dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Log when resyncFiles() fails in FileBackendMultiWrite"
[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
24 /**
25 * @defgroup Parser Parser
26 */
27
28 /**
29 * PHP Parser - Processes wiki markup (which uses a more user-friendly
30 * syntax, such as "[[link]]" for making links), and provides a one-way
31 * transformation of that wiki markup it into (X)HTML output / markup
32 * (which in turn the browser understands, and can display).
33 *
34 * There are seven main entry points into the Parser class:
35 *
36 * - Parser::parse()
37 * produces HTML output
38 * - Parser::preSaveTransform()
39 * produces altered wiki markup
40 * - Parser::preprocess()
41 * removes HTML comments and expands templates
42 * - Parser::cleanSig() and Parser::cleanSigInSig()
43 * cleans a signature before saving it to preferences
44 * - Parser::getSection()
45 * return the content of a section from an article for section editing
46 * - Parser::replaceSection()
47 * replaces a section by number inside an article
48 * - Parser::getPreloadText()
49 * removes <noinclude> sections and <includeonly> tags
50 *
51 * Globals used:
52 * object: $wgContLang
53 *
54 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
55 *
56 * @par Settings:
57 * $wgNamespacesWithSubpages
58 *
59 * @par Settings only within ParserOptions:
60 * $wgAllowExternalImages
61 * $wgAllowSpecialInclusion
62 * $wgInterwikiMagic
63 * $wgMaxArticleSize
64 *
65 * @ingroup Parser
66 */
67 class Parser {
68 /**
69 * Update this version number when the ParserOutput format
70 * changes in an incompatible way, so the parser cache
71 * can automatically discard old data.
72 */
73 const VERSION = '1.6.4';
74
75 /**
76 * Update this version number when the output of serialiseHalfParsedText()
77 * changes in an incompatible way
78 */
79 const HALF_PARSED_VERSION = 2;
80
81 # Flags for Parser::setFunctionHook
82 const SFH_NO_HASH = 1;
83 const SFH_OBJECT_ARGS = 2;
84
85 # Constants needed for external link processing
86 # Everything except bracket, space, or control characters
87 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
88 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
89 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
90 # Simplified expression to match an IPv4 or IPv6 address, or
91 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
92 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
93 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
94 // @codingStandardsIgnoreStart Generic.Files.LineLength
95 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
96 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
97 // @codingStandardsIgnoreEnd
98
99 # Regular expression for a non-newline space
100 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
101
102 # State constants for the definition list colon extraction
103 const COLON_STATE_TEXT = 0;
104 const COLON_STATE_TAG = 1;
105 const COLON_STATE_TAGSTART = 2;
106 const COLON_STATE_CLOSETAG = 3;
107 const COLON_STATE_TAGSLASH = 4;
108 const COLON_STATE_COMMENT = 5;
109 const COLON_STATE_COMMENTDASH = 6;
110 const COLON_STATE_COMMENTDASHDASH = 7;
111
112 # Flags for preprocessToDom
113 const PTD_FOR_INCLUSION = 1;
114
115 # Allowed values for $this->mOutputType
116 # Parameter to startExternalParse().
117 const OT_HTML = 1; # like parse()
118 const OT_WIKI = 2; # like preSaveTransform()
119 const OT_PREPROCESS = 3; # like preprocess()
120 const OT_MSG = 3;
121 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
122
123 /**
124 * @var string Prefix and suffix for temporary replacement strings
125 * for the multipass parser.
126 *
127 * \x7f should never appear in input as it's disallowed in XML.
128 * Using it at the front also gives us a little extra robustness
129 * since it shouldn't match when butted up against identifier-like
130 * string constructs.
131 *
132 * Must not consist of all title characters, or else it will change
133 * the behavior of <nowiki> in a link.
134 */
135 const MARKER_SUFFIX = "-QINU\x7f";
136 const MARKER_PREFIX = "\x7fUNIQ-";
137
138 # Markers used for wrapping the table of contents
139 const TOC_START = '<mw:toc>';
140 const TOC_END = '</mw:toc>';
141
142 # Persistent:
143 public $mTagHooks = [];
144 public $mTransparentTagHooks = [];
145 public $mFunctionHooks = [];
146 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
147 public $mFunctionTagHooks = [];
148 public $mStripList = [];
149 public $mDefaultStripList = [];
150 public $mVarCache = [];
151 public $mImageParams = [];
152 public $mImageParamsMagicArray = [];
153 public $mMarkerIndex = 0;
154 public $mFirstCall = true;
155
156 # Initialised by initialiseVariables()
157
158 /**
159 * @var MagicWordArray
160 */
161 public $mVariables;
162
163 /**
164 * @var MagicWordArray
165 */
166 public $mSubstWords;
167 # Initialised in constructor
168 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
169
170 # Initialized in getPreprocessor()
171 /** @var Preprocessor */
172 public $mPreprocessor;
173
174 # Cleared with clearState():
175 /**
176 * @var ParserOutput
177 */
178 public $mOutput;
179 public $mAutonumber, $mDTopen;
180
181 /**
182 * @var StripState
183 */
184 public $mStripState;
185
186 public $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
187 /**
188 * @var LinkHolderArray
189 */
190 public $mLinkHolders;
191
192 public $mLinkID;
193 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
194 public $mDefaultSort;
195 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
196 public $mExpensiveFunctionCount; # number of expensive parser function calls
197 public $mShowToc, $mForceTocPosition;
198
199 /**
200 * @var User
201 */
202 public $mUser; # User object; only used when doing pre-save transform
203
204 # Temporary
205 # These are variables reset at least once per parse regardless of $clearState
206
207 /**
208 * @var ParserOptions
209 */
210 public $mOptions;
211
212 /**
213 * @var Title
214 */
215 public $mTitle; # Title context, used for self-link rendering and similar things
216 public $mOutputType; # Output type, one of the OT_xxx constants
217 public $ot; # Shortcut alias, see setOutputType()
218 public $mRevisionObject; # The revision object of the specified revision ID
219 public $mRevisionId; # ID to display in {{REVISIONID}} tags
220 public $mRevisionTimestamp; # The timestamp of the specified revision ID
221 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
222 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
223 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
224 public $mInputSize = false; # For {{PAGESIZE}} on current page.
225
226 /**
227 * @var string Deprecated accessor for the strip marker prefix.
228 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
229 **/
230 public $mUniqPrefix = Parser::MARKER_PREFIX;
231
232 /**
233 * @var array Array with the language name of each language link (i.e. the
234 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
235 * duplicate language links to the ParserOutput.
236 */
237 public $mLangLinkLanguages;
238
239 /**
240 * @var MapCacheLRU|null
241 * @since 1.24
242 *
243 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
244 */
245 public $currentRevisionCache;
246
247 /**
248 * @var bool Recursive call protection.
249 * This variable should be treated as if it were private.
250 */
251 public $mInParse = false;
252
253 /** @var SectionProfiler */
254 protected $mProfiler;
255
256 /**
257 * @param array $conf
258 */
259 public function __construct( $conf = [] ) {
260 $this->mConf = $conf;
261 $this->mUrlProtocols = wfUrlProtocols();
262 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
263 self::EXT_LINK_ADDR .
264 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
265 if ( isset( $conf['preprocessorClass'] ) ) {
266 $this->mPreprocessorClass = $conf['preprocessorClass'];
267 } elseif ( defined( 'HPHP_VERSION' ) ) {
268 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
269 $this->mPreprocessorClass = 'Preprocessor_Hash';
270 } elseif ( extension_loaded( 'domxml' ) ) {
271 # PECL extension that conflicts with the core DOM extension (bug 13770)
272 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
273 $this->mPreprocessorClass = 'Preprocessor_Hash';
274 } elseif ( extension_loaded( 'dom' ) ) {
275 $this->mPreprocessorClass = 'Preprocessor_DOM';
276 } else {
277 $this->mPreprocessorClass = 'Preprocessor_Hash';
278 }
279 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
280 }
281
282 /**
283 * Reduce memory usage to reduce the impact of circular references
284 */
285 public function __destruct() {
286 if ( isset( $this->mLinkHolders ) ) {
287 unset( $this->mLinkHolders );
288 }
289 foreach ( $this as $name => $value ) {
290 unset( $this->$name );
291 }
292 }
293
294 /**
295 * Allow extensions to clean up when the parser is cloned
296 */
297 public function __clone() {
298 $this->mInParse = false;
299
300 // Bug 56226: When you create a reference "to" an object field, that
301 // makes the object field itself be a reference too (until the other
302 // reference goes out of scope). When cloning, any field that's a
303 // reference is copied as a reference in the new object. Both of these
304 // are defined PHP5 behaviors, as inconvenient as it is for us when old
305 // hooks from PHP4 days are passing fields by reference.
306 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
307 // Make a non-reference copy of the field, then rebind the field to
308 // reference the new copy.
309 $tmp = $this->$k;
310 $this->$k =& $tmp;
311 unset( $tmp );
312 }
313
314 Hooks::run( 'ParserCloned', [ $this ] );
315 }
316
317 /**
318 * Do various kinds of initialisation on the first call of the parser
319 */
320 public function firstCallInit() {
321 if ( !$this->mFirstCall ) {
322 return;
323 }
324 $this->mFirstCall = false;
325
326 CoreParserFunctions::register( $this );
327 CoreTagHooks::register( $this );
328 $this->initialiseVariables();
329
330 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
331 }
332
333 /**
334 * Clear Parser state
335 *
336 * @private
337 */
338 public function clearState() {
339 if ( $this->mFirstCall ) {
340 $this->firstCallInit();
341 }
342 $this->mOutput = new ParserOutput;
343 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
344 $this->mAutonumber = 0;
345 $this->mLastSection = '';
346 $this->mDTopen = false;
347 $this->mIncludeCount = [];
348 $this->mArgStack = false;
349 $this->mInPre = false;
350 $this->mLinkHolders = new LinkHolderArray( $this );
351 $this->mLinkID = 0;
352 $this->mRevisionObject = $this->mRevisionTimestamp =
353 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
354 $this->mVarCache = [];
355 $this->mUser = null;
356 $this->mLangLinkLanguages = [];
357 $this->currentRevisionCache = null;
358
359 $this->mStripState = new StripState;
360
361 # Clear these on every parse, bug 4549
362 $this->mTplRedirCache = $this->mTplDomCache = [];
363
364 $this->mShowToc = true;
365 $this->mForceTocPosition = false;
366 $this->mIncludeSizes = [
367 'post-expand' => 0,
368 'arg' => 0,
369 ];
370 $this->mPPNodeCount = 0;
371 $this->mGeneratedPPNodeCount = 0;
372 $this->mHighestExpansionDepth = 0;
373 $this->mDefaultSort = false;
374 $this->mHeadings = [];
375 $this->mDoubleUnderscores = [];
376 $this->mExpensiveFunctionCount = 0;
377
378 # Fix cloning
379 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
380 $this->mPreprocessor = null;
381 }
382
383 $this->mProfiler = new SectionProfiler();
384
385 Hooks::run( 'ParserClearState', [ &$this ] );
386 }
387
388 /**
389 * Convert wikitext to HTML
390 * Do not call this function recursively.
391 *
392 * @param string $text Text we want to parse
393 * @param Title $title
394 * @param ParserOptions $options
395 * @param bool $linestart
396 * @param bool $clearState
397 * @param int $revid Number to pass in {{REVISIONID}}
398 * @return ParserOutput A ParserOutput
399 */
400 public function parse( $text, Title $title, ParserOptions $options,
401 $linestart = true, $clearState = true, $revid = null
402 ) {
403 /**
404 * First pass--just handle <nowiki> sections, pass the rest off
405 * to internalParse() which does all the real work.
406 */
407
408 global $wgShowHostnames;
409
410 if ( $clearState ) {
411 // We use U+007F DELETE to construct strip markers, so we have to make
412 // sure that this character does not occur in the input text.
413 $text = strtr( $text, "\x7f", "?" );
414 $magicScopeVariable = $this->lock();
415 }
416
417 $this->startParse( $title, $options, self::OT_HTML, $clearState );
418
419 $this->currentRevisionCache = null;
420 $this->mInputSize = strlen( $text );
421 if ( $this->mOptions->getEnableLimitReport() ) {
422 $this->mOutput->resetParseStartTime();
423 }
424
425 $oldRevisionId = $this->mRevisionId;
426 $oldRevisionObject = $this->mRevisionObject;
427 $oldRevisionTimestamp = $this->mRevisionTimestamp;
428 $oldRevisionUser = $this->mRevisionUser;
429 $oldRevisionSize = $this->mRevisionSize;
430 if ( $revid !== null ) {
431 $this->mRevisionId = $revid;
432 $this->mRevisionObject = null;
433 $this->mRevisionTimestamp = null;
434 $this->mRevisionUser = null;
435 $this->mRevisionSize = null;
436 }
437
438 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
439 # No more strip!
440 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
441 $text = $this->internalParse( $text );
442 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
443
444 $text = $this->internalParseHalfParsed( $text, true, $linestart );
445
446 /**
447 * A converted title will be provided in the output object if title and
448 * content conversion are enabled, the article text does not contain
449 * a conversion-suppressing double-underscore tag, and no
450 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
451 * automatic link conversion.
452 */
453 if ( !( $options->getDisableTitleConversion()
454 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
455 || isset( $this->mDoubleUnderscores['notitleconvert'] )
456 || $this->mOutput->getDisplayTitle() !== false )
457 ) {
458 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
459 if ( $convruletitle ) {
460 $this->mOutput->setTitleText( $convruletitle );
461 } else {
462 $titleText = $this->getConverterLanguage()->convertTitle( $title );
463 $this->mOutput->setTitleText( $titleText );
464 }
465 }
466
467 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
468 $this->limitationWarn( 'expensive-parserfunction',
469 $this->mExpensiveFunctionCount,
470 $this->mOptions->getExpensiveParserFunctionLimit()
471 );
472 }
473
474 # Information on include size limits, for the benefit of users who try to skirt them
475 if ( $this->mOptions->getEnableLimitReport() ) {
476 $max = $this->mOptions->getMaxIncludeSize();
477
478 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
479 if ( $cpuTime !== null ) {
480 $this->mOutput->setLimitReportData( 'limitreport-cputime',
481 sprintf( "%.3f", $cpuTime )
482 );
483 }
484
485 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
486 $this->mOutput->setLimitReportData( 'limitreport-walltime',
487 sprintf( "%.3f", $wallTime )
488 );
489
490 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
491 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
492 );
493 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
494 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
495 );
496 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
497 [ $this->mIncludeSizes['post-expand'], $max ]
498 );
499 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
500 [ $this->mIncludeSizes['arg'], $max ]
501 );
502 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
503 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
504 );
505 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
506 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
507 );
508 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
509
510 $limitReport = "NewPP limit report\n";
511 if ( $wgShowHostnames ) {
512 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
513 }
514 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
515 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
516 $limitReport .= 'Dynamic content: ' .
517 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
518 "\n";
519
520 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
521 if ( Hooks::run( 'ParserLimitReportFormat',
522 [ $key, &$value, &$limitReport, false, false ]
523 ) ) {
524 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
525 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
526 ->inLanguage( 'en' )->useDatabase( false );
527 if ( !$valueMsg->exists() ) {
528 $valueMsg = new RawMessage( '$1' );
529 }
530 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
531 $valueMsg->params( $value );
532 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
533 }
534 }
535 }
536 // Since we're not really outputting HTML, decode the entities and
537 // then re-encode the things that need hiding inside HTML comments.
538 $limitReport = htmlspecialchars_decode( $limitReport );
539 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
540
541 // Sanitize for comment. Note '‐' in the replacement is U+2010,
542 // which looks much like the problematic '-'.
543 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
544 $text .= "\n<!-- \n$limitReport-->\n";
545
546 // Add on template profiling data
547 $dataByFunc = $this->mProfiler->getFunctionStats();
548 uasort( $dataByFunc, function ( $a, $b ) {
549 return $a['real'] < $b['real']; // descending order
550 } );
551 $profileReport = "Transclusion expansion time report (%,ms,calls,template)\n";
552 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
553 $profileReport .= sprintf( "%6.2f%% %8.3f %6d - %s\n",
554 $item['%real'], $item['real'], $item['calls'],
555 htmlspecialchars( $item['name'] ) );
556 }
557 $text .= "\n<!-- \n$profileReport-->\n";
558
559 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
560 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
561 $this->mTitle->getPrefixedDBkey() );
562 }
563 }
564 $this->mOutput->setText( $text );
565
566 $this->mRevisionId = $oldRevisionId;
567 $this->mRevisionObject = $oldRevisionObject;
568 $this->mRevisionTimestamp = $oldRevisionTimestamp;
569 $this->mRevisionUser = $oldRevisionUser;
570 $this->mRevisionSize = $oldRevisionSize;
571 $this->mInputSize = false;
572 $this->currentRevisionCache = null;
573
574 return $this->mOutput;
575 }
576
577 /**
578 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
579 * can be called from an extension tag hook.
580 *
581 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
582 * instead, which means that lists and links have not been fully parsed yet,
583 * and strip markers are still present.
584 *
585 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
586 *
587 * Use this function if you're a parser tag hook and you want to parse
588 * wikitext before or after applying additional transformations, and you
589 * intend to *return the result as hook output*, which will cause it to go
590 * through the rest of parsing process automatically.
591 *
592 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
593 * $text are not expanded
594 *
595 * @param string $text Text extension wants to have parsed
596 * @param bool|PPFrame $frame The frame to use for expanding any template variables
597 * @return string UNSAFE half-parsed HTML
598 */
599 public function recursiveTagParse( $text, $frame = false ) {
600 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
601 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
602 $text = $this->internalParse( $text, false, $frame );
603 return $text;
604 }
605
606 /**
607 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
608 * point can be called from an extension tag hook.
609 *
610 * The output of this function is fully-parsed HTML that is safe for output.
611 * If you're a parser tag hook, you might want to use recursiveTagParse()
612 * instead.
613 *
614 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
615 * $text are not expanded
616 *
617 * @since 1.25
618 *
619 * @param string $text Text extension wants to have parsed
620 * @param bool|PPFrame $frame The frame to use for expanding any template variables
621 * @return string Fully parsed HTML
622 */
623 public function recursiveTagParseFully( $text, $frame = false ) {
624 $text = $this->recursiveTagParse( $text, $frame );
625 $text = $this->internalParseHalfParsed( $text, false );
626 return $text;
627 }
628
629 /**
630 * Expand templates and variables in the text, producing valid, static wikitext.
631 * Also removes comments.
632 * Do not call this function recursively.
633 * @param string $text
634 * @param Title $title
635 * @param ParserOptions $options
636 * @param int|null $revid
637 * @param bool|PPFrame $frame
638 * @return mixed|string
639 */
640 public function preprocess( $text, Title $title = null,
641 ParserOptions $options, $revid = null, $frame = false
642 ) {
643 $magicScopeVariable = $this->lock();
644 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
645 if ( $revid !== null ) {
646 $this->mRevisionId = $revid;
647 }
648 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
649 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
650 $text = $this->replaceVariables( $text, $frame );
651 $text = $this->mStripState->unstripBoth( $text );
652 return $text;
653 }
654
655 /**
656 * Recursive parser entry point that can be called from an extension tag
657 * hook.
658 *
659 * @param string $text Text to be expanded
660 * @param bool|PPFrame $frame The frame to use for expanding any template variables
661 * @return string
662 * @since 1.19
663 */
664 public function recursivePreprocess( $text, $frame = false ) {
665 $text = $this->replaceVariables( $text, $frame );
666 $text = $this->mStripState->unstripBoth( $text );
667 return $text;
668 }
669
670 /**
671 * Process the wikitext for the "?preload=" feature. (bug 5210)
672 *
673 * "<noinclude>", "<includeonly>" etc. are parsed as for template
674 * transclusion, comments, templates, arguments, tags hooks and parser
675 * functions are untouched.
676 *
677 * @param string $text
678 * @param Title $title
679 * @param ParserOptions $options
680 * @param array $params
681 * @return string
682 */
683 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
684 $msg = new RawMessage( $text );
685 $text = $msg->params( $params )->plain();
686
687 # Parser (re)initialisation
688 $magicScopeVariable = $this->lock();
689 $this->startParse( $title, $options, self::OT_PLAIN, true );
690
691 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
692 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
693 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
694 $text = $this->mStripState->unstripBoth( $text );
695 return $text;
696 }
697
698 /**
699 * Get a random string
700 *
701 * @return string
702 * @deprecated since 1.26; use wfRandomString() instead.
703 */
704 public static function getRandomString() {
705 wfDeprecated( __METHOD__, '1.26' );
706 return wfRandomString( 16 );
707 }
708
709 /**
710 * Set the current user.
711 * Should only be used when doing pre-save transform.
712 *
713 * @param User|null $user User object or null (to reset)
714 */
715 public function setUser( $user ) {
716 $this->mUser = $user;
717 }
718
719 /**
720 * Accessor for mUniqPrefix.
721 *
722 * @return string
723 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
724 */
725 public function uniqPrefix() {
726 wfDeprecated( __METHOD__, '1.26' );
727 return self::MARKER_PREFIX;
728 }
729
730 /**
731 * Set the context title
732 *
733 * @param Title $t
734 */
735 public function setTitle( $t ) {
736 if ( !$t ) {
737 $t = Title::newFromText( 'NO TITLE' );
738 }
739
740 if ( $t->hasFragment() ) {
741 # Strip the fragment to avoid various odd effects
742 $this->mTitle = $t->createFragmentTarget( '' );
743 } else {
744 $this->mTitle = $t;
745 }
746 }
747
748 /**
749 * Accessor for the Title object
750 *
751 * @return Title
752 */
753 public function getTitle() {
754 return $this->mTitle;
755 }
756
757 /**
758 * Accessor/mutator for the Title object
759 *
760 * @param Title $x Title object or null to just get the current one
761 * @return Title
762 */
763 public function Title( $x = null ) {
764 return wfSetVar( $this->mTitle, $x );
765 }
766
767 /**
768 * Set the output type
769 *
770 * @param int $ot New value
771 */
772 public function setOutputType( $ot ) {
773 $this->mOutputType = $ot;
774 # Shortcut alias
775 $this->ot = [
776 'html' => $ot == self::OT_HTML,
777 'wiki' => $ot == self::OT_WIKI,
778 'pre' => $ot == self::OT_PREPROCESS,
779 'plain' => $ot == self::OT_PLAIN,
780 ];
781 }
782
783 /**
784 * Accessor/mutator for the output type
785 *
786 * @param int|null $x New value or null to just get the current one
787 * @return int
788 */
789 public function OutputType( $x = null ) {
790 return wfSetVar( $this->mOutputType, $x );
791 }
792
793 /**
794 * Get the ParserOutput object
795 *
796 * @return ParserOutput
797 */
798 public function getOutput() {
799 return $this->mOutput;
800 }
801
802 /**
803 * Get the ParserOptions object
804 *
805 * @return ParserOptions
806 */
807 public function getOptions() {
808 return $this->mOptions;
809 }
810
811 /**
812 * Accessor/mutator for the ParserOptions object
813 *
814 * @param ParserOptions $x New value or null to just get the current one
815 * @return ParserOptions Current ParserOptions object
816 */
817 public function Options( $x = null ) {
818 return wfSetVar( $this->mOptions, $x );
819 }
820
821 /**
822 * @return int
823 */
824 public function nextLinkID() {
825 return $this->mLinkID++;
826 }
827
828 /**
829 * @param int $id
830 */
831 public function setLinkID( $id ) {
832 $this->mLinkID = $id;
833 }
834
835 /**
836 * Get a language object for use in parser functions such as {{FORMATNUM:}}
837 * @return Language
838 */
839 public function getFunctionLang() {
840 return $this->getTargetLanguage();
841 }
842
843 /**
844 * Get the target language for the content being parsed. This is usually the
845 * language that the content is in.
