dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "EtcdConfig: Fix infinite timeout bug, and reduce timeout"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for T21052
92 # \x{FFFD} is the Unicode replacement character, which Preprocessor_DOM
93 # uses to replace invalid HTML characters.
94 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]';
95 # Simplified expression to match an IPv4 or IPv6 address, or
96 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
97 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}])';
98 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
99 // @codingStandardsIgnoreStart Generic.Files.LineLength
100 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}\x{FFFD}]+)
101 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
102 // @codingStandardsIgnoreEnd
103
104 # Regular expression for a non-newline space
105 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
106
107 # Flags for preprocessToDom
108 const PTD_FOR_INCLUSION = 1;
109
110 # Allowed values for $this->mOutputType
111 # Parameter to startExternalParse().
112 const OT_HTML = 1; # like parse()
113 const OT_WIKI = 2; # like preSaveTransform()
114 const OT_PREPROCESS = 3; # like preprocess()
115 const OT_MSG = 3;
116 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
117
118 /**
119 * @var string Prefix and suffix for temporary replacement strings
120 * for the multipass parser.
121 *
122 * \x7f should never appear in input as it's disallowed in XML.
123 * Using it at the front also gives us a little extra robustness
124 * since it shouldn't match when butted up against identifier-like
125 * string constructs.
126 *
127 * Must not consist of all title characters, or else it will change
128 * the behavior of <nowiki> in a link.
129 *
130 * Must have a character that needs escaping in attributes, otherwise
131 * someone could put a strip marker in an attribute, to get around
132 * escaping quote marks, and break out of the attribute. Thus we add
133 * `'".
134 */
135 const MARKER_SUFFIX = "-QINU`\"'\x7f";
136 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
137
138 # Markers used for wrapping the table of contents
139 const TOC_START = '<mw:toc>';
140 const TOC_END = '</mw:toc>';
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 public $mFirstCall = true;
155
156 # Initialised by initialiseVariables()
157
158 /**
159 * @var MagicWordArray
160 */
161 public $mVariables;
162
163 /**
164 * @var MagicWordArray
165 */
166 public $mSubstWords;
167 # Initialised in constructor
168 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
169
170 # Initialized in getPreprocessor()
171 /** @var Preprocessor */
172 public $mPreprocessor;
173
174 # Cleared with clearState():
175 /**
176 * @var ParserOutput
177 */
178 public $mOutput;
179 public $mAutonumber;
180
181 /**
182 * @var StripState
183 */
184 public $mStripState;
185
186 public $mIncludeCount;
187 /**
188 * @var LinkHolderArray
189 */
190 public $mLinkHolders;
191
192 public $mLinkID;
193 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
194 public $mDefaultSort;
195 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
196 public $mExpensiveFunctionCount; # number of expensive parser function calls
197 public $mShowToc, $mForceTocPosition;
198
199 /**
200 * @var User
201 */
202 public $mUser; # User object; only used when doing pre-save transform
203
204 # Temporary
205 # These are variables reset at least once per parse regardless of $clearState
206
207 /**
208 * @var ParserOptions
209 */
210 public $mOptions;
211
212 /**
213 * @var Title
214 */
215 public $mTitle; # Title context, used for self-link rendering and similar things
216 public $mOutputType; # Output type, one of the OT_xxx constants
217 public $ot; # Shortcut alias, see setOutputType()
218 public $mRevisionObject; # The revision object of the specified revision ID
219 public $mRevisionId; # ID to display in {{REVISIONID}} tags
220 public $mRevisionTimestamp; # The timestamp of the specified revision ID
221 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
222 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
223 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
224 public $mInputSize = false; # For {{PAGESIZE}} on current page.
225
226 /**
227 * @var string Deprecated accessor for the strip marker prefix.
228 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
229 */
230 public $mUniqPrefix = Parser::MARKER_PREFIX;
231
232 /**
233 * @var array Array with the language name of each language link (i.e. the
234 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
235 * duplicate language links to the ParserOutput.
236 */
237 public $mLangLinkLanguages;
238
239 /**
240 * @var MapCacheLRU|null
241 * @since 1.24
242 *
243 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
244 */
245 public $currentRevisionCache;
246
247 /**
248 * @var bool Recursive call protection.
249 * This variable should be treated as if it were private.
250 */
251 public $mInParse = false;
252
253 /** @var SectionProfiler */
254 protected $mProfiler;
255
256 /**
257 * @var LinkRenderer
258 */
259 protected $mLinkRenderer;
260
261 /**
262 * @param array $conf
263 */
264 public function __construct( $conf = [] ) {
265 $this->mConf = $conf;
266 $this->mUrlProtocols = wfUrlProtocols();
267 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
268 self::EXT_LINK_ADDR .
269 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F\\x{FFFD}]*?)\]/Su';
270 if ( isset( $conf['preprocessorClass'] ) ) {
271 $this->mPreprocessorClass = $conf['preprocessorClass'];
272 } elseif ( defined( 'HPHP_VERSION' ) ) {
273 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
274 $this->mPreprocessorClass = 'Preprocessor_Hash';
275 } elseif ( extension_loaded( 'domxml' ) ) {
276 # PECL extension that conflicts with the core DOM extension (T15770)
277 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
278 $this->mPreprocessorClass = 'Preprocessor_Hash';
279 } elseif ( extension_loaded( 'dom' ) ) {
280 $this->mPreprocessorClass = 'Preprocessor_DOM';
281 } else {
282 $this->mPreprocessorClass = 'Preprocessor_Hash';
283 }
284 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
285 }
286
287 /**
288 * Reduce memory usage to reduce the impact of circular references
289 */
290 public function __destruct() {
291 if ( isset( $this->mLinkHolders ) ) {
292 unset( $this->mLinkHolders );
293 }
294 foreach ( $this as $name => $value ) {
295 unset( $this->$name );
296 }
297 }
298
299 /**
300 * Allow extensions to clean up when the parser is cloned
301 */
302 public function __clone() {
303 $this->mInParse = false;
304
305 // T58226: When you create a reference "to" an object field, that
306 // makes the object field itself be a reference too (until the other
307 // reference goes out of scope). When cloning, any field that's a
308 // reference is copied as a reference in the new object. Both of these
309 // are defined PHP5 behaviors, as inconvenient as it is for us when old
310 // hooks from PHP4 days are passing fields by reference.
311 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
312 // Make a non-reference copy of the field, then rebind the field to
313 // reference the new copy.
314 $tmp = $this->$k;
315 $this->$k =& $tmp;
316 unset( $tmp );
317 }
318
319 Hooks::run( 'ParserCloned', [ $this ] );
320 }
321
322 /**
323 * Do various kinds of initialisation on the first call of the parser
324 */
325 public function firstCallInit() {
326 if ( !$this->mFirstCall ) {
327 return;
328 }
329 $this->mFirstCall = false;
330
331 CoreParserFunctions::register( $this );
332 CoreTagHooks::register( $this );
333 $this->initialiseVariables();
334
335 // Avoid PHP 7.1 warning from passing $this by reference
336 $parser = $this;
337 Hooks::run( 'ParserFirstCallInit', [ &$parser ] );
338 }
339
340 /**
341 * Clear Parser state
342 *
343 * @private
344 */
345 public function clearState() {
346 if ( $this->mFirstCall ) {
347 $this->firstCallInit();
348 }
349 $this->mOutput = new ParserOutput;
350 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
351 $this->mAutonumber = 0;
352 $this->mIncludeCount = [];
353 $this->mLinkHolders = new LinkHolderArray( $this );
354 $this->mLinkID = 0;
355 $this->mRevisionObject = $this->mRevisionTimestamp =
356 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
357 $this->mVarCache = [];
358 $this->mUser = null;
359 $this->mLangLinkLanguages = [];
360 $this->currentRevisionCache = null;
361
362 $this->mStripState = new StripState;
363
364 # Clear these on every parse, T6549
365 $this->mTplRedirCache = $this->mTplDomCache = [];
366
367 $this->mShowToc = true;
368 $this->mForceTocPosition = false;
369 $this->mIncludeSizes = [
370 'post-expand' => 0,
371 'arg' => 0,
372 ];
373 $this->mPPNodeCount = 0;
374 $this->mGeneratedPPNodeCount = 0;
375 $this->mHighestExpansionDepth = 0;
376 $this->mDefaultSort = false;
377 $this->mHeadings = [];
378 $this->mDoubleUnderscores = [];
379 $this->mExpensiveFunctionCount = 0;
380
381 # Fix cloning
382 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
383 $this->mPreprocessor = null;
384 }
385
386 $this->mProfiler = new SectionProfiler();
387
388 // Avoid PHP 7.1 warning from passing $this by reference
389 $parser = $this;
390 Hooks::run( 'ParserClearState', [ &$parser ] );
391 }
392
393 /**
394 * Convert wikitext to HTML
395 * Do not call this function recursively.
396 *
397 * @param string $text Text we want to parse
398 * @param Title $title
399 * @param ParserOptions $options
400 * @param bool $linestart
401 * @param bool $clearState
402 * @param int $revid Number to pass in {{REVISIONID}}
403 * @return ParserOutput A ParserOutput
404 */
405 public function parse(
406 $text, Title $title, ParserOptions $options,
407 $linestart = true, $clearState = true, $revid = null
408 ) {
409 /**
410 * First pass--just handle <nowiki> sections, pass the rest off
411 * to internalParse() which does all the real work.
412 */
413
414 global $wgShowHostnames;
415
416 if ( $clearState ) {
417 // We use U+007F DELETE to construct strip markers, so we have to make
418 // sure that this character does not occur in the input text.
419 $text = strtr( $text, "\x7f", "?" );
420 $magicScopeVariable = $this->lock();
421 }
422 // Strip U+0000 NULL (T159174)
423 $text = str_replace( "\000", '', $text );
424
425 $this->startParse( $title, $options, self::OT_HTML, $clearState );
426
427 $this->currentRevisionCache = null;
428 $this->mInputSize = strlen( $text );
429 if ( $this->mOptions->getEnableLimitReport() ) {
430 $this->mOutput->resetParseStartTime();
431 }
432
433 $oldRevisionId = $this->mRevisionId;
434 $oldRevisionObject = $this->mRevisionObject;
435 $oldRevisionTimestamp = $this->mRevisionTimestamp;
436 $oldRevisionUser = $this->mRevisionUser;
437 $oldRevisionSize = $this->mRevisionSize;
438 if ( $revid !== null ) {
439 $this->mRevisionId = $revid;
440 $this->mRevisionObject = null;
441 $this->mRevisionTimestamp = null;
442 $this->mRevisionUser = null;
443 $this->mRevisionSize = null;
444 }
445
446 // Avoid PHP 7.1 warning from passing $this by reference
447 $parser = $this;
448 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
449 # No more strip!
450 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
451 $text = $this->internalParse( $text );
452 Hooks::run( 'ParserAfterParse', [ &$parser, &$text, &$this->mStripState ] );
453
454 $text = $this->internalParseHalfParsed( $text, true, $linestart );
455
456 /**
457 * A converted title will be provided in the output object if title and
458 * content conversion are enabled, the article text does not contain
459 * a conversion-suppressing double-underscore tag, and no
460 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
461 * automatic link conversion.
462 */
463 if ( !( $options->getDisableTitleConversion()
464 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
465 || isset( $this->mDoubleUnderscores['notitleconvert'] )
466 || $this->mOutput->getDisplayTitle() !== false )
467 ) {
468 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
469 if ( $convruletitle ) {
470 $this->mOutput->setTitleText( $convruletitle );
471 } else {
472 $titleText = $this->getConverterLanguage()->convertTitle( $title );
473 $this->mOutput->setTitleText( $titleText );
474 }
475 }
476
477 # Done parsing! Compute runtime adaptive expiry if set
478 $this->mOutput->finalizeAdaptiveCacheExpiry();
479
480 # Warn if too many heavyweight parser functions were used
481 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
482 $this->limitationWarn( 'expensive-parserfunction',
483 $this->mExpensiveFunctionCount,
484 $this->mOptions->getExpensiveParserFunctionLimit()
485 );
486 }
487
488 # Information on include size limits, for the benefit of users who try to skirt them
489 if ( $this->mOptions->getEnableLimitReport() ) {
490 $max = $this->mOptions->getMaxIncludeSize();
491
492 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
493 if ( $cpuTime !== null ) {
494 $this->mOutput->setLimitReportData( 'limitreport-cputime',
495 sprintf( "%.3f", $cpuTime )
496 );
497 }
498
499 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
500 $this->mOutput->setLimitReportData( 'limitreport-walltime',
501 sprintf( "%.3f", $wallTime )
502 );
503
504 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
505 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
506 );
507 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
508 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
509 );
510 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
511 [ $this->mIncludeSizes['post-expand'], $max ]
512 );
513 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
514 [ $this->mIncludeSizes['arg'], $max ]
515 );
516 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
517 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
518 );
519 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
520 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
521 );
522 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
523
524 $limitReport = "NewPP limit report\n";
525 if ( $wgShowHostnames ) {
526 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
527 }
528 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
529 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
530 $limitReport .= 'Dynamic content: ' .
531 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
532 "\n";
533
534 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
535 if ( Hooks::run( 'ParserLimitReportFormat',
536 [ $key, &$value, &$limitReport, false, false ]
537 ) ) {
538 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
539 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
540 ->inLanguage( 'en' )->useDatabase( false );
541 if ( !$valueMsg->exists() ) {
542 $valueMsg = new RawMessage( '$1' );
543 }
544 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
545 $valueMsg->params( $value );
546 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
547 }
548 }
549 }
550 // Since we're not really outputting HTML, decode the entities and
551 // then re-encode the things that need hiding inside HTML comments.
552 $limitReport = htmlspecialchars_decode( $limitReport );
553 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
554
555 // Sanitize for comment. Note '‐' in the replacement is U+2010,
556 // which looks much like the problematic '-'.
557 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
558 $text .= "\n<!-- \n$limitReport-->\n";
559
560 // Add on template profiling data in human/machine readable way
561 $dataByFunc = $this->mProfiler->getFunctionStats();
562 uasort( $dataByFunc, function ( $a, $b ) {
563 return $a['real'] < $b['real']; // descending order
564 } );
565 $profileReport = [];
566 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
567 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
568 $item['%real'], $item['real'], $item['calls'],
569 htmlspecialchars( $item['name'] ) );
570 }
571 $text .= "<!--\nTransclusion expansion time report (%,ms,calls,template)\n";
572 $text .= implode( "\n", $profileReport ) . "\n-->\n";
573
574 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
575
576 // Add other cache related metadata
577 if ( $wgShowHostnames ) {
578 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
579 }
580 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
581 $this->mOutput->getCacheTime() );
582 $this->mOutput->setLimitReportData( 'cachereport-ttl',
583 $this->mOutput->getCacheExpiry() );
584 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
585 $this->mOutput->hasDynamicContent() );
586
587 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
588 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
589 $this->mTitle->getPrefixedDBkey() );
590 }
591 }
592 $this->mOutput->setText( $text );
593
594 $this->mRevisionId = $oldRevisionId;
595 $this->mRevisionObject = $oldRevisionObject;
596 $this->mRevisionTimestamp = $oldRevisionTimestamp;
597 $this->mRevisionUser = $oldRevisionUser;
598 $this->mRevisionSize = $oldRevisionSize;
599 $this->mInputSize = false;
600 $this->currentRevisionCache = null;
601
602 return $this->mOutput;
603 }
604
605 /**
606 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
607 * can be called from an extension tag hook.
608 *
609 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
610 * instead, which means that lists and links have not been fully parsed yet,
611 * and strip markers are still present.
612 *
613 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
614 *
615 * Use this function if you're a parser tag hook and you want to parse
616 * wikitext before or after applying additional transformations, and you
617 * intend to *return the result as hook output*, which will cause it to go
618 * through the rest of parsing process automatically.
619 *
620 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
621 * $text are not expanded
622 *
623 * @param string $text Text extension wants to have parsed
624 * @param bool|PPFrame $frame The frame to use for expanding any template variables
625 * @return string UNSAFE half-parsed HTML
626 */
627 public function recursiveTagParse( $text, $frame = false ) {
628 // Avoid PHP 7.1 warning from passing $this by reference
629 $parser = $this;
630 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
631 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
632 $text = $this->internalParse( $text, false, $frame );
633 return $text;
634 }
635
636 /**
637 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
638 * point can be called from an extension tag hook.
639 *
640 * The output of this function is fully-parsed HTML that is safe for output.
641 * If you're a parser tag hook, you might want to use recursiveTagParse()
642 * instead.
643 *
644 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
645 * $text are not expanded
646 *
647 * @since 1.25
648 *
649 * @param string $text Text extension wants to have parsed
650 * @param bool|PPFrame $frame The frame to use for expanding any template variables
651 * @return string Fully parsed HTML
652 */
653 public function recursiveTagParseFully( $text, $frame = false ) {
654 $text = $this->recursiveTagParse( $text, $frame );
655 $text = $this->internalParseHalfParsed( $text, false );
656 return $text;
657 }
658
659 /**
660 * Expand templates and variables in the text, producing valid, static wikitext.
661 * Also removes comments.
662 * Do not call this function recursively.
663 * @param string $text
664 * @param Title $title
665 * @param ParserOptions $options
666 * @param int|null $revid
667 * @param bool|PPFrame $frame
668 * @return mixed|string
669 */
670 public function preprocess( $text, Title $title = null,
671 ParserOptions $options, $revid = null, $frame = false
672 ) {
673 $magicScopeVariable = $this->lock();
674 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
675 if ( $revid !== null ) {
676 $this->mRevisionId = $revid;
677 }
678 // Avoid PHP 7.1 warning from passing $this by reference
679 $parser = $this;
680 Hooks::run( 'ParserBeforeStrip', [ &$parser, &$text, &$this->mStripState ] );
681 Hooks::run( 'ParserAfterStrip', [ &$parser, &$text, &$this->mStripState ] );
682 $text = $this->replaceVariables( $text, $frame );
683 $text = $this->mStripState->unstripBoth( $text );
684 return $text;
685 }
686
687 /**
688 * Recursive parser entry point that can be called from an extension tag
689 * hook.
690 *
691 * @param string $text Text to be expanded
692 * @param bool|PPFrame $frame The frame to use for expanding any template variables
693 * @return string
694 * @since 1.19
695 */
696 public function recursivePreprocess( $text, $frame = false ) {
697 $text = $this->replaceVariables( $text, $frame );
698 $text = $this->mStripState->unstripBoth( $text );
699 return $text;
700 }
701
702 /**
703 * Process the wikitext for the "?preload=" feature. (T7210)
704 *
705 * "<noinclude>", "<includeonly>" etc. are parsed as for template
706 * transclusion, comments, templates, arguments, tags hooks and parser
707 * functions are untouched.
708 *
709 * @param string $text
710 * @param Title $title
711 * @param ParserOptions $options
712 * @param array $params
713 * @return string
714 */
715 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
716 $msg = new RawMessage( $text );
717 $text = $msg->params( $params )->plain();
718
719 # Parser (re)initialisation
720 $magicScopeVariable = $this->lock();
721 $this->startParse( $title, $options, self::OT_PLAIN, true );
722
723 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
724 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
725 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
726 $text = $this->mStripState->unstripBoth( $text );
727 return $text;
728 }
729
730 /**
731 * Get a random string
732 *
733 * @return string
734 * @deprecated since 1.26; use wfRandomString() instead.
735 */
736 public static function getRandomString() {
737 wfDeprecated( __METHOD__, '1.26' );
738 return wfRandomString( 16 );
739 }
740
741 /**
742 * Set the current user.
743 * Should only be used when doing pre-save transform.
744 *
745 * @param User|null $user User object or null (to reset)
746 */
747 public function setUser( $user ) {
748 $this->mUser = $user;
749 }
750
751 /**
752 * Accessor for mUniqPrefix.
753 *
754 * @return string
755 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
756 */
757 public function uniqPrefix() {
758 wfDeprecated( __METHOD__, '1.26' );
759 return self::MARKER_PREFIX;
760 }
761
762 /**
763 * Set the context title
764 *
765 * @param Title $t
766 */
767 public function setTitle( $t ) {
768 if ( !$t ) {
769 $t = Title::newFromText( 'NO TITLE' );
770 }
771
772 if ( $t->hasFragment() ) {
773 # Strip the fragment to avoid various odd effects
774 $this->mTitle = $t->createFragmentTarget( '' );
775 } else {
776 $this->mTitle = $t;
777 }
778 }
779
780 /**
781 * Accessor for the Title object
782 *
783 * @return Title
784 */
785 public function getTitle() {
786 return $this->mTitle;
787 }
788
789 /**
790 * Accessor/mutator for the Title object
791 *
792 * @param Title $x Title object or null to just get the current one
793 * @return Title
794 */
795 public function Title( $x = null ) {
796 return wfSetVar( $this->mTitle, $x );
797 }
798
799 /**
800 * Set the output type
801 *
802 * @param int $ot New value
803 */
804 public function setOutputType( $ot ) {
805 $this->mOutputType = $ot;
806 # Shortcut alias
807 $this->ot = [
808 'html' => $ot == self::OT_HTML,
809 'wiki' => $ot == self::OT_WIKI,
810 'pre' => $ot == self::OT_PREPROCESS,
811 'plain' => $ot == self::OT_PLAIN,
812 ];
813 }
814
815 /**
816 * Accessor/mutator for the output type
817 *
818 * @param int|null $x New value or null to just get the current one
819 * @return int
820 */
821 public function OutputType( $x = null ) {
822 return wfSetVar( $this->mOutputType, $x );
823 }
824
825 /**
826 * Get the ParserOutput object
827 *
828 * @return ParserOutput
829 */
830 public function getOutput() {
831 return $this->mOutput;
832 }
833
834 /**
835 * Get the ParserOptions object
836 *
837 * @return ParserOptions
838 */
839 public function getOptions() {
840 return $this->mOptions;
841 }
842
843 /**
844 * Accessor/mutator for the ParserOptions object
845 *
846 * @param ParserOptions $x New value or null to just get the current one
847 * @return ParserOptions Current ParserOptions object
848 */
849 public function Options( $x = null ) {
850 return wfSetVar( $this->mOptions, $x );
851 }
852
853 /**
854 * @return int
855 */
856 public function nextLinkID() {
857 return $this->mLinkID++;
858 }
859
860 /**
861 * @param int $id
862 */
863 public function setLinkID( $id ) {
864 $this->mLinkID = $id;
865 }
866
867 /**
868 * Get a language object for use in parser functions such as {{FORMATNUM:}}
869 * @return Language
870 */
871 public function getFunctionLang() {
872 return $this->getTargetLanguage();
873 }
874
875 /**
876 * Get the target language for the content being parsed. This is usually the
877 * language that the content is in.
