dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Handle missing namespace prefix in XML dumps more gracefully"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // @codingStandardsIgnoreStart Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102 // @codingStandardsIgnoreEnd
103
104 # Regular expression for a non-newline space
105 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
106
107 # Flags for preprocessToDom
108 const PTD_FOR_INCLUSION = 1;
109
110 # Allowed values for $this->mOutputType
111 # Parameter to startExternalParse().
112 const OT_HTML = 1; # like parse()
113 const OT_WIKI = 2; # like preSaveTransform()
114 const OT_PREPROCESS = 3; # like preprocess()
115 const OT_MSG = 3;
116 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
117
118 /**
119 * @var string Prefix and suffix for temporary replacement strings
120 * for the multipass parser.
121 *
122 * \x7f should never appear in input as it's disallowed in XML.
123 * Using it at the front also gives us a little extra robustness
124 * since it shouldn't match when butted up against identifier-like
125 * string constructs.
126 *
127 * Must not consist of all title characters, or else it will change
128 * the behavior of <nowiki> in a link.
129 *
130 * Must have a character that needs escaping in attributes, otherwise
131 * someone could put a strip marker in an attribute, to get around
132 * escaping quote marks, and break out of the attribute. Thus we add
133 * `'".
134 */
135 const MARKER_SUFFIX = "-QINU`\"'\x7f";
136 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
137
138 # Markers used for wrapping the table of contents
139 const TOC_START = '<mw:toc>';
140 const TOC_END = '</mw:toc>';
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 public $mFirstCall = true;
155
156 # Initialised by initialiseVariables()
157
158 /**
159 * @var MagicWordArray
160 */
161 public $mVariables;
162
163 /**
164 * @var MagicWordArray
165 */
166 public $mSubstWords;
167 # Initialised in constructor
168 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
169
170 # Initialized in getPreprocessor()
171 /** @var Preprocessor */
172 public $mPreprocessor;
173
174 # Cleared with clearState():
175 /**
176 * @var ParserOutput
177 */
178 public $mOutput;
179 public $mAutonumber;
180
181 /**
182 * @var StripState
183 */
184 public $mStripState;
185
186 public $mIncludeCount;
187 /**
188 * @var LinkHolderArray
189 */
190 public $mLinkHolders;
191
192 public $mLinkID;
193 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
194 public $mDefaultSort;
195 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
196 public $mExpensiveFunctionCount; # number of expensive parser function calls
197 public $mShowToc, $mForceTocPosition;
198
199 /**
200 * @var User
201 */
202 public $mUser; # User object; only used when doing pre-save transform
203
204 # Temporary
205 # These are variables reset at least once per parse regardless of $clearState
206
207 /**
208 * @var ParserOptions
209 */
210 public $mOptions;
211
212 /**
213 * @var Title
214 */
215 public $mTitle; # Title context, used for self-link rendering and similar things
216 public $mOutputType; # Output type, one of the OT_xxx constants
217 public $ot; # Shortcut alias, see setOutputType()
218 public $mRevisionObject; # The revision object of the specified revision ID
219 public $mRevisionId; # ID to display in {{REVISIONID}} tags
220 public $mRevisionTimestamp; # The timestamp of the specified revision ID
221 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
222 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
223 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
224 public $mInputSize = false; # For {{PAGESIZE}} on current page.
225
226 /**
227 * @var string Deprecated accessor for the strip marker prefix.
228 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
229 */
230 public $mUniqPrefix = Parser::MARKER_PREFIX;
231
232 /**
233 * @var array Array with the language name of each language link (i.e. the
234 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
235 * duplicate language links to the ParserOutput.
236 */
237 public $mLangLinkLanguages;
238
239 /**
240 * @var MapCacheLRU|null
241 * @since 1.24
242 *
243 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
244 */
245 public $currentRevisionCache;
246
247 /**
248 * @var bool Recursive call protection.
249 * This variable should be treated as if it were private.
250 */
251 public $mInParse = false;
252
253 /** @var SectionProfiler */
254 protected $mProfiler;
255
256 /**
257 * @var LinkRenderer
258 */
259 protected $mLinkRenderer;
260
261 /**
262 * @param array $conf
263 */
264 public function __construct( $conf = [] ) {
265 $this->mConf = $conf;
266 $this->mUrlProtocols = wfUrlProtocols();
267 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
268 self::EXT_LINK_ADDR .
269 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
270 if ( isset( $conf['preprocessorClass'] ) ) {
271 $this->mPreprocessorClass = $conf['preprocessorClass'];
272 } elseif ( defined( 'HPHP_VERSION' ) ) {
273 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
274 $this->mPreprocessorClass = 'Preprocessor_Hash';
275 } elseif ( extension_loaded( 'domxml' ) ) {
276 # PECL extension that conflicts with the core DOM extension (T15770)
277 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
278 $this->mPreprocessorClass = 'Preprocessor_Hash';
279 } elseif ( extension_loaded( 'dom' ) ) {
280 $this->mPreprocessorClass = 'Preprocessor_DOM';
281 } else {
282 $this->mPreprocessorClass = 'Preprocessor_Hash';
283 }
284 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
285 }
286
287 /**
288 * Reduce memory usage to reduce the impact of circular references
289 */
290 public function __destruct() {
291 if ( isset( $this->mLinkHolders ) ) {
292 unset( $this->mLinkHolders );
293 }
294 foreach ( $this as $name => $value ) {
295 unset( $this->$name );
296 }
297 }
298
299 /**
300 * Allow extensions to clean up when the parser is cloned
301 */
302 public function __clone() {
303 $this->mInParse = false;
304
305 // T58226: When you create a reference "to" an object field, that
306 // makes the object field itself be a reference too (until the other
307 // reference goes out of scope). When cloning, any field that's a
308 // reference is copied as a reference in the new object. Both of these
309 // are defined PHP5 behaviors, as inconvenient as it is for us when old
310 // hooks from PHP4 days are passing fields by reference.
311 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
312 // Make a non-reference copy of the field, then rebind the field to
313 // reference the new copy.
314 $tmp = $this->$k;
315 $this->$k =& $tmp;
316 unset( $tmp );
317 }
318
319 Hooks::run( 'ParserCloned', [ $this ] );
320 }
321
322 /**
323 * Do various kinds of initialisation on the first call of the parser
324 */
325 public function firstCallInit() {
326 if ( !$this->mFirstCall ) {
327 return;
328 }
329 $this->mFirstCall = false;
330
331 CoreParserFunctions::register( $this );
332 CoreTagHooks::register( $this );
333 $this->initialiseVariables();
334
335 // Avoid PHP 7.1 warning from passing $this by reference
336 $parser = $this;
337 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
338 }
339
340 /**
341 * Clear Parser state
342 *
343 * @private
344 */
345 public function clearState() {
346 if ( $this->mFirstCall ) {
347 $this->firstCallInit();
348 }
349 $this->mOutput = new ParserOutput;
350 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
351 $this->mAutonumber = 0;
352 $this->mIncludeCount = [];
353 $this->mLinkHolders = new LinkHolderArray( $this );
354 $this->mLinkID = 0;
355 $this->mRevisionObject = $this->mRevisionTimestamp =
356 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
357 $this->mVarCache = [];
358 $this->mUser = null;
359 $this->mLangLinkLanguages = [];
360 $this->currentRevisionCache = null;
361
362 $this->mStripState = new StripState;
363
364 # Clear these on every parse, T6549
365 $this->mTplRedirCache = $this->mTplDomCache = [];
366
367 $this->mShowToc = true;
368 $this->mForceTocPosition = false;
369 $this->mIncludeSizes = [
370 'post-expand' => 0,
371 'arg' => 0,
372 ];
373 $this->mPPNodeCount = 0;
374 $this->mGeneratedPPNodeCount = 0;
375 $this->mHighestExpansionDepth = 0;
376 $this->mDefaultSort = false;
377 $this->mHeadings = [];
378 $this->mDoubleUnderscores = [];
379 $this->mExpensiveFunctionCount = 0;
380
381 # Fix cloning
382 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
383 $this->mPreprocessor = null;
384 }
385
386 $this->mProfiler = new SectionProfiler();
387
388 // Avoid PHP 7.1 warning from passing $this by reference
389 $parser = $this;
390 Hooks::run( 'ParserClearState', [ &$parser ] );
391 }
392
393 /**
394 * Convert wikitext to HTML
395 * Do not call this function recursively.
396 *
397 * @param string $text Text we want to parse
398 * @param Title $title
399 * @param ParserOptions $options
400 * @param bool $linestart
401 * @param bool $clearState
402 * @param int $revid Number to pass in {{REVISIONID}}
403 * @return ParserOutput A ParserOutput
404 */
405 public function parse(
406 $text, Title $title, ParserOptions $options,
407 $linestart = true, $clearState = true, $revid = null
408 ) {
409 /**
410 * First pass--just handle <nowiki> sections, pass the rest off
411 * to internalParse() which does all the real work.
412 */
413
414 global $wgShowHostnames;
415
416 if ( $clearState ) {
417 // We use U+007F DELETE to construct strip markers, so we have to make
418 // sure that this character does not occur in the input text.
419 $text = strtr( $text, "\x7f", "?" );
420 $magicScopeVariable = $this->lock();
421 }
422 // Strip U+0000 NULL (T159174)
423 $text = str_replace( "\000", '', $text );
424
425 $this->startParse( $title, $options, self::OT_HTML, $clearState );
426
427 $this->currentRevisionCache = null;
428 $this->mInputSize = strlen( $text );
429 if ( $this->mOptions->getEnableLimitReport() ) {
430 $this->mOutput->resetParseStartTime();
431 }
432
433 $oldRevisionId = $this->mRevisionId;
434 $oldRevisionObject = $this->mRevisionObject;
435 $oldRevisionTimestamp = $this->mRevisionTimestamp;
436 $oldRevisionUser = $this->mRevisionUser;
437 $oldRevisionSize = $this->mRevisionSize;
438 if ( $revid !== null ) {
439 $this->mRevisionId = $revid;
440 $this->mRevisionObject = null;
441 $this->mRevisionTimestamp = null;
442 $this->mRevisionUser = null;
443 $this->mRevisionSize = null;
444 }
445
446 // Avoid PHP 7.1 warning from passing $this by reference
447 $parser = $this;
448 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
449 # No more strip!
450 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
451 $text = $this->internalParse( $text );
452 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
453
454 $text = $this->internalParseHalfParsed( $text, true, $linestart );
455
456 /**
457 * A converted title will be provided in the output object if title and
458 * content conversion are enabled, the article text does not contain
459 * a conversion-suppressing double-underscore tag, and no
460 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
461 * automatic link conversion.
462 */
463 if ( !( $options->getDisableTitleConversion()
464 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
465 || isset( $this->mDoubleUnderscores['notitleconvert'] )
466 || $this->mOutput->getDisplayTitle() !== false )
467 ) {
468 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
469 if ( $convruletitle ) {
470 $this->mOutput->setTitleText( $convruletitle );
471 } else {
472 $titleText = $this->getConverterLanguage()->convertTitle( $title );
473 $this->mOutput->setTitleText( $titleText );
474 }
475 }
476
477 # Done parsing! Compute runtime adaptive expiry if set
478 $this->mOutput->finalizeAdaptiveCacheExpiry();
479
480 # Warn if too many heavyweight parser functions were used
481 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
482 $this->limitationWarn( 'expensive-parserfunction',
483 $this->mExpensiveFunctionCount,
484 $this->mOptions->getExpensiveParserFunctionLimit()
485 );
486 }
487
488 # Information on include size limits, for the benefit of users who try to skirt them
489 if ( $this->mOptions->getEnableLimitReport() ) {
490 $max = $this->mOptions->getMaxIncludeSize();
491
492 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
493 if ( $cpuTime !== null ) {
494 $this->mOutput->setLimitReportData( 'limitreport-cputime',
495 sprintf( "%.3f", $cpuTime )
496 );
497 }
498
499 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
500 $this->mOutput->setLimitReportData( 'limitreport-walltime',
501 sprintf( "%.3f", $wallTime )
502 );
503
504 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
505 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
506 );
507 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
508 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
509 );
510 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
511 [ $this->mIncludeSizes['post-expand'], $max ]
512 );
513 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
514 [ $this->mIncludeSizes['arg'], $max ]
515 );
516 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
517 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
518 );
519 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
520 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
521 );
522 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
523
524 $limitReport = "NewPP limit report\n";
525 if ( $wgShowHostnames ) {
526 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
527 }
528 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
529 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
530 $limitReport .= 'Dynamic content: ' .
531 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
532 "\n";
533
534 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
535 if ( Hooks::run( 'ParserLimitReportFormat',
536 [ $key, &$value, &$limitReport, false, false ]
537 ) ) {
538 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
539 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
540 ->inLanguage( 'en' )->useDatabase( false );
541 if ( !$valueMsg->exists() ) {
542 $valueMsg = new RawMessage( '$1' );
543 }
544 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
545 $valueMsg->params( $value );
546 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
547 }
548 }
549 }
550 // Since we're not really outputting HTML, decode the entities and
551 // then re-encode the things that need hiding inside HTML comments.
552 $limitReport = htmlspecialchars_decode( $limitReport );
553 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
554
555 // Sanitize for comment. Note '‐' in the replacement is U+2010,
556 // which looks much like the problematic '-'.
557 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
558 $text .= "\n<!-- \n$limitReport-->\n";
559
560 // Add on template profiling data in human/machine readable way
561 $dataByFunc = $this->mProfiler->getFunctionStats();
562 uasort( $dataByFunc, function ( $a, $b ) {
563 return $a['real'] < $b['real']; // descending order
564 } );
565 $profileReport = [];
566 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
567 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
568 $item['%real'], $item['real'], $item['calls'],
569 htmlspecialchars( $item['name'] ) );
570 }
571 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
572 $text .= implode( "\n", $profileReport ) . "\n-->\n";
573
574 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
575
576 // Add other cache related metadata
577 if ( $wgShowHostnames ) {
578 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
579 }
580 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
581 $this->mOutput->getCacheTime() );
582 $this->mOutput->setLimitReportData( 'cachereport-ttl',
583 $this->mOutput->getCacheExpiry() );
584 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
585 $this->mOutput->hasDynamicContent() );
586
587 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
588 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
589 $this->mTitle->getPrefixedDBkey() );
590 }
591 }
592 $this->mOutput->setText( $text );
593
594 $this->mRevisionId = $oldRevisionId;
595 $this->mRevisionObject = $oldRevisionObject;
596 $this->mRevisionTimestamp = $oldRevisionTimestamp;
597 $this->mRevisionUser = $oldRevisionUser;
598 $this->mRevisionSize = $oldRevisionSize;
599 $this->mInputSize = false;
600 $this->currentRevisionCache = null;
601
602 return $this->mOutput;
603 }
604
605 /**
606 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
607 * can be called from an extension tag hook.
608 *
609 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
610 * instead, which means that lists and links have not been fully parsed yet,
611 * and strip markers are still present.
612 *
613 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
614 *
615 * Use this function if you're a parser tag hook and you want to parse
616 * wikitext before or after applying additional transformations, and you
617 * intend to *return the result as hook output*, which will cause it to go
618 * through the rest of parsing process automatically.
619 *
620 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
621 * $text are not expanded
622 *
623 * @param string $text Text extension wants to have parsed
624 * @param bool|PPFrame $frame The frame to use for expanding any template variables
625 * @return string UNSAFE half-parsed HTML
626 */
627 public function recursiveTagParse( $text, $frame = false ) {
628 // Avoid PHP 7.1 warning from passing $this by reference
629 $parser = $this;
630 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
631 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
632 $text = $this->internalParse( $text, false, $frame );
633 return $text;
634 }
635
636 /**
637 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
638 * point can be called from an extension tag hook.
639 *
640 * The output of this function is fully-parsed HTML that is safe for output.
641 * If you're a parser tag hook, you might want to use recursiveTagParse()
642 * instead.
643 *
644 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
645 * $text are not expanded
646 *
647 * @since 1.25
648 *
649 * @param string $text Text extension wants to have parsed
650 * @param bool|PPFrame $frame The frame to use for expanding any template variables
651 * @return string Fully parsed HTML
652 */
653 public function recursiveTagParseFully( $text, $frame = false ) {
654 $text = $this->recursiveTagParse( $text, $frame );
655 $text = $this->internalParseHalfParsed( $text, false );
656 return $text;
657 }
658
659 /**
660 * Expand templates and variables in the text, producing valid, static wikitext.
661 * Also removes comments.
662 * Do not call this function recursively.
663 * @param string $text
664 * @param Title $title
665 * @param ParserOptions $options
666 * @param int|null $revid
667 * @param bool|PPFrame $frame
668 * @return mixed|string
669 */
670 public function preprocess( $text, Title $title = null,
671 ParserOptions $options, $revid = null, $frame = false
672 ) {
673 $magicScopeVariable = $this->lock();
674 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
675 if ( $revid !== null ) {
676 $this->mRevisionId = $revid;
677 }
678 // Avoid PHP 7.1 warning from passing $this by reference
679 $parser = $this;
680 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
681 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
682 $text = $this->replaceVariables( $text, $frame );
683 $text = $this->mStripState->unstripBoth( $text );
684 return $text;
685 }
686
687 /**
688 * Recursive parser entry point that can be called from an extension tag
689 * hook.
690 *
691 * @param string $text Text to be expanded
692 * @param bool|PPFrame $frame The frame to use for expanding any template variables
693 * @return string
694 * @since 1.19
695 */
696 public function recursivePreprocess( $text, $frame = false ) {
697 $text = $this->replaceVariables( $text, $frame );
698 $text = $this->mStripState->unstripBoth( $text );
699 return $text;
700 }
701
702 /**
703 * Process the wikitext for the "?preload=" feature. (T7210)
704 *
705 * "<noinclude>", "<includeonly>" etc. are parsed as for template
706 * transclusion, comments, templates, arguments, tags hooks and parser
707 * functions are untouched.
708 *
709 * @param string $text
710 * @param Title $title
711 * @param ParserOptions $options
712 * @param array $params
713 * @return string
714 */
715 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
716 $msg = new RawMessage( $text );
717 $text = $msg->params( $params )->plain();
718
719 # Parser (re)initialisation
720 $magicScopeVariable = $this->lock();
721 $this->startParse( $title, $options, self::OT_PLAIN, true );
722
723 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
724 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
725 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
726 $text = $this->mStripState->unstripBoth( $text );
727 return $text;
728 }
729
730 /**
731 * Get a random string
732 *
733 * @return string
734 * @deprecated since 1.26; use wfRandomString() instead.
735 */
736 public static function getRandomString() {
737 wfDeprecated( __METHOD__, '1.26' );
738 return wfRandomString( 16 );
739 }
740
741 /**
742 * Set the current user.
743 * Should only be used when doing pre-save transform.
744 *
745 * @param User|null $user User object or null (to reset)
746 */
747 public function setUser( $user ) {
748 $this->mUser = $user;
749 }
750
751 /**
752 * Accessor for mUniqPrefix.
753 *
754 * @return string
755 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
756 */
757 public function uniqPrefix() {
758 wfDeprecated( __METHOD__, '1.26' );
759 return self::MARKER_PREFIX;
760 }
761
762 /**
763 * Set the context title
764 *
765 * @param Title $t
766 */
767 public function setTitle( $t ) {
768 if ( !$t ) {
769 $t = Title::newFromText( 'NO TITLE' );
770 }
771
772 if ( $t->hasFragment() ) {
773 # Strip the fragment to avoid various odd effects
774 $this->mTitle = $t->createFragmentTarget( '' );
775 } else {
776 $this->mTitle = $t;
777 }
778 }
779
780 /**
781 * Accessor for the Title object
782 *
783 * @return Title
784 */
785 public function getTitle() {
786 return $this->mTitle;
787 }
788
789 /**
790 * Accessor/mutator for the Title object
791 *
792 * @param Title $x Title object or null to just get the current one
793 * @return Title
794 */
795 public function Title( $x = null ) {
796 return wfSetVar( $this->mTitle, $x );
797 }
798
799 /**
800 * Set the output type
801 *
802 * @param int $ot New value
803 */
804 public function setOutputType( $ot ) {
805 $this->mOutputType = $ot;
806 # Shortcut alias
807 $this->ot = [
808 'html' => $ot == self::OT_HTML,
809 'wiki' => $ot == self::OT_WIKI,
810 'pre' => $ot == self::OT_PREPROCESS,
811 'plain' => $ot == self::OT_PLAIN,
812 ];
813 }
814
815 /**
816 * Accessor/mutator for the output type
817 *
818 * @param int|null $x New value or null to just get the current one
819 * @return int
820 */
821 public function OutputType( $x = null ) {
822 return wfSetVar( $this->mOutputType, $x );
823 }
824
825 /**
826 * Get the ParserOutput object
827 *
828 * @return ParserOutput
829 */
830 public function getOutput() {
831 return $this->mOutput;
832 }
833
834 /**
835 * Get the ParserOptions object
836 *
837 * @return ParserOptions
838 */
839 public function getOptions() {
840 return $this->mOptions;
841 }
842
843 /**
844 * Accessor/mutator for the ParserOptions object
845 *
846 * @param ParserOptions $x New value or null to just get the current one
847 * @return ParserOptions Current ParserOptions object
848 */
849 public function Options( $x = null ) {
850 return wfSetVar( $this->mOptions, $x );
851 }
852
853 /**
854 * @return int
855 */
856 public function nextLinkID() {
857 return $this->mLinkID++;
858 }
859
860 /**
861 * @param int $id
862 */
863 public function setLinkID( $id ) {
864 $this->mLinkID = $id;
865 }
866
867 /**
868 * Get a language object for use in parser functions such as {{FORMATNUM:}}
869 * @return Language
870 */
871 public function getFunctionLang() {
872 return $this->getTargetLanguage();
873 }
874
875 /**
876 * Get the target language for the content being parsed. This is usually the
877 * language that the content is in.
