dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Move around "ا" to after "آ" and not before"
[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 = self::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|string 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 // Run deprecated hook
554 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
555
556 // Sanitize for comment. Note '‐' in the replacement is U+2010,
557 // which looks much like the problematic '-'.
558 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
559 $text .= "\n<!-- \n$limitReport-->\n";
560
561 // Add on template profiling data in human/machine readable way
562 $dataByFunc = $this->mProfiler->getFunctionStats();
563 uasort( $dataByFunc, function ( $a, $b ) {
564 return $a['real'] < $b['real']; // descending order
565 } );
566 $profileReport = [];
567 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
568 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
569 $item['%real'], $item['real'], $item['calls'],
570 htmlspecialchars( $item['name'] ) );
571 }
572 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
573 $text .= implode( "\n", $profileReport ) . "\n-->\n";
574
575 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
576
577 // Add other cache related metadata
578 if ( $wgShowHostnames ) {
579 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
580 }
581 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
582 $this->mOutput->getCacheTime() );
583 $this->mOutput->setLimitReportData( 'cachereport-ttl',
584 $this->mOutput->getCacheExpiry() );
585 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
586 $this->mOutput->hasDynamicContent() );
587
588 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
589 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
590 $this->mTitle->getPrefixedDBkey() );
591 }
592 }
593
594 # Wrap non-interface parser output in a <div> so it can be targeted
595 # with CSS (T37247)
596 $class = $this->mOptions->getWrapOutputClass();
597 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
598 $text = Html::rawElement( 'div', [ 'class' => $class ], $text );
599 }
600
601 $this->mOutput->setText( $text );
602
603 $this->mRevisionId = $oldRevisionId;
604 $this->mRevisionObject = $oldRevisionObject;
605 $this->mRevisionTimestamp = $oldRevisionTimestamp;
606 $this->mRevisionUser = $oldRevisionUser;
607 $this->mRevisionSize = $oldRevisionSize;
608 $this->mInputSize = false;
609 $this->currentRevisionCache = null;
610
611 return $this->mOutput;
612 }
613
614 /**
615 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
616 * can be called from an extension tag hook.
617 *
618 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
619 * instead, which means that lists and links have not been fully parsed yet,
620 * and strip markers are still present.
621 *
622 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
623 *
624 * Use this function if you're a parser tag hook and you want to parse
625 * wikitext before or after applying additional transformations, and you
626 * intend to *return the result as hook output*, which will cause it to go
627 * through the rest of parsing process automatically.
628 *
629 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
630 * $text are not expanded
631 *
632 * @param string $text Text extension wants to have parsed
633 * @param bool|PPFrame $frame The frame to use for expanding any template variables
634 * @return string UNSAFE half-parsed HTML
635 */
636 public function recursiveTagParse( $text, $frame = false ) {
637 // Avoid PHP 7.1 warning from passing $this by reference
638 $parser = $this;
639 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
640 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
641 $text = $this->internalParse( $text, false, $frame );
642 return $text;
643 }
644
645 /**
646 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
647 * point can be called from an extension tag hook.
648 *
649 * The output of this function is fully-parsed HTML that is safe for output.
650 * If you're a parser tag hook, you might want to use recursiveTagParse()
651 * instead.
652 *
653 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
654 * $text are not expanded
655 *
656 * @since 1.25
657 *
658 * @param string $text Text extension wants to have parsed
659 * @param bool|PPFrame $frame The frame to use for expanding any template variables
660 * @return string Fully parsed HTML
661 */
662 public function recursiveTagParseFully( $text, $frame = false ) {
663 $text = $this->recursiveTagParse( $text, $frame );
664 $text = $this->internalParseHalfParsed( $text, false );
665 return $text;
666 }
667
668 /**
669 * Expand templates and variables in the text, producing valid, static wikitext.
670 * Also removes comments.
671 * Do not call this function recursively.
672 * @param string $text
673 * @param Title $title
674 * @param ParserOptions $options
675 * @param int|null $revid
676 * @param bool|PPFrame $frame
677 * @return mixed|string
678 */
679 public function preprocess( $text, Title $title = null,
680 ParserOptions $options, $revid = null, $frame = false
681 ) {
682 $magicScopeVariable = $this->lock();
683 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
684 if ( $revid !== null ) {
685 $this->mRevisionId = $revid;
686 }
687 // Avoid PHP 7.1 warning from passing $this by reference
688 $parser = $this;
689 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
690 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
691 $text = $this->replaceVariables( $text, $frame );
692 $text = $this->mStripState->unstripBoth( $text );
693 return $text;
694 }
695
696 /**
697 * Recursive parser entry point that can be called from an extension tag
698 * hook.
699 *
700 * @param string $text Text to be expanded
701 * @param bool|PPFrame $frame The frame to use for expanding any template variables
702 * @return string
703 * @since 1.19
704 */
705 public function recursivePreprocess( $text, $frame = false ) {
706 $text = $this->replaceVariables( $text, $frame );
707 $text = $this->mStripState->unstripBoth( $text );
708 return $text;
709 }
710
711 /**
712 * Process the wikitext for the "?preload=" feature. (T7210)
713 *
714 * "<noinclude>", "<includeonly>" etc. are parsed as for template
715 * transclusion, comments, templates, arguments, tags hooks and parser
716 * functions are untouched.
717 *
718 * @param string $text
719 * @param Title $title
720 * @param ParserOptions $options
721 * @param array $params
722 * @return string
723 */
724 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
725 $msg = new RawMessage( $text );
726 $text = $msg->params( $params )->plain();
727
728 # Parser (re)initialisation
729 $magicScopeVariable = $this->lock();
730 $this->startParse( $title, $options, self::OT_PLAIN, true );
731
732 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
733 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
734 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
735 $text = $this->mStripState->unstripBoth( $text );
736 return $text;
737 }
738
739 /**
740 * Set the current user.
741 * Should only be used when doing pre-save transform.
742 *
743 * @param User|null $user User object or null (to reset)
744 */
745 public function setUser( $user ) {
746 $this->mUser = $user;
747 }
748
749 /**
750 * Set the context title
751 *
752 * @param Title $t
753 */
754 public function setTitle( $t ) {
755 if ( !$t ) {
756 $t = Title::newFromText( 'NO TITLE' );
757 }
758
759 if ( $t->hasFragment() ) {
760 # Strip the fragment to avoid various odd effects
761 $this->mTitle = $t->createFragmentTarget( '' );
762 } else {
763 $this->mTitle = $t;
764 }
765 }
766
767 /**
768 * Accessor for the Title object
769 *
770 * @return Title
771 */
772 public function getTitle() {
773 return $this->mTitle;
774 }
775
776 /**
777 * Accessor/mutator for the Title object
778 *
779 * @param Title $x Title object or null to just get the current one
780 * @return Title
781 */
782 public function Title( $x = null ) {
783 return wfSetVar( $this->mTitle, $x );
784 }
785
786 /**
787 * Set the output type
788 *
789 * @param int $ot New value
790 */
791 public function setOutputType( $ot ) {
792 $this->mOutputType = $ot;
793 # Shortcut alias
794 $this->ot = [
795 'html' => $ot == self::OT_HTML,
796 'wiki' => $ot == self::OT_WIKI,
797 'pre' => $ot == self::OT_PREPROCESS,
798 'plain' => $ot == self::OT_PLAIN,
799 ];
800 }
801
802 /**
803 * Accessor/mutator for the output type
804 *
805 * @param int|null $x New value or null to just get the current one
806 * @return int
807 */
808 public function OutputType( $x = null ) {
809 return wfSetVar( $this->mOutputType, $x );
810 }
811
812 /**
813 * Get the ParserOutput object
814 *
815 * @return ParserOutput
816 */
817 public function getOutput() {
818 return $this->mOutput;
819 }
820
821 /**
822 * Get the ParserOptions object
823 *
824 * @return ParserOptions
825 */
826 public function getOptions() {
827 return $this->mOptions;
828 }
829
830 /**
831 * Accessor/mutator for the ParserOptions object
832 *
833 * @param ParserOptions $x New value or null to just get the current one
834 * @return ParserOptions Current ParserOptions object
835 */
836 public function Options( $x = null ) {
837 return wfSetVar( $this->mOptions, $x );
838 }
839
840 /**
841 * @return int
842 */
843 public function nextLinkID() {
844 return $this->mLinkID++;
845 }
846
847 /**
848 * @param int $id
849 */
850 public function setLinkID( $id ) {
851 $this->mLinkID = $id;
852 }
853
854 /**
855 * Get a language object for use in parser functions such as {{FORMATNUM:}}
856 * @return Language
857 */
858 public function getFunctionLang() {
859 return $this->getTargetLanguage();
860 }
861
862 /**
863 * Get the target language for the content being parsed. This is usually the
864 * language that the content is in.
865 *
866 * @since 1.19
867 *
868 * @throws MWException
869 * @return Language
870 */
871 public function getTargetLanguage() {
872 $target = $this->mOptions->getTargetLanguage();
873
874 if ( $target !== null ) {
875 return $target;
876 } elseif ( $this->mOptions->getInterfaceMessage() ) {
877 return $this->mOptions->getUserLangObj();
878 } elseif ( is_null( $this->mTitle ) ) {
879 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
880 }
881
882 return $this->mTitle->getPageLanguage();
883 }
884
885 /**
886 * Get the language object for language conversion
887 * @return Language|null
888 */
889 public function getConverterLanguage() {
890 return $this->getTargetLanguage();
891 }
892
893 /**
894 * Get a User object either from $this->mUser, if set, or from the
895 * ParserOptions object otherwise
896 *
897 * @return User
898 */
899 public function getUser() {
900 if ( !is_null( $this->mUser ) ) {
901 return $this->mUser;
902 }
903 return $this->mOptions->getUser();
904 }
905
906 /**
907 * Get a preprocessor object
908 *
909 * @return Preprocessor
910 */
911 public function getPreprocessor() {
912 if ( !isset( $this->mPreprocessor ) ) {
913 $class = $this->mPreprocessorClass;
914 $this->mPreprocessor = new $class( $this );
915 }
916 return $this->mPreprocessor;
917 }
918
919 /**
920 * Get a LinkRenderer instance to make links with
921 *
922 * @since 1.28
923 * @return LinkRenderer
924 */
925 public function getLinkRenderer() {
926 if ( !$this->mLinkRenderer ) {
927 $this->mLinkRenderer = MediaWikiServices::getInstance()
928 ->getLinkRendererFactory()->create();
929 $this->mLinkRenderer->setStubThreshold(
930 $this->getOptions()->getStubThreshold()
931 );
932 }
933
934 return $this->mLinkRenderer;
935 }
936
937 /**
938 * Replaces all occurrences of HTML-style comments and the given tags
939 * in the text with a random marker and returns the next text. The output
940 * parameter $matches will be an associative array filled with data in
941 * the form:
942 *
943 * @code
944 * 'UNIQ-xxxxx' => [
945 * 'element',
946 * 'tag content',
947 * [ 'param' => 'x' ],
948 * '<element param="x">tag content</element>' ]
949 * @endcode
950 *
951 * @param array $elements List of element names. Comments are always extracted.
952 * @param string $text Source text string.
953 * @param array &$matches Out parameter, Array: extracted tags
954 * @return string Stripped text
955 */
956 public static function extractTagsAndParams( $elements, $text, &$matches ) {
957 static $n = 1;
958 $stripped = '';
959 $matches = [];
960
961 $taglist = implode( '|', $elements );
962 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
963
964 while ( $text != '' ) {
965 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
966 $stripped .= $p[0];
967 if ( count( $p ) < 5 ) {
968 break;
969 }
970 if ( count( $p ) > 5 ) {
971 # comment
972 $element = $p[4];
973 $attributes = '';
974 $close = '';
975 $inside = $p[5];
976 } else {
977 # tag
978 $element = $p[1];
979 $attributes = $p[2];
980 $close = $p[3];
981 $inside = $p[4];
982 }
983
984 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
985 $stripped .= $marker;
986
987 if ( $close === '/>' ) {
988 # Empty element tag, <tag />
989 $content = null;
990 $text = $inside;
991 $tail = null;
992 } else {
993 if ( $element === '!--' ) {
994 $end = '/(-->)/';
995 } else {
996 $end = "/(<\\/$element\\s*>)/i";
997 }
998 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
999 $content = $q[0];
1000 if ( count( $q ) < 3 ) {
1001 # No end tag -- let it run out to the end of the text.
1002 $tail = '';
1003 $text = '';
1004 } else {
1005 $tail = $q[1];
1006 $text = $q[2];
1007 }
1008 }
1009
1010 $matches[$marker] = [ $element,
1011 $content,
1012 Sanitizer::decodeTagAttributes( $attributes ),
1013 "<$element$attributes$close$content$tail" ];
1014 }
1015 return $stripped;
1016 }
1017
1018 /**
1019 * Get a list of strippable XML-like elements
1020 *
1021 * @return array
1022 */
1023 public function getStripList() {
1024 return $this->mStripList;
1025 }
1026
1027 /**
1028 * Add an item to the strip state
1029 * Returns the unique tag which must be inserted into the stripped text
1030 * The tag will be replaced with the original text in unstrip()
1031 *
1032 * @param string $text
1033 *
1034 * @return string
1035 */
1036 public function insertStripItem( $text ) {
1037 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1038 $this->mMarkerIndex++;
1039 $this->mStripState->addGeneral( $marker, $text );
1040 return $marker;
1041 }
1042
1043 /**
1044 * parse the wiki syntax used to render tables
1045 *
1046 * @private
1047 * @param string $text
1048 * @return string
1049 */
1050 public function doTableStuff( $text ) {
1051 $lines = StringUtils::explode( "\n", $text );
1052 $out = '';
1053 $td_history = []; # Is currently a td tag open?
1054 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1055 $tr_history = []; # Is currently a tr tag open?
1056 $tr_attributes = []; # history of tr attributes
1057 $has_opened_tr = []; # Did this table open a <tr> element?
1058 $indent_level = 0; # indent level of the table
1059
1060 foreach ( $lines as $outLine ) {
1061 $line = trim( $outLine );
1062
1063 if ( $line === '' ) { # empty line, go to next line
1064 $out .= $outLine . "\n";
1065 continue;
1066 }
1067
1068 $first_character = $line[0];
1069 $first_two = substr( $line, 0, 2 );
1070 $matches = [];
1071
1072 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1073 # First check if we are starting a new table
1074 $indent_level = strlen( $matches[1] );
1075
1076 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1077 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1078
1079 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1080 array_push( $td_history, false );
1081 array_push( $last_tag_history, '' );
1082 array_push( $tr_history, false );
1083 array_push( $tr_attributes, '' );
1084 array_push( $has_opened_tr, false );
1085 } elseif ( count( $td_history ) == 0 ) {
1086 # Don't do any of the following
1087 $out .= $outLine . "\n";
1088 continue;
1089 } elseif ( $first_two === '|}' ) {
1090 # We are ending a table
1091 $line = '</table>' . substr( $line, 2 );
1092 $last_tag = array_pop( $last_tag_history );
1093
1094 if ( !array_pop( $has_opened_tr ) ) {
1095 $line = "<tr><td></td></tr>{$line}";
1096 }
1097
1098 if ( array_pop( $tr_history ) ) {
1099 $line = "</tr>{$line}";
1100 }
1101
1102 if ( array_pop( $td_history ) ) {
1103 $line = "</{$last_tag}>{$line}";
1104 }
1105 array_pop( $tr_attributes );
1106 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1107 } elseif ( $first_two === '|-' ) {
1108 # Now we have a table row
1109 $line = preg_replace( '#^\|-+#', '', $line );
1110
1111 # Whats after the tag is now only attributes
1112 $attributes = $this->mStripState->unstripBoth( $line );
1113 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1114 array_pop( $tr_attributes );
1115 array_push( $tr_attributes, $attributes );
1116
1117 $line = '';
1118 $last_tag = array_pop( $last_tag_history );
1119 array_pop( $has_opened_tr );
1120 array_push( $has_opened_tr, true );
1121
1122 if ( array_pop( $tr_history ) ) {
1123 $line = '</tr>';
1124 }
1125
1126 if ( array_pop( $td_history ) ) {
1127 $line = "</{$last_tag}>{$line}";
1128 }
1129
1130 $outLine = $line;
1131 array_push( $tr_history, false );
1132 array_push( $td_history, false );
1133 array_push( $last_tag_history, '' );
1134 } elseif ( $first_character === '|'
1135 || $first_character === '!'
1136 || $first_two === '|+'
1137 ) {
1138 # This might be cell elements, td, th or captions
1139 if ( $first_two === '|+' ) {
1140 $first_character = '+';
1141 $line = substr( $line, 2 );
1142 } else {
1143 $line = substr( $line, 1 );
1144 }
1145
1146 // Implies both are valid for table headings.
1147 if ( $first_character === '!' ) {
1148 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1149 }
1150
1151 # Split up multiple cells on the same line.
1152 # FIXME : This can result in improper nesting of tags processed
1153 # by earlier parser steps.
1154 $cells = explode( '||', $line );
1155
1156 $outLine = '';
1157
1158 # Loop through each table cell
1159 foreach ( $cells as $cell ) {
1160 $previous = '';
1161 if ( $first_character !== '+' ) {
1162 $tr_after = array_pop( $tr_attributes );
1163 if ( !array_pop( $tr_history ) ) {
1164 $previous = "<tr{$tr_after}>\n";
1165 }
1166 array_push( $tr_history, true );
1167 array_push( $tr_attributes, '' );
1168 array_pop( $has_opened_tr );
1169 array_push( $has_opened_tr, true );
1170 }
1171
1172 $last_tag = array_pop( $last_tag_history );
1173
1174 if ( array_pop( $td_history ) ) {
1175 $previous = "</{$last_tag}>\n{$previous}";
1176 }
1177
1178 if ( $first_character === '|' ) {
1179 $last_tag = 'td';
1180 } elseif ( $first_character === '!' ) {
1181 $last_tag = 'th';
1182 } elseif ( $first_character === '+' ) {
1183 $last_tag = 'caption';
1184 } else {
1185 $last_tag = '';
1186 }
1187
1188 array_push( $last_tag_history, $last_tag );
1189
1190 # A cell could contain both parameters and data
1191 $cell_data = explode( '|', $cell, 2 );
1192
1193 # T2553: Note that a '|' inside an invalid link should not
1194 # be mistaken as delimiting cell parameters
1195 # Bug T153140: Neither should language converter markup.
1196 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1197 $cell = "{$previous}<{$last_tag}>{$cell}";
1198 } elseif ( count( $cell_data ) == 1 ) {
1199 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1200 } else {
1201 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1202 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1203 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1204 }
1205
1206 $outLine .= $cell;
1207 array_push( $td_history, true );
1208 }
1209 }
1210 $out .= $outLine . "\n";
1211 }
1212
1213 # Closing open td, tr && table
1214 while ( count( $td_history ) > 0 ) {
1215 if ( array_pop( $td_history ) ) {
1216 $out .= "</td>\n";
1217 }
1218 if ( array_pop( $tr_history ) ) {
1219 $out .= "</tr>\n";
1220 }
1221 if ( !array_pop( $has_opened_tr ) ) {
1222 $out .= "<tr><td></td></tr>\n";
1223 }
1224
1225 $out .= "</table>\n";
1226 }
1227
1228 # Remove trailing line-ending (b/c)
1229 if ( substr( $out, -1 ) === "\n" ) {
1230 $out = substr( $out, 0, -1 );
1231 }
1232
1233 # special case: don't return empty table
1234 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1235 $out = '';
1236 }
1237
1238 return $out;
1239 }
1240
1241 /**
1242 * Helper function for parse() that transforms wiki markup into half-parsed
1243 * HTML. Only called for $mOutputType == self::OT_HTML.
1244 *
1245 * @private
1246 *
1247 * @param string $text The text to parse
1248 * @param bool $isMain Whether this is being called from the main parse() function
1249 * @param PPFrame|bool $frame A pre-processor frame
1250 *
1251 * @return string
1252 */
1253 public function internalParse( $text, $isMain = true, $frame = false ) {
1254 $origText = $text;
1255
1256 // Avoid PHP 7.1 warning from passing $this by reference
1257 $parser = $this;
1258
1259 # Hook to suspend the parser in this state
1260 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1261 return $text;
1262 }
1263
1264 # if $frame is provided, then use $frame for replacing any variables
1265 if ( $frame ) {
1266 # use frame depth to infer how include/noinclude tags should be handled
1267 # depth=0 means this is the top-level document; otherwise it's an included document
1268 if ( !$frame->depth ) {
1269 $flag = 0;
1270 } else {
1271 $flag = self::PTD_FOR_INCLUSION;
1272 }
1273 $dom = $this->preprocessToDom( $text, $flag );
1274 $text = $frame->expand( $dom );
1275 } else {
1276 # if $frame is not provided, then use old-style replaceVariables
1277 $text = $this->replaceVariables( $text );
1278 }
1279
1280 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1281 $text = Sanitizer::removeHTMLtags(
1282 $text,
1283 [ $this, 'attributeStripCallback' ],
1284 false,
1285 array_keys( $this->mTransparentTagHooks ),
1286 [],
1287 [ $this, 'addTrackingCategory' ]
1288 );
1289 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1290
1291 # Tables need to come after variable replacement for things to work
1292 # properly; putting them before other transformations should keep
1293 # exciting things like link expansions from showing up in surprising
1294 # places.
