dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Fix link prefix/suffixes around Category and Language links (take 2)."
[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 /**
2300 * Strip the whitespace interwiki links produce, see T10897
2301 */
2302 $s = rtrim( $s . $prefix );
2303 // Special case: strip newlines when only thing between
2304 // this link and next are newlines
2305 $s .= trim( $trail, "\n" ) === '' ? '' : $trail;
2306 continue;
2307 }
2308
2309 if ( $ns == NS_FILE ) {
2310 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2311 if ( $wasblank ) {
2312 # if no parameters were passed, $text
2313 # becomes something like "File:Foo.png",
2314 # which we don't want to pass on to the
2315 # image generator
2316 $text = '';
2317 } else {
2318 # recursively parse links inside the image caption
2319 # actually, this will parse them in any other parameters, too,
2320 # but it might be hard to fix that, and it doesn't matter ATM
2321 $text = $this->replaceExternalLinks( $text );
2322 $holders->merge( $this->replaceInternalLinks2( $text ) );
2323 }
2324 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2325 $s .= $prefix . $this->armorLinks(
2326 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2327 continue;
2328 }
2329 } elseif ( $ns == NS_CATEGORY ) {
2330 /**
2331 * Strip the whitespace Category links produce, see T2087
2332 */
2333 $s = rtrim( $s . $prefix ); # T2087, T87753
2334 // Special case: strip newlines when only thing between
2335 // this link and next are newlines
2336 $s .= trim( $trail, "\n" ) === '' ? '' : $trail;
2337
2338 if ( $wasblank ) {
2339 $sortkey = $this->getDefaultSort();
2340 } else {
2341 $sortkey = $text;
2342 }
2343 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2344 $sortkey = str_replace( "\n", '', $sortkey );
2345 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2346 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2347
2348 continue;
2349 }
2350 }
2351
2352 # Self-link checking. For some languages, variants of the title are checked in
2353 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2354 # for linking to a different variant.
2355 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2356 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2357 continue;
2358 }
2359
2360 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2361 # @todo FIXME: Should do batch file existence checks, see comment below
2362 if ( $ns == NS_MEDIA ) {
2363 # Give extensions a chance to select the file revision for us
2364 $options = [];
2365 $descQuery = false;
2366 Hooks::run( 'BeforeParserFetchFileAndTitle',
2367 [ $this, $nt, &$options, &$descQuery ] );
2368 # Fetch and register the file (file title may be different via hooks)
2369 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2370 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2371 $s .= $prefix . $this->armorLinks(
2372 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2373 continue;
2374 }
2375
2376 # Some titles, such as valid special pages or files in foreign repos, should
2377 # be shown as bluelinks even though they're not included in the page table
2378 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2379 # batch file existence checks for NS_FILE and NS_MEDIA
2380 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2381 $this->mOutput->addLink( $nt );
2382 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2383 } else {
2384 # Links will be added to the output link list after checking
2385 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2386 }
2387 }
2388 return $holders;
2389 }
2390
2391 /**
2392 * Render a forced-blue link inline; protect against double expansion of
2393 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2394 * Since this little disaster has to split off the trail text to avoid
2395 * breaking URLs in the following text without breaking trails on the
2396 * wiki links, it's been made into a horrible function.
2397 *
2398 * @param Title $nt
2399 * @param string $text
2400 * @param string $trail
2401 * @param string $prefix
2402 * @return string HTML-wikitext mix oh yuck
2403 */
2404 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2405 list( $inside, $trail ) = Linker::splitTrail( $trail );
2406
2407 if ( $text == '' ) {
2408 $text = htmlspecialchars( $nt->getPrefixedText() );
2409 }
2410
2411 $link = $this->getLinkRenderer()->makeKnownLink(
2412 $nt, new HtmlArmor( "$prefix$text$inside" )
2413 );
2414
2415 return $this->armorLinks( $link ) . $trail;
2416 }
2417
2418 /**
2419 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2420 * going to go through further parsing steps before inline URL expansion.
2421 *
2422 * Not needed quite as much as it used to be since free links are a bit
2423 * more sensible these days. But bracketed links are still an issue.
2424 *
2425 * @param string $text More-or-less HTML
2426 * @return string Less-or-more HTML with NOPARSE bits
2427 */
2428 public function armorLinks( $text ) {
2429 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2430 self::MARKER_PREFIX . "NOPARSE$1", $text );
2431 }
2432
2433 /**
2434 * Return true if subpage links should be expanded on this page.
2435 * @return bool
2436 */
2437 public function areSubpagesAllowed() {
2438 # Some namespaces don't allow subpages
2439 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2440 }
2441
2442 /**
2443 * Handle link to subpage if necessary
2444 *
2445 * @param string $target The source of the link
2446 * @param string &$text The link text, modified as necessary
2447 * @return string The full name of the link
2448 * @private
2449 */
2450 public function maybeDoSubpageLink( $target, &$text ) {
2451 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2452 }
2453
2454 /**
2455 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2456 *
2457 * @param string $text
2458 * @param bool $linestart Whether or not this is at the start of a line.
2459 * @private
2460 * @return string The lists rendered as HTML
2461 */
2462 public function doBlockLevels( $text, $linestart ) {
2463 return BlockLevelPass::doBlockLevels( $text, $linestart );
2464 }
2465
2466 /**
2467 * Return value of a magic variable (like PAGENAME)
2468 *
2469 * @private
2470 *
2471 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2472 * @param bool|PPFrame $frame
2473 *
2474 * @throws MWException
2475 * @return string
2476 */
2477 public function getVariableValue( $index, $frame = false ) {
2478 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2479 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2480
2481 if ( is_null( $this->mTitle ) ) {
2482 // If no title set, bad things are going to happen
2483 // later. Title should always be set since this
2484 // should only be called in the middle of a parse
2485 // operation (but the unit-tests do funky stuff)
2486 throw new MWException( __METHOD__ . ' Should only be '
2487 . ' called while parsing (no title set)' );
2488 }
2489
2490 // Avoid PHP 7.1 warning from passing $this by reference
2491 $parser = $this;
2492
2493 /**
2494 * Some of these require message or data lookups and can be
2495 * expensive to check many times.
2496 */
2497 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2498 if ( isset( $this->mVarCache[$index] ) ) {
2499 return $this->mVarCache[$index];
2500 }
2501 }
2502
2503 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2504 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2505
2506 $pageLang = $this->getFunctionLang();
2507
2508 switch ( $index ) {
2509 case '!':
2510 $value = '|';
2511 break;
2512 case 'currentmonth':
2513 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2514 break;
2515 case 'currentmonth1':
2516 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2517 break;
2518 case 'currentmonthname':
2519 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2520 break;
2521 case 'currentmonthnamegen':
2522 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2523 break;
2524 case 'currentmonthabbrev':
2525 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2526 break;
2527 case 'currentday':
2528 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2529 break;
2530 case 'currentday2':
2531 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2532 break;
2533 case 'localmonth':
2534 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2535 break;
2536 case 'localmonth1':
2537 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2538 break;
2539 case 'localmonthname':
2540 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2541 break;
2542 case 'localmonthnamegen':
2543 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2544 break;
2545 case 'localmonthabbrev':
2546 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2547 break;
2548 case 'localday':
2549 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2550 break;
2551 case 'localday2':
2552 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2553 break;
2554 case 'pagename':
2555 $value = wfEscapeWikiText( $this->mTitle->getText() );
2556 break;
2557 case 'pagenamee':
2558 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2559 break;
2560 case 'fullpagename':
2561 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2562 break;
2563 case 'fullpagenamee':
2564 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2565 break;
2566 case 'subpagename':
2567 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2568 break;
2569 case 'subpagenamee':
2570 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2571 break;
2572 case 'rootpagename':
2573 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2574 break;
2575 case 'rootpagenamee':
2576 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2577 ' ',
2578 '_',
2579 $this->mTitle->getRootText()
2580 ) ) );
2581 break;
2582 case 'basepagename':
2583 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2584 break;
2585 case 'basepagenamee':
2586 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2587 ' ',
2588 '_',
2589 $this->mTitle->getBaseText()
2590 ) ) );
2591 break;
2592 case 'talkpagename':
2593 if ( $this->mTitle->canHaveTalkPage() ) {
2594 $talkPage = $this->mTitle->getTalkPage();
2595 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2596 } else {
2597 $value = '';
2598 }
2599 break;
2600 case 'talkpagenamee':
2601 if ( $this->mTitle->canHaveTalkPage() ) {
2602 $talkPage = $this->mTitle->getTalkPage();
2603 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2604 } else {
2605 $value = '';
2606 }
2607 break;
2608 case 'subjectpagename':
2609 $subjPage = $this->mTitle->getSubjectPage();
2610 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2611 break;
2612 case 'subjectpagenamee':
2613 $subjPage = $this->mTitle->getSubjectPage();
2614 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2615 break;
2616 case 'pageid': // requested in T25427
2617 $pageid = $this->getTitle()->getArticleID();
2618 if ( $pageid == 0 ) {
2619 # 0 means the page doesn't exist in the database,
2620 # which means the user is previewing a new page.
2621 # The vary-revision flag must be set, because the magic word
2622 # will have a different value once the page is saved.
2623 $this->mOutput->setFlag( 'vary-revision' );
2624 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2625 }
2626 $value = $pageid ? $pageid : null;
2627 break;
2628 case 'revisionid':
2629 # Let the edit saving system know we should parse the page
2630 # *after* a revision ID has been assigned.
2631 $this->mOutput->setFlag( 'vary-revision-id' );
2632 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2633 $value = $this->mRevisionId;
2634 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2635 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2636 $this->mOutput->setSpeculativeRevIdUsed( $value );
2637 }
2638 break;
2639 case 'revisionday':
2640 # Let the edit saving system know we should parse the page
2641 # *after* a revision ID has been assigned. This is for null edits.
2642 $this->mOutput->setFlag( 'vary-revision' );
2643 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2644 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2645 break;
2646 case 'revisionday2':
2647 # Let the edit saving system know we should parse the page
2648 # *after* a revision ID has been assigned. This is for null edits.
2649 $this->mOutput->setFlag( 'vary-revision' );
2650 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2651 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2652 break;
2653 case 'revisionmonth':
2654 # Let the edit saving system know we should parse the page
2655 # *after* a revision ID has been assigned. This is for null edits.
2656 $this->mOutput->setFlag( 'vary-revision' );
2657 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2658 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2659 break;
2660 case 'revisionmonth1':
2661 # Let the edit saving system know we should parse the page
2662 # *after* a revision ID has been assigned. This is for null edits.
2663 $this->mOutput->setFlag( 'vary-revision' );
2664 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2665 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2666 break;
2667 case 'revisionyear':
2668 # Let the edit saving system know we should parse the page
2669 # *after* a revision ID has been assigned. This is for null edits.
2670 $this->mOutput->setFlag( 'vary-revision' );
2671 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2672 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2673 break;
2674 case 'revisiontimestamp':
2675 # Let the edit saving system know we should parse the page
2676 # *after* a revision ID has been assigned. This is for null edits.
2677 $this->mOutput->setFlag( 'vary-revision' );
2678 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2679 $value = $this->getRevisionTimestamp();
2680 break;
2681 case 'revisionuser':
2682 # Let the edit saving system know we should parse the page
2683 # *after* a revision ID has been assigned for null edits.
2684 $this->mOutput->setFlag( 'vary-user' );
2685 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2686 $value = $this->getRevisionUser();
2687 break;
2688 case 'revisionsize':
2689 $value = $this->getRevisionSize();
2690 break;
2691 case 'namespace':
2692 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2693 break;
2694 case 'namespacee':
2695 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2696 break;
2697 case 'namespacenumber':
2698 $value = $this->mTitle->getNamespace();
2699 break;
2700 case 'talkspace':
2701 $value = $this->mTitle->canHaveTalkPage()
2702 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2703 : '';
2704 break;
2705 case 'talkspacee':
2706 $value = $this->mTitle->canHaveTalkPage() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2707 break;
2708 case 'subjectspace':
2709 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2710 break;
2711 case 'subjectspacee':
2712 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2713 break;
2714 case 'currentdayname':
2715 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2716 break;
2717 case 'currentyear':
2718 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2719 break;
2720 case 'currenttime':
2721 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2722 break;
2723 case 'currenthour':
2724 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2725 break;
2726 case 'currentweek':
2727 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2728 # int to remove the padding
2729 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2730 break;
2731 case 'currentdow':
2732 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2733 break;
2734 case 'localdayname':
2735 $value = $pageLang->getWeekdayName(
2736 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2737 );
2738 break;
2739 case 'localyear':
2740 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2741 break;
2742 case 'localtime':
2743 $value = $pageLang->time(
2744 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2745 false,
2746 false
2747 );
2748 break;
2749 case 'localhour':
2750 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2751 break;
2752 case 'localweek':
2753 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2754 # int to remove the padding
2755 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2756 break;
2757 case 'localdow':
2758 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2759 break;
2760 case 'numberofarticles':
2761 $value = $pageLang->formatNum( SiteStats::articles() );
2762 break;
2763 case 'numberoffiles':
2764 $value = $pageLang->formatNum( SiteStats::images() );
2765 break;
2766 case 'numberofusers':
2767 $value = $pageLang->formatNum( SiteStats::users() );
2768 break;
2769 case 'numberofactiveusers':
2770 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2771 break;
2772 case 'numberofpages':
2773 $value = $pageLang->formatNum( SiteStats::pages() );
2774 break;
2775 case 'numberofadmins':
2776 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2777 break;
2778 case 'numberofedits':
2779 $value = $pageLang->formatNum( SiteStats::edits() );
2780 break;
2781 case 'currenttimestamp':
2782 $value = wfTimestamp( TS_MW, $ts );
2783 break;
2784 case 'localtimestamp':
2785 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2786 break;
2787 case 'currentversion':
2788 $value = SpecialVersion::getVersion();
2789 break;
2790 case 'articlepath':
2791 return $wgArticlePath;
2792 case 'sitename':
2793 return $wgSitename;
2794 case 'server':
2795 return $wgServer;
2796 case 'servername':
2797 return $wgServerName;
2798 case 'scriptpath':
2799 return $wgScriptPath;
2800 case 'stylepath':
2801 return $wgStylePath;
2802 case 'directionmark':
2803 return $pageLang->getDirMark();
2804 case 'contentlanguage':
2805 global $wgLanguageCode;
2806 return $wgLanguageCode;
2807 case 'pagelanguage':
2808 $value = $pageLang->getCode();
2809 break;
2810 case 'cascadingsources':
2811 $value = CoreParserFunctions::cascadingsources( $this );
2812 break;
2813 default:
2814 $ret = null;
2815 Hooks::run(
2816 'ParserGetVariableValueSwitch',
2817 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2818 );
2819
2820 return $ret;
2821 }
2822
2823 if ( $index ) {
2824 $this->mVarCache[$index] = $value;
2825 }
2826
2827 return $value;
2828 }
2829
2830 /**
2831 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2832 *
2833 * @private
2834 */
2835 public function initialiseVariables() {
2836 $variableIDs = MagicWord::getVariableIDs();
2837 $substIDs = MagicWord::getSubstIDs();
2838
2839 $this->mVariables = new MagicWordArray( $variableIDs );
2840 $this->mSubstWords = new MagicWordArray( $substIDs );
2841 }
2842
2843 /**
2844 * Preprocess some wikitext and return the document tree.