878 *
879 * @since 1.19
880 *
881 * @throws MWException
882 * @return Language
883 */
884 public function getTargetLanguage() {
885 $target = $this->mOptions->getTargetLanguage();
886
887 if ( $target !== null ) {
888 return $target;
889 } elseif ( $this->mOptions->getInterfaceMessage() ) {
890 return $this->mOptions->getUserLangObj();
891 } elseif ( is_null( $this->mTitle ) ) {
892 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
893 }
894
895 return $this->mTitle->getPageLanguage();
896 }
897
898 /**
899 * Get the language object for language conversion
900 * @return Language|null
901 */
902 public function getConverterLanguage() {
903 return $this->getTargetLanguage();
904 }
905
906 /**
907 * Get a User object either from $this->mUser, if set, or from the
908 * ParserOptions object otherwise
909 *
910 * @return User
911 */
912 public function getUser() {
913 if ( !is_null( $this->mUser ) ) {
914 return $this->mUser;
915 }
916 return $this->mOptions->getUser();
917 }
918
919 /**
920 * Get a preprocessor object
921 *
922 * @return Preprocessor
923 */
924 public function getPreprocessor() {
925 if ( !isset( $this->mPreprocessor ) ) {
926 $class = $this->mPreprocessorClass;
927 $this->mPreprocessor = new $class( $this );
928 }
929 return $this->mPreprocessor;
930 }
931
932 /**
933 * Get a LinkRenderer instance to make links with
934 *
935 * @since 1.28
936 * @return LinkRenderer
937 */
938 public function getLinkRenderer() {
939 if ( !$this->mLinkRenderer ) {
940 $this->mLinkRenderer = MediaWikiServices::getInstance()
941 ->getLinkRendererFactory()->create();
942 $this->mLinkRenderer->setStubThreshold(
943 $this->getOptions()->getStubThreshold()
944 );
945 }
946
947 return $this->mLinkRenderer;
948 }
949
950 /**
951 * Replaces all occurrences of HTML-style comments and the given tags
952 * in the text with a random marker and returns the next text. The output
953 * parameter $matches will be an associative array filled with data in
954 * the form:
955 *
956 * @code
957 * 'UNIQ-xxxxx' => [
958 * 'element',
959 * 'tag content',
960 * [ 'param' => 'x' ],
961 * '<element param="x">tag content</element>' ]
962 * @endcode
963 *
964 * @param array $elements List of element names. Comments are always extracted.
965 * @param string $text Source text string.
966 * @param array $matches Out parameter, Array: extracted tags
967 * @param string|null $uniq_prefix
968 * @return string Stripped text
969 * @since 1.26 The uniq_prefix argument is deprecated.
970 */
971 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
972 if ( $uniq_prefix !== null ) {
973 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
974 }
975 static $n = 1;
976 $stripped = '';
977 $matches = [];
978
979 $taglist = implode( '|', $elements );
980 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
981
982 while ( $text != '' ) {
983 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
984 $stripped .= $p[0];
985 if ( count( $p ) < 5 ) {
986 break;
987 }
988 if ( count( $p ) > 5 ) {
989 # comment
990 $element = $p[4];
991 $attributes = '';
992 $close = '';
993 $inside = $p[5];
994 } else {
995 # tag
996 $element = $p[1];
997 $attributes = $p[2];
998 $close = $p[3];
999 $inside = $p[4];
1000 }
1001
1002 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1003 $stripped .= $marker;
1004
1005 if ( $close === '/>' ) {
1006 # Empty element tag, <tag />
1007 $content = null;
1008 $text = $inside;
1009 $tail = null;
1010 } else {
1011 if ( $element === '!--' ) {
1012 $end = '/(-->)/';
1013 } else {
1014 $end = "/(<\\/$element\\s*>)/i";
1015 }
1016 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1017 $content = $q[0];
1018 if ( count( $q ) < 3 ) {
1019 # No end tag -- let it run out to the end of the text.
1020 $tail = '';
1021 $text = '';
1022 } else {
1023 $tail = $q[1];
1024 $text = $q[2];
1025 }
1026 }
1027
1028 $matches[$marker] = [ $element,
1029 $content,
1030 Sanitizer::decodeTagAttributes( $attributes ),
1031 "<$element$attributes$close$content$tail" ];
1032 }
1033 return $stripped;
1034 }
1035
1036 /**
1037 * Get a list of strippable XML-like elements
1038 *
1039 * @return array
1040 */
1041 public function getStripList() {
1042 return $this->mStripList;
1043 }
1044
1045 /**
1046 * Add an item to the strip state
1047 * Returns the unique tag which must be inserted into the stripped text
1048 * The tag will be replaced with the original text in unstrip()
1049 *
1050 * @param string $text
1051 *
1052 * @return string
1053 */
1054 public function insertStripItem( $text ) {
1055 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1056 $this->mMarkerIndex++;
1057 $this->mStripState->addGeneral( $marker, $text );
1058 return $marker;
1059 }
1060
1061 /**
1062 * parse the wiki syntax used to render tables
1063 *
1064 * @private
1065 * @param string $text
1066 * @return string
1067 */
1068 public function doTableStuff( $text ) {
1069
1070 $lines = StringUtils::explode( "\n", $text );
1071 $out = '';
1072 $td_history = []; # Is currently a td tag open?
1073 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1074 $tr_history = []; # Is currently a tr tag open?
1075 $tr_attributes = []; # history of tr attributes
1076 $has_opened_tr = []; # Did this table open a <tr> element?
1077 $indent_level = 0; # indent level of the table
1078
1079 foreach ( $lines as $outLine ) {
1080 $line = trim( $outLine );
1081
1082 if ( $line === '' ) { # empty line, go to next line
1083 $out .= $outLine . "\n";
1084 continue;
1085 }
1086
1087 $first_character = $line[0];
1088 $first_two = substr( $line, 0, 2 );
1089 $matches = [];
1090
1091 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1092 # First check if we are starting a new table
1093 $indent_level = strlen( $matches[1] );
1094
1095 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1096 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1097
1098 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1099 array_push( $td_history, false );
1100 array_push( $last_tag_history, '' );
1101 array_push( $tr_history, false );
1102 array_push( $tr_attributes, '' );
1103 array_push( $has_opened_tr, false );
1104 } elseif ( count( $td_history ) == 0 ) {
1105 # Don't do any of the following
1106 $out .= $outLine . "\n";
1107 continue;
1108 } elseif ( $first_two === '|}' ) {
1109 # We are ending a table
1110 $line = '</table>' . substr( $line, 2 );
1111 $last_tag = array_pop( $last_tag_history );
1112
1113 if ( !array_pop( $has_opened_tr ) ) {
1114 $line = "<tr><td></td></tr>{$line}";
1115 }
1116
1117 if ( array_pop( $tr_history ) ) {
1118 $line = "</tr>{$line}";
1119 }
1120
1121 if ( array_pop( $td_history ) ) {
1122 $line = "</{$last_tag}>{$line}";
1123 }
1124 array_pop( $tr_attributes );
1125 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1126 } elseif ( $first_two === '|-' ) {
1127 # Now we have a table row
1128 $line = preg_replace( '#^\|-+#', '', $line );
1129
1130 # Whats after the tag is now only attributes
1131 $attributes = $this->mStripState->unstripBoth( $line );
1132 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1133 array_pop( $tr_attributes );
1134 array_push( $tr_attributes, $attributes );
1135
1136 $line = '';
1137 $last_tag = array_pop( $last_tag_history );
1138 array_pop( $has_opened_tr );
1139 array_push( $has_opened_tr, true );
1140
1141 if ( array_pop( $tr_history ) ) {
1142 $line = '</tr>';
1143 }
1144
1145 if ( array_pop( $td_history ) ) {
1146 $line = "</{$last_tag}>{$line}";
1147 }
1148
1149 $outLine = $line;
1150 array_push( $tr_history, false );
1151 array_push( $td_history, false );
1152 array_push( $last_tag_history, '' );
1153 } elseif ( $first_character === '|'
1154 || $first_character === '!'
1155 || $first_two === '|+'
1156 ) {
1157 # This might be cell elements, td, th or captions
1158 if ( $first_two === '|+' ) {
1159 $first_character = '+';
1160 $line = substr( $line, 2 );
1161 } else {
1162 $line = substr( $line, 1 );
1163 }
1164
1165 // Implies both are valid for table headings.
1166 if ( $first_character === '!' ) {
1167 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1168 }
1169
1170 # Split up multiple cells on the same line.
1171 # FIXME : This can result in improper nesting of tags processed
1172 # by earlier parser steps.
1173 $cells = explode( '||', $line );
1174
1175 $outLine = '';
1176
1177 # Loop through each table cell
1178 foreach ( $cells as $cell ) {
1179 $previous = '';
1180 if ( $first_character !== '+' ) {
1181 $tr_after = array_pop( $tr_attributes );
1182 if ( !array_pop( $tr_history ) ) {
1183 $previous = "<tr{$tr_after}>\n";
1184 }
1185 array_push( $tr_history, true );
1186 array_push( $tr_attributes, '' );
1187 array_pop( $has_opened_tr );
1188 array_push( $has_opened_tr, true );
1189 }
1190
1191 $last_tag = array_pop( $last_tag_history );
1192
1193 if ( array_pop( $td_history ) ) {
1194 $previous = "</{$last_tag}>\n{$previous}";
1195 }
1196
1197 if ( $first_character === '|' ) {
1198 $last_tag = 'td';
1199 } elseif ( $first_character === '!' ) {
1200 $last_tag = 'th';
1201 } elseif ( $first_character === '+' ) {
1202 $last_tag = 'caption';
1203 } else {
1204 $last_tag = '';
1205 }
1206
1207 array_push( $last_tag_history, $last_tag );
1208
1209 # A cell could contain both parameters and data
1210 $cell_data = explode( '|', $cell, 2 );
1211
1212 # T2553: Note that a '|' inside an invalid link should not
1213 # be mistaken as delimiting cell parameters
1214 # Bug T153140: Neither should language converter markup.
1215 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1216 $cell = "{$previous}<{$last_tag}>{$cell}";
1217 } elseif ( count( $cell_data ) == 1 ) {
1218 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1219 } else {
1220 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1221 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1222 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1223 }
1224
1225 $outLine .= $cell;
1226 array_push( $td_history, true );
1227 }
1228 }
1229 $out .= $outLine . "\n";
1230 }
1231
1232 # Closing open td, tr && table
1233 while ( count( $td_history ) > 0 ) {
1234 if ( array_pop( $td_history ) ) {
1235 $out .= "</td>\n";
1236 }
1237 if ( array_pop( $tr_history ) ) {
1238 $out .= "</tr>\n";
1239 }
1240 if ( !array_pop( $has_opened_tr ) ) {
1241 $out .= "<tr><td></td></tr>\n";
1242 }
1243
1244 $out .= "</table>\n";
1245 }
1246
1247 # Remove trailing line-ending (b/c)
1248 if ( substr( $out, -1 ) === "\n" ) {
1249 $out = substr( $out, 0, -1 );
1250 }
1251
1252 # special case: don't return empty table
1253 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1254 $out = '';
1255 }
1256
1257 return $out;
1258 }
1259
1260 /**
1261 * Helper function for parse() that transforms wiki markup into half-parsed
1262 * HTML. Only called for $mOutputType == self::OT_HTML.
1263 *
1264 * @private
1265 *
1266 * @param string $text The text to parse
1267 * @param bool $isMain Whether this is being called from the main parse() function
1268 * @param PPFrame|bool $frame A pre-processor frame
1269 *
1270 * @return string
1271 */
1272 public function internalParse( $text, $isMain = true, $frame = false ) {
1273
1274 $origText = $text;
1275
1276 // Avoid PHP 7.1 warning from passing $this by reference
1277 $parser = $this;
1278
1279 # Hook to suspend the parser in this state
1280 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1281 return $text;
1282 }
1283
1284 # if $frame is provided, then use $frame for replacing any variables
1285 if ( $frame ) {
1286 # use frame depth to infer how include/noinclude tags should be handled
1287 # depth=0 means this is the top-level document; otherwise it's an included document
1288 if ( !$frame->depth ) {
1289 $flag = 0;
1290 } else {
1291 $flag = Parser::PTD_FOR_INCLUSION;
1292 }
1293 $dom = $this->preprocessToDom( $text, $flag );
1294 $text = $frame->expand( $dom );
1295 } else {
1296 # if $frame is not provided, then use old-style replaceVariables
1297 $text = $this->replaceVariables( $text );
1298 }
1299
1300 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1301 $text = Sanitizer::removeHTMLtags(
1302 $text,
1303 [ $this, 'attributeStripCallback' ],
1304 false,
1305 array_keys( $this->mTransparentTagHooks ),
1306 [],
1307 [ $this, 'addTrackingCategory' ]
1308 );
1309 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1310
1311 # Tables need to come after variable replacement for things to work
1312 # properly; putting them before other transformations should keep
1313 # exciting things like link expansions from showing up in surprising
1314 # places.
1315 $text = $this->doTableStuff( $text );
1316
1317 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1318
1319 $text = $this->doDoubleUnderscore( $text );
1320
1321 $text = $this->doHeadings( $text );
1322 $text = $this->replaceInternalLinks( $text );
1323 $text = $this->doAllQuotes( $text );
1324 $text = $this->replaceExternalLinks( $text );
1325
1326 # replaceInternalLinks may sometimes leave behind
1327 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1328 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1329
1330 $text = $this->doMagicLinks( $text );
1331 $text = $this->formatHeadings( $text, $origText, $isMain );
1332
1333 return $text;
1334 }
1335
1336 /**
1337 * Helper function for parse() that transforms half-parsed HTML into fully
1338 * parsed HTML.
1339 *
1340 * @param string $text
1341 * @param bool $isMain
1342 * @param bool $linestart
1343 * @return string
1344 */
1345 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1346 $text = $this->mStripState->unstripGeneral( $text );
1347
1348 // Avoid PHP 7.1 warning from passing $this by reference
1349 $parser = $this;
1350
1351 if ( $isMain ) {
1352 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1353 }
1354
1355 # Clean up special characters, only run once, next-to-last before doBlockLevels
1356 $fixtags = [
1357 # French spaces, last one Guillemet-left
1358 # only if there is something before the space
1359 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1360 # french spaces, Guillemet-right
1361 '/(\\302\\253) /' => '\\1&#160;',
1362 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1363 ];
1364 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1365
1366 $text = $this->doBlockLevels( $text, $linestart );
1367
1368 $this->replaceLinkHolders( $text );
1369
1370 /**
1371 * The input doesn't get language converted if
1372 * a) It's disabled
1373 * b) Content isn't converted
1374 * c) It's a conversion table
1375 * d) it is an interface message (which is in the user language)
1376 */
1377 if ( !( $this->mOptions->getDisableContentConversion()
1378 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1379 ) {
1380 if ( !$this->mOptions->getInterfaceMessage() ) {
1381 # The position of the convert() call should not be changed. it
1382 # assumes that the links are all replaced and the only thing left
1383 # is the <nowiki> mark.
1384 $text = $this->getConverterLanguage()->convert( $text );
1385 }
1386 }
1387
1388 $text = $this->mStripState->unstripNoWiki( $text );
1389
1390 if ( $isMain ) {
1391 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1392 }
1393
1394 $text = $this->replaceTransparentTags( $text );
1395 $text = $this->mStripState->unstripGeneral( $text );
1396
1397 $text = Sanitizer::normalizeCharReferences( $text );
1398
1399 if ( MWTidy::isEnabled() ) {
1400 if ( $this->mOptions->getTidy() ) {
1401 $text = MWTidy::tidy( $text );
1402 }
1403 } else {
1404 # attempt to sanitize at least some nesting problems
1405 # (T4702 and quite a few others)
1406 $tidyregs = [
1407 # ''Something [http://www.cool.com cool''] -->
1408 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1409 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1410 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1411 # fix up an anchor inside another anchor, only
1412 # at least for a single single nested link (T5695)
1413 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1414 '\\1\\2</a>\\3</a>\\1\\4</a>',
1415 # fix div inside inline elements- doBlockLevels won't wrap a line which
1416 # contains a div, so fix it up here; replace
1417 # div with escaped text
1418 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1419 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1420 # remove empty italic or bold tag pairs, some
1421 # introduced by rules above
1422 '/<([bi])><\/\\1>/' => '',
1423 ];
1424
1425 $text = preg_replace(
1426 array_keys( $tidyregs ),
1427 array_values( $tidyregs ),
1428 $text );
1429 }
1430
1431 if ( $isMain ) {
1432 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1433 }
1434
1435 return $text;
1436 }
1437
1438 /**
1439 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1440 * magic external links.
1441 *
1442 * DML
1443 * @private
1444 *
1445 * @param string $text
1446 *
1447 * @return string
1448 */
1449 public function doMagicLinks( $text ) {
1450 $prots = wfUrlProtocolsWithoutProtRel();
1451 $urlChar = self::EXT_LINK_URL_CLASS;
1452 $addr = self::EXT_LINK_ADDR;
1453 $space = self::SPACE_NOT_NL; # non-newline space
1454 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1455 $spaces = "$space++"; # possessive match of 1 or more spaces
1456 $text = preg_replace_callback(
1457 '!(?: # Start cases
1458 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1459 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1460 (\b # m[3]: Free external links
1461 (?i:$prots)
1462 ($addr$urlChar*) # m[4]: Post-protocol path
1463 ) |
1464 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1465 ([0-9]+)\b |
1466 \bISBN $spaces ( # m[6]: ISBN, capture number
1467 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1468 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1469 [0-9Xx] # check digit
1470 )\b
1471 )!xu", [ $this, 'magicLinkCallback' ], $text );
1472 return $text;
1473 }
1474
1475 /**
1476 * @throws MWException
1477 * @param array $m
1478 * @return HTML|string
1479 */
1480 public function magicLinkCallback( $m ) {
1481 if ( isset( $m[1] ) && $m[1] !== '' ) {
1482 # Skip anchor
1483 return $m[0];
1484 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1485 # Skip HTML element
1486 return $m[0];
1487 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1488 # Free external link
1489 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1490 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1491 # RFC or PMID
1492 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1493 if ( !$this->mOptions->getMagicRFCLinks() ) {
1494 return $m[0];
1495 }
1496 $keyword = 'RFC';
1497 $urlmsg = 'rfcurl';
1498 $cssClass = 'mw-magiclink-rfc';
1499 $trackingCat = 'magiclink-tracking-rfc';
1500 $id = $m[5];
1501 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1502 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1503 return $m[0];
1504 }
1505 $keyword = 'PMID';
1506 $urlmsg = 'pubmedurl';
1507 $cssClass = 'mw-magiclink-pmid';
1508 $trackingCat = 'magiclink-tracking-pmid';
1509 $id = $m[5];
1510 } else {
1511 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1512 substr( $m[0], 0, 20 ) . '"' );
1513 }
1514 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1515 $this->addTrackingCategory( $trackingCat );
1516 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1517 } elseif ( isset( $m[6] ) && $m[6] !== ''
1518 && $this->mOptions->getMagicISBNLinks()
1519 ) {
1520 # ISBN
1521 $isbn = $m[6];
1522 $space = self::SPACE_NOT_NL; # non-newline space
1523 $isbn = preg_replace( "/$space/", ' ', $isbn );
1524 $num = strtr( $isbn, [
1525 '-' => '',
1526 ' ' => '',
1527 'x' => 'X',
1528 ] );
1529 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1530 return $this->getLinkRenderer()->makeKnownLink(
1531 SpecialPage::getTitleFor( 'Booksources', $num ),
1532 "ISBN $isbn",
1533 [
1534 'class' => 'internal mw-magiclink-isbn',
1535 'title' => false // suppress title attribute
1536 ]
1537 );
1538 } else {
1539 return $m[0];
1540 }
1541 }
1542
1543 /**
1544 * Make a free external link, given a user-supplied URL
1545 *
1546 * @param string $url
1547 * @param int $numPostProto
1548 * The number of characters after the protocol.
1549 * @return string HTML
1550 * @private
1551 */
1552 public function makeFreeExternalLink( $url, $numPostProto ) {
1553 $trail = '';
1554
1555 # The characters '<' and '>' (which were escaped by
1556 # removeHTMLtags()) should not be included in
1557 # URLs, per RFC 2396.
1558 # Make &nbsp; terminate a URL as well (bug T84937)
1559 $m2 = [];
1560 if ( preg_match(
1561 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1562 $url,
1563 $m2,
1564 PREG_OFFSET_CAPTURE
1565 ) ) {
1566 $trail = substr( $url, $m2[0][1] ) . $trail;
1567 $url = substr( $url, 0, $m2[0][1] );
1568 }
1569
1570 # Move trailing punctuation to $trail
1571 $sep = ',;\.:!?';
1572 # If there is no left bracket, then consider right brackets fair game too
1573 if ( strpos( $url, '(' ) === false ) {
1574 $sep .= ')';
1575 }
1576
1577 $urlRev = strrev( $url );
1578 $numSepChars = strspn( $urlRev, $sep );
1579 # Don't break a trailing HTML entity by moving the ; into $trail
1580 # This is in hot code, so use substr_compare to avoid having to
1581 # create a new string object for the comparison
1582 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1583 # more optimization: instead of running preg_match with a $
1584 # anchor, which can be slow, do the match on the reversed
1585 # string starting at the desired offset.
1586 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1587 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1588 $numSepChars--;
1589 }
1590 }
1591 if ( $numSepChars ) {
1592 $trail = substr( $url, -$numSepChars ) . $trail;
1593 $url = substr( $url, 0, -$numSepChars );
1594 }
1595
1596 # Verify that we still have a real URL after trail removal, and
1597 # not just lone protocol
1598 if ( strlen( $trail ) >= $numPostProto ) {
1599 return $url . $trail;
1600 }
1601
1602 $url = Sanitizer::cleanUrl( $url );
1603
1604 # Is this an external image?
1605 $text = $this->maybeMakeExternalImage( $url );
1606 if ( $text === false ) {
1607 # Not an image, make a link
1608 $text = Linker::makeExternalLink( $url,
1609 $this->getConverterLanguage()->markNoConversion( $url, true ),
1610 true, 'free',
1611 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1612 # Register it in the output object...
1613 # Replace unnecessary URL escape codes with their equivalent characters
1614 $pasteurized = self::normalizeLinkUrl( $url );
1615 $this->mOutput->addExternalLink( $pasteurized );
1616 }
1617 return $text . $trail;
1618 }
1619
1620 /**
1621 * Parse headers and return html
1622 *
1623 * @private
1624 *
1625 * @param string $text
1626 *
1627 * @return string
1628 */
1629 public function doHeadings( $text ) {
1630 for ( $i = 6; $i >= 1; --$i ) {
1631 $h = str_repeat( '=', $i );
1632 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1633 }
1634 return $text;
1635 }
1636
1637 /**
1638 * Replace single quotes with HTML markup
1639 * @private
1640 *
1641 * @param string $text
1642 *
1643 * @return string The altered text
1644 */
1645 public function doAllQuotes( $text ) {
1646 $outtext = '';
1647 $lines = StringUtils::explode( "\n", $text );
1648 foreach ( $lines as $line ) {
1649 $outtext .= $this->doQuotes( $line ) . "\n";
1650 }
1651 $outtext = substr( $outtext, 0, -1 );
1652 return $outtext;
1653 }
1654
1655 /**
1656 * Helper function for doAllQuotes()
1657 *
1658 * @param string $text
1659 *
1660 * @return string
1661 */
1662 public function doQuotes( $text ) {
1663 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1664 $countarr = count( $arr );
1665 if ( $countarr == 1 ) {
1666 return $text;
1667 }
1668
1669 // First, do some preliminary work. This may shift some apostrophes from
1670 // being mark-up to being text. It also counts the number of occurrences
1671 // of bold and italics mark-ups.
