dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "mw.Upload.BookletLayout: Better handle error messages from AbuseFilter and...
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25
26 /**
27 * @defgroup Parser Parser
28 */
29
30 /**
31 * PHP Parser - Processes wiki markup (which uses a more user-friendly
32 * syntax, such as "[[link]]" for making links), and provides a one-way
33 * transformation of that wiki markup it into (X)HTML output / markup
34 * (which in turn the browser understands, and can display).
35 *
36 * There are seven main entry points into the Parser class:
37 *
38 * - Parser::parse()
39 * produces HTML output
40 * - Parser::preSaveTransform()
41 * produces altered wiki markup
42 * - Parser::preprocess()
43 * removes HTML comments and expands templates
44 * - Parser::cleanSig() and Parser::cleanSigInSig()
45 * cleans a signature before saving it to preferences
46 * - Parser::getSection()
47 * return the content of a section from an article for section editing
48 * - Parser::replaceSection()
49 * replaces a section by number inside an article
50 * - Parser::getPreloadText()
51 * removes <noinclude> sections and <includeonly> tags
52 *
53 * Globals used:
54 * object: $wgContLang
55 *
56 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
57 *
58 * @par Settings:
59 * $wgNamespacesWithSubpages
60 *
61 * @par Settings only within ParserOptions:
62 * $wgAllowExternalImages
63 * $wgAllowSpecialInclusion
64 * $wgInterwikiMagic
65 * $wgMaxArticleSize
66 *
67 * @ingroup Parser
68 */
69 class Parser {
70 /**
71 * Update this version number when the ParserOutput format
72 * changes in an incompatible way, so the parser cache
73 * can automatically discard old data.
74 */
75 const VERSION = '1.6.4';
76
77 /**
78 * Update this version number when the output of serialiseHalfParsedText()
79 * changes in an incompatible way
80 */
81 const HALF_PARSED_VERSION = 2;
82
83 # Flags for Parser::setFunctionHook
84 const SFH_NO_HASH = 1;
85 const SFH_OBJECT_ARGS = 2;
86
87 # Constants needed for external link processing
88 # Everything except bracket, space, or control characters
89 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
90 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
91 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
92 # Simplified expression to match an IPv4 or IPv6 address, or
93 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
94 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
95 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
96 // @codingStandardsIgnoreStart Generic.Files.LineLength
97 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
98 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
99 // @codingStandardsIgnoreEnd
100
101 # Regular expression for a non-newline space
102 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
103
104 # Flags for preprocessToDom
105 const PTD_FOR_INCLUSION = 1;
106
107 # Allowed values for $this->mOutputType
108 # Parameter to startExternalParse().
109 const OT_HTML = 1; # like parse()
110 const OT_WIKI = 2; # like preSaveTransform()
111 const OT_PREPROCESS = 3; # like preprocess()
112 const OT_MSG = 3;
113 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
114
115 /**
116 * @var string Prefix and suffix for temporary replacement strings
117 * for the multipass parser.
118 *
119 * \x7f should never appear in input as it's disallowed in XML.
120 * Using it at the front also gives us a little extra robustness
121 * since it shouldn't match when butted up against identifier-like
122 * string constructs.
123 *
124 * Must not consist of all title characters, or else it will change
125 * the behavior of <nowiki> in a link.
126 *
127 * Must have a character that needs escaping in attributes, otherwise
128 * someone could put a strip marker in an attribute, to get around
129 * escaping quote marks, and break out of the attribute. Thus we add
130 * `'".
131 */
132 const MARKER_SUFFIX = "-QINU`\"'\x7f";
133 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
134
135 # Markers used for wrapping the table of contents
136 const TOC_START = '<mw:toc>';
137 const TOC_END = '</mw:toc>';
138
139 # Persistent:
140 public $mTagHooks = [];
141 public $mTransparentTagHooks = [];
142 public $mFunctionHooks = [];
143 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
144 public $mFunctionTagHooks = [];
145 public $mStripList = [];
146 public $mDefaultStripList = [];
147 public $mVarCache = [];
148 public $mImageParams = [];
149 public $mImageParamsMagicArray = [];
150 public $mMarkerIndex = 0;
151 public $mFirstCall = true;
152
153 # Initialised by initialiseVariables()
154
155 /**
156 * @var MagicWordArray
157 */
158 public $mVariables;
159
160 /**
161 * @var MagicWordArray
162 */
163 public $mSubstWords;
164 # Initialised in constructor
165 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
166
167 # Initialized in getPreprocessor()
168 /** @var Preprocessor */
169 public $mPreprocessor;
170
171 # Cleared with clearState():
172 /**
173 * @var ParserOutput
174 */
175 public $mOutput;
176 public $mAutonumber;
177
178 /**
179 * @var StripState
180 */
181 public $mStripState;
182
183 public $mIncludeCount;
184 /**
185 * @var LinkHolderArray
186 */
187 public $mLinkHolders;
188
189 public $mLinkID;
190 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
191 public $mDefaultSort;
192 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
193 public $mExpensiveFunctionCount; # number of expensive parser function calls
194 public $mShowToc, $mForceTocPosition;
195
196 /**
197 * @var User
198 */
199 public $mUser; # User object; only used when doing pre-save transform
200
201 # Temporary
202 # These are variables reset at least once per parse regardless of $clearState
203
204 /**
205 * @var ParserOptions
206 */
207 public $mOptions;
208
209 /**
210 * @var Title
211 */
212 public $mTitle; # Title context, used for self-link rendering and similar things
213 public $mOutputType; # Output type, one of the OT_xxx constants
214 public $ot; # Shortcut alias, see setOutputType()
215 public $mRevisionObject; # The revision object of the specified revision ID
216 public $mRevisionId; # ID to display in {{REVISIONID}} tags
217 public $mRevisionTimestamp; # The timestamp of the specified revision ID
218 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
219 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
220 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
221 public $mInputSize = false; # For {{PAGESIZE}} on current page.
222
223 /**
224 * @var string Deprecated accessor for the strip marker prefix.
225 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
226 **/
227 public $mUniqPrefix = Parser::MARKER_PREFIX;
228
229 /**
230 * @var array Array with the language name of each language link (i.e. the
231 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
232 * duplicate language links to the ParserOutput.
233 */
234 public $mLangLinkLanguages;
235
236 /**
237 * @var MapCacheLRU|null
238 * @since 1.24
239 *
240 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
241 */
242 public $currentRevisionCache;
243
244 /**
245 * @var bool Recursive call protection.
246 * This variable should be treated as if it were private.
247 */
248 public $mInParse = false;
249
250 /** @var SectionProfiler */
251 protected $mProfiler;
252
253 /**
254 * @var LinkRenderer
255 */
256 protected $mLinkRenderer;
257
258 /**
259 * @param array $conf
260 */
261 public function __construct( $conf = [] ) {
262 $this->mConf = $conf;
263 $this->mUrlProtocols = wfUrlProtocols();
264 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
265 self::EXT_LINK_ADDR .
266 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
267 if ( isset( $conf['preprocessorClass'] ) ) {
268 $this->mPreprocessorClass = $conf['preprocessorClass'];
269 } elseif ( defined( 'HPHP_VERSION' ) ) {
270 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
271 $this->mPreprocessorClass = 'Preprocessor_Hash';
272 } elseif ( extension_loaded( 'domxml' ) ) {
273 # PECL extension that conflicts with the core DOM extension (bug 13770)
274 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
275 $this->mPreprocessorClass = 'Preprocessor_Hash';
276 } elseif ( extension_loaded( 'dom' ) ) {
277 $this->mPreprocessorClass = 'Preprocessor_DOM';
278 } else {
279 $this->mPreprocessorClass = 'Preprocessor_Hash';
280 }
281 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
282 }
283
284 /**
285 * Reduce memory usage to reduce the impact of circular references
286 */
287 public function __destruct() {
288 if ( isset( $this->mLinkHolders ) ) {
289 unset( $this->mLinkHolders );
290 }
291 foreach ( $this as $name => $value ) {
292 unset( $this->$name );
293 }
294 }
295
296 /**
297 * Allow extensions to clean up when the parser is cloned
298 */
299 public function __clone() {
300 $this->mInParse = false;
301
302 // Bug 56226: When you create a reference "to" an object field, that
303 // makes the object field itself be a reference too (until the other
304 // reference goes out of scope). When cloning, any field that's a
305 // reference is copied as a reference in the new object. Both of these
306 // are defined PHP5 behaviors, as inconvenient as it is for us when old
307 // hooks from PHP4 days are passing fields by reference.
308 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
309 // Make a non-reference copy of the field, then rebind the field to
310 // reference the new copy.
311 $tmp = $this->$k;
312 $this->$k =& $tmp;
313 unset( $tmp );
314 }
315
316 Hooks::run( 'ParserCloned', [ $this ] );
317 }
318
319 /**
320 * Do various kinds of initialisation on the first call of the parser
321 */
322 public function firstCallInit() {
323 if ( !$this->mFirstCall ) {
324 return;
325 }
326 $this->mFirstCall = false;
327
328 CoreParserFunctions::register( $this );
329 CoreTagHooks::register( $this );
330 $this->initialiseVariables();
331
332 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
333 }
334
335 /**
336 * Clear Parser state
337 *
338 * @private
339 */
340 public function clearState() {
341 if ( $this->mFirstCall ) {
342 $this->firstCallInit();
343 }
344 $this->mOutput = new ParserOutput;
345 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
346 $this->mAutonumber = 0;
347 $this->mIncludeCount = [];
348 $this->mLinkHolders = new LinkHolderArray( $this );
349 $this->mLinkID = 0;
350 $this->mRevisionObject = $this->mRevisionTimestamp =
351 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
352 $this->mVarCache = [];
353 $this->mUser = null;
354 $this->mLangLinkLanguages = [];
355 $this->currentRevisionCache = null;
356
357 $this->mStripState = new StripState;
358
359 # Clear these on every parse, bug 4549
360 $this->mTplRedirCache = $this->mTplDomCache = [];
361
362 $this->mShowToc = true;
363 $this->mForceTocPosition = false;
364 $this->mIncludeSizes = [
365 'post-expand' => 0,
366 'arg' => 0,
367 ];
368 $this->mPPNodeCount = 0;
369 $this->mGeneratedPPNodeCount = 0;
370 $this->mHighestExpansionDepth = 0;
371 $this->mDefaultSort = false;
372 $this->mHeadings = [];
373 $this->mDoubleUnderscores = [];
374 $this->mExpensiveFunctionCount = 0;
375
376 # Fix cloning
377 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
378 $this->mPreprocessor = null;
379 }
380
381 $this->mProfiler = new SectionProfiler();
382
383 Hooks::run( 'ParserClearState', [ &$this ] );
384 }
385
386 /**
387 * Convert wikitext to HTML
388 * Do not call this function recursively.
389 *
390 * @param string $text Text we want to parse
391 * @param Title $title
392 * @param ParserOptions $options
393 * @param bool $linestart
394 * @param bool $clearState
395 * @param int $revid Number to pass in {{REVISIONID}}
396 * @return ParserOutput A ParserOutput
397 */
398 public function parse( $text, Title $title, ParserOptions $options,
399 $linestart = true, $clearState = true, $revid = null
400 ) {
401 /**
402 * First pass--just handle <nowiki> sections, pass the rest off
403 * to internalParse() which does all the real work.
404 */
405
406 global $wgShowHostnames;
407
408 if ( $clearState ) {
409 // We use U+007F DELETE to construct strip markers, so we have to make
410 // sure that this character does not occur in the input text.
411 $text = strtr( $text, "\x7f", "?" );
412 $magicScopeVariable = $this->lock();
413 }
414
415 $this->startParse( $title, $options, self::OT_HTML, $clearState );
416
417 $this->currentRevisionCache = null;
418 $this->mInputSize = strlen( $text );
419 if ( $this->mOptions->getEnableLimitReport() ) {
420 $this->mOutput->resetParseStartTime();
421 }
422
423 $oldRevisionId = $this->mRevisionId;
424 $oldRevisionObject = $this->mRevisionObject;
425 $oldRevisionTimestamp = $this->mRevisionTimestamp;
426 $oldRevisionUser = $this->mRevisionUser;
427 $oldRevisionSize = $this->mRevisionSize;
428 if ( $revid !== null ) {
429 $this->mRevisionId = $revid;
430 $this->mRevisionObject = null;
431 $this->mRevisionTimestamp = null;
432 $this->mRevisionUser = null;
433 $this->mRevisionSize = null;
434 }
435
436 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
437 # No more strip!
438 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
439 $text = $this->internalParse( $text );
440 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
441
442 $text = $this->internalParseHalfParsed( $text, true, $linestart );
443
444 /**
445 * A converted title will be provided in the output object if title and
446 * content conversion are enabled, the article text does not contain
447 * a conversion-suppressing double-underscore tag, and no
448 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
449 * automatic link conversion.
450 */
451 if ( !( $options->getDisableTitleConversion()
452 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
453 || isset( $this->mDoubleUnderscores['notitleconvert'] )
454 || $this->mOutput->getDisplayTitle() !== false )
455 ) {
456 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
457 if ( $convruletitle ) {
458 $this->mOutput->setTitleText( $convruletitle );
459 } else {
460 $titleText = $this->getConverterLanguage()->convertTitle( $title );
461 $this->mOutput->setTitleText( $titleText );
462 }
463 }
464
465 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
466 $this->limitationWarn( 'expensive-parserfunction',
467 $this->mExpensiveFunctionCount,
468 $this->mOptions->getExpensiveParserFunctionLimit()
469 );
470 }
471
472 # Information on include size limits, for the benefit of users who try to skirt them
473 if ( $this->mOptions->getEnableLimitReport() ) {
474 $max = $this->mOptions->getMaxIncludeSize();
475
476 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
477 if ( $cpuTime !== null ) {
478 $this->mOutput->setLimitReportData( 'limitreport-cputime',
479 sprintf( "%.3f", $cpuTime )
480 );
481 }
482
483 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
484 $this->mOutput->setLimitReportData( 'limitreport-walltime',
485 sprintf( "%.3f", $wallTime )
486 );
487
488 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
489 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
490 );
491 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
492 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
493 );
494 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
495 [ $this->mIncludeSizes['post-expand'], $max ]
496 );
497 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
498 [ $this->mIncludeSizes['arg'], $max ]
499 );
500 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
501 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
502 );
503 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
504 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
505 );
506 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
507
508 $limitReport = "NewPP limit report\n";
509 if ( $wgShowHostnames ) {
510 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
511 }
512 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
513 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
514 $limitReport .= 'Dynamic content: ' .
515 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
516 "\n";
517
518 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
519 if ( Hooks::run( 'ParserLimitReportFormat',
520 [ $key, &$value, &$limitReport, false, false ]
521 ) ) {
522 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
523 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
524 ->inLanguage( 'en' )->useDatabase( false );
525 if ( !$valueMsg->exists() ) {
526 $valueMsg = new RawMessage( '$1' );
527 }
528 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
529 $valueMsg->params( $value );
530 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
531 }
532 }
533 }
534 // Since we're not really outputting HTML, decode the entities and
535 // then re-encode the things that need hiding inside HTML comments.
536 $limitReport = htmlspecialchars_decode( $limitReport );
537 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
538
539 // Sanitize for comment. Note '‐' in the replacement is U+2010,
540 // which looks much like the problematic '-'.
541 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
542 $text .= "\n<!-- \n$limitReport-->\n";
543
544 // Add on template profiling data
545 $dataByFunc = $this->mProfiler->getFunctionStats();
546 uasort( $dataByFunc, function ( $a, $b ) {
547 return $a['real'] < $b['real']; // descending order
548 } );
549 $profileReport = "Transclusion expansion time report (%,ms,calls,template)\n";
550 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
551 $profileReport .= sprintf( "%6.2f%% %8.3f %6d - %s\n",
552 $item['%real'], $item['real'], $item['calls'],
553 htmlspecialchars( $item['name'] ) );
554 }
555 $text .= "\n<!-- \n$profileReport-->\n";
556
557 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
558 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
559 $this->mTitle->getPrefixedDBkey() );
560 }
561 }
562 $this->mOutput->setText( $text );
563
564 $this->mRevisionId = $oldRevisionId;
565 $this->mRevisionObject = $oldRevisionObject;
566 $this->mRevisionTimestamp = $oldRevisionTimestamp;
567 $this->mRevisionUser = $oldRevisionUser;
568 $this->mRevisionSize = $oldRevisionSize;
569 $this->mInputSize = false;
570 $this->currentRevisionCache = null;
571
572 return $this->mOutput;
573 }
574
575 /**
576 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
577 * can be called from an extension tag hook.
578 *
579 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
580 * instead, which means that lists and links have not been fully parsed yet,
581 * and strip markers are still present.
582 *
583 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
584 *
585 * Use this function if you're a parser tag hook and you want to parse
586 * wikitext before or after applying additional transformations, and you
587 * intend to *return the result as hook output*, which will cause it to go
588 * through the rest of parsing process automatically.
589 *
590 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
591 * $text are not expanded
592 *
593 * @param string $text Text extension wants to have parsed
594 * @param bool|PPFrame $frame The frame to use for expanding any template variables
595 * @return string UNSAFE half-parsed HTML
596 */
597 public function recursiveTagParse( $text, $frame = false ) {
598 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
599 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
600 $text = $this->internalParse( $text, false, $frame );
601 return $text;
602 }
603
604 /**
605 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
606 * point can be called from an extension tag hook.
607 *
608 * The output of this function is fully-parsed HTML that is safe for output.
609 * If you're a parser tag hook, you might want to use recursiveTagParse()
610 * instead.
611 *
612 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
613 * $text are not expanded
614 *
615 * @since 1.25
616 *
617 * @param string $text Text extension wants to have parsed
618 * @param bool|PPFrame $frame The frame to use for expanding any template variables
619 * @return string Fully parsed HTML
620 */
621 public function recursiveTagParseFully( $text, $frame = false ) {
622 $text = $this->recursiveTagParse( $text, $frame );
623 $text = $this->internalParseHalfParsed( $text, false );
624 return $text;
625 }
626
627 /**
628 * Expand templates and variables in the text, producing valid, static wikitext.
629 * Also removes comments.
630 * Do not call this function recursively.
631 * @param string $text
632 * @param Title $title
633 * @param ParserOptions $options
634 * @param int|null $revid
635 * @param bool|PPFrame $frame
636 * @return mixed|string
637 */
638 public function preprocess( $text, Title $title = null,
639 ParserOptions $options, $revid = null, $frame = false
640 ) {
641 $magicScopeVariable = $this->lock();
642 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
643 if ( $revid !== null ) {
644 $this->mRevisionId = $revid;
645 }
646 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
647 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
648 $text = $this->replaceVariables( $text, $frame );
649 $text = $this->mStripState->unstripBoth( $text );
650 return $text;
651 }
652
653 /**
654 * Recursive parser entry point that can be called from an extension tag
655 * hook.
656 *
657 * @param string $text Text to be expanded
658 * @param bool|PPFrame $frame The frame to use for expanding any template variables
659 * @return string
660 * @since 1.19
661 */
662 public function recursivePreprocess( $text, $frame = false ) {
663 $text = $this->replaceVariables( $text, $frame );
664 $text = $this->mStripState->unstripBoth( $text );
665 return $text;
666 }
667
668 /**
669 * Process the wikitext for the "?preload=" feature. (bug 5210)
670 *
671 * "<noinclude>", "<includeonly>" etc. are parsed as for template
672 * transclusion, comments, templates, arguments, tags hooks and parser
673 * functions are untouched.
674 *
675 * @param string $text
676 * @param Title $title
677 * @param ParserOptions $options
678 * @param array $params
679 * @return string
680 */
681 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
682 $msg = new RawMessage( $text );
683 $text = $msg->params( $params )->plain();
684
685 # Parser (re)initialisation
686 $magicScopeVariable = $this->lock();
687 $this->startParse( $title, $options, self::OT_PLAIN, true );
688
689 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
690 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
691 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
692 $text = $this->mStripState->unstripBoth( $text );
693 return $text;
694 }
695
696 /**
697 * Get a random string
698 *
699 * @return string
700 * @deprecated since 1.26; use wfRandomString() instead.
701 */
702 public static function getRandomString() {
703 wfDeprecated( __METHOD__, '1.26' );
704 return wfRandomString( 16 );
705 }
706
707 /**
708 * Set the current user.
709 * Should only be used when doing pre-save transform.
710 *
711 * @param User|null $user User object or null (to reset)
712 */
713 public function setUser( $user ) {
714 $this->mUser = $user;
715 }
716
717 /**
718 * Accessor for mUniqPrefix.
719 *
720 * @return string
721 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
722 */
723 public function uniqPrefix() {
724 wfDeprecated( __METHOD__, '1.26' );
725 return self::MARKER_PREFIX;
726 }
727
728 /**
729 * Set the context title
730 *
731 * @param Title $t
732 */
733 public function setTitle( $t ) {
734 if ( !$t ) {
735 $t = Title::newFromText( 'NO TITLE' );
736 }
737
738 if ( $t->hasFragment() ) {
739 # Strip the fragment to avoid various odd effects
740 $this->mTitle = $t->createFragmentTarget( '' );
741 } else {
742 $this->mTitle = $t;
743 }
744 }
745
746 /**
747 * Accessor for the Title object
748 *
749 * @return Title
750 */
751 public function getTitle() {
752 return $this->mTitle;
753 }
754
755 /**
756 * Accessor/mutator for the Title object
757 *
758 * @param Title $x Title object or null to just get the current one
759 * @return Title
760 */
761 public function Title( $x = null ) {
762 return wfSetVar( $this->mTitle, $x );
763 }
764
765 /**
766 * Set the output type
767 *
768 * @param int $ot New value
769 */
770 public function setOutputType( $ot ) {
771 $this->mOutputType = $ot;
772 # Shortcut alias
773 $this->ot = [
774 'html' => $ot == self::OT_HTML,
775 'wiki' => $ot == self::OT_WIKI,
776 'pre' => $ot == self::OT_PREPROCESS,
777 'plain' => $ot == self::OT_PLAIN,
778 ];
779 }
780
781 /**
782 * Accessor/mutator for the output type
783 *
784 * @param int|null $x New value or null to just get the current one
785 * @return int
786 */
787 public function OutputType( $x = null ) {
788 return wfSetVar( $this->mOutputType, $x );
789 }
790
791 /**
792 * Get the ParserOutput object
793 *
794 * @return ParserOutput
795 */
796 public function getOutput() {
797 return $this->mOutput;
798 }
799
800 /**
801 * Get the ParserOptions object
802 *
803 * @return ParserOptions
804 */
805 public function getOptions() {
806 return $this->mOptions;
807 }
808
809 /**
810 * Accessor/mutator for the ParserOptions object
811 *
812 * @param ParserOptions $x New value or null to just get the current one
813 * @return ParserOptions Current ParserOptions object
814 */
815 public function Options( $x = null ) {
816 return wfSetVar( $this->mOptions, $x );
817 }
818
819 /**
820 * @return int
821 */
822 public function nextLinkID() {
823 return $this->mLinkID++;
824 }
825
826 /**
827 * @param int $id
828 */
829 public function setLinkID( $id ) {
830 $this->mLinkID = $id;
831 }
832
833 /**
834 * Get a language object for use in parser functions such as {{FORMATNUM:}}
835 * @return Language
836 */
837 public function getFunctionLang() {
838 return $this->getTargetLanguage();
839 }
840
841 /**
842 * Get the target language for the content being parsed. This is usually the
843 * language that the content is in.