1295 $text = $this->doTableStuff( $text );
1296
1297 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1298
1299 $text = $this->doDoubleUnderscore( $text );
1300
1301 $text = $this->doHeadings( $text );
1302 $text = $this->replaceInternalLinks( $text );
1303 $text = $this->doAllQuotes( $text );
1304 $text = $this->replaceExternalLinks( $text );
1305
1306 # replaceInternalLinks may sometimes leave behind
1307 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1308 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1309
1310 $text = $this->doMagicLinks( $text );
1311 $text = $this->formatHeadings( $text, $origText, $isMain );
1312
1313 return $text;
1314 }
1315
1316 /**
1317 * Helper function for parse() that transforms half-parsed HTML into fully
1318 * parsed HTML.
1319 *
1320 * @param string $text
1321 * @param bool $isMain
1322 * @param bool $linestart
1323 * @return string
1324 */
1325 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1326 $text = $this->mStripState->unstripGeneral( $text );
1327
1328 // Avoid PHP 7.1 warning from passing $this by reference
1329 $parser = $this;
1330
1331 if ( $isMain ) {
1332 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1333 }
1334
1335 # Clean up special characters, only run once, next-to-last before doBlockLevels
1336 $fixtags = [
1337 # French spaces, last one Guillemet-left
1338 # only if there is something before the space
1339 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1340 # french spaces, Guillemet-right
1341 '/(\\302\\253) /' => '\\1&#160;',
1342 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1343 ];
1344 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1345
1346 $text = $this->doBlockLevels( $text, $linestart );
1347
1348 $this->replaceLinkHolders( $text );
1349
1350 /**
1351 * The input doesn't get language converted if
1352 * a) It's disabled
1353 * b) Content isn't converted
1354 * c) It's a conversion table
1355 * d) it is an interface message (which is in the user language)
1356 */
1357 if ( !( $this->mOptions->getDisableContentConversion()
1358 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1359 ) {
1360 if ( !$this->mOptions->getInterfaceMessage() ) {
1361 # The position of the convert() call should not be changed. it
1362 # assumes that the links are all replaced and the only thing left
1363 # is the <nowiki> mark.
1364 $text = $this->getConverterLanguage()->convert( $text );
1365 }
1366 }
1367
1368 $text = $this->mStripState->unstripNoWiki( $text );
1369
1370 if ( $isMain ) {
1371 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1372 }
1373
1374 $text = $this->replaceTransparentTags( $text );
1375 $text = $this->mStripState->unstripGeneral( $text );
1376
1377 $text = Sanitizer::normalizeCharReferences( $text );
1378
1379 if ( MWTidy::isEnabled() ) {
1380 if ( $this->mOptions->getTidy() ) {
1381 $text = MWTidy::tidy( $text );
1382 }
1383 } else {
1384 # attempt to sanitize at least some nesting problems
1385 # (T4702 and quite a few others)
1386 $tidyregs = [
1387 # ''Something [http://www.cool.com cool''] -->
1388 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1389 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1390 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1391 # fix up an anchor inside another anchor, only
1392 # at least for a single single nested link (T5695)
1393 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1394 '\\1\\2</a>\\3</a>\\1\\4</a>',
1395 # fix div inside inline elements- doBlockLevels won't wrap a line which
1396 # contains a div, so fix it up here; replace
1397 # div with escaped text
1398 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1399 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1400 # remove empty italic or bold tag pairs, some
1401 # introduced by rules above
1402 '/<([bi])><\/\\1>/' => '',
1403 ];
1404
1405 $text = preg_replace(
1406 array_keys( $tidyregs ),
1407 array_values( $tidyregs ),
1408 $text );
1409 }
1410
1411 if ( $isMain ) {
1412 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1413 }
1414
1415 return $text;
1416 }
1417
1418 /**
1419 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1420 * magic external links.
1421 *
1422 * DML
1423 * @private
1424 *
1425 * @param string $text
1426 *
1427 * @return string
1428 */
1429 public function doMagicLinks( $text ) {
1430 $prots = wfUrlProtocolsWithoutProtRel();
1431 $urlChar = self::EXT_LINK_URL_CLASS;
1432 $addr = self::EXT_LINK_ADDR;
1433 $space = self::SPACE_NOT_NL; # non-newline space
1434 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1435 $spaces = "$space++"; # possessive match of 1 or more spaces
1436 $text = preg_replace_callback(
1437 '!(?: # Start cases
1438 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1439 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1440 (\b # m[3]: Free external links
1441 (?i:$prots)
1442 ($addr$urlChar*) # m[4]: Post-protocol path
1443 ) |
1444 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1445 ([0-9]+)\b |
1446 \bISBN $spaces ( # m[6]: ISBN, capture number
1447 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1448 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1449 [0-9Xx] # check digit
1450 )\b
1451 )!xu", [ $this, 'magicLinkCallback' ], $text );
1452 return $text;
1453 }
1454
1455 /**
1456 * @throws MWException
1457 * @param array $m
1458 * @return HTML|string
1459 */
1460 public function magicLinkCallback( $m ) {
1461 if ( isset( $m[1] ) && $m[1] !== '' ) {
1462 # Skip anchor
1463 return $m[0];
1464 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1465 # Skip HTML element
1466 return $m[0];
1467 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1468 # Free external link
1469 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1470 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1471 # RFC or PMID
1472 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1473 if ( !$this->mOptions->getMagicRFCLinks() ) {
1474 return $m[0];
1475 }
1476 $keyword = 'RFC';
1477 $urlmsg = 'rfcurl';
1478 $cssClass = 'mw-magiclink-rfc';
1479 $trackingCat = 'magiclink-tracking-rfc';
1480 $id = $m[5];
1481 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1482 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1483 return $m[0];
1484 }
1485 $keyword = 'PMID';
1486 $urlmsg = 'pubmedurl';
1487 $cssClass = 'mw-magiclink-pmid';
1488 $trackingCat = 'magiclink-tracking-pmid';
1489 $id = $m[5];
1490 } else {
1491 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1492 substr( $m[0], 0, 20 ) . '"' );
1493 }
1494 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1495 $this->addTrackingCategory( $trackingCat );
1496 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1497 } elseif ( isset( $m[6] ) && $m[6] !== ''
1498 && $this->mOptions->getMagicISBNLinks()
1499 ) {
1500 # ISBN
1501 $isbn = $m[6];
1502 $space = self::SPACE_NOT_NL; # non-newline space
1503 $isbn = preg_replace( "/$space/", ' ', $isbn );
1504 $num = strtr( $isbn, [
1505 '-' => '',
1506 ' ' => '',
1507 'x' => 'X',
1508 ] );
1509 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1510 return $this->getLinkRenderer()->makeKnownLink(
1511 SpecialPage::getTitleFor( 'Booksources', $num ),
1512 "ISBN $isbn",
1513 [
1514 'class' => 'internal mw-magiclink-isbn',
1515 'title' => false // suppress title attribute
1516 ]
1517 );
1518 } else {
1519 return $m[0];
1520 }
1521 }
1522
1523 /**
1524 * Make a free external link, given a user-supplied URL
1525 *
1526 * @param string $url
1527 * @param int $numPostProto
1528 * The number of characters after the protocol.
1529 * @return string HTML
1530 * @private
1531 */
1532 public function makeFreeExternalLink( $url, $numPostProto ) {
1533 $trail = '';
1534
1535 # The characters '<' and '>' (which were escaped by
1536 # removeHTMLtags()) should not be included in
1537 # URLs, per RFC 2396.
1538 # Make &nbsp; terminate a URL as well (bug T84937)
1539 $m2 = [];
1540 if ( preg_match(
1541 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1542 $url,
1543 $m2,
1544 PREG_OFFSET_CAPTURE
1545 ) ) {
1546 $trail = substr( $url, $m2[0][1] ) . $trail;
1547 $url = substr( $url, 0, $m2[0][1] );
1548 }
1549
1550 # Move trailing punctuation to $trail
1551 $sep = ',;\.:!?';
1552 # If there is no left bracket, then consider right brackets fair game too
1553 if ( strpos( $url, '(' ) === false ) {
1554 $sep .= ')';
1555 }
1556
1557 $urlRev = strrev( $url );
1558 $numSepChars = strspn( $urlRev, $sep );
1559 # Don't break a trailing HTML entity by moving the ; into $trail
1560 # This is in hot code, so use substr_compare to avoid having to
1561 # create a new string object for the comparison
1562 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1563 # more optimization: instead of running preg_match with a $
1564 # anchor, which can be slow, do the match on the reversed
1565 # string starting at the desired offset.
1566 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1567 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1568 $numSepChars--;
1569 }
1570 }
1571 if ( $numSepChars ) {
1572 $trail = substr( $url, -$numSepChars ) . $trail;
1573 $url = substr( $url, 0, -$numSepChars );
1574 }
1575
1576 # Verify that we still have a real URL after trail removal, and
1577 # not just lone protocol
1578 if ( strlen( $trail ) >= $numPostProto ) {
1579 return $url . $trail;
1580 }
1581
1582 $url = Sanitizer::cleanUrl( $url );
1583
1584 # Is this an external image?
1585 $text = $this->maybeMakeExternalImage( $url );
1586 if ( $text === false ) {
1587 # Not an image, make a link
1588 $text = Linker::makeExternalLink( $url,
1589 $this->getConverterLanguage()->markNoConversion( $url, true ),
1590 true, 'free',
1591 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1592 # Register it in the output object...
1593 $this->mOutput->addExternalLink( $url );
1594 }
1595 return $text . $trail;
1596 }
1597
1598 /**
1599 * Parse headers and return html
1600 *
1601 * @private
1602 *
1603 * @param string $text
1604 *
1605 * @return string
1606 */
1607 public function doHeadings( $text ) {
1608 for ( $i = 6; $i >= 1; --$i ) {
1609 $h = str_repeat( '=', $i );
1610 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1611 }
1612 return $text;
1613 }
1614
1615 /**
1616 * Replace single quotes with HTML markup
1617 * @private
1618 *
1619 * @param string $text
1620 *
1621 * @return string The altered text
1622 */
1623 public function doAllQuotes( $text ) {
1624 $outtext = '';
1625 $lines = StringUtils::explode( "\n", $text );
1626 foreach ( $lines as $line ) {
1627 $outtext .= $this->doQuotes( $line ) . "\n";
1628 }
1629 $outtext = substr( $outtext, 0, -1 );
1630 return $outtext;
1631 }
1632
1633 /**
1634 * Helper function for doAllQuotes()
1635 *
1636 * @param string $text
1637 *
1638 * @return string
1639 */
1640 public function doQuotes( $text ) {
1641 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1642 $countarr = count( $arr );
1643 if ( $countarr == 1 ) {
1644 return $text;
1645 }
1646
1647 // First, do some preliminary work. This may shift some apostrophes from
1648 // being mark-up to being text. It also counts the number of occurrences
1649 // of bold and italics mark-ups.
1650 $numbold = 0;
1651 $numitalics = 0;
1652 for ( $i = 1; $i < $countarr; $i += 2 ) {
1653 $thislen = strlen( $arr[$i] );
1654 // If there are ever four apostrophes, assume the first is supposed to
1655 // be text, and the remaining three constitute mark-up for bold text.
1656 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1657 if ( $thislen == 4 ) {
1658 $arr[$i - 1] .= "'";
1659 $arr[$i] = "'''";
1660 $thislen = 3;
1661 } elseif ( $thislen > 5 ) {
1662 // If there are more than 5 apostrophes in a row, assume they're all
1663 // text except for the last 5.
1664 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1665 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1666 $arr[$i] = "'''''";
1667 $thislen = 5;
1668 }
1669 // Count the number of occurrences of bold and italics mark-ups.
1670 if ( $thislen == 2 ) {
1671 $numitalics++;
1672 } elseif ( $thislen == 3 ) {
1673 $numbold++;
1674 } elseif ( $thislen == 5 ) {
1675 $numitalics++;
1676 $numbold++;
1677 }
1678 }
1679
1680 // If there is an odd number of both bold and italics, it is likely
1681 // that one of the bold ones was meant to be an apostrophe followed
1682 // by italics. Which one we cannot know for certain, but it is more
1683 // likely to be one that has a single-letter word before it.
1684 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1685 $firstsingleletterword = -1;
1686 $firstmultiletterword = -1;
1687 $firstspace = -1;
1688 for ( $i = 1; $i < $countarr; $i += 2 ) {
1689 if ( strlen( $arr[$i] ) == 3 ) {
1690 $x1 = substr( $arr[$i - 1], -1 );
1691 $x2 = substr( $arr[$i - 1], -2, 1 );
1692 if ( $x1 === ' ' ) {
1693 if ( $firstspace == -1 ) {
1694 $firstspace = $i;
1695 }
1696 } elseif ( $x2 === ' ' ) {
1697 $firstsingleletterword = $i;
1698 // if $firstsingleletterword is set, we don't
1699 // look at the other options, so we can bail early.
1700 break;
1701 } else {
1702 if ( $firstmultiletterword == -1 ) {
1703 $firstmultiletterword = $i;
1704 }
1705 }
1706 }
1707 }
1708
1709 // If there is a single-letter word, use it!
1710 if ( $firstsingleletterword > -1 ) {
1711 $arr[$firstsingleletterword] = "''";
1712 $arr[$firstsingleletterword - 1] .= "'";
1713 } elseif ( $firstmultiletterword > -1 ) {
1714 // If not, but there's a multi-letter word, use that one.
1715 $arr[$firstmultiletterword] = "''";
1716 $arr[$firstmultiletterword - 1] .= "'";
1717 } elseif ( $firstspace > -1 ) {
1718 // ... otherwise use the first one that has neither.
1719 // (notice that it is possible for all three to be -1 if, for example,
1720 // there is only one pentuple-apostrophe in the line)
1721 $arr[$firstspace] = "''";
1722 $arr[$firstspace - 1] .= "'";
1723 }
1724 }
1725
1726 // Now let's actually convert our apostrophic mush to HTML!
1727 $output = '';
1728 $buffer = '';
1729 $state = '';
1730 $i = 0;
1731 foreach ( $arr as $r ) {
1732 if ( ( $i % 2 ) == 0 ) {
1733 if ( $state === 'both' ) {
1734 $buffer .= $r;
1735 } else {
1736 $output .= $r;
1737 }
1738 } else {
1739 $thislen = strlen( $r );
1740 if ( $thislen == 2 ) {
1741 if ( $state === 'i' ) {
1742 $output .= '</i>';
1743 $state = '';
1744 } elseif ( $state === 'bi' ) {
1745 $output .= '</i>';
1746 $state = 'b';
1747 } elseif ( $state === 'ib' ) {
1748 $output .= '</b></i><b>';
1749 $state = 'b';
1750 } elseif ( $state === 'both' ) {
1751 $output .= '<b><i>' . $buffer . '</i>';
1752 $state = 'b';
1753 } else { // $state can be 'b' or ''
1754 $output .= '<i>';
1755 $state .= 'i';
1756 }
1757 } elseif ( $thislen == 3 ) {
1758 if ( $state === 'b' ) {
1759 $output .= '</b>';
1760 $state = '';
1761 } elseif ( $state === 'bi' ) {
1762 $output .= '</i></b><i>';
1763 $state = 'i';
1764 } elseif ( $state === 'ib' ) {
1765 $output .= '</b>';
1766 $state = 'i';
1767 } elseif ( $state === 'both' ) {
1768 $output .= '<i><b>' . $buffer . '</b>';
1769 $state = 'i';
1770 } else { // $state can be 'i' or ''
1771 $output .= '<b>';
1772 $state .= 'b';
1773 }
1774 } elseif ( $thislen == 5 ) {
1775 if ( $state === 'b' ) {
1776 $output .= '</b><i>';
1777 $state = 'i';
1778 } elseif ( $state === 'i' ) {
1779 $output .= '</i><b>';
1780 $state = 'b';
1781 } elseif ( $state === 'bi' ) {
1782 $output .= '</i></b>';
1783 $state = '';
1784 } elseif ( $state === 'ib' ) {
1785 $output .= '</b></i>';
1786 $state = '';
1787 } elseif ( $state === 'both' ) {
1788 $output .= '<i><b>' . $buffer . '</b></i>';
1789 $state = '';
1790 } else { // ($state == '')
1791 $buffer = '';
1792 $state = 'both';
1793 }
1794 }
1795 }
1796 $i++;
1797 }
1798 // Now close all remaining tags. Notice that the order is important.
1799 if ( $state === 'b' || $state === 'ib' ) {
1800 $output .= '</b>';
1801 }
1802 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1803 $output .= '</i>';
1804 }
1805 if ( $state === 'bi' ) {
1806 $output .= '</b>';
1807 }
1808 // There might be lonely ''''', so make sure we have a buffer
1809 if ( $state === 'both' && $buffer ) {
1810 $output .= '<b><i>' . $buffer . '</i></b>';
1811 }
1812 return $output;
1813 }
1814
1815 /**
1816 * Replace external links (REL)
1817 *
1818 * Note: this is all very hackish and the order of execution matters a lot.
1819 * Make sure to run tests/parser/parserTests.php if you change this code.
1820 *
1821 * @private
1822 *
1823 * @param string $text
1824 *
1825 * @throws MWException
1826 * @return string
1827 */
1828 public function replaceExternalLinks( $text ) {
1829 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1830 if ( $bits === false ) {
1831 throw new MWException( "PCRE needs to be compiled with "
1832 . "--enable-unicode-properties in order for MediaWiki to function" );
1833 }
1834 $s = array_shift( $bits );
1835
1836 $i = 0;
1837 while ( $i < count( $bits ) ) {
1838 $url = $bits[$i++];
1839 $i++; // protocol
1840 $text = $bits[$i++];
1841 $trail = $bits[$i++];
1842
1843 # The characters '<' and '>' (which were escaped by
1844 # removeHTMLtags()) should not be included in
1845 # URLs, per RFC 2396.
1846 $m2 = [];
1847 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1848 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1849 $url = substr( $url, 0, $m2[0][1] );
1850 }
1851
1852 # If the link text is an image URL, replace it with an <img> tag
1853 # This happened by accident in the original parser, but some people used it extensively
1854 $img = $this->maybeMakeExternalImage( $text );
1855 if ( $img !== false ) {
1856 $text = $img;
1857 }
1858
1859 $dtrail = '';
1860
1861 # Set linktype for CSS - if URL==text, link is essentially free
1862 $linktype = ( $text === $url ) ? 'free' : 'text';
1863
1864 # No link text, e.g. [http://domain.tld/some.link]
1865 if ( $text == '' ) {
1866 # Autonumber
1867 $langObj = $this->getTargetLanguage();
1868 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1869 $linktype = 'autonumber';
1870 } else {
1871 # Have link text, e.g. [http://domain.tld/some.link text]s
1872 # Check for trail
1873 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1874 }
1875
1876 $text = $this->getConverterLanguage()->markNoConversion( $text );
1877
1878 $url = Sanitizer::cleanUrl( $url );
1879
1880 # Use the encoded URL
1881 # This means that users can paste URLs directly into the text
1882 # Funny characters like ö aren't valid in URLs anyway
1883 # This was changed in August 2004
1884 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1885 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1886
1887 # Register link in the output object.
1888 $this->mOutput->addExternalLink( $url );
1889 }
1890
1891 return $s;
1892 }
1893
1894 /**
1895 * Get the rel attribute for a particular external link.
1896 *
1897 * @since 1.21
1898 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1899 * nofollow if appropriate
1900 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1901 * @return string|null Rel attribute for $url
1902 */
1903 public static function getExternalLinkRel( $url = false, $title = null ) {
1904 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1905 $ns = $title ? $title->getNamespace() : false;
1906 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1907 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1908 ) {
1909 return 'nofollow';
1910 }
1911 return null;
1912 }
1913
1914 /**
1915 * Get an associative array of additional HTML attributes appropriate for a
1916 * particular external link. This currently may include rel => nofollow
1917 * (depending on configuration, namespace, and the URL's domain) and/or a
1918 * target attribute (depending on configuration).
1919 *
1920 * @param string $url URL to extract the domain from for rel =>
1921 * nofollow if appropriate
1922 * @return array Associative array of HTML attributes
1923 */
1924 public function getExternalLinkAttribs( $url ) {
1925 $attribs = [];
1926 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1927
1928 $target = $this->mOptions->getExternalLinkTarget();
1929 if ( $target ) {
1930 $attribs['target'] = $target;
1931 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1932 // T133507. New windows can navigate parent cross-origin.
1933 // Including noreferrer due to lacking browser
1934 // support of noopener. Eventually noreferrer should be removed.
