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