2845 * This is the ghost of replace_variables().
2846 *
2847 * @param string $text The text to parse
2848 * @param int $flags Bitwise combination of:
2849 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2850 * included. Default is to assume a direct page view.
2851 *
2852 * The generated DOM tree must depend only on the input text and the flags.
2853 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2854 *
2855 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2856 * change in the DOM tree for a given text, must be passed through the section identifier
2857 * in the section edit link and thus back to extractSections().
2858 *
2859 * The output of this function is currently only cached in process memory, but a persistent
2860 * cache may be implemented at a later date which takes further advantage of these strict
2861 * dependency requirements.
2862 *
2863 * @return PPNode
2864 */
2865 public function preprocessToDom( $text, $flags = 0 ) {
2866 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2867 return $dom;
2868 }
2869
2870 /**
2871 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2872 *
2873 * @param string $s
2874 *
2875 * @return array
2876 */
2877 public static function splitWhitespace( $s ) {
2878 $ltrimmed = ltrim( $s );
2879 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2880 $trimmed = rtrim( $ltrimmed );
2881 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2882 if ( $diff > 0 ) {
2883 $w2 = substr( $ltrimmed, -$diff );
2884 } else {
2885 $w2 = '';
2886 }
2887 return [ $w1, $trimmed, $w2 ];
2888 }
2889
2890 /**
2891 * Replace magic variables, templates, and template arguments
2892 * with the appropriate text. Templates are substituted recursively,
2893 * taking care to avoid infinite loops.
2894 *
2895 * Note that the substitution depends on value of $mOutputType:
2896 * self::OT_WIKI: only {{subst:}} templates
2897 * self::OT_PREPROCESS: templates but not extension tags
2898 * self::OT_HTML: all templates and extension tags
2899 *
2900 * @param string $text The text to transform
2901 * @param bool|PPFrame $frame Object describing the arguments passed to the
2902 * template. Arguments may also be provided as an associative array, as
2903 * was the usual case before MW1.12. Providing arguments this way may be
2904 * useful for extensions wishing to perform variable replacement
2905 * explicitly.
2906 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2907 * double-brace expansion.
2908 * @return string
2909 */
2910 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2911 # Is there any text? Also, Prevent too big inclusions!
2912 $textSize = strlen( $text );
2913 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2914 return $text;
2915 }
2916
2917 if ( $frame === false ) {
2918 $frame = $this->getPreprocessor()->newFrame();
2919 } elseif ( !( $frame instanceof PPFrame ) ) {
2920 wfDebug( __METHOD__ . " called using plain parameters instead of "
2921 . "a PPFrame instance. Creating custom frame.\n" );
2922 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2923 }
2924
2925 $dom = $this->preprocessToDom( $text );
2926 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2927 $text = $frame->expand( $dom, $flags );
2928
2929 return $text;
2930 }
2931
2932 /**
2933 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2934 *
2935 * @param array $args
2936 *
2937 * @return array
2938 */
2939 public static function createAssocArgs( $args ) {
2940 $assocArgs = [];
2941 $index = 1;
2942 foreach ( $args as $arg ) {
2943 $eqpos = strpos( $arg, '=' );
2944 if ( $eqpos === false ) {
2945 $assocArgs[$index++] = $arg;
2946 } else {
2947 $name = trim( substr( $arg, 0, $eqpos ) );
2948 $value = trim( substr( $arg, $eqpos + 1 ) );
2949 if ( $value === false ) {
2950 $value = '';
2951 }
2952 if ( $name !== false ) {
2953 $assocArgs[$name] = $value;
2954 }
2955 }
2956 }
2957
2958 return $assocArgs;
2959 }
2960
2961 /**
2962 * Warn the user when a parser limitation is reached
2963 * Will warn at most once the user per limitation type
2964 *
2965 * The results are shown during preview and run through the Parser (See EditPage.php)
2966 *
2967 * @param string $limitationType Should be one of:
2968 * 'expensive-parserfunction' (corresponding messages:
2969 * 'expensive-parserfunction-warning',
2970 * 'expensive-parserfunction-category')
2971 * 'post-expand-template-argument' (corresponding messages:
2972 * 'post-expand-template-argument-warning',
2973 * 'post-expand-template-argument-category')
2974 * 'post-expand-template-inclusion' (corresponding messages:
2975 * 'post-expand-template-inclusion-warning',
2976 * 'post-expand-template-inclusion-category')
2977 * 'node-count-exceeded' (corresponding messages:
2978 * 'node-count-exceeded-warning',
2979 * 'node-count-exceeded-category')
2980 * 'expansion-depth-exceeded' (corresponding messages:
2981 * 'expansion-depth-exceeded-warning',
2982 * 'expansion-depth-exceeded-category')
2983 * @param string|int|null $current Current value
2984 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2985 * exceeded, provide the values (optional)
2986 */
2987 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2988 # does no harm if $current and $max are present but are unnecessary for the message
2989 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2990 # only during preview, and that would split the parser cache unnecessarily.
2991 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2992 ->text();
2993 $this->mOutput->addWarning( $warning );
2994 $this->addTrackingCategory( "$limitationType-category" );
2995 }
2996
2997 /**
2998 * Return the text of a template, after recursively
2999 * replacing any variables or templates within the template.
3000 *
3001 * @param array $piece The parts of the template
3002 * $piece['title']: the title, i.e. the part before the |
3003 * $piece['parts']: the parameter array
3004 * $piece['lineStart']: whether the brace was at the start of a line
3005 * @param PPFrame $frame The current frame, contains template arguments
3006 * @throws Exception
3007 * @return string The text of the template
3008 */
3009 public function braceSubstitution( $piece, $frame ) {
3010 // Flags
3011
3012 // $text has been filled
3013 $found = false;
3014 // wiki markup in $text should be escaped
3015 $nowiki = false;
3016 // $text is HTML, armour it against wikitext transformation
3017 $isHTML = false;
3018 // Force interwiki transclusion to be done in raw mode not rendered
3019 $forceRawInterwiki = false;
3020 // $text is a DOM node needing expansion in a child frame
3021 $isChildObj = false;
3022 // $text is a DOM node needing expansion in the current frame
3023 $isLocalObj = false;
3024
3025 # Title object, where $text came from
3026 $title = false;
3027
3028 # $part1 is the bit before the first |, and must contain only title characters.
3029 # Various prefixes will be stripped from it later.
3030 $titleWithSpaces = $frame->expand( $piece['title'] );
3031 $part1 = trim( $titleWithSpaces );
3032 $titleText = false;
3033
3034 # Original title text preserved for various purposes
3035 $originalTitle = $part1;
3036
3037 # $args is a list of argument nodes, starting from index 0, not including $part1
3038 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3039 # below won't work b/c this $args isn't an object
3040 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3041
3042 $profileSection = null; // profile templates
3043
3044 # SUBST
3045 if ( !$found ) {
3046 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3047
3048 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3049 # Decide whether to expand template or keep wikitext as-is.
3050 if ( $this->ot['wiki'] ) {
3051 if ( $substMatch === false ) {
3052 $literal = true; # literal when in PST with no prefix
3053 } else {
3054 $literal = false; # expand when in PST with subst: or safesubst:
3055 }
3056 } else {
3057 if ( $substMatch == 'subst' ) {
3058 $literal = true; # literal when not in PST with plain subst:
3059 } else {
3060 $literal = false; # expand when not in PST with safesubst: or no prefix
3061 }
3062 }
3063 if ( $literal ) {
3064 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3065 $isLocalObj = true;
3066 $found = true;
3067 }
3068 }
3069
3070 # Variables
3071 if ( !$found && $args->getLength() == 0 ) {
3072 $id = $this->mVariables->matchStartToEnd( $part1 );
3073 if ( $id !== false ) {
3074 $text = $this->getVariableValue( $id, $frame );
3075 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3076 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3077 }
3078 $found = true;
3079 }
3080 }
3081
3082 # MSG, MSGNW and RAW
3083 if ( !$found ) {
3084 # Check for MSGNW:
3085 $mwMsgnw = MagicWord::get( 'msgnw' );
3086 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3087 $nowiki = true;
3088 } else {
3089 # Remove obsolete MSG:
3090 $mwMsg = MagicWord::get( 'msg' );
3091 $mwMsg->matchStartAndRemove( $part1 );
3092 }
3093
3094 # Check for RAW:
3095 $mwRaw = MagicWord::get( 'raw' );
3096 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3097 $forceRawInterwiki = true;
3098 }
3099 }
3100
3101 # Parser functions
3102 if ( !$found ) {
3103 $colonPos = strpos( $part1, ':' );
3104 if ( $colonPos !== false ) {
3105 $func = substr( $part1, 0, $colonPos );
3106 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3107 $argsLength = $args->getLength();
3108 for ( $i = 0; $i < $argsLength; $i++ ) {
3109 $funcArgs[] = $args->item( $i );
3110 }
3111 try {
3112 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3113 } catch ( Exception $ex ) {
3114 throw $ex;
3115 }
3116
3117 # The interface for parser functions allows for extracting
3118 # flags into the local scope. Extract any forwarded flags
3119 # here.
3120 extract( $result );
3121 }
3122 }
3123
3124 # Finish mangling title and then check for loops.
3125 # Set $title to a Title object and $titleText to the PDBK
3126 if ( !$found ) {
3127 $ns = NS_TEMPLATE;
3128 # Split the title into page and subpage
3129 $subpage = '';
3130 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3131 if ( $part1 !== $relative ) {
3132 $part1 = $relative;
3133 $ns = $this->mTitle->getNamespace();
3134 }
3135 $title = Title::newFromText( $part1, $ns );
3136 if ( $title ) {
3137 $titleText = $title->getPrefixedText();
3138 # Check for language variants if the template is not found
3139 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3140 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3141 }
3142 # Do recursion depth check
3143 $limit = $this->mOptions->getMaxTemplateDepth();
3144 if ( $frame->depth >= $limit ) {
3145 $found = true;
3146 $text = '<span class="error">'
3147 . wfMessage( 'parser-template-recursion-depth-warning' )
3148 ->numParams( $limit )->inContentLanguage()->text()
3149 . '</span>';
3150 }
3151 }
3152 }
3153
3154 # Load from database
3155 if ( !$found && $title ) {
3156 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3157 if ( !$title->isExternal() ) {
3158 if ( $title->isSpecialPage()
3159 && $this->mOptions->getAllowSpecialInclusion()
3160 && $this->ot['html']
3161 ) {
3162 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3163 // Pass the template arguments as URL parameters.
3164 // "uselang" will have no effect since the Language object
3165 // is forced to the one defined in ParserOptions.
3166 $pageArgs = [];
3167 $argsLength = $args->getLength();
3168 for ( $i = 0; $i < $argsLength; $i++ ) {
3169 $bits = $args->item( $i )->splitArg();
3170 if ( strval( $bits['index'] ) === '' ) {
3171 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3172 $value = trim( $frame->expand( $bits['value'] ) );
3173 $pageArgs[$name] = $value;
3174 }
3175 }
3176
3177 // Create a new context to execute the special page
3178 $context = new RequestContext;
3179 $context->setTitle( $title );
3180 $context->setRequest( new FauxRequest( $pageArgs ) );
3181 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3182 $context->setUser( $this->getUser() );
3183 } else {
3184 // If this page is cached, then we better not be per user.
3185 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3186 }
3187 $context->setLanguage( $this->mOptions->getUserLangObj() );
3188 $ret = SpecialPageFactory::capturePath(
3189 $title, $context, $this->getLinkRenderer() );
3190 if ( $ret ) {
3191 $text = $context->getOutput()->getHTML();
3192 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3193 $found = true;
3194 $isHTML = true;
3195 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3196 $this->mOutput->updateRuntimeAdaptiveExpiry(
3197 $specialPage->maxIncludeCacheTime()
3198 );
3199 }
3200 }
3201 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3202 $found = false; # access denied
3203 wfDebug( __METHOD__ . ": template inclusion denied for " .
3204 $title->getPrefixedDBkey() . "\n" );
3205 } else {
3206 list( $text, $title ) = $this->getTemplateDom( $title );
3207 if ( $text !== false ) {
3208 $found = true;
3209 $isChildObj = true;
3210 }
3211 }
3212
3213 # If the title is valid but undisplayable, make a link to it
3214 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3215 $text = "[[:$titleText]]";
3216 $found = true;
3217 }
3218 } elseif ( $title->isTrans() ) {
3219 # Interwiki transclusion
3220 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3221 $text = $this->interwikiTransclude( $title, 'render' );
3222 $isHTML = true;
3223 } else {
3224 $text = $this->interwikiTransclude( $title, 'raw' );
3225 # Preprocess it like a template
3226 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3227 $isChildObj = true;
3228 }
3229 $found = true;
3230 }
3231
3232 # Do infinite loop check
3233 # This has to be done after redirect resolution to avoid infinite loops via redirects
3234 if ( !$frame->loopCheck( $title ) ) {
3235 $found = true;
3236 $text = '<span class="error">'
3237 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3238 . '</span>';
3239 $this->addTrackingCategory( 'template-loop-category' );
3240 $this->mOutput->addWarning( wfMessage( 'template-loop-warning',
3241 wfEscapeWikiText( $titleText ) )->text() );
3242 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3243 }
3244 }
3245
3246 # If we haven't found text to substitute by now, we're done
3247 # Recover the source wikitext and return it
3248 if ( !$found ) {
3249 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3250 if ( $profileSection ) {
3251 $this->mProfiler->scopedProfileOut( $profileSection );
3252 }
3253 return [ 'object' => $text ];
3254 }
3255
3256 # Expand DOM-style return values in a child frame
3257 if ( $isChildObj ) {
3258 # Clean up argument array
3259 $newFrame = $frame->newChild( $args, $title );
3260
3261 if ( $nowiki ) {
3262 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3263 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3264 # Expansion is eligible for the empty-frame cache
3265 $text = $newFrame->cachedExpand( $titleText, $text );
3266 } else {
3267 # Uncached expansion
3268 $text = $newFrame->expand( $text );
3269 }
3270 }
3271 if ( $isLocalObj && $nowiki ) {
3272 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3273 $isLocalObj = false;
3274 }
3275
3276 if ( $profileSection ) {
3277 $this->mProfiler->scopedProfileOut( $profileSection );
3278 }
3279
3280 # Replace raw HTML by a placeholder
3281 if ( $isHTML ) {
3282 $text = $this->insertStripItem( $text );
3283 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3284 # Escape nowiki-style return values
3285 $text = wfEscapeWikiText( $text );
3286 } elseif ( is_string( $text )
3287 && !$piece['lineStart']
3288 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3289 ) {
3290 # T2529: if the template begins with a table or block-level
3291 # element, it should be treated as beginning a new line.
