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