dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Add User::isBot() method"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23
24 /**
25 * @defgroup Parser Parser
26 */
27
28 /**
29 * PHP Parser - Processes wiki markup (which uses a more user-friendly
30 * syntax, such as "[[link]]" for making links), and provides a one-way
31 * transformation of that wiki markup it into (X)HTML output / markup
32 * (which in turn the browser understands, and can display).
33 *
34 * There are seven main entry points into the Parser class:
35 *
36 * - Parser::parse()
37 * produces HTML output
38 * - Parser::preSaveTransform()
39 * produces altered wiki markup
40 * - Parser::preprocess()
41 * removes HTML comments and expands templates
42 * - Parser::cleanSig() and Parser::cleanSigInSig()
43 * cleans a signature before saving it to preferences
44 * - Parser::getSection()
45 * return the content of a section from an article for section editing
46 * - Parser::replaceSection()
47 * replaces a section by number inside an article
48 * - Parser::getPreloadText()
49 * removes <noinclude> sections and <includeonly> tags
50 *
51 * Globals used:
52 * object: $wgContLang
53 *
54 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
55 *
56 * @par Settings:
57 * $wgNamespacesWithSubpages
58 *
59 * @par Settings only within ParserOptions:
60 * $wgAllowExternalImages
61 * $wgAllowSpecialInclusion
62 * $wgInterwikiMagic
63 * $wgMaxArticleSize
64 *
65 * @ingroup Parser
66 */
67 class Parser {
68 /**
69 * Update this version number when the ParserOutput format
70 * changes in an incompatible way, so the parser cache
71 * can automatically discard old data.
72 */
73 const VERSION = '1.6.4';
74
75 /**
76 * Update this version number when the output of serialiseHalfParsedText()
77 * changes in an incompatible way
78 */
79 const HALF_PARSED_VERSION = 2;
80
81 # Flags for Parser::setFunctionHook
82 const SFH_NO_HASH = 1;
83 const SFH_OBJECT_ARGS = 2;
84
85 # Constants needed for external link processing
86 # Everything except bracket, space, or control characters
87 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
88 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
89 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
90 # Simplified expression to match an IPv4 or IPv6 address, or
91 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
92 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
93 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
94 // @codingStandardsIgnoreStart Generic.Files.LineLength
95 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
96 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
97 // @codingStandardsIgnoreEnd
98
99 # Regular expression for a non-newline space
100 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
101
102 # Flags for preprocessToDom
103 const PTD_FOR_INCLUSION = 1;
104
105 # Allowed values for $this->mOutputType
106 # Parameter to startExternalParse().
107 const OT_HTML = 1; # like parse()
108 const OT_WIKI = 2; # like preSaveTransform()
109 const OT_PREPROCESS = 3; # like preprocess()
110 const OT_MSG = 3;
111 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
112
113 /**
114 * @var string Prefix and suffix for temporary replacement strings
115 * for the multipass parser.
116 *
117 * \x7f should never appear in input as it's disallowed in XML.
118 * Using it at the front also gives us a little extra robustness
119 * since it shouldn't match when butted up against identifier-like
120 * string constructs.
121 *
122 * Must not consist of all title characters, or else it will change
123 * the behavior of <nowiki> in a link.
124 */
125 const MARKER_SUFFIX = "-QINU\x7f";
126 const MARKER_PREFIX = "\x7fUNIQ-";
127
128 # Markers used for wrapping the table of contents
129 const TOC_START = '<mw:toc>';
130 const TOC_END = '</mw:toc>';
131
132 # Persistent:
133 public $mTagHooks = [];
134 public $mTransparentTagHooks = [];
135 public $mFunctionHooks = [];
136 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
137 public $mFunctionTagHooks = [];
138 public $mStripList = [];
139 public $mDefaultStripList = [];
140 public $mVarCache = [];
141 public $mImageParams = [];
142 public $mImageParamsMagicArray = [];
143 public $mMarkerIndex = 0;
144 public $mFirstCall = true;
145
146 # Initialised by initialiseVariables()
147
148 /**
149 * @var MagicWordArray
150 */
151 public $mVariables;
152
153 /**
154 * @var MagicWordArray
155 */
156 public $mSubstWords;
157 # Initialised in constructor
158 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
159
160 # Initialized in getPreprocessor()
161 /** @var Preprocessor */
162 public $mPreprocessor;
163
164 # Cleared with clearState():
165 /**
166 * @var ParserOutput
167 */
168 public $mOutput;
169 public $mAutonumber;
170
171 /**
172 * @var StripState
173 */
174 public $mStripState;
175
176 public $mIncludeCount;
177 /**
178 * @var LinkHolderArray
179 */
180 public $mLinkHolders;
181
182 public $mLinkID;
183 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
184 public $mDefaultSort;
185 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
186 public $mExpensiveFunctionCount; # number of expensive parser function calls
187 public $mShowToc, $mForceTocPosition;
188
189 /**
190 * @var User
191 */
192 public $mUser; # User object; only used when doing pre-save transform
193
194 # Temporary
195 # These are variables reset at least once per parse regardless of $clearState
196
197 /**
198 * @var ParserOptions
199 */
200 public $mOptions;
201
202 /**
203 * @var Title
204 */
205 public $mTitle; # Title context, used for self-link rendering and similar things
206 public $mOutputType; # Output type, one of the OT_xxx constants
207 public $ot; # Shortcut alias, see setOutputType()
208 public $mRevisionObject; # The revision object of the specified revision ID
209 public $mRevisionId; # ID to display in {{REVISIONID}} tags
210 public $mRevisionTimestamp; # The timestamp of the specified revision ID
211 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
212 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
213 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
214 public $mInputSize = false; # For {{PAGESIZE}} on current page.
215
216 /**
217 * @var string Deprecated accessor for the strip marker prefix.
218 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
219 **/
220 public $mUniqPrefix = Parser::MARKER_PREFIX;
221
222 /**
223 * @var array Array with the language name of each language link (i.e. the
224 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
225 * duplicate language links to the ParserOutput.
226 */
227 public $mLangLinkLanguages;
228
229 /**
230 * @var MapCacheLRU|null
231 * @since 1.24
232 *
233 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
234 */
235 public $currentRevisionCache;
236
237 /**
238 * @var bool Recursive call protection.
239 * This variable should be treated as if it were private.
240 */
241 public $mInParse = false;
242
243 /** @var SectionProfiler */
244 protected $mProfiler;
245
246 /**
247 * @param array $conf
248 */
249 public function __construct( $conf = [] ) {
250 $this->mConf = $conf;
251 $this->mUrlProtocols = wfUrlProtocols();
252 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
253 self::EXT_LINK_ADDR .
254 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
255 if ( isset( $conf['preprocessorClass'] ) ) {
256 $this->mPreprocessorClass = $conf['preprocessorClass'];
257 } elseif ( defined( 'HPHP_VERSION' ) ) {
258 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
259 $this->mPreprocessorClass = 'Preprocessor_Hash';
260 } elseif ( extension_loaded( 'domxml' ) ) {
261 # PECL extension that conflicts with the core DOM extension (bug 13770)
262 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
263 $this->mPreprocessorClass = 'Preprocessor_Hash';
264 } elseif ( extension_loaded( 'dom' ) ) {
265 $this->mPreprocessorClass = 'Preprocessor_DOM';
266 } else {
267 $this->mPreprocessorClass = 'Preprocessor_Hash';
268 }
269 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
270 }
271
272 /**
273 * Reduce memory usage to reduce the impact of circular references
274 */
275 public function __destruct() {
276 if ( isset( $this->mLinkHolders ) ) {
277 unset( $this->mLinkHolders );
278 }
279 foreach ( $this as $name => $value ) {
280 unset( $this->$name );
281 }
282 }
283
284 /**
285 * Allow extensions to clean up when the parser is cloned
286 */
287 public function __clone() {
288 $this->mInParse = false;
289
290 // Bug 56226: When you create a reference "to" an object field, that
291 // makes the object field itself be a reference too (until the other
292 // reference goes out of scope). When cloning, any field that's a
293 // reference is copied as a reference in the new object. Both of these
294 // are defined PHP5 behaviors, as inconvenient as it is for us when old
295 // hooks from PHP4 days are passing fields by reference.
296 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
297 // Make a non-reference copy of the field, then rebind the field to
298 // reference the new copy.
299 $tmp = $this->$k;
300 $this->$k =& $tmp;
301 unset( $tmp );
302 }
303
304 Hooks::run( 'ParserCloned', [ $this ] );
305 }
306
307 /**
308 * Do various kinds of initialisation on the first call of the parser
309 */
310 public function firstCallInit() {
311 if ( !$this->mFirstCall ) {
312 return;
313 }
314 $this->mFirstCall = false;
315
316 CoreParserFunctions::register( $this );
317 CoreTagHooks::register( $this );
318 $this->initialiseVariables();
319
320 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
321 }
322
323 /**
324 * Clear Parser state
325 *
326 * @private
327 */
328 public function clearState() {
329 if ( $this->mFirstCall ) {
330 $this->firstCallInit();
331 }
332 $this->mOutput = new ParserOutput;
333 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
334 $this->mAutonumber = 0;
335 $this->mIncludeCount = [];
336 $this->mLinkHolders = new LinkHolderArray( $this );
337 $this->mLinkID = 0;
338 $this->mRevisionObject = $this->mRevisionTimestamp =
339 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
340 $this->mVarCache = [];
341 $this->mUser = null;
342 $this->mLangLinkLanguages = [];
343 $this->currentRevisionCache = null;
344
345 $this->mStripState = new StripState;
346
347 # Clear these on every parse, bug 4549
348 $this->mTplRedirCache = $this->mTplDomCache = [];
349
350 $this->mShowToc = true;
351 $this->mForceTocPosition = false;
352 $this->mIncludeSizes = [
353 'post-expand' => 0,
354 'arg' => 0,
355 ];
356 $this->mPPNodeCount = 0;
357 $this->mGeneratedPPNodeCount = 0;
358 $this->mHighestExpansionDepth = 0;
359 $this->mDefaultSort = false;
360 $this->mHeadings = [];
361 $this->mDoubleUnderscores = [];
362 $this->mExpensiveFunctionCount = 0;
363
364 # Fix cloning
365 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
366 $this->mPreprocessor = null;
367 }
368
369 $this->mProfiler = new SectionProfiler();
370
371 Hooks::run( 'ParserClearState', [ &$this ] );
372 }
373
374 /**
375 * Convert wikitext to HTML
376 * Do not call this function recursively.
377 *
378 * @param string $text Text we want to parse
379 * @param Title $title
380 * @param ParserOptions $options
381 * @param bool $linestart
382 * @param bool $clearState
383 * @param int $revid Number to pass in {{REVISIONID}}
384 * @return ParserOutput A ParserOutput
385 */
386 public function parse( $text, Title $title, ParserOptions $options,
387 $linestart = true, $clearState = true, $revid = null
388 ) {
389 /**
390 * First pass--just handle <nowiki> sections, pass the rest off
391 * to internalParse() which does all the real work.
392 */
393
394 global $wgShowHostnames;
395
396 if ( $clearState ) {
397 // We use U+007F DELETE to construct strip markers, so we have to make
398 // sure that this character does not occur in the input text.
399 $text = strtr( $text, "\x7f", "?" );
400 $magicScopeVariable = $this->lock();
401 }
402
403 $this->startParse( $title, $options, self::OT_HTML, $clearState );
404
405 $this->currentRevisionCache = null;
406 $this->mInputSize = strlen( $text );
407 if ( $this->mOptions->getEnableLimitReport() ) {
408 $this->mOutput->resetParseStartTime();
409 }
410
411 $oldRevisionId = $this->mRevisionId;
412 $oldRevisionObject = $this->mRevisionObject;
413 $oldRevisionTimestamp = $this->mRevisionTimestamp;
414 $oldRevisionUser = $this->mRevisionUser;
415 $oldRevisionSize = $this->mRevisionSize;
416 if ( $revid !== null ) {
417 $this->mRevisionId = $revid;
418 $this->mRevisionObject = null;
419 $this->mRevisionTimestamp = null;
420 $this->mRevisionUser = null;
421 $this->mRevisionSize = null;
422 }
423
424 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
425 # No more strip!
426 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
427 $text = $this->internalParse( $text );
428 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
429
430 $text = $this->internalParseHalfParsed( $text, true, $linestart );
431
432 /**
433 * A converted title will be provided in the output object if title and
434 * content conversion are enabled, the article text does not contain
435 * a conversion-suppressing double-underscore tag, and no
436 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
437 * automatic link conversion.
438 */
439 if ( !( $options->getDisableTitleConversion()
440 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
441 || isset( $this->mDoubleUnderscores['notitleconvert'] )
442 || $this->mOutput->getDisplayTitle() !== false )
443 ) {
444 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
445 if ( $convruletitle ) {
446 $this->mOutput->setTitleText( $convruletitle );
447 } else {
448 $titleText = $this->getConverterLanguage()->convertTitle( $title );
449 $this->mOutput->setTitleText( $titleText );
450 }
451 }
452
453 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
454 $this->limitationWarn( 'expensive-parserfunction',
455 $this->mExpensiveFunctionCount,
456 $this->mOptions->getExpensiveParserFunctionLimit()
457 );
458 }
459
460 # Information on include size limits, for the benefit of users who try to skirt them
461 if ( $this->mOptions->getEnableLimitReport() ) {
462 $max = $this->mOptions->getMaxIncludeSize();
463
464 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
465 if ( $cpuTime !== null ) {
466 $this->mOutput->setLimitReportData( 'limitreport-cputime',
467 sprintf( "%.3f", $cpuTime )
468 );
469 }
470
471 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
472 $this->mOutput->setLimitReportData( 'limitreport-walltime',
473 sprintf( "%.3f", $wallTime )
474 );
475
476 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
477 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
478 );
479 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
480 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
481 );
482 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
483 [ $this->mIncludeSizes['post-expand'], $max ]
484 );
485 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
486 [ $this->mIncludeSizes['arg'], $max ]
487 );
488 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
489 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
490 );
491 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
492 [ $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() ]
493 );
494 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
495
496 $limitReport = "NewPP limit report\n";
497 if ( $wgShowHostnames ) {
498 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
499 }
500 $limitReport .= 'Cached time: ' . $this->mOutput->getCacheTime() . "\n";
501 $limitReport .= 'Cache expiry: ' . $this->mOutput->getCacheExpiry() . "\n";
502 $limitReport .= 'Dynamic content: ' .
503 ( $this->mOutput->hasDynamicContent() ? 'true' : 'false' ) .
504 "\n";
505
506 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
507 if ( Hooks::run( 'ParserLimitReportFormat',
508 [ $key, &$value, &$limitReport, false, false ]
509 ) ) {
510 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
511 $valueMsg = wfMessage( [ "$key-value-text", "$key-value" ] )
512 ->inLanguage( 'en' )->useDatabase( false );
513 if ( !$valueMsg->exists() ) {
514 $valueMsg = new RawMessage( '$1' );
515 }
516 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
517 $valueMsg->params( $value );
518 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
519 }
520 }
521 }
522 // Since we're not really outputting HTML, decode the entities and
523 // then re-encode the things that need hiding inside HTML comments.
524 $limitReport = htmlspecialchars_decode( $limitReport );
525 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
526
527 // Sanitize for comment. Note '‐' in the replacement is U+2010,
528 // which looks much like the problematic '-'.
529 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
530 $text .= "\n<!-- \n$limitReport-->\n";
531
532 // Add on template profiling data
533 $dataByFunc = $this->mProfiler->getFunctionStats();
534 uasort( $dataByFunc, function ( $a, $b ) {
535 return $a['real'] < $b['real']; // descending order
536 } );
537 $profileReport = "Transclusion expansion time report (%,ms,calls,template)\n";
538 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
539 $profileReport .= sprintf( "%6.2f%% %8.3f %6d - %s\n",
540 $item['%real'], $item['real'], $item['calls'],
541 htmlspecialchars( $item['name'] ) );
542 }
543 $text .= "\n<!-- \n$profileReport-->\n";
544
545 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
546 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
547 $this->mTitle->getPrefixedDBkey() );
548 }
549 }
550 $this->mOutput->setText( $text );
551
552 $this->mRevisionId = $oldRevisionId;
553 $this->mRevisionObject = $oldRevisionObject;
554 $this->mRevisionTimestamp = $oldRevisionTimestamp;
555 $this->mRevisionUser = $oldRevisionUser;
556 $this->mRevisionSize = $oldRevisionSize;
557 $this->mInputSize = false;
558 $this->currentRevisionCache = null;
559
560 return $this->mOutput;
561 }
562
563 /**
564 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
565 * can be called from an extension tag hook.
566 *
567 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
568 * instead, which means that lists and links have not been fully parsed yet,
569 * and strip markers are still present.
570 *
571 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
572 *
573 * Use this function if you're a parser tag hook and you want to parse
574 * wikitext before or after applying additional transformations, and you
575 * intend to *return the result as hook output*, which will cause it to go
576 * through the rest of parsing process automatically.
577 *
578 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
579 * $text are not expanded
580 *
581 * @param string $text Text extension wants to have parsed
582 * @param bool|PPFrame $frame The frame to use for expanding any template variables
583 * @return string UNSAFE half-parsed HTML
584 */
585 public function recursiveTagParse( $text, $frame = false ) {
586 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
587 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
588 $text = $this->internalParse( $text, false, $frame );
589 return $text;
590 }
591
592 /**
593 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
594 * point can be called from an extension tag hook.
595 *
596 * The output of this function is fully-parsed HTML that is safe for output.
597 * If you're a parser tag hook, you might want to use recursiveTagParse()
598 * instead.
599 *
600 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
601 * $text are not expanded
602 *
603 * @since 1.25
604 *
605 * @param string $text Text extension wants to have parsed
606 * @param bool|PPFrame $frame The frame to use for expanding any template variables
607 * @return string Fully parsed HTML
608 */
609 public function recursiveTagParseFully( $text, $frame = false ) {
610 $text = $this->recursiveTagParse( $text, $frame );
611 $text = $this->internalParseHalfParsed( $text, false );
612 return $text;
613 }
614
615 /**
616 * Expand templates and variables in the text, producing valid, static wikitext.
617 * Also removes comments.
618 * Do not call this function recursively.
619 * @param string $text
620 * @param Title $title
621 * @param ParserOptions $options
622 * @param int|null $revid
623 * @param bool|PPFrame $frame
624 * @return mixed|string
625 */
626 public function preprocess( $text, Title $title = null,
627 ParserOptions $options, $revid = null, $frame = false
628 ) {
629 $magicScopeVariable = $this->lock();
630 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
631 if ( $revid !== null ) {
632 $this->mRevisionId = $revid;
633 }
634 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
635 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
636 $text = $this->replaceVariables( $text, $frame );
637 $text = $this->mStripState->unstripBoth( $text );
638 return $text;
639 }
640
641 /**
642 * Recursive parser entry point that can be called from an extension tag
643 * hook.
644 *
645 * @param string $text Text to be expanded
646 * @param bool|PPFrame $frame The frame to use for expanding any template variables
647 * @return string
648 * @since 1.19
649 */
650 public function recursivePreprocess( $text, $frame = false ) {
651 $text = $this->replaceVariables( $text, $frame );
652 $text = $this->mStripState->unstripBoth( $text );
653 return $text;
654 }
655
656 /**
657 * Process the wikitext for the "?preload=" feature. (bug 5210)
658 *
659 * "<noinclude>", "<includeonly>" etc. are parsed as for template
660 * transclusion, comments, templates, arguments, tags hooks and parser
661 * functions are untouched.
662 *
663 * @param string $text
664 * @param Title $title
665 * @param ParserOptions $options
666 * @param array $params
667 * @return string
668 */
669 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
670 $msg = new RawMessage( $text );
671 $text = $msg->params( $params )->plain();
672
673 # Parser (re)initialisation
674 $magicScopeVariable = $this->lock();
675 $this->startParse( $title, $options, self::OT_PLAIN, true );
676
677 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
678 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
679 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
680 $text = $this->mStripState->unstripBoth( $text );
681 return $text;
682 }
683
684 /**
685 * Get a random string
686 *
687 * @return string
688 * @deprecated since 1.26; use wfRandomString() instead.
689 */
690 public static function getRandomString() {
691 wfDeprecated( __METHOD__, '1.26' );
692 return wfRandomString( 16 );
693 }
694
695 /**
696 * Set the current user.
697 * Should only be used when doing pre-save transform.
698 *
699 * @param User|null $user User object or null (to reset)
700 */
701 public function setUser( $user ) {
702 $this->mUser = $user;
703 }
704
705 /**
706 * Accessor for mUniqPrefix.
707 *
708 * @return string
709 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
710 */
711 public function uniqPrefix() {
712 wfDeprecated( __METHOD__, '1.26' );
713 return self::MARKER_PREFIX;
714 }
715
716 /**
717 * Set the context title
718 *
719 * @param Title $t
720 */
721 public function setTitle( $t ) {
722 if ( !$t ) {
723 $t = Title::newFromText( 'NO TITLE' );
724 }
725
726 if ( $t->hasFragment() ) {
727 # Strip the fragment to avoid various odd effects
728 $this->mTitle = $t->createFragmentTarget( '' );
729 } else {
730 $this->mTitle = $t;
731 }
732 }
733
734 /**
735 * Accessor for the Title object
736 *
737 * @return Title
738 */
739 public function getTitle() {
740 return $this->mTitle;
741 }
742
743 /**
744 * Accessor/mutator for the Title object
745 *
746 * @param Title $x Title object or null to just get the current one
747 * @return Title
748 */
749 public function Title( $x = null ) {
750 return wfSetVar( $this->mTitle, $x );
751 }
752
753 /**
754 * Set the output type
755 *
756 * @param int $ot New value
757 */
758 public function setOutputType( $ot ) {
759 $this->mOutputType = $ot;
760 # Shortcut alias
761 $this->ot = [
762 'html' => $ot == self::OT_HTML,
763 'wiki' => $ot == self::OT_WIKI,
764 'pre' => $ot == self::OT_PREPROCESS,
765 'plain' => $ot == self::OT_PLAIN,
766 ];
767 }
768
769 /**
770 * Accessor/mutator for the output type
771 *
772 * @param int|null $x New value or null to just get the current one
773 * @return int
774 */
775 public function OutputType( $x = null ) {
776 return wfSetVar( $this->mOutputType, $x );
777 }
778
779 /**
780 * Get the ParserOutput object
781 *
782 * @return ParserOutput
783 */
784 public function getOutput() {
785 return $this->mOutput;
786 }
787
788 /**
789 * Get the ParserOptions object
790 *
791 * @return ParserOptions
792 */
793 public function getOptions() {
794 return $this->mOptions;
795 }
796
797 /**
798 * Accessor/mutator for the ParserOptions object
799 *
800 * @param ParserOptions $x New value or null to just get the current one
801 * @return ParserOptions Current ParserOptions object
802 */
803 public function Options( $x = null ) {
804 return wfSetVar( $this->mOptions, $x );
805 }
806
807 /**
808 * @return int
809 */
810 public function nextLinkID() {
811 return $this->mLinkID++;
812 }
813
814 /**
815 * @param int $id
816 */
817 public function setLinkID( $id ) {
818 $this->mLinkID = $id;
819 }
820
821 /**
822 * Get a language object for use in parser functions such as {{FORMATNUM:}}
823 * @return Language
824 */
825 public function getFunctionLang() {
826 return $this->getTargetLanguage();
827 }
828
829 /**
830 * Get the target language for the content being parsed. This is usually the
831 * language that the content is in.