878 *
879 * @since 1.19
880 *
881 * @throws MWException
882 * @return Language
883 */
884 public function getTargetLanguage() {
885 $target = $this->mOptions->getTargetLanguage();
886
887 if ( $target !== null ) {
888 return $target;
889 } elseif ( $this->mOptions->getInterfaceMessage() ) {
890 return $this->mOptions->getUserLangObj();
891 } elseif ( is_null( $this->mTitle ) ) {
892 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
893 }
894
895 return $this->mTitle->getPageLanguage();
896 }
897
898 /**
899 * Get the language object for language conversion
900 * @return Language|null
901 */
902 public function getConverterLanguage() {
903 return $this->getTargetLanguage();
904 }
905
906 /**
907 * Get a User object either from $this->mUser, if set, or from the
908 * ParserOptions object otherwise
909 *
910 * @return User
911 */
912 public function getUser() {
913 if ( !is_null( $this->mUser ) ) {
914 return $this->mUser;
915 }
916 return $this->mOptions->getUser();
917 }
918
919 /**
920 * Get a preprocessor object
921 *
922 * @return Preprocessor
923 */
924 public function getPreprocessor() {
925 if ( !isset( $this->mPreprocessor ) ) {
926 $class = $this->mPreprocessorClass;
927 $this->mPreprocessor = new $class( $this );
928 }
929 return $this->mPreprocessor;
930 }
931
932 /**
933 * Get a LinkRenderer instance to make links with
934 *
935 * @since 1.28
936 * @return LinkRenderer
937 */
938 public function getLinkRenderer() {
939 if ( !$this->mLinkRenderer ) {
940 $this->mLinkRenderer = MediaWikiServices::getInstance()
941 ->getLinkRendererFactory()->create();
942 $this->mLinkRenderer->setStubThreshold(
943 $this->getOptions()->getStubThreshold()
944 );
945 }
946
947 return $this->mLinkRenderer;
948 }
949
950 /**
951 * Replaces all occurrences of HTML-style comments and the given tags
952 * in the text with a random marker and returns the next text. The output
953 * parameter $matches will be an associative array filled with data in
954 * the form:
955 *
956 * @code
957 * 'UNIQ-xxxxx' => [
958 * 'element',
959 * 'tag content',
960 * [ 'param' => 'x' ],
961 * '<element param="x">tag content</element>' ]
962 * @endcode
963 *
964 * @param array $elements List of element names. Comments are always extracted.
965 * @param string $text Source text string.
966 * @param array $matches Out parameter, Array: extracted tags
967 * @param string|null $uniq_prefix
968 * @return string Stripped text
969 * @since 1.26 The uniq_prefix argument is deprecated.
970 */
971 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
972 if ( $uniq_prefix !== null ) {
973 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
974 }
975 static $n = 1;
976 $stripped = '';
977 $matches = [];
978
979 $taglist = implode( '|', $elements );
980 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
981
982 while ( $text != '' ) {
983 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
984 $stripped .= $p[0];
985 if ( count( $p ) < 5 ) {
986 break;
987 }
988 if ( count( $p ) > 5 ) {
989 # comment
990 $element = $p[4];
991 $attributes = '';
992 $close = '';
993 $inside = $p[5];
994 } else {
995 # tag
996 $element = $p[1];
997 $attributes = $p[2];
998 $close = $p[3];
999 $inside = $p[4];
1000 }
1001
1002 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
1003 $stripped .= $marker;
1004
1005 if ( $close === '/>' ) {
1006 # Empty element tag, <tag />
1007 $content = null;
1008 $text = $inside;
1009 $tail = null;
1010 } else {
1011 if ( $element === '!--' ) {
1012 $end = '/(-->)/';
1013 } else {
1014 $end = "/(<\\/$element\\s*>)/i";
1015 }
1016 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
1017 $content = $q[0];
1018 if ( count( $q ) < 3 ) {
1019 # No end tag -- let it run out to the end of the text.
1020 $tail = '';
1021 $text = '';
1022 } else {
1023 $tail = $q[1];
1024 $text = $q[2];
1025 }
1026 }
1027
1028 $matches[$marker] = [ $element,
1029 $content,
1030 Sanitizer::decodeTagAttributes( $attributes ),
1031 "<$element$attributes$close$content$tail" ];
1032 }
1033 return $stripped;
1034 }
1035
1036 /**
1037 * Get a list of strippable XML-like elements
1038 *
1039 * @return array
1040 */
1041 public function getStripList() {
1042 return $this->mStripList;
1043 }
1044
1045 /**
1046 * Add an item to the strip state
1047 * Returns the unique tag which must be inserted into the stripped text
1048 * The tag will be replaced with the original text in unstrip()
1049 *
1050 * @param string $text
1051 *
1052 * @return string
1053 */
1054 public function insertStripItem( $text ) {
1055 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1056 $this->mMarkerIndex++;
1057 $this->mStripState->addGeneral( $marker, $text );
1058 return $marker;
1059 }
1060
1061 /**
1062 * parse the wiki syntax used to render tables
1063 *
1064 * @private
1065 * @param string $text
1066 * @return string
1067 */
1068 public function doTableStuff( $text ) {
1069
1070 $lines = StringUtils::explode( "\n", $text );
1071 $out = '';
1072 $td_history = []; # Is currently a td tag open?
1073 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1074 $tr_history = []; # Is currently a tr tag open?
1075 $tr_attributes = []; # history of tr attributes
1076 $has_opened_tr = []; # Did this table open a <tr> element?
1077 $indent_level = 0; # indent level of the table
1078
1079 foreach ( $lines as $outLine ) {
1080 $line = trim( $outLine );
1081
1082 if ( $line === '' ) { # empty line, go to next line
1083 $out .= $outLine . "\n";
1084 continue;
1085 }
1086
1087 $first_character = $line[0];
1088 $first_two = substr( $line, 0, 2 );
1089 $matches = [];
1090
1091 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1092 # First check if we are starting a new table
1093 $indent_level = strlen( $matches[1] );
1094
1095 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1096 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1097
1098 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1099 array_push( $td_history, false );
1100 array_push( $last_tag_history, '' );
1101 array_push( $tr_history, false );
1102 array_push( $tr_attributes, '' );
1103 array_push( $has_opened_tr, false );
1104 } elseif ( count( $td_history ) == 0 ) {
1105 # Don't do any of the following
1106 $out .= $outLine . "\n";
1107 continue;
1108 } elseif ( $first_two === '|}' ) {
1109 # We are ending a table
1110 $line = '</table>' . substr( $line, 2 );
1111 $last_tag = array_pop( $last_tag_history );
1112
1113 if ( !array_pop( $has_opened_tr ) ) {
1114 $line = "<tr><td></td></tr>{$line}";
1115 }
1116
1117 if ( array_pop( $tr_history ) ) {
1118 $line = "</tr>{$line}";
1119 }
1120
1121 if ( array_pop( $td_history ) ) {
1122 $line = "</{$last_tag}>{$line}";
1123 }
1124 array_pop( $tr_attributes );
1125 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1126 } elseif ( $first_two === '|-' ) {
1127 # Now we have a table row
1128 $line = preg_replace( '#^\|-+#', '', $line );
1129
1130 # Whats after the tag is now only attributes
1131 $attributes = $this->mStripState->unstripBoth( $line );
1132 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1133 array_pop( $tr_attributes );
1134 array_push( $tr_attributes, $attributes );
1135
1136 $line = '';
1137 $last_tag = array_pop( $last_tag_history );
1138 array_pop( $has_opened_tr );
1139 array_push( $has_opened_tr, true );
1140
1141 if ( array_pop( $tr_history ) ) {
1142 $line = '</tr>';
1143 }
1144
1145 if ( array_pop( $td_history ) ) {
1146 $line = "</{$last_tag}>{$line}";
1147 }
1148
1149 $outLine = $line;
1150 array_push( $tr_history, false );
1151 array_push( $td_history, false );
1152 array_push( $last_tag_history, '' );
1153 } elseif ( $first_character === '|'
1154 || $first_character === '!'
1155 || $first_two === '|+'
1156 ) {
1157 # This might be cell elements, td, th or captions
1158 if ( $first_two === '|+' ) {
1159 $first_character = '+';
1160 $line = substr( $line, 2 );
1161 } else {
1162 $line = substr( $line, 1 );
1163 }
1164
1165 // Implies both are valid for table headings.
1166 if ( $first_character === '!' ) {
1167 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1168 }
1169
1170 # Split up multiple cells on the same line.
1171 # FIXME : This can result in improper nesting of tags processed
1172 # by earlier parser steps.
1173 $cells = explode( '||', $line );
1174
1175 $outLine = '';
1176
1177 # Loop through each table cell
1178 foreach ( $cells as $cell ) {
1179 $previous = '';
1180 if ( $first_character !== '+' ) {
1181 $tr_after = array_pop( $tr_attributes );
1182 if ( !array_pop( $tr_history ) ) {
1183 $previous = "<tr{$tr_after}>\n";
1184 }
1185 array_push( $tr_history, true );
1186 array_push( $tr_attributes, '' );
1187 array_pop( $has_opened_tr );
1188 array_push( $has_opened_tr, true );
1189 }
1190
1191 $last_tag = array_pop( $last_tag_history );
1192
1193 if ( array_pop( $td_history ) ) {
1194 $previous = "</{$last_tag}>\n{$previous}";
1195 }
1196
1197 if ( $first_character === '|' ) {
1198 $last_tag = 'td';
1199 } elseif ( $first_character === '!' ) {
1200 $last_tag = 'th';
1201 } elseif ( $first_character === '+' ) {
1202 $last_tag = 'caption';
1203 } else {
1204 $last_tag = '';
1205 }
1206
1207 array_push( $last_tag_history, $last_tag );
1208
1209 # A cell could contain both parameters and data
1210 $cell_data = explode( '|', $cell, 2 );
1211
1212 # T2553: Note that a '|' inside an invalid link should not
1213 # be mistaken as delimiting cell parameters
1214 # Bug T153140: Neither should language converter markup.
1215 if ( preg_match( '/\[\[|-\{/', $cell_data[0] ) === 1 ) {
1216 $cell = "{$previous}<{$last_tag}>{$cell}";
1217 } elseif ( count( $cell_data ) == 1 ) {
1218 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1219 } else {
1220 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1221 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1222 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1223 }
1224
1225 $outLine .= $cell;
1226 array_push( $td_history, true );
1227 }
1228 }
1229 $out .= $outLine . "\n";
1230 }
1231
1232 # Closing open td, tr && table
1233 while ( count( $td_history ) > 0 ) {
1234 if ( array_pop( $td_history ) ) {
1235 $out .= "</td>\n";
1236 }
1237 if ( array_pop( $tr_history ) ) {
1238 $out .= "</tr>\n";
1239 }
1240 if ( !array_pop( $has_opened_tr ) ) {
1241 $out .= "<tr><td></td></tr>\n";
1242 }
1243
1244 $out .= "</table>\n";
1245 }
1246
1247 # Remove trailing line-ending (b/c)
1248 if ( substr( $out, -1 ) === "\n" ) {
1249 $out = substr( $out, 0, -1 );
1250 }
1251
1252 # special case: don't return empty table
1253 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1254 $out = '';
1255 }
1256
1257 return $out;
1258 }
1259
1260 /**
1261 * Helper function for parse() that transforms wiki markup into half-parsed
1262 * HTML. Only called for $mOutputType == self::OT_HTML.
1263 *
1264 * @private
1265 *
1266 * @param string $text The text to parse
1267 * @param bool $isMain Whether this is being called from the main parse() function
1268 * @param PPFrame|bool $frame A pre-processor frame
1269 *
1270 * @return string
1271 */
1272 public function internalParse( $text, $isMain = true, $frame = false ) {
1273
1274 $origText = $text;
1275
1276 // Avoid PHP 7.1 warning from passing $this by reference
1277 $parser = $this;
1278
1279 # Hook to suspend the parser in this state
1280 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$parser, &$text, &$this->mStripState ] ) ) {
1281 return $text;
1282 }
1283
1284 # if $frame is provided, then use $frame for replacing any variables
1285 if ( $frame ) {
1286 # use frame depth to infer how include/noinclude tags should be handled
1287 # depth=0 means this is the top-level document; otherwise it's an included document
1288 if ( !$frame->depth ) {
1289 $flag = 0;
1290 } else {
1291 $flag = Parser::PTD_FOR_INCLUSION;
1292 }
1293 $dom = $this->preprocessToDom( $text, $flag );
1294 $text = $frame->expand( $dom );
1295 } else {
1296 # if $frame is not provided, then use old-style replaceVariables
1297 $text = $this->replaceVariables( $text );
1298 }
1299
1300 Hooks::run( 'InternalParseBeforeSanitize', [ &$parser, &$text, &$this->mStripState ] );
1301 $text = Sanitizer::removeHTMLtags(
1302 $text,
1303 [ $this, 'attributeStripCallback' ],
1304 false,
1305 array_keys( $this->mTransparentTagHooks ),
1306 [],
1307 [ $this, 'addTrackingCategory' ]
1308 );
1309 Hooks::run( 'InternalParseBeforeLinks', [ &$parser, &$text, &$this->mStripState ] );
1310
1311 # Tables need to come after variable replacement for things to work
1312 # properly; putting them before other transformations should keep
1313 # exciting things like link expansions from showing up in surprising
1314 # places.
1315 $text = $this->doTableStuff( $text );
1316
1317 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1318
1319 $text = $this->doDoubleUnderscore( $text );
1320
1321 $text = $this->doHeadings( $text );
1322 $text = $this->replaceInternalLinks( $text );
1323 $text = $this->doAllQuotes( $text );
1324 $text = $this->replaceExternalLinks( $text );
1325
1326 # replaceInternalLinks may sometimes leave behind
1327 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1328 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1329
1330 $text = $this->doMagicLinks( $text );
1331 $text = $this->formatHeadings( $text, $origText, $isMain );
1332
1333 return $text;
1334 }
1335
1336 /**
1337 * Helper function for parse() that transforms half-parsed HTML into fully
1338 * parsed HTML.
1339 *
1340 * @param string $text
1341 * @param bool $isMain
1342 * @param bool $linestart
1343 * @return string
1344 */
1345 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1346 $text = $this->mStripState->unstripGeneral( $text );
1347
1348 // Avoid PHP 7.1 warning from passing $this by reference
1349 $parser = $this;
1350
1351 if ( $isMain ) {
1352 Hooks::run( 'ParserAfterUnstrip', [ &$parser, &$text ] );
1353 }
1354
1355 # Clean up special characters, only run once, next-to-last before doBlockLevels
1356 $fixtags = [
1357 # French spaces, last one Guillemet-left
1358 # only if there is something before the space
1359 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1360 # french spaces, Guillemet-right
1361 '/(\\302\\253) /' => '\\1&#160;',
1362 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, T13874.
1363 ];
1364 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1365
1366 $text = $this->doBlockLevels( $text, $linestart );
1367
1368 $this->replaceLinkHolders( $text );
1369
1370 /**
1371 * The input doesn't get language converted if
1372 * a) It's disabled
1373 * b) Content isn't converted
1374 * c) It's a conversion table
1375 * d) it is an interface message (which is in the user language)
1376 */
1377 if ( !( $this->mOptions->getDisableContentConversion()
1378 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1379 ) {
1380 if ( !$this->mOptions->getInterfaceMessage() ) {
1381 # The position of the convert() call should not be changed. it
1382 # assumes that the links are all replaced and the only thing left
1383 # is the <nowiki> mark.
1384 $text = $this->getConverterLanguage()->convert( $text );
1385 }
1386 }
1387
1388 $text = $this->mStripState->unstripNoWiki( $text );
1389
1390 if ( $isMain ) {
1391 Hooks::run( 'ParserBeforeTidy', [ &$parser, &$text ] );
1392 }
1393
1394 $text = $this->replaceTransparentTags( $text );
1395 $text = $this->mStripState->unstripGeneral( $text );
1396
1397 $text = Sanitizer::normalizeCharReferences( $text );
1398
1399 if ( MWTidy::isEnabled() ) {
1400 if ( $this->mOptions->getTidy() ) {
1401 $text = MWTidy::tidy( $text );
1402 }
1403 } else {
1404 # attempt to sanitize at least some nesting problems
1405 # (T4702 and quite a few others)
1406 $tidyregs = [
1407 # ''Something [http://www.cool.com cool''] -->
1408 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1409 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1410 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1411 # fix up an anchor inside another anchor, only
1412 # at least for a single single nested link (T5695)
1413 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1414 '\\1\\2</a>\\3</a>\\1\\4</a>',
1415 # fix div inside inline elements- doBlockLevels won't wrap a line which
1416 # contains a div, so fix it up here; replace
1417 # div with escaped text
1418 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1419 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1420 # remove empty italic or bold tag pairs, some
1421 # introduced by rules above
1422 '/<([bi])><\/\\1>/' => '',
1423 ];
1424
1425 $text = preg_replace(
1426 array_keys( $tidyregs ),
1427 array_values( $tidyregs ),
1428 $text );
1429 }
1430
1431 if ( $isMain ) {
1432 Hooks::run( 'ParserAfterTidy', [ &$parser, &$text ] );
1433 }
1434
1435 return $text;
1436 }
1437
1438 /**
1439 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1440 * magic external links.
1441 *
1442 * DML
1443 * @private
1444 *
1445 * @param string $text
1446 *
1447 * @return string
1448 */
1449 public function doMagicLinks( $text ) {
1450 $prots = wfUrlProtocolsWithoutProtRel();
1451 $urlChar = self::EXT_LINK_URL_CLASS;
1452 $addr = self::EXT_LINK_ADDR;
1453 $space = self::SPACE_NOT_NL; # non-newline space
1454 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1455 $spaces = "$space++"; # possessive match of 1 or more spaces
1456 $text = preg_replace_callback(
1457 '!(?: # Start cases
1458 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1459 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1460 (\b # m[3]: Free external links
1461 (?i:$prots)
1462 ($addr$urlChar*) # m[4]: Post-protocol path
1463 ) |
1464 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1465 ([0-9]+)\b |
1466 \bISBN $spaces ( # m[6]: ISBN, capture number
1467 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1468 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1469 [0-9Xx] # check digit
1470 )\b
1471 )!xu", [ $this, 'magicLinkCallback' ], $text );
1472 return $text;
1473 }
1474
1475 /**
1476 * @throws MWException
1477 * @param array $m
1478 * @return HTML|string
1479 */
1480 public function magicLinkCallback( $m ) {
1481 if ( isset( $m[1] ) && $m[1] !== '' ) {
1482 # Skip anchor
1483 return $m[0];
1484 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1485 # Skip HTML element
1486 return $m[0];
1487 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1488 # Free external link
1489 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1490 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1491 # RFC or PMID
1492 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1493 if ( !$this->mOptions->getMagicRFCLinks() ) {
1494 return $m[0];
1495 }
1496 $keyword = 'RFC';
1497 $urlmsg = 'rfcurl';
1498 $cssClass = 'mw-magiclink-rfc';
1499 $trackingCat = 'magiclink-tracking-rfc';
1500 $id = $m[5];
1501 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1502 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1503 return $m[0];
1504 }
1505 $keyword = 'PMID';
1506 $urlmsg = 'pubmedurl';
1507 $cssClass = 'mw-magiclink-pmid';
1508 $trackingCat = 'magiclink-tracking-pmid';
1509 $id = $m[5];
1510 } else {
1511 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1512 substr( $m[0], 0, 20 ) . '"' );
1513 }
1514 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1515 $this->addTrackingCategory( $trackingCat );
1516 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1517 } elseif ( isset( $m[6] ) && $m[6] !== ''
1518 && $this->mOptions->getMagicISBNLinks()
1519 ) {
1520 # ISBN
1521 $isbn = $m[6];
1522 $space = self::SPACE_NOT_NL; # non-newline space
1523 $isbn = preg_replace( "/$space/", ' ', $isbn );
1524 $num = strtr( $isbn, [
1525 '-' => '',
1526 ' ' => '',
1527 'x' => 'X',
1528 ] );
1529 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1530 return $this->getLinkRenderer()->makeKnownLink(
1531 SpecialPage::getTitleFor( 'Booksources', $num ),
1532 "ISBN $isbn",
1533 [
1534 'class' => 'internal mw-magiclink-isbn',
1535 'title' => false // suppress title attribute
1536 ]
1537 );
1538 } else {
1539 return $m[0];
1540 }
1541 }
1542
1543 /**
1544 * Make a free external link, given a user-supplied URL
1545 *
1546 * @param string $url
1547 * @param int $numPostProto
1548 * The number of characters after the protocol.
1549 * @return string HTML
1550 * @private
1551 */
1552 public function makeFreeExternalLink( $url, $numPostProto ) {
1553 $trail = '';
1554
1555 # The characters '<' and '>' (which were escaped by
1556 # removeHTMLtags()) should not be included in
1557 # URLs, per RFC 2396.
1558 # Make &nbsp; terminate a URL as well (bug T84937)
1559 $m2 = [];
1560 if ( preg_match(
1561 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1562 $url,
1563 $m2,
1564 PREG_OFFSET_CAPTURE
1565 ) ) {
1566 $trail = substr( $url, $m2[0][1] ) . $trail;
1567 $url = substr( $url, 0, $m2[0][1] );
1568 }
1569
1570 # Move trailing punctuation to $trail
1571 $sep = ',;\.:!?';
1572 # If there is no left bracket, then consider right brackets fair game too
1573 if ( strpos( $url, '(' ) === false ) {
1574 $sep .= ')';
1575 }
1576
1577 $urlRev = strrev( $url );
1578 $numSepChars = strspn( $urlRev, $sep );
1579 # Don't break a trailing HTML entity by moving the ; into $trail
1580 # This is in hot code, so use substr_compare to avoid having to
1581 # create a new string object for the comparison
1582 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1583 # more optimization: instead of running preg_match with a $
1584 # anchor, which can be slow, do the match on the reversed
1585 # string starting at the desired offset.
1586 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1587 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1588 $numSepChars--;
1589 }
1590 }
1591 if ( $numSepChars ) {
1592 $trail = substr( $url, -$numSepChars ) . $trail;
1593 $url = substr( $url, 0, -$numSepChars );
1594 }
1595
1596 # Verify that we still have a real URL after trail removal, and
1597 # not just lone protocol
1598 if ( strlen( $trail ) >= $numPostProto ) {
1599 return $url . $trail;
1600 }
1601
1602 $url = Sanitizer::cleanUrl( $url );
1603
1604 # Is this an external image?
1605 $text = $this->maybeMakeExternalImage( $url );
1606 if ( $text === false ) {
1607 # Not an image, make a link
1608 $text = Linker::makeExternalLink( $url,
1609 $this->getConverterLanguage()->markNoConversion( $url, true ),
1610 true, 'free',
1611 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1612 # Register it in the output object...
