dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Use a low TTL for parser output when special pages are included
[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( $this->mTransparentTagHooks )
1268 );
1269 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1270
1271 # Tables need to come after variable replacement for things to work
1272 # properly; putting them before other transformations should keep
1273 # exciting things like link expansions from showing up in surprising
1274 # places.
1275 $text = $this->doTableStuff( $text );
1276
1277 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1278
1279 $text = $this->doDoubleUnderscore( $text );
1280
1281 $text = $this->doHeadings( $text );
1282 $text = $this->replaceInternalLinks( $text );
1283 $text = $this->doAllQuotes( $text );
1284 $text = $this->replaceExternalLinks( $text );
1285
1286 # replaceInternalLinks may sometimes leave behind
1287 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1288 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1289
1290 $text = $this->doMagicLinks( $text );
1291 $text = $this->formatHeadings( $text, $origText, $isMain );
1292
1293 return $text;
1294 }
1295
1296 /**
1297 * Helper function for parse() that transforms half-parsed HTML into fully
1298 * parsed HTML.
1299 *
1300 * @param string $text
1301 * @param bool $isMain
1302 * @param bool $linestart
1303 * @return string
1304 */
1305 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1306 $text = $this->mStripState->unstripGeneral( $text );
1307
1308 if ( $isMain ) {
1309 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1310 }
1311
1312 # Clean up special characters, only run once, next-to-last before doBlockLevels
1313 $fixtags = [
1314 # french spaces, last one Guillemet-left
1315 # only if there is something before the space
1316 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1317 # french spaces, Guillemet-right
1318 '/(\\302\\253) /' => '\\1&#160;',
1319 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1320 ];
1321 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1322
1323 $text = $this->doBlockLevels( $text, $linestart );
1324
1325 $this->replaceLinkHolders( $text );
1326
1327 /**
1328 * The input doesn't get language converted if
1329 * a) It's disabled
1330 * b) Content isn't converted
1331 * c) It's a conversion table
1332 * d) it is an interface message (which is in the user language)
1333 */
1334 if ( !( $this->mOptions->getDisableContentConversion()
1335 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1336 ) {
1337 if ( !$this->mOptions->getInterfaceMessage() ) {
1338 # The position of the convert() call should not be changed. it
1339 # assumes that the links are all replaced and the only thing left
1340 # is the <nowiki> mark.
1341 $text = $this->getConverterLanguage()->convert( $text );
1342 }
1343 }
1344
1345 $text = $this->mStripState->unstripNoWiki( $text );
1346
1347 if ( $isMain ) {
1348 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1349 }
1350
1351 $text = $this->replaceTransparentTags( $text );
1352 $text = $this->mStripState->unstripGeneral( $text );
1353
1354 $text = Sanitizer::normalizeCharReferences( $text );
1355
1356 if ( MWTidy::isEnabled() && $this->mOptions->getTidy() ) {
1357 $text = MWTidy::tidy( $text );
1358 $this->mOutput->addModuleStyles( MWTidy::getModuleStyles() );
1359 } else {
1360 # attempt to sanitize at least some nesting problems
1361 # (bug #2702 and quite a few others)
1362 $tidyregs = [
1363 # ''Something [http://www.cool.com cool''] -->
1364 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1365 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1366 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1367 # fix up an anchor inside another anchor, only
1368 # at least for a single single nested link (bug 3695)
1369 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1370 '\\1\\2</a>\\3</a>\\1\\4</a>',
1371 # fix div inside inline elements- doBlockLevels won't wrap a line which
1372 # contains a div, so fix it up here; replace
1373 # div with escaped text
1374 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1375 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1376 # remove empty italic or bold tag pairs, some
1377 # introduced by rules above
1378 '/<([bi])><\/\\1>/' => '',
1379 ];
1380
1381 $text = preg_replace(
1382 array_keys( $tidyregs ),
1383 array_values( $tidyregs ),
1384 $text );
1385 }
1386
1387 if ( $isMain ) {
1388 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1389 }
1390
1391 return $text;
1392 }
1393
1394 /**
1395 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1396 * magic external links.
1397 *
1398 * DML
1399 * @private
1400 *
1401 * @param string $text
1402 *
1403 * @return string
1404 */
1405 public function doMagicLinks( $text ) {
1406 $prots = wfUrlProtocolsWithoutProtRel();
1407 $urlChar = self::EXT_LINK_URL_CLASS;
1408 $addr = self::EXT_LINK_ADDR;
1409 $space = self::SPACE_NOT_NL; # non-newline space
1410 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1411 $spaces = "$space++"; # possessive match of 1 or more spaces
1412 $text = preg_replace_callback(
1413 '!(?: # Start cases
1414 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1415 (<.*?>) | # m[2]: Skip stuff inside
1416 # HTML elements' . "
1417 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1418 # m[4]: Post-protocol path
1419 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1420 ([0-9]+)\b |
1421 \bISBN $spaces ( # m[6]: ISBN, capture number
1422 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1423 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1424 [0-9Xx] # check digit
1425 )\b
1426 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1427 return $text;
1428 }
1429
1430 /**
1431 * @throws MWException
1432 * @param array $m
1433 * @return HTML|string
1434 */
1435 public function magicLinkCallback( $m ) {
1436 if ( isset( $m[1] ) && $m[1] !== '' ) {
1437 # Skip anchor
1438 return $m[0];
1439 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1440 # Skip HTML element
1441 return $m[0];
1442 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1443 # Free external link
1444 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1445 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1446 # RFC or PMID
1447 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1448 $keyword = 'RFC';
1449 $urlmsg = 'rfcurl';
1450 $cssClass = 'mw-magiclink-rfc';
1451 $id = $m[5];
1452 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1453 $keyword = 'PMID';
1454 $urlmsg = 'pubmedurl';
1455 $cssClass = 'mw-magiclink-pmid';
1456 $id = $m[5];
1457 } else {
1458 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1459 substr( $m[0], 0, 20 ) . '"' );
1460 }
1461 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1462 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1463 } elseif ( isset( $m[6] ) && $m[6] !== '' ) {
1464 # ISBN
1465 $isbn = $m[6];
1466 $space = self::SPACE_NOT_NL; # non-newline space
1467 $isbn = preg_replace( "/$space/", ' ', $isbn );
1468 $num = strtr( $isbn, [
1469 '-' => '',
1470 ' ' => '',
1471 'x' => 'X',
1472 ] );
1473 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1474 return '<a href="' .
1475 htmlspecialchars( $titleObj->getLocalURL() ) .
1476 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1477 } else {
1478 return $m[0];
1479 }
1480 }
1481
1482 /**
1483 * Make a free external link, given a user-supplied URL
1484 *
1485 * @param string $url
1486 * @param int $numPostProto
1487 * The number of characters after the protocol.
1488 * @return string HTML
1489 * @private
1490 */
1491 public function makeFreeExternalLink( $url, $numPostProto ) {
1492 $trail = '';
1493
1494 # The characters '<' and '>' (which were escaped by
1495 # removeHTMLtags()) should not be included in
1496 # URLs, per RFC 2396.
1497 # Make &nbsp; terminate a URL as well (bug T84937)
1498 $m2 = [];
1499 if ( preg_match(
1500 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1501 $url,
1502 $m2,
1503 PREG_OFFSET_CAPTURE
1504 ) ) {
1505 $trail = substr( $url, $m2[0][1] ) . $trail;
1506 $url = substr( $url, 0, $m2[0][1] );
1507 }
1508
1509 # Move trailing punctuation to $trail
1510 $sep = ',;\.:!?';
1511 # If there is no left bracket, then consider right brackets fair game too
1512 if ( strpos( $url, '(' ) === false ) {
1513 $sep .= ')';
1514 }
1515
1516 $urlRev = strrev( $url );
1517 $numSepChars = strspn( $urlRev, $sep );
1518 # Don't break a trailing HTML entity by moving the ; into $trail
1519 # This is in hot code, so use substr_compare to avoid having to
1520 # create a new string object for the comparison
1521 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1522 # more optimization: instead of running preg_match with a $
1523 # anchor, which can be slow, do the match on the reversed
1524 # string starting at the desired offset.
1525 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1526 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1527 $numSepChars--;
1528 }
1529 }
1530 if ( $numSepChars ) {
1531 $trail = substr( $url, -$numSepChars ) . $trail;
1532 $url = substr( $url, 0, -$numSepChars );
1533 }
1534
1535 # Verify that we still have a real URL after trail removal, and
1536 # not just lone protocol
1537 if ( strlen( $trail ) >= $numPostProto ) {
1538 return $url . $trail;
1539 }
1540
1541 $url = Sanitizer::cleanUrl( $url );
1542
1543 # Is this an external image?
1544 $text = $this->maybeMakeExternalImage( $url );
1545 if ( $text === false ) {
1546 # Not an image, make a link
1547 $text = Linker::makeExternalLink( $url,
1548 $this->getConverterLanguage()->markNoConversion( $url, true ),
1549 true, 'free',
1550 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1551 # Register it in the output object...
1552 # Replace unnecessary URL escape codes with their equivalent characters
1553 $pasteurized = self::normalizeLinkUrl( $url );
1554 $this->mOutput->addExternalLink( $pasteurized );
1555 }
1556 return $text . $trail;
1557 }
1558
1559 /**
1560 * Parse headers and return html
1561 *
1562 * @private
1563 *
1564 * @param string $text
1565 *
1566 * @return string
1567 */
1568 public function doHeadings( $text ) {
1569 for ( $i = 6; $i >= 1; --$i ) {
1570 $h = str_repeat( '=', $i );
1571 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1572 }
1573 return $text;
1574 }
1575
1576 /**
1577 * Replace single quotes with HTML markup
1578 * @private
1579 *
1580 * @param string $text
1581 *
1582 * @return string The altered text
1583 */
1584 public function doAllQuotes( $text ) {
1585 $outtext = '';
1586 $lines = StringUtils::explode( "\n", $text );
1587 foreach ( $lines as $line ) {
1588 $outtext .= $this->doQuotes( $line ) . "\n";
1589 }
1590 $outtext = substr( $outtext, 0, -1 );
1591 return $outtext;
1592 }
1593
1594 /**
1595 * Helper function for doAllQuotes()
1596 *
1597 * @param string $text
1598 *
1599 * @return string
1600 */
1601 public function doQuotes( $text ) {
1602 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1603 $countarr = count( $arr );
1604 if ( $countarr == 1 ) {
1605 return $text;
1606 }
1607
1608 // First, do some preliminary work. This may shift some apostrophes from
1609 // being mark-up to being text. It also counts the number of occurrences
1610 // of bold and italics mark-ups.
1611 $numbold = 0;
1612 $numitalics = 0;
1613 for ( $i = 1; $i < $countarr; $i += 2 ) {
1614 $thislen = strlen( $arr[$i] );
1615 // If there are ever four apostrophes, assume the first is supposed to
1616 // be text, and the remaining three constitute mark-up for bold text.
1617 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1618 if ( $thislen == 4 ) {
1619 $arr[$i - 1] .= "'";
1620 $arr[$i] = "'''";
1621 $thislen = 3;
1622 } elseif ( $thislen > 5 ) {
1623 // If there are more than 5 apostrophes in a row, assume they're all
1624 // text except for the last 5.
1625 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1626 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1627 $arr[$i] = "'''''";
1628 $thislen = 5;
1629 }
1630 // Count the number of occurrences of bold and italics mark-ups.
1631 if ( $thislen == 2 ) {
1632 $numitalics++;
1633 } elseif ( $thislen == 3 ) {
1634 $numbold++;
1635 } elseif ( $thislen == 5 ) {
1636 $numitalics++;
1637 $numbold++;
1638 }
1639 }
1640
1641 // If there is an odd number of both bold and italics, it is likely
1642 // that one of the bold ones was meant to be an apostrophe followed
1643 // by italics. Which one we cannot know for certain, but it is more
1644 // likely to be one that has a single-letter word before it.
1645 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1646 $firstsingleletterword = -1;
1647 $firstmultiletterword = -1;
1648 $firstspace = -1;
1649 for ( $i = 1; $i < $countarr; $i += 2 ) {
1650 if ( strlen( $arr[$i] ) == 3 ) {
1651 $x1 = substr( $arr[$i - 1], -1 );
1652 $x2 = substr( $arr[$i - 1], -2, 1 );
1653 if ( $x1 === ' ' ) {
1654 if ( $firstspace == -1 ) {
1655 $firstspace = $i;
1656 }
1657 } elseif ( $x2 === ' ' ) {
1658 $firstsingleletterword = $i;
1659 // if $firstsingleletterword is set, we don't
1660 // look at the other options, so we can bail early.
1661 break;
1662 } else {
1663 if ( $firstmultiletterword == -1 ) {
1664 $firstmultiletterword = $i;
1665 }
1666 }
1667 }
1668 }
1669
1670 // If there is a single-letter word, use it!
1671 if ( $firstsingleletterword > -1 ) {
1672 $arr[$firstsingleletterword] = "''";
1673 $arr[$firstsingleletterword - 1] .= "'";
1674 } elseif ( $firstmultiletterword > -1 ) {
1675 // If not, but there's a multi-letter word, use that one.
1676 $arr[$firstmultiletterword] = "''";
1677 $arr[$firstmultiletterword - 1] .= "'";
1678 } elseif ( $firstspace > -1 ) {
1679 // ... otherwise use the first one that has neither.
1680 // (notice that it is possible for all three to be -1 if, for example,
1681 // there is only one pentuple-apostrophe in the line)
1682 $arr[$firstspace] = "''";
1683 $arr[$firstspace - 1] .= "'";
1684 }
1685 }
1686
1687 // Now let's actually convert our apostrophic mush to HTML!
1688 $output = '';
1689 $buffer = '';
1690 $state = '';
1691 $i = 0;
1692 foreach ( $arr as $r ) {
1693 if ( ( $i % 2 ) == 0 ) {
1694 if ( $state === 'both' ) {
1695 $buffer .= $r;
1696 } else {
1697 $output .= $r;
1698 }
1699 } else {
1700 $thislen = strlen( $r );
1701 if ( $thislen == 2 ) {
1702 if ( $state === 'i' ) {
1703 $output .= '</i>';
1704 $state = '';
1705 } elseif ( $state === 'bi' ) {
1706 $output .= '</i>';
1707 $state = 'b';
1708 } elseif ( $state === 'ib' ) {
1709 $output .= '</b></i><b>';
1710 $state = 'b';
1711 } elseif ( $state === 'both' ) {
1712 $output .= '<b><i>' . $buffer . '</i>';
1713 $state = 'b';
1714 } else { // $state can be 'b' or ''
1715 $output .= '<i>';
1716 $state .= 'i';
1717 }
1718 } elseif ( $thislen == 3 ) {
1719 if ( $state === 'b' ) {
1720 $output .= '</b>';
1721 $state = '';
1722 } elseif ( $state === 'bi' ) {
1723 $output .= '</i></b><i>';
1724 $state = 'i';
1725 } elseif ( $state === 'ib' ) {
1726 $output .= '</b>';
1727 $state = 'i';
1728 } elseif ( $state === 'both' ) {
1729 $output .= '<i><b>' . $buffer . '</b>';
1730 $state = 'i';
1731 } else { // $state can be 'i' or ''
1732 $output .= '<b>';
1733 $state .= 'b';
1734 }
1735 } elseif ( $thislen == 5 ) {
1736 if ( $state === 'b' ) {
1737 $output .= '</b><i>';
1738 $state = 'i';
1739 } elseif ( $state === 'i' ) {
1740 $output .= '</i><b>';
1741 $state = 'b';
1742 } elseif ( $state === 'bi' ) {
1743 $output .= '</i></b>';
1744 $state = '';
1745 } elseif ( $state === 'ib' ) {
1746 $output .= '</b></i>';
1747 $state = '';
1748 } elseif ( $state === 'both' ) {
1749 $output .= '<i><b>' . $buffer . '</b></i>';
1750 $state = '';
1751 } else { // ($state == '')
1752 $buffer = '';
1753 $state = 'both';
1754 }
1755 }
1756 }
1757 $i++;
1758 }
1759 // Now close all remaining tags. Notice that the order is important.
1760 if ( $state === 'b' || $state === 'ib' ) {
1761 $output .= '</b>';
1762 }
1763 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1764 $output .= '</i>';
1765 }
1766 if ( $state === 'bi' ) {
1767 $output .= '</b>';
1768 }
1769 // There might be lonely ''''', so make sure we have a buffer
1770 if ( $state === 'both' && $buffer ) {
1771 $output .= '<b><i>' . $buffer . '</i></b>';
1772 }
1773 return $output;
1774 }
1775
1776 /**
1777 * Replace external links (REL)
1778 *
1779 * Note: this is all very hackish and the order of execution matters a lot.
1780 * Make sure to run tests/parserTests.php if you change this code.
1781 *
1782 * @private
1783 *
1784 * @param string $text
1785 *
1786 * @throws MWException
1787 * @return string
1788 */
1789 public function replaceExternalLinks( $text ) {
1790
1791 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1792 if ( $bits === false ) {
1793 throw new MWException( "PCRE needs to be compiled with "
1794 . "--enable-unicode-properties in order for MediaWiki to function" );
1795 }
1796 $s = array_shift( $bits );
1797
1798 $i = 0;
1799 while ( $i < count( $bits ) ) {
1800 $url = $bits[$i++];
1801 $i++; // protocol
1802 $text = $bits[$i++];
1803 $trail = $bits[$i++];
1804
1805 # The characters '<' and '>' (which were escaped by
1806 # removeHTMLtags()) should not be included in
1807 # URLs, per RFC 2396.
