dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Move NewPP limit report HTML comments to JS variables
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25
26 /**
27 * @defgroup Parser Parser
28 */
29
30 /**
31 * PHP Parser - Processes wiki markup (which uses a more user-friendly
32 * syntax, such as "[[link]]" for making links), and provides a one-way
33 * transformation of that wiki markup it into (X)HTML output / markup
34 * (which in turn the browser understands, and can display).
35 *
36 * There are seven main entry points into the Parser class:
37 *
38 * - Parser::parse()
39 * produces HTML output
40 * - Parser::preSaveTransform()
41 * produces altered wiki markup
42 * - Parser::preprocess()
43 * removes HTML comments and expands templates
44 * - Parser::cleanSig() and Parser::cleanSigInSig()
45 * cleans a signature before saving it to preferences
46 * - Parser::getSection()
47 * return the content of a section from an article for section editing
48 * - Parser::replaceSection()
49 * replaces a section by number inside an article
50 * - Parser::getPreloadText()
51 * removes <noinclude> sections and <includeonly> tags
52 *
53 * Globals used:
54 * object: $wgContLang
55 *
56 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
57 *
58 * @par Settings:
59 * $wgNamespacesWithSubpages
60 *
61 * @par Settings only within ParserOptions:
62 * $wgAllowExternalImages
63 * $wgAllowSpecialInclusion
64 * $wgInterwikiMagic
65 * $wgMaxArticleSize
66 *
67 * @ingroup Parser
68 */
69 class Parser {
70 /**
71 * Update this version number when the ParserOutput format
72 * changes in an incompatible way, so the parser cache
73 * can automatically discard old data.
74 */
75 const VERSION = '1.6.4';
76
77 /**
78 * Update this version number when the output of serialiseHalfParsedText()
79 * changes in an incompatible way
80 */
81 const HALF_PARSED_VERSION = 2;
82
83 # Flags for Parser::setFunctionHook
84 const SFH_NO_HASH = 1;
85 const SFH_OBJECT_ARGS = 2;
86
87 # Constants needed for external link processing
88 # Everything except bracket, space, or control characters
89 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
90 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
91 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
92 # Simplified expression to match an IPv4 or IPv6 address, or
93 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
94 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
95 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
96 // @codingStandardsIgnoreStart Generic.Files.LineLength
97 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
98 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
99 // @codingStandardsIgnoreEnd
100
101 # Regular expression for a non-newline space
102 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
103
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION = 1;
106
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML = 1; # like parse()
110 const OT_WIKI = 2; # like preSaveTransform()
111 const OT_PREPROCESS = 3; # like preprocess()
112 const OT_MSG = 3;
113 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
114
115 /**
116 * @var string Prefix and suffix for temporary replacement strings
117 * for the multipass parser.
118 *
119 * \x7f should never appear in input as it's disallowed in XML.
120 * Using it at the front also gives us a little extra robustness
121 * since it shouldn't match when butted up against identifier-like
122 * string constructs.
123 *
124 * Must not consist of all title characters, or else it will change
125 * the behavior of <nowiki> in a link.
126 *
127 * Must have a character that needs escaping in attributes, otherwise
128 * someone could put a strip marker in an attribute, to get around
129 * escaping quote marks, and break out of the attribute. Thus we add
130 * `'".
131 */
132 const MARKER_SUFFIX = "-QINU`\"'\x7f";
133 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
134
135 # Markers used for wrapping the table of contents
136 const TOC_START = '<mw:toc>';
137 const TOC_END = '</mw:toc>';
138
139 # Persistent:
140 public $mTagHooks = [];
141 public $mTransparentTagHooks = [];
142 public $mFunctionHooks = [];
143 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
144 public $mFunctionTagHooks = [];
145 public $mStripList = [];
146 public $mDefaultStripList = [];
147 public $mVarCache = [];
148 public $mImageParams = [];
149 public $mImageParamsMagicArray = [];
150 public $mMarkerIndex = 0;
151 public $mFirstCall = true;
152
153 # Initialised by initialiseVariables()
154
155 /**
156 * @var MagicWordArray
157 */
158 public $mVariables;
159
160 /**
161 * @var MagicWordArray
162 */
163 public $mSubstWords;
164 # Initialised in constructor
165 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
166
167 # Initialized in getPreprocessor()
168 /** @var Preprocessor */
169 public $mPreprocessor;
170
171 # Cleared with clearState():
172 /**
173 * @var ParserOutput
174 */
175 public $mOutput;
176 public $mAutonumber;
177
178 /**
179 * @var StripState
180 */
181 public $mStripState;
182
183 public $mIncludeCount;
184 /**
185 * @var LinkHolderArray
186 */
187 public $mLinkHolders;
188
189 public $mLinkID;
190 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
191 public $mDefaultSort;
192 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
193 public $mExpensiveFunctionCount; # number of expensive parser function calls
194 public $mShowToc, $mForceTocPosition;
195
196 /**
197 * @var User
198 */
199 public $mUser; # User object; only used when doing pre-save transform
200
201 # Temporary
202 # These are variables reset at least once per parse regardless of $clearState
203
204 /**
205 * @var ParserOptions
206 */
207 public $mOptions;
208
209 /**
210 * @var Title
211 */
212 public $mTitle; # Title context, used for self-link rendering and similar things
213 public $mOutputType; # Output type, one of the OT_xxx constants
214 public $ot; # Shortcut alias, see setOutputType()
215 public $mRevisionObject; # The revision object of the specified revision ID
216 public $mRevisionId; # ID to display in {{REVISIONID}} tags
217 public $mRevisionTimestamp; # The timestamp of the specified revision ID
218 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
219 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
220 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
221 public $mInputSize = false; # For {{PAGESIZE}} on current page.
222
223 /**
224 * @var string Deprecated accessor for the strip marker prefix.
225 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
226 **/
227 public $mUniqPrefix = Parser::MARKER_PREFIX;
228
229 /**
230 * @var array Array with the language name of each language link (i.e. the
231 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
232 * duplicate language links to the ParserOutput.
233 */
234 public $mLangLinkLanguages;
235
236 /**
237 * @var MapCacheLRU|null
238 * @since 1.24
239 *
240 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
241 */
242 public $currentRevisionCache;
243
244 /**
245 * @var bool Recursive call protection.
246 * This variable should be treated as if it were private.
247 */
248 public $mInParse = false;
249
250 /** @var SectionProfiler */
251 protected $mProfiler;
252
253 /**
254 * @var LinkRenderer
255 */
256 protected $mLinkRenderer;
257
258 /**
259 * @param array $conf
260 */
261 public function __construct( $conf = [] ) {
262 $this->mConf = $conf;
263 $this->mUrlProtocols = wfUrlProtocols();
264 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
265 self::EXT_LINK_ADDR .
266 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
267 if ( isset( $conf['preprocessorClass'] ) ) {
268 $this->mPreprocessorClass = $conf['preprocessorClass'];
269 } elseif ( defined( 'HPHP_VERSION' ) ) {
270 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
271 $this->mPreprocessorClass = 'Preprocessor_Hash';
272 } elseif ( extension_loaded( 'domxml' ) ) {
273 # PECL extension that conflicts with the core DOM extension (bug 13770)
274 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
275 $this->mPreprocessorClass = 'Preprocessor_Hash';
276 } elseif ( extension_loaded( 'dom' ) ) {
277 $this->mPreprocessorClass = 'Preprocessor_DOM';
278 } else {
279 $this->mPreprocessorClass = 'Preprocessor_Hash';
280 }
281 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
282 }
283
284 /**
285 * Reduce memory usage to reduce the impact of circular references
286 */
287 public function __destruct() {
288 if ( isset( $this->mLinkHolders ) ) {
289 unset( $this->mLinkHolders );
290 }
291 foreach ( $this as $name => $value ) {
292 unset( $this->$name );
293 }
294 }
295
296 /**
297 * Allow extensions to clean up when the parser is cloned
298 */
299 public function __clone() {
300 $this->mInParse = false;
301
302 // Bug 56226: When you create a reference "to" an object field, that
303 // makes the object field itself be a reference too (until the other
304 // reference goes out of scope). When cloning, any field that's a
305 // reference is copied as a reference in the new object. Both of these
306 // are defined PHP5 behaviors, as inconvenient as it is for us when old
307 // hooks from PHP4 days are passing fields by reference.
308 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
309 // Make a non-reference copy of the field, then rebind the field to
310 // reference the new copy.
311 $tmp = $this->$k;
312 $this->$k =& $tmp;
313 unset( $tmp );
314 }
315
316 Hooks::run( 'ParserCloned', [ $this ] );
317 }
318
319 /**
320 * Do various kinds of initialisation on the first call of the parser
321 */
322 public function firstCallInit() {
323 if ( !$this->mFirstCall ) {
324 return;
325 }
326 $this->mFirstCall = false;
327
328 CoreParserFunctions::register( $this );
329 CoreTagHooks::register( $this );
330 $this->initialiseVariables();
331
332 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
333 }
334
335 /**
336 * Clear Parser state
337 *
338 * @private
339 */
340 public function clearState() {
341 if ( $this->mFirstCall ) {
342 $this->firstCallInit();
343 }
344 $this->mOutput = new ParserOutput;
345 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
346 $this->mAutonumber = 0;
347 $this->mIncludeCount = [];
348 $this->mLinkHolders = new LinkHolderArray( $this );
349 $this->mLinkID = 0;
350 $this->mRevisionObject = $this->mRevisionTimestamp =
351 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
352 $this->mVarCache = [];
353 $this->mUser = null;
354 $this->mLangLinkLanguages = [];
355 $this->currentRevisionCache = null;
356
357 $this->mStripState = new StripState;
358
359 # Clear these on every parse, bug 4549
360 $this->mTplRedirCache = $this->mTplDomCache = [];
361
362 $this->mShowToc = true;
363 $this->mForceTocPosition = false;
364 $this->mIncludeSizes = [
365 'post-expand' => 0,
366 'arg' => 0,
367 ];
368 $this->mPPNodeCount = 0;
369 $this->mGeneratedPPNodeCount = 0;
370 $this->mHighestExpansionDepth = 0;
371 $this->mDefaultSort = false;
372 $this->mHeadings = [];
373 $this->mDoubleUnderscores = [];
374 $this->mExpensiveFunctionCount = 0;
375
376 # Fix cloning
377 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
378 $this->mPreprocessor = null;
379 }
380
381 $this->mProfiler = new SectionProfiler();
382
383 Hooks::run( 'ParserClearState', [ &$this ] );
384 }
385
386 /**
387 * Convert wikitext to HTML
388 * Do not call this function recursively.
389 *
390 * @param string $text Text we want to parse
391 * @param Title $title
392 * @param ParserOptions $options
393 * @param bool $linestart
394 * @param bool $clearState
395 * @param int $revid Number to pass in {{REVISIONID}}
396 * @return ParserOutput A ParserOutput
397 */
398 public function parse( $text, Title $title, ParserOptions $options,
399 $linestart = true, $clearState = true, $revid = null
400 ) {
401 /**
402 * First pass--just handle <nowiki> sections, pass the rest off
403 * to internalParse() which does all the real work.
404 */
405
406 global $wgShowHostnames;
407
408 if ( $clearState ) {
409 // We use U+007F DELETE to construct strip markers, so we have to make
410 // sure that this character does not occur in the input text.
411 $text = strtr( $text, "\x7f", "?" );
412 $magicScopeVariable = $this->lock();
413 }
414
415 $this->startParse( $title, $options, self::OT_HTML, $clearState );
416
417 $this->currentRevisionCache = null;
418 $this->mInputSize = strlen( $text );
419 if ( $this->mOptions->getEnableLimitReport() ) {
420 $this->mOutput->resetParseStartTime();
421 }
422
423 $oldRevisionId = $this->mRevisionId;
424 $oldRevisionObject = $this->mRevisionObject;
425 $oldRevisionTimestamp = $this->mRevisionTimestamp;
426 $oldRevisionUser = $this->mRevisionUser;
427 $oldRevisionSize = $this->mRevisionSize;
428 if ( $revid !== null ) {
429 $this->mRevisionId = $revid;
430 $this->mRevisionObject = null;
431 $this->mRevisionTimestamp = null;
432 $this->mRevisionUser = null;
433 $this->mRevisionSize = null;
434 }
435
436 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
437 # No more strip!
438 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
439 $text = $this->internalParse( $text );
440 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
441
442 $text = $this->internalParseHalfParsed( $text, true, $linestart );
443
444 /**
445 * A converted title will be provided in the output object if title and
446 * content conversion are enabled, the article text does not contain
447 * a conversion-suppressing double-underscore tag, and no
448 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
449 * automatic link conversion.
450 */
451 if ( !( $options->getDisableTitleConversion()
452 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
453 || isset( $this->mDoubleUnderscores['notitleconvert'] )
454 || $this->mOutput->getDisplayTitle() !== false )
455 ) {
456 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
457 if ( $convruletitle ) {
458 $this->mOutput->setTitleText( $convruletitle );
459 } else {
460 $titleText = $this->getConverterLanguage()->convertTitle( $title );
461 $this->mOutput->setTitleText( $titleText );
462 }
463 }
464
465 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
466 $this->limitationWarn( 'expensive-parserfunction',
467 $this->mExpensiveFunctionCount,
468 $this->mOptions->getExpensiveParserFunctionLimit()
469 );
470 }
471
472 # Information on include size limits, for the benefit of users who try to skirt them
473 if ( $this->mOptions->getEnableLimitReport() ) {
474 $max = $this->mOptions->getMaxIncludeSize();
475
476 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
477 if ( $cpuTime !== null ) {
478 $this->mOutput->setLimitReportData( 'limitreport-cputime',
479 sprintf( "%.3f", $cpuTime )
480 );
481 }
482
483 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
484 $this->mOutput->setLimitReportData( 'limitreport-walltime',
485 sprintf( "%.3f", $wallTime )
486 );
487
488 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
489 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
490 );
491 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
492 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
493 );
494 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
495 [ $this->mIncludeSizes['post-expand'], $max ]
496 );
497 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
498 [ $this->mIncludeSizes['arg'], $max ]
499 );
500 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
501 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
502 );
503 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
504 [ $this->mExpensiveFunctionCount,
505 $this->mOptions->getExpensiveParserFunctionLimit() ]
506 );
507 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
508
509 $limitReport = '';
510 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
511 if ( $limitReport != '' ) {
512 // Sanitize for comment. Note '‐' in the replacement is U+2010,
513 // which looks much like the problematic '-'.
514 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
515 $text .= "\n<!-- \nNewPP limit report\n$limitReport-->\n";
516 }
517
518 // Add on template profiling data in human/machine readable way
519 $dataByFunc = $this->mProfiler->getFunctionStats();
520 uasort( $dataByFunc, function ( $a, $b ) {
521 return $a['real'] < $b['real']; // descending order
522 } );
523 $profileReport = [];
524 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
525 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
526 $item['%real'], $item['real'], $item['calls'], $item['name'] );
527 }
528 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
529
530 // Add other cache related metadata
531 if ( $wgShowHostnames ) {
532 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
533 }
534 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
535 $this->mOutput->getCacheTime() );
536 $this->mOutput->setLimitReportData( 'cachereport-ttl',
537 $this->mOutput->getCacheExpiry() );
538 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
539 $this->mOutput->hasDynamicContent() );
540
541 if ( $this->mGeneratedPPNodeCount
542 > $this->mOptions->getMaxGeneratedPPNodeCount() / 10
543 ) {
544 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
545 $this->mTitle->getPrefixedDBkey() );
546 }
547 }
548 $this->mOutput->setText( $text );
549
550 $this->mRevisionId = $oldRevisionId;
551 $this->mRevisionObject = $oldRevisionObject;
552 $this->mRevisionTimestamp = $oldRevisionTimestamp;
553 $this->mRevisionUser = $oldRevisionUser;
554 $this->mRevisionSize = $oldRevisionSize;
555 $this->mInputSize = false;
556 $this->currentRevisionCache = null;
557
558 return $this->mOutput;
559 }
560
561 /**
562 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
563 * can be called from an extension tag hook.
564 *
565 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
566 * instead, which means that lists and links have not been fully parsed yet,
567 * and strip markers are still present.
568 *
569 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
570 *
571 * Use this function if you're a parser tag hook and you want to parse
572 * wikitext before or after applying additional transformations, and you
573 * intend to *return the result as hook output*, which will cause it to go
574 * through the rest of parsing process automatically.
575 *
576 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
577 * $text are not expanded
578 *
579 * @param string $text Text extension wants to have parsed
580 * @param bool|PPFrame $frame The frame to use for expanding any template variables
581 * @return string UNSAFE half-parsed HTML
582 */
583 public function recursiveTagParse( $text, $frame = false ) {
584 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
585 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
586 $text = $this->internalParse( $text, false, $frame );
587 return $text;
588 }
589
590 /**
591 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
592 * point can be called from an extension tag hook.
593 *
594 * The output of this function is fully-parsed HTML that is safe for output.
595 * If you're a parser tag hook, you might want to use recursiveTagParse()
596 * instead.
597 *
598 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
599 * $text are not expanded
600 *
601 * @since 1.25
602 *
603 * @param string $text Text extension wants to have parsed
604 * @param bool|PPFrame $frame The frame to use for expanding any template variables
605 * @return string Fully parsed HTML
606 */
607 public function recursiveTagParseFully( $text, $frame = false ) {
608 $text = $this->recursiveTagParse( $text, $frame );
609 $text = $this->internalParseHalfParsed( $text, false );
610 return $text;
611 }
612
613 /**
614 * Expand templates and variables in the text, producing valid, static wikitext.
615 * Also removes comments.
616 * Do not call this function recursively.
617 * @param string $text
618 * @param Title $title
619 * @param ParserOptions $options
620 * @param int|null $revid
621 * @param bool|PPFrame $frame
622 * @return mixed|string
623 */
624 public function preprocess( $text, Title $title = null,
625 ParserOptions $options, $revid = null, $frame = false
626 ) {
627 $magicScopeVariable = $this->lock();
628 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
629 if ( $revid !== null ) {
630 $this->mRevisionId = $revid;
631 }
632 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
633 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
634 $text = $this->replaceVariables( $text, $frame );
635 $text = $this->mStripState->unstripBoth( $text );
636 return $text;
637 }
638
639 /**
640 * Recursive parser entry point that can be called from an extension tag
641 * hook.
642 *
643 * @param string $text Text to be expanded
644 * @param bool|PPFrame $frame The frame to use for expanding any template variables
645 * @return string
646 * @since 1.19
647 */
648 public function recursivePreprocess( $text, $frame = false ) {
649 $text = $this->replaceVariables( $text, $frame );
650 $text = $this->mStripState->unstripBoth( $text );
651 return $text;
652 }
653
654 /**
655 * Process the wikitext for the "?preload=" feature. (bug 5210)
656 *
657 * "<noinclude>", "<includeonly>" etc. are parsed as for template
658 * transclusion, comments, templates, arguments, tags hooks and parser
659 * functions are untouched.
660 *
661 * @param string $text
662 * @param Title $title
663 * @param ParserOptions $options
664 * @param array $params
665 * @return string
666 */
667 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
668 $msg = new RawMessage( $text );
669 $text = $msg->params( $params )->plain();
670
671 # Parser (re)initialisation
672 $magicScopeVariable = $this->lock();
673 $this->startParse( $title, $options, self::OT_PLAIN, true );
674
675 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
676 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
677 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
678 $text = $this->mStripState->unstripBoth( $text );
679 return $text;
680 }
681
682 /**
683 * Get a random string
684 *
685 * @return string
686 * @deprecated since 1.26; use wfRandomString() instead.
687 */
688 public static function getRandomString() {
689 wfDeprecated( __METHOD__, '1.26' );
690 return wfRandomString( 16 );
691 }
692
693 /**
694 * Set the current user.
695 * Should only be used when doing pre-save transform.
696 *
697 * @param User|null $user User object or null (to reset)
698 */
699 public function setUser( $user ) {
700 $this->mUser = $user;
701 }
702
703 /**
704 * Accessor for mUniqPrefix.
705 *
706 * @return string
707 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
708 */
709 public function uniqPrefix() {
710 wfDeprecated( __METHOD__, '1.26' );
711 return self::MARKER_PREFIX;
712 }
713
714 /**
715 * Set the context title
716 *
717 * @param Title $t
718 */
719 public function setTitle( $t ) {
720 if ( !$t ) {
721 $t = Title::newFromText( 'NO TITLE' );
722 }
723
724 if ( $t->hasFragment() ) {
725 # Strip the fragment to avoid various odd effects
726 $this->mTitle = $t->createFragmentTarget( '' );
727 } else {
728 $this->mTitle = $t;
729 }
730 }
731
732 /**
733 * Accessor for the Title object
734 *
735 * @return Title
736 */
737 public function getTitle() {
738 return $this->mTitle;
739 }
740
741 /**
742 * Accessor/mutator for the Title object
743 *
744 * @param Title $x Title object or null to just get the current one
745 * @return Title
746 */
747 public function Title( $x = null ) {
748 return wfSetVar( $this->mTitle, $x );
749 }
750
751 /**
752 * Set the output type
753 *
754 * @param int $ot New value
755 */
756 public function setOutputType( $ot ) {
757 $this->mOutputType = $ot;
758 # Shortcut alias
759 $this->ot = [
760 'html' => $ot == self::OT_HTML,
761 'wiki' => $ot == self::OT_WIKI,
762 'pre' => $ot == self::OT_PREPROCESS,
763 'plain' => $ot == self::OT_PLAIN,
764 ];
765 }
766
767 /**
768 * Accessor/mutator for the output type
769 *
770 * @param int|null $x New value or null to just get the current one
771 * @return int
772 */
773 public function OutputType( $x = null ) {
774 return wfSetVar( $this->mOutputType, $x );
775 }
776
777 /**
778 * Get the ParserOutput object
779 *
780 * @return ParserOutput
781 */
782 public function getOutput() {
783 return $this->mOutput;
784 }
785
786 /**
787 * Get the ParserOptions object
788 *
789 * @return ParserOptions
790 */
791 public function getOptions() {
792 return $this->mOptions;
793 }
794
795 /**
796 * Accessor/mutator for the ParserOptions object
797 *
798 * @param ParserOptions $x New value or null to just get the current one
799 * @return ParserOptions Current ParserOptions object
800 */
801 public function Options( $x = null ) {
802 return wfSetVar( $this->mOptions, $x );
803 }
804
805 /**
806 * @return int
807 */
808 public function nextLinkID() {
809 return $this->mLinkID++;
810 }
811
812 /**
813 * @param int $id
814 */
815 public function setLinkID( $id ) {
816 $this->mLinkID = $id;
817 }
818
819 /**
820 * Get a language object for use in parser functions such as {{FORMATNUM:}}
821 * @return Language
822 */
823 public function getFunctionLang() {
824 return $this->getTargetLanguage();
825 }
826
827 /**
828 * Get the target language for the content being parsed. This is usually the
829 * language that the content is in.