3292 # This behavior is somewhat controversial.
3293 $text = "\n" . $text;
3294 }
3295
3296 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3297 # Error, oversize inclusion
3298 if ( $titleText !== false ) {
3299 # Make a working, properly escaped link if possible (T25588)
3300 $text = "[[:$titleText]]";
3301 } else {
3302 # This will probably not be a working link, but at least it may
3303 # provide some hint of where the problem is
3304 preg_replace( '/^:/', '', $originalTitle );
3305 $text = "[[:$originalTitle]]";
3306 }
3307 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3308 . 'post-expand include size too large -->' );
3309 $this->limitationWarn( 'post-expand-template-inclusion' );
3310 }
3311
3312 if ( $isLocalObj ) {
3313 $ret = [ 'object' => $text ];
3314 } else {
3315 $ret = [ 'text' => $text ];
3316 }
3317
3318 return $ret;
3319 }
3320
3321 /**
3322 * Call a parser function and return an array with text and flags.
3323 *
3324 * The returned array will always contain a boolean 'found', indicating
3325 * whether the parser function was found or not. It may also contain the
3326 * following:
3327 * text: string|object, resulting wikitext or PP DOM object
3328 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3329 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3330 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3331 * nowiki: bool, wiki markup in $text should be escaped
3332 *
3333 * @since 1.21
3334 * @param PPFrame $frame The current frame, contains template arguments
3335 * @param string $function Function name
3336 * @param array $args Arguments to the function
3337 * @throws MWException
3338 * @return array
3339 */
3340 public function callParserFunction( $frame, $function, array $args = [] ) {
3341 global $wgContLang;
3342
3343 # Case sensitive functions
3344 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3345 $function = $this->mFunctionSynonyms[1][$function];
3346 } else {
3347 # Case insensitive functions
3348 $function = $wgContLang->lc( $function );
3349 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3350 $function = $this->mFunctionSynonyms[0][$function];
3351 } else {
3352 return [ 'found' => false ];
3353 }
3354 }
3355
3356 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3357
3358 // Avoid PHP 7.1 warning from passing $this by reference
3359 $parser = $this;
3360
3361 $allArgs = [ &$parser ];
3362 if ( $flags & self::SFH_OBJECT_ARGS ) {
3363 # Convert arguments to PPNodes and collect for appending to $allArgs
3364 $funcArgs = [];
3365 foreach ( $args as $k => $v ) {
3366 if ( $v instanceof PPNode || $k === 0 ) {
3367 $funcArgs[] = $v;
3368 } else {
3369 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3370 }
3371 }
3372
3373 # Add a frame parameter, and pass the arguments as an array
3374 $allArgs[] = $frame;
3375 $allArgs[] = $funcArgs;
3376 } else {
3377 # Convert arguments to plain text and append to $allArgs
3378 foreach ( $args as $k => $v ) {
3379 if ( $v instanceof PPNode ) {
3380 $allArgs[] = trim( $frame->expand( $v ) );
3381 } elseif ( is_int( $k ) && $k >= 0 ) {
3382 $allArgs[] = trim( $v );
3383 } else {
3384 $allArgs[] = trim( "$k=$v" );
3385 }
3386 }
3387 }
3388
3389 $result = call_user_func_array( $callback, $allArgs );
3390
3391 # The interface for function hooks allows them to return a wikitext
3392 # string or an array containing the string and any flags. This mungs
3393 # things around to match what this method should return.
3394 if ( !is_array( $result ) ) {
3395 $result = [
3396 'found' => true,
3397 'text' => $result,
3398 ];
3399 } else {
3400 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3401 $result['text'] = $result[0];
3402 }
3403 unset( $result[0] );
3404 $result += [
3405 'found' => true,
3406 ];
3407 }
3408
3409 $noparse = true;
3410 $preprocessFlags = 0;
3411 if ( isset( $result['noparse'] ) ) {
3412 $noparse = $result['noparse'];
3413 }
3414 if ( isset( $result['preprocessFlags'] ) ) {
3415 $preprocessFlags = $result['preprocessFlags'];
3416 }
3417
3418 if ( !$noparse ) {
3419 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3420 $result['isChildObj'] = true;
3421 }
3422
3423 return $result;
3424 }
3425
3426 /**
3427 * Get the semi-parsed DOM representation of a template with a given title,
3428 * and its redirect destination title. Cached.
3429 *
3430 * @param Title $title
3431 *
3432 * @return array
3433 */
3434 public function getTemplateDom( $title ) {
3435 $cacheTitle = $title;
3436 $titleText = $title->getPrefixedDBkey();
3437
3438 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3439 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3440 $title = Title::makeTitle( $ns, $dbk );
3441 $titleText = $title->getPrefixedDBkey();
3442 }
3443 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3444 return [ $this->mTplDomCache[$titleText], $title ];
3445 }
3446
3447 # Cache miss, go to the database
3448 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3449
3450 if ( $text === false ) {
3451 $this->mTplDomCache[$titleText] = false;
3452 return [ false, $title ];
3453 }
3454
3455 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3456 $this->mTplDomCache[$titleText] = $dom;
3457
3458 if ( !$title->equals( $cacheTitle ) ) {
3459 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3460 [ $title->getNamespace(), $title->getDBkey() ];
3461 }
3462
3463 return [ $dom, $title ];
3464 }
3465
3466 /**
3467 * Fetch the current revision of a given title. Note that the revision
3468 * (and even the title) may not exist in the database, so everything
3469 * contributing to the output of the parser should use this method
3470 * where possible, rather than getting the revisions themselves. This
3471 * method also caches its results, so using it benefits performance.
3472 *
3473 * @since 1.24
3474 * @param Title $title
3475 * @return Revision
3476 */
3477 public function fetchCurrentRevisionOfTitle( $title ) {
3478 $cacheKey = $title->getPrefixedDBkey();
3479 if ( !$this->currentRevisionCache ) {
3480 $this->currentRevisionCache = new MapCacheLRU( 100 );
3481 }
3482 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3483 $this->currentRevisionCache->set( $cacheKey,
3484 // Defaults to Parser::statelessFetchRevision()
3485 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3486 );
3487 }
3488 return $this->currentRevisionCache->get( $cacheKey );
3489 }
3490
3491 /**
3492 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3493 * without passing them on to it.
3494 *
3495 * @since 1.24
3496 * @param Title $title
3497 * @param Parser|bool $parser
3498 * @return Revision|bool False if missing
3499 */
3500 public static function statelessFetchRevision( Title $title, $parser = false ) {
3501 $pageId = $title->getArticleID();
3502 $revId = $title->getLatestRevID();
3503
3504 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3505 if ( $rev ) {
3506 $rev->setTitle( $title );
3507 }
3508
3509 return $rev;
3510 }
3511
3512 /**
3513 * Fetch the unparsed text of a template and register a reference to it.
3514 * @param Title $title
3515 * @return array ( string or false, Title )
3516 */
3517 public function fetchTemplateAndTitle( $title ) {
3518 // Defaults to Parser::statelessFetchTemplate()
3519 $templateCb = $this->mOptions->getTemplateCallback();
3520 $stuff = call_user_func( $templateCb, $title, $this );
3521 // We use U+007F DELETE to distinguish strip markers from regular text.
3522 $text = $stuff['text'];
3523 if ( is_string( $stuff['text'] ) ) {
3524 $text = strtr( $text, "\x7f", "?" );
3525 }
3526 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3527 if ( isset( $stuff['deps'] ) ) {
3528 foreach ( $stuff['deps'] as $dep ) {
3529 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3530 if ( $dep['title']->equals( $this->getTitle() ) ) {
3531 // If we transclude ourselves, the final result
3532 // will change based on the new version of the page
3533 $this->mOutput->setFlag( 'vary-revision' );
3534 }
3535 }
3536 }
3537 return [ $text, $finalTitle ];
3538 }
3539
3540 /**
3541 * Fetch the unparsed text of a template and register a reference to it.
3542 * @param Title $title
3543 * @return string|bool
3544 */
3545 public function fetchTemplate( $title ) {
3546 return $this->fetchTemplateAndTitle( $title )[0];
3547 }
3548
3549 /**
3550 * Static function to get a template
3551 * Can be overridden via ParserOptions::setTemplateCallback().
3552 *
3553 * @param Title $title
3554 * @param bool|Parser $parser
3555 *
3556 * @return array
3557 */
3558 public static function statelessFetchTemplate( $title, $parser = false ) {
3559 $text = $skip = false;
3560 $finalTitle = $title;
3561 $deps = [];
3562
3563 # Loop to fetch the article, with up to 1 redirect
3564 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3565 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3566 // @codingStandardsIgnoreEnd
3567 # Give extensions a chance to select the revision instead
3568 $id = false; # Assume current
3569 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3570 [ $parser, $title, &$skip, &$id ] );
3571
3572 if ( $skip ) {
3573 $text = false;
3574 $deps[] = [
3575 'title' => $title,
3576 'page_id' => $title->getArticleID(),
3577 'rev_id' => null
3578 ];
3579 break;
3580 }
3581 # Get the revision
3582 if ( $id ) {
3583 $rev = Revision::newFromId( $id );
3584 } elseif ( $parser ) {
3585 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3586 } else {
3587 $rev = Revision::newFromTitle( $title );
3588 }
3589 $rev_id = $rev ? $rev->getId() : 0;
3590 # If there is no current revision, there is no page
3591 if ( $id === false && !$rev ) {
3592 $linkCache = LinkCache::singleton();
3593 $linkCache->addBadLinkObj( $title );
3594 }
3595
3596 $deps[] = [
3597 'title' => $title,
3598 'page_id' => $title->getArticleID(),
3599 'rev_id' => $rev_id ];
3600 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3601 # We fetched a rev from a different title; register it too...
3602 $deps[] = [
3603 'title' => $rev->getTitle(),
3604 'page_id' => $rev->getPage(),
3605 'rev_id' => $rev_id ];
3606 }
3607
3608 if ( $rev ) {
3609 $content = $rev->getContent();
3610 $text = $content ? $content->getWikitextForTransclusion() : null;
3611
3612 Hooks::run( 'ParserFetchTemplate',
3613 [ $parser, $title, $rev, &$text, &$deps ] );
3614
3615 if ( $text === false || $text === null ) {
3616 $text = false;
3617 break;
3618 }
3619 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3620 global $wgContLang;
3621 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3622 if ( !$message->exists() ) {
3623 $text = false;
3624 break;
3625 }
3626 $content = $message->content();
3627 $text = $message->plain();
3628 } else {
3629 break;
3630 }
3631 if ( !$content ) {
3632 break;
3633 }
3634 # Redirect?
3635 $finalTitle = $title;
3636 $title = $content->getRedirectTarget();
3637 }
3638 return [
3639 'text' => $text,
3640 'finalTitle' => $finalTitle,
3641 'deps' => $deps ];
3642 }
3643
3644 /**
3645 * Fetch a file and its title and register a reference to it.
3646 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3647 * @param Title $title
3648 * @param array $options Array of options to RepoGroup::findFile
3649 * @return File|bool
3650 */
3651 public function fetchFile( $title, $options = [] ) {
3652 return $this->fetchFileAndTitle( $title, $options )[0];
3653 }
3654
3655 /**
3656 * Fetch a file and its title and register a reference to it.
3657 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3658 * @param Title $title
3659 * @param array $options Array of options to RepoGroup::findFile
3660 * @return array ( File or false, Title of file )
3661 */
3662 public function fetchFileAndTitle( $title, $options = [] ) {
3663 $file = $this->fetchFileNoRegister( $title, $options );
3664
3665 $time = $file ? $file->getTimestamp() : false;
3666 $sha1 = $file ? $file->getSha1() : false;
3667 # Register the file as a dependency...
3668 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3669 if ( $file && !$title->equals( $file->getTitle() ) ) {
3670 # Update fetched file title
3671 $title = $file->getTitle();
3672 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3673 }
3674 return [ $file, $title ];
3675 }
3676
3677 /**
3678 * Helper function for fetchFileAndTitle.
3679 *
3680 * Also useful if you need to fetch a file but not use it yet,
3681 * for example to get the file's handler.
3682 *
3683 * @param Title $title
3684 * @param array $options Array of options to RepoGroup::findFile
3685 * @return File|bool
3686 */
3687 protected function fetchFileNoRegister( $title, $options = [] ) {
3688 if ( isset( $options['broken'] ) ) {
3689 $file = false; // broken thumbnail forced by hook
3690 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3691 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3692 } else { // get by (name,timestamp)
3693 $file = wfFindFile( $title, $options );
3694 }
3695 return $file;
3696 }
3697
3698 /**
3699 * Transclude an interwiki link.
3700 *
3701 * @param Title $title
3702 * @param string $action
3703 *
3704 * @return string
3705 */
3706 public function interwikiTransclude( $title, $action ) {
3707 global $wgEnableScaryTranscluding;
3708
3709 if ( !$wgEnableScaryTranscluding ) {
3710 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3711 }
3712
3713 $url = $title->getFullURL( [ 'action' => $action ] );
3714
3715 if ( strlen( $url ) > 255 ) {
3716 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3717 }
3718 return $this->fetchScaryTemplateMaybeFromCache( $url );
3719 }
3720
3721 /**
3722 * @param string $url
3723 * @return mixed|string
3724 */
3725 public function fetchScaryTemplateMaybeFromCache( $url ) {
3726 global $wgTranscludeCacheExpiry;
3727 $dbr = wfGetDB( DB_REPLICA );
3728 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3729 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3730 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3731 if ( $obj ) {
3732 return $obj->tc_contents;
3733 }
3734
3735 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3736 $status = $req->execute(); // Status object
3737 if ( $status->isOK() ) {
3738 $text = $req->getContent();
3739 } elseif ( $req->getStatus() != 200 ) {
3740 // Though we failed to fetch the content, this status is useless.