832 *
833 * @since 1.19
834 *
835 * @throws MWException
836 * @return Language
837 */
838 public function getTargetLanguage() {
839 $target = $this->mOptions->getTargetLanguage();
840
841 if ( $target !== null ) {
842 return $target;
843 } elseif ( $this->mOptions->getInterfaceMessage() ) {
844 return $this->mOptions->getUserLangObj();
845 } elseif ( is_null( $this->mTitle ) ) {
846 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
847 }
848
849 return $this->mTitle->getPageLanguage();
850 }
851
852 /**
853 * Get the language object for language conversion
854 * @return Language|null
855 */
856 public function getConverterLanguage() {
857 return $this->getTargetLanguage();
858 }
859
860 /**
861 * Get a User object either from $this->mUser, if set, or from the
862 * ParserOptions object otherwise
863 *
864 * @return User
865 */
866 public function getUser() {
867 if ( !is_null( $this->mUser ) ) {
868 return $this->mUser;
869 }
870 return $this->mOptions->getUser();
871 }
872
873 /**
874 * Get a preprocessor object
875 *
876 * @return Preprocessor
877 */
878 public function getPreprocessor() {
879 if ( !isset( $this->mPreprocessor ) ) {
880 $class = $this->mPreprocessorClass;
881 $this->mPreprocessor = new $class( $this );
882 }
883 return $this->mPreprocessor;
884 }
885
886 /**
887 * Replaces all occurrences of HTML-style comments and the given tags
888 * in the text with a random marker and returns the next text. The output
889 * parameter $matches will be an associative array filled with data in
890 * the form:
891 *
892 * @code
893 * 'UNIQ-xxxxx' => array(
894 * 'element',
895 * 'tag content',
896 * array( 'param' => 'x' ),
897 * '<element param="x">tag content</element>' ) )
898 * @endcode
899 *
900 * @param array $elements List of element names. Comments are always extracted.
901 * @param string $text Source text string.
902 * @param array $matches Out parameter, Array: extracted tags
903 * @param string|null $uniq_prefix
904 * @return string Stripped text
905 * @since 1.26 The uniq_prefix argument is deprecated.
906 */
907 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
908 if ( $uniq_prefix !== null ) {
909 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
910 }
911 static $n = 1;
912 $stripped = '';
913 $matches = [];
914
915 $taglist = implode( '|', $elements );
916 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
917
918 while ( $text != '' ) {
919 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
920 $stripped .= $p[0];
921 if ( count( $p ) < 5 ) {
922 break;
923 }
924 if ( count( $p ) > 5 ) {
925 # comment
926 $element = $p[4];
927 $attributes = '';
928 $close = '';
929 $inside = $p[5];
930 } else {
931 # tag
932 $element = $p[1];
933 $attributes = $p[2];
934 $close = $p[3];
935 $inside = $p[4];
936 }
937
938 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
939 $stripped .= $marker;
940
941 if ( $close === '/>' ) {
942 # Empty element tag, <tag />
943 $content = null;
944 $text = $inside;
945 $tail = null;
946 } else {
947 if ( $element === '!--' ) {
948 $end = '/(-->)/';
949 } else {
950 $end = "/(<\\/$element\\s*>)/i";
951 }
952 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
953 $content = $q[0];
954 if ( count( $q ) < 3 ) {
955 # No end tag -- let it run out to the end of the text.
956 $tail = '';
957 $text = '';
958 } else {
959 $tail = $q[1];
960 $text = $q[2];
961 }
962 }
963
964 $matches[$marker] = [ $element,
965 $content,
966 Sanitizer::decodeTagAttributes( $attributes ),
967 "<$element$attributes$close$content$tail" ];
968 }
969 return $stripped;
970 }
971
972 /**
973 * Get a list of strippable XML-like elements
974 *
975 * @return array
976 */
977 public function getStripList() {
978 return $this->mStripList;
979 }
980
981 /**
982 * Add an item to the strip state
983 * Returns the unique tag which must be inserted into the stripped text
984 * The tag will be replaced with the original text in unstrip()
985 *
986 * @param string $text
987 *
988 * @return string
989 */
990 public function insertStripItem( $text ) {
991 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
992 $this->mMarkerIndex++;
993 $this->mStripState->addGeneral( $marker, $text );
994 return $marker;
995 }
996
997 /**
998 * parse the wiki syntax used to render tables
999 *
1000 * @private
1001 * @param string $text
1002 * @return string
1003 */
1004 public function doTableStuff( $text ) {
1005
1006 $lines = StringUtils::explode( "\n", $text );
1007 $out = '';
1008 $td_history = []; # Is currently a td tag open?
1009 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1010 $tr_history = []; # Is currently a tr tag open?
1011 $tr_attributes = []; # history of tr attributes
1012 $has_opened_tr = []; # Did this table open a <tr> element?
1013 $indent_level = 0; # indent level of the table
1014
1015 foreach ( $lines as $outLine ) {
1016 $line = trim( $outLine );
1017
1018 if ( $line === '' ) { # empty line, go to next line
1019 $out .= $outLine . "\n";
1020 continue;
1021 }
1022
1023 $first_character = $line[0];
1024 $first_two = substr( $line, 0, 2 );
1025 $matches = [];
1026
1027 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1028 # First check if we are starting a new table
1029 $indent_level = strlen( $matches[1] );
1030
1031 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1032 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1033
1034 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1035 array_push( $td_history, false );
1036 array_push( $last_tag_history, '' );
1037 array_push( $tr_history, false );
1038 array_push( $tr_attributes, '' );
1039 array_push( $has_opened_tr, false );
1040 } elseif ( count( $td_history ) == 0 ) {
1041 # Don't do any of the following
1042 $out .= $outLine . "\n";
1043 continue;
1044 } elseif ( $first_two === '|}' ) {
1045 # We are ending a table
1046 $line = '</table>' . substr( $line, 2 );
1047 $last_tag = array_pop( $last_tag_history );
1048
1049 if ( !array_pop( $has_opened_tr ) ) {
1050 $line = "<tr><td></td></tr>{$line}";
1051 }
1052
1053 if ( array_pop( $tr_history ) ) {
1054 $line = "</tr>{$line}";
1055 }
1056
1057 if ( array_pop( $td_history ) ) {
1058 $line = "</{$last_tag}>{$line}";
1059 }
1060 array_pop( $tr_attributes );
1061 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1062 } elseif ( $first_two === '|-' ) {
1063 # Now we have a table row
1064 $line = preg_replace( '#^\|-+#', '', $line );
1065
1066 # Whats after the tag is now only attributes
1067 $attributes = $this->mStripState->unstripBoth( $line );
1068 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1069 array_pop( $tr_attributes );
1070 array_push( $tr_attributes, $attributes );
1071
1072 $line = '';
1073 $last_tag = array_pop( $last_tag_history );
1074 array_pop( $has_opened_tr );
1075 array_push( $has_opened_tr, true );
1076
1077 if ( array_pop( $tr_history ) ) {
1078 $line = '</tr>';
1079 }
1080
1081 if ( array_pop( $td_history ) ) {
1082 $line = "</{$last_tag}>{$line}";
1083 }
1084
1085 $outLine = $line;
1086 array_push( $tr_history, false );
1087 array_push( $td_history, false );
1088 array_push( $last_tag_history, '' );
1089 } elseif ( $first_character === '|'
1090 || $first_character === '!'
1091 || $first_two === '|+'
1092 ) {
1093 # This might be cell elements, td, th or captions
1094 if ( $first_two === '|+' ) {
1095 $first_character = '+';
1096 $line = substr( $line, 2 );
1097 } else {
1098 $line = substr( $line, 1 );
1099 }
1100
1101 // Implies both are valid for table headings.
1102 if ( $first_character === '!' ) {
1103 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1104 }
1105
1106 # Split up multiple cells on the same line.
1107 # FIXME : This can result in improper nesting of tags processed
1108 # by earlier parser steps.
1109 $cells = explode( '||', $line );
1110
1111 $outLine = '';
1112
1113 # Loop through each table cell
1114 foreach ( $cells as $cell ) {
1115 $previous = '';
1116 if ( $first_character !== '+' ) {
1117 $tr_after = array_pop( $tr_attributes );
1118 if ( !array_pop( $tr_history ) ) {
1119 $previous = "<tr{$tr_after}>\n";
1120 }
1121 array_push( $tr_history, true );
1122 array_push( $tr_attributes, '' );
1123 array_pop( $has_opened_tr );
1124 array_push( $has_opened_tr, true );
1125 }
1126
1127 $last_tag = array_pop( $last_tag_history );
1128
1129 if ( array_pop( $td_history ) ) {
1130 $previous = "</{$last_tag}>\n{$previous}";
1131 }
1132
1133 if ( $first_character === '|' ) {
1134 $last_tag = 'td';
1135 } elseif ( $first_character === '!' ) {
1136 $last_tag = 'th';
1137 } elseif ( $first_character === '+' ) {
1138 $last_tag = 'caption';
1139 } else {
1140 $last_tag = '';
1141 }
1142
1143 array_push( $last_tag_history, $last_tag );
1144
1145 # A cell could contain both parameters and data
1146 $cell_data = explode( '|', $cell, 2 );
1147
1148 # Bug 553: Note that a '|' inside an invalid link should not
1149 # be mistaken as delimiting cell parameters
1150 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1151 $cell = "{$previous}<{$last_tag}>{$cell}";
1152 } elseif ( count( $cell_data ) == 1 ) {
1153 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1154 } else {
1155 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1156 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1157 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1158 }
1159
1160 $outLine .= $cell;
1161 array_push( $td_history, true );
1162 }
1163 }
1164 $out .= $outLine . "\n";
1165 }
1166
1167 # Closing open td, tr && table
1168 while ( count( $td_history ) > 0 ) {
1169 if ( array_pop( $td_history ) ) {
1170 $out .= "</td>\n";
1171 }
1172 if ( array_pop( $tr_history ) ) {
1173 $out .= "</tr>\n";
1174 }
1175 if ( !array_pop( $has_opened_tr ) ) {
1176 $out .= "<tr><td></td></tr>\n";
1177 }
1178
1179 $out .= "</table>\n";
1180 }
1181
1182 # Remove trailing line-ending (b/c)
1183 if ( substr( $out, -1 ) === "\n" ) {
1184 $out = substr( $out, 0, -1 );
1185 }
1186
1187 # special case: don't return empty table
1188 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1189 $out = '';
1190 }
1191
1192 return $out;
1193 }
1194
1195 /**
1196 * Helper function for parse() that transforms wiki markup into half-parsed
1197 * HTML. Only called for $mOutputType == self::OT_HTML.
1198 *
1199 * @private
1200 *
1201 * @param string $text The text to parse
1202 * @param bool $isMain Whether this is being called from the main parse() function
1203 * @param PPFrame|bool $frame A pre-processor frame
1204 *
1205 * @return string
1206 */
1207 public function internalParse( $text, $isMain = true, $frame = false ) {
1208
1209 $origText = $text;
1210
1211 # Hook to suspend the parser in this state
1212 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1213 return $text;
1214 }
1215
1216 # if $frame is provided, then use $frame for replacing any variables
1217 if ( $frame ) {
1218 # use frame depth to infer how include/noinclude tags should be handled
1219 # depth=0 means this is the top-level document; otherwise it's an included document
1220 if ( !$frame->depth ) {
1221 $flag = 0;
1222 } else {
1223 $flag = Parser::PTD_FOR_INCLUSION;
1224 }
1225 $dom = $this->preprocessToDom( $text, $flag );
1226 $text = $frame->expand( $dom );
1227 } else {
1228 # if $frame is not provided, then use old-style replaceVariables
1229 $text = $this->replaceVariables( $text );
1230 }
1231
1232 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1233 $text = Sanitizer::removeHTMLtags(
1234 $text,
1235 [ &$this, 'attributeStripCallback' ],
1236 false,
1237 array_keys( $this->mTransparentTagHooks )
1238 );
1239 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1240
1241 # Tables need to come after variable replacement for things to work
1242 # properly; putting them before other transformations should keep
1243 # exciting things like link expansions from showing up in surprising
1244 # places.
1245 $text = $this->doTableStuff( $text );
1246
1247 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1248
1249 $text = $this->doDoubleUnderscore( $text );
1250
1251 $text = $this->doHeadings( $text );
1252 $text = $this->replaceInternalLinks( $text );
1253 $text = $this->doAllQuotes( $text );
1254 $text = $this->replaceExternalLinks( $text );
1255
1256 # replaceInternalLinks may sometimes leave behind
1257 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1258 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1259
1260 $text = $this->doMagicLinks( $text );
1261 $text = $this->formatHeadings( $text, $origText, $isMain );
1262
1263 return $text;
1264 }
1265
1266 /**
1267 * Helper function for parse() that transforms half-parsed HTML into fully
1268 * parsed HTML.
1269 *
1270 * @param string $text
1271 * @param bool $isMain
1272 * @param bool $linestart
1273 * @return string
1274 */
1275 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1276 $text = $this->mStripState->unstripGeneral( $text );
1277
1278 if ( $isMain ) {
1279 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1280 }
1281
1282 # Clean up special characters, only run once, next-to-last before doBlockLevels
1283 $fixtags = [
1284 # french spaces, last one Guillemet-left
1285 # only if there is something before the space
1286 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1287 # french spaces, Guillemet-right
1288 '/(\\302\\253) /' => '\\1&#160;',
1289 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1290 ];
1291 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1292
1293 $text = $this->doBlockLevels( $text, $linestart );
1294
1295 $this->replaceLinkHolders( $text );
1296
1297 /**
1298 * The input doesn't get language converted if
1299 * a) It's disabled
1300 * b) Content isn't converted
1301 * c) It's a conversion table
1302 * d) it is an interface message (which is in the user language)
1303 */
1304 if ( !( $this->mOptions->getDisableContentConversion()
1305 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1306 ) {
1307 if ( !$this->mOptions->getInterfaceMessage() ) {
1308 # The position of the convert() call should not be changed. it
1309 # assumes that the links are all replaced and the only thing left
1310 # is the <nowiki> mark.
1311 $text = $this->getConverterLanguage()->convert( $text );
1312 }
1313 }
1314
1315 $text = $this->mStripState->unstripNoWiki( $text );
1316
1317 if ( $isMain ) {
1318 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1319 }
1320
1321 $text = $this->replaceTransparentTags( $text );
1322 $text = $this->mStripState->unstripGeneral( $text );
1323
1324 $text = Sanitizer::normalizeCharReferences( $text );
1325
1326 if ( MWTidy::isEnabled() && $this->mOptions->getTidy() ) {
1327 $text = MWTidy::tidy( $text );
1328 $this->mOutput->addModuleStyles( MWTidy::getModuleStyles() );
1329 } else {
1330 # attempt to sanitize at least some nesting problems
1331 # (bug #2702 and quite a few others)
1332 $tidyregs = [
1333 # ''Something [http://www.cool.com cool''] -->
1334 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1335 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1336 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1337 # fix up an anchor inside another anchor, only
1338 # at least for a single single nested link (bug 3695)
1339 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1340 '\\1\\2</a>\\3</a>\\1\\4</a>',
1341 # fix div inside inline elements- doBlockLevels won't wrap a line which
1342 # contains a div, so fix it up here; replace
1343 # div with escaped text
1344 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1345 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1346 # remove empty italic or bold tag pairs, some
1347 # introduced by rules above
1348 '/<([bi])><\/\\1>/' => '',
1349 ];
1350
1351 $text = preg_replace(
1352 array_keys( $tidyregs ),
1353 array_values( $tidyregs ),
1354 $text );
1355 }
1356
1357 if ( $isMain ) {
1358 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1359 }
1360
1361 return $text;
1362 }
1363
1364 /**
1365 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1366 * magic external links.
1367 *
1368 * DML
1369 * @private
1370 *
1371 * @param string $text
1372 *
1373 * @return string
1374 */
1375 public function doMagicLinks( $text ) {
1376 $prots = wfUrlProtocolsWithoutProtRel();
1377 $urlChar = self::EXT_LINK_URL_CLASS;
1378 $addr = self::EXT_LINK_ADDR;
1379 $space = self::SPACE_NOT_NL; # non-newline space
1380 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1381 $spaces = "$space++"; # possessive match of 1 or more spaces
1382 $text = preg_replace_callback(
1383 '!(?: # Start cases
1384 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1385 (<.*?>) | # m[2]: Skip stuff inside
1386 # HTML elements' . "
1387 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1388 # m[4]: Post-protocol path
1389 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1390 ([0-9]+)\b |
1391 \bISBN $spaces ( # m[6]: ISBN, capture number
1392 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1393 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1394 [0-9Xx] # check digit
1395 )\b
1396 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1397 return $text;
1398 }
1399
1400 /**
1401 * @throws MWException
1402 * @param array $m
1403 * @return HTML|string
1404 */
1405 public function magicLinkCallback( $m ) {
1406 if ( isset( $m[1] ) && $m[1] !== '' ) {
1407 # Skip anchor
1408 return $m[0];
1409 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1410 # Skip HTML element
1411 return $m[0];
1412 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1413 # Free external link
1414 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1415 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1416 # RFC or PMID
1417 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1418 $keyword = 'RFC';
1419 $urlmsg = 'rfcurl';
1420 $cssClass = 'mw-magiclink-rfc';
1421 $id = $m[5];
1422 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1423 $keyword = 'PMID';
1424 $urlmsg = 'pubmedurl';
1425 $cssClass = 'mw-magiclink-pmid';
1426 $id = $m[5];
1427 } else {
1428 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1429 substr( $m[0], 0, 20 ) . '"' );
1430 }
1431 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1432 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass );
1433 } elseif ( isset( $m[6] ) && $m[6] !== '' ) {
1434 # ISBN
1435 $isbn = $m[6];
1436 $space = self::SPACE_NOT_NL; # non-newline space
1437 $isbn = preg_replace( "/$space/", ' ', $isbn );
1438 $num = strtr( $isbn, [
1439 '-' => '',
1440 ' ' => '',
1441 'x' => 'X',
1442 ] );
1443 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1444 return '<a href="' .
1445 htmlspecialchars( $titleObj->getLocalURL() ) .
1446 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1447 } else {
1448 return $m[0];
1449 }
1450 }
1451
1452 /**
1453 * Make a free external link, given a user-supplied URL
1454 *
1455 * @param string $url
1456 * @param int $numPostProto
1457 * The number of characters after the protocol.
1458 * @return string HTML
1459 * @private
1460 */
1461 public function makeFreeExternalLink( $url, $numPostProto ) {
1462 $trail = '';
1463
1464 # The characters '<' and '>' (which were escaped by
1465 # removeHTMLtags()) should not be included in
1466 # URLs, per RFC 2396.
1467 # Make &nbsp; terminate a URL as well (bug T84937)
1468 $m2 = [];
1469 if ( preg_match(
1470 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1471 $url,
1472 $m2,
1473 PREG_OFFSET_CAPTURE
1474 ) ) {
1475 $trail = substr( $url, $m2[0][1] ) . $trail;
1476 $url = substr( $url, 0, $m2[0][1] );
1477 }
1478
1479 # Move trailing punctuation to $trail
1480 $sep = ',;\.:!?';
1481 # If there is no left bracket, then consider right brackets fair game too
1482 if ( strpos( $url, '(' ) === false ) {
1483 $sep .= ')';
1484 }
1485
1486 $urlRev = strrev( $url );
1487 $numSepChars = strspn( $urlRev, $sep );
1488 # Don't break a trailing HTML entity by moving the ; into $trail
1489 # This is in hot code, so use substr_compare to avoid having to
1490 # create a new string object for the comparison
1491 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1492 # more optimization: instead of running preg_match with a $
1493 # anchor, which can be slow, do the match on the reversed
1494 # string starting at the desired offset.
1495 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1496 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1497 $numSepChars--;
1498 }
1499 }
1500 if ( $numSepChars ) {
1501 $trail = substr( $url, -$numSepChars ) . $trail;
1502 $url = substr( $url, 0, -$numSepChars );
1503 }
1504
1505 # Verify that we still have a real URL after trail removal, and
1506 # not just lone protocol
1507 if ( strlen( $trail ) >= $numPostProto ) {
1508 return $url . $trail;
1509 }
1510
1511 $url = Sanitizer::cleanUrl( $url );
1512
1513 # Is this an external image?
1514 $text = $this->maybeMakeExternalImage( $url );
1515 if ( $text === false ) {
1516 # Not an image, make a link
1517 $text = Linker::makeExternalLink( $url,
1518 $this->getConverterLanguage()->markNoConversion( $url, true ),
1519 true, 'free',
1520 $this->getExternalLinkAttribs( $url ) );
1521 # Register it in the output object...
1522 # Replace unnecessary URL escape codes with their equivalent characters
1523 $pasteurized = self::normalizeLinkUrl( $url );
1524 $this->mOutput->addExternalLink( $pasteurized );
1525 }
1526 return $text . $trail;
1527 }
1528
1529 /**
1530 * Parse headers and return html
1531 *
1532 * @private
1533 *
1534 * @param string $text
1535 *
1536 * @return string
1537 */
1538 public function doHeadings( $text ) {
1539 for ( $i = 6; $i >= 1; --$i ) {
1540 $h = str_repeat( '=', $i );
1541 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1542 }
1543 return $text;
1544 }
1545
1546 /**
1547 * Replace single quotes with HTML markup
1548 * @private
1549 *
1550 * @param string $text
1551 *
1552 * @return string The altered text
1553 */
1554 public function doAllQuotes( $text ) {
1555 $outtext = '';
1556 $lines = StringUtils::explode( "\n", $text );
1557 foreach ( $lines as $line ) {
1558 $outtext .= $this->doQuotes( $line ) . "\n";
1559 }
1560 $outtext = substr( $outtext, 0, -1 );
1561 return $outtext;
1562 }
1563
1564 /**
1565 * Helper function for doAllQuotes()
1566 *
1567 * @param string $text
1568 *
1569 * @return string
1570 */
1571 public function doQuotes( $text ) {
1572 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1573 $countarr = count( $arr );
1574 if ( $countarr == 1 ) {
1575 return $text;
1576 }
1577
1578 // First, do some preliminary work. This may shift some apostrophes from
1579 // being mark-up to being text. It also counts the number of occurrences
1580 // of bold and italics mark-ups.
