] = $arg;
3311 } else {
3312 $name = trim( substr( $arg, 0, $eqpos ) );
3313 $value = trim( substr( $arg, $eqpos + 1 ) );
3314 if ( $value === false ) {
3315 $value = '';
3316 }
3317 if ( $name !== false ) {
3318 $assocArgs[$name] = $value;
3319 }
3320 }
3321 }
3322
3323 return $assocArgs;
3324 }
3325
3326 /**
3327 * Warn the user when a parser limitation is reached
3328 * Will warn at most once the user per limitation type
3329 *
3330 * The results are shown during preview and run through the Parser (See EditPage.php)
3331 *
3332 * @param string $limitationType Should be one of:
3333 * 'expensive-parserfunction' (corresponding messages:
3334 * 'expensive-parserfunction-warning',
3335 * 'expensive-parserfunction-category')
3336 * 'post-expand-template-argument' (corresponding messages:
3337 * 'post-expand-template-argument-warning',
3338 * 'post-expand-template-argument-category')
3339 * 'post-expand-template-inclusion' (corresponding messages:
3340 * 'post-expand-template-inclusion-warning',
3341 * 'post-expand-template-inclusion-category')
3342 * 'node-count-exceeded' (corresponding messages:
3343 * 'node-count-exceeded-warning',
3344 * 'node-count-exceeded-category')
3345 * 'expansion-depth-exceeded' (corresponding messages:
3346 * 'expansion-depth-exceeded-warning',
3347 * 'expansion-depth-exceeded-category')
3348 * @param string|int|null $current Current value
3349 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3350 * exceeded, provide the values (optional)
3351 */
3352 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3353 # does no harm if $current and $max are present but are unnecessary for the message
3354 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3355 # only during preview, and that would split the parser cache unnecessarily.
3356 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3357 ->text();
3358 $this->mOutput->addWarning( $warning );
3359 $this->addTrackingCategory( "$limitationType-category" );
3360 }
3361
3362 /**
3363 * Return the text of a template, after recursively
3364 * replacing any variables or templates within the template.
3365 *
3366 * @param array $piece The parts of the template
3367 * $piece['title']: the title, i.e. the part before the |
3368 * $piece['parts']: the parameter array
3369 * $piece['lineStart']: whether the brace was at the start of a line
3370 * @param PPFrame $frame The current frame, contains template arguments
3371 * @throws Exception
3372 * @return string|array The text of the template
3373 * @internal
3374 */
3375 public function braceSubstitution( $piece, $frame ) {
3376 // Flags
3377
3378 // $text has been filled
3379 $found = false;
3380 // wiki markup in $text should be escaped
3381 $nowiki = false;
3382 // $text is HTML, armour it against wikitext transformation
3383 $isHTML = false;
3384 // Force interwiki transclusion to be done in raw mode not rendered
3385 $forceRawInterwiki = false;
3386 // $text is a DOM node needing expansion in a child frame
3387 $isChildObj = false;
3388 // $text is a DOM node needing expansion in the current frame
3389 $isLocalObj = false;
3390
3391 # Title object, where $text came from
3392 $title = false;
3393
3394 # $part1 is the bit before the first |, and must contain only title characters.
3395 # Various prefixes will be stripped from it later.
3396 $titleWithSpaces = $frame->expand( $piece['title'] );
3397 $part1 = trim( $titleWithSpaces );
3398 $titleText = false;
3399
3400 # Original title text preserved for various purposes
3401 $originalTitle = $part1;
3402
3403 # $args is a list of argument nodes, starting from index 0, not including $part1
3404 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3405 # below won't work b/c this $args isn't an object
3406 $args = ( $piece['parts'] == null ) ? [] : $piece['parts'];
3407
3408 $profileSection = null; // profile templates
3409
3410 # SUBST
3411 if ( !$found ) {
3412 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3413
3414 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3415 # Decide whether to expand template or keep wikitext as-is.
3416 if ( $this->ot['wiki'] ) {
3417 if ( $substMatch === false ) {
3418 $literal = true; # literal when in PST with no prefix
3419 } else {
3420 $literal = false; # expand when in PST with subst: or safesubst:
3421 }
3422 } else {
3423 if ( $substMatch == 'subst' ) {
3424 $literal = true; # literal when not in PST with plain subst:
3425 } else {
3426 $literal = false; # expand when not in PST with safesubst: or no prefix
3427 }
3428 }
3429 if ( $literal ) {
3430 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3431 $isLocalObj = true;
3432 $found = true;
3433 }
3434 }
3435
3436 # Variables
3437 if ( !$found && $args->getLength() == 0 ) {
3438 $id = $this->mVariables->matchStartToEnd( $part1 );
3439 if ( $id !== false ) {
3440 $text = $this->expandMagicVariable( $id, $frame );
3441 if ( $this->magicWordFactory->getCacheTTL( $id ) > -1 ) {
3442 $this->mOutput->updateCacheExpiry(
3443 $this->magicWordFactory->getCacheTTL( $id ) );
3444 }
3445 $found = true;
3446 }
3447 }
3448
3449 # MSG, MSGNW and RAW
3450 if ( !$found ) {
3451 # Check for MSGNW:
3452 $mwMsgnw = $this->magicWordFactory->get( 'msgnw' );
3453 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3454 $nowiki = true;
3455 } else {
3456 # Remove obsolete MSG:
3457 $mwMsg = $this->magicWordFactory->get( 'msg' );
3458 $mwMsg->matchStartAndRemove( $part1 );
3459 }
3460
3461 # Check for RAW:
3462 $mwRaw = $this->magicWordFactory->get( 'raw' );
3463 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3464 $forceRawInterwiki = true;
3465 }
3466 }
3467
3468 # Parser functions
3469 if ( !$found ) {
3470 $colonPos = strpos( $part1, ':' );
3471 if ( $colonPos !== false ) {
3472 $func = substr( $part1, 0, $colonPos );
3473 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3474 $argsLength = $args->getLength();
3475 for ( $i = 0; $i < $argsLength; $i++ ) {
3476 $funcArgs[] = $args->item( $i );
3477 }
3478
3479 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3480
3481 // Extract any forwarded flags
3482 if ( isset( $result['title'] ) ) {
3483 $title = $result['title'];
3484 }
3485 if ( isset( $result['found'] ) ) {
3486 $found = $result['found'];
3487 }
3488 if ( array_key_exists( 'text', $result ) ) {
3489 // a string or null
3490 $text = $result['text'];
3491 }
3492 if ( isset( $result['nowiki'] ) ) {
3493 $nowiki = $result['nowiki'];
3494 }
3495 if ( isset( $result['isHTML'] ) ) {
3496 $isHTML = $result['isHTML'];
3497 }
3498 if ( isset( $result['forceRawInterwiki'] ) ) {
3499 $forceRawInterwiki = $result['forceRawInterwiki'];
3500 }
3501 if ( isset( $result['isChildObj'] ) ) {
3502 $isChildObj = $result['isChildObj'];
3503 }
3504 if ( isset( $result['isLocalObj'] ) ) {
3505 $isLocalObj = $result['isLocalObj'];
3506 }
3507 }
3508 }
3509
3510 # Finish mangling title and then check for loops.
3511 # Set $title to a Title object and $titleText to the PDBK
3512 if ( !$found ) {
3513 $ns = NS_TEMPLATE;
3514 # Split the title into page and subpage
3515 $subpage = '';
3516 $relative = Linker::normalizeSubpageLink(
3517 $this->mTitle, $part1, $subpage
3518 );
3519 if ( $part1 !== $relative ) {
3520 $part1 = $relative;
3521 $ns = $this->mTitle->getNamespace();
3522 }
3523 $title = Title::newFromText( $part1, $ns );
3524 if ( $title ) {
3525 $titleText = $title->getPrefixedText();
3526 # Check for language variants if the template is not found
3527 if ( $this->getTargetLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3528 $this->getTargetLanguage()->findVariantLink( $part1, $title, true );
3529 }
3530 # Do recursion depth check
3531 $limit = $this->mOptions->getMaxTemplateDepth();
3532 if ( $frame->depth >= $limit ) {
3533 $found = true;
3534 $text = '<span class="error">'
3535 . wfMessage( 'parser-template-recursion-depth-warning' )
3536 ->numParams( $limit )->inContentLanguage()->text()
3537 . '</span>';
3538 }
3539 }
3540 }
3541
3542 # Load from database
3543 if ( !$found && $title ) {
3544 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3545 if ( !$title->isExternal() ) {
3546 if ( $title->isSpecialPage()
3547 && $this->mOptions->getAllowSpecialInclusion()
3548 && $this->ot['html']
3549 ) {
3550 $specialPage = $this->specialPageFactory->getPage( $title->getDBkey() );
3551 // Pass the template arguments as URL parameters.
3552 // "uselang" will have no effect since the Language object
3553 // is forced to the one defined in ParserOptions.
3554 $pageArgs = [];
3555 $argsLength = $args->getLength();
3556 for ( $i = 0; $i < $argsLength; $i++ ) {
3557 $bits = $args->item( $i )->splitArg();
3558 if ( strval( $bits['index'] ) === '' ) {
3559 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3560 $value = trim( $frame->expand( $bits['value'] ) );
3561 $pageArgs[$name] = $value;
3562 }
3563 }
3564
3565 // Create a new context to execute the special page
3566 $context = new RequestContext;
3567 $context->setTitle( $title );
3568 $context->setRequest( new FauxRequest( $pageArgs ) );
3569 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3570 $context->setUser( $this->getUser() );
3571 } else {
3572 // If this page is cached, then we better not be per user.
3573 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3574 }
3575 $context->setLanguage( $this->mOptions->getUserLangObj() );
3576 $ret = $this->specialPageFactory->capturePath( $title, $context, $this->getLinkRenderer() );
3577 if ( $ret ) {
3578 $text = $context->getOutput()->getHTML();
3579 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3580 $found = true;
3581 $isHTML = true;
3582 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3583 $this->mOutput->updateRuntimeAdaptiveExpiry(
3584 $specialPage->maxIncludeCacheTime()
3585 );
3586 }
3587 }
3588 } elseif ( $this->nsInfo->isNonincludable( $title->getNamespace() ) ) {
3589 $found = false; # access denied
3590 $this->logger->debug(
3591 __METHOD__ .
3592 ": template inclusion denied for " . $title->getPrefixedDBkey()
3593 );
3594 } else {
3595 list( $text, $title ) = $this->getTemplateDom( $title );
3596 if ( $text !== false ) {
3597 $found = true;
3598 $isChildObj = true;
3599 }
3600 }
3601
3602 # If the title is valid but undisplayable, make a link to it
3603 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3604 $text = "[[:$titleText]]";
3605 $found = true;
3606 }
3607 } elseif ( $title->isTrans() ) {
3608 # Interwiki transclusion
3609 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3610 $text = $this->interwikiTransclude( $title, 'render' );
3611 $isHTML = true;
3612 } else {
3613 $text = $this->interwikiTransclude( $title, 'raw' );
3614 # Preprocess it like a template
3615 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3616 $isChildObj = true;
3617 }
3618 $found = true;
3619 }
3620
3621 # Do infinite loop check
3622 # This has to be done after redirect resolution to avoid infinite loops via redirects
3623 if ( !$frame->loopCheck( $title ) ) {
3624 $found = true;
3625 $text = '<span class="error">'
3626 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3627 . '</span>';
3628 $this->addTrackingCategory( 'template-loop-category' );
3629 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3630 wfEscapeWikiText( $titleText ) )->text() );
3631 $this->logger->debug( __METHOD__ . ": template loop broken at '$titleText'" );
3632 }
3633 }
3634
3635 # If we haven't found text to substitute by now, we're done
3636 # Recover the source wikitext and return it
3637 if ( !$found ) {
3638 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3639 if ( $profileSection ) {
3640 $this->mProfiler->scopedProfileOut( $profileSection );
3641 }
3642 return [ 'object' => $text ];
3643 }
3644
3645 # Expand DOM-style return values in a child frame
3646 if ( $isChildObj ) {
3647 # Clean up argument array
3648 $newFrame = $frame->newChild( $args, $title );
3649
3650 if ( $nowiki ) {
3651 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3652 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3653 # Expansion is eligible for the empty-frame cache
3654 $text = $newFrame->cachedExpand( $titleText, $text );
3655 } else {
3656 # Uncached expansion
3657 $text = $newFrame->expand( $text );
3658 }
3659 }
3660 if ( $isLocalObj && $nowiki ) {
3661 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3662 $isLocalObj = false;
3663 }
3664
3665 if ( $profileSection ) {
3666 $this->mProfiler->scopedProfileOut( $profileSection );
3667 }
3668
3669 # Replace raw HTML by a placeholder
3670 if ( $isHTML ) {
3671 $text = $this->insertStripItem( $text );
3672 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3673 # Escape nowiki-style return values
3674 $text = wfEscapeWikiText( $text );
3675 } elseif ( is_string( $text )
3676 && !$piece['lineStart']
3677 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3678 ) {
3679 # T2529: if the template begins with a table or block-level
3680 # element, it should be treated as beginning a new line.
3681 # This behavior is somewhat controversial.
3682 $text = "\n" . $text;
3683 }
3684
3685 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3686 # Error, oversize inclusion
3687 if ( $titleText !== false ) {
3688 # Make a working, properly escaped link if possible (T25588)
3689 $text = "[[:$titleText]]";
3690 } else {
3691 # This will probably not be a working link, but at least it may
3692 # provide some hint of where the problem is
3693 preg_replace( '/^:/', '', $originalTitle );
3694 $text = "[[:$originalTitle]]";
3695 }
3696 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3697 . 'post-expand include size too large -->' );
3698 $this->limitationWarn( 'post-expand-template-inclusion' );
3699 }
3700
3701 if ( $isLocalObj ) {
3702 $ret = [ 'object' => $text ];
3703 } else {
3704 $ret = [ 'text' => $text ];
3705 }
3706
3707 return $ret;
3708 }
3709
3710 /**
3711 * Call a parser function and return an array with text and flags.