1672 $numbold = 0;
1673 $numitalics = 0;
1674 for ( $i = 1; $i < $countarr; $i += 2 ) {
1675 $thislen = strlen( $arr[$i] );
1676 // If there are ever four apostrophes, assume the first is supposed to
1677 // be text, and the remaining three constitute mark-up for bold text.
1678 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1679 if ( $thislen == 4 ) {
1680 $arr[$i - 1] .= "'";
1681 $arr[$i] = "'''";
1682 $thislen = 3;
1683 } elseif ( $thislen > 5 ) {
1684 // If there are more than 5 apostrophes in a row, assume they're all
1685 // text except for the last 5.
1686 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1687 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1688 $arr[$i] = "'''''";
1689 $thislen = 5;
1690 }
1691 // Count the number of occurrences of bold and italics mark-ups.
1692 if ( $thislen == 2 ) {
1693 $numitalics++;
1694 } elseif ( $thislen == 3 ) {
1695 $numbold++;
1696 } elseif ( $thislen == 5 ) {
1697 $numitalics++;
1698 $numbold++;
1699 }
1700 }
1701
1702 // If there is an odd number of both bold and italics, it is likely
1703 // that one of the bold ones was meant to be an apostrophe followed
1704 // by italics. Which one we cannot know for certain, but it is more
1705 // likely to be one that has a single-letter word before it.
1706 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1707 $firstsingleletterword = -1;
1708 $firstmultiletterword = -1;
1709 $firstspace = -1;
1710 for ( $i = 1; $i < $countarr; $i += 2 ) {
1711 if ( strlen( $arr[$i] ) == 3 ) {
1712 $x1 = substr( $arr[$i - 1], -1 );
1713 $x2 = substr( $arr[$i - 1], -2, 1 );
1714 if ( $x1 === ' ' ) {
1715 if ( $firstspace == -1 ) {
1716 $firstspace = $i;
1717 }
1718 } elseif ( $x2 === ' ' ) {
1719 $firstsingleletterword = $i;
1720 // if $firstsingleletterword is set, we don't
1721 // look at the other options, so we can bail early.
1722 break;
1723 } else {
1724 if ( $firstmultiletterword == -1 ) {
1725 $firstmultiletterword = $i;
1726 }
1727 }
1728 }
1729 }
1730
1731 // If there is a single-letter word, use it!
1732 if ( $firstsingleletterword > -1 ) {
1733 $arr[$firstsingleletterword] = "''";
1734 $arr[$firstsingleletterword - 1] .= "'";
1735 } elseif ( $firstmultiletterword > -1 ) {
1736 // If not, but there's a multi-letter word, use that one.
1737 $arr[$firstmultiletterword] = "''";
1738 $arr[$firstmultiletterword - 1] .= "'";
1739 } elseif ( $firstspace > -1 ) {
1740 // ... otherwise use the first one that has neither.
1741 // (notice that it is possible for all three to be -1 if, for example,
1742 // there is only one pentuple-apostrophe in the line)
1743 $arr[$firstspace] = "''";
1744 $arr[$firstspace - 1] .= "'";
1745 }
1746 }
1747
1748 // Now let's actually convert our apostrophic mush to HTML!
1749 $output = '';
1750 $buffer = '';
1751 $state = '';
1752 $i = 0;
1753 foreach ( $arr as $r ) {
1754 if ( ( $i % 2 ) == 0 ) {
1755 if ( $state === 'both' ) {
1756 $buffer .= $r;
1757 } else {
1758 $output .= $r;
1759 }
1760 } else {
1761 $thislen = strlen( $r );
1762 if ( $thislen == 2 ) {
1763 if ( $state === 'i' ) {
1764 $output .= '</i>';
1765 $state = '';
1766 } elseif ( $state === 'bi' ) {
1767 $output .= '</i>';
1768 $state = 'b';
1769 } elseif ( $state === 'ib' ) {
1770 $output .= '</b></i><b>';
1771 $state = 'b';
1772 } elseif ( $state === 'both' ) {
1773 $output .= '<b><i>' . $buffer . '</i>';
1774 $state = 'b';
1775 } else { // $state can be 'b' or ''
1776 $output .= '<i>';
1777 $state .= 'i';
1778 }
1779 } elseif ( $thislen == 3 ) {
1780 if ( $state === 'b' ) {
1781 $output .= '</b>';
1782 $state = '';
1783 } elseif ( $state === 'bi' ) {
1784 $output .= '</i></b><i>';
1785 $state = 'i';
1786 } elseif ( $state === 'ib' ) {
1787 $output .= '</b>';
1788 $state = 'i';
1789 } elseif ( $state === 'both' ) {
1790 $output .= '<i><b>' . $buffer . '</b>';
1791 $state = 'i';
1792 } else { // $state can be 'i' or ''
1793 $output .= '<b>';
1794 $state .= 'b';
1795 }
1796 } elseif ( $thislen == 5 ) {
1797 if ( $state === 'b' ) {
1798 $output .= '</b><i>';
1799 $state = 'i';
1800 } elseif ( $state === 'i' ) {
1801 $output .= '</i><b>';
1802 $state = 'b';
1803 } elseif ( $state === 'bi' ) {
1804 $output .= '</i></b>';
1805 $state = '';
1806 } elseif ( $state === 'ib' ) {
1807 $output .= '</b></i>';
1808 $state = '';
1809 } elseif ( $state === 'both' ) {
1810 $output .= '<i><b>' . $buffer . '</b></i>';
1811 $state = '';
1812 } else { // ($state == '')
1813 $buffer = '';
1814 $state = 'both';
1815 }
1816 }
1817 }
1818 $i++;
1819 }
1820 // Now close all remaining tags. Notice that the order is important.
1821 if ( $state === 'b' || $state === 'ib' ) {
1822 $output .= '</b>';
1823 }
1824 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1825 $output .= '</i>';
1826 }
1827 if ( $state === 'bi' ) {
1828 $output .= '</b>';
1829 }
1830 // There might be lonely ''''', so make sure we have a buffer
1831 if ( $state === 'both' && $buffer ) {
1832 $output .= '<b><i>' . $buffer . '</i></b>';
1833 }
1834 return $output;
1835 }
1836
1837 /**
1838 * Replace external links (REL)
1839 *
1840 * Note: this is all very hackish and the order of execution matters a lot.
1841 * Make sure to run tests/parser/parserTests.php if you change this code.
1842 *
1843 * @private
1844 *
1845 * @param string $text
1846 *
1847 * @throws MWException
1848 * @return string
1849 */
1850 public function replaceExternalLinks( $text ) {
1851
1852 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1853 if ( $bits === false ) {
1854 throw new MWException( "PCRE needs to be compiled with "
1855 . "--enable-unicode-properties in order for MediaWiki to function" );
1856 }
1857 $s = array_shift( $bits );
1858
1859 $i = 0;
1860 while ( $i < count( $bits ) ) {
1861 $url = $bits[$i++];
1862 $i++; // protocol
1863 $text = $bits[$i++];
1864 $trail = $bits[$i++];
1865
1866 # The characters '<' and '>' (which were escaped by
1867 # removeHTMLtags()) should not be included in
1868 # URLs, per RFC 2396.
1869 $m2 = [];
1870 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1871 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1872 $url = substr( $url, 0, $m2[0][1] );
1873 }
1874
1875 # If the link text is an image URL, replace it with an <img> tag
1876 # This happened by accident in the original parser, but some people used it extensively
1877 $img = $this->maybeMakeExternalImage( $text );
1878 if ( $img !== false ) {
1879 $text = $img;
1880 }
1881
1882 $dtrail = '';
1883
1884 # Set linktype for CSS - if URL==text, link is essentially free
1885 $linktype = ( $text === $url ) ? 'free' : 'text';
1886
1887 # No link text, e.g. [http://domain.tld/some.link]
1888 if ( $text == '' ) {
1889 # Autonumber
1890 $langObj = $this->getTargetLanguage();
1891 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1892 $linktype = 'autonumber';
1893 } else {
1894 # Have link text, e.g. [http://domain.tld/some.link text]s
1895 # Check for trail
1896 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1897 }
1898
1899 $text = $this->getConverterLanguage()->markNoConversion( $text );
1900
1901 $url = Sanitizer::cleanUrl( $url );
1902
1903 # Use the encoded URL
1904 # This means that users can paste URLs directly into the text
1905 # Funny characters like ö aren't valid in URLs anyway
1906 # This was changed in August 2004
1907 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1908 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1909
1910 # Register link in the output object.
1911 # Replace unnecessary URL escape codes with the referenced character
1912 # This prevents spammers from hiding links from the filters
1913 $pasteurized = self::normalizeLinkUrl( $url );
1914 $this->mOutput->addExternalLink( $pasteurized );
1915 }
1916
1917 return $s;
1918 }
1919
1920 /**
1921 * Get the rel attribute for a particular external link.
1922 *
1923 * @since 1.21
1924 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1925 * nofollow if appropriate
1926 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1927 * @return string|null Rel attribute for $url
1928 */
1929 public static function getExternalLinkRel( $url = false, $title = null ) {
1930 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1931 $ns = $title ? $title->getNamespace() : false;
1932 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1933 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1934 ) {
1935 return 'nofollow';
1936 }
1937 return null;
1938 }
1939
1940 /**
1941 * Get an associative array of additional HTML attributes appropriate for a
1942 * particular external link. This currently may include rel => nofollow
1943 * (depending on configuration, namespace, and the URL's domain) and/or a
1944 * target attribute (depending on configuration).
1945 *
1946 * @param string $url URL to extract the domain from for rel =>
1947 * nofollow if appropriate
1948 * @return array Associative array of HTML attributes
1949 */
1950 public function getExternalLinkAttribs( $url ) {
1951 $attribs = [];
1952 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1953
1954 $target = $this->mOptions->getExternalLinkTarget();
1955 if ( $target ) {
1956 $attribs['target'] = $target;
1957 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1958 // T133507. New windows can navigate parent cross-origin.
1959 // Including noreferrer due to lacking browser
1960 // support of noopener. Eventually noreferrer should be removed.
1961 if ( $rel !== '' ) {
1962 $rel .= ' ';
1963 }
1964 $rel .= 'noreferrer noopener';
1965 }
1966 }
1967 $attribs['rel'] = $rel;
1968 return $attribs;
1969 }
1970
1971 /**
1972 * Replace unusual escape codes in a URL with their equivalent characters
1973 *
1974 * This generally follows the syntax defined in RFC 3986, with special
1975 * consideration for HTTP query strings.
1976 *
1977 * @param string $url
1978 * @return string
1979 */
1980 public static function normalizeLinkUrl( $url ) {
1981 # First, make sure unsafe characters are encoded
1982 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1983 function ( $m ) {
1984 return rawurlencode( $m[0] );
1985 },
1986 $url
1987 );
1988
1989 $ret = '';
1990 $end = strlen( $url );
1991
1992 # Fragment part - 'fragment'
1993 $start = strpos( $url, '#' );
1994 if ( $start !== false && $start < $end ) {
1995 $ret = self::normalizeUrlComponent(
1996 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1997 $end = $start;
1998 }
1999
2000 # Query part - 'query' minus &=+;
2001 $start = strpos( $url, '?' );
2002 if ( $start !== false && $start < $end ) {
2003 $ret = self::normalizeUrlComponent(
2004 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2005 $end = $start;
2006 }
2007
2008 # Scheme and path part - 'pchar'
2009 # (we assume no userinfo or encoded colons in the host)
2010 $ret = self::normalizeUrlComponent(
2011 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2012
2013 return $ret;
2014 }
2015
2016 private static function normalizeUrlComponent( $component, $unsafe ) {
2017 $callback = function ( $matches ) use ( $unsafe ) {
2018 $char = urldecode( $matches[0] );
2019 $ord = ord( $char );
2020 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2021 # Unescape it
2022 return $char;
2023 } else {
2024 # Leave it escaped, but use uppercase for a-f
2025 return strtoupper( $matches[0] );
2026 }
2027 };
2028 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2029 }
2030
2031 /**
2032 * make an image if it's allowed, either through the global
2033 * option, through the exception, or through the on-wiki whitelist
2034 *
2035 * @param string $url
2036 *
2037 * @return string
2038 */
2039 private function maybeMakeExternalImage( $url ) {
2040 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2041 $imagesexception = !empty( $imagesfrom );
2042 $text = false;
2043 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2044 if ( $imagesexception && is_array( $imagesfrom ) ) {
2045 $imagematch = false;
2046 foreach ( $imagesfrom as $match ) {
2047 if ( strpos( $url, $match ) === 0 ) {
2048 $imagematch = true;
2049 break;
2050 }
2051 }
2052 } elseif ( $imagesexception ) {
2053 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2054 } else {
2055 $imagematch = false;
2056 }
2057
2058 if ( $this->mOptions->getAllowExternalImages()
2059 || ( $imagesexception && $imagematch )
2060 ) {
2061 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2062 # Image found
2063 $text = Linker::makeExternalImage( $url );
2064 }
2065 }
2066 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2067 && preg_match( self::EXT_IMAGE_REGEX, $url )
2068 ) {
2069 $whitelist = explode(
2070 "\n",
2071 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2072 );
2073
2074 foreach ( $whitelist as $entry ) {
2075 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2076 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2077 continue;
2078 }
2079 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2080 # Image matches a whitelist entry
2081 $text = Linker::makeExternalImage( $url );
2082 break;
2083 }
2084 }
2085 }
2086 return $text;
2087 }
2088
2089 /**
2090 * Process [[ ]] wikilinks
2091 *
2092 * @param string $s
2093 *
2094 * @return string Processed text
2095 *
2096 * @private
2097 */
2098 public function replaceInternalLinks( $s ) {
2099 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2100 return $s;
2101 }
2102
2103 /**
2104 * Process [[ ]] wikilinks (RIL)
2105 * @param string $s
2106 * @throws MWException
2107 * @return LinkHolderArray
2108 *
2109 * @private
2110 */
2111 public function replaceInternalLinks2( &$s ) {
2112 global $wgExtraInterlanguageLinkPrefixes;
2113
2114 static $tc = false, $e1, $e1_img;
2115 # the % is needed to support urlencoded titles as well
2116 if ( !$tc ) {
2117 $tc = Title::legalChars() . '#%';
2118 # Match a link having the form [[namespace:link|alternate]]trail
2119 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2120 # Match cases where there is no "]]", which might still be images
2121 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2122 }
2123
2124 $holders = new LinkHolderArray( $this );
2125
2126 # split the entire text string on occurrences of [[
2127 $a = StringUtils::explode( '[[', ' ' . $s );
2128 # get the first element (all text up to first [[), and remove the space we added
2129 $s = $a->current();
2130 $a->next();
2131 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2132 $s = substr( $s, 1 );
2133
2134 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2135 $e2 = null;
2136 if ( $useLinkPrefixExtension ) {
2137 # Match the end of a line for a word that's not followed by whitespace,
2138 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2139 global $wgContLang;
2140 $charset = $wgContLang->linkPrefixCharset();
2141 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2142 }
2143
2144 if ( is_null( $this->mTitle ) ) {
2145 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2146 }
2147 $nottalk = !$this->mTitle->isTalkPage();
2148
2149 if ( $useLinkPrefixExtension ) {
2150 $m = [];
2151 if ( preg_match( $e2, $s, $m ) ) {
2152 $first_prefix = $m[2];
2153 } else {
2154 $first_prefix = false;
2155 }
2156 } else {
2157 $prefix = '';
2158 }
2159
2160 $useSubpages = $this->areSubpagesAllowed();
2161
2162 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2163 # Loop for each link
2164 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2165 // @codingStandardsIgnoreEnd
2166
2167 # Check for excessive memory usage
2168 if ( $holders->isBig() ) {
2169 # Too big
2170 # Do the existence check, replace the link holders and clear the array
2171 $holders->replace( $s );
2172 $holders->clear();
2173 }
2174
2175 if ( $useLinkPrefixExtension ) {
2176 if ( preg_match( $e2, $s, $m ) ) {
2177 $prefix = $m[2];
2178 $s = $m[1];
2179 } else {
2180 $prefix = '';
2181 }
2182 # first link
2183 if ( $first_prefix ) {
2184 $prefix = $first_prefix;
2185 $first_prefix = false;
2186 }
2187 }
2188
2189 $might_be_img = false;
2190
2191 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2192 $text = $m[2];
2193 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2194 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2195 # the real problem is with the $e1 regex
2196 # See T1500.
2197 # Still some problems for cases where the ] is meant to be outside punctuation,
2198 # and no image is in sight. See T4095.
2199 if ( $text !== ''
2200 && substr( $m[3], 0, 1 ) === ']'
2201 && strpos( $text, '[' ) !== false
2202 ) {
2203 $text .= ']'; # so that replaceExternalLinks($text) works later
2204 $m[3] = substr( $m[3], 1 );
2205 }
2206 # fix up urlencoded title texts
2207 if ( strpos( $m[1], '%' ) !== false ) {
2208 # Should anchors '#' also be rejected?
2209 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2210 }
2211 $trail = $m[3];
2212 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2213 # Invalid, but might be an image with a link in its caption
2214 $might_be_img = true;
2215 $text = $m[2];
2216 if ( strpos( $m[1], '%' ) !== false ) {
2217 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2218 }
2219 $trail = "";
2220 } else { # Invalid form; output directly
2221 $s .= $prefix . '[[' . $line;
2222 continue;
2223 }
2224
2225 $origLink = ltrim( $m[1], ' ' );
2226
2227 # Don't allow internal links to pages containing
2228 # PROTO: where PROTO is a valid URL protocol; these
2229 # should be external links.
2230 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2231 $s .= $prefix . '[[' . $line;
2232 continue;
2233 }
2234
2235 # Make subpage if necessary
2236 if ( $useSubpages ) {
2237 $link = $this->maybeDoSubpageLink( $origLink, $text );
2238 } else {
2239 $link = $origLink;
2240 }
2241
2242 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2243 if ( !$noforce ) {
2244 # Strip off leading ':'
2245 $link = substr( $link, 1 );
2246 }
2247
2248 $unstrip = $this->mStripState->unstripNoWiki( $link );
2249 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2250 if ( $nt === null ) {
2251 $s .= $prefix . '[[' . $line;
2252 continue;
2253 }
2254
2255 $ns = $nt->getNamespace();
2256 $iw = $nt->getInterwiki();
2257
2258 if ( $might_be_img ) { # if this is actually an invalid link
2259 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2260 $found = false;
2261 while ( true ) {
2262 # look at the next 'line' to see if we can close it there
2263 $a->next();
2264 $next_line = $a->current();
2265 if ( $next_line === false || $next_line === null ) {
2266 break;
2267 }
2268 $m = explode( ']]', $next_line, 3 );
2269 if ( count( $m ) == 3 ) {
2270 # the first ]] closes the inner link, the second the image
2271 $found = true;
2272 $text .= "[[{$m[0]}]]{$m[1]}";
2273 $trail = $m[2];
2274 break;
2275 } elseif ( count( $m ) == 2 ) {
2276 # if there's exactly one ]] that's fine, we'll keep looking
2277 $text .= "[[{$m[0]}]]{$m[1]}";
2278 } else {
2279 # if $next_line is invalid too, we need look no further
2280 $text .= '[[' . $next_line;
2281 break;
2282 }
2283 }
2284 if ( !$found ) {
2285 # we couldn't find the end of this imageLink, so output it raw
2286 # but don't ignore what might be perfectly normal links in the text we've examined
2287 $holders->merge( $this->replaceInternalLinks2( $text ) );
2288 $s .= "{$prefix}[[$link|$text";
2289 # note: no $trail, because without an end, there *is* no trail
2290 continue;
2291 }
2292 } else { # it's not an image, so output it raw
2293 $s .= "{$prefix}[[$link|$text";
2294 # note: no $trail, because without an end, there *is* no trail
2295 continue;
2296 }
2297 }
2298
2299 $wasblank = ( $text == '' );
2300 if ( $wasblank ) {
2301 $text = $link;
2302 } else {
2303 # T6598 madness. Handle the quotes only if they come from the alternate part
2304 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2305 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2306 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2307 $text = $this->doQuotes( $text );
2308 }
2309
2310 # Link not escaped by : , create the various objects
2311 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2312 # Interwikis
2313 if (
2314 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2315 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2316 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2317 )
2318 ) {
2319 # T26502: filter duplicates
2320 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2321 $this->mLangLinkLanguages[$iw] = true;
2322 $this->mOutput->addLanguageLink( $nt->getFullText() );
2323 }
2324
2325 $s = rtrim( $s . $prefix );
2326 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2327 continue;
2328 }
2329
2330 if ( $ns == NS_FILE ) {
2331 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2332 if ( $wasblank ) {
2333 # if no parameters were passed, $text
2334 # becomes something like "File:Foo.png",
2335 # which we don't want to pass on to the
2336 # image generator
2337 $text = '';
2338 } else {
2339 # recursively parse links inside the image caption
2340 # actually, this will parse them in any other parameters, too,
2341 # but it might be hard to fix that, and it doesn't matter ATM
2342 $text = $this->replaceExternalLinks( $text );
2343 $holders->merge( $this->replaceInternalLinks2( $text ) );
2344 }
2345 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2346 $s .= $prefix . $this->armorLinks(
2347 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2348 continue;
2349 }
2350 } elseif ( $ns == NS_CATEGORY ) {
2351 $s = rtrim( $s . "\n" ); # T2087
2352
2353 if ( $wasblank ) {
2354 $sortkey = $this->getDefaultSort();
2355 } else {
2356 $sortkey = $text;
2357 }
2358 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2359 $sortkey = str_replace( "\n", '', $sortkey );
2360 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2361 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2362
2363 /**
2364 * Strip the whitespace Category links produce, see T2087
2365 */
2366 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2367
2368 continue;
2369 }
2370 }
2371
2372 # Self-link checking. For some languages, variants of the title are checked in
2373 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2374 # for linking to a different variant.
