dépôts / lhc / web / wiklou.git / blob
? search:


3f703e3cb554b987d324e5162fbed6d15f864173
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23 use MediaWiki\Linker\LinkRenderer;
24 use MediaWiki\MediaWikiServices;
25 use Wikimedia\ScopedCallback;
26
27 /**
28 * @defgroup Parser Parser
29 */
30
31 /**
32 * PHP Parser - Processes wiki markup (which uses a more user-friendly
33 * syntax, such as "[[link]]" for making links), and provides a one-way
34 * transformation of that wiki markup it into (X)HTML output / markup
35 * (which in turn the browser understands, and can display).
36 *
37 * There are seven main entry points into the Parser class:
38 *
39 * - Parser::parse()
40 * produces HTML output
41 * - Parser::preSaveTransform()
42 * produces altered wiki markup
43 * - Parser::preprocess()
44 * removes HTML comments and expands templates
45 * - Parser::cleanSig() and Parser::cleanSigInSig()
46 * cleans a signature before saving it to preferences
47 * - Parser::getSection()
48 * return the content of a section from an article for section editing
49 * - Parser::replaceSection()
50 * replaces a section by number inside an article
51 * - Parser::getPreloadText()
52 * removes <noinclude> sections and <includeonly> tags
53 *
54 * Globals used:
55 * object: $wgContLang
56 *
57 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
58 *
59 * @par Settings:
60 * $wgNamespacesWithSubpages
61 *
62 * @par Settings only within ParserOptions:
63 * $wgAllowExternalImages
64 * $wgAllowSpecialInclusion
65 * $wgInterwikiMagic
66 * $wgMaxArticleSize
67 *
68 * @ingroup Parser
69 */
70 class Parser {
71 /**
72 * Update this version number when the ParserOutput format
73 * changes in an incompatible way, so the parser cache
74 * can automatically discard old data.
75 */
76 const VERSION = '1.6.4';
77
78 /**
79 * Update this version number when the output of serialiseHalfParsedText()
80 * changes in an incompatible way
81 */
82 const HALF_PARSED_VERSION = 2;
83
84 # Flags for Parser::setFunctionHook
85 const SFH_NO_HASH = 1;
86 const SFH_OBJECT_ARGS = 2;
87
88 # Constants needed for external link processing
89 # Everything except bracket, space, or control characters
90 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
91 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
92 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
93 # Simplified expression to match an IPv4 or IPv6 address, or
94 # at least one character of a host name (embeds EXT_LINK_URL_CLASS)
95 const EXT_LINK_ADDR = '(?:[0-9.]+|\\[(?i:[0-9a-f:.]+)\\]|[^][<>"\\x00-\\x20\\x7F\p{Zs}])';
96 # RegExp to make image URLs (embeds IPv6 part of EXT_LINK_ADDR)
97 // @codingStandardsIgnoreStart Generic.Files.LineLength
98 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)((?:\\[(?i:[0-9a-f:.]+)\\])?[^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
99 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
100 // @codingStandardsIgnoreEnd
101
102 # Regular expression for a non-newline space
103 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
104
105 # Flags for preprocessToDom
106 const PTD_FOR_INCLUSION = 1;
107
108 # Allowed values for $this->mOutputType
109 # Parameter to startExternalParse().
110 const OT_HTML = 1; # like parse()
111 const OT_WIKI = 2; # like preSaveTransform()
112 const OT_PREPROCESS = 3; # like preprocess()
113 const OT_MSG = 3;
114 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
115
116 /**
117 * @var string Prefix and suffix for temporary replacement strings
118 * for the multipass parser.
119 *
120 * \x7f should never appear in input as it's disallowed in XML.
121 * Using it at the front also gives us a little extra robustness
122 * since it shouldn't match when butted up against identifier-like
123 * string constructs.
124 *
125 * Must not consist of all title characters, or else it will change
126 * the behavior of <nowiki> in a link.
127 *
128 * Must have a character that needs escaping in attributes, otherwise
129 * someone could put a strip marker in an attribute, to get around
130 * escaping quote marks, and break out of the attribute. Thus we add
131 * `'".
132 */
133 const MARKER_SUFFIX = "-QINU`\"'\x7f";
134 const MARKER_PREFIX = "\x7f'\"`UNIQ-";
135
136 # Markers used for wrapping the table of contents
137 const TOC_START = '<mw:toc>';
138 const TOC_END = '</mw:toc>';
139
140 # Persistent:
141 public $mTagHooks = [];
142 public $mTransparentTagHooks = [];
143 public $mFunctionHooks = [];
144 public $mFunctionSynonyms = [ 0 => [], 1 => [] ];
145 public $mFunctionTagHooks = [];
146 public $mStripList = [];
147 public $mDefaultStripList = [];
148 public $mVarCache = [];
149 public $mImageParams = [];
150 public $mImageParamsMagicArray = [];
151 public $mMarkerIndex = 0;
152 public $mFirstCall = true;
153
154 # Initialised by initialiseVariables()
155
156 /**
157 * @var MagicWordArray
158 */
159 public $mVariables;
160
161 /**
162 * @var MagicWordArray
163 */
164 public $mSubstWords;
165 # Initialised in constructor
166 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
167
168 # Initialized in getPreprocessor()
169 /** @var Preprocessor */
170 public $mPreprocessor;
171
172 # Cleared with clearState():
173 /**
174 * @var ParserOutput
175 */
176 public $mOutput;
177 public $mAutonumber;
178
179 /**
180 * @var StripState
181 */
182 public $mStripState;
183
184 public $mIncludeCount;
185 /**
186 * @var LinkHolderArray
187 */
188 public $mLinkHolders;
189
190 public $mLinkID;
191 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
192 public $mDefaultSort;
193 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
194 public $mExpensiveFunctionCount; # number of expensive parser function calls
195 public $mShowToc, $mForceTocPosition;
196
197 /**
198 * @var User
199 */
200 public $mUser; # User object; only used when doing pre-save transform
201
202 # Temporary
203 # These are variables reset at least once per parse regardless of $clearState
204
205 /**
206 * @var ParserOptions
207 */
208 public $mOptions;
209
210 /**
211 * @var Title
212 */
213 public $mTitle; # Title context, used for self-link rendering and similar things
214 public $mOutputType; # Output type, one of the OT_xxx constants
215 public $ot; # Shortcut alias, see setOutputType()
216 public $mRevisionObject; # The revision object of the specified revision ID
217 public $mRevisionId; # ID to display in {{REVISIONID}} tags
218 public $mRevisionTimestamp; # The timestamp of the specified revision ID
219 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
220 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
221 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
222 public $mInputSize = false; # For {{PAGESIZE}} on current page.
223
224 /**
225 * @var string Deprecated accessor for the strip marker prefix.
226 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
227 **/
228 public $mUniqPrefix = Parser::MARKER_PREFIX;
229
230 /**
231 * @var array Array with the language name of each language link (i.e. the
232 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
233 * duplicate language links to the ParserOutput.
234 */
235 public $mLangLinkLanguages;
236
237 /**
238 * @var MapCacheLRU|null
239 * @since 1.24
240 *
241 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
242 */
243 public $currentRevisionCache;
244
245 /**
246 * @var bool Recursive call protection.
247 * This variable should be treated as if it were private.
248 */
249 public $mInParse = false;
250
251 /** @var SectionProfiler */
252 protected $mProfiler;
253
254 /**
255 * @var LinkRenderer
256 */
257 protected $mLinkRenderer;
258
259 /**
260 * @param array $conf
261 */
262 public function __construct( $conf = [] ) {
263 $this->mConf = $conf;
264 $this->mUrlProtocols = wfUrlProtocols();
265 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
266 self::EXT_LINK_ADDR .
267 self::EXT_LINK_URL_CLASS . '*)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
268 if ( isset( $conf['preprocessorClass'] ) ) {
269 $this->mPreprocessorClass = $conf['preprocessorClass'];
270 } elseif ( defined( 'HPHP_VERSION' ) ) {
271 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
272 $this->mPreprocessorClass = 'Preprocessor_Hash';
273 } elseif ( extension_loaded( 'domxml' ) ) {
274 # PECL extension that conflicts with the core DOM extension (bug 13770)
275 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
276 $this->mPreprocessorClass = 'Preprocessor_Hash';
277 } elseif ( extension_loaded( 'dom' ) ) {
278 $this->mPreprocessorClass = 'Preprocessor_DOM';
279 } else {
280 $this->mPreprocessorClass = 'Preprocessor_Hash';
281 }
282 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
283 }
284
285 /**
286 * Reduce memory usage to reduce the impact of circular references
287 */
288 public function __destruct() {
289 if ( isset( $this->mLinkHolders ) ) {
290 unset( $this->mLinkHolders );
291 }
292 foreach ( $this as $name => $value ) {
293 unset( $this->$name );
294 }
295 }
296
297 /**
298 * Allow extensions to clean up when the parser is cloned
299 */
300 public function __clone() {
301 $this->mInParse = false;
302
303 // Bug 56226: When you create a reference "to" an object field, that
304 // makes the object field itself be a reference too (until the other
305 // reference goes out of scope). When cloning, any field that's a
306 // reference is copied as a reference in the new object. Both of these
307 // are defined PHP5 behaviors, as inconvenient as it is for us when old
308 // hooks from PHP4 days are passing fields by reference.
309 foreach ( [ 'mStripState', 'mVarCache' ] as $k ) {
310 // Make a non-reference copy of the field, then rebind the field to
311 // reference the new copy.
312 $tmp = $this->$k;
313 $this->$k =& $tmp;
314 unset( $tmp );
315 }
316
317 Hooks::run( 'ParserCloned', [ $this ] );
318 }
319
320 /**
321 * Do various kinds of initialisation on the first call of the parser
322 */
323 public function firstCallInit() {
324 if ( !$this->mFirstCall ) {
325 return;
326 }
327 $this->mFirstCall = false;
328
329 CoreParserFunctions::register( $this );
330 CoreTagHooks::register( $this );
331 $this->initialiseVariables();
332
333 Hooks::run( 'ParserFirstCallInit', [ &$this ] );
334 }
335
336 /**
337 * Clear Parser state
338 *
339 * @private
340 */
341 public function clearState() {
342 if ( $this->mFirstCall ) {
343 $this->firstCallInit();
344 }
345 $this->mOutput = new ParserOutput;
346 $this->mOptions->registerWatcher( [ $this->mOutput, 'recordOption' ] );
347 $this->mAutonumber = 0;
348 $this->mIncludeCount = [];
349 $this->mLinkHolders = new LinkHolderArray( $this );
350 $this->mLinkID = 0;
351 $this->mRevisionObject = $this->mRevisionTimestamp =
352 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
353 $this->mVarCache = [];
354 $this->mUser = null;
355 $this->mLangLinkLanguages = [];
356 $this->currentRevisionCache = null;
357
358 $this->mStripState = new StripState;
359
360 # Clear these on every parse, bug 4549
361 $this->mTplRedirCache = $this->mTplDomCache = [];
362
363 $this->mShowToc = true;
364 $this->mForceTocPosition = false;
365 $this->mIncludeSizes = [
366 'post-expand' => 0,
367 'arg' => 0,
368 ];
369 $this->mPPNodeCount = 0;
370 $this->mGeneratedPPNodeCount = 0;
371 $this->mHighestExpansionDepth = 0;
372 $this->mDefaultSort = false;
373 $this->mHeadings = [];
374 $this->mDoubleUnderscores = [];
375 $this->mExpensiveFunctionCount = 0;
376
377 # Fix cloning
378 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
379 $this->mPreprocessor = null;
380 }
381
382 $this->mProfiler = new SectionProfiler();
383
384 Hooks::run( 'ParserClearState', [ &$this ] );
385 }
386
387 /**
388 * Convert wikitext to HTML
389 * Do not call this function recursively.
390 *
391 * @param string $text Text we want to parse
392 * @param Title $title
393 * @param ParserOptions $options
394 * @param bool $linestart
395 * @param bool $clearState
396 * @param int $revid Number to pass in {{REVISIONID}}
397 * @return ParserOutput A ParserOutput
398 */
399 public function parse(
400 $text, Title $title, ParserOptions $options,
401 $linestart = true, $clearState = true, $revid = null
402 ) {
403 /**
404 * First pass--just handle <nowiki> sections, pass the rest off
405 * to internalParse() which does all the real work.
406 */
407
408 global $wgShowHostnames;
409
410 if ( $clearState ) {
411 // We use U+007F DELETE to construct strip markers, so we have to make
412 // sure that this character does not occur in the input text.