1935 if ( $rel !== '' ) {
1936 $rel .= ' ';
1937 }
1938 $rel .= 'noreferrer noopener';
1939 }
1940 }
1941 $attribs['rel'] = $rel;
1942 return $attribs;
1943 }
1944
1945 /**
1946 * Replace unusual escape codes in a URL with their equivalent characters
1947 *
1948 * This generally follows the syntax defined in RFC 3986, with special
1949 * consideration for HTTP query strings.
1950 *
1951 * @param string $url
1952 * @return string
1953 */
1954 public static function normalizeLinkUrl( $url ) {
1955 # First, make sure unsafe characters are encoded
1956 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1957 function ( $m ) {
1958 return rawurlencode( $m[0] );
1959 },
1960 $url
1961 );
1962
1963 $ret = '';
1964 $end = strlen( $url );
1965
1966 # Fragment part - 'fragment'
1967 $start = strpos( $url, '#' );
1968 if ( $start !== false && $start < $end ) {
1969 $ret = self::normalizeUrlComponent(
1970 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1971 $end = $start;
1972 }
1973
1974 # Query part - 'query' minus &=+;
1975 $start = strpos( $url, '?' );
1976 if ( $start !== false && $start < $end ) {
1977 $ret = self::normalizeUrlComponent(
1978 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1979 $end = $start;
1980 }
1981
1982 # Scheme and path part - 'pchar'
1983 # (we assume no userinfo or encoded colons in the host)
1984 $ret = self::normalizeUrlComponent(
1985 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1986
1987 return $ret;
1988 }
1989
1990 private static function normalizeUrlComponent( $component, $unsafe ) {
1991 $callback = function ( $matches ) use ( $unsafe ) {
1992 $char = urldecode( $matches[0] );
1993 $ord = ord( $char );
1994 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1995 # Unescape it
1996 return $char;
1997 } else {
1998 # Leave it escaped, but use uppercase for a-f
1999 return strtoupper( $matches[0] );
2000 }
2001 };
2002 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2003 }
2004
2005 /**
2006 * make an image if it's allowed, either through the global
2007 * option, through the exception, or through the on-wiki whitelist
2008 *
2009 * @param string $url
2010 *
2011 * @return string
2012 */
2013 private function maybeMakeExternalImage( $url ) {
2014 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2015 $imagesexception = !empty( $imagesfrom );
2016 $text = false;
2017 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2018 if ( $imagesexception && is_array( $imagesfrom ) ) {
2019 $imagematch = false;
2020 foreach ( $imagesfrom as $match ) {
2021 if ( strpos( $url, $match ) === 0 ) {
2022 $imagematch = true;
2023 break;
2024 }
2025 }
2026 } elseif ( $imagesexception ) {
2027 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2028 } else {
2029 $imagematch = false;
2030 }
2031
2032 if ( $this->mOptions->getAllowExternalImages()
2033 || ( $imagesexception && $imagematch )
2034 ) {
2035 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2036 # Image found
2037 $text = Linker::makeExternalImage( $url );
2038 }
2039 }
2040 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2041 && preg_match( self::EXT_IMAGE_REGEX, $url )
2042 ) {
2043 $whitelist = explode(
2044 "\n",
2045 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2046 );
2047
2048 foreach ( $whitelist as $entry ) {
2049 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2050 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2051 continue;
2052 }
2053 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2054 # Image matches a whitelist entry
2055 $text = Linker::makeExternalImage( $url );
2056 break;
2057 }
2058 }
2059 }
2060 return $text;
2061 }
2062
2063 /**
2064 * Process [[ ]] wikilinks
2065 *
2066 * @param string $s
2067 *
2068 * @return string Processed text
2069 *
2070 * @private
2071 */
2072 public function replaceInternalLinks( $s ) {
2073 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2074 return $s;
2075 }
2076
2077 /**
2078 * Process [[ ]] wikilinks (RIL)
2079 * @param string &$s
2080 * @throws MWException
2081 * @return LinkHolderArray
2082 *
2083 * @private
2084 */
2085 public function replaceInternalLinks2( &$s ) {
2086 global $wgExtraInterlanguageLinkPrefixes;
2087
2088 static $tc = false, $e1, $e1_img;
2089 # the % is needed to support urlencoded titles as well
2090 if ( !$tc ) {
2091 $tc = Title::legalChars() . '#%';
2092 # Match a link having the form [[namespace:link|alternate]]trail
2093 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2094 # Match cases where there is no "]]", which might still be images
2095 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2096 }
2097
2098 $holders = new LinkHolderArray( $this );
2099
2100 # split the entire text string on occurrences of [[
2101 $a = StringUtils::explode( '[[', ' ' . $s );
2102 # get the first element (all text up to first [[), and remove the space we added
2103 $s = $a->current();
2104 $a->next();
2105 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2106 $s = substr( $s, 1 );
2107
2108 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2109 $e2 = null;
2110 if ( $useLinkPrefixExtension ) {
2111 # Match the end of a line for a word that's not followed by whitespace,
2112 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2113 global $wgContLang;
2114 $charset = $wgContLang->linkPrefixCharset();
2115 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2116 }
2117
2118 if ( is_null( $this->mTitle ) ) {
2119 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2120 }
2121 $nottalk = !$this->mTitle->isTalkPage();
2122
2123 if ( $useLinkPrefixExtension ) {
2124 $m = [];
2125 if ( preg_match( $e2, $s, $m ) ) {
2126 $first_prefix = $m[2];
2127 } else {
2128 $first_prefix = false;
2129 }
2130 } else {
2131 $prefix = '';
2132 }
2133
2134 $useSubpages = $this->areSubpagesAllowed();
2135
2136 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2137 # Loop for each link
2138 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2139 // @codingStandardsIgnoreEnd
2140
2141 # Check for excessive memory usage
2142 if ( $holders->isBig() ) {
2143 # Too big
2144 # Do the existence check, replace the link holders and clear the array
2145 $holders->replace( $s );
2146 $holders->clear();
2147 }
2148
2149 if ( $useLinkPrefixExtension ) {
2150 if ( preg_match( $e2, $s, $m ) ) {
2151 $prefix = $m[2];
2152 $s = $m[1];
2153 } else {
2154 $prefix = '';
2155 }
2156 # first link
2157 if ( $first_prefix ) {
2158 $prefix = $first_prefix;
2159 $first_prefix = false;
2160 }
2161 }
2162
2163 $might_be_img = false;
2164
2165 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2166 $text = $m[2];
2167 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2168 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2169 # the real problem is with the $e1 regex
2170 # See T1500.
2171 # Still some problems for cases where the ] is meant to be outside punctuation,
2172 # and no image is in sight. See T4095.
2173 if ( $text !== ''
2174 && substr( $m[3], 0, 1 ) === ']'
2175 && strpos( $text, '[' ) !== false
2176 ) {
2177 $text .= ']'; # so that replaceExternalLinks($text) works later
2178 $m[3] = substr( $m[3], 1 );
2179 }
2180 # fix up urlencoded title texts
2181 if ( strpos( $m[1], '%' ) !== false ) {
2182 # Should anchors '#' also be rejected?
2183 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2184 }
2185 $trail = $m[3];
2186 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2187 # Invalid, but might be an image with a link in its caption
2188 $might_be_img = true;
2189 $text = $m[2];
2190 if ( strpos( $m[1], '%' ) !== false ) {
2191 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2192 }
2193 $trail = "";
2194 } else { # Invalid form; output directly
2195 $s .= $prefix . '[[' . $line;
2196 continue;
2197 }
2198
2199 $origLink = ltrim( $m[1], ' ' );
2200
2201 # Don't allow internal links to pages containing
2202 # PROTO: where PROTO is a valid URL protocol; these
2203 # should be external links.
2204 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2205 $s .= $prefix . '[[' . $line;
2206 continue;
2207 }
2208
2209 # Make subpage if necessary
2210 if ( $useSubpages ) {
2211 $link = $this->maybeDoSubpageLink( $origLink, $text );
2212 } else {
2213 $link = $origLink;
2214 }
2215
2216 $unstrip = $this->mStripState->unstripNoWiki( $link );
2217 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2218 if ( $nt === null ) {
2219 $s .= $prefix . '[[' . $line;
2220 continue;
2221 }
2222
2223 $ns = $nt->getNamespace();
2224 $iw = $nt->getInterwiki();
2225
2226 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2227
2228 if ( $might_be_img ) { # if this is actually an invalid link
2229 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2230 $found = false;
2231 while ( true ) {
2232 # look at the next 'line' to see if we can close it there
2233 $a->next();
2234 $next_line = $a->current();
2235 if ( $next_line === false || $next_line === null ) {
2236 break;
2237 }
2238 $m = explode( ']]', $next_line, 3 );
2239 if ( count( $m ) == 3 ) {
2240 # the first ]] closes the inner link, the second the image
2241 $found = true;
2242 $text .= "[[{$m[0]}]]{$m[1]}";
2243 $trail = $m[2];
2244 break;
2245 } elseif ( count( $m ) == 2 ) {
2246 # if there's exactly one ]] that's fine, we'll keep looking
2247 $text .= "[[{$m[0]}]]{$m[1]}";
2248 } else {
2249 # if $next_line is invalid too, we need look no further
2250 $text .= '[[' . $next_line;
2251 break;
2252 }
2253 }
2254 if ( !$found ) {
2255 # we couldn't find the end of this imageLink, so output it raw
2256 # but don't ignore what might be perfectly normal links in the text we've examined
2257 $holders->merge( $this->replaceInternalLinks2( $text ) );
2258 $s .= "{$prefix}[[$link|$text";
2259 # note: no $trail, because without an end, there *is* no trail
2260 continue;
2261 }
2262 } else { # it's not an image, so output it raw
2263 $s .= "{$prefix}[[$link|$text";
2264 # note: no $trail, because without an end, there *is* no trail
2265 continue;
2266 }
2267 }
2268
2269 $wasblank = ( $text == '' );
2270 if ( $wasblank ) {
2271 $text = $link;
2272 if ( !$noforce ) {
2273 # Strip off leading ':'
2274 $text = substr( $text, 1 );
2275 }
2276 } else {
2277 # T6598 madness. Handle the quotes only if they come from the alternate part
2278 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2279 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2280 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2281 $text = $this->doQuotes( $text );
2282 }
2283
2284 # Link not escaped by : , create the various objects
2285 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2286 # Interwikis
2287 if (
2288 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2289 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2290 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2291 )
2292 ) {
2293 # T26502: filter duplicates
2294 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2295 $this->mLangLinkLanguages[$iw] = true;
2296 $this->mOutput->addLanguageLink( $nt->getFullText() );
2297 }
2298
2299 $s = rtrim( $s . $prefix );
2300 $s .= trim( $trail, "\n" ) == '' ? '' : $prefix . $trail;
2301 continue;
2302 }
2303
2304 if ( $ns == NS_FILE ) {
2305 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2306 if ( $wasblank ) {
2307 # if no parameters were passed, $text
2308 # becomes something like "File:Foo.png",
2309 # which we don't want to pass on to the
2310 # image generator
2311 $text = '';
2312 } else {
2313 # recursively parse links inside the image caption
2314 # actually, this will parse them in any other parameters, too,
2315 # but it might be hard to fix that, and it doesn't matter ATM
2316 $text = $this->replaceExternalLinks( $text );
2317 $holders->merge( $this->replaceInternalLinks2( $text ) );
2318 }
2319 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2320 $s .= $prefix . $this->armorLinks(
2321 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2322 continue;
2323 }
2324 } elseif ( $ns == NS_CATEGORY ) {
2325 $s = rtrim( $s . "\n" ); # T2087
2326
2327 if ( $wasblank ) {
2328 $sortkey = $this->getDefaultSort();
2329 } else {
2330 $sortkey = $text;
2331 }
2332 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2333 $sortkey = str_replace( "\n", '', $sortkey );
2334 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2335 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2336
2337 /**
2338 * Strip the whitespace Category links produce, see T2087
2339 */
2340 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2341
2342 continue;
2343 }
2344 }
2345
2346 # Self-link checking. For some languages, variants of the title are checked in
2347 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2348 # for linking to a different variant.
2349 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2350 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2351 continue;
2352 }
2353
2354 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2355 # @todo FIXME: Should do batch file existence checks, see comment below
2356 if ( $ns == NS_MEDIA ) {
2357 # Give extensions a chance to select the file revision for us
2358 $options = [];
2359 $descQuery = false;
2360 Hooks::run( 'BeforeParserFetchFileAndTitle',
2361 [ $this, $nt, &$options, &$descQuery ] );
2362 # Fetch and register the file (file title may be different via hooks)
2363 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2364 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2365 $s .= $prefix . $this->armorLinks(
2366 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2367 continue;
2368 }
2369
2370 # Some titles, such as valid special pages or files in foreign repos, should
2371 # be shown as bluelinks even though they're not included in the page table
2372 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2373 # batch file existence checks for NS_FILE and NS_MEDIA
2374 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2375 $this->mOutput->addLink( $nt );
2376 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2377 } else {
2378 # Links will be added to the output link list after checking
2379 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2380 }
2381 }
2382 return $holders;
2383 }
2384
2385 /**
2386 * Render a forced-blue link inline; protect against double expansion of
2387 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2388 * Since this little disaster has to split off the trail text to avoid
2389 * breaking URLs in the following text without breaking trails on the
2390 * wiki links, it's been made into a horrible function.
2391 *
2392 * @param Title $nt
2393 * @param string $text
2394 * @param string $trail
2395 * @param string $prefix
2396 * @return string HTML-wikitext mix oh yuck
2397 */
2398 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2399 list( $inside, $trail ) = Linker::splitTrail( $trail );
2400
2401 if ( $text == '' ) {
2402 $text = htmlspecialchars( $nt->getPrefixedText() );
2403 }
2404
2405 $link = $this->getLinkRenderer()->makeKnownLink(
2406 $nt, new HtmlArmor( "$prefix$text$inside" )
2407 );
2408
2409 return $this->armorLinks( $link ) . $trail;
2410 }
2411
2412 /**
2413 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2414 * going to go through further parsing steps before inline URL expansion.
2415 *
2416 * Not needed quite as much as it used to be since free links are a bit
2417 * more sensible these days. But bracketed links are still an issue.
2418 *
2419 * @param string $text More-or-less HTML
2420 * @return string Less-or-more HTML with NOPARSE bits
2421 */
2422 public function armorLinks( $text ) {
2423 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2424 self::MARKER_PREFIX . "NOPARSE$1", $text );
2425 }
2426
2427 /**
2428 * Return true if subpage links should be expanded on this page.
2429 * @return bool
2430 */
2431 public function areSubpagesAllowed() {
2432 # Some namespaces don't allow subpages
2433 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2434 }
2435
2436 /**
2437 * Handle link to subpage if necessary
2438 *
2439 * @param string $target The source of the link
2440 * @param string &$text The link text, modified as necessary
2441 * @return string The full name of the link
2442 * @private
2443 */
2444 public function maybeDoSubpageLink( $target, &$text ) {
2445 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2446 }
2447
2448 /**
2449 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2450 *
2451 * @param string $text
2452 * @param bool $linestart Whether or not this is at the start of a line.
2453 * @private
2454 * @return string The lists rendered as HTML
2455 */
2456 public function doBlockLevels( $text, $linestart ) {
2457 return BlockLevelPass::doBlockLevels( $text, $linestart );
2458 }
2459
2460 /**
2461 * Return value of a magic variable (like PAGENAME)
2462 *
2463 * @private
2464 *
2465 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2466 * @param bool|PPFrame $frame
2467 *
2468 * @throws MWException
2469 * @return string
2470 */
2471 public function getVariableValue( $index, $frame = false ) {
2472 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2473 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2474
2475 if ( is_null( $this->mTitle ) ) {
2476 // If no title set, bad things are going to happen
2477 // later. Title should always be set since this
2478 // should only be called in the middle of a parse
2479 // operation (but the unit-tests do funky stuff)
2480 throw new MWException( __METHOD__ . ' Should only be '
2481 . ' called while parsing (no title set)' );
2482 }
2483
2484 // Avoid PHP 7.1 warning from passing $this by reference
2485 $parser = $this;
2486
2487 /**
2488 * Some of these require message or data lookups and can be
2489 * expensive to check many times.
2490 */
2491 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2492 if ( isset( $this->mVarCache[$index] ) ) {
2493 return $this->mVarCache[$index];
2494 }
2495 }
2496
2497 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2498 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2499
2500 $pageLang = $this->getFunctionLang();
2501
2502 switch ( $index ) {
2503 case '!':
2504 $value = '|';
2505 break;
2506 case 'currentmonth':
2507 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2508 break;
2509 case 'currentmonth1':
2510 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2511 break;
2512 case 'currentmonthname':
2513 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2514 break;
2515 case 'currentmonthnamegen':
2516 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2517 break;
2518 case 'currentmonthabbrev':
2519 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2520 break;
2521 case 'currentday':
2522 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2523 break;
2524 case 'currentday2':
2525 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2526 break;
2527 case 'localmonth':
2528 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2529 break;
2530 case 'localmonth1':
2531 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2532 break;
2533 case 'localmonthname':
2534 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2535 break;
2536 case 'localmonthnamegen':
2537 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2538 break;
2539 case 'localmonthabbrev':
2540 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2541 break;
2542 case 'localday':
2543 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2544 break;
2545 case 'localday2':
2546 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2547 break;
2548 case 'pagename':
2549 $value = wfEscapeWikiText( $this->mTitle->getText() );
2550 break;
2551 case 'pagenamee':
2552 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2553 break;
2554 case 'fullpagename':
2555 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2556 break;
2557 case 'fullpagenamee':
2558 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2559 break;
2560 case 'subpagename':
2561 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2562 break;
2563 case 'subpagenamee':
2564 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2565 break;
2566 case 'rootpagename':
2567 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2568 break;
2569 case 'rootpagenamee':
2570 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2571 ' ',
2572 '_',
2573 $this->mTitle->getRootText()
2574 ) ) );
2575 break;
2576 case 'basepagename':
2577 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2578 break;
2579 case 'basepagenamee':
2580 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2581 ' ',
2582 '_',
2583 $this->mTitle->getBaseText()
2584 ) ) );
2585 break;
2586 case 'talkpagename':
2587 if ( $this->mTitle->canHaveTalkPage() ) {
2588 $talkPage = $this->mTitle->getTalkPage();
2589 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2590 } else {
2591 $value = '';
2592 }
2593 break;
2594 case 'talkpagenamee':
2595 if ( $this->mTitle->canHaveTalkPage() ) {
2596 $talkPage = $this->mTitle->getTalkPage();
2597 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2598 } else {
2599 $value = '';
2600 }
2601 break;
2602 case 'subjectpagename':
2603 $subjPage = $this->mTitle->getSubjectPage();
2604 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2605 break;
2606 case 'subjectpagenamee':
2607 $subjPage = $this->mTitle->getSubjectPage();
2608 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2609 break;
2610 case 'pageid': // requested in T25427
2611 $pageid = $this->getTitle()->getArticleID();
2612 if ( $pageid == 0 ) {
2613 # 0 means the page doesn't exist in the database,
2614 # which means the user is previewing a new page.
2615 # The vary-revision flag must be set, because the magic word
2616 # will have a different value once the page is saved.
2617 $this->mOutput->setFlag( 'vary-revision' );
2618 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2619 }
2620 $value = $pageid ? $pageid : null;
2621 break;
2622 case 'revisionid':
2623 # Let the edit saving system know we should parse the page
2624 # *after* a revision ID has been assigned.
2625 $this->mOutput->setFlag( 'vary-revision-id' );
2626 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2627 $value = $this->mRevisionId;
2628 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2629 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2630 $this->mOutput->setSpeculativeRevIdUsed( $value );
2631 }
2632 break;
2633 case 'revisionday':
2634 # Let the edit saving system know we should parse the page
2635 # *after* a revision ID has been assigned. This is for null edits.
2636 $this->mOutput->setFlag( 'vary-revision' );
2637 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2638 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2639 break;
2640 case 'revisionday2':
2641 # Let the edit saving system know we should parse the page
2642 # *after* a revision ID has been assigned. This is for null edits.
2643 $this->mOutput->setFlag( 'vary-revision' );
2644 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2645 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2646 break;
2647 case 'revisionmonth':
2648 # Let the edit saving system know we should parse the page
2649 # *after* a revision ID has been assigned. This is for null edits.
2650 $this->mOutput->setFlag( 'vary-revision' );
2651 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2652 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2653 break;
2654 case 'revisionmonth1':
2655 # Let the edit saving system know we should parse the page
2656 # *after* a revision ID has been assigned. This is for null edits.
2657 $this->mOutput->setFlag( 'vary-revision' );
2658 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2659 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2660 break;
2661 case 'revisionyear':
2662 # Let the edit saving system know we should parse the page
2663 # *after* a revision ID has been assigned. This is for null edits.
2664 $this->mOutput->setFlag( 'vary-revision' );
2665 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2666 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2667 break;
2668 case 'revisiontimestamp':
2669 # Let the edit saving system know we should parse the page
2670 # *after* a revision ID has been assigned. This is for null edits.