1808 $m2 = [];
1809 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1810 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1811 $url = substr( $url, 0, $m2[0][1] );
1812 }
1813
1814 # If the link text is an image URL, replace it with an <img> tag
1815 # This happened by accident in the original parser, but some people used it extensively
1816 $img = $this->maybeMakeExternalImage( $text );
1817 if ( $img !== false ) {
1818 $text = $img;
1819 }
1820
1821 $dtrail = '';
1822
1823 # Set linktype for CSS - if URL==text, link is essentially free
1824 $linktype = ( $text === $url ) ? 'free' : 'text';
1825
1826 # No link text, e.g. [http://domain.tld/some.link]
1827 if ( $text == '' ) {
1828 # Autonumber
1829 $langObj = $this->getTargetLanguage();
1830 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1831 $linktype = 'autonumber';
1832 } else {
1833 # Have link text, e.g. [http://domain.tld/some.link text]s
1834 # Check for trail
1835 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1836 }
1837
1838 $text = $this->getConverterLanguage()->markNoConversion( $text );
1839
1840 $url = Sanitizer::cleanUrl( $url );
1841
1842 # Use the encoded URL
1843 # This means that users can paste URLs directly into the text
1844 # Funny characters like ö aren't valid in URLs anyway
1845 # This was changed in August 2004
1846 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1847 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1848
1849 # Register link in the output object.
1850 # Replace unnecessary URL escape codes with the referenced character
1851 # This prevents spammers from hiding links from the filters
1852 $pasteurized = self::normalizeLinkUrl( $url );
1853 $this->mOutput->addExternalLink( $pasteurized );
1854 }
1855
1856 return $s;
1857 }
1858
1859 /**
1860 * Get the rel attribute for a particular external link.
1861 *
1862 * @since 1.21
1863 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1864 * nofollow if appropriate
1865 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1866 * @return string|null Rel attribute for $url
1867 */
1868 public static function getExternalLinkRel( $url = false, $title = null ) {
1869 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1870 $ns = $title ? $title->getNamespace() : false;
1871 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1872 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1873 ) {
1874 return 'nofollow';
1875 }
1876 return null;
1877 }
1878
1879 /**
1880 * Get an associative array of additional HTML attributes appropriate for a
1881 * particular external link. This currently may include rel => nofollow
1882 * (depending on configuration, namespace, and the URL's domain) and/or a
1883 * target attribute (depending on configuration).
1884 *
1885 * @param string $url URL to extract the domain from for rel =>
1886 * nofollow if appropriate
1887 * @return array Associative array of HTML attributes
1888 */
1889 public function getExternalLinkAttribs( $url ) {
1890 $attribs = [];
1891 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1892
1893 $target = $this->mOptions->getExternalLinkTarget();
1894 if ( $target ) {
1895 $attribs['target'] = $target;
1896 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1897 // T133507. New windows can navigate parent cross-origin.
1898 // Including noreferrer due to lacking browser
1899 // support of noopener. Eventually noreferrer should be removed.
1900 if ( $rel !== '' ) {
1901 $rel .= ' ';
1902 }
1903 $rel .= 'noreferrer noopener';
1904 }
1905 }
1906 $attribs['rel'] = $rel;
1907 return $attribs;
1908 }
1909
1910 /**
1911 * Replace unusual escape codes in a URL with their equivalent characters
1912 *
1913 * @deprecated since 1.24, use normalizeLinkUrl
1914 * @param string $url
1915 * @return string
1916 */
1917 public static function replaceUnusualEscapes( $url ) {
1918 wfDeprecated( __METHOD__, '1.24' );
1919 return self::normalizeLinkUrl( $url );
1920 }
1921
1922 /**
1923 * Replace unusual escape codes in a URL with their equivalent characters
1924 *
1925 * This generally follows the syntax defined in RFC 3986, with special
1926 * consideration for HTTP query strings.
1927 *
1928 * @param string $url
1929 * @return string
1930 */
1931 public static function normalizeLinkUrl( $url ) {
1932 # First, make sure unsafe characters are encoded
1933 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1934 function ( $m ) {
1935 return rawurlencode( $m[0] );
1936 },
1937 $url
1938 );
1939
1940 $ret = '';
1941 $end = strlen( $url );
1942
1943 # Fragment part - 'fragment'
1944 $start = strpos( $url, '#' );
1945 if ( $start !== false && $start < $end ) {
1946 $ret = self::normalizeUrlComponent(
1947 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1948 $end = $start;
1949 }
1950
1951 # Query part - 'query' minus &=+;
1952 $start = strpos( $url, '?' );
1953 if ( $start !== false && $start < $end ) {
1954 $ret = self::normalizeUrlComponent(
1955 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1956 $end = $start;
1957 }
1958
1959 # Scheme and path part - 'pchar'
1960 # (we assume no userinfo or encoded colons in the host)
1961 $ret = self::normalizeUrlComponent(
1962 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1963
1964 return $ret;
1965 }
1966
1967 private static function normalizeUrlComponent( $component, $unsafe ) {
1968 $callback = function ( $matches ) use ( $unsafe ) {
1969 $char = urldecode( $matches[0] );
1970 $ord = ord( $char );
1971 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1972 # Unescape it
1973 return $char;
1974 } else {
1975 # Leave it escaped, but use uppercase for a-f
1976 return strtoupper( $matches[0] );
1977 }
1978 };
1979 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1980 }
1981
1982 /**
1983 * make an image if it's allowed, either through the global
1984 * option, through the exception, or through the on-wiki whitelist
1985 *
1986 * @param string $url
1987 *
1988 * @return string
1989 */
1990 private function maybeMakeExternalImage( $url ) {
1991 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1992 $imagesexception = !empty( $imagesfrom );
1993 $text = false;
1994 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1995 if ( $imagesexception && is_array( $imagesfrom ) ) {
1996 $imagematch = false;
1997 foreach ( $imagesfrom as $match ) {
1998 if ( strpos( $url, $match ) === 0 ) {
1999 $imagematch = true;
2000 break;
2001 }
2002 }
2003 } elseif ( $imagesexception ) {
2004 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2005 } else {
2006 $imagematch = false;
2007 }
2008
2009 if ( $this->mOptions->getAllowExternalImages()
2010 || ( $imagesexception && $imagematch )
2011 ) {
2012 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2013 # Image found
2014 $text = Linker::makeExternalImage( $url );
2015 }
2016 }
2017 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2018 && preg_match( self::EXT_IMAGE_REGEX, $url )
2019 ) {
2020 $whitelist = explode(
2021 "\n",
2022 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2023 );
2024
2025 foreach ( $whitelist as $entry ) {
2026 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2027 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2028 continue;
2029 }
2030 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2031 # Image matches a whitelist entry
2032 $text = Linker::makeExternalImage( $url );
2033 break;
2034 }
2035 }
2036 }
2037 return $text;
2038 }
2039
2040 /**
2041 * Process [[ ]] wikilinks
2042 *
2043 * @param string $s
2044 *
2045 * @return string Processed text
2046 *
2047 * @private
2048 */
2049 public function replaceInternalLinks( $s ) {
2050 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2051 return $s;
2052 }
2053
2054 /**
2055 * Process [[ ]] wikilinks (RIL)
2056 * @param string $s
2057 * @throws MWException
2058 * @return LinkHolderArray
2059 *
2060 * @private
2061 */
2062 public function replaceInternalLinks2( &$s ) {
2063 global $wgExtraInterlanguageLinkPrefixes;
2064
2065 static $tc = false, $e1, $e1_img;
2066 # the % is needed to support urlencoded titles as well
2067 if ( !$tc ) {
2068 $tc = Title::legalChars() . '#%';
2069 # Match a link having the form [[namespace:link|alternate]]trail
2070 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2071 # Match cases where there is no "]]", which might still be images
2072 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2073 }
2074
2075 $holders = new LinkHolderArray( $this );
2076
2077 # split the entire text string on occurrences of [[
2078 $a = StringUtils::explode( '[[', ' ' . $s );
2079 # get the first element (all text up to first [[), and remove the space we added
2080 $s = $a->current();
2081 $a->next();
2082 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2083 $s = substr( $s, 1 );
2084
2085 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2086 $e2 = null;
2087 if ( $useLinkPrefixExtension ) {
2088 # Match the end of a line for a word that's not followed by whitespace,
2089 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2090 global $wgContLang;
2091 $charset = $wgContLang->linkPrefixCharset();
2092 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2093 }
2094
2095 if ( is_null( $this->mTitle ) ) {
2096 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2097 }
2098 $nottalk = !$this->mTitle->isTalkPage();
2099
2100 if ( $useLinkPrefixExtension ) {
2101 $m = [];
2102 if ( preg_match( $e2, $s, $m ) ) {
2103 $first_prefix = $m[2];
2104 } else {
2105 $first_prefix = false;
2106 }
2107 } else {
2108 $prefix = '';
2109 }
2110
2111 $useSubpages = $this->areSubpagesAllowed();
2112
2113 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2114 # Loop for each link
2115 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2116 // @codingStandardsIgnoreEnd
2117
2118 # Check for excessive memory usage
2119 if ( $holders->isBig() ) {
2120 # Too big
2121 # Do the existence check, replace the link holders and clear the array
2122 $holders->replace( $s );
2123 $holders->clear();
2124 }
2125
2126 if ( $useLinkPrefixExtension ) {
2127 if ( preg_match( $e2, $s, $m ) ) {
2128 $prefix = $m[2];
2129 $s = $m[1];
2130 } else {
2131 $prefix = '';
2132 }
2133 # first link
2134 if ( $first_prefix ) {
2135 $prefix = $first_prefix;
2136 $first_prefix = false;
2137 }
2138 }
2139
2140 $might_be_img = false;
2141
2142 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2143 $text = $m[2];
2144 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2145 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2146 # the real problem is with the $e1 regex
2147 # See bug 1300.
2148 # Still some problems for cases where the ] is meant to be outside punctuation,
2149 # and no image is in sight. See bug 2095.
2150 if ( $text !== ''
2151 && substr( $m[3], 0, 1 ) === ']'
2152 && strpos( $text, '[' ) !== false
2153 ) {
2154 $text .= ']'; # so that replaceExternalLinks($text) works later
2155 $m[3] = substr( $m[3], 1 );
2156 }
2157 # fix up urlencoded title texts
2158 if ( strpos( $m[1], '%' ) !== false ) {
2159 # Should anchors '#' also be rejected?
2160 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2161 }
2162 $trail = $m[3];
2163 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2164 # Invalid, but might be an image with a link in its caption
2165 $might_be_img = true;
2166 $text = $m[2];
2167 if ( strpos( $m[1], '%' ) !== false ) {
2168 $m[1] = rawurldecode( $m[1] );
2169 }
2170 $trail = "";
2171 } else { # Invalid form; output directly
2172 $s .= $prefix . '[[' . $line;
2173 continue;
2174 }
2175
2176 $origLink = $m[1];
2177
2178 # Don't allow internal links to pages containing
2179 # PROTO: where PROTO is a valid URL protocol; these
2180 # should be external links.
2181 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2182 $s .= $prefix . '[[' . $line;
2183 continue;
2184 }
2185
2186 # Make subpage if necessary
2187 if ( $useSubpages ) {
2188 $link = $this->maybeDoSubpageLink( $origLink, $text );
2189 } else {
2190 $link = $origLink;
2191 }
2192
2193 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2194 if ( !$noforce ) {
2195 # Strip off leading ':'
2196 $link = substr( $link, 1 );
2197 }
2198
2199 $unstrip = $this->mStripState->unstripNoWiki( $link );
2200 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2201 if ( $nt === null ) {
2202 $s .= $prefix . '[[' . $line;
2203 continue;
2204 }
2205
2206 $ns = $nt->getNamespace();
2207 $iw = $nt->getInterwiki();
2208
2209 if ( $might_be_img ) { # if this is actually an invalid link
2210 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2211 $found = false;
2212 while ( true ) {
2213 # look at the next 'line' to see if we can close it there
2214 $a->next();
2215 $next_line = $a->current();
2216 if ( $next_line === false || $next_line === null ) {
2217 break;
2218 }
2219 $m = explode( ']]', $next_line, 3 );
2220 if ( count( $m ) == 3 ) {
2221 # the first ]] closes the inner link, the second the image
2222 $found = true;
2223 $text .= "[[{$m[0]}]]{$m[1]}";
2224 $trail = $m[2];
2225 break;
2226 } elseif ( count( $m ) == 2 ) {
2227 # if there's exactly one ]] that's fine, we'll keep looking
2228 $text .= "[[{$m[0]}]]{$m[1]}";
2229 } else {
2230 # if $next_line is invalid too, we need look no further
2231 $text .= '[[' . $next_line;
2232 break;
2233 }
2234 }
2235 if ( !$found ) {
2236 # we couldn't find the end of this imageLink, so output it raw
2237 # but don't ignore what might be perfectly normal links in the text we've examined
2238 $holders->merge( $this->replaceInternalLinks2( $text ) );
2239 $s .= "{$prefix}[[$link|$text";
2240 # note: no $trail, because without an end, there *is* no trail
2241 continue;
2242 }
2243 } else { # it's not an image, so output it raw
2244 $s .= "{$prefix}[[$link|$text";
2245 # note: no $trail, because without an end, there *is* no trail
2246 continue;
2247 }
2248 }
2249
2250 $wasblank = ( $text == '' );
2251 if ( $wasblank ) {
2252 $text = $link;
2253 } else {
2254 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2255 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2256 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2257 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2258 $text = $this->doQuotes( $text );
2259 }
2260
2261 # Link not escaped by : , create the various objects
2262 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2263 # Interwikis
2264 if (
2265 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2266 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2267 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2268 )
2269 ) {
2270 # Bug 24502: filter duplicates
2271 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2272 $this->mLangLinkLanguages[$iw] = true;
2273 $this->mOutput->addLanguageLink( $nt->getFullText() );
2274 }
2275
2276 $s = rtrim( $s . $prefix );
2277 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2278 continue;
2279 }
2280
2281 if ( $ns == NS_FILE ) {
2282 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2283 if ( $wasblank ) {
2284 # if no parameters were passed, $text
2285 # becomes something like "File:Foo.png",
2286 # which we don't want to pass on to the
2287 # image generator
2288 $text = '';
2289 } else {
2290 # recursively parse links inside the image caption
2291 # actually, this will parse them in any other parameters, too,
2292 # but it might be hard to fix that, and it doesn't matter ATM
2293 $text = $this->replaceExternalLinks( $text );
2294 $holders->merge( $this->replaceInternalLinks2( $text ) );
2295 }
2296 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2297 $s .= $prefix . $this->armorLinks(
2298 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2299 } else {
2300 $s .= $prefix . $trail;
2301 }
2302 continue;
2303 }
2304
2305 if ( $ns == NS_CATEGORY ) {
2306 $s = rtrim( $s . "\n" ); # bug 87
2307
2308 if ( $wasblank ) {
2309 $sortkey = $this->getDefaultSort();
2310 } else {
2311 $sortkey = $text;
2312 }
2313 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2314 $sortkey = str_replace( "\n", '', $sortkey );
2315 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2316 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2317
2318 /**
2319 * Strip the whitespace Category links produce, see bug 87
2320 */
2321 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2322
2323 continue;
2324 }
2325 }
2326
2327 # Self-link checking. For some languages, variants of the title are checked in
2328 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2329 # for linking to a different variant.
2330 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2331 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2332 continue;
2333 }
2334
2335 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2336 # @todo FIXME: Should do batch file existence checks, see comment below
2337 if ( $ns == NS_MEDIA ) {
2338 # Give extensions a chance to select the file revision for us
2339 $options = [];
2340 $descQuery = false;
2341 Hooks::run( 'BeforeParserFetchFileAndTitle',
2342 [ $this, $nt, &$options, &$descQuery ] );
2343 # Fetch and register the file (file title may be different via hooks)
2344 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2345 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2346 $s .= $prefix . $this->armorLinks(
2347 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2348 continue;
2349 }
2350
2351 # Some titles, such as valid special pages or files in foreign repos, should
2352 # be shown as bluelinks even though they're not included in the page table
2353 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2354 # batch file existence checks for NS_FILE and NS_MEDIA
2355 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2356 $this->mOutput->addLink( $nt );
2357 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2358 } else {
2359 # Links will be added to the output link list after checking
2360 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2361 }
2362 }
2363 return $holders;
2364 }
2365
2366 /**
2367 * Render a forced-blue link inline; protect against double expansion of
2368 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2369 * Since this little disaster has to split off the trail text to avoid
2370 * breaking URLs in the following text without breaking trails on the
2371 * wiki links, it's been made into a horrible function.
2372 *
2373 * @param Title $nt
2374 * @param string $text
2375 * @param string $trail
2376 * @param string $prefix
2377 * @return string HTML-wikitext mix oh yuck
2378 */
2379 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2380 list( $inside, $trail ) = Linker::splitTrail( $trail );
2381
2382 if ( $text == '' ) {
2383 $text = htmlspecialchars( $nt->getPrefixedText() );
2384 }
2385
2386 $link = $this->getLinkRenderer()->makeKnownLink(
2387 $nt, new HtmlArmor( "$prefix$text$inside" )
2388 );
2389
2390 return $this->armorLinks( $link ) . $trail;
2391 }
2392
2393 /**
2394 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2395 * going to go through further parsing steps before inline URL expansion.
2396 *
2397 * Not needed quite as much as it used to be since free links are a bit
2398 * more sensible these days. But bracketed links are still an issue.
2399 *
2400 * @param string $text More-or-less HTML
2401 * @return string Less-or-more HTML with NOPARSE bits
2402 */
2403 public function armorLinks( $text ) {
2404 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2405 self::MARKER_PREFIX . "NOPARSE$1", $text );
2406 }
2407
2408 /**
2409 * Return true if subpage links should be expanded on this page.
2410 * @return bool
2411 */
2412 public function areSubpagesAllowed() {
2413 # Some namespaces don't allow subpages
2414 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2415 }
2416
2417 /**
2418 * Handle link to subpage if necessary
2419 *
2420 * @param string $target The source of the link
2421 * @param string &$text The link text, modified as necessary
2422 * @return string The full name of the link
2423 * @private
2424 */
2425 public function maybeDoSubpageLink( $target, &$text ) {
2426 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2427 }
2428
2429 /**
2430 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2431 *
2432 * @param string $text
2433 * @param bool $linestart Whether or not this is at the start of a line.