3712 *
3713 * The returned array will always contain a boolean 'found', indicating
3714 * whether the parser function was found or not. It may also contain the
3715 * following:
3716 * text: string|object, resulting wikitext or PP DOM object
3717 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3718 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3719 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3720 * nowiki: bool, wiki markup in $text should be escaped
3721 *
3722 * @since 1.21
3723 * @param PPFrame $frame The current frame, contains template arguments
3724 * @param string $function Function name
3725 * @param array $args Arguments to the function
3726 * @throws MWException
3727 * @return array
3728 */
3729 public function callParserFunction( $frame, $function, array $args = [] ) {
3730 # Case sensitive functions
3731 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3732 $function = $this->mFunctionSynonyms[1][$function];
3733 } else {
3734 # Case insensitive functions
3735 $function = $this->contLang->lc( $function );
3736 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3737 $function = $this->mFunctionSynonyms[0][$function];
3738 } else {
3739 return [ 'found' => false ];
3740 }
3741 }
3742
3743 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3744
3745 // Avoid PHP 7.1 warning from passing $this by reference
3746 $parser = $this;
3747
3748 $allArgs = [ &$parser ];
3749 if ( $flags & self::SFH_OBJECT_ARGS ) {
3750 # Convert arguments to PPNodes and collect for appending to $allArgs
3751 $funcArgs = [];
3752 foreach ( $args as $k => $v ) {
3753 if ( $v instanceof PPNode || $k === 0 ) {
3754 $funcArgs[] = $v;
3755 } else {
3756 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3757 }
3758 }
3759
3760 # Add a frame parameter, and pass the arguments as an array
3761 $allArgs[] = $frame;
3762 $allArgs[] = $funcArgs;
3763 } else {
3764 # Convert arguments to plain text and append to $allArgs
3765 foreach ( $args as $k => $v ) {
3766 if ( $v instanceof PPNode ) {
3767 $allArgs[] = trim( $frame->expand( $v ) );
3768 } elseif ( is_int( $k ) && $k >= 0 ) {
3769 $allArgs[] = trim( $v );
3770 } else {
3771 $allArgs[] = trim( "$k=$v" );
3772 }
3773 }
3774 }
3775
3776 $result = $callback( ...$allArgs );
3777
3778 # The interface for function hooks allows them to return a wikitext
3779 # string or an array containing the string and any flags. This mungs
3780 # things around to match what this method should return.
3781 if ( !is_array( $result ) ) {
3782 $result = [
3783 'found' => true,
3784 'text' => $result,
3785 ];
3786 } else {
3787 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3788 $result['text'] = $result[0];
3789 }
3790 unset( $result[0] );
3791 $result += [
3792 'found' => true,
3793 ];
3794 }
3795
3796 $noparse = true;
3797 $preprocessFlags = 0;
3798 if ( isset( $result['noparse'] ) ) {
3799 $noparse = $result['noparse'];
3800 }
3801 if ( isset( $result['preprocessFlags'] ) ) {
3802 $preprocessFlags = $result['preprocessFlags'];
3803 }
3804
3805 if ( !$noparse ) {
3806 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3807 $result['isChildObj'] = true;
3808 }
3809
3810 return $result;
3811 }
3812
3813 /**
3814 * Get the semi-parsed DOM representation of a template with a given title,
3815 * and its redirect destination title. Cached.
3816 *
3817 * @param Title $title
3818 *
3819 * @return array
3820 */
3821 public function getTemplateDom( $title ) {
3822 $cacheTitle = $title;
3823 $titleText = $title->getPrefixedDBkey();
3824
3825 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3826 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3827 $title = Title::makeTitle( $ns, $dbk );
3828 $titleText = $title->getPrefixedDBkey();
3829 }
3830 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3831 return [ $this->mTplDomCache[$titleText], $title ];
3832 }
3833
3834 # Cache miss, go to the database
3835 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3836
3837 if ( $text === false ) {
3838 $this->mTplDomCache[$titleText] = false;
3839 return [ false, $title ];
3840 }
3841
3842 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3843 $this->mTplDomCache[$titleText] = $dom;
3844
3845 if ( !$title->equals( $cacheTitle ) ) {
3846 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3847 [ $title->getNamespace(), $title->getDBkey() ];
3848 }
3849
3850 return [ $dom, $title ];
3851 }
3852
3853 /**
3854 * Fetch the current revision of a given title. Note that the revision
3855 * (and even the title) may not exist in the database, so everything
3856 * contributing to the output of the parser should use this method
3857 * where possible, rather than getting the revisions themselves. This
3858 * method also caches its results, so using it benefits performance.
3859 *
3860 * @since 1.24
3861 * @param Title $title
3862 * @return Revision
3863 */
3864 public function fetchCurrentRevisionOfTitle( $title ) {
3865 $cacheKey = $title->getPrefixedDBkey();
3866 if ( !$this->currentRevisionCache ) {
3867 $this->currentRevisionCache = new MapCacheLRU( 100 );
3868 }
3869 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3870 $this->currentRevisionCache->set( $cacheKey,
3871 // Defaults to Parser::statelessFetchRevision()
3872 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3873 );
3874 }
3875 return $this->currentRevisionCache->get( $cacheKey );
3876 }
3877
3878 /**
3879 * @param Title $title
3880 * @return bool
3881 * @since 1.34
3882 */
3883 public function isCurrentRevisionOfTitleCached( $title ) {
3884 return (
3885 $this->currentRevisionCache &&
3886 $this->currentRevisionCache->has( $title->getPrefixedText() )
3887 );
3888 }
3889
3890 /**
3891 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3892 * without passing them on to it.
3893 *
3894 * @since 1.24
3895 * @param Title $title
3896 * @param Parser|bool $parser
3897 * @return Revision|bool False if missing
3898 */
3899 public static function statelessFetchRevision( Title $title, $parser = false ) {
3900 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $title );
3901
3902 return $rev;
3903 }
3904
3905 /**
3906 * Fetch the unparsed text of a template and register a reference to it.
3907 * @param Title $title
3908 * @return array ( string or false, Title )
3909 */
3910 public function fetchTemplateAndTitle( $title ) {
3911 // Defaults to Parser::statelessFetchTemplate()
3912 $templateCb = $this->mOptions->getTemplateCallback();
3913 $stuff = call_user_func( $templateCb, $title, $this );
3914 $rev = $stuff['revision'] ?? null;
3915 $text = $stuff['text'];
3916 if ( is_string( $stuff['text'] ) ) {
3917 // We use U+007F DELETE to distinguish strip markers from regular text
3918 $text = strtr( $text, "\x7f", "?" );
3919 }
3920 $finalTitle = $stuff['finalTitle'] ?? $title;
3921 foreach ( ( $stuff['deps'] ?? [] ) as $dep ) {
3922 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3923 if ( $dep['title']->equals( $this->getTitle() ) && $rev instanceof Revision ) {
3924 // Self-transclusion; final result may change based on the new page version
3925 $this->setOutputFlag( 'vary-revision-sha1', 'Self transclusion' );
3926 $this->getOutput()->setRevisionUsedSha1Base36( $rev->getSha1() );
3927 }
3928 }
3929
3930 return [ $text, $finalTitle ];
3931 }
3932
3933 /**
3934 * Fetch the unparsed text of a template and register a reference to it.
3935 * @param Title $title
3936 * @return string|bool
3937 */
3938 public function fetchTemplate( $title ) {
3939 return $this->fetchTemplateAndTitle( $title )[0];
3940 }
3941
3942 /**
3943 * Static function to get a template
3944 * Can be overridden via ParserOptions::setTemplateCallback().
3945 *
3946 * @param Title $title
3947 * @param bool|Parser $parser
3948 *
3949 * @return array
3950 */
3951 public static function statelessFetchTemplate( $title, $parser = false ) {
3952 $text = $skip = false;
3953 $finalTitle = $title;
3954 $deps = [];
3955 $rev = null;
3956
3957 # Loop to fetch the article, with up to 1 redirect
3958 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3959 # Give extensions a chance to select the revision instead
3960 $id = false; # Assume current
3961 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3962 [ $parser, $title, &$skip, &$id ] );
3963
3964 if ( $skip ) {
3965 $text = false;
3966 $deps[] = [
3967 'title' => $title,
3968 'page_id' => $title->getArticleID(),
3969 'rev_id' => null
3970 ];
3971 break;
3972 }
3973 # Get the revision
3974 if ( $id ) {
3975 $rev = Revision::newFromId( $id );
3976 } elseif ( $parser ) {
3977 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3978 } else {
3979 $rev = Revision::newFromTitle( $title );
3980 }
3981 $rev_id = $rev ? $rev->getId() : 0;
3982 # If there is no current revision, there is no page
3983 if ( $id === false && !$rev ) {
3984 $linkCache = MediaWikiServices::getInstance()->getLinkCache();
3985 $linkCache->addBadLinkObj( $title );
3986 }
3987
3988 $deps[] = [
3989 'title' => $title,
3990 'page_id' => $title->getArticleID(),
3991 'rev_id' => $rev_id
3992 ];
3993 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3994 # We fetched a rev from a different title; register it too...
3995 $deps[] = [
3996 'title' => $rev->getTitle(),
3997 'page_id' => $rev->getPage(),
3998 'rev_id' => $rev_id
3999 ];
4000 }
4001
4002 if ( $rev ) {
4003 $content = $rev->getContent();
4004 $text = $content ? $content->getWikitextForTransclusion() : null;
4005
4006 Hooks::run( 'ParserFetchTemplate',
4007 [ $parser, $title, $rev, &$text, &$deps ] );
4008
4009 if ( $text === false || $text === null ) {
4010 $text = false;
4011 break;
4012 }
4013 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
4014 $message = wfMessage( MediaWikiServices::getInstance()->getContentLanguage()->
4015 lcfirst( $title->getText() ) )->inContentLanguage();
4016 if ( !$message->exists() ) {
4017 $text = false;
4018 break;
4019 }
4020 $content = $message->content();
4021 $text = $message->plain();
4022 } else {
4023 break;
4024 }
4025 if ( !$content ) {
4026 break;
4027 }
4028 # Redirect?
4029 $finalTitle = $title;
4030 $title = $content->getRedirectTarget();
4031 }
4032 return [
4033 'revision' => $rev,
4034 'text' => $text,
4035 'finalTitle' => $finalTitle,
4036 'deps' => $deps
4037 ];
4038 }
4039
4040 /**
4041 * Fetch a file and its title and register a reference to it.
4042 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
4043 * @param Title $title
4044 * @param array $options Array of options to RepoGroup::findFile
4045 * @return array ( File or false, Title of file )
4046 */
4047 public function fetchFileAndTitle( $title, $options = [] ) {
4048 $file = $this->fetchFileNoRegister( $title, $options );
4049
4050 $time = $file ? $file->getTimestamp() : false;
4051 $sha1 = $file ? $file->getSha1() : false;
4052 # Register the file as a dependency...
4053 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4054 if ( $file && !$title->equals( $file->getTitle() ) ) {
4055 # Update fetched file title
4056 $title = $file->getTitle();
4057 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4058 }
4059 return [ $file, $title ];
4060 }
4061
4062 /**
4063 * Helper function for fetchFileAndTitle.
4064 *
4065 * Also useful if you need to fetch a file but not use it yet,
4066 * for example to get the file's handler.
4067 *
4068 * @param Title $title
4069 * @param array $options Array of options to RepoGroup::findFile
4070 * @return File|bool
4071 */
4072 protected function fetchFileNoRegister( $title, $options = [] ) {
4073 if ( isset( $options['broken'] ) ) {
4074 $file = false; // broken thumbnail forced by hook
4075 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
4076 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
4077 } else { // get by (name,timestamp)
4078 $file = MediaWikiServices::getInstance()->getRepoGroup()->findFile( $title, $options );
4079 }
4080 return $file;
4081 }
4082
4083 /**
4084 * Transclude an interwiki link.
4085 *
4086 * @param Title $title
4087 * @param string $action Usually one of (raw, render)
4088 *
4089 * @return string
4090 */
4091 public function interwikiTransclude( $title, $action ) {
4092 if ( !$this->svcOptions->get( 'EnableScaryTranscluding' ) ) {
4093 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
4094 }
4095
4096 $url = $title->getFullURL( [ 'action' => $action ] );
4097 if ( strlen( $url ) > 1024 ) {
4098 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
4099 }
4100
4101 $wikiId = $title->getTransWikiID(); // remote wiki ID or false
4102
4103 $fname = __METHOD__;
4104 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
4105
4106 $data = $cache->getWithSetCallback(
4107 $cache->makeGlobalKey(
4108 'interwiki-transclude',
4109 ( $wikiId !== false ) ? $wikiId : 'external',
4110 sha1( $url )
4111 ),
4112 $this->svcOptions->get( 'TranscludeCacheExpiry' ),
4113 function ( $oldValue, &$ttl ) use ( $url, $fname, $cache ) {
4114 $req = MWHttpRequest::factory( $url, [], $fname );
4115
4116 $status = $req->execute(); // Status object
4117 if ( !$status->isOK() ) {
4118 $ttl = $cache::TTL_UNCACHEABLE;
4119 } elseif ( $req->getResponseHeader( 'X-Database-Lagged' ) !== null ) {
4120 $ttl = min( $cache::TTL_LAGGED, $ttl );
4121 }
4122
4123 return [
4124 'text' => $status->isOK() ? $req->getContent() : null,
4125 'code' => $req->getStatus()
4126 ];
4127 },
4128 [
4129 'checkKeys' => ( $wikiId !== false )
4130 ? [ $cache->makeGlobalKey( 'interwiki-page', $wikiId, $title->getDBkey() ) ]
4131 : [],
4132 'pcGroup' => 'interwiki-transclude:5',
4133 'pcTTL' => $cache::TTL_PROC_LONG
4134 ]
4135 );
4136
4137 if ( is_string( $data['text'] ) ) {
4138 $text = $data['text'];
4139 } elseif ( $data['code'] != 200 ) {
4140 // Though we failed to fetch the content, this status is useless.
