dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Hide marked empty elements by default (stage 1)
[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
26 /**
27 * @defgroup Parser Parser
28 */
29
30 /**
31 * PHP Parser - Processes wiki markup (which uses a more user-friendly
32 * syntax, such as "[[link]]" for making links), and provides a one-way
33 * transformation of that wiki markup it into (X)HTML output / markup
34 * (which in turn the browser understands, and can display).
35 *
36 * There are seven main entry points into the Parser class:
37 *
38 * - Parser::parse()
39 * produces HTML output
40 * - Parser::preSaveTransform()
41 * produces altered wiki markup
42 * - Parser::preprocess()
43 * removes HTML comments and expands templates
44 * - Parser::cleanSig() and Parser::cleanSigInSig()
45 * cleans a signature before saving it to preferences
46 * - Parser::getSection()
47 * return the content of a section from an article for section editing
48 * - Parser::replaceSection()
49 * replaces a section by number inside an article
50 * - Parser::getPreloadText()
51 * removes <noinclude> sections and <includeonly> tags
52 *
53 * Globals used:
54 * object: $wgContLang
55 *
56 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
57 *
58 * @par Settings:
59 * $wgNamespacesWithSubpages
60 *
61 * @par Settings only within ParserOptions:
62 * $wgAllowExternalImages
63 * $wgAllowSpecialInclusion
64 * $wgInterwikiMagic
65 * $wgMaxArticleSize
66 *
67 * @ingroup Parser
68 */
69 class Parser {
70 /**
71 * Update this version number when the ParserOutput format
72 * changes in an incompatible way, so the parser cache
73 * can automatically discard old data.
74 */
75 const VERSION = '1.6.4';
76
77 /**
78 * Update this version number when the output of serialiseHalfParsedText()
79 * changes in an incompatible way
80 */
81 const HALF_PARSED_VERSION = 2;
82
83 # Flags for Parser::setFunctionHook
84 const SFH_NO_HASH = 1;
85 const SFH_OBJECT_ARGS = 2;
86
87 # Constants needed for external link processing
88 # Everything except bracket, space, or control characters
89 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
90 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
91 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
92 # Simplified expression to match an IPv4 or IPv6 address, or
93 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
94 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
95 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
96 // @codingStandardsIgnoreStart Generic.Files.LineLength
97 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
98 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
99 // @codingStandardsIgnoreEnd
100
101 # Regular expression for a non-newline space
102 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
103
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION = 1;
106
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML = 1; # like parse()
110 const OT_WIKI = 2; # like preSaveTransform()
111 const OT_PREPROCESS = 3; # like preprocess()
112 const OT_MSG = 3;
113 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
114
115 /**
116 * @var string Prefix and suffix for temporary replacement strings
117 * for the multipass parser.
118 *
119 * \x7f should never appear in input as it's disallowed in XML.
120 * Using it at the front also gives us a little extra robustness
121 * since it shouldn't match when butted up against identifier-like
122 * string constructs.
123 *
124 * Must not consist of all title characters, or else it will change
125 * the behavior of <nowiki> in a link.
126 *
127 * Must have a character that needs escaping in attributes, otherwise
128 * someone could put a strip marker in an attribute, to get around
129 * escaping quote marks, and break out of the attribute. Thus we add
130 * `'".
131 */
132 const MARKER_SUFFIX = "-QINU`\"'\x7f";
133 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
134
135 # Markers used for wrapping the table of contents
136 const TOC_START = '<mw:toc>';
137 const TOC_END = '</mw:toc>';
138
139 # Persistent:
140 public $mTagHooks = [];
141 public $mTransparentTagHooks = [];
142 public $mFunctionHooks = [];
143 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
144 public $mFunctionTagHooks = [];
145 public $mStripList = [];
146 public $mDefaultStripList = [];
147 public $mVarCache = [];
148 public $mImageParams = [];
149 public $mImageParamsMagicArray = [];
150 public $mMarkerIndex = 0;
151 public $mFirstCall = true;
152
153 # Initialised by initialiseVariables()
154
155 /**
156 * @var MagicWordArray
157 */
158 public $mVariables;
159
160 /**
161 * @var MagicWordArray
162 */
163 public $mSubstWords;
164 # Initialised in constructor
165 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
166
167 # Initialized in getPreprocessor()
168 /** @var Preprocessor */
169 public $mPreprocessor;
170
171 # Cleared with clearState():
172 /**
173 * @var ParserOutput
174 */
175 public $mOutput;
176 public $mAutonumber;
177
178 /**
179 * @var StripState
180 */
181 public $mStripState;
182
183 public $mIncludeCount;
184 /**
185 * @var LinkHolderArray
186 */
187 public $mLinkHolders;
188
189 public $mLinkID;
190 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
191 public $mDefaultSort;
192 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
193 public $mExpensiveFunctionCount; # number of expensive parser function calls
194 public $mShowToc, $mForceTocPosition;
195
196 /**
197 * @var User
198 */
199 public $mUser; # User object; only used when doing pre-save transform
200
201 # Temporary
202 # These are variables reset at least once per parse regardless of $clearState
203
204 /**
205 * @var ParserOptions
206 */
207 public $mOptions;
208
209 /**
210 * @var Title
211 */
212 public $mTitle; # Title context, used for self-link rendering and similar things
213 public $mOutputType; # Output type, one of the OT_xxx constants
214 public $ot; # Shortcut alias, see setOutputType()
215 public $mRevisionObject; # The revision object of the specified revision ID
216 public $mRevisionId; # ID to display in {{REVISIONID}} tags
217 public $mRevisionTimestamp; # The timestamp of the specified revision ID
218 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
219 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
220 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
221 public $mInputSize = false; # For {{PAGESIZE}} on current page.
222
223 /**
224 * @var string Deprecated accessor for the strip marker prefix.
225 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
226 **/
227 public $mUniqPrefix = Parser::MARKER_PREFIX;
228
229 /**
230 * @var array Array with the language name of each language link (i.e. the
231 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
232 * duplicate language links to the ParserOutput.
233 */
234 public $mLangLinkLanguages;
235
236 /**
237 * @var MapCacheLRU|null
238 * @since 1.24
239 *
240 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
241 */
242 public $currentRevisionCache;
243
244 /**
245 * @var bool Recursive call protection.
246 * This variable should be treated as if it were private.
247 */
248 public $mInParse = false;
249
250 /** @var SectionProfiler */
251 protected $mProfiler;
252
253 /**
254 * @var LinkRenderer
255 */
256 protected $mLinkRenderer;
257
258 /**
259 * @param array $conf
260 */
261 public function __construct( $conf = [] ) {
262 $this->mConf = $conf;
263 $this->mUrlProtocols = wfUrlProtocols();
264 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
265 self::EXT_LINK_ADDR .
266 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
267 if ( isset( $conf['preprocessorClass'] ) ) {
268 $this->mPreprocessorClass = $conf['preprocessorClass'];
269 } elseif ( defined( 'HPHP_VERSION' ) ) {
270 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
271 $this->mPreprocessorClass = 'Preprocessor_Hash';
272 } elseif ( extension_loaded( 'domxml' ) ) {
273 # PECL extension that conflicts with the core DOM extension (bug 13770)
274 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
275 $this->mPreprocessorClass = 'Preprocessor_Hash';
276 } elseif ( extension_loaded( 'dom' ) ) {
277 $this->mPreprocessorClass = 'Preprocessor_DOM';
278 } else {
279 $this->mPreprocessorClass = 'Preprocessor_Hash';
280 }
281 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
282 }
283
284 /**
285 * Reduce memory usage to reduce the impact of circular references
286 */
287 public function __destruct() {
288 if ( isset( $this->mLinkHolders ) ) {
289 unset( $this->mLinkHolders );
290 }
291 foreach ( $this as $name => $value ) {
292 unset( $this->$name );
293 }
294 }
295
296 /**
297 * Allow extensions to clean up when the parser is cloned
298 */
299 public function __clone() {
300 $this->mInParse = false;
301
302 // Bug 56226: When you create a reference "to" an object field, that
303 // makes the object field itself be a reference too (until the other
304 // reference goes out of scope). When cloning, any field that's a
305 // reference is copied as a reference in the new object. Both of these
306 // are defined PHP5 behaviors, as inconvenient as it is for us when old
307 // hooks from PHP4 days are passing fields by reference.
308 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
309 // Make a non-reference copy of the field, then rebind the field to
310 // reference the new copy.
311 $tmp = $this->$k;
312 $this->$k =& $tmp;
313 unset( $tmp );
314 }
315
316 Hooks::run( 'ParserCloned', [ $this ] );
317 }
318
319 /**
320 * Do various kinds of initialisation on the first call of the parser
321 */
322 public function firstCallInit() {
323 if ( !$this->mFirstCall ) {
324 return;
325 }
326 $this->mFirstCall = false;
327
328 CoreParserFunctions::register( $this );
329 CoreTagHooks::register( $this );
330 $this->initialiseVariables();
331
332 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
333 }
334
335 /**
336 * Clear Parser state
337 *
338 * @private
339 */
340 public function clearState() {
341 if ( $this->mFirstCall ) {
342 $this->firstCallInit();
343 }
344 $this->mOutput = new ParserOutput;
345 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
346 $this->mAutonumber = 0;
347 $this->mIncludeCount = [];
348 $this->mLinkHolders = new LinkHolderArray( $this );
349 $this->mLinkID = 0;
350 $this->mRevisionObject = $this->mRevisionTimestamp =
351 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
352 $this->mVarCache = [];
353 $this->mUser = null;
354 $this->mLangLinkLanguages = [];
355 $this->currentRevisionCache = null;
356
357 $this->mStripState = new StripState;
358
359 # Clear these on every parse, bug 4549
360 $this->mTplRedirCache = $this->mTplDomCache = [];
361
362 $this->mShowToc = true;
363 $this->mForceTocPosition = false;
364 $this->mIncludeSizes = [
365 'post-expand' => 0,
366 'arg' => 0,
367 ];
368 $this->mPPNodeCount = 0;
369 $this->mGeneratedPPNodeCount = 0;
370 $this->mHighestExpansionDepth = 0;
371 $this->mDefaultSort = false;
372 $this->mHeadings = [];
373 $this->mDoubleUnderscores = [];
374 $this->mExpensiveFunctionCount = 0;
375
376 # Fix cloning
377 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
378 $this->mPreprocessor = null;
379 }
380
381 $this->mProfiler = new SectionProfiler();
382
383 Hooks::run( 'ParserClearState', [ &$this ] );
384 }
385
386 /**
387 * Convert wikitext to HTML
388 * Do not call this function recursively.
389 *
390 * @param string $text Text we want to parse
391 * @param Title $title
392 * @param ParserOptions $options
393 * @param bool $linestart
394 * @param bool $clearState
395 * @param int $revid Number to pass in {{REVISIONID}}
396 * @return ParserOutput A ParserOutput
397 */
398 public function parse( $text, Title $title, ParserOptions $options,
399 $linestart = true, $clearState = true, $revid = null
400 ) {
401 /**
402 * First pass--just handle <nowiki> sections, pass the rest off
403 * to internalParse() which does all the real work.
404 */
405
406 global $wgShowHostnames;
407
408 if ( $clearState ) {
409 // We use U+007F DELETE to construct strip markers, so we have to make
410 // sure that this character does not occur in the input text.
411 $text = strtr( $text, "\x7f", "?" );
412 $magicScopeVariable = $this->lock();
413 }
414
415 $this->startParse( $title, $options, self::OT_HTML, $clearState );
416
417 $this->currentRevisionCache = null;
418 $this->mInputSize = strlen( $text );
419 if ( $this->mOptions->getEnableLimitReport() ) {
420 $this->mOutput->resetParseStartTime();
421 }
422
423 $oldRevisionId = $this->mRevisionId;
424 $oldRevisionObject = $this->mRevisionObject;
425 $oldRevisionTimestamp = $this->mRevisionTimestamp;
426 $oldRevisionUser = $this->mRevisionUser;
427 $oldRevisionSize = $this->mRevisionSize;
428 if ( $revid !== null ) {
429 $this->mRevisionId = $revid;
430 $this->mRevisionObject = null;
431 $this->mRevisionTimestamp = null;
432 $this->mRevisionUser = null;
433 $this->mRevisionSize = null;
434 }
435
436 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
437 # No more strip!
438 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
439 $text = $this->internalParse( $text );
440 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
441
442 $text = $this->internalParseHalfParsed( $text, true, $linestart );
443
444 /**
445 * A converted title will be provided in the output object if title and
446 * content conversion are enabled, the article text does not contain
447 * a conversion-suppressing double-underscore tag, and no
448 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
449 * automatic link conversion.
450 */
451 if ( !( $options->getDisableTitleConversion()
452 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
453 || isset( $this->mDoubleUnderscores['notitleconvert'] )
454 || $this->mOutput->getDisplayTitle() !== false )
455 ) {
456 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
457 if ( $convruletitle ) {
458 $this->mOutput->setTitleText( $convruletitle );
459 } else {
460 $titleText = $this->getConverterLanguage()->convertTitle( $title );
461 $this->mOutput->setTitleText( $titleText );
462 }
463 }
464
465 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
466 $this->limitationWarn( 'expensive-parserfunction',
467 $this->mExpensiveFunctionCount,
468 $this->mOptions->getExpensiveParserFunctionLimit()
469 );
470 }
471
472 # Information on include size limits, for the benefit of users who try to skirt them
473 if ( $this->mOptions->getEnableLimitReport() ) {
474 $max = $this->mOptions->getMaxIncludeSize();
475
476 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
477 if ( $cpuTime !== null ) {
478 $this->mOutput->setLimitReportData( 'limitreport-cputime',
479 sprintf( "%.3f", $cpuTime )
480 );
481 }
482
483 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
484 $this->mOutput->setLimitReportData( 'limitreport-walltime',
485 sprintf( "%.3f", $wallTime )
486 );
487
488 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
489 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
490 );
491 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
492 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
493 );
494 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
495 [ $this->mIncludeSizes['post-expand'], $max ]
496 );
497 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
498 [ $this->mIncludeSizes['arg'], $max ]
499 );
500 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
501 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
502 );
503 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
504 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
505 );
506 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
507
508 $limitReport = "NewPP limit report\n";
509 if ( $wgShowHostnames ) {
510 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
511 }
512 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
513 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
514 $limitReport .= 'Dynamic content: ' .
515 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
516 "\n";
517
518 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
519 if ( Hooks::run( 'ParserLimitReportFormat',
520 [ $key, &$value, &$limitReport, false, false ]
521 ) ) {
522 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
523 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
524 ->inLanguage( 'en' )->useDatabase( false );
525 if ( !$valueMsg->exists() ) {
526 $valueMsg = new RawMessage( '$1' );
527 }
528 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
529 $valueMsg->params( $value );
530 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
531 }
532 }
533 }
534 // Since we're not really outputting HTML, decode the entities and
535 // then re-encode the things that need hiding inside HTML comments.
536 $limitReport = htmlspecialchars_decode( $limitReport );
537 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
538
539 // Sanitize for comment. Note '‐' in the replacement is U+2010,
540 // which looks much like the problematic '-'.
541 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
542 $text .= "\n<!-- \n$limitReport-->\n";
543
544 // Add on template profiling data
545 $dataByFunc = $this->mProfiler->getFunctionStats();
546 uasort( $dataByFunc, function ( $a, $b ) {
547 return $a['real'] < $b['real']; // descending order
548 } );
549 $profileReport = "Transclusion expansion time report (%,ms,calls,template)\n";
550 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
551 $profileReport .= sprintf( "%6.2f%% %8.3f %6d - %s\n",
552 $item['%real'], $item['real'], $item['calls'],
553 htmlspecialchars( $item['name'] ) );
554 }
555 $text .= "\n<!-- \n$profileReport-->\n";
556
557 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
558 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
559 $this->mTitle->getPrefixedDBkey() );
560 }
561 }
562 $this->mOutput->setText( $text );
563
564 $this->mRevisionId = $oldRevisionId;
565 $this->mRevisionObject = $oldRevisionObject;
566 $this->mRevisionTimestamp = $oldRevisionTimestamp;
567 $this->mRevisionUser = $oldRevisionUser;
568 $this->mRevisionSize = $oldRevisionSize;
569 $this->mInputSize = false;
570 $this->currentRevisionCache = null;
571
572 return $this->mOutput;
573 }
574
575 /**
576 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
577 * can be called from an extension tag hook.
578 *
579 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
580 * instead, which means that lists and links have not been fully parsed yet,
581 * and strip markers are still present.
582 *
583 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
584 *
585 * Use this function if you're a parser tag hook and you want to parse
586 * wikitext before or after applying additional transformations, and you
587 * intend to *return the result as hook output*, which will cause it to go
588 * through the rest of parsing process automatically.
589 *
590 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
591 * $text are not expanded
592 *
593 * @param string $text Text extension wants to have parsed
594 * @param bool|PPFrame $frame The frame to use for expanding any template variables
595 * @return string UNSAFE half-parsed HTML
596 */
597 public function recursiveTagParse( $text, $frame = false ) {
598 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
599 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
600 $text = $this->internalParse( $text, false, $frame );
601 return $text;
602 }
603
604 /**
605 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
606 * point can be called from an extension tag hook.
607 *
608 * The output of this function is fully-parsed HTML that is safe for output.
609 * If you're a parser tag hook, you might want to use recursiveTagParse()
610 * instead.
611 *
612 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
613 * $text are not expanded
614 *
615 * @since 1.25
616 *
617 * @param string $text Text extension wants to have parsed
618 * @param bool|PPFrame $frame The frame to use for expanding any template variables
619 * @return string Fully parsed HTML
620 */
621 public function recursiveTagParseFully( $text, $frame = false ) {
622 $text = $this->recursiveTagParse( $text, $frame );
623 $text = $this->internalParseHalfParsed( $text, false );
624 return $text;
625 }
626
627 /**
628 * Expand templates and variables in the text, producing valid, static wikitext.
629 * Also removes comments.
630 * Do not call this function recursively.
631 * @param string $text
632 * @param Title $title
633 * @param ParserOptions $options
634 * @param int|null $revid
635 * @param bool|PPFrame $frame
636 * @return mixed|string
637 */
638 public function preprocess( $text, Title $title = null,
639 ParserOptions $options, $revid = null, $frame = false
640 ) {
641 $magicScopeVariable = $this->lock();
642 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
643 if ( $revid !== null ) {
644 $this->mRevisionId = $revid;
645 }
646 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
647 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
648 $text = $this->replaceVariables( $text, $frame );
649 $text = $this->mStripState->unstripBoth( $text );
650 return $text;
651 }
652
653 /**
654 * Recursive parser entry point that can be called from an extension tag
655 * hook.
656 *
657 * @param string $text Text to be expanded
658 * @param bool|PPFrame $frame The frame to use for expanding any template variables
659 * @return string
660 * @since 1.19
661 */
662 public function recursivePreprocess( $text, $frame = false ) {
663 $text = $this->replaceVariables( $text, $frame );
664 $text = $this->mStripState->unstripBoth( $text );
665 return $text;
666 }
667
668 /**
669 * Process the wikitext for the "?preload=" feature. (bug 5210)
670 *
671 * "<noinclude>", "<includeonly>" etc. are parsed as for template
672 * transclusion, comments, templates, arguments, tags hooks and parser
673 * functions are untouched.
674 *
675 * @param string $text
676 * @param Title $title
677 * @param ParserOptions $options
678 * @param array $params
679 * @return string
680 */
681 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
682 $msg = new RawMessage( $text );
683 $text = $msg->params( $params )->plain();
684
685 # Parser (re)initialisation
686 $magicScopeVariable = $this->lock();
687 $this->startParse( $title, $options, self::OT_PLAIN, true );
688
689 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
690 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
691 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
692 $text = $this->mStripState->unstripBoth( $text );
693 return $text;
694 }
695
696 /**
697 * Get a random string
698 *
699 * @return string
700 * @deprecated since 1.26; use wfRandomString() instead.
701 */
702 public static function getRandomString() {
703 wfDeprecated( __METHOD__, '1.26' );
704 return wfRandomString( 16 );
705 }
706
707 /**
708 * Set the current user.
709 * Should only be used when doing pre-save transform.
710 *
711 * @param User|null $user User object or null (to reset)
712 */
713 public function setUser( $user ) {
714 $this->mUser = $user;
715 }
716
717 /**
718 * Accessor for mUniqPrefix.
719 *
720 * @return string
721 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
722 */
723 public function uniqPrefix() {
724 wfDeprecated( __METHOD__, '1.26' );
725 return self::MARKER_PREFIX;
726 }
727
728 /**
729 * Set the context title
730 *
731 * @param Title $t
732 */
733 public function setTitle( $t ) {
734 if ( !$t ) {
735 $t = Title::newFromText( 'NO TITLE' );
736 }
737
738 if ( $t->hasFragment() ) {
739 # Strip the fragment to avoid various odd effects
740 $this->mTitle = $t->createFragmentTarget( '' );
741 } else {
742 $this->mTitle = $t;
743 }
744 }
745
746 /**
747 * Accessor for the Title object
748 *
749 * @return Title
750 */
751 public function getTitle() {
752 return $this->mTitle;
753 }
754
755 /**
756 * Accessor/mutator for the Title object
757 *
758 * @param Title $x Title object or null to just get the current one
759 * @return Title
760 */
761 public function Title( $x = null ) {
762 return wfSetVar( $this->mTitle, $x );
763 }
764
765 /**
766 * Set the output type
767 *
768 * @param int $ot New value
769 */
770 public function setOutputType( $ot ) {
771 $this->mOutputType = $ot;
772 # Shortcut alias
773 $this->ot = [
774 'html' => $ot == self::OT_HTML,
775 'wiki' => $ot == self::OT_WIKI,
776 'pre' => $ot == self::OT_PREPROCESS,
777 'plain' => $ot == self::OT_PLAIN,
778 ];
779 }
780
781 /**
782 * Accessor/mutator for the output type
783 *
784 * @param int|null $x New value or null to just get the current one
785 * @return int
786 */
787 public function OutputType( $x = null ) {
788 return wfSetVar( $this->mOutputType, $x );
789 }
790
791 /**
792 * Get the ParserOutput object
793 *
794 * @return ParserOutput
795 */
796 public function getOutput() {
797 return $this->mOutput;
798 }
799
800 /**
801 * Get the ParserOptions object
802 *
803 * @return ParserOptions
804 */
805 public function getOptions() {
806 return $this->mOptions;
807 }
808
809 /**
810 * Accessor/mutator for the ParserOptions object
811 *
812 * @param ParserOptions $x New value or null to just get the current one
813 * @return ParserOptions Current ParserOptions object
814 */
815 public function Options( $x = null ) {
816 return wfSetVar( $this->mOptions, $x );
817 }
818
819 /**
820 * @return int
821 */
822 public function nextLinkID() {
823 return $this->mLinkID++;
824 }
825
826 /**
827 * @param int $id
828 */
829 public function setLinkID( $id ) {
830 $this->mLinkID = $id;
831 }
832
833 /**
834 * Get a language object for use in parser functions such as {{FORMATNUM:}}
835 * @return Language
836 */
837 public function getFunctionLang() {
838 return $this->getTargetLanguage();
839 }
840
841 /**
842 * Get the target language for the content being parsed. This is usually the
843 * language that the content is in.