1613 $this->mOutput->addExternalLink( $url );
1614 }
1615 return $text . $trail;
1616 }
1617
1618 /**
1619 * Parse headers and return html
1620 *
1621 * @private
1622 *
1623 * @param string $text
1624 *
1625 * @return string
1626 */
1627 public function doHeadings( $text ) {
1628 for ( $i = 6; $i >= 1; --$i ) {
1629 $h = str_repeat( '=', $i );
1630 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1631 }
1632 return $text;
1633 }
1634
1635 /**
1636 * Replace single quotes with HTML markup
1637 * @private
1638 *
1639 * @param string $text
1640 *
1641 * @return string The altered text
1642 */
1643 public function doAllQuotes( $text ) {
1644 $outtext = '';
1645 $lines = StringUtils::explode( "\n", $text );
1646 foreach ( $lines as $line ) {
1647 $outtext .= $this->doQuotes( $line ) . "\n";
1648 }
1649 $outtext = substr( $outtext, 0, -1 );
1650 return $outtext;
1651 }
1652
1653 /**
1654 * Helper function for doAllQuotes()
1655 *
1656 * @param string $text
1657 *
1658 * @return string
1659 */
1660 public function doQuotes( $text ) {
1661 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1662 $countarr = count( $arr );
1663 if ( $countarr == 1 ) {
1664 return $text;
1665 }
1666
1667 // First, do some preliminary work. This may shift some apostrophes from
1668 // being mark-up to being text. It also counts the number of occurrences
1669 // of bold and italics mark-ups.
1670 $numbold = 0;
1671 $numitalics = 0;
1672 for ( $i = 1; $i < $countarr; $i += 2 ) {
1673 $thislen = strlen( $arr[$i] );
1674 // If there are ever four apostrophes, assume the first is supposed to
1675 // be text, and the remaining three constitute mark-up for bold text.
1676 // (T15227: ''''foo'''' turns into ' ''' foo ' ''')
1677 if ( $thislen == 4 ) {
1678 $arr[$i - 1] .= "'";
1679 $arr[$i] = "'''";
1680 $thislen = 3;
1681 } elseif ( $thislen > 5 ) {
1682 // If there are more than 5 apostrophes in a row, assume they're all
1683 // text except for the last 5.
1684 // (T15227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1685 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1686 $arr[$i] = "'''''";
1687 $thislen = 5;
1688 }
1689 // Count the number of occurrences of bold and italics mark-ups.
1690 if ( $thislen == 2 ) {
1691 $numitalics++;
1692 } elseif ( $thislen == 3 ) {
1693 $numbold++;
1694 } elseif ( $thislen == 5 ) {
1695 $numitalics++;
1696 $numbold++;
1697 }
1698 }
1699
1700 // If there is an odd number of both bold and italics, it is likely
1701 // that one of the bold ones was meant to be an apostrophe followed
1702 // by italics. Which one we cannot know for certain, but it is more
1703 // likely to be one that has a single-letter word before it.
1704 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1705 $firstsingleletterword = -1;
1706 $firstmultiletterword = -1;
1707 $firstspace = -1;
1708 for ( $i = 1; $i < $countarr; $i += 2 ) {
1709 if ( strlen( $arr[$i] ) == 3 ) {
1710 $x1 = substr( $arr[$i - 1], -1 );
1711 $x2 = substr( $arr[$i - 1], -2, 1 );
1712 if ( $x1 === ' ' ) {
1713 if ( $firstspace == -1 ) {
1714 $firstspace = $i;
1715 }
1716 } elseif ( $x2 === ' ' ) {
1717 $firstsingleletterword = $i;
1718 // if $firstsingleletterword is set, we don't
1719 // look at the other options, so we can bail early.
1720 break;
1721 } else {
1722 if ( $firstmultiletterword == -1 ) {
1723 $firstmultiletterword = $i;
1724 }
1725 }
1726 }
1727 }
1728
1729 // If there is a single-letter word, use it!
1730 if ( $firstsingleletterword > -1 ) {
1731 $arr[$firstsingleletterword] = "''";
1732 $arr[$firstsingleletterword - 1] .= "'";
1733 } elseif ( $firstmultiletterword > -1 ) {
1734 // If not, but there's a multi-letter word, use that one.
1735 $arr[$firstmultiletterword] = "''";
1736 $arr[$firstmultiletterword - 1] .= "'";
1737 } elseif ( $firstspace > -1 ) {
1738 // ... otherwise use the first one that has neither.
1739 // (notice that it is possible for all three to be -1 if, for example,
1740 // there is only one pentuple-apostrophe in the line)
1741 $arr[$firstspace] = "''";
1742 $arr[$firstspace - 1] .= "'";
1743 }
1744 }
1745
1746 // Now let's actually convert our apostrophic mush to HTML!
1747 $output = '';
1748 $buffer = '';
1749 $state = '';
1750 $i = 0;
1751 foreach ( $arr as $r ) {
1752 if ( ( $i % 2 ) == 0 ) {
1753 if ( $state === 'both' ) {
1754 $buffer .= $r;
1755 } else {
1756 $output .= $r;
1757 }
1758 } else {
1759 $thislen = strlen( $r );
1760 if ( $thislen == 2 ) {
1761 if ( $state === 'i' ) {
1762 $output .= '</i>';
1763 $state = '';
1764 } elseif ( $state === 'bi' ) {
1765 $output .= '</i>';
1766 $state = 'b';
1767 } elseif ( $state === 'ib' ) {
1768 $output .= '</b></i><b>';
1769 $state = 'b';
1770 } elseif ( $state === 'both' ) {
1771 $output .= '<b><i>' . $buffer . '</i>';
1772 $state = 'b';
1773 } else { // $state can be 'b' or ''
1774 $output .= '<i>';
1775 $state .= 'i';
1776 }
1777 } elseif ( $thislen == 3 ) {
1778 if ( $state === 'b' ) {
1779 $output .= '</b>';
1780 $state = '';
1781 } elseif ( $state === 'bi' ) {
1782 $output .= '</i></b><i>';
1783 $state = 'i';
1784 } elseif ( $state === 'ib' ) {
1785 $output .= '</b>';
1786 $state = 'i';
1787 } elseif ( $state === 'both' ) {
1788 $output .= '<i><b>' . $buffer . '</b>';
1789 $state = 'i';
1790 } else { // $state can be 'i' or ''
1791 $output .= '<b>';
1792 $state .= 'b';
1793 }
1794 } elseif ( $thislen == 5 ) {
1795 if ( $state === 'b' ) {
1796 $output .= '</b><i>';
1797 $state = 'i';
1798 } elseif ( $state === 'i' ) {
1799 $output .= '</i><b>';
1800 $state = 'b';
1801 } elseif ( $state === 'bi' ) {
1802 $output .= '</i></b>';
1803 $state = '';
1804 } elseif ( $state === 'ib' ) {
1805 $output .= '</b></i>';
1806 $state = '';
1807 } elseif ( $state === 'both' ) {
1808 $output .= '<i><b>' . $buffer . '</b></i>';
1809 $state = '';
1810 } else { // ($state == '')
1811 $buffer = '';
1812 $state = 'both';
1813 }
1814 }
1815 }
1816 $i++;
1817 }
1818 // Now close all remaining tags. Notice that the order is important.
1819 if ( $state === 'b' || $state === 'ib' ) {
1820 $output .= '</b>';
1821 }
1822 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1823 $output .= '</i>';
1824 }
1825 if ( $state === 'bi' ) {
1826 $output .= '</b>';
1827 }
1828 // There might be lonely ''''', so make sure we have a buffer
1829 if ( $state === 'both' && $buffer ) {
1830 $output .= '<b><i>' . $buffer . '</i></b>';
1831 }
1832 return $output;
1833 }
1834
1835 /**
1836 * Replace external links (REL)
1837 *
1838 * Note: this is all very hackish and the order of execution matters a lot.
1839 * Make sure to run tests/parser/parserTests.php if you change this code.
1840 *
1841 * @private
1842 *
1843 * @param string $text
1844 *
1845 * @throws MWException
1846 * @return string
1847 */
1848 public function replaceExternalLinks( $text ) {
1849
1850 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1851 if ( $bits === false ) {
1852 throw new MWException( "PCRE needs to be compiled with "
1853 . "--enable-unicode-properties in order for MediaWiki to function" );
1854 }
1855 $s = array_shift( $bits );
1856
1857 $i = 0;
1858 while ( $i < count( $bits ) ) {
1859 $url = $bits[$i++];
1860 $i++; // protocol
1861 $text = $bits[$i++];
1862 $trail = $bits[$i++];
1863
1864 # The characters '<' and '>' (which were escaped by
1865 # removeHTMLtags()) should not be included in
1866 # URLs, per RFC 2396.
1867 $m2 = [];
1868 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1869 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1870 $url = substr( $url, 0, $m2[0][1] );
1871 }
1872
1873 # If the link text is an image URL, replace it with an <img> tag
1874 # This happened by accident in the original parser, but some people used it extensively
1875 $img = $this->maybeMakeExternalImage( $text );
1876 if ( $img !== false ) {
1877 $text = $img;
1878 }
1879
1880 $dtrail = '';
1881
1882 # Set linktype for CSS - if URL==text, link is essentially free
1883 $linktype = ( $text === $url ) ? 'free' : 'text';
1884
1885 # No link text, e.g. [http://domain.tld/some.link]
1886 if ( $text == '' ) {
1887 # Autonumber
1888 $langObj = $this->getTargetLanguage();
1889 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1890 $linktype = 'autonumber';
1891 } else {
1892 # Have link text, e.g. [http://domain.tld/some.link text]s
1893 # Check for trail
1894 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1895 }
1896
1897 $text = $this->getConverterLanguage()->markNoConversion( $text );
1898
1899 $url = Sanitizer::cleanUrl( $url );
1900
1901 # Use the encoded URL
1902 # This means that users can paste URLs directly into the text
1903 # Funny characters like ö aren't valid in URLs anyway
1904 # This was changed in August 2004
1905 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1906 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1907
1908 # Register link in the output object.
1909 $this->mOutput->addExternalLink( $url );
1910 }
1911
1912 return $s;
1913 }
1914
1915 /**
1916 * Get the rel attribute for a particular external link.
1917 *
1918 * @since 1.21
1919 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1920 * nofollow if appropriate
1921 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1922 * @return string|null Rel attribute for $url
1923 */
1924 public static function getExternalLinkRel( $url = false, $title = null ) {
1925 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1926 $ns = $title ? $title->getNamespace() : false;
1927 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1928 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1929 ) {
1930 return 'nofollow';
1931 }
1932 return null;
1933 }
1934
1935 /**
1936 * Get an associative array of additional HTML attributes appropriate for a
1937 * particular external link. This currently may include rel => nofollow
1938 * (depending on configuration, namespace, and the URL's domain) and/or a
1939 * target attribute (depending on configuration).
1940 *
1941 * @param string $url URL to extract the domain from for rel =>
1942 * nofollow if appropriate
1943 * @return array Associative array of HTML attributes
1944 */
1945 public function getExternalLinkAttribs( $url ) {
1946 $attribs = [];
1947 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1948
1949 $target = $this->mOptions->getExternalLinkTarget();
1950 if ( $target ) {
1951 $attribs['target'] = $target;
1952 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1953 // T133507. New windows can navigate parent cross-origin.
1954 // Including noreferrer due to lacking browser
1955 // support of noopener. Eventually noreferrer should be removed.
1956 if ( $rel !== '' ) {
1957 $rel .= ' ';
1958 }
1959 $rel .= 'noreferrer noopener';
1960 }
1961 }
1962 $attribs['rel'] = $rel;
1963 return $attribs;
1964 }
1965
1966 /**
1967 * Replace unusual escape codes in a URL with their equivalent characters
1968 *
1969 * This generally follows the syntax defined in RFC 3986, with special
1970 * consideration for HTTP query strings.
1971 *
1972 * @param string $url
1973 * @return string
1974 */
1975 public static function normalizeLinkUrl( $url ) {
1976 # First, make sure unsafe characters are encoded
1977 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1978 function ( $m ) {
1979 return rawurlencode( $m[0] );
1980 },
1981 $url
1982 );
1983
1984 $ret = '';
1985 $end = strlen( $url );
1986
1987 # Fragment part - 'fragment'
1988 $start = strpos( $url, '#' );
1989 if ( $start !== false && $start < $end ) {
1990 $ret = self::normalizeUrlComponent(
1991 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1992 $end = $start;
1993 }
1994
1995 # Query part - 'query' minus &=+;
1996 $start = strpos( $url, '?' );
1997 if ( $start !== false && $start < $end ) {
1998 $ret = self::normalizeUrlComponent(
1999 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
2000 $end = $start;
2001 }
2002
2003 # Scheme and path part - 'pchar'
2004 # (we assume no userinfo or encoded colons in the host)
2005 $ret = self::normalizeUrlComponent(
2006 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
2007
2008 return $ret;
2009 }
2010
2011 private static function normalizeUrlComponent( $component, $unsafe ) {
2012 $callback = function ( $matches ) use ( $unsafe ) {
2013 $char = urldecode( $matches[0] );
2014 $ord = ord( $char );
2015 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
2016 # Unescape it
2017 return $char;
2018 } else {
2019 # Leave it escaped, but use uppercase for a-f
2020 return strtoupper( $matches[0] );
2021 }
2022 };
2023 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
2024 }
2025
2026 /**
2027 * make an image if it's allowed, either through the global
2028 * option, through the exception, or through the on-wiki whitelist
2029 *
2030 * @param string $url
2031 *
2032 * @return string
2033 */
2034 private function maybeMakeExternalImage( $url ) {
2035 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2036 $imagesexception = !empty( $imagesfrom );
2037 $text = false;
2038 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2039 if ( $imagesexception && is_array( $imagesfrom ) ) {
2040 $imagematch = false;
2041 foreach ( $imagesfrom as $match ) {
2042 if ( strpos( $url, $match ) === 0 ) {
2043 $imagematch = true;
2044 break;
2045 }
2046 }
2047 } elseif ( $imagesexception ) {
2048 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2049 } else {
2050 $imagematch = false;
2051 }
2052
2053 if ( $this->mOptions->getAllowExternalImages()
2054 || ( $imagesexception && $imagematch )
2055 ) {
2056 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2057 # Image found
2058 $text = Linker::makeExternalImage( $url );
2059 }
2060 }
2061 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2062 && preg_match( self::EXT_IMAGE_REGEX, $url )
2063 ) {
2064 $whitelist = explode(
2065 "\n",
2066 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2067 );
2068
2069 foreach ( $whitelist as $entry ) {
2070 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2071 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2072 continue;
2073 }
2074 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2075 # Image matches a whitelist entry
2076 $text = Linker::makeExternalImage( $url );
2077 break;
2078 }
2079 }
2080 }
2081 return $text;
2082 }
2083
2084 /**
2085 * Process [[ ]] wikilinks
2086 *
2087 * @param string $s
2088 *
2089 * @return string Processed text
2090 *
2091 * @private
2092 */
2093 public function replaceInternalLinks( $s ) {
2094 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2095 return $s;
2096 }
2097
2098 /**
2099 * Process [[ ]] wikilinks (RIL)
2100 * @param string $s
2101 * @throws MWException
2102 * @return LinkHolderArray
2103 *
2104 * @private
2105 */
2106 public function replaceInternalLinks2( &$s ) {
2107 global $wgExtraInterlanguageLinkPrefixes;
2108
2109 static $tc = false, $e1, $e1_img;
2110 # the % is needed to support urlencoded titles as well
2111 if ( !$tc ) {
2112 $tc = Title::legalChars() . '#%';
2113 # Match a link having the form [[namespace:link|alternate]]trail
2114 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2115 # Match cases where there is no "]]", which might still be images
2116 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2117 }
2118
2119 $holders = new LinkHolderArray( $this );
2120
2121 # split the entire text string on occurrences of [[
2122 $a = StringUtils::explode( '[[', ' ' . $s );
2123 # get the first element (all text up to first [[), and remove the space we added
2124 $s = $a->current();
2125 $a->next();
2126 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2127 $s = substr( $s, 1 );
2128
2129 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2130 $e2 = null;
2131 if ( $useLinkPrefixExtension ) {
2132 # Match the end of a line for a word that's not followed by whitespace,
2133 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2134 global $wgContLang;
2135 $charset = $wgContLang->linkPrefixCharset();
2136 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2137 }
2138
2139 if ( is_null( $this->mTitle ) ) {
2140 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2141 }
2142 $nottalk = !$this->mTitle->isTalkPage();
2143
2144 if ( $useLinkPrefixExtension ) {
2145 $m = [];
2146 if ( preg_match( $e2, $s, $m ) ) {
2147 $first_prefix = $m[2];
2148 } else {
2149 $first_prefix = false;
2150 }
2151 } else {
2152 $prefix = '';
2153 }
2154
2155 $useSubpages = $this->areSubpagesAllowed();
2156
2157 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2158 # Loop for each link
2159 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2160 // @codingStandardsIgnoreEnd
2161
2162 # Check for excessive memory usage
2163 if ( $holders->isBig() ) {
2164 # Too big
2165 # Do the existence check, replace the link holders and clear the array
2166 $holders->replace( $s );
2167 $holders->clear();
2168 }
2169
2170 if ( $useLinkPrefixExtension ) {
2171 if ( preg_match( $e2, $s, $m ) ) {
2172 $prefix = $m[2];
2173 $s = $m[1];
2174 } else {
2175 $prefix = '';
2176 }
2177 # first link
2178 if ( $first_prefix ) {
2179 $prefix = $first_prefix;
2180 $first_prefix = false;
2181 }
2182 }
2183
2184 $might_be_img = false;
2185
2186 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2187 $text = $m[2];
2188 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2189 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2190 # the real problem is with the $e1 regex
2191 # See T1500.
2192 # Still some problems for cases where the ] is meant to be outside punctuation,
2193 # and no image is in sight. See T4095.
2194 if ( $text !== ''
2195 && substr( $m[3], 0, 1 ) === ']'
2196 && strpos( $text, '[' ) !== false
2197 ) {
2198 $text .= ']'; # so that replaceExternalLinks($text) works later
2199 $m[3] = substr( $m[3], 1 );
2200 }
2201 # fix up urlencoded title texts
2202 if ( strpos( $m[1], '%' ) !== false ) {
2203 # Should anchors '#' also be rejected?
2204 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2205 }
2206 $trail = $m[3];
2207 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2208 # Invalid, but might be an image with a link in its caption
2209 $might_be_img = true;
2210 $text = $m[2];
2211 if ( strpos( $m[1], '%' ) !== false ) {
2212 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2213 }
2214 $trail = "";
2215 } else { # Invalid form; output directly
2216 $s .= $prefix . '[[' . $line;
2217 continue;
2218 }
2219
2220 $origLink = ltrim( $m[1], ' ' );
2221
2222 # Don't allow internal links to pages containing
2223 # PROTO: where PROTO is a valid URL protocol; these
2224 # should be external links.
2225 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2226 $s .= $prefix . '[[' . $line;
2227 continue;
2228 }
2229
2230 # Make subpage if necessary
2231 if ( $useSubpages ) {
2232 $link = $this->maybeDoSubpageLink( $origLink, $text );
2233 } else {
2234 $link = $origLink;
2235 }
2236
2237 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2238 if ( !$noforce ) {
2239 # Strip off leading ':'
2240 $link = substr( $link, 1 );
2241 }
2242
2243 $unstrip = $this->mStripState->unstripNoWiki( $link );
2244 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2245 if ( $nt === null ) {
2246 $s .= $prefix . '[[' . $line;
2247 continue;
2248 }
2249
2250 $ns = $nt->getNamespace();
2251 $iw = $nt->getInterwiki();
2252
2253 if ( $might_be_img ) { # if this is actually an invalid link
2254 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2255 $found = false;
2256 while ( true ) {
2257 # look at the next 'line' to see if we can close it there
2258 $a->next();
2259 $next_line = $a->current();
2260 if ( $next_line === false || $next_line === null ) {
2261 break;
2262 }
2263 $m = explode( ']]', $next_line, 3 );
2264 if ( count( $m ) == 3 ) {
2265 # the first ]] closes the inner link, the second the image
2266 $found = true;
2267 $text .= "[[{$m[0]}]]{$m[1]}";
2268 $trail = $m[2];
2269 break;
2270 } elseif ( count( $m ) == 2 ) {
2271 # if there's exactly one ]] that's fine, we'll keep looking
2272 $text .= "[[{$m[0]}]]{$m[1]}";
2273 } else {
2274 # if $next_line is invalid too, we need look no further
2275 $text .= '[[' . $next_line;
2276 break;
2277 }
2278 }
2279 if ( !$found ) {
2280 # we couldn't find the end of this imageLink, so output it raw
2281 # but don't ignore what might be perfectly normal links in the text we've examined
2282 $holders->merge( $this->replaceInternalLinks2( $text ) );
2283 $s .= "{$prefix}[[$link|$text";
2284 # note: no $trail, because without an end, there *is* no trail
2285 continue;
2286 }
2287 } else { # it's not an image, so output it raw
2288 $s .= "{$prefix}[[$link|$text";
2289 # note: no $trail, because without an end, there *is* no trail
2290 continue;
2291 }
2292 }
2293
2294 $wasblank = ( $text == '' );
2295 if ( $wasblank ) {
2296 $text = $link;
2297 } else {
2298 # T6598 madness. Handle the quotes only if they come from the alternate part
2299 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2300 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2301 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2302 $text = $this->doQuotes( $text );
2303 }
2304
2305 # Link not escaped by : , create the various objects
2306 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2307 # Interwikis
2308 if (
2309 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2310 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2311 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2312 )
2313 ) {
2314 # T26502: filter duplicates
2315 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2316 $this->mLangLinkLanguages[$iw] = true;
2317 $this->mOutput->addLanguageLink( $nt->getFullText() );
2318 }
2319
2320 $s = rtrim( $s . $prefix );
2321 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2322 continue;
2323 }
2324
2325 if ( $ns == NS_FILE ) {
2326 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2327 if ( $wasblank ) {
2328 # if no parameters were passed, $text
2329 # becomes something like "File:Foo.png",
2330 # which we don't want to pass on to the
2331 # image generator
2332 $text = '';
2333 } else {
2334 # recursively parse links inside the image caption
2335 # actually, this will parse them in any other parameters, too,
2336 # but it might be hard to fix that, and it doesn't matter ATM
2337 $text = $this->replaceExternalLinks( $text );
2338 $holders->merge( $this->replaceInternalLinks2( $text ) );
2339 }
2340 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2341 $s .= $prefix . $this->armorLinks(
2342 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2343 continue;
2344 }
2345 } elseif ( $ns == NS_CATEGORY ) {
2346 $s = rtrim( $s . "\n" ); # T2087
2347
2348 if ( $wasblank ) {
2349 $sortkey = $this->getDefaultSort();
2350 } else {
2351 $sortkey = $text;
2352 }
2353 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2354 $sortkey = str_replace( "\n", '', $sortkey );
2355 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2356 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2357
2358 /**
2359 * Strip the whitespace Category links produce, see T2087
2360 */
2361 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2362
2363 continue;
2364 }
2365 }
2366
2367 # Self-link checking. For some languages, variants of the title are checked in
2368 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2369 # for linking to a different variant.