2671 $this->mOutput->setFlag( 'vary-revision' );
2672 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2673 $value = $this->getRevisionTimestamp();
2674 break;
2675 case 'revisionuser':
2676 # Let the edit saving system know we should parse the page
2677 # *after* a revision ID has been assigned for null edits.
2678 $this->mOutput->setFlag( 'vary-user' );
2679 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2680 $value = $this->getRevisionUser();
2681 break;
2682 case 'revisionsize':
2683 $value = $this->getRevisionSize();
2684 break;
2685 case 'namespace':
2686 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2687 break;
2688 case 'namespacee':
2689 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2690 break;
2691 case 'namespacenumber':
2692 $value = $this->mTitle->getNamespace();
2693 break;
2694 case 'talkspace':
2695 $value = $this->mTitle->canHaveTalkPage()
2696 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2697 : '';
2698 break;
2699 case 'talkspacee':
2700 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2701 break;
2702 case 'subjectspace':
2703 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2704 break;
2705 case 'subjectspacee':
2706 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2707 break;
2708 case 'currentdayname':
2709 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2710 break;
2711 case 'currentyear':
2712 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2713 break;
2714 case 'currenttime':
2715 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2716 break;
2717 case 'currenthour':
2718 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2719 break;
2720 case 'currentweek':
2721 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2722 # int to remove the padding
2723 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2724 break;
2725 case 'currentdow':
2726 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2727 break;
2728 case 'localdayname':
2729 $value = $pageLang->getWeekdayName(
2730 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2731 );
2732 break;
2733 case 'localyear':
2734 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2735 break;
2736 case 'localtime':
2737 $value = $pageLang->time(
2738 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2739 false,
2740 false
2741 );
2742 break;
2743 case 'localhour':
2744 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2745 break;
2746 case 'localweek':
2747 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2748 # int to remove the padding
2749 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2750 break;
2751 case 'localdow':
2752 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2753 break;
2754 case 'numberofarticles':
2755 $value = $pageLang->formatNum( SiteStats::articles() );
2756 break;
2757 case 'numberoffiles':
2758 $value = $pageLang->formatNum( SiteStats::images() );
2759 break;
2760 case 'numberofusers':
2761 $value = $pageLang->formatNum( SiteStats::users() );
2762 break;
2763 case 'numberofactiveusers':
2764 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2765 break;
2766 case 'numberofpages':
2767 $value = $pageLang->formatNum( SiteStats::pages() );
2768 break;
2769 case 'numberofadmins':
2770 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2771 break;
2772 case 'numberofedits':
2773 $value = $pageLang->formatNum( SiteStats::edits() );
2774 break;
2775 case 'currenttimestamp':
2776 $value = wfTimestamp( TS_MW, $ts );
2777 break;
2778 case 'localtimestamp':
2779 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2780 break;
2781 case 'currentversion':
2782 $value = SpecialVersion::getVersion();
2783 break;
2784 case 'articlepath':
2785 return $wgArticlePath;
2786 case 'sitename':
2787 return $wgSitename;
2788 case 'server':
2789 return $wgServer;
2790 case 'servername':
2791 return $wgServerName;
2792 case 'scriptpath':
2793 return $wgScriptPath;
2794 case 'stylepath':
2795 return $wgStylePath;
2796 case 'directionmark':
2797 return $pageLang->getDirMark();
2798 case 'contentlanguage':
2799 global $wgLanguageCode;
2800 return $wgLanguageCode;
2801 case 'pagelanguage':
2802 $value = $pageLang->getCode();
2803 break;
2804 case 'cascadingsources':
2805 $value = CoreParserFunctions::cascadingsources( $this );
2806 break;
2807 default:
2808 $ret = null;
2809 Hooks::run(
2810 'ParserGetVariableValueSwitch',
2811 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2812 );
2813
2814 return $ret;
2815 }
2816
2817 if ( $index ) {
2818 $this->mVarCache[$index] = $value;
2819 }
2820
2821 return $value;
2822 }
2823
2824 /**
2825 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2826 *
2827 * @private
2828 */
2829 public function initialiseVariables() {
2830 $variableIDs = MagicWord::getVariableIDs();
2831 $substIDs = MagicWord::getSubstIDs();
2832
2833 $this->mVariables = new MagicWordArray( $variableIDs );
2834 $this->mSubstWords = new MagicWordArray( $substIDs );
2835 }
2836
2837 /**
2838 * Preprocess some wikitext and return the document tree.
2839 * This is the ghost of replace_variables().
2840 *
2841 * @param string $text The text to parse
2842 * @param int $flags Bitwise combination of:
2843 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2844 * included. Default is to assume a direct page view.
2845 *
2846 * The generated DOM tree must depend only on the input text and the flags.
2847 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2848 *
2849 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2850 * change in the DOM tree for a given text, must be passed through the section identifier
2851 * in the section edit link and thus back to extractSections().
2852 *
2853 * The output of this function is currently only cached in process memory, but a persistent
2854 * cache may be implemented at a later date which takes further advantage of these strict
2855 * dependency requirements.
2856 *
2857 * @return PPNode
2858 */
2859 public function preprocessToDom( $text, $flags = 0 ) {
2860 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2861 return $dom;
2862 }
2863
2864 /**
2865 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2866 *
2867 * @param string $s
2868 *
2869 * @return array
2870 */
2871 public static function splitWhitespace( $s ) {
2872 $ltrimmed = ltrim( $s );
2873 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2874 $trimmed = rtrim( $ltrimmed );
2875 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2876 if ( $diff > 0 ) {
2877 $w2 = substr( $ltrimmed, -$diff );
2878 } else {
2879 $w2 = '';
2880 }
2881 return [ $w1, $trimmed, $w2 ];
2882 }
2883
2884 /**
2885 * Replace magic variables, templates, and template arguments
2886 * with the appropriate text. Templates are substituted recursively,
2887 * taking care to avoid infinite loops.
2888 *
2889 * Note that the substitution depends on value of $mOutputType:
2890 * self::OT_WIKI: only {{subst:}} templates
2891 * self::OT_PREPROCESS: templates but not extension tags
2892 * self::OT_HTML: all templates and extension tags
2893 *
2894 * @param string $text The text to transform
2895 * @param bool|PPFrame $frame Object describing the arguments passed to the
2896 * template. Arguments may also be provided as an associative array, as
2897 * was the usual case before MW1.12. Providing arguments this way may be
2898 * useful for extensions wishing to perform variable replacement
2899 * explicitly.
2900 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2901 * double-brace expansion.
2902 * @return string
2903 */
2904 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2905 # Is there any text? Also, Prevent too big inclusions!
2906 $textSize = strlen( $text );
2907 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2908 return $text;
2909 }
2910
2911 if ( $frame === false ) {
2912 $frame = $this->getPreprocessor()->newFrame();
2913 } elseif ( !( $frame instanceof PPFrame ) ) {
2914 wfDebug( __METHOD__ . " called using plain parameters instead of "
2915 . "a PPFrame instance. Creating custom frame.\n" );
2916 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2917 }
2918
2919 $dom = $this->preprocessToDom( $text );
2920 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2921 $text = $frame->expand( $dom, $flags );
2922
2923 return $text;
2924 }
2925
2926 /**
2927 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2928 *
2929 * @param array $args
2930 *
2931 * @return array
2932 */
2933 public static function createAssocArgs( $args ) {
2934 $assocArgs = [];
2935 $index = 1;
2936 foreach ( $args as $arg ) {
2937 $eqpos = strpos( $arg, '=' );
2938 if ( $eqpos === false ) {
2939 $assocArgs[$index++] = $arg;
2940 } else {
2941 $name = trim( substr( $arg, 0, $eqpos ) );
2942 $value = trim( substr( $arg, $eqpos + 1 ) );
2943 if ( $value === false ) {
2944 $value = '';
2945 }
2946 if ( $name !== false ) {
2947 $assocArgs[$name] = $value;
2948 }
2949 }
2950 }
2951
2952 return $assocArgs;
2953 }
2954
2955 /**
2956 * Warn the user when a parser limitation is reached
2957 * Will warn at most once the user per limitation type
2958 *
2959 * The results are shown during preview and run through the Parser (See EditPage.php)
2960 *
2961 * @param string $limitationType Should be one of:
2962 * 'expensive-parserfunction' (corresponding messages:
2963 * 'expensive-parserfunction-warning',
2964 * 'expensive-parserfunction-category')
2965 * 'post-expand-template-argument' (corresponding messages:
2966 * 'post-expand-template-argument-warning',
2967 * 'post-expand-template-argument-category')
2968 * 'post-expand-template-inclusion' (corresponding messages:
2969 * 'post-expand-template-inclusion-warning',
2970 * 'post-expand-template-inclusion-category')
2971 * 'node-count-exceeded' (corresponding messages:
2972 * 'node-count-exceeded-warning',
2973 * 'node-count-exceeded-category')
2974 * 'expansion-depth-exceeded' (corresponding messages:
2975 * 'expansion-depth-exceeded-warning',
2976 * 'expansion-depth-exceeded-category')
2977 * @param string|int|null $current Current value
2978 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2979 * exceeded, provide the values (optional)
2980 */
2981 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2982 # does no harm if $current and $max are present but are unnecessary for the message
2983 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2984 # only during preview, and that would split the parser cache unnecessarily.
2985 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2986 ->text();
2987 $this->mOutput->addWarning( $warning );
2988 $this->addTrackingCategory( "$limitationType-category" );
2989 }
2990
2991 /**
2992 * Return the text of a template, after recursively
2993 * replacing any variables or templates within the template.
2994 *
2995 * @param array $piece The parts of the template
2996 * $piece['title']: the title, i.e. the part before the |
2997 * $piece['parts']: the parameter array
2998 * $piece['lineStart']: whether the brace was at the start of a line
2999 * @param PPFrame $frame The current frame, contains template arguments
3000 * @throws Exception
3001 * @return string The text of the template
3002 */
3003 public function braceSubstitution( $piece, $frame ) {
3004 // Flags
3005
3006 // $text has been filled
3007 $found = false;
3008 // wiki markup in $text should be escaped
3009 $nowiki = false;
3010 // $text is HTML, armour it against wikitext transformation
3011 $isHTML = false;
3012 // Force interwiki transclusion to be done in raw mode not rendered
3013 $forceRawInterwiki = false;
3014 // $text is a DOM node needing expansion in a child frame
3015 $isChildObj = false;
3016 // $text is a DOM node needing expansion in the current frame
3017 $isLocalObj = false;
3018
3019 # Title object, where $text came from
3020 $title = false;
3021
3022 # $part1 is the bit before the first |, and must contain only title characters.
3023 # Various prefixes will be stripped from it later.
3024 $titleWithSpaces = $frame->expand( $piece['title'] );
3025 $part1 = trim( $titleWithSpaces );
3026 $titleText = false;
3027
3028 # Original title text preserved for various purposes
3029 $originalTitle = $part1;
3030
3031 # $args is a list of argument nodes, starting from index 0, not including $part1
3032 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3033 # below won't work b/c this $args isn't an object
3034 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3035
3036 $profileSection = null; // profile templates
3037
3038 # SUBST
3039 if ( !$found ) {
3040 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3041
3042 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3043 # Decide whether to expand template or keep wikitext as-is.
3044 if ( $this->ot['wiki'] ) {
3045 if ( $substMatch === false ) {
3046 $literal = true; # literal when in PST with no prefix
3047 } else {
3048 $literal = false; # expand when in PST with subst: or safesubst:
3049 }
3050 } else {
3051 if ( $substMatch == 'subst' ) {
3052 $literal = true; # literal when not in PST with plain subst:
3053 } else {
3054 $literal = false; # expand when not in PST with safesubst: or no prefix
3055 }
3056 }
3057 if ( $literal ) {
3058 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3059 $isLocalObj = true;
3060 $found = true;
3061 }
3062 }
3063
3064 # Variables
3065 if ( !$found && $args->getLength() == 0 ) {
3066 $id = $this->mVariables->matchStartToEnd( $part1 );
3067 if ( $id !== false ) {
3068 $text = $this->getVariableValue( $id, $frame );
3069 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3070 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3071 }
3072 $found = true;
3073 }
3074 }
3075
3076 # MSG, MSGNW and RAW
3077 if ( !$found ) {
3078 # Check for MSGNW:
3079 $mwMsgnw = MagicWord::get( 'msgnw' );
3080 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3081 $nowiki = true;
3082 } else {
3083 # Remove obsolete MSG:
3084 $mwMsg = MagicWord::get( 'msg' );
3085 $mwMsg->matchStartAndRemove( $part1 );
3086 }
3087
3088 # Check for RAW:
3089 $mwRaw = MagicWord::get( 'raw' );
3090 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3091 $forceRawInterwiki = true;
3092 }
3093 }
3094
3095 # Parser functions
3096 if ( !$found ) {
3097 $colonPos = strpos( $part1, ':' );
3098 if ( $colonPos !== false ) {
3099 $func = substr( $part1, 0, $colonPos );
3100 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3101 $argsLength = $args->getLength();
3102 for ( $i = 0; $i < $argsLength; $i++ ) {
3103 $funcArgs[] = $args->item( $i );
3104 }
3105 try {
3106 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3107 } catch ( Exception $ex ) {
3108 throw $ex;
3109 }
3110
3111 # The interface for parser functions allows for extracting
3112 # flags into the local scope. Extract any forwarded flags
3113 # here.
3114 extract( $result );
3115 }
3116 }
3117
3118 # Finish mangling title and then check for loops.
3119 # Set $title to a Title object and $titleText to the PDBK
3120 if ( !$found ) {
3121 $ns = NS_TEMPLATE;
3122 # Split the title into page and subpage
3123 $subpage = '';
3124 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3125 if ( $part1 !== $relative ) {
3126 $part1 = $relative;
3127 $ns = $this->mTitle->getNamespace();
3128 }
3129 $title = Title::newFromText( $part1, $ns );
3130 if ( $title ) {
3131 $titleText = $title->getPrefixedText();
3132 # Check for language variants if the template is not found
3133 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3134 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3135 }
3136 # Do recursion depth check
3137 $limit = $this->mOptions->getMaxTemplateDepth();
3138 if ( $frame->depth >= $limit ) {
3139 $found = true;
3140 $text = '<span class="error">'
3141 . wfMessage( 'parser-template-recursion-depth-warning' )
3142 ->numParams( $limit )->inContentLanguage()->text()
3143 . '</span>';
3144 }
3145 }
3146 }
3147
3148 # Load from database
3149 if ( !$found && $title ) {
3150 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3151 if ( !$title->isExternal() ) {
3152 if ( $title->isSpecialPage()
3153 && $this->mOptions->getAllowSpecialInclusion()
3154 && $this->ot['html']
3155 ) {
3156 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3157 // Pass the template arguments as URL parameters.
3158 // "uselang" will have no effect since the Language object
3159 // is forced to the one defined in ParserOptions.
3160 $pageArgs = [];
3161 $argsLength = $args->getLength();
3162 for ( $i = 0; $i < $argsLength; $i++ ) {
3163 $bits = $args->item( $i )->splitArg();
3164 if ( strval( $bits['index'] ) === '' ) {
3165 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3166 $value = trim( $frame->expand( $bits['value'] ) );
3167 $pageArgs[$name] = $value;
3168 }
3169 }
3170
3171 // Create a new context to execute the special page
3172 $context = new RequestContext;
3173 $context->setTitle( $title );
3174 $context->setRequest( new FauxRequest( $pageArgs ) );
3175 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3176 $context->setUser( $this->getUser() );
3177 } else {
3178 // If this page is cached, then we better not be per user.
3179 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3180 }
3181 $context->setLanguage( $this->mOptions->getUserLangObj() );
3182 $ret = SpecialPageFactory::capturePath(
3183 $title, $context, $this->getLinkRenderer() );
3184 if ( $ret ) {
3185 $text = $context->getOutput()->getHTML();
3186 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3187 $found = true;
3188 $isHTML = true;
3189 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3190 $this->mOutput->updateRuntimeAdaptiveExpiry(
3191 $specialPage->maxIncludeCacheTime()
3192 );
3193 }
3194 }
3195 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3196 $found = false; # access denied
3197 wfDebug( __METHOD__ . ": template inclusion denied for " .
3198 $title->getPrefixedDBkey() . "\n" );
3199 } else {
3200 list( $text, $title ) = $this->getTemplateDom( $title );
3201 if ( $text !== false ) {
3202 $found = true;
3203 $isChildObj = true;
3204 }
3205 }
3206
3207 # If the title is valid but undisplayable, make a link to it
3208 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3209 $text = "[[:$titleText]]";
3210 $found = true;
3211 }
3212 } elseif ( $title->isTrans() ) {
3213 # Interwiki transclusion
3214 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3215 $text = $this->interwikiTransclude( $title, 'render' );
3216 $isHTML = true;
3217 } else {
3218 $text = $this->interwikiTransclude( $title, 'raw' );
3219 # Preprocess it like a template
3220 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3221 $isChildObj = true;
3222 }
3223 $found = true;
3224 }
3225
3226 # Do infinite loop check
3227 # This has to be done after redirect resolution to avoid infinite loops via redirects
3228 if ( !$frame->loopCheck( $title ) ) {
3229 $found = true;
3230 $text = '<span class="error">'
3231 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3232 . '</span>';
3233 $this->addTrackingCategory( 'template-loop-category' );
3234 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3235 }
3236 }
3237
3238 # If we haven't found text to substitute by now, we're done
3239 # Recover the source wikitext and return it
3240 if ( !$found ) {
3241 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3242 if ( $profileSection ) {
3243 $this->mProfiler->scopedProfileOut( $profileSection );
3244 }
3245 return [ 'object' => $text ];
3246 }
3247
3248 # Expand DOM-style return values in a child frame
3249 if ( $isChildObj ) {
3250 # Clean up argument array
3251 $newFrame = $frame->newChild( $args, $title );
3252
3253 if ( $nowiki ) {
3254 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3255 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3256 # Expansion is eligible for the empty-frame cache
3257 $text = $newFrame->cachedExpand( $titleText, $text );
3258 } else {
3259 # Uncached expansion
3260 $text = $newFrame->expand( $text );
3261 }
3262 }
3263 if ( $isLocalObj && $nowiki ) {
3264 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3265 $isLocalObj = false;
3266 }
3267
3268 if ( $profileSection ) {
3269 $this->mProfiler->scopedProfileOut( $profileSection );
3270 }
3271
3272 # Replace raw HTML by a placeholder
3273 if ( $isHTML ) {
3274 $text = $this->insertStripItem( $text );
3275 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3276 # Escape nowiki-style return values
3277 $text = wfEscapeWikiText( $text );
3278 } elseif ( is_string( $text )
3279 && !$piece['lineStart']
3280 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3281 ) {
3282 # T2529: if the template begins with a table or block-level
3283 # element, it should be treated as beginning a new line.
3284 # This behavior is somewhat controversial.
3285 $text = "\n" . $text;
3286 }
3287
3288 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3289 # Error, oversize inclusion
3290 if ( $titleText !== false ) {
3291 # Make a working, properly escaped link if possible (T25588)
3292 $text = "[[:$titleText]]";
3293 } else {
3294 # This will probably not be a working link, but at least it may
3295 # provide some hint of where the problem is
3296 preg_replace( '/^:/', '', $originalTitle );
3297 $text = "[[:$originalTitle]]";
3298 }
3299 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3300 . 'post-expand include size too large -->' );
3301 $this->limitationWarn( 'post-expand-template-inclusion' );
3302 }
3303
3304 if ( $isLocalObj ) {
3305 $ret = [ 'object' => $text ];
3306 } else {
3307 $ret = [ 'text' => $text ];
3308 }
3309
3310 return $ret;
3311 }
3312
3313 /**
3314 * Call a parser function and return an array with text and flags.
3315 *
3316 * The returned array will always contain a boolean 'found', indicating
3317 * whether the parser function was found or not. It may also contain the
3318 * following:
3319 * text: string|object, resulting wikitext or PP DOM object
3320 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3321 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3322 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3323 * nowiki: bool, wiki markup in $text should be escaped
3324 *
3325 * @since 1.21
3326 * @param PPFrame $frame The current frame, contains template arguments
3327 * @param string $function Function name
3328 * @param array $args Arguments to the function
3329 * @throws MWException
3330 * @return array
3331 */
3332 public function callParserFunction( $frame, $function, array $args = [] ) {
3333 global $wgContLang;
3334
3335 # Case sensitive functions
3336 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3337 $function = $this->mFunctionSynonyms[1][$function];
3338 } else {
3339 # Case insensitive functions
3340 $function = $wgContLang->lc( $function );
3341 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3342 $function = $this->mFunctionSynonyms[0][$function];
3343 } else {
3344 return [ 'found' => false ];
3345 }
3346 }
3347
3348 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3349
3350 // Avoid PHP 7.1 warning from passing $this by reference
3351 $parser = $this;
3352
3353 $allArgs = [ &$parser ];
3354 if ( $flags & self::SFH_OBJECT_ARGS ) {
3355 # Convert arguments to PPNodes and collect for appending to $allArgs
3356 $funcArgs = [];
3357 foreach ( $args as $k => $v ) {
3358 if ( $v instanceof PPNode || $k === 0 ) {
3359 $funcArgs[] = $v;
3360 } else {
3361 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3362 }
3363 }
3364
3365 # Add a frame parameter, and pass the arguments as an array
3366 $allArgs[] = $frame;
3367 $allArgs[] = $funcArgs;
3368 } else {
3369 # Convert arguments to plain text and append to $allArgs
3370 foreach ( $args as $k => $v ) {
3371 if ( $v instanceof PPNode ) {
3372 $allArgs[] = trim( $frame->expand( $v ) );
3373 } elseif ( is_int( $k ) && $k >= 0 ) {
3374 $allArgs[] = trim( $v );
3375 } else {
3376 $allArgs[] = trim( "$k=$v" );
3377 }
3378 }
3379 }
3380
3381 $result = call_user_func_array( $callback, $allArgs );
3382
3383 # The interface for function hooks allows them to return a wikitext
3384 # string or an array containing the string and any flags. This mungs
3385 # things around to match what this method should return.