2434 * @private
2435 * @return string The lists rendered as HTML
2436 */
2437 public function doBlockLevels( $text, $linestart ) {
2438 return BlockLevelPass::doBlockLevels( $text, $linestart );
2439 }
2440
2441 /**
2442 * Return value of a magic variable (like PAGENAME)
2443 *
2444 * @private
2445 *
2446 * @param int $index
2447 * @param bool|PPFrame $frame
2448 *
2449 * @throws MWException
2450 * @return string
2451 */
2452 public function getVariableValue( $index, $frame = false ) {
2453 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2454 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2455
2456 if ( is_null( $this->mTitle ) ) {
2457 // If no title set, bad things are going to happen
2458 // later. Title should always be set since this
2459 // should only be called in the middle of a parse
2460 // operation (but the unit-tests do funky stuff)
2461 throw new MWException( __METHOD__ . ' Should only be '
2462 . ' called while parsing (no title set)' );
2463 }
2464
2465 /**
2466 * Some of these require message or data lookups and can be
2467 * expensive to check many times.
2468 */
2469 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2470 if ( isset( $this->mVarCache[$index] ) ) {
2471 return $this->mVarCache[$index];
2472 }
2473 }
2474
2475 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2476 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2477
2478 $pageLang = $this->getFunctionLang();
2479
2480 switch ( $index ) {
2481 case '!':
2482 $value = '|';
2483 break;
2484 case 'currentmonth':
2485 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2486 break;
2487 case 'currentmonth1':
2488 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2489 break;
2490 case 'currentmonthname':
2491 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2492 break;
2493 case 'currentmonthnamegen':
2494 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2495 break;
2496 case 'currentmonthabbrev':
2497 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2498 break;
2499 case 'currentday':
2500 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2501 break;
2502 case 'currentday2':
2503 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2504 break;
2505 case 'localmonth':
2506 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2507 break;
2508 case 'localmonth1':
2509 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2510 break;
2511 case 'localmonthname':
2512 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2513 break;
2514 case 'localmonthnamegen':
2515 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2516 break;
2517 case 'localmonthabbrev':
2518 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2519 break;
2520 case 'localday':
2521 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2522 break;
2523 case 'localday2':
2524 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2525 break;
2526 case 'pagename':
2527 $value = wfEscapeWikiText( $this->mTitle->getText() );
2528 break;
2529 case 'pagenamee':
2530 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2531 break;
2532 case 'fullpagename':
2533 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2534 break;
2535 case 'fullpagenamee':
2536 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2537 break;
2538 case 'subpagename':
2539 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2540 break;
2541 case 'subpagenamee':
2542 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2543 break;
2544 case 'rootpagename':
2545 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2546 break;
2547 case 'rootpagenamee':
2548 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2549 ' ',
2550 '_',
2551 $this->mTitle->getRootText()
2552 ) ) );
2553 break;
2554 case 'basepagename':
2555 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2556 break;
2557 case 'basepagenamee':
2558 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2559 ' ',
2560 '_',
2561 $this->mTitle->getBaseText()
2562 ) ) );
2563 break;
2564 case 'talkpagename':
2565 if ( $this->mTitle->canTalk() ) {
2566 $talkPage = $this->mTitle->getTalkPage();
2567 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2568 } else {
2569 $value = '';
2570 }
2571 break;
2572 case 'talkpagenamee':
2573 if ( $this->mTitle->canTalk() ) {
2574 $talkPage = $this->mTitle->getTalkPage();
2575 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2576 } else {
2577 $value = '';
2578 }
2579 break;
2580 case 'subjectpagename':
2581 $subjPage = $this->mTitle->getSubjectPage();
2582 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2583 break;
2584 case 'subjectpagenamee':
2585 $subjPage = $this->mTitle->getSubjectPage();
2586 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2587 break;
2588 case 'pageid': // requested in bug 23427
2589 $pageid = $this->getTitle()->getArticleID();
2590 if ( $pageid == 0 ) {
2591 # 0 means the page doesn't exist in the database,
2592 # which means the user is previewing a new page.
2593 # The vary-revision flag must be set, because the magic word
2594 # will have a different value once the page is saved.
2595 $this->mOutput->setFlag( 'vary-revision' );
2596 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2597 }
2598 $value = $pageid ? $pageid : null;
2599 break;
2600 case 'revisionid':
2601 # Let the edit saving system know we should parse the page
2602 # *after* a revision ID has been assigned.
2603 $this->mOutput->setFlag( 'vary-revision' );
2604 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2605 $value = $this->mRevisionId;
2606 break;
2607 case 'revisionday':
2608 # Let the edit saving system know we should parse the page
2609 # *after* a revision ID has been assigned. This is for null edits.
2610 $this->mOutput->setFlag( 'vary-revision' );
2611 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2612 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2613 break;
2614 case 'revisionday2':
2615 # Let the edit saving system know we should parse the page
2616 # *after* a revision ID has been assigned. This is for null edits.
2617 $this->mOutput->setFlag( 'vary-revision' );
2618 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2619 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2620 break;
2621 case 'revisionmonth':
2622 # Let the edit saving system know we should parse the page
2623 # *after* a revision ID has been assigned. This is for null edits.
2624 $this->mOutput->setFlag( 'vary-revision' );
2625 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2626 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2627 break;
2628 case 'revisionmonth1':
2629 # Let the edit saving system know we should parse the page
2630 # *after* a revision ID has been assigned. This is for null edits.
2631 $this->mOutput->setFlag( 'vary-revision' );
2632 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2633 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2634 break;
2635 case 'revisionyear':
2636 # Let the edit saving system know we should parse the page
2637 # *after* a revision ID has been assigned. This is for null edits.
2638 $this->mOutput->setFlag( 'vary-revision' );
2639 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2640 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2641 break;
2642 case 'revisiontimestamp':
2643 # Let the edit saving system know we should parse the page
2644 # *after* a revision ID has been assigned. This is for null edits.
2645 $this->mOutput->setFlag( 'vary-revision' );
2646 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2647 $value = $this->getRevisionTimestamp();
2648 break;
2649 case 'revisionuser':
2650 # Let the edit saving system know we should parse the page
2651 # *after* a revision ID has been assigned for null edits.
2652 $this->mOutput->setFlag( 'vary-user' );
2653 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2654 $value = $this->getRevisionUser();
2655 break;
2656 case 'revisionsize':
2657 $value = $this->getRevisionSize();
2658 break;
2659 case 'namespace':
2660 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2661 break;
2662 case 'namespacee':
2663 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2664 break;
2665 case 'namespacenumber':
2666 $value = $this->mTitle->getNamespace();
2667 break;
2668 case 'talkspace':
2669 $value = $this->mTitle->canTalk()
2670 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2671 : '';
2672 break;
2673 case 'talkspacee':
2674 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2675 break;
2676 case 'subjectspace':
2677 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2678 break;
2679 case 'subjectspacee':
2680 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2681 break;
2682 case 'currentdayname':
2683 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2684 break;
2685 case 'currentyear':
2686 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2687 break;
2688 case 'currenttime':
2689 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2690 break;
2691 case 'currenthour':
2692 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2693 break;
2694 case 'currentweek':
2695 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2696 # int to remove the padding
2697 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2698 break;
2699 case 'currentdow':
2700 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2701 break;
2702 case 'localdayname':
2703 $value = $pageLang->getWeekdayName(
2704 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2705 );
2706 break;
2707 case 'localyear':
2708 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2709 break;
2710 case 'localtime':
2711 $value = $pageLang->time(
2712 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2713 false,
2714 false
2715 );
2716 break;
2717 case 'localhour':
2718 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2719 break;
2720 case 'localweek':
2721 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2722 # int to remove the padding
2723 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2724 break;
2725 case 'localdow':
2726 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2727 break;
2728 case 'numberofarticles':
2729 $value = $pageLang->formatNum( SiteStats::articles() );
2730 break;
2731 case 'numberoffiles':
2732 $value = $pageLang->formatNum( SiteStats::images() );
2733 break;
2734 case 'numberofusers':
2735 $value = $pageLang->formatNum( SiteStats::users() );
2736 break;
2737 case 'numberofactiveusers':
2738 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2739 break;
2740 case 'numberofpages':
2741 $value = $pageLang->formatNum( SiteStats::pages() );
2742 break;
2743 case 'numberofadmins':
2744 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2745 break;
2746 case 'numberofedits':
2747 $value = $pageLang->formatNum( SiteStats::edits() );
2748 break;
2749 case 'currenttimestamp':
2750 $value = wfTimestamp( TS_MW, $ts );
2751 break;
2752 case 'localtimestamp':
2753 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2754 break;
2755 case 'currentversion':
2756 $value = SpecialVersion::getVersion();
2757 break;
2758 case 'articlepath':
2759 return $wgArticlePath;
2760 case 'sitename':
2761 return $wgSitename;
2762 case 'server':
2763 return $wgServer;
2764 case 'servername':
2765 return $wgServerName;
2766 case 'scriptpath':
2767 return $wgScriptPath;
2768 case 'stylepath':
2769 return $wgStylePath;
2770 case 'directionmark':
2771 return $pageLang->getDirMark();
2772 case 'contentlanguage':
2773 global $wgLanguageCode;
2774 return $wgLanguageCode;
2775 case 'cascadingsources':
2776 $value = CoreParserFunctions::cascadingsources( $this );
2777 break;
2778 default:
2779 $ret = null;
2780 Hooks::run(
2781 'ParserGetVariableValueSwitch',
2782 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2783 );
2784
2785 return $ret;
2786 }
2787
2788 if ( $index ) {
2789 $this->mVarCache[$index] = $value;
2790 }
2791
2792 return $value;
2793 }
2794
2795 /**
2796 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2797 *
2798 * @private
2799 */
2800 public function initialiseVariables() {
2801 $variableIDs = MagicWord::getVariableIDs();
2802 $substIDs = MagicWord::getSubstIDs();
2803
2804 $this->mVariables = new MagicWordArray( $variableIDs );
2805 $this->mSubstWords = new MagicWordArray( $substIDs );
2806 }
2807
2808 /**
2809 * Preprocess some wikitext and return the document tree.
2810 * This is the ghost of replace_variables().
2811 *
2812 * @param string $text The text to parse
2813 * @param int $flags Bitwise combination of:
2814 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2815 * included. Default is to assume a direct page view.
2816 *
2817 * The generated DOM tree must depend only on the input text and the flags.
2818 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2819 *
2820 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2821 * change in the DOM tree for a given text, must be passed through the section identifier
2822 * in the section edit link and thus back to extractSections().
2823 *
2824 * The output of this function is currently only cached in process memory, but a persistent
2825 * cache may be implemented at a later date which takes further advantage of these strict
2826 * dependency requirements.
2827 *
2828 * @return PPNode
2829 */
2830 public function preprocessToDom( $text, $flags = 0 ) {
2831 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2832 return $dom;
2833 }
2834
2835 /**
2836 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2837 *
2838 * @param string $s
2839 *
2840 * @return array
2841 */
2842 public static function splitWhitespace( $s ) {
2843 $ltrimmed = ltrim( $s );
2844 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2845 $trimmed = rtrim( $ltrimmed );
2846 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2847 if ( $diff > 0 ) {
2848 $w2 = substr( $ltrimmed, -$diff );
2849 } else {
2850 $w2 = '';
2851 }
2852 return [ $w1, $trimmed, $w2 ];
2853 }
2854
2855 /**
2856 * Replace magic variables, templates, and template arguments
2857 * with the appropriate text. Templates are substituted recursively,
2858 * taking care to avoid infinite loops.
2859 *
2860 * Note that the substitution depends on value of $mOutputType:
2861 * self::OT_WIKI: only {{subst:}} templates
2862 * self::OT_PREPROCESS: templates but not extension tags
2863 * self::OT_HTML: all templates and extension tags
2864 *
2865 * @param string $text The text to transform
2866 * @param bool|PPFrame $frame Object describing the arguments passed to the
2867 * template. Arguments may also be provided as an associative array, as
2868 * was the usual case before MW1.12. Providing arguments this way may be
2869 * useful for extensions wishing to perform variable replacement
2870 * explicitly.
2871 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2872 * double-brace expansion.
2873 * @return string
2874 */
2875 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2876 # Is there any text? Also, Prevent too big inclusions!
2877 $textSize = strlen( $text );
2878 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2879 return $text;
2880 }
2881
2882 if ( $frame === false ) {
2883 $frame = $this->getPreprocessor()->newFrame();
2884 } elseif ( !( $frame instanceof PPFrame ) ) {
2885 wfDebug( __METHOD__ . " called using plain parameters instead of "
2886 . "a PPFrame instance. Creating custom frame.\n" );
2887 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2888 }
2889
2890 $dom = $this->preprocessToDom( $text );
2891 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2892 $text = $frame->expand( $dom, $flags );
2893
2894 return $text;
2895 }
2896
2897 /**
2898 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2899 *
2900 * @param array $args
2901 *
2902 * @return array
2903 */
2904 public static function createAssocArgs( $args ) {
2905 $assocArgs = [];
2906 $index = 1;
2907 foreach ( $args as $arg ) {
2908 $eqpos = strpos( $arg, '=' );
2909 if ( $eqpos === false ) {
2910 $assocArgs[$index++] = $arg;
2911 } else {
2912 $name = trim( substr( $arg, 0, $eqpos ) );
2913 $value = trim( substr( $arg, $eqpos + 1 ) );
2914 if ( $value === false ) {
2915 $value = '';
2916 }
2917 if ( $name !== false ) {
2918 $assocArgs[$name] = $value;
2919 }
2920 }
2921 }
2922
2923 return $assocArgs;
2924 }
2925
2926 /**
2927 * Warn the user when a parser limitation is reached
2928 * Will warn at most once the user per limitation type
2929 *
2930 * The results are shown during preview and run through the Parser (See EditPage.php)
2931 *
2932 * @param string $limitationType Should be one of:
2933 * 'expensive-parserfunction' (corresponding messages:
2934 * 'expensive-parserfunction-warning',
2935 * 'expensive-parserfunction-category')
2936 * 'post-expand-template-argument' (corresponding messages:
2937 * 'post-expand-template-argument-warning',
2938 * 'post-expand-template-argument-category')
2939 * 'post-expand-template-inclusion' (corresponding messages:
2940 * 'post-expand-template-inclusion-warning',
2941 * 'post-expand-template-inclusion-category')
2942 * 'node-count-exceeded' (corresponding messages:
2943 * 'node-count-exceeded-warning',
2944 * 'node-count-exceeded-category')
2945 * 'expansion-depth-exceeded' (corresponding messages:
2946 * 'expansion-depth-exceeded-warning',
2947 * 'expansion-depth-exceeded-category')
2948 * @param string|int|null $current Current value
2949 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2950 * exceeded, provide the values (optional)
2951 */
2952 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2953 # does no harm if $current and $max are present but are unnecessary for the message
2954 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2955 # only during preview, and that would split the parser cache unnecessarily.
2956 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2957 ->text();
2958 $this->mOutput->addWarning( $warning );
2959 $this->addTrackingCategory( "$limitationType-category" );
2960 }
2961
2962 /**
2963 * Return the text of a template, after recursively
2964 * replacing any variables or templates within the template.
2965 *
2966 * @param array $piece The parts of the template
2967 * $piece['title']: the title, i.e. the part before the |
2968 * $piece['parts']: the parameter array
2969 * $piece['lineStart']: whether the brace was at the start of a line
2970 * @param PPFrame $frame The current frame, contains template arguments
2971 * @throws Exception
2972 * @return string The text of the template
2973 */
2974 public function braceSubstitution( $piece, $frame ) {
2975
2976 // Flags
2977
2978 // $text has been filled
2979 $found = false;
2980 // wiki markup in $text should be escaped
2981 $nowiki = false;
2982 // $text is HTML, armour it against wikitext transformation
2983 $isHTML = false;
2984 // Force interwiki transclusion to be done in raw mode not rendered
2985 $forceRawInterwiki = false;
2986 // $text is a DOM node needing expansion in a child frame
2987 $isChildObj = false;
2988 // $text is a DOM node needing expansion in the current frame
2989 $isLocalObj = false;
2990
2991 # Title object, where $text came from
2992 $title = false;
2993
2994 # $part1 is the bit before the first |, and must contain only title characters.
2995 # Various prefixes will be stripped from it later.
2996 $titleWithSpaces = $frame->expand( $piece['title'] );
2997 $part1 = trim( $titleWithSpaces );
2998 $titleText = false;
2999
3000 # Original title text preserved for various purposes
3001 $originalTitle = $part1;
3002
3003 # $args is a list of argument nodes, starting from index 0, not including $part1
3004 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3005 # below won't work b/c this $args isn't an object
3006 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3007
3008 $profileSection = null; // profile templates
3009
3010 # SUBST
3011 if ( !$found ) {
3012 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3013
3014 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3015 # Decide whether to expand template or keep wikitext as-is.
3016 if ( $this->ot['wiki'] ) {
3017 if ( $substMatch === false ) {
3018 $literal = true; # literal when in PST with no prefix
3019 } else {
3020 $literal = false; # expand when in PST with subst: or safesubst:
3021 }
3022 } else {
3023 if ( $substMatch == 'subst' ) {
3024 $literal = true; # literal when not in PST with plain subst:
3025 } else {
3026 $literal = false; # expand when not in PST with safesubst: or no prefix
3027 }
3028 }
3029 if ( $literal ) {
3030 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3031 $isLocalObj = true;
3032 $found = true;
3033 }
3034 }
3035
3036 # Variables
3037 if ( !$found && $args->getLength() == 0 ) {
3038 $id = $this->mVariables->matchStartToEnd( $part1 );
3039 if ( $id !== false ) {
3040 $text = $this->getVariableValue( $id, $frame );
3041 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3042 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3043 }
3044 $found = true;
3045 }
3046 }
3047
3048 # MSG, MSGNW and RAW
3049 if ( !$found ) {
3050 # Check for MSGNW:
3051 $mwMsgnw = MagicWord::get( 'msgnw' );
3052 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3053 $nowiki = true;
3054 } else {
3055 # Remove obsolete MSG:
3056 $mwMsg = MagicWord::get( 'msg' );
3057 $mwMsg->matchStartAndRemove( $part1 );
3058 }
3059
3060 # Check for RAW:
3061 $mwRaw = MagicWord::get( 'raw' );
3062 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3063 $forceRawInterwiki = true;
3064 }
3065 }
3066
3067 # Parser functions
3068 if ( !$found ) {
3069 $colonPos = strpos( $part1, ':' );
3070 if ( $colonPos !== false ) {
3071 $func = substr( $part1, 0, $colonPos );
3072 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3073 $argsLength = $args->getLength();
3074 for ( $i = 0; $i < $argsLength; $i++ ) {
3075 $funcArgs[] = $args->item( $i );
3076 }
3077 try {
3078 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3079 } catch ( Exception $ex ) {
3080 throw $ex;
3081 }
3082
3083 # The interface for parser functions allows for extracting
3084 # flags into the local scope. Extract any forwarded flags
3085 # here.