844 *
845 * @since 1.19
846 *
847 * @throws MWException
848 * @return Language
849 */
850 public function getTargetLanguage() {
851 $target = $this->mOptions->getTargetLanguage();
852
853 if ( $target !== null ) {
854 return $target;
855 } elseif ( $this->mOptions->getInterfaceMessage() ) {
856 return $this->mOptions->getUserLangObj();
857 } elseif ( is_null( $this->mTitle ) ) {
858 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
859 }
860
861 return $this->mTitle->getPageLanguage();
862 }
863
864 /**
865 * Get the language object for language conversion
866 * @return Language|null
867 */
868 public function getConverterLanguage() {
869 return $this->getTargetLanguage();
870 }
871
872 /**
873 * Get a User object either from $this->mUser, if set, or from the
874 * ParserOptions object otherwise
875 *
876 * @return User
877 */
878 public function getUser() {
879 if ( !is_null( $this->mUser ) ) {
880 return $this->mUser;
881 }
882 return $this->mOptions->getUser();
883 }
884
885 /**
886 * Get a preprocessor object
887 *
888 * @return Preprocessor
889 */
890 public function getPreprocessor() {
891 if ( !isset( $this->mPreprocessor ) ) {
892 $class = $this->mPreprocessorClass;
893 $this->mPreprocessor = new $class( $this );
894 }
895 return $this->mPreprocessor;
896 }
897
898 /**
899 * Get a LinkRenderer instance to make links with
900 *
901 * @since 1.28
902 * @return LinkRenderer
903 */
904 public function getLinkRenderer() {
905 if ( !$this->mLinkRenderer ) {
906 $this->mLinkRenderer = MediaWikiServices::getInstance()
907 ->getLinkRendererFactory()->create();
908 $this->mLinkRenderer->setStubThreshold(
909 $this->getOptions()->getStubThreshold()
910 );
911 }
912
913 return $this->mLinkRenderer;
914 }
915
916 /**
917 * Replaces all occurrences of HTML-style comments and the given tags
918 * in the text with a random marker and returns the next text. The output
919 * parameter $matches will be an associative array filled with data in
920 * the form:
921 *
922 * @code
923 * 'UNIQ-xxxxx' => array(
924 * 'element',
925 * 'tag content',
926 * array( 'param' => 'x' ),
927 * '<element param="x">tag content</element>' ) )
928 * @endcode
929 *
930 * @param array $elements List of element names. Comments are always extracted.
931 * @param string $text Source text string.
932 * @param array $matches Out parameter, Array: extracted tags
933 * @param string|null $uniq_prefix
934 * @return string Stripped text
935 * @since 1.26 The uniq_prefix argument is deprecated.
936 */
937 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
938 if ( $uniq_prefix !== null ) {
939 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
940 }
941 static $n = 1;
942 $stripped = '';
943 $matches = [];
944
945 $taglist = implode( '|', $elements );
946 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
947
948 while ( $text != '' ) {
949 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
950 $stripped .= $p[0];
951 if ( count( $p ) < 5 ) {
952 break;
953 }
954 if ( count( $p ) > 5 ) {
955 # comment
956 $element = $p[4];
957 $attributes = '';
958 $close = '';
959 $inside = $p[5];
960 } else {
961 # tag
962 $element = $p[1];
963 $attributes = $p[2];
964 $close = $p[3];
965 $inside = $p[4];
966 }
967
968 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
969 $stripped .= $marker;
970
971 if ( $close === '/>' ) {
972 # Empty element tag, <tag />
973 $content = null;
974 $text = $inside;
975 $tail = null;
976 } else {
977 if ( $element === '!--' ) {
978 $end = '/(-->)/';
979 } else {
980 $end = "/(<\\/$element\\s*>)/i";
981 }
982 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
983 $content = $q[0];
984 if ( count( $q ) < 3 ) {
985 # No end tag -- let it run out to the end of the text.
986 $tail = '';
987 $text = '';
988 } else {
989 $tail = $q[1];
990 $text = $q[2];
991 }
992 }
993
994 $matches[$marker] = [ $element,
995 $content,
996 Sanitizer::decodeTagAttributes( $attributes ),
997 "<$element$attributes$close$content$tail" ];
998 }
999 return $stripped;
1000 }
1001
1002 /**
1003 * Get a list of strippable XML-like elements
1004 *
1005 * @return array
1006 */
1007 public function getStripList() {
1008 return $this->mStripList;
1009 }
1010
1011 /**
1012 * Add an item to the strip state
1013 * Returns the unique tag which must be inserted into the stripped text
1014 * The tag will be replaced with the original text in unstrip()
1015 *
1016 * @param string $text
1017 *
1018 * @return string
1019 */
1020 public function insertStripItem( $text ) {
1021 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1022 $this->mMarkerIndex++;
1023 $this->mStripState->addGeneral( $marker, $text );
1024 return $marker;
1025 }
1026
1027 /**
1028 * parse the wiki syntax used to render tables
1029 *
1030 * @private
1031 * @param string $text
1032 * @return string
1033 */
1034 public function doTableStuff( $text ) {
1035
1036 $lines = StringUtils::explode( "\n", $text );
1037 $out = '';
1038 $td_history = []; # Is currently a td tag open?
1039 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1040 $tr_history = []; # Is currently a tr tag open?
1041 $tr_attributes = []; # history of tr attributes
1042 $has_opened_tr = []; # Did this table open a <tr> element?
1043 $indent_level = 0; # indent level of the table
1044
1045 foreach ( $lines as $outLine ) {
1046 $line = trim( $outLine );
1047
1048 if ( $line === '' ) { # empty line, go to next line
1049 $out .= $outLine . "\n";
1050 continue;
1051 }
1052
1053 $first_character = $line[0];
1054 $first_two = substr( $line, 0, 2 );
1055 $matches = [];
1056
1057 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1058 # First check if we are starting a new table
1059 $indent_level = strlen( $matches[1] );
1060
1061 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1062 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1063
1064 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1065 array_push( $td_history, false );
1066 array_push( $last_tag_history, '' );
1067 array_push( $tr_history, false );
1068 array_push( $tr_attributes, '' );
1069 array_push( $has_opened_tr, false );
1070 } elseif ( count( $td_history ) == 0 ) {
1071 # Don't do any of the following
1072 $out .= $outLine . "\n";
1073 continue;
1074 } elseif ( $first_two === '|}' ) {
1075 # We are ending a table
1076 $line = '</table>' . substr( $line, 2 );
1077 $last_tag = array_pop( $last_tag_history );
1078
1079 if ( !array_pop( $has_opened_tr ) ) {
1080 $line = "<tr><td></td></tr>{$line}";
1081 }
1082
1083 if ( array_pop( $tr_history ) ) {
1084 $line = "</tr>{$line}";
1085 }
1086
1087 if ( array_pop( $td_history ) ) {
1088 $line = "</{$last_tag}>{$line}";
1089 }
1090 array_pop( $tr_attributes );
1091 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1092 } elseif ( $first_two === '|-' ) {
1093 # Now we have a table row
1094 $line = preg_replace( '#^\|-+#', '', $line );
1095
1096 # Whats after the tag is now only attributes
1097 $attributes = $this->mStripState->unstripBoth( $line );
1098 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1099 array_pop( $tr_attributes );
1100 array_push( $tr_attributes, $attributes );
1101
1102 $line = '';
1103 $last_tag = array_pop( $last_tag_history );
1104 array_pop( $has_opened_tr );
1105 array_push( $has_opened_tr, true );
1106
1107 if ( array_pop( $tr_history ) ) {
1108 $line = '</tr>';
1109 }
1110
1111 if ( array_pop( $td_history ) ) {
1112 $line = "</{$last_tag}>{$line}";
1113 }
1114
1115 $outLine = $line;
1116 array_push( $tr_history, false );
1117 array_push( $td_history, false );
1118 array_push( $last_tag_history, '' );
1119 } elseif ( $first_character === '|'
1120 || $first_character === '!'
1121 || $first_two === '|+'
1122 ) {
1123 # This might be cell elements, td, th or captions
1124 if ( $first_two === '|+' ) {
1125 $first_character = '+';
1126 $line = substr( $line, 2 );
1127 } else {
1128 $line = substr( $line, 1 );
1129 }
1130
1131 // Implies both are valid for table headings.
1132 if ( $first_character === '!' ) {
1133 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1134 }
1135
1136 # Split up multiple cells on the same line.
1137 # FIXME : This can result in improper nesting of tags processed
1138 # by earlier parser steps.
1139 $cells = explode( '||', $line );
1140
1141 $outLine = '';
1142
1143 # Loop through each table cell
1144 foreach ( $cells as $cell ) {
1145 $previous = '';
1146 if ( $first_character !== '+' ) {
1147 $tr_after = array_pop( $tr_attributes );
1148 if ( !array_pop( $tr_history ) ) {
1149 $previous = "<tr{$tr_after}>\n";
1150 }
1151 array_push( $tr_history, true );
1152 array_push( $tr_attributes, '' );
1153 array_pop( $has_opened_tr );
1154 array_push( $has_opened_tr, true );
1155 }
1156
1157 $last_tag = array_pop( $last_tag_history );
1158
1159 if ( array_pop( $td_history ) ) {
1160 $previous = "</{$last_tag}>\n{$previous}";
1161 }
1162
1163 if ( $first_character === '|' ) {
1164 $last_tag = 'td';
1165 } elseif ( $first_character === '!' ) {
1166 $last_tag = 'th';
1167 } elseif ( $first_character === '+' ) {
1168 $last_tag = 'caption';
1169 } else {
1170 $last_tag = '';
1171 }
1172
1173 array_push( $last_tag_history, $last_tag );
1174
1175 # A cell could contain both parameters and data
1176 $cell_data = explode( '|', $cell, 2 );
1177
1178 # Bug 553: Note that a '|' inside an invalid link should not
1179 # be mistaken as delimiting cell parameters
1180 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1181 $cell = "{$previous}<{$last_tag}>{$cell}";
1182 } elseif ( count( $cell_data ) == 1 ) {
1183 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1184 } else {
1185 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1186 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1187 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1188 }
1189
1190 $outLine .= $cell;
1191 array_push( $td_history, true );
1192 }
1193 }
1194 $out .= $outLine . "\n";
1195 }
1196
1197 # Closing open td, tr && table
1198 while ( count( $td_history ) > 0 ) {
1199 if ( array_pop( $td_history ) ) {
1200 $out .= "</td>\n";
1201 }
1202 if ( array_pop( $tr_history ) ) {
1203 $out .= "</tr>\n";
1204 }
1205 if ( !array_pop( $has_opened_tr ) ) {
1206 $out .= "<tr><td></td></tr>\n";
1207 }
1208
1209 $out .= "</table>\n";
1210 }
1211
1212 # Remove trailing line-ending (b/c)
1213 if ( substr( $out, -1 ) === "\n" ) {
1214 $out = substr( $out, 0, -1 );
1215 }
1216
1217 # special case: don't return empty table
1218 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1219 $out = '';
1220 }
1221
1222 return $out;
1223 }
1224
1225 /**
1226 * Helper function for parse() that transforms wiki markup into half-parsed
1227 * HTML. Only called for $mOutputType == self::OT_HTML.
1228 *
1229 * @private
1230 *
1231 * @param string $text The text to parse
1232 * @param bool $isMain Whether this is being called from the main parse() function
1233 * @param PPFrame|bool $frame A pre-processor frame
1234 *
1235 * @return string
1236 */
1237 public function internalParse( $text, $isMain = true, $frame = false ) {
1238
1239 $origText = $text;
1240
1241 # Hook to suspend the parser in this state
1242 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1243 return $text;
1244 }
1245
1246 # if $frame is provided, then use $frame for replacing any variables
1247 if ( $frame ) {
1248 # use frame depth to infer how include/noinclude tags should be handled
1249 # depth=0 means this is the top-level document; otherwise it's an included document
1250 if ( !$frame->depth ) {
1251 $flag = 0;
1252 } else {
1253 $flag = Parser::PTD_FOR_INCLUSION;
1254 }
1255 $dom = $this->preprocessToDom( $text, $flag );
1256 $text = $frame->expand( $dom );
1257 } else {
1258 # if $frame is not provided, then use old-style replaceVariables
1259 $text = $this->replaceVariables( $text );
1260 }
1261
1262 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1263 $text = Sanitizer::removeHTMLtags(
1264 $text,
1265 [ &$this, 'attributeStripCallback' ],
1266 false,
1267 array_keys( $this->mTransparentTagHooks ),
1268 [],
1269 [ &$this, 'addTrackingCategory' ]
1270 );
1271 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1272
1273 # Tables need to come after variable replacement for things to work
1274 # properly; putting them before other transformations should keep
1275 # exciting things like link expansions from showing up in surprising
1276 # places.
1277 $text = $this->doTableStuff( $text );
1278
1279 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1280
1281 $text = $this->doDoubleUnderscore( $text );
1282
1283 $text = $this->doHeadings( $text );
1284 $text = $this->replaceInternalLinks( $text );
1285 $text = $this->doAllQuotes( $text );
1286 $text = $this->replaceExternalLinks( $text );
1287
1288 # replaceInternalLinks may sometimes leave behind
1289 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1290 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1291
1292 $text = $this->doMagicLinks( $text );
1293 $text = $this->formatHeadings( $text, $origText, $isMain );
1294
1295 return $text;
1296 }
1297
1298 /**
1299 * Helper function for parse() that transforms half-parsed HTML into fully
1300 * parsed HTML.
1301 *
1302 * @param string $text
1303 * @param bool $isMain
1304 * @param bool $linestart
1305 * @return string
1306 */
1307 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1308 $text = $this->mStripState->unstripGeneral( $text );
1309
1310 if ( $isMain ) {
1311 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1312 }
1313
1314 # Clean up special characters, only run once, next-to-last before doBlockLevels
1315 $fixtags = [
1316 # french spaces, last one Guillemet-left
1317 # only if there is something before the space
1318 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1319 # french spaces, Guillemet-right
1320 '/(\\302\\253) /' => '\\1&#160;',
1321 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1322 ];
1323 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1324
1325 $text = $this->doBlockLevels( $text, $linestart );
1326
1327 $this->replaceLinkHolders( $text );
1328
1329 /**
1330 * The input doesn't get language converted if
1331 * a) It's disabled
1332 * b) Content isn't converted
1333 * c) It's a conversion table
1334 * d) it is an interface message (which is in the user language)
1335 */
1336 if ( !( $this->mOptions->getDisableContentConversion()
1337 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1338 ) {
1339 if ( !$this->mOptions->getInterfaceMessage() ) {
1340 # The position of the convert() call should not be changed. it
1341 # assumes that the links are all replaced and the only thing left
1342 # is the <nowiki> mark.
1343 $text = $this->getConverterLanguage()->convert( $text );
1344 }
1345 }
1346
1347 $text = $this->mStripState->unstripNoWiki( $text );
1348
1349 if ( $isMain ) {
1350 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1351 }
1352
1353 $text = $this->replaceTransparentTags( $text );
1354 $text = $this->mStripState->unstripGeneral( $text );
1355
1356 $text = Sanitizer::normalizeCharReferences( $text );
1357
1358 if ( MWTidy::isEnabled() && $this->mOptions->getTidy() ) {
1359 $text = MWTidy::tidy( $text );
1360 } else {
1361 # attempt to sanitize at least some nesting problems
1362 # (bug #2702 and quite a few others)
1363 $tidyregs = [
1364 # ''Something [http://www.cool.com cool''] -->
1365 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1366 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1367 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1368 # fix up an anchor inside another anchor, only
1369 # at least for a single single nested link (bug 3695)
1370 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1371 '\\1\\2</a>\\3</a>\\1\\4</a>',
1372 # fix div inside inline elements- doBlockLevels won't wrap a line which
1373 # contains a div, so fix it up here; replace
1374 # div with escaped text
1375 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1376 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1377 # remove empty italic or bold tag pairs, some
1378 # introduced by rules above
1379 '/<([bi])><\/\\1>/' => '',
1380 ];
1381
1382 $text = preg_replace(
1383 array_keys( $tidyregs ),
1384 array_values( $tidyregs ),
1385 $text );
1386 }
1387
1388 if ( $isMain ) {
1389 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1390 }
1391
1392 return $text;
1393 }
1394
1395 /**
1396 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1397 * magic external links.
1398 *
1399 * DML
1400 * @private
1401 *
1402 * @param string $text
1403 *
1404 * @return string
1405 */
1406 public function doMagicLinks( $text ) {
1407 $prots = wfUrlProtocolsWithoutProtRel();
1408 $urlChar = self::EXT_LINK_URL_CLASS;
1409 $addr = self::EXT_LINK_ADDR;
1410 $space = self::SPACE_NOT_NL; # non-newline space
1411 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1412 $spaces = "$space++"; # possessive match of 1 or more spaces
1413 $text = preg_replace_callback(
1414 '!(?: # Start cases
1415 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1416 (<.*?>) | # m[2]: Skip stuff inside
1417 # HTML elements' . "
1418 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1419 # m[4]: Post-protocol path
1420 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1421 ([0-9]+)\b |
1422 \bISBN $spaces ( # m[6]: ISBN, capture number
1423 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1424 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1425 [0-9Xx] # check digit
1426 )\b
1427 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1428 return $text;
1429 }
1430
1431 /**
1432 * @throws MWException
1433 * @param array $m
1434 * @return HTML|string
1435 */
1436 public function magicLinkCallback( $m ) {
1437 if ( isset( $m[1] ) && $m[1] !== '' ) {
1438 # Skip anchor
1439 return $m[0];
1440 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1441 # Skip HTML element
1442 return $m[0];
1443 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1444 # Free external link
1445 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1446 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1447 # RFC or PMID
1448 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1449 $keyword = 'RFC';
1450 $urlmsg = 'rfcurl';
1451 $cssClass = 'mw-magiclink-rfc';
1452 $id = $m[5];
1453 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1454 $keyword = 'PMID';
1455 $urlmsg = 'pubmedurl';
1456 $cssClass = 'mw-magiclink-pmid';
1457 $id = $m[5];
1458 } else {
1459 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1460 substr( $m[0], 0, 20 ) . '"' );
1461 }
1462 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1463 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1464 } elseif ( isset( $m[6] ) && $m[6] !== '' ) {
1465 # ISBN
1466 $isbn = $m[6];
1467 $space = self::SPACE_NOT_NL; # non-newline space
1468 $isbn = preg_replace( "/$space/", ' ', $isbn );
1469 $num = strtr( $isbn, [
1470 '-' => '',
1471 ' ' => '',
1472 'x' => 'X',
1473 ] );
1474 return $this->getLinkRenderer()->makeKnownLink(
1475 SpecialPage::getTitleFor( 'Booksources', $num ),
1476 "ISBN $isbn",
1477 [
1478 'class' => 'internal mw-magiclink-isbn',
1479 'title' => false // suppress title attribute
1480 ]
1481 );
1482 } else {
1483 return $m[0];
1484 }
1485 }
1486
1487 /**
1488 * Make a free external link, given a user-supplied URL
1489 *
1490 * @param string $url
1491 * @param int $numPostProto
1492 * The number of characters after the protocol.
1493 * @return string HTML
1494 * @private
1495 */
1496 public function makeFreeExternalLink( $url, $numPostProto ) {
1497 $trail = '';
1498
1499 # The characters '<' and '>' (which were escaped by
1500 # removeHTMLtags()) should not be included in
1501 # URLs, per RFC 2396.
1502 # Make &nbsp; terminate a URL as well (bug T84937)
1503 $m2 = [];
1504 if ( preg_match(
1505 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1506 $url,
1507 $m2,
1508 PREG_OFFSET_CAPTURE
1509 ) ) {
1510 $trail = substr( $url, $m2[0][1] ) . $trail;
1511 $url = substr( $url, 0, $m2[0][1] );
1512 }
1513
1514 # Move trailing punctuation to $trail
1515 $sep = ',;\.:!?';
1516 # If there is no left bracket, then consider right brackets fair game too
1517 if ( strpos( $url, '(' ) === false ) {
1518 $sep .= ')';
1519 }
1520
1521 $urlRev = strrev( $url );
1522 $numSepChars = strspn( $urlRev, $sep );
1523 # Don't break a trailing HTML entity by moving the ; into $trail
1524 # This is in hot code, so use substr_compare to avoid having to
1525 # create a new string object for the comparison
1526 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1527 # more optimization: instead of running preg_match with a $
1528 # anchor, which can be slow, do the match on the reversed
1529 # string starting at the desired offset.
1530 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1531 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1532 $numSepChars--;
1533 }
1534 }
1535 if ( $numSepChars ) {
1536 $trail = substr( $url, -$numSepChars ) . $trail;
1537 $url = substr( $url, 0, -$numSepChars );
1538 }
1539
1540 # Verify that we still have a real URL after trail removal, and
1541 # not just lone protocol
1542 if ( strlen( $trail ) >= $numPostProto ) {
1543 return $url . $trail;
1544 }
1545
1546 $url = Sanitizer::cleanUrl( $url );
1547
1548 # Is this an external image?
1549 $text = $this->maybeMakeExternalImage( $url );
1550 if ( $text === false ) {
1551 # Not an image, make a link
1552 $text = Linker::makeExternalLink( $url,
1553 $this->getConverterLanguage()->markNoConversion( $url, true ),
1554 true, 'free',
1555 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1556 # Register it in the output object...