413 $text = strtr( $text, "\x7f", "?" );
414 $magicScopeVariable = $this->lock();
415 }
416
417 $this->startParse( $title, $options, self::OT_HTML, $clearState );
418
419 $this->currentRevisionCache = null;
420 $this->mInputSize = strlen( $text );
421 if ( $this->mOptions->getEnableLimitReport() ) {
422 $this->mOutput->resetParseStartTime();
423 }
424
425 $oldRevisionId = $this->mRevisionId;
426 $oldRevisionObject = $this->mRevisionObject;
427 $oldRevisionTimestamp = $this->mRevisionTimestamp;
428 $oldRevisionUser = $this->mRevisionUser;
429 $oldRevisionSize = $this->mRevisionSize;
430 if ( $revid !== null ) {
431 $this->mRevisionId = $revid;
432 $this->mRevisionObject = null;
433 $this->mRevisionTimestamp = null;
434 $this->mRevisionUser = null;
435 $this->mRevisionSize = null;
436 }
437
438 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
439 # No more strip!
440 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
441 $text = $this->internalParse( $text );
442 Hooks::run( 'ParserAfterParse', [ &$this, &$text, &$this->mStripState ] );
443
444 $text = $this->internalParseHalfParsed( $text, true, $linestart );
445
446 /**
447 * A converted title will be provided in the output object if title and
448 * content conversion are enabled, the article text does not contain
449 * a conversion-suppressing double-underscore tag, and no
450 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
451 * automatic link conversion.
452 */
453 if ( !( $options->getDisableTitleConversion()
454 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
455 || isset( $this->mDoubleUnderscores['notitleconvert'] )
456 || $this->mOutput->getDisplayTitle() !== false )
457 ) {
458 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
459 if ( $convruletitle ) {
460 $this->mOutput->setTitleText( $convruletitle );
461 } else {
462 $titleText = $this->getConverterLanguage()->convertTitle( $title );
463 $this->mOutput->setTitleText( $titleText );
464 }
465 }
466
467 # Done parsing! Compute runtime adaptive expiry if set
468 $this->mOutput->finalizeAdaptiveCacheExpiry();
469
470 # Warn if too many heavyweight parser functions were used
471 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
472 $this->limitationWarn( 'expensive-parserfunction',
473 $this->mExpensiveFunctionCount,
474 $this->mOptions->getExpensiveParserFunctionLimit()
475 );
476 }
477
478 # Information on include size limits, for the benefit of users who try to skirt them
479 if ( $this->mOptions->getEnableLimitReport() ) {
480 $max = $this->mOptions->getMaxIncludeSize();
481
482 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
483 if ( $cpuTime !== null ) {
484 $this->mOutput->setLimitReportData( 'limitreport-cputime',
485 sprintf( "%.3f", $cpuTime )
486 );
487 }
488
489 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
490 $this->mOutput->setLimitReportData( 'limitreport-walltime',
491 sprintf( "%.3f", $wallTime )
492 );
493
494 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
495 [ $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() ]
496 );
497 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
498 [ $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() ]
499 );
500 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
501 [ $this->mIncludeSizes['post-expand'], $max ]
502 );
503 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
504 [ $this->mIncludeSizes['arg'], $max ]
505 );
506 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
507 [ $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() ]
508 );
509 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
510 [ $this->mExpensiveFunctionCount,
511 $this->mOptions->getExpensiveParserFunctionLimit() ]
512 );
513 Hooks::run( 'ParserLimitReportPrepare', [ $this, $this->mOutput ] );
514
515 $limitReport = '';
516 Hooks::run( 'ParserLimitReport', [ $this, &$limitReport ] );
517 if ( $limitReport != '' ) {
518 // Sanitize for comment. Note '‐' in the replacement is U+2010,
519 // which looks much like the problematic '-'.
520 $limitReport = str_replace( [ '-', '&' ], [ '‐', '&amp;' ], $limitReport );
521 $text .= "\n<!-- \nNewPP limit report\n$limitReport-->\n";
522 }
523
524 // Add on template profiling data in human/machine readable way
525 $dataByFunc = $this->mProfiler->getFunctionStats();
526 uasort( $dataByFunc, function ( $a, $b ) {
527 return $a['real'] < $b['real']; // descending order
528 } );
529 $profileReport = [];
530 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
531 $profileReport[] = sprintf( "%6.2f%% %8.3f %6d %s",
532 $item['%real'], $item['real'], $item['calls'], $item['name'] );
533 }
534 $this->mOutput->setLimitReportData( 'limitreport-timingprofile', $profileReport );
535
536 // Add other cache related metadata
537 if ( $wgShowHostnames ) {
538 $this->mOutput->setLimitReportData( 'cachereport-origin', wfHostname() );
539 }
540 $this->mOutput->setLimitReportData( 'cachereport-timestamp',
541 $this->mOutput->getCacheTime() );
542 $this->mOutput->setLimitReportData( 'cachereport-ttl',
543 $this->mOutput->getCacheExpiry() );
544 $this->mOutput->setLimitReportData( 'cachereport-transientcontent',
545 $this->mOutput->hasDynamicContent() );
546
547 if ( $this->mGeneratedPPNodeCount
548 > $this->mOptions->getMaxGeneratedPPNodeCount() / 10
549 ) {
550 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
551 $this->mTitle->getPrefixedDBkey() );
552 }
553 }
554 $this->mOutput->setText( $text );
555
556 $this->mRevisionId = $oldRevisionId;
557 $this->mRevisionObject = $oldRevisionObject;
558 $this->mRevisionTimestamp = $oldRevisionTimestamp;
559 $this->mRevisionUser = $oldRevisionUser;
560 $this->mRevisionSize = $oldRevisionSize;
561 $this->mInputSize = false;
562 $this->currentRevisionCache = null;
563
564 return $this->mOutput;
565 }
566
567 /**
568 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
569 * can be called from an extension tag hook.
570 *
571 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
572 * instead, which means that lists and links have not been fully parsed yet,
573 * and strip markers are still present.
574 *
575 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
576 *
577 * Use this function if you're a parser tag hook and you want to parse
578 * wikitext before or after applying additional transformations, and you
579 * intend to *return the result as hook output*, which will cause it to go
580 * through the rest of parsing process automatically.
581 *
582 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
583 * $text are not expanded
584 *
585 * @param string $text Text extension wants to have parsed
586 * @param bool|PPFrame $frame The frame to use for expanding any template variables
587 * @return string UNSAFE half-parsed HTML
588 */
589 public function recursiveTagParse( $text, $frame = false ) {
590 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
591 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
592 $text = $this->internalParse( $text, false, $frame );
593 return $text;
594 }
595
596 /**
597 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
598 * point can be called from an extension tag hook.
599 *
600 * The output of this function is fully-parsed HTML that is safe for output.
601 * If you're a parser tag hook, you might want to use recursiveTagParse()
602 * instead.
603 *
604 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
605 * $text are not expanded
606 *
607 * @since 1.25
608 *
609 * @param string $text Text extension wants to have parsed
610 * @param bool|PPFrame $frame The frame to use for expanding any template variables
611 * @return string Fully parsed HTML
612 */
613 public function recursiveTagParseFully( $text, $frame = false ) {
614 $text = $this->recursiveTagParse( $text, $frame );
615 $text = $this->internalParseHalfParsed( $text, false );
616 return $text;
617 }
618
619 /**
620 * Expand templates and variables in the text, producing valid, static wikitext.
621 * Also removes comments.
622 * Do not call this function recursively.
623 * @param string $text
624 * @param Title $title
625 * @param ParserOptions $options
626 * @param int|null $revid
627 * @param bool|PPFrame $frame
628 * @return mixed|string
629 */
630 public function preprocess( $text, Title $title = null,
631 ParserOptions $options, $revid = null, $frame = false
632 ) {
633 $magicScopeVariable = $this->lock();
634 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
635 if ( $revid !== null ) {
636 $this->mRevisionId = $revid;
637 }
638 Hooks::run( 'ParserBeforeStrip', [ &$this, &$text, &$this->mStripState ] );
639 Hooks::run( 'ParserAfterStrip', [ &$this, &$text, &$this->mStripState ] );
640 $text = $this->replaceVariables( $text, $frame );
641 $text = $this->mStripState->unstripBoth( $text );
642 return $text;
643 }
644
645 /**
646 * Recursive parser entry point that can be called from an extension tag
647 * hook.
648 *
649 * @param string $text Text to be expanded
650 * @param bool|PPFrame $frame The frame to use for expanding any template variables
651 * @return string
652 * @since 1.19
653 */
654 public function recursivePreprocess( $text, $frame = false ) {
655 $text = $this->replaceVariables( $text, $frame );
656 $text = $this->mStripState->unstripBoth( $text );
657 return $text;
658 }
659
660 /**
661 * Process the wikitext for the "?preload=" feature. (bug 5210)
662 *
663 * "<noinclude>", "<includeonly>" etc. are parsed as for template
664 * transclusion, comments, templates, arguments, tags hooks and parser
665 * functions are untouched.
666 *
667 * @param string $text
668 * @param Title $title
669 * @param ParserOptions $options
670 * @param array $params
671 * @return string
672 */
673 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = [] ) {
674 $msg = new RawMessage( $text );
675 $text = $msg->params( $params )->plain();
676
677 # Parser (re)initialisation
678 $magicScopeVariable = $this->lock();
679 $this->startParse( $title, $options, self::OT_PLAIN, true );
680
681 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
682 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
683 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
684 $text = $this->mStripState->unstripBoth( $text );
685 return $text;
686 }
687
688 /**
689 * Get a random string
690 *
691 * @return string
692 * @deprecated since 1.26; use wfRandomString() instead.
693 */
694 public static function getRandomString() {
695 wfDeprecated( __METHOD__, '1.26' );
696 return wfRandomString( 16 );
697 }
698
699 /**
700 * Set the current user.
701 * Should only be used when doing pre-save transform.
702 *
703 * @param User|null $user User object or null (to reset)
704 */
705 public function setUser( $user ) {
706 $this->mUser = $user;
707 }
708
709 /**
710 * Accessor for mUniqPrefix.
711 *
712 * @return string
713 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
714 */
715 public function uniqPrefix() {
716 wfDeprecated( __METHOD__, '1.26' );
717 return self::MARKER_PREFIX;
718 }
719
720 /**
721 * Set the context title
722 *
723 * @param Title $t
724 */
725 public function setTitle( $t ) {
726 if ( !$t ) {
727 $t = Title::newFromText( 'NO TITLE' );
728 }
729
730 if ( $t->hasFragment() ) {
731 # Strip the fragment to avoid various odd effects
732 $this->mTitle = $t->createFragmentTarget( '' );
733 } else {
734 $this->mTitle = $t;
735 }
736 }
737
738 /**
739 * Accessor for the Title object
740 *
741 * @return Title
742 */
743 public function getTitle() {
744 return $this->mTitle;
745 }
746
747 /**
748 * Accessor/mutator for the Title object
749 *
750 * @param Title $x Title object or null to just get the current one
751 * @return Title
752 */
753 public function Title( $x = null ) {
754 return wfSetVar( $this->mTitle, $x );
755 }
756
757 /**
758 * Set the output type
759 *
760 * @param int $ot New value
761 */
762 public function setOutputType( $ot ) {
763 $this->mOutputType = $ot;
764 # Shortcut alias
765 $this->ot = [
766 'html' => $ot == self::OT_HTML,
767 'wiki' => $ot == self::OT_WIKI,
768 'pre' => $ot == self::OT_PREPROCESS,
769 'plain' => $ot == self::OT_PLAIN,
770 ];
771 }
772
773 /**
774 * Accessor/mutator for the output type
775 *
776 * @param int|null $x New value or null to just get the current one
777 * @return int
778 */
779 public function OutputType( $x = null ) {
780 return wfSetVar( $this->mOutputType, $x );
781 }
782
783 /**
784 * Get the ParserOutput object
785 *
786 * @return ParserOutput
787 */
788 public function getOutput() {
789 return $this->mOutput;
790 }
791
792 /**
793 * Get the ParserOptions object
794 *
795 * @return ParserOptions
796 */
797 public function getOptions() {
798 return $this->mOptions;
799 }
800
801 /**
802 * Accessor/mutator for the ParserOptions object
803 *
804 * @param ParserOptions $x New value or null to just get the current one
805 * @return ParserOptions Current ParserOptions object
806 */
807 public function Options( $x = null ) {
808 return wfSetVar( $this->mOptions, $x );
809 }
810
811 /**
812 * @return int
813 */
814 public function nextLinkID() {
815 return $this->mLinkID++;
816 }
817
818 /**
819 * @param int $id
820 */
821 public function setLinkID( $id ) {
822 $this->mLinkID = $id;
823 }
824
825 /**
826 * Get a language object for use in parser functions such as {{FORMATNUM:}}
827 * @return Language
828 */
829 public function getFunctionLang() {
830 return $this->getTargetLanguage();
831 }
832
833 /**
834 * Get the target language for the content being parsed. This is usually the
835 * language that the content is in.