1581 $numbold = 0;
1582 $numitalics = 0;
1583 for ( $i = 1; $i < $countarr; $i += 2 ) {
1584 $thislen = strlen( $arr[$i] );
1585 // If there are ever four apostrophes, assume the first is supposed to
1586 // be text, and the remaining three constitute mark-up for bold text.
1587 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1588 if ( $thislen == 4 ) {
1589 $arr[$i - 1] .= "'";
1590 $arr[$i] = "'''";
1591 $thislen = 3;
1592 } elseif ( $thislen > 5 ) {
1593 // If there are more than 5 apostrophes in a row, assume they're all
1594 // text except for the last 5.
1595 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1596 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1597 $arr[$i] = "'''''";
1598 $thislen = 5;
1599 }
1600 // Count the number of occurrences of bold and italics mark-ups.
1601 if ( $thislen == 2 ) {
1602 $numitalics++;
1603 } elseif ( $thislen == 3 ) {
1604 $numbold++;
1605 } elseif ( $thislen == 5 ) {
1606 $numitalics++;
1607 $numbold++;
1608 }
1609 }
1610
1611 // If there is an odd number of both bold and italics, it is likely
1612 // that one of the bold ones was meant to be an apostrophe followed
1613 // by italics. Which one we cannot know for certain, but it is more
1614 // likely to be one that has a single-letter word before it.
1615 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1616 $firstsingleletterword = -1;
1617 $firstmultiletterword = -1;
1618 $firstspace = -1;
1619 for ( $i = 1; $i < $countarr; $i += 2 ) {
1620 if ( strlen( $arr[$i] ) == 3 ) {
1621 $x1 = substr( $arr[$i - 1], -1 );
1622 $x2 = substr( $arr[$i - 1], -2, 1 );
1623 if ( $x1 === ' ' ) {
1624 if ( $firstspace == -1 ) {
1625 $firstspace = $i;
1626 }
1627 } elseif ( $x2 === ' ' ) {
1628 $firstsingleletterword = $i;
1629 // if $firstsingleletterword is set, we don't
1630 // look at the other options, so we can bail early.
1631 break;
1632 } else {
1633 if ( $firstmultiletterword == -1 ) {
1634 $firstmultiletterword = $i;
1635 }
1636 }
1637 }
1638 }
1639
1640 // If there is a single-letter word, use it!
1641 if ( $firstsingleletterword > -1 ) {
1642 $arr[$firstsingleletterword] = "''";
1643 $arr[$firstsingleletterword - 1] .= "'";
1644 } elseif ( $firstmultiletterword > -1 ) {
1645 // If not, but there's a multi-letter word, use that one.
1646 $arr[$firstmultiletterword] = "''";
1647 $arr[$firstmultiletterword - 1] .= "'";
1648 } elseif ( $firstspace > -1 ) {
1649 // ... otherwise use the first one that has neither.
1650 // (notice that it is possible for all three to be -1 if, for example,
1651 // there is only one pentuple-apostrophe in the line)
1652 $arr[$firstspace] = "''";
1653 $arr[$firstspace - 1] .= "'";
1654 }
1655 }
1656
1657 // Now let's actually convert our apostrophic mush to HTML!
1658 $output = '';
1659 $buffer = '';
1660 $state = '';
1661 $i = 0;
1662 foreach ( $arr as $r ) {
1663 if ( ( $i % 2 ) == 0 ) {
1664 if ( $state === 'both' ) {
1665 $buffer .= $r;
1666 } else {
1667 $output .= $r;
1668 }
1669 } else {
1670 $thislen = strlen( $r );
1671 if ( $thislen == 2 ) {
1672 if ( $state === 'i' ) {
1673 $output .= '</i>';
1674 $state = '';
1675 } elseif ( $state === 'bi' ) {
1676 $output .= '</i>';
1677 $state = 'b';
1678 } elseif ( $state === 'ib' ) {
1679 $output .= '</b></i><b>';
1680 $state = 'b';
1681 } elseif ( $state === 'both' ) {
1682 $output .= '<b><i>' . $buffer . '</i>';
1683 $state = 'b';
1684 } else { // $state can be 'b' or ''
1685 $output .= '<i>';
1686 $state .= 'i';
1687 }
1688 } elseif ( $thislen == 3 ) {
1689 if ( $state === 'b' ) {
1690 $output .= '</b>';
1691 $state = '';
1692 } elseif ( $state === 'bi' ) {
1693 $output .= '</i></b><i>';
1694 $state = 'i';
1695 } elseif ( $state === 'ib' ) {
1696 $output .= '</b>';
1697 $state = 'i';
1698 } elseif ( $state === 'both' ) {
1699 $output .= '<i><b>' . $buffer . '</b>';
1700 $state = 'i';
1701 } else { // $state can be 'i' or ''
1702 $output .= '<b>';
1703 $state .= 'b';
1704 }
1705 } elseif ( $thislen == 5 ) {
1706 if ( $state === 'b' ) {
1707 $output .= '</b><i>';
1708 $state = 'i';
1709 } elseif ( $state === 'i' ) {
1710 $output .= '</i><b>';
1711 $state = 'b';
1712 } elseif ( $state === 'bi' ) {
1713 $output .= '</i></b>';
1714 $state = '';
1715 } elseif ( $state === 'ib' ) {
1716 $output .= '</b></i>';
1717 $state = '';
1718 } elseif ( $state === 'both' ) {
1719 $output .= '<i><b>' . $buffer . '</b></i>';
1720 $state = '';
1721 } else { // ($state == '')
1722 $buffer = '';
1723 $state = 'both';
1724 }
1725 }
1726 }
1727 $i++;
1728 }
1729 // Now close all remaining tags. Notice that the order is important.
1730 if ( $state === 'b' || $state === 'ib' ) {
1731 $output .= '</b>';
1732 }
1733 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1734 $output .= '</i>';
1735 }
1736 if ( $state === 'bi' ) {
1737 $output .= '</b>';
1738 }
1739 // There might be lonely ''''', so make sure we have a buffer
1740 if ( $state === 'both' && $buffer ) {
1741 $output .= '<b><i>' . $buffer . '</i></b>';
1742 }
1743 return $output;
1744 }
1745
1746 /**
1747 * Replace external links (REL)
1748 *
1749 * Note: this is all very hackish and the order of execution matters a lot.
1750 * Make sure to run tests/parserTests.php if you change this code.
1751 *
1752 * @private
1753 *
1754 * @param string $text
1755 *
1756 * @throws MWException
1757 * @return string
1758 */
1759 public function replaceExternalLinks( $text ) {
1760
1761 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1762 if ( $bits === false ) {
1763 throw new MWException( "PCRE needs to be compiled with "
1764 . "--enable-unicode-properties in order for MediaWiki to function" );
1765 }
1766 $s = array_shift( $bits );
1767
1768 $i = 0;
1769 while ( $i < count( $bits ) ) {
1770 $url = $bits[$i++];
1771 $i++; // protocol
1772 $text = $bits[$i++];
1773 $trail = $bits[$i++];
1774
1775 # The characters '<' and '>' (which were escaped by
1776 # removeHTMLtags()) should not be included in
1777 # URLs, per RFC 2396.
1778 $m2 = [];
1779 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1780 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1781 $url = substr( $url, 0, $m2[0][1] );
1782 }
1783
1784 # If the link text is an image URL, replace it with an <img> tag
1785 # This happened by accident in the original parser, but some people used it extensively
1786 $img = $this->maybeMakeExternalImage( $text );
1787 if ( $img !== false ) {
1788 $text = $img;
1789 }
1790
1791 $dtrail = '';
1792
1793 # Set linktype for CSS - if URL==text, link is essentially free
1794 $linktype = ( $text === $url ) ? 'free' : 'text';
1795
1796 # No link text, e.g. [http://domain.tld/some.link]
1797 if ( $text == '' ) {
1798 # Autonumber
1799 $langObj = $this->getTargetLanguage();
1800 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1801 $linktype = 'autonumber';
1802 } else {
1803 # Have link text, e.g. [http://domain.tld/some.link text]s
1804 # Check for trail
1805 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1806 }
1807
1808 $text = $this->getConverterLanguage()->markNoConversion( $text );
1809
1810 $url = Sanitizer::cleanUrl( $url );
1811
1812 # Use the encoded URL
1813 # This means that users can paste URLs directly into the text
1814 # Funny characters like ö aren't valid in URLs anyway
1815 # This was changed in August 2004
1816 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1817 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1818
1819 # Register link in the output object.
1820 # Replace unnecessary URL escape codes with the referenced character
1821 # This prevents spammers from hiding links from the filters
1822 $pasteurized = self::normalizeLinkUrl( $url );
1823 $this->mOutput->addExternalLink( $pasteurized );
1824 }
1825
1826 return $s;
1827 }
1828
1829 /**
1830 * Get the rel attribute for a particular external link.
1831 *
1832 * @since 1.21
1833 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1834 * nofollow if appropriate
1835 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1836 * @return string|null Rel attribute for $url
1837 */
1838 public static function getExternalLinkRel( $url = false, $title = null ) {
1839 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1840 $ns = $title ? $title->getNamespace() : false;
1841 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1842 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1843 ) {
1844 return 'nofollow';
1845 }
1846 return null;
1847 }
1848
1849 /**
1850 * Get an associative array of additional HTML attributes appropriate for a
1851 * particular external link. This currently may include rel => nofollow
1852 * (depending on configuration, namespace, and the URL's domain) and/or a
1853 * target attribute (depending on configuration).
1854 *
1855 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1856 * nofollow if appropriate
1857 * @return array Associative array of HTML attributes
1858 */
1859 public function getExternalLinkAttribs( $url = false ) {
1860 $attribs = [];
1861 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1862
1863 if ( $this->mOptions->getExternalLinkTarget() ) {
1864 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1865 }
1866 return $attribs;
1867 }
1868
1869 /**
1870 * Replace unusual escape codes in a URL with their equivalent characters
1871 *
1872 * @deprecated since 1.24, use normalizeLinkUrl
1873 * @param string $url
1874 * @return string
1875 */
1876 public static function replaceUnusualEscapes( $url ) {
1877 wfDeprecated( __METHOD__, '1.24' );
1878 return self::normalizeLinkUrl( $url );
1879 }
1880
1881 /**
1882 * Replace unusual escape codes in a URL with their equivalent characters
1883 *
1884 * This generally follows the syntax defined in RFC 3986, with special
1885 * consideration for HTTP query strings.
1886 *
1887 * @param string $url
1888 * @return string
1889 */
1890 public static function normalizeLinkUrl( $url ) {
1891 # First, make sure unsafe characters are encoded
1892 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1893 function ( $m ) {
1894 return rawurlencode( $m[0] );
1895 },
1896 $url
1897 );
1898
1899 $ret = '';
1900 $end = strlen( $url );
1901
1902 # Fragment part - 'fragment'
1903 $start = strpos( $url, '#' );
1904 if ( $start !== false && $start < $end ) {
1905 $ret = self::normalizeUrlComponent(
1906 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1907 $end = $start;
1908 }
1909
1910 # Query part - 'query' minus &=+;
1911 $start = strpos( $url, '?' );
1912 if ( $start !== false && $start < $end ) {
1913 $ret = self::normalizeUrlComponent(
1914 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1915 $end = $start;
1916 }
1917
1918 # Scheme and path part - 'pchar'
1919 # (we assume no userinfo or encoded colons in the host)
1920 $ret = self::normalizeUrlComponent(
1921 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1922
1923 return $ret;
1924 }
1925
1926 private static function normalizeUrlComponent( $component, $unsafe ) {
1927 $callback = function ( $matches ) use ( $unsafe ) {
1928 $char = urldecode( $matches[0] );
1929 $ord = ord( $char );
1930 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1931 # Unescape it
1932 return $char;
1933 } else {
1934 # Leave it escaped, but use uppercase for a-f
1935 return strtoupper( $matches[0] );
1936 }
1937 };
1938 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1939 }
1940
1941 /**
1942 * make an image if it's allowed, either through the global
1943 * option, through the exception, or through the on-wiki whitelist
1944 *
1945 * @param string $url
1946 *
1947 * @return string
1948 */
1949 private function maybeMakeExternalImage( $url ) {
1950 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1951 $imagesexception = !empty( $imagesfrom );
1952 $text = false;
1953 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1954 if ( $imagesexception && is_array( $imagesfrom ) ) {
1955 $imagematch = false;
1956 foreach ( $imagesfrom as $match ) {
1957 if ( strpos( $url, $match ) === 0 ) {
1958 $imagematch = true;
1959 break;
1960 }
1961 }
1962 } elseif ( $imagesexception ) {
1963 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1964 } else {
1965 $imagematch = false;
1966 }
1967
1968 if ( $this->mOptions->getAllowExternalImages()
1969 || ( $imagesexception && $imagematch )
1970 ) {
1971 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1972 # Image found
1973 $text = Linker::makeExternalImage( $url );
1974 }
1975 }
1976 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1977 && preg_match( self::EXT_IMAGE_REGEX, $url )
1978 ) {
1979 $whitelist = explode(
1980 "\n",
1981 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
1982 );
1983
1984 foreach ( $whitelist as $entry ) {
1985 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1986 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1987 continue;
1988 }
1989 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1990 # Image matches a whitelist entry
1991 $text = Linker::makeExternalImage( $url );
1992 break;
1993 }
1994 }
1995 }
1996 return $text;
1997 }
1998
1999 /**
2000 * Process [[ ]] wikilinks
2001 *
2002 * @param string $s
2003 *
2004 * @return string Processed text
2005 *
2006 * @private
2007 */
2008 public function replaceInternalLinks( $s ) {
2009 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2010 return $s;
2011 }
2012
2013 /**
2014 * Process [[ ]] wikilinks (RIL)
2015 * @param string $s
2016 * @throws MWException
2017 * @return LinkHolderArray
2018 *
2019 * @private
2020 */
2021 public function replaceInternalLinks2( &$s ) {
2022 global $wgExtraInterlanguageLinkPrefixes;
2023
2024 static $tc = false, $e1, $e1_img;
2025 # the % is needed to support urlencoded titles as well
2026 if ( !$tc ) {
2027 $tc = Title::legalChars() . '#%';
2028 # Match a link having the form [[namespace:link|alternate]]trail
2029 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2030 # Match cases where there is no "]]", which might still be images
2031 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2032 }
2033
2034 $holders = new LinkHolderArray( $this );
2035
2036 # split the entire text string on occurrences of [[
2037 $a = StringUtils::explode( '[[', ' ' . $s );
2038 # get the first element (all text up to first [[), and remove the space we added
2039 $s = $a->current();
2040 $a->next();
2041 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2042 $s = substr( $s, 1 );
2043
2044 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2045 $e2 = null;
2046 if ( $useLinkPrefixExtension ) {
2047 # Match the end of a line for a word that's not followed by whitespace,
2048 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2049 global $wgContLang;
2050 $charset = $wgContLang->linkPrefixCharset();
2051 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2052 }
2053
2054 if ( is_null( $this->mTitle ) ) {
2055 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2056 }
2057 $nottalk = !$this->mTitle->isTalkPage();
2058
2059 if ( $useLinkPrefixExtension ) {
2060 $m = [];
2061 if ( preg_match( $e2, $s, $m ) ) {
2062 $first_prefix = $m[2];
2063 } else {
2064 $first_prefix = false;
2065 }
2066 } else {
2067 $prefix = '';
2068 }
2069
2070 $useSubpages = $this->areSubpagesAllowed();
2071
2072 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2073 # Loop for each link
2074 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2075 // @codingStandardsIgnoreEnd
2076
2077 # Check for excessive memory usage
2078 if ( $holders->isBig() ) {
2079 # Too big
2080 # Do the existence check, replace the link holders and clear the array
2081 $holders->replace( $s );
2082 $holders->clear();
2083 }
2084
2085 if ( $useLinkPrefixExtension ) {
2086 if ( preg_match( $e2, $s, $m ) ) {
2087 $prefix = $m[2];
2088 $s = $m[1];
2089 } else {
2090 $prefix = '';
2091 }
2092 # first link
2093 if ( $first_prefix ) {
2094 $prefix = $first_prefix;
2095 $first_prefix = false;
2096 }
2097 }
2098
2099 $might_be_img = false;
2100
2101 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2102 $text = $m[2];
2103 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2104 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2105 # the real problem is with the $e1 regex
2106 # See bug 1300.
2107 # Still some problems for cases where the ] is meant to be outside punctuation,
2108 # and no image is in sight. See bug 2095.
2109 if ( $text !== ''
2110 && substr( $m[3], 0, 1 ) === ']'
2111 && strpos( $text, '[' ) !== false
2112 ) {
2113 $text .= ']'; # so that replaceExternalLinks($text) works later
2114 $m[3] = substr( $m[3], 1 );
2115 }
2116 # fix up urlencoded title texts
2117 if ( strpos( $m[1], '%' ) !== false ) {
2118 # Should anchors '#' also be rejected?
2119 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2120 }
2121 $trail = $m[3];
2122 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2123 # Invalid, but might be an image with a link in its caption
2124 $might_be_img = true;
2125 $text = $m[2];
2126 if ( strpos( $m[1], '%' ) !== false ) {
2127 $m[1] = rawurldecode( $m[1] );
2128 }
2129 $trail = "";
2130 } else { # Invalid form; output directly
2131 $s .= $prefix . '[[' . $line;
2132 continue;
2133 }
2134
2135 $origLink = $m[1];
2136
2137 # Don't allow internal links to pages containing
2138 # PROTO: where PROTO is a valid URL protocol; these
2139 # should be external links.
2140 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2141 $s .= $prefix . '[[' . $line;
2142 continue;
2143 }
2144
2145 # Make subpage if necessary
2146 if ( $useSubpages ) {
2147 $link = $this->maybeDoSubpageLink( $origLink, $text );
2148 } else {
2149 $link = $origLink;
2150 }
2151
2152 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2153 if ( !$noforce ) {
2154 # Strip off leading ':'
2155 $link = substr( $link, 1 );
2156 }
2157
2158 $unstrip = $this->mStripState->unstripNoWiki( $link );
2159 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2160 if ( $nt === null ) {
2161 $s .= $prefix . '[[' . $line;
2162 continue;
2163 }
2164
2165 $ns = $nt->getNamespace();
2166 $iw = $nt->getInterwiki();
2167
2168 if ( $might_be_img ) { # if this is actually an invalid link
2169 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2170 $found = false;
2171 while ( true ) {
2172 # look at the next 'line' to see if we can close it there
2173 $a->next();
2174 $next_line = $a->current();
2175 if ( $next_line === false || $next_line === null ) {
2176 break;
2177 }
2178 $m = explode( ']]', $next_line, 3 );
2179 if ( count( $m ) == 3 ) {
2180 # the first ]] closes the inner link, the second the image
2181 $found = true;
2182 $text .= "[[{$m[0]}]]{$m[1]}";
2183 $trail = $m[2];
2184 break;
2185 } elseif ( count( $m ) == 2 ) {
2186 # if there's exactly one ]] that's fine, we'll keep looking
2187 $text .= "[[{$m[0]}]]{$m[1]}";
2188 } else {
2189 # if $next_line is invalid too, we need look no further
2190 $text .= '[[' . $next_line;
2191 break;
2192 }
2193 }
2194 if ( !$found ) {
2195 # we couldn't find the end of this imageLink, so output it raw
2196 # but don't ignore what might be perfectly normal links in the text we've examined
2197 $holders->merge( $this->replaceInternalLinks2( $text ) );
2198 $s .= "{$prefix}[[$link|$text";
2199 # note: no $trail, because without an end, there *is* no trail
2200 continue;
2201 }
2202 } else { # it's not an image, so output it raw
2203 $s .= "{$prefix}[[$link|$text";
2204 # note: no $trail, because without an end, there *is* no trail
2205 continue;
2206 }
2207 }
2208
2209 $wasblank = ( $text == '' );
2210 if ( $wasblank ) {
2211 $text = $link;
2212 } else {
2213 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2214 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2215 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2216 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2217 $text = $this->doQuotes( $text );
2218 }
2219
2220 # Link not escaped by : , create the various objects
2221 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2222 # Interwikis
2223 if (
2224 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2225 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2226 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2227 )
2228 ) {
2229 # Bug 24502: filter duplicates
2230 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2231 $this->mLangLinkLanguages[$iw] = true;
2232 $this->mOutput->addLanguageLink( $nt->getFullText() );
2233 }
2234
2235 $s = rtrim( $s . $prefix );
2236 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2237 continue;
2238 }
2239
2240 if ( $ns == NS_FILE ) {
2241 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2242 if ( $wasblank ) {
2243 # if no parameters were passed, $text
2244 # becomes something like "File:Foo.png",
2245 # which we don't want to pass on to the
2246 # image generator
2247 $text = '';
2248 } else {
2249 # recursively parse links inside the image caption
2250 # actually, this will parse them in any other parameters, too,
2251 # but it might be hard to fix that, and it doesn't matter ATM
2252 $text = $this->replaceExternalLinks( $text );
2253 $holders->merge( $this->replaceInternalLinks2( $text ) );
2254 }
2255 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2256 $s .= $prefix . $this->armorLinks(
2257 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2258 } else {
2259 $s .= $prefix . $trail;
2260 }
2261 continue;
2262 }
2263
2264 if ( $ns == NS_CATEGORY ) {
2265 $s = rtrim( $s . "\n" ); # bug 87
2266
2267 if ( $wasblank ) {
2268 $sortkey = $this->getDefaultSort();
2269 } else {
2270 $sortkey = $text;
2271 }
2272 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2273 $sortkey = str_replace( "\n", '', $sortkey );
2274 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2275 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2276
2277 /**
2278 * Strip the whitespace Category links produce, see bug 87
2279 */
2280 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2281
2282 continue;
2283 }
2284 }
2285
2286 # Self-link checking. For some languages, variants of the title are checked in
2287 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2288 # for linking to a different variant.