1557 # Replace unnecessary URL escape codes with their equivalent characters
1558 $pasteurized = self::normalizeLinkUrl( $url );
1559 $this->mOutput->addExternalLink( $pasteurized );
1560 }
1561 return $text . $trail;
1562 }
1563
1564 /**
1565 * Parse headers and return html
1566 *
1567 * @private
1568 *
1569 * @param string $text
1570 *
1571 * @return string
1572 */
1573 public function doHeadings( $text ) {
1574 for ( $i = 6; $i >= 1; --$i ) {
1575 $h = str_repeat( '=', $i );
1576 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1577 }
1578 return $text;
1579 }
1580
1581 /**
1582 * Replace single quotes with HTML markup
1583 * @private
1584 *
1585 * @param string $text
1586 *
1587 * @return string The altered text
1588 */
1589 public function doAllQuotes( $text ) {
1590 $outtext = '';
1591 $lines = StringUtils::explode( "\n", $text );
1592 foreach ( $lines as $line ) {
1593 $outtext .= $this->doQuotes( $line ) . "\n";
1594 }
1595 $outtext = substr( $outtext, 0, -1 );
1596 return $outtext;
1597 }
1598
1599 /**
1600 * Helper function for doAllQuotes()
1601 *
1602 * @param string $text
1603 *
1604 * @return string
1605 */
1606 public function doQuotes( $text ) {
1607 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1608 $countarr = count( $arr );
1609 if ( $countarr == 1 ) {
1610 return $text;
1611 }
1612
1613 // First, do some preliminary work. This may shift some apostrophes from
1614 // being mark-up to being text. It also counts the number of occurrences
1615 // of bold and italics mark-ups.
1616 $numbold = 0;
1617 $numitalics = 0;
1618 for ( $i = 1; $i < $countarr; $i += 2 ) {
1619 $thislen = strlen( $arr[$i] );
1620 // If there are ever four apostrophes, assume the first is supposed to
1621 // be text, and the remaining three constitute mark-up for bold text.
1622 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1623 if ( $thislen == 4 ) {
1624 $arr[$i - 1] .= "'";
1625 $arr[$i] = "'''";
1626 $thislen = 3;
1627 } elseif ( $thislen > 5 ) {
1628 // If there are more than 5 apostrophes in a row, assume they're all
1629 // text except for the last 5.
1630 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1631 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1632 $arr[$i] = "'''''";
1633 $thislen = 5;
1634 }
1635 // Count the number of occurrences of bold and italics mark-ups.
1636 if ( $thislen == 2 ) {
1637 $numitalics++;
1638 } elseif ( $thislen == 3 ) {
1639 $numbold++;
1640 } elseif ( $thislen == 5 ) {
1641 $numitalics++;
1642 $numbold++;
1643 }
1644 }
1645
1646 // If there is an odd number of both bold and italics, it is likely
1647 // that one of the bold ones was meant to be an apostrophe followed
1648 // by italics. Which one we cannot know for certain, but it is more
1649 // likely to be one that has a single-letter word before it.
1650 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1651 $firstsingleletterword = -1;
1652 $firstmultiletterword = -1;
1653 $firstspace = -1;
1654 for ( $i = 1; $i < $countarr; $i += 2 ) {
1655 if ( strlen( $arr[$i] ) == 3 ) {
1656 $x1 = substr( $arr[$i - 1], -1 );
1657 $x2 = substr( $arr[$i - 1], -2, 1 );
1658 if ( $x1 === ' ' ) {
1659 if ( $firstspace == -1 ) {
1660 $firstspace = $i;
1661 }
1662 } elseif ( $x2 === ' ' ) {
1663 $firstsingleletterword = $i;
1664 // if $firstsingleletterword is set, we don't
1665 // look at the other options, so we can bail early.
1666 break;
1667 } else {
1668 if ( $firstmultiletterword == -1 ) {
1669 $firstmultiletterword = $i;
1670 }
1671 }
1672 }
1673 }
1674
1675 // If there is a single-letter word, use it!
1676 if ( $firstsingleletterword > -1 ) {
1677 $arr[$firstsingleletterword] = "''";
1678 $arr[$firstsingleletterword - 1] .= "'";
1679 } elseif ( $firstmultiletterword > -1 ) {
1680 // If not, but there's a multi-letter word, use that one.
1681 $arr[$firstmultiletterword] = "''";
1682 $arr[$firstmultiletterword - 1] .= "'";
1683 } elseif ( $firstspace > -1 ) {
1684 // ... otherwise use the first one that has neither.
1685 // (notice that it is possible for all three to be -1 if, for example,
1686 // there is only one pentuple-apostrophe in the line)
1687 $arr[$firstspace] = "''";
1688 $arr[$firstspace - 1] .= "'";
1689 }
1690 }
1691
1692 // Now let's actually convert our apostrophic mush to HTML!
1693 $output = '';
1694 $buffer = '';
1695 $state = '';
1696 $i = 0;
1697 foreach ( $arr as $r ) {
1698 if ( ( $i % 2 ) == 0 ) {
1699 if ( $state === 'both' ) {
1700 $buffer .= $r;
1701 } else {
1702 $output .= $r;
1703 }
1704 } else {
1705 $thislen = strlen( $r );
1706 if ( $thislen == 2 ) {
1707 if ( $state === 'i' ) {
1708 $output .= '</i>';
1709 $state = '';
1710 } elseif ( $state === 'bi' ) {
1711 $output .= '</i>';
1712 $state = 'b';
1713 } elseif ( $state === 'ib' ) {
1714 $output .= '</b></i><b>';
1715 $state = 'b';
1716 } elseif ( $state === 'both' ) {
1717 $output .= '<b><i>' . $buffer . '</i>';
1718 $state = 'b';
1719 } else { // $state can be 'b' or ''
1720 $output .= '<i>';
1721 $state .= 'i';
1722 }
1723 } elseif ( $thislen == 3 ) {
1724 if ( $state === 'b' ) {
1725 $output .= '</b>';
1726 $state = '';
1727 } elseif ( $state === 'bi' ) {
1728 $output .= '</i></b><i>';
1729 $state = 'i';
1730 } elseif ( $state === 'ib' ) {
1731 $output .= '</b>';
1732 $state = 'i';
1733 } elseif ( $state === 'both' ) {
1734 $output .= '<i><b>' . $buffer . '</b>';
1735 $state = 'i';
1736 } else { // $state can be 'i' or ''
1737 $output .= '<b>';
1738 $state .= 'b';
1739 }
1740 } elseif ( $thislen == 5 ) {
1741 if ( $state === 'b' ) {
1742 $output .= '</b><i>';
1743 $state = 'i';
1744 } elseif ( $state === 'i' ) {
1745 $output .= '</i><b>';
1746 $state = 'b';
1747 } elseif ( $state === 'bi' ) {
1748 $output .= '</i></b>';
1749 $state = '';
1750 } elseif ( $state === 'ib' ) {
1751 $output .= '</b></i>';
1752 $state = '';
1753 } elseif ( $state === 'both' ) {
1754 $output .= '<i><b>' . $buffer . '</b></i>';
1755 $state = '';
1756 } else { // ($state == '')
1757 $buffer = '';
1758 $state = 'both';
1759 }
1760 }
1761 }
1762 $i++;
1763 }
1764 // Now close all remaining tags. Notice that the order is important.
1765 if ( $state === 'b' || $state === 'ib' ) {
1766 $output .= '</b>';
1767 }
1768 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1769 $output .= '</i>';
1770 }
1771 if ( $state === 'bi' ) {
1772 $output .= '</b>';
1773 }
1774 // There might be lonely ''''', so make sure we have a buffer
1775 if ( $state === 'both' && $buffer ) {
1776 $output .= '<b><i>' . $buffer . '</i></b>';
1777 }
1778 return $output;
1779 }
1780
1781 /**
1782 * Replace external links (REL)
1783 *
1784 * Note: this is all very hackish and the order of execution matters a lot.
1785 * Make sure to run tests/parserTests.php if you change this code.
1786 *
1787 * @private
1788 *
1789 * @param string $text
1790 *
1791 * @throws MWException
1792 * @return string
1793 */
1794 public function replaceExternalLinks( $text ) {
1795
1796 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1797 if ( $bits === false ) {
1798 throw new MWException( "PCRE needs to be compiled with "
1799 . "--enable-unicode-properties in order for MediaWiki to function" );
1800 }
1801 $s = array_shift( $bits );
1802
1803 $i = 0;
1804 while ( $i < count( $bits ) ) {
1805 $url = $bits[$i++];
1806 $i++; // protocol
1807 $text = $bits[$i++];
1808 $trail = $bits[$i++];
1809
1810 # The characters '<' and '>' (which were escaped by
1811 # removeHTMLtags()) should not be included in
1812 # URLs, per RFC 2396.
1813 $m2 = [];
1814 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1815 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1816 $url = substr( $url, 0, $m2[0][1] );
1817 }
1818
1819 # If the link text is an image URL, replace it with an <img> tag
1820 # This happened by accident in the original parser, but some people used it extensively
1821 $img = $this->maybeMakeExternalImage( $text );
1822 if ( $img !== false ) {
1823 $text = $img;
1824 }
1825
1826 $dtrail = '';
1827
1828 # Set linktype for CSS - if URL==text, link is essentially free
1829 $linktype = ( $text === $url ) ? 'free' : 'text';
1830
1831 # No link text, e.g. [http://domain.tld/some.link]
1832 if ( $text == '' ) {
1833 # Autonumber
1834 $langObj = $this->getTargetLanguage();
1835 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1836 $linktype = 'autonumber';
1837 } else {
1838 # Have link text, e.g. [http://domain.tld/some.link text]s
1839 # Check for trail
1840 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1841 }
1842
1843 $text = $this->getConverterLanguage()->markNoConversion( $text );
1844
1845 $url = Sanitizer::cleanUrl( $url );
1846
1847 # Use the encoded URL
1848 # This means that users can paste URLs directly into the text
1849 # Funny characters like ö aren't valid in URLs anyway
1850 # This was changed in August 2004
1851 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1852 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1853
1854 # Register link in the output object.
1855 # Replace unnecessary URL escape codes with the referenced character
1856 # This prevents spammers from hiding links from the filters
1857 $pasteurized = self::normalizeLinkUrl( $url );
1858 $this->mOutput->addExternalLink( $pasteurized );
1859 }
1860
1861 return $s;
1862 }
1863
1864 /**
1865 * Get the rel attribute for a particular external link.
1866 *
1867 * @since 1.21
1868 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1869 * nofollow if appropriate
1870 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1871 * @return string|null Rel attribute for $url
1872 */
1873 public static function getExternalLinkRel( $url = false, $title = null ) {
1874 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1875 $ns = $title ? $title->getNamespace() : false;
1876 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1877 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1878 ) {
1879 return 'nofollow';
1880 }
1881 return null;
1882 }
1883
1884 /**
1885 * Get an associative array of additional HTML attributes appropriate for a
1886 * particular external link. This currently may include rel => nofollow
1887 * (depending on configuration, namespace, and the URL's domain) and/or a
1888 * target attribute (depending on configuration).
1889 *
1890 * @param string $url URL to extract the domain from for rel =>
1891 * nofollow if appropriate
1892 * @return array Associative array of HTML attributes
1893 */
1894 public function getExternalLinkAttribs( $url ) {
1895 $attribs = [];
1896 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1897
1898 $target = $this->mOptions->getExternalLinkTarget();
1899 if ( $target ) {
1900 $attribs['target'] = $target;
1901 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1902 // T133507. New windows can navigate parent cross-origin.
1903 // Including noreferrer due to lacking browser
1904 // support of noopener. Eventually noreferrer should be removed.
1905 if ( $rel !== '' ) {
1906 $rel .= ' ';
1907 }
1908 $rel .= 'noreferrer noopener';
1909 }
1910 }
1911 $attribs['rel'] = $rel;
1912 return $attribs;
1913 }
1914
1915 /**
1916 * Replace unusual escape codes in a URL with their equivalent characters
1917 *
1918 * @deprecated since 1.24, use normalizeLinkUrl
1919 * @param string $url
1920 * @return string
1921 */
1922 public static function replaceUnusualEscapes( $url ) {
1923 wfDeprecated( __METHOD__, '1.24' );
1924 return self::normalizeLinkUrl( $url );
1925 }
1926
1927 /**
1928 * Replace unusual escape codes in a URL with their equivalent characters
1929 *
1930 * This generally follows the syntax defined in RFC 3986, with special
1931 * consideration for HTTP query strings.
1932 *
1933 * @param string $url
1934 * @return string
1935 */
1936 public static function normalizeLinkUrl( $url ) {
1937 # First, make sure unsafe characters are encoded
1938 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1939 function ( $m ) {
1940 return rawurlencode( $m[0] );
1941 },
1942 $url
1943 );
1944
1945 $ret = '';
1946 $end = strlen( $url );
1947
1948 # Fragment part - 'fragment'
1949 $start = strpos( $url, '#' );
1950 if ( $start !== false && $start < $end ) {
1951 $ret = self::normalizeUrlComponent(
1952 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1953 $end = $start;
1954 }
1955
1956 # Query part - 'query' minus &=+;
1957 $start = strpos( $url, '?' );
1958 if ( $start !== false && $start < $end ) {
1959 $ret = self::normalizeUrlComponent(
1960 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1961 $end = $start;
1962 }
1963
1964 # Scheme and path part - 'pchar'
1965 # (we assume no userinfo or encoded colons in the host)
1966 $ret = self::normalizeUrlComponent(
1967 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1968
1969 return $ret;
1970 }
1971
1972 private static function normalizeUrlComponent( $component, $unsafe ) {
1973 $callback = function ( $matches ) use ( $unsafe ) {
1974 $char = urldecode( $matches[0] );
1975 $ord = ord( $char );
1976 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1977 # Unescape it
1978 return $char;
1979 } else {
1980 # Leave it escaped, but use uppercase for a-f
1981 return strtoupper( $matches[0] );
1982 }
1983 };
1984 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1985 }
1986
1987 /**
1988 * make an image if it's allowed, either through the global
1989 * option, through the exception, or through the on-wiki whitelist
1990 *
1991 * @param string $url
1992 *
1993 * @return string
1994 */
1995 private function maybeMakeExternalImage( $url ) {
1996 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1997 $imagesexception = !empty( $imagesfrom );
1998 $text = false;
1999 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2000 if ( $imagesexception && is_array( $imagesfrom ) ) {
2001 $imagematch = false;
2002 foreach ( $imagesfrom as $match ) {
2003 if ( strpos( $url, $match ) === 0 ) {
2004 $imagematch = true;
2005 break;
2006 }
2007 }
2008 } elseif ( $imagesexception ) {
2009 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2010 } else {
2011 $imagematch = false;
2012 }
2013
2014 if ( $this->mOptions->getAllowExternalImages()
2015 || ( $imagesexception && $imagematch )
2016 ) {
2017 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2018 # Image found
2019 $text = Linker::makeExternalImage( $url );
2020 }
2021 }
2022 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2023 && preg_match( self::EXT_IMAGE_REGEX, $url )
2024 ) {
2025 $whitelist = explode(
2026 "\n",
2027 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2028 );
2029
2030 foreach ( $whitelist as $entry ) {
2031 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2032 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2033 continue;
2034 }
2035 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2036 # Image matches a whitelist entry
2037 $text = Linker::makeExternalImage( $url );
2038 break;
2039 }
2040 }
2041 }
2042 return $text;
2043 }
2044
2045 /**
2046 * Process [[ ]] wikilinks
2047 *
2048 * @param string $s
2049 *
2050 * @return string Processed text
2051 *
2052 * @private
2053 */
2054 public function replaceInternalLinks( $s ) {
2055 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2056 return $s;
2057 }
2058
2059 /**
2060 * Process [[ ]] wikilinks (RIL)
2061 * @param string $s
2062 * @throws MWException
2063 * @return LinkHolderArray
2064 *
2065 * @private
2066 */
2067 public function replaceInternalLinks2( &$s ) {
2068 global $wgExtraInterlanguageLinkPrefixes;
2069
2070 static $tc = false, $e1, $e1_img;
2071 # the % is needed to support urlencoded titles as well
2072 if ( !$tc ) {
2073 $tc = Title::legalChars() . '#%';
2074 # Match a link having the form [[namespace:link|alternate]]trail
2075 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2076 # Match cases where there is no "]]", which might still be images
2077 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2078 }
2079
2080 $holders = new LinkHolderArray( $this );
2081
2082 # split the entire text string on occurrences of [[
2083 $a = StringUtils::explode( '[[', ' ' . $s );
2084 # get the first element (all text up to first [[), and remove the space we added
2085 $s = $a->current();
2086 $a->next();
2087 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2088 $s = substr( $s, 1 );
2089
2090 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2091 $e2 = null;
2092 if ( $useLinkPrefixExtension ) {
2093 # Match the end of a line for a word that's not followed by whitespace,
2094 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2095 global $wgContLang;
2096 $charset = $wgContLang->linkPrefixCharset();
2097 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2098 }
2099
2100 if ( is_null( $this->mTitle ) ) {
2101 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2102 }
2103 $nottalk = !$this->mTitle->isTalkPage();
2104
2105 if ( $useLinkPrefixExtension ) {
2106 $m = [];
2107 if ( preg_match( $e2, $s, $m ) ) {
2108 $first_prefix = $m[2];
2109 } else {
2110 $first_prefix = false;
2111 }
2112 } else {
2113 $prefix = '';
2114 }
2115
2116 $useSubpages = $this->areSubpagesAllowed();
2117
2118 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2119 # Loop for each link
2120 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2121 // @codingStandardsIgnoreEnd
2122
2123 # Check for excessive memory usage
2124 if ( $holders->isBig() ) {
2125 # Too big
2126 # Do the existence check, replace the link holders and clear the array
2127 $holders->replace( $s );
2128 $holders->clear();
2129 }
2130
2131 if ( $useLinkPrefixExtension ) {
2132 if ( preg_match( $e2, $s, $m ) ) {
2133 $prefix = $m[2];
2134 $s = $m[1];
2135 } else {
2136 $prefix = '';
2137 }
2138 # first link
2139 if ( $first_prefix ) {
2140 $prefix = $first_prefix;
2141 $first_prefix = false;
2142 }
2143 }
2144
2145 $might_be_img = false;
2146
2147 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2148 $text = $m[2];
2149 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2150 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2151 # the real problem is with the $e1 regex
2152 # See bug 1300.
2153 # Still some problems for cases where the ] is meant to be outside punctuation,
2154 # and no image is in sight. See bug 2095.
2155 if ( $text !== ''
2156 && substr( $m[3], 0, 1 ) === ']'
2157 && strpos( $text, '[' ) !== false
2158 ) {
2159 $text .= ']'; # so that replaceExternalLinks($text) works later
2160 $m[3] = substr( $m[3], 1 );
2161 }
2162 # fix up urlencoded title texts
2163 if ( strpos( $m[1], '%' ) !== false ) {
2164 # Should anchors '#' also be rejected?
2165 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2166 }
2167 $trail = $m[3];
2168 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2169 # Invalid, but might be an image with a link in its caption
2170 $might_be_img = true;
2171 $text = $m[2];
2172 if ( strpos( $m[1], '%' ) !== false ) {
2173 $m[1] = rawurldecode( $m[1] );
2174 }
2175 $trail = "";
2176 } else { # Invalid form; output directly
2177 $s .= $prefix . '[[' . $line;
2178 continue;
2179 }
2180
2181 $origLink = $m[1];
2182
2183 # Don't allow internal links to pages containing
2184 # PROTO: where PROTO is a valid URL protocol; these
2185 # should be external links.
2186 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2187 $s .= $prefix . '[[' . $line;
2188 continue;
2189 }
2190
2191 # Make subpage if necessary
2192 if ( $useSubpages ) {
2193 $link = $this->maybeDoSubpageLink( $origLink, $text );
2194 } else {
2195 $link = $origLink;
2196 }
2197
2198 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2199 if ( !$noforce ) {
2200 # Strip off leading ':'
2201 $link = substr( $link, 1 );
2202 }
2203
2204 $unstrip = $this->mStripState->unstripNoWiki( $link );
2205 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2206 if ( $nt === null ) {
2207 $s .= $prefix . '[[' . $line;
2208 continue;
2209 }
2210
2211 $ns = $nt->getNamespace();
2212 $iw = $nt->getInterwiki();
2213
2214 if ( $might_be_img ) { # if this is actually an invalid link
2215 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2216 $found = false;
2217 while ( true ) {
2218 # look at the next 'line' to see if we can close it there
2219 $a->next();
2220 $next_line = $a->current();
2221 if ( $next_line === false || $next_line === null ) {
2222 break;
2223 }
2224 $m = explode( ']]', $next_line, 3 );
2225 if ( count( $m ) == 3 ) {
2226 # the first ]] closes the inner link, the second the image
2227 $found = true;
2228 $text .= "[[{$m[0]}]]{$m[1]}";
2229 $trail = $m[2];
2230 break;
2231 } elseif ( count( $m ) == 2 ) {
2232 # if there's exactly one ]] that's fine, we'll keep looking
2233 $text .= "[[{$m[0]}]]{$m[1]}";
2234 } else {
2235 # if $next_line is invalid too, we need look no further
2236 $text .= '[[' . $next_line;
2237 break;
2238 }
2239 }
2240 if ( !$found ) {
2241 # we couldn't find the end of this imageLink, so output it raw
2242 # but don't ignore what might be perfectly normal links in the text we've examined
2243 $holders->merge( $this->replaceInternalLinks2( $text ) );
2244 $s .= "{$prefix}[[$link|$text";
2245 # note: no $trail, because without an end, there *is* no trail
2246 continue;
2247 }
2248 } else { # it's not an image, so output it raw
2249 $s .= "{$prefix}[[$link|$text";
2250 # note: no $trail, because without an end, there *is* no trail
2251 continue;
2252 }
2253 }
2254
2255 $wasblank = ( $text == '' );
2256 if ( $wasblank ) {
2257 $text = $link;
2258 } else {
2259 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2260 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2261 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2262 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2263 $text = $this->doQuotes( $text );
2264 }
2265
2266 # Link not escaped by : , create the various objects
2267 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2268 # Interwikis
2269 if (
2270 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2271 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2272 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2273 )
2274 ) {
2275 # Bug 24502: filter duplicates
2276 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2277 $this->mLangLinkLanguages[$iw] = true;
2278 $this->mOutput->addLanguageLink( $nt->getFullText() );
2279 }
2280
2281 $s = rtrim( $s . $prefix );
2282 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2283 continue;
2284 }
2285
2286 if ( $ns == NS_FILE ) {
2287 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2288 if ( $wasblank ) {
2289 # if no parameters were passed, $text
2290 # becomes something like "File:Foo.png",
2291 # which we don't want to pass on to the
2292 # image generator
2293 $text = '';
2294 } else {
2295 # recursively parse links inside the image caption
2296 # actually, this will parse them in any other parameters, too,
2297 # but it might be hard to fix that, and it doesn't matter ATM
2298 $text = $this->replaceExternalLinks( $text );
2299 $holders->merge( $this->replaceInternalLinks2( $text ) );
2300 }
2301 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2302 $s .= $prefix . $this->armorLinks(
2303 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2304 continue;
2305 }
2306 } elseif ( $ns == NS_CATEGORY ) {
2307 $s = rtrim( $s . "\n" ); # bug 87
2308
2309 if ( $wasblank ) {
2310 $sortkey = $this->getDefaultSort();
2311 } else {
2312 $sortkey = $text;
2313 }
2314 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2315 $sortkey = str_replace( "\n", '', $sortkey );
2316 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2317 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2318
2319 /**
2320 * Strip the whitespace Category links produce, see bug 87
2321 */
2322 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2323
2324 continue;
2325 }
2326 }
2327
2328 # Self-link checking. For some languages, variants of the title are checked in
2329 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2330 # for linking to a different variant.