844 *
845 * @since 1.19
846 *
847 * @throws MWException
848 * @return Language
849 */
850 public function getTargetLanguage() {
851 $target = $this->mOptions->getTargetLanguage();
852
853 if ( $target !== null ) {
854 return $target;
855 } elseif ( $this->mOptions->getInterfaceMessage() ) {
856 return $this->mOptions->getUserLangObj();
857 } elseif ( is_null( $this->mTitle ) ) {
858 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
859 }
860
861 return $this->mTitle->getPageLanguage();
862 }
863
864 /**
865 * Get the language object for language conversion
866 * @return Language|null
867 */
868 public function getConverterLanguage() {
869 return $this->getTargetLanguage();
870 }
871
872 /**
873 * Get a User object either from $this->mUser, if set, or from the
874 * ParserOptions object otherwise
875 *
876 * @return User
877 */
878 public function getUser() {
879 if ( !is_null( $this->mUser ) ) {
880 return $this->mUser;
881 }
882 return $this->mOptions->getUser();
883 }
884
885 /**
886 * Get a preprocessor object
887 *
888 * @return Preprocessor
889 */
890 public function getPreprocessor() {
891 if ( !isset( $this->mPreprocessor ) ) {
892 $class = $this->mPreprocessorClass;
893 $this->mPreprocessor = new $class( $this );
894 }
895 return $this->mPreprocessor;
896 }
897
898 /**
899 * Get a LinkRenderer instance to make links with
900 *
901 * @since 1.28
902 * @return LinkRenderer
903 */
904 public function getLinkRenderer() {
905 if ( !$this->mLinkRenderer ) {
906 $this->mLinkRenderer = MediaWikiServices::getInstance()
907 ->getLinkRendererFactory()->create();
908 $this->mLinkRenderer->setStubThreshold(
909 $this->getOptions()->getStubThreshold()
910 );
911 }
912
913 return $this->mLinkRenderer;
914 }
915
916 /**
917 * Replaces all occurrences of HTML-style comments and the given tags
918 * in the text with a random marker and returns the next text. The output
919 * parameter $matches will be an associative array filled with data in
920 * the form:
921 *
922 * @code
923 * 'UNIQ-xxxxx' => array(
924 * 'element',
925 * 'tag content',
926 * array( 'param' => 'x' ),
927 * '<element param="x">tag content</element>' ) )
928 * @endcode
929 *
930 * @param array $elements List of element names. Comments are always extracted.
931 * @param string $text Source text string.
932 * @param array $matches Out parameter, Array: extracted tags
933 * @param string|null $uniq_prefix
934 * @return string Stripped text
935 * @since 1.26 The uniq_prefix argument is deprecated.
936 */
937 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
938 if ( $uniq_prefix !== null ) {
939 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
940 }
941 static $n = 1;
942 $stripped = '';
943 $matches = [];
944
945 $taglist = implode( '|', $elements );
946 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
947
948 while ( $text != '' ) {
949 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
950 $stripped .= $p[0];
951 if ( count( $p ) < 5 ) {
952 break;
953 }
954 if ( count( $p ) > 5 ) {
955 # comment
956 $element = $p[4];
957 $attributes = '';
958 $close = '';
959 $inside = $p[5];
960 } else {
961 # tag
962 $element = $p[1];
963 $attributes = $p[2];
964 $close = $p[3];
965 $inside = $p[4];
966 }
967
968 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
969 $stripped .= $marker;
970
971 if ( $close === '/>' ) {
972 # Empty element tag, <tag />
973 $content = null;
974 $text = $inside;
975 $tail = null;
976 } else {
977 if ( $element === '!--' ) {
978 $end = '/(-->)/';
979 } else {
980 $end = "/(<\\/$element\\s*>)/i";
981 }
982 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
983 $content = $q[0];
984 if ( count( $q ) < 3 ) {
985 # No end tag -- let it run out to the end of the text.
986 $tail = '';
987 $text = '';
988 } else {
989 $tail = $q[1];
990 $text = $q[2];
991 }
992 }
993
994 $matches[$marker] = [ $element,
995 $content,
996 Sanitizer::decodeTagAttributes( $attributes ),
997 "<$element$attributes$close$content$tail" ];
998 }
999 return $stripped;
1000 }
1001
1002 /**
1003 * Get a list of strippable XML-like elements
1004 *
1005 * @return array
1006 */
1007 public function getStripList() {
1008 return $this->mStripList;
1009 }
1010
1011 /**
1012 * Add an item to the strip state
1013 * Returns the unique tag which must be inserted into the stripped text
1014 * The tag will be replaced with the original text in unstrip()
1015 *
1016 * @param string $text
1017 *
1018 * @return string
1019 */
1020 public function insertStripItem( $text ) {
1021 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1022 $this->mMarkerIndex++;
1023 $this->mStripState->addGeneral( $marker, $text );
1024 return $marker;
1025 }
1026
1027 /**
1028 * parse the wiki syntax used to render tables
1029 *
1030 * @private
1031 * @param string $text
1032 * @return string
1033 */
1034 public function doTableStuff( $text ) {
1035
1036 $lines = StringUtils::explode( "\n", $text );
1037 $out = '';
1038 $td_history = []; # Is currently a td tag open?
1039 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1040 $tr_history = []; # Is currently a tr tag open?
1041 $tr_attributes = []; # history of tr attributes
1042 $has_opened_tr = []; # Did this table open a <tr> element?
1043 $indent_level = 0; # indent level of the table
1044
1045 foreach ( $lines as $outLine ) {
1046 $line = trim( $outLine );
1047
1048 if ( $line === '' ) { # empty line, go to next line
1049 $out .= $outLine . "\n";
1050 continue;
1051 }
1052
1053 $first_character = $line[0];
1054 $first_two = substr( $line, 0, 2 );
1055 $matches = [];
1056
1057 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1058 # First check if we are starting a new table
1059 $indent_level = strlen( $matches[1] );
1060
1061 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1062 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1063
1064 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1065 array_push( $td_history, false );
1066 array_push( $last_tag_history, '' );
1067 array_push( $tr_history, false );
1068 array_push( $tr_attributes, '' );
1069 array_push( $has_opened_tr, false );
1070 } elseif ( count( $td_history ) == 0 ) {
1071 # Don't do any of the following
1072 $out .= $outLine . "\n";
1073 continue;
1074 } elseif ( $first_two === '|}' ) {
1075 # We are ending a table
1076 $line = '</table>' . substr( $line, 2 );
1077 $last_tag = array_pop( $last_tag_history );
1078
1079 if ( !array_pop( $has_opened_tr ) ) {
1080 $line = "<tr><td></td></tr>{$line}";
1081 }
1082
1083 if ( array_pop( $tr_history ) ) {
1084 $line = "</tr>{$line}";
1085 }
1086
1087 if ( array_pop( $td_history ) ) {
1088 $line = "</{$last_tag}>{$line}";
1089 }
1090 array_pop( $tr_attributes );
1091 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1092 } elseif ( $first_two === '|-' ) {
1093 # Now we have a table row
1094 $line = preg_replace( '#^\|-+#', '', $line );
1095
1096 # Whats after the tag is now only attributes
1097 $attributes = $this->mStripState->unstripBoth( $line );
1098 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1099 array_pop( $tr_attributes );
1100 array_push( $tr_attributes, $attributes );
1101
1102 $line = '';
1103 $last_tag = array_pop( $last_tag_history );
1104 array_pop( $has_opened_tr );
1105 array_push( $has_opened_tr, true );
1106
1107 if ( array_pop( $tr_history ) ) {
1108 $line = '</tr>';
1109 }
1110
1111 if ( array_pop( $td_history ) ) {
1112 $line = "</{$last_tag}>{$line}";
1113 }
1114
1115 $outLine = $line;
1116 array_push( $tr_history, false );
1117 array_push( $td_history, false );
1118 array_push( $last_tag_history, '' );
1119 } elseif ( $first_character === '|'
1120 || $first_character === '!'
1121 || $first_two === '|+'
1122 ) {
1123 # This might be cell elements, td, th or captions
1124 if ( $first_two === '|+' ) {
1125 $first_character = '+';
1126 $line = substr( $line, 2 );
1127 } else {
1128 $line = substr( $line, 1 );
1129 }
1130
1131 // Implies both are valid for table headings.
1132 if ( $first_character === '!' ) {
1133 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1134 }
1135
1136 # Split up multiple cells on the same line.
1137 # FIXME : This can result in improper nesting of tags processed
1138 # by earlier parser steps.
1139 $cells = explode( '||', $line );
1140
1141 $outLine = '';
1142
1143 # Loop through each table cell
1144 foreach ( $cells as $cell ) {
1145 $previous = '';
1146 if ( $first_character !== '+' ) {
1147 $tr_after = array_pop( $tr_attributes );
1148 if ( !array_pop( $tr_history ) ) {
1149 $previous = "<tr{$tr_after}>\n";
1150 }
1151 array_push( $tr_history, true );
1152 array_push( $tr_attributes, '' );
1153 array_pop( $has_opened_tr );
1154 array_push( $has_opened_tr, true );
1155 }
1156
1157 $last_tag = array_pop( $last_tag_history );
1158
1159 if ( array_pop( $td_history ) ) {
1160 $previous = "</{$last_tag}>\n{$previous}";
1161 }
1162
1163 if ( $first_character === '|' ) {
1164 $last_tag = 'td';
1165 } elseif ( $first_character === '!' ) {
1166 $last_tag = 'th';
1167 } elseif ( $first_character === '+' ) {
1168 $last_tag = 'caption';
1169 } else {
1170 $last_tag = '';
1171 }
1172
1173 array_push( $last_tag_history, $last_tag );
1174
1175 # A cell could contain both parameters and data
1176 $cell_data = explode( '|', $cell, 2 );
1177
1178 # Bug 553: Note that a '|' inside an invalid link should not
1179 # be mistaken as delimiting cell parameters
1180 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1181 $cell = "{$previous}<{$last_tag}>{$cell}";
1182 } elseif ( count( $cell_data ) == 1 ) {
1183 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1184 } else {
1185 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1186 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1187 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1188 }
1189
1190 $outLine .= $cell;
1191 array_push( $td_history, true );
1192 }
1193 }
1194 $out .= $outLine . "\n";
1195 }
1196
1197 # Closing open td, tr && table
1198 while ( count( $td_history ) > 0 ) {
1199 if ( array_pop( $td_history ) ) {
1200 $out .= "</td>\n";
1201 }
1202 if ( array_pop( $tr_history ) ) {
1203 $out .= "</tr>\n";
1204 }
1205 if ( !array_pop( $has_opened_tr ) ) {
1206 $out .= "<tr><td></td></tr>\n";
1207 }
1208
1209 $out .= "</table>\n";
1210 }
1211
1212 # Remove trailing line-ending (b/c)
1213 if ( substr( $out, -1 ) === "\n" ) {
1214 $out = substr( $out, 0, -1 );
1215 }
1216
1217 # special case: don't return empty table
1218 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1219 $out = '';
1220 }
1221
1222 return $out;
1223 }
1224
1225 /**
1226 * Helper function for parse() that transforms wiki markup into half-parsed
1227 * HTML. Only called for $mOutputType == self::OT_HTML.
1228 *
1229 * @private
1230 *
1231 * @param string $text The text to parse
1232 * @param bool $isMain Whether this is being called from the main parse() function
1233 * @param PPFrame|bool $frame A pre-processor frame
1234 *
1235 * @return string
1236 */
1237 public function internalParse( $text, $isMain = true, $frame = false ) {
1238
1239 $origText = $text;
1240
1241 # Hook to suspend the parser in this state
1242 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1243 return $text;
1244 }
1245
1246 # if $frame is provided, then use $frame for replacing any variables
1247 if ( $frame ) {
1248 # use frame depth to infer how include/noinclude tags should be handled
1249 # depth=0 means this is the top-level document; otherwise it's an included document
1250 if ( !$frame->depth ) {
1251 $flag = 0;
1252 } else {
1253 $flag = Parser::PTD_FOR_INCLUSION;
1254 }
1255 $dom = $this->preprocessToDom( $text, $flag );
1256 $text = $frame->expand( $dom );
1257 } else {
1258 # if $frame is not provided, then use old-style replaceVariables
1259 $text = $this->replaceVariables( $text );
1260 }
1261
1262 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1263 $text = Sanitizer::removeHTMLtags(
1264 $text,
1265 [ &$this, 'attributeStripCallback' ],
1266 false,
1267 array_keys( $this->mTransparentTagHooks )
1268 );
1269 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1270
1271 # Tables need to come after variable replacement for things to work
1272 # properly; putting them before other transformations should keep
1273 # exciting things like link expansions from showing up in surprising
1274 # places.
1275 $text = $this->doTableStuff( $text );
1276
1277 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1278
1279 $text = $this->doDoubleUnderscore( $text );
1280
1281 $text = $this->doHeadings( $text );
1282 $text = $this->replaceInternalLinks( $text );
1283 $text = $this->doAllQuotes( $text );
1284 $text = $this->replaceExternalLinks( $text );
1285
1286 # replaceInternalLinks may sometimes leave behind
1287 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1288 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1289
1290 $text = $this->doMagicLinks( $text );
1291 $text = $this->formatHeadings( $text, $origText, $isMain );
1292
1293 return $text;
1294 }
1295
1296 /**
1297 * Helper function for parse() that transforms half-parsed HTML into fully
1298 * parsed HTML.
1299 *
1300 * @param string $text
1301 * @param bool $isMain
1302 * @param bool $linestart
1303 * @return string
1304 */
1305 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1306 $text = $this->mStripState->unstripGeneral( $text );
1307
1308 if ( $isMain ) {
1309 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1310 }
1311
1312 # Clean up special characters, only run once, next-to-last before doBlockLevels
1313 $fixtags = [
1314 # french spaces, last one Guillemet-left
1315 # only if there is something before the space
1316 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1317 # french spaces, Guillemet-right
1318 '/(\\302\\253) /' => '\\1&#160;',
1319 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1320 ];
1321 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1322
1323 $text = $this->doBlockLevels( $text, $linestart );
1324
1325 $this->replaceLinkHolders( $text );
1326
1327 /**
1328 * The input doesn't get language converted if
1329 * a) It's disabled
1330 * b) Content isn't converted
1331 * c) It's a conversion table
1332 * d) it is an interface message (which is in the user language)
1333 */
1334 if ( !( $this->mOptions->getDisableContentConversion()
1335 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1336 ) {
1337 if ( !$this->mOptions->getInterfaceMessage() ) {
1338 # The position of the convert() call should not be changed. it
1339 # assumes that the links are all replaced and the only thing left
1340 # is the <nowiki> mark.
1341 $text = $this->getConverterLanguage()->convert( $text );
1342 }
1343 }
1344
1345 $text = $this->mStripState->unstripNoWiki( $text );
1346
1347 if ( $isMain ) {
1348 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1349 }
1350
1351 $text = $this->replaceTransparentTags( $text );
1352 $text = $this->mStripState->unstripGeneral( $text );
1353
1354 $text = Sanitizer::normalizeCharReferences( $text );
1355
1356 if ( MWTidy::isEnabled() && $this->mOptions->getTidy() ) {
1357 $text = MWTidy::tidy( $text );
1358 $this->mOutput->addModuleStyles( MWTidy::getModuleStyles() );
1359 } else {
1360 # attempt to sanitize at least some nesting problems
1361 # (bug #2702 and quite a few others)
1362 $tidyregs = [
1363 # ''Something [http://www.cool.com cool''] -->
1364 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1365 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1366 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1367 # fix up an anchor inside another anchor, only
1368 # at least for a single single nested link (bug 3695)
1369 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1370 '\\1\\2</a>\\3</a>\\1\\4</a>',
1371 # fix div inside inline elements- doBlockLevels won't wrap a line which
1372 # contains a div, so fix it up here; replace
1373 # div with escaped text
1374 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1375 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1376 # remove empty italic or bold tag pairs, some
1377 # introduced by rules above
1378 '/<([bi])><\/\\1>/' => '',
1379 ];
1380
1381 $text = preg_replace(
1382 array_keys( $tidyregs ),
1383 array_values( $tidyregs ),
1384 $text );
1385 }
1386
1387 if ( $isMain ) {
1388 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1389 }
1390
1391 return $text;
1392 }
1393
1394 /**
1395 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1396 * magic external links.
1397 *
1398 * DML
1399 * @private
1400 *
1401 * @param string $text
1402 *
1403 * @return string
1404 */
1405 public function doMagicLinks( $text ) {
1406 $prots = wfUrlProtocolsWithoutProtRel();
1407 $urlChar = self::EXT_LINK_URL_CLASS;
1408 $addr = self::EXT_LINK_ADDR;
1409 $space = self::SPACE_NOT_NL; # non-newline space
1410 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1411 $spaces = "$space++"; # possessive match of 1 or more spaces
1412 $text = preg_replace_callback(
1413 '!(?: # Start cases
1414 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1415 (<.*?>) | # m[2]: Skip stuff inside
1416 # HTML elements' . "
1417 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1418 # m[4]: Post-protocol path
1419 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1420 ([0-9]+)\b |
1421 \bISBN $spaces ( # m[6]: ISBN, capture number
1422 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1423 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1424 [0-9Xx] # check digit
1425 )\b
1426 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1427 return $text;
1428 }
1429
1430 /**
1431 * @throws MWException
1432 * @param array $m
1433 * @return HTML|string
1434 */
1435 public function magicLinkCallback( $m ) {
1436 if ( isset( $m[1] ) && $m[1] !== '' ) {
1437 # Skip anchor
1438 return $m[0];
1439 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1440 # Skip HTML element
1441 return $m[0];
1442 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1443 # Free external link
1444 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1445 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1446 # RFC or PMID
1447 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1448 $keyword = 'RFC';
1449 $urlmsg = 'rfcurl';
1450 $cssClass = 'mw-magiclink-rfc';
1451 $id = $m[5];
1452 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1453 $keyword = 'PMID';
1454 $urlmsg = 'pubmedurl';
1455 $cssClass = 'mw-magiclink-pmid';
1456 $id = $m[5];
1457 } else {
1458 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1459 substr( $m[0], 0, 20 ) . '"' );
1460 }
1461 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1462 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1463 } elseif ( isset( $m[6] ) && $m[6] !== '' ) {
1464 # ISBN
1465 $isbn = $m[6];
1466 $space = self::SPACE_NOT_NL; # non-newline space
1467 $isbn = preg_replace( "/$space/", ' ', $isbn );
1468 $num = strtr( $isbn, [
1469 '-' => '',
1470 ' ' => '',
1471 'x' => 'X',
1472 ] );
1473 return $this->getLinkRenderer()->makeKnownLink(
1474 SpecialPage::getTitleFor( 'Booksources', $num ),
1475 "ISBN $isbn",
1476 [
1477 'class' => 'internal mw-magiclink-isbn',
1478 'title' => false // suppress title attribute
1479 ]
1480 );
1481 } else {
1482 return $m[0];
1483 }
1484 }
1485
1486 /**
1487 * Make a free external link, given a user-supplied URL
1488 *
1489 * @param string $url
1490 * @param int $numPostProto
1491 * The number of characters after the protocol.
1492 * @return string HTML
1493 * @private
1494 */
1495 public function makeFreeExternalLink( $url, $numPostProto ) {
1496 $trail = '';
1497
1498 # The characters '<' and '>' (which were escaped by
1499 # removeHTMLtags()) should not be included in
1500 # URLs, per RFC 2396.
1501 # Make &nbsp; terminate a URL as well (bug T84937)
1502 $m2 = [];
1503 if ( preg_match(
1504 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1505 $url,
1506 $m2,
1507 PREG_OFFSET_CAPTURE
1508 ) ) {
1509 $trail = substr( $url, $m2[0][1] ) . $trail;
1510 $url = substr( $url, 0, $m2[0][1] );
1511 }
1512
1513 # Move trailing punctuation to $trail
1514 $sep = ',;\.:!?';
1515 # If there is no left bracket, then consider right brackets fair game too
1516 if ( strpos( $url, '(' ) === false ) {
1517 $sep .= ')';
1518 }
1519
1520 $urlRev = strrev( $url );
1521 $numSepChars = strspn( $urlRev, $sep );
1522 # Don't break a trailing HTML entity by moving the ; into $trail
1523 # This is in hot code, so use substr_compare to avoid having to
1524 # create a new string object for the comparison
1525 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1526 # more optimization: instead of running preg_match with a $
1527 # anchor, which can be slow, do the match on the reversed
1528 # string starting at the desired offset.
1529 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1530 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1531 $numSepChars--;
1532 }
1533 }
1534 if ( $numSepChars ) {
1535 $trail = substr( $url, -$numSepChars ) . $trail;
1536 $url = substr( $url, 0, -$numSepChars );
1537 }
1538
1539 # Verify that we still have a real URL after trail removal, and
1540 # not just lone protocol
1541 if ( strlen( $trail ) >= $numPostProto ) {
1542 return $url . $trail;
1543 }
1544
1545 $url = Sanitizer::cleanUrl( $url );
1546
1547 # Is this an external image?
1548 $text = $this->maybeMakeExternalImage( $url );
1549 if ( $text === false ) {
1550 # Not an image, make a link
1551 $text = Linker::makeExternalLink( $url,
1552 $this->getConverterLanguage()->markNoConversion( $url, true ),
1553 true, 'free',
1554 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1555 # Register it in the output object...
