dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Parser: Emit deprecation warnings for ParserLimitReport hook
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // @codingStandardsIgnoreStart Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102 // @codingStandardsIgnoreEnd
103
104 # Regular expression for a non-newline space
105 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
106
107 # Flags for preprocessToDom
108 const PTD_FOR_INCLUSION = 1;
109
110 # Allowed values for $this->mOutputType
111 # Parameter to startExternalParse().
112 const OT_HTML = 1; # like parse()
113 const OT_WIKI = 2; # like preSaveTransform()
114 const OT_PREPROCESS = 3; # like preprocess()
115 const OT_MSG = 3;
116 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
117
118 /**
119 * @var string Prefix and suffix for temporary replacement strings
120 * for the multipass parser.
121 *
122 * \x7f should never appear in input as it's disallowed in XML.
123 * Using it at the front also gives us a little extra robustness
124 * since it shouldn't match when butted up against identifier-like
125 * string constructs.
126 *
127 * Must not consist of all title characters, or else it will change
128 * the behavior of <nowiki> in a link.
129 *
130 * Must have a character that needs escaping in attributes, otherwise
131 * someone could put a strip marker in an attribute, to get around
132 * escaping quote marks, and break out of the attribute. Thus we add
133 * `'".
134 */
135 const MARKER_SUFFIX = "-QINU`\"'\x7f";
136 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
137
138 # Markers used for wrapping the table of contents
139 const TOC_START = '<mw:toc>';
140 const TOC_END = '</mw:toc>';
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 public $mFirstCall = true;
155
156 # Initialised by initialiseVariables()
157
158 /**
159 * @var MagicWordArray
160 */
161 public $mVariables;
162
163 /**
164 * @var MagicWordArray
165 */
166 public $mSubstWords;
167 # Initialised in constructor
168 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
169
170 # Initialized in getPreprocessor()
171 /** @var Preprocessor */
172 public $mPreprocessor;
173
174 # Cleared with clearState():
175 /**
176 * @var ParserOutput
177 */
178 public $mOutput;
179 public $mAutonumber;
180
181 /**
182 * @var StripState
183 */
184 public $mStripState;
185
186 public $mIncludeCount;
187 /**
188 * @var LinkHolderArray
189 */
190 public $mLinkHolders;
191
192 public $mLinkID;
193 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
194 public $mDefaultSort;
195 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
196 public $mExpensiveFunctionCount; # number of expensive parser function calls
197 public $mShowToc, $mForceTocPosition;
198
199 /**
200 * @var User
201 */
202 public $mUser; # User object; only used when doing pre-save transform
203
204 # Temporary
205 # These are variables reset at least once per parse regardless of $clearState
206
207 /**
208 * @var ParserOptions
209 */
210 public $mOptions;
211
212 /**
213 * @var Title
214 */
215 public $mTitle; # Title context, used for self-link rendering and similar things
216 public $mOutputType; # Output type, one of the OT_xxx constants
217 public $ot; # Shortcut alias, see setOutputType()
218 public $mRevisionObject; # The revision object of the specified revision ID
219 public $mRevisionId; # ID to display in {{REVISIONID}} tags
220 public $mRevisionTimestamp; # The timestamp of the specified revision ID
221 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
222 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
223 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
224 public $mInputSize = false; # For {{PAGESIZE}} on current page.
225
226 /**
227 * @var string Deprecated accessor for the strip marker prefix.
228 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
229 */
230 public $mUniqPrefix = Parser::MARKER_PREFIX;
231
232 /**
233 * @var array Array with the language name of each language link (i.e. the
234 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
235 * duplicate language links to the ParserOutput.
236 */
237 public $mLangLinkLanguages;
238
239 /**
240 * @var MapCacheLRU|null
241 * @since 1.24
242 *
243 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
244 */
245 public $currentRevisionCache;
246
247 /**
248 * @var bool|string Recursive call protection.
249 * This variable should be treated as if it were private.
250 */
251 public $mInParse = false;
252
253 /** @var SectionProfiler */
254 protected $mProfiler;
255
256 /**
257 * @var LinkRenderer
258 */
259 protected $mLinkRenderer;
260
261 /**
262 * @param array $conf
263 */
264 public function __construct( $conf = [] ) {
265 $this->mConf = $conf;
266 $this->mUrlProtocols = wfUrlProtocols();
267 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
268 self::EXT_LINK_ADDR .
269 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
270 if ( isset( $conf['preprocessorClass'] ) ) {
271 $this->mPreprocessorClass = $conf['preprocessorClass'];
272 } elseif ( defined( 'HPHP_VERSION' ) ) {
273 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
274 $this->mPreprocessorClass = 'Preprocessor_Hash';
275 } elseif ( extension_loaded( 'domxml' ) ) {
276 # PECL extension that conflicts with the core DOM extension (T15770)
277 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
278 $this->mPreprocessorClass = 'Preprocessor_Hash';
279 } elseif ( extension_loaded( 'dom' ) ) {
280 $this->mPreprocessorClass = 'Preprocessor_DOM';
281 } else {
282 $this->mPreprocessorClass = 'Preprocessor_Hash';
283 }
284 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
285 }
286
287 /**
288 * Reduce memory usage to reduce the impact of circular references
289 */
290 public function __destruct() {
291 if ( isset( $this->mLinkHolders ) ) {
292 unset( $this->mLinkHolders );
293 }
294 foreach ( $this as $name => $value ) {
295 unset( $this->$name );
296 }
297 }
298
299 /**
300 * Allow extensions to clean up when the parser is cloned
301 */
302 public function __clone() {
303 $this->mInParse = false;
304
305 // T58226: When you create a reference "to" an object field, that
306 // makes the object field itself be a reference too (until the other
307 // reference goes out of scope). When cloning, any field that's a
308 // reference is copied as a reference in the new object. Both of these
309 // are defined PHP5 behaviors, as inconvenient as it is for us when old
310 // hooks from PHP4 days are passing fields by reference.
311 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
312 // Make a non-reference copy of the field, then rebind the field to
313 // reference the new copy.
314 $tmp = $this->$k;
315 $this->$k =& $tmp;
316 unset( $tmp );
317 }
318
319 Hooks::run( 'ParserCloned', [ $this ] );
320 }
321
322 /**
323 * Do various kinds of initialisation on the first call of the parser
324 */
325 public function firstCallInit() {
326 if ( !$this->mFirstCall ) {
327 return;
328 }
329 $this->mFirstCall = false;
330
331 CoreParserFunctions::register( $this );
332 CoreTagHooks::register( $this );
333 $this->initialiseVariables();
334
335 // Avoid PHP 7.1 warning from passing $this by reference
336 $parser = $this;
337 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
338 }
339
340 /**
341 * Clear Parser state
342 *
343 * @private
344 */
345 public function clearState() {
346 if ( $this->mFirstCall ) {
347 $this->firstCallInit();
348 }
349 $this->mOutput = new ParserOutput;
350 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
351 $this->mAutonumber = 0;
352 $this->mIncludeCount = [];
353 $this->mLinkHolders = new LinkHolderArray( $this );
354 $this->mLinkID = 0;
355 $this->mRevisionObject = $this->mRevisionTimestamp =
356 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
357 $this->mVarCache = [];
358 $this->mUser = null;
359 $this->mLangLinkLanguages = [];
360 $this->currentRevisionCache = null;
361
362 $this->mStripState = new StripState;
363
364 # Clear these on every parse, T6549
365 $this->mTplRedirCache = $this->mTplDomCache = [];
366
367 $this->mShowToc = true;
368 $this->mForceTocPosition = false;
369 $this->mIncludeSizes = [
370 'post-expand' => 0,
371 'arg' => 0,
372 ];
373 $this->mPPNodeCount = 0;
374 $this->mGeneratedPPNodeCount = 0;
375 $this->mHighestExpansionDepth = 0;
376 $this->mDefaultSort = false;
377 $this->mHeadings = [];
378 $this->mDoubleUnderscores = [];
379 $this->mExpensiveFunctionCount = 0;
380
381 # Fix cloning
382 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
383 $this->mPreprocessor = null;
384 }
385
386 $this->mProfiler = new SectionProfiler();
387
388 // Avoid PHP 7.1 warning from passing $this by reference
389 $parser = $this;
390 Hooks::run( 'ParserClearState', [ &$parser ] );
391 }
392
393 /**
394 * Convert wikitext to HTML
395 * Do not call this function recursively.
396 *
397 * @param string $text Text we want to parse
398 * @param Title $title
399 * @param ParserOptions $options
400 * @param bool $linestart
401 * @param bool $clearState
402 * @param int $revid Number to pass in {{REVISIONID}}
403 * @return ParserOutput A ParserOutput
404 */
405 public function parse(
406 $text, Title $title, ParserOptions $options,
407 $linestart = true, $clearState = true, $revid = null
408 ) {
409 /**
410 * First pass--just handle <nowiki> sections, pass the rest off
411 * to internalParse() which does all the real work.
412 */
413
414 global $wgShowHostnames;
415
416 if ( $clearState ) {
417 // We use U+007F DELETE to construct strip markers, so we have to make
418 // sure that this character does not occur in the input text.
419 $text = strtr( $text, "\x7f", "?" );
420 $magicScopeVariable = $this->lock();
421 }
422 // Strip U+0000 NULL (T159174)
423 $text = str_replace( "\000", '', $text );
424
425 $this->startParse( $title, $options, self::OT_HTML, $clearState );
426
427 $this->currentRevisionCache = null;
428 $this->mInputSize = strlen( $text );
429 if ( $this->mOptions->getEnableLimitReport() ) {
430 $this->mOutput->resetParseStartTime();
431 }
432
433 $oldRevisionId = $this->mRevisionId;
434 $oldRevisionObject = $this->mRevisionObject;
435 $oldRevisionTimestamp = $this->mRevisionTimestamp;
436 $oldRevisionUser = $this->mRevisionUser;
437 $oldRevisionSize = $this->mRevisionSize;
438 if ( $revid !== null ) {
439 $this->mRevisionId = $revid;
440 $this->mRevisionObject = null;
441 $this->mRevisionTimestamp = null;
442 $this->mRevisionUser = null;
443 $this->mRevisionSize = null;
444 }
445
446 // Avoid PHP 7.1 warning from passing $this by reference
447 $parser = $this;
448 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
449 # No more strip!
450 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
451 $text = $this->internalParse( $text );
452 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
453
454 $text = $this->internalParseHalfParsed( $text, true, $linestart );
455
456 /**
457 * A converted title will be provided in the output object if title and
458 * content conversion are enabled, the article text does not contain
459 * a conversion-suppressing double-underscore tag, and no
460 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
461 * automatic link conversion.
462 */
463 if ( !( $options->getDisableTitleConversion()
464 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
465 || isset( $this->mDoubleUnderscores['notitleconvert'] )
466 || $this->mOutput->getDisplayTitle() !== false )
467 ) {
468 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
469 if ( $convruletitle ) {
470 $this->mOutput->setTitleText( $convruletitle );
471 } else {
472 $titleText = $this->getConverterLanguage()->convertTitle( $title );
473 $this->mOutput->setTitleText( $titleText );
474 }
475 }
476
477 # Done parsing! Compute runtime adaptive expiry if set
478 $this->mOutput->finalizeAdaptiveCacheExpiry();
479
480 # Warn if too many heavyweight parser functions were used
481 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
482 $this->limitationWarn( 'expensive-parserfunction',
483 $this->mExpensiveFunctionCount,
484 $this->mOptions->getExpensiveParserFunctionLimit()
485 );
486 }
487
488 # Information on include size limits, for the benefit of users who try to skirt them
489 if ( $this->mOptions->getEnableLimitReport() ) {
490 $max = $this->mOptions->getMaxIncludeSize();
491
492 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
493 if ( $cpuTime !== null ) {
494 $this->mOutput->setLimitReportData( 'limitreport-cputime',
495 sprintf( "%.3f", $cpuTime )
496 );
497 }
498
499 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
500 $this->mOutput->setLimitReportData( 'limitreport-walltime',
501 sprintf( "%.3f", $wallTime )
502 );
503
504 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
505 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
506 );
507 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
508 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
509 );
510 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
511 [ $this->mIncludeSizes['post-expand'], $max ]
512 );
513 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
514 [ $this->mIncludeSizes['arg'], $max ]
515 );
516 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
517 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
518 );
519 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
520 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
521 );
522 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
523
524 $limitReport = "NewPP limit report\n";
525 if ( $wgShowHostnames ) {
526 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
527 }
528 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
529 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
530 $limitReport .= 'Dynamic content: ' .
531 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
532 "\n";
533
534 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
535 if ( Hooks::run( 'ParserLimitReportFormat',
536 [ $key, &$value, &$limitReport, false, false ]
537 ) ) {
538 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
539 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
540 ->inLanguage( 'en' )->useDatabase( false );
541 if ( !$valueMsg->exists() ) {
542 $valueMsg = new RawMessage( '$1' );
543 }
544 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
545 $valueMsg->params( $value );
546 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
547 }
548 }
549 }
550 // Since we're not really outputting HTML, decode the entities and
551 // then re-encode the things that need hiding inside HTML comments.
552 $limitReport = htmlspecialchars_decode( $limitReport );
553 // Run deprecated hook
554 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ], '1.22' );
555
556 // Sanitize for comment. Note '‐' in the replacement is U+2010,
557 // which looks much like the problematic '-'.
558 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
559 $text .= "\n<!-- \n$limitReport-->\n";
560
561 // Add on template profiling data in human/machine readable way
562 $dataByFunc = $this->mProfiler->getFunctionStats();
563 uasort( $dataByFunc, function ( $a, $b ) {
564 return $a['real'] < $b['real']; // descending order
565 } );
566 $profileReport = [];
567 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
568 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
569 $item['%real'], $item['real'], $item['calls'],
570 htmlspecialchars( $item['name'] ) );
571 }
572 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
573 $text .= implode( "\n", $profileReport ) . "\n-->\n";
574
575 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
576
577 // Add other cache related metadata
578 if ( $wgShowHostnames ) {
579 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
580 }
581 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
582 $this->mOutput->getCacheTime() );
583 $this->mOutput->setLimitReportData( 'cachereport-ttl',
584 $this->mOutput->getCacheExpiry() );
585 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
586 $this->mOutput->hasDynamicContent() );
587
588 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
589 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
590 $this->mTitle->getPrefixedDBkey() );
591 }
592 }
593
594 # Wrap non-interface parser output in a <div> so it can be targeted
595 # with CSS (T37247)
596 $class = $this->mOptions->getWrapOutputClass();
597 if ( $class !== false && !$this->mOptions->getInterfaceMessage() ) {
598 $text = Html::rawElement( 'div', [ 'class' => $class ], $text );
599 }
600
601 $this->mOutput->setText( $text );
602
603 $this->mRevisionId = $oldRevisionId;
604 $this->mRevisionObject = $oldRevisionObject;
605 $this->mRevisionTimestamp = $oldRevisionTimestamp;
606 $this->mRevisionUser = $oldRevisionUser;
607 $this->mRevisionSize = $oldRevisionSize;
608 $this->mInputSize = false;
609 $this->currentRevisionCache = null;
610
611 return $this->mOutput;
612 }
613
614 /**
615 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
616 * can be called from an extension tag hook.
617 *
618 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
619 * instead, which means that lists and links have not been fully parsed yet,
620 * and strip markers are still present.
621 *
622 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
623 *
624 * Use this function if you're a parser tag hook and you want to parse
625 * wikitext before or after applying additional transformations, and you
626 * intend to *return the result as hook output*, which will cause it to go
627 * through the rest of parsing process automatically.
628 *
629 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
630 * $text are not expanded
631 *
632 * @param string $text Text extension wants to have parsed
633 * @param bool|PPFrame $frame The frame to use for expanding any template variables
634 * @return string UNSAFE half-parsed HTML
635 */
636 public function recursiveTagParse( $text, $frame = false ) {
637 // Avoid PHP 7.1 warning from passing $this by reference
638 $parser = $this;
639 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
640 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
641 $text = $this->internalParse( $text, false, $frame );
642 return $text;
643 }
644
645 /**
646 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
647 * point can be called from an extension tag hook.
648 *
649 * The output of this function is fully-parsed HTML that is safe for output.
650 * If you're a parser tag hook, you might want to use recursiveTagParse()
651 * instead.
652 *
653 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
654 * $text are not expanded
655 *
656 * @since 1.25
657 *
658 * @param string $text Text extension wants to have parsed
659 * @param bool|PPFrame $frame The frame to use for expanding any template variables
660 * @return string Fully parsed HTML
661 */
662 public function recursiveTagParseFully( $text, $frame = false ) {
663 $text = $this->recursiveTagParse( $text, $frame );
664 $text = $this->internalParseHalfParsed( $text, false );
665 return $text;
666 }
667
668 /**
669 * Expand templates and variables in the text, producing valid, static wikitext.
670 * Also removes comments.
671 * Do not call this function recursively.
672 * @param string $text
673 * @param Title $title
674 * @param ParserOptions $options
675 * @param int|null $revid
676 * @param bool|PPFrame $frame
677 * @return mixed|string
678 */
679 public function preprocess( $text, Title $title = null,
680 ParserOptions $options, $revid = null, $frame = false
681 ) {
682 $magicScopeVariable = $this->lock();
683 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
684 if ( $revid !== null ) {
685 $this->mRevisionId = $revid;
686 }
687 // Avoid PHP 7.1 warning from passing $this by reference
688 $parser = $this;
689 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
690 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
691 $text = $this->replaceVariables( $text, $frame );
692 $text = $this->mStripState->unstripBoth( $text );
693 return $text;
694 }
695
696 /**
697 * Recursive parser entry point that can be called from an extension tag
698 * hook.
699 *
700 * @param string $text Text to be expanded
701 * @param bool|PPFrame $frame The frame to use for expanding any template variables
702 * @return string
703 * @since 1.19
704 */
705 public function recursivePreprocess( $text, $frame = false ) {
706 $text = $this->replaceVariables( $text, $frame );
707 $text = $this->mStripState->unstripBoth( $text );
708 return $text;
709 }
710
711 /**
712 * Process the wikitext for the "?preload=" feature. (T7210)
713 *
714 * "<noinclude>", "<includeonly>" etc. are parsed as for template
715 * transclusion, comments, templates, arguments, tags hooks and parser
716 * functions are untouched.
717 *
718 * @param string $text
719 * @param Title $title
720 * @param ParserOptions $options
721 * @param array $params
722 * @return string
723 */
724 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
725 $msg = new RawMessage( $text );
726 $text = $msg->params( $params )->plain();
727
728 # Parser (re)initialisation
729 $magicScopeVariable = $this->lock();
730 $this->startParse( $title, $options, self::OT_PLAIN, true );
731
732 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
733 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
734 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
735 $text = $this->mStripState->unstripBoth( $text );
736 return $text;
737 }
738
739 /**
740 * Get a random string
741 *
742 * @return string
743 * @deprecated since 1.26; use wfRandomString() instead.
744 */
745 public static function getRandomString() {
746 wfDeprecated( __METHOD__, '1.26' );
747 return wfRandomString( 16 );
748 }
749
750 /**
751 * Set the current user.
752 * Should only be used when doing pre-save transform.
753 *
754 * @param User|null $user User object or null (to reset)
755 */
756 public function setUser( $user ) {
757 $this->mUser = $user;
758 }
759
760 /**
761 * Accessor for mUniqPrefix.
762 *
763 * @return string
764 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
765 */
766 public function uniqPrefix() {
767 wfDeprecated( __METHOD__, '1.26' );
768 return self::MARKER_PREFIX;
769 }
770
771 /**
772 * Set the context title
773 *
774 * @param Title $t
775 */
776 public function setTitle( $t ) {
777 if ( !$t ) {
778 $t = Title::newFromText( 'NO TITLE' );
779 }
780
781 if ( $t->hasFragment() ) {
782 # Strip the fragment to avoid various odd effects
783 $this->mTitle = $t->createFragmentTarget( '' );
784 } else {
785 $this->mTitle = $t;
786 }
787 }
788
789 /**
790 * Accessor for the Title object
791 *
792 * @return Title
793 */
794 public function getTitle() {
795 return $this->mTitle;
796 }
797
798 /**
799 * Accessor/mutator for the Title object
800 *
801 * @param Title $x Title object or null to just get the current one
802 * @return Title
803 */
804 public function Title( $x = null ) {
805 return wfSetVar( $this->mTitle, $x );
806 }
807
808 /**
809 * Set the output type
810 *
811 * @param int $ot New value
812 */
813 public function setOutputType( $ot ) {
814 $this->mOutputType = $ot;
815 # Shortcut alias
816 $this->ot = [
817 'html' => $ot == self::OT_HTML,
818 'wiki' => $ot == self::OT_WIKI,
819 'pre' => $ot == self::OT_PREPROCESS,
820 'plain' => $ot == self::OT_PLAIN,
821 ];
822 }
823
824 /**
825 * Accessor/mutator for the output type
826 *
827 * @param int|null $x New value or null to just get the current one
828 * @return int
829 */
830 public function OutputType( $x = null ) {
831 return wfSetVar( $this->mOutputType, $x );
832 }
833
834 /**
835 * Get the ParserOutput object
836 *
837 * @return ParserOutput
838 */
839 public function getOutput() {
840 return $this->mOutput;
841 }
842
843 /**
844 * Get the ParserOptions object
845 *
846 * @return ParserOptions
847 */
848 public function getOptions() {
849 return $this->mOptions;
850 }
851
852 /**
853 * Accessor/mutator for the ParserOptions object
854 *
855 * @param ParserOptions $x New value or null to just get the current one
856 * @return ParserOptions Current ParserOptions object
857 */
858 public function Options( $x = null ) {
859 return wfSetVar( $this->mOptions, $x );
860 }
861
862 /**
863 * @return int
864 */
865 public function nextLinkID() {
866 return $this->mLinkID++;
867 }
868
869 /**
870 * @param int $id
871 */
872 public function setLinkID( $id ) {
873 $this->mLinkID = $id;
874 }
875
876 /**
877 * Get a language object for use in parser functions such as {{FORMATNUM:}}
878 * @return Language
879 */
880 public function getFunctionLang() {
881 return $this->getTargetLanguage();
882 }
883
884 /**
885 * Get the target language for the content being parsed. This is usually the
886 * language that the content is in.