846 *
847 * @since 1.19
848 *
849 * @throws MWException
850 * @return Language
851 */
852 public function getTargetLanguage() {
853 $target = $this->mOptions->getTargetLanguage();
854
855 if ( $target !== null ) {
856 return $target;
857 } elseif ( $this->mOptions->getInterfaceMessage() ) {
858 return $this->mOptions->getUserLangObj();
859 } elseif ( is_null( $this->mTitle ) ) {
860 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
861 }
862
863 return $this->mTitle->getPageLanguage();
864 }
865
866 /**
867 * Get the language object for language conversion
868 * @return Language|null
869 */
870 public function getConverterLanguage() {
871 return $this->getTargetLanguage();
872 }
873
874 /**
875 * Get a User object either from $this->mUser, if set, or from the
876 * ParserOptions object otherwise
877 *
878 * @return User
879 */
880 public function getUser() {
881 if ( !is_null( $this->mUser ) ) {
882 return $this->mUser;
883 }
884 return $this->mOptions->getUser();
885 }
886
887 /**
888 * Get a preprocessor object
889 *
890 * @return Preprocessor
891 */
892 public function getPreprocessor() {
893 if ( !isset( $this->mPreprocessor ) ) {
894 $class = $this->mPreprocessorClass;
895 $this->mPreprocessor = new $class( $this );
896 }
897 return $this->mPreprocessor;
898 }
899
900 /**
901 * Replaces all occurrences of HTML-style comments and the given tags
902 * in the text with a random marker and returns the next text. The output
903 * parameter $matches will be an associative array filled with data in
904 * the form:
905 *
906 * @code
907 * 'UNIQ-xxxxx' => array(
908 * 'element',
909 * 'tag content',
910 * array( 'param' => 'x' ),
911 * '<element param="x">tag content</element>' ) )
912 * @endcode
913 *
914 * @param array $elements List of element names. Comments are always extracted.
915 * @param string $text Source text string.
916 * @param array $matches Out parameter, Array: extracted tags
917 * @param string|null $uniq_prefix
918 * @return string Stripped text
919 * @since 1.26 The uniq_prefix argument is deprecated.
920 */
921 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
922 if ( $uniq_prefix !== null ) {
923 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
924 }
925 static $n = 1;
926 $stripped = '';
927 $matches = [];
928
929 $taglist = implode( '|', $elements );
930 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
931
932 while ( $text != '' ) {
933 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
934 $stripped .= $p[0];
935 if ( count( $p ) < 5 ) {
936 break;
937 }
938 if ( count( $p ) > 5 ) {
939 # comment
940 $element = $p[4];
941 $attributes = '';
942 $close = '';
943 $inside = $p[5];
944 } else {
945 # tag
946 $element = $p[1];
947 $attributes = $p[2];
948 $close = $p[3];
949 $inside = $p[4];
950 }
951
952 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
953 $stripped .= $marker;
954
955 if ( $close === '/>' ) {
956 # Empty element tag, <tag />
957 $content = null;
958 $text = $inside;
959 $tail = null;
960 } else {
961 if ( $element === '!--' ) {
962 $end = '/(-->)/';
963 } else {
964 $end = "/(<\\/$element\\s*>)/i";
965 }
966 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
967 $content = $q[0];
968 if ( count( $q ) < 3 ) {
969 # No end tag -- let it run out to the end of the text.
970 $tail = '';
971 $text = '';
972 } else {
973 $tail = $q[1];
974 $text = $q[2];
975 }
976 }
977
978 $matches[$marker] = [ $element,
979 $content,
980 Sanitizer::decodeTagAttributes( $attributes ),
981 "<$element$attributes$close$content$tail" ];
982 }
983 return $stripped;
984 }
985
986 /**
987 * Get a list of strippable XML-like elements
988 *
989 * @return array
990 */
991 public function getStripList() {
992 return $this->mStripList;
993 }
994
995 /**
996 * Add an item to the strip state
997 * Returns the unique tag which must be inserted into the stripped text
998 * The tag will be replaced with the original text in unstrip()
999 *
1000 * @param string $text
1001 *
1002 * @return string
1003 */
1004 public function insertStripItem( $text ) {
1005 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1006 $this->mMarkerIndex++;
1007 $this->mStripState->addGeneral( $marker, $text );
1008 return $marker;
1009 }
1010
1011 /**
1012 * parse the wiki syntax used to render tables
1013 *
1014 * @private
1015 * @param string $text
1016 * @return string
1017 */
1018 public function doTableStuff( $text ) {
1019
1020 $lines = StringUtils::explode( "\n", $text );
1021 $out = '';
1022 $td_history = []; # Is currently a td tag open?
1023 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1024 $tr_history = []; # Is currently a tr tag open?
1025 $tr_attributes = []; # history of tr attributes
1026 $has_opened_tr = []; # Did this table open a <tr> element?
1027 $indent_level = 0; # indent level of the table
1028
1029 foreach ( $lines as $outLine ) {
1030 $line = trim( $outLine );
1031
1032 if ( $line === '' ) { # empty line, go to next line
1033 $out .= $outLine . "\n";
1034 continue;
1035 }
1036
1037 $first_character = $line[0];
1038 $first_two = substr( $line, 0, 2 );
1039 $matches = [];
1040
1041 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1042 # First check if we are starting a new table
1043 $indent_level = strlen( $matches[1] );
1044
1045 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1046 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1047
1048 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1049 array_push( $td_history, false );
1050 array_push( $last_tag_history, '' );
1051 array_push( $tr_history, false );
1052 array_push( $tr_attributes, '' );
1053 array_push( $has_opened_tr, false );
1054 } elseif ( count( $td_history ) == 0 ) {
1055 # Don't do any of the following
1056 $out .= $outLine . "\n";
1057 continue;
1058 } elseif ( $first_two === '|}' ) {
1059 # We are ending a table
1060 $line = '</table>' . substr( $line, 2 );
1061 $last_tag = array_pop( $last_tag_history );
1062
1063 if ( !array_pop( $has_opened_tr ) ) {
1064 $line = "<tr><td></td></tr>{$line}";
1065 }
1066
1067 if ( array_pop( $tr_history ) ) {
1068 $line = "</tr>{$line}";
1069 }
1070
1071 if ( array_pop( $td_history ) ) {
1072 $line = "</{$last_tag}>{$line}";
1073 }
1074 array_pop( $tr_attributes );
1075 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1076 } elseif ( $first_two === '|-' ) {
1077 # Now we have a table row
1078 $line = preg_replace( '#^\|-+#', '', $line );
1079
1080 # Whats after the tag is now only attributes
1081 $attributes = $this->mStripState->unstripBoth( $line );
1082 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1083 array_pop( $tr_attributes );
1084 array_push( $tr_attributes, $attributes );
1085
1086 $line = '';
1087 $last_tag = array_pop( $last_tag_history );
1088 array_pop( $has_opened_tr );
1089 array_push( $has_opened_tr, true );
1090
1091 if ( array_pop( $tr_history ) ) {
1092 $line = '</tr>';
1093 }
1094
1095 if ( array_pop( $td_history ) ) {
1096 $line = "</{$last_tag}>{$line}";
1097 }
1098
1099 $outLine = $line;
1100 array_push( $tr_history, false );
1101 array_push( $td_history, false );
1102 array_push( $last_tag_history, '' );
1103 } elseif ( $first_character === '|'
1104 || $first_character === '!'
1105 || $first_two === '|+'
1106 ) {
1107 # This might be cell elements, td, th or captions
1108 if ( $first_two === '|+' ) {
1109 $first_character = '+';
1110 $line = substr( $line, 2 );
1111 } else {
1112 $line = substr( $line, 1 );
1113 }
1114
1115 // Implies both are valid for table headings.
1116 if ( $first_character === '!' ) {
1117 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1118 }
1119
1120 # Split up multiple cells on the same line.
1121 # FIXME : This can result in improper nesting of tags processed
1122 # by earlier parser steps.
1123 $cells = explode( '||', $line );
1124
1125 $outLine = '';
1126
1127 # Loop through each table cell
1128 foreach ( $cells as $cell ) {
1129 $previous = '';
1130 if ( $first_character !== '+' ) {
1131 $tr_after = array_pop( $tr_attributes );
1132 if ( !array_pop( $tr_history ) ) {
1133 $previous = "<tr{$tr_after}>\n";
1134 }
1135 array_push( $tr_history, true );
1136 array_push( $tr_attributes, '' );
1137 array_pop( $has_opened_tr );
1138 array_push( $has_opened_tr, true );
1139 }
1140
1141 $last_tag = array_pop( $last_tag_history );
1142
1143 if ( array_pop( $td_history ) ) {
1144 $previous = "</{$last_tag}>\n{$previous}";
1145 }
1146
1147 if ( $first_character === '|' ) {
1148 $last_tag = 'td';
1149 } elseif ( $first_character === '!' ) {
1150 $last_tag = 'th';
1151 } elseif ( $first_character === '+' ) {
1152 $last_tag = 'caption';
1153 } else {
1154 $last_tag = '';
1155 }
1156
1157 array_push( $last_tag_history, $last_tag );
1158
1159 # A cell could contain both parameters and data
1160 $cell_data = explode( '|', $cell, 2 );
1161
1162 # Bug 553: Note that a '|' inside an invalid link should not
1163 # be mistaken as delimiting cell parameters
1164 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1165 $cell = "{$previous}<{$last_tag}>{$cell}";
1166 } elseif ( count( $cell_data ) == 1 ) {
1167 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1168 } else {
1169 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1170 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1171 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1172 }
1173
1174 $outLine .= $cell;
1175 array_push( $td_history, true );
1176 }
1177 }
1178 $out .= $outLine . "\n";
1179 }
1180
1181 # Closing open td, tr && table
1182 while ( count( $td_history ) > 0 ) {
1183 if ( array_pop( $td_history ) ) {
1184 $out .= "</td>\n";
1185 }
1186 if ( array_pop( $tr_history ) ) {
1187 $out .= "</tr>\n";
1188 }
1189 if ( !array_pop( $has_opened_tr ) ) {
1190 $out .= "<tr><td></td></tr>\n";
1191 }
1192
1193 $out .= "</table>\n";
1194 }
1195
1196 # Remove trailing line-ending (b/c)
1197 if ( substr( $out, -1 ) === "\n" ) {
1198 $out = substr( $out, 0, -1 );
1199 }
1200
1201 # special case: don't return empty table
1202 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1203 $out = '';
1204 }
1205
1206 return $out;
1207 }
1208
1209 /**
1210 * Helper function for parse() that transforms wiki markup into half-parsed
1211 * HTML. Only called for $mOutputType == self::OT_HTML.
1212 *
1213 * @private
1214 *
1215 * @param string $text The text to parse
1216 * @param bool $isMain Whether this is being called from the main parse() function
1217 * @param PPFrame|bool $frame A pre-processor frame
1218 *
1219 * @return string
1220 */
1221 public function internalParse( $text, $isMain = true, $frame = false ) {
1222
1223 $origText = $text;
1224
1225 # Hook to suspend the parser in this state
1226 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1227 return $text;
1228 }
1229
1230 # if $frame is provided, then use $frame for replacing any variables
1231 if ( $frame ) {
1232 # use frame depth to infer how include/noinclude tags should be handled
1233 # depth=0 means this is the top-level document; otherwise it's an included document
1234 if ( !$frame->depth ) {
1235 $flag = 0;
1236 } else {
1237 $flag = Parser::PTD_FOR_INCLUSION;
1238 }
1239 $dom = $this->preprocessToDom( $text, $flag );
1240 $text = $frame->expand( $dom );
1241 } else {
1242 # if $frame is not provided, then use old-style replaceVariables
1243 $text = $this->replaceVariables( $text );
1244 }
1245
1246 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1247 $text = Sanitizer::removeHTMLtags(
1248 $text,
1249 [ &$this, 'attributeStripCallback' ],
1250 false,
1251 array_keys( $this->mTransparentTagHooks )
1252 );
1253 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1254
1255 # Tables need to come after variable replacement for things to work
1256 # properly; putting them before other transformations should keep
1257 # exciting things like link expansions from showing up in surprising
1258 # places.
1259 $text = $this->doTableStuff( $text );
1260
1261 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1262
1263 $text = $this->doDoubleUnderscore( $text );
1264
1265 $text = $this->doHeadings( $text );
1266 $text = $this->replaceInternalLinks( $text );
1267 $text = $this->doAllQuotes( $text );
1268 $text = $this->replaceExternalLinks( $text );
1269
1270 # replaceInternalLinks may sometimes leave behind
1271 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1272 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1273
1274 $text = $this->doMagicLinks( $text );
1275 $text = $this->formatHeadings( $text, $origText, $isMain );
1276
1277 return $text;
1278 }
1279
1280 /**
1281 * Helper function for parse() that transforms half-parsed HTML into fully
1282 * parsed HTML.
1283 *
1284 * @param string $text
1285 * @param bool $isMain
1286 * @param bool $linestart
1287 * @return string
1288 */
1289 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1290 $text = $this->mStripState->unstripGeneral( $text );
1291
1292 if ( $isMain ) {
1293 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1294 }
1295
1296 # Clean up special characters, only run once, next-to-last before doBlockLevels
1297 $fixtags = [
1298 # french spaces, last one Guillemet-left
1299 # only if there is something before the space
1300 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1301 # french spaces, Guillemet-right
1302 '/(\\302\\253) /' => '\\1&#160;',
1303 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1304 ];
1305 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1306
1307 $text = $this->doBlockLevels( $text, $linestart );
1308
1309 $this->replaceLinkHolders( $text );
1310
1311 /**
1312 * The input doesn't get language converted if
1313 * a) It's disabled
1314 * b) Content isn't converted
1315 * c) It's a conversion table
1316 * d) it is an interface message (which is in the user language)
1317 */
1318 if ( !( $this->mOptions->getDisableContentConversion()
1319 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1320 ) {
1321 if ( !$this->mOptions->getInterfaceMessage() ) {
1322 # The position of the convert() call should not be changed. it
1323 # assumes that the links are all replaced and the only thing left
1324 # is the <nowiki> mark.
1325 $text = $this->getConverterLanguage()->convert( $text );
1326 }
1327 }
1328
1329 $text = $this->mStripState->unstripNoWiki( $text );
1330
1331 if ( $isMain ) {
1332 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1333 }
1334
1335 $text = $this->replaceTransparentTags( $text );
1336 $text = $this->mStripState->unstripGeneral( $text );
1337
1338 $text = Sanitizer::normalizeCharReferences( $text );
1339
1340 if ( MWTidy::isEnabled() && $this->mOptions->getTidy() ) {
1341 $text = MWTidy::tidy( $text );
1342 $this->mOutput->addModuleStyles( MWTidy::getModuleStyles() );
1343 } else {
1344 # attempt to sanitize at least some nesting problems
1345 # (bug #2702 and quite a few others)
1346 $tidyregs = [
1347 # ''Something [http://www.cool.com cool''] -->
1348 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1349 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1350 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1351 # fix up an anchor inside another anchor, only
1352 # at least for a single single nested link (bug 3695)
1353 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1354 '\\1\\2</a>\\3</a>\\1\\4</a>',
1355 # fix div inside inline elements- doBlockLevels won't wrap a line which
1356 # contains a div, so fix it up here; replace
1357 # div with escaped text
1358 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1359 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1360 # remove empty italic or bold tag pairs, some
1361 # introduced by rules above
1362 '/<([bi])><\/\\1>/' => '',
1363 ];
1364
1365 $text = preg_replace(
1366 array_keys( $tidyregs ),
1367 array_values( $tidyregs ),
1368 $text );
1369 }
1370
1371 if ( $isMain ) {
1372 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1373 }
1374
1375 return $text;
1376 }
1377
1378 /**
1379 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1380 * magic external links.
1381 *
1382 * DML
1383 * @private
1384 *
1385 * @param string $text
1386 *
1387 * @return string
1388 */
1389 public function doMagicLinks( $text ) {
1390 $prots = wfUrlProtocolsWithoutProtRel();
1391 $urlChar = self::EXT_LINK_URL_CLASS;
1392 $addr = self::EXT_LINK_ADDR;
1393 $space = self::SPACE_NOT_NL; # non-newline space
1394 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1395 $spaces = "$space++"; # possessive match of 1 or more spaces
1396 $text = preg_replace_callback(
1397 '!(?: # Start cases
1398 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1399 (<.*?>) | # m[2]: Skip stuff inside
1400 # HTML elements' . "
1401 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1402 # m[4]: Post-protocol path
1403 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1404 ([0-9]+)\b |
1405 \bISBN $spaces ( # m[6]: ISBN, capture number
1406 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1407 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1408 [0-9Xx] # check digit
1409 )\b
1410 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1411 return $text;
1412 }
1413
1414 /**
1415 * @throws MWException
1416 * @param array $m
1417 * @return HTML|string
1418 */
1419 public function magicLinkCallback( $m ) {
1420 if ( isset( $m[1] ) && $m[1] !== '' ) {
1421 # Skip anchor
1422 return $m[0];
1423 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1424 # Skip HTML element
1425 return $m[0];
1426 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1427 # Free external link
1428 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1429 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1430 # RFC or PMID
1431 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1432 $keyword = 'RFC';
1433 $urlmsg = 'rfcurl';
1434 $cssClass = 'mw-magiclink-rfc';
1435 $id = $m[5];
1436 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1437 $keyword = 'PMID';
1438 $urlmsg = 'pubmedurl';
1439 $cssClass = 'mw-magiclink-pmid';
1440 $id = $m[5];
1441 } else {
1442 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1443 substr( $m[0], 0, 20 ) . '"' );
1444 }
1445 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1446 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass );
1447 } elseif ( isset( $m[6] ) && $m[6] !== '' ) {
1448 # ISBN
1449 $isbn = $m[6];
1450 $space = self::SPACE_NOT_NL; # non-newline space
1451 $isbn = preg_replace( "/$space/", ' ', $isbn );
1452 $num = strtr( $isbn, [
1453 '-' => '',
1454 ' ' => '',
1455 'x' => 'X',
1456 ] );
1457 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1458 return '<a href="' .
1459 htmlspecialchars( $titleObj->getLocalURL() ) .
1460 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1461 } else {
1462 return $m[0];
1463 }
1464 }
1465
1466 /**
1467 * Make a free external link, given a user-supplied URL
1468 *
1469 * @param string $url
1470 * @param int $numPostProto
1471 * The number of characters after the protocol.
1472 * @return string HTML
1473 * @private
1474 */
1475 public function makeFreeExternalLink( $url, $numPostProto ) {
1476 $trail = '';
1477
1478 # The characters '<' and '>' (which were escaped by
1479 # removeHTMLtags()) should not be included in
1480 # URLs, per RFC 2396.
1481 # Make &nbsp; terminate a URL as well (bug T84937)
1482 $m2 = [];
1483 if ( preg_match(
1484 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1485 $url,
1486 $m2,
1487 PREG_OFFSET_CAPTURE
1488 ) ) {
1489 $trail = substr( $url, $m2[0][1] ) . $trail;
1490 $url = substr( $url, 0, $m2[0][1] );
1491 }
1492
1493 # Move trailing punctuation to $trail
1494 $sep = ',;\.:!?';
1495 # If there is no left bracket, then consider right brackets fair game too
1496 if ( strpos( $url, '(' ) === false ) {
1497 $sep .= ')';
1498 }
1499
1500 $urlRev = strrev( $url );
1501 $numSepChars = strspn( $urlRev, $sep );
1502 # Don't break a trailing HTML entity by moving the ; into $trail
1503 # This is in hot code, so use substr_compare to avoid having to
1504 # create a new string object for the comparison
1505 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1506 # more optimization: instead of running preg_match with a $
1507 # anchor, which can be slow, do the match on the reversed
1508 # string starting at the desired offset.
1509 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1510 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1511 $numSepChars--;
1512 }
1513 }
1514 if ( $numSepChars ) {
1515 $trail = substr( $url, -$numSepChars ) . $trail;
1516 $url = substr( $url, 0, -$numSepChars );
1517 }
1518
1519 # Verify that we still have a real URL after trail removal, and
1520 # not just lone protocol
1521 if ( strlen( $trail ) >= $numPostProto ) {
1522 return $url . $trail;
1523 }
1524
1525 $url = Sanitizer::cleanUrl( $url );
1526
1527 # Is this an external image?
1528 $text = $this->maybeMakeExternalImage( $url );
1529 if ( $text === false ) {
1530 # Not an image, make a link
1531 $text = Linker::makeExternalLink( $url,
1532 $this->getConverterLanguage()->markNoConversion( $url, true ),
1533 true, 'free',
1534 $this->getExternalLinkAttribs( $url ) );
1535 # Register it in the output object...
1536 # Replace unnecessary URL escape codes with their equivalent characters
1537 $pasteurized = self::normalizeLinkUrl( $url );
1538 $this->mOutput->addExternalLink( $pasteurized );
1539 }
1540 return $text . $trail;
1541 }
1542
1543 /**
1544 * Parse headers and return html
1545 *
1546 * @private
1547 *
1548 * @param string $text
1549 *
1550 * @return string
1551 */
1552 public function doHeadings( $text ) {
1553 for ( $i = 6; $i >= 1; --$i ) {
1554 $h = str_repeat( '=', $i );
1555 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1556 }
1557 return $text;
1558 }
1559
1560 /**
1561 * Replace single quotes with HTML markup
1562 * @private
1563 *
1564 * @param string $text
1565 *
1566 * @return string The altered text
1567 */
1568 public function doAllQuotes( $text ) {
1569 $outtext = '';
1570 $lines = StringUtils::explode( "\n", $text );
1571 foreach ( $lines as $line ) {
1572 $outtext .= $this->doQuotes( $line ) . "\n";
1573 }
1574 $outtext = substr( $outtext, 0, -1 );
1575 return $outtext;
1576 }
1577
1578 /**
1579 * Helper function for doAllQuotes()
1580 *
1581 * @param string $text
1582 *
1583 * @return string
1584 */
1585 public function doQuotes( $text ) {
1586 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1587 $countarr = count( $arr );
1588 if ( $countarr == 1 ) {
1589 return $text;
1590 }
1591
1592 // First, do some preliminary work. This may shift some apostrophes from
1593 // being mark-up to being text. It also counts the number of occurrences
1594 // of bold and italics mark-ups.
1595 $numbold = 0;
1596 $numitalics = 0;
1597 for ( $i = 1; $i < $countarr; $i += 2 ) {
1598 $thislen = strlen( $arr[$i] );
1599 // If there are ever four apostrophes, assume the first is supposed to
1600 // be text, and the remaining three constitute mark-up for bold text.
1601 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1602 if ( $thislen == 4 ) {
1603 $arr[$i - 1] .= "'";
1604 $arr[$i] = "'''";
1605 $thislen = 3;
1606 } elseif ( $thislen > 5 ) {
1607 // If there are more than 5 apostrophes in a row, assume they're all
1608 // text except for the last 5.
1609 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1610 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1611 $arr[$i] = "'''''";
1612 $thislen = 5;
1613 }
1614 // Count the number of occurrences of bold and italics mark-ups.
1615 if ( $thislen == 2 ) {
1616 $numitalics++;
1617 } elseif ( $thislen == 3 ) {
1618 $numbold++;
1619 } elseif ( $thislen == 5 ) {
1620 $numitalics++;
1621 $numbold++;
1622 }
1623 }
1624
1625 // If there is an odd number of both bold and italics, it is likely
1626 // that one of the bold ones was meant to be an apostrophe followed
1627 // by italics. Which one we cannot know for certain, but it is more
1628 // likely to be one that has a single-letter word before it.
1629 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1630 $firstsingleletterword = -1;
1631 $firstmultiletterword = -1;
1632 $firstspace = -1;
1633 for ( $i = 1; $i < $countarr; $i += 2 ) {
1634 if ( strlen( $arr[$i] ) == 3 ) {
1635 $x1 = substr( $arr[$i - 1], -1 );
1636 $x2 = substr( $arr[$i - 1], -2, 1 );
1637 if ( $x1 === ' ' ) {
1638 if ( $firstspace == -1 ) {
1639 $firstspace = $i;
1640 }
1641 } elseif ( $x2 === ' ' ) {
1642 $firstsingleletterword = $i;
1643 // if $firstsingleletterword is set, we don't
1644 // look at the other options, so we can bail early.
1645 break;
1646 } else {
1647 if ( $firstmultiletterword == -1 ) {
1648 $firstmultiletterword = $i;
1649 }
1650 }
1651 }
1652 }
1653
1654 // If there is a single-letter word, use it!
1655 if ( $firstsingleletterword > -1 ) {
1656 $arr[$firstsingleletterword] = "''";
1657 $arr[$firstsingleletterword - 1] .= "'";
1658 } elseif ( $firstmultiletterword > -1 ) {
1659 // If not, but there's a multi-letter word, use that one.
1660 $arr[$firstmultiletterword] = "''";
1661 $arr[$firstmultiletterword - 1] .= "'";
1662 } elseif ( $firstspace > -1 ) {
1663 // ... otherwise use the first one that has neither.