1556 # Replace unnecessary URL escape codes with their equivalent characters
1557 $pasteurized = self::normalizeLinkUrl( $url );
1558 $this->mOutput->addExternalLink( $pasteurized );
1559 }
1560 return $text . $trail;
1561 }
1562
1563 /**
1564 * Parse headers and return html
1565 *
1566 * @private
1567 *
1568 * @param string $text
1569 *
1570 * @return string
1571 */
1572 public function doHeadings( $text ) {
1573 for ( $i = 6; $i >= 1; --$i ) {
1574 $h = str_repeat( '=', $i );
1575 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1576 }
1577 return $text;
1578 }
1579
1580 /**
1581 * Replace single quotes with HTML markup
1582 * @private
1583 *
1584 * @param string $text
1585 *
1586 * @return string The altered text
1587 */
1588 public function doAllQuotes( $text ) {
1589 $outtext = '';
1590 $lines = StringUtils::explode( "\n", $text );
1591 foreach ( $lines as $line ) {
1592 $outtext .= $this->doQuotes( $line ) . "\n";
1593 }
1594 $outtext = substr( $outtext, 0, -1 );
1595 return $outtext;
1596 }
1597
1598 /**
1599 * Helper function for doAllQuotes()
1600 *
1601 * @param string $text
1602 *
1603 * @return string
1604 */
1605 public function doQuotes( $text ) {
1606 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1607 $countarr = count( $arr );
1608 if ( $countarr == 1 ) {
1609 return $text;
1610 }
1611
1612 // First, do some preliminary work. This may shift some apostrophes from
1613 // being mark-up to being text. It also counts the number of occurrences
1614 // of bold and italics mark-ups.
1615 $numbold = 0;
1616 $numitalics = 0;
1617 for ( $i = 1; $i < $countarr; $i += 2 ) {
1618 $thislen = strlen( $arr[$i] );
1619 // If there are ever four apostrophes, assume the first is supposed to
1620 // be text, and the remaining three constitute mark-up for bold text.
1621 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1622 if ( $thislen == 4 ) {
1623 $arr[$i - 1] .= "'";
1624 $arr[$i] = "'''";
1625 $thislen = 3;
1626 } elseif ( $thislen > 5 ) {
1627 // If there are more than 5 apostrophes in a row, assume they're all
1628 // text except for the last 5.
1629 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1630 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1631 $arr[$i] = "'''''";
1632 $thislen = 5;
1633 }
1634 // Count the number of occurrences of bold and italics mark-ups.
1635 if ( $thislen == 2 ) {
1636 $numitalics++;
1637 } elseif ( $thislen == 3 ) {
1638 $numbold++;
1639 } elseif ( $thislen == 5 ) {
1640 $numitalics++;
1641 $numbold++;
1642 }
1643 }
1644
1645 // If there is an odd number of both bold and italics, it is likely
1646 // that one of the bold ones was meant to be an apostrophe followed
1647 // by italics. Which one we cannot know for certain, but it is more
1648 // likely to be one that has a single-letter word before it.
1649 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1650 $firstsingleletterword = -1;
1651 $firstmultiletterword = -1;
1652 $firstspace = -1;
1653 for ( $i = 1; $i < $countarr; $i += 2 ) {
1654 if ( strlen( $arr[$i] ) == 3 ) {
1655 $x1 = substr( $arr[$i - 1], -1 );
1656 $x2 = substr( $arr[$i - 1], -2, 1 );
1657 if ( $x1 === ' ' ) {
1658 if ( $firstspace == -1 ) {
1659 $firstspace = $i;
1660 }
1661 } elseif ( $x2 === ' ' ) {
1662 $firstsingleletterword = $i;
1663 // if $firstsingleletterword is set, we don't
1664 // look at the other options, so we can bail early.
1665 break;
1666 } else {
1667 if ( $firstmultiletterword == -1 ) {
1668 $firstmultiletterword = $i;
1669 }
1670 }
1671 }
1672 }
1673
1674 // If there is a single-letter word, use it!
1675 if ( $firstsingleletterword > -1 ) {
1676 $arr[$firstsingleletterword] = "''";
1677 $arr[$firstsingleletterword - 1] .= "'";
1678 } elseif ( $firstmultiletterword > -1 ) {
1679 // If not, but there's a multi-letter word, use that one.
1680 $arr[$firstmultiletterword] = "''";
1681 $arr[$firstmultiletterword - 1] .= "'";
1682 } elseif ( $firstspace > -1 ) {
1683 // ... otherwise use the first one that has neither.
1684 // (notice that it is possible for all three to be -1 if, for example,
1685 // there is only one pentuple-apostrophe in the line)
1686 $arr[$firstspace] = "''";
1687 $arr[$firstspace - 1] .= "'";
1688 }
1689 }
1690
1691 // Now let's actually convert our apostrophic mush to HTML!
1692 $output = '';
1693 $buffer = '';
1694 $state = '';
1695 $i = 0;
1696 foreach ( $arr as $r ) {
1697 if ( ( $i % 2 ) == 0 ) {
1698 if ( $state === 'both' ) {
1699 $buffer .= $r;
1700 } else {
1701 $output .= $r;
1702 }
1703 } else {
1704 $thislen = strlen( $r );
1705 if ( $thislen == 2 ) {
1706 if ( $state === 'i' ) {
1707 $output .= '</i>';
1708 $state = '';
1709 } elseif ( $state === 'bi' ) {
1710 $output .= '</i>';
1711 $state = 'b';
1712 } elseif ( $state === 'ib' ) {
1713 $output .= '</b></i><b>';
1714 $state = 'b';
1715 } elseif ( $state === 'both' ) {
1716 $output .= '<b><i>' . $buffer . '</i>';
1717 $state = 'b';
1718 } else { // $state can be 'b' or ''
1719 $output .= '<i>';
1720 $state .= 'i';
1721 }
1722 } elseif ( $thislen == 3 ) {
1723 if ( $state === 'b' ) {
1724 $output .= '</b>';
1725 $state = '';
1726 } elseif ( $state === 'bi' ) {
1727 $output .= '</i></b><i>';
1728 $state = 'i';
1729 } elseif ( $state === 'ib' ) {
1730 $output .= '</b>';
1731 $state = 'i';
1732 } elseif ( $state === 'both' ) {
1733 $output .= '<i><b>' . $buffer . '</b>';
1734 $state = 'i';
1735 } else { // $state can be 'i' or ''
1736 $output .= '<b>';
1737 $state .= 'b';
1738 }
1739 } elseif ( $thislen == 5 ) {
1740 if ( $state === 'b' ) {
1741 $output .= '</b><i>';
1742 $state = 'i';
1743 } elseif ( $state === 'i' ) {
1744 $output .= '</i><b>';
1745 $state = 'b';
1746 } elseif ( $state === 'bi' ) {
1747 $output .= '</i></b>';
1748 $state = '';
1749 } elseif ( $state === 'ib' ) {
1750 $output .= '</b></i>';
1751 $state = '';
1752 } elseif ( $state === 'both' ) {
1753 $output .= '<i><b>' . $buffer . '</b></i>';
1754 $state = '';
1755 } else { // ($state == '')
1756 $buffer = '';
1757 $state = 'both';
1758 }
1759 }
1760 }
1761 $i++;
1762 }
1763 // Now close all remaining tags. Notice that the order is important.
1764 if ( $state === 'b' || $state === 'ib' ) {
1765 $output .= '</b>';
1766 }
1767 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1768 $output .= '</i>';
1769 }
1770 if ( $state === 'bi' ) {
1771 $output .= '</b>';
1772 }
1773 // There might be lonely ''''', so make sure we have a buffer
1774 if ( $state === 'both' && $buffer ) {
1775 $output .= '<b><i>' . $buffer . '</i></b>';
1776 }
1777 return $output;
1778 }
1779
1780 /**
1781 * Replace external links (REL)
1782 *
1783 * Note: this is all very hackish and the order of execution matters a lot.
1784 * Make sure to run tests/parserTests.php if you change this code.
1785 *
1786 * @private
1787 *
1788 * @param string $text
1789 *
1790 * @throws MWException
1791 * @return string
1792 */
1793 public function replaceExternalLinks( $text ) {
1794
1795 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1796 if ( $bits === false ) {
1797 throw new MWException( "PCRE needs to be compiled with "
1798 . "--enable-unicode-properties in order for MediaWiki to function" );
1799 }
1800 $s = array_shift( $bits );
1801
1802 $i = 0;
1803 while ( $i < count( $bits ) ) {
1804 $url = $bits[$i++];
1805 $i++; // protocol
1806 $text = $bits[$i++];
1807 $trail = $bits[$i++];
1808
1809 # The characters '<' and '>' (which were escaped by
1810 # removeHTMLtags()) should not be included in
1811 # URLs, per RFC 2396.
1812 $m2 = [];
1813 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1814 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1815 $url = substr( $url, 0, $m2[0][1] );
1816 }
1817
1818 # If the link text is an image URL, replace it with an <img> tag
1819 # This happened by accident in the original parser, but some people used it extensively
1820 $img = $this->maybeMakeExternalImage( $text );
1821 if ( $img !== false ) {
1822 $text = $img;
1823 }
1824
1825 $dtrail = '';
1826
1827 # Set linktype for CSS - if URL==text, link is essentially free
1828 $linktype = ( $text === $url ) ? 'free' : 'text';
1829
1830 # No link text, e.g. [http://domain.tld/some.link]
1831 if ( $text == '' ) {
1832 # Autonumber
1833 $langObj = $this->getTargetLanguage();
1834 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1835 $linktype = 'autonumber';
1836 } else {
1837 # Have link text, e.g. [http://domain.tld/some.link text]s
1838 # Check for trail
1839 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1840 }
1841
1842 $text = $this->getConverterLanguage()->markNoConversion( $text );
1843
1844 $url = Sanitizer::cleanUrl( $url );
1845
1846 # Use the encoded URL
1847 # This means that users can paste URLs directly into the text
1848 # Funny characters like ö aren't valid in URLs anyway
1849 # This was changed in August 2004
1850 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1851 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1852
1853 # Register link in the output object.
1854 # Replace unnecessary URL escape codes with the referenced character
1855 # This prevents spammers from hiding links from the filters
1856 $pasteurized = self::normalizeLinkUrl( $url );
1857 $this->mOutput->addExternalLink( $pasteurized );
1858 }
1859
1860 return $s;
1861 }
1862
1863 /**
1864 * Get the rel attribute for a particular external link.
1865 *
1866 * @since 1.21
1867 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1868 * nofollow if appropriate
1869 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1870 * @return string|null Rel attribute for $url
1871 */
1872 public static function getExternalLinkRel( $url = false, $title = null ) {
1873 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1874 $ns = $title ? $title->getNamespace() : false;
1875 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1876 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1877 ) {
1878 return 'nofollow';
1879 }
1880 return null;
1881 }
1882
1883 /**
1884 * Get an associative array of additional HTML attributes appropriate for a
1885 * particular external link. This currently may include rel => nofollow
1886 * (depending on configuration, namespace, and the URL's domain) and/or a
1887 * target attribute (depending on configuration).
1888 *
1889 * @param string $url URL to extract the domain from for rel =>
1890 * nofollow if appropriate
1891 * @return array Associative array of HTML attributes
1892 */
1893 public function getExternalLinkAttribs( $url ) {
1894 $attribs = [];
1895 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1896
1897 $target = $this->mOptions->getExternalLinkTarget();
1898 if ( $target ) {
1899 $attribs['target'] = $target;
1900 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1901 // T133507. New windows can navigate parent cross-origin.
1902 // Including noreferrer due to lacking browser
1903 // support of noopener. Eventually noreferrer should be removed.
1904 if ( $rel !== '' ) {
1905 $rel .= ' ';
1906 }
1907 $rel .= 'noreferrer noopener';
1908 }
1909 }
1910 $attribs['rel'] = $rel;
1911 return $attribs;
1912 }
1913
1914 /**
1915 * Replace unusual escape codes in a URL with their equivalent characters
1916 *
1917 * @deprecated since 1.24, use normalizeLinkUrl
1918 * @param string $url
1919 * @return string
1920 */
1921 public static function replaceUnusualEscapes( $url ) {
1922 wfDeprecated( __METHOD__, '1.24' );
1923 return self::normalizeLinkUrl( $url );
1924 }
1925
1926 /**
1927 * Replace unusual escape codes in a URL with their equivalent characters
1928 *
1929 * This generally follows the syntax defined in RFC 3986, with special
1930 * consideration for HTTP query strings.
1931 *
1932 * @param string $url
1933 * @return string
1934 */
1935 public static function normalizeLinkUrl( $url ) {
1936 # First, make sure unsafe characters are encoded
1937 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1938 function ( $m ) {
1939 return rawurlencode( $m[0] );
1940 },
1941 $url
1942 );
1943
1944 $ret = '';
1945 $end = strlen( $url );
1946
1947 # Fragment part - 'fragment'
1948 $start = strpos( $url, '#' );
1949 if ( $start !== false && $start < $end ) {
1950 $ret = self::normalizeUrlComponent(
1951 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1952 $end = $start;
1953 }
1954
1955 # Query part - 'query' minus &=+;
1956 $start = strpos( $url, '?' );
1957 if ( $start !== false && $start < $end ) {
1958 $ret = self::normalizeUrlComponent(
1959 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1960 $end = $start;
1961 }
1962
1963 # Scheme and path part - 'pchar'
1964 # (we assume no userinfo or encoded colons in the host)
1965 $ret = self::normalizeUrlComponent(
1966 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1967
1968 return $ret;
1969 }
1970
1971 private static function normalizeUrlComponent( $component, $unsafe ) {
1972 $callback = function ( $matches ) use ( $unsafe ) {
1973 $char = urldecode( $matches[0] );
1974 $ord = ord( $char );
1975 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1976 # Unescape it
1977 return $char;
1978 } else {
1979 # Leave it escaped, but use uppercase for a-f
1980 return strtoupper( $matches[0] );
1981 }
1982 };
1983 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1984 }
1985
1986 /**
1987 * make an image if it's allowed, either through the global
1988 * option, through the exception, or through the on-wiki whitelist
1989 *
1990 * @param string $url
1991 *
1992 * @return string
1993 */
1994 private function maybeMakeExternalImage( $url ) {
1995 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1996 $imagesexception = !empty( $imagesfrom );
1997 $text = false;
1998 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1999 if ( $imagesexception && is_array( $imagesfrom ) ) {
2000 $imagematch = false;
2001 foreach ( $imagesfrom as $match ) {
2002 if ( strpos( $url, $match ) === 0 ) {
2003 $imagematch = true;
2004 break;
2005 }
2006 }
2007 } elseif ( $imagesexception ) {
2008 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2009 } else {
2010 $imagematch = false;
2011 }
2012
2013 if ( $this->mOptions->getAllowExternalImages()
2014 || ( $imagesexception && $imagematch )
2015 ) {
2016 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2017 # Image found
2018 $text = Linker::makeExternalImage( $url );
2019 }
2020 }
2021 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2022 && preg_match( self::EXT_IMAGE_REGEX, $url )
2023 ) {
2024 $whitelist = explode(
2025 "\n",
2026 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2027 );
2028
2029 foreach ( $whitelist as $entry ) {
2030 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2031 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2032 continue;
2033 }
2034 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2035 # Image matches a whitelist entry
2036 $text = Linker::makeExternalImage( $url );
2037 break;
2038 }
2039 }
2040 }
2041 return $text;
2042 }
2043
2044 /**
2045 * Process [[ ]] wikilinks
2046 *
2047 * @param string $s
2048 *
2049 * @return string Processed text
2050 *
2051 * @private
2052 */
2053 public function replaceInternalLinks( $s ) {
2054 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2055 return $s;
2056 }
2057
2058 /**
2059 * Process [[ ]] wikilinks (RIL)
2060 * @param string $s
2061 * @throws MWException
2062 * @return LinkHolderArray
2063 *
2064 * @private
2065 */
2066 public function replaceInternalLinks2( &$s ) {
2067 global $wgExtraInterlanguageLinkPrefixes;
2068
2069 static $tc = false, $e1, $e1_img;
2070 # the % is needed to support urlencoded titles as well
2071 if ( !$tc ) {
2072 $tc = Title::legalChars() . '#%';
2073 # Match a link having the form [[namespace:link|alternate]]trail
2074 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2075 # Match cases where there is no "]]", which might still be images
2076 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2077 }
2078
2079 $holders = new LinkHolderArray( $this );
2080
2081 # split the entire text string on occurrences of [[
2082 $a = StringUtils::explode( '[[', ' ' . $s );
2083 # get the first element (all text up to first [[), and remove the space we added
2084 $s = $a->current();
2085 $a->next();
2086 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2087 $s = substr( $s, 1 );
2088
2089 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2090 $e2 = null;
2091 if ( $useLinkPrefixExtension ) {
2092 # Match the end of a line for a word that's not followed by whitespace,
2093 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2094 global $wgContLang;
2095 $charset = $wgContLang->linkPrefixCharset();
2096 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2097 }
2098
2099 if ( is_null( $this->mTitle ) ) {
2100 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2101 }
2102 $nottalk = !$this->mTitle->isTalkPage();
2103
2104 if ( $useLinkPrefixExtension ) {
2105 $m = [];
2106 if ( preg_match( $e2, $s, $m ) ) {
2107 $first_prefix = $m[2];
2108 } else {
2109 $first_prefix = false;
2110 }
2111 } else {
2112 $prefix = '';
2113 }
2114
2115 $useSubpages = $this->areSubpagesAllowed();
2116
2117 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2118 # Loop for each link
2119 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2120 // @codingStandardsIgnoreEnd
2121
2122 # Check for excessive memory usage
2123 if ( $holders->isBig() ) {
2124 # Too big
2125 # Do the existence check, replace the link holders and clear the array
2126 $holders->replace( $s );
2127 $holders->clear();
2128 }
2129
2130 if ( $useLinkPrefixExtension ) {
2131 if ( preg_match( $e2, $s, $m ) ) {
2132 $prefix = $m[2];
2133 $s = $m[1];
2134 } else {
2135 $prefix = '';
2136 }
2137 # first link
2138 if ( $first_prefix ) {
2139 $prefix = $first_prefix;
2140 $first_prefix = false;
2141 }
2142 }
2143
2144 $might_be_img = false;
2145
2146 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2147 $text = $m[2];
2148 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2149 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2150 # the real problem is with the $e1 regex
2151 # See bug 1300.
2152 # Still some problems for cases where the ] is meant to be outside punctuation,
2153 # and no image is in sight. See bug 2095.
2154 if ( $text !== ''
2155 && substr( $m[3], 0, 1 ) === ']'
2156 && strpos( $text, '[' ) !== false
2157 ) {
2158 $text .= ']'; # so that replaceExternalLinks($text) works later
2159 $m[3] = substr( $m[3], 1 );
2160 }
2161 # fix up urlencoded title texts
2162 if ( strpos( $m[1], '%' ) !== false ) {
2163 # Should anchors '#' also be rejected?
2164 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2165 }
2166 $trail = $m[3];
2167 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2168 # Invalid, but might be an image with a link in its caption
2169 $might_be_img = true;
2170 $text = $m[2];
2171 if ( strpos( $m[1], '%' ) !== false ) {
2172 $m[1] = rawurldecode( $m[1] );
2173 }
2174 $trail = "";
2175 } else { # Invalid form; output directly
2176 $s .= $prefix . '[[' . $line;
2177 continue;
2178 }
2179
2180 $origLink = $m[1];
2181
2182 # Don't allow internal links to pages containing
2183 # PROTO: where PROTO is a valid URL protocol; these
2184 # should be external links.
2185 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2186 $s .= $prefix . '[[' . $line;
2187 continue;
2188 }
2189
2190 # Make subpage if necessary
2191 if ( $useSubpages ) {
2192 $link = $this->maybeDoSubpageLink( $origLink, $text );
2193 } else {
2194 $link = $origLink;
2195 }
2196
2197 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2198 if ( !$noforce ) {
2199 # Strip off leading ':'
2200 $link = substr( $link, 1 );
2201 }
2202
2203 $unstrip = $this->mStripState->unstripNoWiki( $link );
2204 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2205 if ( $nt === null ) {
2206 $s .= $prefix . '[[' . $line;
2207 continue;
2208 }
2209
2210 $ns = $nt->getNamespace();
2211 $iw = $nt->getInterwiki();
2212
2213 if ( $might_be_img ) { # if this is actually an invalid link
2214 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2215 $found = false;
2216 while ( true ) {
2217 # look at the next 'line' to see if we can close it there
2218 $a->next();
2219 $next_line = $a->current();
2220 if ( $next_line === false || $next_line === null ) {
2221 break;
2222 }
2223 $m = explode( ']]', $next_line, 3 );
2224 if ( count( $m ) == 3 ) {
2225 # the first ]] closes the inner link, the second the image
2226 $found = true;
2227 $text .= "[[{$m[0]}]]{$m[1]}";
2228 $trail = $m[2];
2229 break;
2230 } elseif ( count( $m ) == 2 ) {
2231 # if there's exactly one ]] that's fine, we'll keep looking
2232 $text .= "[[{$m[0]}]]{$m[1]}";
2233 } else {
2234 # if $next_line is invalid too, we need look no further
2235 $text .= '[[' . $next_line;
2236 break;
2237 }
2238 }
2239 if ( !$found ) {
2240 # we couldn't find the end of this imageLink, so output it raw
2241 # but don't ignore what might be perfectly normal links in the text we've examined
2242 $holders->merge( $this->replaceInternalLinks2( $text ) );
2243 $s .= "{$prefix}[[$link|$text";
2244 # note: no $trail, because without an end, there *is* no trail
2245 continue;
2246 }
2247 } else { # it's not an image, so output it raw
2248 $s .= "{$prefix}[[$link|$text";
2249 # note: no $trail, because without an end, there *is* no trail
2250 continue;
2251 }
2252 }
2253
2254 $wasblank = ( $text == '' );
2255 if ( $wasblank ) {
2256 $text = $link;
2257 } else {
2258 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2259 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2260 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2261 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2262 $text = $this->doQuotes( $text );
2263 }
2264
2265 # Link not escaped by : , create the various objects
2266 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2267 # Interwikis
2268 if (
2269 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2270 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2271 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2272 )
2273 ) {
2274 # Bug 24502: filter duplicates
2275 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2276 $this->mLangLinkLanguages[$iw] = true;
2277 $this->mOutput->addLanguageLink( $nt->getFullText() );
2278 }
2279
2280 $s = rtrim( $s . $prefix );
2281 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2282 continue;
2283 }
2284
2285 if ( $ns == NS_FILE ) {
2286 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2287 if ( $wasblank ) {
2288 # if no parameters were passed, $text
2289 # becomes something like "File:Foo.png",
2290 # which we don't want to pass on to the
2291 # image generator
2292 $text = '';
2293 } else {
2294 # recursively parse links inside the image caption
2295 # actually, this will parse them in any other parameters, too,
2296 # but it might be hard to fix that, and it doesn't matter ATM
2297 $text = $this->replaceExternalLinks( $text );
2298 $holders->merge( $this->replaceInternalLinks2( $text ) );
2299 }
2300 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2301 $s .= $prefix . $this->armorLinks(
2302 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2303 continue;
2304 }
2305 } elseif ( $ns == NS_CATEGORY ) {
2306 $s = rtrim( $s . "\n" ); # bug 87
2307
2308 if ( $wasblank ) {
2309 $sortkey = $this->getDefaultSort();
2310 } else {
2311 $sortkey = $text;
2312 }
2313 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2314 $sortkey = str_replace( "\n", '', $sortkey );
2315 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2316 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2317
2318 /**
2319 * Strip the whitespace Category links produce, see bug 87
2320 */
2321 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2322
2323 continue;
2324 }
2325 }
2326
2327 # Self-link checking. For some languages, variants of the title are checked in
2328 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2329 # for linking to a different variant.