2289 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2290 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2291 continue;
2292 }
2293
2294 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2295 # @todo FIXME: Should do batch file existence checks, see comment below
2296 if ( $ns == NS_MEDIA ) {
2297 # Give extensions a chance to select the file revision for us
2298 $options = [];
2299 $descQuery = false;
2300 Hooks::run( 'BeforeParserFetchFileAndTitle',
2301 [ $this, $nt, &$options, &$descQuery ] );
2302 # Fetch and register the file (file title may be different via hooks)
2303 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2304 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2305 $s .= $prefix . $this->armorLinks(
2306 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2307 continue;
2308 }
2309
2310 # Some titles, such as valid special pages or files in foreign repos, should
2311 # be shown as bluelinks even though they're not included in the page table
2312 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2313 # batch file existence checks for NS_FILE and NS_MEDIA
2314 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2315 $this->mOutput->addLink( $nt );
2316 $s .= $this->makeKnownLinkHolder( $nt, $text, [], $trail, $prefix );
2317 } else {
2318 # Links will be added to the output link list after checking
2319 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2320 }
2321 }
2322 return $holders;
2323 }
2324
2325 /**
2326 * Render a forced-blue link inline; protect against double expansion of
2327 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2328 * Since this little disaster has to split off the trail text to avoid
2329 * breaking URLs in the following text without breaking trails on the
2330 * wiki links, it's been made into a horrible function.
2331 *
2332 * @param Title $nt
2333 * @param string $text
2334 * @param array|string $query
2335 * @param string $trail
2336 * @param string $prefix
2337 * @return string HTML-wikitext mix oh yuck
2338 */
2339 public function makeKnownLinkHolder( $nt, $text = '', $query = [], $trail = '', $prefix = '' ) {
2340 list( $inside, $trail ) = Linker::splitTrail( $trail );
2341
2342 if ( is_string( $query ) ) {
2343 $query = wfCgiToArray( $query );
2344 }
2345 if ( $text == '' ) {
2346 $text = htmlspecialchars( $nt->getPrefixedText() );
2347 }
2348
2349 $link = Linker::linkKnown( $nt, "$prefix$text$inside", [], $query );
2350
2351 return $this->armorLinks( $link ) . $trail;
2352 }
2353
2354 /**
2355 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2356 * going to go through further parsing steps before inline URL expansion.
2357 *
2358 * Not needed quite as much as it used to be since free links are a bit
2359 * more sensible these days. But bracketed links are still an issue.
2360 *
2361 * @param string $text More-or-less HTML
2362 * @return string Less-or-more HTML with NOPARSE bits
2363 */
2364 public function armorLinks( $text ) {
2365 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2366 self::MARKER_PREFIX . "NOPARSE$1", $text );
2367 }
2368
2369 /**
2370 * Return true if subpage links should be expanded on this page.
2371 * @return bool
2372 */
2373 public function areSubpagesAllowed() {
2374 # Some namespaces don't allow subpages
2375 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2376 }
2377
2378 /**
2379 * Handle link to subpage if necessary
2380 *
2381 * @param string $target The source of the link
2382 * @param string &$text The link text, modified as necessary
2383 * @return string The full name of the link
2384 * @private
2385 */
2386 public function maybeDoSubpageLink( $target, &$text ) {
2387 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2388 }
2389
2390 /**
2391 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2392 *
2393 * @param string $text
2394 * @param bool $linestart Whether or not this is at the start of a line.
2395 * @private
2396 * @return string The lists rendered as HTML
2397 */
2398 public function doBlockLevels( $text, $linestart ) {
2399 return BlockLevelPass::doBlockLevels( $text, $linestart );
2400 }
2401
2402 /**
2403 * Return value of a magic variable (like PAGENAME)
2404 *
2405 * @private
2406 *
2407 * @param int $index
2408 * @param bool|PPFrame $frame
2409 *
2410 * @throws MWException
2411 * @return string
2412 */
2413 public function getVariableValue( $index, $frame = false ) {
2414 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2415 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2416
2417 if ( is_null( $this->mTitle ) ) {
2418 // If no title set, bad things are going to happen
2419 // later. Title should always be set since this
2420 // should only be called in the middle of a parse
2421 // operation (but the unit-tests do funky stuff)
2422 throw new MWException( __METHOD__ . ' Should only be '
2423 . ' called while parsing (no title set)' );
2424 }
2425
2426 /**
2427 * Some of these require message or data lookups and can be
2428 * expensive to check many times.
2429 */
2430 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2431 if ( isset( $this->mVarCache[$index] ) ) {
2432 return $this->mVarCache[$index];
2433 }
2434 }
2435
2436 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2437 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2438
2439 $pageLang = $this->getFunctionLang();
2440
2441 switch ( $index ) {
2442 case '!':
2443 $value = '|';
2444 break;
2445 case 'currentmonth':
2446 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2447 break;
2448 case 'currentmonth1':
2449 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2450 break;
2451 case 'currentmonthname':
2452 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2453 break;
2454 case 'currentmonthnamegen':
2455 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2456 break;
2457 case 'currentmonthabbrev':
2458 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2459 break;
2460 case 'currentday':
2461 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2462 break;
2463 case 'currentday2':
2464 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2465 break;
2466 case 'localmonth':
2467 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2468 break;
2469 case 'localmonth1':
2470 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2471 break;
2472 case 'localmonthname':
2473 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2474 break;
2475 case 'localmonthnamegen':
2476 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2477 break;
2478 case 'localmonthabbrev':
2479 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2480 break;
2481 case 'localday':
2482 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2483 break;
2484 case 'localday2':
2485 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2486 break;
2487 case 'pagename':
2488 $value = wfEscapeWikiText( $this->mTitle->getText() );
2489 break;
2490 case 'pagenamee':
2491 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2492 break;
2493 case 'fullpagename':
2494 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2495 break;
2496 case 'fullpagenamee':
2497 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2498 break;
2499 case 'subpagename':
2500 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2501 break;
2502 case 'subpagenamee':
2503 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2504 break;
2505 case 'rootpagename':
2506 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2507 break;
2508 case 'rootpagenamee':
2509 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2510 ' ',
2511 '_',
2512 $this->mTitle->getRootText()
2513 ) ) );
2514 break;
2515 case 'basepagename':
2516 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2517 break;
2518 case 'basepagenamee':
2519 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2520 ' ',
2521 '_',
2522 $this->mTitle->getBaseText()
2523 ) ) );
2524 break;
2525 case 'talkpagename':
2526 if ( $this->mTitle->canTalk() ) {
2527 $talkPage = $this->mTitle->getTalkPage();
2528 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2529 } else {
2530 $value = '';
2531 }
2532 break;
2533 case 'talkpagenamee':
2534 if ( $this->mTitle->canTalk() ) {
2535 $talkPage = $this->mTitle->getTalkPage();
2536 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2537 } else {
2538 $value = '';
2539 }
2540 break;
2541 case 'subjectpagename':
2542 $subjPage = $this->mTitle->getSubjectPage();
2543 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2544 break;
2545 case 'subjectpagenamee':
2546 $subjPage = $this->mTitle->getSubjectPage();
2547 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2548 break;
2549 case 'pageid': // requested in bug 23427
2550 $pageid = $this->getTitle()->getArticleID();
2551 if ( $pageid == 0 ) {
2552 # 0 means the page doesn't exist in the database,
2553 # which means the user is previewing a new page.
2554 # The vary-revision flag must be set, because the magic word
2555 # will have a different value once the page is saved.
2556 $this->mOutput->setFlag( 'vary-revision' );
2557 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2558 }
2559 $value = $pageid ? $pageid : null;
2560 break;
2561 case 'revisionid':
2562 # Let the edit saving system know we should parse the page
2563 # *after* a revision ID has been assigned.
2564 $this->mOutput->setFlag( 'vary-revision' );
2565 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2566 $value = $this->mRevisionId;
2567 break;
2568 case 'revisionday':
2569 # Let the edit saving system know we should parse the page
2570 # *after* a revision ID has been assigned. This is for null edits.
2571 $this->mOutput->setFlag( 'vary-revision' );
2572 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2573 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2574 break;
2575 case 'revisionday2':
2576 # Let the edit saving system know we should parse the page
2577 # *after* a revision ID has been assigned. This is for null edits.
2578 $this->mOutput->setFlag( 'vary-revision' );
2579 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2580 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2581 break;
2582 case 'revisionmonth':
2583 # Let the edit saving system know we should parse the page
2584 # *after* a revision ID has been assigned. This is for null edits.
2585 $this->mOutput->setFlag( 'vary-revision' );
2586 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2587 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2588 break;
2589 case 'revisionmonth1':
2590 # Let the edit saving system know we should parse the page
2591 # *after* a revision ID has been assigned. This is for null edits.
2592 $this->mOutput->setFlag( 'vary-revision' );
2593 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2594 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2595 break;
2596 case 'revisionyear':
2597 # Let the edit saving system know we should parse the page
2598 # *after* a revision ID has been assigned. This is for null edits.
2599 $this->mOutput->setFlag( 'vary-revision' );
2600 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2601 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2602 break;
2603 case 'revisiontimestamp':
2604 # Let the edit saving system know we should parse the page
2605 # *after* a revision ID has been assigned. This is for null edits.
2606 $this->mOutput->setFlag( 'vary-revision' );
2607 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2608 $value = $this->getRevisionTimestamp();
2609 break;
2610 case 'revisionuser':
2611 # Let the edit saving system know we should parse the page
2612 # *after* a revision ID has been assigned. This is for null edits.
2613 $this->mOutput->setFlag( 'vary-revision' );
2614 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2615 $value = $this->getRevisionUser();
2616 break;
2617 case 'revisionsize':
2618 # Let the edit saving system know we should parse the page
2619 # *after* a revision ID has been assigned. This is for null edits.
2620 $this->mOutput->setFlag( 'vary-revision' );
2621 wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
2622 $value = $this->getRevisionSize();
2623 break;
2624 case 'namespace':
2625 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2626 break;
2627 case 'namespacee':
2628 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2629 break;
2630 case 'namespacenumber':
2631 $value = $this->mTitle->getNamespace();
2632 break;
2633 case 'talkspace':
2634 $value = $this->mTitle->canTalk()
2635 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2636 : '';
2637 break;
2638 case 'talkspacee':
2639 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2640 break;
2641 case 'subjectspace':
2642 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2643 break;
2644 case 'subjectspacee':
2645 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2646 break;
2647 case 'currentdayname':
2648 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2649 break;
2650 case 'currentyear':
2651 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2652 break;
2653 case 'currenttime':
2654 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2655 break;
2656 case 'currenthour':
2657 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2658 break;
2659 case 'currentweek':
2660 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2661 # int to remove the padding
2662 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2663 break;
2664 case 'currentdow':
2665 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2666 break;
2667 case 'localdayname':
2668 $value = $pageLang->getWeekdayName(
2669 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2670 );
2671 break;
2672 case 'localyear':
2673 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2674 break;
2675 case 'localtime':
2676 $value = $pageLang->time(
2677 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2678 false,
2679 false
2680 );
2681 break;
2682 case 'localhour':
2683 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2684 break;
2685 case 'localweek':
2686 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2687 # int to remove the padding
2688 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2689 break;
2690 case 'localdow':
2691 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2692 break;
2693 case 'numberofarticles':
2694 $value = $pageLang->formatNum( SiteStats::articles() );
2695 break;
2696 case 'numberoffiles':
2697 $value = $pageLang->formatNum( SiteStats::images() );
2698 break;
2699 case 'numberofusers':
2700 $value = $pageLang->formatNum( SiteStats::users() );
2701 break;
2702 case 'numberofactiveusers':
2703 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2704 break;
2705 case 'numberofpages':
2706 $value = $pageLang->formatNum( SiteStats::pages() );
2707 break;
2708 case 'numberofadmins':
2709 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2710 break;
2711 case 'numberofedits':
2712 $value = $pageLang->formatNum( SiteStats::edits() );
2713 break;
2714 case 'currenttimestamp':
2715 $value = wfTimestamp( TS_MW, $ts );
2716 break;
2717 case 'localtimestamp':
2718 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2719 break;
2720 case 'currentversion':
2721 $value = SpecialVersion::getVersion();
2722 break;
2723 case 'articlepath':
2724 return $wgArticlePath;
2725 case 'sitename':
2726 return $wgSitename;
2727 case 'server':
2728 return $wgServer;
2729 case 'servername':
2730 return $wgServerName;
2731 case 'scriptpath':
2732 return $wgScriptPath;
2733 case 'stylepath':
2734 return $wgStylePath;
2735 case 'directionmark':
2736 return $pageLang->getDirMark();
2737 case 'contentlanguage':
2738 global $wgLanguageCode;
2739 return $wgLanguageCode;
2740 case 'cascadingsources':
2741 $value = CoreParserFunctions::cascadingsources( $this );
2742 break;
2743 default:
2744 $ret = null;
2745 Hooks::run(
2746 'ParserGetVariableValueSwitch',
2747 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2748 );
2749
2750 return $ret;
2751 }
2752
2753 if ( $index ) {
2754 $this->mVarCache[$index] = $value;
2755 }
2756
2757 return $value;
2758 }
2759
2760 /**
2761 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2762 *
2763 * @private
2764 */
2765 public function initialiseVariables() {
2766 $variableIDs = MagicWord::getVariableIDs();
2767 $substIDs = MagicWord::getSubstIDs();
2768
2769 $this->mVariables = new MagicWordArray( $variableIDs );
2770 $this->mSubstWords = new MagicWordArray( $substIDs );
2771 }
2772
2773 /**
2774 * Preprocess some wikitext and return the document tree.
2775 * This is the ghost of replace_variables().
2776 *
2777 * @param string $text The text to parse
2778 * @param int $flags Bitwise combination of:
2779 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2780 * included. Default is to assume a direct page view.
2781 *
2782 * The generated DOM tree must depend only on the input text and the flags.
2783 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2784 *
2785 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2786 * change in the DOM tree for a given text, must be passed through the section identifier
2787 * in the section edit link and thus back to extractSections().
2788 *
2789 * The output of this function is currently only cached in process memory, but a persistent
2790 * cache may be implemented at a later date which takes further advantage of these strict
2791 * dependency requirements.
2792 *
2793 * @return PPNode
2794 */
2795 public function preprocessToDom( $text, $flags = 0 ) {
2796 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2797 return $dom;
2798 }
2799
2800 /**
2801 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2802 *
2803 * @param string $s
2804 *
2805 * @return array
2806 */
2807 public static function splitWhitespace( $s ) {
2808 $ltrimmed = ltrim( $s );
2809 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2810 $trimmed = rtrim( $ltrimmed );
2811 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2812 if ( $diff > 0 ) {
2813 $w2 = substr( $ltrimmed, -$diff );
2814 } else {
2815 $w2 = '';
2816 }
2817 return [ $w1, $trimmed, $w2 ];
2818 }
2819
2820 /**
2821 * Replace magic variables, templates, and template arguments
2822 * with the appropriate text. Templates are substituted recursively,
2823 * taking care to avoid infinite loops.
2824 *
2825 * Note that the substitution depends on value of $mOutputType:
2826 * self::OT_WIKI: only {{subst:}} templates
2827 * self::OT_PREPROCESS: templates but not extension tags
2828 * self::OT_HTML: all templates and extension tags
2829 *
2830 * @param string $text The text to transform
2831 * @param bool|PPFrame $frame Object describing the arguments passed to the
2832 * template. Arguments may also be provided as an associative array, as
2833 * was the usual case before MW1.12. Providing arguments this way may be
2834 * useful for extensions wishing to perform variable replacement
2835 * explicitly.
2836 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2837 * double-brace expansion.
2838 * @return string
2839 */
2840 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2841 # Is there any text? Also, Prevent too big inclusions!
2842 $textSize = strlen( $text );
2843 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2844 return $text;
2845 }
2846
2847 if ( $frame === false ) {
2848 $frame = $this->getPreprocessor()->newFrame();
2849 } elseif ( !( $frame instanceof PPFrame ) ) {
2850 wfDebug( __METHOD__ . " called using plain parameters instead of "
2851 . "a PPFrame instance. Creating custom frame.\n" );
2852 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2853 }
2854
2855 $dom = $this->preprocessToDom( $text );
2856 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2857 $text = $frame->expand( $dom, $flags );
2858
2859 return $text;
2860 }
2861
2862 /**
2863 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2864 *
2865 * @param array $args
2866 *
2867 * @return array
2868 */
2869 public static function createAssocArgs( $args ) {
2870 $assocArgs = [];
2871 $index = 1;
2872 foreach ( $args as $arg ) {
2873 $eqpos = strpos( $arg, '=' );
2874 if ( $eqpos === false ) {
2875 $assocArgs[$index++] = $arg;
2876 } else {
2877 $name = trim( substr( $arg, 0, $eqpos ) );
2878 $value = trim( substr( $arg, $eqpos + 1 ) );
2879 if ( $value === false ) {
2880 $value = '';
2881 }
2882 if ( $name !== false ) {
2883 $assocArgs[$name] = $value;
2884 }
2885 }
2886 }
2887
2888 return $assocArgs;
2889 }
2890
2891 /**
2892 * Warn the user when a parser limitation is reached
2893 * Will warn at most once the user per limitation type
2894 *
2895 * The results are shown during preview and run through the Parser (See EditPage.php)
2896 *
2897 * @param string $limitationType Should be one of:
2898 * 'expensive-parserfunction' (corresponding messages:
2899 * 'expensive-parserfunction-warning',
2900 * 'expensive-parserfunction-category')
2901 * 'post-expand-template-argument' (corresponding messages:
2902 * 'post-expand-template-argument-warning',
2903 * 'post-expand-template-argument-category')
2904 * 'post-expand-template-inclusion' (corresponding messages:
2905 * 'post-expand-template-inclusion-warning',
2906 * 'post-expand-template-inclusion-category')
2907 * 'node-count-exceeded' (corresponding messages:
2908 * 'node-count-exceeded-warning',
2909 * 'node-count-exceeded-category')
2910 * 'expansion-depth-exceeded' (corresponding messages:
2911 * 'expansion-depth-exceeded-warning',
2912 * 'expansion-depth-exceeded-category')
2913 * @param string|int|null $current Current value
2914 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2915 * exceeded, provide the values (optional)
2916 */
2917 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2918 # does no harm if $current and $max are present but are unnecessary for the message
2919 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2920 # only during preview, and that would split the parser cache unnecessarily.
2921 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2922 ->text();
2923 $this->mOutput->addWarning( $warning );
2924 $this->addTrackingCategory( "$limitationType-category" );
2925 }
2926
2927 /**
2928 * Return the text of a template, after recursively
2929 * replacing any variables or templates within the template.
2930 *
2931 * @param array $piece The parts of the template
2932 * $piece['title']: the title, i.e. the part before the |
2933 * $piece['parts']: the parameter array
2934 * $piece['lineStart']: whether the brace was at the start of a line
2935 * @param PPFrame $frame The current frame, contains template arguments
2936 * @throws Exception
2937 * @return string The text of the template
2938 */
2939 public function braceSubstitution( $piece, $frame ) {
2940
2941 // Flags
2942
2943 // $text has been filled
2944 $found = false;
2945 // wiki markup in $text should be escaped
2946 $nowiki = false;
2947 // $text is HTML, armour it against wikitext transformation
2948 $isHTML = false;
2949 // Force interwiki transclusion to be done in raw mode not rendered
2950 $forceRawInterwiki = false;
2951 // $text is a DOM node needing expansion in a child frame
2952 $isChildObj = false;
2953 // $text is a DOM node needing expansion in the current frame
2954 $isLocalObj = false;
2955
2956 # Title object, where $text came from
2957 $title = false;
2958
2959 # $part1 is the bit before the first |, and must contain only title characters.
2960 # Various prefixes will be stripped from it later.
2961 $titleWithSpaces = $frame->expand( $piece['title'] );
2962 $part1 = trim( $titleWithSpaces );
2963 $titleText = false;
2964
2965 # Original title text preserved for various purposes
2966 $originalTitle = $part1;
2967
2968 # $args is a list of argument nodes, starting from index 0, not including $part1
2969 # @todo FIXME: If piece['parts'] is null then the call to getLength()
2970 # below won't work b/c this $args isn't an object
2971 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
2972
2973 $profileSection = null; // profile templates
2974
2975 # SUBST
2976 if ( !$found ) {
2977 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2978
2979 # Possibilities for substMatch: "subst", "safesubst" or FALSE
2980 # Decide whether to expand template or keep wikitext as-is.
2981 if ( $this->ot['wiki'] ) {
2982 if ( $substMatch === false ) {
2983 $literal = true; # literal when in PST with no prefix
2984 } else {
2985 $literal = false; # expand when in PST with subst: or safesubst:
2986 }
2987 } else {
2988 if ( $substMatch == 'subst' ) {
2989 $literal = true; # literal when not in PST with plain subst:
2990 } else {
2991 $literal = false; # expand when not in PST with safesubst: or no prefix
2992 }
2993 }
2994 if ( $literal ) {
2995 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2996 $isLocalObj = true;
2997 $found = true;
2998 }
2999 }
3000
3001 # Variables
3002 if ( !$found && $args->getLength() == 0 ) {
3003 $id = $this->mVariables->matchStartToEnd( $part1 );
3004 if ( $id !== false ) {
3005 $text = $this->getVariableValue( $id, $frame );
3006 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3007 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3008 }
3009 $found = true;
3010 }
3011 }
3012
3013 # MSG, MSGNW and RAW
3014 if ( !$found ) {
3015 # Check for MSGNW:
3016 $mwMsgnw = MagicWord::get( 'msgnw' );
3017 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3018 $nowiki = true;
3019 } else {
3020 # Remove obsolete MSG:
3021 $mwMsg = MagicWord::get( 'msg' );
3022 $mwMsg->matchStartAndRemove( $part1 );
3023 }
3024
3025 # Check for RAW:
3026 $mwRaw = MagicWord::get( 'raw' );
3027 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3028 $forceRawInterwiki = true;
3029 }
3030 }
3031
3032 # Parser functions
3033 if ( !$found ) {
3034 $colonPos = strpos( $part1, ':' );
3035 if ( $colonPos !== false ) {
3036 $func = substr( $part1, 0, $colonPos );
3037 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3038 $argsLength = $args->getLength();
3039 for ( $i = 0; $i < $argsLength; $i++ ) {
3040 $funcArgs[] = $args->item( $i );
3041 }
3042 try {
3043 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3044 } catch ( Exception $ex ) {
3045 throw $ex;
3046 }
3047
3048 # The interface for parser functions allows for extracting
3049 # flags into the local scope. Extract any forwarded flags
3050 # here.