1664 // (notice that it is possible for all three to be -1 if, for example,
1665 // there is only one pentuple-apostrophe in the line)
1666 $arr[$firstspace] = "''";
1667 $arr[$firstspace - 1] .= "'";
1668 }
1669 }
1670
1671 // Now let's actually convert our apostrophic mush to HTML!
1672 $output = '';
1673 $buffer = '';
1674 $state = '';
1675 $i = 0;
1676 foreach ( $arr as $r ) {
1677 if ( ( $i % 2 ) == 0 ) {
1678 if ( $state === 'both' ) {
1679 $buffer .= $r;
1680 } else {
1681 $output .= $r;
1682 }
1683 } else {
1684 $thislen = strlen( $r );
1685 if ( $thislen == 2 ) {
1686 if ( $state === 'i' ) {
1687 $output .= '</i>';
1688 $state = '';
1689 } elseif ( $state === 'bi' ) {
1690 $output .= '</i>';
1691 $state = 'b';
1692 } elseif ( $state === 'ib' ) {
1693 $output .= '</b></i><b>';
1694 $state = 'b';
1695 } elseif ( $state === 'both' ) {
1696 $output .= '<b><i>' . $buffer . '</i>';
1697 $state = 'b';
1698 } else { // $state can be 'b' or ''
1699 $output .= '<i>';
1700 $state .= 'i';
1701 }
1702 } elseif ( $thislen == 3 ) {
1703 if ( $state === 'b' ) {
1704 $output .= '</b>';
1705 $state = '';
1706 } elseif ( $state === 'bi' ) {
1707 $output .= '</i></b><i>';
1708 $state = 'i';
1709 } elseif ( $state === 'ib' ) {
1710 $output .= '</b>';
1711 $state = 'i';
1712 } elseif ( $state === 'both' ) {
1713 $output .= '<i><b>' . $buffer . '</b>';
1714 $state = 'i';
1715 } else { // $state can be 'i' or ''
1716 $output .= '<b>';
1717 $state .= 'b';
1718 }
1719 } elseif ( $thislen == 5 ) {
1720 if ( $state === 'b' ) {
1721 $output .= '</b><i>';
1722 $state = 'i';
1723 } elseif ( $state === 'i' ) {
1724 $output .= '</i><b>';
1725 $state = 'b';
1726 } elseif ( $state === 'bi' ) {
1727 $output .= '</i></b>';
1728 $state = '';
1729 } elseif ( $state === 'ib' ) {
1730 $output .= '</b></i>';
1731 $state = '';
1732 } elseif ( $state === 'both' ) {
1733 $output .= '<i><b>' . $buffer . '</b></i>';
1734 $state = '';
1735 } else { // ($state == '')
1736 $buffer = '';
1737 $state = 'both';
1738 }
1739 }
1740 }
1741 $i++;
1742 }
1743 // Now close all remaining tags. Notice that the order is important.
1744 if ( $state === 'b' || $state === 'ib' ) {
1745 $output .= '</b>';
1746 }
1747 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1748 $output .= '</i>';
1749 }
1750 if ( $state === 'bi' ) {
1751 $output .= '</b>';
1752 }
1753 // There might be lonely ''''', so make sure we have a buffer
1754 if ( $state === 'both' && $buffer ) {
1755 $output .= '<b><i>' . $buffer . '</i></b>';
1756 }
1757 return $output;
1758 }
1759
1760 /**
1761 * Replace external links (REL)
1762 *
1763 * Note: this is all very hackish and the order of execution matters a lot.
1764 * Make sure to run tests/parserTests.php if you change this code.
1765 *
1766 * @private
1767 *
1768 * @param string $text
1769 *
1770 * @throws MWException
1771 * @return string
1772 */
1773 public function replaceExternalLinks( $text ) {
1774
1775 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1776 if ( $bits === false ) {
1777 throw new MWException( "PCRE needs to be compiled with "
1778 . "--enable-unicode-properties in order for MediaWiki to function" );
1779 }
1780 $s = array_shift( $bits );
1781
1782 $i = 0;
1783 while ( $i < count( $bits ) ) {
1784 $url = $bits[$i++];
1785 $i++; // protocol
1786 $text = $bits[$i++];
1787 $trail = $bits[$i++];
1788
1789 # The characters '<' and '>' (which were escaped by
1790 # removeHTMLtags()) should not be included in
1791 # URLs, per RFC 2396.
1792 $m2 = [];
1793 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1794 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1795 $url = substr( $url, 0, $m2[0][1] );
1796 }
1797
1798 # If the link text is an image URL, replace it with an <img> tag
1799 # This happened by accident in the original parser, but some people used it extensively
1800 $img = $this->maybeMakeExternalImage( $text );
1801 if ( $img !== false ) {
1802 $text = $img;
1803 }
1804
1805 $dtrail = '';
1806
1807 # Set linktype for CSS - if URL==text, link is essentially free
1808 $linktype = ( $text === $url ) ? 'free' : 'text';
1809
1810 # No link text, e.g. [http://domain.tld/some.link]
1811 if ( $text == '' ) {
1812 # Autonumber
1813 $langObj = $this->getTargetLanguage();
1814 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1815 $linktype = 'autonumber';
1816 } else {
1817 # Have link text, e.g. [http://domain.tld/some.link text]s
1818 # Check for trail
1819 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1820 }
1821
1822 $text = $this->getConverterLanguage()->markNoConversion( $text );
1823
1824 $url = Sanitizer::cleanUrl( $url );
1825
1826 # Use the encoded URL
1827 # This means that users can paste URLs directly into the text
1828 # Funny characters like ö aren't valid in URLs anyway
1829 # This was changed in August 2004
1830 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1831 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1832
1833 # Register link in the output object.
1834 # Replace unnecessary URL escape codes with the referenced character
1835 # This prevents spammers from hiding links from the filters
1836 $pasteurized = self::normalizeLinkUrl( $url );
1837 $this->mOutput->addExternalLink( $pasteurized );
1838 }
1839
1840 return $s;
1841 }
1842
1843 /**
1844 * Get the rel attribute for a particular external link.
1845 *
1846 * @since 1.21
1847 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1848 * nofollow if appropriate
1849 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1850 * @return string|null Rel attribute for $url
1851 */
1852 public static function getExternalLinkRel( $url = false, $title = null ) {
1853 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1854 $ns = $title ? $title->getNamespace() : false;
1855 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1856 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1857 ) {
1858 return 'nofollow';
1859 }
1860 return null;
1861 }
1862
1863 /**
1864 * Get an associative array of additional HTML attributes appropriate for a
1865 * particular external link. This currently may include rel => nofollow
1866 * (depending on configuration, namespace, and the URL's domain) and/or a
1867 * target attribute (depending on configuration).
1868 *
1869 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1870 * nofollow if appropriate
1871 * @return array Associative array of HTML attributes
1872 */
1873 public function getExternalLinkAttribs( $url = false ) {
1874 $attribs = [];
1875 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1876
1877 if ( $this->mOptions->getExternalLinkTarget() ) {
1878 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1879 }
1880 return $attribs;
1881 }
1882
1883 /**
1884 * Replace unusual escape codes in a URL with their equivalent characters
1885 *
1886 * @deprecated since 1.24, use normalizeLinkUrl
1887 * @param string $url
1888 * @return string
1889 */
1890 public static function replaceUnusualEscapes( $url ) {
1891 wfDeprecated( __METHOD__, '1.24' );
1892 return self::normalizeLinkUrl( $url );
1893 }
1894
1895 /**
1896 * Replace unusual escape codes in a URL with their equivalent characters
1897 *
1898 * This generally follows the syntax defined in RFC 3986, with special
1899 * consideration for HTTP query strings.
1900 *
1901 * @param string $url
1902 * @return string
1903 */
1904 public static function normalizeLinkUrl( $url ) {
1905 # First, make sure unsafe characters are encoded
1906 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1907 function ( $m ) {
1908 return rawurlencode( $m[0] );
1909 },
1910 $url
1911 );
1912
1913 $ret = '';
1914 $end = strlen( $url );
1915
1916 # Fragment part - 'fragment'
1917 $start = strpos( $url, '#' );
1918 if ( $start !== false && $start < $end ) {
1919 $ret = self::normalizeUrlComponent(
1920 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1921 $end = $start;
1922 }
1923
1924 # Query part - 'query' minus &=+;
1925 $start = strpos( $url, '?' );
1926 if ( $start !== false && $start < $end ) {
1927 $ret = self::normalizeUrlComponent(
1928 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1929 $end = $start;
1930 }
1931
1932 # Scheme and path part - 'pchar'
1933 # (we assume no userinfo or encoded colons in the host)
1934 $ret = self::normalizeUrlComponent(
1935 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1936
1937 return $ret;
1938 }
1939
1940 private static function normalizeUrlComponent( $component, $unsafe ) {
1941 $callback = function ( $matches ) use ( $unsafe ) {
1942 $char = urldecode( $matches[0] );
1943 $ord = ord( $char );
1944 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1945 # Unescape it
1946 return $char;
1947 } else {
1948 # Leave it escaped, but use uppercase for a-f
1949 return strtoupper( $matches[0] );
1950 }
1951 };
1952 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1953 }
1954
1955 /**
1956 * make an image if it's allowed, either through the global
1957 * option, through the exception, or through the on-wiki whitelist
1958 *
1959 * @param string $url
1960 *
1961 * @return string
1962 */
1963 private function maybeMakeExternalImage( $url ) {
1964 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1965 $imagesexception = !empty( $imagesfrom );
1966 $text = false;
1967 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1968 if ( $imagesexception && is_array( $imagesfrom ) ) {
1969 $imagematch = false;
1970 foreach ( $imagesfrom as $match ) {
1971 if ( strpos( $url, $match ) === 0 ) {
1972 $imagematch = true;
1973 break;
1974 }
1975 }
1976 } elseif ( $imagesexception ) {
1977 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1978 } else {
1979 $imagematch = false;
1980 }
1981
1982 if ( $this->mOptions->getAllowExternalImages()
1983 || ( $imagesexception && $imagematch )
1984 ) {
1985 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1986 # Image found
1987 $text = Linker::makeExternalImage( $url );
1988 }
1989 }
1990 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1991 && preg_match( self::EXT_IMAGE_REGEX, $url )
1992 ) {
1993 $whitelist = explode(
1994 "\n",
1995 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
1996 );
1997
1998 foreach ( $whitelist as $entry ) {
1999 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2000 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2001 continue;
2002 }
2003 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2004 # Image matches a whitelist entry
2005 $text = Linker::makeExternalImage( $url );
2006 break;
2007 }
2008 }
2009 }
2010 return $text;
2011 }
2012
2013 /**
2014 * Process [[ ]] wikilinks
2015 *
2016 * @param string $s
2017 *
2018 * @return string Processed text
2019 *
2020 * @private
2021 */
2022 public function replaceInternalLinks( $s ) {
2023 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2024 return $s;
2025 }
2026
2027 /**
2028 * Process [[ ]] wikilinks (RIL)
2029 * @param string $s
2030 * @throws MWException
2031 * @return LinkHolderArray
2032 *
2033 * @private
2034 */
2035 public function replaceInternalLinks2( &$s ) {
2036 global $wgExtraInterlanguageLinkPrefixes;
2037
2038 static $tc = false, $e1, $e1_img;
2039 # the % is needed to support urlencoded titles as well
2040 if ( !$tc ) {
2041 $tc = Title::legalChars() . '#%';
2042 # Match a link having the form [[namespace:link|alternate]]trail
2043 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2044 # Match cases where there is no "]]", which might still be images
2045 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2046 }
2047
2048 $holders = new LinkHolderArray( $this );
2049
2050 # split the entire text string on occurrences of [[
2051 $a = StringUtils::explode( '[[', ' ' . $s );
2052 # get the first element (all text up to first [[), and remove the space we added
2053 $s = $a->current();
2054 $a->next();
2055 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2056 $s = substr( $s, 1 );
2057
2058 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2059 $e2 = null;
2060 if ( $useLinkPrefixExtension ) {
2061 # Match the end of a line for a word that's not followed by whitespace,
2062 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2063 global $wgContLang;
2064 $charset = $wgContLang->linkPrefixCharset();
2065 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2066 }
2067
2068 if ( is_null( $this->mTitle ) ) {
2069 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2070 }
2071 $nottalk = !$this->mTitle->isTalkPage();
2072
2073 if ( $useLinkPrefixExtension ) {
2074 $m = [];
2075 if ( preg_match( $e2, $s, $m ) ) {
2076 $first_prefix = $m[2];
2077 } else {
2078 $first_prefix = false;
2079 }
2080 } else {
2081 $prefix = '';
2082 }
2083
2084 $useSubpages = $this->areSubpagesAllowed();
2085
2086 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2087 # Loop for each link
2088 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2089 // @codingStandardsIgnoreEnd
2090
2091 # Check for excessive memory usage
2092 if ( $holders->isBig() ) {
2093 # Too big
2094 # Do the existence check, replace the link holders and clear the array
2095 $holders->replace( $s );
2096 $holders->clear();
2097 }
2098
2099 if ( $useLinkPrefixExtension ) {
2100 if ( preg_match( $e2, $s, $m ) ) {
2101 $prefix = $m[2];
2102 $s = $m[1];
2103 } else {
2104 $prefix = '';
2105 }
2106 # first link
2107 if ( $first_prefix ) {
2108 $prefix = $first_prefix;
2109 $first_prefix = false;
2110 }
2111 }
2112
2113 $might_be_img = false;
2114
2115 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2116 $text = $m[2];
2117 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2118 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2119 # the real problem is with the $e1 regex
2120 # See bug 1300.
2121 # Still some problems for cases where the ] is meant to be outside punctuation,
2122 # and no image is in sight. See bug 2095.
2123 if ( $text !== ''
2124 && substr( $m[3], 0, 1 ) === ']'
2125 && strpos( $text, '[' ) !== false
2126 ) {
2127 $text .= ']'; # so that replaceExternalLinks($text) works later
2128 $m[3] = substr( $m[3], 1 );
2129 }
2130 # fix up urlencoded title texts
2131 if ( strpos( $m[1], '%' ) !== false ) {
2132 # Should anchors '#' also be rejected?
2133 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2134 }
2135 $trail = $m[3];
2136 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2137 # Invalid, but might be an image with a link in its caption
2138 $might_be_img = true;
2139 $text = $m[2];
2140 if ( strpos( $m[1], '%' ) !== false ) {
2141 $m[1] = rawurldecode( $m[1] );
2142 }
2143 $trail = "";
2144 } else { # Invalid form; output directly
2145 $s .= $prefix . '[[' . $line;
2146 continue;
2147 }
2148
2149 $origLink = $m[1];
2150
2151 # Don't allow internal links to pages containing
2152 # PROTO: where PROTO is a valid URL protocol; these
2153 # should be external links.
2154 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2155 $s .= $prefix . '[[' . $line;
2156 continue;
2157 }
2158
2159 # Make subpage if necessary
2160 if ( $useSubpages ) {
2161 $link = $this->maybeDoSubpageLink( $origLink, $text );
2162 } else {
2163 $link = $origLink;
2164 }
2165
2166 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2167 if ( !$noforce ) {
2168 # Strip off leading ':'
2169 $link = substr( $link, 1 );
2170 }
2171
2172 $unstrip = $this->mStripState->unstripNoWiki( $link );
2173 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2174 if ( $nt === null ) {
2175 $s .= $prefix . '[[' . $line;
2176 continue;
2177 }
2178
2179 $ns = $nt->getNamespace();
2180 $iw = $nt->getInterwiki();
2181
2182 if ( $might_be_img ) { # if this is actually an invalid link
2183 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2184 $found = false;
2185 while ( true ) {
2186 # look at the next 'line' to see if we can close it there
2187 $a->next();
2188 $next_line = $a->current();
2189 if ( $next_line === false || $next_line === null ) {
2190 break;
2191 }
2192 $m = explode( ']]', $next_line, 3 );
2193 if ( count( $m ) == 3 ) {
2194 # the first ]] closes the inner link, the second the image
2195 $found = true;
2196 $text .= "[[{$m[0]}]]{$m[1]}";
2197 $trail = $m[2];
2198 break;
2199 } elseif ( count( $m ) == 2 ) {
2200 # if there's exactly one ]] that's fine, we'll keep looking
2201 $text .= "[[{$m[0]}]]{$m[1]}";
2202 } else {
2203 # if $next_line is invalid too, we need look no further
2204 $text .= '[[' . $next_line;
2205 break;
2206 }
2207 }
2208 if ( !$found ) {
2209 # we couldn't find the end of this imageLink, so output it raw
2210 # but don't ignore what might be perfectly normal links in the text we've examined
2211 $holders->merge( $this->replaceInternalLinks2( $text ) );
2212 $s .= "{$prefix}[[$link|$text";
2213 # note: no $trail, because without an end, there *is* no trail
2214 continue;
2215 }
2216 } else { # it's not an image, so output it raw
2217 $s .= "{$prefix}[[$link|$text";
2218 # note: no $trail, because without an end, there *is* no trail
2219 continue;
2220 }
2221 }
2222
2223 $wasblank = ( $text == '' );
2224 if ( $wasblank ) {
2225 $text = $link;
2226 } else {
2227 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2228 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2229 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2230 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2231 $text = $this->doQuotes( $text );
2232 }
2233
2234 # Link not escaped by : , create the various objects
2235 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2236 # Interwikis
2237 if (
2238 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2239 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2240 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2241 )
2242 ) {
2243 # Bug 24502: filter duplicates
2244 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2245 $this->mLangLinkLanguages[$iw] = true;
2246 $this->mOutput->addLanguageLink( $nt->getFullText() );
2247 }
2248
2249 $s = rtrim( $s . $prefix );
2250 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2251 continue;
2252 }
2253
2254 if ( $ns == NS_FILE ) {
2255 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2256 if ( $wasblank ) {
2257 # if no parameters were passed, $text
2258 # becomes something like "File:Foo.png",
2259 # which we don't want to pass on to the
2260 # image generator
2261 $text = '';
2262 } else {
2263 # recursively parse links inside the image caption
2264 # actually, this will parse them in any other parameters, too,
2265 # but it might be hard to fix that, and it doesn't matter ATM
2266 $text = $this->replaceExternalLinks( $text );
2267 $holders->merge( $this->replaceInternalLinks2( $text ) );
2268 }
2269 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2270 $s .= $prefix . $this->armorLinks(
2271 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2272 } else {
2273 $s .= $prefix . $trail;
2274 }
2275 continue;
2276 }
2277
2278 if ( $ns == NS_CATEGORY ) {
2279 $s = rtrim( $s . "\n" ); # bug 87
2280
2281 if ( $wasblank ) {
2282 $sortkey = $this->getDefaultSort();
2283 } else {
2284 $sortkey = $text;
2285 }
2286 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2287 $sortkey = str_replace( "\n", '', $sortkey );
2288 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2289 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2290
2291 /**
2292 * Strip the whitespace Category links produce, see bug 87
2293 */
2294 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2295
2296 continue;
2297 }
2298 }
2299
2300 # Self-link checking. For some languages, variants of the title are checked in
2301 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2302 # for linking to a different variant.
2303 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2304 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2305 continue;
2306 }
2307
2308 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2309 # @todo FIXME: Should do batch file existence checks, see comment below
2310 if ( $ns == NS_MEDIA ) {
2311 # Give extensions a chance to select the file revision for us
2312 $options = [];
2313 $descQuery = false;
2314 Hooks::run( 'BeforeParserFetchFileAndTitle',
2315 [ $this, $nt, &$options, &$descQuery ] );
2316 # Fetch and register the file (file title may be different via hooks)
2317 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2318 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2319 $s .= $prefix . $this->armorLinks(
2320 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2321 continue;
2322 }
2323
2324 # Some titles, such as valid special pages or files in foreign repos, should
2325 # be shown as bluelinks even though they're not included in the page table
2326 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2327 # batch file existence checks for NS_FILE and NS_MEDIA
2328 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2329 $this->mOutput->addLink( $nt );
2330 $s .= $this->makeKnownLinkHolder( $nt, $text, [], $trail, $prefix );
2331 } else {
2332 # Links will be added to the output link list after checking
2333 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2334 }
2335 }
2336 return $holders;
2337 }
2338
2339 /**
2340 * Render a forced-blue link inline; protect against double expansion of
2341 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2342 * Since this little disaster has to split off the trail text to avoid
2343 * breaking URLs in the following text without breaking trails on the
2344 * wiki links, it's been made into a horrible function.
2345 *
2346 * @param Title $nt
2347 * @param string $text
2348 * @param array|string $query
2349 * @param string $trail
2350 * @param string $prefix
2351 * @return string HTML-wikitext mix oh yuck
2352 */
2353 public function makeKnownLinkHolder( $nt, $text = '', $query = [], $trail = '', $prefix = '' ) {
2354 list( $inside, $trail ) = Linker::splitTrail( $trail );
2355
2356 if ( is_string( $query ) ) {
2357 $query = wfCgiToArray( $query );
2358 }
2359 if ( $text == '' ) {
2360 $text = htmlspecialchars( $nt->getPrefixedText() );
2361 }
2362
2363 $link = Linker::linkKnown( $nt, "$prefix$text$inside", [], $query );
2364
2365 return $this->armorLinks( $link ) . $trail;
2366 }
2367
2368 /**
2369 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2370 * going to go through further parsing steps before inline URL expansion.
2371 *
2372 * Not needed quite as much as it used to be since free links are a bit
2373 * more sensible these days. But bracketed links are still an issue.
2374 *
2375 * @param string $text More-or-less HTML
2376 * @return string Less-or-more HTML with NOPARSE bits
2377 */
2378 public function armorLinks( $text ) {
2379 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2380 self::MARKER_PREFIX . "NOPARSE$1", $text );
2381 }
2382
2383 /**
2384 * Return true if subpage links should be expanded on this page.
2385 * @return bool
2386 */
2387 public function areSubpagesAllowed() {
2388 # Some namespaces don't allow subpages
2389 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2390 }
2391
2392 /**
2393 * Handle link to subpage if necessary
2394 *
2395 * @param string $target The source of the link
2396 * @param string &$text The link text, modified as necessary
2397 * @return string The full name of the link
2398 * @private
2399 */
2400 public function maybeDoSubpageLink( $target, &$text ) {
2401 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2402 }
2403
2404 /**#@+
2405 * Used by doBlockLevels()
2406 * @private
2407 *
2408 * @return string
2409 */
2410 public function closeParagraph() {
2411 $result = '';
2412 if ( $this->mLastSection != '' ) {
2413 $result = '</' . $this->mLastSection . ">\n";
2414 }
2415 $this->mInPre = false;
2416 $this->mLastSection = '';
2417 return $result;
2418 }
2419
2420 /**
2421 * getCommon() returns the length of the longest common substring
2422 * of both arguments, starting at the beginning of both.
2423 * @private
2424 *
2425 * @param string $st1
2426 * @param string $st2
2427 *
2428 * @return int
2429 */
2430 public function getCommon( $st1, $st2 ) {
2431 $fl = strlen( $st1 );
2432 $shorter = strlen( $st2 );
2433 if ( $fl < $shorter ) {
2434 $shorter = $fl;
2435 }
2436
2437 for ( $i = 0; $i < $shorter; ++$i ) {
2438 if ( $st1[$i] != $st2[$i] ) {
2439 break;
2440 }
2441 }
2442 return $i;
2443 }
2444
2445 /**
2446 * These next three functions open, continue, and close the list
2447 * element appropriate to the prefix character passed into them.
2448 * @private
2449 *
2450 * @param string $char
2451 *
2452 * @return string
2453 */
2454 public function openList( $char ) {
2455 $result = $this->closeParagraph();
2456
2457 if ( '*' === $char ) {
2458 $result .= "<ul><li>";
2459 } elseif ( '#' === $char ) {
2460 $result .= "<ol><li>";
2461 } elseif ( ':' === $char ) {
2462 $result .= "<dl><dd>";
2463 } elseif ( ';' === $char ) {
2464 $result .= "<dl><dt>";
2465 $this->mDTopen = true;
2466 } else {
2467 $result = '<!-- ERR 1 -->';
2468 }
2469
2470 return $result;
2471 }
2472
2473 /**
2474 * TODO: document
2475 * @param string $char
2476 * @private
2477 *
2478 * @return string
2479 */
2480 public function nextItem( $char ) {
2481 if ( '*' === $char || '#' === $char ) {
2482 return "</li>\n<li>";
2483 } elseif ( ':' === $char || ';' === $char ) {
2484 $close = "</dd>\n";
2485 if ( $this->mDTopen ) {
2486 $close = "</dt>\n";
2487 }
2488 if ( ';' === $char ) {
2489 $this->mDTopen = true;
2490 return $close . '<dt>';
2491 } else {
2492 $this->mDTopen = false;
2493 return $close . '<dd>';
2494 }
2495 }
2496 return '<!-- ERR 2 -->';
2497 }
2498
2499 /**
2500 * @todo Document
2501 * @param string $char
2502 * @private
2503 *
2504 * @return string
2505 */
2506 public function closeList( $char ) {
2507 if ( '*' === $char ) {
2508 $text = "</li></ul>";
2509 } elseif ( '#' === $char ) {
2510 $text = "</li></ol>";
2511 } elseif ( ':' === $char ) {
2512 if ( $this->mDTopen ) {
2513 $this->mDTopen = false;
2514 $text = "</dt></dl>";
2515 } else {
2516 $text = "</dd></dl>";
2517 }
2518 } else {
2519 return '<!-- ERR 3 -->';
2520 }
2521 return $text;
2522 }
2523 /**#@-*/
2524
2525 /**
2526 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2527 *
2528 * @param string $text
2529 * @param bool $linestart Whether or not this is at the start of a line.