2330 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2331 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2332 continue;
2333 }
2334
2335 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2336 # @todo FIXME: Should do batch file existence checks, see comment below
2337 if ( $ns == NS_MEDIA ) {
2338 # Give extensions a chance to select the file revision for us
2339 $options = [];
2340 $descQuery = false;
2341 Hooks::run( 'BeforeParserFetchFileAndTitle',
2342 [ $this, $nt, &$options, &$descQuery ] );
2343 # Fetch and register the file (file title may be different via hooks)
2344 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2345 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2346 $s .= $prefix . $this->armorLinks(
2347 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2348 continue;
2349 }
2350
2351 # Some titles, such as valid special pages or files in foreign repos, should
2352 # be shown as bluelinks even though they're not included in the page table
2353 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2354 # batch file existence checks for NS_FILE and NS_MEDIA
2355 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2356 $this->mOutput->addLink( $nt );
2357 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2358 } else {
2359 # Links will be added to the output link list after checking
2360 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2361 }
2362 }
2363 return $holders;
2364 }
2365
2366 /**
2367 * Render a forced-blue link inline; protect against double expansion of
2368 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2369 * Since this little disaster has to split off the trail text to avoid
2370 * breaking URLs in the following text without breaking trails on the
2371 * wiki links, it's been made into a horrible function.
2372 *
2373 * @param Title $nt
2374 * @param string $text
2375 * @param string $trail
2376 * @param string $prefix
2377 * @return string HTML-wikitext mix oh yuck
2378 */
2379 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2380 list( $inside, $trail ) = Linker::splitTrail( $trail );
2381
2382 if ( $text == '' ) {
2383 $text = htmlspecialchars( $nt->getPrefixedText() );
2384 }
2385
2386 $link = $this->getLinkRenderer()->makeKnownLink(
2387 $nt, new HtmlArmor( "$prefix$text$inside" )
2388 );
2389
2390 return $this->armorLinks( $link ) . $trail;
2391 }
2392
2393 /**
2394 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2395 * going to go through further parsing steps before inline URL expansion.
2396 *
2397 * Not needed quite as much as it used to be since free links are a bit
2398 * more sensible these days. But bracketed links are still an issue.
2399 *
2400 * @param string $text More-or-less HTML
2401 * @return string Less-or-more HTML with NOPARSE bits
2402 */
2403 public function armorLinks( $text ) {
2404 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2405 self::MARKER_PREFIX . "NOPARSE$1", $text );
2406 }
2407
2408 /**
2409 * Return true if subpage links should be expanded on this page.
2410 * @return bool
2411 */
2412 public function areSubpagesAllowed() {
2413 # Some namespaces don't allow subpages
2414 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2415 }
2416
2417 /**
2418 * Handle link to subpage if necessary
2419 *
2420 * @param string $target The source of the link
2421 * @param string &$text The link text, modified as necessary
2422 * @return string The full name of the link
2423 * @private
2424 */
2425 public function maybeDoSubpageLink( $target, &$text ) {
2426 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2427 }
2428
2429 /**
2430 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2431 *
2432 * @param string $text
2433 * @param bool $linestart Whether or not this is at the start of a line.
2434 * @private
2435 * @return string The lists rendered as HTML
2436 */
2437 public function doBlockLevels( $text, $linestart ) {
2438 return BlockLevelPass::doBlockLevels( $text, $linestart );
2439 }
2440
2441 /**
2442 * Return value of a magic variable (like PAGENAME)
2443 *
2444 * @private
2445 *
2446 * @param int $index
2447 * @param bool|PPFrame $frame
2448 *
2449 * @throws MWException
2450 * @return string
2451 */
2452 public function getVariableValue( $index, $frame = false ) {
2453 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2454 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2455
2456 if ( is_null( $this->mTitle ) ) {
2457 // If no title set, bad things are going to happen
2458 // later. Title should always be set since this
2459 // should only be called in the middle of a parse
2460 // operation (but the unit-tests do funky stuff)
2461 throw new MWException( __METHOD__ . ' Should only be '
2462 . ' called while parsing (no title set)' );
2463 }
2464
2465 /**
2466 * Some of these require message or data lookups and can be
2467 * expensive to check many times.
2468 */
2469 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2470 if ( isset( $this->mVarCache[$index] ) ) {
2471 return $this->mVarCache[$index];
2472 }
2473 }
2474
2475 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2476 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2477
2478 $pageLang = $this->getFunctionLang();
2479
2480 switch ( $index ) {
2481 case '!':
2482 $value = '|';
2483 break;
2484 case 'currentmonth':
2485 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2486 break;
2487 case 'currentmonth1':
2488 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2489 break;
2490 case 'currentmonthname':
2491 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2492 break;
2493 case 'currentmonthnamegen':
2494 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2495 break;
2496 case 'currentmonthabbrev':
2497 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2498 break;
2499 case 'currentday':
2500 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2501 break;
2502 case 'currentday2':
2503 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2504 break;
2505 case 'localmonth':
2506 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2507 break;
2508 case 'localmonth1':
2509 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2510 break;
2511 case 'localmonthname':
2512 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2513 break;
2514 case 'localmonthnamegen':
2515 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2516 break;
2517 case 'localmonthabbrev':
2518 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2519 break;
2520 case 'localday':
2521 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2522 break;
2523 case 'localday2':
2524 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2525 break;
2526 case 'pagename':
2527 $value = wfEscapeWikiText( $this->mTitle->getText() );
2528 break;
2529 case 'pagenamee':
2530 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2531 break;
2532 case 'fullpagename':
2533 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2534 break;
2535 case 'fullpagenamee':
2536 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2537 break;
2538 case 'subpagename':
2539 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2540 break;
2541 case 'subpagenamee':
2542 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2543 break;
2544 case 'rootpagename':
2545 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2546 break;
2547 case 'rootpagenamee':
2548 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2549 ' ',
2550 '_',
2551 $this->mTitle->getRootText()
2552 ) ) );
2553 break;
2554 case 'basepagename':
2555 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2556 break;
2557 case 'basepagenamee':
2558 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2559 ' ',
2560 '_',
2561 $this->mTitle->getBaseText()
2562 ) ) );
2563 break;
2564 case 'talkpagename':
2565 if ( $this->mTitle->canTalk() ) {
2566 $talkPage = $this->mTitle->getTalkPage();
2567 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2568 } else {
2569 $value = '';
2570 }
2571 break;
2572 case 'talkpagenamee':
2573 if ( $this->mTitle->canTalk() ) {
2574 $talkPage = $this->mTitle->getTalkPage();
2575 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2576 } else {
2577 $value = '';
2578 }
2579 break;
2580 case 'subjectpagename':
2581 $subjPage = $this->mTitle->getSubjectPage();
2582 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2583 break;
2584 case 'subjectpagenamee':
2585 $subjPage = $this->mTitle->getSubjectPage();
2586 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2587 break;
2588 case 'pageid': // requested in bug 23427
2589 $pageid = $this->getTitle()->getArticleID();
2590 if ( $pageid == 0 ) {
2591 # 0 means the page doesn't exist in the database,
2592 # which means the user is previewing a new page.
2593 # The vary-revision flag must be set, because the magic word
2594 # will have a different value once the page is saved.
2595 $this->mOutput->setFlag( 'vary-revision' );
2596 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2597 }
2598 $value = $pageid ? $pageid : null;
2599 break;
2600 case 'revisionid':
2601 # Let the edit saving system know we should parse the page
2602 # *after* a revision ID has been assigned.
2603 $this->mOutput->setFlag( 'vary-revision-id' );
2604 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2605 $value = $this->mRevisionId;
2606 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2607 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2608 $this->mOutput->setSpeculativeRevIdUsed( $value );
2609 }
2610 break;
2611 case 'revisionday':
2612 # Let the edit saving system know we should parse the page
2613 # *after* a revision ID has been assigned. This is for null edits.
2614 $this->mOutput->setFlag( 'vary-revision' );
2615 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2616 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2617 break;
2618 case 'revisionday2':
2619 # Let the edit saving system know we should parse the page
2620 # *after* a revision ID has been assigned. This is for null edits.
2621 $this->mOutput->setFlag( 'vary-revision' );
2622 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2623 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2624 break;
2625 case 'revisionmonth':
2626 # Let the edit saving system know we should parse the page
2627 # *after* a revision ID has been assigned. This is for null edits.
2628 $this->mOutput->setFlag( 'vary-revision' );
2629 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2630 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2631 break;
2632 case 'revisionmonth1':
2633 # Let the edit saving system know we should parse the page
2634 # *after* a revision ID has been assigned. This is for null edits.
2635 $this->mOutput->setFlag( 'vary-revision' );
2636 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2637 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2638 break;
2639 case 'revisionyear':
2640 # Let the edit saving system know we should parse the page
2641 # *after* a revision ID has been assigned. This is for null edits.
2642 $this->mOutput->setFlag( 'vary-revision' );
2643 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2644 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2645 break;
2646 case 'revisiontimestamp':
2647 # Let the edit saving system know we should parse the page
2648 # *after* a revision ID has been assigned. This is for null edits.
2649 $this->mOutput->setFlag( 'vary-revision' );
2650 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2651 $value = $this->getRevisionTimestamp();
2652 break;
2653 case 'revisionuser':
2654 # Let the edit saving system know we should parse the page
2655 # *after* a revision ID has been assigned for null edits.
2656 $this->mOutput->setFlag( 'vary-user' );
2657 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2658 $value = $this->getRevisionUser();
2659 break;
2660 case 'revisionsize':
2661 $value = $this->getRevisionSize();
2662 break;
2663 case 'namespace':
2664 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2665 break;
2666 case 'namespacee':
2667 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2668 break;
2669 case 'namespacenumber':
2670 $value = $this->mTitle->getNamespace();
2671 break;
2672 case 'talkspace':
2673 $value = $this->mTitle->canTalk()
2674 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2675 : '';
2676 break;
2677 case 'talkspacee':
2678 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2679 break;
2680 case 'subjectspace':
2681 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2682 break;
2683 case 'subjectspacee':
2684 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2685 break;
2686 case 'currentdayname':
2687 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2688 break;
2689 case 'currentyear':
2690 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2691 break;
2692 case 'currenttime':
2693 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2694 break;
2695 case 'currenthour':
2696 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2697 break;
2698 case 'currentweek':
2699 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2700 # int to remove the padding
2701 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2702 break;
2703 case 'currentdow':
2704 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2705 break;
2706 case 'localdayname':
2707 $value = $pageLang->getWeekdayName(
2708 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2709 );
2710 break;
2711 case 'localyear':
2712 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2713 break;
2714 case 'localtime':
2715 $value = $pageLang->time(
2716 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2717 false,
2718 false
2719 );
2720 break;
2721 case 'localhour':
2722 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2723 break;
2724 case 'localweek':
2725 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2726 # int to remove the padding
2727 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2728 break;
2729 case 'localdow':
2730 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2731 break;
2732 case 'numberofarticles':
2733 $value = $pageLang->formatNum( SiteStats::articles() );
2734 break;
2735 case 'numberoffiles':
2736 $value = $pageLang->formatNum( SiteStats::images() );
2737 break;
2738 case 'numberofusers':
2739 $value = $pageLang->formatNum( SiteStats::users() );
2740 break;
2741 case 'numberofactiveusers':
2742 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2743 break;
2744 case 'numberofpages':
2745 $value = $pageLang->formatNum( SiteStats::pages() );
2746 break;
2747 case 'numberofadmins':
2748 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2749 break;
2750 case 'numberofedits':
2751 $value = $pageLang->formatNum( SiteStats::edits() );
2752 break;
2753 case 'currenttimestamp':
2754 $value = wfTimestamp( TS_MW, $ts );
2755 break;
2756 case 'localtimestamp':
2757 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2758 break;
2759 case 'currentversion':
2760 $value = SpecialVersion::getVersion();
2761 break;
2762 case 'articlepath':
2763 return $wgArticlePath;
2764 case 'sitename':
2765 return $wgSitename;
2766 case 'server':
2767 return $wgServer;
2768 case 'servername':
2769 return $wgServerName;
2770 case 'scriptpath':
2771 return $wgScriptPath;
2772 case 'stylepath':
2773 return $wgStylePath;
2774 case 'directionmark':
2775 return $pageLang->getDirMark();
2776 case 'contentlanguage':
2777 global $wgLanguageCode;
2778 return $wgLanguageCode;
2779 case 'cascadingsources':
2780 $value = CoreParserFunctions::cascadingsources( $this );
2781 break;
2782 default:
2783 $ret = null;
2784 Hooks::run(
2785 'ParserGetVariableValueSwitch',
2786 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2787 );
2788
2789 return $ret;
2790 }
2791
2792 if ( $index ) {
2793 $this->mVarCache[$index] = $value;
2794 }
2795
2796 return $value;
2797 }
2798
2799 /**
2800 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2801 *
2802 * @private
2803 */
2804 public function initialiseVariables() {
2805 $variableIDs = MagicWord::getVariableIDs();
2806 $substIDs = MagicWord::getSubstIDs();
2807
2808 $this->mVariables = new MagicWordArray( $variableIDs );
2809 $this->mSubstWords = new MagicWordArray( $substIDs );
2810 }
2811
2812 /**
2813 * Preprocess some wikitext and return the document tree.
2814 * This is the ghost of replace_variables().
2815 *
2816 * @param string $text The text to parse
2817 * @param int $flags Bitwise combination of:
2818 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2819 * included. Default is to assume a direct page view.
2820 *
2821 * The generated DOM tree must depend only on the input text and the flags.
2822 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2823 *
2824 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2825 * change in the DOM tree for a given text, must be passed through the section identifier
2826 * in the section edit link and thus back to extractSections().
2827 *
2828 * The output of this function is currently only cached in process memory, but a persistent
2829 * cache may be implemented at a later date which takes further advantage of these strict
2830 * dependency requirements.
2831 *
2832 * @return PPNode
2833 */
2834 public function preprocessToDom( $text, $flags = 0 ) {
2835 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2836 return $dom;
2837 }
2838
2839 /**
2840 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2841 *
2842 * @param string $s
2843 *
2844 * @return array
2845 */
2846 public static function splitWhitespace( $s ) {
2847 $ltrimmed = ltrim( $s );
2848 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2849 $trimmed = rtrim( $ltrimmed );
2850 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2851 if ( $diff > 0 ) {
2852 $w2 = substr( $ltrimmed, -$diff );
2853 } else {
2854 $w2 = '';
2855 }
2856 return [ $w1, $trimmed, $w2 ];
2857 }
2858
2859 /**
2860 * Replace magic variables, templates, and template arguments
2861 * with the appropriate text. Templates are substituted recursively,
2862 * taking care to avoid infinite loops.
2863 *
2864 * Note that the substitution depends on value of $mOutputType:
2865 * self::OT_WIKI: only {{subst:}} templates
2866 * self::OT_PREPROCESS: templates but not extension tags
2867 * self::OT_HTML: all templates and extension tags
2868 *
2869 * @param string $text The text to transform
2870 * @param bool|PPFrame $frame Object describing the arguments passed to the
2871 * template. Arguments may also be provided as an associative array, as
2872 * was the usual case before MW1.12. Providing arguments this way may be
2873 * useful for extensions wishing to perform variable replacement
2874 * explicitly.
2875 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2876 * double-brace expansion.
2877 * @return string
2878 */
2879 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2880 # Is there any text? Also, Prevent too big inclusions!
2881 $textSize = strlen( $text );
2882 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2883 return $text;
2884 }
2885
2886 if ( $frame === false ) {
2887 $frame = $this->getPreprocessor()->newFrame();
2888 } elseif ( !( $frame instanceof PPFrame ) ) {
2889 wfDebug( __METHOD__ . " called using plain parameters instead of "
2890 . "a PPFrame instance. Creating custom frame.\n" );
2891 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2892 }
2893
2894 $dom = $this->preprocessToDom( $text );
2895 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2896 $text = $frame->expand( $dom, $flags );
2897
2898 return $text;
2899 }
2900
2901 /**
2902 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2903 *
2904 * @param array $args
2905 *
2906 * @return array
2907 */
2908 public static function createAssocArgs( $args ) {
2909 $assocArgs = [];
2910 $index = 1;
2911 foreach ( $args as $arg ) {
2912 $eqpos = strpos( $arg, '=' );
2913 if ( $eqpos === false ) {
2914 $assocArgs[$index++] = $arg;
2915 } else {
2916 $name = trim( substr( $arg, 0, $eqpos ) );
2917 $value = trim( substr( $arg, $eqpos + 1 ) );
2918 if ( $value === false ) {
2919 $value = '';
2920 }
2921 if ( $name !== false ) {
2922 $assocArgs[$name] = $value;
2923 }
2924 }
2925 }
2926
2927 return $assocArgs;
2928 }
2929
2930 /**
2931 * Warn the user when a parser limitation is reached
2932 * Will warn at most once the user per limitation type
2933 *
2934 * The results are shown during preview and run through the Parser (See EditPage.php)
2935 *
2936 * @param string $limitationType Should be one of:
2937 * 'expensive-parserfunction' (corresponding messages:
2938 * 'expensive-parserfunction-warning',
2939 * 'expensive-parserfunction-category')
2940 * 'post-expand-template-argument' (corresponding messages:
2941 * 'post-expand-template-argument-warning',
2942 * 'post-expand-template-argument-category')
2943 * 'post-expand-template-inclusion' (corresponding messages:
2944 * 'post-expand-template-inclusion-warning',
2945 * 'post-expand-template-inclusion-category')
2946 * 'node-count-exceeded' (corresponding messages:
2947 * 'node-count-exceeded-warning',
2948 * 'node-count-exceeded-category')
2949 * 'expansion-depth-exceeded' (corresponding messages:
2950 * 'expansion-depth-exceeded-warning',
2951 * 'expansion-depth-exceeded-category')
2952 * @param string|int|null $current Current value
2953 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2954 * exceeded, provide the values (optional)
2955 */
2956 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2957 # does no harm if $current and $max are present but are unnecessary for the message
2958 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2959 # only during preview, and that would split the parser cache unnecessarily.
2960 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2961 ->text();
2962 $this->mOutput->addWarning( $warning );
2963 $this->addTrackingCategory( "$limitationType-category" );
2964 }
2965
2966 /**
2967 * Return the text of a template, after recursively
2968 * replacing any variables or templates within the template.
2969 *
2970 * @param array $piece The parts of the template
2971 * $piece['title']: the title, i.e. the part before the |
2972 * $piece['parts']: the parameter array
2973 * $piece['lineStart']: whether the brace was at the start of a line
2974 * @param PPFrame $frame The current frame, contains template arguments
2975 * @throws Exception
2976 * @return string The text of the template
2977 */
2978 public function braceSubstitution( $piece, $frame ) {
2979
2980 // Flags
2981
2982 // $text has been filled
2983 $found = false;
2984 // wiki markup in $text should be escaped
2985 $nowiki = false;
2986 // $text is HTML, armour it against wikitext transformation
2987 $isHTML = false;
2988 // Force interwiki transclusion to be done in raw mode not rendered
2989 $forceRawInterwiki = false;
2990 // $text is a DOM node needing expansion in a child frame
2991 $isChildObj = false;
2992 // $text is a DOM node needing expansion in the current frame
2993 $isLocalObj = false;
2994
2995 # Title object, where $text came from
2996 $title = false;
2997
2998 # $part1 is the bit before the first |, and must contain only title characters.
2999 # Various prefixes will be stripped from it later.
3000 $titleWithSpaces = $frame->expand( $piece['title'] );
3001 $part1 = trim( $titleWithSpaces );
3002 $titleText = false;
3003
3004 # Original title text preserved for various purposes
3005 $originalTitle = $part1;
3006
3007 # $args is a list of argument nodes, starting from index 0, not including $part1
3008 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3009 # below won't work b/c this $args isn't an object
3010 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3011
3012 $profileSection = null; // profile templates
3013
3014 # SUBST
3015 if ( !$found ) {
3016 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3017
3018 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3019 # Decide whether to expand template or keep wikitext as-is.
3020 if ( $this->ot['wiki'] ) {
3021 if ( $substMatch === false ) {
3022 $literal = true; # literal when in PST with no prefix
3023 } else {
3024 $literal = false; # expand when in PST with subst: or safesubst:
3025 }
3026 } else {
3027 if ( $substMatch == 'subst' ) {
3028 $literal = true; # literal when not in PST with plain subst:
3029 } else {
3030 $literal = false; # expand when not in PST with safesubst: or no prefix
3031 }
3032 }
3033 if ( $literal ) {
3034 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3035 $isLocalObj = true;
3036 $found = true;
3037 }
3038 }
3039
3040 # Variables
3041 if ( !$found && $args->getLength() == 0 ) {
3042 $id = $this->mVariables->matchStartToEnd( $part1 );
3043 if ( $id !== false ) {
3044 $text = $this->getVariableValue( $id, $frame );
3045 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3046 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3047 }
3048 $found = true;
3049 }
3050 }
3051
3052 # MSG, MSGNW and RAW
3053 if ( !$found ) {
3054 # Check for MSGNW:
3055 $mwMsgnw = MagicWord::get( 'msgnw' );
3056 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3057 $nowiki = true;
3058 } else {
3059 # Remove obsolete MSG:
3060 $mwMsg = MagicWord::get( 'msg' );
3061 $mwMsg->matchStartAndRemove( $part1 );
3062 }
3063
3064 # Check for RAW:
3065 $mwRaw = MagicWord::get( 'raw' );
3066 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3067 $forceRawInterwiki = true;
3068 }
3069 }
3070
3071 # Parser functions
3072 if ( !$found ) {
3073 $colonPos = strpos( $part1, ':' );
3074 if ( $colonPos !== false ) {
3075 $func = substr( $part1, 0, $colonPos );
3076 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3077 $argsLength = $args->getLength();
3078 for ( $i = 0; $i < $argsLength; $i++ ) {
3079 $funcArgs[] = $args->item( $i );
3080 }
3081 try {
3082 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3083 } catch ( Exception $ex ) {
3084 throw $ex;
3085 }
3086
3087 # The interface for parser functions allows for extracting
3088 # flags into the local scope. Extract any forwarded flags
3089 # here.