3086 extract( $result );
3087 }
3088 }
3089
3090 # Finish mangling title and then check for loops.
3091 # Set $title to a Title object and $titleText to the PDBK
3092 if ( !$found ) {
3093 $ns = NS_TEMPLATE;
3094 # Split the title into page and subpage
3095 $subpage = '';
3096 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3097 if ( $part1 !== $relative ) {
3098 $part1 = $relative;
3099 $ns = $this->mTitle->getNamespace();
3100 }
3101 $title = Title::newFromText( $part1, $ns );
3102 if ( $title ) {
3103 $titleText = $title->getPrefixedText();
3104 # Check for language variants if the template is not found
3105 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3106 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3107 }
3108 # Do recursion depth check
3109 $limit = $this->mOptions->getMaxTemplateDepth();
3110 if ( $frame->depth >= $limit ) {
3111 $found = true;
3112 $text = '<span class="error">'
3113 . wfMessage( 'parser-template-recursion-depth-warning' )
3114 ->numParams( $limit )->inContentLanguage()->text()
3115 . '</span>';
3116 }
3117 }
3118 }
3119
3120 # Load from database
3121 if ( !$found && $title ) {
3122 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3123 if ( !$title->isExternal() ) {
3124 if ( $title->isSpecialPage()
3125 && $this->mOptions->getAllowSpecialInclusion()
3126 && $this->ot['html']
3127 ) {
3128 // Pass the template arguments as URL parameters.
3129 // "uselang" will have no effect since the Language object
3130 // is forced to the one defined in ParserOptions.
3131 $pageArgs = [];
3132 $argsLength = $args->getLength();
3133 for ( $i = 0; $i < $argsLength; $i++ ) {
3134 $bits = $args->item( $i )->splitArg();
3135 if ( strval( $bits['index'] ) === '' ) {
3136 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3137 $value = trim( $frame->expand( $bits['value'] ) );
3138 $pageArgs[$name] = $value;
3139 }
3140 }
3141
3142 // Create a new context to execute the special page
3143 $context = new RequestContext;
3144 $context->setTitle( $title );
3145 $context->setRequest( new FauxRequest( $pageArgs ) );
3146 $context->setUser( $this->getUser() );
3147 $context->setLanguage( $this->mOptions->getUserLangObj() );
3148 $ret = SpecialPageFactory::capturePath( $title, $context );
3149 if ( $ret ) {
3150 $text = $context->getOutput()->getHTML();
3151 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3152 $found = true;
3153 $isHTML = true;
3154 // Severely reduce cache time to keep content dynamic
3155 $this->mOutput->updateCacheExpiry( 30 );
3156 }
3157 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3158 $found = false; # access denied
3159 wfDebug( __METHOD__ . ": template inclusion denied for " .
3160 $title->getPrefixedDBkey() . "\n" );
3161 } else {
3162 list( $text, $title ) = $this->getTemplateDom( $title );
3163 if ( $text !== false ) {
3164 $found = true;
3165 $isChildObj = true;
3166 }
3167 }
3168
3169 # If the title is valid but undisplayable, make a link to it
3170 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3171 $text = "[[:$titleText]]";
3172 $found = true;
3173 }
3174 } elseif ( $title->isTrans() ) {
3175 # Interwiki transclusion
3176 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3177 $text = $this->interwikiTransclude( $title, 'render' );
3178 $isHTML = true;
3179 } else {
3180 $text = $this->interwikiTransclude( $title, 'raw' );
3181 # Preprocess it like a template
3182 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3183 $isChildObj = true;
3184 }
3185 $found = true;
3186 }
3187
3188 # Do infinite loop check
3189 # This has to be done after redirect resolution to avoid infinite loops via redirects
3190 if ( !$frame->loopCheck( $title ) ) {
3191 $found = true;
3192 $text = '<span class="error">'
3193 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3194 . '</span>';
3195 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3196 }
3197 }
3198
3199 # If we haven't found text to substitute by now, we're done
3200 # Recover the source wikitext and return it
3201 if ( !$found ) {
3202 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3203 if ( $profileSection ) {
3204 $this->mProfiler->scopedProfileOut( $profileSection );
3205 }
3206 return [ 'object' => $text ];
3207 }
3208
3209 # Expand DOM-style return values in a child frame
3210 if ( $isChildObj ) {
3211 # Clean up argument array
3212 $newFrame = $frame->newChild( $args, $title );
3213
3214 if ( $nowiki ) {
3215 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3216 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3217 # Expansion is eligible for the empty-frame cache
3218 $text = $newFrame->cachedExpand( $titleText, $text );
3219 } else {
3220 # Uncached expansion
3221 $text = $newFrame->expand( $text );
3222 }
3223 }
3224 if ( $isLocalObj && $nowiki ) {
3225 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3226 $isLocalObj = false;
3227 }
3228
3229 if ( $profileSection ) {
3230 $this->mProfiler->scopedProfileOut( $profileSection );
3231 }
3232
3233 # Replace raw HTML by a placeholder
3234 if ( $isHTML ) {
3235 $text = $this->insertStripItem( $text );
3236 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3237 # Escape nowiki-style return values
3238 $text = wfEscapeWikiText( $text );
3239 } elseif ( is_string( $text )
3240 && !$piece['lineStart']
3241 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3242 ) {
3243 # Bug 529: if the template begins with a table or block-level
3244 # element, it should be treated as beginning a new line.
3245 # This behavior is somewhat controversial.
3246 $text = "\n" . $text;
3247 }
3248
3249 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3250 # Error, oversize inclusion
3251 if ( $titleText !== false ) {
3252 # Make a working, properly escaped link if possible (bug 23588)
3253 $text = "[[:$titleText]]";
3254 } else {
3255 # This will probably not be a working link, but at least it may
3256 # provide some hint of where the problem is
3257 preg_replace( '/^:/', '', $originalTitle );
3258 $text = "[[:$originalTitle]]";
3259 }
3260 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3261 . 'post-expand include size too large -->' );
3262 $this->limitationWarn( 'post-expand-template-inclusion' );
3263 }
3264
3265 if ( $isLocalObj ) {
3266 $ret = [ 'object' => $text ];
3267 } else {
3268 $ret = [ 'text' => $text ];
3269 }
3270
3271 return $ret;
3272 }
3273
3274 /**
3275 * Call a parser function and return an array with text and flags.
3276 *
3277 * The returned array will always contain a boolean 'found', indicating
3278 * whether the parser function was found or not. It may also contain the
3279 * following:
3280 * text: string|object, resulting wikitext or PP DOM object
3281 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3282 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3283 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3284 * nowiki: bool, wiki markup in $text should be escaped
3285 *
3286 * @since 1.21
3287 * @param PPFrame $frame The current frame, contains template arguments
3288 * @param string $function Function name
3289 * @param array $args Arguments to the function
3290 * @throws MWException
3291 * @return array
3292 */
3293 public function callParserFunction( $frame, $function, array $args = [] ) {
3294 global $wgContLang;
3295
3296 # Case sensitive functions
3297 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3298 $function = $this->mFunctionSynonyms[1][$function];
3299 } else {
3300 # Case insensitive functions
3301 $function = $wgContLang->lc( $function );
3302 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3303 $function = $this->mFunctionSynonyms[0][$function];
3304 } else {
3305 return [ 'found' => false ];
3306 }
3307 }
3308
3309 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3310
3311 # Workaround for PHP bug 35229 and similar
3312 if ( !is_callable( $callback ) ) {
3313 throw new MWException( "Tag hook for $function is not callable\n" );
3314 }
3315
3316 $allArgs = [ &$this ];
3317 if ( $flags & self::SFH_OBJECT_ARGS ) {
3318 # Convert arguments to PPNodes and collect for appending to $allArgs
3319 $funcArgs = [];
3320 foreach ( $args as $k => $v ) {
3321 if ( $v instanceof PPNode || $k === 0 ) {
3322 $funcArgs[] = $v;
3323 } else {
3324 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3325 }
3326 }
3327
3328 # Add a frame parameter, and pass the arguments as an array
3329 $allArgs[] = $frame;
3330 $allArgs[] = $funcArgs;
3331 } else {
3332 # Convert arguments to plain text and append to $allArgs
3333 foreach ( $args as $k => $v ) {
3334 if ( $v instanceof PPNode ) {
3335 $allArgs[] = trim( $frame->expand( $v ) );
3336 } elseif ( is_int( $k ) && $k >= 0 ) {
3337 $allArgs[] = trim( $v );
3338 } else {
3339 $allArgs[] = trim( "$k=$v" );
3340 }
3341 }
3342 }
3343
3344 $result = call_user_func_array( $callback, $allArgs );
3345
3346 # The interface for function hooks allows them to return a wikitext
3347 # string or an array containing the string and any flags. This mungs
3348 # things around to match what this method should return.
3349 if ( !is_array( $result ) ) {
3350 $result =[
3351 'found' => true,
3352 'text' => $result,
3353 ];
3354 } else {
3355 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3356 $result['text'] = $result[0];
3357 }
3358 unset( $result[0] );
3359 $result += [
3360 'found' => true,
3361 ];
3362 }
3363
3364 $noparse = true;
3365 $preprocessFlags = 0;
3366 if ( isset( $result['noparse'] ) ) {
3367 $noparse = $result['noparse'];
3368 }
3369 if ( isset( $result['preprocessFlags'] ) ) {
3370 $preprocessFlags = $result['preprocessFlags'];
3371 }
3372
3373 if ( !$noparse ) {
3374 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3375 $result['isChildObj'] = true;
3376 }
3377
3378 return $result;
3379 }
3380
3381 /**
3382 * Get the semi-parsed DOM representation of a template with a given title,
3383 * and its redirect destination title. Cached.
3384 *
3385 * @param Title $title
3386 *
3387 * @return array
3388 */
3389 public function getTemplateDom( $title ) {
3390 $cacheTitle = $title;
3391 $titleText = $title->getPrefixedDBkey();
3392
3393 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3394 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3395 $title = Title::makeTitle( $ns, $dbk );
3396 $titleText = $title->getPrefixedDBkey();
3397 }
3398 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3399 return [ $this->mTplDomCache[$titleText], $title ];
3400 }
3401
3402 # Cache miss, go to the database
3403 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3404
3405 if ( $text === false ) {
3406 $this->mTplDomCache[$titleText] = false;
3407 return [ false, $title ];
3408 }
3409
3410 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3411 $this->mTplDomCache[$titleText] = $dom;
3412
3413 if ( !$title->equals( $cacheTitle ) ) {
3414 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3415 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3416 }
3417
3418 return [ $dom, $title ];
3419 }
3420
3421 /**
3422 * Fetch the current revision of a given title. Note that the revision
3423 * (and even the title) may not exist in the database, so everything
3424 * contributing to the output of the parser should use this method
3425 * where possible, rather than getting the revisions themselves. This
3426 * method also caches its results, so using it benefits performance.
3427 *
3428 * @since 1.24
3429 * @param Title $title
3430 * @return Revision
3431 */
3432 public function fetchCurrentRevisionOfTitle( $title ) {
3433 $cacheKey = $title->getPrefixedDBkey();
3434 if ( !$this->currentRevisionCache ) {
3435 $this->currentRevisionCache = new MapCacheLRU( 100 );
3436 }
3437 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3438 $this->currentRevisionCache->set( $cacheKey,
3439 // Defaults to Parser::statelessFetchRevision()
3440 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3441 );
3442 }
3443 return $this->currentRevisionCache->get( $cacheKey );
3444 }
3445
3446 /**
3447 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3448 * without passing them on to it.
3449 *
3450 * @since 1.24
3451 * @param Title $title
3452 * @param Parser|bool $parser
3453 * @return Revision
3454 */
3455 public static function statelessFetchRevision( $title, $parser = false ) {
3456 return Revision::newFromTitle( $title );
3457 }
3458
3459 /**
3460 * Fetch the unparsed text of a template and register a reference to it.
3461 * @param Title $title
3462 * @return array ( string or false, Title )
3463 */
3464 public function fetchTemplateAndTitle( $title ) {
3465 // Defaults to Parser::statelessFetchTemplate()
3466 $templateCb = $this->mOptions->getTemplateCallback();
3467 $stuff = call_user_func( $templateCb, $title, $this );
3468 // We use U+007F DELETE to distinguish strip markers from regular text.
3469 $text = $stuff['text'];
3470 if ( is_string( $stuff['text'] ) ) {
3471 $text = strtr( $text, "\x7f", "?" );
3472 }
3473 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3474 if ( isset( $stuff['deps'] ) ) {
3475 foreach ( $stuff['deps'] as $dep ) {
3476 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3477 if ( $dep['title']->equals( $this->getTitle() ) ) {
3478 // If we transclude ourselves, the final result
3479 // will change based on the new version of the page
3480 $this->mOutput->setFlag( 'vary-revision' );
3481 }
3482 }
3483 }
3484 return [ $text, $finalTitle ];
3485 }
3486
3487 /**
3488 * Fetch the unparsed text of a template and register a reference to it.
3489 * @param Title $title
3490 * @return string|bool
3491 */
3492 public function fetchTemplate( $title ) {
3493 return $this->fetchTemplateAndTitle( $title )[0];
3494 }
3495
3496 /**
3497 * Static function to get a template
3498 * Can be overridden via ParserOptions::setTemplateCallback().
3499 *
3500 * @param Title $title
3501 * @param bool|Parser $parser
3502 *
3503 * @return array
3504 */
3505 public static function statelessFetchTemplate( $title, $parser = false ) {
3506 $text = $skip = false;
3507 $finalTitle = $title;
3508 $deps = [];
3509
3510 # Loop to fetch the article, with up to 1 redirect
3511 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3512 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3513 // @codingStandardsIgnoreEnd
3514 # Give extensions a chance to select the revision instead
3515 $id = false; # Assume current
3516 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3517 [ $parser, $title, &$skip, &$id ] );
3518
3519 if ( $skip ) {
3520 $text = false;
3521 $deps[] = [
3522 'title' => $title,
3523 'page_id' => $title->getArticleID(),
3524 'rev_id' => null
3525 ];
3526 break;
3527 }
3528 # Get the revision
3529 if ( $id ) {
3530 $rev = Revision::newFromId( $id );
3531 } elseif ( $parser ) {
3532 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3533 } else {
3534 $rev = Revision::newFromTitle( $title );
3535 }
3536 $rev_id = $rev ? $rev->getId() : 0;
3537 # If there is no current revision, there is no page
3538 if ( $id === false && !$rev ) {
3539 $linkCache = LinkCache::singleton();
3540 $linkCache->addBadLinkObj( $title );
3541 }
3542
3543 $deps[] = [
3544 'title' => $title,
3545 'page_id' => $title->getArticleID(),
3546 'rev_id' => $rev_id ];
3547 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3548 # We fetched a rev from a different title; register it too...
3549 $deps[] = [
3550 'title' => $rev->getTitle(),
3551 'page_id' => $rev->getPage(),
3552 'rev_id' => $rev_id ];
3553 }
3554
3555 if ( $rev ) {
3556 $content = $rev->getContent();
3557 $text = $content ? $content->getWikitextForTransclusion() : null;
3558
3559 if ( $text === false || $text === null ) {
3560 $text = false;
3561 break;
3562 }
3563 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3564 global $wgContLang;
3565 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3566 if ( !$message->exists() ) {
3567 $text = false;
3568 break;
3569 }
3570 $content = $message->content();
3571 $text = $message->plain();
3572 } else {
3573 break;
3574 }
3575 if ( !$content ) {
3576 break;
3577 }
3578 # Redirect?
3579 $finalTitle = $title;
3580 $title = $content->getRedirectTarget();
3581 }
3582 return [
3583 'text' => $text,
3584 'finalTitle' => $finalTitle,
3585 'deps' => $deps ];
3586 }
3587
3588 /**
3589 * Fetch a file and its title and register a reference to it.
3590 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3591 * @param Title $title
3592 * @param array $options Array of options to RepoGroup::findFile
3593 * @return File|bool
3594 */
3595 public function fetchFile( $title, $options = [] ) {
3596 return $this->fetchFileAndTitle( $title, $options )[0];
3597 }
3598
3599 /**
3600 * Fetch a file and its title and register a reference to it.
3601 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3602 * @param Title $title
3603 * @param array $options Array of options to RepoGroup::findFile
3604 * @return array ( File or false, Title of file )
3605 */
3606 public function fetchFileAndTitle( $title, $options = [] ) {
3607 $file = $this->fetchFileNoRegister( $title, $options );
3608
3609 $time = $file ? $file->getTimestamp() : false;
3610 $sha1 = $file ? $file->getSha1() : false;
3611 # Register the file as a dependency...
3612 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3613 if ( $file && !$title->equals( $file->getTitle() ) ) {
3614 # Update fetched file title
3615 $title = $file->getTitle();
3616 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3617 }
3618 return [ $file, $title ];
3619 }
3620
3621 /**
3622 * Helper function for fetchFileAndTitle.
3623 *
3624 * Also useful if you need to fetch a file but not use it yet,
3625 * for example to get the file's handler.
3626 *
3627 * @param Title $title
3628 * @param array $options Array of options to RepoGroup::findFile
3629 * @return File|bool
3630 */
3631 protected function fetchFileNoRegister( $title, $options = [] ) {
3632 if ( isset( $options['broken'] ) ) {
3633 $file = false; // broken thumbnail forced by hook
3634 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3635 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3636 } else { // get by (name,timestamp)
3637 $file = wfFindFile( $title, $options );
3638 }
3639 return $file;
3640 }
3641
3642 /**
3643 * Transclude an interwiki link.