2331 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2332 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2333 continue;
2334 }
2335
2336 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2337 # @todo FIXME: Should do batch file existence checks, see comment below
2338 if ( $ns == NS_MEDIA ) {
2339 # Give extensions a chance to select the file revision for us
2340 $options = [];
2341 $descQuery = false;
2342 Hooks::run( 'BeforeParserFetchFileAndTitle',
2343 [ $this, $nt, &$options, &$descQuery ] );
2344 # Fetch and register the file (file title may be different via hooks)
2345 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2346 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2347 $s .= $prefix . $this->armorLinks(
2348 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2349 continue;
2350 }
2351
2352 # Some titles, such as valid special pages or files in foreign repos, should
2353 # be shown as bluelinks even though they're not included in the page table
2354 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2355 # batch file existence checks for NS_FILE and NS_MEDIA
2356 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2357 $this->mOutput->addLink( $nt );
2358 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2359 } else {
2360 # Links will be added to the output link list after checking
2361 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2362 }
2363 }
2364 return $holders;
2365 }
2366
2367 /**
2368 * Render a forced-blue link inline; protect against double expansion of
2369 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2370 * Since this little disaster has to split off the trail text to avoid
2371 * breaking URLs in the following text without breaking trails on the
2372 * wiki links, it's been made into a horrible function.
2373 *
2374 * @param Title $nt
2375 * @param string $text
2376 * @param string $trail
2377 * @param string $prefix
2378 * @return string HTML-wikitext mix oh yuck
2379 */
2380 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2381 list( $inside, $trail ) = Linker::splitTrail( $trail );
2382
2383 if ( $text == '' ) {
2384 $text = htmlspecialchars( $nt->getPrefixedText() );
2385 }
2386
2387 $link = $this->getLinkRenderer()->makeKnownLink(
2388 $nt, new HtmlArmor( "$prefix$text$inside" )
2389 );
2390
2391 return $this->armorLinks( $link ) . $trail;
2392 }
2393
2394 /**
2395 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2396 * going to go through further parsing steps before inline URL expansion.
2397 *
2398 * Not needed quite as much as it used to be since free links are a bit
2399 * more sensible these days. But bracketed links are still an issue.
2400 *
2401 * @param string $text More-or-less HTML
2402 * @return string Less-or-more HTML with NOPARSE bits
2403 */
2404 public function armorLinks( $text ) {
2405 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2406 self::MARKER_PREFIX . "NOPARSE$1", $text );
2407 }
2408
2409 /**
2410 * Return true if subpage links should be expanded on this page.
2411 * @return bool
2412 */
2413 public function areSubpagesAllowed() {
2414 # Some namespaces don't allow subpages
2415 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2416 }
2417
2418 /**
2419 * Handle link to subpage if necessary
2420 *
2421 * @param string $target The source of the link
2422 * @param string &$text The link text, modified as necessary
2423 * @return string The full name of the link
2424 * @private
2425 */
2426 public function maybeDoSubpageLink( $target, &$text ) {
2427 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2428 }
2429
2430 /**
2431 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2432 *
2433 * @param string $text
2434 * @param bool $linestart Whether or not this is at the start of a line.
2435 * @private
2436 * @return string The lists rendered as HTML
2437 */
2438 public function doBlockLevels( $text, $linestart ) {
2439 return BlockLevelPass::doBlockLevels( $text, $linestart );
2440 }
2441
2442 /**
2443 * Return value of a magic variable (like PAGENAME)
2444 *
2445 * @private
2446 *
2447 * @param int $index
2448 * @param bool|PPFrame $frame
2449 *
2450 * @throws MWException
2451 * @return string
2452 */
2453 public function getVariableValue( $index, $frame = false ) {
2454 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2455 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2456
2457 if ( is_null( $this->mTitle ) ) {
2458 // If no title set, bad things are going to happen
2459 // later. Title should always be set since this
2460 // should only be called in the middle of a parse
2461 // operation (but the unit-tests do funky stuff)
2462 throw new MWException( __METHOD__ . ' Should only be '
2463 . ' called while parsing (no title set)' );
2464 }
2465
2466 /**
2467 * Some of these require message or data lookups and can be
2468 * expensive to check many times.
2469 */
2470 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2471 if ( isset( $this->mVarCache[$index] ) ) {
2472 return $this->mVarCache[$index];
2473 }
2474 }
2475
2476 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2477 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2478
2479 $pageLang = $this->getFunctionLang();
2480
2481 switch ( $index ) {
2482 case '!':
2483 $value = '|';
2484 break;
2485 case 'currentmonth':
2486 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2487 break;
2488 case 'currentmonth1':
2489 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2490 break;
2491 case 'currentmonthname':
2492 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2493 break;
2494 case 'currentmonthnamegen':
2495 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2496 break;
2497 case 'currentmonthabbrev':
2498 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2499 break;
2500 case 'currentday':
2501 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2502 break;
2503 case 'currentday2':
2504 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2505 break;
2506 case 'localmonth':
2507 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2508 break;
2509 case 'localmonth1':
2510 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2511 break;
2512 case 'localmonthname':
2513 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2514 break;
2515 case 'localmonthnamegen':
2516 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2517 break;
2518 case 'localmonthabbrev':
2519 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2520 break;
2521 case 'localday':
2522 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2523 break;
2524 case 'localday2':
2525 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2526 break;
2527 case 'pagename':
2528 $value = wfEscapeWikiText( $this->mTitle->getText() );
2529 break;
2530 case 'pagenamee':
2531 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2532 break;
2533 case 'fullpagename':
2534 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2535 break;
2536 case 'fullpagenamee':
2537 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2538 break;
2539 case 'subpagename':
2540 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2541 break;
2542 case 'subpagenamee':
2543 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2544 break;
2545 case 'rootpagename':
2546 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2547 break;
2548 case 'rootpagenamee':
2549 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2550 ' ',
2551 '_',
2552 $this->mTitle->getRootText()
2553 ) ) );
2554 break;
2555 case 'basepagename':
2556 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2557 break;
2558 case 'basepagenamee':
2559 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2560 ' ',
2561 '_',
2562 $this->mTitle->getBaseText()
2563 ) ) );
2564 break;
2565 case 'talkpagename':
2566 if ( $this->mTitle->canTalk() ) {
2567 $talkPage = $this->mTitle->getTalkPage();
2568 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2569 } else {
2570 $value = '';
2571 }
2572 break;
2573 case 'talkpagenamee':
2574 if ( $this->mTitle->canTalk() ) {
2575 $talkPage = $this->mTitle->getTalkPage();
2576 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2577 } else {
2578 $value = '';
2579 }
2580 break;
2581 case 'subjectpagename':
2582 $subjPage = $this->mTitle->getSubjectPage();
2583 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2584 break;
2585 case 'subjectpagenamee':
2586 $subjPage = $this->mTitle->getSubjectPage();
2587 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2588 break;
2589 case 'pageid': // requested in bug 23427
2590 $pageid = $this->getTitle()->getArticleID();
2591 if ( $pageid == 0 ) {
2592 # 0 means the page doesn't exist in the database,
2593 # which means the user is previewing a new page.
2594 # The vary-revision flag must be set, because the magic word
2595 # will have a different value once the page is saved.
2596 $this->mOutput->setFlag( 'vary-revision' );
2597 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2598 }
2599 $value = $pageid ? $pageid : null;
2600 break;
2601 case 'revisionid':
2602 # Let the edit saving system know we should parse the page
2603 # *after* a revision ID has been assigned.
2604 $this->mOutput->setFlag( 'vary-revision-id' );
2605 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2606 $value = $this->mRevisionId;
2607 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2608 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2609 $this->mOutput->setSpeculativeRevIdUsed( $value );
2610 }
2611 break;
2612 case 'revisionday':
2613 # Let the edit saving system know we should parse the page
2614 # *after* a revision ID has been assigned. This is for null edits.
2615 $this->mOutput->setFlag( 'vary-revision' );
2616 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2617 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2618 break;
2619 case 'revisionday2':
2620 # Let the edit saving system know we should parse the page
2621 # *after* a revision ID has been assigned. This is for null edits.
2622 $this->mOutput->setFlag( 'vary-revision' );
2623 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2624 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2625 break;
2626 case 'revisionmonth':
2627 # Let the edit saving system know we should parse the page
2628 # *after* a revision ID has been assigned. This is for null edits.
2629 $this->mOutput->setFlag( 'vary-revision' );
2630 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2631 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2632 break;
2633 case 'revisionmonth1':
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__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2638 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2639 break;
2640 case 'revisionyear':
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__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2645 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2646 break;
2647 case 'revisiontimestamp':
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__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2652 $value = $this->getRevisionTimestamp();
2653 break;
2654 case 'revisionuser':
2655 # Let the edit saving system know we should parse the page
2656 # *after* a revision ID has been assigned for null edits.
2657 $this->mOutput->setFlag( 'vary-user' );
2658 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2659 $value = $this->getRevisionUser();
2660 break;
2661 case 'revisionsize':
2662 $value = $this->getRevisionSize();
2663 break;
2664 case 'namespace':
2665 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2666 break;
2667 case 'namespacee':
2668 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2669 break;
2670 case 'namespacenumber':
2671 $value = $this->mTitle->getNamespace();
2672 break;
2673 case 'talkspace':
2674 $value = $this->mTitle->canTalk()
2675 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2676 : '';
2677 break;
2678 case 'talkspacee':
2679 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2680 break;
2681 case 'subjectspace':
2682 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2683 break;
2684 case 'subjectspacee':
2685 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2686 break;
2687 case 'currentdayname':
2688 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2689 break;
2690 case 'currentyear':
2691 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2692 break;
2693 case 'currenttime':
2694 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2695 break;
2696 case 'currenthour':
2697 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2698 break;
2699 case 'currentweek':
2700 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2701 # int to remove the padding
2702 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2703 break;
2704 case 'currentdow':
2705 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2706 break;
2707 case 'localdayname':
2708 $value = $pageLang->getWeekdayName(
2709 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2710 );
2711 break;
2712 case 'localyear':
2713 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2714 break;
2715 case 'localtime':
2716 $value = $pageLang->time(
2717 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2718 false,
2719 false
2720 );
2721 break;
2722 case 'localhour':
2723 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2724 break;
2725 case 'localweek':
2726 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2727 # int to remove the padding
2728 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2729 break;
2730 case 'localdow':
2731 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2732 break;
2733 case 'numberofarticles':
2734 $value = $pageLang->formatNum( SiteStats::articles() );
2735 break;
2736 case 'numberoffiles':
2737 $value = $pageLang->formatNum( SiteStats::images() );
2738 break;
2739 case 'numberofusers':
2740 $value = $pageLang->formatNum( SiteStats::users() );
2741 break;
2742 case 'numberofactiveusers':
2743 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2744 break;
2745 case 'numberofpages':
2746 $value = $pageLang->formatNum( SiteStats::pages() );
2747 break;
2748 case 'numberofadmins':
2749 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2750 break;
2751 case 'numberofedits':
2752 $value = $pageLang->formatNum( SiteStats::edits() );
2753 break;
2754 case 'currenttimestamp':
2755 $value = wfTimestamp( TS_MW, $ts );
2756 break;
2757 case 'localtimestamp':
2758 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2759 break;
2760 case 'currentversion':
2761 $value = SpecialVersion::getVersion();
2762 break;
2763 case 'articlepath':
2764 return $wgArticlePath;
2765 case 'sitename':
2766 return $wgSitename;
2767 case 'server':
2768 return $wgServer;
2769 case 'servername':
2770 return $wgServerName;
2771 case 'scriptpath':
2772 return $wgScriptPath;
2773 case 'stylepath':
2774 return $wgStylePath;
2775 case 'directionmark':
2776 return $pageLang->getDirMark();
2777 case 'contentlanguage':
2778 global $wgLanguageCode;
2779 return $wgLanguageCode;
2780 case 'cascadingsources':
2781 $value = CoreParserFunctions::cascadingsources( $this );
2782 break;
2783 default:
2784 $ret = null;
2785 Hooks::run(
2786 'ParserGetVariableValueSwitch',
2787 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2788 );
2789
2790 return $ret;
2791 }
2792
2793 if ( $index ) {
2794 $this->mVarCache[$index] = $value;
2795 }
2796
2797 return $value;
2798 }
2799
2800 /**
2801 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2802 *
2803 * @private
2804 */
2805 public function initialiseVariables() {
2806 $variableIDs = MagicWord::getVariableIDs();
2807 $substIDs = MagicWord::getSubstIDs();
2808
2809 $this->mVariables = new MagicWordArray( $variableIDs );
2810 $this->mSubstWords = new MagicWordArray( $substIDs );
2811 }
2812
2813 /**
2814 * Preprocess some wikitext and return the document tree.
2815 * This is the ghost of replace_variables().
2816 *
2817 * @param string $text The text to parse
2818 * @param int $flags Bitwise combination of:
2819 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2820 * included. Default is to assume a direct page view.
2821 *
2822 * The generated DOM tree must depend only on the input text and the flags.
2823 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2824 *
2825 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2826 * change in the DOM tree for a given text, must be passed through the section identifier
2827 * in the section edit link and thus back to extractSections().
2828 *
2829 * The output of this function is currently only cached in process memory, but a persistent
2830 * cache may be implemented at a later date which takes further advantage of these strict
2831 * dependency requirements.
2832 *
2833 * @return PPNode
2834 */
2835 public function preprocessToDom( $text, $flags = 0 ) {
2836 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2837 return $dom;
2838 }
2839
2840 /**
2841 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2842 *
2843 * @param string $s
2844 *
2845 * @return array
2846 */
2847 public static function splitWhitespace( $s ) {
2848 $ltrimmed = ltrim( $s );
2849 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2850 $trimmed = rtrim( $ltrimmed );
2851 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2852 if ( $diff > 0 ) {
2853 $w2 = substr( $ltrimmed, -$diff );
2854 } else {
2855 $w2 = '';
2856 }
2857 return [ $w1, $trimmed, $w2 ];
2858 }
2859
2860 /**
2861 * Replace magic variables, templates, and template arguments
2862 * with the appropriate text. Templates are substituted recursively,
2863 * taking care to avoid infinite loops.
2864 *
2865 * Note that the substitution depends on value of $mOutputType:
2866 * self::OT_WIKI: only {{subst:}} templates
2867 * self::OT_PREPROCESS: templates but not extension tags
2868 * self::OT_HTML: all templates and extension tags
2869 *
2870 * @param string $text The text to transform
2871 * @param bool|PPFrame $frame Object describing the arguments passed to the
2872 * template. Arguments may also be provided as an associative array, as
2873 * was the usual case before MW1.12. Providing arguments this way may be
2874 * useful for extensions wishing to perform variable replacement
2875 * explicitly.
2876 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2877 * double-brace expansion.
2878 * @return string
2879 */
2880 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2881 # Is there any text? Also, Prevent too big inclusions!
2882 $textSize = strlen( $text );
2883 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2884 return $text;
2885 }
2886
2887 if ( $frame === false ) {
2888 $frame = $this->getPreprocessor()->newFrame();
2889 } elseif ( !( $frame instanceof PPFrame ) ) {
2890 wfDebug( __METHOD__ . " called using plain parameters instead of "
2891 . "a PPFrame instance. Creating custom frame.\n" );
2892 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2893 }
2894
2895 $dom = $this->preprocessToDom( $text );
2896 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2897 $text = $frame->expand( $dom, $flags );
2898
2899 return $text;
2900 }
2901
2902 /**
2903 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2904 *
2905 * @param array $args
2906 *
2907 * @return array
2908 */
2909 public static function createAssocArgs( $args ) {
2910 $assocArgs = [];
2911 $index = 1;
2912 foreach ( $args as $arg ) {
2913 $eqpos = strpos( $arg, '=' );
2914 if ( $eqpos === false ) {
2915 $assocArgs[$index++] = $arg;
2916 } else {
2917 $name = trim( substr( $arg, 0, $eqpos ) );
2918 $value = trim( substr( $arg, $eqpos + 1 ) );
2919 if ( $value === false ) {
2920 $value = '';
2921 }
2922 if ( $name !== false ) {
2923 $assocArgs[$name] = $value;
2924 }
2925 }
2926 }
2927
2928 return $assocArgs;
2929 }
2930
2931 /**
2932 * Warn the user when a parser limitation is reached
2933 * Will warn at most once the user per limitation type
2934 *
2935 * The results are shown during preview and run through the Parser (See EditPage.php)
2936 *
2937 * @param string $limitationType Should be one of:
2938 * 'expensive-parserfunction' (corresponding messages:
2939 * 'expensive-parserfunction-warning',
2940 * 'expensive-parserfunction-category')
2941 * 'post-expand-template-argument' (corresponding messages:
2942 * 'post-expand-template-argument-warning',
2943 * 'post-expand-template-argument-category')
2944 * 'post-expand-template-inclusion' (corresponding messages:
2945 * 'post-expand-template-inclusion-warning',
2946 * 'post-expand-template-inclusion-category')
2947 * 'node-count-exceeded' (corresponding messages:
2948 * 'node-count-exceeded-warning',
2949 * 'node-count-exceeded-category')
2950 * 'expansion-depth-exceeded' (corresponding messages:
2951 * 'expansion-depth-exceeded-warning',
2952 * 'expansion-depth-exceeded-category')
2953 * @param string|int|null $current Current value
2954 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2955 * exceeded, provide the values (optional)
2956 */
2957 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2958 # does no harm if $current and $max are present but are unnecessary for the message
2959 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2960 # only during preview, and that would split the parser cache unnecessarily.
2961 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2962 ->text();
2963 $this->mOutput->addWarning( $warning );
2964 $this->addTrackingCategory( "$limitationType-category" );
2965 }
2966
2967 /**
2968 * Return the text of a template, after recursively
2969 * replacing any variables or templates within the template.
2970 *
2971 * @param array $piece The parts of the template
2972 * $piece['title']: the title, i.e. the part before the |
2973 * $piece['parts']: the parameter array
2974 * $piece['lineStart']: whether the brace was at the start of a line
2975 * @param PPFrame $frame The current frame, contains template arguments
2976 * @throws Exception
2977 * @return string The text of the template
2978 */
2979 public function braceSubstitution( $piece, $frame ) {
2980
2981 // Flags
2982
2983 // $text has been filled
2984 $found = false;
2985 // wiki markup in $text should be escaped
2986 $nowiki = false;
2987 // $text is HTML, armour it against wikitext transformation
2988 $isHTML = false;
2989 // Force interwiki transclusion to be done in raw mode not rendered
2990 $forceRawInterwiki = false;
2991 // $text is a DOM node needing expansion in a child frame
2992 $isChildObj = false;
2993 // $text is a DOM node needing expansion in the current frame
2994 $isLocalObj = false;
2995
2996 # Title object, where $text came from
2997 $title = false;
2998
2999 # $part1 is the bit before the first |, and must contain only title characters.
3000 # Various prefixes will be stripped from it later.
3001 $titleWithSpaces = $frame->expand( $piece['title'] );
3002 $part1 = trim( $titleWithSpaces );
3003 $titleText = false;
3004
3005 # Original title text preserved for various purposes
3006 $originalTitle = $part1;
3007
3008 # $args is a list of argument nodes, starting from index 0, not including $part1
3009 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3010 # below won't work b/c this $args isn't an object
3011 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3012
3013 $profileSection = null; // profile templates
3014
3015 # SUBST
3016 if ( !$found ) {
3017 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3018
3019 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3020 # Decide whether to expand template or keep wikitext as-is.
3021 if ( $this->ot['wiki'] ) {
3022 if ( $substMatch === false ) {
3023 $literal = true; # literal when in PST with no prefix
3024 } else {
3025 $literal = false; # expand when in PST with subst: or safesubst:
3026 }
3027 } else {
3028 if ( $substMatch == 'subst' ) {
3029 $literal = true; # literal when not in PST with plain subst:
3030 } else {
3031 $literal = false; # expand when not in PST with safesubst: or no prefix
3032 }
3033 }
3034 if ( $literal ) {
3035 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3036 $isLocalObj = true;
3037 $found = true;
3038 }
3039 }
3040
3041 # Variables
3042 if ( !$found && $args->getLength() == 0 ) {
3043 $id = $this->mVariables->matchStartToEnd( $part1 );
3044 if ( $id !== false ) {
3045 $text = $this->getVariableValue( $id, $frame );
3046 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3047 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3048 }
3049 $found = true;
3050 }
3051 }
3052
3053 # MSG, MSGNW and RAW
3054 if ( !$found ) {
3055 # Check for MSGNW:
3056 $mwMsgnw = MagicWord::get( 'msgnw' );
3057 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3058 $nowiki = true;
3059 } else {
3060 # Remove obsolete MSG:
3061 $mwMsg = MagicWord::get( 'msg' );
3062 $mwMsg->matchStartAndRemove( $part1 );
3063 }
3064
3065 # Check for RAW:
3066 $mwRaw = MagicWord::get( 'raw' );
3067 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3068 $forceRawInterwiki = true;
3069 }
3070 }
3071
3072 # Parser functions
3073 if ( !$found ) {
3074 $colonPos = strpos( $part1, ':' );
3075 if ( $colonPos !== false ) {
3076 $func = substr( $part1, 0, $colonPos );
3077 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3078 $argsLength = $args->getLength();
3079 for ( $i = 0; $i < $argsLength; $i++ ) {
3080 $funcArgs[] = $args->item( $i );
3081 }
3082 try {
3083 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3084 } catch ( Exception $ex ) {
3085 throw $ex;
3086 }
3087
3088 # The interface for parser functions allows for extracting
3089 # flags into the local scope. Extract any forwarded flags
3090 # here.
3091 extract( $result );
3092 }
3093 }
3094
3095 # Finish mangling title and then check for loops.