2370 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2371 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2372 continue;
2373 }
2374
2375 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2376 # @todo FIXME: Should do batch file existence checks, see comment below
2377 if ( $ns == NS_MEDIA ) {
2378 # Give extensions a chance to select the file revision for us
2379 $options = [];
2380 $descQuery = false;
2381 Hooks::run( 'BeforeParserFetchFileAndTitle',
2382 [ $this, $nt, &$options, &$descQuery ] );
2383 # Fetch and register the file (file title may be different via hooks)
2384 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2385 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2386 $s .= $prefix . $this->armorLinks(
2387 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2388 continue;
2389 }
2390
2391 # Some titles, such as valid special pages or files in foreign repos, should
2392 # be shown as bluelinks even though they're not included in the page table
2393 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2394 # batch file existence checks for NS_FILE and NS_MEDIA
2395 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2396 $this->mOutput->addLink( $nt );
2397 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2398 } else {
2399 # Links will be added to the output link list after checking
2400 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2401 }
2402 }
2403 return $holders;
2404 }
2405
2406 /**
2407 * Render a forced-blue link inline; protect against double expansion of
2408 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2409 * Since this little disaster has to split off the trail text to avoid
2410 * breaking URLs in the following text without breaking trails on the
2411 * wiki links, it's been made into a horrible function.
2412 *
2413 * @param Title $nt
2414 * @param string $text
2415 * @param string $trail
2416 * @param string $prefix
2417 * @return string HTML-wikitext mix oh yuck
2418 */
2419 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2420 list( $inside, $trail ) = Linker::splitTrail( $trail );
2421
2422 if ( $text == '' ) {
2423 $text = htmlspecialchars( $nt->getPrefixedText() );
2424 }
2425
2426 $link = $this->getLinkRenderer()->makeKnownLink(
2427 $nt, new HtmlArmor( "$prefix$text$inside" )
2428 );
2429
2430 return $this->armorLinks( $link ) . $trail;
2431 }
2432
2433 /**
2434 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2435 * going to go through further parsing steps before inline URL expansion.
2436 *
2437 * Not needed quite as much as it used to be since free links are a bit
2438 * more sensible these days. But bracketed links are still an issue.
2439 *
2440 * @param string $text More-or-less HTML
2441 * @return string Less-or-more HTML with NOPARSE bits
2442 */
2443 public function armorLinks( $text ) {
2444 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2445 self::MARKER_PREFIX . "NOPARSE$1", $text );
2446 }
2447
2448 /**
2449 * Return true if subpage links should be expanded on this page.
2450 * @return bool
2451 */
2452 public function areSubpagesAllowed() {
2453 # Some namespaces don't allow subpages
2454 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2455 }
2456
2457 /**
2458 * Handle link to subpage if necessary
2459 *
2460 * @param string $target The source of the link
2461 * @param string &$text The link text, modified as necessary
2462 * @return string The full name of the link
2463 * @private
2464 */
2465 public function maybeDoSubpageLink( $target, &$text ) {
2466 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2467 }
2468
2469 /**
2470 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2471 *
2472 * @param string $text
2473 * @param bool $linestart Whether or not this is at the start of a line.
2474 * @private
2475 * @return string The lists rendered as HTML
2476 */
2477 public function doBlockLevels( $text, $linestart ) {
2478 return BlockLevelPass::doBlockLevels( $text, $linestart );
2479 }
2480
2481 /**
2482 * Return value of a magic variable (like PAGENAME)
2483 *
2484 * @private
2485 *
2486 * @param string $index Magic variable identifier as mapped in MagicWord::$mVariableIDs
2487 * @param bool|PPFrame $frame
2488 *
2489 * @throws MWException
2490 * @return string
2491 */
2492 public function getVariableValue( $index, $frame = false ) {
2493 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2494 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2495
2496 if ( is_null( $this->mTitle ) ) {
2497 // If no title set, bad things are going to happen
2498 // later. Title should always be set since this
2499 // should only be called in the middle of a parse
2500 // operation (but the unit-tests do funky stuff)
2501 throw new MWException( __METHOD__ . ' Should only be '
2502 . ' called while parsing (no title set)' );
2503 }
2504
2505 // Avoid PHP 7.1 warning from passing $this by reference
2506 $parser = $this;
2507
2508 /**
2509 * Some of these require message or data lookups and can be
2510 * expensive to check many times.
2511 */
2512 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$parser, &$this->mVarCache ] ) ) {
2513 if ( isset( $this->mVarCache[$index] ) ) {
2514 return $this->mVarCache[$index];
2515 }
2516 }
2517
2518 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2519 Hooks::run( 'ParserGetVariableValueTs', [ &$parser, &$ts ] );
2520
2521 $pageLang = $this->getFunctionLang();
2522
2523 switch ( $index ) {
2524 case '!':
2525 $value = '|';
2526 break;
2527 case 'currentmonth':
2528 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2529 break;
2530 case 'currentmonth1':
2531 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2532 break;
2533 case 'currentmonthname':
2534 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2535 break;
2536 case 'currentmonthnamegen':
2537 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2538 break;
2539 case 'currentmonthabbrev':
2540 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2541 break;
2542 case 'currentday':
2543 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2544 break;
2545 case 'currentday2':
2546 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2547 break;
2548 case 'localmonth':
2549 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2550 break;
2551 case 'localmonth1':
2552 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2553 break;
2554 case 'localmonthname':
2555 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2556 break;
2557 case 'localmonthnamegen':
2558 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2559 break;
2560 case 'localmonthabbrev':
2561 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2562 break;
2563 case 'localday':
2564 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2565 break;
2566 case 'localday2':
2567 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2568 break;
2569 case 'pagename':
2570 $value = wfEscapeWikiText( $this->mTitle->getText() );
2571 break;
2572 case 'pagenamee':
2573 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2574 break;
2575 case 'fullpagename':
2576 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2577 break;
2578 case 'fullpagenamee':
2579 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2580 break;
2581 case 'subpagename':
2582 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2583 break;
2584 case 'subpagenamee':
2585 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2586 break;
2587 case 'rootpagename':
2588 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2589 break;
2590 case 'rootpagenamee':
2591 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2592 ' ',
2593 '_',
2594 $this->mTitle->getRootText()
2595 ) ) );
2596 break;
2597 case 'basepagename':
2598 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2599 break;
2600 case 'basepagenamee':
2601 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2602 ' ',
2603 '_',
2604 $this->mTitle->getBaseText()
2605 ) ) );
2606 break;
2607 case 'talkpagename':
2608 if ( $this->mTitle->canTalk() ) {
2609 $talkPage = $this->mTitle->getTalkPage();
2610 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2611 } else {
2612 $value = '';
2613 }
2614 break;
2615 case 'talkpagenamee':
2616 if ( $this->mTitle->canTalk() ) {
2617 $talkPage = $this->mTitle->getTalkPage();
2618 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2619 } else {
2620 $value = '';
2621 }
2622 break;
2623 case 'subjectpagename':
2624 $subjPage = $this->mTitle->getSubjectPage();
2625 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2626 break;
2627 case 'subjectpagenamee':
2628 $subjPage = $this->mTitle->getSubjectPage();
2629 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2630 break;
2631 case 'pageid': // requested in T25427
2632 $pageid = $this->getTitle()->getArticleID();
2633 if ( $pageid == 0 ) {
2634 # 0 means the page doesn't exist in the database,
2635 # which means the user is previewing a new page.
2636 # The vary-revision flag must be set, because the magic word
2637 # will have a different value once the page is saved.
2638 $this->mOutput->setFlag( 'vary-revision' );
2639 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2640 }
2641 $value = $pageid ? $pageid : null;
2642 break;
2643 case 'revisionid':
2644 # Let the edit saving system know we should parse the page
2645 # *after* a revision ID has been assigned.
2646 $this->mOutput->setFlag( 'vary-revision-id' );
2647 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2648 $value = $this->mRevisionId;
2649 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2650 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2651 $this->mOutput->setSpeculativeRevIdUsed( $value );
2652 }
2653 break;
2654 case 'revisionday':
2655 # Let the edit saving system know we should parse the page
2656 # *after* a revision ID has been assigned. This is for null edits.
2657 $this->mOutput->setFlag( 'vary-revision' );
2658 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2659 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2660 break;
2661 case 'revisionday2':
2662 # Let the edit saving system know we should parse the page
2663 # *after* a revision ID has been assigned. This is for null edits.
2664 $this->mOutput->setFlag( 'vary-revision' );
2665 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2666 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2667 break;
2668 case 'revisionmonth':
2669 # Let the edit saving system know we should parse the page
2670 # *after* a revision ID has been assigned. This is for null edits.
2671 $this->mOutput->setFlag( 'vary-revision' );
2672 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2673 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2674 break;
2675 case 'revisionmonth1':
2676 # Let the edit saving system know we should parse the page
2677 # *after* a revision ID has been assigned. This is for null edits.
2678 $this->mOutput->setFlag( 'vary-revision' );
2679 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2680 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2681 break;
2682 case 'revisionyear':
2683 # Let the edit saving system know we should parse the page
2684 # *after* a revision ID has been assigned. This is for null edits.
2685 $this->mOutput->setFlag( 'vary-revision' );
2686 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2687 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2688 break;
2689 case 'revisiontimestamp':
2690 # Let the edit saving system know we should parse the page
2691 # *after* a revision ID has been assigned. This is for null edits.
2692 $this->mOutput->setFlag( 'vary-revision' );
2693 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2694 $value = $this->getRevisionTimestamp();
2695 break;
2696 case 'revisionuser':
2697 # Let the edit saving system know we should parse the page
2698 # *after* a revision ID has been assigned for null edits.
2699 $this->mOutput->setFlag( 'vary-user' );
2700 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2701 $value = $this->getRevisionUser();
2702 break;
2703 case 'revisionsize':
2704 $value = $this->getRevisionSize();
2705 break;
2706 case 'namespace':
2707 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2708 break;
2709 case 'namespacee':
2710 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2711 break;
2712 case 'namespacenumber':
2713 $value = $this->mTitle->getNamespace();
2714 break;
2715 case 'talkspace':
2716 $value = $this->mTitle->canTalk()
2717 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2718 : '';
2719 break;
2720 case 'talkspacee':
2721 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2722 break;
2723 case 'subjectspace':
2724 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2725 break;
2726 case 'subjectspacee':
2727 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2728 break;
2729 case 'currentdayname':
2730 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2731 break;
2732 case 'currentyear':
2733 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2734 break;
2735 case 'currenttime':
2736 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2737 break;
2738 case 'currenthour':
2739 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2740 break;
2741 case 'currentweek':
2742 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2743 # int to remove the padding
2744 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2745 break;
2746 case 'currentdow':
2747 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2748 break;
2749 case 'localdayname':
2750 $value = $pageLang->getWeekdayName(
2751 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2752 );
2753 break;
2754 case 'localyear':
2755 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2756 break;
2757 case 'localtime':
2758 $value = $pageLang->time(
2759 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2760 false,
2761 false
2762 );
2763 break;
2764 case 'localhour':
2765 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2766 break;
2767 case 'localweek':
2768 # @bug T6594 PHP5 has it zero padded, PHP4 does not, cast to
2769 # int to remove the padding
2770 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2771 break;
2772 case 'localdow':
2773 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2774 break;
2775 case 'numberofarticles':
2776 $value = $pageLang->formatNum( SiteStats::articles() );
2777 break;
2778 case 'numberoffiles':
2779 $value = $pageLang->formatNum( SiteStats::images() );
2780 break;
2781 case 'numberofusers':
2782 $value = $pageLang->formatNum( SiteStats::users() );
2783 break;
2784 case 'numberofactiveusers':
2785 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2786 break;
2787 case 'numberofpages':
2788 $value = $pageLang->formatNum( SiteStats::pages() );
2789 break;
2790 case 'numberofadmins':
2791 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2792 break;
2793 case 'numberofedits':
2794 $value = $pageLang->formatNum( SiteStats::edits() );
2795 break;
2796 case 'currenttimestamp':
2797 $value = wfTimestamp( TS_MW, $ts );
2798 break;
2799 case 'localtimestamp':
2800 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2801 break;
2802 case 'currentversion':
2803 $value = SpecialVersion::getVersion();
2804 break;
2805 case 'articlepath':
2806 return $wgArticlePath;
2807 case 'sitename':
2808 return $wgSitename;
2809 case 'server':
2810 return $wgServer;
2811 case 'servername':
2812 return $wgServerName;
2813 case 'scriptpath':
2814 return $wgScriptPath;
2815 case 'stylepath':
2816 return $wgStylePath;
2817 case 'directionmark':
2818 return $pageLang->getDirMark();
2819 case 'contentlanguage':
2820 global $wgLanguageCode;
2821 return $wgLanguageCode;
2822 case 'pagelanguage':
2823 $value = $pageLang->getCode();
2824 break;
2825 case 'cascadingsources':
2826 $value = CoreParserFunctions::cascadingsources( $this );
2827 break;
2828 default:
2829 $ret = null;
2830 Hooks::run(
2831 'ParserGetVariableValueSwitch',
2832 [ &$parser, &$this->mVarCache, &$index, &$ret, &$frame ]
2833 );
2834
2835 return $ret;
2836 }
2837
2838 if ( $index ) {
2839 $this->mVarCache[$index] = $value;
2840 }
2841
2842 return $value;
2843 }
2844
2845 /**
2846 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2847 *
2848 * @private
2849 */
2850 public function initialiseVariables() {
2851 $variableIDs = MagicWord::getVariableIDs();
2852 $substIDs = MagicWord::getSubstIDs();
2853
2854 $this->mVariables = new MagicWordArray( $variableIDs );
2855 $this->mSubstWords = new MagicWordArray( $substIDs );
2856 }
2857
2858 /**
2859 * Preprocess some wikitext and return the document tree.
2860 * This is the ghost of replace_variables().
2861 *
2862 * @param string $text The text to parse
2863 * @param int $flags Bitwise combination of:
2864 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2865 * included. Default is to assume a direct page view.
2866 *
2867 * The generated DOM tree must depend only on the input text and the flags.
2868 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of T6899.
2869 *
2870 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2871 * change in the DOM tree for a given text, must be passed through the section identifier
2872 * in the section edit link and thus back to extractSections().
2873 *
2874 * The output of this function is currently only cached in process memory, but a persistent
2875 * cache may be implemented at a later date which takes further advantage of these strict
2876 * dependency requirements.
2877 *
2878 * @return PPNode
2879 */
2880 public function preprocessToDom( $text, $flags = 0 ) {
2881 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2882 return $dom;
2883 }
2884
2885 /**
2886 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2887 *
2888 * @param string $s
2889 *
2890 * @return array
2891 */
2892 public static function splitWhitespace( $s ) {
2893 $ltrimmed = ltrim( $s );
2894 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2895 $trimmed = rtrim( $ltrimmed );
2896 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2897 if ( $diff > 0 ) {
2898 $w2 = substr( $ltrimmed, -$diff );
2899 } else {
2900 $w2 = '';
2901 }
2902 return [ $w1, $trimmed, $w2 ];
2903 }
2904
2905 /**
2906 * Replace magic variables, templates, and template arguments
2907 * with the appropriate text. Templates are substituted recursively,
2908 * taking care to avoid infinite loops.
2909 *
2910 * Note that the substitution depends on value of $mOutputType:
2911 * self::OT_WIKI: only {{subst:}} templates
2912 * self::OT_PREPROCESS: templates but not extension tags
2913 * self::OT_HTML: all templates and extension tags
2914 *
2915 * @param string $text The text to transform
2916 * @param bool|PPFrame $frame Object describing the arguments passed to the
2917 * template. Arguments may also be provided as an associative array, as
2918 * was the usual case before MW1.12. Providing arguments this way may be
2919 * useful for extensions wishing to perform variable replacement
2920 * explicitly.
2921 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2922 * double-brace expansion.
2923 * @return string
2924 */
2925 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2926 # Is there any text? Also, Prevent too big inclusions!
2927 $textSize = strlen( $text );
2928 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2929 return $text;
2930 }
2931
2932 if ( $frame === false ) {
2933 $frame = $this->getPreprocessor()->newFrame();
2934 } elseif ( !( $frame instanceof PPFrame ) ) {
2935 wfDebug( __METHOD__ . " called using plain parameters instead of "
2936 . "a PPFrame instance. Creating custom frame.\n" );
2937 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2938 }
2939
2940 $dom = $this->preprocessToDom( $text );
2941 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2942 $text = $frame->expand( $dom, $flags );
2943
2944 return $text;
2945 }
2946
2947 /**
2948 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2949 *
2950 * @param array $args
2951 *
2952 * @return array
2953 */
2954 public static function createAssocArgs( $args ) {
2955 $assocArgs = [];
2956 $index = 1;
2957 foreach ( $args as $arg ) {
2958 $eqpos = strpos( $arg, '=' );
2959 if ( $eqpos === false ) {
2960 $assocArgs[$index++] = $arg;
2961 } else {
2962 $name = trim( substr( $arg, 0, $eqpos ) );
2963 $value = trim( substr( $arg, $eqpos + 1 ) );
2964 if ( $value === false ) {
2965 $value = '';
2966 }
2967 if ( $name !== false ) {
2968 $assocArgs[$name] = $value;
2969 }
2970 }
2971 }
2972
2973 return $assocArgs;
2974 }
2975
2976 /**
2977 * Warn the user when a parser limitation is reached
2978 * Will warn at most once the user per limitation type
2979 *
2980 * The results are shown during preview and run through the Parser (See EditPage.php)
2981 *
2982 * @param string $limitationType Should be one of:
2983 * 'expensive-parserfunction' (corresponding messages:
2984 * 'expensive-parserfunction-warning',
2985 * 'expensive-parserfunction-category')
2986 * 'post-expand-template-argument' (corresponding messages:
2987 * 'post-expand-template-argument-warning',
2988 * 'post-expand-template-argument-category')
2989 * 'post-expand-template-inclusion' (corresponding messages:
2990 * 'post-expand-template-inclusion-warning',
2991 * 'post-expand-template-inclusion-category')
2992 * 'node-count-exceeded' (corresponding messages:
2993 * 'node-count-exceeded-warning',
2994 * 'node-count-exceeded-category')
2995 * 'expansion-depth-exceeded' (corresponding messages:
2996 * 'expansion-depth-exceeded-warning',
2997 * 'expansion-depth-exceeded-category')
2998 * @param string|int|null $current Current value
2999 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3000 * exceeded, provide the values (optional)
3001 */
3002 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3003 # does no harm if $current and $max are present but are unnecessary for the message
3004 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3005 # only during preview, and that would split the parser cache unnecessarily.
3006 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3007 ->text();
3008 $this->mOutput->addWarning( $warning );
3009 $this->addTrackingCategory( "$limitationType-category" );
3010 }
3011
3012 /**
3013 * Return the text of a template, after recursively
3014 * replacing any variables or templates within the template.
3015 *
3016 * @param array $piece The parts of the template
3017 * $piece['title']: the title, i.e. the part before the |
3018 * $piece['parts']: the parameter array
3019 * $piece['lineStart']: whether the brace was at the start of a line
3020 * @param PPFrame $frame The current frame, contains template arguments
3021 * @throws Exception
3022 * @return string The text of the template
3023 */
3024 public function braceSubstitution( $piece, $frame ) {
3025
3026 // Flags
3027
3028 // $text has been filled
3029 $found = false;
3030 // wiki markup in $text should be escaped
3031 $nowiki = false;
3032 // $text is HTML, armour it against wikitext transformation
3033 $isHTML = false;
3034 // Force interwiki transclusion to be done in raw mode not rendered
3035 $forceRawInterwiki = false;
3036 // $text is a DOM node needing expansion in a child frame
3037 $isChildObj = false;
3038 // $text is a DOM node needing expansion in the current frame
3039 $isLocalObj = false;
3040
3041 # Title object, where $text came from
3042 $title = false;
3043
3044 # $part1 is the bit before the first |, and must contain only title characters.
3045 # Various prefixes will be stripped from it later.
3046 $titleWithSpaces = $frame->expand( $piece['title'] );
3047 $part1 = trim( $titleWithSpaces );
3048 $titleText = false;
3049
3050 # Original title text preserved for various purposes
3051 $originalTitle = $part1;
3052
3053 # $args is a list of argument nodes, starting from index 0, not including $part1
3054 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3055 # below won't work b/c this $args isn't an object
3056 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3057
3058 $profileSection = null; // profile templates
3059
3060 # SUBST
3061 if ( !$found ) {
3062 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3063
3064 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3065 # Decide whether to expand template or keep wikitext as-is.
3066 if ( $this->ot['wiki'] ) {
3067 if ( $substMatch === false ) {
3068 $literal = true; # literal when in PST with no prefix
3069 } else {
3070 $literal = false; # expand when in PST with subst: or safesubst:
3071 }
3072 } else {
3073 if ( $substMatch == 'subst' ) {
3074 $literal = true; # literal when not in PST with plain subst:
3075 } else {
3076 $literal = false; # expand when not in PST with safesubst: or no prefix
3077 }
3078 }
3079 if ( $literal ) {
3080 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3081 $isLocalObj = true;
3082 $found = true;
3083 }
3084 }
3085
3086 # Variables
3087 if ( !$found && $args->getLength() == 0 ) {
3088 $id = $this->mVariables->matchStartToEnd( $part1 );
3089 if ( $id !== false ) {
3090 $text = $this->getVariableValue( $id, $frame );
3091 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3092 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3093 }
3094 $found = true;
3095 }
3096 }
3097
3098 # MSG, MSGNW and RAW
3099 if ( !$found ) {
3100 # Check for MSGNW:
3101 $mwMsgnw = MagicWord::get( 'msgnw' );
3102 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3103 $nowiki = true;
3104 } else {
3105 # Remove obsolete MSG:
3106 $mwMsg = MagicWord::get( 'msg' );
3107 $mwMsg->matchStartAndRemove( $part1 );
3108 }
3109
3110 # Check for RAW:
3111 $mwRaw = MagicWord::get( 'raw' );
3112 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3113 $forceRawInterwiki = true;
3114 }
3115 }
3116
3117 # Parser functions
3118 if ( !$found ) {
3119 $colonPos = strpos( $part1, ':' );
3120 if ( $colonPos !== false ) {
3121 $func = substr( $part1, 0, $colonPos );
3122 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3123 $argsLength = $args->getLength();
3124 for ( $i = 0; $i < $argsLength; $i++ ) {
3125 $funcArgs[] = $args->item( $i );
3126 }
3127 try {
3128 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3129 } catch ( Exception $ex ) {
3130 throw $ex;
3131 }
3132
3133 # The interface for parser functions allows for extracting
3134 # flags into the local scope. Extract any forwarded flags
3135 # here.
3136 extract( $result );
3137 }
3138 }
3139
3140 # Finish mangling title and then check for loops.