887 *
888 * @since 1.19
889 *
890 * @throws MWException
891 * @return Language
892 */
893 public function getTargetLanguage() {
894 $target = $this->mOptions->getTargetLanguage();
895
896 if ( $target !== null ) {
897 return $target;
898 } elseif ( $this->mOptions->getInterfaceMessage() ) {
899 return $this->mOptions->getUserLangObj();
900 } elseif ( is_null( $this->mTitle ) ) {
901 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
902 }
903
904 return $this->mTitle->getPageLanguage();
905 }
906
907 /**
908 * Get the language object for language conversion
909 * @return Language|null
910 */
911 public function getConverterLanguage() {
912 return $this->getTargetLanguage();
913 }
914
915 /**
916 * Get a User object either from $this->mUser, if set, or from the
917 * ParserOptions object otherwise
918 *
919 * @return User
920 */
921 public function getUser() {
922 if ( !is_null( $this->mUser ) ) {
923 return $this->mUser;
924 }
925 return $this->mOptions->getUser();
926 }
927
928 /**
929 * Get a preprocessor object
930 *
931 * @return Preprocessor
932 */
933 public function getPreprocessor() {
934 if ( !isset( $this->mPreprocessor ) ) {
935 $class = $this->mPreprocessorClass;
936 $this->mPreprocessor = new $class( $this );
937 }
938 return $this->mPreprocessor;
939 }
940
941 /**
942 * Get a LinkRenderer instance to make links with
943 *
944 * @since 1.28
945 * @return LinkRenderer
946 */
947 public function getLinkRenderer() {
948 if ( !$this->mLinkRenderer ) {
949 $this->mLinkRenderer = MediaWikiServices::getInstance()
950 ->getLinkRendererFactory()->create();
951 $this->mLinkRenderer->setStubThreshold(
952 $this->getOptions()->getStubThreshold()
953 );
954 }
955
956 return $this->mLinkRenderer;
957 }
958
959 /**
960 * Replaces all occurrences of HTML-style comments and the given tags
961 * in the text with a random marker and returns the next text. The output
962 * parameter $matches will be an associative array filled with data in
963 * the form:
964 *
965 * @code
966 * 'UNIQ-xxxxx' => [
967 * 'element',
968 * 'tag content',
969 * [ 'param' => 'x' ],
970 * '<element param="x">tag content</element>' ]
971 * @endcode
972 *
973 * @param array $elements List of element names. Comments are always extracted.
974 * @param string $text Source text string.
975 * @param array $matches Out parameter, Array: extracted tags
976 * @param string|null $uniq_prefix
977 * @return string Stripped text
978 * @since 1.26 The uniq_prefix argument is deprecated.
979 */
980 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
981 if ( $uniq_prefix !== null ) {
982 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
983 }
984 static $n = 1;
985 $stripped = '';
986 $matches = [];
987
988 $taglist = implode( '|', $elements );
989 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
990
991 while ( $text != '' ) {
992 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
993 $stripped .= $p[0];
994 if ( count( $p ) < 5 ) {
995 break;
996 }
997 if ( count( $p ) > 5 ) {
998 # comment
999 $element = $p[4];
1000 $attributes = '';
1001 $close = '';
1002 $inside = $p[5];
1003 } else {
1004 # tag
1005 $element = $p[1];
1006 $attributes = $p[2];
1007 $close = $p[3];
1008 $inside = $p[4];
1009 }
1010
1011 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1012 $stripped .= $marker;
1013
1014 if ( $close === '/>' ) {
1015 # Empty element tag, <tag />
1016 $content = null;
1017 $text = $inside;
1018 $tail = null;
1019 } else {
1020 if ( $element === '!--' ) {
1021 $end = '/(-->)/';
1022 } else {
1023 $end = "/(<\\/$element\\s*>)/i";
1024 }
1025 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1026 $content = $q[0];
1027 if ( count( $q ) < 3 ) {
1028 # No end tag -- let it run out to the end of the text.
1029 $tail = '';
1030 $text = '';
1031 } else {
1032 $tail = $q[1];
1033 $text = $q[2];
1034 }
1035 }
1036
1037 $matches[$marker] = [ $element,
1038 $content,
1039 Sanitizer::decodeTagAttributes( $attributes ),
1040 "<$element$attributes$close$content$tail" ];
1041 }
1042 return $stripped;
1043 }
1044
1045 /**
1046 * Get a list of strippable XML-like elements
1047 *
1048 * @return array
1049 */
1050 public function getStripList() {
1051 return $this->mStripList;
1052 }
1053
1054 /**
1055 * Add an item to the strip state
1056 * Returns the unique tag which must be inserted into the stripped text
1057 * The tag will be replaced with the original text in unstrip()
1058 *
1059 * @param string $text
1060 *
1061 * @return string
1062 */
1063 public function insertStripItem( $text ) {
1064 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1065 $this->mMarkerIndex++;
1066 $this->mStripState->addGeneral( $marker, $text );
1067 return $marker;
1068 }
1069
1070 /**
1071 * parse the wiki syntax used to render tables
1072 *
1073 * @private
1074 * @param string $text
1075 * @return string
1076 */
1077 public function doTableStuff( $text ) {
1078
1079 $lines = StringUtils::explode( "\n", $text );
1080 $out = '';
1081 $td_history = []; # Is currently a td tag open?
1082 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1083 $tr_history = []; # Is currently a tr tag open?
1084 $tr_attributes = []; # history of tr attributes
1085 $has_opened_tr = []; # Did this table open a <tr> element?
1086 $indent_level = 0; # indent level of the table
1087
1088 foreach ( $lines as $outLine ) {
1089 $line = trim( $outLine );
1090
1091 if ( $line === '' ) { # empty line, go to next line
1092 $out .= $outLine . "\n";
1093 continue;
1094 }
1095
1096 $first_character = $line[0];
1097 $first_two = substr( $line, 0, 2 );
1098 $matches = [];
1099
1100 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1101 # First check if we are starting a new table
1102 $indent_level = strlen( $matches[1] );
1103
1104 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1105 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1106
1107 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1108 array_push( $td_history, false );
1109 array_push( $last_tag_history, '' );
1110 array_push( $tr_history, false );
1111 array_push( $tr_attributes, '' );
1112 array_push( $has_opened_tr, false );
1113 } elseif ( count( $td_history ) == 0 ) {
1114 # Don't do any of the following
1115 $out .= $outLine . "\n";
1116 continue;
1117 } elseif ( $first_two === '|}' ) {
1118 # We are ending a table
1119 $line = '</table>' . substr( $line, 2 );
1120 $last_tag = array_pop( $last_tag_history );
1121
1122 if ( !array_pop( $has_opened_tr ) ) {
1123 $line = "<tr><td></td></tr>{$line}";
1124 }
1125
1126 if ( array_pop( $tr_history ) ) {
1127 $line = "</tr>{$line}";
1128 }
1129
1130 if ( array_pop( $td_history ) ) {
1131 $line = "</{$last_tag}>{$line}";
1132 }
1133 array_pop( $tr_attributes );
1134 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1135 } elseif ( $first_two === '|-' ) {
1136 # Now we have a table row
1137 $line = preg_replace( '#^\|-+#', '', $line );
1138
1139 # Whats after the tag is now only attributes
1140 $attributes = $this->mStripState->unstripBoth( $line );
1141 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1142 array_pop( $tr_attributes );
1143 array_push( $tr_attributes, $attributes );
1144
1145 $line = '';
1146 $last_tag = array_pop( $last_tag_history );
1147 array_pop( $has_opened_tr );
1148 array_push( $has_opened_tr, true );
1149
1150 if ( array_pop( $tr_history ) ) {
1151 $line = '</tr>';
1152 }
1153
1154 if ( array_pop( $td_history ) ) {
1155 $line = "</{$last_tag}>{$line}";
1156 }
1157
1158 $outLine = $line;
1159 array_push( $tr_history, false );
1160 array_push( $td_history, false );
1161 array_push( $last_tag_history, '' );
1162 } elseif ( $first_character === '|'
1163 || $first_character === '!'
1164 || $first_two === '|+'
1165 ) {
1166 # This might be cell elements, td, th or captions
1167 if ( $first_two === '|+' ) {
1168 $first_character = '+';
1169 $line = substr( $line, 2 );
1170 } else {
1171 $line = substr( $line, 1 );
1172 }
1173
1174 // Implies both are valid for table headings.
1175 if ( $first_character === '!' ) {
1176 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1177 }
1178
1179 # Split up multiple cells on the same line.
1180 # FIXME : This can result in improper nesting of tags processed
1181 # by earlier parser steps.
1182 $cells = explode( '||', $line );
1183
1184 $outLine = '';
1185
1186 # Loop through each table cell
1187 foreach ( $cells as $cell ) {
1188 $previous = '';
1189 if ( $first_character !== '+' ) {
1190 $tr_after = array_pop( $tr_attributes );
1191 if ( !array_pop( $tr_history ) ) {
1192 $previous = "<tr{$tr_after}>\n";
1193 }
1194 array_push( $tr_history, true );
1195 array_push( $tr_attributes, '' );
1196 array_pop( $has_opened_tr );
1197 array_push( $has_opened_tr, true );
1198 }
1199
1200 $last_tag = array_pop( $last_tag_history );
1201
1202 if ( array_pop( $td_history ) ) {
1203 $previous = "</{$last_tag}>\n{$previous}";
1204 }
1205
1206 if ( $first_character === '|' ) {
1207 $last_tag = 'td';
1208 } elseif ( $first_character === '!' ) {
1209 $last_tag = 'th';
1210 } elseif ( $first_character === '+' ) {
1211 $last_tag = 'caption';
1212 } else {
1213 $last_tag = '';
1214 }
1215
1216 array_push( $last_tag_history, $last_tag );
1217
1218 # A cell could contain both parameters and data
1219 $cell_data = explode( '|', $cell, 2 );
1220
1221 # T2553: Note that a '|' inside an invalid link should not
1222 # be mistaken as delimiting cell parameters
1223 # Bug T153140: Neither should language converter markup.
1224 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1225 $cell = "{$previous}<{$last_tag}>{$cell}";
1226 } elseif ( count( $cell_data ) == 1 ) {
1227 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1228 } else {
1229 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1230 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1231 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1232 }
1233
1234 $outLine .= $cell;
1235 array_push( $td_history, true );
1236 }
1237 }
1238 $out .= $outLine . "\n";
1239 }
1240
1241 # Closing open td, tr && table
1242 while ( count( $td_history ) > 0 ) {
1243 if ( array_pop( $td_history ) ) {
1244 $out .= "</td>\n";
1245 }
1246 if ( array_pop( $tr_history ) ) {
1247 $out .= "</tr>\n";
1248 }
1249 if ( !array_pop( $has_opened_tr ) ) {
1250 $out .= "<tr><td></td></tr>\n";
1251 }
1252
1253 $out .= "</table>\n";
1254 }
1255
1256 # Remove trailing line-ending (b/c)
1257 if ( substr( $out, -1 ) === "\n" ) {
1258 $out = substr( $out, 0, -1 );
1259 }
1260
1261 # special case: don't return empty table
1262 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1263 $out = '';
1264 }
1265
1266 return $out;
1267 }
1268
1269 /**
1270 * Helper function for parse() that transforms wiki markup into half-parsed
1271 * HTML. Only called for $mOutputType == self::OT_HTML.
1272 *
1273 * @private
1274 *
1275 * @param string $text The text to parse
1276 * @param bool $isMain Whether this is being called from the main parse() function
1277 * @param PPFrame|bool $frame A pre-processor frame
1278 *
1279 * @return string
1280 */
1281 public function internalParse( $text, $isMain = true, $frame = false ) {
1282
1283 $origText = $text;
1284
1285 // Avoid PHP 7.1 warning from passing $this by reference
1286 $parser = $this;
1287
1288 # Hook to suspend the parser in this state
1289 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1290 return $text;
1291 }
1292
1293 # if $frame is provided, then use $frame for replacing any variables
1294 if ( $frame ) {
1295 # use frame depth to infer how include/noinclude tags should be handled
1296 # depth=0 means this is the top-level document; otherwise it's an included document
1297 if ( !$frame->depth ) {
1298 $flag = 0;
1299 } else {
1300 $flag = Parser::PTD_FOR_INCLUSION;
1301 }
1302 $dom = $this->preprocessToDom( $text, $flag );
1303 $text = $frame->expand( $dom );
1304 } else {
1305 # if $frame is not provided, then use old-style replaceVariables
1306 $text = $this->replaceVariables( $text );
1307 }
1308
1309 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1310 $text = Sanitizer::removeHTMLtags(
1311 $text,
1312 [ $this, 'attributeStripCallback' ],
1313 false,
1314 array_keys( $this->mTransparentTagHooks ),
1315 [],
1316 [ $this, 'addTrackingCategory' ]
1317 );
1318 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1319
1320 # Tables need to come after variable replacement for things to work
1321 # properly; putting them before other transformations should keep
1322 # exciting things like link expansions from showing up in surprising
1323 # places.
1324 $text = $this->doTableStuff( $text );
1325
1326 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1327
1328 $text = $this->doDoubleUnderscore( $text );
1329
1330 $text = $this->doHeadings( $text );
1331 $text = $this->replaceInternalLinks( $text );
1332 $text = $this->doAllQuotes( $text );
1333 $text = $this->replaceExternalLinks( $text );
1334
1335 # replaceInternalLinks may sometimes leave behind
1336 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1337 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1338
1339 $text = $this->doMagicLinks( $text );
1340 $text = $this->formatHeadings( $text, $origText, $isMain );
1341
1342 return $text;
1343 }
1344
1345 /**
1346 * Helper function for parse() that transforms half-parsed HTML into fully
1347 * parsed HTML.
1348 *
1349 * @param string $text
1350 * @param bool $isMain
1351 * @param bool $linestart
1352 * @return string
1353 */
1354 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1355 $text = $this->mStripState->unstripGeneral( $text );
1356
1357 // Avoid PHP 7.1 warning from passing $this by reference
1358 $parser = $this;
1359
1360 if ( $isMain ) {
1361 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1362 }
1363
1364 # Clean up special characters, only run once, next-to-last before doBlockLevels
1365 $fixtags = [
1366 # French spaces, last one Guillemet-left
1367 # only if there is something before the space
1368 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1369 # french spaces, Guillemet-right
1370 '/(\\302\\253) /' => '\\1&#160;',
1371 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1372 ];
1373 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1374
1375 $text = $this->doBlockLevels( $text, $linestart );
1376
1377 $this->replaceLinkHolders( $text );
1378
1379 /**
1380 * The input doesn't get language converted if
1381 * a) It's disabled
1382 * b) Content isn't converted
1383 * c) It's a conversion table
1384 * d) it is an interface message (which is in the user language)
1385 */
1386 if ( !( $this->mOptions->getDisableContentConversion()
1387 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1388 ) {
1389 if ( !$this->mOptions->getInterfaceMessage() ) {
1390 # The position of the convert() call should not be changed. it
1391 # assumes that the links are all replaced and the only thing left
1392 # is the <nowiki> mark.
1393 $text = $this->getConverterLanguage()->convert( $text );
1394 }
1395 }
1396
1397 $text = $this->mStripState->unstripNoWiki( $text );
1398
1399 if ( $isMain ) {
1400 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1401 }
1402
1403 $text = $this->replaceTransparentTags( $text );
1404 $text = $this->mStripState->unstripGeneral( $text );
1405
1406 $text = Sanitizer::normalizeCharReferences( $text );
1407
1408 if ( MWTidy::isEnabled() ) {
1409 if ( $this->mOptions->getTidy() ) {
1410 $text = MWTidy::tidy( $text );
1411 }
1412 } else {
1413 # attempt to sanitize at least some nesting problems
1414 # (T4702 and quite a few others)
1415 $tidyregs = [
1416 # ''Something [http://www.cool.com cool''] -->
1417 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1418 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1419 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1420 # fix up an anchor inside another anchor, only
1421 # at least for a single single nested link (T5695)
1422 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1423 '\\1\\2</a>\\3</a>\\1\\4</a>',
1424 # fix div inside inline elements- doBlockLevels won't wrap a line which
1425 # contains a div, so fix it up here; replace
1426 # div with escaped text
1427 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1428 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1429 # remove empty italic or bold tag pairs, some
1430 # introduced by rules above
1431 '/<([bi])><\/\\1>/' => '',
1432 ];
1433
1434 $text = preg_replace(
1435 array_keys( $tidyregs ),
1436 array_values( $tidyregs ),
1437 $text );
1438 }
1439
1440 if ( $isMain ) {
1441 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1442 }
1443
1444 return $text;
1445 }
1446
1447 /**
1448 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1449 * magic external links.
1450 *
1451 * DML
1452 * @private
1453 *
1454 * @param string $text
1455 *
1456 * @return string
1457 */
1458 public function doMagicLinks( $text ) {
1459 $prots = wfUrlProtocolsWithoutProtRel();
1460 $urlChar = self::EXT_LINK_URL_CLASS;
1461 $addr = self::EXT_LINK_ADDR;
1462 $space = self::SPACE_NOT_NL; # non-newline space
1463 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1464 $spaces = "$space++"; # possessive match of 1 or more spaces
1465 $text = preg_replace_callback(
1466 '!(?: # Start cases
1467 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1468 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1469 (\b # m[3]: Free external links
1470 (?i:$prots)
1471 ($addr$urlChar*) # m[4]: Post-protocol path
1472 ) |
1473 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1474 ([0-9]+)\b |
1475 \bISBN $spaces ( # m[6]: ISBN, capture number
1476 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1477 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1478 [0-9Xx] # check digit
1479 )\b
1480 )!xu", [ $this, 'magicLinkCallback' ], $text );
1481 return $text;
1482 }
1483
1484 /**
1485 * @throws MWException
1486 * @param array $m
1487 * @return HTML|string
1488 */
1489 public function magicLinkCallback( $m ) {
1490 if ( isset( $m[1] ) && $m[1] !== '' ) {
1491 # Skip anchor
1492 return $m[0];
1493 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1494 # Skip HTML element
1495 return $m[0];
1496 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1497 # Free external link
1498 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1499 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1500 # RFC or PMID
1501 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1502 if ( !$this->mOptions->getMagicRFCLinks() ) {
1503 return $m[0];
1504 }
1505 $keyword = 'RFC';
1506 $urlmsg = 'rfcurl';
1507 $cssClass = 'mw-magiclink-rfc';
1508 $trackingCat = 'magiclink-tracking-rfc';
1509 $id = $m[5];
1510 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1511 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1512 return $m[0];
1513 }
1514 $keyword = 'PMID';
1515 $urlmsg = 'pubmedurl';
1516 $cssClass = 'mw-magiclink-pmid';
1517 $trackingCat = 'magiclink-tracking-pmid';
1518 $id = $m[5];
1519 } else {
1520 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1521 substr( $m[0], 0, 20 ) . '"' );
1522 }
1523 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1524 $this->addTrackingCategory( $trackingCat );
1525 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1526 } elseif ( isset( $m[6] ) && $m[6] !== ''
1527 && $this->mOptions->getMagicISBNLinks()
1528 ) {
1529 # ISBN
1530 $isbn = $m[6];
1531 $space = self::SPACE_NOT_NL; # non-newline space
1532 $isbn = preg_replace( "/$space/", ' ', $isbn );
1533 $num = strtr( $isbn, [
1534 '-' => '',
1535 ' ' => '',
1536 'x' => 'X',
1537 ] );
1538 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1539 return $this->getLinkRenderer()->makeKnownLink(
1540 SpecialPage::getTitleFor( 'Booksources', $num ),
1541 "ISBN $isbn",
1542 [
1543 'class' => 'internal mw-magiclink-isbn',
1544 'title' => false // suppress title attribute
1545 ]
1546 );
1547 } else {
1548 return $m[0];
1549 }
1550 }
1551
1552 /**
1553 * Make a free external link, given a user-supplied URL
1554 *
1555 * @param string $url
1556 * @param int $numPostProto
1557 * The number of characters after the protocol.
1558 * @return string HTML
1559 * @private
1560 */
1561 public function makeFreeExternalLink( $url, $numPostProto ) {
1562 $trail = '';
1563
1564 # The characters '<' and '>' (which were escaped by
1565 # removeHTMLtags()) should not be included in
1566 # URLs, per RFC 2396.
1567 # Make &nbsp; terminate a URL as well (bug T84937)
1568 $m2 = [];
1569 if ( preg_match(
1570 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1571 $url,
1572 $m2,
1573 PREG_OFFSET_CAPTURE
1574 ) ) {
1575 $trail = substr( $url, $m2[0][1] ) . $trail;
1576 $url = substr( $url, 0, $m2[0][1] );
1577 }
1578
1579 # Move trailing punctuation to $trail
1580 $sep = ',;\.:!?';
1581 # If there is no left bracket, then consider right brackets fair game too
1582 if ( strpos( $url, '(' ) === false ) {
1583 $sep .= ')';
1584 }
1585
1586 $urlRev = strrev( $url );
1587 $numSepChars = strspn( $urlRev, $sep );
1588 # Don't break a trailing HTML entity by moving the ; into $trail
1589 # This is in hot code, so use substr_compare to avoid having to
1590 # create a new string object for the comparison
1591 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1592 # more optimization: instead of running preg_match with a $
1593 # anchor, which can be slow, do the match on the reversed
1594 # string starting at the desired offset.
1595 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1596 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1597 $numSepChars--;
1598 }
1599 }
1600 if ( $numSepChars ) {
1601 $trail = substr( $url, -$numSepChars ) . $trail;
1602 $url = substr( $url, 0, -$numSepChars );
1603 }
1604
1605 # Verify that we still have a real URL after trail removal, and
1606 # not just lone protocol
1607 if ( strlen( $trail ) >= $numPostProto ) {
1608 return $url . $trail;
1609 }
1610
1611 $url = Sanitizer::cleanUrl( $url );
1612
1613 # Is this an external image?
1614 $text = $this->maybeMakeExternalImage( $url );
1615 if ( $text === false ) {
1616 # Not an image, make a link
1617 $text = Linker::makeExternalLink( $url,
1618 $this->getConverterLanguage()->markNoConversion( $url, true ),
1619 true, 'free',
1620 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1621 # Register it in the output object...
1622 $this->mOutput->addExternalLink( $url );
1623 }
1624 return $text . $trail;
1625 }
1626
1627 /**
1628 * Parse headers and return html
1629 *
1630 * @private
1631 *
1632 * @param string $text
1633 *
1634 * @return string
1635 */
1636 public function doHeadings( $text ) {
1637 for ( $i = 6; $i >= 1; --$i ) {
1638 $h = str_repeat( '=', $i );
1639 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1640 }
1641 return $text;
1642 }
1643
1644 /**
1645 * Replace single quotes with HTML markup
1646 * @private
1647 *
1648 * @param string $text
1649 *
1650 * @return string The altered text
1651 */
1652 public function doAllQuotes( $text ) {
1653 $outtext = '';
1654 $lines = StringUtils::explode( "\n", $text );
1655 foreach ( $lines as $line ) {
1656 $outtext .= $this->doQuotes( $line ) . "\n";
1657 }
1658 $outtext = substr( $outtext, 0, -1 );
1659 return $outtext;
1660 }
1661
1662 /**
1663 * Helper function for doAllQuotes()
1664 *
1665 * @param string $text
1666 *
1667 * @return string
1668 */
1669 public function doQuotes( $text ) {
1670 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1671 $countarr = count( $arr );
1672 if ( $countarr == 1 ) {
1673 return $text;
1674 }
1675
1676 // First, do some preliminary work. This may shift some apostrophes from
1677 // being mark-up to being text. It also counts the number of occurrences
1678 // of bold and italics mark-ups.