3644 *
3645 * @param Title $title
3646 * @param string $action
3647 *
3648 * @return string
3649 */
3650 public function interwikiTransclude( $title, $action ) {
3651 global $wgEnableScaryTranscluding;
3652
3653 if ( !$wgEnableScaryTranscluding ) {
3654 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3655 }
3656
3657 $url = $title->getFullURL( [ 'action' => $action ] );
3658
3659 if ( strlen( $url ) > 255 ) {
3660 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3661 }
3662 return $this->fetchScaryTemplateMaybeFromCache( $url );
3663 }
3664
3665 /**
3666 * @param string $url
3667 * @return mixed|string
3668 */
3669 public function fetchScaryTemplateMaybeFromCache( $url ) {
3670 global $wgTranscludeCacheExpiry;
3671 $dbr = wfGetDB( DB_SLAVE );
3672 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3673 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3674 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3675 if ( $obj ) {
3676 return $obj->tc_contents;
3677 }
3678
3679 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3680 $status = $req->execute(); // Status object
3681 if ( $status->isOK() ) {
3682 $text = $req->getContent();
3683 } elseif ( $req->getStatus() != 200 ) {
3684 // Though we failed to fetch the content, this status is useless.
3685 return wfMessage( 'scarytranscludefailed-httpstatus' )
3686 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3687 } else {
3688 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3689 }
3690
3691 $dbw = wfGetDB( DB_MASTER );
3692 $dbw->replace( 'transcache', [ 'tc_url' ], [
3693 'tc_url' => $url,
3694 'tc_time' => $dbw->timestamp( time() ),
3695 'tc_contents' => $text
3696 ] );
3697 return $text;
3698 }
3699
3700 /**
3701 * Triple brace replacement -- used for template arguments
3702 * @private
3703 *
3704 * @param array $piece
3705 * @param PPFrame $frame
3706 *
3707 * @return array
3708 */
3709 public function argSubstitution( $piece, $frame ) {
3710
3711 $error = false;
3712 $parts = $piece['parts'];
3713 $nameWithSpaces = $frame->expand( $piece['title'] );
3714 $argName = trim( $nameWithSpaces );
3715 $object = false;
3716 $text = $frame->getArgument( $argName );
3717 if ( $text === false && $parts->getLength() > 0
3718 && ( $this->ot['html']
3719 || $this->ot['pre']
3720 || ( $this->ot['wiki'] && $frame->isTemplate() )
3721 )
3722 ) {
3723 # No match in frame, use the supplied default
3724 $object = $parts->item( 0 )->getChildren();
3725 }
3726 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3727 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3728 $this->limitationWarn( 'post-expand-template-argument' );
3729 }
3730
3731 if ( $text === false && $object === false ) {
3732 # No match anywhere
3733 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3734 }
3735 if ( $error !== false ) {
3736 $text .= $error;
3737 }
3738 if ( $object !== false ) {
3739 $ret = [ 'object' => $object ];
3740 } else {
3741 $ret = [ 'text' => $text ];
3742 }
3743
3744 return $ret;
3745 }
3746
3747 /**
3748 * Return the text to be used for a given extension tag.
3749 * This is the ghost of strip().
3750 *
3751 * @param array $params Associative array of parameters:
3752 * name PPNode for the tag name
3753 * attr PPNode for unparsed text where tag attributes are thought to be
3754 * attributes Optional associative array of parsed attributes
3755 * inner Contents of extension element
3756 * noClose Original text did not have a close tag
3757 * @param PPFrame $frame
3758 *
3759 * @throws MWException
3760 * @return string
3761 */
3762 public function extensionSubstitution( $params, $frame ) {
3763 $name = $frame->expand( $params['name'] );
3764 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3765 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3766 $marker = self::MARKER_PREFIX . "-$name-"
3767 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3768
3769 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3770 ( $this->ot['html'] || $this->ot['pre'] );
3771 if ( $isFunctionTag ) {
3772 $markerType = 'none';
3773 } else {
3774 $markerType = 'general';
3775 }
3776 if ( $this->ot['html'] || $isFunctionTag ) {
3777 $name = strtolower( $name );
3778 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3779 if ( isset( $params['attributes'] ) ) {
3780 $attributes = $attributes + $params['attributes'];
3781 }
3782
3783 if ( isset( $this->mTagHooks[$name] ) ) {
3784 # Workaround for PHP bug 35229 and similar
3785 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3786 throw new MWException( "Tag hook for $name is not callable\n" );
3787 }
3788 $output = call_user_func_array( $this->mTagHooks[$name],
3789 [ $content, $attributes, $this, $frame ] );
3790 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3791 list( $callback, ) = $this->mFunctionTagHooks[$name];
3792 if ( !is_callable( $callback ) ) {
3793 throw new MWException( "Tag hook for $name is not callable\n" );
3794 }
3795
3796 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3797 } else {
3798 $output = '<span class="error">Invalid tag extension name: ' .
3799 htmlspecialchars( $name ) . '</span>';
3800 }
3801
3802 if ( is_array( $output ) ) {
3803 # Extract flags to local scope (to override $markerType)
3804 $flags = $output;
3805 $output = $flags[0];
3806 unset( $flags[0] );
3807 extract( $flags );
3808 }
3809 } else {
3810 if ( is_null( $attrText ) ) {
3811 $attrText = '';
3812 }
3813 if ( isset( $params['attributes'] ) ) {
3814 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3815 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3816 htmlspecialchars( $attrValue ) . '"';
3817 }
3818 }
3819 if ( $content === null ) {
3820 $output = "<$name$attrText/>";
3821 } else {
3822 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3823 $output = "<$name$attrText>$content$close";
3824 }
3825 }
3826
3827 if ( $markerType === 'none' ) {
3828 return $output;
3829 } elseif ( $markerType === 'nowiki' ) {
3830 $this->mStripState->addNoWiki( $marker, $output );
3831 } elseif ( $markerType === 'general' ) {
3832 $this->mStripState->addGeneral( $marker, $output );
3833 } else {
3834 throw new MWException( __METHOD__ . ': invalid marker type' );
3835 }
3836 return $marker;
3837 }
3838
3839 /**
3840 * Increment an include size counter
3841 *
3842 * @param string $type The type of expansion
3843 * @param int $size The size of the text
3844 * @return bool False if this inclusion would take it over the maximum, true otherwise
3845 */
3846 public function incrementIncludeSize( $type, $size ) {
3847 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3848 return false;
3849 } else {
3850 $this->mIncludeSizes[$type] += $size;
3851 return true;
3852 }
3853 }
3854
3855 /**
3856 * Increment the expensive function count
3857 *
3858 * @return bool False if the limit has been exceeded
3859 */
3860 public function incrementExpensiveFunctionCount() {
3861 $this->mExpensiveFunctionCount++;
3862 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3863 }
3864
3865 /**
3866 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3867 * Fills $this->mDoubleUnderscores, returns the modified text
3868 *
3869 * @param string $text
3870 *
3871 * @return string
3872 */
3873 public function doDoubleUnderscore( $text ) {
3874
3875 # The position of __TOC__ needs to be recorded
3876 $mw = MagicWord::get( 'toc' );
3877 if ( $mw->match( $text ) ) {
3878 $this->mShowToc = true;
3879 $this->mForceTocPosition = true;
3880
3881 # Set a placeholder. At the end we'll fill it in with the TOC.
3882 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3883
3884 # Only keep the first one.
3885 $text = $mw->replace( '', $text );
3886 }
3887
3888 # Now match and remove the rest of them
3889 $mwa = MagicWord::getDoubleUnderscoreArray();
3890 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3891
3892 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3893 $this->mOutput->mNoGallery = true;
3894 }
3895 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3896 $this->mShowToc = false;
3897 }
3898 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3899 && $this->mTitle->getNamespace() == NS_CATEGORY
3900 ) {
3901 $this->addTrackingCategory( 'hidden-category-category' );
3902 }
3903 # (bug 8068) Allow control over whether robots index a page.
3904 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3905 # is not desirable, the last one on the page should win.
3906 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3907 $this->mOutput->setIndexPolicy( 'noindex' );
3908 $this->addTrackingCategory( 'noindex-category' );
3909 }
3910 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3911 $this->mOutput->setIndexPolicy( 'index' );
3912 $this->addTrackingCategory( 'index-category' );
3913 }
3914
3915 # Cache all double underscores in the database
3916 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3917 $this->mOutput->setProperty( $key, '' );
3918 }
3919
3920 return $text;
3921 }
3922
3923 /**
3924 * @see ParserOutput::addTrackingCategory()
3925 * @param string $msg Message key
3926 * @return bool Whether the addition was successful
3927 */
3928 public function addTrackingCategory( $msg ) {
3929 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3930 }
3931
3932 /**
3933 * This function accomplishes several tasks:
3934 * 1) Auto-number headings if that option is enabled
3935 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3936 * 3) Add a Table of contents on the top for users who have enabled the option
3937 * 4) Auto-anchor headings
3938 *
3939 * It loops through all headlines, collects the necessary data, then splits up the
3940 * string and re-inserts the newly formatted headlines.
3941 *
3942 * @param string $text
3943 * @param string $origText Original, untouched wikitext
3944 * @param bool $isMain
3945 * @return mixed|string
3946 * @private
3947 */
3948 public function formatHeadings( $text, $origText, $isMain = true ) {
3949 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
3950
3951 # Inhibit editsection links if requested in the page
3952 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3953 $maybeShowEditLink = $showEditLink = false;
3954 } else {
3955 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3956 $showEditLink = $this->mOptions->getEditSection();
3957 }
3958 if ( $showEditLink ) {
3959 $this->mOutput->setEditSectionTokens( true );
3960 }
3961
3962 # Get all headlines for numbering them and adding funky stuff like [edit]
3963 # links - this is for later, but we need the number of headlines right now
3964 $matches = [];
3965 $numMatches = preg_match_all(
3966 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
3967 $text,
3968 $matches
3969 );
3970
3971 # if there are fewer than 4 headlines in the article, do not show TOC
3972 # unless it's been explicitly enabled.
3973 $enoughToc = $this->mShowToc &&
3974 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3975
3976 # Allow user to stipulate that a page should have a "new section"
3977 # link added via __NEWSECTIONLINK__
3978 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3979 $this->mOutput->setNewSection( true );
3980 }
3981
3982 # Allow user to remove the "new section"
3983 # link via __NONEWSECTIONLINK__
3984 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3985 $this->mOutput->hideNewSection( true );
3986 }
3987
3988 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3989 # override above conditions and always show TOC above first header
3990 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3991 $this->mShowToc = true;
3992 $enoughToc = true;
3993 }
3994
3995 # headline counter
3996 $headlineCount = 0;
3997 $numVisible = 0;
3998
3999 # Ugh .. the TOC should have neat indentation levels which can be
4000 # passed to the skin functions. These are determined here
4001 $toc = '';
4002 $full = '';
4003 $head = [];
4004 $sublevelCount = [];
4005 $levelCount = [];
4006 $level = 0;
4007 $prevlevel = 0;
4008 $toclevel = 0;
4009 $prevtoclevel = 0;
4010 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4011 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4012 $oldType = $this->mOutputType;
4013 $this->setOutputType( self::OT_WIKI );
4014 $frame = $this->getPreprocessor()->newFrame();
4015 $root = $this->preprocessToDom( $origText );
4016 $node = $root->getFirstChild();
4017 $byteOffset = 0;
4018 $tocraw = [];
4019 $refers = [];
4020
4021 $headlines = $numMatches !== false ? $matches[3] : [];
4022
4023 foreach ( $headlines as $headline ) {
4024 $isTemplate = false;
4025 $titleText = false;
4026 $sectionIndex = false;
4027 $numbering = '';
4028 $markerMatches = [];
4029 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4030 $serial = $markerMatches[1];
4031 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4032 $isTemplate = ( $titleText != $baseTitleText );
4033 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4034 }
4035
4036 if ( $toclevel ) {
4037 $prevlevel = $level;
4038 }
4039 $level = $matches[1][$headlineCount];
4040
4041 if ( $level > $prevlevel ) {
4042 # Increase TOC level
4043 $toclevel++;
4044 $sublevelCount[$toclevel] = 0;
4045 if ( $toclevel < $wgMaxTocLevel ) {
4046 $prevtoclevel = $toclevel;
4047 $toc .= Linker::tocIndent();
4048 $numVisible++;
4049 }
4050 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4051 # Decrease TOC level, find level to jump to
4052
4053 for ( $i = $toclevel; $i > 0; $i-- ) {
4054 if ( $levelCount[$i] == $level ) {
4055 # Found last matching level
4056 $toclevel = $i;
4057 break;
4058 } elseif ( $levelCount[$i] < $level ) {
4059 # Found first matching level below current level
4060 $toclevel = $i + 1;
4061 break;
4062 }
4063 }
4064 if ( $i == 0 ) {
4065 $toclevel = 1;
4066 }
4067 if ( $toclevel < $wgMaxTocLevel ) {
4068 if ( $prevtoclevel < $wgMaxTocLevel ) {
4069 # Unindent only if the previous toc level was shown :p
4070 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4071 $prevtoclevel = $toclevel;
4072 } else {
4073 $toc .= Linker::tocLineEnd();
4074 }
4075 }
4076 } else {
4077 # No change in level, end TOC line
4078 if ( $toclevel < $wgMaxTocLevel ) {
4079 $toc .= Linker::tocLineEnd();
4080 }
4081 }
4082
4083 $levelCount[$toclevel] = $level;
4084
4085 # count number of headlines for each level
4086 $sublevelCount[$toclevel]++;
4087 $dot = 0;
4088 for ( $i = 1; $i <= $toclevel; $i++ ) {
4089 if ( !empty( $sublevelCount[$i] ) ) {
4090 if ( $dot ) {
4091 $numbering .= '.';
4092 }
4093 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4094 $dot = 1;
4095 }
4096 }
4097
4098 # The safe header is a version of the header text safe to use for links
4099
4100 # Remove link placeholders by the link text.
4101 # <!--LINK number-->
4102 # turns into
4103 # link text with suffix
4104 # Do this before unstrip since link text can contain strip markers
4105 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4106
4107 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4108 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4109
4110 # Strip out HTML (first regex removes any tag not allowed)
4111 # Allowed tags are:
4112 # * <sup> and <sub> (bug 8393)
4113 # * <i> (bug 26375)
4114 # * <b> (r105284)
4115 # * <bdi> (bug 72884)
4116 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4117 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4118 # to allow setting directionality in toc items.
4119 $tocline = preg_replace(
4120 [
4121 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4122 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4123 ],
4124 [ '', '<$1>' ],
4125 $safeHeadline
4126 );
4127
4128 # Strip '<span></span>', which is the result from the above if
4129 # <span id="foo"></span> is used to produce an additional anchor
4130 # for a section.
4131 $tocline = str_replace( '<span></span>', '', $tocline );
4132
4133 $tocline = trim( $tocline );
4134
4135 # For the anchor, strip out HTML-y stuff period
4136 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4137 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4138
4139 # Save headline for section edit hint before it's escaped
4140 $headlineHint = $safeHeadline;
4141
4142 if ( $wgExperimentalHtmlIds ) {
4143 # For reverse compatibility, provide an id that's
4144 # HTML4-compatible, like we used to.
4145 # It may be worth noting, academically, that it's possible for
4146 # the legacy anchor to conflict with a non-legacy headline
4147 # anchor on the page. In this case likely the "correct" thing
4148 # would be to either drop the legacy anchors or make sure
4149 # they're numbered first. However, this would require people
4150 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4151 # manually, so let's not bother worrying about it.
4152 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4153 [ 'noninitial', 'legacy' ] );
4154 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4155
4156 if ( $legacyHeadline == $safeHeadline ) {
4157 # No reason to have both (in fact, we can't)
4158 $legacyHeadline = false;
4159 }
4160 } else {
4161 $legacyHeadline = false;
4162 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4163 'noninitial' );
4164 }
4165
4166 # HTML names must be case-insensitively unique (bug 10721).
4167 # This does not apply to Unicode characters per
4168 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4169 # @todo FIXME: We may be changing them depending on the current locale.
4170 $arrayKey = strtolower( $safeHeadline );
4171 if ( $legacyHeadline === false ) {
4172 $legacyArrayKey = false;
4173 } else {
4174 $legacyArrayKey = strtolower( $legacyHeadline );
4175 }
4176
4177 # Create the anchor for linking from the TOC to the section
4178 $anchor = $safeHeadline;
4179 $legacyAnchor = $legacyHeadline;
4180 if ( isset( $refers[$arrayKey] ) ) {
4181 // @codingStandardsIgnoreStart
4182 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4183 // @codingStandardsIgnoreEnd
4184 $anchor .= "_$i";
4185 $refers["${arrayKey}_$i"] = true;
4186 } else {
4187 $refers[$arrayKey] = true;
4188 }
4189 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4190 // @codingStandardsIgnoreStart
4191 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4192 // @codingStandardsIgnoreEnd
4193 $legacyAnchor .= "_$i";
4194 $refers["${legacyArrayKey}_$i"] = true;
4195 } else {
4196 $refers[$legacyArrayKey] = true;
4197 }
4198
4199 # Don't number the heading if it is the only one (looks silly)
4200 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4201 # the two are different if the line contains a link
4202 $headline = Html::element(
4203 'span',
4204 [ 'class' => 'mw-headline-number' ],
4205 $numbering
4206 ) . ' ' . $headline;
4207 }
4208
4209 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4210 $toc .= Linker::tocLine( $anchor, $tocline,
4211 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4212 }
4213
4214 # Add the section to the section tree
4215 # Find the DOM node for this header
4216 $noOffset = ( $isTemplate || $sectionIndex === false );
4217 while ( $node && !$noOffset ) {
4218 if ( $node->getName() === 'h' ) {
4219 $bits = $node->splitHeading();
4220 if ( $bits['i'] == $sectionIndex ) {
4221 break;
4222 }
4223 }
4224 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4225 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4226 $node = $node->getNextSibling();
4227 }
4228 $tocraw[] = [
4229 'toclevel' => $toclevel,
4230 'level' => $level,
4231 'line' => $tocline,
4232 'number' => $numbering,
4233 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4234 'fromtitle' => $titleText,
4235 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4236 'anchor' => $anchor,
4237 ];
4238
4239 # give headline the correct <h#> tag
4240 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4241 // Output edit section links as markers with styles that can be customized by skins
4242 if ( $isTemplate ) {
4243 # Put a T flag in the section identifier, to indicate to extractSections()
4244 # that sections inside <includeonly> should be counted.