3141 # Set $title to a Title object and $titleText to the PDBK
3142 if ( !$found ) {
3143 $ns = NS_TEMPLATE;
3144 # Split the title into page and subpage
3145 $subpage = '';
3146 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3147 if ( $part1 !== $relative ) {
3148 $part1 = $relative;
3149 $ns = $this->mTitle->getNamespace();
3150 }
3151 $title = Title::newFromText( $part1, $ns );
3152 if ( $title ) {
3153 $titleText = $title->getPrefixedText();
3154 # Check for language variants if the template is not found
3155 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3156 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3157 }
3158 # Do recursion depth check
3159 $limit = $this->mOptions->getMaxTemplateDepth();
3160 if ( $frame->depth >= $limit ) {
3161 $found = true;
3162 $text = '<span class="error">'
3163 . wfMessage( 'parser-template-recursion-depth-warning' )
3164 ->numParams( $limit )->inContentLanguage()->text()
3165 . '</span>';
3166 }
3167 }
3168 }
3169
3170 # Load from database
3171 if ( !$found && $title ) {
3172 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3173 if ( !$title->isExternal() ) {
3174 if ( $title->isSpecialPage()
3175 && $this->mOptions->getAllowSpecialInclusion()
3176 && $this->ot['html']
3177 ) {
3178 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3179 // Pass the template arguments as URL parameters.
3180 // "uselang" will have no effect since the Language object
3181 // is forced to the one defined in ParserOptions.
3182 $pageArgs = [];
3183 $argsLength = $args->getLength();
3184 for ( $i = 0; $i < $argsLength; $i++ ) {
3185 $bits = $args->item( $i )->splitArg();
3186 if ( strval( $bits['index'] ) === '' ) {
3187 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3188 $value = trim( $frame->expand( $bits['value'] ) );
3189 $pageArgs[$name] = $value;
3190 }
3191 }
3192
3193 // Create a new context to execute the special page
3194 $context = new RequestContext;
3195 $context->setTitle( $title );
3196 $context->setRequest( new FauxRequest( $pageArgs ) );
3197 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3198 $context->setUser( $this->getUser() );
3199 } else {
3200 // If this page is cached, then we better not be per user.
3201 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3202 }
3203 $context->setLanguage( $this->mOptions->getUserLangObj() );
3204 $ret = SpecialPageFactory::capturePath(
3205 $title, $context, $this->getLinkRenderer() );
3206 if ( $ret ) {
3207 $text = $context->getOutput()->getHTML();
3208 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3209 $found = true;
3210 $isHTML = true;
3211 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3212 $this->mOutput->updateRuntimeAdaptiveExpiry(
3213 $specialPage->maxIncludeCacheTime()
3214 );
3215 }
3216 }
3217 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3218 $found = false; # access denied
3219 wfDebug( __METHOD__ . ": template inclusion denied for " .
3220 $title->getPrefixedDBkey() . "\n" );
3221 } else {
3222 list( $text, $title ) = $this->getTemplateDom( $title );
3223 if ( $text !== false ) {
3224 $found = true;
3225 $isChildObj = true;
3226 }
3227 }
3228
3229 # If the title is valid but undisplayable, make a link to it
3230 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3231 $text = "[[:$titleText]]";
3232 $found = true;
3233 }
3234 } elseif ( $title->isTrans() ) {
3235 # Interwiki transclusion
3236 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3237 $text = $this->interwikiTransclude( $title, 'render' );
3238 $isHTML = true;
3239 } else {
3240 $text = $this->interwikiTransclude( $title, 'raw' );
3241 # Preprocess it like a template
3242 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3243 $isChildObj = true;
3244 }
3245 $found = true;
3246 }
3247
3248 # Do infinite loop check
3249 # This has to be done after redirect resolution to avoid infinite loops via redirects
3250 if ( !$frame->loopCheck( $title ) ) {
3251 $found = true;
3252 $text = '<span class="error">'
3253 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3254 . '</span>';
3255 $this->addTrackingCategory( 'template-loop-category' );
3256 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3257 }
3258 }
3259
3260 # If we haven't found text to substitute by now, we're done
3261 # Recover the source wikitext and return it
3262 if ( !$found ) {
3263 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3264 if ( $profileSection ) {
3265 $this->mProfiler->scopedProfileOut( $profileSection );
3266 }
3267 return [ 'object' => $text ];
3268 }
3269
3270 # Expand DOM-style return values in a child frame
3271 if ( $isChildObj ) {
3272 # Clean up argument array
3273 $newFrame = $frame->newChild( $args, $title );
3274
3275 if ( $nowiki ) {
3276 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3277 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3278 # Expansion is eligible for the empty-frame cache
3279 $text = $newFrame->cachedExpand( $titleText, $text );
3280 } else {
3281 # Uncached expansion
3282 $text = $newFrame->expand( $text );
3283 }
3284 }
3285 if ( $isLocalObj && $nowiki ) {
3286 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3287 $isLocalObj = false;
3288 }
3289
3290 if ( $profileSection ) {
3291 $this->mProfiler->scopedProfileOut( $profileSection );
3292 }
3293
3294 # Replace raw HTML by a placeholder
3295 if ( $isHTML ) {
3296 $text = $this->insertStripItem( $text );
3297 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3298 # Escape nowiki-style return values
3299 $text = wfEscapeWikiText( $text );
3300 } elseif ( is_string( $text )
3301 && !$piece['lineStart']
3302 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3303 ) {
3304 # T2529: if the template begins with a table or block-level
3305 # element, it should be treated as beginning a new line.
3306 # This behavior is somewhat controversial.
3307 $text = "\n" . $text;
3308 }
3309
3310 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3311 # Error, oversize inclusion
3312 if ( $titleText !== false ) {
3313 # Make a working, properly escaped link if possible (T25588)
3314 $text = "[[:$titleText]]";
3315 } else {
3316 # This will probably not be a working link, but at least it may
3317 # provide some hint of where the problem is
3318 preg_replace( '/^:/', '', $originalTitle );
3319 $text = "[[:$originalTitle]]";
3320 }
3321 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3322 . 'post-expand include size too large -->' );
3323 $this->limitationWarn( 'post-expand-template-inclusion' );
3324 }
3325
3326 if ( $isLocalObj ) {
3327 $ret = [ 'object' => $text ];
3328 } else {
3329 $ret = [ 'text' => $text ];
3330 }
3331
3332 return $ret;
3333 }
3334
3335 /**
3336 * Call a parser function and return an array with text and flags.
3337 *
3338 * The returned array will always contain a boolean 'found', indicating
3339 * whether the parser function was found or not. It may also contain the
3340 * following:
3341 * text: string|object, resulting wikitext or PP DOM object
3342 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3343 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3344 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3345 * nowiki: bool, wiki markup in $text should be escaped
3346 *
3347 * @since 1.21
3348 * @param PPFrame $frame The current frame, contains template arguments
3349 * @param string $function Function name
3350 * @param array $args Arguments to the function
3351 * @throws MWException
3352 * @return array
3353 */
3354 public function callParserFunction( $frame, $function, array $args = [] ) {
3355 global $wgContLang;
3356
3357 # Case sensitive functions
3358 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3359 $function = $this->mFunctionSynonyms[1][$function];
3360 } else {
3361 # Case insensitive functions
3362 $function = $wgContLang->lc( $function );
3363 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3364 $function = $this->mFunctionSynonyms[0][$function];
3365 } else {
3366 return [ 'found' => false ];
3367 }
3368 }
3369
3370 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3371
3372 # Workaround for PHP bug 35229 and similar
3373 if ( !is_callable( $callback ) ) {
3374 throw new MWException( "Tag hook for $function is not callable\n" );
3375 }
3376
3377 // Avoid PHP 7.1 warning from passing $this by reference
3378 $parser = $this;
3379
3380 $allArgs = [ &$parser ];
3381 if ( $flags & self::SFH_OBJECT_ARGS ) {
3382 # Convert arguments to PPNodes and collect for appending to $allArgs
3383 $funcArgs = [];
3384 foreach ( $args as $k => $v ) {
3385 if ( $v instanceof PPNode || $k === 0 ) {
3386 $funcArgs[] = $v;
3387 } else {
3388 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3389 }
3390 }
3391
3392 # Add a frame parameter, and pass the arguments as an array
3393 $allArgs[] = $frame;
3394 $allArgs[] = $funcArgs;
3395 } else {
3396 # Convert arguments to plain text and append to $allArgs
3397 foreach ( $args as $k => $v ) {
3398 if ( $v instanceof PPNode ) {
3399 $allArgs[] = trim( $frame->expand( $v ) );
3400 } elseif ( is_int( $k ) && $k >= 0 ) {
3401 $allArgs[] = trim( $v );
3402 } else {
3403 $allArgs[] = trim( "$k=$v" );
3404 }
3405 }
3406 }
3407
3408 $result = call_user_func_array( $callback, $allArgs );
3409
3410 # The interface for function hooks allows them to return a wikitext
3411 # string or an array containing the string and any flags. This mungs
3412 # things around to match what this method should return.
3413 if ( !is_array( $result ) ) {
3414 $result =[
3415 'found' => true,
3416 'text' => $result,
3417 ];
3418 } else {
3419 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3420 $result['text'] = $result[0];
3421 }
3422 unset( $result[0] );
3423 $result += [
3424 'found' => true,
3425 ];
3426 }
3427
3428 $noparse = true;
3429 $preprocessFlags = 0;
3430 if ( isset( $result['noparse'] ) ) {
3431 $noparse = $result['noparse'];
3432 }
3433 if ( isset( $result['preprocessFlags'] ) ) {
3434 $preprocessFlags = $result['preprocessFlags'];
3435 }
3436
3437 if ( !$noparse ) {
3438 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3439 $result['isChildObj'] = true;
3440 }
3441
3442 return $result;
3443 }
3444
3445 /**
3446 * Get the semi-parsed DOM representation of a template with a given title,
3447 * and its redirect destination title. Cached.
3448 *
3449 * @param Title $title
3450 *
3451 * @return array
3452 */
3453 public function getTemplateDom( $title ) {
3454 $cacheTitle = $title;
3455 $titleText = $title->getPrefixedDBkey();
3456
3457 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3458 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3459 $title = Title::makeTitle( $ns, $dbk );
3460 $titleText = $title->getPrefixedDBkey();
3461 }
3462 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3463 return [ $this->mTplDomCache[$titleText], $title ];
3464 }
3465
3466 # Cache miss, go to the database
3467 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3468
3469 if ( $text === false ) {
3470 $this->mTplDomCache[$titleText] = false;
3471 return [ false, $title ];
3472 }
3473
3474 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3475 $this->mTplDomCache[$titleText] = $dom;
3476
3477 if ( !$title->equals( $cacheTitle ) ) {
3478 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3479 [ $title->getNamespace(), $title->getDBkey() ];
3480 }
3481
3482 return [ $dom, $title ];
3483 }
3484
3485 /**
3486 * Fetch the current revision of a given title. Note that the revision
3487 * (and even the title) may not exist in the database, so everything
3488 * contributing to the output of the parser should use this method
3489 * where possible, rather than getting the revisions themselves. This
3490 * method also caches its results, so using it benefits performance.
3491 *
3492 * @since 1.24
3493 * @param Title $title
3494 * @return Revision
3495 */
3496 public function fetchCurrentRevisionOfTitle( $title ) {
3497 $cacheKey = $title->getPrefixedDBkey();
3498 if ( !$this->currentRevisionCache ) {
3499 $this->currentRevisionCache = new MapCacheLRU( 100 );
3500 }
3501 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3502 $this->currentRevisionCache->set( $cacheKey,
3503 // Defaults to Parser::statelessFetchRevision()
3504 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3505 );
3506 }
3507 return $this->currentRevisionCache->get( $cacheKey );
3508 }
3509
3510 /**
3511 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3512 * without passing them on to it.
3513 *
3514 * @since 1.24
3515 * @param Title $title
3516 * @param Parser|bool $parser
3517 * @return Revision|bool False if missing
3518 */
3519 public static function statelessFetchRevision( Title $title, $parser = false ) {
3520 $pageId = $title->getArticleID();
3521 $revId = $title->getLatestRevID();
3522
3523 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3524 if ( $rev ) {
3525 $rev->setTitle( $title );
3526 }
3527
3528 return $rev;
3529 }
3530
3531 /**
3532 * Fetch the unparsed text of a template and register a reference to it.
3533 * @param Title $title
3534 * @return array ( string or false, Title )
3535 */
3536 public function fetchTemplateAndTitle( $title ) {
3537 // Defaults to Parser::statelessFetchTemplate()
3538 $templateCb = $this->mOptions->getTemplateCallback();
3539 $stuff = call_user_func( $templateCb, $title, $this );
3540 // We use U+007F DELETE to distinguish strip markers from regular text.
3541 $text = $stuff['text'];
3542 if ( is_string( $stuff['text'] ) ) {
3543 $text = strtr( $text, "\x7f", "?" );
3544 }
3545 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3546 if ( isset( $stuff['deps'] ) ) {
3547 foreach ( $stuff['deps'] as $dep ) {
3548 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3549 if ( $dep['title']->equals( $this->getTitle() ) ) {
3550 // If we transclude ourselves, the final result
3551 // will change based on the new version of the page
3552 $this->mOutput->setFlag( 'vary-revision' );
3553 }
3554 }
3555 }
3556 return [ $text, $finalTitle ];
3557 }
3558
3559 /**
3560 * Fetch the unparsed text of a template and register a reference to it.
3561 * @param Title $title
3562 * @return string|bool
3563 */
3564 public function fetchTemplate( $title ) {
3565 return $this->fetchTemplateAndTitle( $title )[0];
3566 }
3567
3568 /**
3569 * Static function to get a template
3570 * Can be overridden via ParserOptions::setTemplateCallback().
3571 *
3572 * @param Title $title
3573 * @param bool|Parser $parser
3574 *
3575 * @return array
3576 */
3577 public static function statelessFetchTemplate( $title, $parser = false ) {
3578 $text = $skip = false;
3579 $finalTitle = $title;
3580 $deps = [];
3581
3582 # Loop to fetch the article, with up to 1 redirect
3583 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3584 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3585 // @codingStandardsIgnoreEnd
3586 # Give extensions a chance to select the revision instead
3587 $id = false; # Assume current
3588 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3589 [ $parser, $title, &$skip, &$id ] );
3590
3591 if ( $skip ) {
3592 $text = false;
3593 $deps[] = [
3594 'title' => $title,
3595 'page_id' => $title->getArticleID(),
3596 'rev_id' => null
3597 ];
3598 break;
3599 }
3600 # Get the revision
3601 if ( $id ) {
3602 $rev = Revision::newFromId( $id );
3603 } elseif ( $parser ) {
3604 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3605 } else {
3606 $rev = Revision::newFromTitle( $title );
3607 }
3608 $rev_id = $rev ? $rev->getId() : 0;
3609 # If there is no current revision, there is no page
3610 if ( $id === false && !$rev ) {
3611 $linkCache = LinkCache::singleton();
3612 $linkCache->addBadLinkObj( $title );
3613 }
3614
3615 $deps[] = [
3616 'title' => $title,
3617 'page_id' => $title->getArticleID(),
3618 'rev_id' => $rev_id ];
3619 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3620 # We fetched a rev from a different title; register it too...
3621 $deps[] = [
3622 'title' => $rev->getTitle(),
3623 'page_id' => $rev->getPage(),
3624 'rev_id' => $rev_id ];
3625 }
3626
3627 if ( $rev ) {
3628 $content = $rev->getContent();
3629 $text = $content ? $content->getWikitextForTransclusion() : null;
3630
3631 Hooks::run( 'ParserFetchTemplate',
3632 [ $parser, $title, $rev, &$text, &$deps ] );
3633
3634 if ( $text === false || $text === null ) {
3635 $text = false;
3636 break;
3637 }
3638 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3639 global $wgContLang;
3640 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3641 if ( !$message->exists() ) {
3642 $text = false;
3643 break;
3644 }
3645 $content = $message->content();
3646 $text = $message->plain();
3647 } else {
3648 break;
3649 }
3650 if ( !$content ) {
3651 break;
3652 }
3653 # Redirect?
3654 $finalTitle = $title;
3655 $title = $content->getRedirectTarget();
3656 }
3657 return [
3658 'text' => $text,
3659 'finalTitle' => $finalTitle,
3660 'deps' => $deps ];
3661 }
3662
3663 /**
3664 * Fetch a file and its title and register a reference to it.
3665 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3666 * @param Title $title
3667 * @param array $options Array of options to RepoGroup::findFile
3668 * @return File|bool
3669 */
3670 public function fetchFile( $title, $options = [] ) {
3671 return $this->fetchFileAndTitle( $title, $options )[0];
3672 }
3673
3674 /**
3675 * Fetch a file and its title and register a reference to it.
3676 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3677 * @param Title $title
3678 * @param array $options Array of options to RepoGroup::findFile
3679 * @return array ( File or false, Title of file )
3680 */
3681 public function fetchFileAndTitle( $title, $options = [] ) {
3682 $file = $this->fetchFileNoRegister( $title, $options );
3683
3684 $time = $file ? $file->getTimestamp() : false;
3685 $sha1 = $file ? $file->getSha1() : false;
3686 # Register the file as a dependency...
3687 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3688 if ( $file && !$title->equals( $file->getTitle() ) ) {
3689 # Update fetched file title
3690 $title = $file->getTitle();
3691 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3692 }
3693 return [ $file, $title ];
3694 }
3695
3696 /**
3697 * Helper function for fetchFileAndTitle.
3698 *
3699 * Also useful if you need to fetch a file but not use it yet,
3700 * for example to get the file's handler.
3701 *
3702 * @param Title $title
3703 * @param array $options Array of options to RepoGroup::findFile
3704 * @return File|bool
3705 */
3706 protected function fetchFileNoRegister( $title, $options = [] ) {
3707 if ( isset( $options['broken'] ) ) {
3708 $file = false; // broken thumbnail forced by hook
3709 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3710 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3711 } else { // get by (name,timestamp)
3712 $file = wfFindFile( $title, $options );
3713 }
3714 return $file;
3715 }
3716
3717 /**
3718 * Transclude an interwiki link.
3719 *
3720 * @param Title $title
3721 * @param string $action
3722 *
3723 * @return string
3724 */
3725 public function interwikiTransclude( $title, $action ) {
3726 global $wgEnableScaryTranscluding;
3727
3728 if ( !$wgEnableScaryTranscluding ) {
3729 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3730 }
3731
3732 $url = $title->getFullURL( [ 'action' => $action ] );
3733
3734 if ( strlen( $url ) > 255 ) {
3735 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3736 }
3737 return $this->fetchScaryTemplateMaybeFromCache( $url );
3738 }
3739
3740 /**
3741 * @param string $url
3742 * @return mixed|string
3743 */
3744 public function fetchScaryTemplateMaybeFromCache( $url ) {
3745 global $wgTranscludeCacheExpiry;
3746 $dbr = wfGetDB( DB_REPLICA );
3747 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3748 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3749 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3750 if ( $obj ) {
3751 return $obj->tc_contents;
3752 }
3753
3754 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3755 $status = $req->execute(); // Status object
3756 if ( $status->isOK() ) {
3757 $text = $req->getContent();
3758 } elseif ( $req->getStatus() != 200 ) {
3759 // Though we failed to fetch the content, this status is useless.
3760 return wfMessage( 'scarytranscludefailed-httpstatus' )
3761 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3762 } else {
3763 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3764 }
3765
3766 $dbw = wfGetDB( DB_MASTER );
3767 $dbw->replace( 'transcache', [ 'tc_url' ], [
3768 'tc_url' => $url,
3769 'tc_time' => $dbw->timestamp( time() ),
3770 'tc_contents' => $text
3771 ] );
3772 return $text;
3773 }
3774
3775 /**
3776 * Triple brace replacement -- used for template arguments
3777 * @private
3778 *
3779 * @param array $piece
3780 * @param PPFrame $frame
3781 *
3782 * @return array
3783 */
3784 public function argSubstitution( $piece, $frame ) {
3785
3786 $error = false;
3787 $parts = $piece['parts'];
3788 $nameWithSpaces = $frame->expand( $piece['title'] );
3789 $argName = trim( $nameWithSpaces );
3790 $object = false;
3791 $text = $frame->getArgument( $argName );
3792 if ( $text === false && $parts->getLength() > 0
3793 && ( $this->ot['html']
3794 || $this->ot['pre']
3795 || ( $this->ot['wiki'] && $frame->isTemplate() )
3796 )
3797 ) {
3798 # No match in frame, use the supplied default
3799 $object = $parts->item( 0 )->getChildren();
3800 }
3801 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3802 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3803 $this->limitationWarn( 'post-expand-template-argument' );
3804 }
3805
3806 if ( $text === false && $object === false ) {
3807 # No match anywhere
3808 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3809 }
3810 if ( $error !== false ) {
3811 $text .= $error;
3812 }
3813 if ( $object !== false ) {
3814 $ret = [ 'object' => $object ];
3815 } else {
3816 $ret = [ 'text' => $text ];
3817 }
3818
3819 return $ret;
3820 }
3821
3822 /**
3823 * Return the text to be used for a given extension tag.
3824 * This is the ghost of strip().
3825 *
3826 * @param array $params Associative array of parameters:
3827 * name PPNode for the tag name
3828 * attr PPNode for unparsed text where tag attributes are thought to be
3829 * attributes Optional associative array of parsed attributes
3830 * inner Contents of extension element
3831 * noClose Original text did not have a close tag
3832 * @param PPFrame $frame
3833 *
3834 * @throws MWException
3835 * @return string
3836 */
3837 public function extensionSubstitution( $params, $frame ) {
3838 static $errorStr = '<span class="error">';
3839 static $errorLen = 20;
3840
3841 $name = $frame->expand( $params['name'] );
3842 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3843 // Probably expansion depth or node count exceeded. Just punt the
3844 // error up.
3845 return $name;
3846 }
3847
3848 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3849 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3850 // See above
3851 return $attrText;
3852 }
3853
3854 // We can't safely check if the expansion for $content resulted in an
3855 // error, because the content could happen to be the error string
3856 // (T149622).
3857 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3858
3859 $marker = self::MARKER_PREFIX . "-$name-"
3860 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3861
3862 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3863 ( $this->ot['html'] || $this->ot['pre'] );
3864 if ( $isFunctionTag ) {
3865 $markerType = 'none';
3866 } else {
3867 $markerType = 'general';
3868 }
3869 if ( $this->ot['html'] || $isFunctionTag ) {
3870 $name = strtolower( $name );
3871 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3872 if ( isset( $params['attributes'] ) ) {
3873 $attributes = $attributes + $params['attributes'];
3874 }
3875
3876 if ( isset( $this->mTagHooks[$name] ) ) {
3877 # Workaround for PHP bug 35229 and similar
3878 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3879 throw new MWException( "Tag hook for $name is not callable\n" );
3880 }
3881 $output = call_user_func_array( $this->mTagHooks[$name],
3882 [ $content, $attributes, $this, $frame ] );
3883 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3884 list( $callback, ) = $this->mFunctionTagHooks[$name];
3885 if ( !is_callable( $callback ) ) {
3886 throw new MWException( "Tag hook for $name is not callable\n" );
3887 }
3888
3889 // Avoid PHP 7.1 warning from passing $this by reference
3890 $parser = $this;
3891 $output = call_user_func_array( $callback, [ &$parser, $frame, $content, $attributes ] );
3892 } else {
3893 $output = '<span class="error">Invalid tag extension name: ' .