1679 $numbold = 0;
1680 $numitalics = 0;
1681 for ( $i = 1; $i < $countarr; $i += 2 ) {
1682 $thislen = strlen( $arr[$i] );
1683 // If there are ever four apostrophes, assume the first is supposed to
1684 // be text, and the remaining three constitute mark-up for bold text.
1685 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1686 if ( $thislen == 4 ) {
1687 $arr[$i - 1] .= "'";
1688 $arr[$i] = "'''";
1689 $thislen = 3;
1690 } elseif ( $thislen > 5 ) {
1691 // If there are more than 5 apostrophes in a row, assume they're all
1692 // text except for the last 5.
1693 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1694 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1695 $arr[$i] = "'''''";
1696 $thislen = 5;
1697 }
1698 // Count the number of occurrences of bold and italics mark-ups.
1699 if ( $thislen == 2 ) {
1700 $numitalics++;
1701 } elseif ( $thislen == 3 ) {
1702 $numbold++;
1703 } elseif ( $thislen == 5 ) {
1704 $numitalics++;
1705 $numbold++;
1706 }
1707 }
1708
1709 // If there is an odd number of both bold and italics, it is likely
1710 // that one of the bold ones was meant to be an apostrophe followed
1711 // by italics. Which one we cannot know for certain, but it is more
1712 // likely to be one that has a single-letter word before it.
1713 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1714 $firstsingleletterword = -1;
1715 $firstmultiletterword = -1;
1716 $firstspace = -1;
1717 for ( $i = 1; $i < $countarr; $i += 2 ) {
1718 if ( strlen( $arr[$i] ) == 3 ) {
1719 $x1 = substr( $arr[$i - 1], -1 );
1720 $x2 = substr( $arr[$i - 1], -2, 1 );
1721 if ( $x1 === ' ' ) {
1722 if ( $firstspace == -1 ) {
1723 $firstspace = $i;
1724 }
1725 } elseif ( $x2 === ' ' ) {
1726 $firstsingleletterword = $i;
1727 // if $firstsingleletterword is set, we don't
1728 // look at the other options, so we can bail early.
1729 break;
1730 } else {
1731 if ( $firstmultiletterword == -1 ) {
1732 $firstmultiletterword = $i;
1733 }
1734 }
1735 }
1736 }
1737
1738 // If there is a single-letter word, use it!
1739 if ( $firstsingleletterword > -1 ) {
1740 $arr[$firstsingleletterword] = "''";
1741 $arr[$firstsingleletterword - 1] .= "'";
1742 } elseif ( $firstmultiletterword > -1 ) {
1743 // If not, but there's a multi-letter word, use that one.
1744 $arr[$firstmultiletterword] = "''";
1745 $arr[$firstmultiletterword - 1] .= "'";
1746 } elseif ( $firstspace > -1 ) {
1747 // ... otherwise use the first one that has neither.
1748 // (notice that it is possible for all three to be -1 if, for example,
1749 // there is only one pentuple-apostrophe in the line)
1750 $arr[$firstspace] = "''";
1751 $arr[$firstspace - 1] .= "'";
1752 }
1753 }
1754
1755 // Now let's actually convert our apostrophic mush to HTML!
1756 $output = '';
1757 $buffer = '';
1758 $state = '';
1759 $i = 0;
1760 foreach ( $arr as $r ) {
1761 if ( ( $i % 2 ) == 0 ) {
1762 if ( $state === 'both' ) {
1763 $buffer .= $r;
1764 } else {
1765 $output .= $r;
1766 }
1767 } else {
1768 $thislen = strlen( $r );
1769 if ( $thislen == 2 ) {
1770 if ( $state === 'i' ) {
1771 $output .= '</i>';
1772 $state = '';
1773 } elseif ( $state === 'bi' ) {
1774 $output .= '</i>';
1775 $state = 'b';
1776 } elseif ( $state === 'ib' ) {
1777 $output .= '</b></i><b>';
1778 $state = 'b';
1779 } elseif ( $state === 'both' ) {
1780 $output .= '<b><i>' . $buffer . '</i>';
1781 $state = 'b';
1782 } else { // $state can be 'b' or ''
1783 $output .= '<i>';
1784 $state .= 'i';
1785 }
1786 } elseif ( $thislen == 3 ) {
1787 if ( $state === 'b' ) {
1788 $output .= '</b>';
1789 $state = '';
1790 } elseif ( $state === 'bi' ) {
1791 $output .= '</i></b><i>';
1792 $state = 'i';
1793 } elseif ( $state === 'ib' ) {
1794 $output .= '</b>';
1795 $state = 'i';
1796 } elseif ( $state === 'both' ) {
1797 $output .= '<i><b>' . $buffer . '</b>';
1798 $state = 'i';
1799 } else { // $state can be 'i' or ''
1800 $output .= '<b>';
1801 $state .= 'b';
1802 }
1803 } elseif ( $thislen == 5 ) {
1804 if ( $state === 'b' ) {
1805 $output .= '</b><i>';
1806 $state = 'i';
1807 } elseif ( $state === 'i' ) {
1808 $output .= '</i><b>';
1809 $state = 'b';
1810 } elseif ( $state === 'bi' ) {
1811 $output .= '</i></b>';
1812 $state = '';
1813 } elseif ( $state === 'ib' ) {
1814 $output .= '</b></i>';
1815 $state = '';
1816 } elseif ( $state === 'both' ) {
1817 $output .= '<i><b>' . $buffer . '</b></i>';
1818 $state = '';
1819 } else { // ($state == '')
1820 $buffer = '';
1821 $state = 'both';
1822 }
1823 }
1824 }
1825 $i++;
1826 }
1827 // Now close all remaining tags. Notice that the order is important.
1828 if ( $state === 'b' || $state === 'ib' ) {
1829 $output .= '</b>';
1830 }
1831 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1832 $output .= '</i>';
1833 }
1834 if ( $state === 'bi' ) {
1835 $output .= '</b>';
1836 }
1837 // There might be lonely ''''', so make sure we have a buffer
1838 if ( $state === 'both' && $buffer ) {
1839 $output .= '<b><i>' . $buffer . '</i></b>';
1840 }
1841 return $output;
1842 }
1843
1844 /**
1845 * Replace external links (REL)
1846 *
1847 * Note: this is all very hackish and the order of execution matters a lot.
1848 * Make sure to run tests/parser/parserTests.php if you change this code.
1849 *
1850 * @private
1851 *
1852 * @param string $text
1853 *
1854 * @throws MWException
1855 * @return string
1856 */
1857 public function replaceExternalLinks( $text ) {
1858
1859 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1860 if ( $bits === false ) {
1861 throw new MWException( "PCRE needs to be compiled with "
1862 . "--enable-unicode-properties in order for MediaWiki to function" );
1863 }
1864 $s = array_shift( $bits );
1865
1866 $i = 0;
1867 while ( $i < count( $bits ) ) {
1868 $url = $bits[$i++];
1869 $i++; // protocol
1870 $text = $bits[$i++];
1871 $trail = $bits[$i++];
1872
1873 # The characters '<' and '>' (which were escaped by
1874 # removeHTMLtags()) should not be included in
1875 # URLs, per RFC 2396.
1876 $m2 = [];
1877 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1878 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1879 $url = substr( $url, 0, $m2[0][1] );
1880 }
1881
1882 # If the link text is an image URL, replace it with an <img> tag
1883 # This happened by accident in the original parser, but some people used it extensively
1884 $img = $this->maybeMakeExternalImage( $text );
1885 if ( $img !== false ) {
1886 $text = $img;
1887 }
1888
1889 $dtrail = '';
1890
1891 # Set linktype for CSS - if URL==text, link is essentially free
1892 $linktype = ( $text === $url ) ? 'free' : 'text';
1893
1894 # No link text, e.g. [http://domain.tld/some.link]
1895 if ( $text == '' ) {
1896 # Autonumber
1897 $langObj = $this->getTargetLanguage();
1898 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1899 $linktype = 'autonumber';
1900 } else {
1901 # Have link text, e.g. [http://domain.tld/some.link text]s
1902 # Check for trail
1903 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1904 }
1905
1906 $text = $this->getConverterLanguage()->markNoConversion( $text );
1907
1908 $url = Sanitizer::cleanUrl( $url );
1909
1910 # Use the encoded URL
1911 # This means that users can paste URLs directly into the text
1912 # Funny characters like ö aren't valid in URLs anyway
1913 # This was changed in August 2004
1914 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1915 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1916
1917 # Register link in the output object.
1918 $this->mOutput->addExternalLink( $url );
1919 }
1920
1921 return $s;
1922 }
1923
1924 /**
1925 * Get the rel attribute for a particular external link.
1926 *
1927 * @since 1.21
1928 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1929 * nofollow if appropriate
1930 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1931 * @return string|null Rel attribute for $url
1932 */
1933 public static function getExternalLinkRel( $url = false, $title = null ) {
1934 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1935 $ns = $title ? $title->getNamespace() : false;
1936 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1937 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1938 ) {
1939 return 'nofollow';
1940 }
1941 return null;
1942 }
1943
1944 /**
1945 * Get an associative array of additional HTML attributes appropriate for a
1946 * particular external link. This currently may include rel => nofollow
1947 * (depending on configuration, namespace, and the URL's domain) and/or a
1948 * target attribute (depending on configuration).
1949 *
1950 * @param string $url URL to extract the domain from for rel =>
1951 * nofollow if appropriate
1952 * @return array Associative array of HTML attributes
1953 */
1954 public function getExternalLinkAttribs( $url ) {
1955 $attribs = [];
1956 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1957
1958 $target = $this->mOptions->getExternalLinkTarget();
1959 if ( $target ) {
1960 $attribs['target'] = $target;
1961 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1962 // T133507. New windows can navigate parent cross-origin.
1963 // Including noreferrer due to lacking browser
1964 // support of noopener. Eventually noreferrer should be removed.
1965 if ( $rel !== '' ) {
1966 $rel .= ' ';
1967 }
1968 $rel .= 'noreferrer noopener';
1969 }
1970 }
1971 $attribs['rel'] = $rel;
1972 return $attribs;
1973 }
1974
1975 /**
1976 * Replace unusual escape codes in a URL with their equivalent characters
1977 *
1978 * This generally follows the syntax defined in RFC 3986, with special
1979 * consideration for HTTP query strings.
1980 *
1981 * @param string $url
1982 * @return string
1983 */
1984 public static function normalizeLinkUrl( $url ) {
1985 # First, make sure unsafe characters are encoded
1986 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1987 function ( $m ) {
1988 return rawurlencode( $m[0] );
1989 },
1990 $url
1991 );
1992
1993 $ret = '';
1994 $end = strlen( $url );
1995
1996 # Fragment part - 'fragment'
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 # Query part - 'query' minus &=+;
2005 $start = strpos( $url, '?' );
2006 if ( $start !== false && $start < $end ) {
2007 $ret = self::normalizeUrlComponent(
2008 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2009 $end = $start;
2010 }
2011
2012 # Scheme and path part - 'pchar'
2013 # (we assume no userinfo or encoded colons in the host)
2014 $ret = self::normalizeUrlComponent(
2015 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2016
2017 return $ret;
2018 }
2019
2020 private static function normalizeUrlComponent( $component, $unsafe ) {
2021 $callback = function ( $matches ) use ( $unsafe ) {
2022 $char = urldecode( $matches[0] );
2023 $ord = ord( $char );
2024 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2025 # Unescape it
2026 return $char;
2027 } else {
2028 # Leave it escaped, but use uppercase for a-f
2029 return strtoupper( $matches[0] );
2030 }
2031 };
2032 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2033 }
2034
2035 /**
2036 * make an image if it's allowed, either through the global
2037 * option, through the exception, or through the on-wiki whitelist
2038 *
2039 * @param string $url
2040 *
2041 * @return string
2042 */
2043 private function maybeMakeExternalImage( $url ) {
2044 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2045 $imagesexception = !empty( $imagesfrom );
2046 $text = false;
2047 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2048 if ( $imagesexception && is_array( $imagesfrom ) ) {
2049 $imagematch = false;
2050 foreach ( $imagesfrom as $match ) {
2051 if ( strpos( $url, $match ) === 0 ) {
2052 $imagematch = true;
2053 break;
2054 }
2055 }
2056 } elseif ( $imagesexception ) {
2057 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2058 } else {
2059 $imagematch = false;
2060 }
2061
2062 if ( $this->mOptions->getAllowExternalImages()
2063 || ( $imagesexception && $imagematch )
2064 ) {
2065 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2066 # Image found
2067 $text = Linker::makeExternalImage( $url );
2068 }
2069 }
2070 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2071 && preg_match( self::EXT_IMAGE_REGEX, $url )
2072 ) {
2073 $whitelist = explode(
2074 "\n",
2075 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2076 );
2077
2078 foreach ( $whitelist as $entry ) {
2079 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2080 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2081 continue;
2082 }
2083 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2084 # Image matches a whitelist entry
2085 $text = Linker::makeExternalImage( $url );
2086 break;
2087 }
2088 }
2089 }
2090 return $text;
2091 }
2092
2093 /**
2094 * Process [[ ]] wikilinks
2095 *
2096 * @param string $s
2097 *
2098 * @return string Processed text
2099 *
2100 * @private
2101 */
2102 public function replaceInternalLinks( $s ) {
2103 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2104 return $s;
2105 }
2106
2107 /**
2108 * Process [[ ]] wikilinks (RIL)
2109 * @param string $s
2110 * @throws MWException
2111 * @return LinkHolderArray
2112 *
2113 * @private
2114 */
2115 public function replaceInternalLinks2( &$s ) {
2116 global $wgExtraInterlanguageLinkPrefixes;
2117
2118 static $tc = false, $e1, $e1_img;
2119 # the % is needed to support urlencoded titles as well
2120 if ( !$tc ) {
2121 $tc = Title::legalChars() . '#%';
2122 # Match a link having the form [[namespace:link|alternate]]trail
2123 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2124 # Match cases where there is no "]]", which might still be images
2125 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2126 }
2127
2128 $holders = new LinkHolderArray( $this );
2129
2130 # split the entire text string on occurrences of [[
2131 $a = StringUtils::explode( '[[', ' ' . $s );
2132 # get the first element (all text up to first [[), and remove the space we added
2133 $s = $a->current();
2134 $a->next();
2135 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2136 $s = substr( $s, 1 );
2137
2138 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2139 $e2 = null;
2140 if ( $useLinkPrefixExtension ) {
2141 # Match the end of a line for a word that's not followed by whitespace,
2142 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2143 global $wgContLang;
2144 $charset = $wgContLang->linkPrefixCharset();
2145 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2146 }
2147
2148 if ( is_null( $this->mTitle ) ) {
2149 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2150 }
2151 $nottalk = !$this->mTitle->isTalkPage();
2152
2153 if ( $useLinkPrefixExtension ) {
2154 $m = [];
2155 if ( preg_match( $e2, $s, $m ) ) {
2156 $first_prefix = $m[2];
2157 } else {
2158 $first_prefix = false;
2159 }
2160 } else {
2161 $prefix = '';
2162 }
2163
2164 $useSubpages = $this->areSubpagesAllowed();
2165
2166 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2167 # Loop for each link
2168 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2169 // @codingStandardsIgnoreEnd
2170
2171 # Check for excessive memory usage
2172 if ( $holders->isBig() ) {
2173 # Too big
2174 # Do the existence check, replace the link holders and clear the array
2175 $holders->replace( $s );
2176 $holders->clear();
2177 }
2178
2179 if ( $useLinkPrefixExtension ) {
2180 if ( preg_match( $e2, $s, $m ) ) {
2181 $prefix = $m[2];
2182 $s = $m[1];
2183 } else {
2184 $prefix = '';
2185 }
2186 # first link
2187 if ( $first_prefix ) {
2188 $prefix = $first_prefix;
2189 $first_prefix = false;
2190 }
2191 }
2192
2193 $might_be_img = false;
2194
2195 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2196 $text = $m[2];
2197 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2198 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2199 # the real problem is with the $e1 regex
2200 # See T1500.
2201 # Still some problems for cases where the ] is meant to be outside punctuation,
2202 # and no image is in sight. See T4095.
2203 if ( $text !== ''
2204 && substr( $m[3], 0, 1 ) === ']'
2205 && strpos( $text, '[' ) !== false
2206 ) {
2207 $text .= ']'; # so that replaceExternalLinks($text) works later
2208 $m[3] = substr( $m[3], 1 );
2209 }
2210 # fix up urlencoded title texts
2211 if ( strpos( $m[1], '%' ) !== false ) {
2212 # Should anchors '#' also be rejected?
2213 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2214 }
2215 $trail = $m[3];
2216 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2217 # Invalid, but might be an image with a link in its caption
2218 $might_be_img = true;
2219 $text = $m[2];
2220 if ( strpos( $m[1], '%' ) !== false ) {
2221 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2222 }
2223 $trail = "";
2224 } else { # Invalid form; output directly
2225 $s .= $prefix . '[[' . $line;
2226 continue;
2227 }
2228
2229 $origLink = ltrim( $m[1], ' ' );
2230
2231 # Don't allow internal links to pages containing
2232 # PROTO: where PROTO is a valid URL protocol; these
2233 # should be external links.
2234 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2235 $s .= $prefix . '[[' . $line;
2236 continue;
2237 }
2238
2239 # Make subpage if necessary
2240 if ( $useSubpages ) {
2241 $link = $this->maybeDoSubpageLink( $origLink, $text );
2242 } else {
2243 $link = $origLink;
2244 }
2245
2246 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2247 if ( !$noforce ) {
2248 # Strip off leading ':'
2249 $link = substr( $link, 1 );
2250 }
2251
2252 $unstrip = $this->mStripState->unstripNoWiki( $link );
2253 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2254 if ( $nt === null ) {
2255 $s .= $prefix . '[[' . $line;
2256 continue;
2257 }
2258
2259 $ns = $nt->getNamespace();
2260 $iw = $nt->getInterwiki();
2261
2262 if ( $might_be_img ) { # if this is actually an invalid link
2263 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2264 $found = false;
2265 while ( true ) {
2266 # look at the next 'line' to see if we can close it there
2267 $a->next();
2268 $next_line = $a->current();
2269 if ( $next_line === false || $next_line === null ) {
2270 break;
2271 }
2272 $m = explode( ']]', $next_line, 3 );
2273 if ( count( $m ) == 3 ) {
2274 # the first ]] closes the inner link, the second the image
2275 $found = true;
2276 $text .= "[[{$m[0]}]]{$m[1]}";
2277 $trail = $m[2];
2278 break;
2279 } elseif ( count( $m ) == 2 ) {
2280 # if there's exactly one ]] that's fine, we'll keep looking
2281 $text .= "[[{$m[0]}]]{$m[1]}";
2282 } else {
2283 # if $next_line is invalid too, we need look no further
2284 $text .= '[[' . $next_line;
2285 break;
2286 }
2287 }
2288 if ( !$found ) {
2289 # we couldn't find the end of this imageLink, so output it raw
2290 # but don't ignore what might be perfectly normal links in the text we've examined
2291 $holders->merge( $this->replaceInternalLinks2( $text ) );
2292 $s .= "{$prefix}[[$link|$text";
2293 # note: no $trail, because without an end, there *is* no trail
2294 continue;
2295 }
2296 } else { # it's not an image, so output it raw
2297 $s .= "{$prefix}[[$link|$text";
2298 # note: no $trail, because without an end, there *is* no trail
2299 continue;
2300 }
2301 }
2302
2303 $wasblank = ( $text == '' );
2304 if ( $wasblank ) {
2305 $text = $link;
2306 } else {
2307 # T6598 madness. Handle the quotes only if they come from the alternate part
2308 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2309 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2310 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2311 $text = $this->doQuotes( $text );
2312 }
2313
2314 # Link not escaped by : , create the various objects
2315 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2316 # Interwikis
2317 if (
2318 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2319 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2320 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2321 )
2322 ) {
2323 # T26502: filter duplicates
2324 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2325 $this->mLangLinkLanguages[$iw] = true;
2326 $this->mOutput->addLanguageLink( $nt->getFullText() );
2327 }
2328
2329 $s = rtrim( $s . $prefix );
2330 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2331 continue;
2332 }
2333
2334 if ( $ns == NS_FILE ) {
2335 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2336 if ( $wasblank ) {
2337 # if no parameters were passed, $text
2338 # becomes something like "File:Foo.png",
2339 # which we don't want to pass on to the
2340 # image generator
2341 $text = '';
2342 } else {
2343 # recursively parse links inside the image caption
2344 # actually, this will parse them in any other parameters, too,
2345 # but it might be hard to fix that, and it doesn't matter ATM
2346 $text = $this->replaceExternalLinks( $text );
2347 $holders->merge( $this->replaceInternalLinks2( $text ) );
2348 }
2349 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2350 $s .= $prefix . $this->armorLinks(
2351 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2352 continue;
2353 }
2354 } elseif ( $ns == NS_CATEGORY ) {
2355 $s = rtrim( $s . "\n" ); # T2087
2356
2357 if ( $wasblank ) {
2358 $sortkey = $this->getDefaultSort();
2359 } else {
2360 $sortkey = $text;
2361 }
2362 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2363 $sortkey = str_replace( "\n", '', $sortkey );
2364 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2365 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2366
2367 /**
2368 * Strip the whitespace Category links produce, see T2087
2369 */
2370 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2371
2372 continue;
2373 }
2374 }
2375
2376 # Self-link checking. For some languages, variants of the title are checked in
2377 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2378 # for linking to a different variant.
2379 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2380 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2381 continue;
2382 }
2383
2384 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2385 # @todo FIXME: Should do batch file existence checks, see comment below
2386 if ( $ns == NS_MEDIA ) {
2387 # Give extensions a chance to select the file revision for us
2388 $options = [];
2389 $descQuery = false;
2390 Hooks::run( 'BeforeParserFetchFileAndTitle',
2391 [ $this, $nt, &$options, &$descQuery ] );
2392 # Fetch and register the file (file title may be different via hooks)
2393 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2394 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2395 $s .= $prefix . $this->armorLinks(
2396 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2397 continue;
2398 }
2399
2400 # Some titles, such as valid special pages or files in foreign repos, should
2401 # be shown as bluelinks even though they're not included in the page table
2402 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2403 # batch file existence checks for NS_FILE and NS_MEDIA
2404 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2405 $this->mOutput->addLink( $nt );
2406 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2407 } else {
2408 # Links will be added to the output link list after checking
2409 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2410 }
2411 }
2412 return $holders;
2413 }
2414
2415 /**
2416 * Render a forced-blue link inline; protect against double expansion of
2417 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2418 * Since this little disaster has to split off the trail text to avoid
2419 * breaking URLs in the following text without breaking trails on the
2420 * wiki links, it's been made into a horrible function.