836 *
837 * @since 1.19
838 *
839 * @throws MWException
840 * @return Language
841 */
842 public function getTargetLanguage() {
843 $target = $this->mOptions->getTargetLanguage();
844
845 if ( $target !== null ) {
846 return $target;
847 } elseif ( $this->mOptions->getInterfaceMessage() ) {
848 return $this->mOptions->getUserLangObj();
849 } elseif ( is_null( $this->mTitle ) ) {
850 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
851 }
852
853 return $this->mTitle->getPageLanguage();
854 }
855
856 /**
857 * Get the language object for language conversion
858 * @return Language|null
859 */
860 public function getConverterLanguage() {
861 return $this->getTargetLanguage();
862 }
863
864 /**
865 * Get a User object either from $this->mUser, if set, or from the
866 * ParserOptions object otherwise
867 *
868 * @return User
869 */
870 public function getUser() {
871 if ( !is_null( $this->mUser ) ) {
872 return $this->mUser;
873 }
874 return $this->mOptions->getUser();
875 }
876
877 /**
878 * Get a preprocessor object
879 *
880 * @return Preprocessor
881 */
882 public function getPreprocessor() {
883 if ( !isset( $this->mPreprocessor ) ) {
884 $class = $this->mPreprocessorClass;
885 $this->mPreprocessor = new $class( $this );
886 }
887 return $this->mPreprocessor;
888 }
889
890 /**
891 * Get a LinkRenderer instance to make links with
892 *
893 * @since 1.28
894 * @return LinkRenderer
895 */
896 public function getLinkRenderer() {
897 if ( !$this->mLinkRenderer ) {
898 $this->mLinkRenderer = MediaWikiServices::getInstance()
899 ->getLinkRendererFactory()->create();
900 $this->mLinkRenderer->setStubThreshold(
901 $this->getOptions()->getStubThreshold()
902 );
903 }
904
905 return $this->mLinkRenderer;
906 }
907
908 /**
909 * Replaces all occurrences of HTML-style comments and the given tags
910 * in the text with a random marker and returns the next text. The output
911 * parameter $matches will be an associative array filled with data in
912 * the form:
913 *
914 * @code
915 * 'UNIQ-xxxxx' => [
916 * 'element',
917 * 'tag content',
918 * [ 'param' => 'x' ],
919 * '<element param="x">tag content</element>' ]
920 * @endcode
921 *
922 * @param array $elements List of element names. Comments are always extracted.
923 * @param string $text Source text string.
924 * @param array $matches Out parameter, Array: extracted tags
925 * @param string|null $uniq_prefix
926 * @return string Stripped text
927 * @since 1.26 The uniq_prefix argument is deprecated.
928 */
929 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
930 if ( $uniq_prefix !== null ) {
931 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
932 }
933 static $n = 1;
934 $stripped = '';
935 $matches = [];
936
937 $taglist = implode( '|', $elements );
938 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
939
940 while ( $text != '' ) {
941 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
942 $stripped .= $p[0];
943 if ( count( $p ) < 5 ) {
944 break;
945 }
946 if ( count( $p ) > 5 ) {
947 # comment
948 $element = $p[4];
949 $attributes = '';
950 $close = '';
951 $inside = $p[5];
952 } else {
953 # tag
954 $element = $p[1];
955 $attributes = $p[2];
956 $close = $p[3];
957 $inside = $p[4];
958 }
959
960 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
961 $stripped .= $marker;
962
963 if ( $close === '/>' ) {
964 # Empty element tag, <tag />
965 $content = null;
966 $text = $inside;
967 $tail = null;
968 } else {
969 if ( $element === '!--' ) {
970 $end = '/(-->)/';
971 } else {
972 $end = "/(<\\/$element\\s*>)/i";
973 }
974 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
975 $content = $q[0];
976 if ( count( $q ) < 3 ) {
977 # No end tag -- let it run out to the end of the text.
978 $tail = '';
979 $text = '';
980 } else {
981 $tail = $q[1];
982 $text = $q[2];
983 }
984 }
985
986 $matches[$marker] = [ $element,
987 $content,
988 Sanitizer::decodeTagAttributes( $attributes ),
989 "<$element$attributes$close$content$tail" ];
990 }
991 return $stripped;
992 }
993
994 /**
995 * Get a list of strippable XML-like elements
996 *
997 * @return array
998 */
999 public function getStripList() {
1000 return $this->mStripList;
1001 }
1002
1003 /**
1004 * Add an item to the strip state
1005 * Returns the unique tag which must be inserted into the stripped text
1006 * The tag will be replaced with the original text in unstrip()
1007 *
1008 * @param string $text
1009 *
1010 * @return string
1011 */
1012 public function insertStripItem( $text ) {
1013 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
1014 $this->mMarkerIndex++;
1015 $this->mStripState->addGeneral( $marker, $text );
1016 return $marker;
1017 }
1018
1019 /**
1020 * parse the wiki syntax used to render tables
1021 *
1022 * @private
1023 * @param string $text
1024 * @return string
1025 */
1026 public function doTableStuff( $text ) {
1027
1028 $lines = StringUtils::explode( "\n", $text );
1029 $out = '';
1030 $td_history = []; # Is currently a td tag open?
1031 $last_tag_history = []; # Save history of last lag activated (td, th or caption)
1032 $tr_history = []; # Is currently a tr tag open?
1033 $tr_attributes = []; # history of tr attributes
1034 $has_opened_tr = []; # Did this table open a <tr> element?
1035 $indent_level = 0; # indent level of the table
1036
1037 foreach ( $lines as $outLine ) {
1038 $line = trim( $outLine );
1039
1040 if ( $line === '' ) { # empty line, go to next line
1041 $out .= $outLine . "\n";
1042 continue;
1043 }
1044
1045 $first_character = $line[0];
1046 $first_two = substr( $line, 0, 2 );
1047 $matches = [];
1048
1049 if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line, $matches ) ) {
1050 # First check if we are starting a new table
1051 $indent_level = strlen( $matches[1] );
1052
1053 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1054 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1055
1056 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1057 array_push( $td_history, false );
1058 array_push( $last_tag_history, '' );
1059 array_push( $tr_history, false );
1060 array_push( $tr_attributes, '' );
1061 array_push( $has_opened_tr, false );
1062 } elseif ( count( $td_history ) == 0 ) {
1063 # Don't do any of the following
1064 $out .= $outLine . "\n";
1065 continue;
1066 } elseif ( $first_two === '|}' ) {
1067 # We are ending a table
1068 $line = '</table>' . substr( $line, 2 );
1069 $last_tag = array_pop( $last_tag_history );
1070
1071 if ( !array_pop( $has_opened_tr ) ) {
1072 $line = "<tr><td></td></tr>{$line}";
1073 }
1074
1075 if ( array_pop( $tr_history ) ) {
1076 $line = "</tr>{$line}";
1077 }
1078
1079 if ( array_pop( $td_history ) ) {
1080 $line = "</{$last_tag}>{$line}";
1081 }
1082 array_pop( $tr_attributes );
1083 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1084 } elseif ( $first_two === '|-' ) {
1085 # Now we have a table row
1086 $line = preg_replace( '#^\|-+#', '', $line );
1087
1088 # Whats after the tag is now only attributes
1089 $attributes = $this->mStripState->unstripBoth( $line );
1090 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1091 array_pop( $tr_attributes );
1092 array_push( $tr_attributes, $attributes );
1093
1094 $line = '';
1095 $last_tag = array_pop( $last_tag_history );
1096 array_pop( $has_opened_tr );
1097 array_push( $has_opened_tr, true );
1098
1099 if ( array_pop( $tr_history ) ) {
1100 $line = '</tr>';
1101 }
1102
1103 if ( array_pop( $td_history ) ) {
1104 $line = "</{$last_tag}>{$line}";
1105 }
1106
1107 $outLine = $line;
1108 array_push( $tr_history, false );
1109 array_push( $td_history, false );
1110 array_push( $last_tag_history, '' );
1111 } elseif ( $first_character === '|'
1112 || $first_character === '!'
1113 || $first_two === '|+'
1114 ) {
1115 # This might be cell elements, td, th or captions
1116 if ( $first_two === '|+' ) {
1117 $first_character = '+';
1118 $line = substr( $line, 2 );
1119 } else {
1120 $line = substr( $line, 1 );
1121 }
1122
1123 // Implies both are valid for table headings.
1124 if ( $first_character === '!' ) {
1125 $line = StringUtils::replaceMarkup( '!!', '||', $line );
1126 }
1127
1128 # Split up multiple cells on the same line.
1129 # FIXME : This can result in improper nesting of tags processed
1130 # by earlier parser steps.
1131 $cells = explode( '||', $line );
1132
1133 $outLine = '';
1134
1135 # Loop through each table cell
1136 foreach ( $cells as $cell ) {
1137 $previous = '';
1138 if ( $first_character !== '+' ) {
1139 $tr_after = array_pop( $tr_attributes );
1140 if ( !array_pop( $tr_history ) ) {
1141 $previous = "<tr{$tr_after}>\n";
1142 }
1143 array_push( $tr_history, true );
1144 array_push( $tr_attributes, '' );
1145 array_pop( $has_opened_tr );
1146 array_push( $has_opened_tr, true );
1147 }
1148
1149 $last_tag = array_pop( $last_tag_history );
1150
1151 if ( array_pop( $td_history ) ) {
1152 $previous = "</{$last_tag}>\n{$previous}";
1153 }
1154
1155 if ( $first_character === '|' ) {
1156 $last_tag = 'td';
1157 } elseif ( $first_character === '!' ) {
1158 $last_tag = 'th';
1159 } elseif ( $first_character === '+' ) {
1160 $last_tag = 'caption';
1161 } else {
1162 $last_tag = '';
1163 }
1164
1165 array_push( $last_tag_history, $last_tag );
1166
1167 # A cell could contain both parameters and data
1168 $cell_data = explode( '|', $cell, 2 );
1169
1170 # Bug 553: Note that a '|' inside an invalid link should not
1171 # be mistaken as delimiting cell parameters
1172 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1173 $cell = "{$previous}<{$last_tag}>{$cell}";
1174 } elseif ( count( $cell_data ) == 1 ) {
1175 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1176 } else {
1177 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1178 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1179 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1180 }
1181
1182 $outLine .= $cell;
1183 array_push( $td_history, true );
1184 }
1185 }
1186 $out .= $outLine . "\n";
1187 }
1188
1189 # Closing open td, tr && table
1190 while ( count( $td_history ) > 0 ) {
1191 if ( array_pop( $td_history ) ) {
1192 $out .= "</td>\n";
1193 }
1194 if ( array_pop( $tr_history ) ) {
1195 $out .= "</tr>\n";
1196 }
1197 if ( !array_pop( $has_opened_tr ) ) {
1198 $out .= "<tr><td></td></tr>\n";
1199 }
1200
1201 $out .= "</table>\n";
1202 }
1203
1204 # Remove trailing line-ending (b/c)
1205 if ( substr( $out, -1 ) === "\n" ) {
1206 $out = substr( $out, 0, -1 );
1207 }
1208
1209 # special case: don't return empty table
1210 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1211 $out = '';
1212 }
1213
1214 return $out;
1215 }
1216
1217 /**
1218 * Helper function for parse() that transforms wiki markup into half-parsed
1219 * HTML. Only called for $mOutputType == self::OT_HTML.
1220 *
1221 * @private
1222 *
1223 * @param string $text The text to parse
1224 * @param bool $isMain Whether this is being called from the main parse() function
1225 * @param PPFrame|bool $frame A pre-processor frame
1226 *
1227 * @return string
1228 */
1229 public function internalParse( $text, $isMain = true, $frame = false ) {
1230
1231 $origText = $text;
1232
1233 # Hook to suspend the parser in this state
1234 if ( !Hooks::run( 'ParserBeforeInternalParse', [ &$this, &$text, &$this->mStripState ] ) ) {
1235 return $text;
1236 }
1237
1238 # if $frame is provided, then use $frame for replacing any variables
1239 if ( $frame ) {
1240 # use frame depth to infer how include/noinclude tags should be handled
1241 # depth=0 means this is the top-level document; otherwise it's an included document
1242 if ( !$frame->depth ) {
1243 $flag = 0;
1244 } else {
1245 $flag = Parser::PTD_FOR_INCLUSION;
1246 }
1247 $dom = $this->preprocessToDom( $text, $flag );
1248 $text = $frame->expand( $dom );
1249 } else {
1250 # if $frame is not provided, then use old-style replaceVariables
1251 $text = $this->replaceVariables( $text );
1252 }
1253
1254 Hooks::run( 'InternalParseBeforeSanitize', [ &$this, &$text, &$this->mStripState ] );
1255 $text = Sanitizer::removeHTMLtags(
1256 $text,
1257 [ &$this, 'attributeStripCallback' ],
1258 false,
1259 array_keys( $this->mTransparentTagHooks ),
1260 [],
1261 [ &$this, 'addTrackingCategory' ]
1262 );
1263 Hooks::run( 'InternalParseBeforeLinks', [ &$this, &$text, &$this->mStripState ] );
1264
1265 # Tables need to come after variable replacement for things to work
1266 # properly; putting them before other transformations should keep
1267 # exciting things like link expansions from showing up in surprising
1268 # places.