3090 extract( $result );
3091 }
3092 }
3093
3094 # Finish mangling title and then check for loops.
3095 # Set $title to a Title object and $titleText to the PDBK
3096 if ( !$found ) {
3097 $ns = NS_TEMPLATE;
3098 # Split the title into page and subpage
3099 $subpage = '';
3100 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3101 if ( $part1 !== $relative ) {
3102 $part1 = $relative;
3103 $ns = $this->mTitle->getNamespace();
3104 }
3105 $title = Title::newFromText( $part1, $ns );
3106 if ( $title ) {
3107 $titleText = $title->getPrefixedText();
3108 # Check for language variants if the template is not found
3109 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3110 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3111 }
3112 # Do recursion depth check
3113 $limit = $this->mOptions->getMaxTemplateDepth();
3114 if ( $frame->depth >= $limit ) {
3115 $found = true;
3116 $text = '<span class="error">'
3117 . wfMessage( 'parser-template-recursion-depth-warning' )
3118 ->numParams( $limit )->inContentLanguage()->text()
3119 . '</span>';
3120 }
3121 }
3122 }
3123
3124 # Load from database
3125 if ( !$found && $title ) {
3126 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3127 if ( !$title->isExternal() ) {
3128 if ( $title->isSpecialPage()
3129 && $this->mOptions->getAllowSpecialInclusion()
3130 && $this->ot['html']
3131 ) {
3132 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3133 // Pass the template arguments as URL parameters.
3134 // "uselang" will have no effect since the Language object
3135 // is forced to the one defined in ParserOptions.
3136 $pageArgs = [];
3137 $argsLength = $args->getLength();
3138 for ( $i = 0; $i < $argsLength; $i++ ) {
3139 $bits = $args->item( $i )->splitArg();
3140 if ( strval( $bits['index'] ) === '' ) {
3141 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3142 $value = trim( $frame->expand( $bits['value'] ) );
3143 $pageArgs[$name] = $value;
3144 }
3145 }
3146
3147 // Create a new context to execute the special page
3148 $context = new RequestContext;
3149 $context->setTitle( $title );
3150 $context->setRequest( new FauxRequest( $pageArgs ) );
3151 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3152 $context->setUser( $this->getUser() );
3153 } else {
3154 // If this page is cached, then we better not be per user.
3155 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3156 }
3157 $context->setLanguage( $this->mOptions->getUserLangObj() );
3158 $ret = SpecialPageFactory::capturePath( $title, $context, $this->getLinkRenderer() );
3159 if ( $ret ) {
3160 $text = $context->getOutput()->getHTML();
3161 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3162 $found = true;
3163 $isHTML = true;
3164 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3165 $this->mOutput->updateCacheExpiry( $specialPage->maxIncludeCacheTime() );
3166 }
3167 }
3168 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3169 $found = false; # access denied
3170 wfDebug( __METHOD__ . ": template inclusion denied for " .
3171 $title->getPrefixedDBkey() . "\n" );
3172 } else {
3173 list( $text, $title ) = $this->getTemplateDom( $title );
3174 if ( $text !== false ) {
3175 $found = true;
3176 $isChildObj = true;
3177 }
3178 }
3179
3180 # If the title is valid but undisplayable, make a link to it
3181 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3182 $text = "[[:$titleText]]";
3183 $found = true;
3184 }
3185 } elseif ( $title->isTrans() ) {
3186 # Interwiki transclusion
3187 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3188 $text = $this->interwikiTransclude( $title, 'render' );
3189 $isHTML = true;
3190 } else {
3191 $text = $this->interwikiTransclude( $title, 'raw' );
3192 # Preprocess it like a template
3193 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3194 $isChildObj = true;
3195 }
3196 $found = true;
3197 }
3198
3199 # Do infinite loop check
3200 # This has to be done after redirect resolution to avoid infinite loops via redirects
3201 if ( !$frame->loopCheck( $title ) ) {
3202 $found = true;
3203 $text = '<span class="error">'
3204 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3205 . '</span>';
3206 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3207 }
3208 }
3209
3210 # If we haven't found text to substitute by now, we're done
3211 # Recover the source wikitext and return it
3212 if ( !$found ) {
3213 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3214 if ( $profileSection ) {
3215 $this->mProfiler->scopedProfileOut( $profileSection );
3216 }
3217 return [ 'object' => $text ];
3218 }
3219
3220 # Expand DOM-style return values in a child frame
3221 if ( $isChildObj ) {
3222 # Clean up argument array
3223 $newFrame = $frame->newChild( $args, $title );
3224
3225 if ( $nowiki ) {
3226 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3227 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3228 # Expansion is eligible for the empty-frame cache
3229 $text = $newFrame->cachedExpand( $titleText, $text );
3230 } else {
3231 # Uncached expansion
3232 $text = $newFrame->expand( $text );
3233 }
3234 }
3235 if ( $isLocalObj && $nowiki ) {
3236 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3237 $isLocalObj = false;
3238 }
3239
3240 if ( $profileSection ) {
3241 $this->mProfiler->scopedProfileOut( $profileSection );
3242 }
3243
3244 # Replace raw HTML by a placeholder
3245 if ( $isHTML ) {
3246 $text = $this->insertStripItem( $text );
3247 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3248 # Escape nowiki-style return values
3249 $text = wfEscapeWikiText( $text );
3250 } elseif ( is_string( $text )
3251 && !$piece['lineStart']
3252 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3253 ) {
3254 # Bug 529: if the template begins with a table or block-level
3255 # element, it should be treated as beginning a new line.
3256 # This behavior is somewhat controversial.
3257 $text = "\n" . $text;
3258 }
3259
3260 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3261 # Error, oversize inclusion
3262 if ( $titleText !== false ) {
3263 # Make a working, properly escaped link if possible (bug 23588)
3264 $text = "[[:$titleText]]";
3265 } else {
3266 # This will probably not be a working link, but at least it may
3267 # provide some hint of where the problem is
3268 preg_replace( '/^:/', '', $originalTitle );
3269 $text = "[[:$originalTitle]]";
3270 }
3271 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3272 . 'post-expand include size too large -->' );
3273 $this->limitationWarn( 'post-expand-template-inclusion' );
3274 }
3275
3276 if ( $isLocalObj ) {
3277 $ret = [ 'object' => $text ];
3278 } else {
3279 $ret = [ 'text' => $text ];
3280 }
3281
3282 return $ret;
3283 }
3284
3285 /**
3286 * Call a parser function and return an array with text and flags.
3287 *
3288 * The returned array will always contain a boolean 'found', indicating
3289 * whether the parser function was found or not. It may also contain the
3290 * following:
3291 * text: string|object, resulting wikitext or PP DOM object
3292 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3293 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3294 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3295 * nowiki: bool, wiki markup in $text should be escaped
3296 *
3297 * @since 1.21
3298 * @param PPFrame $frame The current frame, contains template arguments
3299 * @param string $function Function name
3300 * @param array $args Arguments to the function
3301 * @throws MWException
3302 * @return array
3303 */
3304 public function callParserFunction( $frame, $function, array $args = [] ) {
3305 global $wgContLang;
3306
3307 # Case sensitive functions
3308 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3309 $function = $this->mFunctionSynonyms[1][$function];
3310 } else {
3311 # Case insensitive functions
3312 $function = $wgContLang->lc( $function );
3313 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3314 $function = $this->mFunctionSynonyms[0][$function];
3315 } else {
3316 return [ 'found' => false ];
3317 }
3318 }
3319
3320 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3321
3322 # Workaround for PHP bug 35229 and similar
3323 if ( !is_callable( $callback ) ) {
3324 throw new MWException( "Tag hook for $function is not callable\n" );
3325 }
3326
3327 $allArgs = [ &$this ];
3328 if ( $flags & self::SFH_OBJECT_ARGS ) {
3329 # Convert arguments to PPNodes and collect for appending to $allArgs
3330 $funcArgs = [];
3331 foreach ( $args as $k => $v ) {
3332 if ( $v instanceof PPNode || $k === 0 ) {
3333 $funcArgs[] = $v;
3334 } else {
3335 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3336 }
3337 }
3338
3339 # Add a frame parameter, and pass the arguments as an array
3340 $allArgs[] = $frame;
3341 $allArgs[] = $funcArgs;
3342 } else {
3343 # Convert arguments to plain text and append to $allArgs
3344 foreach ( $args as $k => $v ) {
3345 if ( $v instanceof PPNode ) {
3346 $allArgs[] = trim( $frame->expand( $v ) );
3347 } elseif ( is_int( $k ) && $k >= 0 ) {
3348 $allArgs[] = trim( $v );
3349 } else {
3350 $allArgs[] = trim( "$k=$v" );
3351 }
3352 }
3353 }
3354
3355 $result = call_user_func_array( $callback, $allArgs );
3356
3357 # The interface for function hooks allows them to return a wikitext
3358 # string or an array containing the string and any flags. This mungs
3359 # things around to match what this method should return.
3360 if ( !is_array( $result ) ) {
3361 $result =[
3362 'found' => true,
3363 'text' => $result,
3364 ];
3365 } else {
3366 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3367 $result['text'] = $result[0];
3368 }
3369 unset( $result[0] );
3370 $result += [
3371 'found' => true,
3372 ];
3373 }
3374
3375 $noparse = true;
3376 $preprocessFlags = 0;
3377 if ( isset( $result['noparse'] ) ) {
3378 $noparse = $result['noparse'];
3379 }
3380 if ( isset( $result['preprocessFlags'] ) ) {
3381 $preprocessFlags = $result['preprocessFlags'];
3382 }
3383
3384 if ( !$noparse ) {
3385 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3386 $result['isChildObj'] = true;
3387 }
3388
3389 return $result;
3390 }
3391
3392 /**
3393 * Get the semi-parsed DOM representation of a template with a given title,
3394 * and its redirect destination title. Cached.
3395 *
3396 * @param Title $title
3397 *
3398 * @return array
3399 */
3400 public function getTemplateDom( $title ) {
3401 $cacheTitle = $title;
3402 $titleText = $title->getPrefixedDBkey();
3403
3404 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3405 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3406 $title = Title::makeTitle( $ns, $dbk );
3407 $titleText = $title->getPrefixedDBkey();
3408 }
3409 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3410 return [ $this->mTplDomCache[$titleText], $title ];
3411 }
3412
3413 # Cache miss, go to the database
3414 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3415
3416 if ( $text === false ) {
3417 $this->mTplDomCache[$titleText] = false;
3418 return [ false, $title ];
3419 }
3420
3421 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3422 $this->mTplDomCache[$titleText] = $dom;
3423
3424 if ( !$title->equals( $cacheTitle ) ) {
3425 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3426 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3427 }
3428
3429 return [ $dom, $title ];
3430 }
3431
3432 /**
3433 * Fetch the current revision of a given title. Note that the revision
3434 * (and even the title) may not exist in the database, so everything
3435 * contributing to the output of the parser should use this method
3436 * where possible, rather than getting the revisions themselves. This
3437 * method also caches its results, so using it benefits performance.
3438 *
3439 * @since 1.24
3440 * @param Title $title
3441 * @return Revision
3442 */
3443 public function fetchCurrentRevisionOfTitle( $title ) {
3444 $cacheKey = $title->getPrefixedDBkey();
3445 if ( !$this->currentRevisionCache ) {
3446 $this->currentRevisionCache = new MapCacheLRU( 100 );
3447 }
3448 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3449 $this->currentRevisionCache->set( $cacheKey,
3450 // Defaults to Parser::statelessFetchRevision()
3451 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3452 );
3453 }
3454 return $this->currentRevisionCache->get( $cacheKey );
3455 }
3456
3457 /**
3458 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3459 * without passing them on to it.
3460 *
3461 * @since 1.24
3462 * @param Title $title
3463 * @param Parser|bool $parser
3464 * @return Revision
3465 */
3466 public static function statelessFetchRevision( $title, $parser = false ) {
3467 return Revision::newFromTitle( $title );
3468 }
3469
3470 /**
3471 * Fetch the unparsed text of a template and register a reference to it.
3472 * @param Title $title
3473 * @return array ( string or false, Title )
3474 */
3475 public function fetchTemplateAndTitle( $title ) {
3476 // Defaults to Parser::statelessFetchTemplate()
3477 $templateCb = $this->mOptions->getTemplateCallback();
3478 $stuff = call_user_func( $templateCb, $title, $this );
3479 // We use U+007F DELETE to distinguish strip markers from regular text.
3480 $text = $stuff['text'];
3481 if ( is_string( $stuff['text'] ) ) {
3482 $text = strtr( $text, "\x7f", "?" );
3483 }
3484 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3485 if ( isset( $stuff['deps'] ) ) {
3486 foreach ( $stuff['deps'] as $dep ) {
3487 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3488 if ( $dep['title']->equals( $this->getTitle() ) ) {
3489 // If we transclude ourselves, the final result
3490 // will change based on the new version of the page
3491 $this->mOutput->setFlag( 'vary-revision' );
3492 }
3493 }
3494 }
3495 return [ $text, $finalTitle ];
3496 }
3497
3498 /**
3499 * Fetch the unparsed text of a template and register a reference to it.
3500 * @param Title $title
3501 * @return string|bool
3502 */
3503 public function fetchTemplate( $title ) {
3504 return $this->fetchTemplateAndTitle( $title )[0];
3505 }
3506
3507 /**
3508 * Static function to get a template
3509 * Can be overridden via ParserOptions::setTemplateCallback().
3510 *
3511 * @param Title $title
3512 * @param bool|Parser $parser
3513 *
3514 * @return array
3515 */
3516 public static function statelessFetchTemplate( $title, $parser = false ) {
3517 $text = $skip = false;
3518 $finalTitle = $title;
3519 $deps = [];
3520
3521 # Loop to fetch the article, with up to 1 redirect
3522 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3523 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3524 // @codingStandardsIgnoreEnd
3525 # Give extensions a chance to select the revision instead
3526 $id = false; # Assume current
3527 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3528 [ $parser, $title, &$skip, &$id ] );
3529
3530 if ( $skip ) {
3531 $text = false;
3532 $deps[] = [
3533 'title' => $title,
3534 'page_id' => $title->getArticleID(),
3535 'rev_id' => null
3536 ];
3537 break;
3538 }
3539 # Get the revision
3540 if ( $id ) {
3541 $rev = Revision::newFromId( $id );
3542 } elseif ( $parser ) {
3543 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3544 } else {
3545 $rev = Revision::newFromTitle( $title );
3546 }
3547 $rev_id = $rev ? $rev->getId() : 0;
3548 # If there is no current revision, there is no page
3549 if ( $id === false && !$rev ) {
3550 $linkCache = LinkCache::singleton();
3551 $linkCache->addBadLinkObj( $title );
3552 }
3553
3554 $deps[] = [
3555 'title' => $title,
3556 'page_id' => $title->getArticleID(),
3557 'rev_id' => $rev_id ];
3558 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3559 # We fetched a rev from a different title; register it too...
3560 $deps[] = [
3561 'title' => $rev->getTitle(),
3562 'page_id' => $rev->getPage(),
3563 'rev_id' => $rev_id ];
3564 }
3565
3566 if ( $rev ) {
3567 $content = $rev->getContent();
3568 $text = $content ? $content->getWikitextForTransclusion() : null;
3569
3570 if ( $text === false || $text === null ) {
3571 $text = false;
3572 break;
3573 }
3574 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3575 global $wgContLang;
3576 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3577 if ( !$message->exists() ) {
3578 $text = false;
3579 break;
3580 }
3581 $content = $message->content();
3582 $text = $message->plain();
3583 } else {
3584 break;
3585 }
3586 if ( !$content ) {
3587 break;
3588 }
3589 # Redirect?
3590 $finalTitle = $title;
3591 $title = $content->getRedirectTarget();
3592 }
3593 return [
3594 'text' => $text,
3595 'finalTitle' => $finalTitle,
3596 'deps' => $deps ];
3597 }
3598
3599 /**
3600 * Fetch a file and its title and register a reference to it.
3601 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3602 * @param Title $title
3603 * @param array $options Array of options to RepoGroup::findFile
3604 * @return File|bool
3605 */
3606 public function fetchFile( $title, $options = [] ) {
3607 return $this->fetchFileAndTitle( $title, $options )[0];
3608 }
3609
3610 /**
3611 * Fetch a file and its title and register a reference to it.
3612 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3613 * @param Title $title
3614 * @param array $options Array of options to RepoGroup::findFile
3615 * @return array ( File or false, Title of file )
3616 */
3617 public function fetchFileAndTitle( $title, $options = [] ) {
3618 $file = $this->fetchFileNoRegister( $title, $options );
3619
3620 $time = $file ? $file->getTimestamp() : false;
3621 $sha1 = $file ? $file->getSha1() : false;
3622 # Register the file as a dependency...
3623 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3624 if ( $file && !$title->equals( $file->getTitle() ) ) {
3625 # Update fetched file title
3626 $title = $file->getTitle();
3627 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3628 }
3629 return [ $file, $title ];
3630 }
3631
3632 /**
3633 * Helper function for fetchFileAndTitle.
3634 *
3635 * Also useful if you need to fetch a file but not use it yet,
3636 * for example to get the file's handler.
3637 *
3638 * @param Title $title
3639 * @param array $options Array of options to RepoGroup::findFile
3640 * @return File|bool
3641 */
3642 protected function fetchFileNoRegister( $title, $options = [] ) {
3643 if ( isset( $options['broken'] ) ) {
3644 $file = false; // broken thumbnail forced by hook
3645 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3646 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3647 } else { // get by (name,timestamp)
3648 $file = wfFindFile( $title, $options );
3649 }
3650 return $file;
3651 }
3652
3653 /**
3654 * Transclude an interwiki link.
3655 *
3656 * @param Title $title
3657 * @param string $action
3658 *
3659 * @return string
3660 */
3661 public function interwikiTransclude( $title, $action ) {
3662 global $wgEnableScaryTranscluding;
3663
3664 if ( !$wgEnableScaryTranscluding ) {
3665 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3666 }
3667
3668 $url = $title->getFullURL( [ 'action' => $action ] );
3669
3670 if ( strlen( $url ) > 255 ) {
3671 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3672 }
3673 return $this->fetchScaryTemplateMaybeFromCache( $url );
3674 }
3675
3676 /**
3677 * @param string $url
3678 * @return mixed|string
3679 */
3680 public function fetchScaryTemplateMaybeFromCache( $url ) {
3681 global $wgTranscludeCacheExpiry;
3682 $dbr = wfGetDB( DB_SLAVE );
3683 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3684 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3685 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3686 if ( $obj ) {
3687 return $obj->tc_contents;
3688 }
3689
3690 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3691 $status = $req->execute(); // Status object
3692 if ( $status->isOK() ) {
3693 $text = $req->getContent();
3694 } elseif ( $req->getStatus() != 200 ) {
3695 // Though we failed to fetch the content, this status is useless.
3696 return wfMessage( 'scarytranscludefailed-httpstatus' )
3697 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3698 } else {
3699 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3700 }
3701
3702 $dbw = wfGetDB( DB_MASTER );
3703 $dbw->replace( 'transcache', [ 'tc_url' ], [
3704 'tc_url' => $url,
3705 'tc_time' => $dbw->timestamp( time() ),
3706 'tc_contents' => $text
3707 ] );
3708 return $text;
3709 }
3710
3711 /**
3712 * Triple brace replacement -- used for template arguments
3713 * @private
3714 *
3715 * @param array $piece
3716 * @param PPFrame $frame
3717 *
3718 * @return array
3719 */
3720 public function argSubstitution( $piece, $frame ) {
3721
3722 $error = false;
3723 $parts = $piece['parts'];
3724 $nameWithSpaces = $frame->expand( $piece['title'] );
3725 $argName = trim( $nameWithSpaces );
3726 $object = false;
3727 $text = $frame->getArgument( $argName );
3728 if ( $text === false && $parts->getLength() > 0
3729 && ( $this->ot['html']
3730 || $this->ot['pre']
3731 || ( $this->ot['wiki'] && $frame->isTemplate() )
3732 )
3733 ) {
3734 # No match in frame, use the supplied default
3735 $object = $parts->item( 0 )->getChildren();
3736 }
3737 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3738 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3739 $this->limitationWarn( 'post-expand-template-argument' );
3740 }
3741
3742 if ( $text === false && $object === false ) {
3743 # No match anywhere
3744 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3745 }
3746 if ( $error !== false ) {
3747 $text .= $error;
3748 }
3749 if ( $object !== false ) {
3750 $ret = [ 'object' => $object ];
3751 } else {
3752 $ret = [ 'text' => $text ];
3753 }
3754
3755 return $ret;
3756 }
3757
3758 /**
3759 * Return the text to be used for a given extension tag.
3760 * This is the ghost of strip().
3761 *
3762 * @param array $params Associative array of parameters:
3763 * name PPNode for the tag name
3764 * attr PPNode for unparsed text where tag attributes are thought to be
3765 * attributes Optional associative array of parsed attributes
3766 * inner Contents of extension element
3767 * noClose Original text did not have a close tag
3768 * @param PPFrame $frame
3769 *
3770 * @throws MWException
3771 * @return string
3772 */
3773 public function extensionSubstitution( $params, $frame ) {
3774 $name = $frame->expand( $params['name'] );
3775 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3776 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3777 $marker = self::MARKER_PREFIX . "-$name-"
3778 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3779
3780 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3781 ( $this->ot['html'] || $this->ot['pre'] );
3782 if ( $isFunctionTag ) {
3783 $markerType = 'none';
3784 } else {
3785 $markerType = 'general';
3786 }
3787 if ( $this->ot['html'] || $isFunctionTag ) {
3788 $name = strtolower( $name );
3789 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3790 if ( isset( $params['attributes'] ) ) {
3791 $attributes = $attributes + $params['attributes'];
3792 }
3793
3794 if ( isset( $this->mTagHooks[$name] ) ) {
3795 # Workaround for PHP bug 35229 and similar
3796 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3797 throw new MWException( "Tag hook for $name is not callable\n" );
3798 }
3799 $output = call_user_func_array( $this->mTagHooks[$name],
3800 [ $content, $attributes, $this, $frame ] );
3801 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3802 list( $callback, ) = $this->mFunctionTagHooks[$name];
3803 if ( !is_callable( $callback ) ) {
3804 throw new MWException( "Tag hook for $name is not callable\n" );
3805 }
3806
3807 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3808 } else {
3809 $output = '<span class="error">Invalid tag extension name: ' .