2421 *
2422 * @param Title $nt
2423 * @param string $text
2424 * @param string $trail
2425 * @param string $prefix
2426 * @return string HTML-wikitext mix oh yuck
2427 */
2428 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2429 list( $inside, $trail ) = Linker::splitTrail( $trail );
2430
2431 if ( $text == '' ) {
2432 $text = htmlspecialchars( $nt->getPrefixedText() );
2433 }
2434
2435 $link = $this->getLinkRenderer()->makeKnownLink(
2436 $nt, new HtmlArmor( "$prefix$text$inside" )
2437 );
2438
2439 return $this->armorLinks( $link ) . $trail;
2440 }
2441
2442 /**
2443 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2444 * going to go through further parsing steps before inline URL expansion.
2445 *
2446 * Not needed quite as much as it used to be since free links are a bit
2447 * more sensible these days. But bracketed links are still an issue.
2448 *
2449 * @param string $text More-or-less HTML
2450 * @return string Less-or-more HTML with NOPARSE bits
2451 */
2452 public function armorLinks( $text ) {
2453 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2454 self::MARKER_PREFIX . "NOPARSE$1", $text );
2455 }
2456
2457 /**
2458 * Return true if subpage links should be expanded on this page.
2459 * @return bool
2460 */
2461 public function areSubpagesAllowed() {
2462 # Some namespaces don't allow subpages
2463 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2464 }
2465
2466 /**
2467 * Handle link to subpage if necessary
2468 *
2469 * @param string $target The source of the link
2470 * @param string &$text The link text, modified as necessary
2471 * @return string The full name of the link
2472 * @private
2473 */
2474 public function maybeDoSubpageLink( $target, &$text ) {
2475 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2476 }
2477
2478 /**
2479 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2480 *
2481 * @param string $text
2482 * @param bool $linestart Whether or not this is at the start of a line.
2483 * @private
2484 * @return string The lists rendered as HTML
2485 */
2486 public function doBlockLevels( $text, $linestart ) {
2487 return BlockLevelPass::doBlockLevels( $text, $linestart );
2488 }
2489
2490 /**
2491 * Return value of a magic variable (like PAGENAME)
2492 *
2493 * @private
2494 *
2495 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2496 * @param bool|PPFrame $frame
2497 *
2498 * @throws MWException
2499 * @return string
2500 */
2501 public function getVariableValue( $index, $frame = false ) {
2502 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2503 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2504
2505 if ( is_null( $this->mTitle ) ) {
2506 // If no title set, bad things are going to happen
2507 // later. Title should always be set since this
2508 // should only be called in the middle of a parse
2509 // operation (but the unit-tests do funky stuff)
2510 throw new MWException( __METHOD__ . ' Should only be '
2511 . ' called while parsing (no title set)' );
2512 }
2513
2514 // Avoid PHP 7.1 warning from passing $this by reference
2515 $parser = $this;
2516
2517 /**
2518 * Some of these require message or data lookups and can be
2519 * expensive to check many times.
2520 */
2521 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2522 if ( isset( $this->mVarCache[$index] ) ) {
2523 return $this->mVarCache[$index];
2524 }
2525 }
2526
2527 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2528 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2529
2530 $pageLang = $this->getFunctionLang();
2531
2532 switch ( $index ) {
2533 case '!':
2534 $value = '|';
2535 break;
2536 case 'currentmonth':
2537 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2538 break;
2539 case 'currentmonth1':
2540 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2541 break;
2542 case 'currentmonthname':
2543 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2544 break;
2545 case 'currentmonthnamegen':
2546 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2547 break;
2548 case 'currentmonthabbrev':
2549 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2550 break;
2551 case 'currentday':
2552 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2553 break;
2554 case 'currentday2':
2555 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2556 break;
2557 case 'localmonth':
2558 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2559 break;
2560 case 'localmonth1':
2561 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2562 break;
2563 case 'localmonthname':
2564 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2565 break;
2566 case 'localmonthnamegen':
2567 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2568 break;
2569 case 'localmonthabbrev':
2570 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2571 break;
2572 case 'localday':
2573 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2574 break;
2575 case 'localday2':
2576 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2577 break;
2578 case 'pagename':
2579 $value = wfEscapeWikiText( $this->mTitle->getText() );
2580 break;
2581 case 'pagenamee':
2582 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2583 break;
2584 case 'fullpagename':
2585 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2586 break;
2587 case 'fullpagenamee':
2588 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2589 break;
2590 case 'subpagename':
2591 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2592 break;
2593 case 'subpagenamee':
2594 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2595 break;
2596 case 'rootpagename':
2597 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2598 break;
2599 case 'rootpagenamee':
2600 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2601 ' ',
2602 '_',
2603 $this->mTitle->getRootText()
2604 ) ) );
2605 break;
2606 case 'basepagename':
2607 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2608 break;
2609 case 'basepagenamee':
2610 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2611 ' ',
2612 '_',
2613 $this->mTitle->getBaseText()
2614 ) ) );
2615 break;
2616 case 'talkpagename':
2617 if ( $this->mTitle->canTalk() ) {
2618 $talkPage = $this->mTitle->getTalkPage();
2619 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2620 } else {
2621 $value = '';
2622 }
2623 break;
2624 case 'talkpagenamee':
2625 if ( $this->mTitle->canTalk() ) {
2626 $talkPage = $this->mTitle->getTalkPage();
2627 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2628 } else {
2629 $value = '';
2630 }
2631 break;
2632 case 'subjectpagename':
2633 $subjPage = $this->mTitle->getSubjectPage();
2634 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2635 break;
2636 case 'subjectpagenamee':
2637 $subjPage = $this->mTitle->getSubjectPage();
2638 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2639 break;
2640 case 'pageid': // requested in T25427
2641 $pageid = $this->getTitle()->getArticleID();
2642 if ( $pageid == 0 ) {
2643 # 0 means the page doesn't exist in the database,
2644 # which means the user is previewing a new page.
2645 # The vary-revision flag must be set, because the magic word
2646 # will have a different value once the page is saved.
2647 $this->mOutput->setFlag( 'vary-revision' );
2648 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2649 }
2650 $value = $pageid ? $pageid : null;
2651 break;
2652 case 'revisionid':
2653 # Let the edit saving system know we should parse the page
2654 # *after* a revision ID has been assigned.
2655 $this->mOutput->setFlag( 'vary-revision-id' );
2656 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2657 $value = $this->mRevisionId;
2658 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2659 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2660 $this->mOutput->setSpeculativeRevIdUsed( $value );
2661 }
2662 break;
2663 case 'revisionday':
2664 # Let the edit saving system know we should parse the page
2665 # *after* a revision ID has been assigned. This is for null edits.
2666 $this->mOutput->setFlag( 'vary-revision' );
2667 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2668 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2669 break;
2670 case 'revisionday2':
2671 # Let the edit saving system know we should parse the page
2672 # *after* a revision ID has been assigned. This is for null edits.
2673 $this->mOutput->setFlag( 'vary-revision' );
2674 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2675 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2676 break;
2677 case 'revisionmonth':
2678 # Let the edit saving system know we should parse the page
2679 # *after* a revision ID has been assigned. This is for null edits.
2680 $this->mOutput->setFlag( 'vary-revision' );
2681 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2682 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2683 break;
2684 case 'revisionmonth1':
2685 # Let the edit saving system know we should parse the page
2686 # *after* a revision ID has been assigned. This is for null edits.
2687 $this->mOutput->setFlag( 'vary-revision' );
2688 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2689 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2690 break;
2691 case 'revisionyear':
2692 # Let the edit saving system know we should parse the page
2693 # *after* a revision ID has been assigned. This is for null edits.
2694 $this->mOutput->setFlag( 'vary-revision' );
2695 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2696 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2697 break;
2698 case 'revisiontimestamp':
2699 # Let the edit saving system know we should parse the page
2700 # *after* a revision ID has been assigned. This is for null edits.
2701 $this->mOutput->setFlag( 'vary-revision' );
2702 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2703 $value = $this->getRevisionTimestamp();
2704 break;
2705 case 'revisionuser':
2706 # Let the edit saving system know we should parse the page
2707 # *after* a revision ID has been assigned for null edits.
2708 $this->mOutput->setFlag( 'vary-user' );
2709 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2710 $value = $this->getRevisionUser();
2711 break;
2712 case 'revisionsize':
2713 $value = $this->getRevisionSize();
2714 break;
2715 case 'namespace':
2716 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2717 break;
2718 case 'namespacee':
2719 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2720 break;
2721 case 'namespacenumber':
2722 $value = $this->mTitle->getNamespace();
2723 break;
2724 case 'talkspace':
2725 $value = $this->mTitle->canTalk()
2726 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2727 : '';
2728 break;
2729 case 'talkspacee':
2730 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2731 break;
2732 case 'subjectspace':
2733 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2734 break;
2735 case 'subjectspacee':
2736 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2737 break;
2738 case 'currentdayname':
2739 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2740 break;
2741 case 'currentyear':
2742 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2743 break;
2744 case 'currenttime':
2745 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2746 break;
2747 case 'currenthour':
2748 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2749 break;
2750 case 'currentweek':
2751 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2752 # int to remove the padding
2753 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2754 break;
2755 case 'currentdow':
2756 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2757 break;
2758 case 'localdayname':
2759 $value = $pageLang->getWeekdayName(
2760 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2761 );
2762 break;
2763 case 'localyear':
2764 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2765 break;
2766 case 'localtime':
2767 $value = $pageLang->time(
2768 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2769 false,
2770 false
2771 );
2772 break;
2773 case 'localhour':
2774 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2775 break;
2776 case 'localweek':
2777 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2778 # int to remove the padding
2779 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2780 break;
2781 case 'localdow':
2782 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2783 break;
2784 case 'numberofarticles':
2785 $value = $pageLang->formatNum( SiteStats::articles() );
2786 break;
2787 case 'numberoffiles':
2788 $value = $pageLang->formatNum( SiteStats::images() );
2789 break;
2790 case 'numberofusers':
2791 $value = $pageLang->formatNum( SiteStats::users() );
2792 break;
2793 case 'numberofactiveusers':
2794 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2795 break;
2796 case 'numberofpages':
2797 $value = $pageLang->formatNum( SiteStats::pages() );
2798 break;
2799 case 'numberofadmins':
2800 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2801 break;
2802 case 'numberofedits':
2803 $value = $pageLang->formatNum( SiteStats::edits() );
2804 break;
2805 case 'currenttimestamp':
2806 $value = wfTimestamp( TS_MW, $ts );
2807 break;
2808 case 'localtimestamp':
2809 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2810 break;
2811 case 'currentversion':
2812 $value = SpecialVersion::getVersion();
2813 break;
2814 case 'articlepath':
2815 return $wgArticlePath;
2816 case 'sitename':
2817 return $wgSitename;
2818 case 'server':
2819 return $wgServer;
2820 case 'servername':
2821 return $wgServerName;
2822 case 'scriptpath':
2823 return $wgScriptPath;
2824 case 'stylepath':
2825 return $wgStylePath;
2826 case 'directionmark':
2827 return $pageLang->getDirMark();
2828 case 'contentlanguage':
2829 global $wgLanguageCode;
2830 return $wgLanguageCode;
2831 case 'pagelanguage':
2832 $value = $pageLang->getCode();
2833 break;
2834 case 'cascadingsources':
2835 $value = CoreParserFunctions::cascadingsources( $this );
2836 break;
2837 default:
2838 $ret = null;
2839 Hooks::run(
2840 'ParserGetVariableValueSwitch',
2841 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2842 );
2843
2844 return $ret;
2845 }
2846
2847 if ( $index ) {
2848 $this->mVarCache[$index] = $value;
2849 }
2850
2851 return $value;
2852 }
2853
2854 /**
2855 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2856 *
2857 * @private
2858 */
2859 public function initialiseVariables() {
2860 $variableIDs = MagicWord::getVariableIDs();
2861 $substIDs = MagicWord::getSubstIDs();
2862
2863 $this->mVariables = new MagicWordArray( $variableIDs );
2864 $this->mSubstWords = new MagicWordArray( $substIDs );
2865 }
2866
2867 /**
2868 * Preprocess some wikitext and return the document tree.
2869 * This is the ghost of replace_variables().
2870 *
2871 * @param string $text The text to parse
2872 * @param int $flags Bitwise combination of:
2873 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2874 * included. Default is to assume a direct page view.
2875 *
2876 * The generated DOM tree must depend only on the input text and the flags.
2877 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2878 *
2879 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2880 * change in the DOM tree for a given text, must be passed through the section identifier
2881 * in the section edit link and thus back to extractSections().
2882 *
2883 * The output of this function is currently only cached in process memory, but a persistent
2884 * cache may be implemented at a later date which takes further advantage of these strict
2885 * dependency requirements.
2886 *
2887 * @return PPNode
2888 */
2889 public function preprocessToDom( $text, $flags = 0 ) {
2890 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2891 return $dom;
2892 }
2893
2894 /**
2895 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2896 *
2897 * @param string $s
2898 *
2899 * @return array
2900 */
2901 public static function splitWhitespace( $s ) {
2902 $ltrimmed = ltrim( $s );
2903 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2904 $trimmed = rtrim( $ltrimmed );
2905 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2906 if ( $diff > 0 ) {
2907 $w2 = substr( $ltrimmed, -$diff );
2908 } else {
2909 $w2 = '';
2910 }
2911 return [ $w1, $trimmed, $w2 ];
2912 }
2913
2914 /**
2915 * Replace magic variables, templates, and template arguments
2916 * with the appropriate text. Templates are substituted recursively,
2917 * taking care to avoid infinite loops.
2918 *
2919 * Note that the substitution depends on value of $mOutputType:
2920 * self::OT_WIKI: only {{subst:}} templates
2921 * self::OT_PREPROCESS: templates but not extension tags
2922 * self::OT_HTML: all templates and extension tags
2923 *
2924 * @param string $text The text to transform
2925 * @param bool|PPFrame $frame Object describing the arguments passed to the
2926 * template. Arguments may also be provided as an associative array, as
2927 * was the usual case before MW1.12. Providing arguments this way may be
2928 * useful for extensions wishing to perform variable replacement
2929 * explicitly.
2930 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2931 * double-brace expansion.
2932 * @return string
2933 */
2934 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2935 # Is there any text? Also, Prevent too big inclusions!
2936 $textSize = strlen( $text );
2937 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2938 return $text;
2939 }
2940
2941 if ( $frame === false ) {
2942 $frame = $this->getPreprocessor()->newFrame();
2943 } elseif ( !( $frame instanceof PPFrame ) ) {
2944 wfDebug( __METHOD__ . " called using plain parameters instead of "
2945 . "a PPFrame instance. Creating custom frame.\n" );
2946 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2947 }
2948
2949 $dom = $this->preprocessToDom( $text );
2950 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2951 $text = $frame->expand( $dom, $flags );
2952
2953 return $text;
2954 }
2955
2956 /**
2957 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2958 *
2959 * @param array $args
2960 *
2961 * @return array
2962 */
2963 public static function createAssocArgs( $args ) {
2964 $assocArgs = [];
2965 $index = 1;
2966 foreach ( $args as $arg ) {
2967 $eqpos = strpos( $arg, '=' );
2968 if ( $eqpos === false ) {
2969 $assocArgs[$index++] = $arg;
2970 } else {
2971 $name = trim( substr( $arg, 0, $eqpos ) );
2972 $value = trim( substr( $arg, $eqpos + 1 ) );
2973 if ( $value === false ) {
2974 $value = '';
2975 }
2976 if ( $name !== false ) {
2977 $assocArgs[$name] = $value;
2978 }
2979 }
2980 }
2981
2982 return $assocArgs;
2983 }
2984
2985 /**
2986 * Warn the user when a parser limitation is reached
2987 * Will warn at most once the user per limitation type
2988 *
2989 * The results are shown during preview and run through the Parser (See EditPage.php)
2990 *
2991 * @param string $limitationType Should be one of:
2992 * 'expensive-parserfunction' (corresponding messages:
2993 * 'expensive-parserfunction-warning',
2994 * 'expensive-parserfunction-category')
2995 * 'post-expand-template-argument' (corresponding messages:
2996 * 'post-expand-template-argument-warning',
2997 * 'post-expand-template-argument-category')
2998 * 'post-expand-template-inclusion' (corresponding messages:
2999 * 'post-expand-template-inclusion-warning',
3000 * 'post-expand-template-inclusion-category')
3001 * 'node-count-exceeded' (corresponding messages:
3002 * 'node-count-exceeded-warning',
3003 * 'node-count-exceeded-category')
3004 * 'expansion-depth-exceeded' (corresponding messages:
3005 * 'expansion-depth-exceeded-warning',
3006 * 'expansion-depth-exceeded-category')
3007 * @param string|int|null $current Current value
3008 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3009 * exceeded, provide the values (optional)
3010 */
3011 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3012 # does no harm if $current and $max are present but are unnecessary for the message
3013 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3014 # only during preview, and that would split the parser cache unnecessarily.
3015 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3016 ->text();
3017 $this->mOutput->addWarning( $warning );
3018 $this->addTrackingCategory( "$limitationType-category" );
3019 }
3020
3021 /**
3022 * Return the text of a template, after recursively
3023 * replacing any variables or templates within the template.
3024 *
3025 * @param array $piece The parts of the template
3026 * $piece['title']: the title, i.e. the part before the |
3027 * $piece['parts']: the parameter array
3028 * $piece['lineStart']: whether the brace was at the start of a line
3029 * @param PPFrame $frame The current frame, contains template arguments
3030 * @throws Exception
3031 * @return string The text of the template
3032 */
3033 public function braceSubstitution( $piece, $frame ) {
3034
3035 // Flags
3036
3037 // $text has been filled
3038 $found = false;
3039 // wiki markup in $text should be escaped
3040 $nowiki = false;
3041 // $text is HTML, armour it against wikitext transformation
3042 $isHTML = false;
3043 // Force interwiki transclusion to be done in raw mode not rendered
3044 $forceRawInterwiki = false;
3045 // $text is a DOM node needing expansion in a child frame
3046 $isChildObj = false;
3047 // $text is a DOM node needing expansion in the current frame
3048 $isLocalObj = false;
3049
3050 # Title object, where $text came from
3051 $title = false;
3052
3053 # $part1 is the bit before the first |, and must contain only title characters.
3054 # Various prefixes will be stripped from it later.
3055 $titleWithSpaces = $frame->expand( $piece['title'] );
3056 $part1 = trim( $titleWithSpaces );
3057 $titleText = false;
3058
3059 # Original title text preserved for various purposes
3060 $originalTitle = $part1;
3061
3062 # $args is a list of argument nodes, starting from index 0, not including $part1
3063 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3064 # below won't work b/c this $args isn't an object
3065 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3066
3067 $profileSection = null; // profile templates
3068
3069 # SUBST
3070 if ( !$found ) {
3071 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3072
3073 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3074 # Decide whether to expand template or keep wikitext as-is.
3075 if ( $this->ot['wiki'] ) {
3076 if ( $substMatch === false ) {
3077 $literal = true; # literal when in PST with no prefix
3078 } else {
3079 $literal = false; # expand when in PST with subst: or safesubst:
3080 }
3081 } else {
3082 if ( $substMatch == 'subst' ) {
3083 $literal = true; # literal when not in PST with plain subst:
3084 } else {
3085 $literal = false; # expand when not in PST with safesubst: or no prefix
3086 }
3087 }
3088 if ( $literal ) {
3089 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3090 $isLocalObj = true;
3091 $found = true;
3092 }
3093 }
3094
3095 # Variables
3096 if ( !$found && $args->getLength() == 0 ) {
3097 $id = $this->mVariables->matchStartToEnd( $part1 );
3098 if ( $id !== false ) {
3099 $text = $this->getVariableValue( $id, $frame );
3100 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3101 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3102 }
3103 $found = true;
3104 }
3105 }
3106
3107 # MSG, MSGNW and RAW
3108 if ( !$found ) {
3109 # Check for MSGNW:
3110 $mwMsgnw = MagicWord::get( 'msgnw' );
3111 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3112 $nowiki = true;
3113 } else {
3114 # Remove obsolete MSG:
3115 $mwMsg = MagicWord::get( 'msg' );
3116 $mwMsg->matchStartAndRemove( $part1 );
3117 }
3118
3119 # Check for RAW:
3120 $mwRaw = MagicWord::get( 'raw' );
3121 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3122 $forceRawInterwiki = true;
3123 }
3124 }
3125
3126 # Parser functions
3127 if ( !$found ) {
3128 $colonPos = strpos( $part1, ':' );
3129 if ( $colonPos !== false ) {
3130 $func = substr( $part1, 0, $colonPos );
3131 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3132 $argsLength = $args->getLength();
3133 for ( $i = 0; $i < $argsLength; $i++ ) {
3134 $funcArgs[] = $args->item( $i );
3135 }
3136 try {
3137 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3138 } catch ( Exception $ex ) {
3139 throw $ex;
3140 }
3141
3142 # The interface for parser functions allows for extracting
3143 # flags into the local scope. Extract any forwarded flags
3144 # here.
3145 extract( $result );
3146 }
3147 }
3148
3149 # Finish mangling title and then check for loops.
3150 # Set $title to a Title object and $titleText to the PDBK
3151 if ( !$found ) {
3152 $ns = NS_TEMPLATE;
3153 # Split the title into page and subpage
3154 $subpage = '';
3155 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3156 if ( $part1 !== $relative ) {
3157 $part1 = $relative;
3158 $ns = $this->mTitle->getNamespace();
3159 }
3160 $title = Title::newFromText( $part1, $ns );
3161 if ( $title ) {
3162 $titleText = $title->getPrefixedText();
3163 # Check for language variants if the template is not found
3164 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3165 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3166 }
3167 # Do recursion depth check
3168 $limit = $this->mOptions->getMaxTemplateDepth();
3169 if ( $frame->depth >= $limit ) {
3170 $found = true;
3171 $text = '<span class="error">'
3172 . wfMessage( 'parser-template-recursion-depth-warning' )
3173 ->numParams( $limit )->inContentLanguage()->text()
3174 . '</span>';
3175 }
3176 }
3177 }
3178
3179 # Load from database
3180 if ( !$found && $title ) {
3181 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3182 if ( !$title->isExternal() ) {
3183 if ( $title->isSpecialPage()
3184 && $this->mOptions->getAllowSpecialInclusion()
3185 && $this->ot['html']
3186 ) {
3187 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3188 // Pass the template arguments as URL parameters.