3741 return wfMessage( 'scarytranscludefailed-httpstatus' )
3742 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3743 } else {
3744 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3745 }
3746
3747 $dbw = wfGetDB( DB_MASTER );
3748 $dbw->replace( 'transcache', [ 'tc_url' ], [
3749 'tc_url' => $url,
3750 'tc_time' => $dbw->timestamp( time() ),
3751 'tc_contents' => $text
3752 ] );
3753 return $text;
3754 }
3755
3756 /**
3757 * Triple brace replacement -- used for template arguments
3758 * @private
3759 *
3760 * @param array $piece
3761 * @param PPFrame $frame
3762 *
3763 * @return array
3764 */
3765 public function argSubstitution( $piece, $frame ) {
3766 $error = false;
3767 $parts = $piece['parts'];
3768 $nameWithSpaces = $frame->expand( $piece['title'] );
3769 $argName = trim( $nameWithSpaces );
3770 $object = false;
3771 $text = $frame->getArgument( $argName );
3772 if ( $text === false && $parts->getLength() > 0
3773 && ( $this->ot['html']
3774 || $this->ot['pre']
3775 || ( $this->ot['wiki'] && $frame->isTemplate() )
3776 )
3777 ) {
3778 # No match in frame, use the supplied default
3779 $object = $parts->item( 0 )->getChildren();
3780 }
3781 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3782 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3783 $this->limitationWarn( 'post-expand-template-argument' );
3784 }
3785
3786 if ( $text === false && $object === false ) {
3787 # No match anywhere
3788 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3789 }
3790 if ( $error !== false ) {
3791 $text .= $error;
3792 }
3793 if ( $object !== false ) {
3794 $ret = [ 'object' => $object ];
3795 } else {
3796 $ret = [ 'text' => $text ];
3797 }
3798
3799 return $ret;
3800 }
3801
3802 /**
3803 * Return the text to be used for a given extension tag.
3804 * This is the ghost of strip().
3805 *
3806 * @param array $params Associative array of parameters:
3807 * name PPNode for the tag name
3808 * attr PPNode for unparsed text where tag attributes are thought to be
3809 * attributes Optional associative array of parsed attributes
3810 * inner Contents of extension element
3811 * noClose Original text did not have a close tag
3812 * @param PPFrame $frame
3813 *
3814 * @throws MWException
3815 * @return string
3816 */
3817 public function extensionSubstitution( $params, $frame ) {
3818 static $errorStr = '<span class="error">';
3819 static $errorLen = 20;
3820
3821 $name = $frame->expand( $params['name'] );
3822 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3823 // Probably expansion depth or node count exceeded. Just punt the
3824 // error up.
3825 return $name;
3826 }
3827
3828 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3829 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3830 // See above
3831 return $attrText;
3832 }
3833
3834 // We can't safely check if the expansion for $content resulted in an
3835 // error, because the content could happen to be the error string
3836 // (T149622).
3837 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3838
3839 $marker = self::MARKER_PREFIX . "-$name-"
3840 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3841
3842 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3843 ( $this->ot['html'] || $this->ot['pre'] );
3844 if ( $isFunctionTag ) {
3845 $markerType = 'none';
3846 } else {
3847 $markerType = 'general';
3848 }
3849 if ( $this->ot['html'] || $isFunctionTag ) {
3850 $name = strtolower( $name );
3851 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3852 if ( isset( $params['attributes'] ) ) {
3853 $attributes = $attributes + $params['attributes'];
3854 }
3855
3856 if ( isset( $this->mTagHooks[$name] ) ) {
3857 $output = call_user_func_array( $this->mTagHooks[$name],
3858 [ $content, $attributes, $this, $frame ] );
3859 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3860 list( $callback, ) = $this->mFunctionTagHooks[$name];
3861
3862 // Avoid PHP 7.1 warning from passing $this by reference
3863 $parser = $this;
3864 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3865 } else {
3866 $output = '<span class="error">Invalid tag extension name: ' .
3867 htmlspecialchars( $name ) . '</span>';
3868 }
3869
3870 if ( is_array( $output ) ) {
3871 # Extract flags to local scope (to override $markerType)
3872 $flags = $output;
3873 $output = $flags[0];
3874 unset( $flags[0] );
3875 extract( $flags );
3876 }
3877 } else {
3878 if ( is_null( $attrText ) ) {
3879 $attrText = '';
3880 }
3881 if ( isset( $params['attributes'] ) ) {
3882 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3883 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3884 htmlspecialchars( $attrValue ) . '"';
3885 }
3886 }
3887 if ( $content === null ) {
3888 $output = "<$name$attrText/>";
3889 } else {
3890 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3891 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3892 // See above
3893 return $close;
3894 }
3895 $output = "<$name$attrText>$content$close";
3896 }
3897 }
3898
3899 if ( $markerType === 'none' ) {
3900 return $output;
3901 } elseif ( $markerType === 'nowiki' ) {
3902 $this->mStripState->addNoWiki( $marker, $output );
3903 } elseif ( $markerType === 'general' ) {
3904 $this->mStripState->addGeneral( $marker, $output );
3905 } else {
3906 throw new MWException( __METHOD__ . ': invalid marker type' );
3907 }
3908 return $marker;
3909 }
3910
3911 /**
3912 * Increment an include size counter
3913 *
3914 * @param string $type The type of expansion
3915 * @param int $size The size of the text
3916 * @return bool False if this inclusion would take it over the maximum, true otherwise
3917 */
3918 public function incrementIncludeSize( $type, $size ) {
3919 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3920 return false;
3921 } else {
3922 $this->mIncludeSizes[$type] += $size;
3923 return true;
3924 }
3925 }
3926
3927 /**
3928 * Increment the expensive function count
3929 *
3930 * @return bool False if the limit has been exceeded
3931 */
3932 public function incrementExpensiveFunctionCount() {
3933 $this->mExpensiveFunctionCount++;
3934 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3935 }
3936
3937 /**
3938 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3939 * Fills $this->mDoubleUnderscores, returns the modified text
3940 *
3941 * @param string $text
3942 *
3943 * @return string
3944 */
3945 public function doDoubleUnderscore( $text ) {
3946 # The position of __TOC__ needs to be recorded
3947 $mw = MagicWord::get( 'toc' );
3948 if ( $mw->match( $text ) ) {
3949 $this->mShowToc = true;
3950 $this->mForceTocPosition = true;
3951
3952 # Set a placeholder. At the end we'll fill it in with the TOC.
3953 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3954
3955 # Only keep the first one.
3956 $text = $mw->replace( '', $text );
3957 }
3958
3959 # Now match and remove the rest of them
3960 $mwa = MagicWord::getDoubleUnderscoreArray();
3961 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3962
3963 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3964 $this->mOutput->mNoGallery = true;
3965 }
3966 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3967 $this->mShowToc = false;
3968 }
3969 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3970 && $this->mTitle->getNamespace() == NS_CATEGORY
3971 ) {
3972 $this->addTrackingCategory( 'hidden-category-category' );
3973 }
3974 # (T10068) Allow control over whether robots index a page.
3975 # __INDEX__ always overrides __NOINDEX__, see T16899
3976 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3977 $this->mOutput->setIndexPolicy( 'noindex' );
3978 $this->addTrackingCategory( 'noindex-category' );
3979 }
3980 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3981 $this->mOutput->setIndexPolicy( 'index' );
3982 $this->addTrackingCategory( 'index-category' );
3983 }
3984
3985 # Cache all double underscores in the database
3986 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3987 $this->mOutput->setProperty( $key, '' );
3988 }
3989
3990 return $text;
3991 }
3992
3993 /**
3994 * @see ParserOutput::addTrackingCategory()
3995 * @param string $msg Message key
3996 * @return bool Whether the addition was successful
3997 */
3998 public function addTrackingCategory( $msg ) {
3999 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4000 }
4001
4002 /**
4003 * This function accomplishes several tasks:
4004 * 1) Auto-number headings if that option is enabled
4005 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4006 * 3) Add a Table of contents on the top for users who have enabled the option
4007 * 4) Auto-anchor headings
4008 *
4009 * It loops through all headlines, collects the necessary data, then splits up the
4010 * string and re-inserts the newly formatted headlines.
4011 *
4012 * @param string $text
4013 * @param string $origText Original, untouched wikitext
4014 * @param bool $isMain
4015 * @return mixed|string
4016 * @private
4017 */
4018 public function formatHeadings( $text, $origText, $isMain = true ) {
4019 global $wgMaxTocLevel;
4020
4021 # Inhibit editsection links if requested in the page
4022 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4023 $maybeShowEditLink = $showEditLink = false;
4024 } else {
4025 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4026 $showEditLink = $this->mOptions->getEditSection();
4027 }
4028 if ( $showEditLink ) {
4029 $this->mOutput->setEditSectionTokens( true );
4030 }
4031
4032 # Get all headlines for numbering them and adding funky stuff like [edit]
4033 # links - this is for later, but we need the number of headlines right now
4034 $matches = [];
4035 $numMatches = preg_match_all(
4036 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4037 $text,
4038 $matches
4039 );
4040
4041 # if there are fewer than 4 headlines in the article, do not show TOC
4042 # unless it's been explicitly enabled.
4043 $enoughToc = $this->mShowToc &&
4044 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4045
4046 # Allow user to stipulate that a page should have a "new section"
4047 # link added via __NEWSECTIONLINK__
4048 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4049 $this->mOutput->setNewSection( true );
4050 }
4051
4052 # Allow user to remove the "new section"
4053 # link via __NONEWSECTIONLINK__
4054 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4055 $this->mOutput->hideNewSection( true );
4056 }
4057
4058 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4059 # override above conditions and always show TOC above first header
4060 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4061 $this->mShowToc = true;
4062 $enoughToc = true;
4063 }
4064
4065 # headline counter
4066 $headlineCount = 0;
4067 $numVisible = 0;
4068
4069 # Ugh .. the TOC should have neat indentation levels which can be
4070 # passed to the skin functions. These are determined here
4071 $toc = '';
4072 $full = '';
4073 $head = [];
4074 $sublevelCount = [];
4075 $levelCount = [];
4076 $level = 0;
4077 $prevlevel = 0;
4078 $toclevel = 0;
4079 $prevtoclevel = 0;
4080 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4081 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4082 $oldType = $this->mOutputType;
4083 $this->setOutputType( self::OT_WIKI );
4084 $frame = $this->getPreprocessor()->newFrame();
4085 $root = $this->preprocessToDom( $origText );
4086 $node = $root->getFirstChild();
4087 $byteOffset = 0;
4088 $tocraw = [];
4089 $refers = [];
4090
4091 $headlines = $numMatches !== false ? $matches[3] : [];
4092
4093 foreach ( $headlines as $headline ) {
4094 $isTemplate = false;
4095 $titleText = false;
4096 $sectionIndex = false;
4097 $numbering = '';
4098 $markerMatches = [];
4099 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4100 $serial = $markerMatches[1];
4101 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4102 $isTemplate = ( $titleText != $baseTitleText );
4103 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4104 }
4105
4106 if ( $toclevel ) {
4107 $prevlevel = $level;
4108 }
4109 $level = $matches[1][$headlineCount];
4110
4111 if ( $level > $prevlevel ) {
4112 # Increase TOC level
4113 $toclevel++;
4114 $sublevelCount[$toclevel] = 0;
4115 if ( $toclevel < $wgMaxTocLevel ) {
4116 $prevtoclevel = $toclevel;
4117 $toc .= Linker::tocIndent();
4118 $numVisible++;
4119 }
4120 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4121 # Decrease TOC level, find level to jump to
4122
4123 for ( $i = $toclevel; $i > 0; $i-- ) {
4124 if ( $levelCount[$i] == $level ) {
4125 # Found last matching level
4126 $toclevel = $i;
4127 break;
4128 } elseif ( $levelCount[$i] < $level ) {
4129 # Found first matching level below current level
4130 $toclevel = $i + 1;
4131 break;
4132 }
4133 }
4134 if ( $i == 0 ) {
4135 $toclevel = 1;
4136 }
4137 if ( $toclevel < $wgMaxTocLevel ) {
4138 if ( $prevtoclevel < $wgMaxTocLevel ) {
4139 # Unindent only if the previous toc level was shown :p
4140 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4141 $prevtoclevel = $toclevel;
4142 } else {
4143 $toc .= Linker::tocLineEnd();
4144 }
4145 }
4146 } else {
4147 # No change in level, end TOC line
4148 if ( $toclevel < $wgMaxTocLevel ) {
4149 $toc .= Linker::tocLineEnd();
4150 }
4151 }
4152
4153 $levelCount[$toclevel] = $level;
4154
4155 # count number of headlines for each level
4156 $sublevelCount[$toclevel]++;
4157 $dot = 0;
4158 for ( $i = 1; $i <= $toclevel; $i++ ) {
4159 if ( !empty( $sublevelCount[$i] ) ) {
4160 if ( $dot ) {
4161 $numbering .= '.';
4162 }
4163 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4164 $dot = 1;
4165 }
4166 }
4167
4168 # The safe header is a version of the header text safe to use for links
4169
4170 # Remove link placeholders by the link text.
4171 # <!--LINK number-->
4172 # turns into
4173 # link text with suffix
4174 # Do this before unstrip since link text can contain strip markers
4175 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4176
4177 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4178 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4179
4180 # Strip out HTML (first regex removes any tag not allowed)
4181 # Allowed tags are:
4182 # * <sup> and <sub> (T10393)
4183 # * <i> (T28375)
4184 # * <b> (r105284)
4185 # * <bdi> (T74884)
4186 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4187 # * <s> and <strike> (T35715)
4188 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4189 # to allow setting directionality in toc items.
4190 $tocline = preg_replace(
4191 [
4192 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4193 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4194 ],
4195 [ '', '<$1>' ],
4196 $safeHeadline
4197 );
4198
4199 # Strip '<span></span>', which is the result from the above if
4200 # <span id="foo"></span> is used to produce an additional anchor
4201 # for a section.
4202 $tocline = str_replace( '<span></span>', '', $tocline );
4203
4204 $tocline = trim( $tocline );
4205
4206 # For the anchor, strip out HTML-y stuff period
4207 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4208 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4209
4210 # Save headline for section edit hint before it's escaped
4211 $headlineHint = $safeHeadline;
4212
4213 # Decode HTML entities
4214 $safeHeadline = Sanitizer::decodeCharReferences( $safeHeadline );
4215 $fallbackHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_FALLBACK );
4216 $linkAnchor = Sanitizer::escapeIdForLink( $safeHeadline );
4217 $safeHeadline = Sanitizer::escapeIdForAttribute( $safeHeadline, Sanitizer::ID_PRIMARY );
4218 if ( $fallbackHeadline === $safeHeadline ) {
4219 # No reason to have both (in fact, we can't)
4220 $fallbackHeadline = false;
4221 }
4222
4223 # HTML IDs must be case-insensitively unique for IE compatibility (T12721).
4224 # @todo FIXME: We may be changing them depending on the current locale.