3051 extract( $result );
3052 }
3053 }
3054
3055 # Finish mangling title and then check for loops.
3056 # Set $title to a Title object and $titleText to the PDBK
3057 if ( !$found ) {
3058 $ns = NS_TEMPLATE;
3059 # Split the title into page and subpage
3060 $subpage = '';
3061 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3062 if ( $part1 !== $relative ) {
3063 $part1 = $relative;
3064 $ns = $this->mTitle->getNamespace();
3065 }
3066 $title = Title::newFromText( $part1, $ns );
3067 if ( $title ) {
3068 $titleText = $title->getPrefixedText();
3069 # Check for language variants if the template is not found
3070 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3071 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3072 }
3073 # Do recursion depth check
3074 $limit = $this->mOptions->getMaxTemplateDepth();
3075 if ( $frame->depth >= $limit ) {
3076 $found = true;
3077 $text = '<span class="error">'
3078 . wfMessage( 'parser-template-recursion-depth-warning' )
3079 ->numParams( $limit )->inContentLanguage()->text()
3080 . '</span>';
3081 }
3082 }
3083 }
3084
3085 # Load from database
3086 if ( !$found && $title ) {
3087 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3088 if ( !$title->isExternal() ) {
3089 if ( $title->isSpecialPage()
3090 && $this->mOptions->getAllowSpecialInclusion()
3091 && $this->ot['html']
3092 ) {
3093 // Pass the template arguments as URL parameters.
3094 // "uselang" will have no effect since the Language object
3095 // is forced to the one defined in ParserOptions.
3096 $pageArgs = [];
3097 $argsLength = $args->getLength();
3098 for ( $i = 0; $i < $argsLength; $i++ ) {
3099 $bits = $args->item( $i )->splitArg();
3100 if ( strval( $bits['index'] ) === '' ) {
3101 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3102 $value = trim( $frame->expand( $bits['value'] ) );
3103 $pageArgs[$name] = $value;
3104 }
3105 }
3106
3107 // Create a new context to execute the special page
3108 $context = new RequestContext;
3109 $context->setTitle( $title );
3110 $context->setRequest( new FauxRequest( $pageArgs ) );
3111 $context->setUser( $this->getUser() );
3112 $context->setLanguage( $this->mOptions->getUserLangObj() );
3113 $ret = SpecialPageFactory::capturePath( $title, $context );
3114 if ( $ret ) {
3115 $text = $context->getOutput()->getHTML();
3116 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3117 $found = true;
3118 $isHTML = true;
3119 $this->disableCache();
3120 }
3121 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3122 $found = false; # access denied
3123 wfDebug( __METHOD__ . ": template inclusion denied for " .
3124 $title->getPrefixedDBkey() . "\n" );
3125 } else {
3126 list( $text, $title ) = $this->getTemplateDom( $title );
3127 if ( $text !== false ) {
3128 $found = true;
3129 $isChildObj = true;
3130 }
3131 }
3132
3133 # If the title is valid but undisplayable, make a link to it
3134 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3135 $text = "[[:$titleText]]";
3136 $found = true;
3137 }
3138 } elseif ( $title->isTrans() ) {
3139 # Interwiki transclusion
3140 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3141 $text = $this->interwikiTransclude( $title, 'render' );
3142 $isHTML = true;
3143 } else {
3144 $text = $this->interwikiTransclude( $title, 'raw' );
3145 # Preprocess it like a template
3146 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3147 $isChildObj = true;
3148 }
3149 $found = true;
3150 }
3151
3152 # Do infinite loop check
3153 # This has to be done after redirect resolution to avoid infinite loops via redirects
3154 if ( !$frame->loopCheck( $title ) ) {
3155 $found = true;
3156 $text = '<span class="error">'
3157 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3158 . '</span>';
3159 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3160 }
3161 }
3162
3163 # If we haven't found text to substitute by now, we're done
3164 # Recover the source wikitext and return it
3165 if ( !$found ) {
3166 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3167 if ( $profileSection ) {
3168 $this->mProfiler->scopedProfileOut( $profileSection );
3169 }
3170 return [ 'object' => $text ];
3171 }
3172
3173 # Expand DOM-style return values in a child frame
3174 if ( $isChildObj ) {
3175 # Clean up argument array
3176 $newFrame = $frame->newChild( $args, $title );
3177
3178 if ( $nowiki ) {
3179 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3180 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3181 # Expansion is eligible for the empty-frame cache
3182 $text = $newFrame->cachedExpand( $titleText, $text );
3183 } else {
3184 # Uncached expansion
3185 $text = $newFrame->expand( $text );
3186 }
3187 }
3188 if ( $isLocalObj && $nowiki ) {
3189 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3190 $isLocalObj = false;
3191 }
3192
3193 if ( $profileSection ) {
3194 $this->mProfiler->scopedProfileOut( $profileSection );
3195 }
3196
3197 # Replace raw HTML by a placeholder
3198 if ( $isHTML ) {
3199 $text = $this->insertStripItem( $text );
3200 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3201 # Escape nowiki-style return values
3202 $text = wfEscapeWikiText( $text );
3203 } elseif ( is_string( $text )
3204 && !$piece['lineStart']
3205 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3206 ) {
3207 # Bug 529: if the template begins with a table or block-level
3208 # element, it should be treated as beginning a new line.
3209 # This behavior is somewhat controversial.
3210 $text = "\n" . $text;
3211 }
3212
3213 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3214 # Error, oversize inclusion
3215 if ( $titleText !== false ) {
3216 # Make a working, properly escaped link if possible (bug 23588)
3217 $text = "[[:$titleText]]";
3218 } else {
3219 # This will probably not be a working link, but at least it may
3220 # provide some hint of where the problem is
3221 preg_replace( '/^:/', '', $originalTitle );
3222 $text = "[[:$originalTitle]]";
3223 }
3224 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3225 . 'post-expand include size too large -->' );
3226 $this->limitationWarn( 'post-expand-template-inclusion' );
3227 }
3228
3229 if ( $isLocalObj ) {
3230 $ret = [ 'object' => $text ];
3231 } else {
3232 $ret = [ 'text' => $text ];
3233 }
3234
3235 return $ret;
3236 }
3237
3238 /**
3239 * Call a parser function and return an array with text and flags.
3240 *
3241 * The returned array will always contain a boolean 'found', indicating
3242 * whether the parser function was found or not. It may also contain the
3243 * following:
3244 * text: string|object, resulting wikitext or PP DOM object
3245 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3246 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3247 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3248 * nowiki: bool, wiki markup in $text should be escaped
3249 *
3250 * @since 1.21
3251 * @param PPFrame $frame The current frame, contains template arguments
3252 * @param string $function Function name
3253 * @param array $args Arguments to the function
3254 * @throws MWException
3255 * @return array
3256 */
3257 public function callParserFunction( $frame, $function, array $args = [] ) {
3258 global $wgContLang;
3259
3260 # Case sensitive functions
3261 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3262 $function = $this->mFunctionSynonyms[1][$function];
3263 } else {
3264 # Case insensitive functions
3265 $function = $wgContLang->lc( $function );
3266 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3267 $function = $this->mFunctionSynonyms[0][$function];
3268 } else {
3269 return [ 'found' => false ];
3270 }
3271 }
3272
3273 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3274
3275 # Workaround for PHP bug 35229 and similar
3276 if ( !is_callable( $callback ) ) {
3277 throw new MWException( "Tag hook for $function is not callable\n" );
3278 }
3279
3280 $allArgs = [ &$this ];
3281 if ( $flags & self::SFH_OBJECT_ARGS ) {
3282 # Convert arguments to PPNodes and collect for appending to $allArgs
3283 $funcArgs = [];
3284 foreach ( $args as $k => $v ) {
3285 if ( $v instanceof PPNode || $k === 0 ) {
3286 $funcArgs[] = $v;
3287 } else {
3288 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3289 }
3290 }
3291
3292 # Add a frame parameter, and pass the arguments as an array
3293 $allArgs[] = $frame;
3294 $allArgs[] = $funcArgs;
3295 } else {
3296 # Convert arguments to plain text and append to $allArgs
3297 foreach ( $args as $k => $v ) {
3298 if ( $v instanceof PPNode ) {
3299 $allArgs[] = trim( $frame->expand( $v ) );
3300 } elseif ( is_int( $k ) && $k >= 0 ) {
3301 $allArgs[] = trim( $v );
3302 } else {
3303 $allArgs[] = trim( "$k=$v" );
3304 }
3305 }
3306 }
3307
3308 $result = call_user_func_array( $callback, $allArgs );
3309
3310 # The interface for function hooks allows them to return a wikitext
3311 # string or an array containing the string and any flags. This mungs
3312 # things around to match what this method should return.
3313 if ( !is_array( $result ) ) {
3314 $result =[
3315 'found' => true,
3316 'text' => $result,
3317 ];
3318 } else {
3319 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3320 $result['text'] = $result[0];
3321 }
3322 unset( $result[0] );
3323 $result += [
3324 'found' => true,
3325 ];
3326 }
3327
3328 $noparse = true;
3329 $preprocessFlags = 0;
3330 if ( isset( $result['noparse'] ) ) {
3331 $noparse = $result['noparse'];
3332 }
3333 if ( isset( $result['preprocessFlags'] ) ) {
3334 $preprocessFlags = $result['preprocessFlags'];
3335 }
3336
3337 if ( !$noparse ) {
3338 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3339 $result['isChildObj'] = true;
3340 }
3341
3342 return $result;
3343 }
3344
3345 /**
3346 * Get the semi-parsed DOM representation of a template with a given title,
3347 * and its redirect destination title. Cached.
3348 *
3349 * @param Title $title
3350 *
3351 * @return array
3352 */
3353 public function getTemplateDom( $title ) {
3354 $cacheTitle = $title;
3355 $titleText = $title->getPrefixedDBkey();
3356
3357 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3358 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3359 $title = Title::makeTitle( $ns, $dbk );
3360 $titleText = $title->getPrefixedDBkey();
3361 }
3362 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3363 return [ $this->mTplDomCache[$titleText], $title ];
3364 }
3365
3366 # Cache miss, go to the database
3367 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3368
3369 if ( $text === false ) {
3370 $this->mTplDomCache[$titleText] = false;
3371 return [ false, $title ];
3372 }
3373
3374 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3375 $this->mTplDomCache[$titleText] = $dom;
3376
3377 if ( !$title->equals( $cacheTitle ) ) {
3378 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3379 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3380 }
3381
3382 return [ $dom, $title ];
3383 }
3384
3385 /**
3386 * Fetch the current revision of a given title. Note that the revision
3387 * (and even the title) may not exist in the database, so everything
3388 * contributing to the output of the parser should use this method
3389 * where possible, rather than getting the revisions themselves. This
3390 * method also caches its results, so using it benefits performance.
3391 *
3392 * @since 1.24
3393 * @param Title $title
3394 * @return Revision
3395 */
3396 public function fetchCurrentRevisionOfTitle( $title ) {
3397 $cacheKey = $title->getPrefixedDBkey();
3398 if ( !$this->currentRevisionCache ) {
3399 $this->currentRevisionCache = new MapCacheLRU( 100 );
3400 }
3401 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3402 $this->currentRevisionCache->set( $cacheKey,
3403 // Defaults to Parser::statelessFetchRevision()
3404 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3405 );
3406 }
3407 return $this->currentRevisionCache->get( $cacheKey );
3408 }
3409
3410 /**
3411 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3412 * without passing them on to it.
3413 *
3414 * @since 1.24
3415 * @param Title $title
3416 * @param Parser|bool $parser
3417 * @return Revision
3418 */
3419 public static function statelessFetchRevision( $title, $parser = false ) {
3420 return Revision::newFromTitle( $title );
3421 }
3422
3423 /**
3424 * Fetch the unparsed text of a template and register a reference to it.
3425 * @param Title $title
3426 * @return array ( string or false, Title )
3427 */
3428 public function fetchTemplateAndTitle( $title ) {
3429 // Defaults to Parser::statelessFetchTemplate()
3430 $templateCb = $this->mOptions->getTemplateCallback();
3431 $stuff = call_user_func( $templateCb, $title, $this );
3432 // We use U+007F DELETE to distinguish strip markers from regular text.
3433 $text = $stuff['text'];
3434 if ( is_string( $stuff['text'] ) ) {
3435 $text = strtr( $text, "\x7f", "?" );
3436 }
3437 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3438 if ( isset( $stuff['deps'] ) ) {
3439 foreach ( $stuff['deps'] as $dep ) {
3440 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3441 if ( $dep['title']->equals( $this->getTitle() ) ) {
3442 // If we transclude ourselves, the final result
3443 // will change based on the new version of the page
3444 $this->mOutput->setFlag( 'vary-revision' );
3445 }
3446 }
3447 }
3448 return [ $text, $finalTitle ];
3449 }
3450
3451 /**
3452 * Fetch the unparsed text of a template and register a reference to it.
3453 * @param Title $title
3454 * @return string|bool
3455 */
3456 public function fetchTemplate( $title ) {
3457 return $this->fetchTemplateAndTitle( $title )[0];
3458 }
3459
3460 /**
3461 * Static function to get a template
3462 * Can be overridden via ParserOptions::setTemplateCallback().
3463 *
3464 * @param Title $title
3465 * @param bool|Parser $parser
3466 *
3467 * @return array
3468 */
3469 public static function statelessFetchTemplate( $title, $parser = false ) {
3470 $text = $skip = false;
3471 $finalTitle = $title;
3472 $deps = [];
3473
3474 # Loop to fetch the article, with up to 1 redirect
3475 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3476 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3477 // @codingStandardsIgnoreEnd
3478 # Give extensions a chance to select the revision instead
3479 $id = false; # Assume current
3480 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3481 [ $parser, $title, &$skip, &$id ] );
3482
3483 if ( $skip ) {
3484 $text = false;
3485 $deps[] = [
3486 'title' => $title,
3487 'page_id' => $title->getArticleID(),
3488 'rev_id' => null
3489 ];
3490 break;
3491 }
3492 # Get the revision
3493 if ( $id ) {
3494 $rev = Revision::newFromId( $id );
3495 } elseif ( $parser ) {
3496 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3497 } else {
3498 $rev = Revision::newFromTitle( $title );
3499 }
3500 $rev_id = $rev ? $rev->getId() : 0;
3501 # If there is no current revision, there is no page
3502 if ( $id === false && !$rev ) {
3503 $linkCache = LinkCache::singleton();
3504 $linkCache->addBadLinkObj( $title );
3505 }
3506
3507 $deps[] = [
3508 'title' => $title,
3509 'page_id' => $title->getArticleID(),
3510 'rev_id' => $rev_id ];
3511 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3512 # We fetched a rev from a different title; register it too...
3513 $deps[] = [
3514 'title' => $rev->getTitle(),
3515 'page_id' => $rev->getPage(),
3516 'rev_id' => $rev_id ];
3517 }
3518
3519 if ( $rev ) {
3520 $content = $rev->getContent();
3521 $text = $content ? $content->getWikitextForTransclusion() : null;
3522
3523 if ( $text === false || $text === null ) {
3524 $text = false;
3525 break;
3526 }
3527 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3528 global $wgContLang;
3529 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3530 if ( !$message->exists() ) {
3531 $text = false;
3532 break;
3533 }
3534 $content = $message->content();
3535 $text = $message->plain();
3536 } else {
3537 break;
3538 }
3539 if ( !$content ) {
3540 break;
3541 }
3542 # Redirect?
3543 $finalTitle = $title;
3544 $title = $content->getRedirectTarget();
3545 }
3546 return [
3547 'text' => $text,
3548 'finalTitle' => $finalTitle,
3549 'deps' => $deps ];
3550 }
3551
3552 /**
3553 * Fetch a file and its title and register a reference to it.
3554 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3555 * @param Title $title
3556 * @param array $options Array of options to RepoGroup::findFile
3557 * @return File|bool
3558 */
3559 public function fetchFile( $title, $options = [] ) {
3560 return $this->fetchFileAndTitle( $title, $options )[0];
3561 }
3562
3563 /**
3564 * Fetch a file and its title and register a reference to it.
3565 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3566 * @param Title $title
3567 * @param array $options Array of options to RepoGroup::findFile
3568 * @return array ( File or false, Title of file )
3569 */
3570 public function fetchFileAndTitle( $title, $options = [] ) {
3571 $file = $this->fetchFileNoRegister( $title, $options );
3572
3573 $time = $file ? $file->getTimestamp() : false;
3574 $sha1 = $file ? $file->getSha1() : false;
3575 # Register the file as a dependency...
3576 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3577 if ( $file && !$title->equals( $file->getTitle() ) ) {
3578 # Update fetched file title
3579 $title = $file->getTitle();
3580 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3581 }
3582 return [ $file, $title ];
3583 }
3584
3585 /**
3586 * Helper function for fetchFileAndTitle.
3587 *
3588 * Also useful if you need to fetch a file but not use it yet,
3589 * for example to get the file's handler.
3590 *
3591 * @param Title $title
3592 * @param array $options Array of options to RepoGroup::findFile
3593 * @return File|bool
3594 */
3595 protected function fetchFileNoRegister( $title, $options = [] ) {
3596 if ( isset( $options['broken'] ) ) {
3597 $file = false; // broken thumbnail forced by hook
3598 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3599 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3600 } else { // get by (name,timestamp)
3601 $file = wfFindFile( $title, $options );
3602 }
3603 return $file;
3604 }
3605
3606 /**
3607 * Transclude an interwiki link.
3608 *
3609 * @param Title $title
3610 * @param string $action
3611 *
3612 * @return string
3613 */
3614 public function interwikiTransclude( $title, $action ) {
3615 global $wgEnableScaryTranscluding;
3616
3617 if ( !$wgEnableScaryTranscluding ) {
3618 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3619 }
3620
3621 $url = $title->getFullURL( [ 'action' => $action ] );
3622
3623 if ( strlen( $url ) > 255 ) {
3624 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3625 }
3626 return $this->fetchScaryTemplateMaybeFromCache( $url );
3627 }
3628
3629 /**
3630 * @param string $url
3631 * @return mixed|string
3632 */
3633 public function fetchScaryTemplateMaybeFromCache( $url ) {
3634 global $wgTranscludeCacheExpiry;
3635 $dbr = wfGetDB( DB_SLAVE );
3636 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3637 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3638 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3639 if ( $obj ) {
3640 return $obj->tc_contents;
3641 }
3642
3643 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3644 $status = $req->execute(); // Status object
3645 if ( $status->isOK() ) {
3646 $text = $req->getContent();
3647 } elseif ( $req->getStatus() != 200 ) {
3648 // Though we failed to fetch the content, this status is useless.
3649 return wfMessage( 'scarytranscludefailed-httpstatus' )
3650 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3651 } else {
3652 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3653 }
3654
3655 $dbw = wfGetDB( DB_MASTER );
3656 $dbw->replace( 'transcache', [ 'tc_url' ], [
3657 'tc_url' => $url,
3658 'tc_time' => $dbw->timestamp( time() ),
3659 'tc_contents' => $text
3660 ] );
3661 return $text;
3662 }
3663
3664 /**
3665 * Triple brace replacement -- used for template arguments
3666 * @private
3667 *
3668 * @param array $piece
3669 * @param PPFrame $frame
3670 *
3671 * @return array
3672 */
3673 public function argSubstitution( $piece, $frame ) {
3674
3675 $error = false;
3676 $parts = $piece['parts'];
3677 $nameWithSpaces = $frame->expand( $piece['title'] );
3678 $argName = trim( $nameWithSpaces );
3679 $object = false;
3680 $text = $frame->getArgument( $argName );
3681 if ( $text === false && $parts->getLength() > 0
3682 && ( $this->ot['html']
3683 || $this->ot['pre']
3684 || ( $this->ot['wiki'] && $frame->isTemplate() )
3685 )
3686 ) {
3687 # No match in frame, use the supplied default
3688 $object = $parts->item( 0 )->getChildren();
3689 }
3690 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3691 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3692 $this->limitationWarn( 'post-expand-template-argument' );
3693 }
3694
3695 if ( $text === false && $object === false ) {
3696 # No match anywhere
3697 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3698 }
3699 if ( $error !== false ) {
3700 $text .= $error;
3701 }
3702 if ( $object !== false ) {
3703 $ret = [ 'object' => $object ];
3704 } else {
3705 $ret = [ 'text' => $text ];
3706 }
3707
3708 return $ret;
3709 }
3710
3711 /**
3712 * Return the text to be used for a given extension tag.
3713 * This is the ghost of strip().
3714 *
3715 * @param array $params Associative array of parameters:
3716 * name PPNode for the tag name
3717 * attr PPNode for unparsed text where tag attributes are thought to be
3718 * attributes Optional associative array of parsed attributes
3719 * inner Contents of extension element
3720 * noClose Original text did not have a close tag
3721 * @param PPFrame $frame
3722 *
3723 * @throws MWException
3724 * @return string
3725 */
3726 public function extensionSubstitution( $params, $frame ) {
3727 $name = $frame->expand( $params['name'] );
3728 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3729 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3730 $marker = self::MARKER_PREFIX . "-$name-"
3731 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3732
3733 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3734 ( $this->ot['html'] || $this->ot['pre'] );
3735 if ( $isFunctionTag ) {
3736 $markerType = 'none';
3737 } else {
3738 $markerType = 'general';
3739 }
3740 if ( $this->ot['html'] || $isFunctionTag ) {
3741 $name = strtolower( $name );
3742 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3743 if ( isset( $params['attributes'] ) ) {
3744 $attributes = $attributes + $params['attributes'];
3745 }
3746
3747 if ( isset( $this->mTagHooks[$name] ) ) {
3748 # Workaround for PHP bug 35229 and similar
3749 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3750 throw new MWException( "Tag hook for $name is not callable\n" );
3751 }
3752 $output = call_user_func_array( $this->mTagHooks[$name],
3753 [ $content, $attributes, $this, $frame ] );
3754 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3755 list( $callback, ) = $this->mFunctionTagHooks[$name];
3756 if ( !is_callable( $callback ) ) {
3757 throw new MWException( "Tag hook for $name is not callable\n" );
3758 }
3759
3760 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3761 } else {
3762 $output = '<span class="error">Invalid tag extension name: ' .