3894 htmlspecialchars( $name ) . '</span>';
3895 }
3896
3897 if ( is_array( $output ) ) {
3898 # Extract flags to local scope (to override $markerType)
3899 $flags = $output;
3900 $output = $flags[0];
3901 unset( $flags[0] );
3902 extract( $flags );
3903 }
3904 } else {
3905 if ( is_null( $attrText ) ) {
3906 $attrText = '';
3907 }
3908 if ( isset( $params['attributes'] ) ) {
3909 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3910 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3911 htmlspecialchars( $attrValue ) . '"';
3912 }
3913 }
3914 if ( $content === null ) {
3915 $output = "<$name$attrText/>";
3916 } else {
3917 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3918 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3919 // See above
3920 return $close;
3921 }
3922 $output = "<$name$attrText>$content$close";
3923 }
3924 }
3925
3926 if ( $markerType === 'none' ) {
3927 return $output;
3928 } elseif ( $markerType === 'nowiki' ) {
3929 $this->mStripState->addNoWiki( $marker, $output );
3930 } elseif ( $markerType === 'general' ) {
3931 $this->mStripState->addGeneral( $marker, $output );
3932 } else {
3933 throw new MWException( __METHOD__ . ': invalid marker type' );
3934 }
3935 return $marker;
3936 }
3937
3938 /**
3939 * Increment an include size counter
3940 *
3941 * @param string $type The type of expansion
3942 * @param int $size The size of the text
3943 * @return bool False if this inclusion would take it over the maximum, true otherwise
3944 */
3945 public function incrementIncludeSize( $type, $size ) {
3946 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3947 return false;
3948 } else {
3949 $this->mIncludeSizes[$type] += $size;
3950 return true;
3951 }
3952 }
3953
3954 /**
3955 * Increment the expensive function count
3956 *
3957 * @return bool False if the limit has been exceeded
3958 */
3959 public function incrementExpensiveFunctionCount() {
3960 $this->mExpensiveFunctionCount++;
3961 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3962 }
3963
3964 /**
3965 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3966 * Fills $this->mDoubleUnderscores, returns the modified text
3967 *
3968 * @param string $text
3969 *
3970 * @return string
3971 */
3972 public function doDoubleUnderscore( $text ) {
3973
3974 # The position of __TOC__ needs to be recorded
3975 $mw = MagicWord::get( 'toc' );
3976 if ( $mw->match( $text ) ) {
3977 $this->mShowToc = true;
3978 $this->mForceTocPosition = true;
3979
3980 # Set a placeholder. At the end we'll fill it in with the TOC.
3981 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3982
3983 # Only keep the first one.
3984 $text = $mw->replace( '', $text );
3985 }
3986
3987 # Now match and remove the rest of them
3988 $mwa = MagicWord::getDoubleUnderscoreArray();
3989 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3990
3991 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3992 $this->mOutput->mNoGallery = true;
3993 }
3994 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3995 $this->mShowToc = false;
3996 }
3997 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3998 && $this->mTitle->getNamespace() == NS_CATEGORY
3999 ) {
4000 $this->addTrackingCategory( 'hidden-category-category' );
4001 }
4002 # (T10068) Allow control over whether robots index a page.
4003 # __INDEX__ always overrides __NOINDEX__, see T16899
4004 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4005 $this->mOutput->setIndexPolicy( 'noindex' );
4006 $this->addTrackingCategory( 'noindex-category' );
4007 }
4008 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4009 $this->mOutput->setIndexPolicy( 'index' );
4010 $this->addTrackingCategory( 'index-category' );
4011 }
4012
4013 # Cache all double underscores in the database
4014 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4015 $this->mOutput->setProperty( $key, '' );
4016 }
4017
4018 return $text;
4019 }
4020
4021 /**
4022 * @see ParserOutput::addTrackingCategory()
4023 * @param string $msg Message key
4024 * @return bool Whether the addition was successful
4025 */
4026 public function addTrackingCategory( $msg ) {
4027 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4028 }
4029
4030 /**
4031 * This function accomplishes several tasks:
4032 * 1) Auto-number headings if that option is enabled
4033 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4034 * 3) Add a Table of contents on the top for users who have enabled the option
4035 * 4) Auto-anchor headings
4036 *
4037 * It loops through all headlines, collects the necessary data, then splits up the
4038 * string and re-inserts the newly formatted headlines.
4039 *
4040 * @param string $text
4041 * @param string $origText Original, untouched wikitext
4042 * @param bool $isMain
4043 * @return mixed|string
4044 * @private
4045 */
4046 public function formatHeadings( $text, $origText, $isMain = true ) {
4047 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4048
4049 # Inhibit editsection links if requested in the page
4050 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4051 $maybeShowEditLink = $showEditLink = false;
4052 } else {
4053 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4054 $showEditLink = $this->mOptions->getEditSection();
4055 }
4056 if ( $showEditLink ) {
4057 $this->mOutput->setEditSectionTokens( true );
4058 }
4059
4060 # Get all headlines for numbering them and adding funky stuff like [edit]
4061 # links - this is for later, but we need the number of headlines right now
4062 $matches = [];
4063 $numMatches = preg_match_all(
4064 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4065 $text,
4066 $matches
4067 );
4068
4069 # if there are fewer than 4 headlines in the article, do not show TOC
4070 # unless it's been explicitly enabled.
4071 $enoughToc = $this->mShowToc &&
4072 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4073
4074 # Allow user to stipulate that a page should have a "new section"
4075 # link added via __NEWSECTIONLINK__
4076 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4077 $this->mOutput->setNewSection( true );
4078 }
4079
4080 # Allow user to remove the "new section"
4081 # link via __NONEWSECTIONLINK__
4082 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4083 $this->mOutput->hideNewSection( true );
4084 }
4085
4086 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4087 # override above conditions and always show TOC above first header
4088 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4089 $this->mShowToc = true;
4090 $enoughToc = true;
4091 }
4092
4093 # headline counter
4094 $headlineCount = 0;
4095 $numVisible = 0;
4096
4097 # Ugh .. the TOC should have neat indentation levels which can be
4098 # passed to the skin functions. These are determined here
4099 $toc = '';
4100 $full = '';
4101 $head = [];
4102 $sublevelCount = [];
4103 $levelCount = [];
4104 $level = 0;
4105 $prevlevel = 0;
4106 $toclevel = 0;
4107 $prevtoclevel = 0;
4108 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4109 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4110 $oldType = $this->mOutputType;
4111 $this->setOutputType( self::OT_WIKI );
4112 $frame = $this->getPreprocessor()->newFrame();
4113 $root = $this->preprocessToDom( $origText );
4114 $node = $root->getFirstChild();
4115 $byteOffset = 0;
4116 $tocraw = [];
4117 $refers = [];
4118
4119 $headlines = $numMatches !== false ? $matches[3] : [];
4120
4121 foreach ( $headlines as $headline ) {
4122 $isTemplate = false;
4123 $titleText = false;
4124 $sectionIndex = false;
4125 $numbering = '';
4126 $markerMatches = [];
4127 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4128 $serial = $markerMatches[1];
4129 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4130 $isTemplate = ( $titleText != $baseTitleText );
4131 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4132 }
4133
4134 if ( $toclevel ) {
4135 $prevlevel = $level;
4136 }
4137 $level = $matches[1][$headlineCount];
4138
4139 if ( $level > $prevlevel ) {
4140 # Increase TOC level
4141 $toclevel++;
4142 $sublevelCount[$toclevel] = 0;
4143 if ( $toclevel < $wgMaxTocLevel ) {
4144 $prevtoclevel = $toclevel;
4145 $toc .= Linker::tocIndent();
4146 $numVisible++;
4147 }
4148 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4149 # Decrease TOC level, find level to jump to
4150
4151 for ( $i = $toclevel; $i > 0; $i-- ) {
4152 if ( $levelCount[$i] == $level ) {
4153 # Found last matching level
4154 $toclevel = $i;
4155 break;
4156 } elseif ( $levelCount[$i] < $level ) {
4157 # Found first matching level below current level
4158 $toclevel = $i + 1;
4159 break;
4160 }
4161 }
4162 if ( $i == 0 ) {
4163 $toclevel = 1;
4164 }
4165 if ( $toclevel < $wgMaxTocLevel ) {
4166 if ( $prevtoclevel < $wgMaxTocLevel ) {
4167 # Unindent only if the previous toc level was shown :p
4168 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4169 $prevtoclevel = $toclevel;
4170 } else {
4171 $toc .= Linker::tocLineEnd();
4172 }
4173 }
4174 } else {
4175 # No change in level, end TOC line
4176 if ( $toclevel < $wgMaxTocLevel ) {
4177 $toc .= Linker::tocLineEnd();
4178 }
4179 }
4180
4181 $levelCount[$toclevel] = $level;
4182
4183 # count number of headlines for each level
4184 $sublevelCount[$toclevel]++;
4185 $dot = 0;
4186 for ( $i = 1; $i <= $toclevel; $i++ ) {
4187 if ( !empty( $sublevelCount[$i] ) ) {
4188 if ( $dot ) {
4189 $numbering .= '.';
4190 }
4191 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4192 $dot = 1;
4193 }
4194 }
4195
4196 # The safe header is a version of the header text safe to use for links
4197
4198 # Remove link placeholders by the link text.
4199 # <!--LINK number-->
4200 # turns into
4201 # link text with suffix
4202 # Do this before unstrip since link text can contain strip markers
4203 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4204
4205 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4206 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4207
4208 # Strip out HTML (first regex removes any tag not allowed)
4209 # Allowed tags are:
4210 # * <sup> and <sub> (T10393)
4211 # * <i> (T28375)
4212 # * <b> (r105284)
4213 # * <bdi> (T74884)
4214 # * <span dir="rtl"> and <span dir="ltr"> (T37167)
4215 # * <s> and <strike> (T35715)
4216 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4217 # to allow setting directionality in toc items.
4218 $tocline = preg_replace(
4219 [
4220 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4221 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4222 ],
4223 [ '', '<$1>' ],
4224 $safeHeadline
4225 );
4226
4227 # Strip '<span></span>', which is the result from the above if
4228 # <span id="foo"></span> is used to produce an additional anchor
4229 # for a section.
4230 $tocline = str_replace( '<span></span>', '', $tocline );
4231
4232 $tocline = trim( $tocline );
4233
4234 # For the anchor, strip out HTML-y stuff period
4235 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4236 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4237
4238 # Save headline for section edit hint before it's escaped
4239 $headlineHint = $safeHeadline;
4240
4241 if ( $wgExperimentalHtmlIds ) {
4242 # For reverse compatibility, provide an id that's
4243 # HTML4-compatible, like we used to.
4244 # It may be worth noting, academically, that it's possible for
4245 # the legacy anchor to conflict with a non-legacy headline
4246 # anchor on the page. In this case likely the "correct" thing
4247 # would be to either drop the legacy anchors or make sure
4248 # they're numbered first. However, this would require people
4249 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4250 # manually, so let's not bother worrying about it.
4251 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4252 [ 'noninitial', 'legacy' ] );
4253 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4254
4255 if ( $legacyHeadline == $safeHeadline ) {
4256 # No reason to have both (in fact, we can't)
4257 $legacyHeadline = false;
4258 }
4259 } else {
4260 $legacyHeadline = false;
4261 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4262 'noninitial' );
4263 }
4264
4265 # HTML names must be case-insensitively unique (T12721).
4266 # This does not apply to Unicode characters per
4267 # https://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4268 # @todo FIXME: We may be changing them depending on the current locale.
4269 $arrayKey = strtolower( $safeHeadline );
4270 if ( $legacyHeadline === false ) {
4271 $legacyArrayKey = false;
4272 } else {
4273 $legacyArrayKey = strtolower( $legacyHeadline );
4274 }
4275
4276 # Create the anchor for linking from the TOC to the section
4277 $anchor = $safeHeadline;
4278 $legacyAnchor = $legacyHeadline;
4279 if ( isset( $refers[$arrayKey] ) ) {
4280 // @codingStandardsIgnoreStart
4281 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4282 // @codingStandardsIgnoreEnd
4283 $anchor .= "_$i";
4284 $refers["${arrayKey}_$i"] = true;
4285 } else {
4286 $refers[$arrayKey] = true;
4287 }
4288 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4289 // @codingStandardsIgnoreStart
4290 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4291 // @codingStandardsIgnoreEnd
4292 $legacyAnchor .= "_$i";
4293 $refers["${legacyArrayKey}_$i"] = true;
4294 } else {
4295 $refers[$legacyArrayKey] = true;
4296 }
4297
4298 # Don't number the heading if it is the only one (looks silly)
4299 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4300 # the two are different if the line contains a link
4301 $headline = Html::element(
4302 'span',
4303 [ 'class' => 'mw-headline-number' ],
4304 $numbering
4305 ) . ' ' . $headline;
4306 }
4307
4308 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4309 $toc .= Linker::tocLine( $anchor, $tocline,
4310 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4311 }
4312
4313 # Add the section to the section tree
4314 # Find the DOM node for this header
4315 $noOffset = ( $isTemplate || $sectionIndex === false );
4316 while ( $node && !$noOffset ) {
4317 if ( $node->getName() === 'h' ) {
4318 $bits = $node->splitHeading();
4319 if ( $bits['i'] == $sectionIndex ) {
4320 break;
4321 }
4322 }
4323 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4324 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4325 $node = $node->getNextSibling();
4326 }
4327 $tocraw[] = [
4328 'toclevel' => $toclevel,
4329 'level' => $level,
4330 'line' => $tocline,
4331 'number' => $numbering,
4332 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4333 'fromtitle' => $titleText,
4334 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4335 'anchor' => $anchor,
4336 ];
4337
4338 # give headline the correct <h#> tag
4339 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4340 // Output edit section links as markers with styles that can be customized by skins
4341 if ( $isTemplate ) {
4342 # Put a T flag in the section identifier, to indicate to extractSections()
4343 # that sections inside <includeonly> should be counted.
4344 $editsectionPage = $titleText;
4345 $editsectionSection = "T-$sectionIndex";
4346 $editsectionContent = null;
4347 } else {
4348 $editsectionPage = $this->mTitle->getPrefixedText();
4349 $editsectionSection = $sectionIndex;
4350 $editsectionContent = $headlineHint;
4351 }
4352 // We use a bit of pesudo-xml for editsection markers. The
4353 // language converter is run later on. Using a UNIQ style marker
4354 // leads to the converter screwing up the tokens when it
4355 // converts stuff. And trying to insert strip tags fails too. At
4356 // this point all real inputted tags have already been escaped,
4357 // so we don't have to worry about a user trying to input one of
4358 // these markers directly. We use a page and section attribute
4359 // to stop the language converter from converting these
4360 // important bits of data, but put the headline hint inside a
4361 // content block because the language converter is supposed to
4362 // be able to convert that piece of data.
4363 // Gets replaced with html in ParserOutput::getText
4364 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4365 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4366 if ( $editsectionContent !== null ) {
4367 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4368 } else {
4369 $editlink .= '/>';
4370 }
4371 } else {
4372 $editlink = '';
4373 }
4374 $head[$headlineCount] = Linker::makeHeadline( $level,
4375 $matches['attrib'][$headlineCount], $anchor, $headline,
4376 $editlink, $legacyAnchor );
4377
4378 $headlineCount++;
4379 }
4380
4381 $this->setOutputType( $oldType );
4382
4383 # Never ever show TOC if no headers
4384 if ( $numVisible < 1 ) {
4385 $enoughToc = false;
4386 }
4387
4388 if ( $enoughToc ) {
4389 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4390 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4391 }
4392 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4393 $this->mOutput->setTOCHTML( $toc );
4394 $toc = self::TOC_START . $toc . self::TOC_END;
4395 $this->mOutput->addModules( 'mediawiki.toc' );
4396 }
4397
4398 if ( $isMain ) {
4399 $this->mOutput->setSections( $tocraw );
4400 }
4401
4402 # split up and insert constructed headlines
4403 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4404 $i = 0;
4405
4406 // build an array of document sections
4407 $sections = [];
4408 foreach ( $blocks as $block ) {
4409 // $head is zero-based, sections aren't.
4410 if ( empty( $head[$i - 1] ) ) {
4411 $sections[$i] = $block;
4412 } else {
4413 $sections[$i] = $head[$i - 1] . $block;
4414 }
4415
4416 /**
4417 * Send a hook, one per section.
4418 * The idea here is to be able to make section-level DIVs, but to do so in a
4419 * lower-impact, more correct way than r50769
4420 *
4421 * $this : caller
4422 * $section : the section number
4423 * &$sectionContent : ref to the content of the section
4424 * $showEditLinks : boolean describing whether this section has an edit link
4425 */
4426 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4427
4428 $i++;
4429 }
4430
4431 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4432 // append the TOC at the beginning
4433 // Top anchor now in skin
4434 $sections[0] = $sections[0] . $toc . "\n";
4435 }
4436
4437 $full .= implode( '', $sections );
4438
4439 if ( $this->mForceTocPosition ) {
4440 return str_replace( '<!--MWTOC-->', $toc, $full );
4441 } else {
4442 return $full;
4443 }
4444 }
4445
4446 /**
4447 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4448 * conversion, substituting signatures, {{subst:}} templates, etc.
4449 *
4450 * @param string $text The text to transform
4451 * @param Title $title The Title object for the current article
4452 * @param User $user The User object describing the current user
4453 * @param ParserOptions $options Parsing options
4454 * @param bool $clearState Whether to clear the parser state first
4455 * @return string The altered wiki markup
4456 */
4457 public function preSaveTransform( $text, Title $title, User $user,
4458 ParserOptions $options, $clearState = true
4459 ) {
4460 if ( $clearState ) {
4461 $magicScopeVariable = $this->lock();
4462 }
4463 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4464 $this->setUser( $user );
4465
4466 // Strip U+0000 NULL (T159174)
4467 $text = str_replace( "\000", '', $text );
4468
4469 // We still normalize line endings for backwards-compatibility
4470 // with other code that just calls PST, but this should already
4471 // be handled in TextContent subclasses
4472 $text = TextContent::normalizeLineEndings( $text );
4473
4474 if ( $options->getPreSaveTransform() ) {
4475 $text = $this->pstPass2( $text, $user );
4476 }
4477 $text = $this->mStripState->unstripBoth( $text );
4478
4479 $this->setUser( null ); # Reset
4480
4481 return $text;
4482 }
4483
4484 /**
4485 * Pre-save transform helper function
4486 *
4487 * @param string $text
4488 * @param User $user
4489 *
4490 * @return string
4491 */
4492 private function pstPass2( $text, $user ) {
4493 global $wgContLang;
4494
4495 # Note: This is the timestamp saved as hardcoded wikitext to
4496 # the database, we use $wgContLang here in order to give
4497 # everyone the same signature and use the default one rather
4498 # than the one selected in each user's preferences.
4499 # (see also T14815)
4500 $ts = $this->mOptions->getTimestamp();
4501 $timestamp = MWTimestamp::getLocalInstance( $ts );
4502 $ts = $timestamp->format( 'YmdHis' );
4503 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4504
4505 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4506
4507 # Variable replacement
4508 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4509 $text = $this->replaceVariables( $text );
4510
4511 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4512 # which may corrupt this parser instance via its wfMessage()->text() call-
4513
4514 # Signatures
4515 $sigText = $this->getUserSig( $user );
4516 $text = strtr( $text, [
4517 '~~~~~' => $d,
4518 '~~~~' => "$sigText $d",
4519 '~~~' => $sigText
4520 ] );
4521
4522 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4523 $tc = '[' . Title::legalChars() . ']';
4524 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4525
4526 // [[ns:page (context)|]]
4527 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4528 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4529 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4530 // [[ns:page (context), context|]] (using either single or double-width comma)
4531 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4532 // [[|page]] (reverse pipe trick: add context from page title)
4533 $p2 = "/\[\[\\|($tc+)]]/";
4534
4535 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4536 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4537 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4538 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4539
4540 $t = $this->mTitle->getText();
4541 $m = [];
4542 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4543 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4544 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4545 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4546 } else {
4547 # if there's no context, don't bother duplicating the title
4548 $text = preg_replace( $p2, '[[\\1]]', $text );
4549 }
4550
4551 return $text;
4552 }
4553
4554 /**
4555 * Fetch the user's signature text, if any, and normalize to
4556 * validated, ready-to-insert wikitext.
4557 * If you have pre-fetched the nickname or the fancySig option, you can
4558 * specify them here to save a database query.
4559 * Do not reuse this parser instance after calling getUserSig(),
4560 * as it may have changed if it's the $wgParser.
4561 *
4562 * @param User $user
4563 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4564 * @param bool|null $fancySig whether the nicknname is the complete signature
4565 * or null to use default value
4566 * @return string
4567 */
4568 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4569 global $wgMaxSigChars;
4570
4571 $username = $user->getName();
4572
4573 # If not given, retrieve from the user object.
4574 if ( $nickname === false ) {
4575 $nickname = $user->getOption( 'nickname' );
4576 }
4577
4578 if ( is_null( $fancySig ) ) {
4579 $fancySig = $user->getBoolOption( 'fancysig' );
4580 }
4581
4582 $nickname = $nickname == null ? $username : $nickname;
4583
4584 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4585 $nickname = $username;
4586 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4587 } elseif ( $fancySig !== false ) {
4588 # Sig. might contain markup; validate this
4589 if ( $this->validateSig( $nickname ) !== false ) {
4590 # Validated; clean up (if needed) and return it
4591 return $this->cleanSig( $nickname, true );
4592 } else {
4593 # Failed to validate; fall back to the default
4594 $nickname = $username;
4595 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4596 }
4597 }
4598
4599 # Make sure nickname doesnt get a sig in a sig
4600 $nickname = self::cleanSigInSig( $nickname );
4601
4602 # If we're still here, make it a link to the user page
4603 $userText = wfEscapeWikiText( $username );
4604 $nickText = wfEscapeWikiText( $nickname );
4605 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4606
4607 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4608 ->title( $this->getTitle() )->text();
4609 }
4610
4611 /**
4612 * Check that the user's signature contains no bad XML
4613 *
4614 * @param string $text
4615 * @return string|bool An expanded string, or false if invalid.
4616 */
4617 public function validateSig( $text ) {
4618 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4619 }
4620
4621 /**
4622 * Clean up signature text
4623 *
4624 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4625 * 2) Substitute all transclusions
4626 *
4627 * @param string $text
4628 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4629 * @return string Signature text
4630 */
4631 public function cleanSig( $text, $parsing = false ) {
4632 if ( !$parsing ) {
4633 global $wgTitle;
4634 $magicScopeVariable = $this->lock();
4635 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4636 }
4637
4638 # Option to disable this feature
4639 if ( !$this->mOptions->getCleanSignatures() ) {
4640 return $text;
4641 }
4642
4643 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4644 # => Move this logic to braceSubstitution()
4645 $substWord = MagicWord::get( 'subst' );
4646 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4647 $substText = '{{' . $substWord->getSynonym( 0 );
4648
4649 $text = preg_replace( $substRegex, $substText, $text );
4650 $text = self::cleanSigInSig( $text );
4651 $dom = $this->preprocessToDom( $text );
4652 $frame = $this->getPreprocessor()->newFrame();
4653 $text = $frame->expand( $dom );
4654
4655 if ( !$parsing ) {
4656 $text = $this->mStripState->unstripBoth( $text );
4657 }
4658
4659 return $text;
4660 }
4661
4662 /**
4663 * Strip 3, 4 or 5 tildes out of signatures.