3386 if ( !is_array( $result ) ) {
3387 $result = [
3388 'found' => true,
3389 'text' => $result,
3390 ];
3391 } else {
3392 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3393 $result['text'] = $result[0];
3394 }
3395 unset( $result[0] );
3396 $result += [
3397 'found' => true,
3398 ];
3399 }
3400
3401 $noparse = true;
3402 $preprocessFlags = 0;
3403 if ( isset( $result['noparse'] ) ) {
3404 $noparse = $result['noparse'];
3405 }
3406 if ( isset( $result['preprocessFlags'] ) ) {
3407 $preprocessFlags = $result['preprocessFlags'];
3408 }
3409
3410 if ( !$noparse ) {
3411 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3412 $result['isChildObj'] = true;
3413 }
3414
3415 return $result;
3416 }
3417
3418 /**
3419 * Get the semi-parsed DOM representation of a template with a given title,
3420 * and its redirect destination title. Cached.
3421 *
3422 * @param Title $title
3423 *
3424 * @return array
3425 */
3426 public function getTemplateDom( $title ) {
3427 $cacheTitle = $title;
3428 $titleText = $title->getPrefixedDBkey();
3429
3430 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3431 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3432 $title = Title::makeTitle( $ns, $dbk );
3433 $titleText = $title->getPrefixedDBkey();
3434 }
3435 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3436 return [ $this->mTplDomCache[$titleText], $title ];
3437 }
3438
3439 # Cache miss, go to the database
3440 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3441
3442 if ( $text === false ) {
3443 $this->mTplDomCache[$titleText] = false;
3444 return [ false, $title ];
3445 }
3446
3447 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3448 $this->mTplDomCache[$titleText] = $dom;
3449
3450 if ( !$title->equals( $cacheTitle ) ) {
3451 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3452 [ $title->getNamespace(), $title->getDBkey() ];
3453 }
3454
3455 return [ $dom, $title ];
3456 }
3457
3458 /**
3459 * Fetch the current revision of a given title. Note that the revision
3460 * (and even the title) may not exist in the database, so everything
3461 * contributing to the output of the parser should use this method
3462 * where possible, rather than getting the revisions themselves. This
3463 * method also caches its results, so using it benefits performance.
3464 *
3465 * @since 1.24
3466 * @param Title $title
3467 * @return Revision
3468 */
3469 public function fetchCurrentRevisionOfTitle( $title ) {
3470 $cacheKey = $title->getPrefixedDBkey();
3471 if ( !$this->currentRevisionCache ) {
3472 $this->currentRevisionCache = new MapCacheLRU( 100 );
3473 }
3474 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3475 $this->currentRevisionCache->set( $cacheKey,
3476 // Defaults to Parser::statelessFetchRevision()
3477 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3478 );
3479 }
3480 return $this->currentRevisionCache->get( $cacheKey );
3481 }
3482
3483 /**
3484 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3485 * without passing them on to it.
3486 *
3487 * @since 1.24
3488 * @param Title $title
3489 * @param Parser|bool $parser
3490 * @return Revision|bool False if missing
3491 */
3492 public static function statelessFetchRevision( Title $title, $parser = false ) {
3493 $pageId = $title->getArticleID();
3494 $revId = $title->getLatestRevID();
3495
3496 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3497 if ( $rev ) {
3498 $rev->setTitle( $title );
3499 }
3500
3501 return $rev;
3502 }
3503
3504 /**
3505 * Fetch the unparsed text of a template and register a reference to it.
3506 * @param Title $title
3507 * @return array ( string or false, Title )
3508 */
3509 public function fetchTemplateAndTitle( $title ) {
3510 // Defaults to Parser::statelessFetchTemplate()
3511 $templateCb = $this->mOptions->getTemplateCallback();
3512 $stuff = call_user_func( $templateCb, $title, $this );
3513 // We use U+007F DELETE to distinguish strip markers from regular text.
3514 $text = $stuff['text'];
3515 if ( is_string( $stuff['text'] ) ) {
3516 $text = strtr( $text, "\x7f", "?" );
3517 }
3518 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3519 if ( isset( $stuff['deps'] ) ) {
3520 foreach ( $stuff['deps'] as $dep ) {
3521 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3522 if ( $dep['title']->equals( $this->getTitle() ) ) {
3523 // If we transclude ourselves, the final result
3524 // will change based on the new version of the page
3525 $this->mOutput->setFlag( 'vary-revision' );
3526 }
3527 }
3528 }
3529 return [ $text, $finalTitle ];
3530 }
3531
3532 /**
3533 * Fetch the unparsed text of a template and register a reference to it.
3534 * @param Title $title
3535 * @return string|bool
3536 */
3537 public function fetchTemplate( $title ) {
3538 return $this->fetchTemplateAndTitle( $title )[0];
3539 }
3540
3541 /**
3542 * Static function to get a template
3543 * Can be overridden via ParserOptions::setTemplateCallback().
3544 *
3545 * @param Title $title
3546 * @param bool|Parser $parser
3547 *
3548 * @return array
3549 */
3550 public static function statelessFetchTemplate( $title, $parser = false ) {
3551 $text = $skip = false;
3552 $finalTitle = $title;
3553 $deps = [];
3554
3555 # Loop to fetch the article, with up to 1 redirect
3556 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3557 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3558 // @codingStandardsIgnoreEnd
3559 # Give extensions a chance to select the revision instead
3560 $id = false; # Assume current
3561 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3562 [ $parser, $title, &$skip, &$id ] );
3563
3564 if ( $skip ) {
3565 $text = false;
3566 $deps[] = [
3567 'title' => $title,
3568 'page_id' => $title->getArticleID(),
3569 'rev_id' => null
3570 ];
3571 break;
3572 }
3573 # Get the revision
3574 if ( $id ) {
3575 $rev = Revision::newFromId( $id );
3576 } elseif ( $parser ) {
3577 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3578 } else {
3579 $rev = Revision::newFromTitle( $title );
3580 }
3581 $rev_id = $rev ? $rev->getId() : 0;
3582 # If there is no current revision, there is no page
3583 if ( $id === false && !$rev ) {
3584 $linkCache = LinkCache::singleton();
3585 $linkCache->addBadLinkObj( $title );
3586 }
3587
3588 $deps[] = [
3589 'title' => $title,
3590 'page_id' => $title->getArticleID(),
3591 'rev_id' => $rev_id ];
3592 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3593 # We fetched a rev from a different title; register it too...
3594 $deps[] = [
3595 'title' => $rev->getTitle(),
3596 'page_id' => $rev->getPage(),
3597 'rev_id' => $rev_id ];
3598 }
3599
3600 if ( $rev ) {
3601 $content = $rev->getContent();
3602 $text = $content ? $content->getWikitextForTransclusion() : null;
3603
3604 Hooks::run( 'ParserFetchTemplate',
3605 [ $parser, $title, $rev, &$text, &$deps ] );
3606
3607 if ( $text === false || $text === null ) {
3608 $text = false;
3609 break;
3610 }
3611 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3612 global $wgContLang;
3613 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3614 if ( !$message->exists() ) {
3615 $text = false;
3616 break;
3617 }
3618 $content = $message->content();
3619 $text = $message->plain();
3620 } else {
3621 break;
3622 }
3623 if ( !$content ) {
3624 break;
3625 }
3626 # Redirect?
3627 $finalTitle = $title;
3628 $title = $content->getRedirectTarget();
3629 }
3630 return [
3631 'text' => $text,
3632 'finalTitle' => $finalTitle,
3633 'deps' => $deps ];
3634 }
3635
3636 /**
3637 * Fetch a file and its title and register a reference to it.
3638 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3639 * @param Title $title
3640 * @param array $options Array of options to RepoGroup::findFile
3641 * @return File|bool
3642 */
3643 public function fetchFile( $title, $options = [] ) {
3644 return $this->fetchFileAndTitle( $title, $options )[0];
3645 }
3646
3647 /**
3648 * Fetch a file and its title and register a reference to it.
3649 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3650 * @param Title $title
3651 * @param array $options Array of options to RepoGroup::findFile
3652 * @return array ( File or false, Title of file )
3653 */
3654 public function fetchFileAndTitle( $title, $options = [] ) {
3655 $file = $this->fetchFileNoRegister( $title, $options );
3656
3657 $time = $file ? $file->getTimestamp() : false;
3658 $sha1 = $file ? $file->getSha1() : false;
3659 # Register the file as a dependency...
3660 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3661 if ( $file && !$title->equals( $file->getTitle() ) ) {
3662 # Update fetched file title
3663 $title = $file->getTitle();
3664 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3665 }
3666 return [ $file, $title ];
3667 }
3668
3669 /**
3670 * Helper function for fetchFileAndTitle.
3671 *
3672 * Also useful if you need to fetch a file but not use it yet,
3673 * for example to get the file's handler.
3674 *
3675 * @param Title $title
3676 * @param array $options Array of options to RepoGroup::findFile
3677 * @return File|bool
3678 */
3679 protected function fetchFileNoRegister( $title, $options = [] ) {
3680 if ( isset( $options['broken'] ) ) {
3681 $file = false; // broken thumbnail forced by hook
3682 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3683 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3684 } else { // get by (name,timestamp)
3685 $file = wfFindFile( $title, $options );
3686 }
3687 return $file;
3688 }
3689
3690 /**
3691 * Transclude an interwiki link.
3692 *
3693 * @param Title $title
3694 * @param string $action
3695 *
3696 * @return string
3697 */
3698 public function interwikiTransclude( $title, $action ) {
3699 global $wgEnableScaryTranscluding;
3700
3701 if ( !$wgEnableScaryTranscluding ) {
3702 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3703 }
3704
3705 $url = $title->getFullURL( [ 'action' => $action ] );
3706
3707 if ( strlen( $url ) > 255 ) {
3708 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3709 }
3710 return $this->fetchScaryTemplateMaybeFromCache( $url );
3711 }
3712
3713 /**
3714 * @param string $url
3715 * @return mixed|string
3716 */
3717 public function fetchScaryTemplateMaybeFromCache( $url ) {
3718 global $wgTranscludeCacheExpiry;
3719 $dbr = wfGetDB( DB_REPLICA );
3720 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3721 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3722 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3723 if ( $obj ) {
3724 return $obj->tc_contents;
3725 }
3726
3727 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3728 $status = $req->execute(); // Status object
3729 if ( $status->isOK() ) {
3730 $text = $req->getContent();
3731 } elseif ( $req->getStatus() != 200 ) {
3732 // Though we failed to fetch the content, this status is useless.
3733 return wfMessage( 'scarytranscludefailed-httpstatus' )
3734 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3735 } else {
3736 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3737 }
3738
3739 $dbw = wfGetDB( DB_MASTER );
3740 $dbw->replace( 'transcache', [ 'tc_url' ], [
3741 'tc_url' => $url,
3742 'tc_time' => $dbw->timestamp( time() ),
3743 'tc_contents' => $text
3744 ] );
3745 return $text;
3746 }
3747
3748 /**
3749 * Triple brace replacement -- used for template arguments
3750 * @private
3751 *
3752 * @param array $piece
3753 * @param PPFrame $frame
3754 *
3755 * @return array
3756 */
3757 public function argSubstitution( $piece, $frame ) {
3758 $error = false;
3759 $parts = $piece['parts'];
3760 $nameWithSpaces = $frame->expand( $piece['title'] );
3761 $argName = trim( $nameWithSpaces );
3762 $object = false;
3763 $text = $frame->getArgument( $argName );
3764 if ( $text === false && $parts->getLength() > 0
3765 && ( $this->ot['html']
3766 || $this->ot['pre']
3767 || ( $this->ot['wiki'] && $frame->isTemplate() )
3768 )
3769 ) {
3770 # No match in frame, use the supplied default
3771 $object = $parts->item( 0 )->getChildren();
3772 }
3773 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3774 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3775 $this->limitationWarn( 'post-expand-template-argument' );
3776 }
3777
3778 if ( $text === false && $object === false ) {
3779 # No match anywhere
3780 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3781 }
3782 if ( $error !== false ) {
3783 $text .= $error;
3784 }
3785 if ( $object !== false ) {
3786 $ret = [ 'object' => $object ];
3787 } else {
3788 $ret = [ 'text' => $text ];
3789 }
3790
3791 return $ret;
3792 }
3793
3794 /**
3795 * Return the text to be used for a given extension tag.
3796 * This is the ghost of strip().
3797 *
3798 * @param array $params Associative array of parameters:
3799 * name PPNode for the tag name
3800 * attr PPNode for unparsed text where tag attributes are thought to be
3801 * attributes Optional associative array of parsed attributes
3802 * inner Contents of extension element
3803 * noClose Original text did not have a close tag
3804 * @param PPFrame $frame
3805 *
3806 * @throws MWException
3807 * @return string
3808 */
3809 public function extensionSubstitution( $params, $frame ) {
3810 static $errorStr = '<span class="error">';
3811 static $errorLen = 20;
3812
3813 $name = $frame->expand( $params['name'] );
3814 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3815 // Probably expansion depth or node count exceeded. Just punt the
3816 // error up.
3817 return $name;
3818 }
3819
3820 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3821 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3822 // See above
3823 return $attrText;
3824 }
3825
3826 // We can't safely check if the expansion for $content resulted in an
3827 // error, because the content could happen to be the error string
3828 // (T149622).
3829 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3830
3831 $marker = self::MARKER_PREFIX . "-$name-"
3832 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3833
3834 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3835 ( $this->ot['html'] || $this->ot['pre'] );
3836 if ( $isFunctionTag ) {
3837 $markerType = 'none';
3838 } else {
3839 $markerType = 'general';
3840 }
3841 if ( $this->ot['html'] || $isFunctionTag ) {
3842 $name = strtolower( $name );
3843 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3844 if ( isset( $params['attributes'] ) ) {
3845 $attributes = $attributes + $params['attributes'];
3846 }
3847
3848 if ( isset( $this->mTagHooks[$name] ) ) {
3849 $output = call_user_func_array( $this->mTagHooks[$name],
3850 [ $content, $attributes, $this, $frame ] );
3851 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3852 list( $callback, ) = $this->mFunctionTagHooks[$name];
3853
3854 // Avoid PHP 7.1 warning from passing $this by reference
3855 $parser = $this;
3856 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3857 } else {
3858 $output = '<span class="error">Invalid tag extension name: ' .
3859 htmlspecialchars( $name ) . '</span>';
3860 }
3861
3862 if ( is_array( $output ) ) {
3863 # Extract flags to local scope (to override $markerType)
3864 $flags = $output;
3865 $output = $flags[0];
3866 unset( $flags[0] );
3867 extract( $flags );
3868 }
3869 } else {
3870 if ( is_null( $attrText ) ) {
3871 $attrText = '';
3872 }
3873 if ( isset( $params['attributes'] ) ) {
3874 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3875 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3876 htmlspecialchars( $attrValue ) . '"';
3877 }
3878 }
3879 if ( $content === null ) {
3880 $output = "<$name$attrText/>";
3881 } else {
3882 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3883 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3884 // See above
3885 return $close;
3886 }
3887 $output = "<$name$attrText>$content$close";
3888 }
3889 }
3890
3891 if ( $markerType === 'none' ) {
3892 return $output;
3893 } elseif ( $markerType === 'nowiki' ) {
3894 $this->mStripState->addNoWiki( $marker, $output );
3895 } elseif ( $markerType === 'general' ) {
3896 $this->mStripState->addGeneral( $marker, $output );
3897 } else {
3898 throw new MWException( __METHOD__ . ': invalid marker type' );
3899 }
3900 return $marker;
3901 }
3902
3903 /**
3904 * Increment an include size counter
3905 *
3906 * @param string $type The type of expansion
3907 * @param int $size The size of the text
3908 * @return bool False if this inclusion would take it over the maximum, true otherwise
3909 */
3910 public function incrementIncludeSize( $type, $size ) {
3911 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3912 return false;
3913 } else {
3914 $this->mIncludeSizes[$type] += $size;
3915 return true;
3916 }
3917 }
3918
3919 /**
3920 * Increment the expensive function count
3921 *
3922 * @return bool False if the limit has been exceeded
3923 */
3924 public function incrementExpensiveFunctionCount() {
3925 $this->mExpensiveFunctionCount++;
3926 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3927 }
3928
3929 /**
3930 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3931 * Fills $this->mDoubleUnderscores, returns the modified text
3932 *
3933 * @param string $text
3934 *
3935 * @return string
3936 */
3937 public function doDoubleUnderscore( $text ) {
3938 # The position of __TOC__ needs to be recorded
3939 $mw = MagicWord::get( 'toc' );
3940 if ( $mw->match( $text ) ) {
3941 $this->mShowToc = true;
3942 $this->mForceTocPosition = true;
3943
3944 # Set a placeholder. At the end we'll fill it in with the TOC.
3945 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3946
3947 # Only keep the first one.
3948 $text = $mw->replace( '', $text );
3949 }
3950
3951 # Now match and remove the rest of them
3952 $mwa = MagicWord::getDoubleUnderscoreArray();
3953 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3954
3955 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3956 $this->mOutput->mNoGallery = true;
3957 }
3958 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3959 $this->mShowToc = false;
3960 }
3961 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3962 && $this->mTitle->getNamespace() == NS_CATEGORY
3963 ) {
3964 $this->addTrackingCategory( 'hidden-category-category' );
3965 }
3966 # (T10068) Allow control over whether robots index a page.
3967 # __INDEX__ always overrides __NOINDEX__, see T16899
3968 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3969 $this->mOutput->setIndexPolicy( 'noindex' );
3970 $this->addTrackingCategory( 'noindex-category' );
3971 }
3972 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3973 $this->mOutput->setIndexPolicy( 'index' );
3974 $this->addTrackingCategory( 'index-category' );
3975 }
3976
3977 # Cache all double underscores in the database
3978 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3979 $this->mOutput->setProperty( $key, '' );
3980 }
3981
3982 return $text;
3983 }
3984
3985 /**
3986 * @see ParserOutput::addTrackingCategory()
3987 * @param string $msg Message key
3988 * @return bool Whether the addition was successful
3989 */
3990 public function addTrackingCategory( $msg ) {
3991 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3992 }
3993
3994 /**
3995 * This function accomplishes several tasks:
3996 * 1) Auto-number headings if that option is enabled
3997 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3998 * 3) Add a Table of contents on the top for users who have enabled the option
3999 * 4) Auto-anchor headings
4000 *
4001 * It loops through all headlines, collects the necessary data, then splits up the
4002 * string and re-inserts the newly formatted headlines.
4003 *
4004 * @param string $text
4005 * @param string $origText Original, untouched wikitext
4006 * @param bool $isMain
4007 * @return mixed|string
4008 * @private
4009 */
4010 public function formatHeadings( $text, $origText, $isMain = true ) {
4011 global $wgMaxTocLevel;
4012
4013 # Inhibit editsection links if requested in the page
4014 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4015 $maybeShowEditLink = $showEditLink = false;
4016 } else {
4017 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4018 $showEditLink = $this->mOptions->getEditSection();
4019 }
4020 if ( $showEditLink ) {
4021 $this->mOutput->setEditSectionTokens( true );
4022 }
4023
4024 # Get all headlines for numbering them and adding funky stuff like [edit]
4025 # links - this is for later, but we need the number of headlines right now
4026 $matches = [];
4027 $numMatches = preg_match_all(
4028 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4029 $text,
4030 $matches
4031 );
4032
4033 # if there are fewer than 4 headlines in the article, do not show TOC
4034 # unless it's been explicitly enabled.