4141 $text = wfMessage( 'scarytranscludefailed-httpstatus' )
4142 ->params( $url, $data['code'] )->inContentLanguage()->text();
4143 } else {
4144 $text = wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4145 }
4146
4147 return $text;
4148 }
4149
4150 /**
4151 * Triple brace replacement -- used for template arguments
4152 * @internal
4153 *
4154 * @param array $piece
4155 * @param PPFrame $frame
4156 *
4157 * @return array
4158 */
4159 public function argSubstitution( $piece, $frame ) {
4160 $error = false;
4161 $parts = $piece['parts'];
4162 $nameWithSpaces = $frame->expand( $piece['title'] );
4163 $argName = trim( $nameWithSpaces );
4164 $object = false;
4165 $text = $frame->getArgument( $argName );
4166 if ( $text === false && $parts->getLength() > 0
4167 && ( $this->ot['html']
4168 || $this->ot['pre']
4169 || ( $this->ot['wiki'] && $frame->isTemplate() )
4170 )
4171 ) {
4172 # No match in frame, use the supplied default
4173 $object = $parts->item( 0 )->getChildren();
4174 }
4175 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4176 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4177 $this->limitationWarn( 'post-expand-template-argument' );
4178 }
4179
4180 if ( $text === false && $object === false ) {
4181 # No match anywhere
4182 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4183 }
4184 if ( $error !== false ) {
4185 $text .= $error;
4186 }
4187 if ( $object !== false ) {
4188 $ret = [ 'object' => $object ];
4189 } else {
4190 $ret = [ 'text' => $text ];
4191 }
4192
4193 return $ret;
4194 }
4195
4196 /**
4197 * Return the text to be used for a given extension tag.
4198 * This is the ghost of strip().
4199 *
4200 * @param array $params Associative array of parameters:
4201 * name PPNode for the tag name
4202 * attr PPNode for unparsed text where tag attributes are thought to be
4203 * attributes Optional associative array of parsed attributes
4204 * inner Contents of extension element
4205 * noClose Original text did not have a close tag
4206 * @param PPFrame $frame
4207 *
4208 * @throws MWException
4209 * @return string
4210 * @internal
4211 */
4212 public function extensionSubstitution( $params, $frame ) {
4213 static $errorStr = '<span class="error">';
4214 static $errorLen = 20;
4215
4216 $name = $frame->expand( $params['name'] );
4217 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
4218 // Probably expansion depth or node count exceeded. Just punt the
4219 // error up.
4220 return $name;
4221 }
4222
4223 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
4224 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
4225 // See above
4226 return $attrText;
4227 }
4228
4229 // We can't safely check if the expansion for $content resulted in an
4230 // error, because the content could happen to be the error string
4231 // (T149622).
4232 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
4233
4234 $marker = self::MARKER_PREFIX . "-$name-"
4235 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
4236
4237 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4238 ( $this->ot['html'] || $this->ot['pre'] );
4239 if ( $isFunctionTag ) {
4240 $markerType = 'none';
4241 } else {
4242 $markerType = 'general';
4243 }
4244 if ( $this->ot['html'] || $isFunctionTag ) {
4245 $name = strtolower( $name );
4246 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4247 if ( isset( $params['attributes'] ) ) {
4248 $attributes += $params['attributes'];
4249 }
4250
4251 if ( isset( $this->mTagHooks[$name] ) ) {
4252 $output = call_user_func_array( $this->mTagHooks[$name],
4253 [ $content, $attributes, $this, $frame ] );
4254 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4255 list( $callback, ) = $this->mFunctionTagHooks[$name];
4256
4257 // Avoid PHP 7.1 warning from passing $this by reference
4258 $parser = $this;
4259 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
4260 } else {
4261 $output = '<span class="error">Invalid tag extension name: ' .
4262 htmlspecialchars( $name ) . '</span>';
4263 }
4264
4265 if ( is_array( $output ) ) {
4266 // Extract flags
4267 $flags = $output;
4268 $output = $flags[0];
4269 if ( isset( $flags['markerType'] ) ) {
4270 $markerType = $flags['markerType'];
4271 }
4272 }
4273 } else {
4274 if ( is_null( $attrText ) ) {
4275 $attrText = '';
4276 }
4277 if ( isset( $params['attributes'] ) ) {
4278 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4279 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4280 htmlspecialchars( $attrValue ) . '"';
4281 }
4282 }
4283 if ( $content === null ) {
4284 $output = "<$name$attrText/>";
4285 } else {
4286 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4287 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
4288 // See above
4289 return $close;
4290 }
4291 $output = "<$name$attrText>$content$close";
4292 }
4293 }
4294
4295 if ( $markerType === 'none' ) {
4296 return $output;
4297 } elseif ( $markerType === 'nowiki' ) {
4298 $this->mStripState->addNoWiki( $marker, $output );
4299 } elseif ( $markerType === 'general' ) {
4300 $this->mStripState->addGeneral( $marker, $output );
4301 } else {
4302 throw new MWException( __METHOD__ . ': invalid marker type' );
4303 }
4304 return $marker;
4305 }
4306
4307 /**
4308 * Increment an include size counter
4309 *
4310 * @param string $type The type of expansion
4311 * @param int $size The size of the text
4312 * @return bool False if this inclusion would take it over the maximum, true otherwise
4313 */
4314 public function incrementIncludeSize( $type, $size ) {
4315 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4316 return false;
4317 } else {
4318 $this->mIncludeSizes[$type] += $size;
4319 return true;
4320 }
4321 }
4322
4323 /**
4324 * Increment the expensive function count
4325 *
4326 * @return bool False if the limit has been exceeded
4327 */
4328 public function incrementExpensiveFunctionCount() {
4329 $this->mExpensiveFunctionCount++;
4330 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4331 }
4332
4333 /**
4334 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4335 * Fills $this->mDoubleUnderscores, returns the modified text
4336 *
4337 * @param string $text
4338 * @return string
4339 * @deprecated since 1.34; should not be used outside parser class.
4340 */
4341 public function doDoubleUnderscore( $text ) {
4342 wfDeprecated( __METHOD__, '1.34' );
4343 return $this->handleDoubleUnderscore( $text );
4344 }
4345
4346 /**
4347 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4348 * Fills $this->mDoubleUnderscores, returns the modified text
4349 *
4350 * @param string $text
4351 * @return string
4352 */
4353 private function handleDoubleUnderscore( $text ) {
4354 # The position of __TOC__ needs to be recorded
4355 $mw = $this->magicWordFactory->get( 'toc' );
4356 if ( $mw->match( $text ) ) {
4357 $this->mShowToc = true;
4358 $this->mForceTocPosition = true;
4359
4360 # Set a placeholder. At the end we'll fill it in with the TOC.
4361 $text = $mw->replace( '<!--MWTOC\'"-->', $text, 1 );
4362
4363 # Only keep the first one.
4364 $text = $mw->replace( '', $text );
4365 }
4366
4367 # Now match and remove the rest of them
4368 $mwa = $this->magicWordFactory->getDoubleUnderscoreArray();
4369 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4370
4371 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4372 $this->mOutput->mNoGallery = true;
4373 }
4374 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4375 $this->mShowToc = false;
4376 }
4377 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4378 && $this->mTitle->getNamespace() == NS_CATEGORY
4379 ) {
4380 $this->addTrackingCategory( 'hidden-category-category' );
4381 }
4382 # (T10068) Allow control over whether robots index a page.
4383 # __INDEX__ always overrides __NOINDEX__, see T16899
4384 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4385 $this->mOutput->setIndexPolicy( 'noindex' );
4386 $this->addTrackingCategory( 'noindex-category' );
4387 }
4388 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4389 $this->mOutput->setIndexPolicy( 'index' );
4390 $this->addTrackingCategory( 'index-category' );
4391 }
4392
4393 # Cache all double underscores in the database
4394 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4395 $this->mOutput->setProperty( $key, '' );
4396 }
4397
4398 return $text;
4399 }
4400
4401 /**
4402 * @see ParserOutput::addTrackingCategory()
4403 * @param string $msg Message key
4404 * @return bool Whether the addition was successful
4405 */
4406 public function addTrackingCategory( $msg ) {
4407 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4408 }
4409
4410 /**
4411 * This function accomplishes several tasks:
4412 * 1) Auto-number headings if that option is enabled
4413 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4414 * 3) Add a Table of contents on the top for users who have enabled the option
4415 * 4) Auto-anchor headings
4416 *
4417 * It loops through all headlines, collects the necessary data, then splits up the
4418 * string and re-inserts the newly formatted headlines.
4419 *
4420 * @param string $text
4421 * @param string $origText Original, untouched wikitext
4422 * @param bool $isMain
4423 * @return mixed|string
4424 * @private
4425 * @deprecated since 1.34; should not be used outside parser class.
4426 */
4427 public function formatHeadings( $text, $origText, $isMain = true ) {
4428 wfDeprecated( __METHOD__, '1.34' );
4429 return $this->finalizeHeadings( $text, $origText, $isMain );
4430 }
4431
4432 /**
4433 * This function accomplishes several tasks:
4434 * 1) Auto-number headings if that option is enabled
4435 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4436 * 3) Add a Table of contents on the top for users who have enabled the option
4437 * 4) Auto-anchor headings
4438 *
4439 * It loops through all headlines, collects the necessary data, then splits up the
4440 * string and re-inserts the newly formatted headlines.
4441 *
4442 * @param string $text
4443 * @param string $origText Original, untouched wikitext
4444 * @param bool $isMain
4445 * @return mixed|string
4446 */
4447 private function finalizeHeadings( $text, $origText, $isMain = true ) {
4448 # Inhibit editsection links if requested in the page
4449 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4450 $maybeShowEditLink = false;
4451 } else {
4452 $maybeShowEditLink = true; /* Actual presence will depend on post-cache transforms */
4453 }
4454
4455 # Get all headlines for numbering them and adding funky stuff like [edit]
4456 # links - this is for later, but we need the number of headlines right now
4457 # NOTE: white space in headings have been trimmed in handleHeadings. They shouldn't
4458 # be trimmed here since whitespace in HTML headings is significant.
4459 $matches = [];
4460 $numMatches = preg_match_all(
4461 '/<H(?P<level>[1-6])(?P<attrib>.*?>)(?P<header>[\s\S]*?)<\/H[1-6] *>/i',
4462 $text,
4463 $matches
4464 );
4465
4466 # if there are fewer than 4 headlines in the article, do not show TOC
4467 # unless it's been explicitly enabled.
4468 $enoughToc = $this->mShowToc &&
4469 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4470
4471 # Allow user to stipulate that a page should have a "new section"
4472 # link added via __NEWSECTIONLINK__
4473 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4474 $this->mOutput->setNewSection( true );
4475 }
4476
4477 # Allow user to remove the "new section"
4478 # link via __NONEWSECTIONLINK__
4479 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4480 $this->mOutput->hideNewSection( true );
4481 }
4482
4483 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4484 # override above conditions and always show TOC above first header
4485 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4486 $this->mShowToc = true;
4487 $enoughToc = true;
4488 }
4489
4490 # headline counter
4491 $headlineCount = 0;
4492 $numVisible = 0;
4493
4494 # Ugh .. the TOC should have neat indentation levels which can be
4495 # passed to the skin functions. These are determined here
4496 $toc = '';
4497 $full = '';
4498 $head = [];
4499 $sublevelCount = [];
4500 $levelCount = [];
4501 $level = 0;
4502 $prevlevel = 0;
4503 $toclevel = 0;
4504 $prevtoclevel = 0;
4505 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4506 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4507 $oldType = $this->mOutputType;
4508 $this->setOutputType( self::OT_WIKI );
4509 $frame = $this->getPreprocessor()->newFrame();
4510 $root = $this->preprocessToDom( $origText );
4511 $node = $root->getFirstChild();
4512 $byteOffset = 0;
4513 $tocraw = [];
4514 $refers = [];
4515
4516 $headlines = $numMatches !== false ? $matches[3] : [];
4517
4518 $maxTocLevel = $this->svcOptions->get( 'MaxTocLevel' );
4519 foreach ( $headlines as $headline ) {
4520 $isTemplate = false;
4521 $titleText = false;
4522 $sectionIndex = false;
4523 $numbering = '';
4524 $markerMatches = [];
4525 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4526 $serial = $markerMatches[1];
4527 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4528 $isTemplate = ( $titleText != $baseTitleText );
4529 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4530 }
4531
4532 if ( $toclevel ) {
4533 $prevlevel = $level;
4534 }
4535 $level = $matches[1][$headlineCount];
4536
4537 if ( $level > $prevlevel ) {
4538 # Increase TOC level
4539 $toclevel++;
4540 $sublevelCount[$toclevel] = 0;
4541 if ( $toclevel < $maxTocLevel ) {
4542 $prevtoclevel = $toclevel;
4543 $toc .= Linker::tocIndent();
4544 $numVisible++;
4545 }
4546 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4547 # Decrease TOC level, find level to jump to
4548
4549 for ( $i = $toclevel; $i > 0; $i-- ) {
4550 if ( $levelCount[$i] == $level ) {
4551 # Found last matching level
4552 $toclevel = $i;
4553 break;
4554 } elseif ( $levelCount[$i] < $level ) {
4555 # Found first matching level below current level
4556 $toclevel = $i + 1;
4557 break;
4558 }
4559 }
4560 if ( $i == 0 ) {
4561 $toclevel = 1;
4562 }
4563 if ( $toclevel < $maxTocLevel ) {
4564 if ( $prevtoclevel < $maxTocLevel ) {
4565 # Unindent only if the previous toc level was shown :p
4566 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4567 $prevtoclevel = $toclevel;
4568 } else {
4569 $toc .= Linker::tocLineEnd();
4570 }
4571 }
4572 } else {
4573 # No change in level, end TOC line
4574 if ( $toclevel < $maxTocLevel ) {
4575 $toc .= Linker::tocLineEnd();
4576 }
4577 }
4578
4579 $levelCount[$toclevel] = $level;
4580
4581 # count number of headlines for each level
4582 $sublevelCount[$toclevel]++;
4583 $dot = 0;
4584 for ( $i = 1; $i <= $toclevel; $i++ ) {
4585 if ( !empty( $sublevelCount[$i] ) ) {
4586 if ( $dot ) {
4587 $numbering .= '.';
4588 }
4589 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4590 $dot = 1;
4591 }
4592 }
4593
4594 # The safe header is a version of the header text safe to use for links
4595
4596 # Remove link placeholders by the link text.