1269 $text = $this->doTableStuff( $text );
1270
1271 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1272
1273 $text = $this->doDoubleUnderscore( $text );
1274
1275 $text = $this->doHeadings( $text );
1276 $text = $this->replaceInternalLinks( $text );
1277 $text = $this->doAllQuotes( $text );
1278 $text = $this->replaceExternalLinks( $text );
1279
1280 # replaceInternalLinks may sometimes leave behind
1281 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1282 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1283
1284 $text = $this->doMagicLinks( $text );
1285 $text = $this->formatHeadings( $text, $origText, $isMain );
1286
1287 return $text;
1288 }
1289
1290 /**
1291 * Helper function for parse() that transforms half-parsed HTML into fully
1292 * parsed HTML.
1293 *
1294 * @param string $text
1295 * @param bool $isMain
1296 * @param bool $linestart
1297 * @return string
1298 */
1299 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1300 $text = $this->mStripState->unstripGeneral( $text );
1301
1302 if ( $isMain ) {
1303 Hooks::run( 'ParserAfterUnstrip', [ &$this, &$text ] );
1304 }
1305
1306 # Clean up special characters, only run once, next-to-last before doBlockLevels
1307 $fixtags = [
1308 # french spaces, last one Guillemet-left
1309 # only if there is something before the space
1310 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1311 # french spaces, Guillemet-right
1312 '/(\\302\\253) /' => '\\1&#160;',
1313 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1314 ];
1315 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1316
1317 $text = $this->doBlockLevels( $text, $linestart );
1318
1319 $this->replaceLinkHolders( $text );
1320
1321 /**
1322 * The input doesn't get language converted if
1323 * a) It's disabled
1324 * b) Content isn't converted
1325 * c) It's a conversion table
1326 * d) it is an interface message (which is in the user language)
1327 */
1328 if ( !( $this->mOptions->getDisableContentConversion()
1329 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1330 ) {
1331 if ( !$this->mOptions->getInterfaceMessage() ) {
1332 # The position of the convert() call should not be changed. it
1333 # assumes that the links are all replaced and the only thing left
1334 # is the <nowiki> mark.
1335 $text = $this->getConverterLanguage()->convert( $text );
1336 }
1337 }
1338
1339 $text = $this->mStripState->unstripNoWiki( $text );
1340
1341 if ( $isMain ) {
1342 Hooks::run( 'ParserBeforeTidy', [ &$this, &$text ] );
1343 }
1344
1345 $text = $this->replaceTransparentTags( $text );
1346 $text = $this->mStripState->unstripGeneral( $text );
1347
1348 $text = Sanitizer::normalizeCharReferences( $text );
1349
1350 if ( MWTidy::isEnabled() ) {
1351 if ( $this->mOptions->getTidy() ) {
1352 $text = MWTidy::tidy( $text );
1353 }
1354 } else {
1355 # attempt to sanitize at least some nesting problems
1356 # (bug #2702 and quite a few others)
1357 $tidyregs = [
1358 # ''Something [http://www.cool.com cool''] -->
1359 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1360 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1361 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1362 # fix up an anchor inside another anchor, only
1363 # at least for a single single nested link (bug 3695)
1364 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1365 '\\1\\2</a>\\3</a>\\1\\4</a>',
1366 # fix div inside inline elements- doBlockLevels won't wrap a line which
1367 # contains a div, so fix it up here; replace
1368 # div with escaped text
1369 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1370 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1371 # remove empty italic or bold tag pairs, some
1372 # introduced by rules above
1373 '/<([bi])><\/\\1>/' => '',
1374 ];
1375
1376 $text = preg_replace(
1377 array_keys( $tidyregs ),
1378 array_values( $tidyregs ),
1379 $text );
1380 }
1381
1382 if ( $isMain ) {
1383 Hooks::run( 'ParserAfterTidy', [ &$this, &$text ] );
1384 }
1385
1386 return $text;
1387 }
1388
1389 /**
1390 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1391 * magic external links.
1392 *
1393 * DML
1394 * @private
1395 *
1396 * @param string $text
1397 *
1398 * @return string
1399 */
1400 public function doMagicLinks( $text ) {
1401 $prots = wfUrlProtocolsWithoutProtRel();
1402 $urlChar = self::EXT_LINK_URL_CLASS;
1403 $addr = self::EXT_LINK_ADDR;
1404 $space = self::SPACE_NOT_NL; # non-newline space
1405 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1406 $spaces = "$space++"; # possessive match of 1 or more spaces
1407 $text = preg_replace_callback(
1408 '!(?: # Start cases
1409 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1410 (<.*?>) | # m[2]: Skip stuff inside
1411 # HTML elements' . "
1412 (\b(?i:$prots)($addr$urlChar*)) | # m[3]: Free external links
1413 # m[4]: Post-protocol path
1414 \b(?:RFC|PMID) $spaces # m[5]: RFC or PMID, capture number
1415 ([0-9]+)\b |
1416 \bISBN $spaces ( # m[6]: ISBN, capture number
1417 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1418 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1419 [0-9Xx] # check digit
1420 )\b
1421 )!xu", [ &$this, 'magicLinkCallback' ], $text );
1422 return $text;
1423 }
1424
1425 /**
1426 * @throws MWException
1427 * @param array $m
1428 * @return HTML|string
1429 */
1430 public function magicLinkCallback( $m ) {
1431 if ( isset( $m[1] ) && $m[1] !== '' ) {
1432 # Skip anchor
1433 return $m[0];
1434 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1435 # Skip HTML element
1436 return $m[0];
1437 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1438 # Free external link
1439 return $this->makeFreeExternalLink( $m[0], strlen( $m[4] ) );
1440 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1441 # RFC or PMID
1442 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1443 if ( !$this->mOptions->getMagicRFCLinks() ) {
1444 return $m[0];
1445 }
1446 $keyword = 'RFC';
1447 $urlmsg = 'rfcurl';
1448 $cssClass = 'mw-magiclink-rfc';
1449 $trackingCat = 'magiclink-tracking-rfc';
1450 $id = $m[5];
1451 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1452 if ( !$this->mOptions->getMagicPMIDLinks() ) {
1453 return $m[0];
1454 }
1455 $keyword = 'PMID';
1456 $urlmsg = 'pubmedurl';
1457 $cssClass = 'mw-magiclink-pmid';
1458 $trackingCat = 'magiclink-tracking-pmid';
1459 $id = $m[5];
1460 } else {
1461 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1462 substr( $m[0], 0, 20 ) . '"' );
1463 }
1464 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1465 $this->addTrackingCategory( $trackingCat );
1466 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass, [], $this->mTitle );
1467 } elseif ( isset( $m[6] ) && $m[6] !== ''
1468 && $this->mOptions->getMagicISBNLinks()
1469 ) {
1470 # ISBN
1471 $isbn = $m[6];
1472 $space = self::SPACE_NOT_NL; # non-newline space
1473 $isbn = preg_replace( "/$space/", ' ', $isbn );
1474 $num = strtr( $isbn, [
1475 '-' => '',
1476 ' ' => '',
1477 'x' => 'X',
1478 ] );
1479 $this->addTrackingCategory( 'magiclink-tracking-isbn' );
1480 return $this->getLinkRenderer()->makeKnownLink(
1481 SpecialPage::getTitleFor( 'Booksources', $num ),
1482 "ISBN $isbn",
1483 [
1484 'class' => 'internal mw-magiclink-isbn',
1485 'title' => false // suppress title attribute
1486 ]
1487 );
1488 } else {
1489 return $m[0];
1490 }
1491 }
1492
1493 /**
1494 * Make a free external link, given a user-supplied URL
1495 *
1496 * @param string $url
1497 * @param int $numPostProto
1498 * The number of characters after the protocol.
1499 * @return string HTML
1500 * @private
1501 */
1502 public function makeFreeExternalLink( $url, $numPostProto ) {
1503 $trail = '';
1504
1505 # The characters '<' and '>' (which were escaped by
1506 # removeHTMLtags()) should not be included in
1507 # URLs, per RFC 2396.
1508 # Make &nbsp; terminate a URL as well (bug T84937)
1509 $m2 = [];
1510 if ( preg_match(
1511 '/&(lt|gt|nbsp|#x0*(3[CcEe]|[Aa]0)|#0*(60|62|160));/',
1512 $url,
1513 $m2,
1514 PREG_OFFSET_CAPTURE
1515 ) ) {
1516 $trail = substr( $url, $m2[0][1] ) . $trail;
1517 $url = substr( $url, 0, $m2[0][1] );
1518 }
1519
1520 # Move trailing punctuation to $trail
1521 $sep = ',;\.:!?';
1522 # If there is no left bracket, then consider right brackets fair game too
1523 if ( strpos( $url, '(' ) === false ) {
1524 $sep .= ')';
1525 }
1526
1527 $urlRev = strrev( $url );
1528 $numSepChars = strspn( $urlRev, $sep );
1529 # Don't break a trailing HTML entity by moving the ; into $trail
1530 # This is in hot code, so use substr_compare to avoid having to
1531 # create a new string object for the comparison
1532 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0 ) {
1533 # more optimization: instead of running preg_match with a $
1534 # anchor, which can be slow, do the match on the reversed
1535 # string starting at the desired offset.
1536 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1537 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1538 $numSepChars--;
1539 }
1540 }
1541 if ( $numSepChars ) {
1542 $trail = substr( $url, -$numSepChars ) . $trail;
1543 $url = substr( $url, 0, -$numSepChars );
1544 }
1545
1546 # Verify that we still have a real URL after trail removal, and
1547 # not just lone protocol
1548 if ( strlen( $trail ) >= $numPostProto ) {
1549 return $url . $trail;
1550 }
1551
1552 $url = Sanitizer::cleanUrl( $url );
1553
1554 # Is this an external image?
1555 $text = $this->maybeMakeExternalImage( $url );
1556 if ( $text === false ) {
1557 # Not an image, make a link
1558 $text = Linker::makeExternalLink( $url,
1559 $this->getConverterLanguage()->markNoConversion( $url, true ),
1560 true, 'free',
1561 $this->getExternalLinkAttribs( $url ), $this->mTitle );
1562 # Register it in the output object...
1563 # Replace unnecessary URL escape codes with their equivalent characters
1564 $pasteurized = self::normalizeLinkUrl( $url );
1565 $this->mOutput->addExternalLink( $pasteurized );
1566 }
1567 return $text . $trail;
1568 }
1569
1570 /**
1571 * Parse headers and return html
1572 *
1573 * @private
1574 *
1575 * @param string $text
1576 *
1577 * @return string
1578 */
1579 public function doHeadings( $text ) {
1580 for ( $i = 6; $i >= 1; --$i ) {
1581 $h = str_repeat( '=', $i );
1582 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1583 }
1584 return $text;
1585 }
1586
1587 /**
1588 * Replace single quotes with HTML markup
1589 * @private
1590 *
1591 * @param string $text
1592 *
1593 * @return string The altered text
1594 */
1595 public function doAllQuotes( $text ) {
1596 $outtext = '';
1597 $lines = StringUtils::explode( "\n", $text );
1598 foreach ( $lines as $line ) {
1599 $outtext .= $this->doQuotes( $line ) . "\n";
1600 }
1601 $outtext = substr( $outtext, 0, -1 );
1602 return $outtext;
1603 }
1604
1605 /**
1606 * Helper function for doAllQuotes()
1607 *
1608 * @param string $text
1609 *
1610 * @return string
1611 */
1612 public function doQuotes( $text ) {
1613 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1614 $countarr = count( $arr );
1615 if ( $countarr == 1 ) {
1616 return $text;
1617 }
1618
1619 // First, do some preliminary work. This may shift some apostrophes from
1620 // being mark-up to being text. It also counts the number of occurrences
1621 // of bold and italics mark-ups.
1622 $numbold = 0;
1623 $numitalics = 0;
1624 for ( $i = 1; $i < $countarr; $i += 2 ) {
1625 $thislen = strlen( $arr[$i] );
1626 // If there are ever four apostrophes, assume the first is supposed to
1627 // be text, and the remaining three constitute mark-up for bold text.
1628 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1629 if ( $thislen == 4 ) {
1630 $arr[$i - 1] .= "'";
1631 $arr[$i] = "'''";
1632 $thislen = 3;
1633 } elseif ( $thislen > 5 ) {
1634 // If there are more than 5 apostrophes in a row, assume they're all
1635 // text except for the last 5.