3189 // "uselang" will have no effect since the Language object
3190 // is forced to the one defined in ParserOptions.
3191 $pageArgs = [];
3192 $argsLength = $args->getLength();
3193 for ( $i = 0; $i < $argsLength; $i++ ) {
3194 $bits = $args->item( $i )->splitArg();
3195 if ( strval( $bits['index'] ) === '' ) {
3196 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3197 $value = trim( $frame->expand( $bits['value'] ) );
3198 $pageArgs[$name] = $value;
3199 }
3200 }
3201
3202 // Create a new context to execute the special page
3203 $context = new RequestContext;
3204 $context->setTitle( $title );
3205 $context->setRequest( new FauxRequest( $pageArgs ) );
3206 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3207 $context->setUser( $this->getUser() );
3208 } else {
3209 // If this page is cached, then we better not be per user.
3210 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3211 }
3212 $context->setLanguage( $this->mOptions->getUserLangObj() );
3213 $ret = SpecialPageFactory::capturePath(
3214 $title, $context, $this->getLinkRenderer() );
3215 if ( $ret ) {
3216 $text = $context->getOutput()->getHTML();
3217 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3218 $found = true;
3219 $isHTML = true;
3220 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3221 $this->mOutput->updateRuntimeAdaptiveExpiry(
3222 $specialPage->maxIncludeCacheTime()
3223 );
3224 }
3225 }
3226 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3227 $found = false; # access denied
3228 wfDebug( __METHOD__ . ": template inclusion denied for " .
3229 $title->getPrefixedDBkey() . "\n" );
3230 } else {
3231 list( $text, $title ) = $this->getTemplateDom( $title );
3232 if ( $text !== false ) {
3233 $found = true;
3234 $isChildObj = true;
3235 }
3236 }
3237
3238 # If the title is valid but undisplayable, make a link to it
3239 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3240 $text = "[[:$titleText]]";
3241 $found = true;
3242 }
3243 } elseif ( $title->isTrans() ) {
3244 # Interwiki transclusion
3245 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3246 $text = $this->interwikiTransclude( $title, 'render' );
3247 $isHTML = true;
3248 } else {
3249 $text = $this->interwikiTransclude( $title, 'raw' );
3250 # Preprocess it like a template
3251 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3252 $isChildObj = true;
3253 }
3254 $found = true;
3255 }
3256
3257 # Do infinite loop check
3258 # This has to be done after redirect resolution to avoid infinite loops via redirects
3259 if ( !$frame->loopCheck( $title ) ) {
3260 $found = true;
3261 $text = '<span class="error">'
3262 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3263 . '</span>';
3264 $this->addTrackingCategory( 'template-loop-category' );
3265 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3266 }
3267 }
3268
3269 # If we haven't found text to substitute by now, we're done
3270 # Recover the source wikitext and return it
3271 if ( !$found ) {
3272 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3273 if ( $profileSection ) {
3274 $this->mProfiler->scopedProfileOut( $profileSection );
3275 }
3276 return [ 'object' => $text ];
3277 }
3278
3279 # Expand DOM-style return values in a child frame
3280 if ( $isChildObj ) {
3281 # Clean up argument array
3282 $newFrame = $frame->newChild( $args, $title );
3283
3284 if ( $nowiki ) {
3285 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3286 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3287 # Expansion is eligible for the empty-frame cache
3288 $text = $newFrame->cachedExpand( $titleText, $text );
3289 } else {
3290 # Uncached expansion
3291 $text = $newFrame->expand( $text );
3292 }
3293 }
3294 if ( $isLocalObj && $nowiki ) {
3295 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3296 $isLocalObj = false;
3297 }
3298
3299 if ( $profileSection ) {
3300 $this->mProfiler->scopedProfileOut( $profileSection );
3301 }
3302
3303 # Replace raw HTML by a placeholder
3304 if ( $isHTML ) {
3305 $text = $this->insertStripItem( $text );
3306 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3307 # Escape nowiki-style return values
3308 $text = wfEscapeWikiText( $text );
3309 } elseif ( is_string( $text )
3310 && !$piece['lineStart']
3311 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3312 ) {
3313 # T2529: if the template begins with a table or block-level
3314 # element, it should be treated as beginning a new line.
3315 # This behavior is somewhat controversial.
3316 $text = "\n" . $text;
3317 }
3318
3319 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3320 # Error, oversize inclusion
3321 if ( $titleText !== false ) {
3322 # Make a working, properly escaped link if possible (T25588)
3323 $text = "[[:$titleText]]";
3324 } else {
3325 # This will probably not be a working link, but at least it may
3326 # provide some hint of where the problem is
3327 preg_replace( '/^:/', '', $originalTitle );
3328 $text = "[[:$originalTitle]]";
3329 }
3330 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3331 . 'post-expand include size too large -->' );
3332 $this->limitationWarn( 'post-expand-template-inclusion' );
3333 }
3334
3335 if ( $isLocalObj ) {
3336 $ret = [ 'object' => $text ];
3337 } else {
3338 $ret = [ 'text' => $text ];
3339 }
3340
3341 return $ret;
3342 }
3343
3344 /**
3345 * Call a parser function and return an array with text and flags.
3346 *
3347 * The returned array will always contain a boolean 'found', indicating
3348 * whether the parser function was found or not. It may also contain the
3349 * following:
3350 * text: string|object, resulting wikitext or PP DOM object
3351 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3352 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3353 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3354 * nowiki: bool, wiki markup in $text should be escaped
3355 *
3356 * @since 1.21
3357 * @param PPFrame $frame The current frame, contains template arguments
3358 * @param string $function Function name
3359 * @param array $args Arguments to the function
3360 * @throws MWException
3361 * @return array
3362 */
3363 public function callParserFunction( $frame, $function, array $args = [] ) {
3364 global $wgContLang;
3365
3366 # Case sensitive functions
3367 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3368 $function = $this->mFunctionSynonyms[1][$function];
3369 } else {
3370 # Case insensitive functions
3371 $function = $wgContLang->lc( $function );
3372 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3373 $function = $this->mFunctionSynonyms[0][$function];
3374 } else {
3375 return [ 'found' => false ];
3376 }
3377 }
3378
3379 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3380
3381 # Workaround for PHP bug 35229 and similar
3382 if ( !is_callable( $callback ) ) {
3383 throw new MWException( "Tag hook for $function is not callable\n" );
3384 }
3385
3386 // Avoid PHP 7.1 warning from passing $this by reference
3387 $parser = $this;
3388
3389 $allArgs = [ &$parser ];
3390 if ( $flags & self::SFH_OBJECT_ARGS ) {
3391 # Convert arguments to PPNodes and collect for appending to $allArgs
3392 $funcArgs = [];
3393 foreach ( $args as $k => $v ) {
3394 if ( $v instanceof PPNode || $k === 0 ) {
3395 $funcArgs[] = $v;
3396 } else {
3397 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3398 }
3399 }
3400
3401 # Add a frame parameter, and pass the arguments as an array
3402 $allArgs[] = $frame;
3403 $allArgs[] = $funcArgs;
3404 } else {
3405 # Convert arguments to plain text and append to $allArgs
3406 foreach ( $args as $k => $v ) {
3407 if ( $v instanceof PPNode ) {
3408 $allArgs[] = trim( $frame->expand( $v ) );
3409 } elseif ( is_int( $k ) && $k >= 0 ) {
3410 $allArgs[] = trim( $v );
3411 } else {
3412 $allArgs[] = trim( "$k=$v" );
3413 }
3414 }
3415 }
3416
3417 $result = call_user_func_array( $callback, $allArgs );
3418
3419 # The interface for function hooks allows them to return a wikitext
3420 # string or an array containing the string and any flags. This mungs
3421 # things around to match what this method should return.
3422 if ( !is_array( $result ) ) {
3423 $result =[
3424 'found' => true,
3425 'text' => $result,
3426 ];
3427 } else {
3428 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3429 $result['text'] = $result[0];
3430 }
3431 unset( $result[0] );
3432 $result += [
3433 'found' => true,
3434 ];
3435 }
3436
3437 $noparse = true;
3438 $preprocessFlags = 0;
3439 if ( isset( $result['noparse'] ) ) {
3440 $noparse = $result['noparse'];
3441 }
3442 if ( isset( $result['preprocessFlags'] ) ) {
3443 $preprocessFlags = $result['preprocessFlags'];
3444 }
3445
3446 if ( !$noparse ) {
3447 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3448 $result['isChildObj'] = true;
3449 }
3450
3451 return $result;
3452 }
3453
3454 /**
3455 * Get the semi-parsed DOM representation of a template with a given title,
3456 * and its redirect destination title. Cached.
3457 *
3458 * @param Title $title
3459 *
3460 * @return array
3461 */
3462 public function getTemplateDom( $title ) {
3463 $cacheTitle = $title;
3464 $titleText = $title->getPrefixedDBkey();
3465
3466 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3467 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3468 $title = Title::makeTitle( $ns, $dbk );
3469 $titleText = $title->getPrefixedDBkey();
3470 }
3471 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3472 return [ $this->mTplDomCache[$titleText], $title ];
3473 }
3474
3475 # Cache miss, go to the database
3476 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3477
3478 if ( $text === false ) {
3479 $this->mTplDomCache[$titleText] = false;
3480 return [ false, $title ];
3481 }
3482
3483 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3484 $this->mTplDomCache[$titleText] = $dom;
3485
3486 if ( !$title->equals( $cacheTitle ) ) {
3487 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3488 [ $title->getNamespace(), $title->getDBkey() ];
3489 }
3490
3491 return [ $dom, $title ];
3492 }
3493
3494 /**
3495 * Fetch the current revision of a given title. Note that the revision
3496 * (and even the title) may not exist in the database, so everything
3497 * contributing to the output of the parser should use this method
3498 * where possible, rather than getting the revisions themselves. This
3499 * method also caches its results, so using it benefits performance.
3500 *
3501 * @since 1.24
3502 * @param Title $title
3503 * @return Revision
3504 */
3505 public function fetchCurrentRevisionOfTitle( $title ) {
3506 $cacheKey = $title->getPrefixedDBkey();
3507 if ( !$this->currentRevisionCache ) {
3508 $this->currentRevisionCache = new MapCacheLRU( 100 );
3509 }
3510 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3511 $this->currentRevisionCache->set( $cacheKey,
3512 // Defaults to Parser::statelessFetchRevision()
3513 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3514 );
3515 }
3516 return $this->currentRevisionCache->get( $cacheKey );
3517 }
3518
3519 /**
3520 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3521 * without passing them on to it.
3522 *
3523 * @since 1.24
3524 * @param Title $title
3525 * @param Parser|bool $parser
3526 * @return Revision|bool False if missing
3527 */
3528 public static function statelessFetchRevision( Title $title, $parser = false ) {
3529 $pageId = $title->getArticleID();
3530 $revId = $title->getLatestRevID();
3531
3532 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3533 if ( $rev ) {
3534 $rev->setTitle( $title );
3535 }
3536
3537 return $rev;
3538 }
3539
3540 /**
3541 * Fetch the unparsed text of a template and register a reference to it.
3542 * @param Title $title
3543 * @return array ( string or false, Title )
3544 */
3545 public function fetchTemplateAndTitle( $title ) {
3546 // Defaults to Parser::statelessFetchTemplate()
3547 $templateCb = $this->mOptions->getTemplateCallback();
3548 $stuff = call_user_func( $templateCb, $title, $this );
3549 // We use U+007F DELETE to distinguish strip markers from regular text.
3550 $text = $stuff['text'];
3551 if ( is_string( $stuff['text'] ) ) {
3552 $text = strtr( $text, "\x7f", "?" );
3553 }
3554 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3555 if ( isset( $stuff['deps'] ) ) {
3556 foreach ( $stuff['deps'] as $dep ) {
3557 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3558 if ( $dep['title']->equals( $this->getTitle() ) ) {
3559 // If we transclude ourselves, the final result
3560 // will change based on the new version of the page
3561 $this->mOutput->setFlag( 'vary-revision' );
3562 }
3563 }
3564 }
3565 return [ $text, $finalTitle ];
3566 }
3567
3568 /**
3569 * Fetch the unparsed text of a template and register a reference to it.
3570 * @param Title $title
3571 * @return string|bool
3572 */
3573 public function fetchTemplate( $title ) {
3574 return $this->fetchTemplateAndTitle( $title )[0];
3575 }
3576
3577 /**
3578 * Static function to get a template
3579 * Can be overridden via ParserOptions::setTemplateCallback().
3580 *
3581 * @param Title $title
3582 * @param bool|Parser $parser
3583 *
3584 * @return array
3585 */
3586 public static function statelessFetchTemplate( $title, $parser = false ) {
3587 $text = $skip = false;
3588 $finalTitle = $title;
3589 $deps = [];
3590
3591 # Loop to fetch the article, with up to 1 redirect
3592 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3593 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3594 // @codingStandardsIgnoreEnd
3595 # Give extensions a chance to select the revision instead
3596 $id = false; # Assume current
3597 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3598 [ $parser, $title, &$skip, &$id ] );
3599
3600 if ( $skip ) {
3601 $text = false;
3602 $deps[] = [
3603 'title' => $title,
3604 'page_id' => $title->getArticleID(),
3605 'rev_id' => null
3606 ];
3607 break;
3608 }
3609 # Get the revision
3610 if ( $id ) {
3611 $rev = Revision::newFromId( $id );
3612 } elseif ( $parser ) {
3613 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3614 } else {
3615 $rev = Revision::newFromTitle( $title );
3616 }
3617 $rev_id = $rev ? $rev->getId() : 0;
3618 # If there is no current revision, there is no page
3619 if ( $id === false && !$rev ) {
3620 $linkCache = LinkCache::singleton();
3621 $linkCache->addBadLinkObj( $title );
3622 }
3623
3624 $deps[] = [
3625 'title' => $title,
3626 'page_id' => $title->getArticleID(),
3627 'rev_id' => $rev_id ];
3628 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3629 # We fetched a rev from a different title; register it too...
3630 $deps[] = [
3631 'title' => $rev->getTitle(),
3632 'page_id' => $rev->getPage(),
3633 'rev_id' => $rev_id ];
3634 }
3635
3636 if ( $rev ) {
3637 $content = $rev->getContent();
3638 $text = $content ? $content->getWikitextForTransclusion() : null;
3639
3640 Hooks::run( 'ParserFetchTemplate',
3641 [ $parser, $title, $rev, &$text, &$deps ] );
3642
3643 if ( $text === false || $text === null ) {
3644 $text = false;
3645 break;
3646 }
3647 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3648 global $wgContLang;
3649 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3650 if ( !$message->exists() ) {
3651 $text = false;
3652 break;
3653 }
3654 $content = $message->content();
3655 $text = $message->plain();
3656 } else {
3657 break;
3658 }
3659 if ( !$content ) {
3660 break;
3661 }
3662 # Redirect?
3663 $finalTitle = $title;
3664 $title = $content->getRedirectTarget();
3665 }
3666 return [
3667 'text' => $text,
3668 'finalTitle' => $finalTitle,
3669 'deps' => $deps ];
3670 }
3671
3672 /**
3673 * Fetch a file and its title and register a reference to it.
3674 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3675 * @param Title $title
3676 * @param array $options Array of options to RepoGroup::findFile
3677 * @return File|bool
3678 */
3679 public function fetchFile( $title, $options = [] ) {
3680 return $this->fetchFileAndTitle( $title, $options )[0];
3681 }
3682
3683 /**
3684 * Fetch a file and its title and register a reference to it.
3685 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3686 * @param Title $title
3687 * @param array $options Array of options to RepoGroup::findFile
3688 * @return array ( File or false, Title of file )
3689 */
3690 public function fetchFileAndTitle( $title, $options = [] ) {
3691 $file = $this->fetchFileNoRegister( $title, $options );
3692
3693 $time = $file ? $file->getTimestamp() : false;
3694 $sha1 = $file ? $file->getSha1() : false;
3695 # Register the file as a dependency...
3696 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3697 if ( $file && !$title->equals( $file->getTitle() ) ) {
3698 # Update fetched file title
3699 $title = $file->getTitle();
3700 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3701 }
3702 return [ $file, $title ];
3703 }
3704
3705 /**
3706 * Helper function for fetchFileAndTitle.
3707 *
3708 * Also useful if you need to fetch a file but not use it yet,
3709 * for example to get the file's handler.
3710 *
3711 * @param Title $title
3712 * @param array $options Array of options to RepoGroup::findFile
3713 * @return File|bool
3714 */
3715 protected function fetchFileNoRegister( $title, $options = [] ) {
3716 if ( isset( $options['broken'] ) ) {
3717 $file = false; // broken thumbnail forced by hook
3718 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3719 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3720 } else { // get by (name,timestamp)
3721 $file = wfFindFile( $title, $options );
3722 }
3723 return $file;
3724 }
3725
3726 /**
3727 * Transclude an interwiki link.
3728 *
3729 * @param Title $title
3730 * @param string $action
3731 *
3732 * @return string
3733 */
3734 public function interwikiTransclude( $title, $action ) {
3735 global $wgEnableScaryTranscluding;
3736
3737 if ( !$wgEnableScaryTranscluding ) {
3738 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3739 }
3740
3741 $url = $title->getFullURL( [ 'action' => $action ] );
3742
3743 if ( strlen( $url ) > 255 ) {
3744 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3745 }
3746 return $this->fetchScaryTemplateMaybeFromCache( $url );
3747 }
3748
3749 /**
3750 * @param string $url
3751 * @return mixed|string
3752 */
3753 public function fetchScaryTemplateMaybeFromCache( $url ) {
3754 global $wgTranscludeCacheExpiry;
3755 $dbr = wfGetDB( DB_REPLICA );
3756 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3757 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3758 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3759 if ( $obj ) {
3760 return $obj->tc_contents;
3761 }
3762
3763 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3764 $status = $req->execute(); // Status object
3765 if ( $status->isOK() ) {
3766 $text = $req->getContent();
3767 } elseif ( $req->getStatus() != 200 ) {
3768 // Though we failed to fetch the content, this status is useless.
3769 return wfMessage( 'scarytranscludefailed-httpstatus' )
3770 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3771 } else {
3772 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3773 }
3774
3775 $dbw = wfGetDB( DB_MASTER );
3776 $dbw->replace( 'transcache', [ 'tc_url' ], [
3777 'tc_url' => $url,
3778 'tc_time' => $dbw->timestamp( time() ),
3779 'tc_contents' => $text
3780 ] );
3781 return $text;
3782 }
3783
3784 /**
3785 * Triple brace replacement -- used for template arguments
3786 * @private
3787 *
3788 * @param array $piece
3789 * @param PPFrame $frame
3790 *
3791 * @return array
3792 */
3793 public function argSubstitution( $piece, $frame ) {
3794
3795 $error = false;
3796 $parts = $piece['parts'];
3797 $nameWithSpaces = $frame->expand( $piece['title'] );
3798 $argName = trim( $nameWithSpaces );
3799 $object = false;
3800 $text = $frame->getArgument( $argName );
3801 if ( $text === false && $parts->getLength() > 0
3802 && ( $this->ot['html']
3803 || $this->ot['pre']
3804 || ( $this->ot['wiki'] && $frame->isTemplate() )
3805 )
3806 ) {
3807 # No match in frame, use the supplied default
3808 $object = $parts->item( 0 )->getChildren();
3809 }
3810 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3811 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3812 $this->limitationWarn( 'post-expand-template-argument' );
3813 }
3814
3815 if ( $text === false && $object === false ) {
3816 # No match anywhere
3817 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3818 }
3819 if ( $error !== false ) {
3820 $text .= $error;
3821 }
3822 if ( $object !== false ) {
3823 $ret = [ 'object' => $object ];
3824 } else {
3825 $ret = [ 'text' => $text ];
3826 }
3827
3828 return $ret;
3829 }
3830
3831 /**
3832 * Return the text to be used for a given extension tag.
3833 * This is the ghost of strip().
3834 *
3835 * @param array $params Associative array of parameters:
3836 * name PPNode for the tag name
3837 * attr PPNode for unparsed text where tag attributes are thought to be
3838 * attributes Optional associative array of parsed attributes
3839 * inner Contents of extension element
3840 * noClose Original text did not have a close tag
3841 * @param PPFrame $frame
3842 *
3843 * @throws MWException
3844 * @return string
3845 */
3846 public function extensionSubstitution( $params, $frame ) {
3847 static $errorStr = '<span class="error">';
3848 static $errorLen = 20;
3849
3850 $name = $frame->expand( $params['name'] );
3851 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3852 // Probably expansion depth or node count exceeded. Just punt the
3853 // error up.
3854 return $name;
3855 }
3856
3857 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3858 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3859 // See above
3860 return $attrText;
3861 }
3862
3863 // We can't safely check if the expansion for $content resulted in an
3864 // error, because the content could happen to be the error string
3865 // (T149622).
3866 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3867
3868 $marker = self::MARKER_PREFIX . "-$name-"
3869 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3870
3871 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3872 ( $this->ot['html'] || $this->ot['pre'] );
3873 if ( $isFunctionTag ) {
3874 $markerType = 'none';
3875 } else {
3876 $markerType = 'general';
3877 }
3878 if ( $this->ot['html'] || $isFunctionTag ) {
3879 $name = strtolower( $name );
3880 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3881 if ( isset( $params['attributes'] ) ) {
3882 $attributes = $attributes + $params['attributes'];
3883 }
3884
3885 if ( isset( $this->mTagHooks[$name] ) ) {
3886 # Workaround for PHP bug 35229 and similar
3887 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3888 throw new MWException( "Tag hook for $name is not callable\n" );
3889 }
3890 $output = call_user_func_array( $this->mTagHooks[$name],
3891 [ $content, $attributes, $this, $frame ] );
3892 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3893 list( $callback, ) = $this->mFunctionTagHooks[$name];
3894 if ( !is_callable( $callback ) ) {
3895 throw new MWException( "Tag hook for $name is not callable\n" );
3896 }
3897
3898 // Avoid PHP 7.1 warning from passing $this by reference
3899 $parser = $this;
3900 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3901 } else {
3902 $output = '<span class="error">Invalid tag extension name: ' .