3763 htmlspecialchars( $name ) . '</span>';
3764 }
3765
3766 if ( is_array( $output ) ) {
3767 # Extract flags to local scope (to override $markerType)
3768 $flags = $output;
3769 $output = $flags[0];
3770 unset( $flags[0] );
3771 extract( $flags );
3772 }
3773 } else {
3774 if ( is_null( $attrText ) ) {
3775 $attrText = '';
3776 }
3777 if ( isset( $params['attributes'] ) ) {
3778 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3779 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3780 htmlspecialchars( $attrValue ) . '"';
3781 }
3782 }
3783 if ( $content === null ) {
3784 $output = "<$name$attrText/>";
3785 } else {
3786 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3787 $output = "<$name$attrText>$content$close";
3788 }
3789 }
3790
3791 if ( $markerType === 'none' ) {
3792 return $output;
3793 } elseif ( $markerType === 'nowiki' ) {
3794 $this->mStripState->addNoWiki( $marker, $output );
3795 } elseif ( $markerType === 'general' ) {
3796 $this->mStripState->addGeneral( $marker, $output );
3797 } else {
3798 throw new MWException( __METHOD__ . ': invalid marker type' );
3799 }
3800 return $marker;
3801 }
3802
3803 /**
3804 * Increment an include size counter
3805 *
3806 * @param string $type The type of expansion
3807 * @param int $size The size of the text
3808 * @return bool False if this inclusion would take it over the maximum, true otherwise
3809 */
3810 public function incrementIncludeSize( $type, $size ) {
3811 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3812 return false;
3813 } else {
3814 $this->mIncludeSizes[$type] += $size;
3815 return true;
3816 }
3817 }
3818
3819 /**
3820 * Increment the expensive function count
3821 *
3822 * @return bool False if the limit has been exceeded
3823 */
3824 public function incrementExpensiveFunctionCount() {
3825 $this->mExpensiveFunctionCount++;
3826 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3827 }
3828
3829 /**
3830 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3831 * Fills $this->mDoubleUnderscores, returns the modified text
3832 *
3833 * @param string $text
3834 *
3835 * @return string
3836 */
3837 public function doDoubleUnderscore( $text ) {
3838
3839 # The position of __TOC__ needs to be recorded
3840 $mw = MagicWord::get( 'toc' );
3841 if ( $mw->match( $text ) ) {
3842 $this->mShowToc = true;
3843 $this->mForceTocPosition = true;
3844
3845 # Set a placeholder. At the end we'll fill it in with the TOC.
3846 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3847
3848 # Only keep the first one.
3849 $text = $mw->replace( '', $text );
3850 }
3851
3852 # Now match and remove the rest of them
3853 $mwa = MagicWord::getDoubleUnderscoreArray();
3854 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3855
3856 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3857 $this->mOutput->mNoGallery = true;
3858 }
3859 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3860 $this->mShowToc = false;
3861 }
3862 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3863 && $this->mTitle->getNamespace() == NS_CATEGORY
3864 ) {
3865 $this->addTrackingCategory( 'hidden-category-category' );
3866 }
3867 # (bug 8068) Allow control over whether robots index a page.
3868 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3869 # is not desirable, the last one on the page should win.
3870 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3871 $this->mOutput->setIndexPolicy( 'noindex' );
3872 $this->addTrackingCategory( 'noindex-category' );
3873 }
3874 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3875 $this->mOutput->setIndexPolicy( 'index' );
3876 $this->addTrackingCategory( 'index-category' );
3877 }
3878
3879 # Cache all double underscores in the database
3880 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3881 $this->mOutput->setProperty( $key, '' );
3882 }
3883
3884 return $text;
3885 }
3886
3887 /**
3888 * @see ParserOutput::addTrackingCategory()
3889 * @param string $msg Message key
3890 * @return bool Whether the addition was successful
3891 */
3892 public function addTrackingCategory( $msg ) {
3893 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3894 }
3895
3896 /**
3897 * This function accomplishes several tasks:
3898 * 1) Auto-number headings if that option is enabled
3899 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3900 * 3) Add a Table of contents on the top for users who have enabled the option
3901 * 4) Auto-anchor headings
3902 *
3903 * It loops through all headlines, collects the necessary data, then splits up the
3904 * string and re-inserts the newly formatted headlines.
3905 *
3906 * @param string $text
3907 * @param string $origText Original, untouched wikitext
3908 * @param bool $isMain
3909 * @return mixed|string
3910 * @private
3911 */
3912 public function formatHeadings( $text, $origText, $isMain = true ) {
3913 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
3914
3915 # Inhibit editsection links if requested in the page
3916 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3917 $maybeShowEditLink = $showEditLink = false;
3918 } else {
3919 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
3920 $showEditLink = $this->mOptions->getEditSection();
3921 }
3922 if ( $showEditLink ) {
3923 $this->mOutput->setEditSectionTokens( true );
3924 }
3925
3926 # Get all headlines for numbering them and adding funky stuff like [edit]
3927 # links - this is for later, but we need the number of headlines right now
3928 $matches = [];
3929 $numMatches = preg_match_all(
3930 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
3931 $text,
3932 $matches
3933 );
3934
3935 # if there are fewer than 4 headlines in the article, do not show TOC
3936 # unless it's been explicitly enabled.
3937 $enoughToc = $this->mShowToc &&
3938 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3939
3940 # Allow user to stipulate that a page should have a "new section"
3941 # link added via __NEWSECTIONLINK__
3942 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3943 $this->mOutput->setNewSection( true );
3944 }
3945
3946 # Allow user to remove the "new section"
3947 # link via __NONEWSECTIONLINK__
3948 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3949 $this->mOutput->hideNewSection( true );
3950 }
3951
3952 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3953 # override above conditions and always show TOC above first header
3954 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3955 $this->mShowToc = true;
3956 $enoughToc = true;
3957 }
3958
3959 # headline counter
3960 $headlineCount = 0;
3961 $numVisible = 0;
3962
3963 # Ugh .. the TOC should have neat indentation levels which can be
3964 # passed to the skin functions. These are determined here
3965 $toc = '';
3966 $full = '';
3967 $head = [];
3968 $sublevelCount = [];
3969 $levelCount = [];
3970 $level = 0;
3971 $prevlevel = 0;
3972 $toclevel = 0;
3973 $prevtoclevel = 0;
3974 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
3975 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3976 $oldType = $this->mOutputType;
3977 $this->setOutputType( self::OT_WIKI );
3978 $frame = $this->getPreprocessor()->newFrame();
3979 $root = $this->preprocessToDom( $origText );
3980 $node = $root->getFirstChild();
3981 $byteOffset = 0;
3982 $tocraw = [];
3983 $refers = [];
3984
3985 $headlines = $numMatches !== false ? $matches[3] : [];
3986
3987 foreach ( $headlines as $headline ) {
3988 $isTemplate = false;
3989 $titleText = false;
3990 $sectionIndex = false;
3991 $numbering = '';
3992 $markerMatches = [];
3993 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
3994 $serial = $markerMatches[1];
3995 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3996 $isTemplate = ( $titleText != $baseTitleText );
3997 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
3998 }
3999
4000 if ( $toclevel ) {
4001 $prevlevel = $level;
4002 }
4003 $level = $matches[1][$headlineCount];
4004
4005 if ( $level > $prevlevel ) {
4006 # Increase TOC level
4007 $toclevel++;
4008 $sublevelCount[$toclevel] = 0;
4009 if ( $toclevel < $wgMaxTocLevel ) {
4010 $prevtoclevel = $toclevel;
4011 $toc .= Linker::tocIndent();
4012 $numVisible++;
4013 }
4014 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4015 # Decrease TOC level, find level to jump to
4016
4017 for ( $i = $toclevel; $i > 0; $i-- ) {
4018 if ( $levelCount[$i] == $level ) {
4019 # Found last matching level
4020 $toclevel = $i;
4021 break;
4022 } elseif ( $levelCount[$i] < $level ) {
4023 # Found first matching level below current level
4024 $toclevel = $i + 1;
4025 break;
4026 }
4027 }
4028 if ( $i == 0 ) {
4029 $toclevel = 1;
4030 }
4031 if ( $toclevel < $wgMaxTocLevel ) {
4032 if ( $prevtoclevel < $wgMaxTocLevel ) {
4033 # Unindent only if the previous toc level was shown :p
4034 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4035 $prevtoclevel = $toclevel;
4036 } else {
4037 $toc .= Linker::tocLineEnd();
4038 }
4039 }
4040 } else {
4041 # No change in level, end TOC line
4042 if ( $toclevel < $wgMaxTocLevel ) {
4043 $toc .= Linker::tocLineEnd();
4044 }
4045 }
4046
4047 $levelCount[$toclevel] = $level;
4048
4049 # count number of headlines for each level
4050 $sublevelCount[$toclevel]++;
4051 $dot = 0;
4052 for ( $i = 1; $i <= $toclevel; $i++ ) {
4053 if ( !empty( $sublevelCount[$i] ) ) {
4054 if ( $dot ) {
4055 $numbering .= '.';
4056 }
4057 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4058 $dot = 1;
4059 }
4060 }
4061
4062 # The safe header is a version of the header text safe to use for links
4063
4064 # Remove link placeholders by the link text.
4065 # <!--LINK number-->
4066 # turns into
4067 # link text with suffix
4068 # Do this before unstrip since link text can contain strip markers
4069 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4070
4071 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4072 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4073
4074 # Strip out HTML (first regex removes any tag not allowed)
4075 # Allowed tags are:
4076 # * <sup> and <sub> (bug 8393)
4077 # * <i> (bug 26375)
4078 # * <b> (r105284)
4079 # * <bdi> (bug 72884)
4080 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4081 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4082 # to allow setting directionality in toc items.
4083 $tocline = preg_replace(
4084 [
4085 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4086 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4087 ],
4088 [ '', '<$1>' ],
4089 $safeHeadline
4090 );
4091
4092 # Strip '<span></span>', which is the result from the above if
4093 # <span id="foo"></span> is used to produce an additional anchor
4094 # for a section.
4095 $tocline = str_replace( '<span></span>', '', $tocline );
4096
4097 $tocline = trim( $tocline );
4098
4099 # For the anchor, strip out HTML-y stuff period
4100 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4101 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4102
4103 # Save headline for section edit hint before it's escaped
4104 $headlineHint = $safeHeadline;
4105
4106 if ( $wgExperimentalHtmlIds ) {
4107 # For reverse compatibility, provide an id that's
4108 # HTML4-compatible, like we used to.
4109 # It may be worth noting, academically, that it's possible for
4110 # the legacy anchor to conflict with a non-legacy headline
4111 # anchor on the page. In this case likely the "correct" thing
4112 # would be to either drop the legacy anchors or make sure
4113 # they're numbered first. However, this would require people
4114 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4115 # manually, so let's not bother worrying about it.
4116 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4117 [ 'noninitial', 'legacy' ] );
4118 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4119
4120 if ( $legacyHeadline == $safeHeadline ) {
4121 # No reason to have both (in fact, we can't)
4122 $legacyHeadline = false;
4123 }
4124 } else {
4125 $legacyHeadline = false;
4126 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4127 'noninitial' );
4128 }
4129
4130 # HTML names must be case-insensitively unique (bug 10721).
4131 # This does not apply to Unicode characters per
4132 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4133 # @todo FIXME: We may be changing them depending on the current locale.
4134 $arrayKey = strtolower( $safeHeadline );
4135 if ( $legacyHeadline === false ) {
4136 $legacyArrayKey = false;
4137 } else {
4138 $legacyArrayKey = strtolower( $legacyHeadline );
4139 }
4140
4141 # Create the anchor for linking from the TOC to the section
4142 $anchor = $safeHeadline;
4143 $legacyAnchor = $legacyHeadline;
4144 if ( isset( $refers[$arrayKey] ) ) {
4145 // @codingStandardsIgnoreStart
4146 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4147 // @codingStandardsIgnoreEnd
4148 $anchor .= "_$i";
4149 $refers["${arrayKey}_$i"] = true;
4150 } else {
4151 $refers[$arrayKey] = true;
4152 }
4153 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4154 // @codingStandardsIgnoreStart
4155 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4156 // @codingStandardsIgnoreEnd
4157 $legacyAnchor .= "_$i";
4158 $refers["${legacyArrayKey}_$i"] = true;
4159 } else {
4160 $refers[$legacyArrayKey] = true;
4161 }
4162
4163 # Don't number the heading if it is the only one (looks silly)
4164 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4165 # the two are different if the line contains a link
4166 $headline = Html::element(
4167 'span',
4168 [ 'class' => 'mw-headline-number' ],
4169 $numbering
4170 ) . ' ' . $headline;
4171 }
4172
4173 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4174 $toc .= Linker::tocLine( $anchor, $tocline,
4175 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4176 }
4177
4178 # Add the section to the section tree
4179 # Find the DOM node for this header
4180 $noOffset = ( $isTemplate || $sectionIndex === false );
4181 while ( $node && !$noOffset ) {
4182 if ( $node->getName() === 'h' ) {
4183 $bits = $node->splitHeading();
4184 if ( $bits['i'] == $sectionIndex ) {
4185 break;
4186 }
4187 }
4188 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4189 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4190 $node = $node->getNextSibling();
4191 }
4192 $tocraw[] = [
4193 'toclevel' => $toclevel,
4194 'level' => $level,
4195 'line' => $tocline,
4196 'number' => $numbering,
4197 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4198 'fromtitle' => $titleText,
4199 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4200 'anchor' => $anchor,
4201 ];
4202
4203 # give headline the correct <h#> tag
4204 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4205 // Output edit section links as markers with styles that can be customized by skins
4206 if ( $isTemplate ) {
4207 # Put a T flag in the section identifier, to indicate to extractSections()
4208 # that sections inside <includeonly> should be counted.
4209 $editsectionPage = $titleText;
4210 $editsectionSection = "T-$sectionIndex";
4211 $editsectionContent = null;
4212 } else {
4213 $editsectionPage = $this->mTitle->getPrefixedText();
4214 $editsectionSection = $sectionIndex;
4215 $editsectionContent = $headlineHint;
4216 }
4217 // We use a bit of pesudo-xml for editsection markers. The
4218 // language converter is run later on. Using a UNIQ style marker
4219 // leads to the converter screwing up the tokens when it
4220 // converts stuff. And trying to insert strip tags fails too. At
4221 // this point all real inputted tags have already been escaped,
4222 // so we don't have to worry about a user trying to input one of
4223 // these markers directly. We use a page and section attribute
4224 // to stop the language converter from converting these
4225 // important bits of data, but put the headline hint inside a
4226 // content block because the language converter is supposed to
4227 // be able to convert that piece of data.
4228 // Gets replaced with html in ParserOutput::getText
4229 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4230 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4231 if ( $editsectionContent !== null ) {
4232 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4233 } else {
4234 $editlink .= '/>';
4235 }
4236 } else {
4237 $editlink = '';
4238 }
4239 $head[$headlineCount] = Linker::makeHeadline( $level,
4240 $matches['attrib'][$headlineCount], $anchor, $headline,
4241 $editlink, $legacyAnchor );
4242
4243 $headlineCount++;
4244 }
4245
4246 $this->setOutputType( $oldType );
4247
4248 # Never ever show TOC if no headers
4249 if ( $numVisible < 1 ) {
4250 $enoughToc = false;
4251 }
4252
4253 if ( $enoughToc ) {
4254 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4255 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4256 }
4257 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4258 $this->mOutput->setTOCHTML( $toc );
4259 $toc = self::TOC_START . $toc . self::TOC_END;
4260 $this->mOutput->addModules( 'mediawiki.toc' );
4261 }
4262
4263 if ( $isMain ) {
4264 $this->mOutput->setSections( $tocraw );
4265 }
4266
4267 # split up and insert constructed headlines
4268 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4269 $i = 0;
4270
4271 // build an array of document sections
4272 $sections = [];
4273 foreach ( $blocks as $block ) {
4274 // $head is zero-based, sections aren't.
4275 if ( empty( $head[$i - 1] ) ) {
4276 $sections[$i] = $block;
4277 } else {
4278 $sections[$i] = $head[$i - 1] . $block;
4279 }
4280
4281 /**
4282 * Send a hook, one per section.
4283 * The idea here is to be able to make section-level DIVs, but to do so in a
4284 * lower-impact, more correct way than r50769
4285 *
4286 * $this : caller
4287 * $section : the section number
4288 * &$sectionContent : ref to the content of the section
4289 * $showEditLinks : boolean describing whether this section has an edit link
4290 */
4291 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4292
4293 $i++;
4294 }
4295
4296 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4297 // append the TOC at the beginning
4298 // Top anchor now in skin
4299 $sections[0] = $sections[0] . $toc . "\n";
4300 }
4301
4302 $full .= implode( '', $sections );
4303
4304 if ( $this->mForceTocPosition ) {
4305 return str_replace( '<!--MWTOC-->', $toc, $full );
4306 } else {
4307 return $full;
4308 }
4309 }
4310
4311 /**
4312 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4313 * conversion, substituting signatures, {{subst:}} templates, etc.
4314 *
4315 * @param string $text The text to transform
4316 * @param Title $title The Title object for the current article
4317 * @param User $user The User object describing the current user
4318 * @param ParserOptions $options Parsing options
4319 * @param bool $clearState Whether to clear the parser state first
4320 * @return string The altered wiki markup
4321 */
4322 public function preSaveTransform( $text, Title $title, User $user,
4323 ParserOptions $options, $clearState = true
4324 ) {
4325 if ( $clearState ) {
4326 $magicScopeVariable = $this->lock();
4327 }
4328 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4329 $this->setUser( $user );
4330
4331 $pairs = [
4332 "\r\n" => "\n",
4333 "\r" => "\n",
4334 ];
4335 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4336 if ( $options->getPreSaveTransform() ) {
4337 $text = $this->pstPass2( $text, $user );
4338 }
4339 $text = $this->mStripState->unstripBoth( $text );
4340
4341 $this->setUser( null ); # Reset
4342
4343 return $text;
4344 }
4345
4346 /**
4347 * Pre-save transform helper function
4348 *
4349 * @param string $text
4350 * @param User $user
4351 *
4352 * @return string
4353 */
4354 private function pstPass2( $text, $user ) {
4355 global $wgContLang;
4356
4357 # Note: This is the timestamp saved as hardcoded wikitext to
4358 # the database, we use $wgContLang here in order to give
4359 # everyone the same signature and use the default one rather
4360 # than the one selected in each user's preferences.
4361 # (see also bug 12815)
4362 $ts = $this->mOptions->getTimestamp();
4363 $timestamp = MWTimestamp::getLocalInstance( $ts );
4364 $ts = $timestamp->format( 'YmdHis' );
4365 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4366
4367 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4368
4369 # Variable replacement
4370 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4371 $text = $this->replaceVariables( $text );
4372
4373 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4374 # which may corrupt this parser instance via its wfMessage()->text() call-
4375
4376 # Signatures
4377 $sigText = $this->getUserSig( $user );
4378 $text = strtr( $text, [
4379 '~~~~~' => $d,
4380 '~~~~' => "$sigText $d",
4381 '~~~' => $sigText
4382 ] );
4383
4384 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4385 $tc = '[' . Title::legalChars() . ']';
4386 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4387
4388 // [[ns:page (context)|]]
4389 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4390 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4391 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4392 // [[ns:page (context), context|]] (using either single or double-width comma)
4393 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4394 // [[|page]] (reverse pipe trick: add context from page title)
4395 $p2 = "/\[\[\\|($tc+)]]/";
4396
4397 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4398 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4399 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4400 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4401
4402 $t = $this->mTitle->getText();
4403 $m = [];
4404 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4405 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4406 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4407 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4408 } else {
4409 # if there's no context, don't bother duplicating the title
4410 $text = preg_replace( $p2, '[[\\1]]', $text );
4411 }
4412
4413 # Trim trailing whitespace
4414 $text = rtrim( $text );
4415
4416 return $text;
4417 }
4418
4419 /**
4420 * Fetch the user's signature text, if any, and normalize to
4421 * validated, ready-to-insert wikitext.
4422 * If you have pre-fetched the nickname or the fancySig option, you can
4423 * specify them here to save a database query.
4424 * Do not reuse this parser instance after calling getUserSig(),
4425 * as it may have changed if it's the $wgParser.
4426 *
4427 * @param User $user
4428 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4429 * @param bool|null $fancySig whether the nicknname is the complete signature
4430 * or null to use default value
4431 * @return string
4432 */
4433 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4434 global $wgMaxSigChars;
4435
4436 $username = $user->getName();
4437
4438 # If not given, retrieve from the user object.
4439 if ( $nickname === false ) {
4440 $nickname = $user->getOption( 'nickname' );
4441 }
4442
4443 if ( is_null( $fancySig ) ) {
4444 $fancySig = $user->getBoolOption( 'fancysig' );
4445 }
4446
4447 $nickname = $nickname == null ? $username : $nickname;
4448
4449 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4450 $nickname = $username;
4451 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4452 } elseif ( $fancySig !== false ) {
4453 # Sig. might contain markup; validate this
4454 if ( $this->validateSig( $nickname ) !== false ) {
4455 # Validated; clean up (if needed) and return it
4456 return $this->cleanSig( $nickname, true );
4457 } else {
4458 # Failed to validate; fall back to the default
4459 $nickname = $username;
4460 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4461 }
4462 }
4463
4464 # Make sure nickname doesnt get a sig in a sig
4465 $nickname = self::cleanSigInSig( $nickname );
4466
4467 # If we're still here, make it a link to the user page
4468 $userText = wfEscapeWikiText( $username );
4469 $nickText = wfEscapeWikiText( $nickname );
4470 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4471
4472 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4473 ->title( $this->getTitle() )->text();
4474 }
4475
4476 /**
4477 * Check that the user's signature contains no bad XML
4478 *
4479 * @param string $text
4480 * @return string|bool An expanded string, or false if invalid.
4481 */
4482 public function validateSig( $text ) {
4483 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4484 }
4485
4486 /**
4487 * Clean up signature text
4488 *
4489 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4490 * 2) Substitute all transclusions
4491 *
4492 * @param string $text
4493 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4494 * @return string Signature text
4495 */
4496 public function cleanSig( $text, $parsing = false ) {
4497 if ( !$parsing ) {
4498 global $wgTitle;
4499 $magicScopeVariable = $this->lock();
4500 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4501 }
4502
4503 # Option to disable this feature
4504 if ( !$this->mOptions->getCleanSignatures() ) {
4505 return $text;
4506 }
4507
4508 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4509 # => Move this logic to braceSubstitution()
4510 $substWord = MagicWord::get( 'subst' );
4511 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4512 $substText = '{{' . $substWord->getSynonym( 0 );
4513
4514 $text = preg_replace( $substRegex, $substText, $text );
4515 $text = self::cleanSigInSig( $text );
4516 $dom = $this->preprocessToDom( $text );
4517 $frame = $this->getPreprocessor()->newFrame();
4518 $text = $frame->expand( $dom );
4519
4520 if ( !$parsing ) {
4521 $text = $this->mStripState->unstripBoth( $text );
4522 }
4523
4524 return $text;
4525 }
4526
4527 /**
4528 * Strip 3, 4 or 5 tildes out of signatures.