830 *
831 * @since 1.19
832 *
833 * @throws MWException
834 * @return Language
835 */
836 public function getTargetLanguage() {
837 $target = $this->mOptions->getTargetLanguage();
838
839 if ( $target !== null ) {
840 return $target;
841 } elseif ( $this->mOptions->getInterfaceMessage() ) {
842 return $this->mOptions->getUserLangObj();
843 } elseif ( is_null( $this->mTitle ) ) {
844 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
845 }
846
847 return $this->mTitle->getPageLanguage();
848 }
849
850 /**
851 * Get the language object for language conversion
852 * @return Language|null
853 */
854 public function getConverterLanguage() {
855 return $this->getTargetLanguage();
856 }
857
858 /**
859 * Get a User object either from $this->mUser, if set, or from the
860 * ParserOptions object otherwise
861 *
862 * @return User
863 */
864 public function getUser() {
865 if ( !is_null( $this->mUser ) ) {
866 return $this->mUser;
867 }
868 return $this->mOptions->getUser();
869 }
870
871 /**
872 * Get a preprocessor object
873 *
874 * @return Preprocessor
875 */
876 public function getPreprocessor() {
877 if ( !isset( $this->mPreprocessor ) ) {
878 $class = $this->mPreprocessorClass;
879 $this->mPreprocessor = new $class( $this );
880 }
881 return $this->mPreprocessor;
882 }
883
884 /**
885 * Get a LinkRenderer instance to make links with
886 *
887 * @since 1.28
888 * @return LinkRenderer
889 */
890 public function getLinkRenderer() {
891 if ( !$this->mLinkRenderer ) {
892 $this->mLinkRenderer = MediaWikiServices::getInstance()
893 ->getLinkRendererFactory()->create();
894 $this->mLinkRenderer->setStubThreshold(
895 $this->getOptions()->getStubThreshold()
896 );
897 }
898
899 return $this->mLinkRenderer;
900 }
901
902 /**
903 * Replaces all occurrences of HTML-style comments and the given tags
904 * in the text with a random marker and returns the next text. The output
905 * parameter $matches will be an associative array filled with data in
906 * the form:
907 *
908 * @code
909 * 'UNIQ-xxxxx' => array(
910 * 'element',
911 * 'tag content',
912 * array( 'param' => 'x' ),
913 * '<element param="x">tag content</element>' ) )
914 * @endcode
915 *
916 * @param array $elements List of element names. Comments are always extracted.
917 * @param string $text Source text string.
918 * @param array $matches Out parameter, Array: extracted tags
919 * @param string|null $uniq_prefix
920 * @return string Stripped text
921 * @since 1.26 The uniq_prefix argument is deprecated.
922 */
923 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
924 if ( $uniq_prefix !== null ) {
925 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
926 }
927 static $n = 1;
928 $stripped = '';
929 $matches = [];
930
931 $taglist = implode( '|', $elements );
932 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
933
934 while ( $text != '' ) {
935 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
936 $stripped .= $p[0];
937 if ( count( $p ) < 5 ) {
938 break;
939 }
940 if ( count( $p ) > 5 ) {
941 # comment
942 $element = $p[4];
943 $attributes = '';
944 $close = '';
945 $inside = $p[5];
946 } else {
947 # tag
948 $element = $p[1];
949 $attributes = $p[2];
950 $close = $p[3];
951 $inside = $p[4];
952 }
953
954 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
955 $stripped .= $marker;
956
957 if ( $close === '/>' ) {
958 # Empty element tag, <tag />
959 $content = null;
960 $text = $inside;
961 $tail = null;
962 } else {
963 if ( $element === '!--' ) {
964 $end = '/(-->)/';
965 } else {
966 $end = "/(<\\/$element\\s*>)/i";
967 }
968 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
969 $content = $q[0];
970 if ( count( $q ) < 3 ) {
971 # No end tag -- let it run out to the end of the text.
972 $tail = '';
973 $text = '';
974 } else {
975 $tail = $q[1];
976 $text = $q[2];
977 }
978 }
979
980 $matches[$marker] = [ $element,
981 $content,
982 Sanitizer::decodeTagAttributes( $attributes ),
983 "<$element$attributes$close$content$tail" ];
984 }
985 return $stripped;
986 }
987
988 /**
989 * Get a list of strippable XML-like elements
990 *
991 * @return array
992 */
993 public function getStripList() {
994 return $this->mStripList;
995 }
996
997 /**
998 * Add an item to the strip state
999 * Returns the unique tag which must be inserted into the stripped text
1000 * The tag will be replaced with the original text in unstrip()
1001 *
1002 * @param string $text
1003 *
1004 * @return string
1005 */
1006 public function insertStripItem( $text ) {
1007 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1008 $this->mMarkerIndex++;
1009 $this->mStripState->addGeneral( $marker, $text );
1010 return $marker;
1011 }
1012
1013 /**
1014 * parse the wiki syntax used to render tables
1015 *
1016 * @private
1017 * @param string $text
1018 * @return string
1019 */
1020 public function doTableStuff( $text ) {
1021
1022 $lines = StringUtils::explode( "\n", $text );
1023 $out = '';
1024 $td_history = []; # Is currently a td tag open?
1025 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1026 $tr_history = []; # Is currently a tr tag open?
1027 $tr_attributes = []; # history of tr attributes
1028 $has_opened_tr = []; # Did this table open a <tr> element?
1029 $indent_level = 0; # indent level of the table
1030
1031 foreach ( $lines as $outLine ) {
1032 $line = trim( $outLine );
1033
1034 if ( $line === '' ) { # empty line, go to next line
1035 $out .= $outLine . "\n";
1036 continue;
1037 }
1038
1039 $first_character = $line[0];
1040 $first_two = substr( $line, 0, 2 );
1041 $matches = [];
1042
1043 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1044 # First check if we are starting a new table
1045 $indent_level = strlen( $matches[1] );
1046
1047 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1048 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1049
1050 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1051 array_push( $td_history, false );
1052 array_push( $last_tag_history, '' );
1053 array_push( $tr_history, false );
1054 array_push( $tr_attributes, '' );
1055 array_push( $has_opened_tr, false );
1056 } elseif ( count( $td_history ) == 0 ) {
1057 # Don't do any of the following
1058 $out .= $outLine . "\n";
1059 continue;
1060 } elseif ( $first_two === '|}' ) {
1061 # We are ending a table
1062 $line = '</table>' . substr( $line, 2 );
1063 $last_tag = array_pop( $last_tag_history );
1064
1065 if ( !array_pop( $has_opened_tr ) ) {
1066 $line = "<tr><td></td></tr>{$line}";
1067 }
1068
1069 if ( array_pop( $tr_history ) ) {
1070 $line = "</tr>{$line}";
1071 }
1072
1073 if ( array_pop( $td_history ) ) {
1074 $line = "</{$last_tag}>{$line}";
1075 }
1076 array_pop( $tr_attributes );
1077 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1078 } elseif ( $first_two === '|-' ) {
1079 # Now we have a table row
1080 $line = preg_replace( '#^\|-+#', '', $line );
1081
1082 # Whats after the tag is now only attributes
1083 $attributes = $this->mStripState->unstripBoth( $line );
1084 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1085 array_pop( $tr_attributes );
1086 array_push( $tr_attributes, $attributes );
1087
1088 $line = '';
1089 $last_tag = array_pop( $last_tag_history );
1090 array_pop( $has_opened_tr );
1091 array_push( $has_opened_tr, true );
1092
1093 if ( array_pop( $tr_history ) ) {
1094 $line = '</tr>';
1095 }
1096
1097 if ( array_pop( $td_history ) ) {
1098 $line = "</{$last_tag}>{$line}";
1099 }
1100
1101 $outLine = $line;
1102 array_push( $tr_history, false );
1103 array_push( $td_history, false );
1104 array_push( $last_tag_history, '' );
1105 } elseif ( $first_character === '|'
1106 || $first_character === '!'
1107 || $first_two === '|+'
1108 ) {
1109 # This might be cell elements, td, th or captions
1110 if ( $first_two === '|+' ) {
1111 $first_character = '+';
1112 $line = substr( $line, 2 );
1113 } else {
1114 $line = substr( $line, 1 );
1115 }
1116
1117 // Implies both are valid for table headings.
1118 if ( $first_character === '!' ) {
1119 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1120 }
1121
1122 # Split up multiple cells on the same line.
1123 # FIXME : This can result in improper nesting of tags processed
1124 # by earlier parser steps.
1125 $cells = explode( '||', $line );
1126
1127 $outLine = '';
1128
1129 # Loop through each table cell
1130 foreach ( $cells as $cell ) {
1131 $previous = '';
1132 if ( $first_character !== '+' ) {
1133 $tr_after = array_pop( $tr_attributes );
1134 if ( !array_pop( $tr_history ) ) {
1135 $previous = "<tr{$tr_after}>\n";
1136 }
1137 array_push( $tr_history, true );
1138 array_push( $tr_attributes, '' );
1139 array_pop( $has_opened_tr );
1140 array_push( $has_opened_tr, true );
1141 }
1142
1143 $last_tag = array_pop( $last_tag_history );
1144
1145 if ( array_pop( $td_history ) ) {
1146 $previous = "</{$last_tag}>\n{$previous}";
1147 }
1148
1149 if ( $first_character === '|' ) {
1150 $last_tag = 'td';
1151 } elseif ( $first_character === '!' ) {
1152 $last_tag = 'th';
1153 } elseif ( $first_character === '+' ) {
1154 $last_tag = 'caption';
1155 } else {
1156 $last_tag = '';
1157 }
1158
1159 array_push( $last_tag_history, $last_tag );
1160
1161 # A cell could contain both parameters and data
1162 $cell_data = explode( '|', $cell, 2 );
1163
1164 # Bug 553: Note that a '|' inside an invalid link should not
1165 # be mistaken as delimiting cell parameters
1166 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1167 $cell = "{$previous}<{$last_tag}>{$cell}";
1168 } elseif ( count( $cell_data ) == 1 ) {
1169 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1170 } else {
1171 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1172 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1173 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1174 }
1175
1176 $outLine .= $cell;
1177 array_push( $td_history, true );
1178 }
1179 }
1180 $out .= $outLine . "\n";
1181 }
1182
1183 # Closing open td, tr && table
1184 while ( count( $td_history ) > 0 ) {
1185 if ( array_pop( $td_history ) ) {
1186 $out .= "</td>\n";
1187 }
1188 if ( array_pop( $tr_history ) ) {
1189 $out .= "</tr>\n";
1190 }
1191 if ( !array_pop( $has_opened_tr ) ) {
1192 $out .= "<tr><td></td></tr>\n";
1193 }
1194
1195 $out .= "</table>\n";
1196 }
1197
1198 # Remove trailing line-ending (b/c)
1199 if ( substr( $out, -1 ) === "\n" ) {
1200 $out = substr( $out, 0, -1 );
1201 }
1202
1203 # special case: don't return empty table
1204 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1205 $out = '';
1206 }
1207
1208 return $out;
1209 }
1210
1211 /**
1212 * Helper function for parse() that transforms wiki markup into half-parsed
1213 * HTML. Only called for $mOutputType == self::OT_HTML.
1214 *
1215 * @private
1216 *
1217 * @param string $text The text to parse
1218 * @param bool $isMain Whether this is being called from the main parse() function
1219 * @param PPFrame|bool $frame A pre-processor frame
1220 *
1221 * @return string
1222 */
1223 public function internalParse( $text, $isMain = true, $frame = false ) {
1224
1225 $origText = $text;
1226
1227 # Hook to suspend the parser in this state
1228 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1229 return $text;
1230 }
1231
1232 # if $frame is provided, then use $frame for replacing any variables
1233 if ( $frame ) {
1234 # use frame depth to infer how include/noinclude tags should be handled
1235 # depth=0 means this is the top-level document; otherwise it's an included document
1236 if ( !$frame->depth ) {
1237 $flag = 0;
1238 } else {
1239 $flag = Parser::PTD_FOR_INCLUSION;
1240 }
1241 $dom = $this->preprocessToDom( $text, $flag );
1242 $text = $frame->expand( $dom );
1243 } else {
1244 # if $frame is not provided, then use old-style replaceVariables
1245 $text = $this->replaceVariables( $text );
1246 }
1247
1248 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1249 $text = Sanitizer::removeHTMLtags(
1250 $text,
1251 [ &$this, 'attributeStripCallback' ],
1252 false,
1253 array_keys( $this->mTransparentTagHooks ),
1254 [],
1255 [ &$this, 'addTrackingCategory' ]
1256 );
1257 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1258
1259 # Tables need to come after variable replacement for things to work
1260 # properly; putting them before other transformations should keep
1261 # exciting things like link expansions from showing up in surprising
1262 # places.
1263 $text = $this->doTableStuff( $text );
1264
1265 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1266
1267 $text = $this->doDoubleUnderscore( $text );
1268
1269 $text = $this->doHeadings( $text );
1270 $text = $this->replaceInternalLinks( $text );
1271 $text = $this->doAllQuotes( $text );
1272 $text = $this->replaceExternalLinks( $text );
1273
1274 # replaceInternalLinks may sometimes leave behind
1275 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1276 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1277
1278 $text = $this->doMagicLinks( $text );
1279 $text = $this->formatHeadings( $text, $origText, $isMain );
1280
1281 return $text;
1282 }
1283
1284 /**
1285 * Helper function for parse() that transforms half-parsed HTML into fully
1286 * parsed HTML.
1287 *
1288 * @param string $text
1289 * @param bool $isMain
1290 * @param bool $linestart
1291 * @return string
1292 */
1293 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1294 $text = $this->mStripState->unstripGeneral( $text );
1295
1296 if ( $isMain ) {
1297 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1298 }
1299
1300 # Clean up special characters, only run once, next-to-last before doBlockLevels
1301 $fixtags = [
1302 # french spaces, last one Guillemet-left
1303 # only if there is something before the space
1304 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1305 # french spaces, Guillemet-right
1306 '/(\\302\\253) /' => '\\1&#160;',
1307 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1308 ];
1309 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1310
1311 $text = $this->doBlockLevels( $text, $linestart );
1312
1313 $this->replaceLinkHolders( $text );
1314
1315 /**
1316 * The input doesn't get language converted if
1317 * a) It's disabled
1318 * b) Content isn't converted
1319 * c) It's a conversion table
1320 * d) it is an interface message (which is in the user language)
1321 */
1322 if ( !( $this->mOptions->getDisableContentConversion()
1323 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1324 ) {
1325 if ( !$this->mOptions->getInterfaceMessage() ) {
1326 # The position of the convert() call should not be changed. it
1327 # assumes that the links are all replaced and the only thing left
1328 # is the <nowiki> mark.
1329 $text = $this->getConverterLanguage()->convert( $text );
1330 }
1331 }
1332
1333 $text = $this->mStripState->unstripNoWiki( $text );
1334
1335 if ( $isMain ) {
1336 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1337 }
1338
1339 $text = $this->replaceTransparentTags( $text );
1340 $text = $this->mStripState->unstripGeneral( $text );
1341
1342 $text = Sanitizer::normalizeCharReferences( $text );
1343
1344 if ( MWTidy::isEnabled() && $this->mOptions->getTidy() ) {
1345 $text = MWTidy::tidy( $text );
1346 } else {
1347 # attempt to sanitize at least some nesting problems
1348 # (bug #2702 and quite a few others)
1349 $tidyregs = [
1350 # ''Something [http://www.cool.com cool''] -->
1351 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1352 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1353 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1354 # fix up an anchor inside another anchor, only
1355 # at least for a single single nested link (bug 3695)
1356 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1357 '\\1\\2</a>\\3</a>\\1\\4</a>',
1358 # fix div inside inline elements- doBlockLevels won't wrap a line which
1359 # contains a div, so fix it up here; replace
1360 # div with escaped text
1361 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1362 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1363 # remove empty italic or bold tag pairs, some
1364 # introduced by rules above
1365 '/<([bi])><\/\\1>/' => '',
1366 ];
1367
1368 $text = preg_replace(
1369 array_keys( $tidyregs ),
1370 array_values( $tidyregs ),
1371 $text );
1372 }
1373
1374 if ( $isMain ) {
1375 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1376 }
1377
1378 return $text;
1379 }
1380
1381 /**
1382 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1383 * magic external links.
1384 *
1385 * DML
1386 * @private
1387 *
1388 * @param string $text
1389 *
1390 * @return string
1391 */
1392 public function doMagicLinks( $text ) {
1393 $prots = wfUrlProtocolsWithoutProtRel();
1394 $urlChar = self::EXT_LINK_URL_CLASS;
1395 $addr = self::EXT_LINK_ADDR;
1396 $space = self::SPACE_NOT_NL; # non-newline space
1397 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1398 $spaces = "$space++"; # possessive match of 1 or more spaces
1399 $text = preg_replace_callback(
1400 '!(?: # Start cases
1401 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1402 (<.*?>) | # m[2]: Skip stuff inside
1403 # HTML elements' . "
1404 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1405 # m[4]: Post-protocol path
1406 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1407 ([0-9]+)\b |
1408 \bISBN $spaces ( # m[6]: ISBN, capture number
1409 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1410 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1411 [0-9Xx] # check digit
1412 )\b
1413 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1414 return $text;
1415 }
1416
1417 /**
1418 * @throws MWException
1419 * @param array $m
1420 * @return HTML|string
1421 */
1422 public function magicLinkCallback( $m ) {
1423 if ( isset( $m[1] ) && $m[1] !== '' ) {
1424 # Skip anchor
1425 return $m[0];
1426 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1427 # Skip HTML element
1428 return $m[0];
1429 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1430 # Free external link
1431 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1432 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1433 # RFC or PMID
1434 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1435 $keyword = 'RFC';
1436 $urlmsg = 'rfcurl';
1437 $cssClass = 'mw-magiclink-rfc';
1438 $id = $m[5];
1439 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1440 $keyword = 'PMID';
1441 $urlmsg = 'pubmedurl';
1442 $cssClass = 'mw-magiclink-pmid';
1443 $id = $m[5];
1444 } else {
1445 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1446 substr( $m[0], 0, 20 ) . '"' );
1447 }
1448 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1449 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1450 } elseif ( isset( $m[6] ) && $m[6] !== '' ) {
1451 # ISBN
1452 $isbn = $m[6];
1453 $space = self::SPACE_NOT_NL; # non-newline space
1454 $isbn = preg_replace( "/$space/", ' ', $isbn );
1455 $num = strtr( $isbn, [
1456 '-' => '',
1457 ' ' => '',
1458 'x' => 'X',
1459 ] );
1460 return $this->getLinkRenderer()->makeKnownLink(
1461 SpecialPage::getTitleFor( 'Booksources', $num ),
1462 "ISBN $isbn",
1463 [
1464 'class' => 'internal mw-magiclink-isbn',
1465 'title' => false // suppress title attribute
1466 ]
1467 );
1468 } else {
1469 return $m[0];
1470 }
1471 }
1472
1473 /**
1474 * Make a free external link, given a user-supplied URL
1475 *
1476 * @param string $url
1477 * @param int $numPostProto
1478 * The number of characters after the protocol.
1479 * @return string HTML
1480 * @private
1481 */
1482 public function makeFreeExternalLink( $url, $numPostProto ) {
1483 $trail = '';
1484
1485 # The characters '<' and '>' (which were escaped by
1486 # removeHTMLtags()) should not be included in
1487 # URLs, per RFC 2396.
1488 # Make &nbsp; terminate a URL as well (bug T84937)
1489 $m2 = [];
1490 if ( preg_match(
1491 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1492 $url,
1493 $m2,
1494 PREG_OFFSET_CAPTURE
1495 ) ) {
1496 $trail = substr( $url, $m2[0][1] ) . $trail;
1497 $url = substr( $url, 0, $m2[0][1] );
1498 }
1499
1500 # Move trailing punctuation to $trail
1501 $sep = ',;\.:!?';
1502 # If there is no left bracket, then consider right brackets fair game too
1503 if ( strpos( $url, '(' ) === false ) {
1504 $sep .= ')';
1505 }
1506
1507 $urlRev = strrev( $url );
1508 $numSepChars = strspn( $urlRev, $sep );
1509 # Don't break a trailing HTML entity by moving the ; into $trail
1510 # This is in hot code, so use substr_compare to avoid having to
1511 # create a new string object for the comparison
1512 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1513 # more optimization: instead of running preg_match with a $
1514 # anchor, which can be slow, do the match on the reversed
1515 # string starting at the desired offset.
1516 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1517 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1518 $numSepChars--;
1519 }
1520 }
1521 if ( $numSepChars ) {
1522 $trail = substr( $url, -$numSepChars ) . $trail;
1523 $url = substr( $url, 0, -$numSepChars );
1524 }
1525
1526 # Verify that we still have a real URL after trail removal, and
1527 # not just lone protocol
1528 if ( strlen( $trail ) >= $numPostProto ) {
1529 return $url . $trail;
1530 }
1531
1532 $url = Sanitizer::cleanUrl( $url );
1533
1534 # Is this an external image?
1535 $text = $this->maybeMakeExternalImage( $url );
1536 if ( $text === false ) {
1537 # Not an image, make a link
1538 $text = Linker::makeExternalLink( $url,
1539 $this->getConverterLanguage()->markNoConversion( $url, true ),
1540 true, 'free',
1541 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1542 # Register it in the output object...
1543 # Replace unnecessary URL escape codes with their equivalent characters
1544 $pasteurized = self::normalizeLinkUrl( $url );
1545 $this->mOutput->addExternalLink( $pasteurized );
1546 }
1547 return $text . $trail;
1548 }
1549
1550 /**
1551 * Parse headers and return html
1552 *
1553 * @private
1554 *
1555 * @param string $text
1556 *
1557 * @return string
1558 */
1559 public function doHeadings( $text ) {
1560 for ( $i = 6; $i >= 1; --$i ) {
1561 $h = str_repeat( '=', $i );
1562 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1563 }
1564 return $text;
1565 }
1566
1567 /**
1568 * Replace single quotes with HTML markup
1569 * @private
1570 *
1571 * @param string $text
1572 *
1573 * @return string The altered text
1574 */
1575 public function doAllQuotes( $text ) {
1576 $outtext = '';
1577 $lines = StringUtils::explode( "\n", $text );
1578 foreach ( $lines as $line ) {
1579 $outtext .= $this->doQuotes( $line ) . "\n";
1580 }
1581 $outtext = substr( $outtext, 0, -1 );
1582 return $outtext;
1583 }
1584
1585 /**
1586 * Helper function for doAllQuotes()
1587 *
1588 * @param string $text
1589 *
1590 * @return string
1591 */
1592 public function doQuotes( $text ) {
1593 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1594 $countarr = count( $arr );
1595 if ( $countarr == 1 ) {
1596 return $text;
1597 }
1598
1599 // First, do some preliminary work. This may shift some apostrophes from
1600 // being mark-up to being text. It also counts the number of occurrences
1601 // of bold and italics mark-ups.