3096 # Set $title to a Title object and $titleText to the PDBK
3097 if ( !$found ) {
3098 $ns = NS_TEMPLATE;
3099 # Split the title into page and subpage
3100 $subpage = '';
3101 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3102 if ( $part1 !== $relative ) {
3103 $part1 = $relative;
3104 $ns = $this->mTitle->getNamespace();
3105 }
3106 $title = Title::newFromText( $part1, $ns );
3107 if ( $title ) {
3108 $titleText = $title->getPrefixedText();
3109 # Check for language variants if the template is not found
3110 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3111 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3112 }
3113 # Do recursion depth check
3114 $limit = $this->mOptions->getMaxTemplateDepth();
3115 if ( $frame->depth >= $limit ) {
3116 $found = true;
3117 $text = '<span class="error">'
3118 . wfMessage( 'parser-template-recursion-depth-warning' )
3119 ->numParams( $limit )->inContentLanguage()->text()
3120 . '</span>';
3121 }
3122 }
3123 }
3124
3125 # Load from database
3126 if ( !$found && $title ) {
3127 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3128 if ( !$title->isExternal() ) {
3129 if ( $title->isSpecialPage()
3130 && $this->mOptions->getAllowSpecialInclusion()
3131 && $this->ot['html']
3132 ) {
3133 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3134 // Pass the template arguments as URL parameters.
3135 // "uselang" will have no effect since the Language object
3136 // is forced to the one defined in ParserOptions.
3137 $pageArgs = [];
3138 $argsLength = $args->getLength();
3139 for ( $i = 0; $i < $argsLength; $i++ ) {
3140 $bits = $args->item( $i )->splitArg();
3141 if ( strval( $bits['index'] ) === '' ) {
3142 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3143 $value = trim( $frame->expand( $bits['value'] ) );
3144 $pageArgs[$name] = $value;
3145 }
3146 }
3147
3148 // Create a new context to execute the special page
3149 $context = new RequestContext;
3150 $context->setTitle( $title );
3151 $context->setRequest( new FauxRequest( $pageArgs ) );
3152 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3153 $context->setUser( $this->getUser() );
3154 } else {
3155 // If this page is cached, then we better not be per user.
3156 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3157 }
3158 $context->setLanguage( $this->mOptions->getUserLangObj() );
3159 $ret = SpecialPageFactory::capturePath( $title, $context, $this->getLinkRenderer() );
3160 if ( $ret ) {
3161 $text = $context->getOutput()->getHTML();
3162 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3163 $found = true;
3164 $isHTML = true;
3165 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3166 $this->mOutput->updateCacheExpiry( $specialPage->maxIncludeCacheTime() );
3167 }
3168 }
3169 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3170 $found = false; # access denied
3171 wfDebug( __METHOD__ . ": template inclusion denied for " .
3172 $title->getPrefixedDBkey() . "\n" );
3173 } else {
3174 list( $text, $title ) = $this->getTemplateDom( $title );
3175 if ( $text !== false ) {
3176 $found = true;
3177 $isChildObj = true;
3178 }
3179 }
3180
3181 # If the title is valid but undisplayable, make a link to it
3182 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3183 $text = "[[:$titleText]]";
3184 $found = true;
3185 }
3186 } elseif ( $title->isTrans() ) {
3187 # Interwiki transclusion
3188 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3189 $text = $this->interwikiTransclude( $title, 'render' );
3190 $isHTML = true;
3191 } else {
3192 $text = $this->interwikiTransclude( $title, 'raw' );
3193 # Preprocess it like a template
3194 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3195 $isChildObj = true;
3196 }
3197 $found = true;
3198 }
3199
3200 # Do infinite loop check
3201 # This has to be done after redirect resolution to avoid infinite loops via redirects
3202 if ( !$frame->loopCheck( $title ) ) {
3203 $found = true;
3204 $text = '<span class="error">'
3205 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3206 . '</span>';
3207 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3208 }
3209 }
3210
3211 # If we haven't found text to substitute by now, we're done
3212 # Recover the source wikitext and return it
3213 if ( !$found ) {
3214 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3215 if ( $profileSection ) {
3216 $this->mProfiler->scopedProfileOut( $profileSection );
3217 }
3218 return [ 'object' => $text ];
3219 }
3220
3221 # Expand DOM-style return values in a child frame
3222 if ( $isChildObj ) {
3223 # Clean up argument array
3224 $newFrame = $frame->newChild( $args, $title );
3225
3226 if ( $nowiki ) {
3227 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3228 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3229 # Expansion is eligible for the empty-frame cache
3230 $text = $newFrame->cachedExpand( $titleText, $text );
3231 } else {
3232 # Uncached expansion
3233 $text = $newFrame->expand( $text );
3234 }
3235 }
3236 if ( $isLocalObj && $nowiki ) {
3237 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3238 $isLocalObj = false;
3239 }
3240
3241 if ( $profileSection ) {
3242 $this->mProfiler->scopedProfileOut( $profileSection );
3243 }
3244
3245 # Replace raw HTML by a placeholder
3246 if ( $isHTML ) {
3247 $text = $this->insertStripItem( $text );
3248 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3249 # Escape nowiki-style return values
3250 $text = wfEscapeWikiText( $text );
3251 } elseif ( is_string( $text )
3252 && !$piece['lineStart']
3253 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3254 ) {
3255 # Bug 529: if the template begins with a table or block-level
3256 # element, it should be treated as beginning a new line.
3257 # This behavior is somewhat controversial.
3258 $text = "\n" . $text;
3259 }
3260
3261 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3262 # Error, oversize inclusion
3263 if ( $titleText !== false ) {
3264 # Make a working, properly escaped link if possible (bug 23588)
3265 $text = "[[:$titleText]]";
3266 } else {
3267 # This will probably not be a working link, but at least it may
3268 # provide some hint of where the problem is
3269 preg_replace( '/^:/', '', $originalTitle );
3270 $text = "[[:$originalTitle]]";
3271 }
3272 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3273 . 'post-expand include size too large -->' );
3274 $this->limitationWarn( 'post-expand-template-inclusion' );
3275 }
3276
3277 if ( $isLocalObj ) {
3278 $ret = [ 'object' => $text ];
3279 } else {
3280 $ret = [ 'text' => $text ];
3281 }
3282
3283 return $ret;
3284 }
3285
3286 /**
3287 * Call a parser function and return an array with text and flags.
3288 *
3289 * The returned array will always contain a boolean 'found', indicating
3290 * whether the parser function was found or not. It may also contain the
3291 * following:
3292 * text: string|object, resulting wikitext or PP DOM object
3293 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3294 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3295 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3296 * nowiki: bool, wiki markup in $text should be escaped
3297 *
3298 * @since 1.21
3299 * @param PPFrame $frame The current frame, contains template arguments
3300 * @param string $function Function name
3301 * @param array $args Arguments to the function
3302 * @throws MWException
3303 * @return array
3304 */
3305 public function callParserFunction( $frame, $function, array $args = [] ) {
3306 global $wgContLang;
3307
3308 # Case sensitive functions
3309 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3310 $function = $this->mFunctionSynonyms[1][$function];
3311 } else {
3312 # Case insensitive functions
3313 $function = $wgContLang->lc( $function );
3314 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3315 $function = $this->mFunctionSynonyms[0][$function];
3316 } else {
3317 return [ 'found' => false ];
3318 }
3319 }
3320
3321 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3322
3323 # Workaround for PHP bug 35229 and similar
3324 if ( !is_callable( $callback ) ) {
3325 throw new MWException( "Tag hook for $function is not callable\n" );
3326 }
3327
3328 $allArgs = [ &$this ];
3329 if ( $flags & self::SFH_OBJECT_ARGS ) {
3330 # Convert arguments to PPNodes and collect for appending to $allArgs
3331 $funcArgs = [];
3332 foreach ( $args as $k => $v ) {
3333 if ( $v instanceof PPNode || $k === 0 ) {
3334 $funcArgs[] = $v;
3335 } else {
3336 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3337 }
3338 }
3339
3340 # Add a frame parameter, and pass the arguments as an array
3341 $allArgs[] = $frame;
3342 $allArgs[] = $funcArgs;
3343 } else {
3344 # Convert arguments to plain text and append to $allArgs
3345 foreach ( $args as $k => $v ) {
3346 if ( $v instanceof PPNode ) {
3347 $allArgs[] = trim( $frame->expand( $v ) );
3348 } elseif ( is_int( $k ) && $k >= 0 ) {
3349 $allArgs[] = trim( $v );
3350 } else {
3351 $allArgs[] = trim( "$k=$v" );
3352 }
3353 }
3354 }
3355
3356 $result = call_user_func_array( $callback, $allArgs );
3357
3358 # The interface for function hooks allows them to return a wikitext
3359 # string or an array containing the string and any flags. This mungs
3360 # things around to match what this method should return.
3361 if ( !is_array( $result ) ) {
3362 $result =[
3363 'found' => true,
3364 'text' => $result,
3365 ];
3366 } else {
3367 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3368 $result['text'] = $result[0];
3369 }
3370 unset( $result[0] );
3371 $result += [
3372 'found' => true,
3373 ];
3374 }
3375
3376 $noparse = true;
3377 $preprocessFlags = 0;
3378 if ( isset( $result['noparse'] ) ) {
3379 $noparse = $result['noparse'];
3380 }
3381 if ( isset( $result['preprocessFlags'] ) ) {
3382 $preprocessFlags = $result['preprocessFlags'];
3383 }
3384
3385 if ( !$noparse ) {
3386 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3387 $result['isChildObj'] = true;
3388 }
3389
3390 return $result;
3391 }
3392
3393 /**
3394 * Get the semi-parsed DOM representation of a template with a given title,
3395 * and its redirect destination title. Cached.
3396 *
3397 * @param Title $title
3398 *
3399 * @return array
3400 */
3401 public function getTemplateDom( $title ) {
3402 $cacheTitle = $title;
3403 $titleText = $title->getPrefixedDBkey();
3404
3405 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3406 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3407 $title = Title::makeTitle( $ns, $dbk );
3408 $titleText = $title->getPrefixedDBkey();
3409 }
3410 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3411 return [ $this->mTplDomCache[$titleText], $title ];
3412 }
3413
3414 # Cache miss, go to the database
3415 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3416
3417 if ( $text === false ) {
3418 $this->mTplDomCache[$titleText] = false;
3419 return [ false, $title ];
3420 }
3421
3422 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3423 $this->mTplDomCache[$titleText] = $dom;
3424
3425 if ( !$title->equals( $cacheTitle ) ) {
3426 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3427 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3428 }
3429
3430 return [ $dom, $title ];
3431 }
3432
3433 /**
3434 * Fetch the current revision of a given title. Note that the revision
3435 * (and even the title) may not exist in the database, so everything
3436 * contributing to the output of the parser should use this method
3437 * where possible, rather than getting the revisions themselves. This
3438 * method also caches its results, so using it benefits performance.
3439 *
3440 * @since 1.24
3441 * @param Title $title
3442 * @return Revision
3443 */
3444 public function fetchCurrentRevisionOfTitle( $title ) {
3445 $cacheKey = $title->getPrefixedDBkey();
3446 if ( !$this->currentRevisionCache ) {
3447 $this->currentRevisionCache = new MapCacheLRU( 100 );
3448 }
3449 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3450 $this->currentRevisionCache->set( $cacheKey,
3451 // Defaults to Parser::statelessFetchRevision()
3452 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3453 );
3454 }
3455 return $this->currentRevisionCache->get( $cacheKey );
3456 }
3457
3458 /**
3459 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3460 * without passing them on to it.
3461 *
3462 * @since 1.24
3463 * @param Title $title
3464 * @param Parser|bool $parser
3465 * @return Revision
3466 */
3467 public static function statelessFetchRevision( $title, $parser = false ) {
3468 return Revision::newFromTitle( $title );
3469 }
3470
3471 /**
3472 * Fetch the unparsed text of a template and register a reference to it.
3473 * @param Title $title
3474 * @return array ( string or false, Title )
3475 */
3476 public function fetchTemplateAndTitle( $title ) {
3477 // Defaults to Parser::statelessFetchTemplate()
3478 $templateCb = $this->mOptions->getTemplateCallback();
3479 $stuff = call_user_func( $templateCb, $title, $this );
3480 // We use U+007F DELETE to distinguish strip markers from regular text.
3481 $text = $stuff['text'];
3482 if ( is_string( $stuff['text'] ) ) {
3483 $text = strtr( $text, "\x7f", "?" );
3484 }
3485 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3486 if ( isset( $stuff['deps'] ) ) {
3487 foreach ( $stuff['deps'] as $dep ) {
3488 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3489 if ( $dep['title']->equals( $this->getTitle() ) ) {
3490 // If we transclude ourselves, the final result
3491 // will change based on the new version of the page
3492 $this->mOutput->setFlag( 'vary-revision' );
3493 }
3494 }
3495 }
3496 return [ $text, $finalTitle ];
3497 }
3498
3499 /**
3500 * Fetch the unparsed text of a template and register a reference to it.
3501 * @param Title $title
3502 * @return string|bool
3503 */
3504 public function fetchTemplate( $title ) {
3505 return $this->fetchTemplateAndTitle( $title )[0];
3506 }
3507
3508 /**
3509 * Static function to get a template
3510 * Can be overridden via ParserOptions::setTemplateCallback().
3511 *
3512 * @param Title $title
3513 * @param bool|Parser $parser
3514 *
3515 * @return array
3516 */
3517 public static function statelessFetchTemplate( $title, $parser = false ) {
3518 $text = $skip = false;
3519 $finalTitle = $title;
3520 $deps = [];
3521
3522 # Loop to fetch the article, with up to 1 redirect
3523 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3524 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3525 // @codingStandardsIgnoreEnd
3526 # Give extensions a chance to select the revision instead
3527 $id = false; # Assume current
3528 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3529 [ $parser, $title, &$skip, &$id ] );
3530
3531 if ( $skip ) {
3532 $text = false;
3533 $deps[] = [
3534 'title' => $title,
3535 'page_id' => $title->getArticleID(),
3536 'rev_id' => null
3537 ];
3538 break;
3539 }
3540 # Get the revision
3541 if ( $id ) {
3542 $rev = Revision::newFromId( $id );
3543 } elseif ( $parser ) {
3544 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3545 } else {
3546 $rev = Revision::newFromTitle( $title );
3547 }
3548 $rev_id = $rev ? $rev->getId() : 0;
3549 # If there is no current revision, there is no page
3550 if ( $id === false && !$rev ) {
3551 $linkCache = LinkCache::singleton();
3552 $linkCache->addBadLinkObj( $title );
3553 }
3554
3555 $deps[] = [
3556 'title' => $title,
3557 'page_id' => $title->getArticleID(),
3558 'rev_id' => $rev_id ];
3559 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3560 # We fetched a rev from a different title; register it too...
3561 $deps[] = [
3562 'title' => $rev->getTitle(),
3563 'page_id' => $rev->getPage(),
3564 'rev_id' => $rev_id ];
3565 }
3566
3567 if ( $rev ) {
3568 $content = $rev->getContent();
3569 $text = $content ? $content->getWikitextForTransclusion() : null;
3570
3571 if ( $text === false || $text === null ) {
3572 $text = false;
3573 break;
3574 }
3575 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3576 global $wgContLang;
3577 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3578 if ( !$message->exists() ) {
3579 $text = false;
3580 break;
3581 }
3582 $content = $message->content();
3583 $text = $message->plain();
3584 } else {
3585 break;
3586 }
3587 if ( !$content ) {
3588 break;
3589 }
3590 # Redirect?
3591 $finalTitle = $title;
3592 $title = $content->getRedirectTarget();
3593 }
3594 return [
3595 'text' => $text,
3596 'finalTitle' => $finalTitle,
3597 'deps' => $deps ];
3598 }
3599
3600 /**
3601 * Fetch a file and its title and register a reference to it.
3602 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3603 * @param Title $title
3604 * @param array $options Array of options to RepoGroup::findFile
3605 * @return File|bool
3606 */
3607 public function fetchFile( $title, $options = [] ) {
3608 return $this->fetchFileAndTitle( $title, $options )[0];
3609 }
3610
3611 /**
3612 * Fetch a file and its title and register a reference to it.
3613 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3614 * @param Title $title
3615 * @param array $options Array of options to RepoGroup::findFile
3616 * @return array ( File or false, Title of file )
3617 */
3618 public function fetchFileAndTitle( $title, $options = [] ) {
3619 $file = $this->fetchFileNoRegister( $title, $options );
3620
3621 $time = $file ? $file->getTimestamp() : false;
3622 $sha1 = $file ? $file->getSha1() : false;
3623 # Register the file as a dependency...
3624 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3625 if ( $file && !$title->equals( $file->getTitle() ) ) {
3626 # Update fetched file title
3627 $title = $file->getTitle();
3628 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3629 }
3630 return [ $file, $title ];
3631 }
3632
3633 /**
3634 * Helper function for fetchFileAndTitle.
3635 *
3636 * Also useful if you need to fetch a file but not use it yet,
3637 * for example to get the file's handler.
3638 *
3639 * @param Title $title
3640 * @param array $options Array of options to RepoGroup::findFile
3641 * @return File|bool
3642 */
3643 protected function fetchFileNoRegister( $title, $options = [] ) {
3644 if ( isset( $options['broken'] ) ) {
3645 $file = false; // broken thumbnail forced by hook
3646 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3647 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3648 } else { // get by (name,timestamp)
3649 $file = wfFindFile( $title, $options );
3650 }
3651 return $file;
3652 }
3653
3654 /**
3655 * Transclude an interwiki link.
3656 *
3657 * @param Title $title
3658 * @param string $action
3659 *
3660 * @return string
3661 */
3662 public function interwikiTransclude( $title, $action ) {
3663 global $wgEnableScaryTranscluding;
3664
3665 if ( !$wgEnableScaryTranscluding ) {
3666 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3667 }
3668
3669 $url = $title->getFullURL( [ 'action' => $action ] );
3670
3671 if ( strlen( $url ) > 255 ) {
3672 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3673 }
3674 return $this->fetchScaryTemplateMaybeFromCache( $url );
3675 }
3676
3677 /**
3678 * @param string $url
3679 * @return mixed|string
3680 */
3681 public function fetchScaryTemplateMaybeFromCache( $url ) {
3682 global $wgTranscludeCacheExpiry;
3683 $dbr = wfGetDB( DB_SLAVE );
3684 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3685 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3686 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3687 if ( $obj ) {
3688 return $obj->tc_contents;
3689 }
3690
3691 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3692 $status = $req->execute(); // Status object
3693 if ( $status->isOK() ) {
3694 $text = $req->getContent();
3695 } elseif ( $req->getStatus() != 200 ) {
3696 // Though we failed to fetch the content, this status is useless.
3697 return wfMessage( 'scarytranscludefailed-httpstatus' )
3698 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3699 } else {
3700 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3701 }
3702
3703 $dbw = wfGetDB( DB_MASTER );
3704 $dbw->replace( 'transcache', [ 'tc_url' ], [
3705 'tc_url' => $url,
3706 'tc_time' => $dbw->timestamp( time() ),
3707 'tc_contents' => $text
3708 ] );
3709 return $text;
3710 }
3711
3712 /**
3713 * Triple brace replacement -- used for template arguments
3714 * @private
3715 *
3716 * @param array $piece
3717 * @param PPFrame $frame
3718 *
3719 * @return array
3720 */
3721 public function argSubstitution( $piece, $frame ) {
3722
3723 $error = false;
3724 $parts = $piece['parts'];
3725 $nameWithSpaces = $frame->expand( $piece['title'] );
3726 $argName = trim( $nameWithSpaces );
3727 $object = false;
3728 $text = $frame->getArgument( $argName );
3729 if ( $text === false && $parts->getLength() > 0
3730 && ( $this->ot['html']
3731 || $this->ot['pre']
3732 || ( $this->ot['wiki'] && $frame->isTemplate() )
3733 )
3734 ) {
3735 # No match in frame, use the supplied default
3736 $object = $parts->item( 0 )->getChildren();
3737 }
3738 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3739 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3740 $this->limitationWarn( 'post-expand-template-argument' );
3741 }
3742
3743 if ( $text === false && $object === false ) {
3744 # No match anywhere
3745 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3746 }
3747 if ( $error !== false ) {
3748 $text .= $error;
3749 }
3750 if ( $object !== false ) {
3751 $ret = [ 'object' => $object ];
3752 } else {
3753 $ret = [ 'text' => $text ];
3754 }
3755
3756 return $ret;
3757 }
3758
3759 /**
3760 * Return the text to be used for a given extension tag.
3761 * This is the ghost of strip().
3762 *
3763 * @param array $params Associative array of parameters:
3764 * name PPNode for the tag name
3765 * attr PPNode for unparsed text where tag attributes are thought to be
3766 * attributes Optional associative array of parsed attributes
3767 * inner Contents of extension element
3768 * noClose Original text did not have a close tag
3769 * @param PPFrame $frame
3770 *
3771 * @throws MWException
3772 * @return string
3773 */
3774 public function extensionSubstitution( $params, $frame ) {
3775 $name = $frame->expand( $params['name'] );
3776 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3777 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3778 $marker = self::MARKER_PREFIX . "-$name-"
3779 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3780
3781 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3782 ( $this->ot['html'] || $this->ot['pre'] );
3783 if ( $isFunctionTag ) {
3784 $markerType = 'none';
3785 } else {
3786 $markerType = 'general';
3787 }
3788 if ( $this->ot['html'] || $isFunctionTag ) {
3789 $name = strtolower( $name );
3790 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3791 if ( isset( $params['attributes'] ) ) {
3792 $attributes = $attributes + $params['attributes'];
3793 }
3794
3795 if ( isset( $this->mTagHooks[$name] ) ) {
3796 # Workaround for PHP bug 35229 and similar
3797 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3798 throw new MWException( "Tag hook for $name is not callable\n" );
3799 }
3800 $output = call_user_func_array( $this->mTagHooks[$name],
3801 [ $content, $attributes, $this, $frame ] );
3802 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3803 list( $callback, ) = $this->mFunctionTagHooks[$name];
3804 if ( !is_callable( $callback ) ) {
3805 throw new MWException( "Tag hook for $name is not callable\n" );
3806 }
3807
3808 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3809 } else {
3810 $output = '<span class="error">Invalid tag extension name: ' .