4529 *
4530 * @param string $text
4531 * @return string Signature text with /~{3,5}/ removed
4532 */
4533 public static function cleanSigInSig( $text ) {
4534 $text = preg_replace( '/~{3,5}/', '', $text );
4535 return $text;
4536 }
4537
4538 /**
4539 * Set up some variables which are usually set up in parse()
4540 * so that an external function can call some class members with confidence
4541 *
4542 * @param Title|null $title
4543 * @param ParserOptions $options
4544 * @param int $outputType
4545 * @param bool $clearState
4546 */
4547 public function startExternalParse( Title $title = null, ParserOptions $options,
4548 $outputType, $clearState = true
4549 ) {
4550 $this->startParse( $title, $options, $outputType, $clearState );
4551 }
4552
4553 /**
4554 * @param Title|null $title
4555 * @param ParserOptions $options
4556 * @param int $outputType
4557 * @param bool $clearState
4558 */
4559 private function startParse( Title $title = null, ParserOptions $options,
4560 $outputType, $clearState = true
4561 ) {
4562 $this->setTitle( $title );
4563 $this->mOptions = $options;
4564 $this->setOutputType( $outputType );
4565 if ( $clearState ) {
4566 $this->clearState();
4567 }
4568 }
4569
4570 /**
4571 * Wrapper for preprocess()
4572 *
4573 * @param string $text The text to preprocess
4574 * @param ParserOptions $options Options
4575 * @param Title|null $title Title object or null to use $wgTitle
4576 * @return string
4577 */
4578 public function transformMsg( $text, $options, $title = null ) {
4579 static $executing = false;
4580
4581 # Guard against infinite recursion
4582 if ( $executing ) {
4583 return $text;
4584 }
4585 $executing = true;
4586
4587 if ( !$title ) {
4588 global $wgTitle;
4589 $title = $wgTitle;
4590 }
4591
4592 $text = $this->preprocess( $text, $title, $options );
4593
4594 $executing = false;
4595 return $text;
4596 }
4597
4598 /**
4599 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4600 * The callback should have the following form:
4601 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4602 *
4603 * Transform and return $text. Use $parser for any required context, e.g. use
4604 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4605 *
4606 * Hooks may return extended information by returning an array, of which the
4607 * first numbered element (index 0) must be the return string, and all other
4608 * entries are extracted into local variables within an internal function
4609 * in the Parser class.
4610 *
4611 * This interface (introduced r61913) appears to be undocumented, but
4612 * 'markerType' is used by some core tag hooks to override which strip
4613 * array their results are placed in. **Use great caution if attempting
4614 * this interface, as it is not documented and injudicious use could smash
4615 * private variables.**
4616 *
4617 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4618 * @param callable $callback The callback function (and object) to use for the tag
4619 * @throws MWException
4620 * @return callable|null The old value of the mTagHooks array associated with the hook
4621 */
4622 public function setHook( $tag, $callback ) {
4623 $tag = strtolower( $tag );
4624 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4625 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4626 }
4627 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4628 $this->mTagHooks[$tag] = $callback;
4629 if ( !in_array( $tag, $this->mStripList ) ) {
4630 $this->mStripList[] = $tag;
4631 }
4632
4633 return $oldVal;
4634 }
4635
4636 /**
4637 * As setHook(), but letting the contents be parsed.
4638 *
4639 * Transparent tag hooks are like regular XML-style tag hooks, except they
4640 * operate late in the transformation sequence, on HTML instead of wikitext.
4641 *
4642 * This is probably obsoleted by things dealing with parser frames?
4643 * The only extension currently using it is geoserver.
4644 *
4645 * @since 1.10
4646 * @todo better document or deprecate this
4647 *
4648 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4649 * @param callable $callback The callback function (and object) to use for the tag
4650 * @throws MWException
4651 * @return callable|null The old value of the mTagHooks array associated with the hook
4652 */
4653 public function setTransparentTagHook( $tag, $callback ) {
4654 $tag = strtolower( $tag );
4655 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4656 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4657 }
4658 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4659 $this->mTransparentTagHooks[$tag] = $callback;
4660
4661 return $oldVal;
4662 }
4663
4664 /**
4665 * Remove all tag hooks
4666 */
4667 public function clearTagHooks() {
4668 $this->mTagHooks = [];
4669 $this->mFunctionTagHooks = [];
4670 $this->mStripList = $this->mDefaultStripList;
4671 }
4672
4673 /**
4674 * Create a function, e.g. {{sum:1|2|3}}
4675 * The callback function should have the form:
4676 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4677 *
4678 * Or with Parser::SFH_OBJECT_ARGS:
4679 * function myParserFunction( $parser, $frame, $args ) { ... }
4680 *
4681 * The callback may either return the text result of the function, or an array with the text
4682 * in element 0, and a number of flags in the other elements. The names of the flags are
4683 * specified in the keys. Valid flags are:
4684 * found The text returned is valid, stop processing the template. This
4685 * is on by default.
4686 * nowiki Wiki markup in the return value should be escaped
4687 * isHTML The returned text is HTML, armour it against wikitext transformation
4688 *
4689 * @param string $id The magic word ID
4690 * @param callable $callback The callback function (and object) to use
4691 * @param int $flags A combination of the following flags:
4692 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4693 *
4694 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4695 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4696 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4697 * the arguments, and to control the way they are expanded.
4698 *
4699 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4700 * arguments, for instance:
4701 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4702 *
4703 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4704 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4705 * working if/when this is changed.
4706 *
4707 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4708 * expansion.
4709 *
4710 * Please read the documentation in includes/parser/Preprocessor.php for more information
4711 * about the methods available in PPFrame and PPNode.
4712 *
4713 * @throws MWException
4714 * @return string|callable The old callback function for this name, if any
4715 */
4716 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4717 global $wgContLang;
4718
4719 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4720 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4721
4722 # Add to function cache
4723 $mw = MagicWord::get( $id );
4724 if ( !$mw ) {
4725 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4726 }
4727
4728 $synonyms = $mw->getSynonyms();
4729 $sensitive = intval( $mw->isCaseSensitive() );
4730
4731 foreach ( $synonyms as $syn ) {
4732 # Case
4733 if ( !$sensitive ) {
4734 $syn = $wgContLang->lc( $syn );
4735 }
4736 # Add leading hash
4737 if ( !( $flags & self::SFH_NO_HASH ) ) {
4738 $syn = '#' . $syn;
4739 }
4740 # Remove trailing colon
4741 if ( substr( $syn, -1, 1 ) === ':' ) {
4742 $syn = substr( $syn, 0, -1 );
4743 }
4744 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4745 }
4746 return $oldVal;
4747 }
4748
4749 /**
4750 * Get all registered function hook identifiers
4751 *
4752 * @return array
4753 */
4754 public function getFunctionHooks() {
4755 return array_keys( $this->mFunctionHooks );
4756 }
4757
4758 /**
4759 * Create a tag function, e.g. "<test>some stuff</test>".
4760 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4761 * Unlike parser functions, their content is not preprocessed.
4762 * @param string $tag
4763 * @param callable $callback
4764 * @param int $flags
4765 * @throws MWException
4766 * @return null
4767 */
4768 public function setFunctionTagHook( $tag, $callback, $flags ) {
4769 $tag = strtolower( $tag );
4770 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4771 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4772 }
4773 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4774 $this->mFunctionTagHooks[$tag] : null;
4775 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4776
4777 if ( !in_array( $tag, $this->mStripList ) ) {
4778 $this->mStripList[] = $tag;
4779 }
4780
4781 return $old;
4782 }
4783
4784 /**
4785 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4786 * Placeholders created in Linker::link()
4787 *
4788 * @param string $text
4789 * @param int $options
4790 */
4791 public function replaceLinkHolders( &$text, $options = 0 ) {
4792 $this->mLinkHolders->replace( $text );
4793 }
4794
4795 /**
4796 * Replace "<!--LINK-->" link placeholders with plain text of links
4797 * (not HTML-formatted).
4798 *
4799 * @param string $text
4800 * @return string
4801 */
4802 public function replaceLinkHoldersText( $text ) {
4803 return $this->mLinkHolders->replaceText( $text );
4804 }
4805
4806 /**
4807 * Renders an image gallery from a text with one line per image.
4808 * text labels may be given by using |-style alternative text. E.g.
4809 * Image:one.jpg|The number "1"
4810 * Image:tree.jpg|A tree
4811 * given as text will return the HTML of a gallery with two images,
4812 * labeled 'The number "1"' and
4813 * 'A tree'.
4814 *
4815 * @param string $text
4816 * @param array $params
4817 * @return string HTML
4818 */
4819 public function renderImageGallery( $text, $params ) {
4820
4821 $mode = false;
4822 if ( isset( $params['mode'] ) ) {
4823 $mode = $params['mode'];
4824 }
4825
4826 try {
4827 $ig = ImageGalleryBase::factory( $mode );
4828 } catch ( Exception $e ) {
4829 // If invalid type set, fallback to default.
4830 $ig = ImageGalleryBase::factory( false );
4831 }
4832
4833 $ig->setContextTitle( $this->mTitle );
4834 $ig->setShowBytes( false );
4835 $ig->setShowFilename( false );
4836 $ig->setParser( $this );
4837 $ig->setHideBadImages();
4838 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4839
4840 if ( isset( $params['showfilename'] ) ) {
4841 $ig->setShowFilename( true );
4842 } else {
4843 $ig->setShowFilename( false );
4844 }
4845 if ( isset( $params['caption'] ) ) {
4846 $caption = $params['caption'];
4847 $caption = htmlspecialchars( $caption );
4848 $caption = $this->replaceInternalLinks( $caption );
4849 $ig->setCaptionHtml( $caption );
4850 }
4851 if ( isset( $params['perrow'] ) ) {
4852 $ig->setPerRow( $params['perrow'] );
4853 }
4854 if ( isset( $params['widths'] ) ) {
4855 $ig->setWidths( $params['widths'] );
4856 }
4857 if ( isset( $params['heights'] ) ) {
4858 $ig->setHeights( $params['heights'] );
4859 }
4860 $ig->setAdditionalOptions( $params );
4861
4862 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4863
4864 $lines = StringUtils::explode( "\n", $text );
4865 foreach ( $lines as $line ) {
4866 # match lines like these:
4867 # Image:someimage.jpg|This is some image
4868 $matches = [];
4869 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4870 # Skip empty lines
4871 if ( count( $matches ) == 0 ) {
4872 continue;
4873 }
4874
4875 if ( strpos( $matches[0], '%' ) !== false ) {
4876 $matches[1] = rawurldecode( $matches[1] );
4877 }
4878 $title = Title::newFromText( $matches[1], NS_FILE );
4879 if ( is_null( $title ) ) {
4880 # Bogus title. Ignore these so we don't bomb out later.
4881 continue;
4882 }
4883
4884 # We need to get what handler the file uses, to figure out parameters.
4885 # Note, a hook can overide the file name, and chose an entirely different
4886 # file (which potentially could be of a different type and have different handler).
4887 $options = [];
4888 $descQuery = false;
4889 Hooks::run( 'BeforeParserFetchFileAndTitle',
4890 [ $this, $title, &$options, &$descQuery ] );
4891 # Don't register it now, as ImageGallery does that later.
4892 $file = $this->fetchFileNoRegister( $title, $options );
4893 $handler = $file ? $file->getHandler() : false;
4894
4895 $paramMap = [
4896 'img_alt' => 'gallery-internal-alt',
4897 'img_link' => 'gallery-internal-link',
4898 ];
4899 if ( $handler ) {
4900 $paramMap = $paramMap + $handler->getParamMap();
4901 // We don't want people to specify per-image widths.
4902 // Additionally the width parameter would need special casing anyhow.
4903 unset( $paramMap['img_width'] );
4904 }
4905
4906 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4907
4908 $label = '';
4909 $alt = '';
4910 $link = '';
4911 $handlerOptions = [];
4912 if ( isset( $matches[3] ) ) {
4913 // look for an |alt= definition while trying not to break existing
4914 // captions with multiple pipes (|) in it, until a more sensible grammar
4915 // is defined for images in galleries
4916
4917 // FIXME: Doing recursiveTagParse at this stage, and the trim before
4918 // splitting on '|' is a bit odd, and different from makeImage.
4919 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4920 $parameterMatches = StringUtils::explode( '|', $matches[3] );
4921
4922 foreach ( $parameterMatches as $parameterMatch ) {
4923 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
4924 if ( $magicName ) {
4925 $paramName = $paramMap[$magicName];
4926
4927 switch ( $paramName ) {
4928 case 'gallery-internal-alt':
4929 $alt = $this->stripAltText( $match, false );
4930 break;
4931 case 'gallery-internal-link':
4932 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
4933 $chars = self::EXT_LINK_URL_CLASS;
4934 $addr = self::EXT_LINK_ADDR;
4935 $prots = $this->mUrlProtocols;
4936 // check to see if link matches an absolute url, if not then it must be a wiki link.
4937 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
4938 $link = $linkValue;
4939 } else {
4940 $localLinkTitle = Title::newFromText( $linkValue );
4941 if ( $localLinkTitle !== null ) {
4942 $link = $localLinkTitle->getLinkURL();
4943 }
4944 }
4945 break;
4946 default:
4947 // Must be a handler specific parameter.
4948 if ( $handler->validateParam( $paramName, $match ) ) {
4949 $handlerOptions[$paramName] = $match;
4950 } else {
4951 // Guess not, consider it as caption.
4952 wfDebug( "$parameterMatch failed parameter validation\n" );
4953 $label = '|' . $parameterMatch;
4954 }
4955 }
4956
4957 } else {
4958 // Last pipe wins.
4959 $label = '|' . $parameterMatch;
4960 }
4961 }
4962 // Remove the pipe.
4963 $label = substr( $label, 1 );
4964 }
4965
4966 $ig->add( $title, $label, $alt, $link, $handlerOptions );
4967 }
4968 $html = $ig->toHTML();
4969 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
4970 return $html;
4971 }
4972
4973 /**
4974 * @param MediaHandler $handler
4975 * @return array
4976 */
4977 public function getImageParams( $handler ) {
4978 if ( $handler ) {
4979 $handlerClass = get_class( $handler );
4980 } else {
4981 $handlerClass = '';
4982 }
4983 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
4984 # Initialise static lists
4985 static $internalParamNames = [
4986 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
4987 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4988 'bottom', 'text-bottom' ],
4989 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
4990 'upright', 'border', 'link', 'alt', 'class' ],
4991 ];
4992 static $internalParamMap;
4993 if ( !$internalParamMap ) {
4994 $internalParamMap = [];
4995 foreach ( $internalParamNames as $type => $names ) {
4996 foreach ( $names as $name ) {
4997 $magicName = str_replace( '-', '_', "img_$name" );
4998 $internalParamMap[$magicName] = [ $type, $name ];
4999 }
5000 }
5001 }
5002
5003 # Add handler params
5004 $paramMap = $internalParamMap;
5005 if ( $handler ) {
5006 $handlerParamMap = $handler->getParamMap();
5007 foreach ( $handlerParamMap as $magic => $paramName ) {
5008 $paramMap[$magic] = [ 'handler', $paramName ];
5009 }
5010 }
5011 $this->mImageParams[$handlerClass] = $paramMap;
5012 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5013 }
5014 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5015 }
5016
5017 /**
5018 * Parse image options text and use it to make an image
5019 *
5020 * @param Title $title
5021 * @param string $options
5022 * @param LinkHolderArray|bool $holders
5023 * @return string HTML
5024 */
5025 public function makeImage( $title, $options, $holders = false ) {
5026 # Check if the options text is of the form "options|alt text"
5027 # Options are:
5028 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5029 # * left no resizing, just left align. label is used for alt= only
5030 # * right same, but right aligned
5031 # * none same, but not aligned
5032 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5033 # * center center the image
5034 # * frame Keep original image size, no magnify-button.
5035 # * framed Same as "frame"
5036 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5037 # * upright reduce width for upright images, rounded to full __0 px
5038 # * border draw a 1px border around the image
5039 # * alt Text for HTML alt attribute (defaults to empty)
5040 # * class Set a class for img node
5041 # * link Set the target of the image link. Can be external, interwiki, or local
5042 # vertical-align values (no % or length right now):
5043 # * baseline
5044 # * sub
5045 # * super
5046 # * top
5047 # * text-top
5048 # * middle
5049 # * bottom
5050 # * text-bottom
5051
5052 $parts = StringUtils::explode( "|", $options );
5053
5054 # Give extensions a chance to select the file revision for us
5055 $options = [];
5056 $descQuery = false;
5057 Hooks::run( 'BeforeParserFetchFileAndTitle',
5058 [ $this, $title, &$options, &$descQuery ] );
5059 # Fetch and register the file (file title may be different via hooks)
5060 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5061
5062 # Get parameter map
5063 $handler = $file ? $file->getHandler() : false;
5064
5065 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5066
5067 if ( !$file ) {
5068 $this->addTrackingCategory( 'broken-file-category' );
5069 }
5070
5071 # Process the input parameters
5072 $caption = '';
5073 $params = [ 'frame' => [], 'handler' => [],
5074 'horizAlign' => [], 'vertAlign' => [] ];
5075 $seenformat = false;
5076 foreach ( $parts as $part ) {
5077 $part = trim( $part );
5078 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5079 $validated = false;
5080 if ( isset( $paramMap[$magicName] ) ) {
5081 list( $type, $paramName ) = $paramMap[$magicName];
5082
5083 # Special case; width and height come in one variable together
5084 if ( $type === 'handler' && $paramName === 'width' ) {
5085 $parsedWidthParam = $this->parseWidthParam( $value );
5086 if ( isset( $parsedWidthParam['width'] ) ) {
5087 $width = $parsedWidthParam['width'];
5088 if ( $handler->validateParam( 'width', $width ) ) {
5089 $params[$type]['width'] = $width;
5090 $validated = true;
5091 }
5092 }
5093 if ( isset( $parsedWidthParam['height'] ) ) {
5094 $height = $parsedWidthParam['height'];
5095 if ( $handler->validateParam( 'height', $height ) ) {
5096 $params[$type]['height'] = $height;
5097 $validated = true;
5098 }
5099 }
5100 # else no validation -- bug 13436
5101 } else {
5102 if ( $type === 'handler' ) {
5103 # Validate handler parameter
5104 $validated = $handler->validateParam( $paramName, $value );
5105 } else {
5106 # Validate internal parameters
5107 switch ( $paramName ) {
5108 case 'manualthumb':
5109 case 'alt':
5110 case 'class':
5111 # @todo FIXME: Possibly check validity here for
5112 # manualthumb? downstream behavior seems odd with
5113 # missing manual thumbs.
5114 $validated = true;
5115 $value = $this->stripAltText( $value, $holders );
5116 break;
5117 case 'link':
5118 $chars = self::EXT_LINK_URL_CLASS;
5119 $addr = self::EXT_LINK_ADDR;
5120 $prots = $this->mUrlProtocols;
5121 if ( $value === '' ) {
5122 $paramName = 'no-link';
5123 $value = true;
5124 $validated = true;
5125 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5126 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5127 $paramName = 'link-url';
5128 $this->mOutput->addExternalLink( $value );
5129 if ( $this->mOptions->getExternalLinkTarget() ) {
5130 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5131 }
5132 $validated = true;
5133 }
5134 } else {
5135 $linkTitle = Title::newFromText( $value );
5136 if ( $linkTitle ) {
5137 $paramName = 'link-title';
5138 $value = $linkTitle;
5139 $this->mOutput->addLink( $linkTitle );
5140 $validated = true;
5141 }
5142 }
5143 break;
5144 case 'frameless':
5145 case 'framed':
5146 case 'thumbnail':
5147 // use first appearing option, discard others.
5148 $validated = ! $seenformat;
5149 $seenformat = true;
5150 break;
5151 default:
5152 # Most other things appear to be empty or numeric...
5153 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5154 }
5155 }
5156
5157 if ( $validated ) {
5158 $params[$type][$paramName] = $value;
5159 }
5160 }
5161 }
5162 if ( !$validated ) {
5163 $caption = $part;
5164 }
5165 }
5166
5167 # Process alignment parameters
5168 if ( $params['horizAlign'] ) {
5169 $params['frame']['align'] = key( $params['horizAlign'] );
5170 }
5171 if ( $params['vertAlign'] ) {
5172 $params['frame']['valign'] = key( $params['vertAlign'] );
5173 }
5174
5175 $params['frame']['caption'] = $caption;
5176
5177 # Will the image be presented in a frame, with the caption below?
5178 $imageIsFramed = isset( $params['frame']['frame'] )
5179 || isset( $params['frame']['framed'] )
5180 || isset( $params['frame']['thumbnail'] )
5181 || isset( $params['frame']['manualthumb'] );
5182
5183 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5184 # came to also set the caption, ordinary text after the image -- which
5185 # makes no sense, because that just repeats the text multiple times in
5186 # screen readers. It *also* came to set the title attribute.
5187 # Now that we have an alt attribute, we should not set the alt text to
5188 # equal the caption: that's worse than useless, it just repeats the
5189 # text. This is the framed/thumbnail case. If there's no caption, we
5190 # use the unnamed parameter for alt text as well, just for the time be-
5191 # ing, if the unnamed param is set and the alt param is not.
5192 # For the future, we need to figure out if we want to tweak this more,
5193 # e.g., introducing a title= parameter for the title; ignoring the un-
5194 # named parameter entirely for images without a caption; adding an ex-
5195 # plicit caption= parameter and preserving the old magic unnamed para-
5196 # meter for BC; ...