4245 $editsectionPage = $titleText;
4246 $editsectionSection = "T-$sectionIndex";
4247 $editsectionContent = null;
4248 } else {
4249 $editsectionPage = $this->mTitle->getPrefixedText();
4250 $editsectionSection = $sectionIndex;
4251 $editsectionContent = $headlineHint;
4252 }
4253 // We use a bit of pesudo-xml for editsection markers. The
4254 // language converter is run later on. Using a UNIQ style marker
4255 // leads to the converter screwing up the tokens when it
4256 // converts stuff. And trying to insert strip tags fails too. At
4257 // this point all real inputted tags have already been escaped,
4258 // so we don't have to worry about a user trying to input one of
4259 // these markers directly. We use a page and section attribute
4260 // to stop the language converter from converting these
4261 // important bits of data, but put the headline hint inside a
4262 // content block because the language converter is supposed to
4263 // be able to convert that piece of data.
4264 // Gets replaced with html in ParserOutput::getText
4265 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4266 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4267 if ( $editsectionContent !== null ) {
4268 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4269 } else {
4270 $editlink .= '/>';
4271 }
4272 } else {
4273 $editlink = '';
4274 }
4275 $head[$headlineCount] = Linker::makeHeadline( $level,
4276 $matches['attrib'][$headlineCount], $anchor, $headline,
4277 $editlink, $legacyAnchor );
4278
4279 $headlineCount++;
4280 }
4281
4282 $this->setOutputType( $oldType );
4283
4284 # Never ever show TOC if no headers
4285 if ( $numVisible < 1 ) {
4286 $enoughToc = false;
4287 }
4288
4289 if ( $enoughToc ) {
4290 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4291 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4292 }
4293 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4294 $this->mOutput->setTOCHTML( $toc );
4295 $toc = self::TOC_START . $toc . self::TOC_END;
4296 $this->mOutput->addModules( 'mediawiki.toc' );
4297 }
4298
4299 if ( $isMain ) {
4300 $this->mOutput->setSections( $tocraw );
4301 }
4302
4303 # split up and insert constructed headlines
4304 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4305 $i = 0;
4306
4307 // build an array of document sections
4308 $sections = [];
4309 foreach ( $blocks as $block ) {
4310 // $head is zero-based, sections aren't.
4311 if ( empty( $head[$i - 1] ) ) {
4312 $sections[$i] = $block;
4313 } else {
4314 $sections[$i] = $head[$i - 1] . $block;
4315 }
4316
4317 /**
4318 * Send a hook, one per section.
4319 * The idea here is to be able to make section-level DIVs, but to do so in a
4320 * lower-impact, more correct way than r50769
4321 *
4322 * $this : caller
4323 * $section : the section number
4324 * &$sectionContent : ref to the content of the section
4325 * $showEditLinks : boolean describing whether this section has an edit link
4326 */
4327 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4328
4329 $i++;
4330 }
4331
4332 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4333 // append the TOC at the beginning
4334 // Top anchor now in skin
4335 $sections[0] = $sections[0] . $toc . "\n";
4336 }
4337
4338 $full .= implode( '', $sections );
4339
4340 if ( $this->mForceTocPosition ) {
4341 return str_replace( '<!--MWTOC-->', $toc, $full );
4342 } else {
4343 return $full;
4344 }
4345 }
4346
4347 /**
4348 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4349 * conversion, substituting signatures, {{subst:}} templates, etc.
4350 *
4351 * @param string $text The text to transform
4352 * @param Title $title The Title object for the current article
4353 * @param User $user The User object describing the current user
4354 * @param ParserOptions $options Parsing options
4355 * @param bool $clearState Whether to clear the parser state first
4356 * @return string The altered wiki markup
4357 */
4358 public function preSaveTransform( $text, Title $title, User $user,
4359 ParserOptions $options, $clearState = true
4360 ) {
4361 if ( $clearState ) {
4362 $magicScopeVariable = $this->lock();
4363 }
4364 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4365 $this->setUser( $user );
4366
4367 $pairs = [
4368 "\r\n" => "\n",
4369 "\r" => "\n",
4370 ];
4371 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4372 if ( $options->getPreSaveTransform() ) {
4373 $text = $this->pstPass2( $text, $user );
4374 }
4375 $text = $this->mStripState->unstripBoth( $text );
4376
4377 $this->setUser( null ); # Reset
4378
4379 return $text;
4380 }
4381
4382 /**
4383 * Pre-save transform helper function
4384 *
4385 * @param string $text
4386 * @param User $user
4387 *
4388 * @return string
4389 */
4390 private function pstPass2( $text, $user ) {
4391 global $wgContLang;
4392
4393 # Note: This is the timestamp saved as hardcoded wikitext to
4394 # the database, we use $wgContLang here in order to give
4395 # everyone the same signature and use the default one rather
4396 # than the one selected in each user's preferences.
4397 # (see also bug 12815)
4398 $ts = $this->mOptions->getTimestamp();
4399 $timestamp = MWTimestamp::getLocalInstance( $ts );
4400 $ts = $timestamp->format( 'YmdHis' );
4401 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4402
4403 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4404
4405 # Variable replacement
4406 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4407 $text = $this->replaceVariables( $text );
4408
4409 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4410 # which may corrupt this parser instance via its wfMessage()->text() call-
4411
4412 # Signatures
4413 $sigText = $this->getUserSig( $user );
4414 $text = strtr( $text, [
4415 '~~~~~' => $d,
4416 '~~~~' => "$sigText $d",
4417 '~~~' => $sigText
4418 ] );
4419
4420 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4421 $tc = '[' . Title::legalChars() . ']';
4422 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4423
4424 // [[ns:page (context)|]]
4425 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4426 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4427 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4428 // [[ns:page (context), context|]] (using either single or double-width comma)
4429 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4430 // [[|page]] (reverse pipe trick: add context from page title)
4431 $p2 = "/\[\[\\|($tc+)]]/";
4432
4433 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4434 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4435 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4436 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4437
4438 $t = $this->mTitle->getText();
4439 $m = [];
4440 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4441 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4442 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4443 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4444 } else {
4445 # if there's no context, don't bother duplicating the title
4446 $text = preg_replace( $p2, '[[\\1]]', $text );
4447 }
4448
4449 # Trim trailing whitespace
4450 $text = rtrim( $text );
4451
4452 return $text;
4453 }
4454
4455 /**
4456 * Fetch the user's signature text, if any, and normalize to
4457 * validated, ready-to-insert wikitext.
4458 * If you have pre-fetched the nickname or the fancySig option, you can
4459 * specify them here to save a database query.
4460 * Do not reuse this parser instance after calling getUserSig(),
4461 * as it may have changed if it's the $wgParser.
4462 *
4463 * @param User $user
4464 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4465 * @param bool|null $fancySig whether the nicknname is the complete signature
4466 * or null to use default value
4467 * @return string
4468 */
4469 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4470 global $wgMaxSigChars;
4471
4472 $username = $user->getName();
4473
4474 # If not given, retrieve from the user object.
4475 if ( $nickname === false ) {
4476 $nickname = $user->getOption( 'nickname' );
4477 }
4478
4479 if ( is_null( $fancySig ) ) {
4480 $fancySig = $user->getBoolOption( 'fancysig' );
4481 }
4482
4483 $nickname = $nickname == null ? $username : $nickname;
4484
4485 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4486 $nickname = $username;
4487 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4488 } elseif ( $fancySig !== false ) {
4489 # Sig. might contain markup; validate this
4490 if ( $this->validateSig( $nickname ) !== false ) {
4491 # Validated; clean up (if needed) and return it
4492 return $this->cleanSig( $nickname, true );
4493 } else {
4494 # Failed to validate; fall back to the default
4495 $nickname = $username;
4496 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4497 }
4498 }
4499
4500 # Make sure nickname doesnt get a sig in a sig
4501 $nickname = self::cleanSigInSig( $nickname );
4502
4503 # If we're still here, make it a link to the user page
4504 $userText = wfEscapeWikiText( $username );
4505 $nickText = wfEscapeWikiText( $nickname );
4506 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4507
4508 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4509 ->title( $this->getTitle() )->text();
4510 }
4511
4512 /**
4513 * Check that the user's signature contains no bad XML
4514 *
4515 * @param string $text
4516 * @return string|bool An expanded string, or false if invalid.
4517 */
4518 public function validateSig( $text ) {
4519 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4520 }
4521
4522 /**
4523 * Clean up signature text
4524 *
4525 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4526 * 2) Substitute all transclusions
4527 *
4528 * @param string $text
4529 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4530 * @return string Signature text
4531 */
4532 public function cleanSig( $text, $parsing = false ) {
4533 if ( !$parsing ) {
4534 global $wgTitle;
4535 $magicScopeVariable = $this->lock();
4536 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4537 }
4538
4539 # Option to disable this feature
4540 if ( !$this->mOptions->getCleanSignatures() ) {
4541 return $text;
4542 }
4543
4544 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4545 # => Move this logic to braceSubstitution()
4546 $substWord = MagicWord::get( 'subst' );
4547 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4548 $substText = '{{' . $substWord->getSynonym( 0 );
4549
4550 $text = preg_replace( $substRegex, $substText, $text );
4551 $text = self::cleanSigInSig( $text );
4552 $dom = $this->preprocessToDom( $text );
4553 $frame = $this->getPreprocessor()->newFrame();
4554 $text = $frame->expand( $dom );
4555
4556 if ( !$parsing ) {
4557 $text = $this->mStripState->unstripBoth( $text );
4558 }
4559
4560 return $text;
4561 }
4562
4563 /**
4564 * Strip 3, 4 or 5 tildes out of signatures.
4565 *
4566 * @param string $text
4567 * @return string Signature text with /~{3,5}/ removed
4568 */
4569 public static function cleanSigInSig( $text ) {
4570 $text = preg_replace( '/~{3,5}/', '', $text );
4571 return $text;
4572 }
4573
4574 /**
4575 * Set up some variables which are usually set up in parse()
4576 * so that an external function can call some class members with confidence
4577 *
4578 * @param Title|null $title
4579 * @param ParserOptions $options
4580 * @param int $outputType
4581 * @param bool $clearState
4582 */
4583 public function startExternalParse( Title $title = null, ParserOptions $options,
4584 $outputType, $clearState = true
4585 ) {
4586 $this->startParse( $title, $options, $outputType, $clearState );
4587 }
4588
4589 /**
4590 * @param Title|null $title
4591 * @param ParserOptions $options
4592 * @param int $outputType
4593 * @param bool $clearState
4594 */
4595 private function startParse( Title $title = null, ParserOptions $options,
4596 $outputType, $clearState = true
4597 ) {
4598 $this->setTitle( $title );
4599 $this->mOptions = $options;
4600 $this->setOutputType( $outputType );
4601 if ( $clearState ) {
4602 $this->clearState();
4603 }
4604 }
4605
4606 /**
4607 * Wrapper for preprocess()
4608 *
4609 * @param string $text The text to preprocess
4610 * @param ParserOptions $options Options
4611 * @param Title|null $title Title object or null to use $wgTitle
4612 * @return string
4613 */
4614 public function transformMsg( $text, $options, $title = null ) {
4615 static $executing = false;
4616
4617 # Guard against infinite recursion
4618 if ( $executing ) {
4619 return $text;
4620 }
4621 $executing = true;
4622
4623 if ( !$title ) {
4624 global $wgTitle;
4625 $title = $wgTitle;
4626 }
4627
4628 $text = $this->preprocess( $text, $title, $options );
4629
4630 $executing = false;
4631 return $text;
4632 }
4633
4634 /**
4635 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4636 * The callback should have the following form:
4637 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4638 *
4639 * Transform and return $text. Use $parser for any required context, e.g. use
4640 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4641 *
4642 * Hooks may return extended information by returning an array, of which the
4643 * first numbered element (index 0) must be the return string, and all other
4644 * entries are extracted into local variables within an internal function
4645 * in the Parser class.
4646 *
4647 * This interface (introduced r61913) appears to be undocumented, but
4648 * 'markerType' is used by some core tag hooks to override which strip
4649 * array their results are placed in. **Use great caution if attempting
4650 * this interface, as it is not documented and injudicious use could smash
4651 * private variables.**
4652 *
4653 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4654 * @param callable $callback The callback function (and object) to use for the tag
4655 * @throws MWException
4656 * @return callable|null The old value of the mTagHooks array associated with the hook
4657 */
4658 public function setHook( $tag, $callback ) {
4659 $tag = strtolower( $tag );
4660 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4661 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4662 }
4663 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4664 $this->mTagHooks[$tag] = $callback;
4665 if ( !in_array( $tag, $this->mStripList ) ) {
4666 $this->mStripList[] = $tag;
4667 }
4668
4669 return $oldVal;
4670 }
4671
4672 /**
4673 * As setHook(), but letting the contents be parsed.
4674 *
4675 * Transparent tag hooks are like regular XML-style tag hooks, except they
4676 * operate late in the transformation sequence, on HTML instead of wikitext.
4677 *
4678 * This is probably obsoleted by things dealing with parser frames?
4679 * The only extension currently using it is geoserver.
4680 *
4681 * @since 1.10
4682 * @todo better document or deprecate this
4683 *
4684 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4685 * @param callable $callback The callback function (and object) to use for the tag
4686 * @throws MWException
4687 * @return callable|null The old value of the mTagHooks array associated with the hook
4688 */
4689 public function setTransparentTagHook( $tag, $callback ) {
4690 $tag = strtolower( $tag );
4691 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4692 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4693 }
4694 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4695 $this->mTransparentTagHooks[$tag] = $callback;
4696
4697 return $oldVal;
4698 }
4699
4700 /**
4701 * Remove all tag hooks
4702 */
4703 public function clearTagHooks() {
4704 $this->mTagHooks = [];
4705 $this->mFunctionTagHooks = [];
4706 $this->mStripList = $this->mDefaultStripList;
4707 }
4708
4709 /**
4710 * Create a function, e.g. {{sum:1|2|3}}
4711 * The callback function should have the form:
4712 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4713 *
4714 * Or with Parser::SFH_OBJECT_ARGS:
4715 * function myParserFunction( $parser, $frame, $args ) { ... }
4716 *
4717 * The callback may either return the text result of the function, or an array with the text
4718 * in element 0, and a number of flags in the other elements. The names of the flags are
4719 * specified in the keys. Valid flags are:
4720 * found The text returned is valid, stop processing the template. This
4721 * is on by default.
4722 * nowiki Wiki markup in the return value should be escaped
4723 * isHTML The returned text is HTML, armour it against wikitext transformation
4724 *
4725 * @param string $id The magic word ID
4726 * @param callable $callback The callback function (and object) to use
4727 * @param int $flags A combination of the following flags:
4728 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4729 *
4730 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4731 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4732 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4733 * the arguments, and to control the way they are expanded.
4734 *
4735 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4736 * arguments, for instance:
4737 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4738 *
4739 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4740 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4741 * working if/when this is changed.
4742 *
4743 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4744 * expansion.
4745 *
4746 * Please read the documentation in includes/parser/Preprocessor.php for more information
4747 * about the methods available in PPFrame and PPNode.
4748 *
4749 * @throws MWException
4750 * @return string|callable The old callback function for this name, if any
4751 */
4752 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4753 global $wgContLang;
4754
4755 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4756 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4757
4758 # Add to function cache
4759 $mw = MagicWord::get( $id );
4760 if ( !$mw ) {
4761 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4762 }
4763
4764 $synonyms = $mw->getSynonyms();
4765 $sensitive = intval( $mw->isCaseSensitive() );
4766
4767 foreach ( $synonyms as $syn ) {
4768 # Case
4769 if ( !$sensitive ) {
4770 $syn = $wgContLang->lc( $syn );
4771 }
4772 # Add leading hash
4773 if ( !( $flags & self::SFH_NO_HASH ) ) {
4774 $syn = '#' . $syn;
4775 }
4776 # Remove trailing colon
4777 if ( substr( $syn, -1, 1 ) === ':' ) {
4778 $syn = substr( $syn, 0, -1 );
4779 }
4780 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4781 }
4782 return $oldVal;
4783 }
4784
4785 /**
4786 * Get all registered function hook identifiers
4787 *
4788 * @return array
4789 */
4790 public function getFunctionHooks() {
4791 return array_keys( $this->mFunctionHooks );
4792 }
4793
4794 /**
4795 * Create a tag function, e.g. "<test>some stuff</test>".
4796 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4797 * Unlike parser functions, their content is not preprocessed.
4798 * @param string $tag
4799 * @param callable $callback
4800 * @param int $flags
4801 * @throws MWException
4802 * @return null
4803 */
4804 public function setFunctionTagHook( $tag, $callback, $flags ) {
4805 $tag = strtolower( $tag );
4806 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4807 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4808 }
4809 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4810 $this->mFunctionTagHooks[$tag] : null;
4811 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4812
4813 if ( !in_array( $tag, $this->mStripList ) ) {
4814 $this->mStripList[] = $tag;
4815 }
4816
4817 return $old;
4818 }
4819
4820 /**
4821 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4822 * Placeholders created in Linker::link()
4823 *
4824 * @param string $text
4825 * @param int $options
4826 */
4827 public function replaceLinkHolders( &$text, $options = 0 ) {
4828 $this->mLinkHolders->replace( $text );
4829 }
4830
4831 /**
4832 * Replace "<!--LINK-->" link placeholders with plain text of links
4833 * (not HTML-formatted).
4834 *
4835 * @param string $text
4836 * @return string
4837 */
4838 public function replaceLinkHoldersText( $text ) {
4839 return $this->mLinkHolders->replaceText( $text );
4840 }
4841
4842 /**
4843 * Renders an image gallery from a text with one line per image.
4844 * text labels may be given by using |-style alternative text. E.g.
4845 * Image:one.jpg|The number "1"
4846 * Image:tree.jpg|A tree
4847 * given as text will return the HTML of a gallery with two images,
4848 * labeled 'The number "1"' and
4849 * 'A tree'.