2530 * @private
2531 * @return string The lists rendered as HTML
2532 */
2533 public function doBlockLevels( $text, $linestart ) {
2534
2535 # Parsing through the text line by line. The main thing
2536 # happening here is handling of block-level elements p, pre,
2537 # and making lists from lines starting with * # : etc.
2538 $textLines = StringUtils::explode( "\n", $text );
2539
2540 $lastPrefix = $output = '';
2541 $this->mDTopen = $inBlockElem = false;
2542 $prefixLength = 0;
2543 $paragraphStack = false;
2544 $inBlockquote = false;
2545
2546 foreach ( $textLines as $oLine ) {
2547 # Fix up $linestart
2548 if ( !$linestart ) {
2549 $output .= $oLine;
2550 $linestart = true;
2551 continue;
2552 }
2553 # * = ul
2554 # # = ol
2555 # ; = dt
2556 # : = dd
2557
2558 $lastPrefixLength = strlen( $lastPrefix );
2559 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2560 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2561 # If not in a <pre> element, scan for and figure out what prefixes are there.
2562 if ( !$this->mInPre ) {
2563 # Multiple prefixes may abut each other for nested lists.
2564 $prefixLength = strspn( $oLine, '*#:;' );
2565 $prefix = substr( $oLine, 0, $prefixLength );
2566
2567 # eh?
2568 # ; and : are both from definition-lists, so they're equivalent
2569 # for the purposes of determining whether or not we need to open/close
2570 # elements.
2571 $prefix2 = str_replace( ';', ':', $prefix );
2572 $t = substr( $oLine, $prefixLength );
2573 $this->mInPre = (bool)$preOpenMatch;
2574 } else {
2575 # Don't interpret any other prefixes in preformatted text
2576 $prefixLength = 0;
2577 $prefix = $prefix2 = '';
2578 $t = $oLine;
2579 }
2580
2581 # List generation
2582 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2583 # Same as the last item, so no need to deal with nesting or opening stuff
2584 $output .= $this->nextItem( substr( $prefix, -1 ) );
2585 $paragraphStack = false;
2586
2587 if ( substr( $prefix, -1 ) === ';' ) {
2588 # The one nasty exception: definition lists work like this:
2589 # ; title : definition text
2590 # So we check for : in the remainder text to split up the
2591 # title and definition, without b0rking links.
2592 $term = $t2 = '';
2593 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2594 $t = $t2;
2595 $output .= $term . $this->nextItem( ':' );
2596 }
2597 }
2598 } elseif ( $prefixLength || $lastPrefixLength ) {
2599 # We need to open or close prefixes, or both.
2600
2601 # Either open or close a level...
2602 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2603 $paragraphStack = false;
2604
2605 # Close all the prefixes which aren't shared.
2606 while ( $commonPrefixLength < $lastPrefixLength ) {
2607 $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
2608 --$lastPrefixLength;
2609 }
2610
2611 # Continue the current prefix if appropriate.
2612 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2613 $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
2614 }
2615
2616 # Open prefixes where appropriate.
2617 if ( $lastPrefix && $prefixLength > $commonPrefixLength ) {
2618 $output .= "\n";
2619 }
2620 while ( $prefixLength > $commonPrefixLength ) {
2621 $char = substr( $prefix, $commonPrefixLength, 1 );
2622 $output .= $this->openList( $char );
2623
2624 if ( ';' === $char ) {
2625 # @todo FIXME: This is dupe of code above
2626 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2627 $t = $t2;
2628 $output .= $term . $this->nextItem( ':' );
2629 }
2630 }
2631 ++$commonPrefixLength;
2632 }
2633 if ( !$prefixLength && $lastPrefix ) {
2634 $output .= "\n";
2635 }
2636 $lastPrefix = $prefix2;
2637 }
2638
2639 # If we have no prefixes, go to paragraph mode.
2640 if ( 0 == $prefixLength ) {
2641 # No prefix (not in list)--go to paragraph mode
2642 # XXX: use a stack for nestable elements like span, table and div
2643 $openmatch = preg_match(
2644 '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|'
2645 . '<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS',
2646 $t
2647 );
2648 $closematch = preg_match(
2649 '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'
2650 . '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|<\\/mw:|'
2651 . self::MARKER_PREFIX
2652 . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS',
2653 $t
2654 );
2655
2656 if ( $openmatch || $closematch ) {
2657 $paragraphStack = false;
2658 # @todo bug 5718: paragraph closed
2659 $output .= $this->closeParagraph();
2660 if ( $preOpenMatch && !$preCloseMatch ) {
2661 $this->mInPre = true;
2662 }
2663 $bqOffset = 0;
2664 while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t,
2665 $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset )
2666 ) {
2667 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
2668 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
2669 }
2670 $inBlockElem = !$closematch;
2671 } elseif ( !$inBlockElem && !$this->mInPre ) {
2672 if ( ' ' == substr( $t, 0, 1 )
2673 && ( $this->mLastSection === 'pre' || trim( $t ) != '' )
2674 && !$inBlockquote
2675 ) {
2676 # pre
2677 if ( $this->mLastSection !== 'pre' ) {
2678 $paragraphStack = false;
2679 $output .= $this->closeParagraph() . '<pre>';
2680 $this->mLastSection = 'pre';
2681 }
2682 $t = substr( $t, 1 );
2683 } else {
2684 # paragraph
2685 if ( trim( $t ) === '' ) {
2686 if ( $paragraphStack ) {
2687 $output .= $paragraphStack . '<br />';
2688 $paragraphStack = false;
2689 $this->mLastSection = 'p';
2690 } else {
2691 if ( $this->mLastSection !== 'p' ) {
2692 $output .= $this->closeParagraph();
2693 $this->mLastSection = '';
2694 $paragraphStack = '<p>';
2695 } else {
2696 $paragraphStack = '</p><p>';
2697 }
2698 }
2699 } else {
2700 if ( $paragraphStack ) {
2701 $output .= $paragraphStack;
2702 $paragraphStack = false;
2703 $this->mLastSection = 'p';
2704 } elseif ( $this->mLastSection !== 'p' ) {
2705 $output .= $this->closeParagraph() . '<p>';
2706 $this->mLastSection = 'p';
2707 }
2708 }
2709 }
2710 }
2711 }
2712 # somewhere above we forget to get out of pre block (bug 785)
2713 if ( $preCloseMatch && $this->mInPre ) {
2714 $this->mInPre = false;
2715 }
2716 if ( $paragraphStack === false ) {
2717 $output .= $t;
2718 if ( $prefixLength === 0 ) {
2719 $output .= "\n";
2720 }
2721 }
2722 }
2723 while ( $prefixLength ) {
2724 $output .= $this->closeList( $prefix2[$prefixLength - 1] );
2725 --$prefixLength;
2726 if ( !$prefixLength ) {
2727 $output .= "\n";
2728 }
2729 }
2730 if ( $this->mLastSection != '' ) {
2731 $output .= '</' . $this->mLastSection . '>';
2732 $this->mLastSection = '';
2733 }
2734
2735 return $output;
2736 }
2737
2738 /**
2739 * Split up a string on ':', ignoring any occurrences inside tags
2740 * to prevent illegal overlapping.
2741 *
2742 * @param string $str The string to split
2743 * @param string &$before Set to everything before the ':'
2744 * @param string &$after Set to everything after the ':'
2745 * @throws MWException
2746 * @return string The position of the ':', or false if none found
2747 */
2748 public function findColonNoLinks( $str, &$before, &$after ) {
2749
2750 $pos = strpos( $str, ':' );
2751 if ( $pos === false ) {
2752 # Nothing to find!
2753 return false;
2754 }
2755
2756 $lt = strpos( $str, '<' );
2757 if ( $lt === false || $lt > $pos ) {
2758 # Easy; no tag nesting to worry about
2759 $before = substr( $str, 0, $pos );
2760 $after = substr( $str, $pos + 1 );
2761 return $pos;
2762 }
2763
2764 # Ugly state machine to walk through avoiding tags.
2765 $state = self::COLON_STATE_TEXT;
2766 $stack = 0;
2767 $len = strlen( $str );
2768 for ( $i = 0; $i < $len; $i++ ) {
2769 $c = $str[$i];
2770
2771 switch ( $state ) {
2772 # (Using the number is a performance hack for common cases)
2773 case 0: # self::COLON_STATE_TEXT:
2774 switch ( $c ) {
2775 case "<":
2776 # Could be either a <start> tag or an </end> tag
2777 $state = self::COLON_STATE_TAGSTART;
2778 break;
2779 case ":":
2780 if ( $stack == 0 ) {
2781 # We found it!
2782 $before = substr( $str, 0, $i );
2783 $after = substr( $str, $i + 1 );
2784 return $i;
2785 }
2786 # Embedded in a tag; don't break it.
2787 break;
2788 default:
2789 # Skip ahead looking for something interesting
2790 $colon = strpos( $str, ':', $i );
2791 if ( $colon === false ) {
2792 # Nothing else interesting
2793 return false;
2794 }
2795 $lt = strpos( $str, '<', $i );
2796 if ( $stack === 0 ) {
2797 if ( $lt === false || $colon < $lt ) {
2798 # We found it!
2799 $before = substr( $str, 0, $colon );
2800 $after = substr( $str, $colon + 1 );
2801 return $i;
2802 }
2803 }
2804 if ( $lt === false ) {
2805 # Nothing else interesting to find; abort!
2806 # We're nested, but there's no close tags left. Abort!
2807 break 2;
2808 }
2809 # Skip ahead to next tag start
2810 $i = $lt;
2811 $state = self::COLON_STATE_TAGSTART;
2812 }
2813 break;
2814 case 1: # self::COLON_STATE_TAG:
2815 # In a <tag>
2816 switch ( $c ) {
2817 case ">":
2818 $stack++;
2819 $state = self::COLON_STATE_TEXT;
2820 break;
2821 case "/":
2822 # Slash may be followed by >?
2823 $state = self::COLON_STATE_TAGSLASH;
2824 break;
2825 default:
2826 # ignore
2827 }
2828 break;
2829 case 2: # self::COLON_STATE_TAGSTART:
2830 switch ( $c ) {
2831 case "/":
2832 $state = self::COLON_STATE_CLOSETAG;
2833 break;
2834 case "!":
2835 $state = self::COLON_STATE_COMMENT;
2836 break;
2837 case ">":
2838 # Illegal early close? This shouldn't happen D:
2839 $state = self::COLON_STATE_TEXT;
2840 break;
2841 default:
2842 $state = self::COLON_STATE_TAG;
2843 }
2844 break;
2845 case 3: # self::COLON_STATE_CLOSETAG:
2846 # In a </tag>
2847 if ( $c === ">" ) {
2848 $stack--;
2849 if ( $stack < 0 ) {
2850 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
2851 return false;
2852 }
2853 $state = self::COLON_STATE_TEXT;
2854 }
2855 break;
2856 case self::COLON_STATE_TAGSLASH:
2857 if ( $c === ">" ) {
2858 # Yes, a self-closed tag <blah/>
2859 $state = self::COLON_STATE_TEXT;
2860 } else {
2861 # Probably we're jumping the gun, and this is an attribute
2862 $state = self::COLON_STATE_TAG;
2863 }
2864 break;
2865 case 5: # self::COLON_STATE_COMMENT:
2866 if ( $c === "-" ) {
2867 $state = self::COLON_STATE_COMMENTDASH;
2868 }
2869 break;
2870 case self::COLON_STATE_COMMENTDASH:
2871 if ( $c === "-" ) {
2872 $state = self::COLON_STATE_COMMENTDASHDASH;
2873 } else {
2874 $state = self::COLON_STATE_COMMENT;
2875 }
2876 break;
2877 case self::COLON_STATE_COMMENTDASHDASH:
2878 if ( $c === ">" ) {
2879 $state = self::COLON_STATE_TEXT;
2880 } else {
2881 $state = self::COLON_STATE_COMMENT;
2882 }
2883 break;
2884 default:
2885 throw new MWException( "State machine error in " . __METHOD__ );
2886 }
2887 }
2888 if ( $stack > 0 ) {
2889 wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2890 return false;
2891 }
2892 return false;
2893 }
2894
2895 /**
2896 * Return value of a magic variable (like PAGENAME)
2897 *
2898 * @private
2899 *
2900 * @param int $index
2901 * @param bool|PPFrame $frame
2902 *
2903 * @throws MWException
2904 * @return string
2905 */
2906 public function getVariableValue( $index, $frame = false ) {
2907 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2908 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2909
2910 if ( is_null( $this->mTitle ) ) {
2911 // If no title set, bad things are going to happen
2912 // later. Title should always be set since this
2913 // should only be called in the middle of a parse
2914 // operation (but the unit-tests do funky stuff)
2915 throw new MWException( __METHOD__ . ' Should only be '
2916 . ' called while parsing (no title set)' );
2917 }
2918
2919 /**
2920 * Some of these require message or data lookups and can be
2921 * expensive to check many times.
2922 */
2923 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2924 if ( isset( $this->mVarCache[$index] ) ) {
2925 return $this->mVarCache[$index];
2926 }
2927 }
2928
2929 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2930 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2931
2932 $pageLang = $this->getFunctionLang();
2933
2934 switch ( $index ) {
2935 case '!':
2936 $value = '|';
2937 break;
2938 case 'currentmonth':
2939 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2940 break;
2941 case 'currentmonth1':
2942 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2943 break;
2944 case 'currentmonthname':
2945 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2946 break;
2947 case 'currentmonthnamegen':
2948 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2949 break;
2950 case 'currentmonthabbrev':
2951 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2952 break;
2953 case 'currentday':
2954 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2955 break;
2956 case 'currentday2':
2957 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2958 break;
2959 case 'localmonth':
2960 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2961 break;
2962 case 'localmonth1':
2963 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2964 break;
2965 case 'localmonthname':
2966 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2967 break;
2968 case 'localmonthnamegen':
2969 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2970 break;
2971 case 'localmonthabbrev':
2972 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2973 break;
2974 case 'localday':
2975 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2976 break;
2977 case 'localday2':
2978 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2979 break;
2980 case 'pagename':
2981 $value = wfEscapeWikiText( $this->mTitle->getText() );
2982 break;
2983 case 'pagenamee':
2984 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2985 break;
2986 case 'fullpagename':
2987 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2988 break;
2989 case 'fullpagenamee':
2990 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2991 break;
2992 case 'subpagename':
2993 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2994 break;
2995 case 'subpagenamee':
2996 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2997 break;
2998 case 'rootpagename':
2999 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
3000 break;
3001 case 'rootpagenamee':
3002 $value = wfEscapeWikiText( wfUrlencode( str_replace(
3003 ' ',
3004 '_',
3005 $this->mTitle->getRootText()
3006 ) ) );
3007 break;
3008 case 'basepagename':
3009 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
3010 break;
3011 case 'basepagenamee':
3012 $value = wfEscapeWikiText( wfUrlencode( str_replace(
3013 ' ',
3014 '_',
3015 $this->mTitle->getBaseText()
3016 ) ) );
3017 break;
3018 case 'talkpagename':
3019 if ( $this->mTitle->canTalk() ) {
3020 $talkPage = $this->mTitle->getTalkPage();
3021 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
3022 } else {
3023 $value = '';
3024 }
3025 break;
3026 case 'talkpagenamee':
3027 if ( $this->mTitle->canTalk() ) {
3028 $talkPage = $this->mTitle->getTalkPage();
3029 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
3030 } else {
3031 $value = '';
3032 }
3033 break;
3034 case 'subjectpagename':
3035 $subjPage = $this->mTitle->getSubjectPage();
3036 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
3037 break;
3038 case 'subjectpagenamee':
3039 $subjPage = $this->mTitle->getSubjectPage();
3040 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
3041 break;
3042 case 'pageid': // requested in bug 23427
3043 $pageid = $this->getTitle()->getArticleID();
3044 if ( $pageid == 0 ) {
3045 # 0 means the page doesn't exist in the database,
3046 # which means the user is previewing a new page.
3047 # The vary-revision flag must be set, because the magic word
3048 # will have a different value once the page is saved.
3049 $this->mOutput->setFlag( 'vary-revision' );
3050 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
3051 }
3052 $value = $pageid ? $pageid : null;
3053 break;
3054 case 'revisionid':
3055 # Let the edit saving system know we should parse the page
3056 # *after* a revision ID has been assigned.
3057 $this->mOutput->setFlag( 'vary-revision' );
3058 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
3059 $value = $this->mRevisionId;
3060 break;
3061 case 'revisionday':
3062 # Let the edit saving system know we should parse the page
3063 # *after* a revision ID has been assigned. This is for null edits.
3064 $this->mOutput->setFlag( 'vary-revision' );
3065 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
3066 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
3067 break;
3068 case 'revisionday2':
3069 # Let the edit saving system know we should parse the page
3070 # *after* a revision ID has been assigned. This is for null edits.
3071 $this->mOutput->setFlag( 'vary-revision' );
3072 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
3073 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
3074 break;
3075 case 'revisionmonth':
3076 # Let the edit saving system know we should parse the page
3077 # *after* a revision ID has been assigned. This is for null edits.
3078 $this->mOutput->setFlag( 'vary-revision' );
3079 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
3080 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
3081 break;
3082 case 'revisionmonth1':
3083 # Let the edit saving system know we should parse the page
3084 # *after* a revision ID has been assigned. This is for null edits.
3085 $this->mOutput->setFlag( 'vary-revision' );
3086 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
3087 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
3088 break;
3089 case 'revisionyear':
3090 # Let the edit saving system know we should parse the page
3091 # *after* a revision ID has been assigned. This is for null edits.
3092 $this->mOutput->setFlag( 'vary-revision' );
3093 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
3094 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
3095 break;
3096 case 'revisiontimestamp':
3097 # Let the edit saving system know we should parse the page
3098 # *after* a revision ID has been assigned. This is for null edits.
3099 $this->mOutput->setFlag( 'vary-revision' );
3100 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
3101 $value = $this->getRevisionTimestamp();
3102 break;
3103 case 'revisionuser':
3104 # Let the edit saving system know we should parse the page
3105 # *after* a revision ID has been assigned. This is for null edits.
3106 $this->mOutput->setFlag( 'vary-revision' );
3107 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
3108 $value = $this->getRevisionUser();
3109 break;
3110 case 'revisionsize':
3111 # Let the edit saving system know we should parse the page
3112 # *after* a revision ID has been assigned. This is for null edits.
3113 $this->mOutput->setFlag( 'vary-revision' );
3114 wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
3115 $value = $this->getRevisionSize();
3116 break;
3117 case 'namespace':
3118 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
3119 break;
3120 case 'namespacee':
3121 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
3122 break;
3123 case 'namespacenumber':
3124 $value = $this->mTitle->getNamespace();
3125 break;
3126 case 'talkspace':
3127 $value = $this->mTitle->canTalk()
3128 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
3129 : '';
3130 break;
3131 case 'talkspacee':
3132 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
3133 break;
3134 case 'subjectspace':
3135 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
3136 break;
3137 case 'subjectspacee':
3138 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
3139 break;
3140 case 'currentdayname':
3141 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
3142 break;
3143 case 'currentyear':
3144 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
3145 break;
3146 case 'currenttime':
3147 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
3148 break;
3149 case 'currenthour':
3150 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
3151 break;
3152 case 'currentweek':
3153 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3154 # int to remove the padding
3155 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
3156 break;
3157 case 'currentdow':
3158 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
3159 break;
3160 case 'localdayname':
3161 $value = $pageLang->getWeekdayName(
3162 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
3163 );
3164 break;
3165 case 'localyear':
3166 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
3167 break;
3168 case 'localtime':
3169 $value = $pageLang->time(
3170 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
3171 false,
3172 false
3173 );
3174 break;
3175 case 'localhour':
3176 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
3177 break;
3178 case 'localweek':
3179 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3180 # int to remove the padding
3181 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
3182 break;
3183 case 'localdow':
3184 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
3185 break;
3186 case 'numberofarticles':
3187 $value = $pageLang->formatNum( SiteStats::articles() );
3188 break;
3189 case 'numberoffiles':
3190 $value = $pageLang->formatNum( SiteStats::images() );
3191 break;
3192 case 'numberofusers':
3193 $value = $pageLang->formatNum( SiteStats::users() );
3194 break;
3195 case 'numberofactiveusers':
3196 $value = $pageLang->formatNum( SiteStats::activeUsers() );
3197 break;
3198 case 'numberofpages':
3199 $value = $pageLang->formatNum( SiteStats::pages() );
3200 break;
3201 case 'numberofadmins':
3202 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
3203 break;
3204 case 'numberofedits':
3205 $value = $pageLang->formatNum( SiteStats::edits() );
3206 break;
3207 case 'currenttimestamp':
3208 $value = wfTimestamp( TS_MW, $ts );
3209 break;
3210 case 'localtimestamp':
3211 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
3212 break;
3213 case 'currentversion':
3214 $value = SpecialVersion::getVersion();
3215 break;
3216 case 'articlepath':
3217 return $wgArticlePath;
3218 case 'sitename':
3219 return $wgSitename;
3220 case 'server':
3221 return $wgServer;
3222 case 'servername':
3223 return $wgServerName;
3224 case 'scriptpath':
3225 return $wgScriptPath;
3226 case 'stylepath':
3227 return $wgStylePath;
3228 case 'directionmark':
3229 return $pageLang->getDirMark();
3230 case 'contentlanguage':
3231 global $wgLanguageCode;
3232 return $wgLanguageCode;
3233 case 'cascadingsources':
3234 $value = CoreParserFunctions::cascadingsources( $this );
3235 break;
3236 default:
3237 $ret = null;
3238 Hooks::run(
3239 'ParserGetVariableValueSwitch',
3240 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
3241 );
3242
3243 return $ret;
3244 }
3245
3246 if ( $index ) {
3247 $this->mVarCache[$index] = $value;
3248 }
3249
3250 return $value;
3251 }
3252
3253 /**
3254 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3255 *
3256 * @private
3257 */
3258 public function initialiseVariables() {
3259 $variableIDs = MagicWord::getVariableIDs();
3260 $substIDs = MagicWord::getSubstIDs();
3261
3262 $this->mVariables = new MagicWordArray( $variableIDs );
3263 $this->mSubstWords = new MagicWordArray( $substIDs );
3264 }
3265
3266 /**
3267 * Preprocess some wikitext and return the document tree.
3268 * This is the ghost of replace_variables().
3269 *
3270 * @param string $text The text to parse
3271 * @param int $flags Bitwise combination of:
3272 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3273 * included. Default is to assume a direct page view.
3274 *
3275 * The generated DOM tree must depend only on the input text and the flags.
3276 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
3277 *
3278 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3279 * change in the DOM tree for a given text, must be passed through the section identifier
3280 * in the section edit link and thus back to extractSections().
3281 *
3282 * The output of this function is currently only cached in process memory, but a persistent
3283 * cache may be implemented at a later date which takes further advantage of these strict
3284 * dependency requirements.
3285 *
3286 * @return PPNode
3287 */
3288 public function preprocessToDom( $text, $flags = 0 ) {
3289 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3290 return $dom;
3291 }
3292
3293 /**
3294 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3295 *
3296 * @param string $s
3297 *
3298 * @return array
3299 */
3300 public static function splitWhitespace( $s ) {
3301 $ltrimmed = ltrim( $s );
3302 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3303 $trimmed = rtrim( $ltrimmed );
3304 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3305 if ( $diff > 0 ) {
3306 $w2 = substr( $ltrimmed, -$diff );
3307 } else {
3308 $w2 = '';
3309 }
3310 return [ $w1, $trimmed, $w2 ];
3311 }
3312
3313 /**
3314 * Replace magic variables, templates, and template arguments
3315 * with the appropriate text. Templates are substituted recursively,
3316 * taking care to avoid infinite loops.