1602 $numbold = 0;
1603 $numitalics = 0;
1604 for ( $i = 1; $i < $countarr; $i += 2 ) {
1605 $thislen = strlen( $arr[$i] );
1606 // If there are ever four apostrophes, assume the first is supposed to
1607 // be text, and the remaining three constitute mark-up for bold text.
1608 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1609 if ( $thislen == 4 ) {
1610 $arr[$i - 1] .= "'";
1611 $arr[$i] = "'''";
1612 $thislen = 3;
1613 } elseif ( $thislen > 5 ) {
1614 // If there are more than 5 apostrophes in a row, assume they're all
1615 // text except for the last 5.
1616 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1617 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1618 $arr[$i] = "'''''";
1619 $thislen = 5;
1620 }
1621 // Count the number of occurrences of bold and italics mark-ups.
1622 if ( $thislen == 2 ) {
1623 $numitalics++;
1624 } elseif ( $thislen == 3 ) {
1625 $numbold++;
1626 } elseif ( $thislen == 5 ) {
1627 $numitalics++;
1628 $numbold++;
1629 }
1630 }
1631
1632 // If there is an odd number of both bold and italics, it is likely
1633 // that one of the bold ones was meant to be an apostrophe followed
1634 // by italics. Which one we cannot know for certain, but it is more
1635 // likely to be one that has a single-letter word before it.
1636 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1637 $firstsingleletterword = -1;
1638 $firstmultiletterword = -1;
1639 $firstspace = -1;
1640 for ( $i = 1; $i < $countarr; $i += 2 ) {
1641 if ( strlen( $arr[$i] ) == 3 ) {
1642 $x1 = substr( $arr[$i - 1], -1 );
1643 $x2 = substr( $arr[$i - 1], -2, 1 );
1644 if ( $x1 === ' ' ) {
1645 if ( $firstspace == -1 ) {
1646 $firstspace = $i;
1647 }
1648 } elseif ( $x2 === ' ' ) {
1649 $firstsingleletterword = $i;
1650 // if $firstsingleletterword is set, we don't
1651 // look at the other options, so we can bail early.
1652 break;
1653 } else {
1654 if ( $firstmultiletterword == -1 ) {
1655 $firstmultiletterword = $i;
1656 }
1657 }
1658 }
1659 }
1660
1661 // If there is a single-letter word, use it!
1662 if ( $firstsingleletterword > -1 ) {
1663 $arr[$firstsingleletterword] = "''";
1664 $arr[$firstsingleletterword - 1] .= "'";
1665 } elseif ( $firstmultiletterword > -1 ) {
1666 // If not, but there's a multi-letter word, use that one.
1667 $arr[$firstmultiletterword] = "''";
1668 $arr[$firstmultiletterword - 1] .= "'";
1669 } elseif ( $firstspace > -1 ) {
1670 // ... otherwise use the first one that has neither.
1671 // (notice that it is possible for all three to be -1 if, for example,
1672 // there is only one pentuple-apostrophe in the line)
1673 $arr[$firstspace] = "''";
1674 $arr[$firstspace - 1] .= "'";
1675 }
1676 }
1677
1678 // Now let's actually convert our apostrophic mush to HTML!
1679 $output = '';
1680 $buffer = '';
1681 $state = '';
1682 $i = 0;
1683 foreach ( $arr as $r ) {
1684 if ( ( $i % 2 ) == 0 ) {
1685 if ( $state === 'both' ) {
1686 $buffer .= $r;
1687 } else {
1688 $output .= $r;
1689 }
1690 } else {
1691 $thislen = strlen( $r );
1692 if ( $thislen == 2 ) {
1693 if ( $state === 'i' ) {
1694 $output .= '</i>';
1695 $state = '';
1696 } elseif ( $state === 'bi' ) {
1697 $output .= '</i>';
1698 $state = 'b';
1699 } elseif ( $state === 'ib' ) {
1700 $output .= '</b></i><b>';
1701 $state = 'b';
1702 } elseif ( $state === 'both' ) {
1703 $output .= '<b><i>' . $buffer . '</i>';
1704 $state = 'b';
1705 } else { // $state can be 'b' or ''
1706 $output .= '<i>';
1707 $state .= 'i';
1708 }
1709 } elseif ( $thislen == 3 ) {
1710 if ( $state === 'b' ) {
1711 $output .= '</b>';
1712 $state = '';
1713 } elseif ( $state === 'bi' ) {
1714 $output .= '</i></b><i>';
1715 $state = 'i';
1716 } elseif ( $state === 'ib' ) {
1717 $output .= '</b>';
1718 $state = 'i';
1719 } elseif ( $state === 'both' ) {
1720 $output .= '<i><b>' . $buffer . '</b>';
1721 $state = 'i';
1722 } else { // $state can be 'i' or ''
1723 $output .= '<b>';
1724 $state .= 'b';
1725 }
1726 } elseif ( $thislen == 5 ) {
1727 if ( $state === 'b' ) {
1728 $output .= '</b><i>';
1729 $state = 'i';
1730 } elseif ( $state === 'i' ) {
1731 $output .= '</i><b>';
1732 $state = 'b';
1733 } elseif ( $state === 'bi' ) {
1734 $output .= '</i></b>';
1735 $state = '';
1736 } elseif ( $state === 'ib' ) {
1737 $output .= '</b></i>';
1738 $state = '';
1739 } elseif ( $state === 'both' ) {
1740 $output .= '<i><b>' . $buffer . '</b></i>';
1741 $state = '';
1742 } else { // ($state == '')
1743 $buffer = '';
1744 $state = 'both';
1745 }
1746 }
1747 }
1748 $i++;
1749 }
1750 // Now close all remaining tags. Notice that the order is important.
1751 if ( $state === 'b' || $state === 'ib' ) {
1752 $output .= '</b>';
1753 }
1754 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1755 $output .= '</i>';
1756 }
1757 if ( $state === 'bi' ) {
1758 $output .= '</b>';
1759 }
1760 // There might be lonely ''''', so make sure we have a buffer
1761 if ( $state === 'both' && $buffer ) {
1762 $output .= '<b><i>' . $buffer . '</i></b>';
1763 }
1764 return $output;
1765 }
1766
1767 /**
1768 * Replace external links (REL)
1769 *
1770 * Note: this is all very hackish and the order of execution matters a lot.
1771 * Make sure to run tests/parserTests.php if you change this code.
1772 *
1773 * @private
1774 *
1775 * @param string $text
1776 *
1777 * @throws MWException
1778 * @return string
1779 */
1780 public function replaceExternalLinks( $text ) {
1781
1782 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1783 if ( $bits === false ) {
1784 throw new MWException( "PCRE needs to be compiled with "
1785 . "--enable-unicode-properties in order for MediaWiki to function" );
1786 }
1787 $s = array_shift( $bits );
1788
1789 $i = 0;
1790 while ( $i < count( $bits ) ) {
1791 $url = $bits[$i++];
1792 $i++; // protocol
1793 $text = $bits[$i++];
1794 $trail = $bits[$i++];
1795
1796 # The characters '<' and '>' (which were escaped by
1797 # removeHTMLtags()) should not be included in
1798 # URLs, per RFC 2396.
1799 $m2 = [];
1800 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1801 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1802 $url = substr( $url, 0, $m2[0][1] );
1803 }
1804
1805 # If the link text is an image URL, replace it with an <img> tag
1806 # This happened by accident in the original parser, but some people used it extensively
1807 $img = $this->maybeMakeExternalImage( $text );
1808 if ( $img !== false ) {
1809 $text = $img;
1810 }
1811
1812 $dtrail = '';
1813
1814 # Set linktype for CSS - if URL==text, link is essentially free
1815 $linktype = ( $text === $url ) ? 'free' : 'text';
1816
1817 # No link text, e.g. [http://domain.tld/some.link]
1818 if ( $text == '' ) {
1819 # Autonumber
1820 $langObj = $this->getTargetLanguage();
1821 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1822 $linktype = 'autonumber';
1823 } else {
1824 # Have link text, e.g. [http://domain.tld/some.link text]s
1825 # Check for trail
1826 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1827 }
1828
1829 $text = $this->getConverterLanguage()->markNoConversion( $text );
1830
1831 $url = Sanitizer::cleanUrl( $url );
1832
1833 # Use the encoded URL
1834 # This means that users can paste URLs directly into the text
1835 # Funny characters like ö aren't valid in URLs anyway
1836 # This was changed in August 2004
1837 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1838 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1839
1840 # Register link in the output object.
1841 # Replace unnecessary URL escape codes with the referenced character
1842 # This prevents spammers from hiding links from the filters
1843 $pasteurized = self::normalizeLinkUrl( $url );
1844 $this->mOutput->addExternalLink( $pasteurized );
1845 }
1846
1847 return $s;
1848 }
1849
1850 /**
1851 * Get the rel attribute for a particular external link.
1852 *
1853 * @since 1.21
1854 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1855 * nofollow if appropriate
1856 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1857 * @return string|null Rel attribute for $url
1858 */
1859 public static function getExternalLinkRel( $url = false, $title = null ) {
1860 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1861 $ns = $title ? $title->getNamespace() : false;
1862 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1863 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1864 ) {
1865 return 'nofollow';
1866 }
1867 return null;
1868 }
1869
1870 /**
1871 * Get an associative array of additional HTML attributes appropriate for a
1872 * particular external link. This currently may include rel => nofollow
1873 * (depending on configuration, namespace, and the URL's domain) and/or a
1874 * target attribute (depending on configuration).
1875 *
1876 * @param string $url URL to extract the domain from for rel =>
1877 * nofollow if appropriate
1878 * @return array Associative array of HTML attributes
1879 */
1880 public function getExternalLinkAttribs( $url ) {
1881 $attribs = [];
1882 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1883
1884 $target = $this->mOptions->getExternalLinkTarget();
1885 if ( $target ) {
1886 $attribs['target'] = $target;
1887 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1888 // T133507. New windows can navigate parent cross-origin.
1889 // Including noreferrer due to lacking browser
1890 // support of noopener. Eventually noreferrer should be removed.
1891 if ( $rel !== '' ) {
1892 $rel .= ' ';
1893 }
1894 $rel .= 'noreferrer noopener';
1895 }
1896 }
1897 $attribs['rel'] = $rel;
1898 return $attribs;
1899 }
1900
1901 /**
1902 * Replace unusual escape codes in a URL with their equivalent characters
1903 *
1904 * @deprecated since 1.24, use normalizeLinkUrl
1905 * @param string $url
1906 * @return string
1907 */
1908 public static function replaceUnusualEscapes( $url ) {
1909 wfDeprecated( __METHOD__, '1.24' );
1910 return self::normalizeLinkUrl( $url );
1911 }
1912
1913 /**
1914 * Replace unusual escape codes in a URL with their equivalent characters
1915 *
1916 * This generally follows the syntax defined in RFC 3986, with special
1917 * consideration for HTTP query strings.
1918 *
1919 * @param string $url
1920 * @return string
1921 */
1922 public static function normalizeLinkUrl( $url ) {
1923 # First, make sure unsafe characters are encoded
1924 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1925 function ( $m ) {
1926 return rawurlencode( $m[0] );
1927 },
1928 $url
1929 );
1930
1931 $ret = '';
1932 $end = strlen( $url );
1933
1934 # Fragment part - 'fragment'
1935 $start = strpos( $url, '#' );
1936 if ( $start !== false && $start < $end ) {
1937 $ret = self::normalizeUrlComponent(
1938 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1939 $end = $start;
1940 }
1941
1942 # Query part - 'query' minus &=+;
1943 $start = strpos( $url, '?' );
1944 if ( $start !== false && $start < $end ) {
1945 $ret = self::normalizeUrlComponent(
1946 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1947 $end = $start;
1948 }
1949
1950 # Scheme and path part - 'pchar'
1951 # (we assume no userinfo or encoded colons in the host)
1952 $ret = self::normalizeUrlComponent(
1953 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1954
1955 return $ret;
1956 }
1957
1958 private static function normalizeUrlComponent( $component, $unsafe ) {
1959 $callback = function ( $matches ) use ( $unsafe ) {
1960 $char = urldecode( $matches[0] );
1961 $ord = ord( $char );
1962 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1963 # Unescape it
1964 return $char;
1965 } else {
1966 # Leave it escaped, but use uppercase for a-f
1967 return strtoupper( $matches[0] );
1968 }
1969 };
1970 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1971 }
1972
1973 /**
1974 * make an image if it's allowed, either through the global
1975 * option, through the exception, or through the on-wiki whitelist
1976 *
1977 * @param string $url
1978 *
1979 * @return string
1980 */
1981 private function maybeMakeExternalImage( $url ) {
1982 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1983 $imagesexception = !empty( $imagesfrom );
1984 $text = false;
1985 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1986 if ( $imagesexception && is_array( $imagesfrom ) ) {
1987 $imagematch = false;
1988 foreach ( $imagesfrom as $match ) {
1989 if ( strpos( $url, $match ) === 0 ) {
1990 $imagematch = true;
1991 break;
1992 }
1993 }
1994 } elseif ( $imagesexception ) {
1995 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1996 } else {
1997 $imagematch = false;
1998 }
1999
2000 if ( $this->mOptions->getAllowExternalImages()
2001 || ( $imagesexception && $imagematch )
2002 ) {
2003 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2004 # Image found
2005 $text = Linker::makeExternalImage( $url );
2006 }
2007 }
2008 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2009 && preg_match( self::EXT_IMAGE_REGEX, $url )
2010 ) {
2011 $whitelist = explode(
2012 "\n",
2013 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2014 );
2015
2016 foreach ( $whitelist as $entry ) {
2017 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2018 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2019 continue;
2020 }
2021 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2022 # Image matches a whitelist entry
2023 $text = Linker::makeExternalImage( $url );
2024 break;
2025 }
2026 }
2027 }
2028 return $text;
2029 }
2030
2031 /**
2032 * Process [[ ]] wikilinks
2033 *
2034 * @param string $s
2035 *
2036 * @return string Processed text
2037 *
2038 * @private
2039 */
2040 public function replaceInternalLinks( $s ) {
2041 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2042 return $s;
2043 }
2044
2045 /**
2046 * Process [[ ]] wikilinks (RIL)
2047 * @param string $s
2048 * @throws MWException
2049 * @return LinkHolderArray
2050 *
2051 * @private
2052 */
2053 public function replaceInternalLinks2( &$s ) {
2054 global $wgExtraInterlanguageLinkPrefixes;
2055
2056 static $tc = false, $e1, $e1_img;
2057 # the % is needed to support urlencoded titles as well
2058 if ( !$tc ) {
2059 $tc = Title::legalChars() . '#%';
2060 # Match a link having the form [[namespace:link|alternate]]trail
2061 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2062 # Match cases where there is no "]]", which might still be images
2063 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2064 }
2065
2066 $holders = new LinkHolderArray( $this );
2067
2068 # split the entire text string on occurrences of [[
2069 $a = StringUtils::explode( '[[', ' ' . $s );
2070 # get the first element (all text up to first [[), and remove the space we added
2071 $s = $a->current();
2072 $a->next();
2073 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2074 $s = substr( $s, 1 );
2075
2076 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2077 $e2 = null;
2078 if ( $useLinkPrefixExtension ) {
2079 # Match the end of a line for a word that's not followed by whitespace,
2080 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2081 global $wgContLang;
2082 $charset = $wgContLang->linkPrefixCharset();
2083 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2084 }
2085
2086 if ( is_null( $this->mTitle ) ) {
2087 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2088 }
2089 $nottalk = !$this->mTitle->isTalkPage();
2090
2091 if ( $useLinkPrefixExtension ) {
2092 $m = [];
2093 if ( preg_match( $e2, $s, $m ) ) {
2094 $first_prefix = $m[2];
2095 } else {
2096 $first_prefix = false;
2097 }
2098 } else {
2099 $prefix = '';
2100 }
2101
2102 $useSubpages = $this->areSubpagesAllowed();
2103
2104 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2105 # Loop for each link
2106 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2107 // @codingStandardsIgnoreEnd
2108
2109 # Check for excessive memory usage
2110 if ( $holders->isBig() ) {
2111 # Too big
2112 # Do the existence check, replace the link holders and clear the array
2113 $holders->replace( $s );
2114 $holders->clear();
2115 }
2116
2117 if ( $useLinkPrefixExtension ) {
2118 if ( preg_match( $e2, $s, $m ) ) {
2119 $prefix = $m[2];
2120 $s = $m[1];
2121 } else {
2122 $prefix = '';
2123 }
2124 # first link
2125 if ( $first_prefix ) {
2126 $prefix = $first_prefix;
2127 $first_prefix = false;
2128 }
2129 }
2130
2131 $might_be_img = false;
2132
2133 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2134 $text = $m[2];
2135 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2136 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2137 # the real problem is with the $e1 regex
2138 # See bug 1300.
2139 # Still some problems for cases where the ] is meant to be outside punctuation,
2140 # and no image is in sight. See bug 2095.
2141 if ( $text !== ''
2142 && substr( $m[3], 0, 1 ) === ']'
2143 && strpos( $text, '[' ) !== false
2144 ) {
2145 $text .= ']'; # so that replaceExternalLinks($text) works later
2146 $m[3] = substr( $m[3], 1 );
2147 }
2148 # fix up urlencoded title texts
2149 if ( strpos( $m[1], '%' ) !== false ) {
2150 # Should anchors '#' also be rejected?
2151 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2152 }
2153 $trail = $m[3];
2154 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2155 # Invalid, but might be an image with a link in its caption
2156 $might_be_img = true;
2157 $text = $m[2];
2158 if ( strpos( $m[1], '%' ) !== false ) {
2159 $m[1] = rawurldecode( $m[1] );
2160 }
2161 $trail = "";
2162 } else { # Invalid form; output directly
2163 $s .= $prefix . '[[' . $line;
2164 continue;
2165 }
2166
2167 $origLink = $m[1];
2168
2169 # Don't allow internal links to pages containing
2170 # PROTO: where PROTO is a valid URL protocol; these
2171 # should be external links.
2172 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2173 $s .= $prefix . '[[' . $line;
2174 continue;
2175 }
2176
2177 # Make subpage if necessary
2178 if ( $useSubpages ) {
2179 $link = $this->maybeDoSubpageLink( $origLink, $text );
2180 } else {
2181 $link = $origLink;
2182 }
2183
2184 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2185 if ( !$noforce ) {
2186 # Strip off leading ':'
2187 $link = substr( $link, 1 );
2188 }
2189
2190 $unstrip = $this->mStripState->unstripNoWiki( $link );
2191 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2192 if ( $nt === null ) {
2193 $s .= $prefix . '[[' . $line;
2194 continue;
2195 }
2196
2197 $ns = $nt->getNamespace();
2198 $iw = $nt->getInterwiki();
2199
2200 if ( $might_be_img ) { # if this is actually an invalid link
2201 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2202 $found = false;
2203 while ( true ) {
2204 # look at the next 'line' to see if we can close it there
2205 $a->next();
2206 $next_line = $a->current();
2207 if ( $next_line === false || $next_line === null ) {
2208 break;
2209 }
2210 $m = explode( ']]', $next_line, 3 );
2211 if ( count( $m ) == 3 ) {
2212 # the first ]] closes the inner link, the second the image
2213 $found = true;
2214 $text .= "[[{$m[0]}]]{$m[1]}";
2215 $trail = $m[2];
2216 break;
2217 } elseif ( count( $m ) == 2 ) {
2218 # if there's exactly one ]] that's fine, we'll keep looking
2219 $text .= "[[{$m[0]}]]{$m[1]}";
2220 } else {
2221 # if $next_line is invalid too, we need look no further
2222 $text .= '[[' . $next_line;
2223 break;
2224 }
2225 }
2226 if ( !$found ) {
2227 # we couldn't find the end of this imageLink, so output it raw
2228 # but don't ignore what might be perfectly normal links in the text we've examined
2229 $holders->merge( $this->replaceInternalLinks2( $text ) );
2230 $s .= "{$prefix}[[$link|$text";
2231 # note: no $trail, because without an end, there *is* no trail
2232 continue;
2233 }
2234 } else { # it's not an image, so output it raw
2235 $s .= "{$prefix}[[$link|$text";
2236 # note: no $trail, because without an end, there *is* no trail
2237 continue;
2238 }
2239 }
2240
2241 $wasblank = ( $text == '' );
2242 if ( $wasblank ) {
2243 $text = $link;
2244 } else {
2245 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2246 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2247 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2248 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2249 $text = $this->doQuotes( $text );
2250 }
2251
2252 # Link not escaped by : , create the various objects
2253 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2254 # Interwikis
2255 if (
2256 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2257 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2258 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2259 )
2260 ) {
2261 # Bug 24502: filter duplicates
2262 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2263 $this->mLangLinkLanguages[$iw] = true;
2264 $this->mOutput->addLanguageLink( $nt->getFullText() );
2265 }
2266
2267 $s = rtrim( $s . $prefix );
2268 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2269 continue;
2270 }
2271
2272 if ( $ns == NS_FILE ) {
2273 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2274 if ( $wasblank ) {
2275 # if no parameters were passed, $text
2276 # becomes something like "File:Foo.png",
2277 # which we don't want to pass on to the
2278 # image generator
2279 $text = '';
2280 } else {
2281 # recursively parse links inside the image caption
2282 # actually, this will parse them in any other parameters, too,
2283 # but it might be hard to fix that, and it doesn't matter ATM
2284 $text = $this->replaceExternalLinks( $text );
2285 $holders->merge( $this->replaceInternalLinks2( $text ) );
2286 }
2287 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2288 $s .= $prefix . $this->armorLinks(
2289 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2290 continue;
2291 }
2292 } elseif ( $ns == NS_CATEGORY ) {
2293 $s = rtrim( $s . "\n" ); # bug 87
2294
2295 if ( $wasblank ) {
2296 $sortkey = $this->getDefaultSort();
2297 } else {
2298 $sortkey = $text;
2299 }
2300 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2301 $sortkey = str_replace( "\n", '', $sortkey );
2302 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2303 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2304
2305 /**
2306 * Strip the whitespace Category links produce, see bug 87
2307 */
2308 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2309
2310 continue;
2311 }
2312 }
2313
2314 # Self-link checking. For some languages, variants of the title are checked in
2315 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2316 # for linking to a different variant.