4225 $arrayKey = strtolower( $safeHeadline );
4226 if ( $fallbackHeadline === false ) {
4227 $fallbackArrayKey = false;
4228 } else {
4229 $fallbackArrayKey = strtolower( $fallbackHeadline );
4230 }
4231
4232 # Create the anchor for linking from the TOC to the section
4233 $anchor = $safeHeadline;
4234 $fallbackAnchor = $fallbackHeadline;
4235 if ( isset( $refers[$arrayKey] ) ) {
4236 // @codingStandardsIgnoreStart
4237 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4238 // @codingStandardsIgnoreEnd
4239 $anchor .= "_$i";
4240 $linkAnchor .= "_$i";
4241 $refers["${arrayKey}_$i"] = true;
4242 } else {
4243 $refers[$arrayKey] = true;
4244 }
4245 if ( $fallbackHeadline !== false && isset( $refers[$fallbackArrayKey] ) ) {
4246 // @codingStandardsIgnoreStart
4247 for ( $i = 2; isset( $refers["${fallbackArrayKey}_$i"] ); ++$i );
4248 // @codingStandardsIgnoreEnd
4249 $fallbackAnchor .= "_$i";
4250 $refers["${fallbackArrayKey}_$i"] = true;
4251 } else {
4252 $refers[$fallbackArrayKey] = true;
4253 }
4254
4255 # Don't number the heading if it is the only one (looks silly)
4256 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4257 # the two are different if the line contains a link
4258 $headline = Html::element(
4259 'span',
4260 [ 'class' => 'mw-headline-number' ],
4261 $numbering
4262 ) . ' ' . $headline;
4263 }
4264
4265 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4266 $toc .= Linker::tocLine( $linkAnchor, $tocline,
4267 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4268 }
4269
4270 # Add the section to the section tree
4271 # Find the DOM node for this header
4272 $noOffset = ( $isTemplate || $sectionIndex === false );
4273 while ( $node && !$noOffset ) {
4274 if ( $node->getName() === 'h' ) {
4275 $bits = $node->splitHeading();
4276 if ( $bits['i'] == $sectionIndex ) {
4277 break;
4278 }
4279 }
4280 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4281 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4282 $node = $node->getNextSibling();
4283 }
4284 $tocraw[] = [
4285 'toclevel' => $toclevel,
4286 'level' => $level,
4287 'line' => $tocline,
4288 'number' => $numbering,
4289 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4290 'fromtitle' => $titleText,
4291 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4292 'anchor' => $anchor,
4293 ];
4294
4295 # give headline the correct <h#> tag
4296 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4297 // Output edit section links as markers with styles that can be customized by skins
4298 if ( $isTemplate ) {
4299 # Put a T flag in the section identifier, to indicate to extractSections()
4300 # that sections inside <includeonly> should be counted.
4301 $editsectionPage = $titleText;
4302 $editsectionSection = "T-$sectionIndex";
4303 $editsectionContent = null;
4304 } else {
4305 $editsectionPage = $this->mTitle->getPrefixedText();
4306 $editsectionSection = $sectionIndex;
4307 $editsectionContent = $headlineHint;
4308 }
4309 // We use a bit of pesudo-xml for editsection markers. The
4310 // language converter is run later on. Using a UNIQ style marker
4311 // leads to the converter screwing up the tokens when it
4312 // converts stuff. And trying to insert strip tags fails too. At
4313 // this point all real inputted tags have already been escaped,
4314 // so we don't have to worry about a user trying to input one of
4315 // these markers directly. We use a page and section attribute
4316 // to stop the language converter from converting these
4317 // important bits of data, but put the headline hint inside a
4318 // content block because the language converter is supposed to
4319 // be able to convert that piece of data.
4320 // Gets replaced with html in ParserOutput::getText
4321 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4322 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4323 if ( $editsectionContent !== null ) {
4324 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4325 } else {
4326 $editlink .= '/>';
4327 }
4328 } else {
4329 $editlink = '';
4330 }
4331 $head[$headlineCount] = Linker::makeHeadline( $level,
4332 $matches['attrib'][$headlineCount], $anchor, $headline,
4333 $editlink, $fallbackAnchor );
4334
4335 $headlineCount++;
4336 }
4337
4338 $this->setOutputType( $oldType );
4339
4340 # Never ever show TOC if no headers
4341 if ( $numVisible < 1 ) {
4342 $enoughToc = false;
4343 }
4344
4345 if ( $enoughToc ) {
4346 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4347 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4348 }
4349 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4350 $this->mOutput->setTOCHTML( $toc );
4351 $toc = self::TOC_START . $toc . self::TOC_END;
4352 }
4353
4354 if ( $isMain ) {
4355 $this->mOutput->setSections( $tocraw );
4356 }
4357
4358 # split up and insert constructed headlines
4359 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4360 $i = 0;
4361
4362 // build an array of document sections
4363 $sections = [];
4364 foreach ( $blocks as $block ) {
4365 // $head is zero-based, sections aren't.
4366 if ( empty( $head[$i - 1] ) ) {
4367 $sections[$i] = $block;
4368 } else {
4369 $sections[$i] = $head[$i - 1] . $block;
4370 }
4371
4372 /**
4373 * Send a hook, one per section.
4374 * The idea here is to be able to make section-level DIVs, but to do so in a
4375 * lower-impact, more correct way than r50769
4376 *
4377 * $this : caller
4378 * $section : the section number
4379 * &$sectionContent : ref to the content of the section
4380 * $showEditLinks : boolean describing whether this section has an edit link
4381 */
4382 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4383
4384 $i++;
4385 }
4386
4387 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4388 // append the TOC at the beginning
4389 // Top anchor now in skin
4390 $sections[0] = $sections[0] . $toc . "\n";
4391 }
4392
4393 $full .= implode( '', $sections );
4394
4395 if ( $this->mForceTocPosition ) {
4396 return str_replace( '<!--MWTOC-->', $toc, $full );
4397 } else {
4398 return $full;
4399 }
4400 }
4401
4402 /**
4403 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4404 * conversion, substituting signatures, {{subst:}} templates, etc.
4405 *
4406 * @param string $text The text to transform
4407 * @param Title $title The Title object for the current article
4408 * @param User $user The User object describing the current user
4409 * @param ParserOptions $options Parsing options
4410 * @param bool $clearState Whether to clear the parser state first
4411 * @return string The altered wiki markup
4412 */
4413 public function preSaveTransform( $text, Title $title, User $user,
4414 ParserOptions $options, $clearState = true
4415 ) {
4416 if ( $clearState ) {
4417 $magicScopeVariable = $this->lock();
4418 }
4419 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4420 $this->setUser( $user );
4421
4422 // Strip U+0000 NULL (T159174)
4423 $text = str_replace( "\000", '', $text );
4424
4425 // We still normalize line endings for backwards-compatibility
4426 // with other code that just calls PST, but this should already
4427 // be handled in TextContent subclasses
4428 $text = TextContent::normalizeLineEndings( $text );
4429
4430 if ( $options->getPreSaveTransform() ) {
4431 $text = $this->pstPass2( $text, $user );
4432 }
4433 $text = $this->mStripState->unstripBoth( $text );
4434
4435 $this->setUser( null ); # Reset
4436
4437 return $text;
4438 }
4439
4440 /**
4441 * Pre-save transform helper function
4442 *
4443 * @param string $text
4444 * @param User $user
4445 *
4446 * @return string
4447 */
4448 private function pstPass2( $text, $user ) {
4449 global $wgContLang;
4450
4451 # Note: This is the timestamp saved as hardcoded wikitext to
4452 # the database, we use $wgContLang here in order to give
4453 # everyone the same signature and use the default one rather
4454 # than the one selected in each user's preferences.
4455 # (see also T14815)
4456 $ts = $this->mOptions->getTimestamp();
4457 $timestamp = MWTimestamp::getLocalInstance( $ts );
4458 $ts = $timestamp->format( 'YmdHis' );
4459 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4460
4461 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4462
4463 # Variable replacement
4464 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4465 $text = $this->replaceVariables( $text );
4466
4467 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4468 # which may corrupt this parser instance via its wfMessage()->text() call-
4469
4470 # Signatures
4471 if ( strpos( $text, '~~~' ) !== false ) {
4472 $sigText = $this->getUserSig( $user );
4473 $text = strtr( $text, [
4474 '~~~~~' => $d,
4475 '~~~~' => "$sigText $d",
4476 '~~~' => $sigText
4477 ] );
4478 # The main two signature forms used above are time-sensitive
4479 $this->mOutput->setFlag( 'user-signature' );
4480 }
4481
4482 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4483 $tc = '[' . Title::legalChars() . ']';
4484 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4485
4486 // [[ns:page (context)|]]
4487 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4488 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4489 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4490 // [[ns:page (context), context|]] (using either single or double-width comma)
4491 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4492 // [[|page]] (reverse pipe trick: add context from page title)
4493 $p2 = "/\[\[\\|($tc+)]]/";
4494
4495 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4496 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4497 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4498 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4499
4500 $t = $this->mTitle->getText();
4501 $m = [];
4502 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4503 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4504 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4505 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4506 } else {
4507 # if there's no context, don't bother duplicating the title
4508 $text = preg_replace( $p2, '[[\\1]]', $text );
4509 }
4510
4511 return $text;
4512 }
4513
4514 /**
4515 * Fetch the user's signature text, if any, and normalize to
4516 * validated, ready-to-insert wikitext.
4517 * If you have pre-fetched the nickname or the fancySig option, you can
4518 * specify them here to save a database query.
4519 * Do not reuse this parser instance after calling getUserSig(),
4520 * as it may have changed if it's the $wgParser.
4521 *
4522 * @param User &$user
4523 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4524 * @param bool|null $fancySig whether the nicknname is the complete signature
4525 * or null to use default value
4526 * @return string
4527 */
4528 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4529 global $wgMaxSigChars;
4530
4531 $username = $user->getName();
4532
4533 # If not given, retrieve from the user object.
4534 if ( $nickname === false ) {
4535 $nickname = $user->getOption( 'nickname' );
4536 }
4537
4538 if ( is_null( $fancySig ) ) {
4539 $fancySig = $user->getBoolOption( 'fancysig' );
4540 }
4541
4542 $nickname = $nickname == null ? $username : $nickname;
4543
4544 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4545 $nickname = $username;
4546 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4547 } elseif ( $fancySig !== false ) {
4548 # Sig. might contain markup; validate this
4549 if ( $this->validateSig( $nickname ) !== false ) {
4550 # Validated; clean up (if needed) and return it
4551 return $this->cleanSig( $nickname, true );
4552 } else {
4553 # Failed to validate; fall back to the default
4554 $nickname = $username;
4555 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4556 }
4557 }
4558
4559 # Make sure nickname doesnt get a sig in a sig
4560 $nickname = self::cleanSigInSig( $nickname );
4561
4562 # If we're still here, make it a link to the user page
4563 $userText = wfEscapeWikiText( $username );
4564 $nickText = wfEscapeWikiText( $nickname );
4565 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4566
4567 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4568 ->title( $this->getTitle() )->text();
4569 }
4570
4571 /**
4572 * Check that the user's signature contains no bad XML
4573 *
4574 * @param string $text
4575 * @return string|bool An expanded string, or false if invalid.
4576 */
4577 public function validateSig( $text ) {
4578 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4579 }
4580
4581 /**
4582 * Clean up signature text
4583 *
4584 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4585 * 2) Substitute all transclusions
4586 *
4587 * @param string $text
4588 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4589 * @return string Signature text
4590 */
4591 public function cleanSig( $text, $parsing = false ) {
4592 if ( !$parsing ) {
4593 global $wgTitle;
4594 $magicScopeVariable = $this->lock();
4595 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4596 }
4597
4598 # Option to disable this feature
4599 if ( !$this->mOptions->getCleanSignatures() ) {
4600 return $text;
4601 }
4602
4603 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4604 # => Move this logic to braceSubstitution()
4605 $substWord = MagicWord::get( 'subst' );
4606 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4607 $substText = '{{' . $substWord->getSynonym( 0 );
4608
4609 $text = preg_replace( $substRegex, $substText, $text );
4610 $text = self::cleanSigInSig( $text );
4611 $dom = $this->preprocessToDom( $text );
4612 $frame = $this->getPreprocessor()->newFrame();
4613 $text = $frame->expand( $dom );
4614
4615 if ( !$parsing ) {
4616 $text = $this->mStripState->unstripBoth( $text );
4617 }
4618
4619 return $text;
4620 }
4621
4622 /**
4623 * Strip 3, 4 or 5 tildes out of signatures.
4624 *
4625 * @param string $text
4626 * @return string Signature text with /~{3,5}/ removed
4627 */
4628 public static function cleanSigInSig( $text ) {
4629 $text = preg_replace( '/~{3,5}/', '', $text );
4630 return $text;
4631 }
4632
4633 /**
4634 * Set up some variables which are usually set up in parse()
4635 * so that an external function can call some class members with confidence
4636 *
4637 * @param Title|null $title
4638 * @param ParserOptions $options
4639 * @param int $outputType
4640 * @param bool $clearState
4641 */
4642 public function startExternalParse( Title $title = null, ParserOptions $options,
4643 $outputType, $clearState = true
4644 ) {
4645 $this->startParse( $title, $options, $outputType, $clearState );
4646 }
4647
4648 /**
4649 * @param Title|null $title
4650 * @param ParserOptions $options
4651 * @param int $outputType
4652 * @param bool $clearState
4653 */
4654 private function startParse( Title $title = null, ParserOptions $options,
4655 $outputType, $clearState = true
4656 ) {
4657 $this->setTitle( $title );
4658 $this->mOptions = $options;
4659 $this->setOutputType( $outputType );
4660 if ( $clearState ) {
4661 $this->clearState();
4662 }
4663 }
4664
4665 /**
4666 * Wrapper for preprocess()
4667 *
4668 * @param string $text The text to preprocess
4669 * @param ParserOptions $options Options
4670 * @param Title|null $title Title object or null to use $wgTitle
4671 * @return string
4672 */
4673 public function transformMsg( $text, $options, $title = null ) {
4674 static $executing = false;
4675
4676 # Guard against infinite recursion
4677 if ( $executing ) {
4678 return $text;
4679 }
4680 $executing = true;
4681
4682 if ( !$title ) {
4683 global $wgTitle;
4684 $title = $wgTitle;
4685 }
4686
4687 $text = $this->preprocess( $text, $title, $options );
4688
4689 $executing = false;
4690 return $text;
4691 }
4692
4693 /**
4694 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4695 * The callback should have the following form:
4696 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4697 *
4698 * Transform and return $text. Use $parser for any required context, e.g. use
4699 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4700 *
4701 * Hooks may return extended information by returning an array, of which the
4702 * first numbered element (index 0) must be the return string, and all other
4703 * entries are extracted into local variables within an internal function
4704 * in the Parser class.