3317 *
3318 * Note that the substitution depends on value of $mOutputType:
3319 * self::OT_WIKI: only {{subst:}} templates
3320 * self::OT_PREPROCESS: templates but not extension tags
3321 * self::OT_HTML: all templates and extension tags
3322 *
3323 * @param string $text The text to transform
3324 * @param bool|PPFrame $frame Object describing the arguments passed to the
3325 * template. Arguments may also be provided as an associative array, as
3326 * was the usual case before MW1.12. Providing arguments this way may be
3327 * useful for extensions wishing to perform variable replacement
3328 * explicitly.
3329 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3330 * double-brace expansion.
3331 * @return string
3332 */
3333 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3334 # Is there any text? Also, Prevent too big inclusions!
3335 $textSize = strlen( $text );
3336 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
3337 return $text;
3338 }
3339
3340 if ( $frame === false ) {
3341 $frame = $this->getPreprocessor()->newFrame();
3342 } elseif ( !( $frame instanceof PPFrame ) ) {
3343 wfDebug( __METHOD__ . " called using plain parameters instead of "
3344 . "a PPFrame instance. Creating custom frame.\n" );
3345 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3346 }
3347
3348 $dom = $this->preprocessToDom( $text );
3349 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3350 $text = $frame->expand( $dom, $flags );
3351
3352 return $text;
3353 }
3354
3355 /**
3356 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3357 *
3358 * @param array $args
3359 *
3360 * @return array
3361 */
3362 public static function createAssocArgs( $args ) {
3363 $assocArgs = [];
3364 $index = 1;
3365 foreach ( $args as $arg ) {
3366 $eqpos = strpos( $arg, '=' );
3367 if ( $eqpos === false ) {
3368 $assocArgs[$index++] = $arg;
3369 } else {
3370 $name = trim( substr( $arg, 0, $eqpos ) );
3371 $value = trim( substr( $arg, $eqpos + 1 ) );
3372 if ( $value === false ) {
3373 $value = '';
3374 }
3375 if ( $name !== false ) {
3376 $assocArgs[$name] = $value;
3377 }
3378 }
3379 }
3380
3381 return $assocArgs;
3382 }
3383
3384 /**
3385 * Warn the user when a parser limitation is reached
3386 * Will warn at most once the user per limitation type
3387 *
3388 * The results are shown during preview and run through the Parser (See EditPage.php)
3389 *
3390 * @param string $limitationType Should be one of:
3391 * 'expensive-parserfunction' (corresponding messages:
3392 * 'expensive-parserfunction-warning',
3393 * 'expensive-parserfunction-category')
3394 * 'post-expand-template-argument' (corresponding messages:
3395 * 'post-expand-template-argument-warning',
3396 * 'post-expand-template-argument-category')
3397 * 'post-expand-template-inclusion' (corresponding messages:
3398 * 'post-expand-template-inclusion-warning',
3399 * 'post-expand-template-inclusion-category')
3400 * 'node-count-exceeded' (corresponding messages:
3401 * 'node-count-exceeded-warning',
3402 * 'node-count-exceeded-category')
3403 * 'expansion-depth-exceeded' (corresponding messages:
3404 * 'expansion-depth-exceeded-warning',
3405 * 'expansion-depth-exceeded-category')
3406 * @param string|int|null $current Current value
3407 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3408 * exceeded, provide the values (optional)
3409 */
3410 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3411 # does no harm if $current and $max are present but are unnecessary for the message
3412 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
3413 # only during preview, and that would split the parser cache unnecessarily.
3414 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3415 ->text();
3416 $this->mOutput->addWarning( $warning );
3417 $this->addTrackingCategory( "$limitationType-category" );
3418 }
3419
3420 /**
3421 * Return the text of a template, after recursively
3422 * replacing any variables or templates within the template.
3423 *
3424 * @param array $piece The parts of the template
3425 * $piece['title']: the title, i.e. the part before the |
3426 * $piece['parts']: the parameter array
3427 * $piece['lineStart']: whether the brace was at the start of a line
3428 * @param PPFrame $frame The current frame, contains template arguments
3429 * @throws Exception
3430 * @return string The text of the template
3431 */
3432 public function braceSubstitution( $piece, $frame ) {
3433
3434 // Flags
3435
3436 // $text has been filled
3437 $found = false;
3438 // wiki markup in $text should be escaped
3439 $nowiki = false;
3440 // $text is HTML, armour it against wikitext transformation
3441 $isHTML = false;
3442 // Force interwiki transclusion to be done in raw mode not rendered
3443 $forceRawInterwiki = false;
3444 // $text is a DOM node needing expansion in a child frame
3445 $isChildObj = false;
3446 // $text is a DOM node needing expansion in the current frame
3447 $isLocalObj = false;
3448
3449 # Title object, where $text came from
3450 $title = false;
3451
3452 # $part1 is the bit before the first |, and must contain only title characters.
3453 # Various prefixes will be stripped from it later.
3454 $titleWithSpaces = $frame->expand( $piece['title'] );
3455 $part1 = trim( $titleWithSpaces );
3456 $titleText = false;
3457
3458 # Original title text preserved for various purposes
3459 $originalTitle = $part1;
3460
3461 # $args is a list of argument nodes, starting from index 0, not including $part1
3462 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3463 # below won't work b/c this $args isn't an object
3464 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3465
3466 $profileSection = null; // profile templates
3467
3468 # SUBST
3469 if ( !$found ) {
3470 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3471
3472 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3473 # Decide whether to expand template or keep wikitext as-is.
3474 if ( $this->ot['wiki'] ) {
3475 if ( $substMatch === false ) {
3476 $literal = true; # literal when in PST with no prefix
3477 } else {
3478 $literal = false; # expand when in PST with subst: or safesubst:
3479 }
3480 } else {
3481 if ( $substMatch == 'subst' ) {
3482 $literal = true; # literal when not in PST with plain subst:
3483 } else {
3484 $literal = false; # expand when not in PST with safesubst: or no prefix
3485 }
3486 }
3487 if ( $literal ) {
3488 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3489 $isLocalObj = true;
3490 $found = true;
3491 }
3492 }
3493
3494 # Variables
3495 if ( !$found && $args->getLength() == 0 ) {
3496 $id = $this->mVariables->matchStartToEnd( $part1 );
3497 if ( $id !== false ) {
3498 $text = $this->getVariableValue( $id, $frame );
3499 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3500 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3501 }
3502 $found = true;
3503 }
3504 }
3505
3506 # MSG, MSGNW and RAW
3507 if ( !$found ) {
3508 # Check for MSGNW:
3509 $mwMsgnw = MagicWord::get( 'msgnw' );
3510 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3511 $nowiki = true;
3512 } else {
3513 # Remove obsolete MSG:
3514 $mwMsg = MagicWord::get( 'msg' );
3515 $mwMsg->matchStartAndRemove( $part1 );
3516 }
3517
3518 # Check for RAW:
3519 $mwRaw = MagicWord::get( 'raw' );
3520 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3521 $forceRawInterwiki = true;
3522 }
3523 }
3524
3525 # Parser functions
3526 if ( !$found ) {
3527 $colonPos = strpos( $part1, ':' );
3528 if ( $colonPos !== false ) {
3529 $func = substr( $part1, 0, $colonPos );
3530 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3531 $argsLength = $args->getLength();
3532 for ( $i = 0; $i < $argsLength; $i++ ) {
3533 $funcArgs[] = $args->item( $i );
3534 }
3535 try {
3536 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3537 } catch ( Exception $ex ) {
3538 throw $ex;
3539 }
3540
3541 # The interface for parser functions allows for extracting
3542 # flags into the local scope. Extract any forwarded flags
3543 # here.
3544 extract( $result );
3545 }
3546 }
3547
3548 # Finish mangling title and then check for loops.
3549 # Set $title to a Title object and $titleText to the PDBK
3550 if ( !$found ) {
3551 $ns = NS_TEMPLATE;
3552 # Split the title into page and subpage
3553 $subpage = '';
3554 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3555 if ( $part1 !== $relative ) {
3556 $part1 = $relative;
3557 $ns = $this->mTitle->getNamespace();
3558 }
3559 $title = Title::newFromText( $part1, $ns );
3560 if ( $title ) {
3561 $titleText = $title->getPrefixedText();
3562 # Check for language variants if the template is not found
3563 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3564 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3565 }
3566 # Do recursion depth check
3567 $limit = $this->mOptions->getMaxTemplateDepth();
3568 if ( $frame->depth >= $limit ) {
3569 $found = true;
3570 $text = '<span class="error">'
3571 . wfMessage( 'parser-template-recursion-depth-warning' )
3572 ->numParams( $limit )->inContentLanguage()->text()
3573 . '</span>';
3574 }
3575 }
3576 }
3577
3578 # Load from database
3579 if ( !$found && $title ) {
3580 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3581 if ( !$title->isExternal() ) {
3582 if ( $title->isSpecialPage()
3583 && $this->mOptions->getAllowSpecialInclusion()
3584 && $this->ot['html']
3585 ) {
3586 // Pass the template arguments as URL parameters.
3587 // "uselang" will have no effect since the Language object
3588 // is forced to the one defined in ParserOptions.
3589 $pageArgs = [];
3590 $argsLength = $args->getLength();
3591 for ( $i = 0; $i < $argsLength; $i++ ) {
3592 $bits = $args->item( $i )->splitArg();
3593 if ( strval( $bits['index'] ) === '' ) {
3594 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3595 $value = trim( $frame->expand( $bits['value'] ) );
3596 $pageArgs[$name] = $value;
3597 }
3598 }
3599
3600 // Create a new context to execute the special page
3601 $context = new RequestContext;
3602 $context->setTitle( $title );
3603 $context->setRequest( new FauxRequest( $pageArgs ) );
3604 $context->setUser( $this->getUser() );
3605 $context->setLanguage( $this->mOptions->getUserLangObj() );
3606 $ret = SpecialPageFactory::capturePath( $title, $context );
3607 if ( $ret ) {
3608 $text = $context->getOutput()->getHTML();
3609 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3610 $found = true;
3611 $isHTML = true;
3612 $this->disableCache();
3613 }
3614 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3615 $found = false; # access denied
3616 wfDebug( __METHOD__ . ": template inclusion denied for " .
3617 $title->getPrefixedDBkey() . "\n" );
3618 } else {
3619 list( $text, $title ) = $this->getTemplateDom( $title );
3620 if ( $text !== false ) {
3621 $found = true;
3622 $isChildObj = true;
3623 }
3624 }
3625
3626 # If the title is valid but undisplayable, make a link to it
3627 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3628 $text = "[[:$titleText]]";
3629 $found = true;
3630 }
3631 } elseif ( $title->isTrans() ) {
3632 # Interwiki transclusion
3633 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3634 $text = $this->interwikiTransclude( $title, 'render' );
3635 $isHTML = true;
3636 } else {
3637 $text = $this->interwikiTransclude( $title, 'raw' );
3638 # Preprocess it like a template
3639 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3640 $isChildObj = true;
3641 }
3642 $found = true;
3643 }
3644
3645 # Do infinite loop check
3646 # This has to be done after redirect resolution to avoid infinite loops via redirects
3647 if ( !$frame->loopCheck( $title ) ) {
3648 $found = true;
3649 $text = '<span class="error">'
3650 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3651 . '</span>';
3652 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3653 }
3654 }
3655
3656 # If we haven't found text to substitute by now, we're done
3657 # Recover the source wikitext and return it
3658 if ( !$found ) {
3659 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3660 if ( $profileSection ) {
3661 $this->mProfiler->scopedProfileOut( $profileSection );
3662 }
3663 return [ 'object' => $text ];
3664 }
3665
3666 # Expand DOM-style return values in a child frame
3667 if ( $isChildObj ) {
3668 # Clean up argument array
3669 $newFrame = $frame->newChild( $args, $title );
3670
3671 if ( $nowiki ) {
3672 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3673 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3674 # Expansion is eligible for the empty-frame cache
3675 $text = $newFrame->cachedExpand( $titleText, $text );
3676 } else {
3677 # Uncached expansion
3678 $text = $newFrame->expand( $text );
3679 }
3680 }
3681 if ( $isLocalObj && $nowiki ) {
3682 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3683 $isLocalObj = false;
3684 }
3685
3686 if ( $profileSection ) {
3687 $this->mProfiler->scopedProfileOut( $profileSection );
3688 }
3689
3690 # Replace raw HTML by a placeholder
3691 if ( $isHTML ) {
3692 $text = $this->insertStripItem( $text );
3693 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3694 # Escape nowiki-style return values
3695 $text = wfEscapeWikiText( $text );
3696 } elseif ( is_string( $text )
3697 && !$piece['lineStart']
3698 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3699 ) {
3700 # Bug 529: if the template begins with a table or block-level
3701 # element, it should be treated as beginning a new line.
3702 # This behavior is somewhat controversial.
3703 $text = "\n" . $text;
3704 }
3705
3706 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3707 # Error, oversize inclusion
3708 if ( $titleText !== false ) {
3709 # Make a working, properly escaped link if possible (bug 23588)
3710 $text = "[[:$titleText]]";
3711 } else {
3712 # This will probably not be a working link, but at least it may
3713 # provide some hint of where the problem is
3714 preg_replace( '/^:/', '', $originalTitle );
3715 $text = "[[:$originalTitle]]";
3716 }
3717 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3718 . 'post-expand include size too large -->' );
3719 $this->limitationWarn( 'post-expand-template-inclusion' );
3720 }
3721
3722 if ( $isLocalObj ) {
3723 $ret = [ 'object' => $text ];
3724 } else {
3725 $ret = [ 'text' => $text ];
3726 }
3727
3728 return $ret;
3729 }
3730
3731 /**
3732 * Call a parser function and return an array with text and flags.
3733 *
3734 * The returned array will always contain a boolean 'found', indicating
3735 * whether the parser function was found or not. It may also contain the
3736 * following:
3737 * text: string|object, resulting wikitext or PP DOM object
3738 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3739 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3740 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3741 * nowiki: bool, wiki markup in $text should be escaped
3742 *
3743 * @since 1.21
3744 * @param PPFrame $frame The current frame, contains template arguments
3745 * @param string $function Function name
3746 * @param array $args Arguments to the function
3747 * @throws MWException
3748 * @return array
3749 */
3750 public function callParserFunction( $frame, $function, array $args = [] ) {
3751 global $wgContLang;
3752
3753 # Case sensitive functions
3754 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3755 $function = $this->mFunctionSynonyms[1][$function];
3756 } else {
3757 # Case insensitive functions
3758 $function = $wgContLang->lc( $function );
3759 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3760 $function = $this->mFunctionSynonyms[0][$function];
3761 } else {
3762 return [ 'found' => false ];
3763 }
3764 }
3765
3766 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3767
3768 # Workaround for PHP bug 35229 and similar
3769 if ( !is_callable( $callback ) ) {
3770 throw new MWException( "Tag hook for $function is not callable\n" );
3771 }
3772
3773 $allArgs = [ &$this ];
3774 if ( $flags & self::SFH_OBJECT_ARGS ) {
3775 # Convert arguments to PPNodes and collect for appending to $allArgs
3776 $funcArgs = [];
3777 foreach ( $args as $k => $v ) {
3778 if ( $v instanceof PPNode || $k === 0 ) {
3779 $funcArgs[] = $v;
3780 } else {
3781 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3782 }
3783 }
3784
3785 # Add a frame parameter, and pass the arguments as an array
3786 $allArgs[] = $frame;
3787 $allArgs[] = $funcArgs;
3788 } else {
3789 # Convert arguments to plain text and append to $allArgs
3790 foreach ( $args as $k => $v ) {
3791 if ( $v instanceof PPNode ) {
3792 $allArgs[] = trim( $frame->expand( $v ) );
3793 } elseif ( is_int( $k ) && $k >= 0 ) {
3794 $allArgs[] = trim( $v );
3795 } else {
3796 $allArgs[] = trim( "$k=$v" );
3797 }
3798 }
3799 }
3800
3801 $result = call_user_func_array( $callback, $allArgs );
3802
3803 # The interface for function hooks allows them to return a wikitext
3804 # string or an array containing the string and any flags. This mungs
3805 # things around to match what this method should return.
3806 if ( !is_array( $result ) ) {
3807 $result =[
3808 'found' => true,
3809 'text' => $result,
3810 ];
3811 } else {
3812 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3813 $result['text'] = $result[0];
3814 }
3815 unset( $result[0] );
3816 $result += [
3817 'found' => true,
3818 ];
3819 }
3820
3821 $noparse = true;
3822 $preprocessFlags = 0;
3823 if ( isset( $result['noparse'] ) ) {
3824 $noparse = $result['noparse'];
3825 }
3826 if ( isset( $result['preprocessFlags'] ) ) {
3827 $preprocessFlags = $result['preprocessFlags'];
3828 }
3829
3830 if ( !$noparse ) {
3831 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3832 $result['isChildObj'] = true;
3833 }
3834
3835 return $result;
3836 }
3837
3838 /**
3839 * Get the semi-parsed DOM representation of a template with a given title,
3840 * and its redirect destination title. Cached.
3841 *
3842 * @param Title $title
3843 *
3844 * @return array
3845 */
3846 public function getTemplateDom( $title ) {
3847 $cacheTitle = $title;
3848 $titleText = $title->getPrefixedDBkey();
3849
3850 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3851 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3852 $title = Title::makeTitle( $ns, $dbk );
3853 $titleText = $title->getPrefixedDBkey();
3854 }
3855 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3856 return [ $this->mTplDomCache[$titleText], $title ];
3857 }
3858
3859 # Cache miss, go to the database
3860 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3861
3862 if ( $text === false ) {
3863 $this->mTplDomCache[$titleText] = false;
3864 return [ false, $title ];
3865 }
3866
3867 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3868 $this->mTplDomCache[$titleText] = $dom;
3869
3870 if ( !$title->equals( $cacheTitle ) ) {
3871 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3872 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3873 }
3874
3875 return [ $dom, $title ];
3876 }
3877
3878 /**
3879 * Fetch the current revision of a given title. Note that the revision
3880 * (and even the title) may not exist in the database, so everything
3881 * contributing to the output of the parser should use this method
3882 * where possible, rather than getting the revisions themselves. This
3883 * method also caches its results, so using it benefits performance.
3884 *
3885 * @since 1.24
3886 * @param Title $title
3887 * @return Revision
3888 */
3889 public function fetchCurrentRevisionOfTitle( $title ) {
3890 $cacheKey = $title->getPrefixedDBkey();
3891 if ( !$this->currentRevisionCache ) {
3892 $this->currentRevisionCache = new MapCacheLRU( 100 );
3893 }
3894 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3895 $this->currentRevisionCache->set( $cacheKey,
3896 // Defaults to Parser::statelessFetchRevision()
3897 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3898 );
3899 }
3900 return $this->currentRevisionCache->get( $cacheKey );
3901 }
3902
3903 /**
3904 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3905 * without passing them on to it.
3906 *
3907 * @since 1.24
3908 * @param Title $title
3909 * @param Parser|bool $parser
3910 * @return Revision
3911 */
3912 public static function statelessFetchRevision( $title, $parser = false ) {
3913 return Revision::newFromTitle( $title );
3914 }
3915
3916 /**
3917 * Fetch the unparsed text of a template and register a reference to it.
3918 * @param Title $title
3919 * @return array ( string or false, Title )
3920 */
3921 public function fetchTemplateAndTitle( $title ) {
3922 // Defaults to Parser::statelessFetchTemplate()
3923 $templateCb = $this->mOptions->getTemplateCallback();
3924 $stuff = call_user_func( $templateCb, $title, $this );
3925 // We use U+007F DELETE to distinguish strip markers from regular text.
3926 $text = $stuff['text'];
3927 if ( is_string( $stuff['text'] ) ) {
3928 $text = strtr( $text, "\x7f", "?" );
3929 }
3930 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3931 if ( isset( $stuff['deps'] ) ) {
3932 foreach ( $stuff['deps'] as $dep ) {
3933 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3934 if ( $dep['title']->equals( $this->getTitle() ) ) {
3935 // If we transclude ourselves, the final result
3936 // will change based on the new version of the page
3937 $this->mOutput->setFlag( 'vary-revision' );
3938 }
3939 }
3940 }
3941 return [ $text, $finalTitle ];
3942 }
3943
3944 /**
3945 * Fetch the unparsed text of a template and register a reference to it.
3946 * @param Title $title
3947 * @return string|bool
3948 */
3949 public function fetchTemplate( $title ) {
3950 return $this->fetchTemplateAndTitle( $title )[0];
3951 }
3952
3953 /**
3954 * Static function to get a template
3955 * Can be overridden via ParserOptions::setTemplateCallback().
3956 *
3957 * @param Title $title
3958 * @param bool|Parser $parser
3959 *
3960 * @return array
3961 */
3962 public static function statelessFetchTemplate( $title, $parser = false ) {
3963 $text = $skip = false;
3964 $finalTitle = $title;
3965 $deps = [];
3966
3967 # Loop to fetch the article, with up to 1 redirect
3968 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3969 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3970 // @codingStandardsIgnoreEnd
3971 # Give extensions a chance to select the revision instead
3972 $id = false; # Assume current
3973 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3974 [ $parser, $title, &$skip, &$id ] );
3975
3976 if ( $skip ) {
3977 $text = false;
3978 $deps[] = [
3979 'title' => $title,
3980 'page_id' => $title->getArticleID(),
3981 'rev_id' => null
3982 ];
3983 break;
3984 }
3985 # Get the revision
3986 if ( $id ) {
3987 $rev = Revision::newFromId( $id );
3988 } elseif ( $parser ) {
3989 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3990 } else {
3991 $rev = Revision::newFromTitle( $title );
3992 }
3993 $rev_id = $rev ? $rev->getId() : 0;
3994 # If there is no current revision, there is no page
3995 if ( $id === false && !$rev ) {
3996 $linkCache = LinkCache::singleton();
3997 $linkCache->addBadLinkObj( $title );
3998 }
3999
4000 $deps[] = [
4001 'title' => $title,
4002 'page_id' => $title->getArticleID(),
4003 'rev_id' => $rev_id ];
4004 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
4005 # We fetched a rev from a different title; register it too...
4006 $deps[] = [
4007 'title' => $rev->getTitle(),
4008 'page_id' => $rev->getPage(),
4009 'rev_id' => $rev_id ];
4010 }
4011
4012 if ( $rev ) {
4013 $content = $rev->getContent();
4014 $text = $content ? $content->getWikitextForTransclusion() : null;
4015
4016 if ( $text === false || $text === null ) {
4017 $text = false;
4018 break;
4019 }
4020 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
4021 global $wgContLang;
4022 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
4023 if ( !$message->exists() ) {
4024 $text = false;
4025 break;
4026 }
4027 $content = $message->content();
4028 $text = $message->plain();
4029 } else {
4030 break;
4031 }
4032 if ( !$content ) {
4033 break;
4034 }
4035 # Redirect?
4036 $finalTitle = $title;
4037 $title = $content->getRedirectTarget();
4038 }
4039 return [
4040 'text' => $text,
4041 'finalTitle' => $finalTitle,
4042 'deps' => $deps ];
4043 }
4044
4045 /**
4046 * Fetch a file and its title and register a reference to it.
4047 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
4048 * @param Title $title
4049 * @param array $options Array of options to RepoGroup::findFile
4050 * @return File|bool
4051 */
4052 public function fetchFile( $title, $options = [] ) {
4053 return $this->fetchFileAndTitle( $title, $options )[0];
4054 }
4055
4056 /**
4057 * Fetch a file and its title and register a reference to it.
4058 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
4059 * @param Title $title
4060 * @param array $options Array of options to RepoGroup::findFile
4061 * @return array ( File or false, Title of file )
4062 */
4063 public function fetchFileAndTitle( $title, $options = [] ) {
4064 $file = $this->fetchFileNoRegister( $title, $options );
4065
4066 $time = $file ? $file->getTimestamp() : false;
4067 $sha1 = $file ? $file->getSha1() : false;
4068 # Register the file as a dependency...
4069 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4070 if ( $file && !$title->equals( $file->getTitle() ) ) {
4071 # Update fetched file title
4072 $title = $file->getTitle();
4073 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4074 }
4075 return [ $file, $title ];
4076 }
4077
4078 /**
4079 * Helper function for fetchFileAndTitle.