3903 htmlspecialchars( $name ) . '</span>';
3904 }
3905
3906 if ( is_array( $output ) ) {
3907 # Extract flags to local scope (to override $markerType)
3908 $flags = $output;
3909 $output = $flags[0];
3910 unset( $flags[0] );
3911 extract( $flags );
3912 }
3913 } else {
3914 if ( is_null( $attrText ) ) {
3915 $attrText = '';
3916 }
3917 if ( isset( $params['attributes'] ) ) {
3918 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3919 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3920 htmlspecialchars( $attrValue ) . '"';
3921 }
3922 }
3923 if ( $content === null ) {
3924 $output = "<$name$attrText/>";
3925 } else {
3926 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3927 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3928 // See above
3929 return $close;
3930 }
3931 $output = "<$name$attrText>$content$close";
3932 }
3933 }
3934
3935 if ( $markerType === 'none' ) {
3936 return $output;
3937 } elseif ( $markerType === 'nowiki' ) {
3938 $this->mStripState->addNoWiki( $marker, $output );
3939 } elseif ( $markerType === 'general' ) {
3940 $this->mStripState->addGeneral( $marker, $output );
3941 } else {
3942 throw new MWException( __METHOD__ . ': invalid marker type' );
3943 }
3944 return $marker;
3945 }
3946
3947 /**
3948 * Increment an include size counter
3949 *
3950 * @param string $type The type of expansion
3951 * @param int $size The size of the text
3952 * @return bool False if this inclusion would take it over the maximum, true otherwise
3953 */
3954 public function incrementIncludeSize( $type, $size ) {
3955 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3956 return false;
3957 } else {
3958 $this->mIncludeSizes[$type] += $size;
3959 return true;
3960 }
3961 }
3962
3963 /**
3964 * Increment the expensive function count
3965 *
3966 * @return bool False if the limit has been exceeded
3967 */
3968 public function incrementExpensiveFunctionCount() {
3969 $this->mExpensiveFunctionCount++;
3970 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3971 }
3972
3973 /**
3974 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3975 * Fills $this->mDoubleUnderscores, returns the modified text
3976 *
3977 * @param string $text
3978 *
3979 * @return string
3980 */
3981 public function doDoubleUnderscore( $text ) {
3982
3983 # The position of __TOC__ needs to be recorded
3984 $mw = MagicWord::get( 'toc' );
3985 if ( $mw->match( $text ) ) {
3986 $this->mShowToc = true;
3987 $this->mForceTocPosition = true;
3988
3989 # Set a placeholder. At the end we'll fill it in with the TOC.
3990 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3991
3992 # Only keep the first one.
3993 $text = $mw->replace( '', $text );
3994 }
3995
3996 # Now match and remove the rest of them
3997 $mwa = MagicWord::getDoubleUnderscoreArray();
3998 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3999
4000 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4001 $this->mOutput->mNoGallery = true;
4002 }
4003 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4004 $this->mShowToc = false;
4005 }
4006 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4007 && $this->mTitle->getNamespace() == NS_CATEGORY
4008 ) {
4009 $this->addTrackingCategory( 'hidden-category-category' );
4010 }
4011 # (T10068) Allow control over whether robots index a page.
4012 # __INDEX__ always overrides __NOINDEX__, see T16899
4013 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4014 $this->mOutput->setIndexPolicy( 'noindex' );
4015 $this->addTrackingCategory( 'noindex-category' );
4016 }
4017 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4018 $this->mOutput->setIndexPolicy( 'index' );
4019 $this->addTrackingCategory( 'index-category' );
4020 }
4021
4022 # Cache all double underscores in the database
4023 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4024 $this->mOutput->setProperty( $key, '' );
4025 }
4026
4027 return $text;
4028 }
4029
4030 /**
4031 * @see ParserOutput::addTrackingCategory()
4032 * @param string $msg Message key
4033 * @return bool Whether the addition was successful
4034 */
4035 public function addTrackingCategory( $msg ) {
4036 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4037 }
4038
4039 /**
4040 * This function accomplishes several tasks:
4041 * 1) Auto-number headings if that option is enabled
4042 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4043 * 3) Add a Table of contents on the top for users who have enabled the option
4044 * 4) Auto-anchor headings
4045 *
4046 * It loops through all headlines, collects the necessary data, then splits up the
4047 * string and re-inserts the newly formatted headlines.
4048 *
4049 * @param string $text
4050 * @param string $origText Original, untouched wikitext
4051 * @param bool $isMain
4052 * @return mixed|string
4053 * @private
4054 */
4055 public function formatHeadings( $text, $origText, $isMain = true ) {
4056 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4057
4058 # Inhibit editsection links if requested in the page
4059 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4060 $maybeShowEditLink = $showEditLink = false;
4061 } else {
4062 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4063 $showEditLink = $this->mOptions->getEditSection();
4064 }
4065 if ( $showEditLink ) {
4066 $this->mOutput->setEditSectionTokens( true );
4067 }
4068
4069 # Get all headlines for numbering them and adding funky stuff like [edit]
4070 # links - this is for later, but we need the number of headlines right now
4071 $matches = [];
4072 $numMatches = preg_match_all(
4073 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4074 $text,
4075 $matches
4076 );
4077
4078 # if there are fewer than 4 headlines in the article, do not show TOC
4079 # unless it's been explicitly enabled.
4080 $enoughToc = $this->mShowToc &&
4081 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4082
4083 # Allow user to stipulate that a page should have a "new section"
4084 # link added via __NEWSECTIONLINK__
4085 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4086 $this->mOutput->setNewSection( true );
4087 }
4088
4089 # Allow user to remove the "new section"
4090 # link via __NONEWSECTIONLINK__
4091 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4092 $this->mOutput->hideNewSection( true );
4093 }
4094
4095 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4096 # override above conditions and always show TOC above first header
4097 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4098 $this->mShowToc = true;
4099 $enoughToc = true;
4100 }
4101
4102 # headline counter
4103 $headlineCount = 0;
4104 $numVisible = 0;
4105
4106 # Ugh .. the TOC should have neat indentation levels which can be
4107 # passed to the skin functions. These are determined here
4108 $toc = '';
4109 $full = '';
4110 $head = [];
4111 $sublevelCount = [];
4112 $levelCount = [];
4113 $level = 0;
4114 $prevlevel = 0;
4115 $toclevel = 0;
4116 $prevtoclevel = 0;
4117 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4118 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4119 $oldType = $this->mOutputType;
4120 $this->setOutputType( self::OT_WIKI );
4121 $frame = $this->getPreprocessor()->newFrame();
4122 $root = $this->preprocessToDom( $origText );
4123 $node = $root->getFirstChild();
4124 $byteOffset = 0;
4125 $tocraw = [];
4126 $refers = [];
4127
4128 $headlines = $numMatches !== false ? $matches[3] : [];
4129
4130 foreach ( $headlines as $headline ) {
4131 $isTemplate = false;
4132 $titleText = false;
4133 $sectionIndex = false;
4134 $numbering = '';
4135 $markerMatches = [];
4136 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4137 $serial = $markerMatches[1];
4138 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4139 $isTemplate = ( $titleText != $baseTitleText );
4140 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4141 }
4142
4143 if ( $toclevel ) {
4144 $prevlevel = $level;
4145 }
4146 $level = $matches[1][$headlineCount];
4147
4148 if ( $level > $prevlevel ) {
4149 # Increase TOC level
4150 $toclevel++;
4151 $sublevelCount[$toclevel] = 0;
4152 if ( $toclevel < $wgMaxTocLevel ) {
4153 $prevtoclevel = $toclevel;
4154 $toc .= Linker::tocIndent();
4155 $numVisible++;
4156 }
4157 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4158 # Decrease TOC level, find level to jump to
4159
4160 for ( $i = $toclevel; $i > 0; $i-- ) {
4161 if ( $levelCount[$i] == $level ) {
4162 # Found last matching level
4163 $toclevel = $i;
4164 break;
4165 } elseif ( $levelCount[$i] < $level ) {
4166 # Found first matching level below current level
4167 $toclevel = $i + 1;
4168 break;
4169 }
4170 }
4171 if ( $i == 0 ) {
4172 $toclevel = 1;
4173 }
4174 if ( $toclevel < $wgMaxTocLevel ) {
4175 if ( $prevtoclevel < $wgMaxTocLevel ) {
4176 # Unindent only if the previous toc level was shown :p
4177 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4178 $prevtoclevel = $toclevel;
4179 } else {
4180 $toc .= Linker::tocLineEnd();
4181 }
4182 }
4183 } else {
4184 # No change in level, end TOC line
4185 if ( $toclevel < $wgMaxTocLevel ) {
4186 $toc .= Linker::tocLineEnd();
4187 }
4188 }
4189
4190 $levelCount[$toclevel] = $level;
4191
4192 # count number of headlines for each level
4193 $sublevelCount[$toclevel]++;
4194 $dot = 0;
4195 for ( $i = 1; $i <= $toclevel; $i++ ) {
4196 if ( !empty( $sublevelCount[$i] ) ) {
4197 if ( $dot ) {
4198 $numbering .= '.';
4199 }
4200 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4201 $dot = 1;
4202 }
4203 }
4204
4205 # The safe header is a version of the header text safe to use for links
4206
4207 # Remove link placeholders by the link text.
4208 # <!--LINK number-->
4209 # turns into
4210 # link text with suffix
4211 # Do this before unstrip since link text can contain strip markers
4212 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4213
4214 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4215 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4216
4217 # Strip out HTML (first regex removes any tag not allowed)
4218 # Allowed tags are:
4219 # * <sup> and <sub> (T10393)
4220 # * <i> (T28375)
4221 # * <b> (r105284)
4222 # * <bdi> (T74884)
4223 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4224 # * <s> and <strike> (T35715)
4225 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4226 # to allow setting directionality in toc items.
4227 $tocline = preg_replace(
4228 [
4229 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4230 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4231 ],
4232 [ '', '<$1>' ],
4233 $safeHeadline
4234 );
4235
4236 # Strip '<span></span>', which is the result from the above if
4237 # <span id="foo"></span> is used to produce an additional anchor
4238 # for a section.
4239 $tocline = str_replace( '<span></span>', '', $tocline );
4240
4241 $tocline = trim( $tocline );
4242
4243 # For the anchor, strip out HTML-y stuff period
4244 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4245 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4246
4247 # Save headline for section edit hint before it's escaped
4248 $headlineHint = $safeHeadline;
4249
4250 if ( $wgExperimentalHtmlIds ) {
4251 # For reverse compatibility, provide an id that's
4252 # HTML4-compatible, like we used to.
4253 # It may be worth noting, academically, that it's possible for
4254 # the legacy anchor to conflict with a non-legacy headline
4255 # anchor on the page. In this case likely the "correct" thing
4256 # would be to either drop the legacy anchors or make sure
4257 # they're numbered first. However, this would require people
4258 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4259 # manually, so let's not bother worrying about it.
4260 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4261 [ 'noninitial', 'legacy' ] );
4262 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4263
4264 if ( $legacyHeadline == $safeHeadline ) {
4265 # No reason to have both (in fact, we can't)
4266 $legacyHeadline = false;
4267 }
4268 } else {
4269 $legacyHeadline = false;
4270 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4271 'noninitial' );
4272 }
4273
4274 # HTML names must be case-insensitively unique (T12721).
4275 # This does not apply to Unicode characters per
4276 # https://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4277 # @todo FIXME: We may be changing them depending on the current locale.
4278 $arrayKey = strtolower( $safeHeadline );
4279 if ( $legacyHeadline === false ) {
4280 $legacyArrayKey = false;
4281 } else {
4282 $legacyArrayKey = strtolower( $legacyHeadline );
4283 }
4284
4285 # Create the anchor for linking from the TOC to the section
4286 $anchor = $safeHeadline;
4287 $legacyAnchor = $legacyHeadline;
4288 if ( isset( $refers[$arrayKey] ) ) {
4289 // @codingStandardsIgnoreStart
4290 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4291 // @codingStandardsIgnoreEnd
4292 $anchor .= "_$i";
4293 $refers["${arrayKey}_$i"] = true;
4294 } else {
4295 $refers[$arrayKey] = true;
4296 }
4297 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4298 // @codingStandardsIgnoreStart
4299 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4300 // @codingStandardsIgnoreEnd
4301 $legacyAnchor .= "_$i";
4302 $refers["${legacyArrayKey}_$i"] = true;
4303 } else {
4304 $refers[$legacyArrayKey] = true;
4305 }
4306
4307 # Don't number the heading if it is the only one (looks silly)
4308 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4309 # the two are different if the line contains a link
4310 $headline = Html::element(
4311 'span',
4312 [ 'class' => 'mw-headline-number' ],
4313 $numbering
4314 ) . ' ' . $headline;
4315 }
4316
4317 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4318 $toc .= Linker::tocLine( $anchor, $tocline,
4319 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4320 }
4321
4322 # Add the section to the section tree
4323 # Find the DOM node for this header
4324 $noOffset = ( $isTemplate || $sectionIndex === false );
4325 while ( $node && !$noOffset ) {
4326 if ( $node->getName() === 'h' ) {
4327 $bits = $node->splitHeading();
4328 if ( $bits['i'] == $sectionIndex ) {
4329 break;
4330 }
4331 }
4332 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4333 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4334 $node = $node->getNextSibling();
4335 }
4336 $tocraw[] = [
4337 'toclevel' => $toclevel,
4338 'level' => $level,
4339 'line' => $tocline,
4340 'number' => $numbering,
4341 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4342 'fromtitle' => $titleText,
4343 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4344 'anchor' => $anchor,
4345 ];
4346
4347 # give headline the correct <h#> tag
4348 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4349 // Output edit section links as markers with styles that can be customized by skins
4350 if ( $isTemplate ) {
4351 # Put a T flag in the section identifier, to indicate to extractSections()
4352 # that sections inside <includeonly> should be counted.
4353 $editsectionPage = $titleText;
4354 $editsectionSection = "T-$sectionIndex";
4355 $editsectionContent = null;
4356 } else {
4357 $editsectionPage = $this->mTitle->getPrefixedText();
4358 $editsectionSection = $sectionIndex;
4359 $editsectionContent = $headlineHint;
4360 }
4361 // We use a bit of pesudo-xml for editsection markers. The
4362 // language converter is run later on. Using a UNIQ style marker
4363 // leads to the converter screwing up the tokens when it
4364 // converts stuff. And trying to insert strip tags fails too. At
4365 // this point all real inputted tags have already been escaped,
4366 // so we don't have to worry about a user trying to input one of
4367 // these markers directly. We use a page and section attribute
4368 // to stop the language converter from converting these
4369 // important bits of data, but put the headline hint inside a
4370 // content block because the language converter is supposed to
4371 // be able to convert that piece of data.
4372 // Gets replaced with html in ParserOutput::getText
4373 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4374 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4375 if ( $editsectionContent !== null ) {
4376 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4377 } else {
4378 $editlink .= '/>';
4379 }
4380 } else {
4381 $editlink = '';
4382 }
4383 $head[$headlineCount] = Linker::makeHeadline( $level,
4384 $matches['attrib'][$headlineCount], $anchor, $headline,
4385 $editlink, $legacyAnchor );
4386
4387 $headlineCount++;
4388 }
4389
4390 $this->setOutputType( $oldType );
4391
4392 # Never ever show TOC if no headers
4393 if ( $numVisible < 1 ) {
4394 $enoughToc = false;
4395 }
4396
4397 if ( $enoughToc ) {
4398 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4399 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4400 }
4401 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4402 $this->mOutput->setTOCHTML( $toc );
4403 $toc = self::TOC_START . $toc . self::TOC_END;
4404 }
4405
4406 if ( $isMain ) {
4407 $this->mOutput->setSections( $tocraw );
4408 }
4409
4410 # split up and insert constructed headlines
4411 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4412 $i = 0;
4413
4414 // build an array of document sections
4415 $sections = [];
4416 foreach ( $blocks as $block ) {
4417 // $head is zero-based, sections aren't.
4418 if ( empty( $head[$i - 1] ) ) {
4419 $sections[$i] = $block;
4420 } else {
4421 $sections[$i] = $head[$i - 1] . $block;
4422 }
4423
4424 /**
4425 * Send a hook, one per section.
4426 * The idea here is to be able to make section-level DIVs, but to do so in a
4427 * lower-impact, more correct way than r50769
4428 *
4429 * $this : caller
4430 * $section : the section number
4431 * &$sectionContent : ref to the content of the section
4432 * $showEditLinks : boolean describing whether this section has an edit link
4433 */
4434 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4435
4436 $i++;
4437 }
4438
4439 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4440 // append the TOC at the beginning
4441 // Top anchor now in skin
4442 $sections[0] = $sections[0] . $toc . "\n";
4443 }
4444
4445 $full .= implode( '', $sections );
4446
4447 if ( $this->mForceTocPosition ) {
4448 return str_replace( '<!--MWTOC-->', $toc, $full );
4449 } else {
4450 return $full;
4451 }
4452 }
4453
4454 /**
4455 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4456 * conversion, substituting signatures, {{subst:}} templates, etc.
4457 *
4458 * @param string $text The text to transform
4459 * @param Title $title The Title object for the current article
4460 * @param User $user The User object describing the current user
4461 * @param ParserOptions $options Parsing options
4462 * @param bool $clearState Whether to clear the parser state first
4463 * @return string The altered wiki markup
4464 */
4465 public function preSaveTransform( $text, Title $title, User $user,
4466 ParserOptions $options, $clearState = true
4467 ) {
4468 if ( $clearState ) {
4469 $magicScopeVariable = $this->lock();
4470 }
4471 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4472 $this->setUser( $user );
4473
4474 // Strip U+0000 NULL (T159174)
4475 $text = str_replace( "\000", '', $text );
4476
4477 // We still normalize line endings for backwards-compatibility
4478 // with other code that just calls PST, but this should already
4479 // be handled in TextContent subclasses
4480 $text = TextContent::normalizeLineEndings( $text );
4481
4482 if ( $options->getPreSaveTransform() ) {
4483 $text = $this->pstPass2( $text, $user );
4484 }
4485 $text = $this->mStripState->unstripBoth( $text );
4486
4487 $this->setUser( null ); # Reset
4488
4489 return $text;
4490 }
4491
4492 /**
4493 * Pre-save transform helper function
4494 *
4495 * @param string $text
4496 * @param User $user
4497 *
4498 * @return string
4499 */
4500 private function pstPass2( $text, $user ) {
4501 global $wgContLang;
4502
4503 # Note: This is the timestamp saved as hardcoded wikitext to
4504 # the database, we use $wgContLang here in order to give
4505 # everyone the same signature and use the default one rather
4506 # than the one selected in each user's preferences.
4507 # (see also T14815)
4508 $ts = $this->mOptions->getTimestamp();
4509 $timestamp = MWTimestamp::getLocalInstance( $ts );
4510 $ts = $timestamp->format( 'YmdHis' );
4511 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4512
4513 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4514
4515 # Variable replacement
4516 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4517 $text = $this->replaceVariables( $text );
4518
4519 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4520 # which may corrupt this parser instance via its wfMessage()->text() call-
4521
4522 # Signatures
4523 $sigText = $this->getUserSig( $user );
4524 $text = strtr( $text, [
4525 '~~~~~' => $d,
4526 '~~~~' => "$sigText $d",
4527 '~~~' => $sigText
4528 ] );
4529
4530 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4531 $tc = '[' . Title::legalChars() . ']';
4532 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4533
4534 // [[ns:page (context)|]]
4535 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4536 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4537 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4538 // [[ns:page (context), context|]] (using either single or double-width comma)
4539 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4540 // [[|page]] (reverse pipe trick: add context from page title)
4541 $p2 = "/\[\[\\|($tc+)]]/";
4542
4543 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4544 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4545 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4546 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4547
4548 $t = $this->mTitle->getText();
4549 $m = [];
4550 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4551 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4552 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4553 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4554 } else {
4555 # if there's no context, don't bother duplicating the title
4556 $text = preg_replace( $p2, '[[\\1]]', $text );
4557 }
4558
4559 return $text;
4560 }
4561
4562 /**
4563 * Fetch the user's signature text, if any, and normalize to
4564 * validated, ready-to-insert wikitext.
4565 * If you have pre-fetched the nickname or the fancySig option, you can
4566 * specify them here to save a database query.
4567 * Do not reuse this parser instance after calling getUserSig(),
4568 * as it may have changed if it's the $wgParser.
4569 *
4570 * @param User $user
4571 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4572 * @param bool|null $fancySig whether the nicknname is the complete signature
4573 * or null to use default value
4574 * @return string
4575 */
4576 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4577 global $wgMaxSigChars;
4578
4579 $username = $user->getName();
4580
4581 # If not given, retrieve from the user object.
4582 if ( $nickname === false ) {
4583 $nickname = $user->getOption( 'nickname' );
4584 }
4585
4586 if ( is_null( $fancySig ) ) {
4587 $fancySig = $user->getBoolOption( 'fancysig' );
4588 }
4589
4590 $nickname = $nickname == null ? $username : $nickname;
4591
4592 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4593 $nickname = $username;
4594 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4595 } elseif ( $fancySig !== false ) {
4596 # Sig. might contain markup; validate this
4597 if ( $this->validateSig( $nickname ) !== false ) {
4598 # Validated; clean up (if needed) and return it
4599 return $this->cleanSig( $nickname, true );
4600 } else {
4601 # Failed to validate; fall back to the default
4602 $nickname = $username;
4603 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4604 }
4605 }
4606
4607 # Make sure nickname doesnt get a sig in a sig
4608 $nickname = self::cleanSigInSig( $nickname );
4609
4610 # If we're still here, make it a link to the user page
4611 $userText = wfEscapeWikiText( $username );
4612 $nickText = wfEscapeWikiText( $nickname );
4613 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4614
4615 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4616 ->title( $this->getTitle() )->text();
4617 }
4618
4619 /**
4620 * Check that the user's signature contains no bad XML
4621 *
4622 * @param string $text
4623 * @return string|bool An expanded string, or false if invalid.
4624 */
4625 public function validateSig( $text ) {
4626 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4627 }
4628
4629 /**
4630 * Clean up signature text
4631 *
4632 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4633 * 2) Substitute all transclusions
4634 *
4635 * @param string $text
4636 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4637 * @return string Signature text
4638 */
4639 public function cleanSig( $text, $parsing = false ) {
4640 if ( !$parsing ) {
4641 global $wgTitle;
4642 $magicScopeVariable = $this->lock();
4643 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4644 }
4645
4646 # Option to disable this feature
4647 if ( !$this->mOptions->getCleanSignatures() ) {
4648 return $text;
4649 }
4650
4651 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4652 # => Move this logic to braceSubstitution()
4653 $substWord = MagicWord::get( 'subst' );
4654 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4655 $substText = '{{' . $substWord->getSynonym( 0 );
4656
4657 $text = preg_replace( $substRegex, $substText, $text );
4658 $text = self::cleanSigInSig( $text );
4659 $dom = $this->preprocessToDom( $text );
4660 $frame = $this->getPreprocessor()->newFrame();
4661 $text = $frame->expand( $dom );
4662
4663 if ( !$parsing ) {
4664 $text = $this->mStripState->unstripBoth( $text );
4665 }
4666
4667 return $text;
4668 }
4669
4670 /**
4671 * Strip 3, 4 or 5 tildes out of signatures.