4597 # <!--LINK number-->
4598 # turns into
4599 # link text with suffix
4600 # Do this before unstrip since link text can contain strip markers
4601 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4602
4603 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4604 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4605
4606 # Remove any <style> or <script> tags (T198618)
4607 $safeHeadline = preg_replace(
4608 '#<(style|script)(?: [^>]*[^>/])?>.*?</\1>#is',
4609 '',
4610 $safeHeadline
4611 );
4612
4613 # Strip out HTML (first regex removes any tag not allowed)
4614 # Allowed tags are:
4615 # * <sup> and <sub> (T10393)
4616 # * <i> (T28375)
4617 # * <b> (r105284)
4618 # * <bdi> (T74884)
4619 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4620 # * <s> and <strike> (T35715)
4621 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4622 # to allow setting directionality in toc items.
4623 $tocline = preg_replace(
4624 [
4625 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4626 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4627 ],
4628 [ '', '<$1>' ],
4629 $safeHeadline
4630 );
4631
4632 # Strip '<span></span>', which is the result from the above if
4633 # <span id="foo"></span> is used to produce an additional anchor
4634 # for a section.
4635 $tocline = str_replace( '<span></span>', '', $tocline );
4636
4637 $tocline = trim( $tocline );
4638
4639 # For the anchor, strip out HTML-y stuff period
4640 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4641 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4642
4643 # Save headline for section edit hint before it's escaped
4644 $headlineHint = $safeHeadline;
4645
4646 # Decode HTML entities
4647 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4648
4649 $safeHeadline = self::normalizeSectionName( $safeHeadline );
4650
4651 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4652 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4653 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4654 if ( $fallbackHeadline === $safeHeadline ) {
4655 # No reason to have both (in fact, we can't)
4656 $fallbackHeadline = false;
4657 }
4658
4659 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4660 # @todo FIXME: We may be changing them depending on the current locale.
4661 $arrayKey = strtolower( $safeHeadline );
4662 if ( $fallbackHeadline === false ) {
4663 $fallbackArrayKey = false;
4664 } else {
4665 $fallbackArrayKey = strtolower( $fallbackHeadline );
4666 }
4667
4668 # Create the anchor for linking from the TOC to the section
4669 $anchor = $safeHeadline;
4670 $fallbackAnchor = $fallbackHeadline;
4671 if ( isset( $refers[$arrayKey] ) ) {
4672 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4673 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4674 $anchor .= "_$i";
4675 $linkAnchor .= "_$i";
4676 $refers["${arrayKey}_$i"] = true;
4677 } else {
4678 $refers[$arrayKey] = true;
4679 }
4680 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4681 // phpcs:ignore Generic.Formatting.DisallowMultipleStatements
4682 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4683 $fallbackAnchor .= "_$i";
4684 $refers["${fallbackArrayKey}_$i"] = true;
4685 } else {
4686 $refers[$fallbackArrayKey] = true;
4687 }
4688
4689 # Don't number the heading if it is the only one (looks silly)
4690 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4691 # the two are different if the line contains a link
4692 $headline = Html::element(
4693 'span',
4694 [ 'class' => 'mw-headline-number' ],
4695 $numbering
4696 ) . ' ' . $headline;
4697 }
4698
4699 if ( $enoughToc && ( !isset( $maxTocLevel ) || $toclevel < $maxTocLevel ) ) {
4700 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4701 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4702 }
4703
4704 # Add the section to the section tree
4705 # Find the DOM node for this header
4706 $noOffset = ( $isTemplate || $sectionIndex === false );
4707 while ( $node && !$noOffset ) {
4708 if ( $node->getName() === 'h' ) {
4709 $bits = $node->splitHeading();
4710 if ( $bits['i'] == $sectionIndex ) {
4711 break;
4712 }
4713 }
4714 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4715 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4716 $node = $node->getNextSibling();
4717 }
4718 $tocraw[] = [
4719 'toclevel' => $toclevel,
4720 'level' => $level,
4721 'line' => $tocline,
4722 'number' => $numbering,
4723 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4724 'fromtitle' => $titleText,
4725 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4726 'anchor' => $anchor,
4727 ];
4728
4729 # give headline the correct <h#> tag
4730 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4731 // Output edit section links as markers with styles that can be customized by skins
4732 if ( $isTemplate ) {
4733 # Put a T flag in the section identifier, to indicate to extractSections()
4734 # that sections inside <includeonly> should be counted.
4735 $editsectionPage = $titleText;
4736 $editsectionSection = "T-$sectionIndex";
4737 $editsectionContent = null;
4738 } else {
4739 $editsectionPage = $this->mTitle->getPrefixedText();
4740 $editsectionSection = $sectionIndex;
4741 $editsectionContent = $headlineHint;
4742 }
4743 // We use a bit of pesudo-xml for editsection markers. The
4744 // language converter is run later on. Using a UNIQ style marker
4745 // leads to the converter screwing up the tokens when it
4746 // converts stuff. And trying to insert strip tags fails too. At
4747 // this point all real inputted tags have already been escaped,
4748 // so we don't have to worry about a user trying to input one of
4749 // these markers directly. We use a page and section attribute
4750 // to stop the language converter from converting these
4751 // important bits of data, but put the headline hint inside a
4752 // content block because the language converter is supposed to
4753 // be able to convert that piece of data.
4754 // Gets replaced with html in ParserOutput::getText
4755 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4756 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4757 if ( $editsectionContent !== null ) {
4758 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4759 } else {
4760 $editlink .= '/>';
4761 }
4762 } else {
4763 $editlink = '';
4764 }
4765 $head[$headlineCount] = Linker::makeHeadline( $level,
4766 $matches['attrib'][$headlineCount], $anchor, $headline,
4767 $editlink, $fallbackAnchor );
4768
4769 $headlineCount++;
4770 }
4771
4772 $this->setOutputType( $oldType );
4773
4774 # Never ever show TOC if no headers
4775 if ( $numVisible < 1 ) {
4776 $enoughToc = false;
4777 }
4778
4779 if ( $enoughToc ) {
4780 if ( $prevtoclevel > 0 && $prevtoclevel < $maxTocLevel ) {
4781 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4782 }
4783 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4784 $this->mOutput->setTOCHTML( $toc );
4785 $toc = self::TOC_START . $toc . self::TOC_END;
4786 }
4787
4788 if ( $isMain ) {
4789 $this->mOutput->setSections( $tocraw );
4790 }
4791
4792 # split up and insert constructed headlines
4793 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4794 $i = 0;
4795
4796 // build an array of document sections
4797 $sections = [];
4798 foreach ( $blocks as $block ) {
4799 // $head is zero-based, sections aren't.
4800 if ( empty( $head[$i - 1] ) ) {
4801 $sections[$i] = $block;
4802 } else {
4803 $sections[$i] = $head[$i - 1] . $block;
4804 }
4805
4806 /**
4807 * Send a hook, one per section.
4808 * The idea here is to be able to make section-level DIVs, but to do so in a
4809 * lower-impact, more correct way than r50769
4810 *
4811 * $this : caller
4812 * $section : the section number
4813 * &$sectionContent : ref to the content of the section
4814 * $maybeShowEditLinks : boolean describing whether this section has an edit link
4815 */
4816 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $maybeShowEditLink ] );
4817
4818 $i++;
4819 }
4820
4821 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4822 // append the TOC at the beginning
4823 // Top anchor now in skin
4824 $sections[0] .= $toc . "\n";
4825 }
4826
4827 $full .= implode( '', $sections );
4828
4829 if ( $this->mForceTocPosition ) {
4830 return str_replace( '<!--MWTOC\'"-->', $toc, $full );
4831 } else {
4832 return $full;
4833 }
4834 }
4835
4836 /**
4837 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4838 * conversion, substituting signatures, {{subst:}} templates, etc.
4839 *
4840 * @param string $text The text to transform
4841 * @param Title $title The Title object for the current article
4842 * @param User $user The User object describing the current user
4843 * @param ParserOptions $options Parsing options
4844 * @param bool $clearState Whether to clear the parser state first
4845 * @return string The altered wiki markup
4846 */
4847 public function preSaveTransform( $text, Title $title, User $user,
4848 ParserOptions $options, $clearState = true
4849 ) {
4850 if ( $clearState ) {
4851 $magicScopeVariable = $this->lock();
4852 }
4853 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4854 $this->setUser( $user );
4855
4856 // Strip U+0000 NULL (T159174)
4857 $text = str_replace( "\000", '', $text );
4858
4859 // We still normalize line endings for backwards-compatibility
4860 // with other code that just calls PST, but this should already
4861 // be handled in TextContent subclasses
4862 $text = TextContent::normalizeLineEndings( $text );
4863
4864 if ( $options->getPreSaveTransform() ) {
4865 $text = $this->pstPass2( $text, $user );
4866 }
4867 $text = $this->mStripState->unstripBoth( $text );
4868
4869 $this->setUser( null ); # Reset
4870
4871 return $text;
4872 }
4873
4874 /**
4875 * Pre-save transform helper function
4876 *
4877 * @param string $text
4878 * @param User $user
4879 *
4880 * @return string
4881 */
4882 private function pstPass2( $text, $user ) {
4883 # Note: This is the timestamp saved as hardcoded wikitext to the database, we use
4884 # $this->contLang here in order to give everyone the same signature and use the default one
4885 # rather than the one selected in each user's preferences. (see also T14815)
4886 $ts = $this->mOptions->getTimestamp();
4887 $timestamp = MWTimestamp::getLocalInstance( $ts );
4888 $ts = $timestamp->format( 'YmdHis' );
4889 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4890
4891 $d = $this->contLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4892
4893 # Variable replacement
4894 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4895 $text = $this->replaceVariables( $text );
4896
4897 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4898 # which may corrupt this parser instance via its wfMessage()->text() call-
4899
4900 # Signatures
4901 if ( strpos( $text, '~~~' ) !== false ) {
4902 $sigText = $this->getUserSig( $user );
4903 $text = strtr( $text, [
4904 '~~~~~' => $d,
4905 '~~~~' => "$sigText $d",
4906 '~~~' => $sigText
4907 ] );
4908 # The main two signature forms used above are time-sensitive
4909 $this->setOutputFlag( 'user-signature', 'User signature detected' );
4910 }
4911
4912 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4913 $tc = '[' . Title::legalChars() . ']';
4914 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4915
4916 // [[ns:page (context)|]]
4917 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4918 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4919 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4920 // [[ns:page (context), context|]] (using either single or double-width comma)
4921 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4922 // [[|page]] (reverse pipe trick: add context from page title)
4923 $p2 = "/\[\[\\|($tc+)]]/";
4924
4925 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4926 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4927 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4928 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4929
4930 $t = $this->mTitle->getText();
4931 $m = [];
4932 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4933 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4934 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4935 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4936 } else {
4937 # if there's no context, don't bother duplicating the title
4938 $text = preg_replace( $p2, '[[\\1]]', $text );
4939 }
4940
4941 return $text;
4942 }
4943
4944 /**
4945 * Fetch the user's signature text, if any, and normalize to
4946 * validated, ready-to-insert wikitext.
4947 * If you have pre-fetched the nickname or the fancySig option, you can
4948 * specify them here to save a database query.
4949 * Do not reuse this parser instance after calling getUserSig(),
4950 * as it may have changed.
4951 *
4952 * @param User &$user
4953 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4954 * @param bool|null $fancySig whether the nicknname is the complete signature
4955 * or null to use default value
4956 * @return string
4957 */
4958 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4959 $username = $user->getName();
4960
4961 # If not given, retrieve from the user object.
4962 if ( $nickname === false ) {
4963 $nickname = $user->getOption( 'nickname' );
4964 }
4965
4966 if ( is_null( $fancySig ) ) {
4967 $fancySig = $user->getBoolOption( 'fancysig' );
4968 }
4969
4970 $nickname = $nickname == null ? $username : $nickname;
4971
4972 if ( mb_strlen( $nickname ) > $this->svcOptions->get( 'MaxSigChars' ) ) {
4973 $nickname = $username;
4974 $this->logger->debug( __METHOD__ . ": $username has overlong signature." );
4975 } elseif ( $fancySig !== false ) {
4976 # Sig. might contain markup; validate this
4977 if ( $this->validateSig( $nickname ) !== false ) {
4978 # Validated; clean up (if needed) and return it
4979 return $this->cleanSig( $nickname, true );
4980 } else {
4981 # Failed to validate; fall back to the default
4982 $nickname = $username;
4983 $this->logger->debug( __METHOD__ . ": $username has bad XML tags in signature." );
4984 }
4985 }
4986
4987 # Make sure nickname doesnt get a sig in a sig
4988 $nickname = self::cleanSigInSig( $nickname );
4989
4990 # If we're still here, make it a link to the user page
4991 $userText = wfEscapeWikiText( $username );
4992 $nickText = wfEscapeWikiText( $nickname );
4993 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4994
4995 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4996 ->title( $this->getTitle() )->text();
4997 }
4998
4999 /**
5000 * Check that the user's signature contains no bad XML
5001 *
5002 * @param string $text
5003 * @return string|bool An expanded string, or false if invalid.
5004 */
5005 public function validateSig( $text ) {
5006 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
5007 }
5008
5009 /**
5010 * Clean up signature text
5011 *
5012 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
5013 * 2) Substitute all transclusions
5014 *
5015 * @param string $text
5016 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
5017 * @return string Signature text
5018 */
5019 public function cleanSig( $text, $parsing = false ) {
5020 if ( !$parsing ) {
5021 global $wgTitle;
5022 $magicScopeVariable = $this->lock();
5023 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
5024 }
5025
5026 # Option to disable this feature
5027 if ( !$this->mOptions->getCleanSignatures() ) {
5028 return $text;
5029 }
5030
5031 # @todo FIXME: Regex doesn't respect extension tags or nowiki
5032 # => Move this logic to braceSubstitution()
5033 $substWord = $this->magicWordFactory->get( 'subst' );
5034 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
5035 $substText = '{{' . $substWord->getSynonym( 0 );
5036
5037 $text = preg_replace( $substRegex, $substText, $text );
5038 $text = self::cleanSigInSig( $text );
5039 $dom = $this->preprocessToDom( $text );
5040 $frame = $this->getPreprocessor()->newFrame();
5041 $text = $frame->expand( $dom );
5042
5043 if ( !$parsing ) {
5044 $text = $this->mStripState->unstripBoth( $text );
5045 }
5046
5047 return $text;
5048 }
5049
5050 /**
5051 * Strip 3, 4 or 5 tildes out of signatures.