4080 *
4081 * Also useful if you need to fetch a file but not use it yet,
4082 * for example to get the file's handler.
4083 *
4084 * @param Title $title
4085 * @param array $options Array of options to RepoGroup::findFile
4086 * @return File|bool
4087 */
4088 protected function fetchFileNoRegister( $title, $options = [] ) {
4089 if ( isset( $options['broken'] ) ) {
4090 $file = false; // broken thumbnail forced by hook
4091 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
4092 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
4093 } else { // get by (name,timestamp)
4094 $file = wfFindFile( $title, $options );
4095 }
4096 return $file;
4097 }
4098
4099 /**
4100 * Transclude an interwiki link.
4101 *
4102 * @param Title $title
4103 * @param string $action
4104 *
4105 * @return string
4106 */
4107 public function interwikiTransclude( $title, $action ) {
4108 global $wgEnableScaryTranscluding;
4109
4110 if ( !$wgEnableScaryTranscluding ) {
4111 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
4112 }
4113
4114 $url = $title->getFullURL( [ 'action' => $action ] );
4115
4116 if ( strlen( $url ) > 255 ) {
4117 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
4118 }
4119 return $this->fetchScaryTemplateMaybeFromCache( $url );
4120 }
4121
4122 /**
4123 * @param string $url
4124 * @return mixed|string
4125 */
4126 public function fetchScaryTemplateMaybeFromCache( $url ) {
4127 global $wgTranscludeCacheExpiry;
4128 $dbr = wfGetDB( DB_SLAVE );
4129 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
4130 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
4131 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
4132 if ( $obj ) {
4133 return $obj->tc_contents;
4134 }
4135
4136 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
4137 $status = $req->execute(); // Status object
4138 if ( $status->isOK() ) {
4139 $text = $req->getContent();
4140 } elseif ( $req->getStatus() != 200 ) {
4141 // Though we failed to fetch the content, this status is useless.
4142 return wfMessage( 'scarytranscludefailed-httpstatus' )
4143 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
4144 } else {
4145 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4146 }
4147
4148 $dbw = wfGetDB( DB_MASTER );
4149 $dbw->replace( 'transcache', [ 'tc_url' ], [
4150 'tc_url' => $url,
4151 'tc_time' => $dbw->timestamp( time() ),
4152 'tc_contents' => $text
4153 ] );
4154 return $text;
4155 }
4156
4157 /**
4158 * Triple brace replacement -- used for template arguments
4159 * @private
4160 *
4161 * @param array $piece
4162 * @param PPFrame $frame
4163 *
4164 * @return array
4165 */
4166 public function argSubstitution( $piece, $frame ) {
4167
4168 $error = false;
4169 $parts = $piece['parts'];
4170 $nameWithSpaces = $frame->expand( $piece['title'] );
4171 $argName = trim( $nameWithSpaces );
4172 $object = false;
4173 $text = $frame->getArgument( $argName );
4174 if ( $text === false && $parts->getLength() > 0
4175 && ( $this->ot['html']
4176 || $this->ot['pre']
4177 || ( $this->ot['wiki'] && $frame->isTemplate() )
4178 )
4179 ) {
4180 # No match in frame, use the supplied default
4181 $object = $parts->item( 0 )->getChildren();
4182 }
4183 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4184 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4185 $this->limitationWarn( 'post-expand-template-argument' );
4186 }
4187
4188 if ( $text === false && $object === false ) {
4189 # No match anywhere
4190 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4191 }
4192 if ( $error !== false ) {
4193 $text .= $error;
4194 }
4195 if ( $object !== false ) {
4196 $ret = [ 'object' => $object ];
4197 } else {
4198 $ret = [ 'text' => $text ];
4199 }
4200
4201 return $ret;
4202 }
4203
4204 /**
4205 * Return the text to be used for a given extension tag.
4206 * This is the ghost of strip().
4207 *
4208 * @param array $params Associative array of parameters:
4209 * name PPNode for the tag name
4210 * attr PPNode for unparsed text where tag attributes are thought to be
4211 * attributes Optional associative array of parsed attributes
4212 * inner Contents of extension element
4213 * noClose Original text did not have a close tag
4214 * @param PPFrame $frame
4215 *
4216 * @throws MWException
4217 * @return string
4218 */
4219 public function extensionSubstitution( $params, $frame ) {
4220 $name = $frame->expand( $params['name'] );
4221 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
4222 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
4223 $marker = self::MARKER_PREFIX . "-$name-"
4224 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
4225
4226 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4227 ( $this->ot['html'] || $this->ot['pre'] );
4228 if ( $isFunctionTag ) {
4229 $markerType = 'none';
4230 } else {
4231 $markerType = 'general';
4232 }
4233 if ( $this->ot['html'] || $isFunctionTag ) {
4234 $name = strtolower( $name );
4235 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4236 if ( isset( $params['attributes'] ) ) {
4237 $attributes = $attributes + $params['attributes'];
4238 }
4239
4240 if ( isset( $this->mTagHooks[$name] ) ) {
4241 # Workaround for PHP bug 35229 and similar
4242 if ( !is_callable( $this->mTagHooks[$name] ) ) {
4243 throw new MWException( "Tag hook for $name is not callable\n" );
4244 }
4245 $output = call_user_func_array( $this->mTagHooks[$name],
4246 [ $content, $attributes, $this, $frame ] );
4247 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4248 list( $callback, ) = $this->mFunctionTagHooks[$name];
4249 if ( !is_callable( $callback ) ) {
4250 throw new MWException( "Tag hook for $name is not callable\n" );
4251 }
4252
4253 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
4254 } else {
4255 $output = '<span class="error">Invalid tag extension name: ' .
4256 htmlspecialchars( $name ) . '</span>';
4257 }
4258
4259 if ( is_array( $output ) ) {
4260 # Extract flags to local scope (to override $markerType)
4261 $flags = $output;
4262 $output = $flags[0];
4263 unset( $flags[0] );
4264 extract( $flags );
4265 }
4266 } else {
4267 if ( is_null( $attrText ) ) {
4268 $attrText = '';
4269 }
4270 if ( isset( $params['attributes'] ) ) {
4271 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4272 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4273 htmlspecialchars( $attrValue ) . '"';
4274 }
4275 }
4276 if ( $content === null ) {
4277 $output = "<$name$attrText/>";
4278 } else {
4279 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4280 $output = "<$name$attrText>$content$close";
4281 }
4282 }
4283
4284 if ( $markerType === 'none' ) {
4285 return $output;
4286 } elseif ( $markerType === 'nowiki' ) {
4287 $this->mStripState->addNoWiki( $marker, $output );
4288 } elseif ( $markerType === 'general' ) {
4289 $this->mStripState->addGeneral( $marker, $output );
4290 } else {
4291 throw new MWException( __METHOD__ . ': invalid marker type' );
4292 }
4293 return $marker;
4294 }
4295
4296 /**
4297 * Increment an include size counter
4298 *
4299 * @param string $type The type of expansion
4300 * @param int $size The size of the text
4301 * @return bool False if this inclusion would take it over the maximum, true otherwise
4302 */
4303 public function incrementIncludeSize( $type, $size ) {
4304 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4305 return false;
4306 } else {
4307 $this->mIncludeSizes[$type] += $size;
4308 return true;
4309 }
4310 }
4311
4312 /**
4313 * Increment the expensive function count
4314 *
4315 * @return bool False if the limit has been exceeded
4316 */
4317 public function incrementExpensiveFunctionCount() {
4318 $this->mExpensiveFunctionCount++;
4319 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4320 }
4321
4322 /**
4323 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4324 * Fills $this->mDoubleUnderscores, returns the modified text
4325 *
4326 * @param string $text
4327 *
4328 * @return string
4329 */
4330 public function doDoubleUnderscore( $text ) {
4331
4332 # The position of __TOC__ needs to be recorded
4333 $mw = MagicWord::get( 'toc' );
4334 if ( $mw->match( $text ) ) {
4335 $this->mShowToc = true;
4336 $this->mForceTocPosition = true;
4337
4338 # Set a placeholder. At the end we'll fill it in with the TOC.
4339 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
4340
4341 # Only keep the first one.
4342 $text = $mw->replace( '', $text );
4343 }
4344
4345 # Now match and remove the rest of them
4346 $mwa = MagicWord::getDoubleUnderscoreArray();
4347 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4348
4349 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4350 $this->mOutput->mNoGallery = true;
4351 }
4352 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4353 $this->mShowToc = false;
4354 }
4355 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4356 && $this->mTitle->getNamespace() == NS_CATEGORY
4357 ) {
4358 $this->addTrackingCategory( 'hidden-category-category' );
4359 }
4360 # (bug 8068) Allow control over whether robots index a page.
4361 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
4362 # is not desirable, the last one on the page should win.
4363 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4364 $this->mOutput->setIndexPolicy( 'noindex' );
4365 $this->addTrackingCategory( 'noindex-category' );
4366 }
4367 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4368 $this->mOutput->setIndexPolicy( 'index' );
4369 $this->addTrackingCategory( 'index-category' );
4370 }
4371
4372 # Cache all double underscores in the database
4373 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4374 $this->mOutput->setProperty( $key, '' );
4375 }
4376
4377 return $text;
4378 }
4379
4380 /**
4381 * @see ParserOutput::addTrackingCategory()
4382 * @param string $msg Message key
4383 * @return bool Whether the addition was successful
4384 */
4385 public function addTrackingCategory( $msg ) {
4386 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4387 }
4388
4389 /**
4390 * This function accomplishes several tasks:
4391 * 1) Auto-number headings if that option is enabled
4392 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4393 * 3) Add a Table of contents on the top for users who have enabled the option
4394 * 4) Auto-anchor headings
4395 *
4396 * It loops through all headlines, collects the necessary data, then splits up the
4397 * string and re-inserts the newly formatted headlines.
4398 *
4399 * @param string $text
4400 * @param string $origText Original, untouched wikitext
4401 * @param bool $isMain
4402 * @return mixed|string
4403 * @private
4404 */
4405 public function formatHeadings( $text, $origText, $isMain = true ) {
4406 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4407
4408 # Inhibit editsection links if requested in the page
4409 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4410 $maybeShowEditLink = $showEditLink = false;
4411 } else {
4412 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4413 $showEditLink = $this->mOptions->getEditSection();
4414 }
4415 if ( $showEditLink ) {
4416 $this->mOutput->setEditSectionTokens( true );
4417 }
4418
4419 # Get all headlines for numbering them and adding funky stuff like [edit]
4420 # links - this is for later, but we need the number of headlines right now
4421 $matches = [];
4422 $numMatches = preg_match_all(
4423 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4424 $text,
4425 $matches
4426 );
4427
4428 # if there are fewer than 4 headlines in the article, do not show TOC
4429 # unless it's been explicitly enabled.
4430 $enoughToc = $this->mShowToc &&
4431 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4432
4433 # Allow user to stipulate that a page should have a "new section"
4434 # link added via __NEWSECTIONLINK__
4435 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4436 $this->mOutput->setNewSection( true );
4437 }
4438
4439 # Allow user to remove the "new section"
4440 # link via __NONEWSECTIONLINK__
4441 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4442 $this->mOutput->hideNewSection( true );
4443 }
4444
4445 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4446 # override above conditions and always show TOC above first header
4447 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4448 $this->mShowToc = true;
4449 $enoughToc = true;
4450 }
4451
4452 # headline counter
4453 $headlineCount = 0;
4454 $numVisible = 0;
4455
4456 # Ugh .. the TOC should have neat indentation levels which can be
4457 # passed to the skin functions. These are determined here
4458 $toc = '';
4459 $full = '';
4460 $head = [];
4461 $sublevelCount = [];
4462 $levelCount = [];
4463 $level = 0;
4464 $prevlevel = 0;
4465 $toclevel = 0;
4466 $prevtoclevel = 0;
4467 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4468 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4469 $oldType = $this->mOutputType;
4470 $this->setOutputType( self::OT_WIKI );
4471 $frame = $this->getPreprocessor()->newFrame();
4472 $root = $this->preprocessToDom( $origText );
4473 $node = $root->getFirstChild();
4474 $byteOffset = 0;
4475 $tocraw = [];
4476 $refers = [];
4477
4478 $headlines = $numMatches !== false ? $matches[3] : [];
4479
4480 foreach ( $headlines as $headline ) {
4481 $isTemplate = false;
4482 $titleText = false;
4483 $sectionIndex = false;
4484 $numbering = '';
4485 $markerMatches = [];
4486 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4487 $serial = $markerMatches[1];
4488 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4489 $isTemplate = ( $titleText != $baseTitleText );
4490 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4491 }
4492
4493 if ( $toclevel ) {
4494 $prevlevel = $level;
4495 }
4496 $level = $matches[1][$headlineCount];
4497
4498 if ( $level > $prevlevel ) {
4499 # Increase TOC level
4500 $toclevel++;
4501 $sublevelCount[$toclevel] = 0;
4502 if ( $toclevel < $wgMaxTocLevel ) {
4503 $prevtoclevel = $toclevel;
4504 $toc .= Linker::tocIndent();
4505 $numVisible++;
4506 }
4507 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4508 # Decrease TOC level, find level to jump to
4509
4510 for ( $i = $toclevel; $i > 0; $i-- ) {
4511 if ( $levelCount[$i] == $level ) {
4512 # Found last matching level
4513 $toclevel = $i;
4514 break;
4515 } elseif ( $levelCount[$i] < $level ) {
4516 # Found first matching level below current level
4517 $toclevel = $i + 1;
4518 break;
4519 }
4520 }
4521 if ( $i == 0 ) {
4522 $toclevel = 1;
4523 }
4524 if ( $toclevel < $wgMaxTocLevel ) {
4525 if ( $prevtoclevel < $wgMaxTocLevel ) {
4526 # Unindent only if the previous toc level was shown :p
4527 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4528 $prevtoclevel = $toclevel;
4529 } else {
4530 $toc .= Linker::tocLineEnd();
4531 }
4532 }
4533 } else {
4534 # No change in level, end TOC line
4535 if ( $toclevel < $wgMaxTocLevel ) {
4536 $toc .= Linker::tocLineEnd();
4537 }
4538 }
4539
4540 $levelCount[$toclevel] = $level;
4541
4542 # count number of headlines for each level
4543 $sublevelCount[$toclevel]++;
4544 $dot = 0;
4545 for ( $i = 1; $i <= $toclevel; $i++ ) {
4546 if ( !empty( $sublevelCount[$i] ) ) {
4547 if ( $dot ) {
4548 $numbering .= '.';
4549 }
4550 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4551 $dot = 1;
4552 }
4553 }
4554
4555 # The safe header is a version of the header text safe to use for links
4556
4557 # Remove link placeholders by the link text.
4558 # <!--LINK number-->
4559 # turns into
4560 # link text with suffix
4561 # Do this before unstrip since link text can contain strip markers
4562 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4563
4564 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4565 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4566
4567 # Strip out HTML (first regex removes any tag not allowed)
4568 # Allowed tags are:
4569 # * <sup> and <sub> (bug 8393)
4570 # * <i> (bug 26375)
4571 # * <b> (r105284)
4572 # * <bdi> (bug 72884)
4573 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4574 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4575 # to allow setting directionality in toc items.
4576 $tocline = preg_replace(
4577 [
4578 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4579 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4580 ],
4581 [ '', '<$1>' ],
4582 $safeHeadline
4583 );
4584
4585 # Strip '<span></span>', which is the result from the above if
4586 # <span id="foo"></span> is used to produce an additional anchor
4587 # for a section.
4588 $tocline = str_replace( '<span></span>', '', $tocline );
4589
4590 $tocline = trim( $tocline );
4591
4592 # For the anchor, strip out HTML-y stuff period
4593 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4594 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4595
4596 # Save headline for section edit hint before it's escaped
4597 $headlineHint = $safeHeadline;
4598
4599 if ( $wgExperimentalHtmlIds ) {
4600 # For reverse compatibility, provide an id that's
4601 # HTML4-compatible, like we used to.
4602 # It may be worth noting, academically, that it's possible for
4603 # the legacy anchor to conflict with a non-legacy headline
4604 # anchor on the page. In this case likely the "correct" thing
4605 # would be to either drop the legacy anchors or make sure
4606 # they're numbered first. However, this would require people
4607 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4608 # manually, so let's not bother worrying about it.
4609 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4610 [ 'noninitial', 'legacy' ] );
4611 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4612
4613 if ( $legacyHeadline == $safeHeadline ) {
4614 # No reason to have both (in fact, we can't)
4615 $legacyHeadline = false;
4616 }
4617 } else {
4618 $legacyHeadline = false;
4619 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4620 'noninitial' );
4621 }
4622
4623 # HTML names must be case-insensitively unique (bug 10721).
4624 # This does not apply to Unicode characters per
4625 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4626 # @todo FIXME: We may be changing them depending on the current locale.
4627 $arrayKey = strtolower( $safeHeadline );
4628 if ( $legacyHeadline === false ) {
4629 $legacyArrayKey = false;
4630 } else {
4631 $legacyArrayKey = strtolower( $legacyHeadline );
4632 }
4633
4634 # Create the anchor for linking from the TOC to the section
4635 $anchor = $safeHeadline;
4636 $legacyAnchor = $legacyHeadline;
4637 if ( isset( $refers[$arrayKey] ) ) {
4638 // @codingStandardsIgnoreStart
4639 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4640 // @codingStandardsIgnoreEnd
4641 $anchor .= "_$i";
4642 $refers["${arrayKey}_$i"] = true;
4643 } else {
4644 $refers[$arrayKey] = true;
4645 }
4646 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4647 // @codingStandardsIgnoreStart
4648 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4649 // @codingStandardsIgnoreEnd
4650 $legacyAnchor .= "_$i";
4651 $refers["${legacyArrayKey}_$i"] = true;
4652 } else {
4653 $refers[$legacyArrayKey] = true;
4654 }
4655
4656 # Don't number the heading if it is the only one (looks silly)
4657 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4658 # the two are different if the line contains a link
4659 $headline = Html::element(
4660 'span',
4661 [ 'class' => 'mw-headline-number' ],
4662 $numbering
4663 ) . ' ' . $headline;
4664 }
4665
4666 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4667 $toc .= Linker::tocLine( $anchor, $tocline,
4668 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4669 }
4670
4671 # Add the section to the section tree
4672 # Find the DOM node for this header
4673 $noOffset = ( $isTemplate || $sectionIndex === false );
4674 while ( $node && !$noOffset ) {
4675 if ( $node->getName() === 'h' ) {
4676 $bits = $node->splitHeading();
4677 if ( $bits['i'] == $sectionIndex ) {
4678 break;
4679 }
4680 }
4681 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4682 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4683 $node = $node->getNextSibling();
4684 }
4685 $tocraw[] = [
4686 'toclevel' => $toclevel,
4687 'level' => $level,
4688 'line' => $tocline,
4689 'number' => $numbering,
4690 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4691 'fromtitle' => $titleText,
4692 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4693 'anchor' => $anchor,
4694 ];
4695
4696 # give headline the correct <h#> tag
4697 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4698 // Output edit section links as markers with styles that can be customized by skins
4699 if ( $isTemplate ) {
4700 # Put a T flag in the section identifier, to indicate to extractSections()
4701 # that sections inside <includeonly> should be counted.
4702 $editsectionPage = $titleText;
4703 $editsectionSection = "T-$sectionIndex";
4704 $editsectionContent = null;
4705 } else {
4706 $editsectionPage = $this->mTitle->getPrefixedText();
4707 $editsectionSection = $sectionIndex;
4708 $editsectionContent = $headlineHint;
4709 }
4710 // We use a bit of pesudo-xml for editsection markers. The
4711 // language converter is run later on. Using a UNIQ style marker
4712 // leads to the converter screwing up the tokens when it
4713 // converts stuff. And trying to insert strip tags fails too. At
4714 // this point all real inputted tags have already been escaped,
4715 // so we don't have to worry about a user trying to input one of
4716 // these markers directly. We use a page and section attribute
4717 // to stop the language converter from converting these
4718 // important bits of data, but put the headline hint inside a
4719 // content block because the language converter is supposed to
4720 // be able to convert that piece of data.
4721 // Gets replaced with html in ParserOutput::getText
4722 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4723 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4724 if ( $editsectionContent !== null ) {
4725 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4726 } else {
4727 $editlink .= '/>';
4728 }
4729 } else {
4730 $editlink = '';
4731 }
4732 $head[$headlineCount] = Linker::makeHeadline( $level,
4733 $matches['attrib'][$headlineCount], $anchor, $headline,
4734 $editlink, $legacyAnchor );
4735
4736 $headlineCount++;
4737 }
4738
4739 $this->setOutputType( $oldType );
4740
4741 # Never ever show TOC if no headers
4742 if ( $numVisible < 1 ) {
4743 $enoughToc = false;
4744 }
4745
4746 if ( $enoughToc ) {
4747 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4748 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4749 }
4750 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4751 $this->mOutput->setTOCHTML( $toc );
4752 $toc = self::TOC_START . $toc . self::TOC_END;
4753 $this->mOutput->addModules( 'mediawiki.toc' );
4754 }
4755
4756 if ( $isMain ) {
4757 $this->mOutput->setSections( $tocraw );
4758 }
4759
4760 # split up and insert constructed headlines
4761 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4762 $i = 0;
4763
4764 // build an array of document sections
4765 $sections = [];
4766 foreach ( $blocks as $block ) {
4767 // $head is zero-based, sections aren't.
4768 if ( empty( $head[$i - 1] ) ) {
4769 $sections[$i] = $block;
4770 } else {
4771 $sections[$i] = $head[$i - 1] . $block;
4772 }
4773
4774 /**
4775 * Send a hook, one per section.
4776 * The idea here is to be able to make section-level DIVs, but to do so in a
4777 * lower-impact, more correct way than r50769
4778 *
4779 * $this : caller
4780 * $section : the section number
4781 * &$sectionContent : ref to the content of the section
4782 * $showEditLinks : boolean describing whether this section has an edit link
4783 */
4784 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4785
4786 $i++;
4787 }
4788
4789 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4790 // append the TOC at the beginning
4791 // Top anchor now in skin
4792 $sections[0] = $sections[0] . $toc . "\n";
4793 }
4794
4795 $full .= implode( '', $sections );
4796
4797 if ( $this->mForceTocPosition ) {
4798 return str_replace( '<!--MWTOC-->', $toc, $full );
4799 } else {
4800 return $full;
4801 }
4802 }
4803
4804 /**
4805 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4806 * conversion, substituting signatures, {{subst:}} templates, etc.
4807 *
4808 * @param string $text The text to transform
4809 * @param Title $title The Title object for the current article
4810 * @param User $user The User object describing the current user
4811 * @param ParserOptions $options Parsing options
4812 * @param bool $clearState Whether to clear the parser state first
4813 * @return string The altered wiki markup
4814 */
4815 public function preSaveTransform( $text, Title $title, User $user,
4816 ParserOptions $options, $clearState = true
4817 ) {
4818 if ( $clearState ) {
4819 $magicScopeVariable = $this->lock();
4820 }
4821 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4822 $this->setUser( $user );
4823
4824 $pairs = [
4825 "\r\n" => "\n",
4826 "\r" => "\n",
4827 ];
4828 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4829 if ( $options->getPreSaveTransform() ) {
4830 $text = $this->pstPass2( $text, $user );
4831 }
4832 $text = $this->mStripState->unstripBoth( $text );
4833
4834 $this->setUser( null ); # Reset
4835
4836 return $text;
4837 }
4838
4839 /**
4840 * Pre-save transform helper function
4841 *
4842 * @param string $text
4843 * @param User $user
4844 *
4845 * @return string
4846 */
4847 private function pstPass2( $text, $user ) {
4848 global $wgContLang;
4849
4850 # Note: This is the timestamp saved as hardcoded wikitext to
4851 # the database, we use $wgContLang here in order to give
4852 # everyone the same signature and use the default one rather
4853 # than the one selected in each user's preferences.