2375 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2376 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2377 continue;
2378 }
2379
2380 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2381 # @todo FIXME: Should do batch file existence checks, see comment below
2382 if ( $ns == NS_MEDIA ) {
2383 # Give extensions a chance to select the file revision for us
2384 $options = [];
2385 $descQuery = false;
2386 Hooks::run( 'BeforeParserFetchFileAndTitle',
2387 [ $this, $nt, &$options, &$descQuery ] );
2388 # Fetch and register the file (file title may be different via hooks)
2389 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2390 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2391 $s .= $prefix . $this->armorLinks(
2392 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2393 continue;
2394 }
2395
2396 # Some titles, such as valid special pages or files in foreign repos, should
2397 # be shown as bluelinks even though they're not included in the page table
2398 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2399 # batch file existence checks for NS_FILE and NS_MEDIA
2400 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2401 $this->mOutput->addLink( $nt );
2402 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2403 } else {
2404 # Links will be added to the output link list after checking
2405 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2406 }
2407 }
2408 return $holders;
2409 }
2410
2411 /**
2412 * Render a forced-blue link inline; protect against double expansion of
2413 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2414 * Since this little disaster has to split off the trail text to avoid
2415 * breaking URLs in the following text without breaking trails on the
2416 * wiki links, it's been made into a horrible function.
2417 *
2418 * @param Title $nt
2419 * @param string $text
2420 * @param string $trail
2421 * @param string $prefix
2422 * @return string HTML-wikitext mix oh yuck
2423 */
2424 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2425 list( $inside, $trail ) = Linker::splitTrail( $trail );
2426
2427 if ( $text == '' ) {
2428 $text = htmlspecialchars( $nt->getPrefixedText() );
2429 }
2430
2431 $link = $this->getLinkRenderer()->makeKnownLink(
2432 $nt, new HtmlArmor( "$prefix$text$inside" )
2433 );
2434
2435 return $this->armorLinks( $link ) . $trail;
2436 }
2437
2438 /**
2439 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2440 * going to go through further parsing steps before inline URL expansion.
2441 *
2442 * Not needed quite as much as it used to be since free links are a bit
2443 * more sensible these days. But bracketed links are still an issue.
2444 *
2445 * @param string $text More-or-less HTML
2446 * @return string Less-or-more HTML with NOPARSE bits
2447 */
2448 public function armorLinks( $text ) {
2449 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2450 self::MARKER_PREFIX . "NOPARSE$1", $text );
2451 }
2452
2453 /**
2454 * Return true if subpage links should be expanded on this page.
2455 * @return bool
2456 */
2457 public function areSubpagesAllowed() {
2458 # Some namespaces don't allow subpages
2459 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2460 }
2461
2462 /**
2463 * Handle link to subpage if necessary
2464 *
2465 * @param string $target The source of the link
2466 * @param string &$text The link text, modified as necessary
2467 * @return string The full name of the link
2468 * @private
2469 */
2470 public function maybeDoSubpageLink( $target, &$text ) {
2471 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2472 }
2473
2474 /**
2475 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2476 *
2477 * @param string $text
2478 * @param bool $linestart Whether or not this is at the start of a line.
2479 * @private
2480 * @return string The lists rendered as HTML
2481 */
2482 public function doBlockLevels( $text, $linestart ) {
2483 return BlockLevelPass::doBlockLevels( $text, $linestart );
2484 }
2485
2486 /**
2487 * Return value of a magic variable (like PAGENAME)
2488 *
2489 * @private
2490 *
2491 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2492 * @param bool|PPFrame $frame
2493 *
2494 * @throws MWException
2495 * @return string
2496 */
2497 public function getVariableValue( $index, $frame = false ) {
2498 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2499 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2500
2501 if ( is_null( $this->mTitle ) ) {
2502 // If no title set, bad things are going to happen
2503 // later. Title should always be set since this
2504 // should only be called in the middle of a parse
2505 // operation (but the unit-tests do funky stuff)
2506 throw new MWException( __METHOD__ . ' Should only be '
2507 . ' called while parsing (no title set)' );
2508 }
2509
2510 // Avoid PHP 7.1 warning from passing $this by reference
2511 $parser = $this;
2512
2513 /**
2514 * Some of these require message or data lookups and can be
2515 * expensive to check many times.
2516 */
2517 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2518 if ( isset( $this->mVarCache[$index] ) ) {
2519 return $this->mVarCache[$index];
2520 }
2521 }
2522
2523 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2524 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2525
2526 $pageLang = $this->getFunctionLang();
2527
2528 switch ( $index ) {
2529 case '!':
2530 $value = '|';
2531 break;
2532 case 'currentmonth':
2533 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2534 break;
2535 case 'currentmonth1':
2536 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2537 break;
2538 case 'currentmonthname':
2539 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2540 break;
2541 case 'currentmonthnamegen':
2542 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2543 break;
2544 case 'currentmonthabbrev':
2545 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2546 break;
2547 case 'currentday':
2548 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2549 break;
2550 case 'currentday2':
2551 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2552 break;
2553 case 'localmonth':
2554 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2555 break;
2556 case 'localmonth1':
2557 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2558 break;
2559 case 'localmonthname':
2560 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2561 break;
2562 case 'localmonthnamegen':
2563 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2564 break;
2565 case 'localmonthabbrev':
2566 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2567 break;
2568 case 'localday':
2569 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2570 break;
2571 case 'localday2':
2572 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2573 break;
2574 case 'pagename':
2575 $value = wfEscapeWikiText( $this->mTitle->getText() );
2576 break;
2577 case 'pagenamee':
2578 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2579 break;
2580 case 'fullpagename':
2581 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2582 break;
2583 case 'fullpagenamee':
2584 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2585 break;
2586 case 'subpagename':
2587 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2588 break;
2589 case 'subpagenamee':
2590 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2591 break;
2592 case 'rootpagename':
2593 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2594 break;
2595 case 'rootpagenamee':
2596 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2597 ' ',
2598 '_',
2599 $this->mTitle->getRootText()
2600 ) ) );
2601 break;
2602 case 'basepagename':
2603 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2604 break;
2605 case 'basepagenamee':
2606 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2607 ' ',
2608 '_',
2609 $this->mTitle->getBaseText()
2610 ) ) );
2611 break;
2612 case 'talkpagename':
2613 if ( $this->mTitle->canTalk() ) {
2614 $talkPage = $this->mTitle->getTalkPage();
2615 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2616 } else {
2617 $value = '';
2618 }
2619 break;
2620 case 'talkpagenamee':
2621 if ( $this->mTitle->canTalk() ) {
2622 $talkPage = $this->mTitle->getTalkPage();
2623 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2624 } else {
2625 $value = '';
2626 }
2627 break;
2628 case 'subjectpagename':
2629 $subjPage = $this->mTitle->getSubjectPage();
2630 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2631 break;
2632 case 'subjectpagenamee':
2633 $subjPage = $this->mTitle->getSubjectPage();
2634 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2635 break;
2636 case 'pageid': // requested in T25427
2637 $pageid = $this->getTitle()->getArticleID();
2638 if ( $pageid == 0 ) {
2639 # 0 means the page doesn't exist in the database,
2640 # which means the user is previewing a new page.
2641 # The vary-revision flag must be set, because the magic word
2642 # will have a different value once the page is saved.
2643 $this->mOutput->setFlag( 'vary-revision' );
2644 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2645 }
2646 $value = $pageid ? $pageid : null;
2647 break;
2648 case 'revisionid':
2649 # Let the edit saving system know we should parse the page
2650 # *after* a revision ID has been assigned.
2651 $this->mOutput->setFlag( 'vary-revision-id' );
2652 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2653 $value = $this->mRevisionId;
2654 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2655 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2656 $this->mOutput->setSpeculativeRevIdUsed( $value );
2657 }
2658 break;
2659 case 'revisionday':
2660 # Let the edit saving system know we should parse the page
2661 # *after* a revision ID has been assigned. This is for null edits.
2662 $this->mOutput->setFlag( 'vary-revision' );
2663 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2664 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2665 break;
2666 case 'revisionday2':
2667 # Let the edit saving system know we should parse the page
2668 # *after* a revision ID has been assigned. This is for null edits.
2669 $this->mOutput->setFlag( 'vary-revision' );
2670 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2671 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2672 break;
2673 case 'revisionmonth':
2674 # Let the edit saving system know we should parse the page
2675 # *after* a revision ID has been assigned. This is for null edits.
2676 $this->mOutput->setFlag( 'vary-revision' );
2677 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2678 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2679 break;
2680 case 'revisionmonth1':
2681 # Let the edit saving system know we should parse the page
2682 # *after* a revision ID has been assigned. This is for null edits.
2683 $this->mOutput->setFlag( 'vary-revision' );
2684 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2685 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2686 break;
2687 case 'revisionyear':
2688 # Let the edit saving system know we should parse the page
2689 # *after* a revision ID has been assigned. This is for null edits.
2690 $this->mOutput->setFlag( 'vary-revision' );
2691 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2692 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2693 break;
2694 case 'revisiontimestamp':
2695 # Let the edit saving system know we should parse the page
2696 # *after* a revision ID has been assigned. This is for null edits.
2697 $this->mOutput->setFlag( 'vary-revision' );
2698 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2699 $value = $this->getRevisionTimestamp();
2700 break;
2701 case 'revisionuser':
2702 # Let the edit saving system know we should parse the page
2703 # *after* a revision ID has been assigned for null edits.
2704 $this->mOutput->setFlag( 'vary-user' );
2705 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2706 $value = $this->getRevisionUser();
2707 break;
2708 case 'revisionsize':
2709 $value = $this->getRevisionSize();
2710 break;
2711 case 'namespace':
2712 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2713 break;
2714 case 'namespacee':
2715 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2716 break;
2717 case 'namespacenumber':
2718 $value = $this->mTitle->getNamespace();
2719 break;
2720 case 'talkspace':
2721 $value = $this->mTitle->canTalk()
2722 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2723 : '';
2724 break;
2725 case 'talkspacee':
2726 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2727 break;
2728 case 'subjectspace':
2729 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2730 break;
2731 case 'subjectspacee':
2732 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2733 break;
2734 case 'currentdayname':
2735 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2736 break;
2737 case 'currentyear':
2738 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2739 break;
2740 case 'currenttime':
2741 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2742 break;
2743 case 'currenthour':
2744 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2745 break;
2746 case 'currentweek':
2747 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2748 # int to remove the padding
2749 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2750 break;
2751 case 'currentdow':
2752 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2753 break;
2754 case 'localdayname':
2755 $value = $pageLang->getWeekdayName(
2756 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2757 );
2758 break;
2759 case 'localyear':
2760 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2761 break;
2762 case 'localtime':
2763 $value = $pageLang->time(
2764 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2765 false,
2766 false
2767 );
2768 break;
2769 case 'localhour':
2770 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2771 break;
2772 case 'localweek':
2773 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2774 # int to remove the padding
2775 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2776 break;
2777 case 'localdow':
2778 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2779 break;
2780 case 'numberofarticles':
2781 $value = $pageLang->formatNum( SiteStats::articles() );
2782 break;
2783 case 'numberoffiles':
2784 $value = $pageLang->formatNum( SiteStats::images() );
2785 break;
2786 case 'numberofusers':
2787 $value = $pageLang->formatNum( SiteStats::users() );
2788 break;
2789 case 'numberofactiveusers':
2790 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2791 break;
2792 case 'numberofpages':
2793 $value = $pageLang->formatNum( SiteStats::pages() );
2794 break;
2795 case 'numberofadmins':
2796 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2797 break;
2798 case 'numberofedits':
2799 $value = $pageLang->formatNum( SiteStats::edits() );
2800 break;
2801 case 'currenttimestamp':
2802 $value = wfTimestamp( TS_MW, $ts );
2803 break;
2804 case 'localtimestamp':
2805 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2806 break;
2807 case 'currentversion':
2808 $value = SpecialVersion::getVersion();
2809 break;
2810 case 'articlepath':
2811 return $wgArticlePath;
2812 case 'sitename':
2813 return $wgSitename;
2814 case 'server':
2815 return $wgServer;
2816 case 'servername':
2817 return $wgServerName;
2818 case 'scriptpath':
2819 return $wgScriptPath;
2820 case 'stylepath':
2821 return $wgStylePath;
2822 case 'directionmark':
2823 return $pageLang->getDirMark();
2824 case 'contentlanguage':
2825 global $wgLanguageCode;
2826 return $wgLanguageCode;
2827 case 'pagelanguage':
2828 $value = $pageLang->getCode();
2829 break;
2830 case 'cascadingsources':
2831 $value = CoreParserFunctions::cascadingsources( $this );
2832 break;
2833 default:
2834 $ret = null;
2835 Hooks::run(
2836 'ParserGetVariableValueSwitch',
2837 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2838 );
2839
2840 return $ret;
2841 }
2842
2843 if ( $index ) {
2844 $this->mVarCache[$index] = $value;
2845 }
2846
2847 return $value;
2848 }
2849
2850 /**
2851 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2852 *
2853 * @private
2854 */
2855 public function initialiseVariables() {
2856 $variableIDs = MagicWord::getVariableIDs();
2857 $substIDs = MagicWord::getSubstIDs();
2858
2859 $this->mVariables = new MagicWordArray( $variableIDs );
2860 $this->mSubstWords = new MagicWordArray( $substIDs );
2861 }
2862
2863 /**
2864 * Preprocess some wikitext and return the document tree.
2865 * This is the ghost of replace_variables().
2866 *
2867 * @param string $text The text to parse
2868 * @param int $flags Bitwise combination of:
2869 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2870 * included. Default is to assume a direct page view.
2871 *
2872 * The generated DOM tree must depend only on the input text and the flags.
2873 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2874 *
2875 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2876 * change in the DOM tree for a given text, must be passed through the section identifier
2877 * in the section edit link and thus back to extractSections().
2878 *
2879 * The output of this function is currently only cached in process memory, but a persistent
2880 * cache may be implemented at a later date which takes further advantage of these strict
2881 * dependency requirements.
2882 *
2883 * @return PPNode
2884 */
2885 public function preprocessToDom( $text, $flags = 0 ) {
2886 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2887 return $dom;
2888 }
2889
2890 /**
2891 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2892 *
2893 * @param string $s
2894 *
2895 * @return array
2896 */
2897 public static function splitWhitespace( $s ) {
2898 $ltrimmed = ltrim( $s );
2899 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2900 $trimmed = rtrim( $ltrimmed );
2901 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2902 if ( $diff > 0 ) {
2903 $w2 = substr( $ltrimmed, -$diff );
2904 } else {
2905 $w2 = '';
2906 }
2907 return [ $w1, $trimmed, $w2 ];
2908 }
2909
2910 /**
2911 * Replace magic variables, templates, and template arguments
2912 * with the appropriate text. Templates are substituted recursively,
2913 * taking care to avoid infinite loops.
2914 *
2915 * Note that the substitution depends on value of $mOutputType:
2916 * self::OT_WIKI: only {{subst:}} templates
2917 * self::OT_PREPROCESS: templates but not extension tags
2918 * self::OT_HTML: all templates and extension tags
2919 *
2920 * @param string $text The text to transform
2921 * @param bool|PPFrame $frame Object describing the arguments passed to the
2922 * template. Arguments may also be provided as an associative array, as
2923 * was the usual case before MW1.12. Providing arguments this way may be
2924 * useful for extensions wishing to perform variable replacement
2925 * explicitly.
2926 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2927 * double-brace expansion.
2928 * @return string
2929 */
2930 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2931 # Is there any text? Also, Prevent too big inclusions!
2932 $textSize = strlen( $text );
2933 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2934 return $text;
2935 }
2936
2937 if ( $frame === false ) {
2938 $frame = $this->getPreprocessor()->newFrame();
2939 } elseif ( !( $frame instanceof PPFrame ) ) {
2940 wfDebug( __METHOD__ . " called using plain parameters instead of "
2941 . "a PPFrame instance. Creating custom frame.\n" );
2942 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2943 }
2944
2945 $dom = $this->preprocessToDom( $text );
2946 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2947 $text = $frame->expand( $dom, $flags );
2948
2949 return $text;
2950 }
2951
2952 /**
2953 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2954 *
2955 * @param array $args
2956 *
2957 * @return array
2958 */
2959 public static function createAssocArgs( $args ) {
2960 $assocArgs = [];
2961 $index = 1;
2962 foreach ( $args as $arg ) {
2963 $eqpos = strpos( $arg, '=' );
2964 if ( $eqpos === false ) {
2965 $assocArgs[$index++] = $arg;
2966 } else {
2967 $name = trim( substr( $arg, 0, $eqpos ) );
2968 $value = trim( substr( $arg, $eqpos + 1 ) );
2969 if ( $value === false ) {
2970 $value = '';
2971 }
2972 if ( $name !== false ) {
2973 $assocArgs[$name] = $value;
2974 }
2975 }
2976 }
2977
2978 return $assocArgs;
2979 }
2980
2981 /**
2982 * Warn the user when a parser limitation is reached
2983 * Will warn at most once the user per limitation type
2984 *
2985 * The results are shown during preview and run through the Parser (See EditPage.php)
2986 *
2987 * @param string $limitationType Should be one of:
2988 * 'expensive-parserfunction' (corresponding messages:
2989 * 'expensive-parserfunction-warning',
2990 * 'expensive-parserfunction-category')
2991 * 'post-expand-template-argument' (corresponding messages:
2992 * 'post-expand-template-argument-warning',
2993 * 'post-expand-template-argument-category')
2994 * 'post-expand-template-inclusion' (corresponding messages:
2995 * 'post-expand-template-inclusion-warning',
2996 * 'post-expand-template-inclusion-category')
2997 * 'node-count-exceeded' (corresponding messages:
2998 * 'node-count-exceeded-warning',
2999 * 'node-count-exceeded-category')
3000 * 'expansion-depth-exceeded' (corresponding messages:
3001 * 'expansion-depth-exceeded-warning',
3002 * 'expansion-depth-exceeded-category')
3003 * @param string|int|null $current Current value
3004 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3005 * exceeded, provide the values (optional)
3006 */
3007 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3008 # does no harm if $current and $max are present but are unnecessary for the message
3009 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3010 # only during preview, and that would split the parser cache unnecessarily.
3011 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3012 ->text();
3013 $this->mOutput->addWarning( $warning );
3014 $this->addTrackingCategory( "$limitationType-category" );
3015 }
3016
3017 /**
3018 * Return the text of a template, after recursively
3019 * replacing any variables or templates within the template.
3020 *
3021 * @param array $piece The parts of the template
3022 * $piece['title']: the title, i.e. the part before the |
3023 * $piece['parts']: the parameter array
3024 * $piece['lineStart']: whether the brace was at the start of a line
3025 * @param PPFrame $frame The current frame, contains template arguments
3026 * @throws Exception
3027 * @return string The text of the template
3028 */
3029 public function braceSubstitution( $piece, $frame ) {
3030
3031 // Flags
3032
3033 // $text has been filled
3034 $found = false;
3035 // wiki markup in $text should be escaped
3036 $nowiki = false;
3037 // $text is HTML, armour it against wikitext transformation
3038 $isHTML = false;
3039 // Force interwiki transclusion to be done in raw mode not rendered
3040 $forceRawInterwiki = false;
3041 // $text is a DOM node needing expansion in a child frame
3042 $isChildObj = false;
3043 // $text is a DOM node needing expansion in the current frame
3044 $isLocalObj = false;
3045
3046 # Title object, where $text came from
3047 $title = false;
3048
3049 # $part1 is the bit before the first |, and must contain only title characters.
3050 # Various prefixes will be stripped from it later.
3051 $titleWithSpaces = $frame->expand( $piece['title'] );
3052 $part1 = trim( $titleWithSpaces );
3053 $titleText = false;
3054
3055 # Original title text preserved for various purposes
3056 $originalTitle = $part1;
3057
3058 # $args is a list of argument nodes, starting from index 0, not including $part1
3059 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3060 # below won't work b/c this $args isn't an object
3061 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3062
3063 $profileSection = null; // profile templates
3064
3065 # SUBST
3066 if ( !$found ) {
3067 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3068
3069 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3070 # Decide whether to expand template or keep wikitext as-is.
3071 if ( $this->ot['wiki'] ) {
3072 if ( $substMatch === false ) {
3073 $literal = true; # literal when in PST with no prefix
3074 } else {
3075 $literal = false; # expand when in PST with subst: or safesubst:
3076 }
3077 } else {
3078 if ( $substMatch == 'subst' ) {
3079 $literal = true; # literal when not in PST with plain subst:
3080 } else {
3081 $literal = false; # expand when not in PST with safesubst: or no prefix
3082 }
3083 }
3084 if ( $literal ) {
3085 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3086 $isLocalObj = true;
3087 $found = true;
3088 }
3089 }
3090
3091 # Variables
3092 if ( !$found && $args->getLength() == 0 ) {
3093 $id = $this->mVariables->matchStartToEnd( $part1 );
3094 if ( $id !== false ) {
3095 $text = $this->getVariableValue( $id, $frame );
3096 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3097 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3098 }
3099 $found = true;
3100 }
3101 }
3102
3103 # MSG, MSGNW and RAW
3104 if ( !$found ) {
3105 # Check for MSGNW:
3106 $mwMsgnw = MagicWord::get( 'msgnw' );
3107 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3108 $nowiki = true;
3109 } else {
3110 # Remove obsolete MSG:
3111 $mwMsg = MagicWord::get( 'msg' );
3112 $mwMsg->matchStartAndRemove( $part1 );
3113 }
3114
3115 # Check for RAW:
3116 $mwRaw = MagicWord::get( 'raw' );
3117 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3118 $forceRawInterwiki = true;
3119 }
3120 }
3121
3122 # Parser functions
3123 if ( !$found ) {
3124 $colonPos = strpos( $part1, ':' );
3125 if ( $colonPos !== false ) {
3126 $func = substr( $part1, 0, $colonPos );
3127 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3128 $argsLength = $args->getLength();
3129 for ( $i = 0; $i < $argsLength; $i++ ) {
3130 $funcArgs[] = $args->item( $i );
3131 }
3132 try {
3133 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3134 } catch ( Exception $ex ) {
3135 throw $ex;
3136 }
3137
3138 # The interface for parser functions allows for extracting
3139 # flags into the local scope. Extract any forwarded flags
3140 # here.
3141 extract( $result );
3142 }
3143 }
3144
3145 # Finish mangling title and then check for loops.
3146 # Set $title to a Title object and $titleText to the PDBK
3147 if ( !$found ) {
3148 $ns = NS_TEMPLATE;
3149 # Split the title into page and subpage
3150 $subpage = '';
3151 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3152 if ( $part1 !== $relative ) {
3153 $part1 = $relative;
3154 $ns = $this->mTitle->getNamespace();
3155 }
3156 $title = Title::newFromText( $part1, $ns );
3157 if ( $title ) {
3158 $titleText = $title->getPrefixedText();
3159 # Check for language variants if the template is not found
3160 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3161 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3162 }
3163 # Do recursion depth check
3164 $limit = $this->mOptions->getMaxTemplateDepth();
3165 if ( $frame->depth >= $limit ) {
3166 $found = true;
3167 $text = '<span class="error">'
3168 . wfMessage( 'parser-template-recursion-depth-warning' )
3169 ->numParams( $limit )->inContentLanguage()->text()
3170 . '</span>';
3171 }
3172 }
3173 }
3174
3175 # Load from database
3176 if ( !$found && $title ) {
3177 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3178 if ( !$title->isExternal() ) {
3179 if ( $title->isSpecialPage()
3180 && $this->mOptions->getAllowSpecialInclusion()
3181 && $this->ot['html']
3182 ) {
3183 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3184 // Pass the template arguments as URL parameters.
