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parser: Update outdated comment about ImageGallery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