4035 $enoughToc = $this->mShowToc &&
4036 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4037
4038 # Allow user to stipulate that a page should have a "new section"
4039 # link added via __NEWSECTIONLINK__
4040 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4041 $this->mOutput->setNewSection( true );
4042 }
4043
4044 # Allow user to remove the "new section"
4045 # link via __NONEWSECTIONLINK__
4046 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4047 $this->mOutput->hideNewSection( true );
4048 }
4049
4050 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4051 # override above conditions and always show TOC above first header
4052 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4053 $this->mShowToc = true;
4054 $enoughToc = true;
4055 }
4056
4057 # headline counter
4058 $headlineCount = 0;
4059 $numVisible = 0;
4060
4061 # Ugh .. the TOC should have neat indentation levels which can be
4062 # passed to the skin functions. These are determined here
4063 $toc = '';
4064 $full = '';
4065 $head = [];
4066 $sublevelCount = [];
4067 $levelCount = [];
4068 $level = 0;
4069 $prevlevel = 0;
4070 $toclevel = 0;
4071 $prevtoclevel = 0;
4072 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4073 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4074 $oldType = $this->mOutputType;
4075 $this->setOutputType( self::OT_WIKI );
4076 $frame = $this->getPreprocessor()->newFrame();
4077 $root = $this->preprocessToDom( $origText );
4078 $node = $root->getFirstChild();
4079 $byteOffset = 0;
4080 $tocraw = [];
4081 $refers = [];
4082
4083 $headlines = $numMatches !== false ? $matches[3] : [];
4084
4085 foreach ( $headlines as $headline ) {
4086 $isTemplate = false;
4087 $titleText = false;
4088 $sectionIndex = false;
4089 $numbering = '';
4090 $markerMatches = [];
4091 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4092 $serial = $markerMatches[1];
4093 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4094 $isTemplate = ( $titleText != $baseTitleText );
4095 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4096 }
4097
4098 if ( $toclevel ) {
4099 $prevlevel = $level;
4100 }
4101 $level = $matches[1][$headlineCount];
4102
4103 if ( $level > $prevlevel ) {
4104 # Increase TOC level
4105 $toclevel++;
4106 $sublevelCount[$toclevel] = 0;
4107 if ( $toclevel < $wgMaxTocLevel ) {
4108 $prevtoclevel = $toclevel;
4109 $toc .= Linker::tocIndent();
4110 $numVisible++;
4111 }
4112 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4113 # Decrease TOC level, find level to jump to
4114
4115 for ( $i = $toclevel; $i > 0; $i-- ) {
4116 if ( $levelCount[$i] == $level ) {
4117 # Found last matching level
4118 $toclevel = $i;
4119 break;
4120 } elseif ( $levelCount[$i] < $level ) {
4121 # Found first matching level below current level
4122 $toclevel = $i + 1;
4123 break;
4124 }
4125 }
4126 if ( $i == 0 ) {
4127 $toclevel = 1;
4128 }
4129 if ( $toclevel < $wgMaxTocLevel ) {
4130 if ( $prevtoclevel < $wgMaxTocLevel ) {
4131 # Unindent only if the previous toc level was shown :p
4132 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4133 $prevtoclevel = $toclevel;
4134 } else {
4135 $toc .= Linker::tocLineEnd();
4136 }
4137 }
4138 } else {
4139 # No change in level, end TOC line
4140 if ( $toclevel < $wgMaxTocLevel ) {
4141 $toc .= Linker::tocLineEnd();
4142 }
4143 }
4144
4145 $levelCount[$toclevel] = $level;
4146
4147 # count number of headlines for each level
4148 $sublevelCount[$toclevel]++;
4149 $dot = 0;
4150 for ( $i = 1; $i <= $toclevel; $i++ ) {
4151 if ( !empty( $sublevelCount[$i] ) ) {
4152 if ( $dot ) {
4153 $numbering .= '.';
4154 }
4155 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4156 $dot = 1;
4157 }
4158 }
4159
4160 # The safe header is a version of the header text safe to use for links
4161
4162 # Remove link placeholders by the link text.
4163 # <!--LINK number-->
4164 # turns into
4165 # link text with suffix
4166 # Do this before unstrip since link text can contain strip markers
4167 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4168
4169 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4170 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4171
4172 # Strip out HTML (first regex removes any tag not allowed)
4173 # Allowed tags are:
4174 # * <sup> and <sub> (T10393)
4175 # * <i> (T28375)
4176 # * <b> (r105284)
4177 # * <bdi> (T74884)
4178 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4179 # * <s> and <strike> (T35715)
4180 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4181 # to allow setting directionality in toc items.
4182 $tocline = preg_replace(
4183 [
4184 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4185 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4186 ],
4187 [ '', '<$1>' ],
4188 $safeHeadline
4189 );
4190
4191 # Strip '<span></span>', which is the result from the above if
4192 # <span id="foo"></span> is used to produce an additional anchor
4193 # for a section.
4194 $tocline = str_replace( '<span></span>', '', $tocline );
4195
4196 $tocline = trim( $tocline );
4197
4198 # For the anchor, strip out HTML-y stuff period
4199 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4200 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4201
4202 # Save headline for section edit hint before it's escaped
4203 $headlineHint = $safeHeadline;
4204
4205 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4206 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4207 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4208 if ( $fallbackHeadline === $safeHeadline ) {
4209 # No reason to have both (in fact, we can't)
4210 $fallbackHeadline = false;
4211 }
4212
4213 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4214 # @todo FIXME: We may be changing them depending on the current locale.
4215 $arrayKey = strtolower( $safeHeadline );
4216 if ( $fallbackHeadline === false ) {
4217 $fallbackArrayKey = false;
4218 } else {
4219 $fallbackArrayKey = strtolower( $fallbackHeadline );
4220 }
4221
4222 # Create the anchor for linking from the TOC to the section
4223 $anchor = $safeHeadline;
4224 $fallbackAnchor = $fallbackHeadline;
4225 if ( isset( $refers[$arrayKey] ) ) {
4226 // @codingStandardsIgnoreStart
4227 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4228 // @codingStandardsIgnoreEnd
4229 $anchor .= "_$i";
4230 $linkAnchor .= "_$i";
4231 $refers["${arrayKey}_$i"] = true;
4232 } else {
4233 $refers[$arrayKey] = true;
4234 }
4235 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4236 // @codingStandardsIgnoreStart
4237 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4238 // @codingStandardsIgnoreEnd
4239 $fallbackAnchor .= "_$i";
4240 $refers["${fallbackArrayKey}_$i"] = true;
4241 } else {
4242 $refers[$fallbackArrayKey] = true;
4243 }
4244
4245 # Don't number the heading if it is the only one (looks silly)
4246 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4247 # the two are different if the line contains a link
4248 $headline = Html::element(
4249 'span',
4250 [ 'class' => 'mw-headline-number' ],
4251 $numbering
4252 ) . ' ' . $headline;
4253 }
4254
4255 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4256 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4257 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4258 }
4259
4260 # Add the section to the section tree
4261 # Find the DOM node for this header
4262 $noOffset = ( $isTemplate || $sectionIndex === false );
4263 while ( $node && !$noOffset ) {
4264 if ( $node->getName() === 'h' ) {
4265 $bits = $node->splitHeading();
4266 if ( $bits['i'] == $sectionIndex ) {
4267 break;
4268 }
4269 }
4270 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4271 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4272 $node = $node->getNextSibling();
4273 }
4274 $tocraw[] = [
4275 'toclevel' => $toclevel,
4276 'level' => $level,
4277 'line' => $tocline,
4278 'number' => $numbering,
4279 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4280 'fromtitle' => $titleText,
4281 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4282 'anchor' => $anchor,
4283 ];
4284
4285 # give headline the correct <h#> tag
4286 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4287 // Output edit section links as markers with styles that can be customized by skins
4288 if ( $isTemplate ) {
4289 # Put a T flag in the section identifier, to indicate to extractSections()
4290 # that sections inside <includeonly> should be counted.
4291 $editsectionPage = $titleText;
4292 $editsectionSection = "T-$sectionIndex";
4293 $editsectionContent = null;
4294 } else {
4295 $editsectionPage = $this->mTitle->getPrefixedText();
4296 $editsectionSection = $sectionIndex;
4297 $editsectionContent = $headlineHint;
4298 }
4299 // We use a bit of pesudo-xml for editsection markers. The
4300 // language converter is run later on. Using a UNIQ style marker
4301 // leads to the converter screwing up the tokens when it
4302 // converts stuff. And trying to insert strip tags fails too. At
4303 // this point all real inputted tags have already been escaped,
4304 // so we don't have to worry about a user trying to input one of
4305 // these markers directly. We use a page and section attribute
4306 // to stop the language converter from converting these
4307 // important bits of data, but put the headline hint inside a
4308 // content block because the language converter is supposed to
4309 // be able to convert that piece of data.
4310 // Gets replaced with html in ParserOutput::getText
4311 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4312 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4313 if ( $editsectionContent !== null ) {
4314 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4315 } else {
4316 $editlink .= '/>';
4317 }
4318 } else {
4319 $editlink = '';
4320 }
4321 $head[$headlineCount] = Linker::makeHeadline( $level,
4322 $matches['attrib'][$headlineCount], $anchor, $headline,
4323 $editlink, $fallbackAnchor );
4324
4325 $headlineCount++;
4326 }
4327
4328 $this->setOutputType( $oldType );
4329
4330 # Never ever show TOC if no headers
4331 if ( $numVisible < 1 ) {
4332 $enoughToc = false;
4333 }
4334
4335 if ( $enoughToc ) {
4336 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4337 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4338 }
4339 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4340 $this->mOutput->setTOCHTML( $toc );
4341 $toc = self::TOC_START . $toc . self::TOC_END;
4342 }
4343
4344 if ( $isMain ) {
4345 $this->mOutput->setSections( $tocraw );
4346 }
4347
4348 # split up and insert constructed headlines
4349 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4350 $i = 0;
4351
4352 // build an array of document sections
4353 $sections = [];
4354 foreach ( $blocks as $block ) {
4355 // $head is zero-based, sections aren't.
4356 if ( empty( $head[$i - 1] ) ) {
4357 $sections[$i] = $block;
4358 } else {
4359 $sections[$i] = $head[$i - 1] . $block;
4360 }
4361
4362 /**
4363 * Send a hook, one per section.
4364 * The idea here is to be able to make section-level DIVs, but to do so in a
4365 * lower-impact, more correct way than r50769
4366 *
4367 * $this : caller
4368 * $section : the section number
4369 * &$sectionContent : ref to the content of the section
4370 * $showEditLinks : boolean describing whether this section has an edit link
4371 */
4372 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4373
4374 $i++;
4375 }
4376
4377 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4378 // append the TOC at the beginning
4379 // Top anchor now in skin
4380 $sections[0] = $sections[0] . $toc . "\n";
4381 }
4382
4383 $full .= implode( '', $sections );
4384
4385 if ( $this->mForceTocPosition ) {
4386 return str_replace( '<!--MWTOC-->', $toc, $full );
4387 } else {
4388 return $full;
4389 }
4390 }
4391
4392 /**
4393 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4394 * conversion, substituting signatures, {{subst:}} templates, etc.
4395 *
4396 * @param string $text The text to transform
4397 * @param Title $title The Title object for the current article
4398 * @param User $user The User object describing the current user
4399 * @param ParserOptions $options Parsing options
4400 * @param bool $clearState Whether to clear the parser state first
4401 * @return string The altered wiki markup
4402 */
4403 public function preSaveTransform( $text, Title $title, User $user,
4404 ParserOptions $options, $clearState = true
4405 ) {
4406 if ( $clearState ) {
4407 $magicScopeVariable = $this->lock();
4408 }
4409 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4410 $this->setUser( $user );
4411
4412 // Strip U+0000 NULL (T159174)
4413 $text = str_replace( "\000", '', $text );
4414
4415 // We still normalize line endings for backwards-compatibility
4416 // with other code that just calls PST, but this should already
4417 // be handled in TextContent subclasses
4418 $text = TextContent::normalizeLineEndings( $text );
4419
4420 if ( $options->getPreSaveTransform() ) {
4421 $text = $this->pstPass2( $text, $user );
4422 }
4423 $text = $this->mStripState->unstripBoth( $text );
4424
4425 $this->setUser( null ); # Reset
4426
4427 return $text;
4428 }
4429
4430 /**
4431 * Pre-save transform helper function
4432 *
4433 * @param string $text
4434 * @param User $user
4435 *
4436 * @return string
4437 */
4438 private function pstPass2( $text, $user ) {
4439 global $wgContLang;
4440
4441 # Note: This is the timestamp saved as hardcoded wikitext to
4442 # the database, we use $wgContLang here in order to give
4443 # everyone the same signature and use the default one rather
4444 # than the one selected in each user's preferences.
4445 # (see also T14815)
4446 $ts = $this->mOptions->getTimestamp();
4447 $timestamp = MWTimestamp::getLocalInstance( $ts );
4448 $ts = $timestamp->format( 'YmdHis' );
4449 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4450
4451 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4452
4453 # Variable replacement
4454 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4455 $text = $this->replaceVariables( $text );
4456
4457 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4458 # which may corrupt this parser instance via its wfMessage()->text() call-
4459
4460 # Signatures
4461 if ( strpos( $text, '~~~' ) !== false ) {
4462 $sigText = $this->getUserSig( $user );
4463 $text = strtr( $text, [
4464 '~~~~~' => $d,
4465 '~~~~' => "$sigText $d",
4466 '~~~' => $sigText
4467 ] );
4468 # The main two signature forms used above are time-sensitive
4469 $this->mOutput->setFlag( 'user-signature' );
4470 }
4471
4472 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4473 $tc = '[' . Title::legalChars() . ']';
4474 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4475
4476 // [[ns:page (context)|]]
4477 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4478 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4479 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4480 // [[ns:page (context), context|]] (using either single or double-width comma)
4481 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4482 // [[|page]] (reverse pipe trick: add context from page title)
4483 $p2 = "/\[\[\\|($tc+)]]/";
4484
4485 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4486 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4487 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4488 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4489
4490 $t = $this->mTitle->getText();
4491 $m = [];
4492 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4493 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4494 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4495 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4496 } else {
4497 # if there's no context, don't bother duplicating the title
4498 $text = preg_replace( $p2, '[[\\1]]', $text );
4499 }
4500
4501 return $text;
4502 }
4503
4504 /**
4505 * Fetch the user's signature text, if any, and normalize to
4506 * validated, ready-to-insert wikitext.
4507 * If you have pre-fetched the nickname or the fancySig option, you can
4508 * specify them here to save a database query.
4509 * Do not reuse this parser instance after calling getUserSig(),
4510 * as it may have changed if it's the $wgParser.
4511 *
4512 * @param User &$user
4513 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4514 * @param bool|null $fancySig whether the nicknname is the complete signature
4515 * or null to use default value
4516 * @return string
4517 */
4518 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4519 global $wgMaxSigChars;
4520
4521 $username = $user->getName();
4522
4523 # If not given, retrieve from the user object.
4524 if ( $nickname === false ) {
4525 $nickname = $user->getOption( 'nickname' );
4526 }
4527
4528 if ( is_null( $fancySig ) ) {
4529 $fancySig = $user->getBoolOption( 'fancysig' );
4530 }
4531
4532 $nickname = $nickname == null ? $username : $nickname;
4533
4534 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4535 $nickname = $username;
4536 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4537 } elseif ( $fancySig !== false ) {
4538 # Sig. might contain markup; validate this
4539 if ( $this->validateSig( $nickname ) !== false ) {
4540 # Validated; clean up (if needed) and return it
4541 return $this->cleanSig( $nickname, true );
4542 } else {
4543 # Failed to validate; fall back to the default
4544 $nickname = $username;
4545 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4546 }
4547 }
4548
4549 # Make sure nickname doesnt get a sig in a sig
4550 $nickname = self::cleanSigInSig( $nickname );
4551
4552 # If we're still here, make it a link to the user page
4553 $userText = wfEscapeWikiText( $username );
4554 $nickText = wfEscapeWikiText( $nickname );
4555 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4556
4557 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4558 ->title( $this->getTitle() )->text();
4559 }
4560
4561 /**
4562 * Check that the user's signature contains no bad XML
4563 *
4564 * @param string $text
4565 * @return string|bool An expanded string, or false if invalid.
4566 */
4567 public function validateSig( $text ) {
4568 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4569 }
4570
4571 /**
4572 * Clean up signature text
4573 *
4574 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4575 * 2) Substitute all transclusions
4576 *
4577 * @param string $text
4578 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4579 * @return string Signature text
4580 */
4581 public function cleanSig( $text, $parsing = false ) {
4582 if ( !$parsing ) {
4583 global $wgTitle;
4584 $magicScopeVariable = $this->lock();
4585 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4586 }
4587
4588 # Option to disable this feature
4589 if ( !$this->mOptions->getCleanSignatures() ) {
4590 return $text;
4591 }
4592
4593 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4594 # => Move this logic to braceSubstitution()
4595 $substWord = MagicWord::get( 'subst' );
4596 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4597 $substText = '{{' . $substWord->getSynonym( 0 );
4598
4599 $text = preg_replace( $substRegex, $substText, $text );
4600 $text = self::cleanSigInSig( $text );
4601 $dom = $this->preprocessToDom( $text );
4602 $frame = $this->getPreprocessor()->newFrame();
4603 $text = $frame->expand( $dom );
4604
4605 if ( !$parsing ) {
4606 $text = $this->mStripState->unstripBoth( $text );
4607 }
4608
4609 return $text;
4610 }
4611
4612 /**
4613 * Strip 3, 4 or 5 tildes out of signatures.
4614 *
4615 * @param string $text
4616 * @return string Signature text with /~{3,5}/ removed
4617 */
4618 public static function cleanSigInSig( $text ) {
4619 $text = preg_replace( '/~{3,5}/', '', $text );
4620 return $text;
4621 }
4622
4623 /**
4624 * Set up some variables which are usually set up in parse()
4625 * so that an external function can call some class members with confidence
4626 *
4627 * @param Title|null $title
4628 * @param ParserOptions $options
4629 * @param int $outputType
4630 * @param bool $clearState
4631 */
4632 public function startExternalParse( Title $title = null, ParserOptions $options,
4633 $outputType, $clearState = true
4634 ) {
4635 $this->startParse( $title, $options, $outputType, $clearState );
4636 }
4637
4638 /**
4639 * @param Title|null $title
4640 * @param ParserOptions $options
4641 * @param int $outputType
4642 * @param bool $clearState
4643 */
4644 private function startParse( Title $title = null, ParserOptions $options,
4645 $outputType, $clearState = true
4646 ) {
4647 $this->setTitle( $title );
4648 $this->mOptions = $options;
4649 $this->setOutputType( $outputType );
4650 if ( $clearState ) {
4651 $this->clearState();
4652 }
4653 }
4654
4655 /**
4656 * Wrapper for preprocess()
4657 *
4658 * @param string $text The text to preprocess
4659 * @param ParserOptions $options Options
4660 * @param Title|null $title Title object or null to use $wgTitle
4661 * @return string
4662 */
4663 public function transformMsg( $text, $options, $title = null ) {
4664 static $executing = false;
4665
4666 # Guard against infinite recursion
4667 if ( $executing ) {
4668 return $text;
4669 }
4670 $executing = true;
4671
4672 if ( !$title ) {
4673 global $wgTitle;
4674 $title = $wgTitle;
4675 }
4676
4677 $text = $this->preprocess( $text, $title, $options );
4678
4679 $executing = false;
4680 return $text;
4681 }
4682
4683 /**
4684 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4685 * The callback should have the following form:
4686 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4687 *
4688 * Transform and return $text. Use $parser for any required context, e.g. use
4689 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4690 *
4691 * Hooks may return extended information by returning an array, of which the
4692 * first numbered element (index 0) must be the return string, and all other
4693 * entries are extracted into local variables within an internal function
4694 * in the Parser class.
4695 *
4696 * This interface (introduced r61913) appears to be undocumented, but
4697 * 'markerType' is used by some core tag hooks to override which strip
4698 * array their results are placed in. **Use great caution if attempting
4699 * this interface, as it is not documented and injudicious use could smash
4700 * private variables.**
4701 *
4702 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4703 * @param callable $callback The callback function (and object) to use for the tag
4704 * @throws MWException
4705 * @return callable|null The old value of the mTagHooks array associated with the hook
4706 */
4707 public function setHook( $tag, callable $callback ) {
4708 $tag = strtolower( $tag );
4709 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4710 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4711 }
4712 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4713 $this->mTagHooks[$tag] = $callback;
4714 if ( !in_array( $tag, $this->mStripList ) ) {
4715 $this->mStripList[] = $tag;
4716 }
4717
4718 return $oldVal;
4719 }
4720
4721 /**
4722 * As setHook(), but letting the contents be parsed.
4723 *
4724 * Transparent tag hooks are like regular XML-style tag hooks, except they
4725 * operate late in the transformation sequence, on HTML instead of wikitext.
4726 *
4727 * This is probably obsoleted by things dealing with parser frames?
4728 * The only extension currently using it is geoserver.