3811 htmlspecialchars( $name ) . '</span>';
3812 }
3813
3814 if ( is_array( $output ) ) {
3815 # Extract flags to local scope (to override $markerType)
3816 $flags = $output;
3817 $output = $flags[0];
3818 unset( $flags[0] );
3819 extract( $flags );
3820 }
3821 } else {
3822 if ( is_null( $attrText ) ) {
3823 $attrText = '';
3824 }
3825 if ( isset( $params['attributes'] ) ) {
3826 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3827 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3828 htmlspecialchars( $attrValue ) . '"';
3829 }
3830 }
3831 if ( $content === null ) {
3832 $output = "<$name$attrText/>";
3833 } else {
3834 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3835 $output = "<$name$attrText>$content$close";
3836 }
3837 }
3838
3839 if ( $markerType === 'none' ) {
3840 return $output;
3841 } elseif ( $markerType === 'nowiki' ) {
3842 $this->mStripState->addNoWiki( $marker, $output );
3843 } elseif ( $markerType === 'general' ) {
3844 $this->mStripState->addGeneral( $marker, $output );
3845 } else {
3846 throw new MWException( __METHOD__ . ': invalid marker type' );
3847 }
3848 return $marker;
3849 }
3850
3851 /**
3852 * Increment an include size counter
3853 *
3854 * @param string $type The type of expansion
3855 * @param int $size The size of the text
3856 * @return bool False if this inclusion would take it over the maximum, true otherwise
3857 */
3858 public function incrementIncludeSize( $type, $size ) {
3859 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3860 return false;
3861 } else {
3862 $this->mIncludeSizes[$type] += $size;
3863 return true;
3864 }
3865 }
3866
3867 /**
3868 * Increment the expensive function count
3869 *
3870 * @return bool False if the limit has been exceeded
3871 */
3872 public function incrementExpensiveFunctionCount() {
3873 $this->mExpensiveFunctionCount++;
3874 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3875 }
3876
3877 /**
3878 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3879 * Fills $this->mDoubleUnderscores, returns the modified text
3880 *
3881 * @param string $text
3882 *
3883 * @return string
3884 */
3885 public function doDoubleUnderscore( $text ) {
3886
3887 # The position of __TOC__ needs to be recorded
3888 $mw = MagicWord::get( 'toc' );
3889 if ( $mw->match( $text ) ) {
3890 $this->mShowToc = true;
3891 $this->mForceTocPosition = true;
3892
3893 # Set a placeholder. At the end we'll fill it in with the TOC.
3894 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3895
3896 # Only keep the first one.
3897 $text = $mw->replace( '', $text );
3898 }
3899
3900 # Now match and remove the rest of them
3901 $mwa = MagicWord::getDoubleUnderscoreArray();
3902 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3903
3904 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3905 $this->mOutput->mNoGallery = true;
3906 }
3907 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3908 $this->mShowToc = false;
3909 }
3910 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3911 && $this->mTitle->getNamespace() == NS_CATEGORY
3912 ) {
3913 $this->addTrackingCategory( 'hidden-category-category' );
3914 }
3915 # (bug 8068) Allow control over whether robots index a page.
3916 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3917 # is not desirable, the last one on the page should win.
3918 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3919 $this->mOutput->setIndexPolicy( 'noindex' );
3920 $this->addTrackingCategory( 'noindex-category' );
3921 }
3922 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3923 $this->mOutput->setIndexPolicy( 'index' );
3924 $this->addTrackingCategory( 'index-category' );
3925 }
3926
3927 # Cache all double underscores in the database
3928 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3929 $this->mOutput->setProperty( $key, '' );
3930 }
3931
3932 return $text;
3933 }
3934
3935 /**
3936 * @see ParserOutput::addTrackingCategory()
3937 * @param string $msg Message key
3938 * @return bool Whether the addition was successful
3939 */
3940 public function addTrackingCategory( $msg ) {
3941 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3942 }
3943
3944 /**
3945 * This function accomplishes several tasks:
3946 * 1) Auto-number headings if that option is enabled
3947 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3948 * 3) Add a Table of contents on the top for users who have enabled the option
3949 * 4) Auto-anchor headings
3950 *
3951 * It loops through all headlines, collects the necessary data, then splits up the
3952 * string and re-inserts the newly formatted headlines.
3953 *
3954 * @param string $text
3955 * @param string $origText Original, untouched wikitext
3956 * @param bool $isMain
3957 * @return mixed|string
3958 * @private
3959 */
3960 public function formatHeadings( $text, $origText, $isMain = true ) {
3961 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
3962
3963 # Inhibit editsection links if requested in the page
3964 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3965 $maybeShowEditLink = $showEditLink = false;
3966 } else {
3967 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3968 $showEditLink = $this->mOptions->getEditSection();
3969 }
3970 if ( $showEditLink ) {
3971 $this->mOutput->setEditSectionTokens( true );
3972 }
3973
3974 # Get all headlines for numbering them and adding funky stuff like [edit]
3975 # links - this is for later, but we need the number of headlines right now
3976 $matches = [];
3977 $numMatches = preg_match_all(
3978 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
3979 $text,
3980 $matches
3981 );
3982
3983 # if there are fewer than 4 headlines in the article, do not show TOC
3984 # unless it's been explicitly enabled.
3985 $enoughToc = $this->mShowToc &&
3986 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3987
3988 # Allow user to stipulate that a page should have a "new section"
3989 # link added via __NEWSECTIONLINK__
3990 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3991 $this->mOutput->setNewSection( true );
3992 }
3993
3994 # Allow user to remove the "new section"
3995 # link via __NONEWSECTIONLINK__
3996 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3997 $this->mOutput->hideNewSection( true );
3998 }
3999
4000 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4001 # override above conditions and always show TOC above first header
4002 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4003 $this->mShowToc = true;
4004 $enoughToc = true;
4005 }
4006
4007 # headline counter
4008 $headlineCount = 0;
4009 $numVisible = 0;
4010
4011 # Ugh .. the TOC should have neat indentation levels which can be
4012 # passed to the skin functions. These are determined here
4013 $toc = '';
4014 $full = '';
4015 $head = [];
4016 $sublevelCount = [];
4017 $levelCount = [];
4018 $level = 0;
4019 $prevlevel = 0;
4020 $toclevel = 0;
4021 $prevtoclevel = 0;
4022 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4023 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4024 $oldType = $this->mOutputType;
4025 $this->setOutputType( self::OT_WIKI );
4026 $frame = $this->getPreprocessor()->newFrame();
4027 $root = $this->preprocessToDom( $origText );
4028 $node = $root->getFirstChild();
4029 $byteOffset = 0;
4030 $tocraw = [];
4031 $refers = [];
4032
4033 $headlines = $numMatches !== false ? $matches[3] : [];
4034
4035 foreach ( $headlines as $headline ) {
4036 $isTemplate = false;
4037 $titleText = false;
4038 $sectionIndex = false;
4039 $numbering = '';
4040 $markerMatches = [];
4041 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4042 $serial = $markerMatches[1];
4043 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4044 $isTemplate = ( $titleText != $baseTitleText );
4045 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4046 }
4047
4048 if ( $toclevel ) {
4049 $prevlevel = $level;
4050 }
4051 $level = $matches[1][$headlineCount];
4052
4053 if ( $level > $prevlevel ) {
4054 # Increase TOC level
4055 $toclevel++;
4056 $sublevelCount[$toclevel] = 0;
4057 if ( $toclevel < $wgMaxTocLevel ) {
4058 $prevtoclevel = $toclevel;
4059 $toc .= Linker::tocIndent();
4060 $numVisible++;
4061 }
4062 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4063 # Decrease TOC level, find level to jump to
4064
4065 for ( $i = $toclevel; $i > 0; $i-- ) {
4066 if ( $levelCount[$i] == $level ) {
4067 # Found last matching level
4068 $toclevel = $i;
4069 break;
4070 } elseif ( $levelCount[$i] < $level ) {
4071 # Found first matching level below current level
4072 $toclevel = $i + 1;
4073 break;
4074 }
4075 }
4076 if ( $i == 0 ) {
4077 $toclevel = 1;
4078 }
4079 if ( $toclevel < $wgMaxTocLevel ) {
4080 if ( $prevtoclevel < $wgMaxTocLevel ) {
4081 # Unindent only if the previous toc level was shown :p
4082 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4083 $prevtoclevel = $toclevel;
4084 } else {
4085 $toc .= Linker::tocLineEnd();
4086 }
4087 }
4088 } else {
4089 # No change in level, end TOC line
4090 if ( $toclevel < $wgMaxTocLevel ) {
4091 $toc .= Linker::tocLineEnd();
4092 }
4093 }
4094
4095 $levelCount[$toclevel] = $level;
4096
4097 # count number of headlines for each level
4098 $sublevelCount[$toclevel]++;
4099 $dot = 0;
4100 for ( $i = 1; $i <= $toclevel; $i++ ) {
4101 if ( !empty( $sublevelCount[$i] ) ) {
4102 if ( $dot ) {
4103 $numbering .= '.';
4104 }
4105 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4106 $dot = 1;
4107 }
4108 }
4109
4110 # The safe header is a version of the header text safe to use for links
4111
4112 # Remove link placeholders by the link text.
4113 # <!--LINK number-->
4114 # turns into
4115 # link text with suffix
4116 # Do this before unstrip since link text can contain strip markers
4117 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4118
4119 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4120 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4121
4122 # Strip out HTML (first regex removes any tag not allowed)
4123 # Allowed tags are:
4124 # * <sup> and <sub> (bug 8393)
4125 # * <i> (bug 26375)
4126 # * <b> (r105284)
4127 # * <bdi> (bug 72884)
4128 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4129 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4130 # to allow setting directionality in toc items.
4131 $tocline = preg_replace(
4132 [
4133 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4134 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4135 ],
4136 [ '', '<$1>' ],
4137 $safeHeadline
4138 );
4139
4140 # Strip '<span></span>', which is the result from the above if
4141 # <span id="foo"></span> is used to produce an additional anchor
4142 # for a section.
4143 $tocline = str_replace( '<span></span>', '', $tocline );
4144
4145 $tocline = trim( $tocline );
4146
4147 # For the anchor, strip out HTML-y stuff period
4148 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4149 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4150
4151 # Save headline for section edit hint before it's escaped
4152 $headlineHint = $safeHeadline;
4153
4154 if ( $wgExperimentalHtmlIds ) {
4155 # For reverse compatibility, provide an id that's
4156 # HTML4-compatible, like we used to.
4157 # It may be worth noting, academically, that it's possible for
4158 # the legacy anchor to conflict with a non-legacy headline
4159 # anchor on the page. In this case likely the "correct" thing
4160 # would be to either drop the legacy anchors or make sure
4161 # they're numbered first. However, this would require people
4162 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4163 # manually, so let's not bother worrying about it.
4164 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4165 [ 'noninitial', 'legacy' ] );
4166 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4167
4168 if ( $legacyHeadline == $safeHeadline ) {
4169 # No reason to have both (in fact, we can't)
4170 $legacyHeadline = false;
4171 }
4172 } else {
4173 $legacyHeadline = false;
4174 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4175 'noninitial' );
4176 }
4177
4178 # HTML names must be case-insensitively unique (bug 10721).
4179 # This does not apply to Unicode characters per
4180 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4181 # @todo FIXME: We may be changing them depending on the current locale.
4182 $arrayKey = strtolower( $safeHeadline );
4183 if ( $legacyHeadline === false ) {
4184 $legacyArrayKey = false;
4185 } else {
4186 $legacyArrayKey = strtolower( $legacyHeadline );
4187 }
4188
4189 # Create the anchor for linking from the TOC to the section
4190 $anchor = $safeHeadline;
4191 $legacyAnchor = $legacyHeadline;
4192 if ( isset( $refers[$arrayKey] ) ) {
4193 // @codingStandardsIgnoreStart
4194 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4195 // @codingStandardsIgnoreEnd
4196 $anchor .= "_$i";
4197 $refers["${arrayKey}_$i"] = true;
4198 } else {
4199 $refers[$arrayKey] = true;
4200 }
4201 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4202 // @codingStandardsIgnoreStart
4203 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4204 // @codingStandardsIgnoreEnd
4205 $legacyAnchor .= "_$i";
4206 $refers["${legacyArrayKey}_$i"] = true;
4207 } else {
4208 $refers[$legacyArrayKey] = true;
4209 }
4210
4211 # Don't number the heading if it is the only one (looks silly)
4212 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4213 # the two are different if the line contains a link
4214 $headline = Html::element(
4215 'span',
4216 [ 'class' => 'mw-headline-number' ],
4217 $numbering
4218 ) . ' ' . $headline;
4219 }
4220
4221 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4222 $toc .= Linker::tocLine( $anchor, $tocline,
4223 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4224 }
4225
4226 # Add the section to the section tree
4227 # Find the DOM node for this header
4228 $noOffset = ( $isTemplate || $sectionIndex === false );
4229 while ( $node && !$noOffset ) {
4230 if ( $node->getName() === 'h' ) {
4231 $bits = $node->splitHeading();
4232 if ( $bits['i'] == $sectionIndex ) {
4233 break;
4234 }
4235 }
4236 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4237 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4238 $node = $node->getNextSibling();
4239 }
4240 $tocraw[] = [
4241 'toclevel' => $toclevel,
4242 'level' => $level,
4243 'line' => $tocline,
4244 'number' => $numbering,
4245 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4246 'fromtitle' => $titleText,
4247 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4248 'anchor' => $anchor,
4249 ];
4250
4251 # give headline the correct <h#> tag
4252 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4253 // Output edit section links as markers with styles that can be customized by skins
4254 if ( $isTemplate ) {
4255 # Put a T flag in the section identifier, to indicate to extractSections()
4256 # that sections inside <includeonly> should be counted.
4257 $editsectionPage = $titleText;
4258 $editsectionSection = "T-$sectionIndex";
4259 $editsectionContent = null;
4260 } else {
4261 $editsectionPage = $this->mTitle->getPrefixedText();
4262 $editsectionSection = $sectionIndex;
4263 $editsectionContent = $headlineHint;
4264 }
4265 // We use a bit of pesudo-xml for editsection markers. The
4266 // language converter is run later on. Using a UNIQ style marker
4267 // leads to the converter screwing up the tokens when it
4268 // converts stuff. And trying to insert strip tags fails too. At
4269 // this point all real inputted tags have already been escaped,
4270 // so we don't have to worry about a user trying to input one of
4271 // these markers directly. We use a page and section attribute
4272 // to stop the language converter from converting these
4273 // important bits of data, but put the headline hint inside a
4274 // content block because the language converter is supposed to
4275 // be able to convert that piece of data.
4276 // Gets replaced with html in ParserOutput::getText
4277 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4278 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4279 if ( $editsectionContent !== null ) {
4280 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4281 } else {
4282 $editlink .= '/>';
4283 }
4284 } else {
4285 $editlink = '';
4286 }
4287 $head[$headlineCount] = Linker::makeHeadline( $level,
4288 $matches['attrib'][$headlineCount], $anchor, $headline,
4289 $editlink, $legacyAnchor );
4290
4291 $headlineCount++;
4292 }
4293
4294 $this->setOutputType( $oldType );
4295
4296 # Never ever show TOC if no headers
4297 if ( $numVisible < 1 ) {
4298 $enoughToc = false;
4299 }
4300
4301 if ( $enoughToc ) {
4302 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4303 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4304 }
4305 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4306 $this->mOutput->setTOCHTML( $toc );
4307 $toc = self::TOC_START . $toc . self::TOC_END;
4308 $this->mOutput->addModules( 'mediawiki.toc' );
4309 }
4310
4311 if ( $isMain ) {
4312 $this->mOutput->setSections( $tocraw );
4313 }
4314
4315 # split up and insert constructed headlines
4316 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4317 $i = 0;
4318
4319 // build an array of document sections
4320 $sections = [];
4321 foreach ( $blocks as $block ) {
4322 // $head is zero-based, sections aren't.
4323 if ( empty( $head[$i - 1] ) ) {
4324 $sections[$i] = $block;
4325 } else {
4326 $sections[$i] = $head[$i - 1] . $block;
4327 }
4328
4329 /**
4330 * Send a hook, one per section.
4331 * The idea here is to be able to make section-level DIVs, but to do so in a
4332 * lower-impact, more correct way than r50769
4333 *
4334 * $this : caller
4335 * $section : the section number
4336 * &$sectionContent : ref to the content of the section
4337 * $showEditLinks : boolean describing whether this section has an edit link
4338 */
4339 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4340
4341 $i++;
4342 }
4343
4344 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4345 // append the TOC at the beginning
4346 // Top anchor now in skin
4347 $sections[0] = $sections[0] . $toc . "\n";
4348 }
4349
4350 $full .= implode( '', $sections );
4351
4352 if ( $this->mForceTocPosition ) {
4353 return str_replace( '<!--MWTOC-->', $toc, $full );
4354 } else {
4355 return $full;
4356 }
4357 }
4358
4359 /**
4360 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4361 * conversion, substituting signatures, {{subst:}} templates, etc.
4362 *
4363 * @param string $text The text to transform
4364 * @param Title $title The Title object for the current article
4365 * @param User $user The User object describing the current user
4366 * @param ParserOptions $options Parsing options
4367 * @param bool $clearState Whether to clear the parser state first
4368 * @return string The altered wiki markup
4369 */
4370 public function preSaveTransform( $text, Title $title, User $user,
4371 ParserOptions $options, $clearState = true
4372 ) {
4373 if ( $clearState ) {
4374 $magicScopeVariable = $this->lock();
4375 }
4376 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4377 $this->setUser( $user );
4378
4379 $pairs = [
4380 "\r\n" => "\n",
4381 "\r" => "\n",
4382 ];
4383 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4384 if ( $options->getPreSaveTransform() ) {
4385 $text = $this->pstPass2( $text, $user );
4386 }
4387 $text = $this->mStripState->unstripBoth( $text );
4388
4389 $this->setUser( null ); # Reset
4390
4391 return $text;
4392 }
4393
4394 /**
4395 * Pre-save transform helper function
4396 *
4397 * @param string $text
4398 * @param User $user
4399 *
4400 * @return string
4401 */
4402 private function pstPass2( $text, $user ) {
4403 global $wgContLang;
4404
4405 # Note: This is the timestamp saved as hardcoded wikitext to
4406 # the database, we use $wgContLang here in order to give
4407 # everyone the same signature and use the default one rather
4408 # than the one selected in each user's preferences.
4409 # (see also bug 12815)
4410 $ts = $this->mOptions->getTimestamp();
4411 $timestamp = MWTimestamp::getLocalInstance( $ts );
4412 $ts = $timestamp->format( 'YmdHis' );
4413 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4414
4415 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4416
4417 # Variable replacement
4418 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4419 $text = $this->replaceVariables( $text );
4420
4421 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4422 # which may corrupt this parser instance via its wfMessage()->text() call-
4423
4424 # Signatures
4425 $sigText = $this->getUserSig( $user );
4426 $text = strtr( $text, [
4427 '~~~~~' => $d,
4428 '~~~~' => "$sigText $d",
4429 '~~~' => $sigText
4430 ] );
4431
4432 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4433 $tc = '[' . Title::legalChars() . ']';
4434 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4435
4436 // [[ns:page (context)|]]
4437 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4438 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4439 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4440 // [[ns:page (context), context|]] (using either single or double-width comma)
4441 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4442 // [[|page]] (reverse pipe trick: add context from page title)
4443 $p2 = "/\[\[\\|($tc+)]]/";
4444
4445 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4446 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4447 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4448 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4449
4450 $t = $this->mTitle->getText();
4451 $m = [];
4452 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4453 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4454 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4455 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4456 } else {
4457 # if there's no context, don't bother duplicating the title
4458 $text = preg_replace( $p2, '[[\\1]]', $text );
4459 }
4460
4461 # Trim trailing whitespace
4462 $text = rtrim( $text );
4463
4464 return $text;
4465 }
4466
4467 /**
4468 * Fetch the user's signature text, if any, and normalize to
4469 * validated, ready-to-insert wikitext.
4470 * If you have pre-fetched the nickname or the fancySig option, you can
4471 * specify them here to save a database query.
4472 * Do not reuse this parser instance after calling getUserSig(),
4473 * as it may have changed if it's the $wgParser.
4474 *
4475 * @param User $user
4476 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4477 * @param bool|null $fancySig whether the nicknname is the complete signature
4478 * or null to use default value
4479 * @return string
4480 */
4481 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4482 global $wgMaxSigChars;
4483
4484 $username = $user->getName();
4485
4486 # If not given, retrieve from the user object.
4487 if ( $nickname === false ) {
4488 $nickname = $user->getOption( 'nickname' );
4489 }
4490
4491 if ( is_null( $fancySig ) ) {
4492 $fancySig = $user->getBoolOption( 'fancysig' );
4493 }
4494
4495 $nickname = $nickname == null ? $username : $nickname;
4496
4497 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4498 $nickname = $username;
4499 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4500 } elseif ( $fancySig !== false ) {
4501 # Sig. might contain markup; validate this
4502 if ( $this->validateSig( $nickname ) !== false ) {
4503 # Validated; clean up (if needed) and return it
4504 return $this->cleanSig( $nickname, true );
4505 } else {
4506 # Failed to validate; fall back to the default
4507 $nickname = $username;
4508 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4509 }
4510 }
4511
4512 # Make sure nickname doesnt get a sig in a sig
4513 $nickname = self::cleanSigInSig( $nickname );
4514
4515 # If we're still here, make it a link to the user page
4516 $userText = wfEscapeWikiText( $username );
4517 $nickText = wfEscapeWikiText( $nickname );
4518 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4519
4520 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4521 ->title( $this->getTitle() )->text();
4522 }
4523
4524 /**
4525 * Check that the user's signature contains no bad XML
4526 *
4527 * @param string $text
4528 * @return string|bool An expanded string, or false if invalid.
4529 */
4530 public function validateSig( $text ) {
4531 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4532 }
4533
4534 /**
4535 * Clean up signature text
4536 *
4537 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4538 * 2) Substitute all transclusions
4539 *
4540 * @param string $text
4541 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4542 * @return string Signature text
4543 */
4544 public function cleanSig( $text, $parsing = false ) {
4545 if ( !$parsing ) {
4546 global $wgTitle;
4547 $magicScopeVariable = $this->lock();
4548 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4549 }
4550
4551 # Option to disable this feature
4552 if ( !$this->mOptions->getCleanSignatures() ) {
4553 return $text;
4554 }
4555
4556 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4557 # => Move this logic to braceSubstitution()
4558 $substWord = MagicWord::get( 'subst' );
4559 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4560 $substText = '{{' . $substWord->getSynonym( 0 );
4561
4562 $text = preg_replace( $substRegex, $substText, $text );
4563 $text = self::cleanSigInSig( $text );
4564 $dom = $this->preprocessToDom( $text );
4565 $frame = $this->getPreprocessor()->newFrame();
4566 $text = $frame->expand( $dom );
4567
4568 if ( !$parsing ) {
4569 $text = $this->mStripState->unstripBoth( $text );
4570 }
4571
4572 return $text;
4573 }
4574
4575 /**
4576 * Strip 3, 4 or 5 tildes out of signatures.