4705 *
4706 * This interface (introduced r61913) appears to be undocumented, but
4707 * 'markerType' is used by some core tag hooks to override which strip
4708 * array their results are placed in. **Use great caution if attempting
4709 * this interface, as it is not documented and injudicious use could smash
4710 * private variables.**
4711 *
4712 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4713 * @param callable $callback The callback function (and object) to use for the tag
4714 * @throws MWException
4715 * @return callable|null The old value of the mTagHooks array associated with the hook
4716 */
4717 public function setHook( $tag, callable $callback ) {
4718 $tag = strtolower( $tag );
4719 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4720 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4721 }
4722 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4723 $this->mTagHooks[$tag] = $callback;
4724 if ( !in_array( $tag, $this->mStripList ) ) {
4725 $this->mStripList[] = $tag;
4726 }
4727
4728 return $oldVal;
4729 }
4730
4731 /**
4732 * As setHook(), but letting the contents be parsed.
4733 *
4734 * Transparent tag hooks are like regular XML-style tag hooks, except they
4735 * operate late in the transformation sequence, on HTML instead of wikitext.
4736 *
4737 * This is probably obsoleted by things dealing with parser frames?
4738 * The only extension currently using it is geoserver.
4739 *
4740 * @since 1.10
4741 * @todo better document or deprecate this
4742 *
4743 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4744 * @param callable $callback The callback function (and object) to use for the tag
4745 * @throws MWException
4746 * @return callable|null The old value of the mTagHooks array associated with the hook
4747 */
4748 public function setTransparentTagHook( $tag, callable $callback ) {
4749 $tag = strtolower( $tag );
4750 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4751 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4752 }
4753 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4754 $this->mTransparentTagHooks[$tag] = $callback;
4755
4756 return $oldVal;
4757 }
4758
4759 /**
4760 * Remove all tag hooks
4761 */
4762 public function clearTagHooks() {
4763 $this->mTagHooks = [];
4764 $this->mFunctionTagHooks = [];
4765 $this->mStripList = $this->mDefaultStripList;
4766 }
4767
4768 /**
4769 * Create a function, e.g. {{sum:1|2|3}}
4770 * The callback function should have the form:
4771 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4772 *
4773 * Or with Parser::SFH_OBJECT_ARGS:
4774 * function myParserFunction( $parser, $frame, $args ) { ... }
4775 *
4776 * The callback may either return the text result of the function, or an array with the text
4777 * in element 0, and a number of flags in the other elements. The names of the flags are
4778 * specified in the keys. Valid flags are:
4779 * found The text returned is valid, stop processing the template. This
4780 * is on by default.
4781 * nowiki Wiki markup in the return value should be escaped
4782 * isHTML The returned text is HTML, armour it against wikitext transformation
4783 *
4784 * @param string $id The magic word ID
4785 * @param callable $callback The callback function (and object) to use
4786 * @param int $flags A combination of the following flags:
4787 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4788 *
4789 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4790 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4791 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4792 * the arguments, and to control the way they are expanded.
4793 *
4794 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4795 * arguments, for instance:
4796 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4797 *
4798 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4799 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4800 * working if/when this is changed.
4801 *
4802 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4803 * expansion.
4804 *
4805 * Please read the documentation in includes/parser/Preprocessor.php for more information
4806 * about the methods available in PPFrame and PPNode.
4807 *
4808 * @throws MWException
4809 * @return string|callable The old callback function for this name, if any
4810 */
4811 public function setFunctionHook( $id, callable $callback, $flags = 0 ) {
4812 global $wgContLang;
4813
4814 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4815 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4816
4817 # Add to function cache
4818 $mw = MagicWord::get( $id );
4819 if ( !$mw ) {
4820 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4821 }
4822
4823 $synonyms = $mw->getSynonyms();
4824 $sensitive = intval( $mw->isCaseSensitive() );
4825
4826 foreach ( $synonyms as $syn ) {
4827 # Case
4828 if ( !$sensitive ) {
4829 $syn = $wgContLang->lc( $syn );
4830 }
4831 # Add leading hash
4832 if ( !( $flags & self::SFH_NO_HASH ) ) {
4833 $syn = '#' . $syn;
4834 }
4835 # Remove trailing colon
4836 if ( substr( $syn, -1, 1 ) === ':' ) {
4837 $syn = substr( $syn, 0, -1 );
4838 }
4839 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4840 }
4841 return $oldVal;
4842 }
4843
4844 /**
4845 * Get all registered function hook identifiers
4846 *
4847 * @return array
4848 */
4849 public function getFunctionHooks() {
4850 return array_keys( $this->mFunctionHooks );
4851 }
4852
4853 /**
4854 * Create a tag function, e.g. "<test>some stuff</test>".
4855 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4856 * Unlike parser functions, their content is not preprocessed.
4857 * @param string $tag
4858 * @param callable $callback
4859 * @param int $flags
4860 * @throws MWException
4861 * @return null
4862 */
4863 public function setFunctionTagHook( $tag, callable $callback, $flags ) {
4864 $tag = strtolower( $tag );
4865 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4866 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4867 }
4868 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4869 $this->mFunctionTagHooks[$tag] : null;
4870 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4871
4872 if ( !in_array( $tag, $this->mStripList ) ) {
4873 $this->mStripList[] = $tag;
4874 }
4875
4876 return $old;
4877 }
4878
4879 /**
4880 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4881 * Placeholders created in Linker::link()
4882 *
4883 * @param string &$text
4884 * @param int $options
4885 */
4886 public function replaceLinkHolders( &$text, $options = 0 ) {
4887 $this->mLinkHolders->replace( $text );
4888 }
4889
4890 /**
4891 * Replace "<!--LINK-->" link placeholders with plain text of links
4892 * (not HTML-formatted).
4893 *
4894 * @param string $text
4895 * @return string
4896 */
4897 public function replaceLinkHoldersText( $text ) {
4898 return $this->mLinkHolders->replaceText( $text );
4899 }
4900
4901 /**
4902 * Renders an image gallery from a text with one line per image.
4903 * text labels may be given by using |-style alternative text. E.g.
4904 * Image:one.jpg|The number "1"
4905 * Image:tree.jpg|A tree
4906 * given as text will return the HTML of a gallery with two images,
4907 * labeled 'The number "1"' and
4908 * 'A tree'.
4909 *
4910 * @param string $text
4911 * @param array $params
4912 * @return string HTML
4913 */
4914 public function renderImageGallery( $text, $params ) {
4915 $mode = false;
4916 if ( isset( $params['mode'] ) ) {
4917 $mode = $params['mode'];
4918 }
4919
4920 try {
4921 $ig = ImageGalleryBase::factory( $mode );
4922 } catch ( Exception $e ) {
4923 // If invalid type set, fallback to default.
4924 $ig = ImageGalleryBase::factory( false );
4925 }
4926
4927 $ig->setContextTitle( $this->mTitle );
4928 $ig->setShowBytes( false );
4929 $ig->setShowDimensions( false );
4930 $ig->setShowFilename( false );
4931 $ig->setParser( $this );
4932 $ig->setHideBadImages();
4933 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4934
4935 if ( isset( $params['showfilename'] ) ) {
4936 $ig->setShowFilename( true );
4937 } else {
4938 $ig->setShowFilename( false );
4939 }
4940 if ( isset( $params['caption'] ) ) {
4941 $caption = $params['caption'];
4942 $caption = htmlspecialchars( $caption );
4943 $caption = $this->replaceInternalLinks( $caption );
4944 $ig->setCaptionHtml( $caption );
4945 }
4946 if ( isset( $params['perrow'] ) ) {
4947 $ig->setPerRow( $params['perrow'] );
4948 }
4949 if ( isset( $params['widths'] ) ) {
4950 $ig->setWidths( $params['widths'] );
4951 }
4952 if ( isset( $params['heights'] ) ) {
4953 $ig->setHeights( $params['heights'] );
4954 }
4955 $ig->setAdditionalOptions( $params );
4956
4957 // Avoid PHP 7.1 warning from passing $this by reference
4958 $parser = $this;
4959 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
4960
4961 $lines = StringUtils::explode( "\n", $text );
4962 foreach ( $lines as $line ) {
4963 # match lines like these:
4964 # Image:someimage.jpg|This is some image
4965 $matches = [];
4966 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4967 # Skip empty lines
4968 if ( count( $matches ) == 0 ) {
4969 continue;
4970 }
4971
4972 if ( strpos( $matches[0], '%' ) !== false ) {
4973 $matches[1] = rawurldecode( $matches[1] );
4974 }
4975 $title = Title::newFromText( $matches[1], NS_FILE );
4976 if ( is_null( $title ) ) {
4977 # Bogus title. Ignore these so we don't bomb out later.
4978 continue;
4979 }
4980
4981 # We need to get what handler the file uses, to figure out parameters.
4982 # Note, a hook can overide the file name, and chose an entirely different
4983 # file (which potentially could be of a different type and have different handler).
4984 $options = [];
4985 $descQuery = false;
4986 Hooks::run( 'BeforeParserFetchFileAndTitle',
4987 [ $this, $title, &$options, &$descQuery ] );
4988 # Don't register it now, as TraditionalImageGallery does that later.
4989 $file = $this->fetchFileNoRegister( $title, $options );
4990 $handler = $file ? $file->getHandler() : false;
4991
4992 $paramMap = [
4993 'img_alt' => 'gallery-internal-alt',
4994 'img_link' => 'gallery-internal-link',
4995 ];
4996 if ( $handler ) {
4997 $paramMap = $paramMap + $handler->getParamMap();
4998 // We don't want people to specify per-image widths.
4999 // Additionally the width parameter would need special casing anyhow.
5000 unset( $paramMap['img_width'] );
5001 }
5002
5003 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5004
5005 $label = '';
5006 $alt = '';
5007 $link = '';
5008 $handlerOptions = [];
5009 if ( isset( $matches[3] ) ) {
5010 // look for an |alt= definition while trying not to break existing
5011 // captions with multiple pipes (|) in it, until a more sensible grammar
5012 // is defined for images in galleries
5013
5014 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5015 // splitting on '|' is a bit odd, and different from makeImage.
5016 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5017 // Protect LanguageConverter markup
5018 $parameterMatches = StringUtils::delimiterExplode(
5019 '-{', '}-', '|', $matches[3], true /* nested */
5020 );
5021
5022 foreach ( $parameterMatches as $parameterMatch ) {
5023 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5024 if ( $magicName ) {
5025 $paramName = $paramMap[$magicName];
5026
5027 switch ( $paramName ) {
5028 case 'gallery-internal-alt':
5029 $alt = $this->stripAltText( $match, false );
5030 break;
5031 case 'gallery-internal-link':
5032 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5033 $chars = self::EXT_LINK_URL_CLASS;
5034 $addr = self::EXT_LINK_ADDR;
5035 $prots = $this->mUrlProtocols;
5036 // check to see if link matches an absolute url, if not then it must be a wiki link.
5037 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5038 // Result of LanguageConverter::markNoConversion
5039 // invoked on an external link.
5040 $linkValue = substr( $linkValue, 4, -2 );
5041 }
5042 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5043 $link = $linkValue;
5044 $this->mOutput->addExternalLink( $link );
5045 } else {
5046 $localLinkTitle = Title::newFromText( $linkValue );
5047 if ( $localLinkTitle !== null ) {
5048 $this->mOutput->addLink( $localLinkTitle );
5049 $link = $localLinkTitle->getLinkURL();
5050 }
5051 }
5052 break;
5053 default:
5054 // Must be a handler specific parameter.
5055 if ( $handler->validateParam( $paramName, $match ) ) {
5056 $handlerOptions[$paramName] = $match;
5057 } else {
5058 // Guess not, consider it as caption.
5059 wfDebug( "$parameterMatch failed parameter validation\n" );
5060 $label = '|' . $parameterMatch;
5061 }
5062 }
5063
5064 } else {
5065 // Last pipe wins.
5066 $label = '|' . $parameterMatch;
5067 }
5068 }
5069 // Remove the pipe.
5070 $label = substr( $label, 1 );
5071 }
5072
5073 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5074 }
5075 $html = $ig->toHTML();
5076 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5077 return $html;
5078 }
5079
5080 /**
5081 * @param MediaHandler $handler
5082 * @return array
5083 */
5084 public function getImageParams( $handler ) {
5085 if ( $handler ) {
5086 $handlerClass = get_class( $handler );
5087 } else {
5088 $handlerClass = '';
5089 }
5090 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5091 # Initialise static lists
5092 static $internalParamNames = [
5093 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5094 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5095 'bottom', 'text-bottom' ],
5096 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5097 'upright', 'border', 'link', 'alt', 'class' ],
5098 ];
5099 static $internalParamMap;
5100 if ( !$internalParamMap ) {
5101 $internalParamMap = [];
5102 foreach ( $internalParamNames as $type => $names ) {
5103 foreach ( $names as $name ) {
5104 // For grep: img_left, img_right, img_center, img_none,
5105 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5106 // img_bottom, img_text_bottom,
5107 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5108 // img_border, img_link, img_alt, img_class
5109 $magicName = str_replace( '-', '_', "img_$name" );
5110 $internalParamMap[$magicName] = [ $type, $name ];
5111 }
5112 }
5113 }
5114
5115 # Add handler params
5116 $paramMap = $internalParamMap;
5117 if ( $handler ) {
5118 $handlerParamMap = $handler->getParamMap();
5119 foreach ( $handlerParamMap as $magic => $paramName ) {
5120 $paramMap[$magic] = [ 'handler', $paramName ];
5121 }
5122 }
5123 $this->mImageParams[$handlerClass] = $paramMap;
5124 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5125 }
5126 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5127 }
5128
5129 /**
5130 * Parse image options text and use it to make an image
5131 *
5132 * @param Title $title
5133 * @param string $options
5134 * @param LinkHolderArray|bool $holders
5135 * @return string HTML
5136 */
5137 public function makeImage( $title, $options, $holders = false ) {
5138 # Check if the options text is of the form "options|alt text"
5139 # Options are:
5140 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5141 # * left no resizing, just left align. label is used for alt= only
5142 # * right same, but right aligned
5143 # * none same, but not aligned
5144 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5145 # * center center the image
5146 # * frame Keep original image size, no magnify-button.