4729 *
4730 * @since 1.10
4731 * @todo better document or deprecate this
4732 *
4733 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4734 * @param callable $callback The callback function (and object) to use for the tag
4735 * @throws MWException
4736 * @return callable|null The old value of the mTagHooks array associated with the hook
4737 */
4738 public function setTransparentTagHook( $tag, callable $callback ) {
4739 $tag = strtolower( $tag );
4740 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4741 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4742 }
4743 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4744 $this->mTransparentTagHooks[$tag] = $callback;
4745
4746 return $oldVal;
4747 }
4748
4749 /**
4750 * Remove all tag hooks
4751 */
4752 public function clearTagHooks() {
4753 $this->mTagHooks = [];
4754 $this->mFunctionTagHooks = [];
4755 $this->mStripList = $this->mDefaultStripList;
4756 }
4757
4758 /**
4759 * Create a function, e.g. {{sum:1|2|3}}
4760 * The callback function should have the form:
4761 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4762 *
4763 * Or with Parser::SFH_OBJECT_ARGS:
4764 * function myParserFunction( $parser, $frame, $args ) { ... }
4765 *
4766 * The callback may either return the text result of the function, or an array with the text
4767 * in element 0, and a number of flags in the other elements. The names of the flags are
4768 * specified in the keys. Valid flags are:
4769 * found The text returned is valid, stop processing the template. This
4770 * is on by default.
4771 * nowiki Wiki markup in the return value should be escaped
4772 * isHTML The returned text is HTML, armour it against wikitext transformation
4773 *
4774 * @param string $id The magic word ID
4775 * @param callable $callback The callback function (and object) to use
4776 * @param int $flags A combination of the following flags:
4777 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4778 *
4779 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4780 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4781 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4782 * the arguments, and to control the way they are expanded.
4783 *
4784 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4785 * arguments, for instance:
4786 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4787 *
4788 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4789 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4790 * working if/when this is changed.
4791 *
4792 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4793 * expansion.
4794 *
4795 * Please read the documentation in includes/parser/Preprocessor.php for more information
4796 * about the methods available in PPFrame and PPNode.
4797 *
4798 * @throws MWException
4799 * @return string|callable The old callback function for this name, if any
4800 */
4801 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4802 global $wgContLang;
4803
4804 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4805 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4806
4807 # Add to function cache
4808 $mw = MagicWord::get( $id );
4809 if ( !$mw ) {
4810 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4811 }
4812
4813 $synonyms = $mw->getSynonyms();
4814 $sensitive = intval( $mw->isCaseSensitive() );
4815
4816 foreach ( $synonyms as $syn ) {
4817 # Case
4818 if ( !$sensitive ) {
4819 $syn = $wgContLang->lc( $syn );
4820 }
4821 # Add leading hash
4822 if ( !( $flags & self::SFH_NO_HASH ) ) {
4823 $syn = '#' . $syn;
4824 }
4825 # Remove trailing colon
4826 if ( substr( $syn, -1, 1 ) === ':' ) {
4827 $syn = substr( $syn, 0, -1 );
4828 }
4829 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4830 }
4831 return $oldVal;
4832 }
4833
4834 /**
4835 * Get all registered function hook identifiers
4836 *
4837 * @return array
4838 */
4839 public function getFunctionHooks() {
4840 return array_keys( $this->mFunctionHooks );
4841 }
4842
4843 /**
4844 * Create a tag function, e.g. "<test>some stuff</test>".
4845 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4846 * Unlike parser functions, their content is not preprocessed.
4847 * @param string $tag
4848 * @param callable $callback
4849 * @param int $flags
4850 * @throws MWException
4851 * @return null
4852 */
4853 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4854 $tag = strtolower( $tag );
4855 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4856 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4857 }
4858 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4859 $this->mFunctionTagHooks[$tag] : null;
4860 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4861
4862 if ( !in_array( $tag, $this->mStripList ) ) {
4863 $this->mStripList[] = $tag;
4864 }
4865
4866 return $old;
4867 }
4868
4869 /**
4870 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4871 * Placeholders created in Linker::link()
4872 *
4873 * @param string &$text
4874 * @param int $options
4875 */
4876 public function replaceLinkHolders( &$text, $options = 0 ) {
4877 $this->mLinkHolders->replace( $text );
4878 }
4879
4880 /**
4881 * Replace "<!--LINK-->" link placeholders with plain text of links
4882 * (not HTML-formatted).
4883 *
4884 * @param string $text
4885 * @return string
4886 */
4887 public function replaceLinkHoldersText( $text ) {
4888 return $this->mLinkHolders->replaceText( $text );
4889 }
4890
4891 /**
4892 * Renders an image gallery from a text with one line per image.
4893 * text labels may be given by using |-style alternative text. E.g.
4894 * Image:one.jpg|The number "1"
4895 * Image:tree.jpg|A tree
4896 * given as text will return the HTML of a gallery with two images,
4897 * labeled 'The number "1"' and
4898 * 'A tree'.
4899 *
4900 * @param string $text
4901 * @param array $params
4902 * @return string HTML
4903 */
4904 public function renderImageGallery( $text, $params ) {
4905 $mode = false;
4906 if ( isset( $params['mode'] ) ) {
4907 $mode = $params['mode'];
4908 }
4909
4910 try {
4911 $ig = ImageGalleryBase::factory( $mode );
4912 } catch ( Exception $e ) {
4913 // If invalid type set, fallback to default.
4914 $ig = ImageGalleryBase::factory( false );
4915 }
4916
4917 $ig->setContextTitle( $this->mTitle );
4918 $ig->setShowBytes( false );
4919 $ig->setShowDimensions( false );
4920 $ig->setShowFilename( false );
4921 $ig->setParser( $this );
4922 $ig->setHideBadImages();
4923 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4924
4925 if ( isset( $params['showfilename'] ) ) {
4926 $ig->setShowFilename( true );
4927 } else {
4928 $ig->setShowFilename( false );
4929 }
4930 if ( isset( $params['caption'] ) ) {
4931 $caption = $params['caption'];
4932 $caption = htmlspecialchars( $caption );
4933 $caption = $this->replaceInternalLinks( $caption );
4934 $ig->setCaptionHtml( $caption );
4935 }
4936 if ( isset( $params['perrow'] ) ) {
4937 $ig->setPerRow( $params['perrow'] );
4938 }
4939 if ( isset( $params['widths'] ) ) {
4940 $ig->setWidths( $params['widths'] );
4941 }
4942 if ( isset( $params['heights'] ) ) {
4943 $ig->setHeights( $params['heights'] );
4944 }
4945 $ig->setAdditionalOptions( $params );
4946
4947 // Avoid PHP 7.1 warning from passing $this by reference
4948 $parser = $this;
4949 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
4950
4951 $lines = StringUtils::explode( "\n", $text );
4952 foreach ( $lines as $line ) {
4953 # match lines like these:
4954 # Image:someimage.jpg|This is some image
4955 $matches = [];
4956 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4957 # Skip empty lines
4958 if ( count( $matches ) == 0 ) {
4959 continue;
4960 }
4961
4962 if ( strpos( $matches[0], '%' ) !== false ) {
4963 $matches[1] = rawurldecode( $matches[1] );
4964 }
4965 $title = Title::newFromText( $matches[1], NS_FILE );
4966 if ( is_null( $title ) ) {
4967 # Bogus title. Ignore these so we don't bomb out later.
4968 continue;
4969 }
4970
4971 # We need to get what handler the file uses, to figure out parameters.
4972 # Note, a hook can overide the file name, and chose an entirely different
4973 # file (which potentially could be of a different type and have different handler).
4974 $options = [];
4975 $descQuery = false;
4976 Hooks::run( 'BeforeParserFetchFileAndTitle',
4977 [ $this, $title, &$options, &$descQuery ] );
4978 # Don't register it now, as TraditionalImageGallery does that later.
4979 $file = $this->fetchFileNoRegister( $title, $options );
4980 $handler = $file ? $file->getHandler() : false;
4981
4982 $paramMap = [
4983 'img_alt' => 'gallery-internal-alt',
4984 'img_link' => 'gallery-internal-link',
4985 ];
4986 if ( $handler ) {
4987 $paramMap = $paramMap + $handler->getParamMap();
4988 // We don't want people to specify per-image widths.
4989 // Additionally the width parameter would need special casing anyhow.
4990 unset( $paramMap['img_width'] );
4991 }
4992
4993 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4994
4995 $label = '';
4996 $alt = '';
4997 $link = '';
4998 $handlerOptions = [];
4999 if ( isset( $matches[3] ) ) {
5000 // look for an |alt= definition while trying not to break existing
5001 // captions with multiple pipes (|) in it, until a more sensible grammar
5002 // is defined for images in galleries
5003
5004 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5005 // splitting on '|' is a bit odd, and different from makeImage.
5006 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5007 // Protect LanguageConverter markup
5008 $parameterMatches = StringUtils::delimiterExplode(
5009 '-{', '}-', '|', $matches[3], true /* nested */
5010 );
5011
5012 foreach ( $parameterMatches as $parameterMatch ) {
5013 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5014 if ( $magicName ) {
5015 $paramName = $paramMap[$magicName];
5016
5017 switch ( $paramName ) {
5018 case 'gallery-internal-alt':
5019 $alt = $this->stripAltText( $match, false );
5020 break;
5021 case 'gallery-internal-link':
5022 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5023 $chars = self::EXT_LINK_URL_CLASS;
5024 $addr = self::EXT_LINK_ADDR;
5025 $prots = $this->mUrlProtocols;
5026 // check to see if link matches an absolute url, if not then it must be a wiki link.
5027 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5028 // Result of LanguageConverter::markNoConversion
5029 // invoked on an external link.
5030 $linkValue = substr( $linkValue, 4, -2 );
5031 }
5032 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5033 $link = $linkValue;
5034 $this->mOutput->addExternalLink( $link );
5035 } else {
5036 $localLinkTitle = Title::newFromText( $linkValue );
5037 if ( $localLinkTitle !== null ) {
5038 $this->mOutput->addLink( $localLinkTitle );
5039 $link = $localLinkTitle->getLinkURL();
5040 }
5041 }
5042 break;
5043 default:
5044 // Must be a handler specific parameter.
5045 if ( $handler->validateParam( $paramName, $match ) ) {
5046 $handlerOptions[$paramName] = $match;
5047 } else {
5048 // Guess not, consider it as caption.
5049 wfDebug( "$parameterMatch failed parameter validation\n" );
5050 $label = '|' . $parameterMatch;
5051 }
5052 }
5053
5054 } else {
5055 // Last pipe wins.
5056 $label = '|' . $parameterMatch;
5057 }
5058 }
5059 // Remove the pipe.
5060 $label = substr( $label, 1 );
5061 }
5062
5063 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5064 }
5065 $html = $ig->toHTML();
5066 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5067 return $html;
5068 }
5069
5070 /**
5071 * @param MediaHandler $handler
5072 * @return array
5073 */
5074 public function getImageParams( $handler ) {
5075 if ( $handler ) {
5076 $handlerClass = get_class( $handler );
5077 } else {
5078 $handlerClass = '';
5079 }
5080 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5081 # Initialise static lists
5082 static $internalParamNames = [
5083 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5084 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5085 'bottom', 'text-bottom' ],
5086 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5087 'upright', 'border', 'link', 'alt', 'class' ],
5088 ];
5089 static $internalParamMap;
5090 if ( !$internalParamMap ) {
5091 $internalParamMap = [];
5092 foreach ( $internalParamNames as $type => $names ) {
5093 foreach ( $names as $name ) {
5094 // For grep: img_left, img_right, img_center, img_none,
5095 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5096 // img_bottom, img_text_bottom,
5097 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5098 // img_border, img_link, img_alt, img_class
5099 $magicName = str_replace( '-', '_', "img_$name" );
5100 $internalParamMap[$magicName] = [ $type, $name ];
5101 }
5102 }
5103 }
5104
5105 # Add handler params
5106 $paramMap = $internalParamMap;
5107 if ( $handler ) {
5108 $handlerParamMap = $handler->getParamMap();
5109 foreach ( $handlerParamMap as $magic => $paramName ) {
5110 $paramMap[$magic] = [ 'handler', $paramName ];
5111 }
5112 }
5113 $this->mImageParams[$handlerClass] = $paramMap;
5114 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5115 }
5116 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5117 }
5118
5119 /**
5120 * Parse image options text and use it to make an image
5121 *
5122 * @param Title $title
5123 * @param string $options
5124 * @param LinkHolderArray|bool $holders
5125 * @return string HTML
5126 */
5127 public function makeImage( $title, $options, $holders = false ) {
5128 # Check if the options text is of the form "options|alt text"
5129 # Options are:
5130 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5131 # * left no resizing, just left align. label is used for alt= only
5132 # * right same, but right aligned
5133 # * none same, but not aligned
5134 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5135 # * center center the image
5136 # * frame Keep original image size, no magnify-button.
5137 # * framed Same as "frame"
5138 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5139 # * upright reduce width for upright images, rounded to full __0 px
5140 # * border draw a 1px border around the image
5141 # * alt Text for HTML alt attribute (defaults to empty)
5142 # * class Set a class for img node
5143 # * link Set the target of the image link. Can be external, interwiki, or local
5144 # vertical-align values (no % or length right now):
5145 # * baseline
5146 # * sub
5147 # * super
5148 # * top
5149 # * text-top
5150 # * middle
5151 # * bottom
5152 # * text-bottom
5153
5154 # Protect LanguageConverter markup when splitting into parts
5155 $parts = StringUtils::delimiterExplode(
5156 '-{', '}-', '|', $options, true /* allow nesting */
5157 );
5158
5159 # Give extensions a chance to select the file revision for us
5160 $options = [];
5161 $descQuery = false;
5162 Hooks::run( 'BeforeParserFetchFileAndTitle',
5163 [ $this, $title, &$options, &$descQuery ] );
5164 # Fetch and register the file (file title may be different via hooks)
5165 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5166
5167 # Get parameter map
5168 $handler = $file ? $file->getHandler() : false;
5169
5170 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5171
5172 if ( !$file ) {
5173 $this->addTrackingCategory( 'broken-file-category' );
5174 }
5175
5176 # Process the input parameters
5177 $caption = '';
5178 $params = [ 'frame' => [], 'handler' => [],
5179 'horizAlign' => [], 'vertAlign' => [] ];
5180 $seenformat = false;
5181 foreach ( $parts as $part ) {
5182 $part = trim( $part );
5183 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5184 $validated = false;
5185 if ( isset( $paramMap[$magicName] ) ) {
5186 list( $type, $paramName ) = $paramMap[$magicName];
5187
5188 # Special case; width and height come in one variable together
5189 if ( $type === 'handler' && $paramName === 'width' ) {
5190 $parsedWidthParam = $this->parseWidthParam( $value );
5191 if ( isset( $parsedWidthParam['width'] ) ) {
5192 $width = $parsedWidthParam['width'];
5193 if ( $handler->validateParam( 'width', $width ) ) {
5194 $params[$type]['width'] = $width;
5195 $validated = true;
5196 }
5197 }
5198 if ( isset( $parsedWidthParam['height'] ) ) {
5199 $height = $parsedWidthParam['height'];
5200 if ( $handler->validateParam( 'height', $height ) ) {
5201 $params[$type]['height'] = $height;
5202 $validated = true;
5203 }
5204 }
5205 # else no validation -- T15436
5206 } else {
5207 if ( $type === 'handler' ) {
5208 # Validate handler parameter
5209 $validated = $handler->validateParam( $paramName, $value );
5210 } else {
5211 # Validate internal parameters
5212 switch ( $paramName ) {
5213 case 'manualthumb':
5214 case 'alt':
5215 case 'class':
5216 # @todo FIXME: Possibly check validity here for
5217 # manualthumb? downstream behavior seems odd with
5218 # missing manual thumbs.
5219 $validated = true;
5220 $value = $this->stripAltText( $value, $holders );
5221 break;
5222 case 'link':
5223 $chars = self::EXT_LINK_URL_CLASS;
5224 $addr = self::EXT_LINK_ADDR;
5225 $prots = $this->mUrlProtocols;
5226 if ( $value === '' ) {
5227 $paramName = 'no-link';
5228 $value = true;
5229 $validated = true;
5230 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5231 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5232 $paramName = 'link-url';
5233 $this->mOutput->addExternalLink( $value );
5234 if ( $this->mOptions->getExternalLinkTarget() ) {
5235 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5236 }
5237 $validated = true;
5238 }
5239 } else {
5240 $linkTitle = Title::newFromText( $value );
5241 if ( $linkTitle ) {
5242 $paramName = 'link-title';
5243 $value = $linkTitle;
5244 $this->mOutput->addLink( $linkTitle );
5245 $validated = true;
5246 }
5247 }
5248 break;
5249 case 'frameless':
5250 case 'framed':
5251 case 'thumbnail':
5252 // use first appearing option, discard others.
5253 $validated = !$seenformat;
5254 $seenformat = true;
5255 break;
5256 default:
5257 # Most other things appear to be empty or numeric...
5258 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5259 }
5260 }
5261
5262 if ( $validated ) {
5263 $params[$type][$paramName] = $value;
5264 }
5265 }
5266 }
5267 if ( !$validated ) {
5268 $caption = $part;
5269 }
5270 }
5271
5272 # Process alignment parameters
5273 if ( $params['horizAlign'] ) {
5274 $params['frame']['align'] = key( $params['horizAlign'] );
5275 }
5276 if ( $params['vertAlign'] ) {
5277 $params['frame']['valign'] = key( $params['vertAlign'] );
5278 }
5279
5280 $params['frame']['caption'] = $caption;
5281
5282 # Will the image be presented in a frame, with the caption below?
5283 $imageIsFramed = isset( $params['frame']['frame'] )
5284 || isset( $params['frame']['framed'] )
5285 || isset( $params['frame']['thumbnail'] )
5286 || isset( $params['frame']['manualthumb'] );
5287
5288 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5289 # came to also set the caption, ordinary text after the image -- which
5290 # makes no sense, because that just repeats the text multiple times in
5291 # screen readers. It *also* came to set the title attribute.
5292 # Now that we have an alt attribute, we should not set the alt text to
5293 # equal the caption: that's worse than useless, it just repeats the
5294 # text. This is the framed/thumbnail case. If there's no caption, we
5295 # use the unnamed parameter for alt text as well, just for the time be-
5296 # ing, if the unnamed param is set and the alt param is not.
5297 # For the future, we need to figure out if we want to tweak this more,
5298 # e.g., introducing a title= parameter for the title; ignoring the un-
5299 # named parameter entirely for images without a caption; adding an ex-
5300 # plicit caption= parameter and preserving the old magic unnamed para-
5301 # meter for BC; ...
5302 if ( $imageIsFramed ) { # Framed image
5303 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5304 # No caption or alt text, add the filename as the alt text so
5305 # that screen readers at least get some description of the image
5306 $params['frame']['alt'] = $title->getText();
5307 }
5308 # Do not set $params['frame']['title'] because tooltips don't make sense
5309 # for framed images
5310 } else { # Inline image
5311 if ( !isset( $params['frame']['alt'] ) ) {
5312 # No alt text, use the "caption" for the alt text
5313 if ( $caption !== '' ) {
5314 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5315 } else {
5316 # No caption, fall back to using the filename for the
5317 # alt text
5318 $params['frame']['alt'] = $title->getText();
5319 }
5320 }
5321 # Use the "caption" for the tooltip text
5322 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5323 }
5324
5325 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5326
5327 # Linker does the rest
5328 $time = isset( $options['time'] ) ? $options['time'] : false;
5329 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5330 $time, $descQuery, $this->mOptions->getThumbSize() );
5331
5332 # Give the handler a chance to modify the parser object
5333 if ( $handler ) {
5334 $handler->parserTransformHook( $this, $file );
5335 }
5336
5337 return $ret;
5338 }
5339
5340 /**
5341 * @param string $caption
5342 * @param LinkHolderArray|bool $holders
5343 * @return mixed|string
5344 */
5345 protected function stripAltText( $caption, $holders ) {
5346 # Strip bad stuff out of the title (tooltip). We can't just use
5347 # replaceLinkHoldersText() here, because if this function is called
5348 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5349 if ( $holders ) {
5350 $tooltip = $holders->replaceText( $caption );
5351 } else {
5352 $tooltip = $this->replaceLinkHoldersText( $caption );
5353 }
5354
5355 # make sure there are no placeholders in thumbnail attributes
5356 # that are later expanded to html- so expand them now and
5357 # remove the tags
5358 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5359 $tooltip = Sanitizer::stripAllTags( $tooltip );
5360
5361 return $tooltip;
5362 }
5363
5364 /**
5365 * Set a flag in the output object indicating that the content is dynamic and
5366 * shouldn't be cached.
5367 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5368 */
5369 public function disableCache() {
5370 wfDebug( "Parser output marked as uncacheable.\n" );
5371 if ( !$this->mOutput ) {
5372 throw new MWException( __METHOD__ .
5373 " can only be called when actually parsing something" );
5374 }
5375 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5376 }
5377
5378 /**
5379 * Callback from the Sanitizer for expanding items found in HTML attribute
5380 * values, so they can be safely tested and escaped.