2317 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2318 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2319 continue;
2320 }
2321
2322 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2323 # @todo FIXME: Should do batch file existence checks, see comment below
2324 if ( $ns == NS_MEDIA ) {
2325 # Give extensions a chance to select the file revision for us
2326 $options = [];
2327 $descQuery = false;
2328 Hooks::run( 'BeforeParserFetchFileAndTitle',
2329 [ $this, $nt, &$options, &$descQuery ] );
2330 # Fetch and register the file (file title may be different via hooks)
2331 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2332 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2333 $s .= $prefix . $this->armorLinks(
2334 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2335 continue;
2336 }
2337
2338 # Some titles, such as valid special pages or files in foreign repos, should
2339 # be shown as bluelinks even though they're not included in the page table
2340 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2341 # batch file existence checks for NS_FILE and NS_MEDIA
2342 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2343 $this->mOutput->addLink( $nt );
2344 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2345 } else {
2346 # Links will be added to the output link list after checking
2347 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2348 }
2349 }
2350 return $holders;
2351 }
2352
2353 /**
2354 * Render a forced-blue link inline; protect against double expansion of
2355 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2356 * Since this little disaster has to split off the trail text to avoid
2357 * breaking URLs in the following text without breaking trails on the
2358 * wiki links, it's been made into a horrible function.
2359 *
2360 * @param Title $nt
2361 * @param string $text
2362 * @param string $trail
2363 * @param string $prefix
2364 * @return string HTML-wikitext mix oh yuck
2365 */
2366 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2367 list( $inside, $trail ) = Linker::splitTrail( $trail );
2368
2369 if ( $text == '' ) {
2370 $text = htmlspecialchars( $nt->getPrefixedText() );
2371 }
2372
2373 $link = $this->getLinkRenderer()->makeKnownLink(
2374 $nt, new HtmlArmor( "$prefix$text$inside" )
2375 );
2376
2377 return $this->armorLinks( $link ) . $trail;
2378 }
2379
2380 /**
2381 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2382 * going to go through further parsing steps before inline URL expansion.
2383 *
2384 * Not needed quite as much as it used to be since free links are a bit
2385 * more sensible these days. But bracketed links are still an issue.
2386 *
2387 * @param string $text More-or-less HTML
2388 * @return string Less-or-more HTML with NOPARSE bits
2389 */
2390 public function armorLinks( $text ) {
2391 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2392 self::MARKER_PREFIX . "NOPARSE$1", $text );
2393 }
2394
2395 /**
2396 * Return true if subpage links should be expanded on this page.
2397 * @return bool
2398 */
2399 public function areSubpagesAllowed() {
2400 # Some namespaces don't allow subpages
2401 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2402 }
2403
2404 /**
2405 * Handle link to subpage if necessary
2406 *
2407 * @param string $target The source of the link
2408 * @param string &$text The link text, modified as necessary
2409 * @return string The full name of the link
2410 * @private
2411 */
2412 public function maybeDoSubpageLink( $target, &$text ) {
2413 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2414 }
2415
2416 /**
2417 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2418 *
2419 * @param string $text
2420 * @param bool $linestart Whether or not this is at the start of a line.
2421 * @private
2422 * @return string The lists rendered as HTML
2423 */
2424 public function doBlockLevels( $text, $linestart ) {
2425 return BlockLevelPass::doBlockLevels( $text, $linestart );
2426 }
2427
2428 /**
2429 * Return value of a magic variable (like PAGENAME)
2430 *
2431 * @private
2432 *
2433 * @param int $index
2434 * @param bool|PPFrame $frame
2435 *
2436 * @throws MWException
2437 * @return string
2438 */
2439 public function getVariableValue( $index, $frame = false ) {
2440 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2441 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2442
2443 if ( is_null( $this->mTitle ) ) {
2444 // If no title set, bad things are going to happen
2445 // later. Title should always be set since this
2446 // should only be called in the middle of a parse
2447 // operation (but the unit-tests do funky stuff)
2448 throw new MWException( __METHOD__ . ' Should only be '
2449 . ' called while parsing (no title set)' );
2450 }
2451
2452 /**
2453 * Some of these require message or data lookups and can be
2454 * expensive to check many times.
2455 */
2456 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2457 if ( isset( $this->mVarCache[$index] ) ) {
2458 return $this->mVarCache[$index];
2459 }
2460 }
2461
2462 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2463 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2464
2465 $pageLang = $this->getFunctionLang();
2466
2467 switch ( $index ) {
2468 case '!':
2469 $value = '|';
2470 break;
2471 case 'currentmonth':
2472 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2473 break;
2474 case 'currentmonth1':
2475 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2476 break;
2477 case 'currentmonthname':
2478 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2479 break;
2480 case 'currentmonthnamegen':
2481 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2482 break;
2483 case 'currentmonthabbrev':
2484 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2485 break;
2486 case 'currentday':
2487 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2488 break;
2489 case 'currentday2':
2490 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2491 break;
2492 case 'localmonth':
2493 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2494 break;
2495 case 'localmonth1':
2496 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2497 break;
2498 case 'localmonthname':
2499 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2500 break;
2501 case 'localmonthnamegen':
2502 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2503 break;
2504 case 'localmonthabbrev':
2505 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2506 break;
2507 case 'localday':
2508 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2509 break;
2510 case 'localday2':
2511 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2512 break;
2513 case 'pagename':
2514 $value = wfEscapeWikiText( $this->mTitle->getText() );
2515 break;
2516 case 'pagenamee':
2517 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2518 break;
2519 case 'fullpagename':
2520 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2521 break;
2522 case 'fullpagenamee':
2523 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2524 break;
2525 case 'subpagename':
2526 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2527 break;
2528 case 'subpagenamee':
2529 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2530 break;
2531 case 'rootpagename':
2532 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2533 break;
2534 case 'rootpagenamee':
2535 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2536 ' ',
2537 '_',
2538 $this->mTitle->getRootText()
2539 ) ) );
2540 break;
2541 case 'basepagename':
2542 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2543 break;
2544 case 'basepagenamee':
2545 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2546 ' ',
2547 '_',
2548 $this->mTitle->getBaseText()
2549 ) ) );
2550 break;
2551 case 'talkpagename':
2552 if ( $this->mTitle->canTalk() ) {
2553 $talkPage = $this->mTitle->getTalkPage();
2554 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2555 } else {
2556 $value = '';
2557 }
2558 break;
2559 case 'talkpagenamee':
2560 if ( $this->mTitle->canTalk() ) {
2561 $talkPage = $this->mTitle->getTalkPage();
2562 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2563 } else {
2564 $value = '';
2565 }
2566 break;
2567 case 'subjectpagename':
2568 $subjPage = $this->mTitle->getSubjectPage();
2569 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2570 break;
2571 case 'subjectpagenamee':
2572 $subjPage = $this->mTitle->getSubjectPage();
2573 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2574 break;
2575 case 'pageid': // requested in bug 23427
2576 $pageid = $this->getTitle()->getArticleID();
2577 if ( $pageid == 0 ) {
2578 # 0 means the page doesn't exist in the database,
2579 # which means the user is previewing a new page.
2580 # The vary-revision flag must be set, because the magic word
2581 # will have a different value once the page is saved.
2582 $this->mOutput->setFlag( 'vary-revision' );
2583 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2584 }
2585 $value = $pageid ? $pageid : null;
2586 break;
2587 case 'revisionid':
2588 # Let the edit saving system know we should parse the page
2589 # *after* a revision ID has been assigned.
2590 $this->mOutput->setFlag( 'vary-revision-id' );
2591 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2592 $value = $this->mRevisionId;
2593 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2594 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2595 $this->mOutput->setSpeculativeRevIdUsed( $value );
2596 }
2597 break;
2598 case 'revisionday':
2599 # Let the edit saving system know we should parse the page
2600 # *after* a revision ID has been assigned. This is for null edits.
2601 $this->mOutput->setFlag( 'vary-revision' );
2602 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2603 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2604 break;
2605 case 'revisionday2':
2606 # Let the edit saving system know we should parse the page
2607 # *after* a revision ID has been assigned. This is for null edits.
2608 $this->mOutput->setFlag( 'vary-revision' );
2609 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2610 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2611 break;
2612 case 'revisionmonth':
2613 # Let the edit saving system know we should parse the page
2614 # *after* a revision ID has been assigned. This is for null edits.
2615 $this->mOutput->setFlag( 'vary-revision' );
2616 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2617 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2618 break;
2619 case 'revisionmonth1':
2620 # Let the edit saving system know we should parse the page
2621 # *after* a revision ID has been assigned. This is for null edits.
2622 $this->mOutput->setFlag( 'vary-revision' );
2623 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2624 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2625 break;
2626 case 'revisionyear':
2627 # Let the edit saving system know we should parse the page
2628 # *after* a revision ID has been assigned. This is for null edits.
2629 $this->mOutput->setFlag( 'vary-revision' );
2630 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2631 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2632 break;
2633 case 'revisiontimestamp':
2634 # Let the edit saving system know we should parse the page
2635 # *after* a revision ID has been assigned. This is for null edits.
2636 $this->mOutput->setFlag( 'vary-revision' );
2637 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2638 $value = $this->getRevisionTimestamp();
2639 break;
2640 case 'revisionuser':
2641 # Let the edit saving system know we should parse the page
2642 # *after* a revision ID has been assigned for null edits.
2643 $this->mOutput->setFlag( 'vary-user' );
2644 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2645 $value = $this->getRevisionUser();
2646 break;
2647 case 'revisionsize':
2648 $value = $this->getRevisionSize();
2649 break;
2650 case 'namespace':
2651 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2652 break;
2653 case 'namespacee':
2654 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2655 break;
2656 case 'namespacenumber':
2657 $value = $this->mTitle->getNamespace();
2658 break;
2659 case 'talkspace':
2660 $value = $this->mTitle->canTalk()
2661 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2662 : '';
2663 break;
2664 case 'talkspacee':
2665 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2666 break;
2667 case 'subjectspace':
2668 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2669 break;
2670 case 'subjectspacee':
2671 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2672 break;
2673 case 'currentdayname':
2674 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2675 break;
2676 case 'currentyear':
2677 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2678 break;
2679 case 'currenttime':
2680 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2681 break;
2682 case 'currenthour':
2683 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2684 break;
2685 case 'currentweek':
2686 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2687 # int to remove the padding
2688 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2689 break;
2690 case 'currentdow':
2691 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2692 break;
2693 case 'localdayname':
2694 $value = $pageLang->getWeekdayName(
2695 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2696 );
2697 break;
2698 case 'localyear':
2699 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2700 break;
2701 case 'localtime':
2702 $value = $pageLang->time(
2703 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2704 false,
2705 false
2706 );
2707 break;
2708 case 'localhour':
2709 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2710 break;
2711 case 'localweek':
2712 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2713 # int to remove the padding
2714 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2715 break;
2716 case 'localdow':
2717 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2718 break;
2719 case 'numberofarticles':
2720 $value = $pageLang->formatNum( SiteStats::articles() );
2721 break;
2722 case 'numberoffiles':
2723 $value = $pageLang->formatNum( SiteStats::images() );
2724 break;
2725 case 'numberofusers':
2726 $value = $pageLang->formatNum( SiteStats::users() );
2727 break;
2728 case 'numberofactiveusers':
2729 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2730 break;
2731 case 'numberofpages':
2732 $value = $pageLang->formatNum( SiteStats::pages() );
2733 break;
2734 case 'numberofadmins':
2735 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2736 break;
2737 case 'numberofedits':
2738 $value = $pageLang->formatNum( SiteStats::edits() );
2739 break;
2740 case 'currenttimestamp':
2741 $value = wfTimestamp( TS_MW, $ts );
2742 break;
2743 case 'localtimestamp':
2744 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2745 break;
2746 case 'currentversion':
2747 $value = SpecialVersion::getVersion();
2748 break;
2749 case 'articlepath':
2750 return $wgArticlePath;
2751 case 'sitename':
2752 return $wgSitename;
2753 case 'server':
2754 return $wgServer;
2755 case 'servername':
2756 return $wgServerName;
2757 case 'scriptpath':
2758 return $wgScriptPath;
2759 case 'stylepath':
2760 return $wgStylePath;
2761 case 'directionmark':
2762 return $pageLang->getDirMark();
2763 case 'contentlanguage':
2764 global $wgLanguageCode;
2765 return $wgLanguageCode;
2766 case 'cascadingsources':
2767 $value = CoreParserFunctions::cascadingsources( $this );
2768 break;
2769 default:
2770 $ret = null;
2771 Hooks::run(
2772 'ParserGetVariableValueSwitch',
2773 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2774 );
2775
2776 return $ret;
2777 }
2778
2779 if ( $index ) {
2780 $this->mVarCache[$index] = $value;
2781 }
2782
2783 return $value;
2784 }
2785
2786 /**
2787 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2788 *
2789 * @private
2790 */
2791 public function initialiseVariables() {
2792 $variableIDs = MagicWord::getVariableIDs();
2793 $substIDs = MagicWord::getSubstIDs();
2794
2795 $this->mVariables = new MagicWordArray( $variableIDs );
2796 $this->mSubstWords = new MagicWordArray( $substIDs );
2797 }
2798
2799 /**
2800 * Preprocess some wikitext and return the document tree.
2801 * This is the ghost of replace_variables().
2802 *
2803 * @param string $text The text to parse
2804 * @param int $flags Bitwise combination of:
2805 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2806 * included. Default is to assume a direct page view.
2807 *
2808 * The generated DOM tree must depend only on the input text and the flags.
2809 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2810 *
2811 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2812 * change in the DOM tree for a given text, must be passed through the section identifier
2813 * in the section edit link and thus back to extractSections().
2814 *
2815 * The output of this function is currently only cached in process memory, but a persistent
2816 * cache may be implemented at a later date which takes further advantage of these strict
2817 * dependency requirements.
2818 *
2819 * @return PPNode
2820 */
2821 public function preprocessToDom( $text, $flags = 0 ) {
2822 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2823 return $dom;
2824 }
2825
2826 /**
2827 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2828 *
2829 * @param string $s
2830 *
2831 * @return array
2832 */
2833 public static function splitWhitespace( $s ) {
2834 $ltrimmed = ltrim( $s );
2835 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2836 $trimmed = rtrim( $ltrimmed );
2837 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2838 if ( $diff > 0 ) {
2839 $w2 = substr( $ltrimmed, -$diff );
2840 } else {
2841 $w2 = '';
2842 }
2843 return [ $w1, $trimmed, $w2 ];
2844 }
2845
2846 /**
2847 * Replace magic variables, templates, and template arguments
2848 * with the appropriate text. Templates are substituted recursively,
2849 * taking care to avoid infinite loops.
2850 *
2851 * Note that the substitution depends on value of $mOutputType:
2852 * self::OT_WIKI: only {{subst:}} templates
2853 * self::OT_PREPROCESS: templates but not extension tags
2854 * self::OT_HTML: all templates and extension tags
2855 *
2856 * @param string $text The text to transform
2857 * @param bool|PPFrame $frame Object describing the arguments passed to the
2858 * template. Arguments may also be provided as an associative array, as
2859 * was the usual case before MW1.12. Providing arguments this way may be
2860 * useful for extensions wishing to perform variable replacement
2861 * explicitly.
2862 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2863 * double-brace expansion.
2864 * @return string
2865 */
2866 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2867 # Is there any text? Also, Prevent too big inclusions!
2868 $textSize = strlen( $text );
2869 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2870 return $text;
2871 }
2872
2873 if ( $frame === false ) {
2874 $frame = $this->getPreprocessor()->newFrame();
2875 } elseif ( !( $frame instanceof PPFrame ) ) {
2876 wfDebug( __METHOD__ . " called using plain parameters instead of "
2877 . "a PPFrame instance. Creating custom frame.\n" );
2878 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2879 }
2880
2881 $dom = $this->preprocessToDom( $text );
2882 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2883 $text = $frame->expand( $dom, $flags );
2884
2885 return $text;
2886 }
2887
2888 /**
2889 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2890 *
2891 * @param array $args
2892 *
2893 * @return array
2894 */
2895 public static function createAssocArgs( $args ) {
2896 $assocArgs = [];
2897 $index = 1;
2898 foreach ( $args as $arg ) {
2899 $eqpos = strpos( $arg, '=' );
2900 if ( $eqpos === false ) {
2901 $assocArgs[$index++] = $arg;
2902 } else {
2903 $name = trim( substr( $arg, 0, $eqpos ) );
2904 $value = trim( substr( $arg, $eqpos + 1 ) );
2905 if ( $value === false ) {
2906 $value = '';
2907 }
2908 if ( $name !== false ) {
2909 $assocArgs[$name] = $value;
2910 }
2911 }
2912 }
2913
2914 return $assocArgs;
2915 }
2916
2917 /**
2918 * Warn the user when a parser limitation is reached
2919 * Will warn at most once the user per limitation type
2920 *
2921 * The results are shown during preview and run through the Parser (See EditPage.php)
2922 *
2923 * @param string $limitationType Should be one of:
2924 * 'expensive-parserfunction' (corresponding messages:
2925 * 'expensive-parserfunction-warning',
2926 * 'expensive-parserfunction-category')
2927 * 'post-expand-template-argument' (corresponding messages:
2928 * 'post-expand-template-argument-warning',
2929 * 'post-expand-template-argument-category')
2930 * 'post-expand-template-inclusion' (corresponding messages:
2931 * 'post-expand-template-inclusion-warning',
2932 * 'post-expand-template-inclusion-category')
2933 * 'node-count-exceeded' (corresponding messages:
2934 * 'node-count-exceeded-warning',
2935 * 'node-count-exceeded-category')
2936 * 'expansion-depth-exceeded' (corresponding messages:
2937 * 'expansion-depth-exceeded-warning',
2938 * 'expansion-depth-exceeded-category')
2939 * @param string|int|null $current Current value
2940 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2941 * exceeded, provide the values (optional)
2942 */
2943 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2944 # does no harm if $current and $max are present but are unnecessary for the message
2945 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2946 # only during preview, and that would split the parser cache unnecessarily.
2947 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2948 ->text();
2949 $this->mOutput->addWarning( $warning );
2950 $this->addTrackingCategory( "$limitationType-category" );
2951 }
2952
2953 /**
2954 * Return the text of a template, after recursively
2955 * replacing any variables or templates within the template.
2956 *
2957 * @param array $piece The parts of the template
2958 * $piece['title']: the title, i.e. the part before the |
2959 * $piece['parts']: the parameter array
2960 * $piece['lineStart']: whether the brace was at the start of a line
2961 * @param PPFrame $frame The current frame, contains template arguments
2962 * @throws Exception
2963 * @return string The text of the template
2964 */
2965 public function braceSubstitution( $piece, $frame ) {
2966
2967 // Flags
2968
2969 // $text has been filled
2970 $found = false;
2971 // wiki markup in $text should be escaped
2972 $nowiki = false;
2973 // $text is HTML, armour it against wikitext transformation
2974 $isHTML = false;
2975 // Force interwiki transclusion to be done in raw mode not rendered
2976 $forceRawInterwiki = false;
2977 // $text is a DOM node needing expansion in a child frame
2978 $isChildObj = false;
2979 // $text is a DOM node needing expansion in the current frame
2980 $isLocalObj = false;
2981
2982 # Title object, where $text came from
2983 $title = false;
2984
2985 # $part1 is the bit before the first |, and must contain only title characters.
2986 # Various prefixes will be stripped from it later.
2987 $titleWithSpaces = $frame->expand( $piece['title'] );
2988 $part1 = trim( $titleWithSpaces );
2989 $titleText = false;
2990
2991 # Original title text preserved for various purposes
2992 $originalTitle = $part1;
2993
2994 # $args is a list of argument nodes, starting from index 0, not including $part1
2995 # @todo FIXME: If piece['parts'] is null then the call to getLength()
2996 # below won't work b/c this $args isn't an object
2997 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
2998
2999 $profileSection = null; // profile templates
3000
3001 # SUBST
3002 if ( !$found ) {
3003 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3004
3005 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3006 # Decide whether to expand template or keep wikitext as-is.
3007 if ( $this->ot['wiki'] ) {
3008 if ( $substMatch === false ) {
3009 $literal = true; # literal when in PST with no prefix
3010 } else {
3011 $literal = false; # expand when in PST with subst: or safesubst:
3012 }
3013 } else {
3014 if ( $substMatch == 'subst' ) {
3015 $literal = true; # literal when not in PST with plain subst:
3016 } else {
3017 $literal = false; # expand when not in PST with safesubst: or no prefix
3018 }
3019 }
3020 if ( $literal ) {
3021 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3022 $isLocalObj = true;
3023 $found = true;
3024 }
3025 }
3026
3027 # Variables
3028 if ( !$found && $args->getLength() == 0 ) {
3029 $id = $this->mVariables->matchStartToEnd( $part1 );
3030 if ( $id !== false ) {
3031 $text = $this->getVariableValue( $id, $frame );
3032 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3033 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3034 }
3035 $found = true;
3036 }
3037 }
3038
3039 # MSG, MSGNW and RAW
3040 if ( !$found ) {
3041 # Check for MSGNW:
3042 $mwMsgnw = MagicWord::get( 'msgnw' );
3043 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3044 $nowiki = true;
3045 } else {
3046 # Remove obsolete MSG:
3047 $mwMsg = MagicWord::get( 'msg' );
3048 $mwMsg->matchStartAndRemove( $part1 );
3049 }
3050
3051 # Check for RAW:
3052 $mwRaw = MagicWord::get( 'raw' );
3053 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3054 $forceRawInterwiki = true;
3055 }
3056 }
3057
3058 # Parser functions
3059 if ( !$found ) {
3060 $colonPos = strpos( $part1, ':' );
3061 if ( $colonPos !== false ) {
3062 $func = substr( $part1, 0, $colonPos );
3063 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3064 $argsLength = $args->getLength();
3065 for ( $i = 0; $i < $argsLength; $i++ ) {
3066 $funcArgs[] = $args->item( $i );
3067 }
3068 try {
3069 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3070 } catch ( Exception $ex ) {
3071 throw $ex;
3072 }
3073
3074 # The interface for parser functions allows for extracting
3075 # flags into the local scope. Extract any forwarded flags
3076 # here.