4664 *
4665 * @param string $text
4666 * @return string Signature text with /~{3,5}/ removed
4667 */
4668 public static function cleanSigInSig( $text ) {
4669 $text = preg_replace( '/~{3,5}/', '', $text );
4670 return $text;
4671 }
4672
4673 /**
4674 * Set up some variables which are usually set up in parse()
4675 * so that an external function can call some class members with confidence
4676 *
4677 * @param Title|null $title
4678 * @param ParserOptions $options
4679 * @param int $outputType
4680 * @param bool $clearState
4681 */
4682 public function startExternalParse( Title $title = null, ParserOptions $options,
4683 $outputType, $clearState = true
4684 ) {
4685 $this->startParse( $title, $options, $outputType, $clearState );
4686 }
4687
4688 /**
4689 * @param Title|null $title
4690 * @param ParserOptions $options
4691 * @param int $outputType
4692 * @param bool $clearState
4693 */
4694 private function startParse( Title $title = null, ParserOptions $options,
4695 $outputType, $clearState = true
4696 ) {
4697 $this->setTitle( $title );
4698 $this->mOptions = $options;
4699 $this->setOutputType( $outputType );
4700 if ( $clearState ) {
4701 $this->clearState();
4702 }
4703 }
4704
4705 /**
4706 * Wrapper for preprocess()
4707 *
4708 * @param string $text The text to preprocess
4709 * @param ParserOptions $options Options
4710 * @param Title|null $title Title object or null to use $wgTitle
4711 * @return string
4712 */
4713 public function transformMsg( $text, $options, $title = null ) {
4714 static $executing = false;
4715
4716 # Guard against infinite recursion
4717 if ( $executing ) {
4718 return $text;
4719 }
4720 $executing = true;
4721
4722 if ( !$title ) {
4723 global $wgTitle;
4724 $title = $wgTitle;
4725 }
4726
4727 $text = $this->preprocess( $text, $title, $options );
4728
4729 $executing = false;
4730 return $text;
4731 }
4732
4733 /**
4734 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4735 * The callback should have the following form:
4736 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4737 *
4738 * Transform and return $text. Use $parser for any required context, e.g. use
4739 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4740 *
4741 * Hooks may return extended information by returning an array, of which the
4742 * first numbered element (index 0) must be the return string, and all other
4743 * entries are extracted into local variables within an internal function
4744 * in the Parser class.
4745 *
4746 * This interface (introduced r61913) appears to be undocumented, but
4747 * 'markerType' is used by some core tag hooks to override which strip
4748 * array their results are placed in. **Use great caution if attempting
4749 * this interface, as it is not documented and injudicious use could smash
4750 * private variables.**
4751 *
4752 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4753 * @param callable $callback The callback function (and object) to use for the tag
4754 * @throws MWException
4755 * @return callable|null The old value of the mTagHooks array associated with the hook
4756 */
4757 public function setHook( $tag, $callback ) {
4758 $tag = strtolower( $tag );
4759 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4760 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4761 }
4762 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4763 $this->mTagHooks[$tag] = $callback;
4764 if ( !in_array( $tag, $this->mStripList ) ) {
4765 $this->mStripList[] = $tag;
4766 }
4767
4768 return $oldVal;
4769 }
4770
4771 /**
4772 * As setHook(), but letting the contents be parsed.
4773 *
4774 * Transparent tag hooks are like regular XML-style tag hooks, except they
4775 * operate late in the transformation sequence, on HTML instead of wikitext.
4776 *
4777 * This is probably obsoleted by things dealing with parser frames?
4778 * The only extension currently using it is geoserver.
4779 *
4780 * @since 1.10
4781 * @todo better document or deprecate this
4782 *
4783 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4784 * @param callable $callback The callback function (and object) to use for the tag
4785 * @throws MWException
4786 * @return callable|null The old value of the mTagHooks array associated with the hook
4787 */
4788 public function setTransparentTagHook( $tag, $callback ) {
4789 $tag = strtolower( $tag );
4790 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4791 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4792 }
4793 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4794 $this->mTransparentTagHooks[$tag] = $callback;
4795
4796 return $oldVal;
4797 }
4798
4799 /**
4800 * Remove all tag hooks
4801 */
4802 public function clearTagHooks() {
4803 $this->mTagHooks = [];
4804 $this->mFunctionTagHooks = [];
4805 $this->mStripList = $this->mDefaultStripList;
4806 }
4807
4808 /**
4809 * Create a function, e.g. {{sum:1|2|3}}
4810 * The callback function should have the form:
4811 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4812 *
4813 * Or with Parser::SFH_OBJECT_ARGS:
4814 * function myParserFunction( $parser, $frame, $args ) { ... }
4815 *
4816 * The callback may either return the text result of the function, or an array with the text
4817 * in element 0, and a number of flags in the other elements. The names of the flags are
4818 * specified in the keys. Valid flags are:
4819 * found The text returned is valid, stop processing the template. This
4820 * is on by default.
4821 * nowiki Wiki markup in the return value should be escaped
4822 * isHTML The returned text is HTML, armour it against wikitext transformation
4823 *
4824 * @param string $id The magic word ID
4825 * @param callable $callback The callback function (and object) to use
4826 * @param int $flags A combination of the following flags:
4827 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4828 *
4829 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4830 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4831 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4832 * the arguments, and to control the way they are expanded.
4833 *
4834 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4835 * arguments, for instance:
4836 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4837 *
4838 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4839 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4840 * working if/when this is changed.
4841 *
4842 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4843 * expansion.
4844 *
4845 * Please read the documentation in includes/parser/Preprocessor.php for more information
4846 * about the methods available in PPFrame and PPNode.
4847 *
4848 * @throws MWException
4849 * @return string|callable The old callback function for this name, if any
4850 */
4851 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4852 global $wgContLang;
4853
4854 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4855 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4856
4857 # Add to function cache
4858 $mw = MagicWord::get( $id );
4859 if ( !$mw ) {
4860 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4861 }
4862
4863 $synonyms = $mw->getSynonyms();
4864 $sensitive = intval( $mw->isCaseSensitive() );
4865
4866 foreach ( $synonyms as $syn ) {
4867 # Case
4868 if ( !$sensitive ) {
4869 $syn = $wgContLang->lc( $syn );
4870 }
4871 # Add leading hash
4872 if ( !( $flags & self::SFH_NO_HASH ) ) {
4873 $syn = '#' . $syn;
4874 }
4875 # Remove trailing colon
4876 if ( substr( $syn, -1, 1 ) === ':' ) {
4877 $syn = substr( $syn, 0, -1 );
4878 }
4879 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4880 }
4881 return $oldVal;
4882 }
4883
4884 /**
4885 * Get all registered function hook identifiers
4886 *
4887 * @return array
4888 */
4889 public function getFunctionHooks() {
4890 return array_keys( $this->mFunctionHooks );
4891 }
4892
4893 /**
4894 * Create a tag function, e.g. "<test>some stuff</test>".
4895 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4896 * Unlike parser functions, their content is not preprocessed.
4897 * @param string $tag
4898 * @param callable $callback
4899 * @param int $flags
4900 * @throws MWException
4901 * @return null
4902 */
4903 public function setFunctionTagHook( $tag, $callback, $flags ) {
4904 $tag = strtolower( $tag );
4905 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4906 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4907 }
4908 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4909 $this->mFunctionTagHooks[$tag] : null;
4910 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4911
4912 if ( !in_array( $tag, $this->mStripList ) ) {
4913 $this->mStripList[] = $tag;
4914 }
4915
4916 return $old;
4917 }
4918
4919 /**
4920 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4921 * Placeholders created in Linker::link()
4922 *
4923 * @param string $text
4924 * @param int $options
4925 */
4926 public function replaceLinkHolders( &$text, $options = 0 ) {
4927 $this->mLinkHolders->replace( $text );
4928 }
4929
4930 /**
4931 * Replace "<!--LINK-->" link placeholders with plain text of links
4932 * (not HTML-formatted).
4933 *
4934 * @param string $text
4935 * @return string
4936 */
4937 public function replaceLinkHoldersText( $text ) {
4938 return $this->mLinkHolders->replaceText( $text );
4939 }
4940
4941 /**
4942 * Renders an image gallery from a text with one line per image.
4943 * text labels may be given by using |-style alternative text. E.g.
4944 * Image:one.jpg|The number "1"
4945 * Image:tree.jpg|A tree
4946 * given as text will return the HTML of a gallery with two images,
4947 * labeled 'The number "1"' and
4948 * 'A tree'.
4949 *
4950 * @param string $text
4951 * @param array $params
4952 * @return string HTML
4953 */
4954 public function renderImageGallery( $text, $params ) {
4955
4956 $mode = false;
4957 if ( isset( $params['mode'] ) ) {
4958 $mode = $params['mode'];
4959 }
4960
4961 try {
4962 $ig = ImageGalleryBase::factory( $mode );
4963 } catch ( Exception $e ) {
4964 // If invalid type set, fallback to default.
4965 $ig = ImageGalleryBase::factory( false );
4966 }
4967
4968 $ig->setContextTitle( $this->mTitle );
4969 $ig->setShowBytes( false );
4970 $ig->setShowFilename( false );
4971 $ig->setParser( $this );
4972 $ig->setHideBadImages();
4973 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'ul' ) );
4974
4975 if ( isset( $params['showfilename'] ) ) {
4976 $ig->setShowFilename( true );
4977 } else {
4978 $ig->setShowFilename( false );
4979 }
4980 if ( isset( $params['caption'] ) ) {
4981 $caption = $params['caption'];
4982 $caption = htmlspecialchars( $caption );
4983 $caption = $this->replaceInternalLinks( $caption );
4984 $ig->setCaptionHtml( $caption );
4985 }
4986 if ( isset( $params['perrow'] ) ) {
4987 $ig->setPerRow( $params['perrow'] );
4988 }
4989 if ( isset( $params['widths'] ) ) {
4990 $ig->setWidths( $params['widths'] );
4991 }
4992 if ( isset( $params['heights'] ) ) {
4993 $ig->setHeights( $params['heights'] );
4994 }
4995 $ig->setAdditionalOptions( $params );
4996
4997 // Avoid PHP 7.1 warning from passing $this by reference
4998 $parser = $this;
4999 Hooks::run( 'BeforeParserrenderImageGallery', [ &$parser, &$ig ] );
5000
5001 $lines = StringUtils::explode( "\n", $text );
5002 foreach ( $lines as $line ) {
5003 # match lines like these:
5004 # Image:someimage.jpg|This is some image
5005 $matches = [];
5006 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5007 # Skip empty lines
5008 if ( count( $matches ) == 0 ) {
5009 continue;
5010 }
5011
5012 if ( strpos( $matches[0], '%' ) !== false ) {
5013 $matches[1] = rawurldecode( $matches[1] );
5014 }
5015 $title = Title::newFromText( $matches[1], NS_FILE );
5016 if ( is_null( $title ) ) {
5017 # Bogus title. Ignore these so we don't bomb out later.
5018 continue;
5019 }
5020
5021 # We need to get what handler the file uses, to figure out parameters.
5022 # Note, a hook can overide the file name, and chose an entirely different
5023 # file (which potentially could be of a different type and have different handler).
5024 $options = [];
5025 $descQuery = false;
5026 Hooks::run( 'BeforeParserFetchFileAndTitle',
5027 [ $this, $title, &$options, &$descQuery ] );
5028 # Don't register it now, as TraditionalImageGallery does that later.
5029 $file = $this->fetchFileNoRegister( $title, $options );
5030 $handler = $file ? $file->getHandler() : false;
5031
5032 $paramMap = [
5033 'img_alt' => 'gallery-internal-alt',
5034 'img_link' => 'gallery-internal-link',
5035 ];
5036 if ( $handler ) {
5037 $paramMap = $paramMap + $handler->getParamMap();
5038 // We don't want people to specify per-image widths.
5039 // Additionally the width parameter would need special casing anyhow.
5040 unset( $paramMap['img_width'] );
5041 }
5042
5043 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5044
5045 $label = '';
5046 $alt = '';
5047 $link = '';
5048 $handlerOptions = [];
5049 if ( isset( $matches[3] ) ) {
5050 // look for an |alt= definition while trying not to break existing
5051 // captions with multiple pipes (|) in it, until a more sensible grammar
5052 // is defined for images in galleries
5053
5054 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5055 // splitting on '|' is a bit odd, and different from makeImage.
5056 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5057 // Protect LanguageConverter markup
5058 $parameterMatches = StringUtils::delimiterExplode(
5059 '-{', '}-', '|', $matches[3], true /* nested */
5060 );
5061
5062 foreach ( $parameterMatches as $parameterMatch ) {
5063 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5064 if ( $magicName ) {
5065 $paramName = $paramMap[$magicName];
5066
5067 switch ( $paramName ) {
5068 case 'gallery-internal-alt':
5069 $alt = $this->stripAltText( $match, false );
5070 break;
5071 case 'gallery-internal-link':
5072 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5073 $chars = self::EXT_LINK_URL_CLASS;
5074 $addr = self::EXT_LINK_ADDR;
5075 $prots = $this->mUrlProtocols;
5076 // check to see if link matches an absolute url, if not then it must be a wiki link.
5077 if ( preg_match( '/^-{R|(.*)}-$/', $linkValue ) ) {
5078 // Result of LanguageConverter::markNoConversion
5079 // invoked on an external link.
5080 $linkValue = substr( $linkValue, 4, -2 );
5081 }
5082 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5083 $link = $linkValue;
5084 $this->mOutput->addExternalLink( $link );
5085 } else {
5086 $localLinkTitle = Title::newFromText( $linkValue );
5087 if ( $localLinkTitle !== null ) {
5088 $this->mOutput->addLink( $localLinkTitle );
5089 $link = $localLinkTitle->getLinkURL();
5090 }
5091 }
5092 break;
5093 default:
5094 // Must be a handler specific parameter.
5095 if ( $handler->validateParam( $paramName, $match ) ) {
5096 $handlerOptions[$paramName] = $match;
5097 } else {
5098 // Guess not, consider it as caption.
5099 wfDebug( "$parameterMatch failed parameter validation\n" );
5100 $label = '|' . $parameterMatch;
5101 }
5102 }
5103
5104 } else {
5105 // Last pipe wins.
5106 $label = '|' . $parameterMatch;
5107 }
5108 }
5109 // Remove the pipe.
5110 $label = substr( $label, 1 );
5111 }
5112
5113 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5114 }
5115 $html = $ig->toHTML();
5116 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5117 return $html;
5118 }
5119
5120 /**
5121 * @param MediaHandler $handler
5122 * @return array
5123 */
5124 public function getImageParams( $handler ) {
5125 if ( $handler ) {
5126 $handlerClass = get_class( $handler );
5127 } else {
5128 $handlerClass = '';
5129 }
5130 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5131 # Initialise static lists
5132 static $internalParamNames = [
5133 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5134 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5135 'bottom', 'text-bottom' ],
5136 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5137 'upright', 'border', 'link', 'alt', 'class' ],
5138 ];
5139 static $internalParamMap;
5140 if ( !$internalParamMap ) {
5141 $internalParamMap = [];
5142 foreach ( $internalParamNames as $type => $names ) {
5143 foreach ( $names as $name ) {
5144 // For grep: img_left, img_right, img_center, img_none,
5145 // img_baseline, img_sub, img_super, img_top, img_text_top, img_middle,
5146 // img_bottom, img_text_bottom,
5147 // img_thumbnail, img_manualthumb, img_framed, img_frameless, img_upright,
5148 // img_border, img_link, img_alt, img_class
5149 $magicName = str_replace( '-', '_', "img_$name" );
5150 $internalParamMap[$magicName] = [ $type, $name ];
5151 }
5152 }
5153 }
5154
5155 # Add handler params
5156 $paramMap = $internalParamMap;
5157 if ( $handler ) {
5158 $handlerParamMap = $handler->getParamMap();
5159 foreach ( $handlerParamMap as $magic => $paramName ) {
5160 $paramMap[$magic] = [ 'handler', $paramName ];
5161 }
5162 }
5163 $this->mImageParams[$handlerClass] = $paramMap;
5164 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5165 }
5166 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5167 }
5168
5169 /**
5170 * Parse image options text and use it to make an image
5171 *
5172 * @param Title $title
5173 * @param string $options
5174 * @param LinkHolderArray|bool $holders
5175 * @return string HTML
5176 */
5177 public function makeImage( $title, $options, $holders = false ) {
5178 # Check if the options text is of the form "options|alt text"
5179 # Options are:
5180 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5181 # * left no resizing, just left align. label is used for alt= only
5182 # * right same, but right aligned
5183 # * none same, but not aligned
5184 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5185 # * center center the image
5186 # * frame Keep original image size, no magnify-button.
5187 # * framed Same as "frame"
5188 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5189 # * upright reduce width for upright images, rounded to full __0 px
5190 # * border draw a 1px border around the image
5191 # * alt Text for HTML alt attribute (defaults to empty)
5192 # * class Set a class for img node
5193 # * link Set the target of the image link. Can be external, interwiki, or local
5194 # vertical-align values (no % or length right now):
5195 # * baseline
5196 # * sub
5197 # * super
5198 # * top
5199 # * text-top
5200 # * middle
5201 # * bottom
5202 # * text-bottom
5203
5204 # Protect LanguageConverter markup when splitting into parts
5205 $parts = StringUtils::delimiterExplode(
5206 '-{', '}-', '|', $options, true /* allow nesting */
5207 );
5208
5209 # Give extensions a chance to select the file revision for us
5210 $options = [];
5211 $descQuery = false;
5212 Hooks::run( 'BeforeParserFetchFileAndTitle',
5213 [ $this, $title, &$options, &$descQuery ] );
5214 # Fetch and register the file (file title may be different via hooks)
5215 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5216
5217 # Get parameter map
5218 $handler = $file ? $file->getHandler() : false;
5219
5220 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5221
5222 if ( !$file ) {
5223 $this->addTrackingCategory( 'broken-file-category' );
5224 }
5225
5226 # Process the input parameters
5227 $caption = '';
5228 $params = [ 'frame' => [], 'handler' => [],
5229 'horizAlign' => [], 'vertAlign' => [] ];
5230 $seenformat = false;
5231 foreach ( $parts as $part ) {
5232 $part = trim( $part );
5233 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5234 $validated = false;
5235 if ( isset( $paramMap[$magicName] ) ) {
5236 list( $type, $paramName ) = $paramMap[$magicName];
5237
5238 # Special case; width and height come in one variable together
5239 if ( $type === 'handler' && $paramName === 'width' ) {
5240 $parsedWidthParam = $this->parseWidthParam( $value );
5241 if ( isset( $parsedWidthParam['width'] ) ) {
5242 $width = $parsedWidthParam['width'];
5243 if ( $handler->validateParam( 'width', $width ) ) {
5244 $params[$type]['width'] = $width;
5245 $validated = true;
5246 }
5247 }
5248 if ( isset( $parsedWidthParam['height'] ) ) {
5249 $height = $parsedWidthParam['height'];
5250 if ( $handler->validateParam( 'height', $height ) ) {
5251 $params[$type]['height'] = $height;
5252 $validated = true;
5253 }
5254 }
5255 # else no validation -- T15436
5256 } else {
5257 if ( $type === 'handler' ) {
5258 # Validate handler parameter
5259 $validated = $handler->validateParam( $paramName, $value );
5260 } else {
5261 # Validate internal parameters
5262 switch ( $paramName ) {
5263 case 'manualthumb':
5264 case 'alt':
5265 case 'class':
5266 # @todo FIXME: Possibly check validity here for
5267 # manualthumb? downstream behavior seems odd with
5268 # missing manual thumbs.
5269 $validated = true;
5270 $value = $this->stripAltText( $value, $holders );
5271 break;
5272 case 'link':
5273 $chars = self::EXT_LINK_URL_CLASS;
5274 $addr = self::EXT_LINK_ADDR;
5275 $prots = $this->mUrlProtocols;
5276 if ( $value === '' ) {
5277 $paramName = 'no-link';
5278 $value = true;
5279 $validated = true;
5280 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5281 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5282 $paramName = 'link-url';
5283 $this->mOutput->addExternalLink( $value );
5284 if ( $this->mOptions->getExternalLinkTarget() ) {
5285 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5286 }
5287 $validated = true;
5288 }
5289 } else {
5290 $linkTitle = Title::newFromText( $value );
5291 if ( $linkTitle ) {
5292 $paramName = 'link-title';
5293 $value = $linkTitle;
5294 $this->mOutput->addLink( $linkTitle );
5295 $validated = true;
5296 }
5297 }
5298 break;
5299 case 'frameless':
5300 case 'framed':
5301 case 'thumbnail':
5302 // use first appearing option, discard others.
5303 $validated = !$seenformat;
5304 $seenformat = true;
5305 break;
5306 default:
5307 # Most other things appear to be empty or numeric...
5308 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5309 }
5310 }
5311
5312 if ( $validated ) {
5313 $params[$type][$paramName] = $value;
5314 }
5315 }
5316 }
5317 if ( !$validated ) {
5318 $caption = $part;
5319 }
5320 }
5321
5322 # Process alignment parameters
5323 if ( $params['horizAlign'] ) {
5324 $params['frame']['align'] = key( $params['horizAlign'] );
5325 }
5326 if ( $params['vertAlign'] ) {
5327 $params['frame']['valign'] = key( $params['vertAlign'] );
5328 }
5329
5330 $params['frame']['caption'] = $caption;
5331
5332 # Will the image be presented in a frame, with the caption below?
5333 $imageIsFramed = isset( $params['frame']['frame'] )
5334 || isset( $params['frame']['framed'] )
5335 || isset( $params['frame']['thumbnail'] )
5336 || isset( $params['frame']['manualthumb'] );
5337
5338 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5339 # came to also set the caption, ordinary text after the image -- which
5340 # makes no sense, because that just repeats the text multiple times in
5341 # screen readers. It *also* came to set the title attribute.
5342 # Now that we have an alt attribute, we should not set the alt text to
5343 # equal the caption: that's worse than useless, it just repeats the
5344 # text. This is the framed/thumbnail case. If there's no caption, we
5345 # use the unnamed parameter for alt text as well, just for the time be-
5346 # ing, if the unnamed param is set and the alt param is not.
5347 # For the future, we need to figure out if we want to tweak this more,
5348 # e.g., introducing a title= parameter for the title; ignoring the un-
5349 # named parameter entirely for images without a caption; adding an ex-
5350 # plicit caption= parameter and preserving the old magic unnamed para-
5351 # meter for BC; ...