1636 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1637 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1638 $arr[$i] = "'''''";
1639 $thislen = 5;
1640 }
1641 // Count the number of occurrences of bold and italics mark-ups.
1642 if ( $thislen == 2 ) {
1643 $numitalics++;
1644 } elseif ( $thislen == 3 ) {
1645 $numbold++;
1646 } elseif ( $thislen == 5 ) {
1647 $numitalics++;
1648 $numbold++;
1649 }
1650 }
1651
1652 // If there is an odd number of both bold and italics, it is likely
1653 // that one of the bold ones was meant to be an apostrophe followed
1654 // by italics. Which one we cannot know for certain, but it is more
1655 // likely to be one that has a single-letter word before it.
1656 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1657 $firstsingleletterword = -1;
1658 $firstmultiletterword = -1;
1659 $firstspace = -1;
1660 for ( $i = 1; $i < $countarr; $i += 2 ) {
1661 if ( strlen( $arr[$i] ) == 3 ) {
1662 $x1 = substr( $arr[$i - 1], -1 );
1663 $x2 = substr( $arr[$i - 1], -2, 1 );
1664 if ( $x1 === ' ' ) {
1665 if ( $firstspace == -1 ) {
1666 $firstspace = $i;
1667 }
1668 } elseif ( $x2 === ' ' ) {
1669 $firstsingleletterword = $i;
1670 // if $firstsingleletterword is set, we don't
1671 // look at the other options, so we can bail early.
1672 break;
1673 } else {
1674 if ( $firstmultiletterword == -1 ) {
1675 $firstmultiletterword = $i;
1676 }
1677 }
1678 }
1679 }
1680
1681 // If there is a single-letter word, use it!
1682 if ( $firstsingleletterword > -1 ) {
1683 $arr[$firstsingleletterword] = "''";
1684 $arr[$firstsingleletterword - 1] .= "'";
1685 } elseif ( $firstmultiletterword > -1 ) {
1686 // If not, but there's a multi-letter word, use that one.
1687 $arr[$firstmultiletterword] = "''";
1688 $arr[$firstmultiletterword - 1] .= "'";
1689 } elseif ( $firstspace > -1 ) {
1690 // ... otherwise use the first one that has neither.
1691 // (notice that it is possible for all three to be -1 if, for example,
1692 // there is only one pentuple-apostrophe in the line)
1693 $arr[$firstspace] = "''";
1694 $arr[$firstspace - 1] .= "'";
1695 }
1696 }
1697
1698 // Now let's actually convert our apostrophic mush to HTML!
1699 $output = '';
1700 $buffer = '';
1701 $state = '';
1702 $i = 0;
1703 foreach ( $arr as $r ) {
1704 if ( ( $i % 2 ) == 0 ) {
1705 if ( $state === 'both' ) {
1706 $buffer .= $r;
1707 } else {
1708 $output .= $r;
1709 }
1710 } else {
1711 $thislen = strlen( $r );
1712 if ( $thislen == 2 ) {
1713 if ( $state === 'i' ) {
1714 $output .= '</i>';
1715 $state = '';
1716 } elseif ( $state === 'bi' ) {
1717 $output .= '</i>';
1718 $state = 'b';
1719 } elseif ( $state === 'ib' ) {
1720 $output .= '</b></i><b>';
1721 $state = 'b';
1722 } elseif ( $state === 'both' ) {
1723 $output .= '<b><i>' . $buffer . '</i>';
1724 $state = 'b';
1725 } else { // $state can be 'b' or ''
1726 $output .= '<i>';
1727 $state .= 'i';
1728 }
1729 } elseif ( $thislen == 3 ) {
1730 if ( $state === 'b' ) {
1731 $output .= '</b>';
1732 $state = '';
1733 } elseif ( $state === 'bi' ) {
1734 $output .= '</i></b><i>';
1735 $state = 'i';
1736 } elseif ( $state === 'ib' ) {
1737 $output .= '</b>';
1738 $state = 'i';
1739 } elseif ( $state === 'both' ) {
1740 $output .= '<i><b>' . $buffer . '</b>';
1741 $state = 'i';
1742 } else { // $state can be 'i' or ''
1743 $output .= '<b>';
1744 $state .= 'b';
1745 }
1746 } elseif ( $thislen == 5 ) {
1747 if ( $state === 'b' ) {
1748 $output .= '</b><i>';
1749 $state = 'i';
1750 } elseif ( $state === 'i' ) {
1751 $output .= '</i><b>';
1752 $state = 'b';
1753 } elseif ( $state === 'bi' ) {
1754 $output .= '</i></b>';
1755 $state = '';
1756 } elseif ( $state === 'ib' ) {
1757 $output .= '</b></i>';
1758 $state = '';
1759 } elseif ( $state === 'both' ) {
1760 $output .= '<i><b>' . $buffer . '</b></i>';
1761 $state = '';
1762 } else { // ($state == '')
1763 $buffer = '';
1764 $state = 'both';
1765 }
1766 }
1767 }
1768 $i++;
1769 }
1770 // Now close all remaining tags. Notice that the order is important.
1771 if ( $state === 'b' || $state === 'ib' ) {
1772 $output .= '</b>';
1773 }
1774 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1775 $output .= '</i>';
1776 }
1777 if ( $state === 'bi' ) {
1778 $output .= '</b>';
1779 }
1780 // There might be lonely ''''', so make sure we have a buffer
1781 if ( $state === 'both' && $buffer ) {
1782 $output .= '<b><i>' . $buffer . '</i></b>';
1783 }
1784 return $output;
1785 }
1786
1787 /**
1788 * Replace external links (REL)
1789 *
1790 * Note: this is all very hackish and the order of execution matters a lot.
1791 * Make sure to run tests/parser/parserTests.php if you change this code.
1792 *
1793 * @private
1794 *
1795 * @param string $text
1796 *
1797 * @throws MWException
1798 * @return string
1799 */
1800 public function replaceExternalLinks( $text ) {
1801
1802 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1803 if ( $bits === false ) {
1804 throw new MWException( "PCRE needs to be compiled with "
1805 . "--enable-unicode-properties in order for MediaWiki to function" );
1806 }
1807 $s = array_shift( $bits );
1808
1809 $i = 0;
1810 while ( $i < count( $bits ) ) {
1811 $url = $bits[$i++];
1812 $i++; // protocol
1813 $text = $bits[$i++];
1814 $trail = $bits[$i++];
1815
1816 # The characters '<' and '>' (which were escaped by
1817 # removeHTMLtags()) should not be included in
1818 # URLs, per RFC 2396.
1819 $m2 = [];
1820 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1821 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1822 $url = substr( $url, 0, $m2[0][1] );
1823 }
1824
1825 # If the link text is an image URL, replace it with an <img> tag
1826 # This happened by accident in the original parser, but some people used it extensively
1827 $img = $this->maybeMakeExternalImage( $text );
1828 if ( $img !== false ) {
1829 $text = $img;
1830 }
1831
1832 $dtrail = '';
1833
1834 # Set linktype for CSS - if URL==text, link is essentially free
1835 $linktype = ( $text === $url ) ? 'free' : 'text';
1836
1837 # No link text, e.g. [http://domain.tld/some.link]
1838 if ( $text == '' ) {
1839 # Autonumber
1840 $langObj = $this->getTargetLanguage();
1841 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1842 $linktype = 'autonumber';
1843 } else {
1844 # Have link text, e.g. [http://domain.tld/some.link text]s
1845 # Check for trail
1846 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1847 }
1848
1849 $text = $this->getConverterLanguage()->markNoConversion( $text );
1850
1851 $url = Sanitizer::cleanUrl( $url );
1852
1853 # Use the encoded URL
1854 # This means that users can paste URLs directly into the text
1855 # Funny characters like ö aren't valid in URLs anyway
1856 # This was changed in August 2004
1857 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1858 $this->getExternalLinkAttribs( $url ), $this->mTitle ) . $dtrail . $trail;
1859
1860 # Register link in the output object.
1861 # Replace unnecessary URL escape codes with the referenced character
1862 # This prevents spammers from hiding links from the filters
1863 $pasteurized = self::normalizeLinkUrl( $url );
1864 $this->mOutput->addExternalLink( $pasteurized );
1865 }
1866
1867 return $s;
1868 }
1869
1870 /**
1871 * Get the rel attribute for a particular external link.
1872 *
1873 * @since 1.21
1874 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1875 * nofollow if appropriate
1876 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1877 * @return string|null Rel attribute for $url
1878 */
1879 public static function getExternalLinkRel( $url = false, $title = null ) {
1880 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1881 $ns = $title ? $title->getNamespace() : false;
1882 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1883 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1884 ) {
1885 return 'nofollow';
1886 }
1887 return null;
1888 }
1889
1890 /**
1891 * Get an associative array of additional HTML attributes appropriate for a
1892 * particular external link. This currently may include rel => nofollow
1893 * (depending on configuration, namespace, and the URL's domain) and/or a
1894 * target attribute (depending on configuration).
1895 *
1896 * @param string $url URL to extract the domain from for rel =>
1897 * nofollow if appropriate
1898 * @return array Associative array of HTML attributes
1899 */
1900 public function getExternalLinkAttribs( $url ) {
1901 $attribs = [];
1902 $rel = self::getExternalLinkRel( $url, $this->mTitle );
1903
1904 $target = $this->mOptions->getExternalLinkTarget();
1905 if ( $target ) {
1906 $attribs['target'] = $target;
1907 if ( !in_array( $target, [ '_self', '_parent', '_top' ] ) ) {
1908 // T133507. New windows can navigate parent cross-origin.
1909 // Including noreferrer due to lacking browser
1910 // support of noopener. Eventually noreferrer should be removed.
1911 if ( $rel !== '' ) {
1912 $rel .= ' ';
1913 }
1914 $rel .= 'noreferrer noopener';
1915 }
1916 }
1917 $attribs['rel'] = $rel;
1918 return $attribs;
1919 }
1920
1921 /**
1922 * Replace unusual escape codes in a URL with their equivalent characters
1923 *
1924 * @deprecated since 1.24, use normalizeLinkUrl
1925 * @param string $url
1926 * @return string
1927 */
1928 public static function replaceUnusualEscapes( $url ) {
1929 wfDeprecated( __METHOD__, '1.24' );
1930 return self::normalizeLinkUrl( $url );
1931 }
1932
1933 /**
1934 * Replace unusual escape codes in a URL with their equivalent characters
1935 *
1936 * This generally follows the syntax defined in RFC 3986, with special
1937 * consideration for HTTP query strings.