5052 *
5053 * @param string $text
5054 * @return string Signature text with /~{3,5}/ removed
5055 */
5056 public static function cleanSigInSig( $text ) {
5057 $text = preg_replace( '/~{3,5}/', '', $text );
5058 return $text;
5059 }
5060
5061 /**
5062 * Set up some variables which are usually set up in parse()
5063 * so that an external function can call some class members with confidence
5064 *
5065 * @param Title|null $title
5066 * @param ParserOptions $options
5067 * @param int $outputType
5068 * @param bool $clearState
5069 * @param int|null $revId
5070 */
5071 public function startExternalParse( Title $title = null, ParserOptions $options,
5072 $outputType, $clearState = true, $revId = null
5073 ) {
5074 $this->startParse( $title, $options, $outputType, $clearState );
5075 if ( $revId !== null ) {
5076 $this->mRevisionId = $revId;
5077 }
5078 }
5079
5080 /**
5081 * @param Title|null $title
5082 * @param ParserOptions $options
5083 * @param int $outputType
5084 * @param bool $clearState
5085 */
5086 private function startParse( Title $title = null, ParserOptions $options,
5087 $outputType, $clearState = true
5088 ) {
5089 $this->setTitle( $title );
5090 $this->mOptions = $options;
5091 $this->setOutputType( $outputType );
5092 if ( $clearState ) {
5093 $this->clearState();
5094 }
5095 }
5096
5097 /**
5098 * Wrapper for preprocess()
5099 *
5100 * @param string $text The text to preprocess
5101 * @param ParserOptions $options
5102 * @param Title|null $title Title object or null to use $wgTitle
5103 * @return string
5104 */
5105 public function transformMsg( $text, $options, $title = null ) {
5106 static $executing = false;
5107
5108 # Guard against infinite recursion
5109 if ( $executing ) {
5110 return $text;
5111 }
5112 $executing = true;
5113
5114 if ( !$title ) {
5115 global $wgTitle;
5116 $title = $wgTitle;
5117 }
5118
5119 $text = $this->preprocess( $text, $title, $options );
5120
5121 $executing = false;
5122 return $text;
5123 }
5124
5125 /**
5126 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
5127 * The callback should have the following form:
5128 * function myParserHook( $text, $params, $parser, $frame ) { ... }
5129 *
5130 * Transform and return $text. Use $parser for any required context, e.g. use
5131 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
5132 *
5133 * Hooks may return extended information by returning an array, of which the
5134 * first numbered element (index 0) must be the return string, and all other
5135 * entries are extracted into local variables within an internal function
5136 * in the Parser class.
5137 *
5138 * This interface (introduced r61913) appears to be undocumented, but
5139 * 'markerType' is used by some core tag hooks to override which strip
5140 * array their results are placed in. **Use great caution if attempting
5141 * this interface, as it is not documented and injudicious use could smash
5142 * private variables.**
5143 *
5144 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5145 * @param callable $callback The callback function (and object) to use for the tag
5146 * @throws MWException
5147 * @return callable|null The old value of the mTagHooks array associated with the hook
5148 */
5149 public function setHook( $tag, callable $callback ) {
5150 $tag = strtolower( $tag );
5151 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5152 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
5153 }
5154 $oldVal = $this->mTagHooks[$tag] ?? null;
5155 $this->mTagHooks[$tag] = $callback;
5156 if ( !in_array( $tag, $this->mStripList ) ) {
5157 $this->mStripList[] = $tag;
5158 }
5159
5160 return $oldVal;
5161 }
5162
5163 /**
5164 * As setHook(), but letting the contents be parsed.
5165 *
5166 * Transparent tag hooks are like regular XML-style tag hooks, except they
5167 * operate late in the transformation sequence, on HTML instead of wikitext.
5168 *
5169 * This is probably obsoleted by things dealing with parser frames?
5170 * The only extension currently using it is geoserver.
5171 *
5172 * @since 1.10
5173 * @todo better document or deprecate this
5174 *
5175 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5176 * @param callable $callback The callback function (and object) to use for the tag
5177 * @throws MWException
5178 * @return callable|null The old value of the mTagHooks array associated with the hook
5179 */
5180 public function setTransparentTagHook( $tag, callable $callback ) {
5181 $tag = strtolower( $tag );
5182 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5183 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5184 }
5185 $oldVal = $this->mTransparentTagHooks[$tag] ?? null;
5186 $this->mTransparentTagHooks[$tag] = $callback;
5187
5188 return $oldVal;
5189 }
5190
5191 /**
5192 * Remove all tag hooks
5193 */
5194 public function clearTagHooks() {
5195 $this->mTagHooks = [];
5196 $this->mFunctionTagHooks = [];
5197 $this->mStripList = $this->mDefaultStripList;
5198 }
5199
5200 /**
5201 * Create a function, e.g. {{sum:1|2|3}}
5202 * The callback function should have the form:
5203 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5204 *
5205 * Or with Parser::SFH_OBJECT_ARGS:
5206 * function myParserFunction( $parser, $frame, $args ) { ... }
5207 *
5208 * The callback may either return the text result of the function, or an array with the text
5209 * in element 0, and a number of flags in the other elements. The names of the flags are
5210 * specified in the keys. Valid flags are:
5211 * found The text returned is valid, stop processing the template. This
5212 * is on by default.
5213 * nowiki Wiki markup in the return value should be escaped
5214 * isHTML The returned text is HTML, armour it against wikitext transformation
5215 *
5216 * @param string $id The magic word ID
5217 * @param callable $callback The callback function (and object) to use
5218 * @param int $flags A combination of the following flags:
5219 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5220 *
5221 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5222 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5223 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5224 * the arguments, and to control the way they are expanded.
5225 *
5226 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5227 * arguments, for instance:
5228 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5229 *
5230 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5231 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5232 * working if/when this is changed.
5233 *
5234 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5235 * expansion.
5236 *
5237 * Please read the documentation in includes/parser/Preprocessor.php for more information
5238 * about the methods available in PPFrame and PPNode.
5239 *
5240 * @throws MWException
5241 * @return string|callable The old callback function for this name, if any
5242 */
5243 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
5244 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
5245 $this->mFunctionHooks[$id] = [ $callback, $flags ];
5246
5247 # Add to function cache
5248 $mw = $this->magicWordFactory->get( $id );
5249 if ( !$mw ) {
5250 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
5251 }
5252
5253 $synonyms = $mw->getSynonyms();
5254 $sensitive = intval( $mw->isCaseSensitive() );
5255
5256 foreach ( $synonyms as $syn ) {
5257 # Case
5258 if ( !$sensitive ) {
5259 $syn = $this->contLang->lc( $syn );
5260 }
5261 # Add leading hash
5262 if ( !( $flags & self::SFH_NO_HASH ) ) {
5263 $syn = '#' . $syn;
5264 }
5265 # Remove trailing colon
5266 if ( substr( $syn, -1, 1 ) === ':' ) {
5267 $syn = substr( $syn, 0, -1 );
5268 }
5269 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5270 }
5271 return $oldVal;
5272 }
5273
5274 /**
5275 * Get all registered function hook identifiers
5276 *
5277 * @return array
5278 */
5279 public function getFunctionHooks() {
5280 $this->firstCallInit();
5281 return array_keys( $this->mFunctionHooks );
5282 }
5283
5284 /**
5285 * Create a tag function, e.g. "<test>some stuff</test>".
5286 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5287 * Unlike parser functions, their content is not preprocessed.
5288 * @param string $tag
5289 * @param callable $callback
5290 * @param int $flags
5291 * @throws MWException
5292 * @return null
5293 */
5294 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
5295 $tag = strtolower( $tag );
5296 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5297 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5298 }
5299 $old = $this->mFunctionTagHooks[$tag] ?? null;
5300 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
5301
5302 if ( !in_array( $tag, $this->mStripList ) ) {
5303 $this->mStripList[] = $tag;
5304 }
5305
5306 return $old;
5307 }
5308
5309 /**
5310 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5311 * Placeholders created in Linker::link()
5312 *
5313 * @param string &$text
5314 * @param int $options
5315 */
5316 public function replaceLinkHolders( &$text, $options = 0 ) {
5317 $this->mLinkHolders->replace( $text );
5318 }
5319
5320 /**
5321 * Replace "<!--LINK-->" link placeholders with plain text of links
5322 * (not HTML-formatted).
5323 *
5324 * @param string $text
5325 * @return string
5326 */
5327 public function replaceLinkHoldersText( $text ) {
5328 return $this->mLinkHolders->replaceText( $text );
5329 }
5330
5331 /**
5332 * Renders an image gallery from a text with one line per image.
5333 * text labels may be given by using |-style alternative text. E.g.
5334 * Image:one.jpg|The number "1"
5335 * Image:tree.jpg|A tree
5336 * given as text will return the HTML of a gallery with two images,
5337 * labeled 'The number "1"' and
5338 * 'A tree'.
5339 *
5340 * @param string $text
5341 * @param array $params
5342 * @return string HTML
5343 */
5344 public function renderImageGallery( $text, $params ) {
5345 $mode = false;
5346 if ( isset( $params['mode'] ) ) {
5347 $mode = $params['mode'];
5348 }
5349
5350 try {
5351 $ig = ImageGalleryBase::factory( $mode );
5352 } catch ( Exception $e ) {
5353 // If invalid type set, fallback to default.
5354 $ig = ImageGalleryBase::factory( false );
5355 }
5356
5357 $ig->setContextTitle( $this->mTitle );
5358 $ig->setShowBytes( false );
5359 $ig->setShowDimensions( false );
5360 $ig->setShowFilename( false );
5361 $ig->setParser( $this );
5362 $ig->setHideBadImages();
5363 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
5364
5365 if ( isset( $params['showfilename'] ) ) {
5366 $ig->setShowFilename( true );
5367 } else {
5368 $ig->setShowFilename( false );
5369 }
5370 if ( isset( $params['caption'] ) ) {
5371 // NOTE: We aren't passing a frame here or below. Frame info
5372 // is currently opaque to Parsoid, which acts on OT_PREPROCESS.
5373 // See T107332#4030581
5374 $caption = $this->recursiveTagParse( $params['caption'] );
5375 $ig->setCaptionHtml( $caption );
5376 }
5377 if ( isset( $params['perrow'] ) ) {
5378 $ig->setPerRow( $params['perrow'] );
5379 }
5380 if ( isset( $params['widths'] ) ) {
5381 $ig->setWidths( $params['widths'] );
5382 }
5383 if ( isset( $params['heights'] ) ) {
5384 $ig->setHeights( $params['heights'] );
5385 }
5386 $ig->setAdditionalOptions( $params );
5387
5388 // Avoid PHP 7.1 warning from passing $this by reference
5389 $parser = $this;
5390 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5391
5392 $lines = StringUtils::explode( "\n", $text );
5393 foreach ( $lines as $line ) {
5394 # match lines like these:
5395 # Image:someimage.jpg|This is some image
5396 $matches = [];
5397 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5398 # Skip empty lines
5399 if ( count( $matches ) == 0 ) {
5400 continue;
5401 }
5402
5403 if ( strpos( $matches[0], '%' ) !== false ) {
5404 $matches[1] = rawurldecode( $matches[1] );
5405 }
5406 $title = Title::newFromText( $matches[1], NS_FILE );
5407 if ( is_null( $title ) ) {
5408 # Bogus title. Ignore these so we don't bomb out later.
5409 continue;
5410 }
5411
5412 # We need to get what handler the file uses, to figure out parameters.
5413 # Note, a hook can overide the file name, and chose an entirely different
5414 # file (which potentially could be of a different type and have different handler).
5415 $options = [];
5416 $descQuery = false;
5417 Hooks::run( 'BeforeParserFetchFileAndTitle',
5418 [ $this, $title, &$options, &$descQuery ] );
5419 # Don't register it now, as TraditionalImageGallery does that later.
5420 $file = $this->fetchFileNoRegister( $title, $options );
5421 $handler = $file ? $file->getHandler() : false;
5422
5423 $paramMap = [
5424 'img_alt' => 'gallery-internal-alt',
5425 'img_link' => 'gallery-internal-link',
5426 ];
5427 if ( $handler ) {
5428 $paramMap += $handler->getParamMap();
5429 // We don't want people to specify per-image widths.
5430 // Additionally the width parameter would need special casing anyhow.
5431 unset( $paramMap['img_width'] );
5432 }
5433
5434 $mwArray = $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5435
5436 $label = '';
5437 $alt = '';
5438 $link = '';
5439 $handlerOptions = [];
5440 if ( isset( $matches[3] ) ) {
5441 // look for an |alt= definition while trying not to break existing
5442 // captions with multiple pipes (|) in it, until a more sensible grammar
5443 // is defined for images in galleries
5444
5445 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5446 // splitting on '|' is a bit odd, and different from makeImage.
5447 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5448 // Protect LanguageConverter markup
5449 $parameterMatches = StringUtils::delimiterExplode(
5450 '-{', '}-', '|', $matches[3], true /* nested */
5451 );
5452
5453 foreach ( $parameterMatches as $parameterMatch ) {
5454 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5455 if ( $magicName ) {
5456 $paramName = $paramMap[$magicName];
5457
5458 switch ( $paramName ) {
5459 case 'gallery-internal-alt':
5460 $alt = $this->stripAltText( $match, false );
5461 break;
5462 case 'gallery-internal-link':
5463 $linkValue = $this->stripAltText( $match, false );
5464 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5465 // Result of LanguageConverter::markNoConversion
5466 // invoked on an external link.
5467 $linkValue = substr( $linkValue, 4, -2 );
5468 }
5469 list( $type, $target ) = $this->parseLinkParameter( $linkValue );
5470 if ( $type === 'link-url' ) {
5471 $link = $target;
5472 $this->mOutput->addExternalLink( $target );
5473 } elseif ( $type === 'link-title' ) {
5474 $link = $target->getLinkURL();
5475 $this->mOutput->addLink( $target );
5476 }
5477 break;
5478 default:
5479 // Must be a handler specific parameter.