5381 *
5382 * @param string &$text
5383 * @param bool|PPFrame $frame
5384 * @return string
5385 */
5386 public function attributeStripCallback( &$text, $frame = false ) {
5387 $text = $this->replaceVariables( $text, $frame );
5388 $text = $this->mStripState->unstripBoth( $text );
5389 return $text;
5390 }
5391
5392 /**
5393 * Accessor
5394 *
5395 * @return array
5396 */
5397 public function getTags() {
5398 return array_merge(
5399 array_keys( $this->mTransparentTagHooks ),
5400 array_keys( $this->mTagHooks ),
5401 array_keys( $this->mFunctionTagHooks )
5402 );
5403 }
5404
5405 /**
5406 * Replace transparent tags in $text with the values given by the callbacks.
5407 *
5408 * Transparent tag hooks are like regular XML-style tag hooks, except they
5409 * operate late in the transformation sequence, on HTML instead of wikitext.
5410 *
5411 * @param string $text
5412 *
5413 * @return string
5414 */
5415 public function replaceTransparentTags( $text ) {
5416 $matches = [];
5417 $elements = array_keys( $this->mTransparentTagHooks );
5418 $text = self::extractTagsAndParams( $elements, $text, $matches );
5419 $replacements = [];
5420
5421 foreach ( $matches as $marker => $data ) {
5422 list( $element, $content, $params, $tag ) = $data;
5423 $tagName = strtolower( $element );
5424 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5425 $output = call_user_func_array(
5426 $this->mTransparentTagHooks[$tagName],
5427 [ $content, $params, $this ]
5428 );
5429 } else {
5430 $output = $tag;
5431 }
5432 $replacements[$marker] = $output;
5433 }
5434 return strtr( $text, $replacements );
5435 }
5436
5437 /**
5438 * Break wikitext input into sections, and either pull or replace
5439 * some particular section's text.
5440 *
5441 * External callers should use the getSection and replaceSection methods.
5442 *
5443 * @param string $text Page wikitext
5444 * @param string|int $sectionId A section identifier string of the form:
5445 * "<flag1> - <flag2> - ... - <section number>"
5446 *
5447 * Currently the only recognised flag is "T", which means the target section number
5448 * was derived during a template inclusion parse, in other words this is a template
5449 * section edit link. If no flags are given, it was an ordinary section edit link.
5450 * This flag is required to avoid a section numbering mismatch when a section is
5451 * enclosed by "<includeonly>" (T8563).
5452 *
5453 * The section number 0 pulls the text before the first heading; other numbers will
5454 * pull the given section along with its lower-level subsections. If the section is
5455 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5456 *
5457 * Section 0 is always considered to exist, even if it only contains the empty
5458 * string. If $text is the empty string and section 0 is replaced, $newText is
5459 * returned.
5460 *
5461 * @param string $mode One of "get" or "replace"
5462 * @param string $newText Replacement text for section data.
5463 * @return string For "get", the extracted section text.
5464 * for "replace", the whole page with the section replaced.
5465 */
5466 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5467 global $wgTitle; # not generally used but removes an ugly failure mode
5468
5469 $magicScopeVariable = $this->lock();
5470 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5471 $outText = '';
5472 $frame = $this->getPreprocessor()->newFrame();
5473
5474 # Process section extraction flags
5475 $flags = 0;
5476 $sectionParts = explode( '-', $sectionId );
5477 $sectionIndex = array_pop( $sectionParts );
5478 foreach ( $sectionParts as $part ) {
5479 if ( $part === 'T' ) {
5480 $flags |= self::PTD_FOR_INCLUSION;
5481 }
5482 }
5483
5484 # Check for empty input
5485 if ( strval( $text ) === '' ) {
5486 # Only sections 0 and T-0 exist in an empty document
5487 if ( $sectionIndex == 0 ) {
5488 if ( $mode === 'get' ) {
5489 return '';
5490 } else {
5491 return $newText;
5492 }
5493 } else {
5494 if ( $mode === 'get' ) {
5495 return $newText;
5496 } else {
5497 return $text;
5498 }
5499 }
5500 }
5501
5502 # Preprocess the text
5503 $root = $this->preprocessToDom( $text, $flags );
5504
5505 # <h> nodes indicate section breaks
5506 # They can only occur at the top level, so we can find them by iterating the root's children
5507 $node = $root->getFirstChild();
5508
5509 # Find the target section
5510 if ( $sectionIndex == 0 ) {
5511 # Section zero doesn't nest, level=big
5512 $targetLevel = 1000;
5513 } else {
5514 while ( $node ) {
5515 if ( $node->getName() === 'h' ) {
5516 $bits = $node->splitHeading();
5517 if ( $bits['i'] == $sectionIndex ) {
5518 $targetLevel = $bits['level'];
5519 break;
5520 }
5521 }
5522 if ( $mode === 'replace' ) {
5523 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5524 }
5525 $node = $node->getNextSibling();
5526 }
5527 }
5528
5529 if ( !$node ) {
5530 # Not found
5531 if ( $mode === 'get' ) {
5532 return $newText;
5533 } else {
5534 return $text;
5535 }
5536 }
5537
5538 # Find the end of the section, including nested sections
5539 do {
5540 if ( $node->getName() === 'h' ) {
5541 $bits = $node->splitHeading();
5542 $curLevel = $bits['level'];
5543 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5544 break;
5545 }
5546 }
5547 if ( $mode === 'get' ) {
5548 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5549 }
5550 $node = $node->getNextSibling();
5551 } while ( $node );
5552
5553 # Write out the remainder (in replace mode only)
5554 if ( $mode === 'replace' ) {
5555 # Output the replacement text
5556 # Add two newlines on -- trailing whitespace in $newText is conventionally
5557 # stripped by the editor, so we need both newlines to restore the paragraph gap
5558 # Only add trailing whitespace if there is newText
5559 if ( $newText != "" ) {
5560 $outText .= $newText . "\n\n";
5561 }
5562
5563 while ( $node ) {
5564 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5565 $node = $node->getNextSibling();
5566 }
5567 }
5568
5569 if ( is_string( $outText ) ) {
5570 # Re-insert stripped tags
5571 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5572 }
5573
5574 return $outText;
5575 }
5576
5577 /**
5578 * This function returns the text of a section, specified by a number ($section).
5579 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5580 * the first section before any such heading (section 0).
5581 *
5582 * If a section contains subsections, these are also returned.
5583 *
5584 * @param string $text Text to look in
5585 * @param string|int $sectionId Section identifier as a number or string
5586 * (e.g. 0, 1 or 'T-1').
5587 * @param string $defaultText Default to return if section is not found
5588 *
5589 * @return string Text of the requested section
5590 */
5591 public function getSection( $text, $sectionId, $defaultText = '' ) {
5592 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5593 }
5594
5595 /**
5596 * This function returns $oldtext after the content of the section
5597 * specified by $section has been replaced with $text. If the target
5598 * section does not exist, $oldtext is returned unchanged.
5599 *
5600 * @param string $oldText Former text of the article
5601 * @param string|int $sectionId Section identifier as a number or string
5602 * (e.g. 0, 1 or 'T-1').
5603 * @param string $newText Replacing text
5604 *
5605 * @return string Modified text
5606 */
5607 public function replaceSection( $oldText, $sectionId, $newText ) {
5608 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5609 }
5610
5611 /**
5612 * Get the ID of the revision we are parsing
5613 *
5614 * @return int|null
5615 */
5616 public function getRevisionId() {
5617 return $this->mRevisionId;
5618 }
5619
5620 /**
5621 * Get the revision object for $this->mRevisionId
5622 *
5623 * @return Revision|null Either a Revision object or null
5624 * @since 1.23 (public since 1.23)
5625 */
5626 public function getRevisionObject() {
5627 if ( !is_null( $this->mRevisionObject ) ) {
5628 return $this->mRevisionObject;
5629 }
5630 if ( is_null( $this->mRevisionId ) ) {
5631 return null;
5632 }
5633
5634 $rev = call_user_func(
5635 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5636 );
5637
5638 # If the parse is for a new revision, then the callback should have
5639 # already been set to force the object and should match mRevisionId.
5640 # If not, try to fetch by mRevisionId for sanity.
5641 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5642 $rev = Revision::newFromId( $this->mRevisionId );
5643 }
5644
5645 $this->mRevisionObject = $rev;
5646
5647 return $this->mRevisionObject;
5648 }
5649
5650 /**
5651 * Get the timestamp associated with the current revision, adjusted for
5652 * the default server-local timestamp
5653 * @return string
5654 */
5655 public function getRevisionTimestamp() {
5656 if ( is_null( $this->mRevisionTimestamp ) ) {
5657 global $wgContLang;
5658
5659 $revObject = $this->getRevisionObject();
5660 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5661
5662 # The cryptic '' timezone parameter tells to use the site-default
5663 # timezone offset instead of the user settings.
5664 # Since this value will be saved into the parser cache, served
5665 # to other users, and potentially even used inside links and such,
5666 # it needs to be consistent for all visitors.
5667 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5668
5669 }
5670 return $this->mRevisionTimestamp;
5671 }
5672
5673 /**
5674 * Get the name of the user that edited the last revision
5675 *
5676 * @return string User name
5677 */
5678 public function getRevisionUser() {
5679 if ( is_null( $this->mRevisionUser ) ) {
5680 $revObject = $this->getRevisionObject();
5681
5682 # if this template is subst: the revision id will be blank,
5683 # so just use the current user's name
5684 if ( $revObject ) {
5685 $this->mRevisionUser = $revObject->getUserText();
5686 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5687 $this->mRevisionUser = $this->getUser()->getName();
5688 }
5689 }
5690 return $this->mRevisionUser;
5691 }
5692
5693 /**
5694 * Get the size of the revision
5695 *
5696 * @return int|null Revision size
5697 */
5698 public function getRevisionSize() {
5699 if ( is_null( $this->mRevisionSize ) ) {
5700 $revObject = $this->getRevisionObject();
5701
5702 # if this variable is subst: the revision id will be blank,
5703 # so just use the parser input size, because the own substituation
5704 # will change the size.
5705 if ( $revObject ) {
5706 $this->mRevisionSize = $revObject->getSize();
5707 } else {
5708 $this->mRevisionSize = $this->mInputSize;
5709 }
5710 }
5711 return $this->mRevisionSize;
5712 }
5713
5714 /**
5715 * Mutator for $mDefaultSort
5716 *
5717 * @param string $sort New value
5718 */
5719 public function setDefaultSort( $sort ) {
5720 $this->mDefaultSort = $sort;
5721 $this->mOutput->setProperty( 'defaultsort', $sort );
5722 }
5723
5724 /**
5725 * Accessor for $mDefaultSort
5726 * Will use the empty string if none is set.
5727 *
5728 * This value is treated as a prefix, so the
5729 * empty string is equivalent to sorting by
5730 * page name.
5731 *
5732 * @return string
5733 */
5734 public function getDefaultSort() {
5735 if ( $this->mDefaultSort !== false ) {
5736 return $this->mDefaultSort;
5737 } else {
5738 return '';
5739 }
5740 }
5741
5742 /**
5743 * Accessor for $mDefaultSort
5744 * Unlike getDefaultSort(), will return false if none is set
5745 *
5746 * @return string|bool
5747 */
5748 public function getCustomDefaultSort() {
5749 return $this->mDefaultSort;
5750 }
5751
5752 /**
5753 * Try to guess the section anchor name based on a wikitext fragment
5754 * presumably extracted from a heading, for example "Header" from
5755 * "== Header ==".
5756 *
5757 * @param string $text
5758 *
5759 * @return string
5760 */
5761 public function guessSectionNameFromWikiText( $text ) {
5762 # Strip out wikitext links(they break the anchor)
5763 $text = $this->stripSectionName( $text );
5764 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5765 return '#' . Sanitizer::escapeIdForLink( $text );
5766 }
5767
5768 /**
5769 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5770 * instead, if possible. For use in redirects, since various versions
5771 * of Microsoft browsers interpret Location: headers as something other
5772 * than UTF-8, resulting in breakage.
5773 *
5774 * @param string $text The section name
5775 * @return string An anchor
5776 */
5777 public function guessLegacySectionNameFromWikiText( $text ) {
5778 global $wgFragmentMode;
5779
5780 # Strip out wikitext links(they break the anchor)
5781 $text = $this->stripSectionName( $text );
5782 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5783
5784 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5785 // ForAttribute() and ForLink() are the same for legacy encoding
5786 $id = Sanitizer::escapeIdForAttribute( $text, Sanitizer::ID_FALLBACK );
5787 } else {
5788 $id = Sanitizer::escapeIdForLink( $text );
5789 }
5790
5791 return "#$id";
5792 }
5793
5794 /**
5795 * Strips a text string of wikitext for use in a section anchor
5796 *
5797 * Accepts a text string and then removes all wikitext from the
5798 * string and leaves only the resultant text (i.e. the result of
5799 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5800 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5801 * to create valid section anchors by mimicing the output of the
5802 * parser when headings are parsed.
5803 *
5804 * @param string $text Text string to be stripped of wikitext
5805 * for use in a Section anchor
5806 * @return string Filtered text string
5807 */
5808 public function stripSectionName( $text ) {
5809 # Strip internal link markup
5810 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5811 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5812
5813 # Strip external link markup
5814 # @todo FIXME: Not tolerant to blank link text
5815 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5816 # on how many empty links there are on the page - need to figure that out.
5817 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5818
5819 # Parse wikitext quotes (italics & bold)
5820 $text = $this->doQuotes( $text );
5821
5822 # Strip HTML tags
5823 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5824 return $text;
5825 }
5826
5827 /**
5828 * strip/replaceVariables/unstrip for preprocessor regression testing
5829 *
5830 * @param string $text
5831 * @param Title $title
5832 * @param ParserOptions $options
5833 * @param int $outputType
5834 *
5835 * @return string
5836 */
5837 public function testSrvus( $text, Title $title, ParserOptions $options,
5838 $outputType = self::OT_HTML
5839 ) {
5840 $magicScopeVariable = $this->lock();
5841 $this->startParse( $title, $options, $outputType, true );
5842
5843 $text = $this->replaceVariables( $text );
5844 $text = $this->mStripState->unstripBoth( $text );
5845 $text = Sanitizer::removeHTMLtags( $text );
5846 return $text;
5847 }
5848
5849 /**
5850 * @param string $text
5851 * @param Title $title
5852 * @param ParserOptions $options
5853 * @return string
5854 */
5855 public function testPst( $text, Title $title, ParserOptions $options ) {
5856 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5857 }
5858
5859 /**
5860 * @param string $text
5861 * @param Title $title
5862 * @param ParserOptions $options
5863 * @return string
5864 */
5865 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5866 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5867 }
5868
5869 /**
5870 * Call a callback function on all regions of the given text that are not
5871 * inside strip markers, and replace those regions with the return value
5872 * of the callback. For example, with input:
5873 *
5874 * aaa<MARKER>bbb
5875 *
5876 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5877 * two strings will be replaced with the value returned by the callback in
5878 * each case.
5879 *
5880 * @param string $s
5881 * @param callable $callback
5882 *
5883 * @return string
5884 */
5885 public function markerSkipCallback( $s, $callback ) {
5886 $i = 0;
5887 $out = '';
5888 while ( $i < strlen( $s ) ) {
5889 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5890 if ( $markerStart === false ) {
5891 $out .= call_user_func( $callback, substr( $s, $i ) );
5892 break;
5893 } else {
5894 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5895 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5896 if ( $markerEnd === false ) {
5897 $out .= substr( $s, $markerStart );
5898 break;
5899 } else {
5900 $markerEnd += strlen( self::MARKER_SUFFIX );
5901 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5902 $i = $markerEnd;
5903 }
5904 }
5905 }
5906 return $out;
5907 }
5908
5909 /**
5910 * Remove any strip markers found in the given text.
5911 *
5912 * @param string $text Input string
5913 * @return string
5914 */
5915 public function killMarkers( $text ) {
5916 return $this->mStripState->killMarkers( $text );
5917 }
5918
5919 /**
5920 * Save the parser state required to convert the given half-parsed text to
5921 * HTML. "Half-parsed" in this context means the output of
5922 * recursiveTagParse() or internalParse(). This output has strip markers
5923 * from replaceVariables (extensionSubstitution() etc.), and link
5924 * placeholders from replaceLinkHolders().
5925 *
5926 * Returns an array which can be serialized and stored persistently. This
5927 * array can later be loaded into another parser instance with
5928 * unserializeHalfParsedText(). The text can then be safely incorporated into
5929 * the return value of a parser hook.
5930 *
5931 * @param string $text
5932 *
5933 * @return array
5934 */
5935 public function serializeHalfParsedText( $text ) {
5936 $data = [
5937 'text' => $text,
5938 'version' => self::HALF_PARSED_VERSION,
5939 'stripState' => $this->mStripState->getSubState( $text ),
5940 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5941 ];
5942 return $data;
5943 }
5944
5945 /**
5946 * Load the parser state given in the $data array, which is assumed to
5947 * have been generated by serializeHalfParsedText(). The text contents is
5948 * extracted from the array, and its markers are transformed into markers
5949 * appropriate for the current Parser instance. This transformed text is
5950 * returned, and can be safely included in the return value of a parser
5951 * hook.
5952 *
5953 * If the $data array has been stored persistently, the caller should first
5954 * check whether it is still valid, by calling isValidHalfParsedText().
5955 *
5956 * @param array $data Serialized data
5957 * @throws MWException
5958 * @return string
5959 */
5960 public function unserializeHalfParsedText( $data ) {
5961 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5962 throw new MWException( __METHOD__ . ': invalid version' );
5963 }
5964
5965 # First, extract the strip state.
5966 $texts = [ $data['text'] ];
5967 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5968
5969 # Now renumber links
5970 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5971
5972 # Should be good to go.
5973 return $texts[0];
5974 }
5975
5976 /**
5977 * Returns true if the given array, presumed to be generated by
5978 * serializeHalfParsedText(), is compatible with the current version of the
5979 * parser.
5980 *
5981 * @param array $data
5982 *
5983 * @return bool
5984 */
5985 public function isValidHalfParsedText( $data ) {
5986 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5987 }
5988
5989 /**
5990 * Parsed a width param of imagelink like 300px or 200x300px
5991 *
5992 * @param string $value
5993 *
5994 * @return array
5995 * @since 1.20
5996 */
5997 public function parseWidthParam( $value ) {
5998 $parsedWidthParam = [];
5999 if ( $value === '' ) {
6000 return $parsedWidthParam;
6001 }
6002 $m = [];
6003 # (T15500) In both cases (width/height and width only),
6004 # permit trailing "px" for backward compatibility.
6005 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6006 $width = intval( $m[1] );
6007 $height = intval( $m[2] );
6008 $parsedWidthParam['width'] = $width;
6009 $parsedWidthParam['height'] = $height;
6010 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6011 $width = intval( $value );
6012 $parsedWidthParam['width'] = $width;
6013 }
6014 return $parsedWidthParam;
6015 }
6016
6017 /**
6018 * Lock the current instance of the parser.
6019 *
6020 * This is meant to stop someone from calling the parser
6021 * recursively and messing up all the strip state.
6022 *
6023 * @throws MWException If parser is in a parse
6024 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6025 */
6026 protected function lock() {
6027 if ( $this->mInParse ) {
6028 throw new MWException( "Parser state cleared while parsing. "
6029 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6030 }
6031
6032 // Save the backtrace when locking, so that if some code tries locking again,
6033 // we can print the lock owner's backtrace for easier debugging
6034 $e = new Exception;
6035 $this->mInParse = $e->getTraceAsString();
6036
6037 $recursiveCheck = new ScopedCallback( function () {
6038 $this->mInParse = false;
6039 } );
6040
6041 return $recursiveCheck;
6042 }
6043
6044 /**
6045 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6046 *
6047 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6048 * or if there is more than one <p/> tag in the input HTML.
6049 *
6050 * @param string $html
6051 * @return string
6052 * @since 1.24
6053 */
6054 public static function stripOuterParagraph( $html ) {
6055 $m = [];
6056 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6057 if ( strpos( $m[1], '</p>' ) === false ) {
6058 $html = $m[1];
6059 }
6060 }
6061
6062 return $html;
6063 }
6064
6065 /**
6066 * Return this parser if it is not doing anything, otherwise
6067 * get a fresh parser. You can use this method by doing
6068 * $myParser = $wgParser->getFreshParser(), or more simply
6069 * $wgParser->getFreshParser()->parse( ... );
6070 * if you're unsure if $wgParser is safe to use.
6071 *
6072 * @since 1.24
6073 * @return Parser A parser object that is not parsing anything
6074 */
6075 public function getFreshParser() {
6076 global $wgParserConf;
6077 if ( $this->mInParse ) {
6078 return new $wgParserConf['class']( $wgParserConf );
6079 } else {
6080 return $this;
6081 }
6082 }
6083
6084 /**
6085 * Set's up the PHP implementation of OOUI for use in this request
6086 * and instructs OutputPage to enable OOUI for itself.
6087 *
6088 * @since 1.26
6089 */
6090 public function enableOOUI() {
6091 OutputPage::setupOOUI();
6092 $this->mOutput->setEnableOOUI( true );
6093 }
6094 }
Wiklou, le Wiki du Wiklou