3077 extract( $result );
3078 }
3079 }
3080
3081 # Finish mangling title and then check for loops.
3082 # Set $title to a Title object and $titleText to the PDBK
3083 if ( !$found ) {
3084 $ns = NS_TEMPLATE;
3085 # Split the title into page and subpage
3086 $subpage = '';
3087 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3088 if ( $part1 !== $relative ) {
3089 $part1 = $relative;
3090 $ns = $this->mTitle->getNamespace();
3091 }
3092 $title = Title::newFromText( $part1, $ns );
3093 if ( $title ) {
3094 $titleText = $title->getPrefixedText();
3095 # Check for language variants if the template is not found
3096 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3097 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3098 }
3099 # Do recursion depth check
3100 $limit = $this->mOptions->getMaxTemplateDepth();
3101 if ( $frame->depth >= $limit ) {
3102 $found = true;
3103 $text = '<span class="error">'
3104 . wfMessage( 'parser-template-recursion-depth-warning' )
3105 ->numParams( $limit )->inContentLanguage()->text()
3106 . '</span>';
3107 }
3108 }
3109 }
3110
3111 # Load from database
3112 if ( !$found && $title ) {
3113 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3114 if ( !$title->isExternal() ) {
3115 if ( $title->isSpecialPage()
3116 && $this->mOptions->getAllowSpecialInclusion()
3117 && $this->ot['html']
3118 ) {
3119 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3120 // Pass the template arguments as URL parameters.
3121 // "uselang" will have no effect since the Language object
3122 // is forced to the one defined in ParserOptions.
3123 $pageArgs = [];
3124 $argsLength = $args->getLength();
3125 for ( $i = 0; $i < $argsLength; $i++ ) {
3126 $bits = $args->item( $i )->splitArg();
3127 if ( strval( $bits['index'] ) === '' ) {
3128 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3129 $value = trim( $frame->expand( $bits['value'] ) );
3130 $pageArgs[$name] = $value;
3131 }
3132 }
3133
3134 // Create a new context to execute the special page
3135 $context = new RequestContext;
3136 $context->setTitle( $title );
3137 $context->setRequest( new FauxRequest( $pageArgs ) );
3138 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3139 $context->setUser( $this->getUser() );
3140 } else {
3141 // If this page is cached, then we better not be per user.
3142 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3143 }
3144 $context->setLanguage( $this->mOptions->getUserLangObj() );
3145 $ret = SpecialPageFactory::capturePath( $title, $context, $this->getLinkRenderer() );
3146 if ( $ret ) {
3147 $text = $context->getOutput()->getHTML();
3148 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3149 $found = true;
3150 $isHTML = true;
3151 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3152 $this->mOutput->updateCacheExpiry( $specialPage->maxIncludeCacheTime() );
3153 }
3154 }
3155 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3156 $found = false; # access denied
3157 wfDebug( __METHOD__ . ": template inclusion denied for " .
3158 $title->getPrefixedDBkey() . "\n" );
3159 } else {
3160 list( $text, $title ) = $this->getTemplateDom( $title );
3161 if ( $text !== false ) {
3162 $found = true;
3163 $isChildObj = true;
3164 }
3165 }
3166
3167 # If the title is valid but undisplayable, make a link to it
3168 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3169 $text = "[[:$titleText]]";
3170 $found = true;
3171 }
3172 } elseif ( $title->isTrans() ) {
3173 # Interwiki transclusion
3174 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3175 $text = $this->interwikiTransclude( $title, 'render' );
3176 $isHTML = true;
3177 } else {
3178 $text = $this->interwikiTransclude( $title, 'raw' );
3179 # Preprocess it like a template
3180 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3181 $isChildObj = true;
3182 }
3183 $found = true;
3184 }
3185
3186 # Do infinite loop check
3187 # This has to be done after redirect resolution to avoid infinite loops via redirects
3188 if ( !$frame->loopCheck( $title ) ) {
3189 $found = true;
3190 $text = '<span class="error">'
3191 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3192 . '</span>';
3193 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3194 }
3195 }
3196
3197 # If we haven't found text to substitute by now, we're done
3198 # Recover the source wikitext and return it
3199 if ( !$found ) {
3200 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3201 if ( $profileSection ) {
3202 $this->mProfiler->scopedProfileOut( $profileSection );
3203 }
3204 return [ 'object' => $text ];
3205 }
3206
3207 # Expand DOM-style return values in a child frame
3208 if ( $isChildObj ) {
3209 # Clean up argument array
3210 $newFrame = $frame->newChild( $args, $title );
3211
3212 if ( $nowiki ) {
3213 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3214 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3215 # Expansion is eligible for the empty-frame cache
3216 $text = $newFrame->cachedExpand( $titleText, $text );
3217 } else {
3218 # Uncached expansion
3219 $text = $newFrame->expand( $text );
3220 }
3221 }
3222 if ( $isLocalObj && $nowiki ) {
3223 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3224 $isLocalObj = false;
3225 }
3226
3227 if ( $profileSection ) {
3228 $this->mProfiler->scopedProfileOut( $profileSection );
3229 }
3230
3231 # Replace raw HTML by a placeholder
3232 if ( $isHTML ) {
3233 $text = $this->insertStripItem( $text );
3234 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3235 # Escape nowiki-style return values
3236 $text = wfEscapeWikiText( $text );
3237 } elseif ( is_string( $text )
3238 && !$piece['lineStart']
3239 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3240 ) {
3241 # Bug 529: if the template begins with a table or block-level
3242 # element, it should be treated as beginning a new line.
3243 # This behavior is somewhat controversial.
3244 $text = "\n" . $text;
3245 }
3246
3247 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3248 # Error, oversize inclusion
3249 if ( $titleText !== false ) {
3250 # Make a working, properly escaped link if possible (bug 23588)
3251 $text = "[[:$titleText]]";
3252 } else {
3253 # This will probably not be a working link, but at least it may
3254 # provide some hint of where the problem is
3255 preg_replace( '/^:/', '', $originalTitle );
3256 $text = "[[:$originalTitle]]";
3257 }
3258 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3259 . 'post-expand include size too large -->' );
3260 $this->limitationWarn( 'post-expand-template-inclusion' );
3261 }
3262
3263 if ( $isLocalObj ) {
3264 $ret = [ 'object' => $text ];
3265 } else {
3266 $ret = [ 'text' => $text ];
3267 }
3268
3269 return $ret;
3270 }
3271
3272 /**
3273 * Call a parser function and return an array with text and flags.
3274 *
3275 * The returned array will always contain a boolean 'found', indicating
3276 * whether the parser function was found or not. It may also contain the
3277 * following:
3278 * text: string|object, resulting wikitext or PP DOM object
3279 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3280 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3281 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3282 * nowiki: bool, wiki markup in $text should be escaped
3283 *
3284 * @since 1.21
3285 * @param PPFrame $frame The current frame, contains template arguments
3286 * @param string $function Function name
3287 * @param array $args Arguments to the function
3288 * @throws MWException
3289 * @return array
3290 */
3291 public function callParserFunction( $frame, $function, array $args = [] ) {
3292 global $wgContLang;
3293
3294 # Case sensitive functions
3295 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3296 $function = $this->mFunctionSynonyms[1][$function];
3297 } else {
3298 # Case insensitive functions
3299 $function = $wgContLang->lc( $function );
3300 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3301 $function = $this->mFunctionSynonyms[0][$function];
3302 } else {
3303 return [ 'found' => false ];
3304 }
3305 }
3306
3307 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3308
3309 # Workaround for PHP bug 35229 and similar
3310 if ( !is_callable( $callback ) ) {
3311 throw new MWException( "Tag hook for $function is not callable\n" );
3312 }
3313
3314 $allArgs = [ &$this ];
3315 if ( $flags & self::SFH_OBJECT_ARGS ) {
3316 # Convert arguments to PPNodes and collect for appending to $allArgs
3317 $funcArgs = [];
3318 foreach ( $args as $k => $v ) {
3319 if ( $v instanceof PPNode || $k === 0 ) {
3320 $funcArgs[] = $v;
3321 } else {
3322 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3323 }
3324 }
3325
3326 # Add a frame parameter, and pass the arguments as an array
3327 $allArgs[] = $frame;
3328 $allArgs[] = $funcArgs;
3329 } else {
3330 # Convert arguments to plain text and append to $allArgs
3331 foreach ( $args as $k => $v ) {
3332 if ( $v instanceof PPNode ) {
3333 $allArgs[] = trim( $frame->expand( $v ) );
3334 } elseif ( is_int( $k ) && $k >= 0 ) {
3335 $allArgs[] = trim( $v );
3336 } else {
3337 $allArgs[] = trim( "$k=$v" );
3338 }
3339 }
3340 }
3341
3342 $result = call_user_func_array( $callback, $allArgs );
3343
3344 # The interface for function hooks allows them to return a wikitext
3345 # string or an array containing the string and any flags. This mungs
3346 # things around to match what this method should return.
3347 if ( !is_array( $result ) ) {
3348 $result =[
3349 'found' => true,
3350 'text' => $result,
3351 ];
3352 } else {
3353 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3354 $result['text'] = $result[0];
3355 }
3356 unset( $result[0] );
3357 $result += [
3358 'found' => true,
3359 ];
3360 }
3361
3362 $noparse = true;
3363 $preprocessFlags = 0;
3364 if ( isset( $result['noparse'] ) ) {
3365 $noparse = $result['noparse'];
3366 }
3367 if ( isset( $result['preprocessFlags'] ) ) {
3368 $preprocessFlags = $result['preprocessFlags'];
3369 }
3370
3371 if ( !$noparse ) {
3372 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3373 $result['isChildObj'] = true;
3374 }
3375
3376 return $result;
3377 }
3378
3379 /**
3380 * Get the semi-parsed DOM representation of a template with a given title,
3381 * and its redirect destination title. Cached.
3382 *
3383 * @param Title $title
3384 *
3385 * @return array
3386 */
3387 public function getTemplateDom( $title ) {
3388 $cacheTitle = $title;
3389 $titleText = $title->getPrefixedDBkey();
3390
3391 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3392 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3393 $title = Title::makeTitle( $ns, $dbk );
3394 $titleText = $title->getPrefixedDBkey();
3395 }
3396 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3397 return [ $this->mTplDomCache[$titleText], $title ];
3398 }
3399
3400 # Cache miss, go to the database
3401 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3402
3403 if ( $text === false ) {
3404 $this->mTplDomCache[$titleText] = false;
3405 return [ false, $title ];
3406 }
3407
3408 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3409 $this->mTplDomCache[$titleText] = $dom;
3410
3411 if ( !$title->equals( $cacheTitle ) ) {
3412 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3413 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3414 }
3415
3416 return [ $dom, $title ];
3417 }
3418
3419 /**
3420 * Fetch the current revision of a given title. Note that the revision
3421 * (and even the title) may not exist in the database, so everything
3422 * contributing to the output of the parser should use this method
3423 * where possible, rather than getting the revisions themselves. This
3424 * method also caches its results, so using it benefits performance.
3425 *
3426 * @since 1.24
3427 * @param Title $title
3428 * @return Revision
3429 */
3430 public function fetchCurrentRevisionOfTitle( $title ) {
3431 $cacheKey = $title->getPrefixedDBkey();
3432 if ( !$this->currentRevisionCache ) {
3433 $this->currentRevisionCache = new MapCacheLRU( 100 );
3434 }
3435 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3436 $this->currentRevisionCache->set( $cacheKey,
3437 // Defaults to Parser::statelessFetchRevision()
3438 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3439 );
3440 }
3441 return $this->currentRevisionCache->get( $cacheKey );
3442 }
3443
3444 /**
3445 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3446 * without passing them on to it.
3447 *
3448 * @since 1.24
3449 * @param Title $title
3450 * @param Parser|bool $parser
3451 * @return Revision
3452 */
3453 public static function statelessFetchRevision( $title, $parser = false ) {
3454 return Revision::newFromTitle( $title );
3455 }
3456
3457 /**
3458 * Fetch the unparsed text of a template and register a reference to it.
3459 * @param Title $title
3460 * @return array ( string or false, Title )
3461 */
3462 public function fetchTemplateAndTitle( $title ) {
3463 // Defaults to Parser::statelessFetchTemplate()
3464 $templateCb = $this->mOptions->getTemplateCallback();
3465 $stuff = call_user_func( $templateCb, $title, $this );
3466 // We use U+007F DELETE to distinguish strip markers from regular text.
3467 $text = $stuff['text'];
3468 if ( is_string( $stuff['text'] ) ) {
3469 $text = strtr( $text, "\x7f", "?" );
3470 }
3471 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3472 if ( isset( $stuff['deps'] ) ) {
3473 foreach ( $stuff['deps'] as $dep ) {
3474 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3475 if ( $dep['title']->equals( $this->getTitle() ) ) {
3476 // If we transclude ourselves, the final result
3477 // will change based on the new version of the page
3478 $this->mOutput->setFlag( 'vary-revision' );
3479 }
3480 }
3481 }
3482 return [ $text, $finalTitle ];
3483 }
3484
3485 /**
3486 * Fetch the unparsed text of a template and register a reference to it.
3487 * @param Title $title
3488 * @return string|bool
3489 */
3490 public function fetchTemplate( $title ) {
3491 return $this->fetchTemplateAndTitle( $title )[0];
3492 }
3493
3494 /**
3495 * Static function to get a template
3496 * Can be overridden via ParserOptions::setTemplateCallback().
3497 *
3498 * @param Title $title
3499 * @param bool|Parser $parser
3500 *
3501 * @return array
3502 */
3503 public static function statelessFetchTemplate( $title, $parser = false ) {
3504 $text = $skip = false;
3505 $finalTitle = $title;
3506 $deps = [];
3507
3508 # Loop to fetch the article, with up to 1 redirect
3509 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3510 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3511 // @codingStandardsIgnoreEnd
3512 # Give extensions a chance to select the revision instead
3513 $id = false; # Assume current
3514 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3515 [ $parser, $title, &$skip, &$id ] );
3516
3517 if ( $skip ) {
3518 $text = false;
3519 $deps[] = [
3520 'title' => $title,
3521 'page_id' => $title->getArticleID(),
3522 'rev_id' => null
3523 ];
3524 break;
3525 }
3526 # Get the revision
3527 if ( $id ) {
3528 $rev = Revision::newFromId( $id );
3529 } elseif ( $parser ) {
3530 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3531 } else {
3532 $rev = Revision::newFromTitle( $title );
3533 }
3534 $rev_id = $rev ? $rev->getId() : 0;
3535 # If there is no current revision, there is no page
3536 if ( $id === false && !$rev ) {
3537 $linkCache = LinkCache::singleton();
3538 $linkCache->addBadLinkObj( $title );
3539 }
3540
3541 $deps[] = [
3542 'title' => $title,
3543 'page_id' => $title->getArticleID(),
3544 'rev_id' => $rev_id ];
3545 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3546 # We fetched a rev from a different title; register it too...
3547 $deps[] = [
3548 'title' => $rev->getTitle(),
3549 'page_id' => $rev->getPage(),
3550 'rev_id' => $rev_id ];
3551 }
3552
3553 if ( $rev ) {
3554 $content = $rev->getContent();
3555 $text = $content ? $content->getWikitextForTransclusion() : null;
3556
3557 if ( $text === false || $text === null ) {
3558 $text = false;
3559 break;
3560 }
3561 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3562 global $wgContLang;
3563 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3564 if ( !$message->exists() ) {
3565 $text = false;
3566 break;
3567 }
3568 $content = $message->content();
3569 $text = $message->plain();
3570 } else {
3571 break;
3572 }
3573 if ( !$content ) {
3574 break;
3575 }
3576 # Redirect?
3577 $finalTitle = $title;
3578 $title = $content->getRedirectTarget();
3579 }
3580 return [
3581 'text' => $text,
3582 'finalTitle' => $finalTitle,
3583 'deps' => $deps ];
3584 }
3585
3586 /**
3587 * Fetch a file and its title and register a reference to it.
3588 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3589 * @param Title $title
3590 * @param array $options Array of options to RepoGroup::findFile
3591 * @return File|bool
3592 */
3593 public function fetchFile( $title, $options = [] ) {
3594 return $this->fetchFileAndTitle( $title, $options )[0];
3595 }
3596
3597 /**
3598 * Fetch a file and its title and register a reference to it.
3599 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3600 * @param Title $title
3601 * @param array $options Array of options to RepoGroup::findFile
3602 * @return array ( File or false, Title of file )
3603 */
3604 public function fetchFileAndTitle( $title, $options = [] ) {
3605 $file = $this->fetchFileNoRegister( $title, $options );
3606
3607 $time = $file ? $file->getTimestamp() : false;
3608 $sha1 = $file ? $file->getSha1() : false;
3609 # Register the file as a dependency...
3610 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3611 if ( $file && !$title->equals( $file->getTitle() ) ) {
3612 # Update fetched file title
3613 $title = $file->getTitle();
3614 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3615 }
3616 return [ $file, $title ];
3617 }
3618
3619 /**
3620 * Helper function for fetchFileAndTitle.
3621 *
3622 * Also useful if you need to fetch a file but not use it yet,
3623 * for example to get the file's handler.
3624 *
3625 * @param Title $title
3626 * @param array $options Array of options to RepoGroup::findFile
3627 * @return File|bool
3628 */
3629 protected function fetchFileNoRegister( $title, $options = [] ) {
3630 if ( isset( $options['broken'] ) ) {
3631 $file = false; // broken thumbnail forced by hook
3632 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3633 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3634 } else { // get by (name,timestamp)
3635 $file = wfFindFile( $title, $options );
3636 }
3637 return $file;
3638 }
3639
3640 /**
3641 * Transclude an interwiki link.
3642 *
3643 * @param Title $title
3644 * @param string $action
3645 *
3646 * @return string
3647 */
3648 public function interwikiTransclude( $title, $action ) {
3649 global $wgEnableScaryTranscluding;
3650
3651 if ( !$wgEnableScaryTranscluding ) {
3652 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3653 }
3654
3655 $url = $title->getFullURL( [ 'action' => $action ] );
3656
3657 if ( strlen( $url ) > 255 ) {
3658 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3659 }
3660 return $this->fetchScaryTemplateMaybeFromCache( $url );
3661 }
3662
3663 /**
3664 * @param string $url
3665 * @return mixed|string
3666 */
3667 public function fetchScaryTemplateMaybeFromCache( $url ) {
3668 global $wgTranscludeCacheExpiry;
3669 $dbr = wfGetDB( DB_SLAVE );
3670 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3671 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3672 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3673 if ( $obj ) {
3674 return $obj->tc_contents;
3675 }
3676
3677 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3678 $status = $req->execute(); // Status object
3679 if ( $status->isOK() ) {
3680 $text = $req->getContent();
3681 } elseif ( $req->getStatus() != 200 ) {
3682 // Though we failed to fetch the content, this status is useless.
3683 return wfMessage( 'scarytranscludefailed-httpstatus' )
3684 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3685 } else {
3686 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3687 }
3688
3689 $dbw = wfGetDB( DB_MASTER );
3690 $dbw->replace( 'transcache', [ 'tc_url' ], [
3691 'tc_url' => $url,
3692 'tc_time' => $dbw->timestamp( time() ),
3693 'tc_contents' => $text
3694 ] );
3695 return $text;
3696 }
3697
3698 /**
3699 * Triple brace replacement -- used for template arguments
3700 * @private
3701 *
3702 * @param array $piece
3703 * @param PPFrame $frame
3704 *
3705 * @return array
3706 */
3707 public function argSubstitution( $piece, $frame ) {
3708
3709 $error = false;
3710 $parts = $piece['parts'];
3711 $nameWithSpaces = $frame->expand( $piece['title'] );
3712 $argName = trim( $nameWithSpaces );
3713 $object = false;
3714 $text = $frame->getArgument( $argName );
3715 if ( $text === false && $parts->getLength() > 0
3716 && ( $this->ot['html']
3717 || $this->ot['pre']
3718 || ( $this->ot['wiki'] && $frame->isTemplate() )
3719 )
3720 ) {
3721 # No match in frame, use the supplied default
3722 $object = $parts->item( 0 )->getChildren();
3723 }
3724 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3725 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3726 $this->limitationWarn( 'post-expand-template-argument' );
3727 }
3728
3729 if ( $text === false && $object === false ) {
3730 # No match anywhere
3731 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3732 }
3733 if ( $error !== false ) {
3734 $text .= $error;
3735 }
3736 if ( $object !== false ) {
3737 $ret = [ 'object' => $object ];
3738 } else {
3739 $ret = [ 'text' => $text ];
3740 }
3741
3742 return $ret;
3743 }
3744
3745 /**
3746 * Return the text to be used for a given extension tag.
3747 * This is the ghost of strip().
3748 *
3749 * @param array $params Associative array of parameters:
3750 * name PPNode for the tag name
3751 * attr PPNode for unparsed text where tag attributes are thought to be
3752 * attributes Optional associative array of parsed attributes
3753 * inner Contents of extension element
3754 * noClose Original text did not have a close tag
3755 * @param PPFrame $frame
3756 *
3757 * @throws MWException
3758 * @return string
3759 */
3760 public function extensionSubstitution( $params, $frame ) {
3761 $name = $frame->expand( $params['name'] );
3762 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3763 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3764 $marker = self::MARKER_PREFIX . "-$name-"
3765 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3766
3767 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3768 ( $this->ot['html'] || $this->ot['pre'] );
3769 if ( $isFunctionTag ) {
3770 $markerType = 'none';
3771 } else {
3772 $markerType = 'general';
3773 }
3774 if ( $this->ot['html'] || $isFunctionTag ) {
3775 $name = strtolower( $name );
3776 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3777 if ( isset( $params['attributes'] ) ) {
3778 $attributes = $attributes + $params['attributes'];
3779 }
3780
3781 if ( isset( $this->mTagHooks[$name] ) ) {
3782 # Workaround for PHP bug 35229 and similar
3783 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3784 throw new MWException( "Tag hook for $name is not callable\n" );
3785 }
3786 $output = call_user_func_array( $this->mTagHooks[$name],
3787 [ $content, $attributes, $this, $frame ] );
3788 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3789 list( $callback, ) = $this->mFunctionTagHooks[$name];
3790 if ( !is_callable( $callback ) ) {
3791 throw new MWException( "Tag hook for $name is not callable\n" );
3792 }
3793
3794 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3795 } else {
3796 $output = '<span class="error">Invalid tag extension name: ' .