5480 if ( $handler->validateParam( $paramName, $match ) ) {
5481 $handlerOptions[$paramName] = $match;
5482 } else {
5483 // Guess not, consider it as caption.
5484 $this->logger->debug(
5485 "$parameterMatch failed parameter validation" );
5486 $label = $parameterMatch;
5487 }
5488 }
5489
5490 } else {
5491 // Last pipe wins.
5492 $label = $parameterMatch;
5493 }
5494 }
5495 }
5496
5497 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5498 }
5499 $html = $ig->toHTML();
5500 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5501 return $html;
5502 }
5503
5504 /**
5505 * @param MediaHandler $handler
5506 * @return array
5507 */
5508 public function getImageParams( $handler ) {
5509 if ( $handler ) {
5510 $handlerClass = get_class( $handler );
5511 } else {
5512 $handlerClass = '';
5513 }
5514 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5515 # Initialise static lists
5516 static $internalParamNames = [
5517 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5518 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5519 'bottom', 'text-bottom' ],
5520 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5521 'upright', 'border', 'link', 'alt', 'class' ],
5522 ];
5523 static $internalParamMap;
5524 if ( !$internalParamMap ) {
5525 $internalParamMap = [];
5526 foreach ( $internalParamNames as $type => $names ) {
5527 foreach ( $names as $name ) {
5528 // For grep: img_left, img_right, img_center, img_none,
5529 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5530 // img_bottom, img_text_bottom,
5531 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5532 // img_border, img_link, img_alt, img_class
5533 $magicName = str_replace( '-', '_', "img_$name" );
5534 $internalParamMap[$magicName] = [ $type, $name ];
5535 }
5536 }
5537 }
5538
5539 # Add handler params
5540 $paramMap = $internalParamMap;
5541 if ( $handler ) {
5542 $handlerParamMap = $handler->getParamMap();
5543 foreach ( $handlerParamMap as $magic => $paramName ) {
5544 $paramMap[$magic] = [ 'handler', $paramName ];
5545 }
5546 }
5547 $this->mImageParams[$handlerClass] = $paramMap;
5548 $this->mImageParamsMagicArray[$handlerClass] =
5549 $this->magicWordFactory->newArray( array_keys( $paramMap ) );
5550 }
5551 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5552 }
5553
5554 /**
5555 * Parse image options text and use it to make an image
5556 *
5557 * @param Title $title
5558 * @param string $options
5559 * @param LinkHolderArray|bool $holders
5560 * @return string HTML
5561 */
5562 public function makeImage( $title, $options, $holders = false ) {
5563 # Check if the options text is of the form "options|alt text"
5564 # Options are:
5565 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5566 # * left no resizing, just left align. label is used for alt= only
5567 # * right same, but right aligned
5568 # * none same, but not aligned
5569 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5570 # * center center the image
5571 # * frame Keep original image size, no magnify-button.
5572 # * framed Same as "frame"
5573 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5574 # * upright reduce width for upright images, rounded to full __0 px
5575 # * border draw a 1px border around the image
5576 # * alt Text for HTML alt attribute (defaults to empty)
5577 # * class Set a class for img node
5578 # * link Set the target of the image link. Can be external, interwiki, or local
5579 # vertical-align values (no % or length right now):
5580 # * baseline
5581 # * sub
5582 # * super
5583 # * top
5584 # * text-top
5585 # * middle
5586 # * bottom
5587 # * text-bottom
5588
5589 # Protect LanguageConverter markup when splitting into parts
5590 $parts = StringUtils::delimiterExplode(
5591 '-{', '}-', '|', $options, true /* allow nesting */
5592 );
5593
5594 # Give extensions a chance to select the file revision for us
5595 $options = [];
5596 $descQuery = false;
5597 Hooks::run( 'BeforeParserFetchFileAndTitle',
5598 [ $this, $title, &$options, &$descQuery ] );
5599 # Fetch and register the file (file title may be different via hooks)
5600 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5601
5602 # Get parameter map
5603 $handler = $file ? $file->getHandler() : false;
5604
5605 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5606
5607 if ( !$file ) {
5608 $this->addTrackingCategory( 'broken-file-category' );
5609 }
5610
5611 # Process the input parameters
5612 $caption = '';
5613 $params = [ 'frame' => [], 'handler' => [],
5614 'horizAlign' => [], 'vertAlign' => [] ];
5615 $seenformat = false;
5616 foreach ( $parts as $part ) {
5617 $part = trim( $part );
5618 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5619 $validated = false;
5620 if ( isset( $paramMap[$magicName] ) ) {
5621 list( $type, $paramName ) = $paramMap[$magicName];
5622
5623 # Special case; width and height come in one variable together
5624 if ( $type === 'handler' && $paramName === 'width' ) {
5625 $parsedWidthParam = self::parseWidthParam( $value );
5626 if ( isset( $parsedWidthParam['width'] ) ) {
5627 $width = $parsedWidthParam['width'];
5628 if ( $handler->validateParam( 'width', $width ) ) {
5629 $params[$type]['width'] = $width;
5630 $validated = true;
5631 }
5632 }
5633 if ( isset( $parsedWidthParam['height'] ) ) {
5634 $height = $parsedWidthParam['height'];
5635 if ( $handler->validateParam( 'height', $height ) ) {
5636 $params[$type]['height'] = $height;
5637 $validated = true;
5638 }
5639 }
5640 # else no validation -- T15436
5641 } else {
5642 if ( $type === 'handler' ) {
5643 # Validate handler parameter
5644 $validated = $handler->validateParam( $paramName, $value );
5645 } else {
5646 # Validate internal parameters
5647 switch ( $paramName ) {
5648 case 'manualthumb':
5649 case 'alt':
5650 case 'class':
5651 # @todo FIXME: Possibly check validity here for
5652 # manualthumb? downstream behavior seems odd with
5653 # missing manual thumbs.
5654 $validated = true;
5655 $value = $this->stripAltText( $value, $holders );
5656 break;
5657 case 'link':
5658 list( $paramName, $value ) =
5659 $this->parseLinkParameter(
5660 $this->stripAltText( $value, $holders )
5661 );
5662 if ( $paramName ) {
5663 $validated = true;
5664 if ( $paramName === 'no-link' ) {
5665 $value = true;
5666 }
5667 if ( ( $paramName === 'link-url' ) && $this->mOptions->getExternalLinkTarget() ) {
5668 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5669 }
5670 }
5671 break;
5672 case 'frameless':
5673 case 'framed':
5674 case 'thumbnail':
5675 // use first appearing option, discard others.
5676 $validated = !$seenformat;
5677 $seenformat = true;
5678 break;
5679 default:
5680 # Most other things appear to be empty or numeric...
5681 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5682 }
5683 }
5684
5685 if ( $validated ) {
5686 $params[$type][$paramName] = $value;
5687 }
5688 }
5689 }
5690 if ( !$validated ) {
5691 $caption = $part;
5692 }
5693 }
5694
5695 # Process alignment parameters
5696 if ( $params['horizAlign'] ) {
5697 $params['frame']['align'] = key( $params['horizAlign'] );
5698 }
5699 if ( $params['vertAlign'] ) {
5700 $params['frame']['valign'] = key( $params['vertAlign'] );
5701 }
5702
5703 $params['frame']['caption'] = $caption;
5704
5705 # Will the image be presented in a frame, with the caption below?
5706 $imageIsFramed = isset( $params['frame']['frame'] )
5707 || isset( $params['frame']['framed'] )
5708 || isset( $params['frame']['thumbnail'] )
5709 || isset( $params['frame']['manualthumb'] );
5710
5711 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5712 # came to also set the caption, ordinary text after the image -- which
5713 # makes no sense, because that just repeats the text multiple times in
5714 # screen readers. It *also* came to set the title attribute.
5715 # Now that we have an alt attribute, we should not set the alt text to
5716 # equal the caption: that's worse than useless, it just repeats the
5717 # text. This is the framed/thumbnail case. If there's no caption, we
5718 # use the unnamed parameter for alt text as well, just for the time be-
5719 # ing, if the unnamed param is set and the alt param is not.
5720 # For the future, we need to figure out if we want to tweak this more,
5721 # e.g., introducing a title= parameter for the title; ignoring the un-
5722 # named parameter entirely for images without a caption; adding an ex-
5723 # plicit caption= parameter and preserving the old magic unnamed para-
5724 # meter for BC; ...
5725 if ( $imageIsFramed ) { # Framed image
5726 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5727 # No caption or alt text, add the filename as the alt text so
5728 # that screen readers at least get some description of the image
5729 $params['frame']['alt'] = $title->getText();
5730 }
5731 # Do not set $params['frame']['title'] because tooltips don't make sense
5732 # for framed images
5733 } else { # Inline image
5734 if ( !isset( $params['frame']['alt'] ) ) {
5735 # No alt text, use the "caption" for the alt text
5736 if ( $caption !== '' ) {
5737 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5738 } else {
5739 # No caption, fall back to using the filename for the
5740 # alt text
5741 $params['frame']['alt'] = $title->getText();
5742 }
5743 }
5744 # Use the "caption" for the tooltip text
5745 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5746 }
5747 $params['handler']['targetlang'] = $this->getTargetLanguage()->getCode();
5748
5749 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5750
5751 # Linker does the rest
5752 $time = $options['time'] ?? false;
5753 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5754 $time, $descQuery, $this->mOptions->getThumbSize() );
5755
5756 # Give the handler a chance to modify the parser object
5757 if ( $handler ) {
5758 $handler->parserTransformHook( $this, $file );
5759 }
5760
5761 return $ret;
5762 }
5763
5764 /**
5765 * Parse the value of 'link' parameter in image syntax (`[[File:Foo.jpg|link=<value>]]`).
5766 *
5767 * Adds an entry to appropriate link tables.
5768 *
5769 * @since 1.32
5770 * @param string $value
5771 * @return array of `[ type, target ]`, where:
5772 * - `type` is one of:
5773 * - `null`: Given value is not a valid link target, use default
5774 * - `'no-link'`: Given value is empty, do not generate a link
5775 * - `'link-url'`: Given value is a valid external link
5776 * - `'link-title'`: Given value is a valid internal link
5777 * - `target` is:
5778 * - When `type` is `null` or `'no-link'`: `false`
5779 * - When `type` is `'link-url'`: URL string corresponding to given value
5780 * - When `type` is `'link-title'`: Title object corresponding to given value
5781 */
5782 public function parseLinkParameter( $value ) {
5783 $chars = self::EXT_LINK_URL_CLASS;
5784 $addr = self::EXT_LINK_ADDR;
5785 $prots = $this->mUrlProtocols;
5786 $type = null;
5787 $target = false;
5788 if ( $value === '' ) {
5789 $type = 'no-link';
5790 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5791 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5792 $this->mOutput->addExternalLink( $value );
5793 $type = 'link-url';
5794 $target = $value;
5795 }
5796 } else {
5797 $linkTitle = Title::newFromText( $value );
5798 if ( $linkTitle ) {
5799 $this->mOutput->addLink( $linkTitle );
5800 $type = 'link-title';
5801 $target = $linkTitle;
5802 }
5803 }
5804 return [ $type, $target ];
5805 }
5806
5807 /**
5808 * @param string $caption
5809 * @param LinkHolderArray|bool $holders
5810 * @return mixed|string
5811 */
5812 protected function stripAltText( $caption, $holders ) {
5813 # Strip bad stuff out of the title (tooltip). We can't just use
5814 # replaceLinkHoldersText() here, because if this function is called
5815 # from handleInternalLinks2(), mLinkHolders won't be up-to-date.
5816 if ( $holders ) {
5817 $tooltip = $holders->replaceText( $caption );
5818 } else {
5819 $tooltip = $this->replaceLinkHoldersText( $caption );
5820 }
5821
5822 # make sure there are no placeholders in thumbnail attributes
5823 # that are later expanded to html- so expand them now and
5824 # remove the tags
5825 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5826 # Compatibility hack! In HTML certain entity references not terminated
5827 # by a semicolon are decoded (but not if we're in an attribute; that's
5828 # how link URLs get away without properly escaping & in queries).
5829 # But wikitext has always required semicolon-termination of entities,
5830 # so encode & where needed to avoid decode of semicolon-less entities.
5831 # See T209236 and
5832 # https://www.w3.org/TR/html5/syntax.html#named-character-references
5833 # T210437 discusses moving this workaround to Sanitizer::stripAllTags.
5834 $tooltip = preg_replace( "/
5835 & # 1. entity prefix
5836 (?= # 2. followed by:
5837 (?: # a. one of the legacy semicolon-less named entities
5838 A(?:Elig|MP|acute|circ|grave|ring|tilde|uml)|
5839 C(?:OPY|cedil)|E(?:TH|acute|circ|grave|uml)|
5840 GT|I(?:acute|circ|grave|uml)|LT|Ntilde|
5841 O(?:acute|circ|grave|slash|tilde|uml)|QUOT|REG|THORN|
5842 U(?:acute|circ|grave|uml)|Yacute|
5843 a(?:acute|c(?:irc|ute)|elig|grave|mp|ring|tilde|uml)|brvbar|
5844 c(?:cedil|edil|urren)|cent(?!erdot;)|copy(?!sr;)|deg|
5845 divide(?!ontimes;)|e(?:acute|circ|grave|th|uml)|
5846 frac(?:1(?:2|4)|34)|
5847 gt(?!c(?:c|ir)|dot|lPar|quest|r(?:a(?:pprox|rr)|dot|eq(?:less|qless)|less|sim);)|
5848 i(?:acute|circ|excl|grave|quest|uml)|laquo|
5849 lt(?!c(?:c|ir)|dot|hree|imes|larr|quest|r(?:Par|i(?:e|f|));)|
5850 m(?:acr|i(?:cro|ddot))|n(?:bsp|tilde)|
5851 not(?!in(?:E|dot|v(?:a|b|c)|)|ni(?:v(?:a|b|c)|);)|
5852 o(?:acute|circ|grave|rd(?:f|m)|slash|tilde|uml)|
5853 p(?:lusmn|ound)|para(?!llel;)|quot|r(?:aquo|eg)|
5854 s(?:ect|hy|up(?:1|2|3)|zlig)|thorn|times(?!b(?:ar|)|d;)|
5855 u(?:acute|circ|grave|ml|uml)|y(?:acute|en|uml)
5856 )
5857 (?:[^;]|$)) # b. and not followed by a semicolon
5858 # S = study, for efficiency
5859 /Sx", '&amp;', $tooltip );
5860 $tooltip = Sanitizer::stripAllTags( $tooltip );
5861
5862 return $tooltip;
5863 }
5864
5865 /**
5866 * Set a flag in the output object indicating that the content is dynamic and
5867 * shouldn't be cached.