5352 if ( $imageIsFramed ) { # Framed image
5353 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5354 # No caption or alt text, add the filename as the alt text so
5355 # that screen readers at least get some description of the image
5356 $params['frame']['alt'] = $title->getText();
5357 }
5358 # Do not set $params['frame']['title'] because tooltips don't make sense
5359 # for framed images
5360 } else { # Inline image
5361 if ( !isset( $params['frame']['alt'] ) ) {
5362 # No alt text, use the "caption" for the alt text
5363 if ( $caption !== '' ) {
5364 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5365 } else {
5366 # No caption, fall back to using the filename for the
5367 # alt text
5368 $params['frame']['alt'] = $title->getText();
5369 }
5370 }
5371 # Use the "caption" for the tooltip text
5372 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5373 }
5374
5375 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5376
5377 # Linker does the rest
5378 $time = isset( $options['time'] ) ? $options['time'] : false;
5379 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5380 $time, $descQuery, $this->mOptions->getThumbSize() );
5381
5382 # Give the handler a chance to modify the parser object
5383 if ( $handler ) {
5384 $handler->parserTransformHook( $this, $file );
5385 }
5386
5387 return $ret;
5388 }
5389
5390 /**
5391 * @param string $caption
5392 * @param LinkHolderArray|bool $holders
5393 * @return mixed|string
5394 */
5395 protected function stripAltText( $caption, $holders ) {
5396 # Strip bad stuff out of the title (tooltip). We can't just use
5397 # replaceLinkHoldersText() here, because if this function is called
5398 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5399 if ( $holders ) {
5400 $tooltip = $holders->replaceText( $caption );
5401 } else {
5402 $tooltip = $this->replaceLinkHoldersText( $caption );
5403 }
5404
5405 # make sure there are no placeholders in thumbnail attributes
5406 # that are later expanded to html- so expand them now and
5407 # remove the tags
5408 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5409 $tooltip = Sanitizer::stripAllTags( $tooltip );
5410
5411 return $tooltip;
5412 }
5413
5414 /**
5415 * Set a flag in the output object indicating that the content is dynamic and
5416 * shouldn't be cached.
5417 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5418 */
5419 public function disableCache() {
5420 wfDebug( "Parser output marked as uncacheable.\n" );
5421 if ( !$this->mOutput ) {
5422 throw new MWException( __METHOD__ .
5423 " can only be called when actually parsing something" );
5424 }
5425 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5426 }
5427
5428 /**
5429 * Callback from the Sanitizer for expanding items found in HTML attribute
5430 * values, so they can be safely tested and escaped.
5431 *
5432 * @param string $text
5433 * @param bool|PPFrame $frame
5434 * @return string
5435 */
5436 public function attributeStripCallback( &$text, $frame = false ) {
5437 $text = $this->replaceVariables( $text, $frame );
5438 $text = $this->mStripState->unstripBoth( $text );
5439 return $text;
5440 }
5441
5442 /**
5443 * Accessor
5444 *
5445 * @return array
5446 */
5447 public function getTags() {
5448 return array_merge(
5449 array_keys( $this->mTransparentTagHooks ),
5450 array_keys( $this->mTagHooks ),
5451 array_keys( $this->mFunctionTagHooks )
5452 );
5453 }
5454
5455 /**
5456 * Replace transparent tags in $text with the values given by the callbacks.
5457 *
5458 * Transparent tag hooks are like regular XML-style tag hooks, except they
5459 * operate late in the transformation sequence, on HTML instead of wikitext.
5460 *
5461 * @param string $text
5462 *
5463 * @return string
5464 */
5465 public function replaceTransparentTags( $text ) {
5466 $matches = [];
5467 $elements = array_keys( $this->mTransparentTagHooks );
5468 $text = self::extractTagsAndParams( $elements, $text, $matches );
5469 $replacements = [];
5470
5471 foreach ( $matches as $marker => $data ) {
5472 list( $element, $content, $params, $tag ) = $data;
5473 $tagName = strtolower( $element );
5474 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5475 $output = call_user_func_array(
5476 $this->mTransparentTagHooks[$tagName],
5477 [ $content, $params, $this ]
5478 );
5479 } else {
5480 $output = $tag;
5481 }
5482 $replacements[$marker] = $output;
5483 }
5484 return strtr( $text, $replacements );
5485 }
5486
5487 /**
5488 * Break wikitext input into sections, and either pull or replace
5489 * some particular section's text.
5490 *
5491 * External callers should use the getSection and replaceSection methods.
5492 *
5493 * @param string $text Page wikitext
5494 * @param string|int $sectionId A section identifier string of the form:
5495 * "<flag1> - <flag2> - ... - <section number>"
5496 *
5497 * Currently the only recognised flag is "T", which means the target section number
5498 * was derived during a template inclusion parse, in other words this is a template
5499 * section edit link. If no flags are given, it was an ordinary section edit link.
5500 * This flag is required to avoid a section numbering mismatch when a section is
5501 * enclosed by "<includeonly>" (T8563).
5502 *
5503 * The section number 0 pulls the text before the first heading; other numbers will
5504 * pull the given section along with its lower-level subsections. If the section is
5505 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5506 *
5507 * Section 0 is always considered to exist, even if it only contains the empty
5508 * string. If $text is the empty string and section 0 is replaced, $newText is
5509 * returned.
5510 *
5511 * @param string $mode One of "get" or "replace"
5512 * @param string $newText Replacement text for section data.
5513 * @return string For "get", the extracted section text.
5514 * for "replace", the whole page with the section replaced.
5515 */
5516 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5517 global $wgTitle; # not generally used but removes an ugly failure mode
5518
5519 $magicScopeVariable = $this->lock();
5520 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5521 $outText = '';
5522 $frame = $this->getPreprocessor()->newFrame();
5523
5524 # Process section extraction flags
5525 $flags = 0;
5526 $sectionParts = explode( '-', $sectionId );
5527 $sectionIndex = array_pop( $sectionParts );
5528 foreach ( $sectionParts as $part ) {
5529 if ( $part === 'T' ) {
5530 $flags |= self::PTD_FOR_INCLUSION;
5531 }
5532 }
5533
5534 # Check for empty input
5535 if ( strval( $text ) === '' ) {
5536 # Only sections 0 and T-0 exist in an empty document
5537 if ( $sectionIndex == 0 ) {
5538 if ( $mode === 'get' ) {
5539 return '';
5540 } else {
5541 return $newText;
5542 }
5543 } else {
5544 if ( $mode === 'get' ) {
5545 return $newText;
5546 } else {
5547 return $text;
5548 }
5549 }
5550 }
5551
5552 # Preprocess the text
5553 $root = $this->preprocessToDom( $text, $flags );
5554
5555 # <h> nodes indicate section breaks
5556 # They can only occur at the top level, so we can find them by iterating the root's children
5557 $node = $root->getFirstChild();
5558
5559 # Find the target section
5560 if ( $sectionIndex == 0 ) {
5561 # Section zero doesn't nest, level=big
5562 $targetLevel = 1000;
5563 } else {
5564 while ( $node ) {
5565 if ( $node->getName() === 'h' ) {
5566 $bits = $node->splitHeading();
5567 if ( $bits['i'] == $sectionIndex ) {
5568 $targetLevel = $bits['level'];
5569 break;
5570 }
5571 }
5572 if ( $mode === 'replace' ) {
5573 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5574 }
5575 $node = $node->getNextSibling();
5576 }
5577 }
5578
5579 if ( !$node ) {
5580 # Not found
5581 if ( $mode === 'get' ) {
5582 return $newText;
5583 } else {
5584 return $text;
5585 }
5586 }
5587
5588 # Find the end of the section, including nested sections
5589 do {
5590 if ( $node->getName() === 'h' ) {
5591 $bits = $node->splitHeading();
5592 $curLevel = $bits['level'];
5593 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5594 break;
5595 }
5596 }
5597 if ( $mode === 'get' ) {
5598 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5599 }
5600 $node = $node->getNextSibling();
5601 } while ( $node );
5602
5603 # Write out the remainder (in replace mode only)
5604 if ( $mode === 'replace' ) {
5605 # Output the replacement text
5606 # Add two newlines on -- trailing whitespace in $newText is conventionally
5607 # stripped by the editor, so we need both newlines to restore the paragraph gap
5608 # Only add trailing whitespace if there is newText
5609 if ( $newText != "" ) {
5610 $outText .= $newText . "\n\n";
5611 }
5612
5613 while ( $node ) {
5614 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5615 $node = $node->getNextSibling();
5616 }
5617 }
5618
5619 if ( is_string( $outText ) ) {
5620 # Re-insert stripped tags
5621 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5622 }
5623
5624 return $outText;
5625 }
5626
5627 /**
5628 * This function returns the text of a section, specified by a number ($section).
5629 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5630 * the first section before any such heading (section 0).
5631 *
5632 * If a section contains subsections, these are also returned.
5633 *
5634 * @param string $text Text to look in
5635 * @param string|int $sectionId Section identifier as a number or string
5636 * (e.g. 0, 1 or 'T-1').
5637 * @param string $defaultText Default to return if section is not found
5638 *
5639 * @return string Text of the requested section
5640 */
5641 public function getSection( $text, $sectionId, $defaultText = '' ) {
5642 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5643 }
5644
5645 /**
5646 * This function returns $oldtext after the content of the section
5647 * specified by $section has been replaced with $text. If the target
5648 * section does not exist, $oldtext is returned unchanged.
5649 *
5650 * @param string $oldText Former text of the article
5651 * @param string|int $sectionId Section identifier as a number or string
5652 * (e.g. 0, 1 or 'T-1').
5653 * @param string $newText Replacing text
5654 *
5655 * @return string Modified text
5656 */
5657 public function replaceSection( $oldText, $sectionId, $newText ) {
5658 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5659 }
5660
5661 /**
5662 * Get the ID of the revision we are parsing
5663 *
5664 * @return int|null
5665 */
5666 public function getRevisionId() {
5667 return $this->mRevisionId;
5668 }
5669
5670 /**
5671 * Get the revision object for $this->mRevisionId
5672 *
5673 * @return Revision|null Either a Revision object or null
5674 * @since 1.23 (public since 1.23)
5675 */
5676 public function getRevisionObject() {
5677 if ( !is_null( $this->mRevisionObject ) ) {
5678 return $this->mRevisionObject;
5679 }
5680 if ( is_null( $this->mRevisionId ) ) {
5681 return null;
5682 }
5683
5684 $rev = call_user_func(
5685 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5686 );
5687
5688 # If the parse is for a new revision, then the callback should have
5689 # already been set to force the object and should match mRevisionId.
5690 # If not, try to fetch by mRevisionId for sanity.
5691 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5692 $rev = Revision::newFromId( $this->mRevisionId );
5693 }
5694
5695 $this->mRevisionObject = $rev;
5696
5697 return $this->mRevisionObject;
5698 }
5699
5700 /**
5701 * Get the timestamp associated with the current revision, adjusted for
5702 * the default server-local timestamp
5703 * @return string
5704 */
5705 public function getRevisionTimestamp() {
5706 if ( is_null( $this->mRevisionTimestamp ) ) {
5707 global $wgContLang;
5708
5709 $revObject = $this->getRevisionObject();
5710 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5711
5712 # The cryptic '' timezone parameter tells to use the site-default
5713 # timezone offset instead of the user settings.
5714 # Since this value will be saved into the parser cache, served
5715 # to other users, and potentially even used inside links and such,
5716 # it needs to be consistent for all visitors.
5717 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5718
5719 }
5720 return $this->mRevisionTimestamp;
5721 }
5722
5723 /**
5724 * Get the name of the user that edited the last revision
5725 *
5726 * @return string User name
5727 */
5728 public function getRevisionUser() {
5729 if ( is_null( $this->mRevisionUser ) ) {
5730 $revObject = $this->getRevisionObject();
5731
5732 # if this template is subst: the revision id will be blank,
5733 # so just use the current user's name
5734 if ( $revObject ) {
5735 $this->mRevisionUser = $revObject->getUserText();
5736 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5737 $this->mRevisionUser = $this->getUser()->getName();
5738 }
5739 }
5740 return $this->mRevisionUser;
5741 }
5742
5743 /**
5744 * Get the size of the revision
5745 *
5746 * @return int|null Revision size
5747 */
5748 public function getRevisionSize() {
5749 if ( is_null( $this->mRevisionSize ) ) {
5750 $revObject = $this->getRevisionObject();
5751
5752 # if this variable is subst: the revision id will be blank,
5753 # so just use the parser input size, because the own substituation
5754 # will change the size.
5755 if ( $revObject ) {
5756 $this->mRevisionSize = $revObject->getSize();
5757 } else {
5758 $this->mRevisionSize = $this->mInputSize;
5759 }
5760 }
5761 return $this->mRevisionSize;
5762 }
5763
5764 /**
5765 * Mutator for $mDefaultSort
5766 *
5767 * @param string $sort New value
5768 */
5769 public function setDefaultSort( $sort ) {
5770 $this->mDefaultSort = $sort;
5771 $this->mOutput->setProperty( 'defaultsort', $sort );
5772 }
5773
5774 /**
5775 * Accessor for $mDefaultSort
5776 * Will use the empty string if none is set.
5777 *
5778 * This value is treated as a prefix, so the
5779 * empty string is equivalent to sorting by
5780 * page name.
5781 *
5782 * @return string
5783 */
5784 public function getDefaultSort() {
5785 if ( $this->mDefaultSort !== false ) {
5786 return $this->mDefaultSort;
5787 } else {
5788 return '';
5789 }
5790 }
5791
5792 /**
5793 * Accessor for $mDefaultSort
5794 * Unlike getDefaultSort(), will return false if none is set
5795 *
5796 * @return string|bool
5797 */
5798 public function getCustomDefaultSort() {
5799 return $this->mDefaultSort;
5800 }
5801
5802 /**
5803 * Try to guess the section anchor name based on a wikitext fragment
5804 * presumably extracted from a heading, for example "Header" from
5805 * "== Header ==".
5806 *
5807 * @param string $text
5808 *
5809 * @return string
5810 */
5811 public function guessSectionNameFromWikiText( $text ) {
5812 # Strip out wikitext links(they break the anchor)
5813 $text = $this->stripSectionName( $text );
5814 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5815 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5816 }
5817
5818 /**
5819 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5820 * instead. For use in redirects, since IE6 interprets Redirect: headers
5821 * as something other than UTF-8 (apparently?), resulting in breakage.
5822 *
5823 * @param string $text The section name
5824 * @return string An anchor
5825 */
5826 public function guessLegacySectionNameFromWikiText( $text ) {
5827 # Strip out wikitext links(they break the anchor)
5828 $text = $this->stripSectionName( $text );
5829 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5830 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5831 }
5832
5833 /**
5834 * Strips a text string of wikitext for use in a section anchor
5835 *
5836 * Accepts a text string and then removes all wikitext from the
5837 * string and leaves only the resultant text (i.e. the result of
5838 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5839 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5840 * to create valid section anchors by mimicing the output of the
5841 * parser when headings are parsed.
5842 *
5843 * @param string $text Text string to be stripped of wikitext
5844 * for use in a Section anchor
5845 * @return string Filtered text string
5846 */
5847 public function stripSectionName( $text ) {
5848 # Strip internal link markup
5849 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5850 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5851
5852 # Strip external link markup
5853 # @todo FIXME: Not tolerant to blank link text
5854 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5855 # on how many empty links there are on the page - need to figure that out.
5856 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5857
5858 # Parse wikitext quotes (italics & bold)
5859 $text = $this->doQuotes( $text );
5860
5861 # Strip HTML tags
5862 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5863 return $text;
5864 }
5865
5866 /**
5867 * strip/replaceVariables/unstrip for preprocessor regression testing
5868 *
5869 * @param string $text
5870 * @param Title $title
5871 * @param ParserOptions $options
5872 * @param int $outputType
5873 *
5874 * @return string
5875 */
5876 public function testSrvus( $text, Title $title, ParserOptions $options,
5877 $outputType = self::OT_HTML
5878 ) {
5879 $magicScopeVariable = $this->lock();
5880 $this->startParse( $title, $options, $outputType, true );
5881
5882 $text = $this->replaceVariables( $text );
5883 $text = $this->mStripState->unstripBoth( $text );
5884 $text = Sanitizer::removeHTMLtags( $text );
5885 return $text;
5886 }
5887
5888 /**
5889 * @param string $text
5890 * @param Title $title
5891 * @param ParserOptions $options
5892 * @return string
5893 */
5894 public function testPst( $text, Title $title, ParserOptions $options ) {
5895 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5896 }
5897
5898 /**
5899 * @param string $text
5900 * @param Title $title
5901 * @param ParserOptions $options
5902 * @return string
5903 */
5904 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5905 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5906 }
5907
5908 /**
5909 * Call a callback function on all regions of the given text that are not
5910 * inside strip markers, and replace those regions with the return value
5911 * of the callback. For example, with input:
5912 *
5913 * aaa<MARKER>bbb
5914 *
5915 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5916 * two strings will be replaced with the value returned by the callback in
5917 * each case.
5918 *
5919 * @param string $s
5920 * @param callable $callback
5921 *
5922 * @return string
5923 */
5924 public function markerSkipCallback( $s, $callback ) {
5925 $i = 0;
5926 $out = '';
5927 while ( $i < strlen( $s ) ) {
5928 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5929 if ( $markerStart === false ) {
5930 $out .= call_user_func( $callback, substr( $s, $i ) );
5931 break;
5932 } else {
5933 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5934 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5935 if ( $markerEnd === false ) {
5936 $out .= substr( $s, $markerStart );
5937 break;
5938 } else {
5939 $markerEnd += strlen( self::MARKER_SUFFIX );
5940 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5941 $i = $markerEnd;
5942 }
5943 }
5944 }
5945 return $out;
5946 }
5947
5948 /**
5949 * Remove any strip markers found in the given text.
5950 *
5951 * @param string $text Input string
5952 * @return string
5953 */
5954 public function killMarkers( $text ) {
5955 return $this->mStripState->killMarkers( $text );
5956 }
5957
5958 /**
5959 * Save the parser state required to convert the given half-parsed text to
5960 * HTML. "Half-parsed" in this context means the output of
5961 * recursiveTagParse() or internalParse(). This output has strip markers
5962 * from replaceVariables (extensionSubstitution() etc.), and link
5963 * placeholders from replaceLinkHolders().
5964 *
5965 * Returns an array which can be serialized and stored persistently. This
5966 * array can later be loaded into another parser instance with
5967 * unserializeHalfParsedText(). The text can then be safely incorporated into
5968 * the return value of a parser hook.
5969 *
5970 * @param string $text
5971 *
5972 * @return array
5973 */
5974 public function serializeHalfParsedText( $text ) {
5975 $data = [
5976 'text' => $text,
5977 'version' => self::HALF_PARSED_VERSION,
5978 'stripState' => $this->mStripState->getSubState( $text ),
5979 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5980 ];
5981 return $data;
5982 }
5983
5984 /**
5985 * Load the parser state given in the $data array, which is assumed to
5986 * have been generated by serializeHalfParsedText(). The text contents is
5987 * extracted from the array, and its markers are transformed into markers
5988 * appropriate for the current Parser instance. This transformed text is
5989 * returned, and can be safely included in the return value of a parser
5990 * hook.
5991 *
5992 * If the $data array has been stored persistently, the caller should first
5993 * check whether it is still valid, by calling isValidHalfParsedText().
5994 *
5995 * @param array $data Serialized data
5996 * @throws MWException
5997 * @return string
5998 */
5999 public function unserializeHalfParsedText( $data ) {
6000 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6001 throw new MWException( __METHOD__ . ': invalid version' );
6002 }
6003
6004 # First, extract the strip state.
6005 $texts = [ $data['text'] ];
6006 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6007
6008 # Now renumber links
6009 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6010
6011 # Should be good to go.
6012 return $texts[0];
6013 }
6014
6015 /**
6016 * Returns true if the given array, presumed to be generated by
6017 * serializeHalfParsedText(), is compatible with the current version of the
6018 * parser.
6019 *
6020 * @param array $data
6021 *
6022 * @return bool
6023 */
6024 public function isValidHalfParsedText( $data ) {
6025 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6026 }
6027
6028 /**
6029 * Parsed a width param of imagelink like 300px or 200x300px
6030 *
6031 * @param string $value
6032 *
6033 * @return array
6034 * @since 1.20
6035 */
6036 public function parseWidthParam( $value ) {
6037 $parsedWidthParam = [];
6038 if ( $value === '' ) {
6039 return $parsedWidthParam;
6040 }
6041 $m = [];
6042 # (T15500) In both cases (width/height and width only),
6043 # permit trailing "px" for backward compatibility.
6044 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6045 $width = intval( $m[1] );
6046 $height = intval( $m[2] );
6047 $parsedWidthParam['width'] = $width;
6048 $parsedWidthParam['height'] = $height;
6049 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6050 $width = intval( $value );
6051 $parsedWidthParam['width'] = $width;
6052 }
6053 return $parsedWidthParam;
6054 }
6055
6056 /**
6057 * Lock the current instance of the parser.
6058 *
6059 * This is meant to stop someone from calling the parser
6060 * recursively and messing up all the strip state.
6061 *
6062 * @throws MWException If parser is in a parse
6063 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6064 */
6065 protected function lock() {
6066 if ( $this->mInParse ) {
6067 throw new MWException( "Parser state cleared while parsing. "
6068 . "Did you call Parser::parse recursively?" );
6069 }
6070 $this->mInParse = true;
6071
6072 $recursiveCheck = new ScopedCallback( function() {
6073 $this->mInParse = false;
6074 } );
6075
6076 return $recursiveCheck;
6077 }
6078
6079 /**
6080 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6081 *
6082 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6083 * or if there is more than one <p/> tag in the input HTML.
6084 *
6085 * @param string $html
6086 * @return string
6087 * @since 1.24
6088 */
6089 public static function stripOuterParagraph( $html ) {
6090 $m = [];
6091 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6092 if ( strpos( $m[1], '</p>' ) === false ) {
6093 $html = $m[1];
6094 }
6095 }
6096
6097 return $html;
6098 }
6099
6100 /**
6101 * Return this parser if it is not doing anything, otherwise
6102 * get a fresh parser. You can use this method by doing
6103 * $myParser = $wgParser->getFreshParser(), or more simply
6104 * $wgParser->getFreshParser()->parse( ... );
6105 * if you're unsure if $wgParser is safe to use.
6106 *
6107 * @since 1.24
6108 * @return Parser A parser object that is not parsing anything
6109 */
6110 public function getFreshParser() {
6111 global $wgParserConf;
6112 if ( $this->mInParse ) {
6113 return new $wgParserConf['class']( $wgParserConf );
6114 } else {
6115 return $this;
6116 }
6117 }
6118
6119 /**
6120 * Set's up the PHP implementation of OOUI for use in this request
6121 * and instructs OutputPage to enable OOUI for itself.
6122 *
6123 * @since 1.26
6124 */
6125 public function enableOOUI() {
6126 OutputPage::setupOOUI();
6127 $this->mOutput->setEnableOOUI( true );
6128 }
6129 }
Wiklou, le Wiki du Wiklou