1938 *
1939 * @param string $url
1940 * @return string
1941 */
1942 public static function normalizeLinkUrl( $url ) {
1943 # First, make sure unsafe characters are encoded
1944 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1945 function ( $m ) {
1946 return rawurlencode( $m[0] );
1947 },
1948 $url
1949 );
1950
1951 $ret = '';
1952 $end = strlen( $url );
1953
1954 # Fragment part - 'fragment'
1955 $start = strpos( $url, '#' );
1956 if ( $start !== false && $start < $end ) {
1957 $ret = self::normalizeUrlComponent(
1958 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1959 $end = $start;
1960 }
1961
1962 # Query part - 'query' minus &=+;
1963 $start = strpos( $url, '?' );
1964 if ( $start !== false && $start < $end ) {
1965 $ret = self::normalizeUrlComponent(
1966 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1967 $end = $start;
1968 }
1969
1970 # Scheme and path part - 'pchar'
1971 # (we assume no userinfo or encoded colons in the host)
1972 $ret = self::normalizeUrlComponent(
1973 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1974
1975 return $ret;
1976 }
1977
1978 private static function normalizeUrlComponent( $component, $unsafe ) {
1979 $callback = function ( $matches ) use ( $unsafe ) {
1980 $char = urldecode( $matches[0] );
1981 $ord = ord( $char );
1982 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1983 # Unescape it
1984 return $char;
1985 } else {
1986 # Leave it escaped, but use uppercase for a-f
1987 return strtoupper( $matches[0] );
1988 }
1989 };
1990 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1991 }
1992
1993 /**
1994 * make an image if it's allowed, either through the global
1995 * option, through the exception, or through the on-wiki whitelist
1996 *
1997 * @param string $url
1998 *
1999 * @return string
2000 */
2001 private function maybeMakeExternalImage( $url ) {
2002 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
2003 $imagesexception = !empty( $imagesfrom );
2004 $text = false;
2005 # $imagesfrom could be either a single string or an array of strings, parse out the latter
2006 if ( $imagesexception && is_array( $imagesfrom ) ) {
2007 $imagematch = false;
2008 foreach ( $imagesfrom as $match ) {
2009 if ( strpos( $url, $match ) === 0 ) {
2010 $imagematch = true;
2011 break;
2012 }
2013 }
2014 } elseif ( $imagesexception ) {
2015 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
2016 } else {
2017 $imagematch = false;
2018 }
2019
2020 if ( $this->mOptions->getAllowExternalImages()
2021 || ( $imagesexception && $imagematch )
2022 ) {
2023 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
2024 # Image found
2025 $text = Linker::makeExternalImage( $url );
2026 }
2027 }
2028 if ( !$text && $this->mOptions->getEnableImageWhitelist()
2029 && preg_match( self::EXT_IMAGE_REGEX, $url )
2030 ) {
2031 $whitelist = explode(
2032 "\n",
2033 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
2034 );
2035
2036 foreach ( $whitelist as $entry ) {
2037 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
2038 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
2039 continue;
2040 }
2041 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
2042 # Image matches a whitelist entry
2043 $text = Linker::makeExternalImage( $url );
2044 break;
2045 }
2046 }
2047 }
2048 return $text;
2049 }
2050
2051 /**
2052 * Process [[ ]] wikilinks
2053 *
2054 * @param string $s
2055 *
2056 * @return string Processed text
2057 *
2058 * @private
2059 */
2060 public function replaceInternalLinks( $s ) {
2061 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
2062 return $s;
2063 }
2064
2065 /**
2066 * Process [[ ]] wikilinks (RIL)
2067 * @param string $s
2068 * @throws MWException
2069 * @return LinkHolderArray
2070 *
2071 * @private
2072 */
2073 public function replaceInternalLinks2( &$s ) {
2074 global $wgExtraInterlanguageLinkPrefixes;
2075
2076 static $tc = false, $e1, $e1_img;
2077 # the % is needed to support urlencoded titles as well
2078 if ( !$tc ) {
2079 $tc = Title::legalChars() . '#%';
2080 # Match a link having the form [[namespace:link|alternate]]trail
2081 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2082 # Match cases where there is no "]]", which might still be images
2083 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2084 }
2085
2086 $holders = new LinkHolderArray( $this );
2087
2088 # split the entire text string on occurrences of [[
2089 $a = StringUtils::explode( '[[', ' ' . $s );
2090 # get the first element (all text up to first [[), and remove the space we added
2091 $s = $a->current();
2092 $a->next();
2093 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2094 $s = substr( $s, 1 );
2095
2096 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2097 $e2 = null;
2098 if ( $useLinkPrefixExtension ) {
2099 # Match the end of a line for a word that's not followed by whitespace,
2100 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2101 global $wgContLang;
2102 $charset = $wgContLang->linkPrefixCharset();
2103 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2104 }
2105
2106 if ( is_null( $this->mTitle ) ) {
2107 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2108 }
2109 $nottalk = !$this->mTitle->isTalkPage();
2110
2111 if ( $useLinkPrefixExtension ) {
2112 $m = [];
2113 if ( preg_match( $e2, $s, $m ) ) {
2114 $first_prefix = $m[2];
2115 } else {
2116 $first_prefix = false;
2117 }
2118 } else {
2119 $prefix = '';
2120 }
2121
2122 $useSubpages = $this->areSubpagesAllowed();
2123
2124 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2125 # Loop for each link
2126 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2127 // @codingStandardsIgnoreEnd
2128
2129 # Check for excessive memory usage
2130 if ( $holders->isBig() ) {
2131 # Too big
2132 # Do the existence check, replace the link holders and clear the array
2133 $holders->replace( $s );
2134 $holders->clear();
2135 }
2136
2137 if ( $useLinkPrefixExtension ) {
2138 if ( preg_match( $e2, $s, $m ) ) {
2139 $prefix = $m[2];
2140 $s = $m[1];
2141 } else {
2142 $prefix = '';
2143 }
2144 # first link
2145 if ( $first_prefix ) {
2146 $prefix = $first_prefix;
2147 $first_prefix = false;
2148 }
2149 }
2150
2151 $might_be_img = false;
2152
2153 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2154 $text = $m[2];
2155 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2156 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2157 # the real problem is with the $e1 regex
2158 # See bug 1300.
2159 # Still some problems for cases where the ] is meant to be outside punctuation,
2160 # and no image is in sight. See bug 2095.
2161 if ( $text !== ''
2162 && substr( $m[3], 0, 1 ) === ']'
2163 && strpos( $text, '[' ) !== false
2164 ) {
2165 $text .= ']'; # so that replaceExternalLinks($text) works later
2166 $m[3] = substr( $m[3], 1 );
2167 }
2168 # fix up urlencoded title texts
2169 if ( strpos( $m[1], '%' ) !== false ) {
2170 # Should anchors '#' also be rejected?
2171 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2172 }
2173 $trail = $m[3];
2174 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2175 # Invalid, but might be an image with a link in its caption
2176 $might_be_img = true;
2177 $text = $m[2];
2178 if ( strpos( $m[1], '%' ) !== false ) {
2179 $m[1] = str_replace( [ '<', '>' ], [ '&lt;', '&gt;' ], rawurldecode( $m[1] ) );
2180 }
2181 $trail = "";
2182 } else { # Invalid form; output directly
2183 $s .= $prefix . '[[' . $line;
2184 continue;
2185 }
2186
2187 $origLink = $m[1];
2188
2189 # Don't allow internal links to pages containing
2190 # PROTO: where PROTO is a valid URL protocol; these
2191 # should be external links.
2192 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2193 $s .= $prefix . '[[' . $line;
2194 continue;
2195 }
2196
2197 # Make subpage if necessary
2198 if ( $useSubpages ) {
2199 $link = $this->maybeDoSubpageLink( $origLink, $text );
2200 } else {
2201 $link = $origLink;
2202 }
2203
2204 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2205 if ( !$noforce ) {
2206 # Strip off leading ':'
2207 $link = substr( $link, 1 );
2208 }
2209
2210 $unstrip = $this->mStripState->unstripNoWiki( $link );
2211 $nt = is_string( $unstrip ) ? Title::newFromText( $unstrip ) : null;
2212 if ( $nt === null ) {
2213 $s .= $prefix . '[[' . $line;
2214 continue;
2215 }
2216
2217 $ns = $nt->getNamespace();
2218 $iw = $nt->getInterwiki();
2219
2220 if ( $might_be_img ) { # if this is actually an invalid link
2221 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2222 $found = false;
2223 while ( true ) {
2224 # look at the next 'line' to see if we can close it there
2225 $a->next();
2226 $next_line = $a->current();
2227 if ( $next_line === false || $next_line === null ) {
2228 break;
2229 }
2230 $m = explode( ']]', $next_line, 3 );
2231 if ( count( $m ) == 3 ) {
2232 # the first ]] closes the inner link, the second the image
2233 $found = true;
2234 $text .= "[[{$m[0]}]]{$m[1]}";
2235 $trail = $m[2];
2236 break;
2237 } elseif ( count( $m ) == 2 ) {
2238 # if there's exactly one ]] that's fine, we'll keep looking
2239 $text .= "[[{$m[0]}]]{$m[1]}";
2240 } else {
2241 # if $next_line is invalid too, we need look no further
2242 $text .= '[[' . $next_line;
2243 break;
2244 }
2245 }
2246 if ( !$found ) {
2247 # we couldn't find the end of this imageLink, so output it raw
2248 # but don't ignore what might be perfectly normal links in the text we've examined
2249 $holders->merge( $this->replaceInternalLinks2( $text ) );
2250 $s .= "{$prefix}[[$link|$text";
2251 # note: no $trail, because without an end, there *is* no trail
2252 continue;
2253 }
2254 } else { # it's not an image, so output it raw
2255 $s .= "{$prefix}[[$link|$text";
2256 # note: no $trail, because without an end, there *is* no trail
2257 continue;
2258 }
2259 }
2260
2261 $wasblank = ( $text == '' );
2262 if ( $wasblank ) {
2263 $text = $link;
2264 } else {
2265 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2266 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2267 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2268 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2269 $text = $this->doQuotes( $text );
2270 }
2271
2272 # Link not escaped by : , create the various objects
2273 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2274 # Interwikis
2275 if (
2276 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2277 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2278 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2279 )
2280 ) {
2281 # Bug 24502: filter duplicates
2282 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2283 $this->mLangLinkLanguages[$iw] = true;
2284 $this->mOutput->addLanguageLink( $nt->getFullText() );
2285 }
2286
2287 $s = rtrim( $s . $prefix );
2288 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2289 continue;
2290 }
2291
2292 if ( $ns == NS_FILE ) {
2293 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2294 if ( $wasblank ) {
2295 # if no parameters were passed, $text
2296 # becomes something like "File:Foo.png",
2297 # which we don't want to pass on to the
2298 # image generator
2299 $text = '';
2300 } else {
2301 # recursively parse links inside the image caption
2302 # actually, this will parse them in any other parameters, too,
2303 # but it might be hard to fix that, and it doesn't matter ATM
2304 $text = $this->replaceExternalLinks( $text );
2305 $holders->merge( $this->replaceInternalLinks2( $text ) );
2306 }
2307 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2308 $s .= $prefix . $this->armorLinks(
2309 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2310 continue;
2311 }
2312 } elseif ( $ns == NS_CATEGORY ) {
2313 $s = rtrim( $s . "\n" ); # bug 87
2314
2315 if ( $wasblank ) {
2316 $sortkey = $this->getDefaultSort();
2317 } else {
2318 $sortkey = $text;
2319 }
2320 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2321 $sortkey = str_replace( "\n", '', $sortkey );
2322 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2323 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2324
2325 /**
2326 * Strip the whitespace Category links produce, see bug 87
2327 */
2328 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2329
2330 continue;
2331 }
2332 }
2333
2334 # Self-link checking. For some languages, variants of the title are checked in
2335 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2336 # for linking to a different variant.
2337 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2338 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2339 continue;
2340 }
2341
2342 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2343 # @todo FIXME: Should do batch file existence checks, see comment below
2344 if ( $ns == NS_MEDIA ) {
2345 # Give extensions a chance to select the file revision for us
2346 $options = [];
2347 $descQuery = false;
2348 Hooks::run( 'BeforeParserFetchFileAndTitle',
2349 [ $this, $nt, &$options, &$descQuery ] );
2350 # Fetch and register the file (file title may be different via hooks)
2351 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2352 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2353 $s .= $prefix . $this->armorLinks(
2354 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2355 continue;
2356 }
2357
2358 # Some titles, such as valid special pages or files in foreign repos, should
2359 # be shown as bluelinks even though they're not included in the page table
2360 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2361 # batch file existence checks for NS_FILE and NS_MEDIA
2362 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2363 $this->mOutput->addLink( $nt );
2364 $s .= $this->makeKnownLinkHolder( $nt, $text, $trail, $prefix );
2365 } else {
2366 # Links will be added to the output link list after checking
2367 $s .= $holders->makeHolder( $nt, $text, [], $trail, $prefix );
2368 }
2369 }
2370 return $holders;
2371 }
2372
2373 /**
2374 * Render a forced-blue link inline; protect against double expansion of
2375 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2376 * Since this little disaster has to split off the trail text to avoid
2377 * breaking URLs in the following text without breaking trails on the
2378 * wiki links, it's been made into a horrible function.
2379 *
2380 * @param Title $nt
2381 * @param string $text
2382 * @param string $trail
2383 * @param string $prefix
2384 * @return string HTML-wikitext mix oh yuck
2385 */
2386 protected function makeKnownLinkHolder( $nt, $text = '', $trail = '', $prefix = '' ) {
2387 list( $inside, $trail ) = Linker::splitTrail( $trail );
2388
2389 if ( $text == '' ) {
2390 $text = htmlspecialchars( $nt->getPrefixedText() );
2391 }
2392
2393 $link = $this->getLinkRenderer()->makeKnownLink(
2394 $nt, new HtmlArmor( "$prefix$text$inside" )
2395 );
2396
2397 return $this->armorLinks( $link ) . $trail;
2398 }
2399
2400 /**
2401 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2402 * going to go through further parsing steps before inline URL expansion.
2403 *
2404 * Not needed quite as much as it used to be since free links are a bit
2405 * more sensible these days. But bracketed links are still an issue.
2406 *
2407 * @param string $text More-or-less HTML
2408 * @return string Less-or-more HTML with NOPARSE bits
2409 */
2410 public function armorLinks( $text ) {
2411 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2412 self::MARKER_PREFIX . "NOPARSE$1", $text );
2413 }
2414
2415 /**
2416 * Return true if subpage links should be expanded on this page.
2417 * @return bool
2418 */
2419 public function areSubpagesAllowed() {
2420 # Some namespaces don't allow subpages
2421 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2422 }
2423
2424 /**
2425 * Handle link to subpage if necessary
2426 *
2427 * @param string $target The source of the link
2428 * @param string &$text The link text, modified as necessary
2429 * @return string The full name of the link
2430 * @private
2431 */
2432 public function maybeDoSubpageLink( $target, &$text ) {
2433 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2434 }
2435
2436 /**
2437 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2438 *
2439 * @param string $text
2440 * @param bool $linestart Whether or not this is at the start of a line.