3810 htmlspecialchars( $name ) . '</span>';
3811 }
3812
3813 if ( is_array( $output ) ) {
3814 # Extract flags to local scope (to override $markerType)
3815 $flags = $output;
3816 $output = $flags[0];
3817 unset( $flags[0] );
3818 extract( $flags );
3819 }
3820 } else {
3821 if ( is_null( $attrText ) ) {
3822 $attrText = '';
3823 }
3824 if ( isset( $params['attributes'] ) ) {
3825 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3826 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3827 htmlspecialchars( $attrValue ) . '"';
3828 }
3829 }
3830 if ( $content === null ) {
3831 $output = "<$name$attrText/>";
3832 } else {
3833 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3834 $output = "<$name$attrText>$content$close";
3835 }
3836 }
3837
3838 if ( $markerType === 'none' ) {
3839 return $output;
3840 } elseif ( $markerType === 'nowiki' ) {
3841 $this->mStripState->addNoWiki( $marker, $output );
3842 } elseif ( $markerType === 'general' ) {
3843 $this->mStripState->addGeneral( $marker, $output );
3844 } else {
3845 throw new MWException( __METHOD__ . ': invalid marker type' );
3846 }
3847 return $marker;
3848 }
3849
3850 /**
3851 * Increment an include size counter
3852 *
3853 * @param string $type The type of expansion
3854 * @param int $size The size of the text
3855 * @return bool False if this inclusion would take it over the maximum, true otherwise
3856 */
3857 public function incrementIncludeSize( $type, $size ) {
3858 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3859 return false;
3860 } else {
3861 $this->mIncludeSizes[$type] += $size;
3862 return true;
3863 }
3864 }
3865
3866 /**
3867 * Increment the expensive function count
3868 *
3869 * @return bool False if the limit has been exceeded
3870 */
3871 public function incrementExpensiveFunctionCount() {
3872 $this->mExpensiveFunctionCount++;
3873 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3874 }
3875
3876 /**
3877 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3878 * Fills $this->mDoubleUnderscores, returns the modified text
3879 *
3880 * @param string $text
3881 *
3882 * @return string
3883 */
3884 public function doDoubleUnderscore( $text ) {
3885
3886 # The position of __TOC__ needs to be recorded
3887 $mw = MagicWord::get( 'toc' );
3888 if ( $mw->match( $text ) ) {
3889 $this->mShowToc = true;
3890 $this->mForceTocPosition = true;
3891
3892 # Set a placeholder. At the end we'll fill it in with the TOC.
3893 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3894
3895 # Only keep the first one.
3896 $text = $mw->replace( '', $text );
3897 }
3898
3899 # Now match and remove the rest of them
3900 $mwa = MagicWord::getDoubleUnderscoreArray();
3901 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3902
3903 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3904 $this->mOutput->mNoGallery = true;
3905 }
3906 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3907 $this->mShowToc = false;
3908 }
3909 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3910 && $this->mTitle->getNamespace() == NS_CATEGORY
3911 ) {
3912 $this->addTrackingCategory( 'hidden-category-category' );
3913 }
3914 # (bug 8068) Allow control over whether robots index a page.
3915 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3916 # is not desirable, the last one on the page should win.
3917 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3918 $this->mOutput->setIndexPolicy( 'noindex' );
3919 $this->addTrackingCategory( 'noindex-category' );
3920 }
3921 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3922 $this->mOutput->setIndexPolicy( 'index' );
3923 $this->addTrackingCategory( 'index-category' );
3924 }
3925
3926 # Cache all double underscores in the database
3927 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3928 $this->mOutput->setProperty( $key, '' );
3929 }
3930
3931 return $text;
3932 }
3933
3934 /**
3935 * @see ParserOutput::addTrackingCategory()
3936 * @param string $msg Message key
3937 * @return bool Whether the addition was successful
3938 */
3939 public function addTrackingCategory( $msg ) {
3940 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3941 }
3942
3943 /**
3944 * This function accomplishes several tasks:
3945 * 1) Auto-number headings if that option is enabled
3946 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3947 * 3) Add a Table of contents on the top for users who have enabled the option
3948 * 4) Auto-anchor headings
3949 *
3950 * It loops through all headlines, collects the necessary data, then splits up the
3951 * string and re-inserts the newly formatted headlines.
3952 *
3953 * @param string $text
3954 * @param string $origText Original, untouched wikitext
3955 * @param bool $isMain
3956 * @return mixed|string
3957 * @private
3958 */
3959 public function formatHeadings( $text, $origText, $isMain = true ) {
3960 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
3961
3962 # Inhibit editsection links if requested in the page
3963 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3964 $maybeShowEditLink = $showEditLink = false;
3965 } else {
3966 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3967 $showEditLink = $this->mOptions->getEditSection();
3968 }
3969 if ( $showEditLink ) {
3970 $this->mOutput->setEditSectionTokens( true );
3971 }
3972
3973 # Get all headlines for numbering them and adding funky stuff like [edit]
3974 # links - this is for later, but we need the number of headlines right now
3975 $matches = [];
3976 $numMatches = preg_match_all(
3977 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
3978 $text,
3979 $matches
3980 );
3981
3982 # if there are fewer than 4 headlines in the article, do not show TOC
3983 # unless it's been explicitly enabled.
3984 $enoughToc = $this->mShowToc &&
3985 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3986
3987 # Allow user to stipulate that a page should have a "new section"
3988 # link added via __NEWSECTIONLINK__
3989 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3990 $this->mOutput->setNewSection( true );
3991 }
3992
3993 # Allow user to remove the "new section"
3994 # link via __NONEWSECTIONLINK__
3995 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3996 $this->mOutput->hideNewSection( true );
3997 }
3998
3999 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4000 # override above conditions and always show TOC above first header
4001 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4002 $this->mShowToc = true;
4003 $enoughToc = true;
4004 }
4005
4006 # headline counter
4007 $headlineCount = 0;
4008 $numVisible = 0;
4009
4010 # Ugh .. the TOC should have neat indentation levels which can be
4011 # passed to the skin functions. These are determined here
4012 $toc = '';
4013 $full = '';
4014 $head = [];
4015 $sublevelCount = [];
4016 $levelCount = [];
4017 $level = 0;
4018 $prevlevel = 0;
4019 $toclevel = 0;
4020 $prevtoclevel = 0;
4021 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4022 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4023 $oldType = $this->mOutputType;
4024 $this->setOutputType( self::OT_WIKI );
4025 $frame = $this->getPreprocessor()->newFrame();
4026 $root = $this->preprocessToDom( $origText );
4027 $node = $root->getFirstChild();
4028 $byteOffset = 0;
4029 $tocraw = [];
4030 $refers = [];
4031
4032 $headlines = $numMatches !== false ? $matches[3] : [];
4033
4034 foreach ( $headlines as $headline ) {
4035 $isTemplate = false;
4036 $titleText = false;
4037 $sectionIndex = false;
4038 $numbering = '';
4039 $markerMatches = [];
4040 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4041 $serial = $markerMatches[1];
4042 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4043 $isTemplate = ( $titleText != $baseTitleText );
4044 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4045 }
4046
4047 if ( $toclevel ) {
4048 $prevlevel = $level;
4049 }
4050 $level = $matches[1][$headlineCount];
4051
4052 if ( $level > $prevlevel ) {
4053 # Increase TOC level
4054 $toclevel++;
4055 $sublevelCount[$toclevel] = 0;
4056 if ( $toclevel < $wgMaxTocLevel ) {
4057 $prevtoclevel = $toclevel;
4058 $toc .= Linker::tocIndent();
4059 $numVisible++;
4060 }
4061 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4062 # Decrease TOC level, find level to jump to
4063
4064 for ( $i = $toclevel; $i > 0; $i-- ) {
4065 if ( $levelCount[$i] == $level ) {
4066 # Found last matching level
4067 $toclevel = $i;
4068 break;
4069 } elseif ( $levelCount[$i] < $level ) {
4070 # Found first matching level below current level
4071 $toclevel = $i + 1;
4072 break;
4073 }
4074 }
4075 if ( $i == 0 ) {
4076 $toclevel = 1;
4077 }
4078 if ( $toclevel < $wgMaxTocLevel ) {
4079 if ( $prevtoclevel < $wgMaxTocLevel ) {
4080 # Unindent only if the previous toc level was shown :p
4081 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4082 $prevtoclevel = $toclevel;
4083 } else {
4084 $toc .= Linker::tocLineEnd();
4085 }
4086 }
4087 } else {
4088 # No change in level, end TOC line
4089 if ( $toclevel < $wgMaxTocLevel ) {
4090 $toc .= Linker::tocLineEnd();
4091 }
4092 }
4093
4094 $levelCount[$toclevel] = $level;
4095
4096 # count number of headlines for each level
4097 $sublevelCount[$toclevel]++;
4098 $dot = 0;
4099 for ( $i = 1; $i <= $toclevel; $i++ ) {
4100 if ( !empty( $sublevelCount[$i] ) ) {
4101 if ( $dot ) {
4102 $numbering .= '.';
4103 }
4104 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4105 $dot = 1;
4106 }
4107 }
4108
4109 # The safe header is a version of the header text safe to use for links
4110
4111 # Remove link placeholders by the link text.
4112 # <!--LINK number-->
4113 # turns into
4114 # link text with suffix
4115 # Do this before unstrip since link text can contain strip markers
4116 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4117
4118 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4119 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4120
4121 # Strip out HTML (first regex removes any tag not allowed)
4122 # Allowed tags are:
4123 # * <sup> and <sub> (bug 8393)
4124 # * <i> (bug 26375)
4125 # * <b> (r105284)
4126 # * <bdi> (bug 72884)
4127 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4128 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4129 # to allow setting directionality in toc items.
4130 $tocline = preg_replace(
4131 [
4132 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4133 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4134 ],
4135 [ '', '<$1>' ],
4136 $safeHeadline
4137 );
4138
4139 # Strip '<span></span>', which is the result from the above if
4140 # <span id="foo"></span> is used to produce an additional anchor
4141 # for a section.
4142 $tocline = str_replace( '<span></span>', '', $tocline );
4143
4144 $tocline = trim( $tocline );
4145
4146 # For the anchor, strip out HTML-y stuff period
4147 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4148 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4149
4150 # Save headline for section edit hint before it's escaped
4151 $headlineHint = $safeHeadline;
4152
4153 if ( $wgExperimentalHtmlIds ) {
4154 # For reverse compatibility, provide an id that's
4155 # HTML4-compatible, like we used to.
4156 # It may be worth noting, academically, that it's possible for
4157 # the legacy anchor to conflict with a non-legacy headline
4158 # anchor on the page. In this case likely the "correct" thing
4159 # would be to either drop the legacy anchors or make sure
4160 # they're numbered first. However, this would require people
4161 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4162 # manually, so let's not bother worrying about it.
4163 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4164 [ 'noninitial', 'legacy' ] );
4165 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4166
4167 if ( $legacyHeadline == $safeHeadline ) {
4168 # No reason to have both (in fact, we can't)
4169 $legacyHeadline = false;
4170 }
4171 } else {
4172 $legacyHeadline = false;
4173 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4174 'noninitial' );
4175 }
4176
4177 # HTML names must be case-insensitively unique (bug 10721).
4178 # This does not apply to Unicode characters per
4179 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4180 # @todo FIXME: We may be changing them depending on the current locale.
4181 $arrayKey = strtolower( $safeHeadline );
4182 if ( $legacyHeadline === false ) {
4183 $legacyArrayKey = false;
4184 } else {
4185 $legacyArrayKey = strtolower( $legacyHeadline );
4186 }
4187
4188 # Create the anchor for linking from the TOC to the section
4189 $anchor = $safeHeadline;
4190 $legacyAnchor = $legacyHeadline;
4191 if ( isset( $refers[$arrayKey] ) ) {
4192 // @codingStandardsIgnoreStart
4193 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4194 // @codingStandardsIgnoreEnd
4195 $anchor .= "_$i";
4196 $refers["${arrayKey}_$i"] = true;
4197 } else {
4198 $refers[$arrayKey] = true;
4199 }
4200 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4201 // @codingStandardsIgnoreStart
4202 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4203 // @codingStandardsIgnoreEnd
4204 $legacyAnchor .= "_$i";
4205 $refers["${legacyArrayKey}_$i"] = true;
4206 } else {
4207 $refers[$legacyArrayKey] = true;
4208 }
4209
4210 # Don't number the heading if it is the only one (looks silly)
4211 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4212 # the two are different if the line contains a link
4213 $headline = Html::element(
4214 'span',
4215 [ 'class' => 'mw-headline-number' ],
4216 $numbering
4217 ) . ' ' . $headline;
4218 }
4219
4220 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4221 $toc .= Linker::tocLine( $anchor, $tocline,
4222 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4223 }
4224
4225 # Add the section to the section tree
4226 # Find the DOM node for this header
4227 $noOffset = ( $isTemplate || $sectionIndex === false );
4228 while ( $node && !$noOffset ) {
4229 if ( $node->getName() === 'h' ) {
4230 $bits = $node->splitHeading();
4231 if ( $bits['i'] == $sectionIndex ) {
4232 break;
4233 }
4234 }
4235 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4236 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4237 $node = $node->getNextSibling();
4238 }
4239 $tocraw[] = [
4240 'toclevel' => $toclevel,
4241 'level' => $level,
4242 'line' => $tocline,
4243 'number' => $numbering,
4244 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4245 'fromtitle' => $titleText,
4246 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4247 'anchor' => $anchor,
4248 ];
4249
4250 # give headline the correct <h#> tag
4251 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4252 // Output edit section links as markers with styles that can be customized by skins
4253 if ( $isTemplate ) {
4254 # Put a T flag in the section identifier, to indicate to extractSections()
4255 # that sections inside <includeonly> should be counted.
4256 $editsectionPage = $titleText;
4257 $editsectionSection = "T-$sectionIndex";
4258 $editsectionContent = null;
4259 } else {
4260 $editsectionPage = $this->mTitle->getPrefixedText();
4261 $editsectionSection = $sectionIndex;
4262 $editsectionContent = $headlineHint;
4263 }
4264 // We use a bit of pesudo-xml for editsection markers. The
4265 // language converter is run later on. Using a UNIQ style marker
4266 // leads to the converter screwing up the tokens when it
4267 // converts stuff. And trying to insert strip tags fails too. At
4268 // this point all real inputted tags have already been escaped,
4269 // so we don't have to worry about a user trying to input one of
4270 // these markers directly. We use a page and section attribute
4271 // to stop the language converter from converting these
4272 // important bits of data, but put the headline hint inside a
4273 // content block because the language converter is supposed to
4274 // be able to convert that piece of data.
4275 // Gets replaced with html in ParserOutput::getText
4276 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4277 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4278 if ( $editsectionContent !== null ) {
4279 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4280 } else {
4281 $editlink .= '/>';
4282 }
4283 } else {
4284 $editlink = '';
4285 }
4286 $head[$headlineCount] = Linker::makeHeadline( $level,
4287 $matches['attrib'][$headlineCount], $anchor, $headline,
4288 $editlink, $legacyAnchor );
4289
4290 $headlineCount++;
4291 }
4292
4293 $this->setOutputType( $oldType );
4294
4295 # Never ever show TOC if no headers
4296 if ( $numVisible < 1 ) {
4297 $enoughToc = false;
4298 }
4299
4300 if ( $enoughToc ) {
4301 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4302 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4303 }
4304 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4305 $this->mOutput->setTOCHTML( $toc );
4306 $toc = self::TOC_START . $toc . self::TOC_END;
4307 $this->mOutput->addModules( 'mediawiki.toc' );
4308 }
4309
4310 if ( $isMain ) {
4311 $this->mOutput->setSections( $tocraw );
4312 }
4313
4314 # split up and insert constructed headlines
4315 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4316 $i = 0;
4317
4318 // build an array of document sections
4319 $sections = [];
4320 foreach ( $blocks as $block ) {
4321 // $head is zero-based, sections aren't.
4322 if ( empty( $head[$i - 1] ) ) {
4323 $sections[$i] = $block;
4324 } else {
4325 $sections[$i] = $head[$i - 1] . $block;
4326 }
4327
4328 /**
4329 * Send a hook, one per section.
4330 * The idea here is to be able to make section-level DIVs, but to do so in a
4331 * lower-impact, more correct way than r50769
4332 *
4333 * $this : caller
4334 * $section : the section number
4335 * &$sectionContent : ref to the content of the section
4336 * $showEditLinks : boolean describing whether this section has an edit link
4337 */
4338 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4339
4340 $i++;
4341 }
4342
4343 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4344 // append the TOC at the beginning
4345 // Top anchor now in skin
4346 $sections[0] = $sections[0] . $toc . "\n";
4347 }
4348
4349 $full .= implode( '', $sections );
4350
4351 if ( $this->mForceTocPosition ) {
4352 return str_replace( '<!--MWTOC-->', $toc, $full );
4353 } else {
4354 return $full;
4355 }
4356 }
4357
4358 /**
4359 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4360 * conversion, substituting signatures, {{subst:}} templates, etc.
4361 *
4362 * @param string $text The text to transform
4363 * @param Title $title The Title object for the current article
4364 * @param User $user The User object describing the current user
4365 * @param ParserOptions $options Parsing options
4366 * @param bool $clearState Whether to clear the parser state first
4367 * @return string The altered wiki markup
4368 */
4369 public function preSaveTransform( $text, Title $title, User $user,
4370 ParserOptions $options, $clearState = true
4371 ) {
4372 if ( $clearState ) {
4373 $magicScopeVariable = $this->lock();
4374 }
4375 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4376 $this->setUser( $user );
4377
4378 $pairs = [
4379 "\r\n" => "\n",
4380 "\r" => "\n",
4381 ];
4382 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4383 if ( $options->getPreSaveTransform() ) {
4384 $text = $this->pstPass2( $text, $user );
4385 }
4386 $text = $this->mStripState->unstripBoth( $text );
4387
4388 $this->setUser( null ); # Reset
4389
4390 return $text;
4391 }
4392
4393 /**
4394 * Pre-save transform helper function
4395 *
4396 * @param string $text
4397 * @param User $user
4398 *
4399 * @return string
4400 */
4401 private function pstPass2( $text, $user ) {
4402 global $wgContLang;
4403
4404 # Note: This is the timestamp saved as hardcoded wikitext to
4405 # the database, we use $wgContLang here in order to give
4406 # everyone the same signature and use the default one rather
4407 # than the one selected in each user's preferences.
4408 # (see also bug 12815)
4409 $ts = $this->mOptions->getTimestamp();
4410 $timestamp = MWTimestamp::getLocalInstance( $ts );
4411 $ts = $timestamp->format( 'YmdHis' );
4412 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4413
4414 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4415
4416 # Variable replacement
4417 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4418 $text = $this->replaceVariables( $text );
4419
4420 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4421 # which may corrupt this parser instance via its wfMessage()->text() call-
4422
4423 # Signatures
4424 $sigText = $this->getUserSig( $user );
4425 $text = strtr( $text, [
4426 '~~~~~' => $d,
4427 '~~~~' => "$sigText $d",
4428 '~~~' => $sigText
4429 ] );
4430
4431 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4432 $tc = '[' . Title::legalChars() . ']';
4433 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4434
4435 // [[ns:page (context)|]]
4436 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4437 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4438 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4439 // [[ns:page (context), context|]] (using either single or double-width comma)
4440 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4441 // [[|page]] (reverse pipe trick: add context from page title)
4442 $p2 = "/\[\[\\|($tc+)]]/";
4443
4444 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4445 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4446 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4447 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4448
4449 $t = $this->mTitle->getText();
4450 $m = [];
4451 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4452 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4453 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4454 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4455 } else {
4456 # if there's no context, don't bother duplicating the title
4457 $text = preg_replace( $p2, '[[\\1]]', $text );
4458 }
4459
4460 # Trim trailing whitespace
4461 $text = rtrim( $text );
4462
4463 return $text;
4464 }
4465
4466 /**
4467 * Fetch the user's signature text, if any, and normalize to
4468 * validated, ready-to-insert wikitext.
4469 * If you have pre-fetched the nickname or the fancySig option, you can
4470 * specify them here to save a database query.
4471 * Do not reuse this parser instance after calling getUserSig(),
4472 * as it may have changed if it's the $wgParser.
4473 *
4474 * @param User $user
4475 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4476 * @param bool|null $fancySig whether the nicknname is the complete signature
4477 * or null to use default value
4478 * @return string
4479 */
4480 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4481 global $wgMaxSigChars;
4482
4483 $username = $user->getName();
4484
4485 # If not given, retrieve from the user object.
4486 if ( $nickname === false ) {
4487 $nickname = $user->getOption( 'nickname' );
4488 }
4489
4490 if ( is_null( $fancySig ) ) {
4491 $fancySig = $user->getBoolOption( 'fancysig' );
4492 }
4493
4494 $nickname = $nickname == null ? $username : $nickname;
4495
4496 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4497 $nickname = $username;
4498 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4499 } elseif ( $fancySig !== false ) {
4500 # Sig. might contain markup; validate this
4501 if ( $this->validateSig( $nickname ) !== false ) {
4502 # Validated; clean up (if needed) and return it
4503 return $this->cleanSig( $nickname, true );
4504 } else {
4505 # Failed to validate; fall back to the default
4506 $nickname = $username;
4507 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4508 }
4509 }
4510
4511 # Make sure nickname doesnt get a sig in a sig
4512 $nickname = self::cleanSigInSig( $nickname );
4513
4514 # If we're still here, make it a link to the user page
4515 $userText = wfEscapeWikiText( $username );
4516 $nickText = wfEscapeWikiText( $nickname );
4517 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4518
4519 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4520 ->title( $this->getTitle() )->text();
4521 }
4522
4523 /**
4524 * Check that the user's signature contains no bad XML
4525 *
4526 * @param string $text
4527 * @return string|bool An expanded string, or false if invalid.
4528 */
4529 public function validateSig( $text ) {
4530 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4531 }
4532
4533 /**
4534 * Clean up signature text
4535 *
4536 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4537 * 2) Substitute all transclusions
4538 *
4539 * @param string $text
4540 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4541 * @return string Signature text
4542 */
4543 public function cleanSig( $text, $parsing = false ) {
4544 if ( !$parsing ) {
4545 global $wgTitle;
4546 $magicScopeVariable = $this->lock();
4547 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4548 }
4549
4550 # Option to disable this feature
4551 if ( !$this->mOptions->getCleanSignatures() ) {
4552 return $text;
4553 }
4554
4555 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4556 # => Move this logic to braceSubstitution()
4557 $substWord = MagicWord::get( 'subst' );
4558 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4559 $substText = '{{' . $substWord->getSynonym( 0 );
4560
4561 $text = preg_replace( $substRegex, $substText, $text );
4562 $text = self::cleanSigInSig( $text );
4563 $dom = $this->preprocessToDom( $text );
4564 $frame = $this->getPreprocessor()->newFrame();
4565 $text = $frame->expand( $dom );
4566
4567 if ( !$parsing ) {
4568 $text = $this->mStripState->unstripBoth( $text );
4569 }
4570
4571 return $text;
4572 }
4573
4574 /**
4575 * Strip 3, 4 or 5 tildes out of signatures.