3185 // "uselang" will have no effect since the Language object
3186 // is forced to the one defined in ParserOptions.
3187 $pageArgs = [];
3188 $argsLength = $args->getLength();
3189 for ( $i = 0; $i < $argsLength; $i++ ) {
3190 $bits = $args->item( $i )->splitArg();
3191 if ( strval( $bits['index'] ) === '' ) {
3192 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3193 $value = trim( $frame->expand( $bits['value'] ) );
3194 $pageArgs[$name] = $value;
3195 }
3196 }
3197
3198 // Create a new context to execute the special page
3199 $context = new RequestContext;
3200 $context->setTitle( $title );
3201 $context->setRequest( new FauxRequest( $pageArgs ) );
3202 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3203 $context->setUser( $this->getUser() );
3204 } else {
3205 // If this page is cached, then we better not be per user.
3206 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3207 }
3208 $context->setLanguage( $this->mOptions->getUserLangObj() );
3209 $ret = SpecialPageFactory::capturePath(
3210 $title, $context, $this->getLinkRenderer() );
3211 if ( $ret ) {
3212 $text = $context->getOutput()->getHTML();
3213 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3214 $found = true;
3215 $isHTML = true;
3216 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3217 $this->mOutput->updateRuntimeAdaptiveExpiry(
3218 $specialPage->maxIncludeCacheTime()
3219 );
3220 }
3221 }
3222 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3223 $found = false; # access denied
3224 wfDebug( __METHOD__ . ": template inclusion denied for " .
3225 $title->getPrefixedDBkey() . "\n" );
3226 } else {
3227 list( $text, $title ) = $this->getTemplateDom( $title );
3228 if ( $text !== false ) {
3229 $found = true;
3230 $isChildObj = true;
3231 }
3232 }
3233
3234 # If the title is valid but undisplayable, make a link to it
3235 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3236 $text = "[[:$titleText]]";
3237 $found = true;
3238 }
3239 } elseif ( $title->isTrans() ) {
3240 # Interwiki transclusion
3241 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3242 $text = $this->interwikiTransclude( $title, 'render' );
3243 $isHTML = true;
3244 } else {
3245 $text = $this->interwikiTransclude( $title, 'raw' );
3246 # Preprocess it like a template
3247 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3248 $isChildObj = true;
3249 }
3250 $found = true;
3251 }
3252
3253 # Do infinite loop check
3254 # This has to be done after redirect resolution to avoid infinite loops via redirects
3255 if ( !$frame->loopCheck( $title ) ) {
3256 $found = true;
3257 $text = '<span class="error">'
3258 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3259 . '</span>';
3260 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3261 }
3262 }
3263
3264 # If we haven't found text to substitute by now, we're done
3265 # Recover the source wikitext and return it
3266 if ( !$found ) {
3267 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3268 if ( $profileSection ) {
3269 $this->mProfiler->scopedProfileOut( $profileSection );
3270 }
3271 return [ 'object' => $text ];
3272 }
3273
3274 # Expand DOM-style return values in a child frame
3275 if ( $isChildObj ) {
3276 # Clean up argument array
3277 $newFrame = $frame->newChild( $args, $title );
3278
3279 if ( $nowiki ) {
3280 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3281 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3282 # Expansion is eligible for the empty-frame cache
3283 $text = $newFrame->cachedExpand( $titleText, $text );
3284 } else {
3285 # Uncached expansion
3286 $text = $newFrame->expand( $text );
3287 }
3288 }
3289 if ( $isLocalObj && $nowiki ) {
3290 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3291 $isLocalObj = false;
3292 }
3293
3294 if ( $profileSection ) {
3295 $this->mProfiler->scopedProfileOut( $profileSection );
3296 }
3297
3298 # Replace raw HTML by a placeholder
3299 if ( $isHTML ) {
3300 $text = $this->insertStripItem( $text );
3301 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3302 # Escape nowiki-style return values
3303 $text = wfEscapeWikiText( $text );
3304 } elseif ( is_string( $text )
3305 && !$piece['lineStart']
3306 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3307 ) {
3308 # T2529: if the template begins with a table or block-level
3309 # element, it should be treated as beginning a new line.
3310 # This behavior is somewhat controversial.
3311 $text = "\n" . $text;
3312 }
3313
3314 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3315 # Error, oversize inclusion
3316 if ( $titleText !== false ) {
3317 # Make a working, properly escaped link if possible (T25588)
3318 $text = "[[:$titleText]]";
3319 } else {
3320 # This will probably not be a working link, but at least it may
3321 # provide some hint of where the problem is
3322 preg_replace( '/^:/', '', $originalTitle );
3323 $text = "[[:$originalTitle]]";
3324 }
3325 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3326 . 'post-expand include size too large -->' );
3327 $this->limitationWarn( 'post-expand-template-inclusion' );
3328 }
3329
3330 if ( $isLocalObj ) {
3331 $ret = [ 'object' => $text ];
3332 } else {
3333 $ret = [ 'text' => $text ];
3334 }
3335
3336 return $ret;
3337 }
3338
3339 /**
3340 * Call a parser function and return an array with text and flags.
3341 *
3342 * The returned array will always contain a boolean 'found', indicating
3343 * whether the parser function was found or not. It may also contain the
3344 * following:
3345 * text: string|object, resulting wikitext or PP DOM object
3346 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3347 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3348 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3349 * nowiki: bool, wiki markup in $text should be escaped
3350 *
3351 * @since 1.21
3352 * @param PPFrame $frame The current frame, contains template arguments
3353 * @param string $function Function name
3354 * @param array $args Arguments to the function
3355 * @throws MWException
3356 * @return array
3357 */
3358 public function callParserFunction( $frame, $function, array $args = [] ) {
3359 global $wgContLang;
3360
3361 # Case sensitive functions
3362 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3363 $function = $this->mFunctionSynonyms[1][$function];
3364 } else {
3365 # Case insensitive functions
3366 $function = $wgContLang->lc( $function );
3367 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3368 $function = $this->mFunctionSynonyms[0][$function];
3369 } else {
3370 return [ 'found' => false ];
3371 }
3372 }
3373
3374 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3375
3376 # Workaround for PHP bug 35229 and similar
3377 if ( !is_callable( $callback ) ) {
3378 throw new MWException( "Tag hook for $function is not callable\n" );
3379 }
3380
3381 // Avoid PHP 7.1 warning from passing $this by reference
3382 $parser = $this;
3383
3384 $allArgs = [ &$parser ];
3385 if ( $flags & self::SFH_OBJECT_ARGS ) {
3386 # Convert arguments to PPNodes and collect for appending to $allArgs
3387 $funcArgs = [];
3388 foreach ( $args as $k => $v ) {
3389 if ( $v instanceof PPNode || $k === 0 ) {
3390 $funcArgs[] = $v;
3391 } else {
3392 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3393 }
3394 }
3395
3396 # Add a frame parameter, and pass the arguments as an array
3397 $allArgs[] = $frame;
3398 $allArgs[] = $funcArgs;
3399 } else {
3400 # Convert arguments to plain text and append to $allArgs
3401 foreach ( $args as $k => $v ) {
3402 if ( $v instanceof PPNode ) {
3403 $allArgs[] = trim( $frame->expand( $v ) );
3404 } elseif ( is_int( $k ) && $k >= 0 ) {
3405 $allArgs[] = trim( $v );
3406 } else {
3407 $allArgs[] = trim( "$k=$v" );
3408 }
3409 }
3410 }
3411
3412 $result = call_user_func_array( $callback, $allArgs );
3413
3414 # The interface for function hooks allows them to return a wikitext
3415 # string or an array containing the string and any flags. This mungs
3416 # things around to match what this method should return.
3417 if ( !is_array( $result ) ) {
3418 $result =[
3419 'found' => true,
3420 'text' => $result,
3421 ];
3422 } else {
3423 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3424 $result['text'] = $result[0];
3425 }
3426 unset( $result[0] );
3427 $result += [
3428 'found' => true,
3429 ];
3430 }
3431
3432 $noparse = true;
3433 $preprocessFlags = 0;
3434 if ( isset( $result['noparse'] ) ) {
3435 $noparse = $result['noparse'];
3436 }
3437 if ( isset( $result['preprocessFlags'] ) ) {
3438 $preprocessFlags = $result['preprocessFlags'];
3439 }
3440
3441 if ( !$noparse ) {
3442 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3443 $result['isChildObj'] = true;
3444 }
3445
3446 return $result;
3447 }
3448
3449 /**
3450 * Get the semi-parsed DOM representation of a template with a given title,
3451 * and its redirect destination title. Cached.
3452 *
3453 * @param Title $title
3454 *
3455 * @return array
3456 */
3457 public function getTemplateDom( $title ) {
3458 $cacheTitle = $title;
3459 $titleText = $title->getPrefixedDBkey();
3460
3461 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3462 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3463 $title = Title::makeTitle( $ns, $dbk );
3464 $titleText = $title->getPrefixedDBkey();
3465 }
3466 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3467 return [ $this->mTplDomCache[$titleText], $title ];
3468 }
3469
3470 # Cache miss, go to the database
3471 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3472
3473 if ( $text === false ) {
3474 $this->mTplDomCache[$titleText] = false;
3475 return [ false, $title ];
3476 }
3477
3478 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3479 $this->mTplDomCache[$titleText] = $dom;
3480
3481 if ( !$title->equals( $cacheTitle ) ) {
3482 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3483 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3484 }
3485
3486 return [ $dom, $title ];
3487 }
3488
3489 /**
3490 * Fetch the current revision of a given title. Note that the revision
3491 * (and even the title) may not exist in the database, so everything
3492 * contributing to the output of the parser should use this method
3493 * where possible, rather than getting the revisions themselves. This
3494 * method also caches its results, so using it benefits performance.
3495 *
3496 * @since 1.24
3497 * @param Title $title
3498 * @return Revision
3499 */
3500 public function fetchCurrentRevisionOfTitle( $title ) {
3501 $cacheKey = $title->getPrefixedDBkey();
3502 if ( !$this->currentRevisionCache ) {
3503 $this->currentRevisionCache = new MapCacheLRU( 100 );
3504 }
3505 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3506 $this->currentRevisionCache->set( $cacheKey,
3507 // Defaults to Parser::statelessFetchRevision()
3508 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3509 );
3510 }
3511 return $this->currentRevisionCache->get( $cacheKey );
3512 }
3513
3514 /**
3515 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3516 * without passing them on to it.
3517 *
3518 * @since 1.24
3519 * @param Title $title
3520 * @param Parser|bool $parser
3521 * @return Revision|bool False if missing
3522 */
3523 public static function statelessFetchRevision( Title $title, $parser = false ) {
3524 $pageId = $title->getArticleID();
3525 $revId = $title->getLatestRevID();
3526
3527 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3528 if ( $rev ) {
3529 $rev->setTitle( $title );
3530 }
3531
3532 return $rev;
3533 }
3534
3535 /**
3536 * Fetch the unparsed text of a template and register a reference to it.
3537 * @param Title $title
3538 * @return array ( string or false, Title )
3539 */
3540 public function fetchTemplateAndTitle( $title ) {
3541 // Defaults to Parser::statelessFetchTemplate()
3542 $templateCb = $this->mOptions->getTemplateCallback();
3543 $stuff = call_user_func( $templateCb, $title, $this );
3544 // We use U+007F DELETE to distinguish strip markers from regular text.
3545 $text = $stuff['text'];
3546 if ( is_string( $stuff['text'] ) ) {
3547 $text = strtr( $text, "\x7f", "?" );
3548 }
3549 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3550 if ( isset( $stuff['deps'] ) ) {
3551 foreach ( $stuff['deps'] as $dep ) {
3552 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3553 if ( $dep['title']->equals( $this->getTitle() ) ) {
3554 // If we transclude ourselves, the final result
3555 // will change based on the new version of the page
3556 $this->mOutput->setFlag( 'vary-revision' );
3557 }
3558 }
3559 }
3560 return [ $text, $finalTitle ];
3561 }
3562
3563 /**
3564 * Fetch the unparsed text of a template and register a reference to it.
3565 * @param Title $title
3566 * @return string|bool
3567 */
3568 public function fetchTemplate( $title ) {
3569 return $this->fetchTemplateAndTitle( $title )[0];
3570 }
3571
3572 /**
3573 * Static function to get a template
3574 * Can be overridden via ParserOptions::setTemplateCallback().
3575 *
3576 * @param Title $title
3577 * @param bool|Parser $parser
3578 *
3579 * @return array
3580 */
3581 public static function statelessFetchTemplate( $title, $parser = false ) {
3582 $text = $skip = false;
3583 $finalTitle = $title;
3584 $deps = [];
3585
3586 # Loop to fetch the article, with up to 1 redirect
3587 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3588 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3589 // @codingStandardsIgnoreEnd
3590 # Give extensions a chance to select the revision instead
3591 $id = false; # Assume current
3592 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3593 [ $parser, $title, &$skip, &$id ] );
3594
3595 if ( $skip ) {
3596 $text = false;
3597 $deps[] = [
3598 'title' => $title,
3599 'page_id' => $title->getArticleID(),
3600 'rev_id' => null
3601 ];
3602 break;
3603 }
3604 # Get the revision
3605 if ( $id ) {
3606 $rev = Revision::newFromId( $id );
3607 } elseif ( $parser ) {
3608 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3609 } else {
3610 $rev = Revision::newFromTitle( $title );
3611 }
3612 $rev_id = $rev ? $rev->getId() : 0;
3613 # If there is no current revision, there is no page
3614 if ( $id === false && !$rev ) {
3615 $linkCache = LinkCache::singleton();
3616 $linkCache->addBadLinkObj( $title );
3617 }
3618
3619 $deps[] = [
3620 'title' => $title,
3621 'page_id' => $title->getArticleID(),
3622 'rev_id' => $rev_id ];
3623 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3624 # We fetched a rev from a different title; register it too...
3625 $deps[] = [
3626 'title' => $rev->getTitle(),
3627 'page_id' => $rev->getPage(),
3628 'rev_id' => $rev_id ];
3629 }
3630
3631 if ( $rev ) {
3632 $content = $rev->getContent();
3633 $text = $content ? $content->getWikitextForTransclusion() : null;
3634
3635 Hooks::run( 'ParserFetchTemplate',
3636 [ $parser, $title, $rev, &$text, &$deps ] );
3637
3638 if ( $text === false || $text === null ) {
3639 $text = false;
3640 break;
3641 }
3642 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3643 global $wgContLang;
3644 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3645 if ( !$message->exists() ) {
3646 $text = false;
3647 break;
3648 }
3649 $content = $message->content();
3650 $text = $message->plain();
3651 } else {
3652 break;
3653 }
3654 if ( !$content ) {
3655 break;
3656 }
3657 # Redirect?
3658 $finalTitle = $title;
3659 $title = $content->getRedirectTarget();
3660 }
3661 return [
3662 'text' => $text,
3663 'finalTitle' => $finalTitle,
3664 'deps' => $deps ];
3665 }
3666
3667 /**
3668 * Fetch a file and its title and register a reference to it.
3669 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3670 * @param Title $title
3671 * @param array $options Array of options to RepoGroup::findFile
3672 * @return File|bool
3673 */
3674 public function fetchFile( $title, $options = [] ) {
3675 return $this->fetchFileAndTitle( $title, $options )[0];
3676 }
3677
3678 /**
3679 * Fetch a file and its title and register a reference to it.
3680 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3681 * @param Title $title
3682 * @param array $options Array of options to RepoGroup::findFile
3683 * @return array ( File or false, Title of file )
3684 */
3685 public function fetchFileAndTitle( $title, $options = [] ) {
3686 $file = $this->fetchFileNoRegister( $title, $options );
3687
3688 $time = $file ? $file->getTimestamp() : false;
3689 $sha1 = $file ? $file->getSha1() : false;
3690 # Register the file as a dependency...
3691 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3692 if ( $file && !$title->equals( $file->getTitle() ) ) {
3693 # Update fetched file title
3694 $title = $file->getTitle();
3695 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3696 }
3697 return [ $file, $title ];
3698 }
3699
3700 /**
3701 * Helper function for fetchFileAndTitle.
3702 *
3703 * Also useful if you need to fetch a file but not use it yet,
3704 * for example to get the file's handler.
3705 *
3706 * @param Title $title
3707 * @param array $options Array of options to RepoGroup::findFile
3708 * @return File|bool
3709 */
3710 protected function fetchFileNoRegister( $title, $options = [] ) {
3711 if ( isset( $options['broken'] ) ) {
3712 $file = false; // broken thumbnail forced by hook
3713 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3714 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3715 } else { // get by (name,timestamp)
3716 $file = wfFindFile( $title, $options );
3717 }
3718 return $file;
3719 }
3720
3721 /**
3722 * Transclude an interwiki link.
3723 *
3724 * @param Title $title
3725 * @param string $action
3726 *
3727 * @return string
3728 */
3729 public function interwikiTransclude( $title, $action ) {
3730 global $wgEnableScaryTranscluding;
3731
3732 if ( !$wgEnableScaryTranscluding ) {
3733 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3734 }
3735
3736 $url = $title->getFullURL( [ 'action' => $action ] );
3737
3738 if ( strlen( $url ) > 255 ) {
3739 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3740 }
3741 return $this->fetchScaryTemplateMaybeFromCache( $url );
3742 }
3743
3744 /**
3745 * @param string $url
3746 * @return mixed|string
3747 */
3748 public function fetchScaryTemplateMaybeFromCache( $url ) {
3749 global $wgTranscludeCacheExpiry;
3750 $dbr = wfGetDB( DB_REPLICA );
3751 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3752 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3753 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3754 if ( $obj ) {
3755 return $obj->tc_contents;
3756 }
3757
3758 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3759 $status = $req->execute(); // Status object
3760 if ( $status->isOK() ) {
3761 $text = $req->getContent();
3762 } elseif ( $req->getStatus() != 200 ) {
3763 // Though we failed to fetch the content, this status is useless.
3764 return wfMessage( 'scarytranscludefailed-httpstatus' )
3765 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3766 } else {
3767 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3768 }
3769
3770 $dbw = wfGetDB( DB_MASTER );
3771 $dbw->replace( 'transcache', [ 'tc_url' ], [
3772 'tc_url' => $url,
3773 'tc_time' => $dbw->timestamp( time() ),
3774 'tc_contents' => $text
3775 ] );
3776 return $text;
3777 }
3778
3779 /**
3780 * Triple brace replacement -- used for template arguments
3781 * @private
3782 *
3783 * @param array $piece
3784 * @param PPFrame $frame
3785 *
3786 * @return array
3787 */
3788 public function argSubstitution( $piece, $frame ) {
3789
3790 $error = false;
3791 $parts = $piece['parts'];
3792 $nameWithSpaces = $frame->expand( $piece['title'] );
3793 $argName = trim( $nameWithSpaces );
3794 $object = false;
3795 $text = $frame->getArgument( $argName );
3796 if ( $text === false && $parts->getLength() > 0
3797 && ( $this->ot['html']
3798 || $this->ot['pre']
3799 || ( $this->ot['wiki'] && $frame->isTemplate() )
3800 )
3801 ) {
3802 # No match in frame, use the supplied default
3803 $object = $parts->item( 0 )->getChildren();
3804 }
3805 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3806 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3807 $this->limitationWarn( 'post-expand-template-argument' );
3808 }
3809
3810 if ( $text === false && $object === false ) {
3811 # No match anywhere
3812 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3813 }
3814 if ( $error !== false ) {
3815 $text .= $error;
3816 }
3817 if ( $object !== false ) {
3818 $ret = [ 'object' => $object ];
3819 } else {
3820 $ret = [ 'text' => $text ];
3821 }
3822
3823 return $ret;
3824 }
3825
3826 /**
3827 * Return the text to be used for a given extension tag.
3828 * This is the ghost of strip().
3829 *
3830 * @param array $params Associative array of parameters:
3831 * name PPNode for the tag name
3832 * attr PPNode for unparsed text where tag attributes are thought to be
3833 * attributes Optional associative array of parsed attributes
3834 * inner Contents of extension element
3835 * noClose Original text did not have a close tag
3836 * @param PPFrame $frame
3837 *
3838 * @throws MWException
3839 * @return string
3840 */
3841 public function extensionSubstitution( $params, $frame ) {
3842 static $errorStr = '<span class="error">';
3843 static $errorLen = 20;
3844
3845 $name = $frame->expand( $params['name'] );
3846 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3847 // Probably expansion depth or node count exceeded. Just punt the
3848 // error up.
3849 return $name;
3850 }
3851
3852 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3853 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3854 // See above
3855 return $attrText;
3856 }
3857
3858 // We can't safely check if the expansion for $content resulted in an
3859 // error, because the content could happen to be the error string
3860 // (T149622).
3861 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3862
3863 $marker = self::MARKER_PREFIX . "-$name-"
3864 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3865
3866 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3867 ( $this->ot['html'] || $this->ot['pre'] );
3868 if ( $isFunctionTag ) {
3869 $markerType = 'none';
3870 } else {
3871 $markerType = 'general';
3872 }
3873 if ( $this->ot['html'] || $isFunctionTag ) {
3874 $name = strtolower( $name );
3875 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3876 if ( isset( $params['attributes'] ) ) {
3877 $attributes = $attributes + $params['attributes'];
3878 }
3879
3880 if ( isset( $this->mTagHooks[$name] ) ) {
3881 # Workaround for PHP bug 35229 and similar
3882 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3883 throw new MWException( "Tag hook for $name is not callable\n" );
3884 }
3885 $output = call_user_func_array( $this->mTagHooks[$name],
3886 [ $content, $attributes, $this, $frame ] );
3887 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3888 list( $callback, ) = $this->mFunctionTagHooks[$name];
3889 if ( !is_callable( $callback ) ) {
3890 throw new MWException( "Tag hook for $name is not callable\n" );
3891 }
3892
3893 // Avoid PHP 7.1 warning from passing $this by reference
3894 $parser = $this;
3895 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3896 } else {
3897 $output = '<span class="error">Invalid tag extension name: ' .