5147 # * framed Same as "frame"
5148 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5149 # * upright reduce width for upright images, rounded to full __0 px
5150 # * border draw a 1px border around the image
5151 # * alt Text for HTML alt attribute (defaults to empty)
5152 # * class Set a class for img node
5153 # * link Set the target of the image link. Can be external, interwiki, or local
5154 # vertical-align values (no % or length right now):
5155 # * baseline
5156 # * sub
5157 # * super
5158 # * top
5159 # * text-top
5160 # * middle
5161 # * bottom
5162 # * text-bottom
5163
5164 # Protect LanguageConverter markup when splitting into parts
5165 $parts = StringUtils::delimiterExplode(
5166 '-{', '}-', '|', $options, true /* allow nesting */
5167 );
5168
5169 # Give extensions a chance to select the file revision for us
5170 $options = [];
5171 $descQuery = false;
5172 Hooks::run( 'BeforeParserFetchFileAndTitle',
5173 [ $this, $title, &$options, &$descQuery ] );
5174 # Fetch and register the file (file title may be different via hooks)
5175 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5176
5177 # Get parameter map
5178 $handler = $file ? $file->getHandler() : false;
5179
5180 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5181
5182 if ( !$file ) {
5183 $this->addTrackingCategory( 'broken-file-category' );
5184 }
5185
5186 # Process the input parameters
5187 $caption = '';
5188 $params = [ 'frame' => [], 'handler' => [],
5189 'horizAlign' => [], 'vertAlign' => [] ];
5190 $seenformat = false;
5191 foreach ( $parts as $part ) {
5192 $part = trim( $part );
5193 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5194 $validated = false;
5195 if ( isset( $paramMap[$magicName] ) ) {
5196 list( $type, $paramName ) = $paramMap[$magicName];
5197
5198 # Special case; width and height come in one variable together
5199 if ( $type === 'handler' && $paramName === 'width' ) {
5200 $parsedWidthParam = $this->parseWidthParam( $value );
5201 if ( isset( $parsedWidthParam['width'] ) ) {
5202 $width = $parsedWidthParam['width'];
5203 if ( $handler->validateParam( 'width', $width ) ) {
5204 $params[$type]['width'] = $width;
5205 $validated = true;
5206 }
5207 }
5208 if ( isset( $parsedWidthParam['height'] ) ) {
5209 $height = $parsedWidthParam['height'];
5210 if ( $handler->validateParam( 'height', $height ) ) {
5211 $params[$type]['height'] = $height;
5212 $validated = true;
5213 }
5214 }
5215 # else no validation -- T15436
5216 } else {
5217 if ( $type === 'handler' ) {
5218 # Validate handler parameter
5219 $validated = $handler->validateParam( $paramName, $value );
5220 } else {
5221 # Validate internal parameters
5222 switch ( $paramName ) {
5223 case 'manualthumb':
5224 case 'alt':
5225 case 'class':
5226 # @todo FIXME: Possibly check validity here for
5227 # manualthumb? downstream behavior seems odd with
5228 # missing manual thumbs.
5229 $validated = true;
5230 $value = $this->stripAltText( $value, $holders );
5231 break;
5232 case 'link':
5233 $chars = self::EXT_LINK_URL_CLASS;
5234 $addr = self::EXT_LINK_ADDR;
5235 $prots = $this->mUrlProtocols;
5236 if ( $value === '' ) {
5237 $paramName = 'no-link';
5238 $value = true;
5239 $validated = true;
5240 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5241 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5242 $paramName = 'link-url';
5243 $this->mOutput->addExternalLink( $value );
5244 if ( $this->mOptions->getExternalLinkTarget() ) {
5245 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5246 }
5247 $validated = true;
5248 }
5249 } else {
5250 $linkTitle = Title::newFromText( $value );
5251 if ( $linkTitle ) {
5252 $paramName = 'link-title';
5253 $value = $linkTitle;
5254 $this->mOutput->addLink( $linkTitle );
5255 $validated = true;
5256 }
5257 }
5258 break;
5259 case 'frameless':
5260 case 'framed':
5261 case 'thumbnail':
5262 // use first appearing option, discard others.
5263 $validated = !$seenformat;
5264 $seenformat = true;
5265 break;
5266 default:
5267 # Most other things appear to be empty or numeric...
5268 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5269 }
5270 }
5271
5272 if ( $validated ) {
5273 $params[$type][$paramName] = $value;
5274 }
5275 }
5276 }
5277 if ( !$validated ) {
5278 $caption = $part;
5279 }
5280 }
5281
5282 # Process alignment parameters
5283 if ( $params['horizAlign'] ) {
5284 $params['frame']['align'] = key( $params['horizAlign'] );
5285 }
5286 if ( $params['vertAlign'] ) {
5287 $params['frame']['valign'] = key( $params['vertAlign'] );
5288 }
5289
5290 $params['frame']['caption'] = $caption;
5291
5292 # Will the image be presented in a frame, with the caption below?
5293 $imageIsFramed = isset( $params['frame']['frame'] )
5294 || isset( $params['frame']['framed'] )
5295 || isset( $params['frame']['thumbnail'] )
5296 || isset( $params['frame']['manualthumb'] );
5297
5298 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5299 # came to also set the caption, ordinary text after the image -- which
5300 # makes no sense, because that just repeats the text multiple times in
5301 # screen readers. It *also* came to set the title attribute.
5302 # Now that we have an alt attribute, we should not set the alt text to
5303 # equal the caption: that's worse than useless, it just repeats the
5304 # text. This is the framed/thumbnail case. If there's no caption, we
5305 # use the unnamed parameter for alt text as well, just for the time be-
5306 # ing, if the unnamed param is set and the alt param is not.
5307 # For the future, we need to figure out if we want to tweak this more,
5308 # e.g., introducing a title= parameter for the title; ignoring the un-
5309 # named parameter entirely for images without a caption; adding an ex-
5310 # plicit caption= parameter and preserving the old magic unnamed para-
5311 # meter for BC; ...
5312 if ( $imageIsFramed ) { # Framed image
5313 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5314 # No caption or alt text, add the filename as the alt text so
5315 # that screen readers at least get some description of the image
5316 $params['frame']['alt'] = $title->getText();
5317 }
5318 # Do not set $params['frame']['title'] because tooltips don't make sense
5319 # for framed images
5320 } else { # Inline image
5321 if ( !isset( $params['frame']['alt'] ) ) {
5322 # No alt text, use the "caption" for the alt text
5323 if ( $caption !== '' ) {
5324 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5325 } else {
5326 # No caption, fall back to using the filename for the
5327 # alt text
5328 $params['frame']['alt'] = $title->getText();
5329 }
5330 }
5331 # Use the "caption" for the tooltip text
5332 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5333 }
5334
5335 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5336
5337 # Linker does the rest
5338 $time = isset( $options['time'] ) ? $options['time'] : false;
5339 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5340 $time, $descQuery, $this->mOptions->getThumbSize() );
5341
5342 # Give the handler a chance to modify the parser object
5343 if ( $handler ) {
5344 $handler->parserTransformHook( $this, $file );
5345 }
5346
5347 return $ret;
5348 }
5349
5350 /**
5351 * @param string $caption
5352 * @param LinkHolderArray|bool $holders
5353 * @return mixed|string
5354 */
5355 protected function stripAltText( $caption, $holders ) {
5356 # Strip bad stuff out of the title (tooltip). We can't just use
5357 # replaceLinkHoldersText() here, because if this function is called
5358 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5359 if ( $holders ) {
5360 $tooltip = $holders->replaceText( $caption );
5361 } else {
5362 $tooltip = $this->replaceLinkHoldersText( $caption );
5363 }
5364
5365 # make sure there are no placeholders in thumbnail attributes
5366 # that are later expanded to html- so expand them now and
5367 # remove the tags
5368 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5369 $tooltip = Sanitizer::stripAllTags( $tooltip );
5370
5371 return $tooltip;
5372 }
5373
5374 /**
5375 * Set a flag in the output object indicating that the content is dynamic and
5376 * shouldn't be cached.
5377 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5378 */
5379 public function disableCache() {
5380 wfDebug( "Parser output marked as uncacheable.\n" );
5381 if ( !$this->mOutput ) {
5382 throw new MWException( __METHOD__ .
5383 " can only be called when actually parsing something" );
5384 }
5385 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5386 }
5387
5388 /**
5389 * Callback from the Sanitizer for expanding items found in HTML attribute
5390 * values, so they can be safely tested and escaped.
5391 *
5392 * @param string &$text
5393 * @param bool|PPFrame $frame
5394 * @return string
5395 */
5396 public function attributeStripCallback( &$text, $frame = false ) {
5397 $text = $this->replaceVariables( $text, $frame );
5398 $text = $this->mStripState->unstripBoth( $text );
5399 return $text;
5400 }
5401
5402 /**
5403 * Accessor
5404 *
5405 * @return array
5406 */
5407 public function getTags() {
5408 return array_merge(
5409 array_keys( $this->mTransparentTagHooks ),
5410 array_keys( $this->mTagHooks ),
5411 array_keys( $this->mFunctionTagHooks )
5412 );
5413 }
5414
5415 /**
5416 * Replace transparent tags in $text with the values given by the callbacks.
5417 *
5418 * Transparent tag hooks are like regular XML-style tag hooks, except they
5419 * operate late in the transformation sequence, on HTML instead of wikitext.
5420 *
5421 * @param string $text
5422 *
5423 * @return string
5424 */
5425 public function replaceTransparentTags( $text ) {
5426 $matches = [];
5427 $elements = array_keys( $this->mTransparentTagHooks );
5428 $text = self::extractTagsAndParams( $elements, $text, $matches );
5429 $replacements = [];
5430
5431 foreach ( $matches as $marker => $data ) {
5432 list( $element, $content, $params, $tag ) = $data;
5433 $tagName = strtolower( $element );
5434 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5435 $output = call_user_func_array(
5436 $this->mTransparentTagHooks[$tagName],
5437 [ $content, $params, $this ]
5438 );
5439 } else {
5440 $output = $tag;
5441 }
5442 $replacements[$marker] = $output;
5443 }
5444 return strtr( $text, $replacements );
5445 }
5446
5447 /**
5448 * Break wikitext input into sections, and either pull or replace
5449 * some particular section's text.
5450 *
5451 * External callers should use the getSection and replaceSection methods.
5452 *
5453 * @param string $text Page wikitext
5454 * @param string|int $sectionId A section identifier string of the form:
5455 * "<flag1> - <flag2> - ... - <section number>"
5456 *
5457 * Currently the only recognised flag is "T", which means the target section number
5458 * was derived during a template inclusion parse, in other words this is a template
5459 * section edit link. If no flags are given, it was an ordinary section edit link.
5460 * This flag is required to avoid a section numbering mismatch when a section is
5461 * enclosed by "<includeonly>" (T8563).
5462 *
5463 * The section number 0 pulls the text before the first heading; other numbers will
5464 * pull the given section along with its lower-level subsections. If the section is
5465 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5466 *
5467 * Section 0 is always considered to exist, even if it only contains the empty
5468 * string. If $text is the empty string and section 0 is replaced, $newText is
5469 * returned.
5470 *
5471 * @param string $mode One of "get" or "replace"
5472 * @param string $newText Replacement text for section data.
5473 * @return string For "get", the extracted section text.
5474 * for "replace", the whole page with the section replaced.
5475 */
5476 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5477 global $wgTitle; # not generally used but removes an ugly failure mode
5478
5479 $magicScopeVariable = $this->lock();
5480 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5481 $outText = '';
5482 $frame = $this->getPreprocessor()->newFrame();
5483
5484 # Process section extraction flags
5485 $flags = 0;
5486 $sectionParts = explode( '-', $sectionId );
5487 $sectionIndex = array_pop( $sectionParts );
5488 foreach ( $sectionParts as $part ) {
5489 if ( $part === 'T' ) {
5490 $flags |= self::PTD_FOR_INCLUSION;
5491 }
5492 }
5493
5494 # Check for empty input
5495 if ( strval( $text ) === '' ) {
5496 # Only sections 0 and T-0 exist in an empty document
5497 if ( $sectionIndex == 0 ) {
5498 if ( $mode === 'get' ) {
5499 return '';
5500 } else {
5501 return $newText;
5502 }
5503 } else {
5504 if ( $mode === 'get' ) {
5505 return $newText;
5506 } else {
5507 return $text;
5508 }
5509 }
5510 }
5511
5512 # Preprocess the text
5513 $root = $this->preprocessToDom( $text, $flags );
5514
5515 # <h> nodes indicate section breaks
5516 # They can only occur at the top level, so we can find them by iterating the root's children
5517 $node = $root->getFirstChild();
5518
5519 # Find the target section
5520 if ( $sectionIndex == 0 ) {
5521 # Section zero doesn't nest, level=big
5522 $targetLevel = 1000;
5523 } else {
5524 while ( $node ) {
5525 if ( $node->getName() === 'h' ) {
5526 $bits = $node->splitHeading();
5527 if ( $bits['i'] == $sectionIndex ) {
5528 $targetLevel = $bits['level'];
5529 break;
5530 }
5531 }
5532 if ( $mode === 'replace' ) {
5533 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5534 }
5535 $node = $node->getNextSibling();
5536 }
5537 }
5538
5539 if ( !$node ) {
5540 # Not found
5541 if ( $mode === 'get' ) {
5542 return $newText;
5543 } else {
5544 return $text;
5545 }
5546 }
5547
5548 # Find the end of the section, including nested sections
5549 do {
5550 if ( $node->getName() === 'h' ) {
5551 $bits = $node->splitHeading();
5552 $curLevel = $bits['level'];
5553 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5554 break;
5555 }
5556 }
5557 if ( $mode === 'get' ) {
5558 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5559 }
5560 $node = $node->getNextSibling();
5561 } while ( $node );
5562
5563 # Write out the remainder (in replace mode only)
5564 if ( $mode === 'replace' ) {
5565 # Output the replacement text
5566 # Add two newlines on -- trailing whitespace in $newText is conventionally
5567 # stripped by the editor, so we need both newlines to restore the paragraph gap
5568 # Only add trailing whitespace if there is newText
5569 if ( $newText != "" ) {
5570 $outText .= $newText . "\n\n";
5571 }
5572
5573 while ( $node ) {
5574 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5575 $node = $node->getNextSibling();
5576 }
5577 }
5578
5579 if ( is_string( $outText ) ) {
5580 # Re-insert stripped tags
5581 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5582 }
5583
5584 return $outText;
5585 }
5586
5587 /**
5588 * This function returns the text of a section, specified by a number ($section).
5589 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5590 * the first section before any such heading (section 0).
5591 *
5592 * If a section contains subsections, these are also returned.
5593 *
5594 * @param string $text Text to look in
5595 * @param string|int $sectionId Section identifier as a number or string
5596 * (e.g. 0, 1 or 'T-1').
5597 * @param string $defaultText Default to return if section is not found
5598 *
5599 * @return string Text of the requested section
5600 */
5601 public function getSection( $text, $sectionId, $defaultText = '' ) {
5602 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5603 }
5604
5605 /**
5606 * This function returns $oldtext after the content of the section
5607 * specified by $section has been replaced with $text. If the target
5608 * section does not exist, $oldtext is returned unchanged.