2441 * @private
2442 * @return string The lists rendered as HTML
2443 */
2444 public function doBlockLevels( $text, $linestart ) {
2445 return BlockLevelPass::doBlockLevels( $text, $linestart );
2446 }
2447
2448 /**
2449 * Return value of a magic variable (like PAGENAME)
2450 *
2451 * @private
2452 *
2453 * @param int $index
2454 * @param bool|PPFrame $frame
2455 *
2456 * @throws MWException
2457 * @return string
2458 */
2459 public function getVariableValue( $index, $frame = false ) {
2460 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2461 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2462
2463 if ( is_null( $this->mTitle ) ) {
2464 // If no title set, bad things are going to happen
2465 // later. Title should always be set since this
2466 // should only be called in the middle of a parse
2467 // operation (but the unit-tests do funky stuff)
2468 throw new MWException( __METHOD__ . ' Should only be '
2469 . ' called while parsing (no title set)' );
2470 }
2471
2472 /**
2473 * Some of these require message or data lookups and can be
2474 * expensive to check many times.
2475 */
2476 if ( Hooks::run( 'ParserGetVariableValueVarCache', [ &$this, &$this->mVarCache ] ) ) {
2477 if ( isset( $this->mVarCache[$index] ) ) {
2478 return $this->mVarCache[$index];
2479 }
2480 }
2481
2482 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2483 Hooks::run( 'ParserGetVariableValueTs', [ &$this, &$ts ] );
2484
2485 $pageLang = $this->getFunctionLang();
2486
2487 switch ( $index ) {
2488 case '!':
2489 $value = '|';
2490 break;
2491 case 'currentmonth':
2492 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2493 break;
2494 case 'currentmonth1':
2495 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2496 break;
2497 case 'currentmonthname':
2498 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2499 break;
2500 case 'currentmonthnamegen':
2501 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2502 break;
2503 case 'currentmonthabbrev':
2504 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2505 break;
2506 case 'currentday':
2507 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2508 break;
2509 case 'currentday2':
2510 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2511 break;
2512 case 'localmonth':
2513 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2514 break;
2515 case 'localmonth1':
2516 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2517 break;
2518 case 'localmonthname':
2519 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2520 break;
2521 case 'localmonthnamegen':
2522 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2523 break;
2524 case 'localmonthabbrev':
2525 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2526 break;
2527 case 'localday':
2528 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2529 break;
2530 case 'localday2':
2531 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2532 break;
2533 case 'pagename':
2534 $value = wfEscapeWikiText( $this->mTitle->getText() );
2535 break;
2536 case 'pagenamee':
2537 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2538 break;
2539 case 'fullpagename':
2540 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2541 break;
2542 case 'fullpagenamee':
2543 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2544 break;
2545 case 'subpagename':
2546 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2547 break;
2548 case 'subpagenamee':
2549 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2550 break;
2551 case 'rootpagename':
2552 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2553 break;
2554 case 'rootpagenamee':
2555 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2556 ' ',
2557 '_',
2558 $this->mTitle->getRootText()
2559 ) ) );
2560 break;
2561 case 'basepagename':
2562 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2563 break;
2564 case 'basepagenamee':
2565 $value = wfEscapeWikiText( wfUrlencode( str_replace(
2566 ' ',
2567 '_',
2568 $this->mTitle->getBaseText()
2569 ) ) );
2570 break;
2571 case 'talkpagename':
2572 if ( $this->mTitle->canTalk() ) {
2573 $talkPage = $this->mTitle->getTalkPage();
2574 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2575 } else {
2576 $value = '';
2577 }
2578 break;
2579 case 'talkpagenamee':
2580 if ( $this->mTitle->canTalk() ) {
2581 $talkPage = $this->mTitle->getTalkPage();
2582 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2583 } else {
2584 $value = '';
2585 }
2586 break;
2587 case 'subjectpagename':
2588 $subjPage = $this->mTitle->getSubjectPage();
2589 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2590 break;
2591 case 'subjectpagenamee':
2592 $subjPage = $this->mTitle->getSubjectPage();
2593 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2594 break;
2595 case 'pageid': // requested in bug 23427
2596 $pageid = $this->getTitle()->getArticleID();
2597 if ( $pageid == 0 ) {
2598 # 0 means the page doesn't exist in the database,
2599 # which means the user is previewing a new page.
2600 # The vary-revision flag must be set, because the magic word
2601 # will have a different value once the page is saved.
2602 $this->mOutput->setFlag( 'vary-revision' );
2603 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2604 }
2605 $value = $pageid ? $pageid : null;
2606 break;
2607 case 'revisionid':
2608 # Let the edit saving system know we should parse the page
2609 # *after* a revision ID has been assigned.
2610 $this->mOutput->setFlag( 'vary-revision-id' );
2611 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision-id...\n" );
2612 $value = $this->mRevisionId;
2613 if ( !$value && $this->mOptions->getSpeculativeRevIdCallback() ) {
2614 $value = call_user_func( $this->mOptions->getSpeculativeRevIdCallback() );
2615 $this->mOutput->setSpeculativeRevIdUsed( $value );
2616 }
2617 break;
2618 case 'revisionday':
2619 # Let the edit saving system know we should parse the page
2620 # *after* a revision ID has been assigned. This is for null edits.
2621 $this->mOutput->setFlag( 'vary-revision' );
2622 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2623 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2624 break;
2625 case 'revisionday2':
2626 # Let the edit saving system know we should parse the page
2627 # *after* a revision ID has been assigned. This is for null edits.
2628 $this->mOutput->setFlag( 'vary-revision' );
2629 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2630 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2631 break;
2632 case 'revisionmonth':
2633 # Let the edit saving system know we should parse the page
2634 # *after* a revision ID has been assigned. This is for null edits.
2635 $this->mOutput->setFlag( 'vary-revision' );
2636 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2637 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2638 break;
2639 case 'revisionmonth1':
2640 # Let the edit saving system know we should parse the page
2641 # *after* a revision ID has been assigned. This is for null edits.
2642 $this->mOutput->setFlag( 'vary-revision' );
2643 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2644 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2645 break;
2646 case 'revisionyear':
2647 # Let the edit saving system know we should parse the page
2648 # *after* a revision ID has been assigned. This is for null edits.
2649 $this->mOutput->setFlag( 'vary-revision' );
2650 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2651 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2652 break;
2653 case 'revisiontimestamp':
2654 # Let the edit saving system know we should parse the page
2655 # *after* a revision ID has been assigned. This is for null edits.
2656 $this->mOutput->setFlag( 'vary-revision' );
2657 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2658 $value = $this->getRevisionTimestamp();
2659 break;
2660 case 'revisionuser':
2661 # Let the edit saving system know we should parse the page
2662 # *after* a revision ID has been assigned for null edits.
2663 $this->mOutput->setFlag( 'vary-user' );
2664 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-user...\n" );
2665 $value = $this->getRevisionUser();
2666 break;
2667 case 'revisionsize':
2668 $value = $this->getRevisionSize();
2669 break;
2670 case 'namespace':
2671 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2672 break;
2673 case 'namespacee':
2674 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2675 break;
2676 case 'namespacenumber':
2677 $value = $this->mTitle->getNamespace();
2678 break;
2679 case 'talkspace':
2680 $value = $this->mTitle->canTalk()
2681 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
2682 : '';
2683 break;
2684 case 'talkspacee':
2685 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2686 break;
2687 case 'subjectspace':
2688 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2689 break;
2690 case 'subjectspacee':
2691 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2692 break;
2693 case 'currentdayname':
2694 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2695 break;
2696 case 'currentyear':
2697 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2698 break;
2699 case 'currenttime':
2700 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2701 break;
2702 case 'currenthour':
2703 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2704 break;
2705 case 'currentweek':
2706 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2707 # int to remove the padding
2708 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2709 break;
2710 case 'currentdow':
2711 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2712 break;
2713 case 'localdayname':
2714 $value = $pageLang->getWeekdayName(
2715 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
2716 );
2717 break;
2718 case 'localyear':
2719 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2720 break;
2721 case 'localtime':
2722 $value = $pageLang->time(
2723 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
2724 false,
2725 false
2726 );
2727 break;
2728 case 'localhour':
2729 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2730 break;
2731 case 'localweek':
2732 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2733 # int to remove the padding
2734 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2735 break;
2736 case 'localdow':
2737 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2738 break;
2739 case 'numberofarticles':
2740 $value = $pageLang->formatNum( SiteStats::articles() );
2741 break;
2742 case 'numberoffiles':
2743 $value = $pageLang->formatNum( SiteStats::images() );
2744 break;
2745 case 'numberofusers':
2746 $value = $pageLang->formatNum( SiteStats::users() );
2747 break;
2748 case 'numberofactiveusers':
2749 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2750 break;
2751 case 'numberofpages':
2752 $value = $pageLang->formatNum( SiteStats::pages() );
2753 break;
2754 case 'numberofadmins':
2755 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2756 break;
2757 case 'numberofedits':
2758 $value = $pageLang->formatNum( SiteStats::edits() );
2759 break;
2760 case 'currenttimestamp':
2761 $value = wfTimestamp( TS_MW, $ts );
2762 break;
2763 case 'localtimestamp':
2764 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2765 break;
2766 case 'currentversion':
2767 $value = SpecialVersion::getVersion();
2768 break;
2769 case 'articlepath':
2770 return $wgArticlePath;
2771 case 'sitename':
2772 return $wgSitename;
2773 case 'server':
2774 return $wgServer;
2775 case 'servername':
2776 return $wgServerName;
2777 case 'scriptpath':
2778 return $wgScriptPath;
2779 case 'stylepath':
2780 return $wgStylePath;
2781 case 'directionmark':
2782 return $pageLang->getDirMark();
2783 case 'contentlanguage':
2784 global $wgLanguageCode;
2785 return $wgLanguageCode;
2786 case 'cascadingsources':
2787 $value = CoreParserFunctions::cascadingsources( $this );
2788 break;
2789 default:
2790 $ret = null;
2791 Hooks::run(
2792 'ParserGetVariableValueSwitch',
2793 [ &$this, &$this->mVarCache, &$index, &$ret, &$frame ]
2794 );
2795
2796 return $ret;
2797 }
2798
2799 if ( $index ) {
2800 $this->mVarCache[$index] = $value;
2801 }
2802
2803 return $value;
2804 }
2805
2806 /**
2807 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2808 *
2809 * @private
2810 */
2811 public function initialiseVariables() {
2812 $variableIDs = MagicWord::getVariableIDs();
2813 $substIDs = MagicWord::getSubstIDs();
2814
2815 $this->mVariables = new MagicWordArray( $variableIDs );
2816 $this->mSubstWords = new MagicWordArray( $substIDs );
2817 }
2818
2819 /**
2820 * Preprocess some wikitext and return the document tree.
2821 * This is the ghost of replace_variables().
2822 *
2823 * @param string $text The text to parse
2824 * @param int $flags Bitwise combination of:
2825 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
2826 * included. Default is to assume a direct page view.
2827 *
2828 * The generated DOM tree must depend only on the input text and the flags.
2829 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2830 *
2831 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2832 * change in the DOM tree for a given text, must be passed through the section identifier
2833 * in the section edit link and thus back to extractSections().
2834 *
2835 * The output of this function is currently only cached in process memory, but a persistent
2836 * cache may be implemented at a later date which takes further advantage of these strict
2837 * dependency requirements.
2838 *
2839 * @return PPNode
2840 */
2841 public function preprocessToDom( $text, $flags = 0 ) {
2842 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2843 return $dom;
2844 }
2845
2846 /**
2847 * Return a three-element array: leading whitespace, string contents, trailing whitespace
2848 *
2849 * @param string $s
2850 *
2851 * @return array
2852 */
2853 public static function splitWhitespace( $s ) {
2854 $ltrimmed = ltrim( $s );
2855 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2856 $trimmed = rtrim( $ltrimmed );
2857 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2858 if ( $diff > 0 ) {
2859 $w2 = substr( $ltrimmed, -$diff );
2860 } else {
2861 $w2 = '';
2862 }
2863 return [ $w1, $trimmed, $w2 ];
2864 }
2865
2866 /**
2867 * Replace magic variables, templates, and template arguments
2868 * with the appropriate text. Templates are substituted recursively,
2869 * taking care to avoid infinite loops.