3898 htmlspecialchars( $name ) . '</span>';
3899 }
3900
3901 if ( is_array( $output ) ) {
3902 # Extract flags to local scope (to override $markerType)
3903 $flags = $output;
3904 $output = $flags[0];
3905 unset( $flags[0] );
3906 extract( $flags );
3907 }
3908 } else {
3909 if ( is_null( $attrText ) ) {
3910 $attrText = '';
3911 }
3912 if ( isset( $params['attributes'] ) ) {
3913 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3914 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3915 htmlspecialchars( $attrValue ) . '"';
3916 }
3917 }
3918 if ( $content === null ) {
3919 $output = "<$name$attrText/>";
3920 } else {
3921 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3922 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3923 // See above
3924 return $close;
3925 }
3926 $output = "<$name$attrText>$content$close";
3927 }
3928 }
3929
3930 if ( $markerType === 'none' ) {
3931 return $output;
3932 } elseif ( $markerType === 'nowiki' ) {
3933 $this->mStripState->addNoWiki( $marker, $output );
3934 } elseif ( $markerType === 'general' ) {
3935 $this->mStripState->addGeneral( $marker, $output );
3936 } else {
3937 throw new MWException( __METHOD__ . ': invalid marker type' );
3938 }
3939 return $marker;
3940 }
3941
3942 /**
3943 * Increment an include size counter
3944 *
3945 * @param string $type The type of expansion
3946 * @param int $size The size of the text
3947 * @return bool False if this inclusion would take it over the maximum, true otherwise
3948 */
3949 public function incrementIncludeSize( $type, $size ) {
3950 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3951 return false;
3952 } else {
3953 $this->mIncludeSizes[$type] += $size;
3954 return true;
3955 }
3956 }
3957
3958 /**
3959 * Increment the expensive function count
3960 *
3961 * @return bool False if the limit has been exceeded
3962 */
3963 public function incrementExpensiveFunctionCount() {
3964 $this->mExpensiveFunctionCount++;
3965 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3966 }
3967
3968 /**
3969 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3970 * Fills $this->mDoubleUnderscores, returns the modified text
3971 *
3972 * @param string $text
3973 *
3974 * @return string
3975 */
3976 public function doDoubleUnderscore( $text ) {
3977
3978 # The position of __TOC__ needs to be recorded
3979 $mw = MagicWord::get( 'toc' );
3980 if ( $mw->match( $text ) ) {
3981 $this->mShowToc = true;
3982 $this->mForceTocPosition = true;
3983
3984 # Set a placeholder. At the end we'll fill it in with the TOC.
3985 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3986
3987 # Only keep the first one.
3988 $text = $mw->replace( '', $text );
3989 }
3990
3991 # Now match and remove the rest of them
3992 $mwa = MagicWord::getDoubleUnderscoreArray();
3993 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3994
3995 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3996 $this->mOutput->mNoGallery = true;
3997 }
3998 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3999 $this->mShowToc = false;
4000 }
4001 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4002 && $this->mTitle->getNamespace() == NS_CATEGORY
4003 ) {
4004 $this->addTrackingCategory( 'hidden-category-category' );
4005 }
4006 # (T10068) Allow control over whether robots index a page.
4007 # __INDEX__ always overrides __NOINDEX__, see T16899
4008 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4009 $this->mOutput->setIndexPolicy( 'noindex' );
4010 $this->addTrackingCategory( 'noindex-category' );
4011 }
4012 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4013 $this->mOutput->setIndexPolicy( 'index' );
4014 $this->addTrackingCategory( 'index-category' );
4015 }
4016
4017 # Cache all double underscores in the database
4018 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4019 $this->mOutput->setProperty( $key, '' );
4020 }
4021
4022 return $text;
4023 }
4024
4025 /**
4026 * @see ParserOutput::addTrackingCategory()
4027 * @param string $msg Message key
4028 * @return bool Whether the addition was successful
4029 */
4030 public function addTrackingCategory( $msg ) {
4031 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4032 }
4033
4034 /**
4035 * This function accomplishes several tasks:
4036 * 1) Auto-number headings if that option is enabled
4037 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4038 * 3) Add a Table of contents on the top for users who have enabled the option
4039 * 4) Auto-anchor headings
4040 *
4041 * It loops through all headlines, collects the necessary data, then splits up the
4042 * string and re-inserts the newly formatted headlines.
4043 *
4044 * @param string $text
4045 * @param string $origText Original, untouched wikitext
4046 * @param bool $isMain
4047 * @return mixed|string
4048 * @private
4049 */
4050 public function formatHeadings( $text, $origText, $isMain = true ) {
4051 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4052
4053 # Inhibit editsection links if requested in the page
4054 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4055 $maybeShowEditLink = $showEditLink = false;
4056 } else {
4057 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4058 $showEditLink = $this->mOptions->getEditSection();
4059 }
4060 if ( $showEditLink ) {
4061 $this->mOutput->setEditSectionTokens( true );
4062 }
4063
4064 # Get all headlines for numbering them and adding funky stuff like [edit]
4065 # links - this is for later, but we need the number of headlines right now
4066 $matches = [];
4067 $numMatches = preg_match_all(
4068 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4069 $text,
4070 $matches
4071 );
4072
4073 # if there are fewer than 4 headlines in the article, do not show TOC
4074 # unless it's been explicitly enabled.
4075 $enoughToc = $this->mShowToc &&
4076 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4077
4078 # Allow user to stipulate that a page should have a "new section"
4079 # link added via __NEWSECTIONLINK__
4080 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4081 $this->mOutput->setNewSection( true );
4082 }
4083
4084 # Allow user to remove the "new section"
4085 # link via __NONEWSECTIONLINK__
4086 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4087 $this->mOutput->hideNewSection( true );
4088 }
4089
4090 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4091 # override above conditions and always show TOC above first header
4092 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4093 $this->mShowToc = true;
4094 $enoughToc = true;
4095 }
4096
4097 # headline counter
4098 $headlineCount = 0;
4099 $numVisible = 0;
4100
4101 # Ugh .. the TOC should have neat indentation levels which can be
4102 # passed to the skin functions. These are determined here
4103 $toc = '';
4104 $full = '';
4105 $head = [];
4106 $sublevelCount = [];
4107 $levelCount = [];
4108 $level = 0;
4109 $prevlevel = 0;
4110 $toclevel = 0;
4111 $prevtoclevel = 0;
4112 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4113 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4114 $oldType = $this->mOutputType;
4115 $this->setOutputType( self::OT_WIKI );
4116 $frame = $this->getPreprocessor()->newFrame();
4117 $root = $this->preprocessToDom( $origText );
4118 $node = $root->getFirstChild();
4119 $byteOffset = 0;
4120 $tocraw = [];
4121 $refers = [];
4122
4123 $headlines = $numMatches !== false ? $matches[3] : [];
4124
4125 foreach ( $headlines as $headline ) {
4126 $isTemplate = false;
4127 $titleText = false;
4128 $sectionIndex = false;
4129 $numbering = '';
4130 $markerMatches = [];
4131 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4132 $serial = $markerMatches[1];
4133 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4134 $isTemplate = ( $titleText != $baseTitleText );
4135 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4136 }
4137
4138 if ( $toclevel ) {
4139 $prevlevel = $level;
4140 }
4141 $level = $matches[1][$headlineCount];
4142
4143 if ( $level > $prevlevel ) {
4144 # Increase TOC level
4145 $toclevel++;
4146 $sublevelCount[$toclevel] = 0;
4147 if ( $toclevel < $wgMaxTocLevel ) {
4148 $prevtoclevel = $toclevel;
4149 $toc .= Linker::tocIndent();
4150 $numVisible++;
4151 }
4152 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4153 # Decrease TOC level, find level to jump to
4154
4155 for ( $i = $toclevel; $i > 0; $i-- ) {
4156 if ( $levelCount[$i] == $level ) {
4157 # Found last matching level
4158 $toclevel = $i;
4159 break;
4160 } elseif ( $levelCount[$i] < $level ) {
4161 # Found first matching level below current level
4162 $toclevel = $i + 1;
4163 break;
4164 }
4165 }
4166 if ( $i == 0 ) {
4167 $toclevel = 1;
4168 }
4169 if ( $toclevel < $wgMaxTocLevel ) {
4170 if ( $prevtoclevel < $wgMaxTocLevel ) {
4171 # Unindent only if the previous toc level was shown :p
4172 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4173 $prevtoclevel = $toclevel;
4174 } else {
4175 $toc .= Linker::tocLineEnd();
4176 }
4177 }
4178 } else {
4179 # No change in level, end TOC line
4180 if ( $toclevel < $wgMaxTocLevel ) {
4181 $toc .= Linker::tocLineEnd();
4182 }
4183 }
4184
4185 $levelCount[$toclevel] = $level;
4186
4187 # count number of headlines for each level
4188 $sublevelCount[$toclevel]++;
4189 $dot = 0;
4190 for ( $i = 1; $i <= $toclevel; $i++ ) {
4191 if ( !empty( $sublevelCount[$i] ) ) {
4192 if ( $dot ) {
4193 $numbering .= '.';
4194 }
4195 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4196 $dot = 1;
4197 }
4198 }
4199
4200 # The safe header is a version of the header text safe to use for links
4201
4202 # Remove link placeholders by the link text.
4203 # <!--LINK number-->
4204 # turns into
4205 # link text with suffix
4206 # Do this before unstrip since link text can contain strip markers
4207 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4208
4209 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4210 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4211
4212 # Strip out HTML (first regex removes any tag not allowed)
4213 # Allowed tags are:
4214 # * <sup> and <sub> (T10393)
4215 # * <i> (T28375)
4216 # * <b> (r105284)
4217 # * <bdi> (T74884)
4218 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4219 # * <s> and <strike> (T35715)
4220 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4221 # to allow setting directionality in toc items.
4222 $tocline = preg_replace(
4223 [
4224 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4225 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4226 ],
4227 [ '', '<$1>' ],
4228 $safeHeadline
4229 );
4230
4231 # Strip '<span></span>', which is the result from the above if
4232 # <span id="foo"></span> is used to produce an additional anchor
4233 # for a section.
4234 $tocline = str_replace( '<span></span>', '', $tocline );
4235
4236 $tocline = trim( $tocline );
4237
4238 # For the anchor, strip out HTML-y stuff period
4239 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4240 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4241
4242 # Save headline for section edit hint before it's escaped
4243 $headlineHint = $safeHeadline;
4244
4245 if ( $wgExperimentalHtmlIds ) {
4246 # For reverse compatibility, provide an id that's
4247 # HTML4-compatible, like we used to.
4248 # It may be worth noting, academically, that it's possible for
4249 # the legacy anchor to conflict with a non-legacy headline
4250 # anchor on the page. In this case likely the "correct" thing
4251 # would be to either drop the legacy anchors or make sure
4252 # they're numbered first. However, this would require people
4253 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4254 # manually, so let's not bother worrying about it.
4255 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4256 [ 'noninitial', 'legacy' ] );
4257 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4258
4259 if ( $legacyHeadline == $safeHeadline ) {
4260 # No reason to have both (in fact, we can't)
4261 $legacyHeadline = false;
4262 }
4263 } else {
4264 $legacyHeadline = false;
4265 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4266 'noninitial' );
4267 }
4268
4269 # HTML names must be case-insensitively unique (T12721).
4270 # This does not apply to Unicode characters per
4271 # https://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4272 # @todo FIXME: We may be changing them depending on the current locale.
4273 $arrayKey = strtolower( $safeHeadline );
4274 if ( $legacyHeadline === false ) {
4275 $legacyArrayKey = false;
4276 } else {
4277 $legacyArrayKey = strtolower( $legacyHeadline );
4278 }
4279
4280 # Create the anchor for linking from the TOC to the section
4281 $anchor = $safeHeadline;
4282 $legacyAnchor = $legacyHeadline;
4283 if ( isset( $refers[$arrayKey] ) ) {
4284 // @codingStandardsIgnoreStart
4285 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4286 // @codingStandardsIgnoreEnd
4287 $anchor .= "_$i";
4288 $refers["${arrayKey}_$i"] = true;
4289 } else {
4290 $refers[$arrayKey] = true;
4291 }
4292 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4293 // @codingStandardsIgnoreStart
4294 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4295 // @codingStandardsIgnoreEnd
4296 $legacyAnchor .= "_$i";
4297 $refers["${legacyArrayKey}_$i"] = true;
4298 } else {
4299 $refers[$legacyArrayKey] = true;
4300 }
4301
4302 # Don't number the heading if it is the only one (looks silly)
4303 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4304 # the two are different if the line contains a link
4305 $headline = Html::element(
4306 'span',
4307 [ 'class' => 'mw-headline-number' ],
4308 $numbering
4309 ) . ' ' . $headline;
4310 }
4311
4312 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4313 $toc .= Linker::tocLine( $anchor, $tocline,
4314 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4315 }
4316
4317 # Add the section to the section tree
4318 # Find the DOM node for this header
4319 $noOffset = ( $isTemplate || $sectionIndex === false );
4320 while ( $node && !$noOffset ) {
4321 if ( $node->getName() === 'h' ) {
4322 $bits = $node->splitHeading();
4323 if ( $bits['i'] == $sectionIndex ) {
4324 break;
4325 }
4326 }
4327 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4328 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4329 $node = $node->getNextSibling();
4330 }
4331 $tocraw[] = [
4332 'toclevel' => $toclevel,
4333 'level' => $level,
4334 'line' => $tocline,
4335 'number' => $numbering,
4336 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4337 'fromtitle' => $titleText,
4338 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4339 'anchor' => $anchor,
4340 ];
4341
4342 # give headline the correct <h#> tag
4343 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4344 // Output edit section links as markers with styles that can be customized by skins
4345 if ( $isTemplate ) {
4346 # Put a T flag in the section identifier, to indicate to extractSections()
4347 # that sections inside <includeonly> should be counted.
4348 $editsectionPage = $titleText;
4349 $editsectionSection = "T-$sectionIndex";
4350 $editsectionContent = null;
4351 } else {
4352 $editsectionPage = $this->mTitle->getPrefixedText();
4353 $editsectionSection = $sectionIndex;
4354 $editsectionContent = $headlineHint;
4355 }
4356 // We use a bit of pesudo-xml for editsection markers. The
4357 // language converter is run later on. Using a UNIQ style marker
4358 // leads to the converter screwing up the tokens when it
4359 // converts stuff. And trying to insert strip tags fails too. At
4360 // this point all real inputted tags have already been escaped,
4361 // so we don't have to worry about a user trying to input one of
4362 // these markers directly. We use a page and section attribute
4363 // to stop the language converter from converting these
4364 // important bits of data, but put the headline hint inside a
4365 // content block because the language converter is supposed to
4366 // be able to convert that piece of data.
4367 // Gets replaced with html in ParserOutput::getText
4368 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4369 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4370 if ( $editsectionContent !== null ) {
4371 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4372 } else {
4373 $editlink .= '/>';
4374 }
4375 } else {
4376 $editlink = '';
4377 }
4378 $head[$headlineCount] = Linker::makeHeadline( $level,
4379 $matches['attrib'][$headlineCount], $anchor, $headline,
4380 $editlink, $legacyAnchor );
4381
4382 $headlineCount++;
4383 }
4384
4385 $this->setOutputType( $oldType );
4386
4387 # Never ever show TOC if no headers
4388 if ( $numVisible < 1 ) {
4389 $enoughToc = false;
4390 }
4391
4392 if ( $enoughToc ) {
4393 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4394 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4395 }
4396 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4397 $this->mOutput->setTOCHTML( $toc );
4398 $toc = self::TOC_START . $toc . self::TOC_END;
4399 $this->mOutput->addModules( 'mediawiki.toc' );
4400 }
4401
4402 if ( $isMain ) {
4403 $this->mOutput->setSections( $tocraw );
4404 }
4405
4406 # split up and insert constructed headlines
4407 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4408 $i = 0;
4409
4410 // build an array of document sections
4411 $sections = [];
4412 foreach ( $blocks as $block ) {
4413 // $head is zero-based, sections aren't.
4414 if ( empty( $head[$i - 1] ) ) {
4415 $sections[$i] = $block;
4416 } else {
4417 $sections[$i] = $head[$i - 1] . $block;
4418 }
4419
4420 /**
4421 * Send a hook, one per section.
4422 * The idea here is to be able to make section-level DIVs, but to do so in a
4423 * lower-impact, more correct way than r50769
4424 *
4425 * $this : caller
4426 * $section : the section number
4427 * &$sectionContent : ref to the content of the section
4428 * $showEditLinks : boolean describing whether this section has an edit link
4429 */
4430 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4431
4432 $i++;
4433 }
4434
4435 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4436 // append the TOC at the beginning
4437 // Top anchor now in skin
4438 $sections[0] = $sections[0] . $toc . "\n";
4439 }
4440
4441 $full .= implode( '', $sections );
4442
4443 if ( $this->mForceTocPosition ) {
4444 return str_replace( '<!--MWTOC-->', $toc, $full );
4445 } else {
4446 return $full;
4447 }
4448 }
4449
4450 /**
4451 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4452 * conversion, substituting signatures, {{subst:}} templates, etc.
4453 *
4454 * @param string $text The text to transform
4455 * @param Title $title The Title object for the current article
4456 * @param User $user The User object describing the current user
4457 * @param ParserOptions $options Parsing options
4458 * @param bool $clearState Whether to clear the parser state first
4459 * @return string The altered wiki markup
4460 */
4461 public function preSaveTransform( $text, Title $title, User $user,
4462 ParserOptions $options, $clearState = true
4463 ) {
4464 if ( $clearState ) {
4465 $magicScopeVariable = $this->lock();
4466 }
4467 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4468 $this->setUser( $user );
4469
4470 // Strip U+0000 NULL (T159174)
4471 $text = str_replace( "\000", '', $text );
4472
4473 // We still normalize line endings for backwards-compatibility
4474 // with other code that just calls PST, but this should already
4475 // be handled in TextContent subclasses
4476 $text = TextContent::normalizeLineEndings( $text );
4477
4478 if ( $options->getPreSaveTransform() ) {
4479 $text = $this->pstPass2( $text, $user );
4480 }
4481 $text = $this->mStripState->unstripBoth( $text );
4482
4483 $this->setUser( null ); # Reset
4484
4485 return $text;
4486 }
4487
4488 /**
4489 * Pre-save transform helper function
4490 *
4491 * @param string $text
4492 * @param User $user
4493 *
4494 * @return string
4495 */
4496 private function pstPass2( $text, $user ) {
4497 global $wgContLang;
4498
4499 # Note: This is the timestamp saved as hardcoded wikitext to
4500 # the database, we use $wgContLang here in order to give
4501 # everyone the same signature and use the default one rather
4502 # than the one selected in each user's preferences.
4503 # (see also T14815)
4504 $ts = $this->mOptions->getTimestamp();
4505 $timestamp = MWTimestamp::getLocalInstance( $ts );
4506 $ts = $timestamp->format( 'YmdHis' );
4507 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4508
4509 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4510
4511 # Variable replacement
4512 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4513 $text = $this->replaceVariables( $text );
4514
4515 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4516 # which may corrupt this parser instance via its wfMessage()->text() call-
4517
4518 # Signatures
4519 $sigText = $this->getUserSig( $user );
4520 $text = strtr( $text, [
4521 '~~~~~' => $d,
4522 '~~~~' => "$sigText $d",
4523 '~~~' => $sigText
4524 ] );
4525
4526 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4527 $tc = '[' . Title::legalChars() . ']';
4528 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4529
4530 // [[ns:page (context)|]]
4531 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4532 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4533 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4534 // [[ns:page (context), context|]] (using either single or double-width comma)
4535 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4536 // [[|page]] (reverse pipe trick: add context from page title)
4537 $p2 = "/\[\[\\|($tc+)]]/";
4538
4539 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4540 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4541 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4542 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4543
4544 $t = $this->mTitle->getText();
4545 $m = [];
4546 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4547 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4548 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4549 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4550 } else {
4551 # if there's no context, don't bother duplicating the title
4552 $text = preg_replace( $p2, '[[\\1]]', $text );
4553 }
4554
4555 return $text;
4556 }
4557
4558 /**
4559 * Fetch the user's signature text, if any, and normalize to
4560 * validated, ready-to-insert wikitext.
4561 * If you have pre-fetched the nickname or the fancySig option, you can
4562 * specify them here to save a database query.
4563 * Do not reuse this parser instance after calling getUserSig(),
4564 * as it may have changed if it's the $wgParser.
4565 *
4566 * @param User $user
4567 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4568 * @param bool|null $fancySig whether the nicknname is the complete signature
4569 * or null to use default value
4570 * @return string
4571 */
4572 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4573 global $wgMaxSigChars;
4574
4575 $username = $user->getName();
4576
4577 # If not given, retrieve from the user object.
4578 if ( $nickname === false ) {
4579 $nickname = $user->getOption( 'nickname' );
4580 }
4581
4582 if ( is_null( $fancySig ) ) {
4583 $fancySig = $user->getBoolOption( 'fancysig' );
4584 }
4585
4586 $nickname = $nickname == null ? $username : $nickname;
4587
4588 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4589 $nickname = $username;
4590 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4591 } elseif ( $fancySig !== false ) {
4592 # Sig. might contain markup; validate this
4593 if ( $this->validateSig( $nickname ) !== false ) {
4594 # Validated; clean up (if needed) and return it
4595 return $this->cleanSig( $nickname, true );
4596 } else {
4597 # Failed to validate; fall back to the default
4598 $nickname = $username;
4599 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4600 }
4601 }
4602
4603 # Make sure nickname doesnt get a sig in a sig
4604 $nickname = self::cleanSigInSig( $nickname );
4605
4606 # If we're still here, make it a link to the user page
4607 $userText = wfEscapeWikiText( $username );
4608 $nickText = wfEscapeWikiText( $nickname );
4609 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4610
4611 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4612 ->title( $this->getTitle() )->text();
4613 }
4614
4615 /**
4616 * Check that the user's signature contains no bad XML
4617 *
4618 * @param string $text
4619 * @return string|bool An expanded string, or false if invalid.
4620 */
4621 public function validateSig( $text ) {
4622 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4623 }
4624
4625 /**
4626 * Clean up signature text
4627 *
4628 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4629 * 2) Substitute all transclusions
4630 *
4631 * @param string $text
4632 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4633 * @return string Signature text
4634 */
4635 public function cleanSig( $text, $parsing = false ) {
4636 if ( !$parsing ) {
4637 global $wgTitle;
4638 $magicScopeVariable = $this->lock();
4639 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4640 }
4641
4642 # Option to disable this feature
4643 if ( !$this->mOptions->getCleanSignatures() ) {
4644 return $text;
4645 }
4646
4647 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4648 # => Move this logic to braceSubstitution()
4649 $substWord = MagicWord::get( 'subst' );
4650 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4651 $substText = '{{' . $substWord->getSynonym( 0 );
4652
4653 $text = preg_replace( $substRegex, $substText, $text );
4654 $text = self::cleanSigInSig( $text );
4655 $dom = $this->preprocessToDom( $text );
4656 $frame = $this->getPreprocessor()->newFrame();
4657 $text = $frame->expand( $dom );
4658
4659 if ( !$parsing ) {
4660 $text = $this->mStripState->unstripBoth( $text );
4661 }
4662
4663 return $text;
4664 }
4665
4666 /**
4667 * Strip 3, 4 or 5 tildes out of signatures.