4576 *
4577 * @param string $text
4578 * @return string Signature text with /~{3,5}/ removed
4579 */
4580 public static function cleanSigInSig( $text ) {
4581 $text = preg_replace( '/~{3,5}/', '', $text );
4582 return $text;
4583 }
4584
4585 /**
4586 * Set up some variables which are usually set up in parse()
4587 * so that an external function can call some class members with confidence
4588 *
4589 * @param Title|null $title
4590 * @param ParserOptions $options
4591 * @param int $outputType
4592 * @param bool $clearState
4593 */
4594 public function startExternalParse( Title $title = null, ParserOptions $options,
4595 $outputType, $clearState = true
4596 ) {
4597 $this->startParse( $title, $options, $outputType, $clearState );
4598 }
4599
4600 /**
4601 * @param Title|null $title
4602 * @param ParserOptions $options
4603 * @param int $outputType
4604 * @param bool $clearState
4605 */
4606 private function startParse( Title $title = null, ParserOptions $options,
4607 $outputType, $clearState = true
4608 ) {
4609 $this->setTitle( $title );
4610 $this->mOptions = $options;
4611 $this->setOutputType( $outputType );
4612 if ( $clearState ) {
4613 $this->clearState();
4614 }
4615 }
4616
4617 /**
4618 * Wrapper for preprocess()
4619 *
4620 * @param string $text The text to preprocess
4621 * @param ParserOptions $options Options
4622 * @param Title|null $title Title object or null to use $wgTitle
4623 * @return string
4624 */
4625 public function transformMsg( $text, $options, $title = null ) {
4626 static $executing = false;
4627
4628 # Guard against infinite recursion
4629 if ( $executing ) {
4630 return $text;
4631 }
4632 $executing = true;
4633
4634 if ( !$title ) {
4635 global $wgTitle;
4636 $title = $wgTitle;
4637 }
4638
4639 $text = $this->preprocess( $text, $title, $options );
4640
4641 $executing = false;
4642 return $text;
4643 }
4644
4645 /**
4646 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4647 * The callback should have the following form:
4648 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4649 *
4650 * Transform and return $text. Use $parser for any required context, e.g. use
4651 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4652 *
4653 * Hooks may return extended information by returning an array, of which the
4654 * first numbered element (index 0) must be the return string, and all other
4655 * entries are extracted into local variables within an internal function
4656 * in the Parser class.
4657 *
4658 * This interface (introduced r61913) appears to be undocumented, but
4659 * 'markerType' is used by some core tag hooks to override which strip
4660 * array their results are placed in. **Use great caution if attempting
4661 * this interface, as it is not documented and injudicious use could smash
4662 * private variables.**
4663 *
4664 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4665 * @param callable $callback The callback function (and object) to use for the tag
4666 * @throws MWException
4667 * @return callable|null The old value of the mTagHooks array associated with the hook
4668 */
4669 public function setHook( $tag, $callback ) {
4670 $tag = strtolower( $tag );
4671 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4672 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4673 }
4674 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4675 $this->mTagHooks[$tag] = $callback;
4676 if ( !in_array( $tag, $this->mStripList ) ) {
4677 $this->mStripList[] = $tag;
4678 }
4679
4680 return $oldVal;
4681 }
4682
4683 /**
4684 * As setHook(), but letting the contents be parsed.
4685 *
4686 * Transparent tag hooks are like regular XML-style tag hooks, except they
4687 * operate late in the transformation sequence, on HTML instead of wikitext.
4688 *
4689 * This is probably obsoleted by things dealing with parser frames?
4690 * The only extension currently using it is geoserver.
4691 *
4692 * @since 1.10
4693 * @todo better document or deprecate this
4694 *
4695 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4696 * @param callable $callback The callback function (and object) to use for the tag
4697 * @throws MWException
4698 * @return callable|null The old value of the mTagHooks array associated with the hook
4699 */
4700 public function setTransparentTagHook( $tag, $callback ) {
4701 $tag = strtolower( $tag );
4702 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4703 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4704 }
4705 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4706 $this->mTransparentTagHooks[$tag] = $callback;
4707
4708 return $oldVal;
4709 }
4710
4711 /**
4712 * Remove all tag hooks
4713 */
4714 public function clearTagHooks() {
4715 $this->mTagHooks = [];
4716 $this->mFunctionTagHooks = [];
4717 $this->mStripList = $this->mDefaultStripList;
4718 }
4719
4720 /**
4721 * Create a function, e.g. {{sum:1|2|3}}
4722 * The callback function should have the form:
4723 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4724 *
4725 * Or with Parser::SFH_OBJECT_ARGS:
4726 * function myParserFunction( $parser, $frame, $args ) { ... }
4727 *
4728 * The callback may either return the text result of the function, or an array with the text
4729 * in element 0, and a number of flags in the other elements. The names of the flags are
4730 * specified in the keys. Valid flags are:
4731 * found The text returned is valid, stop processing the template. This
4732 * is on by default.
4733 * nowiki Wiki markup in the return value should be escaped
4734 * isHTML The returned text is HTML, armour it against wikitext transformation
4735 *
4736 * @param string $id The magic word ID
4737 * @param callable $callback The callback function (and object) to use
4738 * @param int $flags A combination of the following flags:
4739 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4740 *
4741 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4742 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4743 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4744 * the arguments, and to control the way they are expanded.
4745 *
4746 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4747 * arguments, for instance:
4748 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4749 *
4750 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4751 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4752 * working if/when this is changed.
4753 *
4754 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4755 * expansion.
4756 *
4757 * Please read the documentation in includes/parser/Preprocessor.php for more information
4758 * about the methods available in PPFrame and PPNode.
4759 *
4760 * @throws MWException
4761 * @return string|callable The old callback function for this name, if any
4762 */
4763 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4764 global $wgContLang;
4765
4766 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4767 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4768
4769 # Add to function cache
4770 $mw = MagicWord::get( $id );
4771 if ( !$mw ) {
4772 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4773 }
4774
4775 $synonyms = $mw->getSynonyms();
4776 $sensitive = intval( $mw->isCaseSensitive() );
4777
4778 foreach ( $synonyms as $syn ) {
4779 # Case
4780 if ( !$sensitive ) {
4781 $syn = $wgContLang->lc( $syn );
4782 }
4783 # Add leading hash
4784 if ( !( $flags & self::SFH_NO_HASH ) ) {
4785 $syn = '#' . $syn;
4786 }
4787 # Remove trailing colon
4788 if ( substr( $syn, -1, 1 ) === ':' ) {
4789 $syn = substr( $syn, 0, -1 );
4790 }
4791 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4792 }
4793 return $oldVal;
4794 }
4795
4796 /**
4797 * Get all registered function hook identifiers
4798 *
4799 * @return array
4800 */
4801 public function getFunctionHooks() {
4802 return array_keys( $this->mFunctionHooks );
4803 }
4804
4805 /**
4806 * Create a tag function, e.g. "<test>some stuff</test>".
4807 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4808 * Unlike parser functions, their content is not preprocessed.
4809 * @param string $tag
4810 * @param callable $callback
4811 * @param int $flags
4812 * @throws MWException
4813 * @return null
4814 */
4815 public function setFunctionTagHook( $tag, $callback, $flags ) {
4816 $tag = strtolower( $tag );
4817 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4818 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4819 }
4820 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4821 $this->mFunctionTagHooks[$tag] : null;
4822 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4823
4824 if ( !in_array( $tag, $this->mStripList ) ) {
4825 $this->mStripList[] = $tag;
4826 }
4827
4828 return $old;
4829 }
4830
4831 /**
4832 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4833 * Placeholders created in Linker::link()
4834 *
4835 * @param string $text
4836 * @param int $options
4837 */
4838 public function replaceLinkHolders( &$text, $options = 0 ) {
4839 $this->mLinkHolders->replace( $text );
4840 }
4841
4842 /**
4843 * Replace "<!--LINK-->" link placeholders with plain text of links
4844 * (not HTML-formatted).
4845 *
4846 * @param string $text
4847 * @return string
4848 */
4849 public function replaceLinkHoldersText( $text ) {
4850 return $this->mLinkHolders->replaceText( $text );
4851 }
4852
4853 /**
4854 * Renders an image gallery from a text with one line per image.
4855 * text labels may be given by using |-style alternative text. E.g.
4856 * Image:one.jpg|The number "1"
4857 * Image:tree.jpg|A tree
4858 * given as text will return the HTML of a gallery with two images,
4859 * labeled 'The number "1"' and
4860 * 'A tree'.
4861 *
4862 * @param string $text
4863 * @param array $params
4864 * @return string HTML
4865 */
4866 public function renderImageGallery( $text, $params ) {
4867
4868 $mode = false;
4869 if ( isset( $params['mode'] ) ) {
4870 $mode = $params['mode'];
4871 }
4872
4873 try {
4874 $ig = ImageGalleryBase::factory( $mode );
4875 } catch ( Exception $e ) {
4876 // If invalid type set, fallback to default.
4877 $ig = ImageGalleryBase::factory( false );
4878 }
4879
4880 $ig->setContextTitle( $this->mTitle );
4881 $ig->setShowBytes( false );
4882 $ig->setShowFilename( false );
4883 $ig->setParser( $this );
4884 $ig->setHideBadImages();
4885 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4886
4887 if ( isset( $params['showfilename'] ) ) {
4888 $ig->setShowFilename( true );
4889 } else {
4890 $ig->setShowFilename( false );
4891 }
4892 if ( isset( $params['caption'] ) ) {
4893 $caption = $params['caption'];
4894 $caption = htmlspecialchars( $caption );
4895 $caption = $this->replaceInternalLinks( $caption );
4896 $ig->setCaptionHtml( $caption );
4897 }
4898 if ( isset( $params['perrow'] ) ) {
4899 $ig->setPerRow( $params['perrow'] );
4900 }
4901 if ( isset( $params['widths'] ) ) {
4902 $ig->setWidths( $params['widths'] );
4903 }
4904 if ( isset( $params['heights'] ) ) {
4905 $ig->setHeights( $params['heights'] );
4906 }
4907 $ig->setAdditionalOptions( $params );
4908
4909 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4910
4911 $lines = StringUtils::explode( "\n", $text );
4912 foreach ( $lines as $line ) {
4913 # match lines like these:
4914 # Image:someimage.jpg|This is some image
4915 $matches = [];
4916 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4917 # Skip empty lines
4918 if ( count( $matches ) == 0 ) {
4919 continue;
4920 }
4921
4922 if ( strpos( $matches[0], '%' ) !== false ) {
4923 $matches[1] = rawurldecode( $matches[1] );
4924 }
4925 $title = Title::newFromText( $matches[1], NS_FILE );
4926 if ( is_null( $title ) ) {
4927 # Bogus title. Ignore these so we don't bomb out later.
4928 continue;
4929 }
4930
4931 # We need to get what handler the file uses, to figure out parameters.
4932 # Note, a hook can overide the file name, and chose an entirely different
4933 # file (which potentially could be of a different type and have different handler).
4934 $options = [];
4935 $descQuery = false;
4936 Hooks::run( 'BeforeParserFetchFileAndTitle',
4937 [ $this, $title, &$options, &$descQuery ] );
4938 # Don't register it now, as ImageGallery does that later.
4939 $file = $this->fetchFileNoRegister( $title, $options );
4940 $handler = $file ? $file->getHandler() : false;
4941
4942 $paramMap = [
4943 'img_alt' => 'gallery-internal-alt',
4944 'img_link' => 'gallery-internal-link',
4945 ];
4946 if ( $handler ) {
4947 $paramMap = $paramMap + $handler->getParamMap();
4948 // We don't want people to specify per-image widths.
4949 // Additionally the width parameter would need special casing anyhow.
4950 unset( $paramMap['img_width'] );
4951 }
4952
4953 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4954
4955 $label = '';
4956 $alt = '';
4957 $link = '';
4958 $handlerOptions = [];
4959 if ( isset( $matches[3] ) ) {
4960 // look for an |alt= definition while trying not to break existing
4961 // captions with multiple pipes (|) in it, until a more sensible grammar
4962 // is defined for images in galleries
4963
4964 // FIXME: Doing recursiveTagParse at this stage, and the trim before
4965 // splitting on '|' is a bit odd, and different from makeImage.
4966 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4967 $parameterMatches = StringUtils::explode( '|', $matches[3] );
4968
4969 foreach ( $parameterMatches as $parameterMatch ) {
4970 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
4971 if ( $magicName ) {
4972 $paramName = $paramMap[$magicName];
4973
4974 switch ( $paramName ) {
4975 case 'gallery-internal-alt':
4976 $alt = $this->stripAltText( $match, false );
4977 break;
4978 case 'gallery-internal-link':
4979 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
4980 $chars = self::EXT_LINK_URL_CLASS;
4981 $addr = self::EXT_LINK_ADDR;
4982 $prots = $this->mUrlProtocols;
4983 // check to see if link matches an absolute url, if not then it must be a wiki link.
4984 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
4985 $link = $linkValue;
4986 } else {
4987 $localLinkTitle = Title::newFromText( $linkValue );
4988 if ( $localLinkTitle !== null ) {
4989 $link = $localLinkTitle->getLinkURL();
4990 }
4991 }
4992 break;
4993 default:
4994 // Must be a handler specific parameter.
4995 if ( $handler->validateParam( $paramName, $match ) ) {
4996 $handlerOptions[$paramName] = $match;
4997 } else {
4998 // Guess not, consider it as caption.
4999 wfDebug( "$parameterMatch failed parameter validation\n" );
5000 $label = '|' . $parameterMatch;
5001 }
5002 }
5003
5004 } else {
5005 // Last pipe wins.
5006 $label = '|' . $parameterMatch;
5007 }
5008 }
5009 // Remove the pipe.
5010 $label = substr( $label, 1 );
5011 }
5012
5013 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5014 }
5015 $html = $ig->toHTML();
5016 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5017 return $html;
5018 }
5019
5020 /**
5021 * @param MediaHandler $handler
5022 * @return array
5023 */
5024 public function getImageParams( $handler ) {
5025 if ( $handler ) {
5026 $handlerClass = get_class( $handler );
5027 } else {
5028 $handlerClass = '';
5029 }
5030 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5031 # Initialise static lists
5032 static $internalParamNames = [
5033 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5034 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5035 'bottom', 'text-bottom' ],
5036 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5037 'upright', 'border', 'link', 'alt', 'class' ],
5038 ];
5039 static $internalParamMap;
5040 if ( !$internalParamMap ) {
5041 $internalParamMap = [];
5042 foreach ( $internalParamNames as $type => $names ) {
5043 foreach ( $names as $name ) {
5044 $magicName = str_replace( '-', '_', "img_$name" );
5045 $internalParamMap[$magicName] = [ $type, $name ];
5046 }
5047 }
5048 }
5049
5050 # Add handler params
5051 $paramMap = $internalParamMap;
5052 if ( $handler ) {
5053 $handlerParamMap = $handler->getParamMap();
5054 foreach ( $handlerParamMap as $magic => $paramName ) {
5055 $paramMap[$magic] = [ 'handler', $paramName ];
5056 }
5057 }
5058 $this->mImageParams[$handlerClass] = $paramMap;
5059 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5060 }
5061 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5062 }
5063
5064 /**
5065 * Parse image options text and use it to make an image
5066 *
5067 * @param Title $title
5068 * @param string $options
5069 * @param LinkHolderArray|bool $holders
5070 * @return string HTML
5071 */
5072 public function makeImage( $title, $options, $holders = false ) {
5073 # Check if the options text is of the form "options|alt text"
5074 # Options are:
5075 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5076 # * left no resizing, just left align. label is used for alt= only
5077 # * right same, but right aligned
5078 # * none same, but not aligned
5079 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5080 # * center center the image
5081 # * frame Keep original image size, no magnify-button.
5082 # * framed Same as "frame"
5083 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5084 # * upright reduce width for upright images, rounded to full __0 px
5085 # * border draw a 1px border around the image
5086 # * alt Text for HTML alt attribute (defaults to empty)
5087 # * class Set a class for img node
5088 # * link Set the target of the image link. Can be external, interwiki, or local
5089 # vertical-align values (no % or length right now):
5090 # * baseline
5091 # * sub
5092 # * super
5093 # * top
5094 # * text-top
5095 # * middle
5096 # * bottom
5097 # * text-bottom
5098
5099 $parts = StringUtils::explode( "|", $options );
5100
5101 # Give extensions a chance to select the file revision for us
5102 $options = [];
5103 $descQuery = false;
5104 Hooks::run( 'BeforeParserFetchFileAndTitle',
5105 [ $this, $title, &$options, &$descQuery ] );
5106 # Fetch and register the file (file title may be different via hooks)
5107 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5108
5109 # Get parameter map
5110 $handler = $file ? $file->getHandler() : false;
5111
5112 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5113
5114 if ( !$file ) {
5115 $this->addTrackingCategory( 'broken-file-category' );
5116 }
5117
5118 # Process the input parameters
5119 $caption = '';
5120 $params = [ 'frame' => [], 'handler' => [],
5121 'horizAlign' => [], 'vertAlign' => [] ];
5122 $seenformat = false;
5123 foreach ( $parts as $part ) {
5124 $part = trim( $part );
5125 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5126 $validated = false;
5127 if ( isset( $paramMap[$magicName] ) ) {
5128 list( $type, $paramName ) = $paramMap[$magicName];
5129
5130 # Special case; width and height come in one variable together
5131 if ( $type === 'handler' && $paramName === 'width' ) {
5132 $parsedWidthParam = $this->parseWidthParam( $value );
5133 if ( isset( $parsedWidthParam['width'] ) ) {
5134 $width = $parsedWidthParam['width'];
5135 if ( $handler->validateParam( 'width', $width ) ) {
5136 $params[$type]['width'] = $width;
5137 $validated = true;
5138 }
5139 }
5140 if ( isset( $parsedWidthParam['height'] ) ) {
5141 $height = $parsedWidthParam['height'];
5142 if ( $handler->validateParam( 'height', $height ) ) {
5143 $params[$type]['height'] = $height;
5144 $validated = true;
5145 }
5146 }
5147 # else no validation -- bug 13436
5148 } else {
5149 if ( $type === 'handler' ) {
5150 # Validate handler parameter
5151 $validated = $handler->validateParam( $paramName, $value );
5152 } else {
5153 # Validate internal parameters
5154 switch ( $paramName ) {
5155 case 'manualthumb':
5156 case 'alt':
5157 case 'class':
5158 # @todo FIXME: Possibly check validity here for
5159 # manualthumb? downstream behavior seems odd with
5160 # missing manual thumbs.
5161 $validated = true;
5162 $value = $this->stripAltText( $value, $holders );
5163 break;
5164 case 'link':
5165 $chars = self::EXT_LINK_URL_CLASS;
5166 $addr = self::EXT_LINK_ADDR;
5167 $prots = $this->mUrlProtocols;
5168 if ( $value === '' ) {
5169 $paramName = 'no-link';
5170 $value = true;
5171 $validated = true;
5172 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5173 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5174 $paramName = 'link-url';
5175 $this->mOutput->addExternalLink( $value );
5176 if ( $this->mOptions->getExternalLinkTarget() ) {
5177 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5178 }
5179 $validated = true;
5180 }
5181 } else {
5182 $linkTitle = Title::newFromText( $value );
5183 if ( $linkTitle ) {
5184 $paramName = 'link-title';
5185 $value = $linkTitle;
5186 $this->mOutput->addLink( $linkTitle );
5187 $validated = true;
5188 }
5189 }
5190 break;
5191 case 'frameless':
5192 case 'framed':
5193 case 'thumbnail':
5194 // use first appearing option, discard others.
5195 $validated = ! $seenformat;
5196 $seenformat = true;
5197 break;
5198 default:
5199 # Most other things appear to be empty or numeric...
5200 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5201 }
5202 }
5203
5204 if ( $validated ) {
5205 $params[$type][$paramName] = $value;
5206 }
5207 }
5208 }
5209 if ( !$validated ) {
5210 $caption = $part;
5211 }
5212 }
5213
5214 # Process alignment parameters
5215 if ( $params['horizAlign'] ) {
5216 $params['frame']['align'] = key( $params['horizAlign'] );
5217 }
5218 if ( $params['vertAlign'] ) {
5219 $params['frame']['valign'] = key( $params['vertAlign'] );
5220 }
5221
5222 $params['frame']['caption'] = $caption;
5223
5224 # Will the image be presented in a frame, with the caption below?
5225 $imageIsFramed = isset( $params['frame']['frame'] )
5226 || isset( $params['frame']['framed'] )
5227 || isset( $params['frame']['thumbnail'] )
5228 || isset( $params['frame']['manualthumb'] );
5229
5230 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5231 # came to also set the caption, ordinary text after the image -- which
5232 # makes no sense, because that just repeats the text multiple times in
5233 # screen readers. It *also* came to set the title attribute.
5234 # Now that we have an alt attribute, we should not set the alt text to
5235 # equal the caption: that's worse than useless, it just repeats the
5236 # text. This is the framed/thumbnail case. If there's no caption, we
5237 # use the unnamed parameter for alt text as well, just for the time be-
5238 # ing, if the unnamed param is set and the alt param is not.
5239 # For the future, we need to figure out if we want to tweak this more,
5240 # e.g., introducing a title= parameter for the title; ignoring the un-
5241 # named parameter entirely for images without a caption; adding an ex-
5242 # plicit caption= parameter and preserving the old magic unnamed para-
5243 # meter for BC; ...