5609 *
5610 * @param string $oldText Former text of the article
5611 * @param string|int $sectionId Section identifier as a number or string
5612 * (e.g. 0, 1 or 'T-1').
5613 * @param string $newText Replacing text
5614 *
5615 * @return string Modified text
5616 */
5617 public function replaceSection( $oldText, $sectionId, $newText ) {
5618 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5619 }
5620
5621 /**
5622 * Get the ID of the revision we are parsing
5623 *
5624 * @return int|null
5625 */
5626 public function getRevisionId() {
5627 return $this->mRevisionId;
5628 }
5629
5630 /**
5631 * Get the revision object for $this->mRevisionId
5632 *
5633 * @return Revision|null Either a Revision object or null
5634 * @since 1.23 (public since 1.23)
5635 */
5636 public function getRevisionObject() {
5637 if ( !is_null( $this->mRevisionObject ) ) {
5638 return $this->mRevisionObject;
5639 }
5640 if ( is_null( $this->mRevisionId ) ) {
5641 return null;
5642 }
5643
5644 $rev = call_user_func(
5645 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5646 );
5647
5648 # If the parse is for a new revision, then the callback should have
5649 # already been set to force the object and should match mRevisionId.
5650 # If not, try to fetch by mRevisionId for sanity.
5651 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5652 $rev = Revision::newFromId( $this->mRevisionId );
5653 }
5654
5655 $this->mRevisionObject = $rev;
5656
5657 return $this->mRevisionObject;
5658 }
5659
5660 /**
5661 * Get the timestamp associated with the current revision, adjusted for
5662 * the default server-local timestamp
5663 * @return string
5664 */
5665 public function getRevisionTimestamp() {
5666 if ( is_null( $this->mRevisionTimestamp ) ) {
5667 global $wgContLang;
5668
5669 $revObject = $this->getRevisionObject();
5670 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5671
5672 # The cryptic '' timezone parameter tells to use the site-default
5673 # timezone offset instead of the user settings.
5674 # Since this value will be saved into the parser cache, served
5675 # to other users, and potentially even used inside links and such,
5676 # it needs to be consistent for all visitors.
5677 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5678
5679 }
5680 return $this->mRevisionTimestamp;
5681 }
5682
5683 /**
5684 * Get the name of the user that edited the last revision
5685 *
5686 * @return string User name
5687 */
5688 public function getRevisionUser() {
5689 if ( is_null( $this->mRevisionUser ) ) {
5690 $revObject = $this->getRevisionObject();
5691
5692 # if this template is subst: the revision id will be blank,
5693 # so just use the current user's name
5694 if ( $revObject ) {
5695 $this->mRevisionUser = $revObject->getUserText();
5696 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5697 $this->mRevisionUser = $this->getUser()->getName();
5698 }
5699 }
5700 return $this->mRevisionUser;
5701 }
5702
5703 /**
5704 * Get the size of the revision
5705 *
5706 * @return int|null Revision size
5707 */
5708 public function getRevisionSize() {
5709 if ( is_null( $this->mRevisionSize ) ) {
5710 $revObject = $this->getRevisionObject();
5711
5712 # if this variable is subst: the revision id will be blank,
5713 # so just use the parser input size, because the own substituation
5714 # will change the size.
5715 if ( $revObject ) {
5716 $this->mRevisionSize = $revObject->getSize();
5717 } else {
5718 $this->mRevisionSize = $this->mInputSize;
5719 }
5720 }
5721 return $this->mRevisionSize;
5722 }
5723
5724 /**
5725 * Mutator for $mDefaultSort
5726 *
5727 * @param string $sort New value
5728 */
5729 public function setDefaultSort( $sort ) {
5730 $this->mDefaultSort = $sort;
5731 $this->mOutput->setProperty( 'defaultsort', $sort );
5732 }
5733
5734 /**
5735 * Accessor for $mDefaultSort
5736 * Will use the empty string if none is set.
5737 *
5738 * This value is treated as a prefix, so the
5739 * empty string is equivalent to sorting by
5740 * page name.
5741 *
5742 * @return string
5743 */
5744 public function getDefaultSort() {
5745 if ( $this->mDefaultSort !== false ) {
5746 return $this->mDefaultSort;
5747 } else {
5748 return '';
5749 }
5750 }
5751
5752 /**
5753 * Accessor for $mDefaultSort
5754 * Unlike getDefaultSort(), will return false if none is set
5755 *
5756 * @return string|bool
5757 */
5758 public function getCustomDefaultSort() {
5759 return $this->mDefaultSort;
5760 }
5761
5762 /**
5763 * Try to guess the section anchor name based on a wikitext fragment
5764 * presumably extracted from a heading, for example "Header" from
5765 * "== Header ==".
5766 *
5767 * @param string $text
5768 *
5769 * @return string
5770 */
5771 public function guessSectionNameFromWikiText( $text ) {
5772 # Strip out wikitext links(they break the anchor)
5773 $text = $this->stripSectionName( $text );
5774 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5775 $text = Sanitizer::decodeCharReferences( $text );
5776 return '#' . Sanitizer::escapeIdForLink( $text );
5777 }
5778
5779 /**
5780 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5781 * instead, if possible. For use in redirects, since various versions
5782 * of Microsoft browsers interpret Location: headers as something other
5783 * than UTF-8, resulting in breakage.
5784 *
5785 * @param string $text The section name
5786 * @return string An anchor
5787 */
5788 public function guessLegacySectionNameFromWikiText( $text ) {
5789 global $wgFragmentMode;
5790
5791 # Strip out wikitext links(they break the anchor)
5792 $text = $this->stripSectionName( $text );
5793 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5794 $text = Sanitizer::decodeCharReferences( $text );
5795
5796 if ( isset( $wgFragmentMode[1] ) && $wgFragmentMode[1] === 'legacy' ) {
5797 // ForAttribute() and ForLink() are the same for legacy encoding
5798 $id = Sanitizer::escapeIdForAttribute( $text, Sanitizer::ID_FALLBACK );
5799 } else {
5800 $id = Sanitizer::escapeIdForLink( $text );
5801 }
5802
5803 return "#$id";
5804 }
5805
5806 /**
5807 * Strips a text string of wikitext for use in a section anchor
5808 *
5809 * Accepts a text string and then removes all wikitext from the
5810 * string and leaves only the resultant text (i.e. the result of
5811 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5812 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5813 * to create valid section anchors by mimicing the output of the
5814 * parser when headings are parsed.
5815 *
5816 * @param string $text Text string to be stripped of wikitext
5817 * for use in a Section anchor
5818 * @return string Filtered text string
5819 */
5820 public function stripSectionName( $text ) {
5821 # Strip internal link markup
5822 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5823 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5824
5825 # Strip external link markup
5826 # @todo FIXME: Not tolerant to blank link text
5827 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5828 # on how many empty links there are on the page - need to figure that out.
5829 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5830
5831 # Parse wikitext quotes (italics & bold)
5832 $text = $this->doQuotes( $text );
5833
5834 # Strip HTML tags
5835 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5836 return $text;
5837 }
5838
5839 /**
5840 * strip/replaceVariables/unstrip for preprocessor regression testing
5841 *
5842 * @param string $text
5843 * @param Title $title
5844 * @param ParserOptions $options
5845 * @param int $outputType
5846 *
5847 * @return string
5848 */
5849 public function testSrvus( $text, Title $title, ParserOptions $options,
5850 $outputType = self::OT_HTML
5851 ) {
5852 $magicScopeVariable = $this->lock();
5853 $this->startParse( $title, $options, $outputType, true );
5854
5855 $text = $this->replaceVariables( $text );
5856 $text = $this->mStripState->unstripBoth( $text );
5857 $text = Sanitizer::removeHTMLtags( $text );
5858 return $text;
5859 }
5860
5861 /**
5862 * @param string $text
5863 * @param Title $title
5864 * @param ParserOptions $options
5865 * @return string
5866 */
5867 public function testPst( $text, Title $title, ParserOptions $options ) {
5868 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5869 }
5870
5871 /**
5872 * @param string $text
5873 * @param Title $title
5874 * @param ParserOptions $options
5875 * @return string
5876 */
5877 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5878 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5879 }
5880
5881 /**
5882 * Call a callback function on all regions of the given text that are not
5883 * inside strip markers, and replace those regions with the return value
5884 * of the callback. For example, with input:
5885 *
5886 * aaa<MARKER>bbb
5887 *
5888 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5889 * two strings will be replaced with the value returned by the callback in
5890 * each case.
5891 *
5892 * @param string $s
5893 * @param callable $callback
5894 *
5895 * @return string
5896 */
5897 public function markerSkipCallback( $s, $callback ) {
5898 $i = 0;
5899 $out = '';
5900 while ( $i < strlen( $s ) ) {
5901 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5902 if ( $markerStart === false ) {
5903 $out .= call_user_func( $callback, substr( $s, $i ) );
5904 break;
5905 } else {
5906 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5907 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5908 if ( $markerEnd === false ) {
5909 $out .= substr( $s, $markerStart );
5910 break;
5911 } else {
5912 $markerEnd += strlen( self::MARKER_SUFFIX );
5913 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5914 $i = $markerEnd;
5915 }
5916 }
5917 }
5918 return $out;
5919 }
5920
5921 /**
5922 * Remove any strip markers found in the given text.
5923 *
5924 * @param string $text Input string
5925 * @return string
5926 */
5927 public function killMarkers( $text ) {
5928 return $this->mStripState->killMarkers( $text );
5929 }
5930
5931 /**
5932 * Save the parser state required to convert the given half-parsed text to
5933 * HTML. "Half-parsed" in this context means the output of
5934 * recursiveTagParse() or internalParse(). This output has strip markers
5935 * from replaceVariables (extensionSubstitution() etc.), and link
5936 * placeholders from replaceLinkHolders().
5937 *
5938 * Returns an array which can be serialized and stored persistently. This
5939 * array can later be loaded into another parser instance with
5940 * unserializeHalfParsedText(). The text can then be safely incorporated into
5941 * the return value of a parser hook.
5942 *
5943 * @param string $text
5944 *
5945 * @return array
5946 */
5947 public function serializeHalfParsedText( $text ) {
5948 $data = [
5949 'text' => $text,
5950 'version' => self::HALF_PARSED_VERSION,
5951 'stripState' => $this->mStripState->getSubState( $text ),
5952 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5953 ];
5954 return $data;
5955 }
5956
5957 /**
5958 * Load the parser state given in the $data array, which is assumed to
5959 * have been generated by serializeHalfParsedText(). The text contents is
5960 * extracted from the array, and its markers are transformed into markers
5961 * appropriate for the current Parser instance. This transformed text is
5962 * returned, and can be safely included in the return value of a parser
5963 * hook.
5964 *
5965 * If the $data array has been stored persistently, the caller should first
5966 * check whether it is still valid, by calling isValidHalfParsedText().
5967 *
5968 * @param array $data Serialized data
5969 * @throws MWException
5970 * @return string
5971 */
5972 public function unserializeHalfParsedText( $data ) {
5973 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5974 throw new MWException( __METHOD__ . ': invalid version' );
5975 }
5976
5977 # First, extract the strip state.
5978 $texts = [ $data['text'] ];
5979 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5980
5981 # Now renumber links
5982 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5983
5984 # Should be good to go.
5985 return $texts[0];
5986 }
5987
5988 /**
5989 * Returns true if the given array, presumed to be generated by
5990 * serializeHalfParsedText(), is compatible with the current version of the
5991 * parser.
5992 *
5993 * @param array $data
5994 *
5995 * @return bool
5996 */
5997 public function isValidHalfParsedText( $data ) {
5998 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5999 }
6000
6001 /**
6002 * Parsed a width param of imagelink like 300px or 200x300px
6003 *
6004 * @param string $value
6005 *
6006 * @return array
6007 * @since 1.20
6008 */
6009 public function parseWidthParam( $value ) {
6010 $parsedWidthParam = [];
6011 if ( $value === '' ) {
6012 return $parsedWidthParam;
6013 }
6014 $m = [];
6015 # (T15500) In both cases (width/height and width only),
6016 # permit trailing "px" for backward compatibility.
6017 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6018 $width = intval( $m[1] );
6019 $height = intval( $m[2] );
6020 $parsedWidthParam['width'] = $width;
6021 $parsedWidthParam['height'] = $height;
6022 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6023 $width = intval( $value );
6024 $parsedWidthParam['width'] = $width;
6025 }
6026 return $parsedWidthParam;
6027 }
6028
6029 /**
6030 * Lock the current instance of the parser.
6031 *
6032 * This is meant to stop someone from calling the parser
6033 * recursively and messing up all the strip state.
6034 *
6035 * @throws MWException If parser is in a parse
6036 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6037 */
6038 protected function lock() {
6039 if ( $this->mInParse ) {
6040 throw new MWException( "Parser state cleared while parsing. "
6041 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6042 }
6043
6044 // Save the backtrace when locking, so that if some code tries locking again,
6045 // we can print the lock owner's backtrace for easier debugging
6046 $e = new Exception;
6047 $this->mInParse = $e->getTraceAsString();
6048
6049 $recursiveCheck = new ScopedCallback( function () {
6050 $this->mInParse = false;
6051 } );
6052
6053 return $recursiveCheck;
6054 }
6055
6056 /**
6057 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6058 *
6059 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6060 * or if there is more than one <p/> tag in the input HTML.
6061 *
6062 * @param string $html
6063 * @return string
6064 * @since 1.24
6065 */
6066 public static function stripOuterParagraph( $html ) {
6067 $m = [];
6068 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6069 if ( strpos( $m[1], '</p>' ) === false ) {
6070 $html = $m[1];
6071 }
6072 }
6073
6074 return $html;
6075 }
6076
6077 /**
6078 * Return this parser if it is not doing anything, otherwise
6079 * get a fresh parser. You can use this method by doing
6080 * $myParser = $wgParser->getFreshParser(), or more simply
6081 * $wgParser->getFreshParser()->parse( ... );
6082 * if you're unsure if $wgParser is safe to use.
6083 *
6084 * @since 1.24
6085 * @return Parser A parser object that is not parsing anything
6086 */
6087 public function getFreshParser() {
6088 global $wgParserConf;
6089 if ( $this->mInParse ) {
6090 return new $wgParserConf['class']( $wgParserConf );
6091 } else {
6092 return $this;
6093 }
6094 }
6095
6096 /**
6097 * Set's up the PHP implementation of OOUI for use in this request
6098 * and instructs OutputPage to enable OOUI for itself.
6099 *
6100 * @since 1.26
6101 */
6102 public function enableOOUI() {
6103 OutputPage::setupOOUI();
6104 $this->mOutput->setEnableOOUI( true );
6105 }
6106 }
Wiklou, le Wiki du Wiklou