4668 *
4669 * @param string $text
4670 * @return string Signature text with /~{3,5}/ removed
4671 */
4672 public static function cleanSigInSig( $text ) {
4673 $text = preg_replace( '/~{3,5}/', '', $text );
4674 return $text;
4675 }
4676
4677 /**
4678 * Set up some variables which are usually set up in parse()
4679 * so that an external function can call some class members with confidence
4680 *
4681 * @param Title|null $title
4682 * @param ParserOptions $options
4683 * @param int $outputType
4684 * @param bool $clearState
4685 */
4686 public function startExternalParse( Title $title = null, ParserOptions $options,
4687 $outputType, $clearState = true
4688 ) {
4689 $this->startParse( $title, $options, $outputType, $clearState );
4690 }
4691
4692 /**
4693 * @param Title|null $title
4694 * @param ParserOptions $options
4695 * @param int $outputType
4696 * @param bool $clearState
4697 */
4698 private function startParse( Title $title = null, ParserOptions $options,
4699 $outputType, $clearState = true
4700 ) {
4701 $this->setTitle( $title );
4702 $this->mOptions = $options;
4703 $this->setOutputType( $outputType );
4704 if ( $clearState ) {
4705 $this->clearState();
4706 }
4707 }
4708
4709 /**
4710 * Wrapper for preprocess()
4711 *
4712 * @param string $text The text to preprocess
4713 * @param ParserOptions $options Options
4714 * @param Title|null $title Title object or null to use $wgTitle
4715 * @return string
4716 */
4717 public function transformMsg( $text, $options, $title = null ) {
4718 static $executing = false;
4719
4720 # Guard against infinite recursion
4721 if ( $executing ) {
4722 return $text;
4723 }
4724 $executing = true;
4725
4726 if ( !$title ) {
4727 global $wgTitle;
4728 $title = $wgTitle;
4729 }
4730
4731 $text = $this->preprocess( $text, $title, $options );
4732
4733 $executing = false;
4734 return $text;
4735 }
4736
4737 /**
4738 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4739 * The callback should have the following form:
4740 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4741 *
4742 * Transform and return $text. Use $parser for any required context, e.g. use
4743 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4744 *
4745 * Hooks may return extended information by returning an array, of which the
4746 * first numbered element (index 0) must be the return string, and all other
4747 * entries are extracted into local variables within an internal function
4748 * in the Parser class.
4749 *
4750 * This interface (introduced r61913) appears to be undocumented, but
4751 * 'markerType' is used by some core tag hooks to override which strip
4752 * array their results are placed in. **Use great caution if attempting
4753 * this interface, as it is not documented and injudicious use could smash
4754 * private variables.**
4755 *
4756 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4757 * @param callable $callback The callback function (and object) to use for the tag
4758 * @throws MWException
4759 * @return callable|null The old value of the mTagHooks array associated with the hook
4760 */
4761 public function setHook( $tag, $callback ) {
4762 $tag = strtolower( $tag );
4763 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4764 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4765 }
4766 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4767 $this->mTagHooks[$tag] = $callback;
4768 if ( !in_array( $tag, $this->mStripList ) ) {
4769 $this->mStripList[] = $tag;
4770 }
4771
4772 return $oldVal;
4773 }
4774
4775 /**
4776 * As setHook(), but letting the contents be parsed.
4777 *
4778 * Transparent tag hooks are like regular XML-style tag hooks, except they
4779 * operate late in the transformation sequence, on HTML instead of wikitext.
4780 *
4781 * This is probably obsoleted by things dealing with parser frames?
4782 * The only extension currently using it is geoserver.
4783 *
4784 * @since 1.10
4785 * @todo better document or deprecate this
4786 *
4787 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4788 * @param callable $callback The callback function (and object) to use for the tag
4789 * @throws MWException
4790 * @return callable|null The old value of the mTagHooks array associated with the hook
4791 */
4792 public function setTransparentTagHook( $tag, $callback ) {
4793 $tag = strtolower( $tag );
4794 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4795 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4796 }
4797 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4798 $this->mTransparentTagHooks[$tag] = $callback;
4799
4800 return $oldVal;
4801 }
4802
4803 /**
4804 * Remove all tag hooks
4805 */
4806 public function clearTagHooks() {
4807 $this->mTagHooks = [];
4808 $this->mFunctionTagHooks = [];
4809 $this->mStripList = $this->mDefaultStripList;
4810 }
4811
4812 /**
4813 * Create a function, e.g. {{sum:1|2|3}}
4814 * The callback function should have the form:
4815 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4816 *
4817 * Or with Parser::SFH_OBJECT_ARGS:
4818 * function myParserFunction( $parser, $frame, $args ) { ... }
4819 *
4820 * The callback may either return the text result of the function, or an array with the text
4821 * in element 0, and a number of flags in the other elements. The names of the flags are
4822 * specified in the keys. Valid flags are:
4823 * found The text returned is valid, stop processing the template. This
4824 * is on by default.
4825 * nowiki Wiki markup in the return value should be escaped
4826 * isHTML The returned text is HTML, armour it against wikitext transformation
4827 *
4828 * @param string $id The magic word ID
4829 * @param callable $callback The callback function (and object) to use
4830 * @param int $flags A combination of the following flags:
4831 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4832 *
4833 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4834 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4835 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4836 * the arguments, and to control the way they are expanded.
4837 *
4838 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4839 * arguments, for instance:
4840 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4841 *
4842 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4843 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4844 * working if/when this is changed.
4845 *
4846 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4847 * expansion.
4848 *
4849 * Please read the documentation in includes/parser/Preprocessor.php for more information
4850 * about the methods available in PPFrame and PPNode.
4851 *
4852 * @throws MWException
4853 * @return string|callable The old callback function for this name, if any
4854 */
4855 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4856 global $wgContLang;
4857
4858 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4859 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4860
4861 # Add to function cache
4862 $mw = MagicWord::get( $id );
4863 if ( !$mw ) {
4864 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4865 }
4866
4867 $synonyms = $mw->getSynonyms();
4868 $sensitive = intval( $mw->isCaseSensitive() );
4869
4870 foreach ( $synonyms as $syn ) {
4871 # Case
4872 if ( !$sensitive ) {
4873 $syn = $wgContLang->lc( $syn );
4874 }
4875 # Add leading hash
4876 if ( !( $flags & self::SFH_NO_HASH ) ) {
4877 $syn = '#' . $syn;
4878 }
4879 # Remove trailing colon
4880 if ( substr( $syn, -1, 1 ) === ':' ) {
4881 $syn = substr( $syn, 0, -1 );
4882 }
4883 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4884 }
4885 return $oldVal;
4886 }
4887
4888 /**
4889 * Get all registered function hook identifiers
4890 *
4891 * @return array
4892 */
4893 public function getFunctionHooks() {
4894 return array_keys( $this->mFunctionHooks );
4895 }
4896
4897 /**
4898 * Create a tag function, e.g. "<test>some stuff</test>".
4899 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4900 * Unlike parser functions, their content is not preprocessed.
4901 * @param string $tag
4902 * @param callable $callback
4903 * @param int $flags
4904 * @throws MWException
4905 * @return null
4906 */
4907 public function setFunctionTagHook( $tag, $callback, $flags ) {
4908 $tag = strtolower( $tag );
4909 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4910 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4911 }
4912 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4913 $this->mFunctionTagHooks[$tag] : null;
4914 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4915
4916 if ( !in_array( $tag, $this->mStripList ) ) {
4917 $this->mStripList[] = $tag;
4918 }
4919
4920 return $old;
4921 }
4922
4923 /**
4924 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4925 * Placeholders created in Linker::link()
4926 *
4927 * @param string $text
4928 * @param int $options
4929 */
4930 public function replaceLinkHolders( &$text, $options = 0 ) {
4931 $this->mLinkHolders->replace( $text );
4932 }
4933
4934 /**
4935 * Replace "<!--LINK-->" link placeholders with plain text of links
4936 * (not HTML-formatted).
4937 *
4938 * @param string $text
4939 * @return string
4940 */
4941 public function replaceLinkHoldersText( $text ) {
4942 return $this->mLinkHolders->replaceText( $text );
4943 }
4944
4945 /**
4946 * Renders an image gallery from a text with one line per image.
4947 * text labels may be given by using |-style alternative text. E.g.
4948 * Image:one.jpg|The number "1"
4949 * Image:tree.jpg|A tree
4950 * given as text will return the HTML of a gallery with two images,
4951 * labeled 'The number "1"' and
4952 * 'A tree'.
4953 *
4954 * @param string $text
4955 * @param array $params
4956 * @return string HTML
4957 */
4958 public function renderImageGallery( $text, $params ) {
4959
4960 $mode = false;
4961 if ( isset( $params['mode'] ) ) {
4962 $mode = $params['mode'];
4963 }
4964
4965 try {
4966 $ig = ImageGalleryBase::factory( $mode );
4967 } catch ( Exception $e ) {
4968 // If invalid type set, fallback to default.
4969 $ig = ImageGalleryBase::factory( false );
4970 }
4971
4972 $ig->setContextTitle( $this->mTitle );
4973 $ig->setShowBytes( false );
4974 $ig->setShowFilename( false );
4975 $ig->setParser( $this );
4976 $ig->setHideBadImages();
4977 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4978
4979 if ( isset( $params['showfilename'] ) ) {
4980 $ig->setShowFilename( true );
4981 } else {
4982 $ig->setShowFilename( false );
4983 }
4984 if ( isset( $params['caption'] ) ) {
4985 $caption = $params['caption'];
4986 $caption = htmlspecialchars( $caption );
4987 $caption = $this->replaceInternalLinks( $caption );
4988 $ig->setCaptionHtml( $caption );
4989 }
4990 if ( isset( $params['perrow'] ) ) {
4991 $ig->setPerRow( $params['perrow'] );
4992 }
4993 if ( isset( $params['widths'] ) ) {
4994 $ig->setWidths( $params['widths'] );
4995 }
4996 if ( isset( $params['heights'] ) ) {
4997 $ig->setHeights( $params['heights'] );
4998 }
4999 $ig->setAdditionalOptions( $params );
5000
5001 // Avoid PHP 7.1 warning from passing $this by reference
5002 $parser = $this;
5003 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5004
5005 $lines = StringUtils::explode( "\n", $text );
5006 foreach ( $lines as $line ) {
5007 # match lines like these:
5008 # Image:someimage.jpg|This is some image
5009 $matches = [];
5010 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5011 # Skip empty lines
5012 if ( count( $matches ) == 0 ) {
5013 continue;
5014 }
5015
5016 if ( strpos( $matches[0], '%' ) !== false ) {
5017 $matches[1] = rawurldecode( $matches[1] );
5018 }
5019 $title = Title::newFromText( $matches[1], NS_FILE );
5020 if ( is_null( $title ) ) {
5021 # Bogus title. Ignore these so we don't bomb out later.
5022 continue;
5023 }
5024
5025 # We need to get what handler the file uses, to figure out parameters.
5026 # Note, a hook can overide the file name, and chose an entirely different
5027 # file (which potentially could be of a different type and have different handler).
5028 $options = [];
5029 $descQuery = false;
5030 Hooks::run( 'BeforeParserFetchFileAndTitle',
5031 [ $this, $title, &$options, &$descQuery ] );
5032 # Don't register it now, as TraditionalImageGallery does that later.
5033 $file = $this->fetchFileNoRegister( $title, $options );
5034 $handler = $file ? $file->getHandler() : false;
5035
5036 $paramMap = [
5037 'img_alt' => 'gallery-internal-alt',
5038 'img_link' => 'gallery-internal-link',
5039 ];
5040 if ( $handler ) {
5041 $paramMap = $paramMap + $handler->getParamMap();
5042 // We don't want people to specify per-image widths.
5043 // Additionally the width parameter would need special casing anyhow.
5044 unset( $paramMap['img_width'] );
5045 }
5046
5047 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5048
5049 $label = '';
5050 $alt = '';
5051 $link = '';
5052 $handlerOptions = [];
5053 if ( isset( $matches[3] ) ) {
5054 // look for an |alt= definition while trying not to break existing
5055 // captions with multiple pipes (|) in it, until a more sensible grammar
5056 // is defined for images in galleries
5057
5058 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5059 // splitting on '|' is a bit odd, and different from makeImage.
5060 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5061 // Protect LanguageConverter markup
5062 $parameterMatches = StringUtils::delimiterExplode(
5063 '-{', '}-', '|', $matches[3], true /* nested */
5064 );
5065
5066 foreach ( $parameterMatches as $parameterMatch ) {
5067 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5068 if ( $magicName ) {
5069 $paramName = $paramMap[$magicName];
5070
5071 switch ( $paramName ) {
5072 case 'gallery-internal-alt':
5073 $alt = $this->stripAltText( $match, false );
5074 break;
5075 case 'gallery-internal-link':
5076 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5077 $chars = self::EXT_LINK_URL_CLASS;
5078 $addr = self::EXT_LINK_ADDR;
5079 $prots = $this->mUrlProtocols;
5080 // check to see if link matches an absolute url, if not then it must be a wiki link.
5081 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5082 // Result of LanguageConverter::markNoConversion
5083 // invoked on an external link.
5084 $linkValue = substr( $linkValue, 4, -2 );
5085 }
5086 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5087 $link = $linkValue;
5088 } else {
5089 $localLinkTitle = Title::newFromText( $linkValue );
5090 if ( $localLinkTitle !== null ) {
5091 $link = $localLinkTitle->getLinkURL();
5092 }
5093 }
5094 break;
5095 default:
5096 // Must be a handler specific parameter.
5097 if ( $handler->validateParam( $paramName, $match ) ) {
5098 $handlerOptions[$paramName] = $match;
5099 } else {
5100 // Guess not, consider it as caption.
5101 wfDebug( "$parameterMatch failed parameter validation\n" );
5102 $label = '|' . $parameterMatch;
5103 }
5104 }
5105
5106 } else {
5107 // Last pipe wins.
5108 $label = '|' . $parameterMatch;
5109 }
5110 }
5111 // Remove the pipe.
5112 $label = substr( $label, 1 );
5113 }
5114
5115 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5116 }
5117 $html = $ig->toHTML();
5118 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5119 return $html;
5120 }
5121
5122 /**
5123 * @param MediaHandler $handler
5124 * @return array
5125 */
5126 public function getImageParams( $handler ) {
5127 if ( $handler ) {
5128 $handlerClass = get_class( $handler );
5129 } else {
5130 $handlerClass = '';
5131 }
5132 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5133 # Initialise static lists
5134 static $internalParamNames = [
5135 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5136 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5137 'bottom', 'text-bottom' ],
5138 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5139 'upright', 'border', 'link', 'alt', 'class' ],
5140 ];
5141 static $internalParamMap;
5142 if ( !$internalParamMap ) {
5143 $internalParamMap = [];
5144 foreach ( $internalParamNames as $type => $names ) {
5145 foreach ( $names as $name ) {
5146 $magicName = str_replace( '-', '_', "img_$name" );
5147 $internalParamMap[$magicName] = [ $type, $name ];
5148 }
5149 }
5150 }
5151
5152 # Add handler params
5153 $paramMap = $internalParamMap;
5154 if ( $handler ) {
5155 $handlerParamMap = $handler->getParamMap();
5156 foreach ( $handlerParamMap as $magic => $paramName ) {
5157 $paramMap[$magic] = [ 'handler', $paramName ];
5158 }
5159 }
5160 $this->mImageParams[$handlerClass] = $paramMap;
5161 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5162 }
5163 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5164 }
5165
5166 /**
5167 * Parse image options text and use it to make an image
5168 *
5169 * @param Title $title
5170 * @param string $options
5171 * @param LinkHolderArray|bool $holders
5172 * @return string HTML
5173 */
5174 public function makeImage( $title, $options, $holders = false ) {
5175 # Check if the options text is of the form "options|alt text"
5176 # Options are:
5177 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5178 # * left no resizing, just left align. label is used for alt= only
5179 # * right same, but right aligned
5180 # * none same, but not aligned
5181 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5182 # * center center the image
5183 # * frame Keep original image size, no magnify-button.
5184 # * framed Same as "frame"
5185 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5186 # * upright reduce width for upright images, rounded to full __0 px
5187 # * border draw a 1px border around the image
5188 # * alt Text for HTML alt attribute (defaults to empty)
5189 # * class Set a class for img node
5190 # * link Set the target of the image link. Can be external, interwiki, or local
5191 # vertical-align values (no % or length right now):
5192 # * baseline
5193 # * sub
5194 # * super
5195 # * top
5196 # * text-top
5197 # * middle
5198 # * bottom
5199 # * text-bottom
5200
5201 # Protect LanguageConverter markup when splitting into parts
5202 $parts = StringUtils::delimiterExplode(
5203 '-{', '}-', '|', $options, true /* allow nesting */
5204 );
5205
5206 # Give extensions a chance to select the file revision for us
5207 $options = [];
5208 $descQuery = false;
5209 Hooks::run( 'BeforeParserFetchFileAndTitle',
5210 [ $this, $title, &$options, &$descQuery ] );
5211 # Fetch and register the file (file title may be different via hooks)
5212 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5213
5214 # Get parameter map
5215 $handler = $file ? $file->getHandler() : false;
5216
5217 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5218
5219 if ( !$file ) {
5220 $this->addTrackingCategory( 'broken-file-category' );
5221 }
5222
5223 # Process the input parameters
5224 $caption = '';
5225 $params = [ 'frame' => [], 'handler' => [],
5226 'horizAlign' => [], 'vertAlign' => [] ];
5227 $seenformat = false;
5228 foreach ( $parts as $part ) {
5229 $part = trim( $part );
5230 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5231 $validated = false;
5232 if ( isset( $paramMap[$magicName] ) ) {
5233 list( $type, $paramName ) = $paramMap[$magicName];
5234
5235 # Special case; width and height come in one variable together
5236 if ( $type === 'handler' && $paramName === 'width' ) {
5237 $parsedWidthParam = $this->parseWidthParam( $value );
5238 if ( isset( $parsedWidthParam['width'] ) ) {
5239 $width = $parsedWidthParam['width'];
5240 if ( $handler->validateParam( 'width', $width ) ) {
5241 $params[$type]['width'] = $width;
5242 $validated = true;
5243 }
5244 }
5245 if ( isset( $parsedWidthParam['height'] ) ) {
5246 $height = $parsedWidthParam['height'];
5247 if ( $handler->validateParam( 'height', $height ) ) {
5248 $params[$type]['height'] = $height;
5249 $validated = true;
5250 }
5251 }
5252 # else no validation -- T15436
5253 } else {
5254 if ( $type === 'handler' ) {
5255 # Validate handler parameter
5256 $validated = $handler->validateParam( $paramName, $value );
5257 } else {
5258 # Validate internal parameters
5259 switch ( $paramName ) {
5260 case 'manualthumb':
5261 case 'alt':
5262 case 'class':
5263 # @todo FIXME: Possibly check validity here for
5264 # manualthumb? downstream behavior seems odd with
5265 # missing manual thumbs.
5266 $validated = true;
5267 $value = $this->stripAltText( $value, $holders );
5268 break;
5269 case 'link':
5270 $chars = self::EXT_LINK_URL_CLASS;
5271 $addr = self::EXT_LINK_ADDR;
5272 $prots = $this->mUrlProtocols;
5273 if ( $value === '' ) {
5274 $paramName = 'no-link';
5275 $value = true;
5276 $validated = true;
5277 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5278 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5279 $paramName = 'link-url';
5280 $this->mOutput->addExternalLink( $value );
5281 if ( $this->mOptions->getExternalLinkTarget() ) {
5282 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5283 }
5284 $validated = true;
5285 }
5286 } else {
5287 $linkTitle = Title::newFromText( $value );
5288 if ( $linkTitle ) {
5289 $paramName = 'link-title';
5290 $value = $linkTitle;
5291 $this->mOutput->addLink( $linkTitle );
5292 $validated = true;
5293 }
5294 }
5295 break;
5296 case 'frameless':
5297 case 'framed':
5298 case 'thumbnail':
5299 // use first appearing option, discard others.
5300 $validated = !$seenformat;
5301 $seenformat = true;
5302 break;
5303 default:
5304 # Most other things appear to be empty or numeric...
5305 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5306 }
5307 }
5308
5309 if ( $validated ) {
5310 $params[$type][$paramName] = $value;
5311 }
5312 }
5313 }
5314 if ( !$validated ) {
5315 $caption = $part;
5316 }
5317 }
5318
5319 # Process alignment parameters
5320 if ( $params['horizAlign'] ) {
5321 $params['frame']['align'] = key( $params['horizAlign'] );
5322 }
5323 if ( $params['vertAlign'] ) {
5324 $params['frame']['valign'] = key( $params['vertAlign'] );
5325 }
5326
5327 $params['frame']['caption'] = $caption;
5328
5329 # Will the image be presented in a frame, with the caption below?
5330 $imageIsFramed = isset( $params['frame']['frame'] )
5331 || isset( $params['frame']['framed'] )
5332 || isset( $params['frame']['thumbnail'] )
5333 || isset( $params['frame']['manualthumb'] );
5334
5335 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5336 # came to also set the caption, ordinary text after the image -- which
5337 # makes no sense, because that just repeats the text multiple times in
5338 # screen readers. It *also* came to set the title attribute.
5339 # Now that we have an alt attribute, we should not set the alt text to
5340 # equal the caption: that's worse than useless, it just repeats the
5341 # text. This is the framed/thumbnail case. If there's no caption, we
5342 # use the unnamed parameter for alt text as well, just for the time be-
5343 # ing, if the unnamed param is set and the alt param is not.
5344 # For the future, we need to figure out if we want to tweak this more,
5345 # e.g., introducing a title= parameter for the title; ignoring the un-
5346 # named parameter entirely for images without a caption; adding an ex-
5347 # plicit caption= parameter and preserving the old magic unnamed para-
5348 # meter for BC; ...