5244 if ( $imageIsFramed ) { # Framed image
5245 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5246 # No caption or alt text, add the filename as the alt text so
5247 # that screen readers at least get some description of the image
5248 $params['frame']['alt'] = $title->getText();
5249 }
5250 # Do not set $params['frame']['title'] because tooltips don't make sense
5251 # for framed images
5252 } else { # Inline image
5253 if ( !isset( $params['frame']['alt'] ) ) {
5254 # No alt text, use the "caption" for the alt text
5255 if ( $caption !== '' ) {
5256 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5257 } else {
5258 # No caption, fall back to using the filename for the
5259 # alt text
5260 $params['frame']['alt'] = $title->getText();
5261 }
5262 }
5263 # Use the "caption" for the tooltip text
5264 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5265 }
5266
5267 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5268
5269 # Linker does the rest
5270 $time = isset( $options['time'] ) ? $options['time'] : false;
5271 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5272 $time, $descQuery, $this->mOptions->getThumbSize() );
5273
5274 # Give the handler a chance to modify the parser object
5275 if ( $handler ) {
5276 $handler->parserTransformHook( $this, $file );
5277 }
5278
5279 return $ret;
5280 }
5281
5282 /**
5283 * @param string $caption
5284 * @param LinkHolderArray|bool $holders
5285 * @return mixed|string
5286 */
5287 protected function stripAltText( $caption, $holders ) {
5288 # Strip bad stuff out of the title (tooltip). We can't just use
5289 # replaceLinkHoldersText() here, because if this function is called
5290 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5291 if ( $holders ) {
5292 $tooltip = $holders->replaceText( $caption );
5293 } else {
5294 $tooltip = $this->replaceLinkHoldersText( $caption );
5295 }
5296
5297 # make sure there are no placeholders in thumbnail attributes
5298 # that are later expanded to html- so expand them now and
5299 # remove the tags
5300 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5301 $tooltip = Sanitizer::stripAllTags( $tooltip );
5302
5303 return $tooltip;
5304 }
5305
5306 /**
5307 * Set a flag in the output object indicating that the content is dynamic and
5308 * shouldn't be cached.
5309 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5310 */
5311 public function disableCache() {
5312 wfDebug( "Parser output marked as uncacheable.\n" );
5313 if ( !$this->mOutput ) {
5314 throw new MWException( __METHOD__ .
5315 " can only be called when actually parsing something" );
5316 }
5317 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5318 }
5319
5320 /**
5321 * Callback from the Sanitizer for expanding items found in HTML attribute
5322 * values, so they can be safely tested and escaped.
5323 *
5324 * @param string $text
5325 * @param bool|PPFrame $frame
5326 * @return string
5327 */
5328 public function attributeStripCallback( &$text, $frame = false ) {
5329 $text = $this->replaceVariables( $text, $frame );
5330 $text = $this->mStripState->unstripBoth( $text );
5331 return $text;
5332 }
5333
5334 /**
5335 * Accessor
5336 *
5337 * @return array
5338 */
5339 public function getTags() {
5340 return array_merge(
5341 array_keys( $this->mTransparentTagHooks ),
5342 array_keys( $this->mTagHooks ),
5343 array_keys( $this->mFunctionTagHooks )
5344 );
5345 }
5346
5347 /**
5348 * Replace transparent tags in $text with the values given by the callbacks.
5349 *
5350 * Transparent tag hooks are like regular XML-style tag hooks, except they
5351 * operate late in the transformation sequence, on HTML instead of wikitext.
5352 *
5353 * @param string $text
5354 *
5355 * @return string
5356 */
5357 public function replaceTransparentTags( $text ) {
5358 $matches = [];
5359 $elements = array_keys( $this->mTransparentTagHooks );
5360 $text = self::extractTagsAndParams( $elements, $text, $matches );
5361 $replacements = [];
5362
5363 foreach ( $matches as $marker => $data ) {
5364 list( $element, $content, $params, $tag ) = $data;
5365 $tagName = strtolower( $element );
5366 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5367 $output = call_user_func_array(
5368 $this->mTransparentTagHooks[$tagName],
5369 [ $content, $params, $this ]
5370 );
5371 } else {
5372 $output = $tag;
5373 }
5374 $replacements[$marker] = $output;
5375 }
5376 return strtr( $text, $replacements );
5377 }
5378
5379 /**
5380 * Break wikitext input into sections, and either pull or replace
5381 * some particular section's text.
5382 *
5383 * External callers should use the getSection and replaceSection methods.
5384 *
5385 * @param string $text Page wikitext
5386 * @param string|number $sectionId A section identifier string of the form:
5387 * "<flag1> - <flag2> - ... - <section number>"
5388 *
5389 * Currently the only recognised flag is "T", which means the target section number
5390 * was derived during a template inclusion parse, in other words this is a template
5391 * section edit link. If no flags are given, it was an ordinary section edit link.
5392 * This flag is required to avoid a section numbering mismatch when a section is
5393 * enclosed by "<includeonly>" (bug 6563).
5394 *
5395 * The section number 0 pulls the text before the first heading; other numbers will
5396 * pull the given section along with its lower-level subsections. If the section is
5397 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5398 *
5399 * Section 0 is always considered to exist, even if it only contains the empty
5400 * string. If $text is the empty string and section 0 is replaced, $newText is
5401 * returned.
5402 *
5403 * @param string $mode One of "get" or "replace"
5404 * @param string $newText Replacement text for section data.
5405 * @return string For "get", the extracted section text.
5406 * for "replace", the whole page with the section replaced.
5407 */
5408 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5409 global $wgTitle; # not generally used but removes an ugly failure mode
5410
5411 $magicScopeVariable = $this->lock();
5412 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5413 $outText = '';
5414 $frame = $this->getPreprocessor()->newFrame();
5415
5416 # Process section extraction flags
5417 $flags = 0;
5418 $sectionParts = explode( '-', $sectionId );
5419 $sectionIndex = array_pop( $sectionParts );
5420 foreach ( $sectionParts as $part ) {
5421 if ( $part === 'T' ) {
5422 $flags |= self::PTD_FOR_INCLUSION;
5423 }
5424 }
5425
5426 # Check for empty input
5427 if ( strval( $text ) === '' ) {
5428 # Only sections 0 and T-0 exist in an empty document
5429 if ( $sectionIndex == 0 ) {
5430 if ( $mode === 'get' ) {
5431 return '';
5432 } else {
5433 return $newText;
5434 }
5435 } else {
5436 if ( $mode === 'get' ) {
5437 return $newText;
5438 } else {
5439 return $text;
5440 }
5441 }
5442 }
5443
5444 # Preprocess the text
5445 $root = $this->preprocessToDom( $text, $flags );
5446
5447 # <h> nodes indicate section breaks
5448 # They can only occur at the top level, so we can find them by iterating the root's children
5449 $node = $root->getFirstChild();
5450
5451 # Find the target section
5452 if ( $sectionIndex == 0 ) {
5453 # Section zero doesn't nest, level=big
5454 $targetLevel = 1000;
5455 } else {
5456 while ( $node ) {
5457 if ( $node->getName() === 'h' ) {
5458 $bits = $node->splitHeading();
5459 if ( $bits['i'] == $sectionIndex ) {
5460 $targetLevel = $bits['level'];
5461 break;
5462 }
5463 }
5464 if ( $mode === 'replace' ) {
5465 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5466 }
5467 $node = $node->getNextSibling();
5468 }
5469 }
5470
5471 if ( !$node ) {
5472 # Not found
5473 if ( $mode === 'get' ) {
5474 return $newText;
5475 } else {
5476 return $text;
5477 }
5478 }
5479
5480 # Find the end of the section, including nested sections
5481 do {
5482 if ( $node->getName() === 'h' ) {
5483 $bits = $node->splitHeading();
5484 $curLevel = $bits['level'];
5485 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5486 break;
5487 }
5488 }
5489 if ( $mode === 'get' ) {
5490 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5491 }
5492 $node = $node->getNextSibling();
5493 } while ( $node );
5494
5495 # Write out the remainder (in replace mode only)
5496 if ( $mode === 'replace' ) {
5497 # Output the replacement text
5498 # Add two newlines on -- trailing whitespace in $newText is conventionally
5499 # stripped by the editor, so we need both newlines to restore the paragraph gap
5500 # Only add trailing whitespace if there is newText
5501 if ( $newText != "" ) {
5502 $outText .= $newText . "\n\n";
5503 }
5504
5505 while ( $node ) {
5506 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5507 $node = $node->getNextSibling();
5508 }
5509 }
5510
5511 if ( is_string( $outText ) ) {
5512 # Re-insert stripped tags
5513 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5514 }
5515
5516 return $outText;
5517 }
5518
5519 /**
5520 * This function returns the text of a section, specified by a number ($section).
5521 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5522 * the first section before any such heading (section 0).
5523 *
5524 * If a section contains subsections, these are also returned.
5525 *
5526 * @param string $text Text to look in
5527 * @param string|number $sectionId Section identifier as a number or string
5528 * (e.g. 0, 1 or 'T-1').
5529 * @param string $defaultText Default to return if section is not found
5530 *
5531 * @return string Text of the requested section
5532 */
5533 public function getSection( $text, $sectionId, $defaultText = '' ) {
5534 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5535 }
5536
5537 /**
5538 * This function returns $oldtext after the content of the section
5539 * specified by $section has been replaced with $text. If the target
5540 * section does not exist, $oldtext is returned unchanged.
5541 *
5542 * @param string $oldText Former text of the article
5543 * @param string|number $sectionId Section identifier as a number or string
5544 * (e.g. 0, 1 or 'T-1').
5545 * @param string $newText Replacing text
5546 *
5547 * @return string Modified text
5548 */
5549 public function replaceSection( $oldText, $sectionId, $newText ) {
5550 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5551 }
5552
5553 /**
5554 * Get the ID of the revision we are parsing
5555 *
5556 * @return int|null
5557 */
5558 public function getRevisionId() {
5559 return $this->mRevisionId;
5560 }
5561
5562 /**
5563 * Get the revision object for $this->mRevisionId
5564 *
5565 * @return Revision|null Either a Revision object or null
5566 * @since 1.23 (public since 1.23)
5567 */
5568 public function getRevisionObject() {
5569 if ( !is_null( $this->mRevisionObject ) ) {
5570 return $this->mRevisionObject;
5571 }
5572 if ( is_null( $this->mRevisionId ) ) {
5573 return null;
5574 }
5575
5576 $rev = call_user_func(
5577 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5578 );
5579
5580 # If the parse is for a new revision, then the callback should have
5581 # already been set to force the object and should match mRevisionId.
5582 # If not, try to fetch by mRevisionId for sanity.
5583 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5584 $rev = Revision::newFromId( $this->mRevisionId );
5585 }
5586
5587 $this->mRevisionObject = $rev;
5588
5589 return $this->mRevisionObject;
5590 }
5591
5592 /**
5593 * Get the timestamp associated with the current revision, adjusted for
5594 * the default server-local timestamp
5595 * @return string
5596 */
5597 public function getRevisionTimestamp() {
5598 if ( is_null( $this->mRevisionTimestamp ) ) {
5599 global $wgContLang;
5600
5601 $revObject = $this->getRevisionObject();
5602 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5603
5604 # The cryptic '' timezone parameter tells to use the site-default
5605 # timezone offset instead of the user settings.
5606 # Since this value will be saved into the parser cache, served
5607 # to other users, and potentially even used inside links and such,
5608 # it needs to be consistent for all visitors.
5609 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5610
5611 }
5612 return $this->mRevisionTimestamp;
5613 }
5614
5615 /**
5616 * Get the name of the user that edited the last revision
5617 *
5618 * @return string User name
5619 */
5620 public function getRevisionUser() {
5621 if ( is_null( $this->mRevisionUser ) ) {
5622 $revObject = $this->getRevisionObject();
5623
5624 # if this template is subst: the revision id will be blank,
5625 # so just use the current user's name
5626 if ( $revObject ) {
5627 $this->mRevisionUser = $revObject->getUserText();
5628 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5629 $this->mRevisionUser = $this->getUser()->getName();
5630 }
5631 }
5632 return $this->mRevisionUser;
5633 }
5634
5635 /**
5636 * Get the size of the revision
5637 *
5638 * @return int|null Revision size
5639 */
5640 public function getRevisionSize() {
5641 if ( is_null( $this->mRevisionSize ) ) {
5642 $revObject = $this->getRevisionObject();
5643
5644 # if this variable is subst: the revision id will be blank,
5645 # so just use the parser input size, because the own substituation
5646 # will change the size.
5647 if ( $revObject ) {
5648 $this->mRevisionSize = $revObject->getSize();
5649 } else {
5650 $this->mRevisionSize = $this->mInputSize;
5651 }
5652 }
5653 return $this->mRevisionSize;
5654 }
5655
5656 /**
5657 * Mutator for $mDefaultSort
5658 *
5659 * @param string $sort New value
5660 */
5661 public function setDefaultSort( $sort ) {
5662 $this->mDefaultSort = $sort;
5663 $this->mOutput->setProperty( 'defaultsort', $sort );
5664 }
5665
5666 /**
5667 * Accessor for $mDefaultSort
5668 * Will use the empty string if none is set.
5669 *
5670 * This value is treated as a prefix, so the
5671 * empty string is equivalent to sorting by
5672 * page name.
5673 *
5674 * @return string
5675 */
5676 public function getDefaultSort() {
5677 if ( $this->mDefaultSort !== false ) {
5678 return $this->mDefaultSort;
5679 } else {
5680 return '';
5681 }
5682 }
5683
5684 /**
5685 * Accessor for $mDefaultSort
5686 * Unlike getDefaultSort(), will return false if none is set
5687 *
5688 * @return string|bool
5689 */
5690 public function getCustomDefaultSort() {
5691 return $this->mDefaultSort;
5692 }
5693
5694 /**
5695 * Try to guess the section anchor name based on a wikitext fragment
5696 * presumably extracted from a heading, for example "Header" from
5697 * "== Header ==".
5698 *
5699 * @param string $text
5700 *
5701 * @return string
5702 */
5703 public function guessSectionNameFromWikiText( $text ) {
5704 # Strip out wikitext links(they break the anchor)
5705 $text = $this->stripSectionName( $text );
5706 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5707 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5708 }
5709
5710 /**
5711 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5712 * instead. For use in redirects, since IE6 interprets Redirect: headers
5713 * as something other than UTF-8 (apparently?), resulting in breakage.
5714 *
5715 * @param string $text The section name
5716 * @return string An anchor
5717 */
5718 public function guessLegacySectionNameFromWikiText( $text ) {
5719 # Strip out wikitext links(they break the anchor)
5720 $text = $this->stripSectionName( $text );
5721 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5722 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5723 }
5724
5725 /**
5726 * Strips a text string of wikitext for use in a section anchor
5727 *
5728 * Accepts a text string and then removes all wikitext from the
5729 * string and leaves only the resultant text (i.e. the result of
5730 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5731 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5732 * to create valid section anchors by mimicing the output of the
5733 * parser when headings are parsed.
5734 *
5735 * @param string $text Text string to be stripped of wikitext
5736 * for use in a Section anchor
5737 * @return string Filtered text string
5738 */
5739 public function stripSectionName( $text ) {
5740 # Strip internal link markup
5741 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5742 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5743
5744 # Strip external link markup
5745 # @todo FIXME: Not tolerant to blank link text
5746 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5747 # on how many empty links there are on the page - need to figure that out.
5748 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5749
5750 # Parse wikitext quotes (italics & bold)
5751 $text = $this->doQuotes( $text );
5752
5753 # Strip HTML tags
5754 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5755 return $text;
5756 }
5757
5758 /**
5759 * strip/replaceVariables/unstrip for preprocessor regression testing
5760 *
5761 * @param string $text
5762 * @param Title $title
5763 * @param ParserOptions $options
5764 * @param int $outputType
5765 *
5766 * @return string
5767 */
5768 public function testSrvus( $text, Title $title, ParserOptions $options,
5769 $outputType = self::OT_HTML
5770 ) {
5771 $magicScopeVariable = $this->lock();
5772 $this->startParse( $title, $options, $outputType, true );
5773
5774 $text = $this->replaceVariables( $text );
5775 $text = $this->mStripState->unstripBoth( $text );
5776 $text = Sanitizer::removeHTMLtags( $text );
5777 return $text;
5778 }
5779
5780 /**
5781 * @param string $text
5782 * @param Title $title
5783 * @param ParserOptions $options
5784 * @return string
5785 */
5786 public function testPst( $text, Title $title, ParserOptions $options ) {
5787 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5788 }
5789
5790 /**
5791 * @param string $text
5792 * @param Title $title
5793 * @param ParserOptions $options
5794 * @return string
5795 */
5796 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5797 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5798 }
5799
5800 /**
5801 * Call a callback function on all regions of the given text that are not
5802 * inside strip markers, and replace those regions with the return value
5803 * of the callback. For example, with input:
5804 *
5805 * aaa<MARKER>bbb
5806 *
5807 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5808 * two strings will be replaced with the value returned by the callback in
5809 * each case.
5810 *
5811 * @param string $s
5812 * @param callable $callback
5813 *
5814 * @return string
5815 */
5816 public function markerSkipCallback( $s, $callback ) {
5817 $i = 0;
5818 $out = '';
5819 while ( $i < strlen( $s ) ) {
5820 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5821 if ( $markerStart === false ) {
5822 $out .= call_user_func( $callback, substr( $s, $i ) );
5823 break;
5824 } else {
5825 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5826 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5827 if ( $markerEnd === false ) {
5828 $out .= substr( $s, $markerStart );
5829 break;
5830 } else {
5831 $markerEnd += strlen( self::MARKER_SUFFIX );
5832 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5833 $i = $markerEnd;
5834 }
5835 }
5836 }
5837 return $out;
5838 }
5839
5840 /**
5841 * Remove any strip markers found in the given text.
5842 *
5843 * @param string $text Input string
5844 * @return string
5845 */
5846 public function killMarkers( $text ) {
5847 return $this->mStripState->killMarkers( $text );
5848 }
5849
5850 /**
5851 * Save the parser state required to convert the given half-parsed text to
5852 * HTML. "Half-parsed" in this context means the output of
5853 * recursiveTagParse() or internalParse(). This output has strip markers
5854 * from replaceVariables (extensionSubstitution() etc.), and link
5855 * placeholders from replaceLinkHolders().
5856 *
5857 * Returns an array which can be serialized and stored persistently. This
5858 * array can later be loaded into another parser instance with
5859 * unserializeHalfParsedText(). The text can then be safely incorporated into
5860 * the return value of a parser hook.
5861 *
5862 * @param string $text
5863 *
5864 * @return array
5865 */
5866 public function serializeHalfParsedText( $text ) {
5867 $data = [
5868 'text' => $text,
5869 'version' => self::HALF_PARSED_VERSION,
5870 'stripState' => $this->mStripState->getSubState( $text ),
5871 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5872 ];
5873 return $data;
5874 }
5875
5876 /**
5877 * Load the parser state given in the $data array, which is assumed to
5878 * have been generated by serializeHalfParsedText(). The text contents is
5879 * extracted from the array, and its markers are transformed into markers
5880 * appropriate for the current Parser instance. This transformed text is
5881 * returned, and can be safely included in the return value of a parser
5882 * hook.
5883 *
5884 * If the $data array has been stored persistently, the caller should first
5885 * check whether it is still valid, by calling isValidHalfParsedText().
5886 *
5887 * @param array $data Serialized data
5888 * @throws MWException
5889 * @return string
5890 */
5891 public function unserializeHalfParsedText( $data ) {
5892 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5893 throw new MWException( __METHOD__ . ': invalid version' );
5894 }
5895
5896 # First, extract the strip state.
5897 $texts = [ $data['text'] ];
5898 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5899
5900 # Now renumber links
5901 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5902
5903 # Should be good to go.
5904 return $texts[0];
5905 }
5906
5907 /**
5908 * Returns true if the given array, presumed to be generated by
5909 * serializeHalfParsedText(), is compatible with the current version of the
5910 * parser.
5911 *
5912 * @param array $data
5913 *
5914 * @return bool
5915 */
5916 public function isValidHalfParsedText( $data ) {
5917 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5918 }
5919
5920 /**
5921 * Parsed a width param of imagelink like 300px or 200x300px
5922 *
5923 * @param string $value
5924 *
5925 * @return array
5926 * @since 1.20
5927 */
5928 public function parseWidthParam( $value ) {
5929 $parsedWidthParam = [];
5930 if ( $value === '' ) {
5931 return $parsedWidthParam;
5932 }
5933 $m = [];
5934 # (bug 13500) In both cases (width/height and width only),
5935 # permit trailing "px" for backward compatibility.
5936 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5937 $width = intval( $m[1] );
5938 $height = intval( $m[2] );
5939 $parsedWidthParam['width'] = $width;
5940 $parsedWidthParam['height'] = $height;
5941 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5942 $width = intval( $value );
5943 $parsedWidthParam['width'] = $width;
5944 }
5945 return $parsedWidthParam;
5946 }
5947
5948 /**
5949 * Lock the current instance of the parser.
5950 *
5951 * This is meant to stop someone from calling the parser
5952 * recursively and messing up all the strip state.
5953 *
5954 * @throws MWException If parser is in a parse
5955 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5956 */
5957 protected function lock() {
5958 if ( $this->mInParse ) {
5959 throw new MWException( "Parser state cleared while parsing. "
5960 . "Did you call Parser::parse recursively?" );
5961 }
5962 $this->mInParse = true;
5963
5964 $recursiveCheck = new ScopedCallback( function() {
5965 $this->mInParse = false;
5966 } );
5967
5968 return $recursiveCheck;
5969 }
5970
5971 /**
5972 * Strip outer <p></p> tag from the HTML source of a single paragraph.
5973 *
5974 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
5975 * or if there is more than one <p/> tag in the input HTML.
5976 *
5977 * @param string $html
5978 * @return string
5979 * @since 1.24
5980 */
5981 public static function stripOuterParagraph( $html ) {
5982 $m = [];
5983 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
5984 if ( strpos( $m[1], '</p>' ) === false ) {
5985 $html = $m[1];
5986 }
5987 }
5988
5989 return $html;
5990 }
5991
5992 /**
5993 * Return this parser if it is not doing anything, otherwise
5994 * get a fresh parser. You can use this method by doing
5995 * $myParser = $wgParser->getFreshParser(), or more simply
5996 * $wgParser->getFreshParser()->parse( ... );
5997 * if you're unsure if $wgParser is safe to use.
5998 *
5999 * @since 1.24
6000 * @return Parser A parser object that is not parsing anything
6001 */
6002 public function getFreshParser() {
6003 global $wgParserConf;
6004 if ( $this->mInParse ) {
6005 return new $wgParserConf['class']( $wgParserConf );
6006 } else {
6007 return $this;
6008 }
6009 }
6010
6011 /**
6012 * Set's up the PHP implementation of OOUI for use in this request
6013 * and instructs OutputPage to enable OOUI for itself.
6014 *
6015 * @since 1.26
6016 */
6017 public function enableOOUI() {
6018 OutputPage::setupOOUI();
6019 $this->mOutput->setEnableOOUI( true );
6020 }
6021 }
Wiklou, le Wiki du Wiklou