4854 # (see also bug 12815)
4855 $ts = $this->mOptions->getTimestamp();
4856 $timestamp = MWTimestamp::getLocalInstance( $ts );
4857 $ts = $timestamp->format( 'YmdHis' );
4858 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4859
4860 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4861
4862 # Variable replacement
4863 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4864 $text = $this->replaceVariables( $text );
4865
4866 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4867 # which may corrupt this parser instance via its wfMessage()->text() call-
4868
4869 # Signatures
4870 $sigText = $this->getUserSig( $user );
4871 $text = strtr( $text, [
4872 '~~~~~' => $d,
4873 '~~~~' => "$sigText $d",
4874 '~~~' => $sigText
4875 ] );
4876
4877 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4878 $tc = '[' . Title::legalChars() . ']';
4879 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4880
4881 // [[ns:page (context)|]]
4882 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4883 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4884 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4885 // [[ns:page (context), context|]] (using either single or double-width comma)
4886 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4887 // [[|page]] (reverse pipe trick: add context from page title)
4888 $p2 = "/\[\[\\|($tc+)]]/";
4889
4890 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4891 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4892 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4893 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4894
4895 $t = $this->mTitle->getText();
4896 $m = [];
4897 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4898 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4899 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4900 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4901 } else {
4902 # if there's no context, don't bother duplicating the title
4903 $text = preg_replace( $p2, '[[\\1]]', $text );
4904 }
4905
4906 # Trim trailing whitespace
4907 $text = rtrim( $text );
4908
4909 return $text;
4910 }
4911
4912 /**
4913 * Fetch the user's signature text, if any, and normalize to
4914 * validated, ready-to-insert wikitext.
4915 * If you have pre-fetched the nickname or the fancySig option, you can
4916 * specify them here to save a database query.
4917 * Do not reuse this parser instance after calling getUserSig(),
4918 * as it may have changed if it's the $wgParser.
4919 *
4920 * @param User $user
4921 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4922 * @param bool|null $fancySig whether the nicknname is the complete signature
4923 * or null to use default value
4924 * @return string
4925 */
4926 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4927 global $wgMaxSigChars;
4928
4929 $username = $user->getName();
4930
4931 # If not given, retrieve from the user object.
4932 if ( $nickname === false ) {
4933 $nickname = $user->getOption( 'nickname' );
4934 }
4935
4936 if ( is_null( $fancySig ) ) {
4937 $fancySig = $user->getBoolOption( 'fancysig' );
4938 }
4939
4940 $nickname = $nickname == null ? $username : $nickname;
4941
4942 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4943 $nickname = $username;
4944 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4945 } elseif ( $fancySig !== false ) {
4946 # Sig. might contain markup; validate this
4947 if ( $this->validateSig( $nickname ) !== false ) {
4948 # Validated; clean up (if needed) and return it
4949 return $this->cleanSig( $nickname, true );
4950 } else {
4951 # Failed to validate; fall back to the default
4952 $nickname = $username;
4953 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4954 }
4955 }
4956
4957 # Make sure nickname doesnt get a sig in a sig
4958 $nickname = self::cleanSigInSig( $nickname );
4959
4960 # If we're still here, make it a link to the user page
4961 $userText = wfEscapeWikiText( $username );
4962 $nickText = wfEscapeWikiText( $nickname );
4963 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4964
4965 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4966 ->title( $this->getTitle() )->text();
4967 }
4968
4969 /**
4970 * Check that the user's signature contains no bad XML
4971 *
4972 * @param string $text
4973 * @return string|bool An expanded string, or false if invalid.
4974 */
4975 public function validateSig( $text ) {
4976 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4977 }
4978
4979 /**
4980 * Clean up signature text
4981 *
4982 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4983 * 2) Substitute all transclusions
4984 *
4985 * @param string $text
4986 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4987 * @return string Signature text
4988 */
4989 public function cleanSig( $text, $parsing = false ) {
4990 if ( !$parsing ) {
4991 global $wgTitle;
4992 $magicScopeVariable = $this->lock();
4993 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4994 }
4995
4996 # Option to disable this feature
4997 if ( !$this->mOptions->getCleanSignatures() ) {
4998 return $text;
4999 }
5000
5001 # @todo FIXME: Regex doesn't respect extension tags or nowiki
5002 # => Move this logic to braceSubstitution()
5003 $substWord = MagicWord::get( 'subst' );
5004 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
5005 $substText = '{{' . $substWord->getSynonym( 0 );
5006
5007 $text = preg_replace( $substRegex, $substText, $text );
5008 $text = self::cleanSigInSig( $text );
5009 $dom = $this->preprocessToDom( $text );
5010 $frame = $this->getPreprocessor()->newFrame();
5011 $text = $frame->expand( $dom );
5012
5013 if ( !$parsing ) {
5014 $text = $this->mStripState->unstripBoth( $text );
5015 }
5016
5017 return $text;
5018 }
5019
5020 /**
5021 * Strip 3, 4 or 5 tildes out of signatures.
5022 *
5023 * @param string $text
5024 * @return string Signature text with /~{3,5}/ removed
5025 */
5026 public static function cleanSigInSig( $text ) {
5027 $text = preg_replace( '/~{3,5}/', '', $text );
5028 return $text;
5029 }
5030
5031 /**
5032 * Set up some variables which are usually set up in parse()
5033 * so that an external function can call some class members with confidence
5034 *
5035 * @param Title|null $title
5036 * @param ParserOptions $options
5037 * @param int $outputType
5038 * @param bool $clearState
5039 */
5040 public function startExternalParse( Title $title = null, ParserOptions $options,
5041 $outputType, $clearState = true
5042 ) {
5043 $this->startParse( $title, $options, $outputType, $clearState );
5044 }
5045
5046 /**
5047 * @param Title|null $title
5048 * @param ParserOptions $options
5049 * @param int $outputType
5050 * @param bool $clearState
5051 */
5052 private function startParse( Title $title = null, ParserOptions $options,
5053 $outputType, $clearState = true
5054 ) {
5055 $this->setTitle( $title );
5056 $this->mOptions = $options;
5057 $this->setOutputType( $outputType );
5058 if ( $clearState ) {
5059 $this->clearState();
5060 }
5061 }
5062
5063 /**
5064 * Wrapper for preprocess()
5065 *
5066 * @param string $text The text to preprocess
5067 * @param ParserOptions $options Options
5068 * @param Title|null $title Title object or null to use $wgTitle
5069 * @return string
5070 */
5071 public function transformMsg( $text, $options, $title = null ) {
5072 static $executing = false;
5073
5074 # Guard against infinite recursion
5075 if ( $executing ) {
5076 return $text;
5077 }
5078 $executing = true;
5079
5080 if ( !$title ) {
5081 global $wgTitle;
5082 $title = $wgTitle;
5083 }
5084
5085 $text = $this->preprocess( $text, $title, $options );
5086
5087 $executing = false;
5088 return $text;
5089 }
5090
5091 /**
5092 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
5093 * The callback should have the following form:
5094 * function myParserHook( $text, $params, $parser, $frame ) { ... }
5095 *
5096 * Transform and return $text. Use $parser for any required context, e.g. use
5097 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
5098 *
5099 * Hooks may return extended information by returning an array, of which the
5100 * first numbered element (index 0) must be the return string, and all other
5101 * entries are extracted into local variables within an internal function
5102 * in the Parser class.
5103 *
5104 * This interface (introduced r61913) appears to be undocumented, but
5105 * 'markerType' is used by some core tag hooks to override which strip
5106 * array their results are placed in. **Use great caution if attempting
5107 * this interface, as it is not documented and injudicious use could smash
5108 * private variables.**
5109 *
5110 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5111 * @param callable $callback The callback function (and object) to use for the tag
5112 * @throws MWException
5113 * @return callable|null The old value of the mTagHooks array associated with the hook
5114 */
5115 public function setHook( $tag, $callback ) {
5116 $tag = strtolower( $tag );
5117 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5118 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
5119 }
5120 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
5121 $this->mTagHooks[$tag] = $callback;
5122 if ( !in_array( $tag, $this->mStripList ) ) {
5123 $this->mStripList[] = $tag;
5124 }
5125
5126 return $oldVal;
5127 }
5128
5129 /**
5130 * As setHook(), but letting the contents be parsed.
5131 *
5132 * Transparent tag hooks are like regular XML-style tag hooks, except they
5133 * operate late in the transformation sequence, on HTML instead of wikitext.
5134 *
5135 * This is probably obsoleted by things dealing with parser frames?
5136 * The only extension currently using it is geoserver.
5137 *
5138 * @since 1.10
5139 * @todo better document or deprecate this
5140 *
5141 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5142 * @param callable $callback The callback function (and object) to use for the tag
5143 * @throws MWException
5144 * @return callable|null The old value of the mTagHooks array associated with the hook
5145 */
5146 public function setTransparentTagHook( $tag, $callback ) {
5147 $tag = strtolower( $tag );
5148 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5149 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5150 }
5151 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
5152 $this->mTransparentTagHooks[$tag] = $callback;
5153
5154 return $oldVal;
5155 }
5156
5157 /**
5158 * Remove all tag hooks
5159 */
5160 public function clearTagHooks() {
5161 $this->mTagHooks = [];
5162 $this->mFunctionTagHooks = [];
5163 $this->mStripList = $this->mDefaultStripList;
5164 }
5165
5166 /**
5167 * Create a function, e.g. {{sum:1|2|3}}
5168 * The callback function should have the form:
5169 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5170 *
5171 * Or with Parser::SFH_OBJECT_ARGS:
5172 * function myParserFunction( $parser, $frame, $args ) { ... }
5173 *
5174 * The callback may either return the text result of the function, or an array with the text
5175 * in element 0, and a number of flags in the other elements. The names of the flags are
5176 * specified in the keys. Valid flags are:
5177 * found The text returned is valid, stop processing the template. This
5178 * is on by default.
5179 * nowiki Wiki markup in the return value should be escaped
5180 * isHTML The returned text is HTML, armour it against wikitext transformation
5181 *
5182 * @param string $id The magic word ID
5183 * @param callable $callback The callback function (and object) to use
5184 * @param int $flags A combination of the following flags:
5185 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5186 *
5187 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5188 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5189 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5190 * the arguments, and to control the way they are expanded.
5191 *
5192 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5193 * arguments, for instance:
5194 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5195 *
5196 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5197 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5198 * working if/when this is changed.
5199 *
5200 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5201 * expansion.
5202 *
5203 * Please read the documentation in includes/parser/Preprocessor.php for more information
5204 * about the methods available in PPFrame and PPNode.
5205 *
5206 * @throws MWException
5207 * @return string|callable The old callback function for this name, if any
5208 */
5209 public function setFunctionHook( $id, $callback, $flags = 0 ) {
5210 global $wgContLang;
5211
5212 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
5213 $this->mFunctionHooks[$id] = [ $callback, $flags ];
5214
5215 # Add to function cache
5216 $mw = MagicWord::get( $id );
5217 if ( !$mw ) {
5218 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
5219 }
5220
5221 $synonyms = $mw->getSynonyms();
5222 $sensitive = intval( $mw->isCaseSensitive() );
5223
5224 foreach ( $synonyms as $syn ) {
5225 # Case
5226 if ( !$sensitive ) {
5227 $syn = $wgContLang->lc( $syn );
5228 }
5229 # Add leading hash
5230 if ( !( $flags & self::SFH_NO_HASH ) ) {
5231 $syn = '#' . $syn;
5232 }
5233 # Remove trailing colon
5234 if ( substr( $syn, -1, 1 ) === ':' ) {
5235 $syn = substr( $syn, 0, -1 );
5236 }
5237 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5238 }
5239 return $oldVal;
5240 }
5241
5242 /**
5243 * Get all registered function hook identifiers
5244 *
5245 * @return array
5246 */
5247 public function getFunctionHooks() {
5248 return array_keys( $this->mFunctionHooks );
5249 }
5250
5251 /**
5252 * Create a tag function, e.g. "<test>some stuff</test>".
5253 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5254 * Unlike parser functions, their content is not preprocessed.
5255 * @param string $tag
5256 * @param callable $callback
5257 * @param int $flags
5258 * @throws MWException
5259 * @return null
5260 */
5261 public function setFunctionTagHook( $tag, $callback, $flags ) {
5262 $tag = strtolower( $tag );
5263 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5264 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5265 }
5266 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
5267 $this->mFunctionTagHooks[$tag] : null;
5268 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
5269
5270 if ( !in_array( $tag, $this->mStripList ) ) {
5271 $this->mStripList[] = $tag;
5272 }
5273
5274 return $old;
5275 }
5276
5277 /**
5278 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5279 * Placeholders created in Linker::link()
5280 *
5281 * @param string $text
5282 * @param int $options
5283 */
5284 public function replaceLinkHolders( &$text, $options = 0 ) {
5285 $this->mLinkHolders->replace( $text );
5286 }
5287
5288 /**
5289 * Replace "<!--LINK-->" link placeholders with plain text of links
5290 * (not HTML-formatted).
5291 *
5292 * @param string $text
5293 * @return string
5294 */
5295 public function replaceLinkHoldersText( $text ) {
5296 return $this->mLinkHolders->replaceText( $text );
5297 }
5298
5299 /**
5300 * Renders an image gallery from a text with one line per image.
5301 * text labels may be given by using |-style alternative text. E.g.
5302 * Image:one.jpg|The number "1"
5303 * Image:tree.jpg|A tree
5304 * given as text will return the HTML of a gallery with two images,
5305 * labeled 'The number "1"' and
5306 * 'A tree'.
5307 *
5308 * @param string $text
5309 * @param array $params
5310 * @return string HTML
5311 */
5312 public function renderImageGallery( $text, $params ) {
5313
5314 $mode = false;
5315 if ( isset( $params['mode'] ) ) {
5316 $mode = $params['mode'];
5317 }
5318
5319 try {
5320 $ig = ImageGalleryBase::factory( $mode );
5321 } catch ( Exception $e ) {
5322 // If invalid type set, fallback to default.
5323 $ig = ImageGalleryBase::factory( false );
5324 }
5325
5326 $ig->setContextTitle( $this->mTitle );
5327 $ig->setShowBytes( false );
5328 $ig->setShowFilename( false );
5329 $ig->setParser( $this );
5330 $ig->setHideBadImages();
5331 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
5332
5333 if ( isset( $params['showfilename'] ) ) {
5334 $ig->setShowFilename( true );
5335 } else {
5336 $ig->setShowFilename( false );
5337 }
5338 if ( isset( $params['caption'] ) ) {
5339 $caption = $params['caption'];
5340 $caption = htmlspecialchars( $caption );
5341 $caption = $this->replaceInternalLinks( $caption );
5342 $ig->setCaptionHtml( $caption );
5343 }
5344 if ( isset( $params['perrow'] ) ) {
5345 $ig->setPerRow( $params['perrow'] );
5346 }
5347 if ( isset( $params['widths'] ) ) {
5348 $ig->setWidths( $params['widths'] );
5349 }
5350 if ( isset( $params['heights'] ) ) {
5351 $ig->setHeights( $params['heights'] );
5352 }
5353 $ig->setAdditionalOptions( $params );
5354
5355 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
5356
5357 $lines = StringUtils::explode( "\n", $text );
5358 foreach ( $lines as $line ) {
5359 # match lines like these:
5360 # Image:someimage.jpg|This is some image
5361 $matches = [];
5362 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5363 # Skip empty lines
5364 if ( count( $matches ) == 0 ) {
5365 continue;
5366 }
5367
5368 if ( strpos( $matches[0], '%' ) !== false ) {
5369 $matches[1] = rawurldecode( $matches[1] );
5370 }
5371 $title = Title::newFromText( $matches[1], NS_FILE );
5372 if ( is_null( $title ) ) {
5373 # Bogus title. Ignore these so we don't bomb out later.
5374 continue;
5375 }
5376
5377 # We need to get what handler the file uses, to figure out parameters.
5378 # Note, a hook can overide the file name, and chose an entirely different
5379 # file (which potentially could be of a different type and have different handler).
5380 $options = [];
5381 $descQuery = false;
5382 Hooks::run( 'BeforeParserFetchFileAndTitle',
5383 [ $this, $title, &$options, &$descQuery ] );
5384 # Don't register it now, as ImageGallery does that later.
5385 $file = $this->fetchFileNoRegister( $title, $options );
5386 $handler = $file ? $file->getHandler() : false;
5387
5388 $paramMap = [
5389 'img_alt' => 'gallery-internal-alt',
5390 'img_link' => 'gallery-internal-link',
5391 ];
5392 if ( $handler ) {
5393 $paramMap = $paramMap + $handler->getParamMap();
5394 // We don't want people to specify per-image widths.
5395 // Additionally the width parameter would need special casing anyhow.
5396 unset( $paramMap['img_width'] );
5397 }
5398
5399 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5400
5401 $label = '';
5402 $alt = '';
5403 $link = '';
5404 $handlerOptions = [];
5405 if ( isset( $matches[3] ) ) {
5406 // look for an |alt= definition while trying not to break existing
5407 // captions with multiple pipes (|) in it, until a more sensible grammar
5408 // is defined for images in galleries
5409
5410 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5411 // splitting on '|' is a bit odd, and different from makeImage.
5412 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5413 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5414
5415 foreach ( $parameterMatches as $parameterMatch ) {
5416 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5417 if ( $magicName ) {
5418 $paramName = $paramMap[$magicName];
5419
5420 switch ( $paramName ) {
5421 case 'gallery-internal-alt':
5422 $alt = $this->stripAltText( $match, false );
5423 break;
5424 case 'gallery-internal-link':
5425 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5426 $chars = self::EXT_LINK_URL_CLASS;
5427 $addr = self::EXT_LINK_ADDR;
5428 $prots = $this->mUrlProtocols;
5429 // check to see if link matches an absolute url, if not then it must be a wiki link.
5430 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5431 $link = $linkValue;
5432 } else {
5433 $localLinkTitle = Title::newFromText( $linkValue );
5434 if ( $localLinkTitle !== null ) {
5435 $link = $localLinkTitle->getLinkURL();
5436 }
5437 }
5438 break;
5439 default:
5440 // Must be a handler specific parameter.
5441 if ( $handler->validateParam( $paramName, $match ) ) {
5442 $handlerOptions[$paramName] = $match;
5443 } else {
5444 // Guess not, consider it as caption.
5445 wfDebug( "$parameterMatch failed parameter validation\n" );
5446 $label = '|' . $parameterMatch;
5447 }
5448 }
5449
5450 } else {
5451 // Last pipe wins.
5452 $label = '|' . $parameterMatch;
5453 }
5454 }
5455 // Remove the pipe.
5456 $label = substr( $label, 1 );
5457 }
5458
5459 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5460 }
5461 $html = $ig->toHTML();
5462 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5463 return $html;
5464 }
5465
5466 /**
5467 * @param MediaHandler $handler
5468 * @return array
5469 */
5470 public function getImageParams( $handler ) {
5471 if ( $handler ) {
5472 $handlerClass = get_class( $handler );
5473 } else {
5474 $handlerClass = '';
5475 }
5476 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5477 # Initialise static lists
5478 static $internalParamNames = [
5479 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5480 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5481 'bottom', 'text-bottom' ],
5482 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5483 'upright', 'border', 'link', 'alt', 'class' ],
5484 ];
5485 static $internalParamMap;
5486 if ( !$internalParamMap ) {
5487 $internalParamMap = [];
5488 foreach ( $internalParamNames as $type => $names ) {
5489 foreach ( $names as $name ) {
5490 $magicName = str_replace( '-', '_', "img_$name" );
5491 $internalParamMap[$magicName] = [ $type, $name ];
5492 }
5493 }
5494 }
5495
5496 # Add handler params
5497 $paramMap = $internalParamMap;
5498 if ( $handler ) {
5499 $handlerParamMap = $handler->getParamMap();
5500 foreach ( $handlerParamMap as $magic => $paramName ) {
5501 $paramMap[$magic] = [ 'handler', $paramName ];
5502 }
5503 }
5504 $this->mImageParams[$handlerClass] = $paramMap;
5505 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5506 }
5507 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5508 }
5509
5510 /**
5511 * Parse image options text and use it to make an image
5512 *
5513 * @param Title $title
5514 * @param string $options
5515 * @param LinkHolderArray|bool $holders
5516 * @return string HTML
5517 */
5518 public function makeImage( $title, $options, $holders = false ) {
5519 # Check if the options text is of the form "options|alt text"
5520 # Options are:
5521 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5522 # * left no resizing, just left align. label is used for alt= only
5523 # * right same, but right aligned
5524 # * none same, but not aligned
5525 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5526 # * center center the image
5527 # * frame Keep original image size, no magnify-button.
5528 # * framed Same as "frame"
5529 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5530 # * upright reduce width for upright images, rounded to full __0 px
5531 # * border draw a 1px border around the image
5532 # * alt Text for HTML alt attribute (defaults to empty)
5533 # * class Set a class for img node
5534 # * link Set the target of the image link. Can be external, interwiki, or local
5535 # vertical-align values (no % or length right now):
5536 # * baseline
5537 # * sub
5538 # * super
5539 # * top
5540 # * text-top
5541 # * middle
5542 # * bottom
5543 # * text-bottom
5544
5545 $parts = StringUtils::explode( "|", $options );
5546
5547 # Give extensions a chance to select the file revision for us
5548 $options = [];
5549 $descQuery = false;
5550 Hooks::run( 'BeforeParserFetchFileAndTitle',
5551 [ $this, $title, &$options, &$descQuery ] );
5552 # Fetch and register the file (file title may be different via hooks)
5553 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5554
5555 # Get parameter map
5556 $handler = $file ? $file->getHandler() : false;
5557
5558 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5559
5560 if ( !$file ) {
5561 $this->addTrackingCategory( 'broken-file-category' );
5562 }
5563
5564 # Process the input parameters
5565 $caption = '';
5566 $params = [ 'frame' => [], 'handler' => [],
5567 'horizAlign' => [], 'vertAlign' => [] ];
5568 $seenformat = false;
5569 foreach ( $parts as $part ) {
5570 $part = trim( $part );
5571 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5572 $validated = false;
5573 if ( isset( $paramMap[$magicName] ) ) {
5574 list( $type, $paramName ) = $paramMap[$magicName];
5575
5576 # Special case; width and height come in one variable together
5577 if ( $type === 'handler' && $paramName === 'width' ) {
5578 $parsedWidthParam = $this->parseWidthParam( $value );
5579 if ( isset( $parsedWidthParam['width'] ) ) {
5580 $width = $parsedWidthParam['width'];
5581 if ( $handler->validateParam( 'width', $width ) ) {
5582 $params[$type]['width'] = $width;
5583 $validated = true;
5584 }
5585 }
5586 if ( isset( $parsedWidthParam['height'] ) ) {
5587 $height = $parsedWidthParam['height'];
5588 if ( $handler->validateParam( 'height', $height ) ) {
5589 $params[$type]['height'] = $height;
5590 $validated = true;
5591 }
5592 }
5593 # else no validation -- bug 13436
5594 } else {
5595 if ( $type === 'handler' ) {
5596 # Validate handler parameter
5597 $validated = $handler->validateParam( $paramName, $value );
5598 } else {
5599 # Validate internal parameters
5600 switch ( $paramName ) {
5601 case 'manualthumb':
5602 case 'alt':
5603 case 'class':
5604 # @todo FIXME: Possibly check validity here for
5605 # manualthumb? downstream behavior seems odd with
5606 # missing manual thumbs.
5607 $validated = true;
5608 $value = $this->stripAltText( $value, $holders );
5609 break;
5610 case 'link':
5611 $chars = self::EXT_LINK_URL_CLASS;
5612 $addr = self::EXT_LINK_ADDR;
5613 $prots = $this->mUrlProtocols;
5614 if ( $value === '' ) {
5615 $paramName = 'no-link';
5616 $value = true;
5617 $validated = true;
5618 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5619 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5620 $paramName = 'link-url';
5621 $this->mOutput->addExternalLink( $value );
5622 if ( $this->mOptions->getExternalLinkTarget() ) {
5623 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5624 }
5625 $validated = true;
5626 }
5627 } else {
5628 $linkTitle = Title::newFromText( $value );
5629 if ( $linkTitle ) {
5630 $paramName = 'link-title';
5631 $value = $linkTitle;
5632 $this->mOutput->addLink( $linkTitle );
5633 $validated = true;
5634 }
5635 }
5636 break;
5637 case 'frameless':
5638 case 'framed':
5639 case 'thumbnail':
5640 // use first appearing option, discard others.
5641 $validated = ! $seenformat;
5642 $seenformat = true;
5643 break;
5644 default:
5645 # Most other things appear to be empty or numeric...
5646 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5647 }
5648 }
5649
5650 if ( $validated ) {
5651 $params[$type][$paramName] = $value;
5652 }
5653 }
5654 }
5655 if ( !$validated ) {
5656 $caption = $part;
5657 }
5658 }
5659
5660 # Process alignment parameters
5661 if ( $params['horizAlign'] ) {
5662 $params['frame']['align'] = key( $params['horizAlign'] );
5663 }
5664 if ( $params['vertAlign'] ) {
5665 $params['frame']['valign'] = key( $params['vertAlign'] );
5666 }
5667
5668 $params['frame']['caption'] = $caption;
5669
5670 # Will the image be presented in a frame, with the caption below?