4672 *
4673 * @param string $text
4674 * @return string Signature text with /~{3,5}/ removed
4675 */
4676 public static function cleanSigInSig( $text ) {
4677 $text = preg_replace( '/~{3,5}/', '', $text );
4678 return $text;
4679 }
4680
4681 /**
4682 * Set up some variables which are usually set up in parse()
4683 * so that an external function can call some class members with confidence
4684 *
4685 * @param Title|null $title
4686 * @param ParserOptions $options
4687 * @param int $outputType
4688 * @param bool $clearState
4689 */
4690 public function startExternalParse( Title $title = null, ParserOptions $options,
4691 $outputType, $clearState = true
4692 ) {
4693 $this->startParse( $title, $options, $outputType, $clearState );
4694 }
4695
4696 /**
4697 * @param Title|null $title
4698 * @param ParserOptions $options
4699 * @param int $outputType
4700 * @param bool $clearState
4701 */
4702 private function startParse( Title $title = null, ParserOptions $options,
4703 $outputType, $clearState = true
4704 ) {
4705 $this->setTitle( $title );
4706 $this->mOptions = $options;
4707 $this->setOutputType( $outputType );
4708 if ( $clearState ) {
4709 $this->clearState();
4710 }
4711 }
4712
4713 /**
4714 * Wrapper for preprocess()
4715 *
4716 * @param string $text The text to preprocess
4717 * @param ParserOptions $options Options
4718 * @param Title|null $title Title object or null to use $wgTitle
4719 * @return string
4720 */
4721 public function transformMsg( $text, $options, $title = null ) {
4722 static $executing = false;
4723
4724 # Guard against infinite recursion
4725 if ( $executing ) {
4726 return $text;
4727 }
4728 $executing = true;
4729
4730 if ( !$title ) {
4731 global $wgTitle;
4732 $title = $wgTitle;
4733 }
4734
4735 $text = $this->preprocess( $text, $title, $options );
4736
4737 $executing = false;
4738 return $text;
4739 }
4740
4741 /**
4742 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4743 * The callback should have the following form:
4744 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4745 *
4746 * Transform and return $text. Use $parser for any required context, e.g. use
4747 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4748 *
4749 * Hooks may return extended information by returning an array, of which the
4750 * first numbered element (index 0) must be the return string, and all other
4751 * entries are extracted into local variables within an internal function
4752 * in the Parser class.
4753 *
4754 * This interface (introduced r61913) appears to be undocumented, but
4755 * 'markerType' is used by some core tag hooks to override which strip
4756 * array their results are placed in. **Use great caution if attempting
4757 * this interface, as it is not documented and injudicious use could smash
4758 * private variables.**
4759 *
4760 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4761 * @param callable $callback The callback function (and object) to use for the tag
4762 * @throws MWException
4763 * @return callable|null The old value of the mTagHooks array associated with the hook
4764 */
4765 public function setHook( $tag, $callback ) {
4766 $tag = strtolower( $tag );
4767 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4768 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4769 }
4770 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4771 $this->mTagHooks[$tag] = $callback;
4772 if ( !in_array( $tag, $this->mStripList ) ) {
4773 $this->mStripList[] = $tag;
4774 }
4775
4776 return $oldVal;
4777 }
4778
4779 /**
4780 * As setHook(), but letting the contents be parsed.
4781 *
4782 * Transparent tag hooks are like regular XML-style tag hooks, except they
4783 * operate late in the transformation sequence, on HTML instead of wikitext.
4784 *
4785 * This is probably obsoleted by things dealing with parser frames?
4786 * The only extension currently using it is geoserver.
4787 *
4788 * @since 1.10
4789 * @todo better document or deprecate this
4790 *
4791 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4792 * @param callable $callback The callback function (and object) to use for the tag
4793 * @throws MWException
4794 * @return callable|null The old value of the mTagHooks array associated with the hook
4795 */
4796 public function setTransparentTagHook( $tag, $callback ) {
4797 $tag = strtolower( $tag );
4798 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4799 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4800 }
4801 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4802 $this->mTransparentTagHooks[$tag] = $callback;
4803
4804 return $oldVal;
4805 }
4806
4807 /**
4808 * Remove all tag hooks
4809 */
4810 public function clearTagHooks() {
4811 $this->mTagHooks = [];
4812 $this->mFunctionTagHooks = [];
4813 $this->mStripList = $this->mDefaultStripList;
4814 }
4815
4816 /**
4817 * Create a function, e.g. {{sum:1|2|3}}
4818 * The callback function should have the form:
4819 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4820 *
4821 * Or with Parser::SFH_OBJECT_ARGS:
4822 * function myParserFunction( $parser, $frame, $args ) { ... }
4823 *
4824 * The callback may either return the text result of the function, or an array with the text
4825 * in element 0, and a number of flags in the other elements. The names of the flags are
4826 * specified in the keys. Valid flags are:
4827 * found The text returned is valid, stop processing the template. This
4828 * is on by default.
4829 * nowiki Wiki markup in the return value should be escaped
4830 * isHTML The returned text is HTML, armour it against wikitext transformation
4831 *
4832 * @param string $id The magic word ID
4833 * @param callable $callback The callback function (and object) to use
4834 * @param int $flags A combination of the following flags:
4835 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4836 *
4837 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4838 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4839 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4840 * the arguments, and to control the way they are expanded.
4841 *
4842 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4843 * arguments, for instance:
4844 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4845 *
4846 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4847 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4848 * working if/when this is changed.
4849 *
4850 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4851 * expansion.
4852 *
4853 * Please read the documentation in includes/parser/Preprocessor.php for more information
4854 * about the methods available in PPFrame and PPNode.
4855 *
4856 * @throws MWException
4857 * @return string|callable The old callback function for this name, if any
4858 */
4859 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4860 global $wgContLang;
4861
4862 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4863 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4864
4865 # Add to function cache
4866 $mw = MagicWord::get( $id );
4867 if ( !$mw ) {
4868 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4869 }
4870
4871 $synonyms = $mw->getSynonyms();
4872 $sensitive = intval( $mw->isCaseSensitive() );
4873
4874 foreach ( $synonyms as $syn ) {
4875 # Case
4876 if ( !$sensitive ) {
4877 $syn = $wgContLang->lc( $syn );
4878 }
4879 # Add leading hash
4880 if ( !( $flags & self::SFH_NO_HASH ) ) {
4881 $syn = '#' . $syn;
4882 }
4883 # Remove trailing colon
4884 if ( substr( $syn, -1, 1 ) === ':' ) {
4885 $syn = substr( $syn, 0, -1 );
4886 }
4887 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4888 }
4889 return $oldVal;
4890 }
4891
4892 /**
4893 * Get all registered function hook identifiers
4894 *
4895 * @return array
4896 */
4897 public function getFunctionHooks() {
4898 return array_keys( $this->mFunctionHooks );
4899 }
4900
4901 /**
4902 * Create a tag function, e.g. "<test>some stuff</test>".
4903 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4904 * Unlike parser functions, their content is not preprocessed.
4905 * @param string $tag
4906 * @param callable $callback
4907 * @param int $flags
4908 * @throws MWException
4909 * @return null
4910 */
4911 public function setFunctionTagHook( $tag, $callback, $flags ) {
4912 $tag = strtolower( $tag );
4913 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4914 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4915 }
4916 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4917 $this->mFunctionTagHooks[$tag] : null;
4918 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4919
4920 if ( !in_array( $tag, $this->mStripList ) ) {
4921 $this->mStripList[] = $tag;
4922 }
4923
4924 return $old;
4925 }
4926
4927 /**
4928 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4929 * Placeholders created in Linker::link()
4930 *
4931 * @param string $text
4932 * @param int $options
4933 */
4934 public function replaceLinkHolders( &$text, $options = 0 ) {
4935 $this->mLinkHolders->replace( $text );
4936 }
4937
4938 /**
4939 * Replace "<!--LINK-->" link placeholders with plain text of links
4940 * (not HTML-formatted).
4941 *
4942 * @param string $text
4943 * @return string
4944 */
4945 public function replaceLinkHoldersText( $text ) {
4946 return $this->mLinkHolders->replaceText( $text );
4947 }
4948
4949 /**
4950 * Renders an image gallery from a text with one line per image.
4951 * text labels may be given by using |-style alternative text. E.g.
4952 * Image:one.jpg|The number "1"
4953 * Image:tree.jpg|A tree
4954 * given as text will return the HTML of a gallery with two images,
4955 * labeled 'The number "1"' and
4956 * 'A tree'.
4957 *
4958 * @param string $text
4959 * @param array $params
4960 * @return string HTML
4961 */
4962 public function renderImageGallery( $text, $params ) {
4963
4964 $mode = false;
4965 if ( isset( $params['mode'] ) ) {
4966 $mode = $params['mode'];
4967 }
4968
4969 try {
4970 $ig = ImageGalleryBase::factory( $mode );
4971 } catch ( Exception $e ) {
4972 // If invalid type set, fallback to default.
4973 $ig = ImageGalleryBase::factory( false );
4974 }
4975
4976 $ig->setContextTitle( $this->mTitle );
4977 $ig->setShowBytes( false );
4978 $ig->setShowDimensions( false );
4979 $ig->setShowFilename( false );
4980 $ig->setParser( $this );
4981 $ig->setHideBadImages();
4982 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4983
4984 if ( isset( $params['showfilename'] ) ) {
4985 $ig->setShowFilename( true );
4986 } else {
4987 $ig->setShowFilename( false );
4988 }
4989 if ( isset( $params['caption'] ) ) {
4990 $caption = $params['caption'];
4991 $caption = htmlspecialchars( $caption );
4992 $caption = $this->replaceInternalLinks( $caption );
4993 $ig->setCaptionHtml( $caption );
4994 }
4995 if ( isset( $params['perrow'] ) ) {
4996 $ig->setPerRow( $params['perrow'] );
4997 }
4998 if ( isset( $params['widths'] ) ) {
4999 $ig->setWidths( $params['widths'] );
5000 }
5001 if ( isset( $params['heights'] ) ) {
5002 $ig->setHeights( $params['heights'] );
5003 }
5004 $ig->setAdditionalOptions( $params );
5005
5006 // Avoid PHP 7.1 warning from passing $this by reference
5007 $parser = $this;
5008 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5009
5010 $lines = StringUtils::explode( "\n", $text );
5011 foreach ( $lines as $line ) {
5012 # match lines like these:
5013 # Image:someimage.jpg|This is some image
5014 $matches = [];
5015 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5016 # Skip empty lines
5017 if ( count( $matches ) == 0 ) {
5018 continue;
5019 }
5020
5021 if ( strpos( $matches[0], '%' ) !== false ) {
5022 $matches[1] = rawurldecode( $matches[1] );
5023 }
5024 $title = Title::newFromText( $matches[1], NS_FILE );
5025 if ( is_null( $title ) ) {
5026 # Bogus title. Ignore these so we don't bomb out later.
5027 continue;
5028 }
5029
5030 # We need to get what handler the file uses, to figure out parameters.
5031 # Note, a hook can overide the file name, and chose an entirely different
5032 # file (which potentially could be of a different type and have different handler).
5033 $options = [];
5034 $descQuery = false;
5035 Hooks::run( 'BeforeParserFetchFileAndTitle',
5036 [ $this, $title, &$options, &$descQuery ] );
5037 # Don't register it now, as TraditionalImageGallery does that later.
5038 $file = $this->fetchFileNoRegister( $title, $options );
5039 $handler = $file ? $file->getHandler() : false;
5040
5041 $paramMap = [
5042 'img_alt' => 'gallery-internal-alt',
5043 'img_link' => 'gallery-internal-link',
5044 ];
5045 if ( $handler ) {
5046 $paramMap = $paramMap + $handler->getParamMap();
5047 // We don't want people to specify per-image widths.
5048 // Additionally the width parameter would need special casing anyhow.
5049 unset( $paramMap['img_width'] );
5050 }
5051
5052 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5053
5054 $label = '';
5055 $alt = '';
5056 $link = '';
5057 $handlerOptions = [];
5058 if ( isset( $matches[3] ) ) {
5059 // look for an |alt= definition while trying not to break existing
5060 // captions with multiple pipes (|) in it, until a more sensible grammar
5061 // is defined for images in galleries
5062
5063 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5064 // splitting on '|' is a bit odd, and different from makeImage.
5065 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5066 // Protect LanguageConverter markup
5067 $parameterMatches = StringUtils::delimiterExplode(
5068 '-{', '}-', '|', $matches[3], true /* nested */
5069 );
5070
5071 foreach ( $parameterMatches as $parameterMatch ) {
5072 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5073 if ( $magicName ) {
5074 $paramName = $paramMap[$magicName];
5075
5076 switch ( $paramName ) {
5077 case 'gallery-internal-alt':
5078 $alt = $this->stripAltText( $match, false );
5079 break;
5080 case 'gallery-internal-link':
5081 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5082 $chars = self::EXT_LINK_URL_CLASS;
5083 $addr = self::EXT_LINK_ADDR;
5084 $prots = $this->mUrlProtocols;
5085 // check to see if link matches an absolute url, if not then it must be a wiki link.
5086 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5087 // Result of LanguageConverter::markNoConversion
5088 // invoked on an external link.
5089 $linkValue = substr( $linkValue, 4, -2 );
5090 }
5091 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5092 $link = $linkValue;
5093 $this->mOutput->addExternalLink( $link );
5094 } else {
5095 $localLinkTitle = Title::newFromText( $linkValue );
5096 if ( $localLinkTitle !== null ) {
5097 $this->mOutput->addLink( $localLinkTitle );
5098 $link = $localLinkTitle->getLinkURL();
5099 }
5100 }
5101 break;
5102 default:
5103 // Must be a handler specific parameter.
5104 if ( $handler->validateParam( $paramName, $match ) ) {
5105 $handlerOptions[$paramName] = $match;
5106 } else {
5107 // Guess not, consider it as caption.
5108 wfDebug( "$parameterMatch failed parameter validation\n" );
5109 $label = '|' . $parameterMatch;
5110 }
5111 }
5112
5113 } else {
5114 // Last pipe wins.
5115 $label = '|' . $parameterMatch;
5116 }
5117 }
5118 // Remove the pipe.
5119 $label = substr( $label, 1 );
5120 }
5121
5122 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5123 }
5124 $html = $ig->toHTML();
5125 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5126 return $html;
5127 }
5128
5129 /**
5130 * @param MediaHandler $handler
5131 * @return array
5132 */
5133 public function getImageParams( $handler ) {
5134 if ( $handler ) {
5135 $handlerClass = get_class( $handler );
5136 } else {
5137 $handlerClass = '';
5138 }
5139 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5140 # Initialise static lists
5141 static $internalParamNames = [
5142 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5143 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5144 'bottom', 'text-bottom' ],
5145 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5146 'upright', 'border', 'link', 'alt', 'class' ],
5147 ];
5148 static $internalParamMap;
5149 if ( !$internalParamMap ) {
5150 $internalParamMap = [];
5151 foreach ( $internalParamNames as $type => $names ) {
5152 foreach ( $names as $name ) {
5153 // For grep: img_left, img_right, img_center, img_none,
5154 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5155 // img_bottom, img_text_bottom,
5156 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5157 // img_border, img_link, img_alt, img_class
5158 $magicName = str_replace( '-', '_', "img_$name" );
5159 $internalParamMap[$magicName] = [ $type, $name ];
5160 }
5161 }
5162 }
5163
5164 # Add handler params
5165 $paramMap = $internalParamMap;
5166 if ( $handler ) {
5167 $handlerParamMap = $handler->getParamMap();
5168 foreach ( $handlerParamMap as $magic => $paramName ) {
5169 $paramMap[$magic] = [ 'handler', $paramName ];
5170 }
5171 }
5172 $this->mImageParams[$handlerClass] = $paramMap;
5173 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5174 }
5175 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5176 }
5177
5178 /**
5179 * Parse image options text and use it to make an image
5180 *
5181 * @param Title $title
5182 * @param string $options
5183 * @param LinkHolderArray|bool $holders
5184 * @return string HTML
5185 */
5186 public function makeImage( $title, $options, $holders = false ) {
5187 # Check if the options text is of the form "options|alt text"
5188 # Options are:
5189 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5190 # * left no resizing, just left align. label is used for alt= only
5191 # * right same, but right aligned
5192 # * none same, but not aligned
5193 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5194 # * center center the image
5195 # * frame Keep original image size, no magnify-button.
5196 # * framed Same as "frame"
5197 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5198 # * upright reduce width for upright images, rounded to full __0 px
5199 # * border draw a 1px border around the image
5200 # * alt Text for HTML alt attribute (defaults to empty)
5201 # * class Set a class for img node
5202 # * link Set the target of the image link. Can be external, interwiki, or local
5203 # vertical-align values (no % or length right now):
5204 # * baseline
5205 # * sub
5206 # * super
5207 # * top
5208 # * text-top
5209 # * middle
5210 # * bottom
5211 # * text-bottom
5212
5213 # Protect LanguageConverter markup when splitting into parts
5214 $parts = StringUtils::delimiterExplode(
5215 '-{', '}-', '|', $options, true /* allow nesting */
5216 );
5217
5218 # Give extensions a chance to select the file revision for us
5219 $options = [];
5220 $descQuery = false;
5221 Hooks::run( 'BeforeParserFetchFileAndTitle',
5222 [ $this, $title, &$options, &$descQuery ] );
5223 # Fetch and register the file (file title may be different via hooks)
5224 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5225
5226 # Get parameter map
5227 $handler = $file ? $file->getHandler() : false;
5228
5229 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5230
5231 if ( !$file ) {
5232 $this->addTrackingCategory( 'broken-file-category' );
5233 }
5234
5235 # Process the input parameters
5236 $caption = '';
5237 $params = [ 'frame' => [], 'handler' => [],
5238 'horizAlign' => [], 'vertAlign' => [] ];
5239 $seenformat = false;
5240 foreach ( $parts as $part ) {
5241 $part = trim( $part );
5242 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5243 $validated = false;
5244 if ( isset( $paramMap[$magicName] ) ) {
5245 list( $type, $paramName ) = $paramMap[$magicName];
5246
5247 # Special case; width and height come in one variable together
5248 if ( $type === 'handler' && $paramName === 'width' ) {
5249 $parsedWidthParam = $this->parseWidthParam( $value );
5250 if ( isset( $parsedWidthParam['width'] ) ) {
5251 $width = $parsedWidthParam['width'];
5252 if ( $handler->validateParam( 'width', $width ) ) {
5253 $params[$type]['width'] = $width;
5254 $validated = true;
5255 }
5256 }
5257 if ( isset( $parsedWidthParam['height'] ) ) {
5258 $height = $parsedWidthParam['height'];
5259 if ( $handler->validateParam( 'height', $height ) ) {
5260 $params[$type]['height'] = $height;
5261 $validated = true;
5262 }
5263 }
5264 # else no validation -- T15436
5265 } else {
5266 if ( $type === 'handler' ) {
5267 # Validate handler parameter
5268 $validated = $handler->validateParam( $paramName, $value );
5269 } else {
5270 # Validate internal parameters
5271 switch ( $paramName ) {
5272 case 'manualthumb':
5273 case 'alt':
5274 case 'class':
5275 # @todo FIXME: Possibly check validity here for
5276 # manualthumb? downstream behavior seems odd with
5277 # missing manual thumbs.
5278 $validated = true;
5279 $value = $this->stripAltText( $value, $holders );
5280 break;
5281 case 'link':
5282 $chars = self::EXT_LINK_URL_CLASS;
5283 $addr = self::EXT_LINK_ADDR;
5284 $prots = $this->mUrlProtocols;
5285 if ( $value === '' ) {
5286 $paramName = 'no-link';
5287 $value = true;
5288 $validated = true;
5289 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5290 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5291 $paramName = 'link-url';
5292 $this->mOutput->addExternalLink( $value );
5293 if ( $this->mOptions->getExternalLinkTarget() ) {
5294 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5295 }
5296 $validated = true;
5297 }
5298 } else {
5299 $linkTitle = Title::newFromText( $value );
5300 if ( $linkTitle ) {
5301 $paramName = 'link-title';
5302 $value = $linkTitle;
5303 $this->mOutput->addLink( $linkTitle );
5304 $validated = true;
5305 }
5306 }
5307 break;
5308 case 'frameless':
5309 case 'framed':
5310 case 'thumbnail':
5311 // use first appearing option, discard others.
5312 $validated = !$seenformat;
5313 $seenformat = true;
5314 break;
5315 default:
5316 # Most other things appear to be empty or numeric...
5317 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5318 }
5319 }
5320
5321 if ( $validated ) {
5322 $params[$type][$paramName] = $value;
5323 }
5324 }
5325 }
5326 if ( !$validated ) {
5327 $caption = $part;
5328 }
5329 }
5330
5331 # Process alignment parameters
5332 if ( $params['horizAlign'] ) {
5333 $params['frame']['align'] = key( $params['horizAlign'] );
5334 }
5335 if ( $params['vertAlign'] ) {
5336 $params['frame']['valign'] = key( $params['vertAlign'] );
5337 }
5338
5339 $params['frame']['caption'] = $caption;
5340
5341 # Will the image be presented in a frame, with the caption below?
5342 $imageIsFramed = isset( $params['frame']['frame'] )
5343 || isset( $params['frame']['framed'] )
5344 || isset( $params['frame']['thumbnail'] )
5345 || isset( $params['frame']['manualthumb'] );
5346
5347 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5348 # came to also set the caption, ordinary text after the image -- which
5349 # makes no sense, because that just repeats the text multiple times in
5350 # screen readers. It *also* came to set the title attribute.
5351 # Now that we have an alt attribute, we should not set the alt text to
5352 # equal the caption: that's worse than useless, it just repeats the
5353 # text. This is the framed/thumbnail case. If there's no caption, we
5354 # use the unnamed parameter for alt text as well, just for the time be-
5355 # ing, if the unnamed param is set and the alt param is not.
5356 # For the future, we need to figure out if we want to tweak this more,
5357 # e.g., introducing a title= parameter for the title; ignoring the un-
5358 # named parameter entirely for images without a caption; adding an ex-
5359 # plicit caption= parameter and preserving the old magic unnamed para-
5360 # meter for BC; ...