4850 *
4851 * @param string $text
4852 * @param array $params
4853 * @return string HTML
4854 */
4855 public function renderImageGallery( $text, $params ) {
4856
4857 $mode = false;
4858 if ( isset( $params['mode'] ) ) {
4859 $mode = $params['mode'];
4860 }
4861
4862 try {
4863 $ig = ImageGalleryBase::factory( $mode );
4864 } catch ( Exception $e ) {
4865 // If invalid type set, fallback to default.
4866 $ig = ImageGalleryBase::factory( false );
4867 }
4868
4869 $ig->setContextTitle( $this->mTitle );
4870 $ig->setShowBytes( false );
4871 $ig->setShowFilename( false );
4872 $ig->setParser( $this );
4873 $ig->setHideBadImages();
4874 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4875
4876 if ( isset( $params['showfilename'] ) ) {
4877 $ig->setShowFilename( true );
4878 } else {
4879 $ig->setShowFilename( false );
4880 }
4881 if ( isset( $params['caption'] ) ) {
4882 $caption = $params['caption'];
4883 $caption = htmlspecialchars( $caption );
4884 $caption = $this->replaceInternalLinks( $caption );
4885 $ig->setCaptionHtml( $caption );
4886 }
4887 if ( isset( $params['perrow'] ) ) {
4888 $ig->setPerRow( $params['perrow'] );
4889 }
4890 if ( isset( $params['widths'] ) ) {
4891 $ig->setWidths( $params['widths'] );
4892 }
4893 if ( isset( $params['heights'] ) ) {
4894 $ig->setHeights( $params['heights'] );
4895 }
4896 $ig->setAdditionalOptions( $params );
4897
4898 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4899
4900 $lines = StringUtils::explode( "\n", $text );
4901 foreach ( $lines as $line ) {
4902 # match lines like these:
4903 # Image:someimage.jpg|This is some image
4904 $matches = [];
4905 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4906 # Skip empty lines
4907 if ( count( $matches ) == 0 ) {
4908 continue;
4909 }
4910
4911 if ( strpos( $matches[0], '%' ) !== false ) {
4912 $matches[1] = rawurldecode( $matches[1] );
4913 }
4914 $title = Title::newFromText( $matches[1], NS_FILE );
4915 if ( is_null( $title ) ) {
4916 # Bogus title. Ignore these so we don't bomb out later.
4917 continue;
4918 }
4919
4920 # We need to get what handler the file uses, to figure out parameters.
4921 # Note, a hook can overide the file name, and chose an entirely different
4922 # file (which potentially could be of a different type and have different handler).
4923 $options = [];
4924 $descQuery = false;
4925 Hooks::run( 'BeforeParserFetchFileAndTitle',
4926 [ $this, $title, &$options, &$descQuery ] );
4927 # Don't register it now, as ImageGallery does that later.
4928 $file = $this->fetchFileNoRegister( $title, $options );
4929 $handler = $file ? $file->getHandler() : false;
4930
4931 $paramMap = [
4932 'img_alt' => 'gallery-internal-alt',
4933 'img_link' => 'gallery-internal-link',
4934 ];
4935 if ( $handler ) {
4936 $paramMap = $paramMap + $handler->getParamMap();
4937 // We don't want people to specify per-image widths.
4938 // Additionally the width parameter would need special casing anyhow.
4939 unset( $paramMap['img_width'] );
4940 }
4941
4942 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4943
4944 $label = '';
4945 $alt = '';
4946 $link = '';
4947 $handlerOptions = [];
4948 if ( isset( $matches[3] ) ) {
4949 // look for an |alt= definition while trying not to break existing
4950 // captions with multiple pipes (|) in it, until a more sensible grammar
4951 // is defined for images in galleries
4952
4953 // FIXME: Doing recursiveTagParse at this stage, and the trim before
4954 // splitting on '|' is a bit odd, and different from makeImage.
4955 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4956 $parameterMatches = StringUtils::explode( '|', $matches[3] );
4957
4958 foreach ( $parameterMatches as $parameterMatch ) {
4959 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
4960 if ( $magicName ) {
4961 $paramName = $paramMap[$magicName];
4962
4963 switch ( $paramName ) {
4964 case 'gallery-internal-alt':
4965 $alt = $this->stripAltText( $match, false );
4966 break;
4967 case 'gallery-internal-link':
4968 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
4969 $chars = self::EXT_LINK_URL_CLASS;
4970 $addr = self::EXT_LINK_ADDR;
4971 $prots = $this->mUrlProtocols;
4972 // check to see if link matches an absolute url, if not then it must be a wiki link.
4973 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
4974 $link = $linkValue;
4975 } else {
4976 $localLinkTitle = Title::newFromText( $linkValue );
4977 if ( $localLinkTitle !== null ) {
4978 $link = $localLinkTitle->getLinkURL();
4979 }
4980 }
4981 break;
4982 default:
4983 // Must be a handler specific parameter.
4984 if ( $handler->validateParam( $paramName, $match ) ) {
4985 $handlerOptions[$paramName] = $match;
4986 } else {
4987 // Guess not, consider it as caption.
4988 wfDebug( "$parameterMatch failed parameter validation\n" );
4989 $label = '|' . $parameterMatch;
4990 }
4991 }
4992
4993 } else {
4994 // Last pipe wins.
4995 $label = '|' . $parameterMatch;
4996 }
4997 }
4998 // Remove the pipe.
4999 $label = substr( $label, 1 );
5000 }
5001
5002 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5003 }
5004 $html = $ig->toHTML();
5005 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5006 return $html;
5007 }
5008
5009 /**
5010 * @param MediaHandler $handler
5011 * @return array
5012 */
5013 public function getImageParams( $handler ) {
5014 if ( $handler ) {
5015 $handlerClass = get_class( $handler );
5016 } else {
5017 $handlerClass = '';
5018 }
5019 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5020 # Initialise static lists
5021 static $internalParamNames = [
5022 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5023 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5024 'bottom', 'text-bottom' ],
5025 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5026 'upright', 'border', 'link', 'alt', 'class' ],
5027 ];
5028 static $internalParamMap;
5029 if ( !$internalParamMap ) {
5030 $internalParamMap = [];
5031 foreach ( $internalParamNames as $type => $names ) {
5032 foreach ( $names as $name ) {
5033 $magicName = str_replace( '-', '_', "img_$name" );
5034 $internalParamMap[$magicName] = [ $type, $name ];
5035 }
5036 }
5037 }
5038
5039 # Add handler params
5040 $paramMap = $internalParamMap;
5041 if ( $handler ) {
5042 $handlerParamMap = $handler->getParamMap();
5043 foreach ( $handlerParamMap as $magic => $paramName ) {
5044 $paramMap[$magic] = [ 'handler', $paramName ];
5045 }
5046 }
5047 $this->mImageParams[$handlerClass] = $paramMap;
5048 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5049 }
5050 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5051 }
5052
5053 /**
5054 * Parse image options text and use it to make an image
5055 *
5056 * @param Title $title
5057 * @param string $options
5058 * @param LinkHolderArray|bool $holders
5059 * @return string HTML
5060 */
5061 public function makeImage( $title, $options, $holders = false ) {
5062 # Check if the options text is of the form "options|alt text"
5063 # Options are:
5064 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5065 # * left no resizing, just left align. label is used for alt= only
5066 # * right same, but right aligned
5067 # * none same, but not aligned
5068 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5069 # * center center the image
5070 # * frame Keep original image size, no magnify-button.
5071 # * framed Same as "frame"
5072 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5073 # * upright reduce width for upright images, rounded to full __0 px
5074 # * border draw a 1px border around the image
5075 # * alt Text for HTML alt attribute (defaults to empty)
5076 # * class Set a class for img node
5077 # * link Set the target of the image link. Can be external, interwiki, or local
5078 # vertical-align values (no % or length right now):
5079 # * baseline
5080 # * sub
5081 # * super
5082 # * top
5083 # * text-top
5084 # * middle
5085 # * bottom
5086 # * text-bottom
5087
5088 $parts = StringUtils::explode( "|", $options );
5089
5090 # Give extensions a chance to select the file revision for us
5091 $options = [];
5092 $descQuery = false;
5093 Hooks::run( 'BeforeParserFetchFileAndTitle',
5094 [ $this, $title, &$options, &$descQuery ] );
5095 # Fetch and register the file (file title may be different via hooks)
5096 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5097
5098 # Get parameter map
5099 $handler = $file ? $file->getHandler() : false;
5100
5101 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5102
5103 if ( !$file ) {
5104 $this->addTrackingCategory( 'broken-file-category' );
5105 }
5106
5107 # Process the input parameters
5108 $caption = '';
5109 $params = [ 'frame' => [], 'handler' => [],
5110 'horizAlign' => [], 'vertAlign' => [] ];
5111 $seenformat = false;
5112 foreach ( $parts as $part ) {
5113 $part = trim( $part );
5114 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5115 $validated = false;
5116 if ( isset( $paramMap[$magicName] ) ) {
5117 list( $type, $paramName ) = $paramMap[$magicName];
5118
5119 # Special case; width and height come in one variable together
5120 if ( $type === 'handler' && $paramName === 'width' ) {
5121 $parsedWidthParam = $this->parseWidthParam( $value );
5122 if ( isset( $parsedWidthParam['width'] ) ) {
5123 $width = $parsedWidthParam['width'];
5124 if ( $handler->validateParam( 'width', $width ) ) {
5125 $params[$type]['width'] = $width;
5126 $validated = true;
5127 }
5128 }
5129 if ( isset( $parsedWidthParam['height'] ) ) {
5130 $height = $parsedWidthParam['height'];
5131 if ( $handler->validateParam( 'height', $height ) ) {
5132 $params[$type]['height'] = $height;
5133 $validated = true;
5134 }
5135 }
5136 # else no validation -- bug 13436
5137 } else {
5138 if ( $type === 'handler' ) {
5139 # Validate handler parameter
5140 $validated = $handler->validateParam( $paramName, $value );
5141 } else {
5142 # Validate internal parameters
5143 switch ( $paramName ) {
5144 case 'manualthumb':
5145 case 'alt':
5146 case 'class':
5147 # @todo FIXME: Possibly check validity here for
5148 # manualthumb? downstream behavior seems odd with
5149 # missing manual thumbs.
5150 $validated = true;
5151 $value = $this->stripAltText( $value, $holders );
5152 break;
5153 case 'link':
5154 $chars = self::EXT_LINK_URL_CLASS;
5155 $addr = self::EXT_LINK_ADDR;
5156 $prots = $this->mUrlProtocols;
5157 if ( $value === '' ) {
5158 $paramName = 'no-link';
5159 $value = true;
5160 $validated = true;
5161 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5162 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5163 $paramName = 'link-url';
5164 $this->mOutput->addExternalLink( $value );
5165 if ( $this->mOptions->getExternalLinkTarget() ) {
5166 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5167 }
5168 $validated = true;
5169 }
5170 } else {
5171 $linkTitle = Title::newFromText( $value );
5172 if ( $linkTitle ) {
5173 $paramName = 'link-title';
5174 $value = $linkTitle;
5175 $this->mOutput->addLink( $linkTitle );
5176 $validated = true;
5177 }
5178 }
5179 break;
5180 case 'frameless':
5181 case 'framed':
5182 case 'thumbnail':
5183 // use first appearing option, discard others.
5184 $validated = ! $seenformat;
5185 $seenformat = true;
5186 break;
5187 default:
5188 # Most other things appear to be empty or numeric...
5189 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5190 }
5191 }
5192
5193 if ( $validated ) {
5194 $params[$type][$paramName] = $value;
5195 }
5196 }
5197 }
5198 if ( !$validated ) {
5199 $caption = $part;
5200 }
5201 }
5202
5203 # Process alignment parameters
5204 if ( $params['horizAlign'] ) {
5205 $params['frame']['align'] = key( $params['horizAlign'] );
5206 }
5207 if ( $params['vertAlign'] ) {
5208 $params['frame']['valign'] = key( $params['vertAlign'] );
5209 }
5210
5211 $params['frame']['caption'] = $caption;
5212
5213 # Will the image be presented in a frame, with the caption below?
5214 $imageIsFramed = isset( $params['frame']['frame'] )
5215 || isset( $params['frame']['framed'] )
5216 || isset( $params['frame']['thumbnail'] )
5217 || isset( $params['frame']['manualthumb'] );
5218
5219 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5220 # came to also set the caption, ordinary text after the image -- which
5221 # makes no sense, because that just repeats the text multiple times in
5222 # screen readers. It *also* came to set the title attribute.
5223 # Now that we have an alt attribute, we should not set the alt text to
5224 # equal the caption: that's worse than useless, it just repeats the
5225 # text. This is the framed/thumbnail case. If there's no caption, we
5226 # use the unnamed parameter for alt text as well, just for the time be-
5227 # ing, if the unnamed param is set and the alt param is not.
5228 # For the future, we need to figure out if we want to tweak this more,
5229 # e.g., introducing a title= parameter for the title; ignoring the un-
5230 # named parameter entirely for images without a caption; adding an ex-
5231 # plicit caption= parameter and preserving the old magic unnamed para-
5232 # meter for BC; ...
5233 if ( $imageIsFramed ) { # Framed image
5234 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5235 # No caption or alt text, add the filename as the alt text so
5236 # that screen readers at least get some description of the image
5237 $params['frame']['alt'] = $title->getText();
5238 }
5239 # Do not set $params['frame']['title'] because tooltips don't make sense
5240 # for framed images
5241 } else { # Inline image
5242 if ( !isset( $params['frame']['alt'] ) ) {
5243 # No alt text, use the "caption" for the alt text
5244 if ( $caption !== '' ) {
5245 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5246 } else {
5247 # No caption, fall back to using the filename for the
5248 # alt text
5249 $params['frame']['alt'] = $title->getText();
5250 }
5251 }
5252 # Use the "caption" for the tooltip text
5253 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5254 }
5255
5256 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5257
5258 # Linker does the rest
5259 $time = isset( $options['time'] ) ? $options['time'] : false;
5260 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5261 $time, $descQuery, $this->mOptions->getThumbSize() );
5262
5263 # Give the handler a chance to modify the parser object
5264 if ( $handler ) {
5265 $handler->parserTransformHook( $this, $file );
5266 }
5267
5268 return $ret;
5269 }
5270
5271 /**
5272 * @param string $caption
5273 * @param LinkHolderArray|bool $holders
5274 * @return mixed|string
5275 */
5276 protected function stripAltText( $caption, $holders ) {
5277 # Strip bad stuff out of the title (tooltip). We can't just use
5278 # replaceLinkHoldersText() here, because if this function is called
5279 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5280 if ( $holders ) {
5281 $tooltip = $holders->replaceText( $caption );
5282 } else {
5283 $tooltip = $this->replaceLinkHoldersText( $caption );
5284 }
5285
5286 # make sure there are no placeholders in thumbnail attributes
5287 # that are later expanded to html- so expand them now and
5288 # remove the tags
5289 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5290 $tooltip = Sanitizer::stripAllTags( $tooltip );
5291
5292 return $tooltip;
5293 }
5294
5295 /**
5296 * Set a flag in the output object indicating that the content is dynamic and
5297 * shouldn't be cached.
5298 */
5299 public function disableCache() {
5300 wfDebug( "Parser output marked as uncacheable.\n" );
5301 if ( !$this->mOutput ) {
5302 throw new MWException( __METHOD__ .
5303 " can only be called when actually parsing something" );
5304 }
5305 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5306 }
5307
5308 /**
5309 * Callback from the Sanitizer for expanding items found in HTML attribute
5310 * values, so they can be safely tested and escaped.
5311 *
5312 * @param string $text
5313 * @param bool|PPFrame $frame
5314 * @return string
5315 */
5316 public function attributeStripCallback( &$text, $frame = false ) {
5317 $text = $this->replaceVariables( $text, $frame );
5318 $text = $this->mStripState->unstripBoth( $text );
5319 return $text;
5320 }
5321
5322 /**
5323 * Accessor
5324 *
5325 * @return array
5326 */
5327 public function getTags() {
5328 return array_merge(
5329 array_keys( $this->mTransparentTagHooks ),
5330 array_keys( $this->mTagHooks ),
5331 array_keys( $this->mFunctionTagHooks )
5332 );
5333 }
5334
5335 /**
5336 * Replace transparent tags in $text with the values given by the callbacks.
5337 *
5338 * Transparent tag hooks are like regular XML-style tag hooks, except they
5339 * operate late in the transformation sequence, on HTML instead of wikitext.
5340 *
5341 * @param string $text
5342 *
5343 * @return string
5344 */
5345 public function replaceTransparentTags( $text ) {
5346 $matches = [];
5347 $elements = array_keys( $this->mTransparentTagHooks );
5348 $text = self::extractTagsAndParams( $elements, $text, $matches );
5349 $replacements = [];
5350
5351 foreach ( $matches as $marker => $data ) {
5352 list( $element, $content, $params, $tag ) = $data;
5353 $tagName = strtolower( $element );
5354 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5355 $output = call_user_func_array(
5356 $this->mTransparentTagHooks[$tagName],
5357 [ $content, $params, $this ]
5358 );
5359 } else {
5360 $output = $tag;
5361 }
5362 $replacements[$marker] = $output;
5363 }
5364 return strtr( $text, $replacements );
5365 }
5366
5367 /**
5368 * Break wikitext input into sections, and either pull or replace
5369 * some particular section's text.
5370 *
5371 * External callers should use the getSection and replaceSection methods.
5372 *
5373 * @param string $text Page wikitext
5374 * @param string|number $sectionId A section identifier string of the form:
5375 * "<flag1> - <flag2> - ... - <section number>"
5376 *
5377 * Currently the only recognised flag is "T", which means the target section number
5378 * was derived during a template inclusion parse, in other words this is a template
5379 * section edit link. If no flags are given, it was an ordinary section edit link.
5380 * This flag is required to avoid a section numbering mismatch when a section is
5381 * enclosed by "<includeonly>" (bug 6563).
5382 *
5383 * The section number 0 pulls the text before the first heading; other numbers will
5384 * pull the given section along with its lower-level subsections. If the section is
5385 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5386 *
5387 * Section 0 is always considered to exist, even if it only contains the empty
5388 * string. If $text is the empty string and section 0 is replaced, $newText is
5389 * returned.