5349 if ( $imageIsFramed ) { # Framed image
5350 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5351 # No caption or alt text, add the filename as the alt text so
5352 # that screen readers at least get some description of the image
5353 $params['frame']['alt'] = $title->getText();
5354 }
5355 # Do not set $params['frame']['title'] because tooltips don't make sense
5356 # for framed images
5357 } else { # Inline image
5358 if ( !isset( $params['frame']['alt'] ) ) {
5359 # No alt text, use the "caption" for the alt text
5360 if ( $caption !== '' ) {
5361 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5362 } else {
5363 # No caption, fall back to using the filename for the
5364 # alt text
5365 $params['frame']['alt'] = $title->getText();
5366 }
5367 }
5368 # Use the "caption" for the tooltip text
5369 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5370 }
5371
5372 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5373
5374 # Linker does the rest
5375 $time = isset( $options['time'] ) ? $options['time'] : false;
5376 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5377 $time, $descQuery, $this->mOptions->getThumbSize() );
5378
5379 # Give the handler a chance to modify the parser object
5380 if ( $handler ) {
5381 $handler->parserTransformHook( $this, $file );
5382 }
5383
5384 return $ret;
5385 }
5386
5387 /**
5388 * @param string $caption
5389 * @param LinkHolderArray|bool $holders
5390 * @return mixed|string
5391 */
5392 protected function stripAltText( $caption, $holders ) {
5393 # Strip bad stuff out of the title (tooltip). We can't just use
5394 # replaceLinkHoldersText() here, because if this function is called
5395 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5396 if ( $holders ) {
5397 $tooltip = $holders->replaceText( $caption );
5398 } else {
5399 $tooltip = $this->replaceLinkHoldersText( $caption );
5400 }
5401
5402 # make sure there are no placeholders in thumbnail attributes
5403 # that are later expanded to html- so expand them now and
5404 # remove the tags
5405 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5406 $tooltip = Sanitizer::stripAllTags( $tooltip );
5407
5408 return $tooltip;
5409 }
5410
5411 /**
5412 * Set a flag in the output object indicating that the content is dynamic and
5413 * shouldn't be cached.
5414 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5415 */
5416 public function disableCache() {
5417 wfDebug( "Parser output marked as uncacheable.\n" );
5418 if ( !$this->mOutput ) {
5419 throw new MWException( __METHOD__ .
5420 " can only be called when actually parsing something" );
5421 }
5422 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5423 }
5424
5425 /**
5426 * Callback from the Sanitizer for expanding items found in HTML attribute
5427 * values, so they can be safely tested and escaped.
5428 *
5429 * @param string $text
5430 * @param bool|PPFrame $frame
5431 * @return string
5432 */
5433 public function attributeStripCallback( &$text, $frame = false ) {
5434 $text = $this->replaceVariables( $text, $frame );
5435 $text = $this->mStripState->unstripBoth( $text );
5436 return $text;
5437 }
5438
5439 /**
5440 * Accessor
5441 *
5442 * @return array
5443 */
5444 public function getTags() {
5445 return array_merge(
5446 array_keys( $this->mTransparentTagHooks ),
5447 array_keys( $this->mTagHooks ),
5448 array_keys( $this->mFunctionTagHooks )
5449 );
5450 }
5451
5452 /**
5453 * Replace transparent tags in $text with the values given by the callbacks.
5454 *
5455 * Transparent tag hooks are like regular XML-style tag hooks, except they
5456 * operate late in the transformation sequence, on HTML instead of wikitext.
5457 *
5458 * @param string $text
5459 *
5460 * @return string
5461 */
5462 public function replaceTransparentTags( $text ) {
5463 $matches = [];
5464 $elements = array_keys( $this->mTransparentTagHooks );
5465 $text = self::extractTagsAndParams( $elements, $text, $matches );
5466 $replacements = [];
5467
5468 foreach ( $matches as $marker => $data ) {
5469 list( $element, $content, $params, $tag ) = $data;
5470 $tagName = strtolower( $element );
5471 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5472 $output = call_user_func_array(
5473 $this->mTransparentTagHooks[$tagName],
5474 [ $content, $params, $this ]
5475 );
5476 } else {
5477 $output = $tag;
5478 }
5479 $replacements[$marker] = $output;
5480 }
5481 return strtr( $text, $replacements );
5482 }
5483
5484 /**
5485 * Break wikitext input into sections, and either pull or replace
5486 * some particular section's text.
5487 *
5488 * External callers should use the getSection and replaceSection methods.
5489 *
5490 * @param string $text Page wikitext
5491 * @param string|int $sectionId A section identifier string of the form:
5492 * "<flag1> - <flag2> - ... - <section number>"
5493 *
5494 * Currently the only recognised flag is "T", which means the target section number
5495 * was derived during a template inclusion parse, in other words this is a template
5496 * section edit link. If no flags are given, it was an ordinary section edit link.
5497 * This flag is required to avoid a section numbering mismatch when a section is
5498 * enclosed by "<includeonly>" (T8563).
5499 *
5500 * The section number 0 pulls the text before the first heading; other numbers will
5501 * pull the given section along with its lower-level subsections. If the section is
5502 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5503 *
5504 * Section 0 is always considered to exist, even if it only contains the empty
5505 * string. If $text is the empty string and section 0 is replaced, $newText is
5506 * returned.
5507 *
5508 * @param string $mode One of "get" or "replace"
5509 * @param string $newText Replacement text for section data.
5510 * @return string For "get", the extracted section text.
5511 * for "replace", the whole page with the section replaced.
5512 */
5513 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5514 global $wgTitle; # not generally used but removes an ugly failure mode
5515
5516 $magicScopeVariable = $this->lock();
5517 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5518 $outText = '';
5519 $frame = $this->getPreprocessor()->newFrame();
5520
5521 # Process section extraction flags
5522 $flags = 0;
5523 $sectionParts = explode( '-', $sectionId );
5524 $sectionIndex = array_pop( $sectionParts );
5525 foreach ( $sectionParts as $part ) {
5526 if ( $part === 'T' ) {
5527 $flags |= self::PTD_FOR_INCLUSION;
5528 }
5529 }
5530
5531 # Check for empty input
5532 if ( strval( $text ) === '' ) {
5533 # Only sections 0 and T-0 exist in an empty document
5534 if ( $sectionIndex == 0 ) {
5535 if ( $mode === 'get' ) {
5536 return '';
5537 } else {
5538 return $newText;
5539 }
5540 } else {
5541 if ( $mode === 'get' ) {
5542 return $newText;
5543 } else {
5544 return $text;
5545 }
5546 }
5547 }
5548
5549 # Preprocess the text
5550 $root = $this->preprocessToDom( $text, $flags );
5551
5552 # <h> nodes indicate section breaks
5553 # They can only occur at the top level, so we can find them by iterating the root's children
5554 $node = $root->getFirstChild();
5555
5556 # Find the target section
5557 if ( $sectionIndex == 0 ) {
5558 # Section zero doesn't nest, level=big
5559 $targetLevel = 1000;
5560 } else {
5561 while ( $node ) {
5562 if ( $node->getName() === 'h' ) {
5563 $bits = $node->splitHeading();
5564 if ( $bits['i'] == $sectionIndex ) {
5565 $targetLevel = $bits['level'];
5566 break;
5567 }
5568 }
5569 if ( $mode === 'replace' ) {
5570 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5571 }
5572 $node = $node->getNextSibling();
5573 }
5574 }
5575
5576 if ( !$node ) {
5577 # Not found
5578 if ( $mode === 'get' ) {
5579 return $newText;
5580 } else {
5581 return $text;
5582 }
5583 }
5584
5585 # Find the end of the section, including nested sections
5586 do {
5587 if ( $node->getName() === 'h' ) {
5588 $bits = $node->splitHeading();
5589 $curLevel = $bits['level'];
5590 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5591 break;
5592 }
5593 }
5594 if ( $mode === 'get' ) {
5595 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5596 }
5597 $node = $node->getNextSibling();
5598 } while ( $node );
5599
5600 # Write out the remainder (in replace mode only)
5601 if ( $mode === 'replace' ) {
5602 # Output the replacement text
5603 # Add two newlines on -- trailing whitespace in $newText is conventionally
5604 # stripped by the editor, so we need both newlines to restore the paragraph gap
5605 # Only add trailing whitespace if there is newText
5606 if ( $newText != "" ) {
5607 $outText .= $newText . "\n\n";
5608 }
5609
5610 while ( $node ) {
5611 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5612 $node = $node->getNextSibling();
5613 }
5614 }
5615
5616 if ( is_string( $outText ) ) {
5617 # Re-insert stripped tags
5618 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5619 }
5620
5621 return $outText;
5622 }
5623
5624 /**
5625 * This function returns the text of a section, specified by a number ($section).
5626 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5627 * the first section before any such heading (section 0).
5628 *
5629 * If a section contains subsections, these are also returned.
5630 *
5631 * @param string $text Text to look in
5632 * @param string|int $sectionId Section identifier as a number or string
5633 * (e.g. 0, 1 or 'T-1').
5634 * @param string $defaultText Default to return if section is not found
5635 *
5636 * @return string Text of the requested section
5637 */
5638 public function getSection( $text, $sectionId, $defaultText = '' ) {
5639 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5640 }
5641
5642 /**
5643 * This function returns $oldtext after the content of the section
5644 * specified by $section has been replaced with $text. If the target
5645 * section does not exist, $oldtext is returned unchanged.
5646 *
5647 * @param string $oldText Former text of the article
5648 * @param string|int $sectionId Section identifier as a number or string
5649 * (e.g. 0, 1 or 'T-1').
5650 * @param string $newText Replacing text
5651 *
5652 * @return string Modified text
5653 */
5654 public function replaceSection( $oldText, $sectionId, $newText ) {
5655 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5656 }
5657
5658 /**
5659 * Get the ID of the revision we are parsing
5660 *
5661 * @return int|null
5662 */
5663 public function getRevisionId() {
5664 return $this->mRevisionId;
5665 }
5666
5667 /**
5668 * Get the revision object for $this->mRevisionId
5669 *
5670 * @return Revision|null Either a Revision object or null
5671 * @since 1.23 (public since 1.23)
5672 */
5673 public function getRevisionObject() {
5674 if ( !is_null( $this->mRevisionObject ) ) {
5675 return $this->mRevisionObject;
5676 }
5677 if ( is_null( $this->mRevisionId ) ) {
5678 return null;
5679 }
5680
5681 $rev = call_user_func(
5682 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5683 );
5684
5685 # If the parse is for a new revision, then the callback should have
5686 # already been set to force the object and should match mRevisionId.
5687 # If not, try to fetch by mRevisionId for sanity.
5688 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5689 $rev = Revision::newFromId( $this->mRevisionId );
5690 }
5691
5692 $this->mRevisionObject = $rev;
5693
5694 return $this->mRevisionObject;
5695 }
5696
5697 /**
5698 * Get the timestamp associated with the current revision, adjusted for
5699 * the default server-local timestamp
5700 * @return string
5701 */
5702 public function getRevisionTimestamp() {
5703 if ( is_null( $this->mRevisionTimestamp ) ) {
5704 global $wgContLang;
5705
5706 $revObject = $this->getRevisionObject();
5707 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5708
5709 # The cryptic '' timezone parameter tells to use the site-default
5710 # timezone offset instead of the user settings.
5711 # Since this value will be saved into the parser cache, served
5712 # to other users, and potentially even used inside links and such,
5713 # it needs to be consistent for all visitors.
5714 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5715
5716 }
5717 return $this->mRevisionTimestamp;
5718 }
5719
5720 /**
5721 * Get the name of the user that edited the last revision
5722 *
5723 * @return string User name
5724 */
5725 public function getRevisionUser() {
5726 if ( is_null( $this->mRevisionUser ) ) {
5727 $revObject = $this->getRevisionObject();
5728
5729 # if this template is subst: the revision id will be blank,
5730 # so just use the current user's name
5731 if ( $revObject ) {
5732 $this->mRevisionUser = $revObject->getUserText();
5733 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5734 $this->mRevisionUser = $this->getUser()->getName();
5735 }
5736 }
5737 return $this->mRevisionUser;
5738 }
5739
5740 /**
5741 * Get the size of the revision
5742 *
5743 * @return int|null Revision size
5744 */
5745 public function getRevisionSize() {
5746 if ( is_null( $this->mRevisionSize ) ) {
5747 $revObject = $this->getRevisionObject();
5748
5749 # if this variable is subst: the revision id will be blank,
5750 # so just use the parser input size, because the own substituation
5751 # will change the size.
5752 if ( $revObject ) {
5753 $this->mRevisionSize = $revObject->getSize();
5754 } else {
5755 $this->mRevisionSize = $this->mInputSize;
5756 }
5757 }
5758 return $this->mRevisionSize;
5759 }
5760
5761 /**
5762 * Mutator for $mDefaultSort
5763 *
5764 * @param string $sort New value
5765 */
5766 public function setDefaultSort( $sort ) {
5767 $this->mDefaultSort = $sort;
5768 $this->mOutput->setProperty( 'defaultsort', $sort );
5769 }
5770
5771 /**
5772 * Accessor for $mDefaultSort
5773 * Will use the empty string if none is set.
5774 *
5775 * This value is treated as a prefix, so the
5776 * empty string is equivalent to sorting by
5777 * page name.
5778 *
5779 * @return string
5780 */
5781 public function getDefaultSort() {
5782 if ( $this->mDefaultSort !== false ) {
5783 return $this->mDefaultSort;
5784 } else {
5785 return '';
5786 }
5787 }
5788
5789 /**
5790 * Accessor for $mDefaultSort
5791 * Unlike getDefaultSort(), will return false if none is set
5792 *
5793 * @return string|bool
5794 */
5795 public function getCustomDefaultSort() {
5796 return $this->mDefaultSort;
5797 }
5798
5799 /**
5800 * Try to guess the section anchor name based on a wikitext fragment
5801 * presumably extracted from a heading, for example "Header" from
5802 * "== Header ==".
5803 *
5804 * @param string $text
5805 *
5806 * @return string
5807 */
5808 public function guessSectionNameFromWikiText( $text ) {
5809 # Strip out wikitext links(they break the anchor)
5810 $text = $this->stripSectionName( $text );
5811 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5812 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5813 }
5814
5815 /**
5816 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5817 * instead. For use in redirects, since IE6 interprets Redirect: headers
5818 * as something other than UTF-8 (apparently?), resulting in breakage.
5819 *
5820 * @param string $text The section name
5821 * @return string An anchor
5822 */
5823 public function guessLegacySectionNameFromWikiText( $text ) {
5824 # Strip out wikitext links(they break the anchor)
5825 $text = $this->stripSectionName( $text );
5826 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5827 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5828 }
5829
5830 /**
5831 * Strips a text string of wikitext for use in a section anchor
5832 *
5833 * Accepts a text string and then removes all wikitext from the
5834 * string and leaves only the resultant text (i.e. the result of
5835 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5836 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5837 * to create valid section anchors by mimicing the output of the
5838 * parser when headings are parsed.
5839 *
5840 * @param string $text Text string to be stripped of wikitext
5841 * for use in a Section anchor
5842 * @return string Filtered text string
5843 */
5844 public function stripSectionName( $text ) {
5845 # Strip internal link markup
5846 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5847 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5848
5849 # Strip external link markup
5850 # @todo FIXME: Not tolerant to blank link text
5851 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5852 # on how many empty links there are on the page - need to figure that out.
5853 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5854
5855 # Parse wikitext quotes (italics & bold)
5856 $text = $this->doQuotes( $text );
5857
5858 # Strip HTML tags
5859 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5860 return $text;
5861 }
5862
5863 /**
5864 * strip/replaceVariables/unstrip for preprocessor regression testing
5865 *
5866 * @param string $text
5867 * @param Title $title
5868 * @param ParserOptions $options
5869 * @param int $outputType
5870 *
5871 * @return string
5872 */
5873 public function testSrvus( $text, Title $title, ParserOptions $options,
5874 $outputType = self::OT_HTML
5875 ) {
5876 $magicScopeVariable = $this->lock();
5877 $this->startParse( $title, $options, $outputType, true );
5878
5879 $text = $this->replaceVariables( $text );
5880 $text = $this->mStripState->unstripBoth( $text );
5881 $text = Sanitizer::removeHTMLtags( $text );
5882 return $text;
5883 }
5884
5885 /**
5886 * @param string $text
5887 * @param Title $title
5888 * @param ParserOptions $options
5889 * @return string
5890 */
5891 public function testPst( $text, Title $title, ParserOptions $options ) {
5892 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5893 }
5894
5895 /**
5896 * @param string $text
5897 * @param Title $title
5898 * @param ParserOptions $options
5899 * @return string
5900 */
5901 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5902 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5903 }
5904
5905 /**
5906 * Call a callback function on all regions of the given text that are not
5907 * inside strip markers, and replace those regions with the return value
5908 * of the callback. For example, with input:
5909 *
5910 * aaa<MARKER>bbb
5911 *
5912 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5913 * two strings will be replaced with the value returned by the callback in
5914 * each case.
5915 *
5916 * @param string $s
5917 * @param callable $callback
5918 *
5919 * @return string
5920 */
5921 public function markerSkipCallback( $s, $callback ) {
5922 $i = 0;
5923 $out = '';
5924 while ( $i < strlen( $s ) ) {
5925 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5926 if ( $markerStart === false ) {
5927 $out .= call_user_func( $callback, substr( $s, $i ) );
5928 break;
5929 } else {
5930 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5931 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5932 if ( $markerEnd === false ) {
5933 $out .= substr( $s, $markerStart );
5934 break;
5935 } else {
5936 $markerEnd += strlen( self::MARKER_SUFFIX );
5937 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5938 $i = $markerEnd;
5939 }
5940 }
5941 }
5942 return $out;
5943 }
5944
5945 /**
5946 * Remove any strip markers found in the given text.
5947 *
5948 * @param string $text Input string
5949 * @return string
5950 */
5951 public function killMarkers( $text ) {
5952 return $this->mStripState->killMarkers( $text );
5953 }
5954
5955 /**
5956 * Save the parser state required to convert the given half-parsed text to
5957 * HTML. "Half-parsed" in this context means the output of
5958 * recursiveTagParse() or internalParse(). This output has strip markers
5959 * from replaceVariables (extensionSubstitution() etc.), and link
5960 * placeholders from replaceLinkHolders().
5961 *
5962 * Returns an array which can be serialized and stored persistently. This
5963 * array can later be loaded into another parser instance with
5964 * unserializeHalfParsedText(). The text can then be safely incorporated into
5965 * the return value of a parser hook.
5966 *
5967 * @param string $text
5968 *
5969 * @return array
5970 */
5971 public function serializeHalfParsedText( $text ) {
5972 $data = [
5973 'text' => $text,
5974 'version' => self::HALF_PARSED_VERSION,
5975 'stripState' => $this->mStripState->getSubState( $text ),
5976 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5977 ];
5978 return $data;
5979 }
5980
5981 /**
5982 * Load the parser state given in the $data array, which is assumed to
5983 * have been generated by serializeHalfParsedText(). The text contents is
5984 * extracted from the array, and its markers are transformed into markers
5985 * appropriate for the current Parser instance. This transformed text is
5986 * returned, and can be safely included in the return value of a parser
5987 * hook.
5988 *
5989 * If the $data array has been stored persistently, the caller should first
5990 * check whether it is still valid, by calling isValidHalfParsedText().
5991 *
5992 * @param array $data Serialized data
5993 * @throws MWException
5994 * @return string
5995 */
5996 public function unserializeHalfParsedText( $data ) {
5997 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5998 throw new MWException( __METHOD__ . ': invalid version' );
5999 }
6000
6001 # First, extract the strip state.
6002 $texts = [ $data['text'] ];
6003 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6004
6005 # Now renumber links
6006 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6007
6008 # Should be good to go.
6009 return $texts[0];
6010 }
6011
6012 /**
6013 * Returns true if the given array, presumed to be generated by
6014 * serializeHalfParsedText(), is compatible with the current version of the
6015 * parser.
6016 *
6017 * @param array $data
6018 *
6019 * @return bool
6020 */
6021 public function isValidHalfParsedText( $data ) {
6022 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6023 }
6024
6025 /**
6026 * Parsed a width param of imagelink like 300px or 200x300px
6027 *
6028 * @param string $value
6029 *
6030 * @return array
6031 * @since 1.20
6032 */
6033 public function parseWidthParam( $value ) {
6034 $parsedWidthParam = [];
6035 if ( $value === '' ) {
6036 return $parsedWidthParam;
6037 }
6038 $m = [];
6039 # (T15500) In both cases (width/height and width only),
6040 # permit trailing "px" for backward compatibility.
6041 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6042 $width = intval( $m[1] );
6043 $height = intval( $m[2] );
6044 $parsedWidthParam['width'] = $width;
6045 $parsedWidthParam['height'] = $height;
6046 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6047 $width = intval( $value );
6048 $parsedWidthParam['width'] = $width;
6049 }
6050 return $parsedWidthParam;
6051 }
6052
6053 /**
6054 * Lock the current instance of the parser.
6055 *
6056 * This is meant to stop someone from calling the parser
6057 * recursively and messing up all the strip state.
6058 *
6059 * @throws MWException If parser is in a parse
6060 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6061 */
6062 protected function lock() {
6063 if ( $this->mInParse ) {
6064 throw new MWException( "Parser state cleared while parsing. "
6065 . "Did you call Parser::parse recursively?" );
6066 }
6067 $this->mInParse = true;
6068
6069 $recursiveCheck = new ScopedCallback( function() {
6070 $this->mInParse = false;
6071 } );
6072
6073 return $recursiveCheck;
6074 }
6075
6076 /**
6077 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6078 *
6079 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6080 * or if there is more than one <p/> tag in the input HTML.
6081 *
6082 * @param string $html
6083 * @return string
6084 * @since 1.24
6085 */
6086 public static function stripOuterParagraph( $html ) {
6087 $m = [];
6088 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6089 if ( strpos( $m[1], '</p>' ) === false ) {
6090 $html = $m[1];
6091 }
6092 }
6093
6094 return $html;
6095 }
6096
6097 /**
6098 * Return this parser if it is not doing anything, otherwise
6099 * get a fresh parser. You can use this method by doing
6100 * $myParser = $wgParser->getFreshParser(), or more simply
6101 * $wgParser->getFreshParser()->parse( ... );
6102 * if you're unsure if $wgParser is safe to use.
6103 *
6104 * @since 1.24
6105 * @return Parser A parser object that is not parsing anything
6106 */
6107 public function getFreshParser() {
6108 global $wgParserConf;
6109 if ( $this->mInParse ) {
6110 return new $wgParserConf['class']( $wgParserConf );
6111 } else {
6112 return $this;
6113 }
6114 }
6115
6116 /**
6117 * Set's up the PHP implementation of OOUI for use in this request
6118 * and instructs OutputPage to enable OOUI for itself.
6119 *
6120 * @since 1.26
6121 */
6122 public function enableOOUI() {
6123 OutputPage::setupOOUI();
6124 $this->mOutput->setEnableOOUI( true );
6125 }
6126 }
Wiklou, le Wiki du Wiklou