5671 $imageIsFramed = isset( $params['frame']['frame'] )
5672 || isset( $params['frame']['framed'] )
5673 || isset( $params['frame']['thumbnail'] )
5674 || isset( $params['frame']['manualthumb'] );
5675
5676 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5677 # came to also set the caption, ordinary text after the image -- which
5678 # makes no sense, because that just repeats the text multiple times in
5679 # screen readers. It *also* came to set the title attribute.
5680 # Now that we have an alt attribute, we should not set the alt text to
5681 # equal the caption: that's worse than useless, it just repeats the
5682 # text. This is the framed/thumbnail case. If there's no caption, we
5683 # use the unnamed parameter for alt text as well, just for the time be-
5684 # ing, if the unnamed param is set and the alt param is not.
5685 # For the future, we need to figure out if we want to tweak this more,
5686 # e.g., introducing a title= parameter for the title; ignoring the un-
5687 # named parameter entirely for images without a caption; adding an ex-
5688 # plicit caption= parameter and preserving the old magic unnamed para-
5689 # meter for BC; ...
5690 if ( $imageIsFramed ) { # Framed image
5691 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5692 # No caption or alt text, add the filename as the alt text so
5693 # that screen readers at least get some description of the image
5694 $params['frame']['alt'] = $title->getText();
5695 }
5696 # Do not set $params['frame']['title'] because tooltips don't make sense
5697 # for framed images
5698 } else { # Inline image
5699 if ( !isset( $params['frame']['alt'] ) ) {
5700 # No alt text, use the "caption" for the alt text
5701 if ( $caption !== '' ) {
5702 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5703 } else {
5704 # No caption, fall back to using the filename for the
5705 # alt text
5706 $params['frame']['alt'] = $title->getText();
5707 }
5708 }
5709 # Use the "caption" for the tooltip text
5710 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5711 }
5712
5713 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5714
5715 # Linker does the rest
5716 $time = isset( $options['time'] ) ? $options['time'] : false;
5717 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5718 $time, $descQuery, $this->mOptions->getThumbSize() );
5719
5720 # Give the handler a chance to modify the parser object
5721 if ( $handler ) {
5722 $handler->parserTransformHook( $this, $file );
5723 }
5724
5725 return $ret;
5726 }
5727
5728 /**
5729 * @param string $caption
5730 * @param LinkHolderArray|bool $holders
5731 * @return mixed|string
5732 */
5733 protected function stripAltText( $caption, $holders ) {
5734 # Strip bad stuff out of the title (tooltip). We can't just use
5735 # replaceLinkHoldersText() here, because if this function is called
5736 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5737 if ( $holders ) {
5738 $tooltip = $holders->replaceText( $caption );
5739 } else {
5740 $tooltip = $this->replaceLinkHoldersText( $caption );
5741 }
5742
5743 # make sure there are no placeholders in thumbnail attributes
5744 # that are later expanded to html- so expand them now and
5745 # remove the tags
5746 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5747 $tooltip = Sanitizer::stripAllTags( $tooltip );
5748
5749 return $tooltip;
5750 }
5751
5752 /**
5753 * Set a flag in the output object indicating that the content is dynamic and
5754 * shouldn't be cached.
5755 */
5756 public function disableCache() {
5757 wfDebug( "Parser output marked as uncacheable.\n" );
5758 if ( !$this->mOutput ) {
5759 throw new MWException( __METHOD__ .
5760 " can only be called when actually parsing something" );
5761 }
5762 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5763 }
5764
5765 /**
5766 * Callback from the Sanitizer for expanding items found in HTML attribute
5767 * values, so they can be safely tested and escaped.
5768 *
5769 * @param string $text
5770 * @param bool|PPFrame $frame
5771 * @return string
5772 */
5773 public function attributeStripCallback( &$text, $frame = false ) {
5774 $text = $this->replaceVariables( $text, $frame );
5775 $text = $this->mStripState->unstripBoth( $text );
5776 return $text;
5777 }
5778
5779 /**
5780 * Accessor
5781 *
5782 * @return array
5783 */
5784 public function getTags() {
5785 return array_merge(
5786 array_keys( $this->mTransparentTagHooks ),
5787 array_keys( $this->mTagHooks ),
5788 array_keys( $this->mFunctionTagHooks )
5789 );
5790 }
5791
5792 /**
5793 * Replace transparent tags in $text with the values given by the callbacks.
5794 *
5795 * Transparent tag hooks are like regular XML-style tag hooks, except they
5796 * operate late in the transformation sequence, on HTML instead of wikitext.
5797 *
5798 * @param string $text
5799 *
5800 * @return string
5801 */
5802 public function replaceTransparentTags( $text ) {
5803 $matches = [];
5804 $elements = array_keys( $this->mTransparentTagHooks );
5805 $text = self::extractTagsAndParams( $elements, $text, $matches );
5806 $replacements = [];
5807
5808 foreach ( $matches as $marker => $data ) {
5809 list( $element, $content, $params, $tag ) = $data;
5810 $tagName = strtolower( $element );
5811 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5812 $output = call_user_func_array(
5813 $this->mTransparentTagHooks[$tagName],
5814 [ $content, $params, $this ]
5815 );
5816 } else {
5817 $output = $tag;
5818 }
5819 $replacements[$marker] = $output;
5820 }
5821 return strtr( $text, $replacements );
5822 }
5823
5824 /**
5825 * Break wikitext input into sections, and either pull or replace
5826 * some particular section's text.
5827 *
5828 * External callers should use the getSection and replaceSection methods.
5829 *
5830 * @param string $text Page wikitext
5831 * @param string|number $sectionId A section identifier string of the form:
5832 * "<flag1> - <flag2> - ... - <section number>"
5833 *
5834 * Currently the only recognised flag is "T", which means the target section number
5835 * was derived during a template inclusion parse, in other words this is a template
5836 * section edit link. If no flags are given, it was an ordinary section edit link.
5837 * This flag is required to avoid a section numbering mismatch when a section is
5838 * enclosed by "<includeonly>" (bug 6563).
5839 *
5840 * The section number 0 pulls the text before the first heading; other numbers will
5841 * pull the given section along with its lower-level subsections. If the section is
5842 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5843 *
5844 * Section 0 is always considered to exist, even if it only contains the empty
5845 * string. If $text is the empty string and section 0 is replaced, $newText is
5846 * returned.
5847 *
5848 * @param string $mode One of "get" or "replace"
5849 * @param string $newText Replacement text for section data.
5850 * @return string For "get", the extracted section text.
5851 * for "replace", the whole page with the section replaced.
5852 */
5853 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5854 global $wgTitle; # not generally used but removes an ugly failure mode
5855
5856 $magicScopeVariable = $this->lock();
5857 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5858 $outText = '';
5859 $frame = $this->getPreprocessor()->newFrame();
5860
5861 # Process section extraction flags
5862 $flags = 0;
5863 $sectionParts = explode( '-', $sectionId );
5864 $sectionIndex = array_pop( $sectionParts );
5865 foreach ( $sectionParts as $part ) {
5866 if ( $part === 'T' ) {
5867 $flags |= self::PTD_FOR_INCLUSION;
5868 }
5869 }
5870
5871 # Check for empty input
5872 if ( strval( $text ) === '' ) {
5873 # Only sections 0 and T-0 exist in an empty document
5874 if ( $sectionIndex == 0 ) {
5875 if ( $mode === 'get' ) {
5876 return '';
5877 } else {
5878 return $newText;
5879 }
5880 } else {
5881 if ( $mode === 'get' ) {
5882 return $newText;
5883 } else {
5884 return $text;
5885 }
5886 }
5887 }
5888
5889 # Preprocess the text
5890 $root = $this->preprocessToDom( $text, $flags );
5891
5892 # <h> nodes indicate section breaks
5893 # They can only occur at the top level, so we can find them by iterating the root's children
5894 $node = $root->getFirstChild();
5895
5896 # Find the target section
5897 if ( $sectionIndex == 0 ) {
5898 # Section zero doesn't nest, level=big
5899 $targetLevel = 1000;
5900 } else {
5901 while ( $node ) {
5902 if ( $node->getName() === 'h' ) {
5903 $bits = $node->splitHeading();
5904 if ( $bits['i'] == $sectionIndex ) {
5905 $targetLevel = $bits['level'];
5906 break;
5907 }
5908 }
5909 if ( $mode === 'replace' ) {
5910 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5911 }
5912 $node = $node->getNextSibling();
5913 }
5914 }
5915
5916 if ( !$node ) {
5917 # Not found
5918 if ( $mode === 'get' ) {
5919 return $newText;
5920 } else {
5921 return $text;
5922 }
5923 }
5924
5925 # Find the end of the section, including nested sections
5926 do {
5927 if ( $node->getName() === 'h' ) {
5928 $bits = $node->splitHeading();
5929 $curLevel = $bits['level'];
5930 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5931 break;
5932 }
5933 }
5934 if ( $mode === 'get' ) {
5935 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5936 }
5937 $node = $node->getNextSibling();
5938 } while ( $node );
5939
5940 # Write out the remainder (in replace mode only)
5941 if ( $mode === 'replace' ) {
5942 # Output the replacement text
5943 # Add two newlines on -- trailing whitespace in $newText is conventionally
5944 # stripped by the editor, so we need both newlines to restore the paragraph gap
5945 # Only add trailing whitespace if there is newText
5946 if ( $newText != "" ) {
5947 $outText .= $newText . "\n\n";
5948 }
5949
5950 while ( $node ) {
5951 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5952 $node = $node->getNextSibling();
5953 }
5954 }
5955
5956 if ( is_string( $outText ) ) {
5957 # Re-insert stripped tags
5958 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5959 }
5960
5961 return $outText;
5962 }
5963
5964 /**
5965 * This function returns the text of a section, specified by a number ($section).
5966 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5967 * the first section before any such heading (section 0).
5968 *
5969 * If a section contains subsections, these are also returned.
5970 *
5971 * @param string $text Text to look in
5972 * @param string|number $sectionId Section identifier as a number or string
5973 * (e.g. 0, 1 or 'T-1').
5974 * @param string $defaultText Default to return if section is not found
5975 *
5976 * @return string Text of the requested section
5977 */
5978 public function getSection( $text, $sectionId, $defaultText = '' ) {
5979 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5980 }
5981
5982 /**
5983 * This function returns $oldtext after the content of the section
5984 * specified by $section has been replaced with $text. If the target
5985 * section does not exist, $oldtext is returned unchanged.
5986 *
5987 * @param string $oldText Former text of the article
5988 * @param string|number $sectionId Section identifier as a number or string
5989 * (e.g. 0, 1 or 'T-1').
5990 * @param string $newText Replacing text
5991 *
5992 * @return string Modified text
5993 */
5994 public function replaceSection( $oldText, $sectionId, $newText ) {
5995 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5996 }
5997
5998 /**
5999 * Get the ID of the revision we are parsing
6000 *
6001 * @return int|null
6002 */
6003 public function getRevisionId() {
6004 return $this->mRevisionId;
6005 }
6006
6007 /**
6008 * Get the revision object for $this->mRevisionId
6009 *
6010 * @return Revision|null Either a Revision object or null
6011 * @since 1.23 (public since 1.23)
6012 */
6013 public function getRevisionObject() {
6014 if ( !is_null( $this->mRevisionObject ) ) {
6015 return $this->mRevisionObject;
6016 }
6017 if ( is_null( $this->mRevisionId ) ) {
6018 return null;
6019 }
6020
6021 $rev = call_user_func(
6022 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
6023 );
6024
6025 # If the parse is for a new revision, then the callback should have
6026 # already been set to force the object and should match mRevisionId.
6027 # If not, try to fetch by mRevisionId for sanity.
6028 if ( $rev && $rev->getId() != $this->mRevisionId ) {
6029 $rev = Revision::newFromId( $this->mRevisionId );
6030 }
6031
6032 $this->mRevisionObject = $rev;
6033
6034 return $this->mRevisionObject;
6035 }
6036
6037 /**
6038 * Get the timestamp associated with the current revision, adjusted for
6039 * the default server-local timestamp
6040 * @return string
6041 */
6042 public function getRevisionTimestamp() {
6043 if ( is_null( $this->mRevisionTimestamp ) ) {
6044 global $wgContLang;
6045
6046 $revObject = $this->getRevisionObject();
6047 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
6048
6049 # The cryptic '' timezone parameter tells to use the site-default
6050 # timezone offset instead of the user settings.
6051 # Since this value will be saved into the parser cache, served
6052 # to other users, and potentially even used inside links and such,
6053 # it needs to be consistent for all visitors.
6054 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
6055
6056 }
6057 return $this->mRevisionTimestamp;
6058 }
6059
6060 /**
6061 * Get the name of the user that edited the last revision
6062 *
6063 * @return string User name
6064 */
6065 public function getRevisionUser() {
6066 if ( is_null( $this->mRevisionUser ) ) {
6067 $revObject = $this->getRevisionObject();
6068
6069 # if this template is subst: the revision id will be blank,
6070 # so just use the current user's name
6071 if ( $revObject ) {
6072 $this->mRevisionUser = $revObject->getUserText();
6073 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6074 $this->mRevisionUser = $this->getUser()->getName();
6075 }
6076 }
6077 return $this->mRevisionUser;
6078 }
6079
6080 /**
6081 * Get the size of the revision
6082 *
6083 * @return int|null Revision size
6084 */
6085 public function getRevisionSize() {
6086 if ( is_null( $this->mRevisionSize ) ) {
6087 $revObject = $this->getRevisionObject();
6088
6089 # if this variable is subst: the revision id will be blank,
6090 # so just use the parser input size, because the own substituation
6091 # will change the size.
6092 if ( $revObject ) {
6093 $this->mRevisionSize = $revObject->getSize();
6094 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6095 $this->mRevisionSize = $this->mInputSize;
6096 }
6097 }
6098 return $this->mRevisionSize;
6099 }
6100
6101 /**
6102 * Mutator for $mDefaultSort
6103 *
6104 * @param string $sort New value
6105 */
6106 public function setDefaultSort( $sort ) {
6107 $this->mDefaultSort = $sort;
6108 $this->mOutput->setProperty( 'defaultsort', $sort );
6109 }
6110
6111 /**
6112 * Accessor for $mDefaultSort
6113 * Will use the empty string if none is set.
6114 *
6115 * This value is treated as a prefix, so the
6116 * empty string is equivalent to sorting by
6117 * page name.
6118 *
6119 * @return string
6120 */
6121 public function getDefaultSort() {
6122 if ( $this->mDefaultSort !== false ) {
6123 return $this->mDefaultSort;
6124 } else {
6125 return '';
6126 }
6127 }
6128
6129 /**
6130 * Accessor for $mDefaultSort
6131 * Unlike getDefaultSort(), will return false if none is set
6132 *
6133 * @return string|bool
6134 */
6135 public function getCustomDefaultSort() {
6136 return $this->mDefaultSort;
6137 }
6138
6139 /**
6140 * Try to guess the section anchor name based on a wikitext fragment
6141 * presumably extracted from a heading, for example "Header" from
6142 * "== Header ==".
6143 *
6144 * @param string $text
6145 *
6146 * @return string
6147 */
6148 public function guessSectionNameFromWikiText( $text ) {
6149 # Strip out wikitext links(they break the anchor)
6150 $text = $this->stripSectionName( $text );
6151 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6152 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
6153 }
6154
6155 /**
6156 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6157 * instead. For use in redirects, since IE6 interprets Redirect: headers
6158 * as something other than UTF-8 (apparently?), resulting in breakage.
6159 *
6160 * @param string $text The section name
6161 * @return string An anchor
6162 */
6163 public function guessLegacySectionNameFromWikiText( $text ) {
6164 # Strip out wikitext links(they break the anchor)
6165 $text = $this->stripSectionName( $text );
6166 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6167 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
6168 }
6169
6170 /**
6171 * Strips a text string of wikitext for use in a section anchor
6172 *
6173 * Accepts a text string and then removes all wikitext from the
6174 * string and leaves only the resultant text (i.e. the result of
6175 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6176 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6177 * to create valid section anchors by mimicing the output of the
6178 * parser when headings are parsed.
6179 *
6180 * @param string $text Text string to be stripped of wikitext
6181 * for use in a Section anchor
6182 * @return string Filtered text string
6183 */
6184 public function stripSectionName( $text ) {
6185 # Strip internal link markup
6186 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6187 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6188
6189 # Strip external link markup
6190 # @todo FIXME: Not tolerant to blank link text
6191 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6192 # on how many empty links there are on the page - need to figure that out.
6193 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6194
6195 # Parse wikitext quotes (italics & bold)
6196 $text = $this->doQuotes( $text );
6197
6198 # Strip HTML tags
6199 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6200 return $text;
6201 }
6202
6203 /**
6204 * strip/replaceVariables/unstrip for preprocessor regression testing
6205 *
6206 * @param string $text
6207 * @param Title $title
6208 * @param ParserOptions $options
6209 * @param int $outputType
6210 *
6211 * @return string
6212 */
6213 public function testSrvus( $text, Title $title, ParserOptions $options,
6214 $outputType = self::OT_HTML
6215 ) {
6216 $magicScopeVariable = $this->lock();
6217 $this->startParse( $title, $options, $outputType, true );
6218
6219 $text = $this->replaceVariables( $text );
6220 $text = $this->mStripState->unstripBoth( $text );
6221 $text = Sanitizer::removeHTMLtags( $text );
6222 return $text;
6223 }
6224
6225 /**
6226 * @param string $text
6227 * @param Title $title
6228 * @param ParserOptions $options
6229 * @return string
6230 */
6231 public function testPst( $text, Title $title, ParserOptions $options ) {
6232 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6233 }
6234
6235 /**
6236 * @param string $text
6237 * @param Title $title
6238 * @param ParserOptions $options
6239 * @return string
6240 */
6241 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6242 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6243 }
6244
6245 /**
6246 * Call a callback function on all regions of the given text that are not
6247 * inside strip markers, and replace those regions with the return value
6248 * of the callback. For example, with input:
6249 *
6250 * aaa<MARKER>bbb
6251 *
6252 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6253 * two strings will be replaced with the value returned by the callback in
6254 * each case.
6255 *
6256 * @param string $s
6257 * @param callable $callback
6258 *
6259 * @return string
6260 */
6261 public function markerSkipCallback( $s, $callback ) {
6262 $i = 0;
6263 $out = '';
6264 while ( $i < strlen( $s ) ) {
6265 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6266 if ( $markerStart === false ) {
6267 $out .= call_user_func( $callback, substr( $s, $i ) );
6268 break;
6269 } else {
6270 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6271 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6272 if ( $markerEnd === false ) {
6273 $out .= substr( $s, $markerStart );
6274 break;
6275 } else {
6276 $markerEnd += strlen( self::MARKER_SUFFIX );
6277 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6278 $i = $markerEnd;
6279 }
6280 }
6281 }
6282 return $out;
6283 }
6284
6285 /**
6286 * Remove any strip markers found in the given text.
6287 *
6288 * @param string $text Input string
6289 * @return string
6290 */
6291 public function killMarkers( $text ) {
6292 return $this->mStripState->killMarkers( $text );
6293 }
6294
6295 /**
6296 * Save the parser state required to convert the given half-parsed text to
6297 * HTML. "Half-parsed" in this context means the output of
6298 * recursiveTagParse() or internalParse(). This output has strip markers
6299 * from replaceVariables (extensionSubstitution() etc.), and link
6300 * placeholders from replaceLinkHolders().
6301 *
6302 * Returns an array which can be serialized and stored persistently. This
6303 * array can later be loaded into another parser instance with
6304 * unserializeHalfParsedText(). The text can then be safely incorporated into
6305 * the return value of a parser hook.
6306 *
6307 * @param string $text
6308 *
6309 * @return array
6310 */
6311 public function serializeHalfParsedText( $text ) {
6312 $data = [
6313 'text' => $text,
6314 'version' => self::HALF_PARSED_VERSION,
6315 'stripState' => $this->mStripState->getSubState( $text ),
6316 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6317 ];
6318 return $data;
6319 }
6320
6321 /**
6322 * Load the parser state given in the $data array, which is assumed to
6323 * have been generated by serializeHalfParsedText(). The text contents is
6324 * extracted from the array, and its markers are transformed into markers
6325 * appropriate for the current Parser instance. This transformed text is
6326 * returned, and can be safely included in the return value of a parser
6327 * hook.
6328 *
6329 * If the $data array has been stored persistently, the caller should first
6330 * check whether it is still valid, by calling isValidHalfParsedText().
6331 *
6332 * @param array $data Serialized data
6333 * @throws MWException
6334 * @return string
6335 */
6336 public function unserializeHalfParsedText( $data ) {
6337 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6338 throw new MWException( __METHOD__ . ': invalid version' );
6339 }
6340
6341 # First, extract the strip state.
6342 $texts = [ $data['text'] ];
6343 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6344
6345 # Now renumber links
6346 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6347
6348 # Should be good to go.
6349 return $texts[0];
6350 }
6351
6352 /**
6353 * Returns true if the given array, presumed to be generated by
6354 * serializeHalfParsedText(), is compatible with the current version of the
6355 * parser.
6356 *
6357 * @param array $data
6358 *
6359 * @return bool
6360 */
6361 public function isValidHalfParsedText( $data ) {
6362 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6363 }
6364
6365 /**
6366 * Parsed a width param of imagelink like 300px or 200x300px
6367 *
6368 * @param string $value
6369 *
6370 * @return array
6371 * @since 1.20
6372 */
6373 public function parseWidthParam( $value ) {
6374 $parsedWidthParam = [];
6375 if ( $value === '' ) {
6376 return $parsedWidthParam;
6377 }
6378 $m = [];
6379 # (bug 13500) In both cases (width/height and width only),
6380 # permit trailing "px" for backward compatibility.
6381 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6382 $width = intval( $m[1] );
6383 $height = intval( $m[2] );
6384 $parsedWidthParam['width'] = $width;
6385 $parsedWidthParam['height'] = $height;
6386 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6387 $width = intval( $value );
6388 $parsedWidthParam['width'] = $width;
6389 }
6390 return $parsedWidthParam;
6391 }
6392
6393 /**
6394 * Lock the current instance of the parser.
6395 *
6396 * This is meant to stop someone from calling the parser
6397 * recursively and messing up all the strip state.
6398 *
6399 * @throws MWException If parser is in a parse
6400 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6401 */
6402 protected function lock() {
6403 if ( $this->mInParse ) {
6404 throw new MWException( "Parser state cleared while parsing. "
6405 . "Did you call Parser::parse recursively?" );
6406 }
6407 $this->mInParse = true;
6408
6409 $recursiveCheck = new ScopedCallback( function() {
6410 $this->mInParse = false;
6411 } );
6412
6413 return $recursiveCheck;
6414 }
6415
6416 /**
6417 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6418 *
6419 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6420 * or if there is more than one <p/> tag in the input HTML.
6421 *
6422 * @param string $html
6423 * @return string
6424 * @since 1.24
6425 */
6426 public static function stripOuterParagraph( $html ) {
6427 $m = [];
6428 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6429 if ( strpos( $m[1], '</p>' ) === false ) {
6430 $html = $m[1];
6431 }
6432 }
6433
6434 return $html;
6435 }
6436
6437 /**
6438 * Return this parser if it is not doing anything, otherwise
6439 * get a fresh parser. You can use this method by doing
6440 * $myParser = $wgParser->getFreshParser(), or more simply
6441 * $wgParser->getFreshParser()->parse( ... );
6442 * if you're unsure if $wgParser is safe to use.
6443 *
6444 * @since 1.24
6445 * @return Parser A parser object that is not parsing anything
6446 */
6447 public function getFreshParser() {
6448 global $wgParserConf;
6449 if ( $this->mInParse ) {
6450 return new $wgParserConf['class']( $wgParserConf );
6451 } else {
6452 return $this;
6453 }
6454 }
6455
6456 /**
6457 * Set's up the PHP implementation of OOUI for use in this request
6458 * and instructs OutputPage to enable OOUI for itself.
6459 *
6460 * @since 1.26
6461 */
6462 public function enableOOUI() {
6463 OutputPage::setupOOUI();
6464 $this->mOutput->setEnableOOUI( true );
6465 }
6466 }
Wiklou, le Wiki du Wiklou