4577 *
4578 * @param string $text
4579 * @return string Signature text with /~{3,5}/ removed
4580 */
4581 public static function cleanSigInSig( $text ) {
4582 $text = preg_replace( '/~{3,5}/', '', $text );
4583 return $text;
4584 }
4585
4586 /**
4587 * Set up some variables which are usually set up in parse()
4588 * so that an external function can call some class members with confidence
4589 *
4590 * @param Title|null $title
4591 * @param ParserOptions $options
4592 * @param int $outputType
4593 * @param bool $clearState
4594 */
4595 public function startExternalParse( Title $title = null, ParserOptions $options,
4596 $outputType, $clearState = true
4597 ) {
4598 $this->startParse( $title, $options, $outputType, $clearState );
4599 }
4600
4601 /**
4602 * @param Title|null $title
4603 * @param ParserOptions $options
4604 * @param int $outputType
4605 * @param bool $clearState
4606 */
4607 private function startParse( Title $title = null, ParserOptions $options,
4608 $outputType, $clearState = true
4609 ) {
4610 $this->setTitle( $title );
4611 $this->mOptions = $options;
4612 $this->setOutputType( $outputType );
4613 if ( $clearState ) {
4614 $this->clearState();
4615 }
4616 }
4617
4618 /**
4619 * Wrapper for preprocess()
4620 *
4621 * @param string $text The text to preprocess
4622 * @param ParserOptions $options Options
4623 * @param Title|null $title Title object or null to use $wgTitle
4624 * @return string
4625 */
4626 public function transformMsg( $text, $options, $title = null ) {
4627 static $executing = false;
4628
4629 # Guard against infinite recursion
4630 if ( $executing ) {
4631 return $text;
4632 }
4633 $executing = true;
4634
4635 if ( !$title ) {
4636 global $wgTitle;
4637 $title = $wgTitle;
4638 }
4639
4640 $text = $this->preprocess( $text, $title, $options );
4641
4642 $executing = false;
4643 return $text;
4644 }
4645
4646 /**
4647 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4648 * The callback should have the following form:
4649 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4650 *
4651 * Transform and return $text. Use $parser for any required context, e.g. use
4652 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4653 *
4654 * Hooks may return extended information by returning an array, of which the
4655 * first numbered element (index 0) must be the return string, and all other
4656 * entries are extracted into local variables within an internal function
4657 * in the Parser class.
4658 *
4659 * This interface (introduced r61913) appears to be undocumented, but
4660 * 'markerType' is used by some core tag hooks to override which strip
4661 * array their results are placed in. **Use great caution if attempting
4662 * this interface, as it is not documented and injudicious use could smash
4663 * private variables.**
4664 *
4665 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4666 * @param callable $callback The callback function (and object) to use for the tag
4667 * @throws MWException
4668 * @return callable|null The old value of the mTagHooks array associated with the hook
4669 */
4670 public function setHook( $tag, $callback ) {
4671 $tag = strtolower( $tag );
4672 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4673 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4674 }
4675 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4676 $this->mTagHooks[$tag] = $callback;
4677 if ( !in_array( $tag, $this->mStripList ) ) {
4678 $this->mStripList[] = $tag;
4679 }
4680
4681 return $oldVal;
4682 }
4683
4684 /**
4685 * As setHook(), but letting the contents be parsed.
4686 *
4687 * Transparent tag hooks are like regular XML-style tag hooks, except they
4688 * operate late in the transformation sequence, on HTML instead of wikitext.
4689 *
4690 * This is probably obsoleted by things dealing with parser frames?
4691 * The only extension currently using it is geoserver.
4692 *
4693 * @since 1.10
4694 * @todo better document or deprecate this
4695 *
4696 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4697 * @param callable $callback The callback function (and object) to use for the tag
4698 * @throws MWException
4699 * @return callable|null The old value of the mTagHooks array associated with the hook
4700 */
4701 public function setTransparentTagHook( $tag, $callback ) {
4702 $tag = strtolower( $tag );
4703 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4704 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4705 }
4706 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4707 $this->mTransparentTagHooks[$tag] = $callback;
4708
4709 return $oldVal;
4710 }
4711
4712 /**
4713 * Remove all tag hooks
4714 */
4715 public function clearTagHooks() {
4716 $this->mTagHooks = [];
4717 $this->mFunctionTagHooks = [];
4718 $this->mStripList = $this->mDefaultStripList;
4719 }
4720
4721 /**
4722 * Create a function, e.g. {{sum:1|2|3}}
4723 * The callback function should have the form:
4724 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4725 *
4726 * Or with Parser::SFH_OBJECT_ARGS:
4727 * function myParserFunction( $parser, $frame, $args ) { ... }
4728 *
4729 * The callback may either return the text result of the function, or an array with the text
4730 * in element 0, and a number of flags in the other elements. The names of the flags are
4731 * specified in the keys. Valid flags are:
4732 * found The text returned is valid, stop processing the template. This
4733 * is on by default.
4734 * nowiki Wiki markup in the return value should be escaped
4735 * isHTML The returned text is HTML, armour it against wikitext transformation
4736 *
4737 * @param string $id The magic word ID
4738 * @param callable $callback The callback function (and object) to use
4739 * @param int $flags A combination of the following flags:
4740 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4741 *
4742 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4743 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4744 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4745 * the arguments, and to control the way they are expanded.
4746 *
4747 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4748 * arguments, for instance:
4749 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4750 *
4751 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4752 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4753 * working if/when this is changed.
4754 *
4755 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4756 * expansion.
4757 *
4758 * Please read the documentation in includes/parser/Preprocessor.php for more information
4759 * about the methods available in PPFrame and PPNode.
4760 *
4761 * @throws MWException
4762 * @return string|callable The old callback function for this name, if any
4763 */
4764 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4765 global $wgContLang;
4766
4767 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4768 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4769
4770 # Add to function cache
4771 $mw = MagicWord::get( $id );
4772 if ( !$mw ) {
4773 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4774 }
4775
4776 $synonyms = $mw->getSynonyms();
4777 $sensitive = intval( $mw->isCaseSensitive() );
4778
4779 foreach ( $synonyms as $syn ) {
4780 # Case
4781 if ( !$sensitive ) {
4782 $syn = $wgContLang->lc( $syn );
4783 }
4784 # Add leading hash
4785 if ( !( $flags & self::SFH_NO_HASH ) ) {
4786 $syn = '#' . $syn;
4787 }
4788 # Remove trailing colon
4789 if ( substr( $syn, -1, 1 ) === ':' ) {
4790 $syn = substr( $syn, 0, -1 );
4791 }
4792 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4793 }
4794 return $oldVal;
4795 }
4796
4797 /**
4798 * Get all registered function hook identifiers
4799 *
4800 * @return array
4801 */
4802 public function getFunctionHooks() {
4803 return array_keys( $this->mFunctionHooks );
4804 }
4805
4806 /**
4807 * Create a tag function, e.g. "<test>some stuff</test>".
4808 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4809 * Unlike parser functions, their content is not preprocessed.
4810 * @param string $tag
4811 * @param callable $callback
4812 * @param int $flags
4813 * @throws MWException
4814 * @return null
4815 */
4816 public function setFunctionTagHook( $tag, $callback, $flags ) {
4817 $tag = strtolower( $tag );
4818 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4819 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4820 }
4821 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4822 $this->mFunctionTagHooks[$tag] : null;
4823 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4824
4825 if ( !in_array( $tag, $this->mStripList ) ) {
4826 $this->mStripList[] = $tag;
4827 }
4828
4829 return $old;
4830 }
4831
4832 /**
4833 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4834 * Placeholders created in Linker::link()
4835 *
4836 * @param string $text
4837 * @param int $options
4838 */
4839 public function replaceLinkHolders( &$text, $options = 0 ) {
4840 $this->mLinkHolders->replace( $text );
4841 }
4842
4843 /**
4844 * Replace "<!--LINK-->" link placeholders with plain text of links
4845 * (not HTML-formatted).
4846 *
4847 * @param string $text
4848 * @return string
4849 */
4850 public function replaceLinkHoldersText( $text ) {
4851 return $this->mLinkHolders->replaceText( $text );
4852 }
4853
4854 /**
4855 * Renders an image gallery from a text with one line per image.
4856 * text labels may be given by using |-style alternative text. E.g.
4857 * Image:one.jpg|The number "1"
4858 * Image:tree.jpg|A tree
4859 * given as text will return the HTML of a gallery with two images,
4860 * labeled 'The number "1"' and
4861 * 'A tree'.
4862 *
4863 * @param string $text
4864 * @param array $params
4865 * @return string HTML
4866 */
4867 public function renderImageGallery( $text, $params ) {
4868
4869 $mode = false;
4870 if ( isset( $params['mode'] ) ) {
4871 $mode = $params['mode'];
4872 }
4873
4874 try {
4875 $ig = ImageGalleryBase::factory( $mode );
4876 } catch ( Exception $e ) {
4877 // If invalid type set, fallback to default.
4878 $ig = ImageGalleryBase::factory( false );
4879 }
4880
4881 $ig->setContextTitle( $this->mTitle );
4882 $ig->setShowBytes( false );
4883 $ig->setShowFilename( false );
4884 $ig->setParser( $this );
4885 $ig->setHideBadImages();
4886 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4887
4888 if ( isset( $params['showfilename'] ) ) {
4889 $ig->setShowFilename( true );
4890 } else {
4891 $ig->setShowFilename( false );
4892 }
4893 if ( isset( $params['caption'] ) ) {
4894 $caption = $params['caption'];
4895 $caption = htmlspecialchars( $caption );
4896 $caption = $this->replaceInternalLinks( $caption );
4897 $ig->setCaptionHtml( $caption );
4898 }
4899 if ( isset( $params['perrow'] ) ) {
4900 $ig->setPerRow( $params['perrow'] );
4901 }
4902 if ( isset( $params['widths'] ) ) {
4903 $ig->setWidths( $params['widths'] );
4904 }
4905 if ( isset( $params['heights'] ) ) {
4906 $ig->setHeights( $params['heights'] );
4907 }
4908 $ig->setAdditionalOptions( $params );
4909
4910 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4911
4912 $lines = StringUtils::explode( "\n", $text );
4913 foreach ( $lines as $line ) {
4914 # match lines like these:
4915 # Image:someimage.jpg|This is some image
4916 $matches = [];
4917 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4918 # Skip empty lines
4919 if ( count( $matches ) == 0 ) {
4920 continue;
4921 }
4922
4923 if ( strpos( $matches[0], '%' ) !== false ) {
4924 $matches[1] = rawurldecode( $matches[1] );
4925 }
4926 $title = Title::newFromText( $matches[1], NS_FILE );
4927 if ( is_null( $title ) ) {
4928 # Bogus title. Ignore these so we don't bomb out later.
4929 continue;
4930 }
4931
4932 # We need to get what handler the file uses, to figure out parameters.
4933 # Note, a hook can overide the file name, and chose an entirely different
4934 # file (which potentially could be of a different type and have different handler).
4935 $options = [];
4936 $descQuery = false;
4937 Hooks::run( 'BeforeParserFetchFileAndTitle',
4938 [ $this, $title, &$options, &$descQuery ] );
4939 # Don't register it now, as ImageGallery does that later.
4940 $file = $this->fetchFileNoRegister( $title, $options );
4941 $handler = $file ? $file->getHandler() : false;
4942
4943 $paramMap = [
4944 'img_alt' => 'gallery-internal-alt',
4945 'img_link' => 'gallery-internal-link',
4946 ];
4947 if ( $handler ) {
4948 $paramMap = $paramMap + $handler->getParamMap();
4949 // We don't want people to specify per-image widths.
4950 // Additionally the width parameter would need special casing anyhow.
4951 unset( $paramMap['img_width'] );
4952 }
4953
4954 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4955
4956 $label = '';
4957 $alt = '';
4958 $link = '';
4959 $handlerOptions = [];
4960 if ( isset( $matches[3] ) ) {
4961 // look for an |alt= definition while trying not to break existing
4962 // captions with multiple pipes (|) in it, until a more sensible grammar
4963 // is defined for images in galleries
4964
4965 // FIXME: Doing recursiveTagParse at this stage, and the trim before
4966 // splitting on '|' is a bit odd, and different from makeImage.
4967 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4968 $parameterMatches = StringUtils::explode( '|', $matches[3] );
4969
4970 foreach ( $parameterMatches as $parameterMatch ) {
4971 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
4972 if ( $magicName ) {
4973 $paramName = $paramMap[$magicName];
4974
4975 switch ( $paramName ) {
4976 case 'gallery-internal-alt':
4977 $alt = $this->stripAltText( $match, false );
4978 break;
4979 case 'gallery-internal-link':
4980 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
4981 $chars = self::EXT_LINK_URL_CLASS;
4982 $addr = self::EXT_LINK_ADDR;
4983 $prots = $this->mUrlProtocols;
4984 // check to see if link matches an absolute url, if not then it must be a wiki link.
4985 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
4986 $link = $linkValue;
4987 } else {
4988 $localLinkTitle = Title::newFromText( $linkValue );
4989 if ( $localLinkTitle !== null ) {
4990 $link = $localLinkTitle->getLinkURL();
4991 }
4992 }
4993 break;
4994 default:
4995 // Must be a handler specific parameter.
4996 if ( $handler->validateParam( $paramName, $match ) ) {
4997 $handlerOptions[$paramName] = $match;
4998 } else {
4999 // Guess not, consider it as caption.
5000 wfDebug( "$parameterMatch failed parameter validation\n" );
5001 $label = '|' . $parameterMatch;
5002 }
5003 }
5004
5005 } else {
5006 // Last pipe wins.
5007 $label = '|' . $parameterMatch;
5008 }
5009 }
5010 // Remove the pipe.
5011 $label = substr( $label, 1 );
5012 }
5013
5014 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5015 }
5016 $html = $ig->toHTML();
5017 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5018 return $html;
5019 }
5020
5021 /**
5022 * @param MediaHandler $handler
5023 * @return array
5024 */
5025 public function getImageParams( $handler ) {
5026 if ( $handler ) {
5027 $handlerClass = get_class( $handler );
5028 } else {
5029 $handlerClass = '';
5030 }
5031 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5032 # Initialise static lists
5033 static $internalParamNames = [
5034 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5035 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5036 'bottom', 'text-bottom' ],
5037 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5038 'upright', 'border', 'link', 'alt', 'class' ],
5039 ];
5040 static $internalParamMap;
5041 if ( !$internalParamMap ) {
5042 $internalParamMap = [];
5043 foreach ( $internalParamNames as $type => $names ) {
5044 foreach ( $names as $name ) {
5045 $magicName = str_replace( '-', '_', "img_$name" );
5046 $internalParamMap[$magicName] = [ $type, $name ];
5047 }
5048 }
5049 }
5050
5051 # Add handler params
5052 $paramMap = $internalParamMap;
5053 if ( $handler ) {
5054 $handlerParamMap = $handler->getParamMap();
5055 foreach ( $handlerParamMap as $magic => $paramName ) {
5056 $paramMap[$magic] = [ 'handler', $paramName ];
5057 }
5058 }
5059 $this->mImageParams[$handlerClass] = $paramMap;
5060 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5061 }
5062 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5063 }
5064
5065 /**
5066 * Parse image options text and use it to make an image
5067 *
5068 * @param Title $title
5069 * @param string $options
5070 * @param LinkHolderArray|bool $holders
5071 * @return string HTML
5072 */
5073 public function makeImage( $title, $options, $holders = false ) {
5074 # Check if the options text is of the form "options|alt text"
5075 # Options are:
5076 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5077 # * left no resizing, just left align. label is used for alt= only
5078 # * right same, but right aligned
5079 # * none same, but not aligned
5080 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5081 # * center center the image
5082 # * frame Keep original image size, no magnify-button.
5083 # * framed Same as "frame"
5084 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5085 # * upright reduce width for upright images, rounded to full __0 px
5086 # * border draw a 1px border around the image
5087 # * alt Text for HTML alt attribute (defaults to empty)
5088 # * class Set a class for img node
5089 # * link Set the target of the image link. Can be external, interwiki, or local
5090 # vertical-align values (no % or length right now):
5091 # * baseline
5092 # * sub
5093 # * super
5094 # * top
5095 # * text-top
5096 # * middle
5097 # * bottom
5098 # * text-bottom
5099
5100 $parts = StringUtils::explode( "|", $options );
5101
5102 # Give extensions a chance to select the file revision for us
5103 $options = [];
5104 $descQuery = false;
5105 Hooks::run( 'BeforeParserFetchFileAndTitle',
5106 [ $this, $title, &$options, &$descQuery ] );
5107 # Fetch and register the file (file title may be different via hooks)
5108 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5109
5110 # Get parameter map
5111 $handler = $file ? $file->getHandler() : false;
5112
5113 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5114
5115 if ( !$file ) {
5116 $this->addTrackingCategory( 'broken-file-category' );
5117 }
5118
5119 # Process the input parameters
5120 $caption = '';
5121 $params = [ 'frame' => [], 'handler' => [],
5122 'horizAlign' => [], 'vertAlign' => [] ];
5123 $seenformat = false;
5124 foreach ( $parts as $part ) {
5125 $part = trim( $part );
5126 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5127 $validated = false;
5128 if ( isset( $paramMap[$magicName] ) ) {
5129 list( $type, $paramName ) = $paramMap[$magicName];
5130
5131 # Special case; width and height come in one variable together
5132 if ( $type === 'handler' && $paramName === 'width' ) {
5133 $parsedWidthParam = $this->parseWidthParam( $value );
5134 if ( isset( $parsedWidthParam['width'] ) ) {
5135 $width = $parsedWidthParam['width'];
5136 if ( $handler->validateParam( 'width', $width ) ) {
5137 $params[$type]['width'] = $width;
5138 $validated = true;
5139 }
5140 }
5141 if ( isset( $parsedWidthParam['height'] ) ) {
5142 $height = $parsedWidthParam['height'];
5143 if ( $handler->validateParam( 'height', $height ) ) {
5144 $params[$type]['height'] = $height;
5145 $validated = true;
5146 }
5147 }
5148 # else no validation -- bug 13436
5149 } else {
5150 if ( $type === 'handler' ) {
5151 # Validate handler parameter
5152 $validated = $handler->validateParam( $paramName, $value );
5153 } else {
5154 # Validate internal parameters
5155 switch ( $paramName ) {
5156 case 'manualthumb':
5157 case 'alt':
5158 case 'class':
5159 # @todo FIXME: Possibly check validity here for
5160 # manualthumb? downstream behavior seems odd with
5161 # missing manual thumbs.
5162 $validated = true;
5163 $value = $this->stripAltText( $value, $holders );
5164 break;
5165 case 'link':
5166 $chars = self::EXT_LINK_URL_CLASS;
5167 $addr = self::EXT_LINK_ADDR;
5168 $prots = $this->mUrlProtocols;
5169 if ( $value === '' ) {
5170 $paramName = 'no-link';
5171 $value = true;
5172 $validated = true;
5173 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5174 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5175 $paramName = 'link-url';
5176 $this->mOutput->addExternalLink( $value );
5177 if ( $this->mOptions->getExternalLinkTarget() ) {
5178 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5179 }
5180 $validated = true;
5181 }
5182 } else {
5183 $linkTitle = Title::newFromText( $value );
5184 if ( $linkTitle ) {
5185 $paramName = 'link-title';
5186 $value = $linkTitle;
5187 $this->mOutput->addLink( $linkTitle );
5188 $validated = true;
5189 }
5190 }
5191 break;
5192 case 'frameless':
5193 case 'framed':
5194 case 'thumbnail':
5195 // use first appearing option, discard others.
5196 $validated = ! $seenformat;
5197 $seenformat = true;
5198 break;
5199 default:
5200 # Most other things appear to be empty or numeric...
5201 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5202 }
5203 }
5204
5205 if ( $validated ) {
5206 $params[$type][$paramName] = $value;
5207 }
5208 }
5209 }
5210 if ( !$validated ) {
5211 $caption = $part;
5212 }
5213 }
5214
5215 # Process alignment parameters
5216 if ( $params['horizAlign'] ) {
5217 $params['frame']['align'] = key( $params['horizAlign'] );
5218 }
5219 if ( $params['vertAlign'] ) {
5220 $params['frame']['valign'] = key( $params['vertAlign'] );
5221 }
5222
5223 $params['frame']['caption'] = $caption;
5224
5225 # Will the image be presented in a frame, with the caption below?
5226 $imageIsFramed = isset( $params['frame']['frame'] )
5227 || isset( $params['frame']['framed'] )
5228 || isset( $params['frame']['thumbnail'] )
5229 || isset( $params['frame']['manualthumb'] );
5230
5231 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5232 # came to also set the caption, ordinary text after the image -- which
5233 # makes no sense, because that just repeats the text multiple times in
5234 # screen readers. It *also* came to set the title attribute.
5235 # Now that we have an alt attribute, we should not set the alt text to
5236 # equal the caption: that's worse than useless, it just repeats the
5237 # text. This is the framed/thumbnail case. If there's no caption, we
5238 # use the unnamed parameter for alt text as well, just for the time be-
5239 # ing, if the unnamed param is set and the alt param is not.
5240 # For the future, we need to figure out if we want to tweak this more,
5241 # e.g., introducing a title= parameter for the title; ignoring the un-
5242 # named parameter entirely for images without a caption; adding an ex-
5243 # plicit caption= parameter and preserving the old magic unnamed para-
5244 # meter for BC; ...