5361 if ( $imageIsFramed ) { # Framed image
5362 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5363 # No caption or alt text, add the filename as the alt text so
5364 # that screen readers at least get some description of the image
5365 $params['frame']['alt'] = $title->getText();
5366 }
5367 # Do not set $params['frame']['title'] because tooltips don't make sense
5368 # for framed images
5369 } else { # Inline image
5370 if ( !isset( $params['frame']['alt'] ) ) {
5371 # No alt text, use the "caption" for the alt text
5372 if ( $caption !== '' ) {
5373 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5374 } else {
5375 # No caption, fall back to using the filename for the
5376 # alt text
5377 $params['frame']['alt'] = $title->getText();
5378 }
5379 }
5380 # Use the "caption" for the tooltip text
5381 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5382 }
5383
5384 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5385
5386 # Linker does the rest
5387 $time = isset( $options['time'] ) ? $options['time'] : false;
5388 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5389 $time, $descQuery, $this->mOptions->getThumbSize() );
5390
5391 # Give the handler a chance to modify the parser object
5392 if ( $handler ) {
5393 $handler->parserTransformHook( $this, $file );
5394 }
5395
5396 return $ret;
5397 }
5398
5399 /**
5400 * @param string $caption
5401 * @param LinkHolderArray|bool $holders
5402 * @return mixed|string
5403 */
5404 protected function stripAltText( $caption, $holders ) {
5405 # Strip bad stuff out of the title (tooltip). We can't just use
5406 # replaceLinkHoldersText() here, because if this function is called
5407 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5408 if ( $holders ) {
5409 $tooltip = $holders->replaceText( $caption );
5410 } else {
5411 $tooltip = $this->replaceLinkHoldersText( $caption );
5412 }
5413
5414 # make sure there are no placeholders in thumbnail attributes
5415 # that are later expanded to html- so expand them now and
5416 # remove the tags
5417 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5418 $tooltip = Sanitizer::stripAllTags( $tooltip );
5419
5420 return $tooltip;
5421 }
5422
5423 /**
5424 * Set a flag in the output object indicating that the content is dynamic and
5425 * shouldn't be cached.
5426 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5427 */
5428 public function disableCache() {
5429 wfDebug( "Parser output marked as uncacheable.\n" );
5430 if ( !$this->mOutput ) {
5431 throw new MWException( __METHOD__ .
5432 " can only be called when actually parsing something" );
5433 }
5434 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5435 }
5436
5437 /**
5438 * Callback from the Sanitizer for expanding items found in HTML attribute
5439 * values, so they can be safely tested and escaped.
5440 *
5441 * @param string $text
5442 * @param bool|PPFrame $frame
5443 * @return string
5444 */
5445 public function attributeStripCallback( &$text, $frame = false ) {
5446 $text = $this->replaceVariables( $text, $frame );
5447 $text = $this->mStripState->unstripBoth( $text );
5448 return $text;
5449 }
5450
5451 /**
5452 * Accessor
5453 *
5454 * @return array
5455 */
5456 public function getTags() {
5457 return array_merge(
5458 array_keys( $this->mTransparentTagHooks ),
5459 array_keys( $this->mTagHooks ),
5460 array_keys( $this->mFunctionTagHooks )
5461 );
5462 }
5463
5464 /**
5465 * Replace transparent tags in $text with the values given by the callbacks.
5466 *
5467 * Transparent tag hooks are like regular XML-style tag hooks, except they
5468 * operate late in the transformation sequence, on HTML instead of wikitext.
5469 *
5470 * @param string $text
5471 *
5472 * @return string
5473 */
5474 public function replaceTransparentTags( $text ) {
5475 $matches = [];
5476 $elements = array_keys( $this->mTransparentTagHooks );
5477 $text = self::extractTagsAndParams( $elements, $text, $matches );
5478 $replacements = [];
5479
5480 foreach ( $matches as $marker => $data ) {
5481 list( $element, $content, $params, $tag ) = $data;
5482 $tagName = strtolower( $element );
5483 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5484 $output = call_user_func_array(
5485 $this->mTransparentTagHooks[$tagName],
5486 [ $content, $params, $this ]
5487 );
5488 } else {
5489 $output = $tag;
5490 }
5491 $replacements[$marker] = $output;
5492 }
5493 return strtr( $text, $replacements );
5494 }
5495
5496 /**
5497 * Break wikitext input into sections, and either pull or replace
5498 * some particular section's text.
5499 *
5500 * External callers should use the getSection and replaceSection methods.
5501 *
5502 * @param string $text Page wikitext
5503 * @param string|int $sectionId A section identifier string of the form:
5504 * "<flag1> - <flag2> - ... - <section number>"
5505 *
5506 * Currently the only recognised flag is "T", which means the target section number
5507 * was derived during a template inclusion parse, in other words this is a template
5508 * section edit link. If no flags are given, it was an ordinary section edit link.
5509 * This flag is required to avoid a section numbering mismatch when a section is
5510 * enclosed by "<includeonly>" (T8563).
5511 *
5512 * The section number 0 pulls the text before the first heading; other numbers will
5513 * pull the given section along with its lower-level subsections. If the section is
5514 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5515 *
5516 * Section 0 is always considered to exist, even if it only contains the empty
5517 * string. If $text is the empty string and section 0 is replaced, $newText is
5518 * returned.
5519 *
5520 * @param string $mode One of "get" or "replace"
5521 * @param string $newText Replacement text for section data.
5522 * @return string For "get", the extracted section text.
5523 * for "replace", the whole page with the section replaced.
5524 */
5525 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5526 global $wgTitle; # not generally used but removes an ugly failure mode
5527
5528 $magicScopeVariable = $this->lock();
5529 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5530 $outText = '';
5531 $frame = $this->getPreprocessor()->newFrame();
5532
5533 # Process section extraction flags
5534 $flags = 0;
5535 $sectionParts = explode( '-', $sectionId );
5536 $sectionIndex = array_pop( $sectionParts );
5537 foreach ( $sectionParts as $part ) {
5538 if ( $part === 'T' ) {
5539 $flags |= self::PTD_FOR_INCLUSION;
5540 }
5541 }
5542
5543 # Check for empty input
5544 if ( strval( $text ) === '' ) {
5545 # Only sections 0 and T-0 exist in an empty document
5546 if ( $sectionIndex == 0 ) {
5547 if ( $mode === 'get' ) {
5548 return '';
5549 } else {
5550 return $newText;
5551 }
5552 } else {
5553 if ( $mode === 'get' ) {
5554 return $newText;
5555 } else {
5556 return $text;
5557 }
5558 }
5559 }
5560
5561 # Preprocess the text
5562 $root = $this->preprocessToDom( $text, $flags );
5563
5564 # <h> nodes indicate section breaks
5565 # They can only occur at the top level, so we can find them by iterating the root's children
5566 $node = $root->getFirstChild();
5567
5568 # Find the target section
5569 if ( $sectionIndex == 0 ) {
5570 # Section zero doesn't nest, level=big
5571 $targetLevel = 1000;
5572 } else {
5573 while ( $node ) {
5574 if ( $node->getName() === 'h' ) {
5575 $bits = $node->splitHeading();
5576 if ( $bits['i'] == $sectionIndex ) {
5577 $targetLevel = $bits['level'];
5578 break;
5579 }
5580 }
5581 if ( $mode === 'replace' ) {
5582 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5583 }
5584 $node = $node->getNextSibling();
5585 }
5586 }
5587
5588 if ( !$node ) {
5589 # Not found
5590 if ( $mode === 'get' ) {
5591 return $newText;
5592 } else {
5593 return $text;
5594 }
5595 }
5596
5597 # Find the end of the section, including nested sections
5598 do {
5599 if ( $node->getName() === 'h' ) {
5600 $bits = $node->splitHeading();
5601 $curLevel = $bits['level'];
5602 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5603 break;
5604 }
5605 }
5606 if ( $mode === 'get' ) {
5607 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5608 }
5609 $node = $node->getNextSibling();
5610 } while ( $node );
5611
5612 # Write out the remainder (in replace mode only)
5613 if ( $mode === 'replace' ) {
5614 # Output the replacement text
5615 # Add two newlines on -- trailing whitespace in $newText is conventionally
5616 # stripped by the editor, so we need both newlines to restore the paragraph gap
5617 # Only add trailing whitespace if there is newText
5618 if ( $newText != "" ) {
5619 $outText .= $newText . "\n\n";
5620 }
5621
5622 while ( $node ) {
5623 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5624 $node = $node->getNextSibling();
5625 }
5626 }
5627
5628 if ( is_string( $outText ) ) {
5629 # Re-insert stripped tags
5630 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5631 }
5632
5633 return $outText;
5634 }
5635
5636 /**
5637 * This function returns the text of a section, specified by a number ($section).
5638 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5639 * the first section before any such heading (section 0).
5640 *
5641 * If a section contains subsections, these are also returned.
5642 *
5643 * @param string $text Text to look in
5644 * @param string|int $sectionId Section identifier as a number or string
5645 * (e.g. 0, 1 or 'T-1').
5646 * @param string $defaultText Default to return if section is not found
5647 *
5648 * @return string Text of the requested section
5649 */
5650 public function getSection( $text, $sectionId, $defaultText = '' ) {
5651 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5652 }
5653
5654 /**
5655 * This function returns $oldtext after the content of the section
5656 * specified by $section has been replaced with $text. If the target
5657 * section does not exist, $oldtext is returned unchanged.
5658 *
5659 * @param string $oldText Former text of the article
5660 * @param string|int $sectionId Section identifier as a number or string
5661 * (e.g. 0, 1 or 'T-1').
5662 * @param string $newText Replacing text
5663 *
5664 * @return string Modified text
5665 */
5666 public function replaceSection( $oldText, $sectionId, $newText ) {
5667 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5668 }
5669
5670 /**
5671 * Get the ID of the revision we are parsing
5672 *
5673 * @return int|null
5674 */
5675 public function getRevisionId() {
5676 return $this->mRevisionId;
5677 }
5678
5679 /**
5680 * Get the revision object for $this->mRevisionId
5681 *
5682 * @return Revision|null Either a Revision object or null
5683 * @since 1.23 (public since 1.23)
5684 */
5685 public function getRevisionObject() {
5686 if ( !is_null( $this->mRevisionObject ) ) {
5687 return $this->mRevisionObject;
5688 }
5689 if ( is_null( $this->mRevisionId ) ) {
5690 return null;
5691 }
5692
5693 $rev = call_user_func(
5694 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5695 );
5696
5697 # If the parse is for a new revision, then the callback should have
5698 # already been set to force the object and should match mRevisionId.
5699 # If not, try to fetch by mRevisionId for sanity.
5700 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5701 $rev = Revision::newFromId( $this->mRevisionId );
5702 }
5703
5704 $this->mRevisionObject = $rev;
5705
5706 return $this->mRevisionObject;
5707 }
5708
5709 /**
5710 * Get the timestamp associated with the current revision, adjusted for
5711 * the default server-local timestamp
5712 * @return string
5713 */
5714 public function getRevisionTimestamp() {
5715 if ( is_null( $this->mRevisionTimestamp ) ) {
5716 global $wgContLang;
5717
5718 $revObject = $this->getRevisionObject();
5719 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5720
5721 # The cryptic '' timezone parameter tells to use the site-default
5722 # timezone offset instead of the user settings.
5723 # Since this value will be saved into the parser cache, served
5724 # to other users, and potentially even used inside links and such,
5725 # it needs to be consistent for all visitors.
5726 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5727
5728 }
5729 return $this->mRevisionTimestamp;
5730 }
5731
5732 /**
5733 * Get the name of the user that edited the last revision
5734 *
5735 * @return string User name
5736 */
5737 public function getRevisionUser() {
5738 if ( is_null( $this->mRevisionUser ) ) {
5739 $revObject = $this->getRevisionObject();
5740
5741 # if this template is subst: the revision id will be blank,
5742 # so just use the current user's name
5743 if ( $revObject ) {
5744 $this->mRevisionUser = $revObject->getUserText();
5745 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5746 $this->mRevisionUser = $this->getUser()->getName();
5747 }
5748 }
5749 return $this->mRevisionUser;
5750 }
5751
5752 /**
5753 * Get the size of the revision
5754 *
5755 * @return int|null Revision size
5756 */
5757 public function getRevisionSize() {
5758 if ( is_null( $this->mRevisionSize ) ) {
5759 $revObject = $this->getRevisionObject();
5760
5761 # if this variable is subst: the revision id will be blank,
5762 # so just use the parser input size, because the own substituation
5763 # will change the size.
5764 if ( $revObject ) {
5765 $this->mRevisionSize = $revObject->getSize();
5766 } else {
5767 $this->mRevisionSize = $this->mInputSize;
5768 }
5769 }
5770 return $this->mRevisionSize;
5771 }
5772
5773 /**
5774 * Mutator for $mDefaultSort
5775 *
5776 * @param string $sort New value
5777 */
5778 public function setDefaultSort( $sort ) {
5779 $this->mDefaultSort = $sort;
5780 $this->mOutput->setProperty( 'defaultsort', $sort );
5781 }
5782
5783 /**
5784 * Accessor for $mDefaultSort
5785 * Will use the empty string if none is set.
5786 *
5787 * This value is treated as a prefix, so the
5788 * empty string is equivalent to sorting by
5789 * page name.
5790 *
5791 * @return string
5792 */
5793 public function getDefaultSort() {
5794 if ( $this->mDefaultSort !== false ) {
5795 return $this->mDefaultSort;
5796 } else {
5797 return '';
5798 }
5799 }
5800
5801 /**
5802 * Accessor for $mDefaultSort
5803 * Unlike getDefaultSort(), will return false if none is set
5804 *
5805 * @return string|bool
5806 */
5807 public function getCustomDefaultSort() {
5808 return $this->mDefaultSort;
5809 }
5810
5811 /**
5812 * Try to guess the section anchor name based on a wikitext fragment
5813 * presumably extracted from a heading, for example "Header" from
5814 * "== Header ==".
5815 *
5816 * @param string $text
5817 *
5818 * @return string
5819 */
5820 public function guessSectionNameFromWikiText( $text ) {
5821 # Strip out wikitext links(they break the anchor)
5822 $text = $this->stripSectionName( $text );
5823 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5824 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5825 }
5826
5827 /**
5828 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5829 * instead. For use in redirects, since IE6 interprets Redirect: headers
5830 * as something other than UTF-8 (apparently?), resulting in breakage.
5831 *
5832 * @param string $text The section name
5833 * @return string An anchor
5834 */
5835 public function guessLegacySectionNameFromWikiText( $text ) {
5836 # Strip out wikitext links(they break the anchor)
5837 $text = $this->stripSectionName( $text );
5838 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5839 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5840 }
5841
5842 /**
5843 * Strips a text string of wikitext for use in a section anchor
5844 *
5845 * Accepts a text string and then removes all wikitext from the
5846 * string and leaves only the resultant text (i.e. the result of
5847 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5848 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5849 * to create valid section anchors by mimicing the output of the
5850 * parser when headings are parsed.
5851 *
5852 * @param string $text Text string to be stripped of wikitext
5853 * for use in a Section anchor
5854 * @return string Filtered text string
5855 */
5856 public function stripSectionName( $text ) {
5857 # Strip internal link markup
5858 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5859 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5860
5861 # Strip external link markup
5862 # @todo FIXME: Not tolerant to blank link text
5863 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5864 # on how many empty links there are on the page - need to figure that out.
5865 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5866
5867 # Parse wikitext quotes (italics & bold)
5868 $text = $this->doQuotes( $text );
5869
5870 # Strip HTML tags
5871 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5872 return $text;
5873 }
5874
5875 /**
5876 * strip/replaceVariables/unstrip for preprocessor regression testing
5877 *
5878 * @param string $text
5879 * @param Title $title
5880 * @param ParserOptions $options
5881 * @param int $outputType
5882 *
5883 * @return string
5884 */
5885 public function testSrvus( $text, Title $title, ParserOptions $options,
5886 $outputType = self::OT_HTML
5887 ) {
5888 $magicScopeVariable = $this->lock();
5889 $this->startParse( $title, $options, $outputType, true );
5890
5891 $text = $this->replaceVariables( $text );
5892 $text = $this->mStripState->unstripBoth( $text );
5893 $text = Sanitizer::removeHTMLtags( $text );
5894 return $text;
5895 }
5896
5897 /**
5898 * @param string $text
5899 * @param Title $title
5900 * @param ParserOptions $options
5901 * @return string
5902 */
5903 public function testPst( $text, Title $title, ParserOptions $options ) {
5904 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5905 }
5906
5907 /**
5908 * @param string $text
5909 * @param Title $title
5910 * @param ParserOptions $options
5911 * @return string
5912 */
5913 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5914 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5915 }
5916
5917 /**
5918 * Call a callback function on all regions of the given text that are not
5919 * inside strip markers, and replace those regions with the return value
5920 * of the callback. For example, with input:
5921 *
5922 * aaa<MARKER>bbb
5923 *
5924 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5925 * two strings will be replaced with the value returned by the callback in
5926 * each case.
5927 *
5928 * @param string $s
5929 * @param callable $callback
5930 *
5931 * @return string
5932 */
5933 public function markerSkipCallback( $s, $callback ) {
5934 $i = 0;
5935 $out = '';
5936 while ( $i < strlen( $s ) ) {
5937 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5938 if ( $markerStart === false ) {
5939 $out .= call_user_func( $callback, substr( $s, $i ) );
5940 break;
5941 } else {
5942 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5943 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5944 if ( $markerEnd === false ) {
5945 $out .= substr( $s, $markerStart );
5946 break;
5947 } else {
5948 $markerEnd += strlen( self::MARKER_SUFFIX );
5949 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5950 $i = $markerEnd;
5951 }
5952 }
5953 }
5954 return $out;
5955 }
5956
5957 /**
5958 * Remove any strip markers found in the given text.
5959 *
5960 * @param string $text Input string
5961 * @return string
5962 */
5963 public function killMarkers( $text ) {
5964 return $this->mStripState->killMarkers( $text );
5965 }
5966
5967 /**
5968 * Save the parser state required to convert the given half-parsed text to
5969 * HTML. "Half-parsed" in this context means the output of
5970 * recursiveTagParse() or internalParse(). This output has strip markers
5971 * from replaceVariables (extensionSubstitution() etc.), and link
5972 * placeholders from replaceLinkHolders().
5973 *
5974 * Returns an array which can be serialized and stored persistently. This
5975 * array can later be loaded into another parser instance with
5976 * unserializeHalfParsedText(). The text can then be safely incorporated into
5977 * the return value of a parser hook.
5978 *
5979 * @param string $text
5980 *
5981 * @return array
5982 */
5983 public function serializeHalfParsedText( $text ) {
5984 $data = [
5985 'text' => $text,
5986 'version' => self::HALF_PARSED_VERSION,
5987 'stripState' => $this->mStripState->getSubState( $text ),
5988 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5989 ];
5990 return $data;
5991 }
5992
5993 /**
5994 * Load the parser state given in the $data array, which is assumed to
5995 * have been generated by serializeHalfParsedText(). The text contents is
5996 * extracted from the array, and its markers are transformed into markers
5997 * appropriate for the current Parser instance. This transformed text is
5998 * returned, and can be safely included in the return value of a parser
5999 * hook.
6000 *
6001 * If the $data array has been stored persistently, the caller should first
6002 * check whether it is still valid, by calling isValidHalfParsedText().
6003 *
6004 * @param array $data Serialized data
6005 * @throws MWException
6006 * @return string
6007 */
6008 public function unserializeHalfParsedText( $data ) {
6009 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6010 throw new MWException( __METHOD__ . ': invalid version' );
6011 }
6012
6013 # First, extract the strip state.
6014 $texts = [ $data['text'] ];
6015 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6016
6017 # Now renumber links
6018 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6019
6020 # Should be good to go.
6021 return $texts[0];
6022 }
6023
6024 /**
6025 * Returns true if the given array, presumed to be generated by
6026 * serializeHalfParsedText(), is compatible with the current version of the
6027 * parser.
6028 *
6029 * @param array $data
6030 *
6031 * @return bool
6032 */
6033 public function isValidHalfParsedText( $data ) {
6034 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6035 }
6036
6037 /**
6038 * Parsed a width param of imagelink like 300px or 200x300px
6039 *
6040 * @param string $value
6041 *
6042 * @return array
6043 * @since 1.20
6044 */
6045 public function parseWidthParam( $value ) {
6046 $parsedWidthParam = [];
6047 if ( $value === '' ) {
6048 return $parsedWidthParam;
6049 }
6050 $m = [];
6051 # (T15500) In both cases (width/height and width only),
6052 # permit trailing "px" for backward compatibility.
6053 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6054 $width = intval( $m[1] );
6055 $height = intval( $m[2] );
6056 $parsedWidthParam['width'] = $width;
6057 $parsedWidthParam['height'] = $height;
6058 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6059 $width = intval( $value );
6060 $parsedWidthParam['width'] = $width;
6061 }
6062 return $parsedWidthParam;
6063 }
6064
6065 /**
6066 * Lock the current instance of the parser.
6067 *
6068 * This is meant to stop someone from calling the parser
6069 * recursively and messing up all the strip state.
6070 *
6071 * @throws MWException If parser is in a parse
6072 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6073 */
6074 protected function lock() {
6075 if ( $this->mInParse ) {
6076 throw new MWException( "Parser state cleared while parsing. "
6077 . "Did you call Parser::parse recursively? Lock is held by: " . $this->mInParse );
6078 }
6079
6080 // Save the backtrace when locking, so that if some code tries locking again,
6081 // we can print the lock owner's backtrace for easier debugging
6082 $e = new Exception;
6083 $this->mInParse = $e->getTraceAsString();
6084
6085 $recursiveCheck = new ScopedCallback( function() {
6086 $this->mInParse = false;
6087 } );
6088
6089 return $recursiveCheck;
6090 }
6091
6092 /**
6093 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6094 *
6095 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6096 * or if there is more than one <p/> tag in the input HTML.
6097 *
6098 * @param string $html
6099 * @return string
6100 * @since 1.24
6101 */
6102 public static function stripOuterParagraph( $html ) {
6103 $m = [];
6104 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6105 if ( strpos( $m[1], '</p>' ) === false ) {
6106 $html = $m[1];
6107 }
6108 }
6109
6110 return $html;
6111 }
6112
6113 /**
6114 * Return this parser if it is not doing anything, otherwise
6115 * get a fresh parser. You can use this method by doing
6116 * $myParser = $wgParser->getFreshParser(), or more simply
6117 * $wgParser->getFreshParser()->parse( ... );
6118 * if you're unsure if $wgParser is safe to use.
6119 *
6120 * @since 1.24
6121 * @return Parser A parser object that is not parsing anything
6122 */
6123 public function getFreshParser() {
6124 global $wgParserConf;
6125 if ( $this->mInParse ) {
6126 return new $wgParserConf['class']( $wgParserConf );
6127 } else {
6128 return $this;
6129 }
6130 }
6131
6132 /**
6133 * Set's up the PHP implementation of OOUI for use in this request
6134 * and instructs OutputPage to enable OOUI for itself.
6135 *
6136 * @since 1.26
6137 */
6138 public function enableOOUI() {
6139 OutputPage::setupOOUI();
6140 $this->mOutput->setEnableOOUI( true );
6141 }
6142 }
Wiklou, le Wiki du Wiklou