5868 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5869 */
5870 public function disableCache() {
5871 $this->logger->debug( "Parser output marked as uncacheable." );
5872 if ( !$this->mOutput ) {
5873 throw new MWException( __METHOD__ .
5874 " can only be called when actually parsing something" );
5875 }
5876 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5877 }
5878
5879 /**
5880 * Callback from the Sanitizer for expanding items found in HTML attribute
5881 * values, so they can be safely tested and escaped.
5882 *
5883 * @param string &$text
5884 * @param bool|PPFrame $frame
5885 * @return string
5886 */
5887 public function attributeStripCallback( &$text, $frame = false ) {
5888 $text = $this->replaceVariables( $text, $frame );
5889 $text = $this->mStripState->unstripBoth( $text );
5890 return $text;
5891 }
5892
5893 /**
5894 * Accessor
5895 *
5896 * @return array
5897 */
5898 public function getTags() {
5899 $this->firstCallInit();
5900 return array_merge(
5901 array_keys( $this->mTransparentTagHooks ),
5902 array_keys( $this->mTagHooks ),
5903 array_keys( $this->mFunctionTagHooks )
5904 );
5905 }
5906
5907 /**
5908 * @since 1.32
5909 * @return array
5910 */
5911 public function getFunctionSynonyms() {
5912 $this->firstCallInit();
5913 return $this->mFunctionSynonyms;
5914 }
5915
5916 /**
5917 * @since 1.32
5918 * @return string
5919 */
5920 public function getUrlProtocols() {
5921 return $this->mUrlProtocols;
5922 }
5923
5924 /**
5925 * Replace transparent tags in $text with the values given by the callbacks.
5926 *
5927 * Transparent tag hooks are like regular XML-style tag hooks, except they
5928 * operate late in the transformation sequence, on HTML instead of wikitext.
5929 *
5930 * @param string $text
5931 *
5932 * @return string
5933 */
5934 public function replaceTransparentTags( $text ) {
5935 $matches = [];
5936 $elements = array_keys( $this->mTransparentTagHooks );
5937 $text = self::extractTagsAndParams( $elements, $text, $matches );
5938 $replacements = [];
5939
5940 foreach ( $matches as $marker => $data ) {
5941 list( $element, $content, $params, $tag ) = $data;
5942 $tagName = strtolower( $element );
5943 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5944 $output = call_user_func_array(
5945 $this->mTransparentTagHooks[$tagName],
5946 [ $content, $params, $this ]
5947 );
5948 } else {
5949 $output = $tag;
5950 }
5951 $replacements[$marker] = $output;
5952 }
5953 return strtr( $text, $replacements );
5954 }
5955
5956 /**
5957 * Break wikitext input into sections, and either pull or replace
5958 * some particular section's text.
5959 *
5960 * External callers should use the getSection and replaceSection methods.
5961 *
5962 * @param string $text Page wikitext
5963 * @param string|int $sectionId A section identifier string of the form:
5964 * "<flag1> - <flag2> - ... - <section number>"
5965 *
5966 * Currently the only recognised flag is "T", which means the target section number
5967 * was derived during a template inclusion parse, in other words this is a template
5968 * section edit link. If no flags are given, it was an ordinary section edit link.
5969 * This flag is required to avoid a section numbering mismatch when a section is
5970 * enclosed by "<includeonly>" (T8563).
5971 *
5972 * The section number 0 pulls the text before the first heading; other numbers will
5973 * pull the given section along with its lower-level subsections. If the section is
5974 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5975 *
5976 * Section 0 is always considered to exist, even if it only contains the empty
5977 * string. If $text is the empty string and section 0 is replaced, $newText is
5978 * returned.
5979 *
5980 * @param string $mode One of "get" or "replace"
5981 * @param string $newText Replacement text for section data.
5982 * @return string For "get", the extracted section text.
5983 * for "replace", the whole page with the section replaced.
5984 */
5985 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5986 global $wgTitle; # not generally used but removes an ugly failure mode
5987
5988 $magicScopeVariable = $this->lock();
5989 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5990 $outText = '';
5991 $frame = $this->getPreprocessor()->newFrame();
5992
5993 # Process section extraction flags
5994 $flags = 0;
5995 $sectionParts = explode( '-', $sectionId );
5996 $sectionIndex = array_pop( $sectionParts );
5997 foreach ( $sectionParts as $part ) {
5998 if ( $part === 'T' ) {
5999 $flags |= self::PTD_FOR_INCLUSION;
6000 }
6001 }
6002
6003 # Check for empty input
6004 if ( strval( $text ) === '' ) {
6005 # Only sections 0 and T-0 exist in an empty document
6006 if ( $sectionIndex == 0 ) {
6007 if ( $mode === 'get' ) {
6008 return '';
6009 }
6010
6011 return $newText;
6012 } else {
6013 if ( $mode === 'get' ) {
6014 return $newText;
6015 }
6016
6017 return $text;
6018 }
6019 }
6020
6021 # Preprocess the text
6022 $root = $this->preprocessToDom( $text, $flags );
6023
6024 # <h> nodes indicate section breaks
6025 # They can only occur at the top level, so we can find them by iterating the root's children
6026 $node = $root->getFirstChild();
6027
6028 # Find the target section
6029 if ( $sectionIndex == 0 ) {
6030 # Section zero doesn't nest, level=big
6031 $targetLevel = 1000;
6032 } else {
6033 while ( $node ) {
6034 if ( $node->getName() === 'h' ) {
6035 $bits = $node->splitHeading();
6036 if ( $bits['i'] == $sectionIndex ) {
6037 $targetLevel = $bits['level'];
6038 break;
6039 }
6040 }
6041 if ( $mode === 'replace' ) {
6042 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
6043 }
6044 $node = $node->getNextSibling();
6045 }
6046 }
6047
6048 if ( !$node ) {
6049 # Not found
6050 if ( $mode === 'get' ) {
6051 return $newText;
6052 } else {
6053 return $text;
6054 }
6055 }
6056
6057 # Find the end of the section, including nested sections
6058 do {
6059 if ( $node->getName() === 'h' ) {
6060 $bits = $node->splitHeading();
6061 $curLevel = $bits['level'];
6062 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
6063 break;
6064 }
6065 }
6066 if ( $mode === 'get' ) {
6067 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
6068 }
6069 $node = $node->getNextSibling();
6070 } while ( $node );
6071
6072 # Write out the remainder (in replace mode only)
6073 if ( $mode === 'replace' ) {
6074 # Output the replacement text
6075 # Add two newlines on -- trailing whitespace in $newText is conventionally
6076 # stripped by the editor, so we need both newlines to restore the paragraph gap
6077 # Only add trailing whitespace if there is newText
6078 if ( $newText != "" ) {
6079 $outText .= $newText . "\n\n";
6080 }
6081
6082 while ( $node ) {
6083 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
6084 $node = $node->getNextSibling();
6085 }
6086 }
6087
6088 if ( is_string( $outText ) ) {
6089 # Re-insert stripped tags
6090 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
6091 }
6092
6093 return $outText;
6094 }
6095
6096 /**
6097 * This function returns the text of a section, specified by a number ($section).
6098 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
6099 * the first section before any such heading (section 0).
6100 *
6101 * If a section contains subsections, these are also returned.
6102 *
6103 * @param string $text Text to look in
6104 * @param string|int $sectionId Section identifier as a number or string
6105 * (e.g. 0, 1 or 'T-1').
6106 * @param string $defaultText Default to return if section is not found
6107 *
6108 * @return string Text of the requested section
6109 */
6110 public function getSection( $text, $sectionId, $defaultText = '' ) {
6111 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
6112 }
6113
6114 /**
6115 * This function returns $oldtext after the content of the section
6116 * specified by $section has been replaced with $text. If the target
6117 * section does not exist, $oldtext is returned unchanged.
6118 *
6119 * @param string $oldText Former text of the article
6120 * @param string|int $sectionId Section identifier as a number or string
6121 * (e.g. 0, 1 or 'T-1').
6122 * @param string $newText Replacing text
6123 *
6124 * @return string Modified text
6125 */
6126 public function replaceSection( $oldText, $sectionId, $newText ) {
6127 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
6128 }
6129
6130 /**
6131 * Get the ID of the revision we are parsing
6132 *
6133 * The return value will be either:
6134 * - a) Positive, indicating a specific revision ID (current or old)
6135 * - b) Zero, meaning the revision ID is specified by getCurrentRevisionCallback()
6136 * - c) Null, meaning the parse is for preview mode and there is no revision
6137 *
6138 * @return int|null
6139 */
6140 public function getRevisionId() {
6141 return $this->mRevisionId;
6142 }
6143
6144 /**
6145 * Get the revision object for $this->mRevisionId
6146 *
6147 * @return Revision|null Either a Revision object or null
6148 * @since 1.23 (public since 1.23)
6149 */
6150 public function getRevisionObject() {
6151 if ( $this->mRevisionObject ) {
6152 return $this->mRevisionObject;
6153 }
6154
6155 // NOTE: try to get the RevisionObject even if mRevisionId is null.
6156 // This is useful when parsing a revision that has not yet been saved.
6157 // However, if we get back a saved revision even though we are in
6158 // preview mode, we'll have to ignore it, see below.
6159 // NOTE: This callback may be used to inject an OLD revision that was
6160 // already loaded, so "current" is a bit of a misnomer. We can't just
6161 // skip it if mRevisionId is set.
6162 $rev = call_user_func(
6163 $this->mOptions->getCurrentRevisionCallback(),
6164 $this->getTitle(),
6165 $this
6166 );
6167
6168 if ( $this->mRevisionId === null && $rev && $rev->getId() ) {
6169 // We are in preview mode (mRevisionId is null), and the current revision callback
6170 // returned an existing revision. Ignore it and return null, it's probably the page's
6171 // current revision, which is not what we want here. Note that we do want to call the
6172 // callback to allow the unsaved revision to be injected here, e.g. for
6173 // self-transclusion previews.
6174 return null;
6175 }
6176
6177 // If the parse is for a new revision, then the callback should have
6178 // already been set to force the object and should match mRevisionId.
6179 // If not, try to fetch by mRevisionId for sanity.
6180 if ( $this->mRevisionId && $rev && $rev->getId() != $this->mRevisionId ) {
6181 $rev = Revision::newFromId( $this->mRevisionId );
6182 }
6183
6184 $this->mRevisionObject = $rev;
6185
6186 return $this->mRevisionObject;
6187 }
6188
6189 /**
6190 * Get the timestamp associated with the current revision, adjusted for
6191 * the default server-local timestamp
6192 * @return string TS_MW timestamp
6193 */
6194 public function getRevisionTimestamp() {
6195 if ( $this->mRevisionTimestamp !== null ) {
6196 return $this->mRevisionTimestamp;
6197 }
6198
6199 # Use specified revision timestamp, falling back to the current timestamp
6200 $revObject = $this->getRevisionObject();
6201 $timestamp = $revObject ? $revObject->getTimestamp() : $this->mOptions->getTimestamp();
6202 $this->mOutput->setRevisionTimestampUsed( $timestamp ); // unadjusted time zone
6203
6204 # The cryptic '' timezone parameter tells to use the site-default
6205 # timezone offset instead of the user settings.
6206 # Since this value will be saved into the parser cache, served
6207 # to other users, and potentially even used inside links and such,
6208 # it needs to be consistent for all visitors.
6209 $this->mRevisionTimestamp = $this->contLang->userAdjust( $timestamp, '' );
6210
6211 return $this->mRevisionTimestamp;
6212 }
6213
6214 /**
6215 * Get the name of the user that edited the last revision
6216 *
6217 * @return string User name
6218 */
6219 public function getRevisionUser() {
6220 if ( is_null( $this->mRevisionUser ) ) {
6221 $revObject = $this->getRevisionObject();
6222
6223 # if this template is subst: the revision id will be blank,
6224 # so just use the current user's name
6225 if ( $revObject ) {
6226 $this->mRevisionUser = $revObject->getUserText();
6227 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6228 $this->mRevisionUser = $this->getUser()->getName();
6229 }
6230 }
6231 return $this->mRevisionUser;
6232 }
6233
6234 /**
6235 * Get the size of the revision
6236 *
6237 * @return int|null Revision size
6238 */
6239 public function getRevisionSize() {
6240 if ( is_null( $this->mRevisionSize ) ) {
6241 $revObject = $this->getRevisionObject();
6242
6243 # if this variable is subst: the revision id will be blank,
6244 # so just use the parser input size, because the own substituation
6245 # will change the size.
6246 if ( $revObject ) {
6247 $this->mRevisionSize = $revObject->getSize();
6248 } else {
6249 $this->mRevisionSize = $this->mInputSize;
6250 }
6251 }
6252 return $this->mRevisionSize;
6253 }
6254
6255 /**
6256 * Mutator for $mDefaultSort
6257 *
6258 * @param string $sort New value
6259 */
6260 public function setDefaultSort( $sort ) {
6261 $this->mDefaultSort = $sort;
6262 $this->mOutput->setProperty( 'defaultsort', $sort );
6263 }
6264
6265 /**
6266 * Accessor for $mDefaultSort
6267 * Will use the empty string if none is set.
6268 *
6269 * This value is treated as a prefix, so the
6270 * empty string is equivalent to sorting by
6271 * page name.