3797 htmlspecialchars( $name ) . '</span>';
3798 }
3799
3800 if ( is_array( $output ) ) {
3801 # Extract flags to local scope (to override $markerType)
3802 $flags = $output;
3803 $output = $flags[0];
3804 unset( $flags[0] );
3805 extract( $flags );
3806 }
3807 } else {
3808 if ( is_null( $attrText ) ) {
3809 $attrText = '';
3810 }
3811 if ( isset( $params['attributes'] ) ) {
3812 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3813 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3814 htmlspecialchars( $attrValue ) . '"';
3815 }
3816 }
3817 if ( $content === null ) {
3818 $output = "<$name$attrText/>";
3819 } else {
3820 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3821 $output = "<$name$attrText>$content$close";
3822 }
3823 }
3824
3825 if ( $markerType === 'none' ) {
3826 return $output;
3827 } elseif ( $markerType === 'nowiki' ) {
3828 $this->mStripState->addNoWiki( $marker, $output );
3829 } elseif ( $markerType === 'general' ) {
3830 $this->mStripState->addGeneral( $marker, $output );
3831 } else {
3832 throw new MWException( __METHOD__ . ': invalid marker type' );
3833 }
3834 return $marker;
3835 }
3836
3837 /**
3838 * Increment an include size counter
3839 *
3840 * @param string $type The type of expansion
3841 * @param int $size The size of the text
3842 * @return bool False if this inclusion would take it over the maximum, true otherwise
3843 */
3844 public function incrementIncludeSize( $type, $size ) {
3845 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3846 return false;
3847 } else {
3848 $this->mIncludeSizes[$type] += $size;
3849 return true;
3850 }
3851 }
3852
3853 /**
3854 * Increment the expensive function count
3855 *
3856 * @return bool False if the limit has been exceeded
3857 */
3858 public function incrementExpensiveFunctionCount() {
3859 $this->mExpensiveFunctionCount++;
3860 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3861 }
3862
3863 /**
3864 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3865 * Fills $this->mDoubleUnderscores, returns the modified text
3866 *
3867 * @param string $text
3868 *
3869 * @return string
3870 */
3871 public function doDoubleUnderscore( $text ) {
3872
3873 # The position of __TOC__ needs to be recorded
3874 $mw = MagicWord::get( 'toc' );
3875 if ( $mw->match( $text ) ) {
3876 $this->mShowToc = true;
3877 $this->mForceTocPosition = true;
3878
3879 # Set a placeholder. At the end we'll fill it in with the TOC.
3880 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3881
3882 # Only keep the first one.
3883 $text = $mw->replace( '', $text );
3884 }
3885
3886 # Now match and remove the rest of them
3887 $mwa = MagicWord::getDoubleUnderscoreArray();
3888 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3889
3890 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3891 $this->mOutput->mNoGallery = true;
3892 }
3893 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3894 $this->mShowToc = false;
3895 }
3896 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3897 && $this->mTitle->getNamespace() == NS_CATEGORY
3898 ) {
3899 $this->addTrackingCategory( 'hidden-category-category' );
3900 }
3901 # (bug 8068) Allow control over whether robots index a page.
3902 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3903 # is not desirable, the last one on the page should win.
3904 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3905 $this->mOutput->setIndexPolicy( 'noindex' );
3906 $this->addTrackingCategory( 'noindex-category' );
3907 }
3908 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3909 $this->mOutput->setIndexPolicy( 'index' );
3910 $this->addTrackingCategory( 'index-category' );
3911 }
3912
3913 # Cache all double underscores in the database
3914 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3915 $this->mOutput->setProperty( $key, '' );
3916 }
3917
3918 return $text;
3919 }
3920
3921 /**
3922 * @see ParserOutput::addTrackingCategory()
3923 * @param string $msg Message key
3924 * @return bool Whether the addition was successful
3925 */
3926 public function addTrackingCategory( $msg ) {
3927 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3928 }
3929
3930 /**
3931 * This function accomplishes several tasks:
3932 * 1) Auto-number headings if that option is enabled
3933 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3934 * 3) Add a Table of contents on the top for users who have enabled the option
3935 * 4) Auto-anchor headings
3936 *
3937 * It loops through all headlines, collects the necessary data, then splits up the
3938 * string and re-inserts the newly formatted headlines.
3939 *
3940 * @param string $text
3941 * @param string $origText Original, untouched wikitext
3942 * @param bool $isMain
3943 * @return mixed|string
3944 * @private
3945 */
3946 public function formatHeadings( $text, $origText, $isMain = true ) {
3947 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
3948
3949 # Inhibit editsection links if requested in the page
3950 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3951 $maybeShowEditLink = $showEditLink = false;
3952 } else {
3953 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3954 $showEditLink = $this->mOptions->getEditSection();
3955 }
3956 if ( $showEditLink ) {
3957 $this->mOutput->setEditSectionTokens( true );
3958 }
3959
3960 # Get all headlines for numbering them and adding funky stuff like [edit]
3961 # links - this is for later, but we need the number of headlines right now
3962 $matches = [];
3963 $numMatches = preg_match_all(
3964 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
3965 $text,
3966 $matches
3967 );
3968
3969 # if there are fewer than 4 headlines in the article, do not show TOC
3970 # unless it's been explicitly enabled.
3971 $enoughToc = $this->mShowToc &&
3972 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3973
3974 # Allow user to stipulate that a page should have a "new section"
3975 # link added via __NEWSECTIONLINK__
3976 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3977 $this->mOutput->setNewSection( true );
3978 }
3979
3980 # Allow user to remove the "new section"
3981 # link via __NONEWSECTIONLINK__
3982 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3983 $this->mOutput->hideNewSection( true );
3984 }
3985
3986 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3987 # override above conditions and always show TOC above first header
3988 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3989 $this->mShowToc = true;
3990 $enoughToc = true;
3991 }
3992
3993 # headline counter
3994 $headlineCount = 0;
3995 $numVisible = 0;
3996
3997 # Ugh .. the TOC should have neat indentation levels which can be
3998 # passed to the skin functions. These are determined here
3999 $toc = '';
4000 $full = '';
4001 $head = [];
4002 $sublevelCount = [];
4003 $levelCount = [];
4004 $level = 0;
4005 $prevlevel = 0;
4006 $toclevel = 0;
4007 $prevtoclevel = 0;
4008 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4009 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4010 $oldType = $this->mOutputType;
4011 $this->setOutputType( self::OT_WIKI );
4012 $frame = $this->getPreprocessor()->newFrame();
4013 $root = $this->preprocessToDom( $origText );
4014 $node = $root->getFirstChild();
4015 $byteOffset = 0;
4016 $tocraw = [];
4017 $refers = [];
4018
4019 $headlines = $numMatches !== false ? $matches[3] : [];
4020
4021 foreach ( $headlines as $headline ) {
4022 $isTemplate = false;
4023 $titleText = false;
4024 $sectionIndex = false;
4025 $numbering = '';
4026 $markerMatches = [];
4027 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4028 $serial = $markerMatches[1];
4029 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4030 $isTemplate = ( $titleText != $baseTitleText );
4031 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4032 }
4033
4034 if ( $toclevel ) {
4035 $prevlevel = $level;
4036 }
4037 $level = $matches[1][$headlineCount];
4038
4039 if ( $level > $prevlevel ) {
4040 # Increase TOC level
4041 $toclevel++;
4042 $sublevelCount[$toclevel] = 0;
4043 if ( $toclevel < $wgMaxTocLevel ) {
4044 $prevtoclevel = $toclevel;
4045 $toc .= Linker::tocIndent();
4046 $numVisible++;
4047 }
4048 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4049 # Decrease TOC level, find level to jump to
4050
4051 for ( $i = $toclevel; $i > 0; $i-- ) {
4052 if ( $levelCount[$i] == $level ) {
4053 # Found last matching level
4054 $toclevel = $i;
4055 break;
4056 } elseif ( $levelCount[$i] < $level ) {
4057 # Found first matching level below current level
4058 $toclevel = $i + 1;
4059 break;
4060 }
4061 }
4062 if ( $i == 0 ) {
4063 $toclevel = 1;
4064 }
4065 if ( $toclevel < $wgMaxTocLevel ) {
4066 if ( $prevtoclevel < $wgMaxTocLevel ) {
4067 # Unindent only if the previous toc level was shown :p
4068 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4069 $prevtoclevel = $toclevel;
4070 } else {
4071 $toc .= Linker::tocLineEnd();
4072 }
4073 }
4074 } else {
4075 # No change in level, end TOC line
4076 if ( $toclevel < $wgMaxTocLevel ) {
4077 $toc .= Linker::tocLineEnd();
4078 }
4079 }
4080
4081 $levelCount[$toclevel] = $level;
4082
4083 # count number of headlines for each level
4084 $sublevelCount[$toclevel]++;
4085 $dot = 0;
4086 for ( $i = 1; $i <= $toclevel; $i++ ) {
4087 if ( !empty( $sublevelCount[$i] ) ) {
4088 if ( $dot ) {
4089 $numbering .= '.';
4090 }
4091 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4092 $dot = 1;
4093 }
4094 }
4095
4096 # The safe header is a version of the header text safe to use for links
4097
4098 # Remove link placeholders by the link text.
4099 # <!--LINK number-->
4100 # turns into
4101 # link text with suffix
4102 # Do this before unstrip since link text can contain strip markers
4103 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4104
4105 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4106 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4107
4108 # Strip out HTML (first regex removes any tag not allowed)
4109 # Allowed tags are:
4110 # * <sup> and <sub> (bug 8393)
4111 # * <i> (bug 26375)
4112 # * <b> (r105284)
4113 # * <bdi> (bug 72884)
4114 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4115 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4116 # to allow setting directionality in toc items.
4117 $tocline = preg_replace(
4118 [
4119 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4120 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4121 ],
4122 [ '', '<$1>' ],
4123 $safeHeadline
4124 );
4125
4126 # Strip '<span></span>', which is the result from the above if
4127 # <span id="foo"></span> is used to produce an additional anchor
4128 # for a section.
4129 $tocline = str_replace( '<span></span>', '', $tocline );
4130
4131 $tocline = trim( $tocline );
4132
4133 # For the anchor, strip out HTML-y stuff period
4134 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4135 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4136
4137 # Save headline for section edit hint before it's escaped
4138 $headlineHint = $safeHeadline;
4139
4140 if ( $wgExperimentalHtmlIds ) {
4141 # For reverse compatibility, provide an id that's
4142 # HTML4-compatible, like we used to.
4143 # It may be worth noting, academically, that it's possible for
4144 # the legacy anchor to conflict with a non-legacy headline
4145 # anchor on the page. In this case likely the "correct" thing
4146 # would be to either drop the legacy anchors or make sure
4147 # they're numbered first. However, this would require people
4148 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4149 # manually, so let's not bother worrying about it.
4150 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4151 [ 'noninitial', 'legacy' ] );
4152 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4153
4154 if ( $legacyHeadline == $safeHeadline ) {
4155 # No reason to have both (in fact, we can't)
4156 $legacyHeadline = false;
4157 }
4158 } else {
4159 $legacyHeadline = false;
4160 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4161 'noninitial' );
4162 }
4163
4164 # HTML names must be case-insensitively unique (bug 10721).
4165 # This does not apply to Unicode characters per
4166 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4167 # @todo FIXME: We may be changing them depending on the current locale.
4168 $arrayKey = strtolower( $safeHeadline );
4169 if ( $legacyHeadline === false ) {
4170 $legacyArrayKey = false;
4171 } else {
4172 $legacyArrayKey = strtolower( $legacyHeadline );
4173 }
4174
4175 # Create the anchor for linking from the TOC to the section
4176 $anchor = $safeHeadline;
4177 $legacyAnchor = $legacyHeadline;
4178 if ( isset( $refers[$arrayKey] ) ) {
4179 // @codingStandardsIgnoreStart
4180 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4181 // @codingStandardsIgnoreEnd
4182 $anchor .= "_$i";
4183 $refers["${arrayKey}_$i"] = true;
4184 } else {
4185 $refers[$arrayKey] = true;
4186 }
4187 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4188 // @codingStandardsIgnoreStart
4189 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4190 // @codingStandardsIgnoreEnd
4191 $legacyAnchor .= "_$i";
4192 $refers["${legacyArrayKey}_$i"] = true;
4193 } else {
4194 $refers[$legacyArrayKey] = true;
4195 }
4196
4197 # Don't number the heading if it is the only one (looks silly)
4198 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4199 # the two are different if the line contains a link
4200 $headline = Html::element(
4201 'span',
4202 [ 'class' => 'mw-headline-number' ],
4203 $numbering
4204 ) . ' ' . $headline;
4205 }
4206
4207 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4208 $toc .= Linker::tocLine( $anchor, $tocline,
4209 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4210 }
4211
4212 # Add the section to the section tree
4213 # Find the DOM node for this header
4214 $noOffset = ( $isTemplate || $sectionIndex === false );
4215 while ( $node && !$noOffset ) {
4216 if ( $node->getName() === 'h' ) {
4217 $bits = $node->splitHeading();
4218 if ( $bits['i'] == $sectionIndex ) {
4219 break;
4220 }
4221 }
4222 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4223 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4224 $node = $node->getNextSibling();
4225 }
4226 $tocraw[] = [
4227 'toclevel' => $toclevel,
4228 'level' => $level,
4229 'line' => $tocline,
4230 'number' => $numbering,
4231 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4232 'fromtitle' => $titleText,
4233 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4234 'anchor' => $anchor,
4235 ];
4236
4237 # give headline the correct <h#> tag
4238 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4239 // Output edit section links as markers with styles that can be customized by skins
4240 if ( $isTemplate ) {
4241 # Put a T flag in the section identifier, to indicate to extractSections()
4242 # that sections inside <includeonly> should be counted.
4243 $editsectionPage = $titleText;
4244 $editsectionSection = "T-$sectionIndex";
4245 $editsectionContent = null;
4246 } else {
4247 $editsectionPage = $this->mTitle->getPrefixedText();
4248 $editsectionSection = $sectionIndex;
4249 $editsectionContent = $headlineHint;
4250 }
4251 // We use a bit of pesudo-xml for editsection markers. The
4252 // language converter is run later on. Using a UNIQ style marker
4253 // leads to the converter screwing up the tokens when it
4254 // converts stuff. And trying to insert strip tags fails too. At
4255 // this point all real inputted tags have already been escaped,
4256 // so we don't have to worry about a user trying to input one of
4257 // these markers directly. We use a page and section attribute
4258 // to stop the language converter from converting these
4259 // important bits of data, but put the headline hint inside a
4260 // content block because the language converter is supposed to
4261 // be able to convert that piece of data.
4262 // Gets replaced with html in ParserOutput::getText
4263 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4264 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4265 if ( $editsectionContent !== null ) {
4266 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4267 } else {
4268 $editlink .= '/>';
4269 }
4270 } else {
4271 $editlink = '';
4272 }
4273 $head[$headlineCount] = Linker::makeHeadline( $level,
4274 $matches['attrib'][$headlineCount], $anchor, $headline,
4275 $editlink, $legacyAnchor );
4276
4277 $headlineCount++;
4278 }
4279
4280 $this->setOutputType( $oldType );
4281
4282 # Never ever show TOC if no headers
4283 if ( $numVisible < 1 ) {
4284 $enoughToc = false;
4285 }
4286
4287 if ( $enoughToc ) {
4288 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4289 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4290 }
4291 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4292 $this->mOutput->setTOCHTML( $toc );
4293 $toc = self::TOC_START . $toc . self::TOC_END;
4294 $this->mOutput->addModules( 'mediawiki.toc' );
4295 }
4296
4297 if ( $isMain ) {
4298 $this->mOutput->setSections( $tocraw );
4299 }
4300
4301 # split up and insert constructed headlines
4302 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4303 $i = 0;
4304
4305 // build an array of document sections
4306 $sections = [];
4307 foreach ( $blocks as $block ) {
4308 // $head is zero-based, sections aren't.
4309 if ( empty( $head[$i - 1] ) ) {
4310 $sections[$i] = $block;
4311 } else {
4312 $sections[$i] = $head[$i - 1] . $block;
4313 }
4314
4315 /**
4316 * Send a hook, one per section.
4317 * The idea here is to be able to make section-level DIVs, but to do so in a
4318 * lower-impact, more correct way than r50769
4319 *
4320 * $this : caller
4321 * $section : the section number
4322 * &$sectionContent : ref to the content of the section
4323 * $showEditLinks : boolean describing whether this section has an edit link
4324 */
4325 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4326
4327 $i++;
4328 }
4329
4330 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4331 // append the TOC at the beginning
4332 // Top anchor now in skin
4333 $sections[0] = $sections[0] . $toc . "\n";
4334 }
4335
4336 $full .= implode( '', $sections );
4337
4338 if ( $this->mForceTocPosition ) {
4339 return str_replace( '<!--MWTOC-->', $toc, $full );
4340 } else {
4341 return $full;
4342 }
4343 }
4344
4345 /**
4346 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4347 * conversion, substituting signatures, {{subst:}} templates, etc.
4348 *
4349 * @param string $text The text to transform
4350 * @param Title $title The Title object for the current article
4351 * @param User $user The User object describing the current user
4352 * @param ParserOptions $options Parsing options
4353 * @param bool $clearState Whether to clear the parser state first
4354 * @return string The altered wiki markup
4355 */
4356 public function preSaveTransform( $text, Title $title, User $user,
4357 ParserOptions $options, $clearState = true
4358 ) {
4359 if ( $clearState ) {
4360 $magicScopeVariable = $this->lock();
4361 }
4362 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4363 $this->setUser( $user );
4364
4365 $pairs = [
4366 "\r\n" => "\n",
4367 "\r" => "\n",
4368 ];
4369 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4370 if ( $options->getPreSaveTransform() ) {
4371 $text = $this->pstPass2( $text, $user );
4372 }
4373 $text = $this->mStripState->unstripBoth( $text );
4374
4375 $this->setUser( null ); # Reset
4376
4377 return $text;
4378 }
4379
4380 /**
4381 * Pre-save transform helper function
4382 *
4383 * @param string $text
4384 * @param User $user
4385 *
4386 * @return string
4387 */
4388 private function pstPass2( $text, $user ) {
4389 global $wgContLang;
4390
4391 # Note: This is the timestamp saved as hardcoded wikitext to
4392 # the database, we use $wgContLang here in order to give
4393 # everyone the same signature and use the default one rather
4394 # than the one selected in each user's preferences.
4395 # (see also bug 12815)
4396 $ts = $this->mOptions->getTimestamp();
4397 $timestamp = MWTimestamp::getLocalInstance( $ts );
4398 $ts = $timestamp->format( 'YmdHis' );
4399 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4400
4401 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4402
4403 # Variable replacement
4404 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4405 $text = $this->replaceVariables( $text );
4406
4407 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4408 # which may corrupt this parser instance via its wfMessage()->text() call-
4409
4410 # Signatures
4411 $sigText = $this->getUserSig( $user );
4412 $text = strtr( $text, [
4413 '~~~~~' => $d,
4414 '~~~~' => "$sigText $d",
4415 '~~~' => $sigText
4416 ] );
4417
4418 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4419 $tc = '[' . Title::legalChars() . ']';
4420 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4421
4422 // [[ns:page (context)|]]
4423 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4424 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4425 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4426 // [[ns:page (context), context|]] (using either single or double-width comma)
4427 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4428 // [[|page]] (reverse pipe trick: add context from page title)
4429 $p2 = "/\[\[\\|($tc+)]]/";
4430
4431 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4432 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4433 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4434 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4435
4436 $t = $this->mTitle->getText();
4437 $m = [];
4438 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4439 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4440 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4441 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4442 } else {
4443 # if there's no context, don't bother duplicating the title
4444 $text = preg_replace( $p2, '[[\\1]]', $text );
4445 }
4446
4447 # Trim trailing whitespace
4448 $text = rtrim( $text );
4449
4450 return $text;
4451 }
4452
4453 /**
4454 * Fetch the user's signature text, if any, and normalize to
4455 * validated, ready-to-insert wikitext.
4456 * If you have pre-fetched the nickname or the fancySig option, you can
4457 * specify them here to save a database query.
4458 * Do not reuse this parser instance after calling getUserSig(),
4459 * as it may have changed if it's the $wgParser.
4460 *
4461 * @param User $user
4462 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4463 * @param bool|null $fancySig whether the nicknname is the complete signature
4464 * or null to use default value
4465 * @return string
4466 */
4467 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4468 global $wgMaxSigChars;
4469
4470 $username = $user->getName();
4471
4472 # If not given, retrieve from the user object.
4473 if ( $nickname === false ) {
4474 $nickname = $user->getOption( 'nickname' );
4475 }
4476
4477 if ( is_null( $fancySig ) ) {
4478 $fancySig = $user->getBoolOption( 'fancysig' );
4479 }
4480
4481 $nickname = $nickname == null ? $username : $nickname;
4482
4483 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4484 $nickname = $username;
4485 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4486 } elseif ( $fancySig !== false ) {
4487 # Sig. might contain markup; validate this
4488 if ( $this->validateSig( $nickname ) !== false ) {
4489 # Validated; clean up (if needed) and return it
4490 return $this->cleanSig( $nickname, true );
4491 } else {
4492 # Failed to validate; fall back to the default
4493 $nickname = $username;
4494 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4495 }
4496 }
4497
4498 # Make sure nickname doesnt get a sig in a sig
4499 $nickname = self::cleanSigInSig( $nickname );
4500
4501 # If we're still here, make it a link to the user page
4502 $userText = wfEscapeWikiText( $username );
4503 $nickText = wfEscapeWikiText( $nickname );
4504 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4505
4506 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4507 ->title( $this->getTitle() )->text();
4508 }
4509
4510 /**
4511 * Check that the user's signature contains no bad XML
4512 *
4513 * @param string $text
4514 * @return string|bool An expanded string, or false if invalid.
4515 */
4516 public function validateSig( $text ) {
4517 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4518 }
4519
4520 /**
4521 * Clean up signature text
4522 *
4523 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4524 * 2) Substitute all transclusions
4525 *
4526 * @param string $text
4527 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4528 * @return string Signature text
4529 */
4530 public function cleanSig( $text, $parsing = false ) {
4531 if ( !$parsing ) {
4532 global $wgTitle;
4533 $magicScopeVariable = $this->lock();
4534 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4535 }
4536
4537 # Option to disable this feature
4538 if ( !$this->mOptions->getCleanSignatures() ) {
4539 return $text;
4540 }
4541
4542 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4543 # => Move this logic to braceSubstitution()
4544 $substWord = MagicWord::get( 'subst' );
4545 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4546 $substText = '{{' . $substWord->getSynonym( 0 );
4547
4548 $text = preg_replace( $substRegex, $substText, $text );
4549 $text = self::cleanSigInSig( $text );
4550 $dom = $this->preprocessToDom( $text );
4551 $frame = $this->getPreprocessor()->newFrame();
4552 $text = $frame->expand( $dom );
4553
4554 if ( !$parsing ) {
4555 $text = $this->mStripState->unstripBoth( $text );
4556 }
4557
4558 return $text;
4559 }
4560
4561 /**
4562 * Strip 3, 4 or 5 tildes out of signatures.