5245 if ( $imageIsFramed ) { # Framed image
5246 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5247 # No caption or alt text, add the filename as the alt text so
5248 # that screen readers at least get some description of the image
5249 $params['frame']['alt'] = $title->getText();
5250 }
5251 # Do not set $params['frame']['title'] because tooltips don't make sense
5252 # for framed images
5253 } else { # Inline image
5254 if ( !isset( $params['frame']['alt'] ) ) {
5255 # No alt text, use the "caption" for the alt text
5256 if ( $caption !== '' ) {
5257 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5258 } else {
5259 # No caption, fall back to using the filename for the
5260 # alt text
5261 $params['frame']['alt'] = $title->getText();
5262 }
5263 }
5264 # Use the "caption" for the tooltip text
5265 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5266 }
5267
5268 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5269
5270 # Linker does the rest
5271 $time = isset( $options['time'] ) ? $options['time'] : false;
5272 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5273 $time, $descQuery, $this->mOptions->getThumbSize() );
5274
5275 # Give the handler a chance to modify the parser object
5276 if ( $handler ) {
5277 $handler->parserTransformHook( $this, $file );
5278 }
5279
5280 return $ret;
5281 }
5282
5283 /**
5284 * @param string $caption
5285 * @param LinkHolderArray|bool $holders
5286 * @return mixed|string
5287 */
5288 protected function stripAltText( $caption, $holders ) {
5289 # Strip bad stuff out of the title (tooltip). We can't just use
5290 # replaceLinkHoldersText() here, because if this function is called
5291 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5292 if ( $holders ) {
5293 $tooltip = $holders->replaceText( $caption );
5294 } else {
5295 $tooltip = $this->replaceLinkHoldersText( $caption );
5296 }
5297
5298 # make sure there are no placeholders in thumbnail attributes
5299 # that are later expanded to html- so expand them now and
5300 # remove the tags
5301 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5302 $tooltip = Sanitizer::stripAllTags( $tooltip );
5303
5304 return $tooltip;
5305 }
5306
5307 /**
5308 * Set a flag in the output object indicating that the content is dynamic and
5309 * shouldn't be cached.
5310 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5311 */
5312 public function disableCache() {
5313 wfDebug( "Parser output marked as uncacheable.\n" );
5314 if ( !$this->mOutput ) {
5315 throw new MWException( __METHOD__ .
5316 " can only be called when actually parsing something" );
5317 }
5318 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5319 }
5320
5321 /**
5322 * Callback from the Sanitizer for expanding items found in HTML attribute
5323 * values, so they can be safely tested and escaped.
5324 *
5325 * @param string $text
5326 * @param bool|PPFrame $frame
5327 * @return string
5328 */
5329 public function attributeStripCallback( &$text, $frame = false ) {
5330 $text = $this->replaceVariables( $text, $frame );
5331 $text = $this->mStripState->unstripBoth( $text );
5332 return $text;
5333 }
5334
5335 /**
5336 * Accessor
5337 *
5338 * @return array
5339 */
5340 public function getTags() {
5341 return array_merge(
5342 array_keys( $this->mTransparentTagHooks ),
5343 array_keys( $this->mTagHooks ),
5344 array_keys( $this->mFunctionTagHooks )
5345 );
5346 }
5347
5348 /**
5349 * Replace transparent tags in $text with the values given by the callbacks.
5350 *
5351 * Transparent tag hooks are like regular XML-style tag hooks, except they
5352 * operate late in the transformation sequence, on HTML instead of wikitext.
5353 *
5354 * @param string $text
5355 *
5356 * @return string
5357 */
5358 public function replaceTransparentTags( $text ) {
5359 $matches = [];
5360 $elements = array_keys( $this->mTransparentTagHooks );
5361 $text = self::extractTagsAndParams( $elements, $text, $matches );
5362 $replacements = [];
5363
5364 foreach ( $matches as $marker => $data ) {
5365 list( $element, $content, $params, $tag ) = $data;
5366 $tagName = strtolower( $element );
5367 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5368 $output = call_user_func_array(
5369 $this->mTransparentTagHooks[$tagName],
5370 [ $content, $params, $this ]
5371 );
5372 } else {
5373 $output = $tag;
5374 }
5375 $replacements[$marker] = $output;
5376 }
5377 return strtr( $text, $replacements );
5378 }
5379
5380 /**
5381 * Break wikitext input into sections, and either pull or replace
5382 * some particular section's text.
5383 *
5384 * External callers should use the getSection and replaceSection methods.
5385 *
5386 * @param string $text Page wikitext
5387 * @param string|number $sectionId A section identifier string of the form:
5388 * "<flag1> - <flag2> - ... - <section number>"
5389 *
5390 * Currently the only recognised flag is "T", which means the target section number
5391 * was derived during a template inclusion parse, in other words this is a template
5392 * section edit link. If no flags are given, it was an ordinary section edit link.
5393 * This flag is required to avoid a section numbering mismatch when a section is
5394 * enclosed by "<includeonly>" (bug 6563).
5395 *
5396 * The section number 0 pulls the text before the first heading; other numbers will
5397 * pull the given section along with its lower-level subsections. If the section is
5398 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5399 *
5400 * Section 0 is always considered to exist, even if it only contains the empty
5401 * string. If $text is the empty string and section 0 is replaced, $newText is
5402 * returned.
5403 *
5404 * @param string $mode One of "get" or "replace"
5405 * @param string $newText Replacement text for section data.
5406 * @return string For "get", the extracted section text.
5407 * for "replace", the whole page with the section replaced.
5408 */
5409 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5410 global $wgTitle; # not generally used but removes an ugly failure mode
5411
5412 $magicScopeVariable = $this->lock();
5413 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5414 $outText = '';
5415 $frame = $this->getPreprocessor()->newFrame();
5416
5417 # Process section extraction flags
5418 $flags = 0;
5419 $sectionParts = explode( '-', $sectionId );
5420 $sectionIndex = array_pop( $sectionParts );
5421 foreach ( $sectionParts as $part ) {
5422 if ( $part === 'T' ) {
5423 $flags |= self::PTD_FOR_INCLUSION;
5424 }
5425 }
5426
5427 # Check for empty input
5428 if ( strval( $text ) === '' ) {
5429 # Only sections 0 and T-0 exist in an empty document
5430 if ( $sectionIndex == 0 ) {
5431 if ( $mode === 'get' ) {
5432 return '';
5433 } else {
5434 return $newText;
5435 }
5436 } else {
5437 if ( $mode === 'get' ) {
5438 return $newText;
5439 } else {
5440 return $text;
5441 }
5442 }
5443 }
5444
5445 # Preprocess the text
5446 $root = $this->preprocessToDom( $text, $flags );
5447
5448 # <h> nodes indicate section breaks
5449 # They can only occur at the top level, so we can find them by iterating the root's children
5450 $node = $root->getFirstChild();
5451
5452 # Find the target section
5453 if ( $sectionIndex == 0 ) {
5454 # Section zero doesn't nest, level=big
5455 $targetLevel = 1000;
5456 } else {
5457 while ( $node ) {
5458 if ( $node->getName() === 'h' ) {
5459 $bits = $node->splitHeading();
5460 if ( $bits['i'] == $sectionIndex ) {
5461 $targetLevel = $bits['level'];
5462 break;
5463 }
5464 }
5465 if ( $mode === 'replace' ) {
5466 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5467 }
5468 $node = $node->getNextSibling();
5469 }
5470 }
5471
5472 if ( !$node ) {
5473 # Not found
5474 if ( $mode === 'get' ) {
5475 return $newText;
5476 } else {
5477 return $text;
5478 }
5479 }
5480
5481 # Find the end of the section, including nested sections
5482 do {
5483 if ( $node->getName() === 'h' ) {
5484 $bits = $node->splitHeading();
5485 $curLevel = $bits['level'];
5486 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5487 break;
5488 }
5489 }
5490 if ( $mode === 'get' ) {
5491 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5492 }
5493 $node = $node->getNextSibling();
5494 } while ( $node );
5495
5496 # Write out the remainder (in replace mode only)
5497 if ( $mode === 'replace' ) {
5498 # Output the replacement text
5499 # Add two newlines on -- trailing whitespace in $newText is conventionally
5500 # stripped by the editor, so we need both newlines to restore the paragraph gap
5501 # Only add trailing whitespace if there is newText
5502 if ( $newText != "" ) {
5503 $outText .= $newText . "\n\n";
5504 }
5505
5506 while ( $node ) {
5507 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5508 $node = $node->getNextSibling();
5509 }
5510 }
5511
5512 if ( is_string( $outText ) ) {
5513 # Re-insert stripped tags
5514 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5515 }
5516
5517 return $outText;
5518 }
5519
5520 /**
5521 * This function returns the text of a section, specified by a number ($section).
5522 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5523 * the first section before any such heading (section 0).
5524 *
5525 * If a section contains subsections, these are also returned.
5526 *
5527 * @param string $text Text to look in
5528 * @param string|number $sectionId Section identifier as a number or string
5529 * (e.g. 0, 1 or 'T-1').
5530 * @param string $defaultText Default to return if section is not found
5531 *
5532 * @return string Text of the requested section
5533 */
5534 public function getSection( $text, $sectionId, $defaultText = '' ) {
5535 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5536 }
5537
5538 /**
5539 * This function returns $oldtext after the content of the section
5540 * specified by $section has been replaced with $text. If the target
5541 * section does not exist, $oldtext is returned unchanged.
5542 *
5543 * @param string $oldText Former text of the article
5544 * @param string|number $sectionId Section identifier as a number or string
5545 * (e.g. 0, 1 or 'T-1').
5546 * @param string $newText Replacing text
5547 *
5548 * @return string Modified text
5549 */
5550 public function replaceSection( $oldText, $sectionId, $newText ) {
5551 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5552 }
5553
5554 /**
5555 * Get the ID of the revision we are parsing
5556 *
5557 * @return int|null
5558 */
5559 public function getRevisionId() {
5560 return $this->mRevisionId;
5561 }
5562
5563 /**
5564 * Get the revision object for $this->mRevisionId
5565 *
5566 * @return Revision|null Either a Revision object or null
5567 * @since 1.23 (public since 1.23)
5568 */
5569 public function getRevisionObject() {
5570 if ( !is_null( $this->mRevisionObject ) ) {
5571 return $this->mRevisionObject;
5572 }
5573 if ( is_null( $this->mRevisionId ) ) {
5574 return null;
5575 }
5576
5577 $rev = call_user_func(
5578 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5579 );
5580
5581 # If the parse is for a new revision, then the callback should have
5582 # already been set to force the object and should match mRevisionId.
5583 # If not, try to fetch by mRevisionId for sanity.
5584 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5585 $rev = Revision::newFromId( $this->mRevisionId );
5586 }
5587
5588 $this->mRevisionObject = $rev;
5589
5590 return $this->mRevisionObject;
5591 }
5592
5593 /**
5594 * Get the timestamp associated with the current revision, adjusted for
5595 * the default server-local timestamp
5596 * @return string
5597 */
5598 public function getRevisionTimestamp() {
5599 if ( is_null( $this->mRevisionTimestamp ) ) {
5600 global $wgContLang;
5601
5602 $revObject = $this->getRevisionObject();
5603 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5604
5605 # The cryptic '' timezone parameter tells to use the site-default
5606 # timezone offset instead of the user settings.
5607 # Since this value will be saved into the parser cache, served
5608 # to other users, and potentially even used inside links and such,
5609 # it needs to be consistent for all visitors.
5610 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5611
5612 }
5613 return $this->mRevisionTimestamp;
5614 }
5615
5616 /**
5617 * Get the name of the user that edited the last revision
5618 *
5619 * @return string User name
5620 */
5621 public function getRevisionUser() {
5622 if ( is_null( $this->mRevisionUser ) ) {
5623 $revObject = $this->getRevisionObject();
5624
5625 # if this template is subst: the revision id will be blank,
5626 # so just use the current user's name
5627 if ( $revObject ) {
5628 $this->mRevisionUser = $revObject->getUserText();
5629 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5630 $this->mRevisionUser = $this->getUser()->getName();
5631 }
5632 }
5633 return $this->mRevisionUser;
5634 }
5635
5636 /**
5637 * Get the size of the revision
5638 *
5639 * @return int|null Revision size
5640 */
5641 public function getRevisionSize() {
5642 if ( is_null( $this->mRevisionSize ) ) {
5643 $revObject = $this->getRevisionObject();
5644
5645 # if this variable is subst: the revision id will be blank,
5646 # so just use the parser input size, because the own substituation
5647 # will change the size.
5648 if ( $revObject ) {
5649 $this->mRevisionSize = $revObject->getSize();
5650 } else {
5651 $this->mRevisionSize = $this->mInputSize;
5652 }
5653 }
5654 return $this->mRevisionSize;
5655 }
5656
5657 /**
5658 * Mutator for $mDefaultSort
5659 *
5660 * @param string $sort New value
5661 */
5662 public function setDefaultSort( $sort ) {
5663 $this->mDefaultSort = $sort;
5664 $this->mOutput->setProperty( 'defaultsort', $sort );
5665 }
5666
5667 /**
5668 * Accessor for $mDefaultSort
5669 * Will use the empty string if none is set.
5670 *
5671 * This value is treated as a prefix, so the
5672 * empty string is equivalent to sorting by
5673 * page name.
5674 *
5675 * @return string
5676 */
5677 public function getDefaultSort() {
5678 if ( $this->mDefaultSort !== false ) {
5679 return $this->mDefaultSort;
5680 } else {
5681 return '';
5682 }
5683 }
5684
5685 /**
5686 * Accessor for $mDefaultSort
5687 * Unlike getDefaultSort(), will return false if none is set
5688 *
5689 * @return string|bool
5690 */
5691 public function getCustomDefaultSort() {
5692 return $this->mDefaultSort;
5693 }
5694
5695 /**
5696 * Try to guess the section anchor name based on a wikitext fragment
5697 * presumably extracted from a heading, for example "Header" from
5698 * "== Header ==".
5699 *
5700 * @param string $text
5701 *
5702 * @return string
5703 */
5704 public function guessSectionNameFromWikiText( $text ) {
5705 # Strip out wikitext links(they break the anchor)
5706 $text = $this->stripSectionName( $text );
5707 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5708 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5709 }
5710
5711 /**
5712 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5713 * instead. For use in redirects, since IE6 interprets Redirect: headers
5714 * as something other than UTF-8 (apparently?), resulting in breakage.
5715 *
5716 * @param string $text The section name
5717 * @return string An anchor
5718 */
5719 public function guessLegacySectionNameFromWikiText( $text ) {
5720 # Strip out wikitext links(they break the anchor)
5721 $text = $this->stripSectionName( $text );
5722 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5723 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5724 }
5725
5726 /**
5727 * Strips a text string of wikitext for use in a section anchor
5728 *
5729 * Accepts a text string and then removes all wikitext from the
5730 * string and leaves only the resultant text (i.e. the result of
5731 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5732 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5733 * to create valid section anchors by mimicing the output of the
5734 * parser when headings are parsed.
5735 *
5736 * @param string $text Text string to be stripped of wikitext
5737 * for use in a Section anchor
5738 * @return string Filtered text string
5739 */
5740 public function stripSectionName( $text ) {
5741 # Strip internal link markup
5742 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5743 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5744
5745 # Strip external link markup
5746 # @todo FIXME: Not tolerant to blank link text
5747 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5748 # on how many empty links there are on the page - need to figure that out.
5749 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5750
5751 # Parse wikitext quotes (italics & bold)
5752 $text = $this->doQuotes( $text );
5753
5754 # Strip HTML tags
5755 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5756 return $text;
5757 }
5758
5759 /**
5760 * strip/replaceVariables/unstrip for preprocessor regression testing
5761 *
5762 * @param string $text
5763 * @param Title $title
5764 * @param ParserOptions $options
5765 * @param int $outputType
5766 *
5767 * @return string
5768 */
5769 public function testSrvus( $text, Title $title, ParserOptions $options,
5770 $outputType = self::OT_HTML
5771 ) {
5772 $magicScopeVariable = $this->lock();
5773 $this->startParse( $title, $options, $outputType, true );
5774
5775 $text = $this->replaceVariables( $text );
5776 $text = $this->mStripState->unstripBoth( $text );
5777 $text = Sanitizer::removeHTMLtags( $text );
5778 return $text;
5779 }
5780
5781 /**
5782 * @param string $text
5783 * @param Title $title
5784 * @param ParserOptions $options
5785 * @return string
5786 */
5787 public function testPst( $text, Title $title, ParserOptions $options ) {
5788 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5789 }
5790
5791 /**
5792 * @param string $text
5793 * @param Title $title
5794 * @param ParserOptions $options
5795 * @return string
5796 */
5797 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5798 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5799 }
5800
5801 /**
5802 * Call a callback function on all regions of the given text that are not
5803 * inside strip markers, and replace those regions with the return value
5804 * of the callback. For example, with input:
5805 *
5806 * aaa<MARKER>bbb
5807 *
5808 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5809 * two strings will be replaced with the value returned by the callback in
5810 * each case.
5811 *
5812 * @param string $s
5813 * @param callable $callback
5814 *
5815 * @return string
5816 */
5817 public function markerSkipCallback( $s, $callback ) {
5818 $i = 0;
5819 $out = '';
5820 while ( $i < strlen( $s ) ) {
5821 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5822 if ( $markerStart === false ) {
5823 $out .= call_user_func( $callback, substr( $s, $i ) );
5824 break;
5825 } else {
5826 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5827 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5828 if ( $markerEnd === false ) {
5829 $out .= substr( $s, $markerStart );
5830 break;
5831 } else {
5832 $markerEnd += strlen( self::MARKER_SUFFIX );
5833 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5834 $i = $markerEnd;
5835 }
5836 }
5837 }
5838 return $out;
5839 }
5840
5841 /**
5842 * Remove any strip markers found in the given text.
5843 *
5844 * @param string $text Input string
5845 * @return string
5846 */
5847 public function killMarkers( $text ) {
5848 return $this->mStripState->killMarkers( $text );
5849 }
5850
5851 /**
5852 * Save the parser state required to convert the given half-parsed text to
5853 * HTML. "Half-parsed" in this context means the output of
5854 * recursiveTagParse() or internalParse(). This output has strip markers
5855 * from replaceVariables (extensionSubstitution() etc.), and link
5856 * placeholders from replaceLinkHolders().
5857 *
5858 * Returns an array which can be serialized and stored persistently. This
5859 * array can later be loaded into another parser instance with
5860 * unserializeHalfParsedText(). The text can then be safely incorporated into
5861 * the return value of a parser hook.
5862 *
5863 * @param string $text
5864 *
5865 * @return array
5866 */
5867 public function serializeHalfParsedText( $text ) {
5868 $data = [
5869 'text' => $text,
5870 'version' => self::HALF_PARSED_VERSION,
5871 'stripState' => $this->mStripState->getSubState( $text ),
5872 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5873 ];
5874 return $data;
5875 }
5876
5877 /**
5878 * Load the parser state given in the $data array, which is assumed to
5879 * have been generated by serializeHalfParsedText(). The text contents is
5880 * extracted from the array, and its markers are transformed into markers
5881 * appropriate for the current Parser instance. This transformed text is
5882 * returned, and can be safely included in the return value of a parser
5883 * hook.
5884 *
5885 * If the $data array has been stored persistently, the caller should first
5886 * check whether it is still valid, by calling isValidHalfParsedText().
5887 *
5888 * @param array $data Serialized data
5889 * @throws MWException
5890 * @return string
5891 */
5892 public function unserializeHalfParsedText( $data ) {
5893 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5894 throw new MWException( __METHOD__ . ': invalid version' );
5895 }
5896
5897 # First, extract the strip state.
5898 $texts = [ $data['text'] ];
5899 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5900
5901 # Now renumber links
5902 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5903
5904 # Should be good to go.
5905 return $texts[0];
5906 }
5907
5908 /**
5909 * Returns true if the given array, presumed to be generated by
5910 * serializeHalfParsedText(), is compatible with the current version of the
5911 * parser.
5912 *
5913 * @param array $data
5914 *
5915 * @return bool
5916 */
5917 public function isValidHalfParsedText( $data ) {
5918 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5919 }
5920
5921 /**
5922 * Parsed a width param of imagelink like 300px or 200x300px
5923 *
5924 * @param string $value
5925 *
5926 * @return array
5927 * @since 1.20
5928 */
5929 public function parseWidthParam( $value ) {
5930 $parsedWidthParam = [];
5931 if ( $value === '' ) {
5932 return $parsedWidthParam;
5933 }
5934 $m = [];
5935 # (bug 13500) In both cases (width/height and width only),
5936 # permit trailing "px" for backward compatibility.
5937 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5938 $width = intval( $m[1] );
5939 $height = intval( $m[2] );
5940 $parsedWidthParam['width'] = $width;
5941 $parsedWidthParam['height'] = $height;
5942 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5943 $width = intval( $value );
5944 $parsedWidthParam['width'] = $width;
5945 }
5946 return $parsedWidthParam;
5947 }
5948
5949 /**
5950 * Lock the current instance of the parser.
5951 *
5952 * This is meant to stop someone from calling the parser
5953 * recursively and messing up all the strip state.
5954 *
5955 * @throws MWException If parser is in a parse
5956 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5957 */
5958 protected function lock() {
5959 if ( $this->mInParse ) {
5960 throw new MWException( "Parser state cleared while parsing. "
5961 . "Did you call Parser::parse recursively?" );
5962 }
5963 $this->mInParse = true;
5964
5965 $recursiveCheck = new ScopedCallback( function() {
5966 $this->mInParse = false;
5967 } );
5968
5969 return $recursiveCheck;
5970 }
5971
5972 /**
5973 * Strip outer <p></p> tag from the HTML source of a single paragraph.
5974 *
5975 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
5976 * or if there is more than one <p/> tag in the input HTML.
5977 *
5978 * @param string $html
5979 * @return string
5980 * @since 1.24
5981 */
5982 public static function stripOuterParagraph( $html ) {
5983 $m = [];
5984 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
5985 if ( strpos( $m[1], '</p>' ) === false ) {
5986 $html = $m[1];
5987 }
5988 }
5989
5990 return $html;
5991 }
5992
5993 /**
5994 * Return this parser if it is not doing anything, otherwise
5995 * get a fresh parser. You can use this method by doing
5996 * $myParser = $wgParser->getFreshParser(), or more simply
5997 * $wgParser->getFreshParser()->parse( ... );
5998 * if you're unsure if $wgParser is safe to use.
5999 *
6000 * @since 1.24
6001 * @return Parser A parser object that is not parsing anything
6002 */
6003 public function getFreshParser() {
6004 global $wgParserConf;
6005 if ( $this->mInParse ) {
6006 return new $wgParserConf['class']( $wgParserConf );
6007 } else {
6008 return $this;
6009 }
6010 }
6011
6012 /**
6013 * Set's up the PHP implementation of OOUI for use in this request
6014 * and instructs OutputPage to enable OOUI for itself.
6015 *
6016 * @since 1.26
6017 */
6018 public function enableOOUI() {
6019 OutputPage::setupOOUI();
6020 $this->mOutput->setEnableOOUI( true );
6021 }
6022 }
Wiklou, le Wiki du Wiklou