5390 *
5391 * @param string $mode One of "get" or "replace"
5392 * @param string $newText Replacement text for section data.
5393 * @return string For "get", the extracted section text.
5394 * for "replace", the whole page with the section replaced.
5395 */
5396 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5397 global $wgTitle; # not generally used but removes an ugly failure mode
5398
5399 $magicScopeVariable = $this->lock();
5400 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5401 $outText = '';
5402 $frame = $this->getPreprocessor()->newFrame();
5403
5404 # Process section extraction flags
5405 $flags = 0;
5406 $sectionParts = explode( '-', $sectionId );
5407 $sectionIndex = array_pop( $sectionParts );
5408 foreach ( $sectionParts as $part ) {
5409 if ( $part === 'T' ) {
5410 $flags |= self::PTD_FOR_INCLUSION;
5411 }
5412 }
5413
5414 # Check for empty input
5415 if ( strval( $text ) === '' ) {
5416 # Only sections 0 and T-0 exist in an empty document
5417 if ( $sectionIndex == 0 ) {
5418 if ( $mode === 'get' ) {
5419 return '';
5420 } else {
5421 return $newText;
5422 }
5423 } else {
5424 if ( $mode === 'get' ) {
5425 return $newText;
5426 } else {
5427 return $text;
5428 }
5429 }
5430 }
5431
5432 # Preprocess the text
5433 $root = $this->preprocessToDom( $text, $flags );
5434
5435 # <h> nodes indicate section breaks
5436 # They can only occur at the top level, so we can find them by iterating the root's children
5437 $node = $root->getFirstChild();
5438
5439 # Find the target section
5440 if ( $sectionIndex == 0 ) {
5441 # Section zero doesn't nest, level=big
5442 $targetLevel = 1000;
5443 } else {
5444 while ( $node ) {
5445 if ( $node->getName() === 'h' ) {
5446 $bits = $node->splitHeading();
5447 if ( $bits['i'] == $sectionIndex ) {
5448 $targetLevel = $bits['level'];
5449 break;
5450 }
5451 }
5452 if ( $mode === 'replace' ) {
5453 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5454 }
5455 $node = $node->getNextSibling();
5456 }
5457 }
5458
5459 if ( !$node ) {
5460 # Not found
5461 if ( $mode === 'get' ) {
5462 return $newText;
5463 } else {
5464 return $text;
5465 }
5466 }
5467
5468 # Find the end of the section, including nested sections
5469 do {
5470 if ( $node->getName() === 'h' ) {
5471 $bits = $node->splitHeading();
5472 $curLevel = $bits['level'];
5473 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5474 break;
5475 }
5476 }
5477 if ( $mode === 'get' ) {
5478 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5479 }
5480 $node = $node->getNextSibling();
5481 } while ( $node );
5482
5483 # Write out the remainder (in replace mode only)
5484 if ( $mode === 'replace' ) {
5485 # Output the replacement text
5486 # Add two newlines on -- trailing whitespace in $newText is conventionally
5487 # stripped by the editor, so we need both newlines to restore the paragraph gap
5488 # Only add trailing whitespace if there is newText
5489 if ( $newText != "" ) {
5490 $outText .= $newText . "\n\n";
5491 }
5492
5493 while ( $node ) {
5494 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5495 $node = $node->getNextSibling();
5496 }
5497 }
5498
5499 if ( is_string( $outText ) ) {
5500 # Re-insert stripped tags
5501 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5502 }
5503
5504 return $outText;
5505 }
5506
5507 /**
5508 * This function returns the text of a section, specified by a number ($section).
5509 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5510 * the first section before any such heading (section 0).
5511 *
5512 * If a section contains subsections, these are also returned.
5513 *
5514 * @param string $text Text to look in
5515 * @param string|number $sectionId Section identifier as a number or string
5516 * (e.g. 0, 1 or 'T-1').
5517 * @param string $defaultText Default to return if section is not found
5518 *
5519 * @return string Text of the requested section
5520 */
5521 public function getSection( $text, $sectionId, $defaultText = '' ) {
5522 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5523 }
5524
5525 /**
5526 * This function returns $oldtext after the content of the section
5527 * specified by $section has been replaced with $text. If the target
5528 * section does not exist, $oldtext is returned unchanged.
5529 *
5530 * @param string $oldText Former text of the article
5531 * @param string|number $sectionId Section identifier as a number or string
5532 * (e.g. 0, 1 or 'T-1').
5533 * @param string $newText Replacing text
5534 *
5535 * @return string Modified text
5536 */
5537 public function replaceSection( $oldText, $sectionId, $newText ) {
5538 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5539 }
5540
5541 /**
5542 * Get the ID of the revision we are parsing
5543 *
5544 * @return int|null
5545 */
5546 public function getRevisionId() {
5547 return $this->mRevisionId;
5548 }
5549
5550 /**
5551 * Get the revision object for $this->mRevisionId
5552 *
5553 * @return Revision|null Either a Revision object or null
5554 * @since 1.23 (public since 1.23)
5555 */
5556 public function getRevisionObject() {
5557 if ( !is_null( $this->mRevisionObject ) ) {
5558 return $this->mRevisionObject;
5559 }
5560 if ( is_null( $this->mRevisionId ) ) {
5561 return null;
5562 }
5563
5564 $rev = call_user_func(
5565 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5566 );
5567
5568 # If the parse is for a new revision, then the callback should have
5569 # already been set to force the object and should match mRevisionId.
5570 # If not, try to fetch by mRevisionId for sanity.
5571 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5572 $rev = Revision::newFromId( $this->mRevisionId );
5573 }
5574
5575 $this->mRevisionObject = $rev;
5576
5577 return $this->mRevisionObject;
5578 }
5579
5580 /**
5581 * Get the timestamp associated with the current revision, adjusted for
5582 * the default server-local timestamp
5583 * @return string
5584 */
5585 public function getRevisionTimestamp() {
5586 if ( is_null( $this->mRevisionTimestamp ) ) {
5587 global $wgContLang;
5588
5589 $revObject = $this->getRevisionObject();
5590 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5591
5592 # The cryptic '' timezone parameter tells to use the site-default
5593 # timezone offset instead of the user settings.
5594 # Since this value will be saved into the parser cache, served
5595 # to other users, and potentially even used inside links and such,
5596 # it needs to be consistent for all visitors.
5597 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5598
5599 }
5600 return $this->mRevisionTimestamp;
5601 }
5602
5603 /**
5604 * Get the name of the user that edited the last revision
5605 *
5606 * @return string User name
5607 */
5608 public function getRevisionUser() {
5609 if ( is_null( $this->mRevisionUser ) ) {
5610 $revObject = $this->getRevisionObject();
5611
5612 # if this template is subst: the revision id will be blank,
5613 # so just use the current user's name
5614 if ( $revObject ) {
5615 $this->mRevisionUser = $revObject->getUserText();
5616 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5617 $this->mRevisionUser = $this->getUser()->getName();
5618 }
5619 }
5620 return $this->mRevisionUser;
5621 }
5622
5623 /**
5624 * Get the size of the revision
5625 *
5626 * @return int|null Revision size
5627 */
5628 public function getRevisionSize() {
5629 if ( is_null( $this->mRevisionSize ) ) {
5630 $revObject = $this->getRevisionObject();
5631
5632 # if this variable is subst: the revision id will be blank,
5633 # so just use the parser input size, because the own substituation
5634 # will change the size.
5635 if ( $revObject ) {
5636 $this->mRevisionSize = $revObject->getSize();
5637 } else {
5638 $this->mRevisionSize = $this->mInputSize;
5639 }
5640 }
5641 return $this->mRevisionSize;
5642 }
5643
5644 /**
5645 * Mutator for $mDefaultSort
5646 *
5647 * @param string $sort New value
5648 */
5649 public function setDefaultSort( $sort ) {
5650 $this->mDefaultSort = $sort;
5651 $this->mOutput->setProperty( 'defaultsort', $sort );
5652 }
5653
5654 /**
5655 * Accessor for $mDefaultSort
5656 * Will use the empty string if none is set.
5657 *
5658 * This value is treated as a prefix, so the
5659 * empty string is equivalent to sorting by
5660 * page name.
5661 *
5662 * @return string
5663 */
5664 public function getDefaultSort() {
5665 if ( $this->mDefaultSort !== false ) {
5666 return $this->mDefaultSort;
5667 } else {
5668 return '';
5669 }
5670 }
5671
5672 /**
5673 * Accessor for $mDefaultSort
5674 * Unlike getDefaultSort(), will return false if none is set
5675 *
5676 * @return string|bool
5677 */
5678 public function getCustomDefaultSort() {
5679 return $this->mDefaultSort;
5680 }
5681
5682 /**
5683 * Try to guess the section anchor name based on a wikitext fragment
5684 * presumably extracted from a heading, for example "Header" from
5685 * "== Header ==".
5686 *
5687 * @param string $text
5688 *
5689 * @return string
5690 */
5691 public function guessSectionNameFromWikiText( $text ) {
5692 # Strip out wikitext links(they break the anchor)
5693 $text = $this->stripSectionName( $text );
5694 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5695 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5696 }
5697
5698 /**
5699 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5700 * instead. For use in redirects, since IE6 interprets Redirect: headers
5701 * as something other than UTF-8 (apparently?), resulting in breakage.
5702 *
5703 * @param string $text The section name
5704 * @return string An anchor
5705 */
5706 public function guessLegacySectionNameFromWikiText( $text ) {
5707 # Strip out wikitext links(they break the anchor)
5708 $text = $this->stripSectionName( $text );
5709 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5710 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5711 }
5712
5713 /**
5714 * Strips a text string of wikitext for use in a section anchor
5715 *
5716 * Accepts a text string and then removes all wikitext from the
5717 * string and leaves only the resultant text (i.e. the result of
5718 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5719 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5720 * to create valid section anchors by mimicing the output of the
5721 * parser when headings are parsed.
5722 *
5723 * @param string $text Text string to be stripped of wikitext
5724 * for use in a Section anchor
5725 * @return string Filtered text string
5726 */
5727 public function stripSectionName( $text ) {
5728 # Strip internal link markup
5729 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5730 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5731
5732 # Strip external link markup
5733 # @todo FIXME: Not tolerant to blank link text
5734 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5735 # on how many empty links there are on the page - need to figure that out.
5736 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5737
5738 # Parse wikitext quotes (italics & bold)
5739 $text = $this->doQuotes( $text );
5740
5741 # Strip HTML tags
5742 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5743 return $text;
5744 }
5745
5746 /**
5747 * strip/replaceVariables/unstrip for preprocessor regression testing
5748 *
5749 * @param string $text
5750 * @param Title $title
5751 * @param ParserOptions $options
5752 * @param int $outputType
5753 *
5754 * @return string
5755 */
5756 public function testSrvus( $text, Title $title, ParserOptions $options,
5757 $outputType = self::OT_HTML
5758 ) {
5759 $magicScopeVariable = $this->lock();
5760 $this->startParse( $title, $options, $outputType, true );
5761
5762 $text = $this->replaceVariables( $text );
5763 $text = $this->mStripState->unstripBoth( $text );
5764 $text = Sanitizer::removeHTMLtags( $text );
5765 return $text;
5766 }
5767
5768 /**
5769 * @param string $text
5770 * @param Title $title
5771 * @param ParserOptions $options
5772 * @return string
5773 */
5774 public function testPst( $text, Title $title, ParserOptions $options ) {
5775 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5776 }
5777
5778 /**
5779 * @param string $text
5780 * @param Title $title
5781 * @param ParserOptions $options
5782 * @return string
5783 */
5784 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5785 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5786 }
5787
5788 /**
5789 * Call a callback function on all regions of the given text that are not
5790 * inside strip markers, and replace those regions with the return value
5791 * of the callback. For example, with input:
5792 *
5793 * aaa<MARKER>bbb
5794 *
5795 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5796 * two strings will be replaced with the value returned by the callback in
5797 * each case.
5798 *
5799 * @param string $s
5800 * @param callable $callback
5801 *
5802 * @return string
5803 */
5804 public function markerSkipCallback( $s, $callback ) {
5805 $i = 0;
5806 $out = '';
5807 while ( $i < strlen( $s ) ) {
5808 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5809 if ( $markerStart === false ) {
5810 $out .= call_user_func( $callback, substr( $s, $i ) );
5811 break;
5812 } else {
5813 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5814 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5815 if ( $markerEnd === false ) {
5816 $out .= substr( $s, $markerStart );
5817 break;
5818 } else {
5819 $markerEnd += strlen( self::MARKER_SUFFIX );
5820 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5821 $i = $markerEnd;
5822 }
5823 }
5824 }
5825 return $out;
5826 }
5827
5828 /**
5829 * Remove any strip markers found in the given text.
5830 *
5831 * @param string $text Input string
5832 * @return string
5833 */
5834 public function killMarkers( $text ) {
5835 return $this->mStripState->killMarkers( $text );
5836 }
5837
5838 /**
5839 * Save the parser state required to convert the given half-parsed text to
5840 * HTML. "Half-parsed" in this context means the output of
5841 * recursiveTagParse() or internalParse(). This output has strip markers
5842 * from replaceVariables (extensionSubstitution() etc.), and link
5843 * placeholders from replaceLinkHolders().
5844 *
5845 * Returns an array which can be serialized and stored persistently. This
5846 * array can later be loaded into another parser instance with
5847 * unserializeHalfParsedText(). The text can then be safely incorporated into
5848 * the return value of a parser hook.
5849 *
5850 * @param string $text
5851 *
5852 * @return array
5853 */
5854 public function serializeHalfParsedText( $text ) {
5855 $data = [
5856 'text' => $text,
5857 'version' => self::HALF_PARSED_VERSION,
5858 'stripState' => $this->mStripState->getSubState( $text ),
5859 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5860 ];
5861 return $data;
5862 }
5863
5864 /**
5865 * Load the parser state given in the $data array, which is assumed to
5866 * have been generated by serializeHalfParsedText(). The text contents is
5867 * extracted from the array, and its markers are transformed into markers
5868 * appropriate for the current Parser instance. This transformed text is
5869 * returned, and can be safely included in the return value of a parser
5870 * hook.
5871 *
5872 * If the $data array has been stored persistently, the caller should first
5873 * check whether it is still valid, by calling isValidHalfParsedText().
5874 *
5875 * @param array $data Serialized data
5876 * @throws MWException
5877 * @return string
5878 */
5879 public function unserializeHalfParsedText( $data ) {
5880 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5881 throw new MWException( __METHOD__ . ': invalid version' );
5882 }
5883
5884 # First, extract the strip state.
5885 $texts = [ $data['text'] ];
5886 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5887
5888 # Now renumber links
5889 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5890
5891 # Should be good to go.
5892 return $texts[0];
5893 }
5894
5895 /**
5896 * Returns true if the given array, presumed to be generated by
5897 * serializeHalfParsedText(), is compatible with the current version of the
5898 * parser.
5899 *
5900 * @param array $data
5901 *
5902 * @return bool
5903 */
5904 public function isValidHalfParsedText( $data ) {
5905 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5906 }
5907
5908 /**
5909 * Parsed a width param of imagelink like 300px or 200x300px
5910 *
5911 * @param string $value
5912 *
5913 * @return array
5914 * @since 1.20
5915 */
5916 public function parseWidthParam( $value ) {
5917 $parsedWidthParam = [];
5918 if ( $value === '' ) {
5919 return $parsedWidthParam;
5920 }
5921 $m = [];
5922 # (bug 13500) In both cases (width/height and width only),
5923 # permit trailing "px" for backward compatibility.
5924 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5925 $width = intval( $m[1] );
5926 $height = intval( $m[2] );
5927 $parsedWidthParam['width'] = $width;
5928 $parsedWidthParam['height'] = $height;
5929 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5930 $width = intval( $value );
5931 $parsedWidthParam['width'] = $width;
5932 }
5933 return $parsedWidthParam;
5934 }
5935
5936 /**
5937 * Lock the current instance of the parser.
5938 *
5939 * This is meant to stop someone from calling the parser
5940 * recursively and messing up all the strip state.
5941 *
5942 * @throws MWException If parser is in a parse
5943 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5944 */
5945 protected function lock() {
5946 if ( $this->mInParse ) {
5947 throw new MWException( "Parser state cleared while parsing. "
5948 . "Did you call Parser::parse recursively?" );
5949 }
5950 $this->mInParse = true;
5951
5952 $recursiveCheck = new ScopedCallback( function() {
5953 $this->mInParse = false;
5954 } );
5955
5956 return $recursiveCheck;
5957 }
5958
5959 /**
5960 * Strip outer <p></p> tag from the HTML source of a single paragraph.
5961 *
5962 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
5963 * or if there is more than one <p/> tag in the input HTML.
5964 *
5965 * @param string $html
5966 * @return string
5967 * @since 1.24
5968 */
5969 public static function stripOuterParagraph( $html ) {
5970 $m = [];
5971 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
5972 if ( strpos( $m[1], '</p>' ) === false ) {
5973 $html = $m[1];
5974 }
5975 }
5976
5977 return $html;
5978 }
5979
5980 /**
5981 * Return this parser if it is not doing anything, otherwise
5982 * get a fresh parser. You can use this method by doing
5983 * $myParser = $wgParser->getFreshParser(), or more simply
5984 * $wgParser->getFreshParser()->parse( ... );
5985 * if you're unsure if $wgParser is safe to use.
5986 *
5987 * @since 1.24
5988 * @return Parser A parser object that is not parsing anything
5989 */
5990 public function getFreshParser() {
5991 global $wgParserConf;
5992 if ( $this->mInParse ) {
5993 return new $wgParserConf['class']( $wgParserConf );
5994 } else {
5995 return $this;
5996 }
5997 }
5998
5999 /**
6000 * Set's up the PHP implementation of OOUI for use in this request
6001 * and instructs OutputPage to enable OOUI for itself.
6002 *
6003 * @since 1.26
6004 */
6005 public function enableOOUI() {
6006 OutputPage::setupOOUI();
6007 $this->mOutput->setEnableOOUI( true );
6008 }
6009 }
Wiklou, le Wiki du Wiklou