4563 *
4564 * @param string $text
4565 * @return string Signature text with /~{3,5}/ removed
4566 */
4567 public static function cleanSigInSig( $text ) {
4568 $text = preg_replace( '/~{3,5}/', '', $text );
4569 return $text;
4570 }
4571
4572 /**
4573 * Set up some variables which are usually set up in parse()
4574 * so that an external function can call some class members with confidence
4575 *
4576 * @param Title|null $title
4577 * @param ParserOptions $options
4578 * @param int $outputType
4579 * @param bool $clearState
4580 */
4581 public function startExternalParse( Title $title = null, ParserOptions $options,
4582 $outputType, $clearState = true
4583 ) {
4584 $this->startParse( $title, $options, $outputType, $clearState );
4585 }
4586
4587 /**
4588 * @param Title|null $title
4589 * @param ParserOptions $options
4590 * @param int $outputType
4591 * @param bool $clearState
4592 */
4593 private function startParse( Title $title = null, ParserOptions $options,
4594 $outputType, $clearState = true
4595 ) {
4596 $this->setTitle( $title );
4597 $this->mOptions = $options;
4598 $this->setOutputType( $outputType );
4599 if ( $clearState ) {
4600 $this->clearState();
4601 }
4602 }
4603
4604 /**
4605 * Wrapper for preprocess()
4606 *
4607 * @param string $text The text to preprocess
4608 * @param ParserOptions $options Options
4609 * @param Title|null $title Title object or null to use $wgTitle
4610 * @return string
4611 */
4612 public function transformMsg( $text, $options, $title = null ) {
4613 static $executing = false;
4614
4615 # Guard against infinite recursion
4616 if ( $executing ) {
4617 return $text;
4618 }
4619 $executing = true;
4620
4621 if ( !$title ) {
4622 global $wgTitle;
4623 $title = $wgTitle;
4624 }
4625
4626 $text = $this->preprocess( $text, $title, $options );
4627
4628 $executing = false;
4629 return $text;
4630 }
4631
4632 /**
4633 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4634 * The callback should have the following form:
4635 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4636 *
4637 * Transform and return $text. Use $parser for any required context, e.g. use
4638 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4639 *
4640 * Hooks may return extended information by returning an array, of which the
4641 * first numbered element (index 0) must be the return string, and all other
4642 * entries are extracted into local variables within an internal function
4643 * in the Parser class.
4644 *
4645 * This interface (introduced r61913) appears to be undocumented, but
4646 * 'markerType' is used by some core tag hooks to override which strip
4647 * array their results are placed in. **Use great caution if attempting
4648 * this interface, as it is not documented and injudicious use could smash
4649 * private variables.**
4650 *
4651 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4652 * @param callable $callback The callback function (and object) to use for the tag
4653 * @throws MWException
4654 * @return callable|null The old value of the mTagHooks array associated with the hook
4655 */
4656 public function setHook( $tag, $callback ) {
4657 $tag = strtolower( $tag );
4658 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4659 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4660 }
4661 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4662 $this->mTagHooks[$tag] = $callback;
4663 if ( !in_array( $tag, $this->mStripList ) ) {
4664 $this->mStripList[] = $tag;
4665 }
4666
4667 return $oldVal;
4668 }
4669
4670 /**
4671 * As setHook(), but letting the contents be parsed.
4672 *
4673 * Transparent tag hooks are like regular XML-style tag hooks, except they
4674 * operate late in the transformation sequence, on HTML instead of wikitext.
4675 *
4676 * This is probably obsoleted by things dealing with parser frames?
4677 * The only extension currently using it is geoserver.
4678 *
4679 * @since 1.10
4680 * @todo better document or deprecate this
4681 *
4682 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4683 * @param callable $callback The callback function (and object) to use for the tag
4684 * @throws MWException
4685 * @return callable|null The old value of the mTagHooks array associated with the hook
4686 */
4687 public function setTransparentTagHook( $tag, $callback ) {
4688 $tag = strtolower( $tag );
4689 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4690 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4691 }
4692 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4693 $this->mTransparentTagHooks[$tag] = $callback;
4694
4695 return $oldVal;
4696 }
4697
4698 /**
4699 * Remove all tag hooks
4700 */
4701 public function clearTagHooks() {
4702 $this->mTagHooks = [];
4703 $this->mFunctionTagHooks = [];
4704 $this->mStripList = $this->mDefaultStripList;
4705 }
4706
4707 /**
4708 * Create a function, e.g. {{sum:1|2|3}}
4709 * The callback function should have the form:
4710 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4711 *
4712 * Or with Parser::SFH_OBJECT_ARGS:
4713 * function myParserFunction( $parser, $frame, $args ) { ... }
4714 *
4715 * The callback may either return the text result of the function, or an array with the text
4716 * in element 0, and a number of flags in the other elements. The names of the flags are
4717 * specified in the keys. Valid flags are:
4718 * found The text returned is valid, stop processing the template. This
4719 * is on by default.
4720 * nowiki Wiki markup in the return value should be escaped
4721 * isHTML The returned text is HTML, armour it against wikitext transformation
4722 *
4723 * @param string $id The magic word ID
4724 * @param callable $callback The callback function (and object) to use
4725 * @param int $flags A combination of the following flags:
4726 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4727 *
4728 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4729 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4730 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4731 * the arguments, and to control the way they are expanded.
4732 *
4733 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4734 * arguments, for instance:
4735 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4736 *
4737 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4738 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4739 * working if/when this is changed.
4740 *
4741 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4742 * expansion.
4743 *
4744 * Please read the documentation in includes/parser/Preprocessor.php for more information
4745 * about the methods available in PPFrame and PPNode.
4746 *
4747 * @throws MWException
4748 * @return string|callable The old callback function for this name, if any
4749 */
4750 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4751 global $wgContLang;
4752
4753 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4754 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4755
4756 # Add to function cache
4757 $mw = MagicWord::get( $id );
4758 if ( !$mw ) {
4759 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4760 }
4761
4762 $synonyms = $mw->getSynonyms();
4763 $sensitive = intval( $mw->isCaseSensitive() );
4764
4765 foreach ( $synonyms as $syn ) {
4766 # Case
4767 if ( !$sensitive ) {
4768 $syn = $wgContLang->lc( $syn );
4769 }
4770 # Add leading hash
4771 if ( !( $flags & self::SFH_NO_HASH ) ) {
4772 $syn = '#' . $syn;
4773 }
4774 # Remove trailing colon
4775 if ( substr( $syn, -1, 1 ) === ':' ) {
4776 $syn = substr( $syn, 0, -1 );
4777 }
4778 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4779 }
4780 return $oldVal;
4781 }
4782
4783 /**
4784 * Get all registered function hook identifiers
4785 *
4786 * @return array
4787 */
4788 public function getFunctionHooks() {
4789 return array_keys( $this->mFunctionHooks );
4790 }
4791
4792 /**
4793 * Create a tag function, e.g. "<test>some stuff</test>".
4794 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4795 * Unlike parser functions, their content is not preprocessed.
4796 * @param string $tag
4797 * @param callable $callback
4798 * @param int $flags
4799 * @throws MWException
4800 * @return null
4801 */
4802 public function setFunctionTagHook( $tag, $callback, $flags ) {
4803 $tag = strtolower( $tag );
4804 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4805 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4806 }
4807 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4808 $this->mFunctionTagHooks[$tag] : null;
4809 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4810
4811 if ( !in_array( $tag, $this->mStripList ) ) {
4812 $this->mStripList[] = $tag;
4813 }
4814
4815 return $old;
4816 }
4817
4818 /**
4819 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4820 * Placeholders created in Linker::link()
4821 *
4822 * @param string $text
4823 * @param int $options
4824 */
4825 public function replaceLinkHolders( &$text, $options = 0 ) {
4826 $this->mLinkHolders->replace( $text );
4827 }
4828
4829 /**
4830 * Replace "<!--LINK-->" link placeholders with plain text of links
4831 * (not HTML-formatted).
4832 *
4833 * @param string $text
4834 * @return string
4835 */
4836 public function replaceLinkHoldersText( $text ) {
4837 return $this->mLinkHolders->replaceText( $text );
4838 }
4839
4840 /**
4841 * Renders an image gallery from a text with one line per image.
4842 * text labels may be given by using |-style alternative text. E.g.
4843 * Image:one.jpg|The number "1"
4844 * Image:tree.jpg|A tree
4845 * given as text will return the HTML of a gallery with two images,
4846 * labeled 'The number "1"' and
4847 * 'A tree'.
4848 *
4849 * @param string $text
4850 * @param array $params
4851 * @return string HTML
4852 */
4853 public function renderImageGallery( $text, $params ) {
4854
4855 $mode = false;
4856 if ( isset( $params['mode'] ) ) {
4857 $mode = $params['mode'];
4858 }
4859
4860 try {
4861 $ig = ImageGalleryBase::factory( $mode );
4862 } catch ( Exception $e ) {
4863 // If invalid type set, fallback to default.
4864 $ig = ImageGalleryBase::factory( false );
4865 }
4866
4867 $ig->setContextTitle( $this->mTitle );
4868 $ig->setShowBytes( false );
4869 $ig->setShowFilename( false );
4870 $ig->setParser( $this );
4871 $ig->setHideBadImages();
4872 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4873
4874 if ( isset( $params['showfilename'] ) ) {
4875 $ig->setShowFilename( true );
4876 } else {
4877 $ig->setShowFilename( false );
4878 }
4879 if ( isset( $params['caption'] ) ) {
4880 $caption = $params['caption'];
4881 $caption = htmlspecialchars( $caption );
4882 $caption = $this->replaceInternalLinks( $caption );
4883 $ig->setCaptionHtml( $caption );
4884 }
4885 if ( isset( $params['perrow'] ) ) {
4886 $ig->setPerRow( $params['perrow'] );
4887 }
4888 if ( isset( $params['widths'] ) ) {
4889 $ig->setWidths( $params['widths'] );
4890 }
4891 if ( isset( $params['heights'] ) ) {
4892 $ig->setHeights( $params['heights'] );
4893 }
4894 $ig->setAdditionalOptions( $params );
4895
4896 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4897
4898 $lines = StringUtils::explode( "\n", $text );
4899 foreach ( $lines as $line ) {
4900 # match lines like these:
4901 # Image:someimage.jpg|This is some image
4902 $matches = [];
4903 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4904 # Skip empty lines
4905 if ( count( $matches ) == 0 ) {
4906 continue;
4907 }
4908
4909 if ( strpos( $matches[0], '%' ) !== false ) {
4910 $matches[1] = rawurldecode( $matches[1] );
4911 }
4912 $title = Title::newFromText( $matches[1], NS_FILE );
4913 if ( is_null( $title ) ) {
4914 # Bogus title. Ignore these so we don't bomb out later.
4915 continue;
4916 }
4917
4918 # We need to get what handler the file uses, to figure out parameters.
4919 # Note, a hook can overide the file name, and chose an entirely different
4920 # file (which potentially could be of a different type and have different handler).
4921 $options = [];
4922 $descQuery = false;
4923 Hooks::run( 'BeforeParserFetchFileAndTitle',
4924 [ $this, $title, &$options, &$descQuery ] );
4925 # Don't register it now, as ImageGallery does that later.
4926 $file = $this->fetchFileNoRegister( $title, $options );
4927 $handler = $file ? $file->getHandler() : false;
4928
4929 $paramMap = [
4930 'img_alt' => 'gallery-internal-alt',
4931 'img_link' => 'gallery-internal-link',
4932 ];
4933 if ( $handler ) {
4934 $paramMap = $paramMap + $handler->getParamMap();
4935 // We don't want people to specify per-image widths.
4936 // Additionally the width parameter would need special casing anyhow.
4937 unset( $paramMap['img_width'] );
4938 }
4939
4940 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4941
4942 $label = '';
4943 $alt = '';
4944 $link = '';
4945 $handlerOptions = [];
4946 if ( isset( $matches[3] ) ) {
4947 // look for an |alt= definition while trying not to break existing
4948 // captions with multiple pipes (|) in it, until a more sensible grammar
4949 // is defined for images in galleries
4950
4951 // FIXME: Doing recursiveTagParse at this stage, and the trim before
4952 // splitting on '|' is a bit odd, and different from makeImage.
4953 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4954 $parameterMatches = StringUtils::explode( '|', $matches[3] );
4955
4956 foreach ( $parameterMatches as $parameterMatch ) {
4957 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
4958 if ( $magicName ) {
4959 $paramName = $paramMap[$magicName];
4960
4961 switch ( $paramName ) {
4962 case 'gallery-internal-alt':
4963 $alt = $this->stripAltText( $match, false );
4964 break;
4965 case 'gallery-internal-link':
4966 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
4967 $chars = self::EXT_LINK_URL_CLASS;
4968 $addr = self::EXT_LINK_ADDR;
4969 $prots = $this->mUrlProtocols;
4970 // check to see if link matches an absolute url, if not then it must be a wiki link.
4971 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
4972 $link = $linkValue;
4973 } else {
4974 $localLinkTitle = Title::newFromText( $linkValue );
4975 if ( $localLinkTitle !== null ) {
4976 $link = $localLinkTitle->getLinkURL();
4977 }
4978 }
4979 break;
4980 default:
4981 // Must be a handler specific parameter.
4982 if ( $handler->validateParam( $paramName, $match ) ) {
4983 $handlerOptions[$paramName] = $match;
4984 } else {
4985 // Guess not, consider it as caption.
4986 wfDebug( "$parameterMatch failed parameter validation\n" );
4987 $label = '|' . $parameterMatch;
4988 }
4989 }
4990
4991 } else {
4992 // Last pipe wins.
4993 $label = '|' . $parameterMatch;
4994 }
4995 }
4996 // Remove the pipe.
4997 $label = substr( $label, 1 );
4998 }
4999
5000 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5001 }
5002 $html = $ig->toHTML();
5003 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5004 return $html;
5005 }
5006
5007 /**
5008 * @param MediaHandler $handler
5009 * @return array
5010 */
5011 public function getImageParams( $handler ) {
5012 if ( $handler ) {
5013 $handlerClass = get_class( $handler );
5014 } else {
5015 $handlerClass = '';
5016 }
5017 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5018 # Initialise static lists
5019 static $internalParamNames = [
5020 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5021 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5022 'bottom', 'text-bottom' ],
5023 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5024 'upright', 'border', 'link', 'alt', 'class' ],
5025 ];
5026 static $internalParamMap;
5027 if ( !$internalParamMap ) {
5028 $internalParamMap = [];
5029 foreach ( $internalParamNames as $type => $names ) {
5030 foreach ( $names as $name ) {
5031 $magicName = str_replace( '-', '_', "img_$name" );
5032 $internalParamMap[$magicName] = [ $type, $name ];
5033 }
5034 }
5035 }
5036
5037 # Add handler params
5038 $paramMap = $internalParamMap;
5039 if ( $handler ) {
5040 $handlerParamMap = $handler->getParamMap();
5041 foreach ( $handlerParamMap as $magic => $paramName ) {
5042 $paramMap[$magic] = [ 'handler', $paramName ];
5043 }
5044 }
5045 $this->mImageParams[$handlerClass] = $paramMap;
5046 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5047 }
5048 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5049 }
5050
5051 /**
5052 * Parse image options text and use it to make an image
5053 *
5054 * @param Title $title
5055 * @param string $options
5056 * @param LinkHolderArray|bool $holders
5057 * @return string HTML
5058 */
5059 public function makeImage( $title, $options, $holders = false ) {
5060 # Check if the options text is of the form "options|alt text"
5061 # Options are:
5062 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5063 # * left no resizing, just left align. label is used for alt= only
5064 # * right same, but right aligned
5065 # * none same, but not aligned
5066 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5067 # * center center the image
5068 # * frame Keep original image size, no magnify-button.
5069 # * framed Same as "frame"
5070 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5071 # * upright reduce width for upright images, rounded to full __0 px
5072 # * border draw a 1px border around the image
5073 # * alt Text for HTML alt attribute (defaults to empty)
5074 # * class Set a class for img node
5075 # * link Set the target of the image link. Can be external, interwiki, or local
5076 # vertical-align values (no % or length right now):
5077 # * baseline
5078 # * sub
5079 # * super
5080 # * top
5081 # * text-top
5082 # * middle
5083 # * bottom
5084 # * text-bottom
5085
5086 $parts = StringUtils::explode( "|", $options );
5087
5088 # Give extensions a chance to select the file revision for us
5089 $options = [];
5090 $descQuery = false;
5091 Hooks::run( 'BeforeParserFetchFileAndTitle',
5092 [ $this, $title, &$options, &$descQuery ] );
5093 # Fetch and register the file (file title may be different via hooks)
5094 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5095
5096 # Get parameter map
5097 $handler = $file ? $file->getHandler() : false;
5098
5099 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5100
5101 if ( !$file ) {
5102 $this->addTrackingCategory( 'broken-file-category' );
5103 }
5104
5105 # Process the input parameters
5106 $caption = '';
5107 $params = [ 'frame' => [], 'handler' => [],
5108 'horizAlign' => [], 'vertAlign' => [] ];
5109 $seenformat = false;
5110 foreach ( $parts as $part ) {
5111 $part = trim( $part );
5112 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5113 $validated = false;
5114 if ( isset( $paramMap[$magicName] ) ) {
5115 list( $type, $paramName ) = $paramMap[$magicName];
5116
5117 # Special case; width and height come in one variable together
5118 if ( $type === 'handler' && $paramName === 'width' ) {
5119 $parsedWidthParam = $this->parseWidthParam( $value );
5120 if ( isset( $parsedWidthParam['width'] ) ) {
5121 $width = $parsedWidthParam['width'];
5122 if ( $handler->validateParam( 'width', $width ) ) {
5123 $params[$type]['width'] = $width;
5124 $validated = true;
5125 }
5126 }
5127 if ( isset( $parsedWidthParam['height'] ) ) {
5128 $height = $parsedWidthParam['height'];
5129 if ( $handler->validateParam( 'height', $height ) ) {
5130 $params[$type]['height'] = $height;
5131 $validated = true;
5132 }
5133 }
5134 # else no validation -- bug 13436
5135 } else {
5136 if ( $type === 'handler' ) {
5137 # Validate handler parameter
5138 $validated = $handler->validateParam( $paramName, $value );
5139 } else {
5140 # Validate internal parameters
5141 switch ( $paramName ) {
5142 case 'manualthumb':
5143 case 'alt':
5144 case 'class':
5145 # @todo FIXME: Possibly check validity here for
5146 # manualthumb? downstream behavior seems odd with
5147 # missing manual thumbs.
5148 $validated = true;
5149 $value = $this->stripAltText( $value, $holders );
5150 break;
5151 case 'link':
5152 $chars = self::EXT_LINK_URL_CLASS;
5153 $addr = self::EXT_LINK_ADDR;
5154 $prots = $this->mUrlProtocols;
5155 if ( $value === '' ) {
5156 $paramName = 'no-link';
5157 $value = true;
5158 $validated = true;
5159 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5160 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5161 $paramName = 'link-url';
5162 $this->mOutput->addExternalLink( $value );
5163 if ( $this->mOptions->getExternalLinkTarget() ) {
5164 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5165 }
5166 $validated = true;
5167 }
5168 } else {
5169 $linkTitle = Title::newFromText( $value );
5170 if ( $linkTitle ) {
5171 $paramName = 'link-title';
5172 $value = $linkTitle;
5173 $this->mOutput->addLink( $linkTitle );
5174 $validated = true;
5175 }
5176 }
5177 break;
5178 case 'frameless':
5179 case 'framed':
5180 case 'thumbnail':
5181 // use first appearing option, discard others.
5182 $validated = ! $seenformat;
5183 $seenformat = true;
5184 break;
5185 default:
5186 # Most other things appear to be empty or numeric...
5187 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5188 }
5189 }
5190
5191 if ( $validated ) {
5192 $params[$type][$paramName] = $value;
5193 }
5194 }
5195 }
5196 if ( !$validated ) {
5197 $caption = $part;
5198 }
5199 }
5200
5201 # Process alignment parameters
5202 if ( $params['horizAlign'] ) {
5203 $params['frame']['align'] = key( $params['horizAlign'] );
5204 }
5205 if ( $params['vertAlign'] ) {
5206 $params['frame']['valign'] = key( $params['vertAlign'] );
5207 }
5208
5209 $params['frame']['caption'] = $caption;
5210
5211 # Will the image be presented in a frame, with the caption below?
5212 $imageIsFramed = isset( $params['frame']['frame'] )
5213 || isset( $params['frame']['framed'] )
5214 || isset( $params['frame']['thumbnail'] )
5215 || isset( $params['frame']['manualthumb'] );
5216
5217 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5218 # came to also set the caption, ordinary text after the image -- which
5219 # makes no sense, because that just repeats the text multiple times in
5220 # screen readers. It *also* came to set the title attribute.
5221 # Now that we have an alt attribute, we should not set the alt text to
5222 # equal the caption: that's worse than useless, it just repeats the
5223 # text. This is the framed/thumbnail case. If there's no caption, we
5224 # use the unnamed parameter for alt text as well, just for the time be-
5225 # ing, if the unnamed param is set and the alt param is not.
5226 # For the future, we need to figure out if we want to tweak this more,
5227 # e.g., introducing a title= parameter for the title; ignoring the un-
5228 # named parameter entirely for images without a caption; adding an ex-
5229 # plicit caption= parameter and preserving the old magic unnamed para-
5230 # meter for BC; ...