6272 *
6273 * @return string
6274 */
6275 public function getDefaultSort() {
6276 if ( $this->mDefaultSort !== false ) {
6277 return $this->mDefaultSort;
6278 } else {
6279 return '';
6280 }
6281 }
6282
6283 /**
6284 * Accessor for $mDefaultSort
6285 * Unlike getDefaultSort(), will return false if none is set
6286 *
6287 * @return string|bool
6288 */
6289 public function getCustomDefaultSort() {
6290 return $this->mDefaultSort;
6291 }
6292
6293 private static function getSectionNameFromStrippedText( $text ) {
6294 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6295 $text = Sanitizer::decodeCharReferences( $text );
6296 $text = self::normalizeSectionName( $text );
6297 return $text;
6298 }
6299
6300 private static function makeAnchor( $sectionName ) {
6301 return '#' . Sanitizer::escapeIdForLink( $sectionName );
6302 }
6303
6304 private function makeLegacyAnchor( $sectionName ) {
6305 $fragmentMode = $this->svcOptions->get( 'FragmentMode' );
6306 if ( isset( $fragmentMode[1] ) && $fragmentMode[1] === 'legacy' ) {
6307 // ForAttribute() and ForLink() are the same for legacy encoding
6308 $id = Sanitizer::escapeIdForAttribute( $sectionName, Sanitizer::ID_FALLBACK );
6309 } else {
6310 $id = Sanitizer::escapeIdForLink( $sectionName );
6311 }
6312
6313 return "#$id";
6314 }
6315
6316 /**
6317 * Try to guess the section anchor name based on a wikitext fragment
6318 * presumably extracted from a heading, for example "Header" from
6319 * "== Header ==".
6320 *
6321 * @param string $text
6322 * @return string Anchor (starting with '#')
6323 */
6324 public function guessSectionNameFromWikiText( $text ) {
6325 # Strip out wikitext links(they break the anchor)
6326 $text = $this->stripSectionName( $text );
6327 $sectionName = self::getSectionNameFromStrippedText( $text );
6328 return self::makeAnchor( $sectionName );
6329 }
6330
6331 /**
6332 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6333 * instead, if possible. For use in redirects, since various versions
6334 * of Microsoft browsers interpret Location: headers as something other
6335 * than UTF-8, resulting in breakage.
6336 *
6337 * @param string $text The section name
6338 * @return string Anchor (starting with '#')
6339 */
6340 public function guessLegacySectionNameFromWikiText( $text ) {
6341 # Strip out wikitext links(they break the anchor)
6342 $text = $this->stripSectionName( $text );
6343 $sectionName = self::getSectionNameFromStrippedText( $text );
6344 return $this->makeLegacyAnchor( $sectionName );
6345 }
6346
6347 /**
6348 * Like guessSectionNameFromWikiText(), but takes already-stripped text as input.
6349 * @param string $text Section name (plain text)
6350 * @return string Anchor (starting with '#')
6351 */
6352 public static function guessSectionNameFromStrippedText( $text ) {
6353 $sectionName = self::getSectionNameFromStrippedText( $text );
6354 return self::makeAnchor( $sectionName );
6355 }
6356
6357 /**
6358 * Apply the same normalization as code making links to this section would
6359 *
6360 * @param string $text
6361 * @return string
6362 */
6363 private static function normalizeSectionName( $text ) {
6364 # T90902: ensure the same normalization is applied for IDs as to links
6365 /** @var MediaWikiTitleCodec $titleParser */
6366 $titleParser = MediaWikiServices::getInstance()->getTitleParser();
6367 '@phan-var MediaWikiTitleCodec $titleParser';
6368 try {
6369
6370 $parts = $titleParser->splitTitleString( "#$text" );
6371 } catch ( MalformedTitleException $ex ) {
6372 return $text;
6373 }
6374 return $parts['fragment'];
6375 }
6376
6377 /**
6378 * Strips a text string of wikitext for use in a section anchor
6379 *
6380 * Accepts a text string and then removes all wikitext from the
6381 * string and leaves only the resultant text (i.e. the result of
6382 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6383 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6384 * to create valid section anchors by mimicing the output of the
6385 * parser when headings are parsed.
6386 *
6387 * @param string $text Text string to be stripped of wikitext
6388 * for use in a Section anchor
6389 * @return string Filtered text string
6390 */
6391 public function stripSectionName( $text ) {
6392 # Strip internal link markup
6393 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6394 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6395
6396 # Strip external link markup
6397 # @todo FIXME: Not tolerant to blank link text
6398 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6399 # on how many empty links there are on the page - need to figure that out.
6400 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6401
6402 # Parse wikitext quotes (italics & bold)
6403 $text = $this->doQuotes( $text );
6404
6405 # Strip HTML tags
6406 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6407 return $text;
6408 }
6409
6410 /**
6411 * strip/replaceVariables/unstrip for preprocessor regression testing
6412 *
6413 * @param string $text
6414 * @param Title $title
6415 * @param ParserOptions $options
6416 * @param int $outputType
6417 *
6418 * @return string
6419 * @deprecated since 1.34; should not be used outside parser class.
6420 */
6421 public function testSrvus( $text, Title $title, ParserOptions $options,
6422 $outputType = self::OT_HTML
6423 ) {
6424 wfDeprecated( __METHOD__, '1.34' );
6425 return $this->fuzzTestSrvus( $text, $title, $options, $outputType );
6426 }
6427
6428 /**
6429 * Strip/replaceVariables/unstrip for preprocessor regression testing
6430 *
6431 * @param string $text
6432 * @param Title $title
6433 * @param ParserOptions $options
6434 * @param int $outputType
6435 *
6436 * @return string
6437 */
6438 private function fuzzTestSrvus( $text, Title $title, ParserOptions $options,
6439 $outputType = self::OT_HTML
6440 ) {
6441 $magicScopeVariable = $this->lock();
6442 $this->startParse( $title, $options, $outputType, true );
6443
6444 $text = $this->replaceVariables( $text );
6445 $text = $this->mStripState->unstripBoth( $text );
6446 $text = Sanitizer::removeHTMLtags( $text );
6447 return $text;
6448 }
6449
6450 /**
6451 * @param string $text
6452 * @param Title $title
6453 * @param ParserOptions $options
6454 * @return string
6455 * @deprecated since 1.34; should not be used outside parser class.
6456 */
6457 public function testPst( $text, Title $title, ParserOptions $options ) {
6458 wfDeprecated( __METHOD__, '1.34' );
6459 return $this->fuzzTestPst( $text, $title, $options );
6460 }
6461
6462 /**
6463 * @param string $text
6464 * @param Title $title
6465 * @param ParserOptions $options
6466 * @return string
6467 */
6468 private function fuzzTestPst( $text, Title $title, ParserOptions $options ) {
6469 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6470 }
6471
6472 /**
6473 * @param string $text
6474 * @param Title $title
6475 * @param ParserOptions $options
6476 * @return string
6477 * @deprecated since 1.34; should not be used outside parser class.
6478 */
6479 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6480 wfDeprecated( __METHOD__, '1.34' );
6481 return $this->fuzzTestPreprocess( $text, $title, $options );
6482 }
6483
6484 /**
6485 * @param string $text
6486 * @param Title $title
6487 * @param ParserOptions $options
6488 * @return string
6489 */
6490 private function fuzzTestPreprocess( $text, Title $title, ParserOptions $options ) {
6491 return $this->fuzzTestSrvus( $text, $title, $options, self::OT_PREPROCESS );
6492 }
6493
6494 /**
6495 * Call a callback function on all regions of the given text that are not
6496 * inside strip markers, and replace those regions with the return value
6497 * of the callback. For example, with input:
6498 *
6499 * aaa<MARKER>bbb
6500 *
6501 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6502 * two strings will be replaced with the value returned by the callback in
6503 * each case.
6504 *
6505 * @param string $s
6506 * @param callable $callback
6507 *
6508 * @return string
6509 */
6510 public function markerSkipCallback( $s, $callback ) {
6511 $i = 0;
6512 $out = '';
6513 while ( $i < strlen( $s ) ) {
6514 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6515 if ( $markerStart === false ) {
6516 $out .= call_user_func( $callback, substr( $s, $i ) );
6517 break;
6518 } else {
6519 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6520 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6521 if ( $markerEnd === false ) {
6522 $out .= substr( $s, $markerStart );
6523 break;
6524 } else {
6525 $markerEnd += strlen( self::MARKER_SUFFIX );
6526 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6527 $i = $markerEnd;
6528 }
6529 }
6530 }
6531 return $out;
6532 }
6533
6534 /**
6535 * Remove any strip markers found in the given text.
6536 *
6537 * @param string $text
6538 * @return string
6539 */
6540 public function killMarkers( $text ) {
6541 return $this->mStripState->killMarkers( $text );
6542 }
6543
6544 /**
6545 * Save the parser state required to convert the given half-parsed text to
6546 * HTML. "Half-parsed" in this context means the output of
6547 * recursiveTagParse() or internalParse(). This output has strip markers
6548 * from replaceVariables (extensionSubstitution() etc.), and link
6549 * placeholders from replaceLinkHolders().
6550 *
6551 * Returns an array which can be serialized and stored persistently. This
6552 * array can later be loaded into another parser instance with
6553 * unserializeHalfParsedText(). The text can then be safely incorporated into
6554 * the return value of a parser hook.
6555 *
6556 * @deprecated since 1.31
6557 * @param string $text
6558 *
6559 * @return array
6560 */
6561 public function serializeHalfParsedText( $text ) {
6562 wfDeprecated( __METHOD__, '1.31' );
6563 $data = [
6564 'text' => $text,
6565 'version' => self::HALF_PARSED_VERSION,
6566 'stripState' => $this->mStripState->getSubState( $text ),
6567 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6568 ];
6569 return $data;
6570 }
6571
6572 /**
6573 * Load the parser state given in the $data array, which is assumed to
6574 * have been generated by serializeHalfParsedText(). The text contents is
6575 * extracted from the array, and its markers are transformed into markers
6576 * appropriate for the current Parser instance. This transformed text is
6577 * returned, and can be safely included in the return value of a parser
6578 * hook.
6579 *
6580 * If the $data array has been stored persistently, the caller should first
6581 * check whether it is still valid, by calling isValidHalfParsedText().
6582 *
6583 * @deprecated since 1.31
6584 * @param array $data Serialized data
6585 * @throws MWException
6586 * @return string
6587 */
6588 public function unserializeHalfParsedText( $data ) {
6589 wfDeprecated( __METHOD__, '1.31' );
6590 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6591 throw new MWException( __METHOD__ . ': invalid version' );
6592 }
6593
6594 # First, extract the strip state.
6595 $texts = [ $data['text'] ];
6596 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6597
6598 # Now renumber links
6599 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6600
6601 # Should be good to go.
6602 return $texts[0];
6603 }
6604
6605 /**
6606 * Returns true if the given array, presumed to be generated by
6607 * serializeHalfParsedText(), is compatible with the current version of the
6608 * parser.
6609 *
6610 * @deprecated since 1.31
6611 * @param array $data
6612 *
6613 * @return bool
6614 */
6615 public function isValidHalfParsedText( $data ) {
6616 wfDeprecated( __METHOD__, '1.31' );
6617 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6618 }
6619
6620 /**
6621 * Parsed a width param of imagelink like 300px or 200x300px
6622 *
6623 * @param string $value
6624 * @param bool $parseHeight
6625 *
6626 * @return array
6627 * @since 1.20
6628 */
6629 public static function parseWidthParam( $value, $parseHeight = true ) {
6630 $parsedWidthParam = [];
6631 if ( $value === '' ) {
6632 return $parsedWidthParam;
6633 }
6634 $m = [];
6635 # (T15500) In both cases (width/height and width only),
6636 # permit trailing "px" for backward compatibility.
6637 if ( $parseHeight && preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6638 $width = intval( $m[1] );
6639 $height = intval( $m[2] );
6640 $parsedWidthParam['width'] = $width;
6641 $parsedWidthParam['height'] = $height;
6642 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6643 $width = intval( $value );
6644 $parsedWidthParam['width'] = $width;
6645 }
6646 return $parsedWidthParam;
6647 }
6648
6649 /**
6650 * Lock the current instance of the parser.
6651 *
6652 * This is meant to stop someone from calling the parser
6653 * recursively and messing up all the strip state.
6654 *
6655 * @throws MWException If parser is in a parse
6656 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6657 */
6658 protected function lock() {
6659 if ( $this->mInParse ) {
6660 throw new MWException( "Parser state cleared while parsing. "
6661 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6662 }
6663
6664 // Save the backtrace when locking, so that if some code tries locking again,
6665 // we can print the lock owner's backtrace for easier debugging
6666 $e = new Exception;
6667 $this->mInParse = $e->getTraceAsString();
6668
6669 $recursiveCheck = new ScopedCallback( function () {
6670 $this->mInParse = false;
6671 } );
6672
6673 return $recursiveCheck;
6674 }
6675
6676 /**
6677 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6678 *
6679 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6680 * or if there is more than one <p/> tag in the input HTML.
6681 *
6682 * @param string $html
6683 * @return string
6684 * @since 1.24
6685 */
6686 public static function stripOuterParagraph( $html ) {
6687 $m = [];
6688 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) && strpos( $m[1], '</p>' ) === false ) {
6689 $html = $m[1];
6690 }
6691
6692 return $html;
6693 }
6694
6695 /**
6696 * Return this parser if it is not doing anything, otherwise
6697 * get a fresh parser. You can use this method by doing
6698 * $newParser = $oldParser->getFreshParser(), or more simply
6699 * $oldParser->getFreshParser()->parse( ... );
6700 * if you're unsure if $oldParser is safe to use.
6701 *
6702 * @since 1.24
6703 * @return Parser A parser object that is not parsing anything
6704 */
6705 public function getFreshParser() {
6706 if ( $this->mInParse ) {
6707 return $this->factory->create();
6708 } else {
6709 return $this;
6710 }
6711 }
6712
6713 /**
6714 * Set's up the PHP implementation of OOUI for use in this request
6715 * and instructs OutputPage to enable OOUI for itself.
6716 *
6717 * @since 1.26
6718 */
6719 public function enableOOUI() {
6720 OutputPage::setupOOUI();
6721 $this->mOutput->setEnableOOUI( true );
6722 }
6723
6724 /**
6725 * @param string $flag
6726 * @param string $reason
6727 */
6728 protected function setOutputFlag( $flag, $reason ) {
6729 $this->mOutput->setFlag( $flag );
6730 $name = $this->mTitle->getPrefixedText();
6731 $this->logger->debug( __METHOD__ . ": set $flag flag on '$name'; $reason" );
6732 }
6733 }
Wiklou, le Wiki du Wiklou