5197 if ( $imageIsFramed ) { # Framed image
5198 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5199 # No caption or alt text, add the filename as the alt text so
5200 # that screen readers at least get some description of the image
5201 $params['frame']['alt'] = $title->getText();
5202 }
5203 # Do not set $params['frame']['title'] because tooltips don't make sense
5204 # for framed images
5205 } else { # Inline image
5206 if ( !isset( $params['frame']['alt'] ) ) {
5207 # No alt text, use the "caption" for the alt text
5208 if ( $caption !== '' ) {
5209 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5210 } else {
5211 # No caption, fall back to using the filename for the
5212 # alt text
5213 $params['frame']['alt'] = $title->getText();
5214 }
5215 }
5216 # Use the "caption" for the tooltip text
5217 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5218 }
5219
5220 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5221
5222 # Linker does the rest
5223 $time = isset( $options['time'] ) ? $options['time'] : false;
5224 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5225 $time, $descQuery, $this->mOptions->getThumbSize() );
5226
5227 # Give the handler a chance to modify the parser object
5228 if ( $handler ) {
5229 $handler->parserTransformHook( $this, $file );
5230 }
5231
5232 return $ret;
5233 }
5234
5235 /**
5236 * @param string $caption
5237 * @param LinkHolderArray|bool $holders
5238 * @return mixed|string
5239 */
5240 protected function stripAltText( $caption, $holders ) {
5241 # Strip bad stuff out of the title (tooltip). We can't just use
5242 # replaceLinkHoldersText() here, because if this function is called
5243 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5244 if ( $holders ) {
5245 $tooltip = $holders->replaceText( $caption );
5246 } else {
5247 $tooltip = $this->replaceLinkHoldersText( $caption );
5248 }
5249
5250 # make sure there are no placeholders in thumbnail attributes
5251 # that are later expanded to html- so expand them now and
5252 # remove the tags
5253 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5254 $tooltip = Sanitizer::stripAllTags( $tooltip );
5255
5256 return $tooltip;
5257 }
5258
5259 /**
5260 * Set a flag in the output object indicating that the content is dynamic and
5261 * shouldn't be cached.
5262 */
5263 public function disableCache() {
5264 wfDebug( "Parser output marked as uncacheable.\n" );
5265 if ( !$this->mOutput ) {
5266 throw new MWException( __METHOD__ .
5267 " can only be called when actually parsing something" );
5268 }
5269 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5270 }
5271
5272 /**
5273 * Callback from the Sanitizer for expanding items found in HTML attribute
5274 * values, so they can be safely tested and escaped.
5275 *
5276 * @param string $text
5277 * @param bool|PPFrame $frame
5278 * @return string
5279 */
5280 public function attributeStripCallback( &$text, $frame = false ) {
5281 $text = $this->replaceVariables( $text, $frame );
5282 $text = $this->mStripState->unstripBoth( $text );
5283 return $text;
5284 }
5285
5286 /**
5287 * Accessor
5288 *
5289 * @return array
5290 */
5291 public function getTags() {
5292 return array_merge(
5293 array_keys( $this->mTransparentTagHooks ),
5294 array_keys( $this->mTagHooks ),
5295 array_keys( $this->mFunctionTagHooks )
5296 );
5297 }
5298
5299 /**
5300 * Replace transparent tags in $text with the values given by the callbacks.
5301 *
5302 * Transparent tag hooks are like regular XML-style tag hooks, except they
5303 * operate late in the transformation sequence, on HTML instead of wikitext.
5304 *
5305 * @param string $text
5306 *
5307 * @return string
5308 */
5309 public function replaceTransparentTags( $text ) {
5310 $matches = [];
5311 $elements = array_keys( $this->mTransparentTagHooks );
5312 $text = self::extractTagsAndParams( $elements, $text, $matches );
5313 $replacements = [];
5314
5315 foreach ( $matches as $marker => $data ) {
5316 list( $element, $content, $params, $tag ) = $data;
5317 $tagName = strtolower( $element );
5318 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5319 $output = call_user_func_array(
5320 $this->mTransparentTagHooks[$tagName],
5321 [ $content, $params, $this ]
5322 );
5323 } else {
5324 $output = $tag;
5325 }
5326 $replacements[$marker] = $output;
5327 }
5328 return strtr( $text, $replacements );
5329 }
5330
5331 /**
5332 * Break wikitext input into sections, and either pull or replace
5333 * some particular section's text.
5334 *
5335 * External callers should use the getSection and replaceSection methods.
5336 *
5337 * @param string $text Page wikitext
5338 * @param string|number $sectionId A section identifier string of the form:
5339 * "<flag1> - <flag2> - ... - <section number>"
5340 *
5341 * Currently the only recognised flag is "T", which means the target section number
5342 * was derived during a template inclusion parse, in other words this is a template
5343 * section edit link. If no flags are given, it was an ordinary section edit link.
5344 * This flag is required to avoid a section numbering mismatch when a section is
5345 * enclosed by "<includeonly>" (bug 6563).
5346 *
5347 * The section number 0 pulls the text before the first heading; other numbers will
5348 * pull the given section along with its lower-level subsections. If the section is
5349 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5350 *
5351 * Section 0 is always considered to exist, even if it only contains the empty
5352 * string. If $text is the empty string and section 0 is replaced, $newText is
5353 * returned.
5354 *
5355 * @param string $mode One of "get" or "replace"
5356 * @param string $newText Replacement text for section data.
5357 * @return string For "get", the extracted section text.
5358 * for "replace", the whole page with the section replaced.
5359 */
5360 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5361 global $wgTitle; # not generally used but removes an ugly failure mode
5362
5363 $magicScopeVariable = $this->lock();
5364 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5365 $outText = '';
5366 $frame = $this->getPreprocessor()->newFrame();
5367
5368 # Process section extraction flags
5369 $flags = 0;
5370 $sectionParts = explode( '-', $sectionId );
5371 $sectionIndex = array_pop( $sectionParts );
5372 foreach ( $sectionParts as $part ) {
5373 if ( $part === 'T' ) {
5374 $flags |= self::PTD_FOR_INCLUSION;
5375 }
5376 }
5377
5378 # Check for empty input
5379 if ( strval( $text ) === '' ) {
5380 # Only sections 0 and T-0 exist in an empty document
5381 if ( $sectionIndex == 0 ) {
5382 if ( $mode === 'get' ) {
5383 return '';
5384 } else {
5385 return $newText;
5386 }
5387 } else {
5388 if ( $mode === 'get' ) {
5389 return $newText;
5390 } else {
5391 return $text;
5392 }
5393 }
5394 }
5395
5396 # Preprocess the text
5397 $root = $this->preprocessToDom( $text, $flags );
5398
5399 # <h> nodes indicate section breaks
5400 # They can only occur at the top level, so we can find them by iterating the root's children
5401 $node = $root->getFirstChild();
5402
5403 # Find the target section
5404 if ( $sectionIndex == 0 ) {
5405 # Section zero doesn't nest, level=big
5406 $targetLevel = 1000;
5407 } else {
5408 while ( $node ) {
5409 if ( $node->getName() === 'h' ) {
5410 $bits = $node->splitHeading();
5411 if ( $bits['i'] == $sectionIndex ) {
5412 $targetLevel = $bits['level'];
5413 break;
5414 }
5415 }
5416 if ( $mode === 'replace' ) {
5417 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5418 }
5419 $node = $node->getNextSibling();
5420 }
5421 }
5422
5423 if ( !$node ) {
5424 # Not found
5425 if ( $mode === 'get' ) {
5426 return $newText;
5427 } else {
5428 return $text;
5429 }
5430 }
5431
5432 # Find the end of the section, including nested sections
5433 do {
5434 if ( $node->getName() === 'h' ) {
5435 $bits = $node->splitHeading();
5436 $curLevel = $bits['level'];
5437 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5438 break;
5439 }
5440 }
5441 if ( $mode === 'get' ) {
5442 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5443 }
5444 $node = $node->getNextSibling();
5445 } while ( $node );
5446
5447 # Write out the remainder (in replace mode only)
5448 if ( $mode === 'replace' ) {
5449 # Output the replacement text
5450 # Add two newlines on -- trailing whitespace in $newText is conventionally
5451 # stripped by the editor, so we need both newlines to restore the paragraph gap
5452 # Only add trailing whitespace if there is newText
5453 if ( $newText != "" ) {
5454 $outText .= $newText . "\n\n";
5455 }
5456
5457 while ( $node ) {
5458 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5459 $node = $node->getNextSibling();
5460 }
5461 }
5462
5463 if ( is_string( $outText ) ) {
5464 # Re-insert stripped tags
5465 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5466 }
5467
5468 return $outText;
5469 }
5470
5471 /**
5472 * This function returns the text of a section, specified by a number ($section).
5473 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5474 * the first section before any such heading (section 0).
5475 *
5476 * If a section contains subsections, these are also returned.
5477 *
5478 * @param string $text Text to look in
5479 * @param string|number $sectionId Section identifier as a number or string
5480 * (e.g. 0, 1 or 'T-1').
5481 * @param string $defaultText Default to return if section is not found
5482 *
5483 * @return string Text of the requested section
5484 */
5485 public function getSection( $text, $sectionId, $defaultText = '' ) {
5486 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5487 }
5488
5489 /**
5490 * This function returns $oldtext after the content of the section
5491 * specified by $section has been replaced with $text. If the target
5492 * section does not exist, $oldtext is returned unchanged.
5493 *
5494 * @param string $oldText Former text of the article
5495 * @param string|number $sectionId Section identifier as a number or string
5496 * (e.g. 0, 1 or 'T-1').
5497 * @param string $newText Replacing text
5498 *
5499 * @return string Modified text
5500 */
5501 public function replaceSection( $oldText, $sectionId, $newText ) {
5502 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5503 }
5504
5505 /**
5506 * Get the ID of the revision we are parsing
5507 *
5508 * @return int|null
5509 */
5510 public function getRevisionId() {
5511 return $this->mRevisionId;
5512 }
5513
5514 /**
5515 * Get the revision object for $this->mRevisionId
5516 *
5517 * @return Revision|null Either a Revision object or null
5518 * @since 1.23 (public since 1.23)
5519 */
5520 public function getRevisionObject() {
5521 if ( !is_null( $this->mRevisionObject ) ) {
5522 return $this->mRevisionObject;
5523 }
5524 if ( is_null( $this->mRevisionId ) ) {
5525 return null;
5526 }
5527
5528 $rev = call_user_func(
5529 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5530 );
5531
5532 # If the parse is for a new revision, then the callback should have
5533 # already been set to force the object and should match mRevisionId.
5534 # If not, try to fetch by mRevisionId for sanity.
5535 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5536 $rev = Revision::newFromId( $this->mRevisionId );
5537 }
5538
5539 $this->mRevisionObject = $rev;
5540
5541 return $this->mRevisionObject;
5542 }
5543
5544 /**
5545 * Get the timestamp associated with the current revision, adjusted for
5546 * the default server-local timestamp
5547 * @return string
5548 */
5549 public function getRevisionTimestamp() {
5550 if ( is_null( $this->mRevisionTimestamp ) ) {
5551 global $wgContLang;
5552
5553 $revObject = $this->getRevisionObject();
5554 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5555
5556 # The cryptic '' timezone parameter tells to use the site-default
5557 # timezone offset instead of the user settings.
5558 # Since this value will be saved into the parser cache, served
5559 # to other users, and potentially even used inside links and such,
5560 # it needs to be consistent for all visitors.
5561 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5562
5563 }
5564 return $this->mRevisionTimestamp;
5565 }
5566
5567 /**
5568 * Get the name of the user that edited the last revision
5569 *
5570 * @return string User name
5571 */
5572 public function getRevisionUser() {
5573 if ( is_null( $this->mRevisionUser ) ) {
5574 $revObject = $this->getRevisionObject();
5575
5576 # if this template is subst: the revision id will be blank,
5577 # so just use the current user's name
5578 if ( $revObject ) {
5579 $this->mRevisionUser = $revObject->getUserText();
5580 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5581 $this->mRevisionUser = $this->getUser()->getName();
5582 }
5583 }
5584 return $this->mRevisionUser;
5585 }
5586
5587 /**
5588 * Get the size of the revision
5589 *
5590 * @return int|null Revision size
5591 */
5592 public function getRevisionSize() {
5593 if ( is_null( $this->mRevisionSize ) ) {
5594 $revObject = $this->getRevisionObject();
5595
5596 # if this variable is subst: the revision id will be blank,
5597 # so just use the parser input size, because the own substituation
5598 # will change the size.
5599 if ( $revObject ) {
5600 $this->mRevisionSize = $revObject->getSize();
5601 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5602 $this->mRevisionSize = $this->mInputSize;
5603 }
5604 }
5605 return $this->mRevisionSize;
5606 }
5607
5608 /**
5609 * Mutator for $mDefaultSort
5610 *
5611 * @param string $sort New value
5612 */
5613 public function setDefaultSort( $sort ) {
5614 $this->mDefaultSort = $sort;
5615 $this->mOutput->setProperty( 'defaultsort', $sort );
5616 }
5617
5618 /**
5619 * Accessor for $mDefaultSort
5620 * Will use the empty string if none is set.
5621 *
5622 * This value is treated as a prefix, so the
5623 * empty string is equivalent to sorting by
5624 * page name.
5625 *
5626 * @return string
5627 */
5628 public function getDefaultSort() {
5629 if ( $this->mDefaultSort !== false ) {
5630 return $this->mDefaultSort;
5631 } else {
5632 return '';
5633 }
5634 }
5635
5636 /**
5637 * Accessor for $mDefaultSort
5638 * Unlike getDefaultSort(), will return false if none is set
5639 *
5640 * @return string|bool
5641 */
5642 public function getCustomDefaultSort() {
5643 return $this->mDefaultSort;
5644 }
5645
5646 /**
5647 * Try to guess the section anchor name based on a wikitext fragment
5648 * presumably extracted from a heading, for example "Header" from
5649 * "== Header ==".
5650 *
5651 * @param string $text
5652 *
5653 * @return string
5654 */
5655 public function guessSectionNameFromWikiText( $text ) {
5656 # Strip out wikitext links(they break the anchor)
5657 $text = $this->stripSectionName( $text );
5658 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5659 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5660 }
5661
5662 /**
5663 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5664 * instead. For use in redirects, since IE6 interprets Redirect: headers
5665 * as something other than UTF-8 (apparently?), resulting in breakage.
5666 *
5667 * @param string $text The section name
5668 * @return string An anchor
5669 */
5670 public function guessLegacySectionNameFromWikiText( $text ) {
5671 # Strip out wikitext links(they break the anchor)
5672 $text = $this->stripSectionName( $text );
5673 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5674 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5675 }
5676
5677 /**
5678 * Strips a text string of wikitext for use in a section anchor
5679 *
5680 * Accepts a text string and then removes all wikitext from the
5681 * string and leaves only the resultant text (i.e. the result of
5682 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5683 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5684 * to create valid section anchors by mimicing the output of the
5685 * parser when headings are parsed.
5686 *
5687 * @param string $text Text string to be stripped of wikitext
5688 * for use in a Section anchor
5689 * @return string Filtered text string
5690 */
5691 public function stripSectionName( $text ) {
5692 # Strip internal link markup
5693 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5694 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5695
5696 # Strip external link markup
5697 # @todo FIXME: Not tolerant to blank link text
5698 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5699 # on how many empty links there are on the page - need to figure that out.
5700 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5701
5702 # Parse wikitext quotes (italics & bold)
5703 $text = $this->doQuotes( $text );
5704
5705 # Strip HTML tags
5706 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5707 return $text;
5708 }
5709
5710 /**
5711 * strip/replaceVariables/unstrip for preprocessor regression testing
5712 *
5713 * @param string $text
5714 * @param Title $title
5715 * @param ParserOptions $options
5716 * @param int $outputType
5717 *
5718 * @return string
5719 */
5720 public function testSrvus( $text, Title $title, ParserOptions $options,
5721 $outputType = self::OT_HTML
5722 ) {
5723 $magicScopeVariable = $this->lock();
5724 $this->startParse( $title, $options, $outputType, true );
5725
5726 $text = $this->replaceVariables( $text );
5727 $text = $this->mStripState->unstripBoth( $text );
5728 $text = Sanitizer::removeHTMLtags( $text );
5729 return $text;
5730 }
5731
5732 /**
5733 * @param string $text
5734 * @param Title $title
5735 * @param ParserOptions $options
5736 * @return string
5737 */
5738 public function testPst( $text, Title $title, ParserOptions $options ) {
5739 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5740 }
5741
5742 /**
5743 * @param string $text
5744 * @param Title $title
5745 * @param ParserOptions $options
5746 * @return string
5747 */
5748 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5749 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5750 }
5751
5752 /**
5753 * Call a callback function on all regions of the given text that are not
5754 * inside strip markers, and replace those regions with the return value
5755 * of the callback. For example, with input:
5756 *
5757 * aaa<MARKER>bbb
5758 *
5759 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5760 * two strings will be replaced with the value returned by the callback in
5761 * each case.
5762 *
5763 * @param string $s
5764 * @param callable $callback
5765 *
5766 * @return string
5767 */
5768 public function markerSkipCallback( $s, $callback ) {
5769 $i = 0;
5770 $out = '';
5771 while ( $i < strlen( $s ) ) {
5772 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5773 if ( $markerStart === false ) {
5774 $out .= call_user_func( $callback, substr( $s, $i ) );
5775 break;
5776 } else {
5777 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5778 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5779 if ( $markerEnd === false ) {
5780 $out .= substr( $s, $markerStart );
5781 break;
5782 } else {
5783 $markerEnd += strlen( self::MARKER_SUFFIX );
5784 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5785 $i = $markerEnd;
5786 }
5787 }
5788 }
5789 return $out;
5790 }
5791
5792 /**
5793 * Remove any strip markers found in the given text.
5794 *
5795 * @param string $text Input string
5796 * @return string
5797 */
5798 public function killMarkers( $text ) {
5799 return $this->mStripState->killMarkers( $text );
5800 }
5801
5802 /**
5803 * Save the parser state required to convert the given half-parsed text to
5804 * HTML. "Half-parsed" in this context means the output of
5805 * recursiveTagParse() or internalParse(). This output has strip markers
5806 * from replaceVariables (extensionSubstitution() etc.), and link
5807 * placeholders from replaceLinkHolders().
5808 *
5809 * Returns an array which can be serialized and stored persistently. This
5810 * array can later be loaded into another parser instance with
5811 * unserializeHalfParsedText(). The text can then be safely incorporated into
5812 * the return value of a parser hook.
5813 *
5814 * @param string $text
5815 *
5816 * @return array
5817 */
5818 public function serializeHalfParsedText( $text ) {
5819 $data = [
5820 'text' => $text,
5821 'version' => self::HALF_PARSED_VERSION,
5822 'stripState' => $this->mStripState->getSubState( $text ),
5823 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5824 ];
5825 return $data;
5826 }
5827
5828 /**
5829 * Load the parser state given in the $data array, which is assumed to
5830 * have been generated by serializeHalfParsedText(). The text contents is
5831 * extracted from the array, and its markers are transformed into markers
5832 * appropriate for the current Parser instance. This transformed text is
5833 * returned, and can be safely included in the return value of a parser
5834 * hook.
5835 *
5836 * If the $data array has been stored persistently, the caller should first
5837 * check whether it is still valid, by calling isValidHalfParsedText().
5838 *
5839 * @param array $data Serialized data
5840 * @throws MWException
5841 * @return string
5842 */
5843 public function unserializeHalfParsedText( $data ) {
5844 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5845 throw new MWException( __METHOD__ . ': invalid version' );
5846 }
5847
5848 # First, extract the strip state.
5849 $texts = [ $data['text'] ];
5850 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5851
5852 # Now renumber links
5853 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5854
5855 # Should be good to go.
5856 return $texts[0];
5857 }
5858
5859 /**
5860 * Returns true if the given array, presumed to be generated by
5861 * serializeHalfParsedText(), is compatible with the current version of the
5862 * parser.
5863 *
5864 * @param array $data
5865 *
5866 * @return bool
5867 */
5868 public function isValidHalfParsedText( $data ) {
5869 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5870 }
5871
5872 /**
5873 * Parsed a width param of imagelink like 300px or 200x300px
5874 *
5875 * @param string $value
5876 *
5877 * @return array
5878 * @since 1.20
5879 */
5880 public function parseWidthParam( $value ) {
5881 $parsedWidthParam = [];
5882 if ( $value === '' ) {
5883 return $parsedWidthParam;
5884 }
5885 $m = [];
5886 # (bug 13500) In both cases (width/height and width only),
5887 # permit trailing "px" for backward compatibility.
5888 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5889 $width = intval( $m[1] );
5890 $height = intval( $m[2] );
5891 $parsedWidthParam['width'] = $width;
5892 $parsedWidthParam['height'] = $height;
5893 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5894 $width = intval( $value );
5895 $parsedWidthParam['width'] = $width;
5896 }
5897 return $parsedWidthParam;
5898 }
5899
5900 /**
5901 * Lock the current instance of the parser.
5902 *
5903 * This is meant to stop someone from calling the parser
5904 * recursively and messing up all the strip state.
5905 *
5906 * @throws MWException If parser is in a parse
5907 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5908 */
5909 protected function lock() {
5910 if ( $this->mInParse ) {
5911 throw new MWException( "Parser state cleared while parsing. "
5912 . "Did you call Parser::parse recursively?" );
5913 }
5914 $this->mInParse = true;
5915
5916 $recursiveCheck = new ScopedCallback( function() {
5917 $this->mInParse = false;
5918 } );
5919
5920 return $recursiveCheck;
5921 }
5922
5923 /**
5924 * Strip outer <p></p> tag from the HTML source of a single paragraph.
5925 *
5926 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
5927 * or if there is more than one <p/> tag in the input HTML.
5928 *
5929 * @param string $html
5930 * @return string
5931 * @since 1.24
5932 */
5933 public static function stripOuterParagraph( $html ) {
5934 $m = [];
5935 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
5936 if ( strpos( $m[1], '</p>' ) === false ) {
5937 $html = $m[1];
5938 }
5939 }
5940
5941 return $html;
5942 }
5943
5944 /**
5945 * Return this parser if it is not doing anything, otherwise
5946 * get a fresh parser. You can use this method by doing
5947 * $myParser = $wgParser->getFreshParser(), or more simply
5948 * $wgParser->getFreshParser()->parse( ... );
5949 * if you're unsure if $wgParser is safe to use.
5950 *
5951 * @since 1.24
5952 * @return Parser A parser object that is not parsing anything
5953 */
5954 public function getFreshParser() {
5955 global $wgParserConf;
5956 if ( $this->mInParse ) {
5957 return new $wgParserConf['class']( $wgParserConf );
5958 } else {
5959 return $this;
5960 }
5961 }
5962
5963 /**
5964 * Set's up the PHP implementation of OOUI for use in this request
5965 * and instructs OutputPage to enable OOUI for itself.
5966 *
5967 * @since 1.26
5968 */
5969 public function enableOOUI() {
5970 OutputPage::setupOOUI();
5971 $this->mOutput->setEnableOOUI( true );
5972 }
5973 }
Wiklou, le Wiki du Wiklou