5231 if ( $imageIsFramed ) { # Framed image
5232 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5233 # No caption or alt text, add the filename as the alt text so
5234 # that screen readers at least get some description of the image
5235 $params['frame']['alt'] = $title->getText();
5236 }
5237 # Do not set $params['frame']['title'] because tooltips don't make sense
5238 # for framed images
5239 } else { # Inline image
5240 if ( !isset( $params['frame']['alt'] ) ) {
5241 # No alt text, use the "caption" for the alt text
5242 if ( $caption !== '' ) {
5243 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5244 } else {
5245 # No caption, fall back to using the filename for the
5246 # alt text
5247 $params['frame']['alt'] = $title->getText();
5248 }
5249 }
5250 # Use the "caption" for the tooltip text
5251 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5252 }
5253
5254 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5255
5256 # Linker does the rest
5257 $time = isset( $options['time'] ) ? $options['time'] : false;
5258 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5259 $time, $descQuery, $this->mOptions->getThumbSize() );
5260
5261 # Give the handler a chance to modify the parser object
5262 if ( $handler ) {
5263 $handler->parserTransformHook( $this, $file );
5264 }
5265
5266 return $ret;
5267 }
5268
5269 /**
5270 * @param string $caption
5271 * @param LinkHolderArray|bool $holders
5272 * @return mixed|string
5273 */
5274 protected function stripAltText( $caption, $holders ) {
5275 # Strip bad stuff out of the title (tooltip). We can't just use
5276 # replaceLinkHoldersText() here, because if this function is called
5277 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5278 if ( $holders ) {
5279 $tooltip = $holders->replaceText( $caption );
5280 } else {
5281 $tooltip = $this->replaceLinkHoldersText( $caption );
5282 }
5283
5284 # make sure there are no placeholders in thumbnail attributes
5285 # that are later expanded to html- so expand them now and
5286 # remove the tags
5287 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5288 $tooltip = Sanitizer::stripAllTags( $tooltip );
5289
5290 return $tooltip;
5291 }
5292
5293 /**
5294 * Set a flag in the output object indicating that the content is dynamic and
5295 * shouldn't be cached.
5296 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5297 */
5298 public function disableCache() {
5299 wfDebug( "Parser output marked as uncacheable.\n" );
5300 if ( !$this->mOutput ) {
5301 throw new MWException( __METHOD__ .
5302 " can only be called when actually parsing something" );
5303 }
5304 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5305 }
5306
5307 /**
5308 * Callback from the Sanitizer for expanding items found in HTML attribute
5309 * values, so they can be safely tested and escaped.
5310 *
5311 * @param string $text
5312 * @param bool|PPFrame $frame
5313 * @return string
5314 */
5315 public function attributeStripCallback( &$text, $frame = false ) {
5316 $text = $this->replaceVariables( $text, $frame );
5317 $text = $this->mStripState->unstripBoth( $text );
5318 return $text;
5319 }
5320
5321 /**
5322 * Accessor
5323 *
5324 * @return array
5325 */
5326 public function getTags() {
5327 return array_merge(
5328 array_keys( $this->mTransparentTagHooks ),
5329 array_keys( $this->mTagHooks ),
5330 array_keys( $this->mFunctionTagHooks )
5331 );
5332 }
5333
5334 /**
5335 * Replace transparent tags in $text with the values given by the callbacks.
5336 *
5337 * Transparent tag hooks are like regular XML-style tag hooks, except they
5338 * operate late in the transformation sequence, on HTML instead of wikitext.
5339 *
5340 * @param string $text
5341 *
5342 * @return string
5343 */
5344 public function replaceTransparentTags( $text ) {
5345 $matches = [];
5346 $elements = array_keys( $this->mTransparentTagHooks );
5347 $text = self::extractTagsAndParams( $elements, $text, $matches );
5348 $replacements = [];
5349
5350 foreach ( $matches as $marker => $data ) {
5351 list( $element, $content, $params, $tag ) = $data;
5352 $tagName = strtolower( $element );
5353 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5354 $output = call_user_func_array(
5355 $this->mTransparentTagHooks[$tagName],
5356 [ $content, $params, $this ]
5357 );
5358 } else {
5359 $output = $tag;
5360 }
5361 $replacements[$marker] = $output;
5362 }
5363 return strtr( $text, $replacements );
5364 }
5365
5366 /**
5367 * Break wikitext input into sections, and either pull or replace
5368 * some particular section's text.
5369 *
5370 * External callers should use the getSection and replaceSection methods.
5371 *
5372 * @param string $text Page wikitext
5373 * @param string|number $sectionId A section identifier string of the form:
5374 * "<flag1> - <flag2> - ... - <section number>"
5375 *
5376 * Currently the only recognised flag is "T", which means the target section number
5377 * was derived during a template inclusion parse, in other words this is a template
5378 * section edit link. If no flags are given, it was an ordinary section edit link.
5379 * This flag is required to avoid a section numbering mismatch when a section is
5380 * enclosed by "<includeonly>" (bug 6563).
5381 *
5382 * The section number 0 pulls the text before the first heading; other numbers will
5383 * pull the given section along with its lower-level subsections. If the section is
5384 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5385 *
5386 * Section 0 is always considered to exist, even if it only contains the empty
5387 * string. If $text is the empty string and section 0 is replaced, $newText is
5388 * returned.
5389 *
5390 * @param string $mode One of "get" or "replace"
5391 * @param string $newText Replacement text for section data.
5392 * @return string For "get", the extracted section text.
5393 * for "replace", the whole page with the section replaced.
5394 */
5395 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5396 global $wgTitle; # not generally used but removes an ugly failure mode
5397
5398 $magicScopeVariable = $this->lock();
5399 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5400 $outText = '';
5401 $frame = $this->getPreprocessor()->newFrame();
5402
5403 # Process section extraction flags
5404 $flags = 0;
5405 $sectionParts = explode( '-', $sectionId );
5406 $sectionIndex = array_pop( $sectionParts );
5407 foreach ( $sectionParts as $part ) {
5408 if ( $part === 'T' ) {
5409 $flags |= self::PTD_FOR_INCLUSION;
5410 }
5411 }
5412
5413 # Check for empty input
5414 if ( strval( $text ) === '' ) {
5415 # Only sections 0 and T-0 exist in an empty document
5416 if ( $sectionIndex == 0 ) {
5417 if ( $mode === 'get' ) {
5418 return '';
5419 } else {
5420 return $newText;
5421 }
5422 } else {
5423 if ( $mode === 'get' ) {
5424 return $newText;
5425 } else {
5426 return $text;
5427 }
5428 }
5429 }
5430
5431 # Preprocess the text
5432 $root = $this->preprocessToDom( $text, $flags );
5433
5434 # <h> nodes indicate section breaks
5435 # They can only occur at the top level, so we can find them by iterating the root's children
5436 $node = $root->getFirstChild();
5437
5438 # Find the target section
5439 if ( $sectionIndex == 0 ) {
5440 # Section zero doesn't nest, level=big
5441 $targetLevel = 1000;
5442 } else {
5443 while ( $node ) {
5444 if ( $node->getName() === 'h' ) {
5445 $bits = $node->splitHeading();
5446 if ( $bits['i'] == $sectionIndex ) {
5447 $targetLevel = $bits['level'];
5448 break;
5449 }
5450 }
5451 if ( $mode === 'replace' ) {
5452 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5453 }
5454 $node = $node->getNextSibling();
5455 }
5456 }
5457
5458 if ( !$node ) {
5459 # Not found
5460 if ( $mode === 'get' ) {
5461 return $newText;
5462 } else {
5463 return $text;
5464 }
5465 }
5466
5467 # Find the end of the section, including nested sections
5468 do {
5469 if ( $node->getName() === 'h' ) {
5470 $bits = $node->splitHeading();
5471 $curLevel = $bits['level'];
5472 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5473 break;
5474 }
5475 }
5476 if ( $mode === 'get' ) {
5477 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5478 }
5479 $node = $node->getNextSibling();
5480 } while ( $node );
5481
5482 # Write out the remainder (in replace mode only)
5483 if ( $mode === 'replace' ) {
5484 # Output the replacement text
5485 # Add two newlines on -- trailing whitespace in $newText is conventionally
5486 # stripped by the editor, so we need both newlines to restore the paragraph gap
5487 # Only add trailing whitespace if there is newText
5488 if ( $newText != "" ) {
5489 $outText .= $newText . "\n\n";
5490 }
5491
5492 while ( $node ) {
5493 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5494 $node = $node->getNextSibling();
5495 }
5496 }
5497
5498 if ( is_string( $outText ) ) {
5499 # Re-insert stripped tags
5500 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5501 }
5502
5503 return $outText;
5504 }
5505
5506 /**
5507 * This function returns the text of a section, specified by a number ($section).
5508 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5509 * the first section before any such heading (section 0).
5510 *
5511 * If a section contains subsections, these are also returned.
5512 *
5513 * @param string $text Text to look in
5514 * @param string|number $sectionId Section identifier as a number or string
5515 * (e.g. 0, 1 or 'T-1').
5516 * @param string $defaultText Default to return if section is not found
5517 *
5518 * @return string Text of the requested section
5519 */
5520 public function getSection( $text, $sectionId, $defaultText = '' ) {
5521 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5522 }
5523
5524 /**
5525 * This function returns $oldtext after the content of the section
5526 * specified by $section has been replaced with $text. If the target
5527 * section does not exist, $oldtext is returned unchanged.
5528 *
5529 * @param string $oldText Former text of the article
5530 * @param string|number $sectionId Section identifier as a number or string
5531 * (e.g. 0, 1 or 'T-1').
5532 * @param string $newText Replacing text
5533 *
5534 * @return string Modified text
5535 */
5536 public function replaceSection( $oldText, $sectionId, $newText ) {
5537 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5538 }
5539
5540 /**
5541 * Get the ID of the revision we are parsing
5542 *
5543 * @return int|null
5544 */
5545 public function getRevisionId() {
5546 return $this->mRevisionId;
5547 }
5548
5549 /**
5550 * Get the revision object for $this->mRevisionId
5551 *
5552 * @return Revision|null Either a Revision object or null
5553 * @since 1.23 (public since 1.23)
5554 */
5555 public function getRevisionObject() {
5556 if ( !is_null( $this->mRevisionObject ) ) {
5557 return $this->mRevisionObject;
5558 }
5559 if ( is_null( $this->mRevisionId ) ) {
5560 return null;
5561 }
5562
5563 $rev = call_user_func(
5564 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5565 );
5566
5567 # If the parse is for a new revision, then the callback should have
5568 # already been set to force the object and should match mRevisionId.
5569 # If not, try to fetch by mRevisionId for sanity.
5570 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5571 $rev = Revision::newFromId( $this->mRevisionId );
5572 }
5573
5574 $this->mRevisionObject = $rev;
5575
5576 return $this->mRevisionObject;
5577 }
5578
5579 /**
5580 * Get the timestamp associated with the current revision, adjusted for
5581 * the default server-local timestamp
5582 * @return string
5583 */
5584 public function getRevisionTimestamp() {
5585 if ( is_null( $this->mRevisionTimestamp ) ) {
5586 global $wgContLang;
5587
5588 $revObject = $this->getRevisionObject();
5589 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5590
5591 # The cryptic '' timezone parameter tells to use the site-default
5592 # timezone offset instead of the user settings.
5593 # Since this value will be saved into the parser cache, served
5594 # to other users, and potentially even used inside links and such,
5595 # it needs to be consistent for all visitors.
5596 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5597
5598 }
5599 return $this->mRevisionTimestamp;
5600 }
5601
5602 /**
5603 * Get the name of the user that edited the last revision
5604 *
5605 * @return string User name
5606 */
5607 public function getRevisionUser() {
5608 if ( is_null( $this->mRevisionUser ) ) {
5609 $revObject = $this->getRevisionObject();
5610
5611 # if this template is subst: the revision id will be blank,
5612 # so just use the current user's name
5613 if ( $revObject ) {
5614 $this->mRevisionUser = $revObject->getUserText();
5615 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5616 $this->mRevisionUser = $this->getUser()->getName();
5617 }
5618 }
5619 return $this->mRevisionUser;
5620 }
5621
5622 /**
5623 * Get the size of the revision
5624 *
5625 * @return int|null Revision size
5626 */
5627 public function getRevisionSize() {
5628 if ( is_null( $this->mRevisionSize ) ) {
5629 $revObject = $this->getRevisionObject();
5630
5631 # if this variable is subst: the revision id will be blank,
5632 # so just use the parser input size, because the own substituation
5633 # will change the size.
5634 if ( $revObject ) {
5635 $this->mRevisionSize = $revObject->getSize();
5636 } else {
5637 $this->mRevisionSize = $this->mInputSize;
5638 }
5639 }
5640 return $this->mRevisionSize;
5641 }
5642
5643 /**
5644 * Mutator for $mDefaultSort
5645 *
5646 * @param string $sort New value
5647 */
5648 public function setDefaultSort( $sort ) {
5649 $this->mDefaultSort = $sort;
5650 $this->mOutput->setProperty( 'defaultsort', $sort );
5651 }
5652
5653 /**
5654 * Accessor for $mDefaultSort
5655 * Will use the empty string if none is set.
5656 *
5657 * This value is treated as a prefix, so the
5658 * empty string is equivalent to sorting by
5659 * page name.
5660 *
5661 * @return string
5662 */
5663 public function getDefaultSort() {
5664 if ( $this->mDefaultSort !== false ) {
5665 return $this->mDefaultSort;
5666 } else {
5667 return '';
5668 }
5669 }
5670
5671 /**
5672 * Accessor for $mDefaultSort
5673 * Unlike getDefaultSort(), will return false if none is set
5674 *
5675 * @return string|bool
5676 */
5677 public function getCustomDefaultSort() {
5678 return $this->mDefaultSort;
5679 }
5680
5681 /**
5682 * Try to guess the section anchor name based on a wikitext fragment
5683 * presumably extracted from a heading, for example "Header" from
5684 * "== Header ==".
5685 *
5686 * @param string $text
5687 *
5688 * @return string
5689 */
5690 public function guessSectionNameFromWikiText( $text ) {
5691 # Strip out wikitext links(they break the anchor)
5692 $text = $this->stripSectionName( $text );
5693 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5694 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5695 }
5696
5697 /**
5698 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5699 * instead. For use in redirects, since IE6 interprets Redirect: headers
5700 * as something other than UTF-8 (apparently?), resulting in breakage.
5701 *
5702 * @param string $text The section name
5703 * @return string An anchor
5704 */
5705 public function guessLegacySectionNameFromWikiText( $text ) {
5706 # Strip out wikitext links(they break the anchor)
5707 $text = $this->stripSectionName( $text );
5708 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5709 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5710 }
5711
5712 /**
5713 * Strips a text string of wikitext for use in a section anchor
5714 *
5715 * Accepts a text string and then removes all wikitext from the
5716 * string and leaves only the resultant text (i.e. the result of
5717 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5718 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5719 * to create valid section anchors by mimicing the output of the
5720 * parser when headings are parsed.
5721 *
5722 * @param string $text Text string to be stripped of wikitext
5723 * for use in a Section anchor
5724 * @return string Filtered text string
5725 */
5726 public function stripSectionName( $text ) {
5727 # Strip internal link markup
5728 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5729 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5730
5731 # Strip external link markup
5732 # @todo FIXME: Not tolerant to blank link text
5733 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5734 # on how many empty links there are on the page - need to figure that out.
5735 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5736
5737 # Parse wikitext quotes (italics & bold)
5738 $text = $this->doQuotes( $text );
5739
5740 # Strip HTML tags
5741 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5742 return $text;
5743 }
5744
5745 /**
5746 * strip/replaceVariables/unstrip for preprocessor regression testing
5747 *
5748 * @param string $text
5749 * @param Title $title
5750 * @param ParserOptions $options
5751 * @param int $outputType
5752 *
5753 * @return string
5754 */
5755 public function testSrvus( $text, Title $title, ParserOptions $options,
5756 $outputType = self::OT_HTML
5757 ) {
5758 $magicScopeVariable = $this->lock();
5759 $this->startParse( $title, $options, $outputType, true );
5760
5761 $text = $this->replaceVariables( $text );
5762 $text = $this->mStripState->unstripBoth( $text );
5763 $text = Sanitizer::removeHTMLtags( $text );
5764 return $text;
5765 }
5766
5767 /**
5768 * @param string $text
5769 * @param Title $title
5770 * @param ParserOptions $options
5771 * @return string
5772 */
5773 public function testPst( $text, Title $title, ParserOptions $options ) {
5774 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5775 }
5776
5777 /**
5778 * @param string $text
5779 * @param Title $title
5780 * @param ParserOptions $options
5781 * @return string
5782 */
5783 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5784 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5785 }
5786
5787 /**
5788 * Call a callback function on all regions of the given text that are not
5789 * inside strip markers, and replace those regions with the return value
5790 * of the callback. For example, with input:
5791 *
5792 * aaa<MARKER>bbb
5793 *
5794 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5795 * two strings will be replaced with the value returned by the callback in
5796 * each case.
5797 *
5798 * @param string $s
5799 * @param callable $callback
5800 *
5801 * @return string
5802 */
5803 public function markerSkipCallback( $s, $callback ) {
5804 $i = 0;
5805 $out = '';
5806 while ( $i < strlen( $s ) ) {
5807 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5808 if ( $markerStart === false ) {
5809 $out .= call_user_func( $callback, substr( $s, $i ) );
5810 break;
5811 } else {
5812 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5813 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5814 if ( $markerEnd === false ) {
5815 $out .= substr( $s, $markerStart );
5816 break;
5817 } else {
5818 $markerEnd += strlen( self::MARKER_SUFFIX );
5819 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5820 $i = $markerEnd;
5821 }
5822 }
5823 }
5824 return $out;
5825 }
5826
5827 /**
5828 * Remove any strip markers found in the given text.
5829 *
5830 * @param string $text Input string
5831 * @return string
5832 */
5833 public function killMarkers( $text ) {
5834 return $this->mStripState->killMarkers( $text );
5835 }
5836
5837 /**
5838 * Save the parser state required to convert the given half-parsed text to
5839 * HTML. "Half-parsed" in this context means the output of
5840 * recursiveTagParse() or internalParse(). This output has strip markers
5841 * from replaceVariables (extensionSubstitution() etc.), and link
5842 * placeholders from replaceLinkHolders().
5843 *
5844 * Returns an array which can be serialized and stored persistently. This
5845 * array can later be loaded into another parser instance with
5846 * unserializeHalfParsedText(). The text can then be safely incorporated into
5847 * the return value of a parser hook.
5848 *
5849 * @param string $text
5850 *
5851 * @return array
5852 */
5853 public function serializeHalfParsedText( $text ) {
5854 $data = [
5855 'text' => $text,
5856 'version' => self::HALF_PARSED_VERSION,
5857 'stripState' => $this->mStripState->getSubState( $text ),
5858 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5859 ];
5860 return $data;
5861 }
5862
5863 /**
5864 * Load the parser state given in the $data array, which is assumed to
5865 * have been generated by serializeHalfParsedText(). The text contents is
5866 * extracted from the array, and its markers are transformed into markers
5867 * appropriate for the current Parser instance. This transformed text is
5868 * returned, and can be safely included in the return value of a parser
5869 * hook.
5870 *
5871 * If the $data array has been stored persistently, the caller should first
5872 * check whether it is still valid, by calling isValidHalfParsedText().
5873 *
5874 * @param array $data Serialized data
5875 * @throws MWException
5876 * @return string
5877 */
5878 public function unserializeHalfParsedText( $data ) {
5879 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5880 throw new MWException( __METHOD__ . ': invalid version' );
5881 }
5882
5883 # First, extract the strip state.
5884 $texts = [ $data['text'] ];
5885 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5886
5887 # Now renumber links
5888 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5889
5890 # Should be good to go.
5891 return $texts[0];
5892 }
5893
5894 /**
5895 * Returns true if the given array, presumed to be generated by
5896 * serializeHalfParsedText(), is compatible with the current version of the
5897 * parser.
5898 *
5899 * @param array $data
5900 *
5901 * @return bool
5902 */
5903 public function isValidHalfParsedText( $data ) {
5904 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5905 }
5906
5907 /**
5908 * Parsed a width param of imagelink like 300px or 200x300px
5909 *
5910 * @param string $value
5911 *
5912 * @return array
5913 * @since 1.20
5914 */
5915 public function parseWidthParam( $value ) {
5916 $parsedWidthParam = [];
5917 if ( $value === '' ) {
5918 return $parsedWidthParam;
5919 }
5920 $m = [];
5921 # (bug 13500) In both cases (width/height and width only),
5922 # permit trailing "px" for backward compatibility.
5923 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5924 $width = intval( $m[1] );
5925 $height = intval( $m[2] );
5926 $parsedWidthParam['width'] = $width;
5927 $parsedWidthParam['height'] = $height;
5928 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5929 $width = intval( $value );
5930 $parsedWidthParam['width'] = $width;
5931 }
5932 return $parsedWidthParam;
5933 }
5934
5935 /**
5936 * Lock the current instance of the parser.
5937 *
5938 * This is meant to stop someone from calling the parser
5939 * recursively and messing up all the strip state.
5940 *
5941 * @throws MWException If parser is in a parse
5942 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5943 */
5944 protected function lock() {
5945 if ( $this->mInParse ) {
5946 throw new MWException( "Parser state cleared while parsing. "
5947 . "Did you call Parser::parse recursively?" );
5948 }
5949 $this->mInParse = true;
5950
5951 $recursiveCheck = new ScopedCallback( function() {
5952 $this->mInParse = false;
5953 } );
5954
5955 return $recursiveCheck;
5956 }
5957
5958 /**
5959 * Strip outer <p></p> tag from the HTML source of a single paragraph.
5960 *
5961 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
5962 * or if there is more than one <p/> tag in the input HTML.
5963 *
5964 * @param string $html
5965 * @return string
5966 * @since 1.24
5967 */
5968 public static function stripOuterParagraph( $html ) {
5969 $m = [];
5970 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
5971 if ( strpos( $m[1], '</p>' ) === false ) {
5972 $html = $m[1];
5973 }
5974 }
5975
5976 return $html;
5977 }
5978
5979 /**
5980 * Return this parser if it is not doing anything, otherwise
5981 * get a fresh parser. You can use this method by doing
5982 * $myParser = $wgParser->getFreshParser(), or more simply
5983 * $wgParser->getFreshParser()->parse( ... );
5984 * if you're unsure if $wgParser is safe to use.
5985 *
5986 * @since 1.24
5987 * @return Parser A parser object that is not parsing anything
5988 */
5989 public function getFreshParser() {
5990 global $wgParserConf;
5991 if ( $this->mInParse ) {
5992 return new $wgParserConf['class']( $wgParserConf );
5993 } else {
5994 return $this;
5995 }
5996 }
5997
5998 /**
5999 * Set's up the PHP implementation of OOUI for use in this request
6000 * and instructs OutputPage to enable OOUI for itself.
6001 *
6002 * @since 1.26
6003 */
6004 public function enableOOUI() {
6005 OutputPage::setupOOUI();
6006 $this->mOutput->setEnableOOUI( true );
6007 }
6008 }
Wiklou, le Wiki du Wiklou