2870 *
2871 * Note that the substitution depends on value of $mOutputType:
2872 * self::OT_WIKI: only {{subst:}} templates
2873 * self::OT_PREPROCESS: templates but not extension tags
2874 * self::OT_HTML: all templates and extension tags
2875 *
2876 * @param string $text The text to transform
2877 * @param bool|PPFrame $frame Object describing the arguments passed to the
2878 * template. Arguments may also be provided as an associative array, as
2879 * was the usual case before MW1.12. Providing arguments this way may be
2880 * useful for extensions wishing to perform variable replacement
2881 * explicitly.
2882 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
2883 * double-brace expansion.
2884 * @return string
2885 */
2886 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2887 # Is there any text? Also, Prevent too big inclusions!
2888 $textSize = strlen( $text );
2889 if ( $textSize < 1 || $textSize > $this->mOptions->getMaxIncludeSize() ) {
2890 return $text;
2891 }
2892
2893 if ( $frame === false ) {
2894 $frame = $this->getPreprocessor()->newFrame();
2895 } elseif ( !( $frame instanceof PPFrame ) ) {
2896 wfDebug( __METHOD__ . " called using plain parameters instead of "
2897 . "a PPFrame instance. Creating custom frame.\n" );
2898 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2899 }
2900
2901 $dom = $this->preprocessToDom( $text );
2902 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2903 $text = $frame->expand( $dom, $flags );
2904
2905 return $text;
2906 }
2907
2908 /**
2909 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2910 *
2911 * @param array $args
2912 *
2913 * @return array
2914 */
2915 public static function createAssocArgs( $args ) {
2916 $assocArgs = [];
2917 $index = 1;
2918 foreach ( $args as $arg ) {
2919 $eqpos = strpos( $arg, '=' );
2920 if ( $eqpos === false ) {
2921 $assocArgs[$index++] = $arg;
2922 } else {
2923 $name = trim( substr( $arg, 0, $eqpos ) );
2924 $value = trim( substr( $arg, $eqpos + 1 ) );
2925 if ( $value === false ) {
2926 $value = '';
2927 }
2928 if ( $name !== false ) {
2929 $assocArgs[$name] = $value;
2930 }
2931 }
2932 }
2933
2934 return $assocArgs;
2935 }
2936
2937 /**
2938 * Warn the user when a parser limitation is reached
2939 * Will warn at most once the user per limitation type
2940 *
2941 * The results are shown during preview and run through the Parser (See EditPage.php)
2942 *
2943 * @param string $limitationType Should be one of:
2944 * 'expensive-parserfunction' (corresponding messages:
2945 * 'expensive-parserfunction-warning',
2946 * 'expensive-parserfunction-category')
2947 * 'post-expand-template-argument' (corresponding messages:
2948 * 'post-expand-template-argument-warning',
2949 * 'post-expand-template-argument-category')
2950 * 'post-expand-template-inclusion' (corresponding messages:
2951 * 'post-expand-template-inclusion-warning',
2952 * 'post-expand-template-inclusion-category')
2953 * 'node-count-exceeded' (corresponding messages:
2954 * 'node-count-exceeded-warning',
2955 * 'node-count-exceeded-category')
2956 * 'expansion-depth-exceeded' (corresponding messages:
2957 * 'expansion-depth-exceeded-warning',
2958 * 'expansion-depth-exceeded-category')
2959 * @param string|int|null $current Current value
2960 * @param string|int|null $max Maximum allowed, when an explicit limit has been
2961 * exceeded, provide the values (optional)
2962 */
2963 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
2964 # does no harm if $current and $max are present but are unnecessary for the message
2965 # Not doing ->inLanguage( $this->mOptions->getUserLangObj() ), since this is shown
2966 # only during preview, and that would split the parser cache unnecessarily.
2967 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
2968 ->text();
2969 $this->mOutput->addWarning( $warning );
2970 $this->addTrackingCategory( "$limitationType-category" );
2971 }
2972
2973 /**
2974 * Return the text of a template, after recursively
2975 * replacing any variables or templates within the template.
2976 *
2977 * @param array $piece The parts of the template
2978 * $piece['title']: the title, i.e. the part before the |
2979 * $piece['parts']: the parameter array
2980 * $piece['lineStart']: whether the brace was at the start of a line
2981 * @param PPFrame $frame The current frame, contains template arguments
2982 * @throws Exception
2983 * @return string The text of the template
2984 */
2985 public function braceSubstitution( $piece, $frame ) {
2986
2987 // Flags
2988
2989 // $text has been filled
2990 $found = false;
2991 // wiki markup in $text should be escaped
2992 $nowiki = false;
2993 // $text is HTML, armour it against wikitext transformation
2994 $isHTML = false;
2995 // Force interwiki transclusion to be done in raw mode not rendered
2996 $forceRawInterwiki = false;
2997 // $text is a DOM node needing expansion in a child frame
2998 $isChildObj = false;
2999 // $text is a DOM node needing expansion in the current frame
3000 $isLocalObj = false;
3001
3002 # Title object, where $text came from
3003 $title = false;
3004
3005 # $part1 is the bit before the first |, and must contain only title characters.
3006 # Various prefixes will be stripped from it later.
3007 $titleWithSpaces = $frame->expand( $piece['title'] );
3008 $part1 = trim( $titleWithSpaces );
3009 $titleText = false;
3010
3011 # Original title text preserved for various purposes
3012 $originalTitle = $part1;
3013
3014 # $args is a list of argument nodes, starting from index 0, not including $part1
3015 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3016 # below won't work b/c this $args isn't an object
3017 $args = ( null == $piece['parts'] ) ? [] : $piece['parts'];
3018
3019 $profileSection = null; // profile templates
3020
3021 # SUBST
3022 if ( !$found ) {
3023 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3024
3025 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3026 # Decide whether to expand template or keep wikitext as-is.
3027 if ( $this->ot['wiki'] ) {
3028 if ( $substMatch === false ) {
3029 $literal = true; # literal when in PST with no prefix
3030 } else {
3031 $literal = false; # expand when in PST with subst: or safesubst:
3032 }
3033 } else {
3034 if ( $substMatch == 'subst' ) {
3035 $literal = true; # literal when not in PST with plain subst:
3036 } else {
3037 $literal = false; # expand when not in PST with safesubst: or no prefix
3038 }
3039 }
3040 if ( $literal ) {
3041 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3042 $isLocalObj = true;
3043 $found = true;
3044 }
3045 }
3046
3047 # Variables
3048 if ( !$found && $args->getLength() == 0 ) {
3049 $id = $this->mVariables->matchStartToEnd( $part1 );
3050 if ( $id !== false ) {
3051 $text = $this->getVariableValue( $id, $frame );
3052 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3053 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3054 }
3055 $found = true;
3056 }
3057 }
3058
3059 # MSG, MSGNW and RAW
3060 if ( !$found ) {
3061 # Check for MSGNW:
3062 $mwMsgnw = MagicWord::get( 'msgnw' );
3063 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3064 $nowiki = true;
3065 } else {
3066 # Remove obsolete MSG:
3067 $mwMsg = MagicWord::get( 'msg' );
3068 $mwMsg->matchStartAndRemove( $part1 );
3069 }
3070
3071 # Check for RAW:
3072 $mwRaw = MagicWord::get( 'raw' );
3073 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3074 $forceRawInterwiki = true;
3075 }
3076 }
3077
3078 # Parser functions
3079 if ( !$found ) {
3080 $colonPos = strpos( $part1, ':' );
3081 if ( $colonPos !== false ) {
3082 $func = substr( $part1, 0, $colonPos );
3083 $funcArgs = [ trim( substr( $part1, $colonPos + 1 ) ) ];
3084 $argsLength = $args->getLength();
3085 for ( $i = 0; $i < $argsLength; $i++ ) {
3086 $funcArgs[] = $args->item( $i );
3087 }
3088 try {
3089 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3090 } catch ( Exception $ex ) {
3091 throw $ex;
3092 }
3093
3094 # The interface for parser functions allows for extracting
3095 # flags into the local scope. Extract any forwarded flags
3096 # here.
3097 extract( $result );
3098 }
3099 }
3100
3101 # Finish mangling title and then check for loops.
3102 # Set $title to a Title object and $titleText to the PDBK
3103 if ( !$found ) {
3104 $ns = NS_TEMPLATE;
3105 # Split the title into page and subpage
3106 $subpage = '';
3107 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3108 if ( $part1 !== $relative ) {
3109 $part1 = $relative;
3110 $ns = $this->mTitle->getNamespace();
3111 }
3112 $title = Title::newFromText( $part1, $ns );
3113 if ( $title ) {
3114 $titleText = $title->getPrefixedText();
3115 # Check for language variants if the template is not found
3116 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3117 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3118 }
3119 # Do recursion depth check
3120 $limit = $this->mOptions->getMaxTemplateDepth();
3121 if ( $frame->depth >= $limit ) {
3122 $found = true;
3123 $text = '<span class="error">'
3124 . wfMessage( 'parser-template-recursion-depth-warning' )
3125 ->numParams( $limit )->inContentLanguage()->text()
3126 . '</span>';
3127 }
3128 }
3129 }
3130
3131 # Load from database
3132 if ( !$found && $title ) {
3133 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3134 if ( !$title->isExternal() ) {
3135 if ( $title->isSpecialPage()
3136 && $this->mOptions->getAllowSpecialInclusion()
3137 && $this->ot['html']
3138 ) {
3139 $specialPage = SpecialPageFactory::getPage( $title->getDBkey() );
3140 // Pass the template arguments as URL parameters.
3141 // "uselang" will have no effect since the Language object
3142 // is forced to the one defined in ParserOptions.
3143 $pageArgs = [];
3144 $argsLength = $args->getLength();
3145 for ( $i = 0; $i < $argsLength; $i++ ) {
3146 $bits = $args->item( $i )->splitArg();
3147 if ( strval( $bits['index'] ) === '' ) {
3148 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3149 $value = trim( $frame->expand( $bits['value'] ) );
3150 $pageArgs[$name] = $value;
3151 }
3152 }
3153
3154 // Create a new context to execute the special page
3155 $context = new RequestContext;
3156 $context->setTitle( $title );
3157 $context->setRequest( new FauxRequest( $pageArgs ) );
3158 if ( $specialPage && $specialPage->maxIncludeCacheTime() === 0 ) {
3159 $context->setUser( $this->getUser() );
3160 } else {
3161 // If this page is cached, then we better not be per user.
3162 $context->setUser( User::newFromName( '127.0.0.1', false ) );
3163 }
3164 $context->setLanguage( $this->mOptions->getUserLangObj() );
3165 $ret = SpecialPageFactory::capturePath(
3166 $title, $context, $this->getLinkRenderer() );
3167 if ( $ret ) {
3168 $text = $context->getOutput()->getHTML();
3169 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3170 $found = true;
3171 $isHTML = true;
3172 if ( $specialPage && $specialPage->maxIncludeCacheTime() !== false ) {
3173 $this->mOutput->updateRuntimeAdaptiveExpiry(
3174 $specialPage->maxIncludeCacheTime()
3175 );
3176 }
3177 }
3178 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3179 $found = false; # access denied
3180 wfDebug( __METHOD__ . ": template inclusion denied for " .
3181 $title->getPrefixedDBkey() . "\n" );
3182 } else {
3183 list( $text, $title ) = $this->getTemplateDom( $title );
3184 if ( $text !== false ) {
3185 $found = true;
3186 $isChildObj = true;
3187 }
3188 }
3189
3190 # If the title is valid but undisplayable, make a link to it
3191 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3192 $text = "[[:$titleText]]";
3193 $found = true;
3194 }
3195 } elseif ( $title->isTrans() ) {
3196 # Interwiki transclusion
3197 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3198 $text = $this->interwikiTransclude( $title, 'render' );
3199 $isHTML = true;
3200 } else {
3201 $text = $this->interwikiTransclude( $title, 'raw' );
3202 # Preprocess it like a template
3203 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3204 $isChildObj = true;
3205 }
3206 $found = true;
3207 }
3208
3209 # Do infinite loop check
3210 # This has to be done after redirect resolution to avoid infinite loops via redirects
3211 if ( !$frame->loopCheck( $title ) ) {
3212 $found = true;
3213 $text = '<span class="error">'
3214 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3215 . '</span>';
3216 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3217 }
3218 }
3219
3220 # If we haven't found text to substitute by now, we're done
3221 # Recover the source wikitext and return it
3222 if ( !$found ) {
3223 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3224 if ( $profileSection ) {
3225 $this->mProfiler->scopedProfileOut( $profileSection );
3226 }
3227 return [ 'object' => $text ];
3228 }
3229
3230 # Expand DOM-style return values in a child frame
3231 if ( $isChildObj ) {
3232 # Clean up argument array
3233 $newFrame = $frame->newChild( $args, $title );
3234
3235 if ( $nowiki ) {
3236 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3237 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3238 # Expansion is eligible for the empty-frame cache
3239 $text = $newFrame->cachedExpand( $titleText, $text );
3240 } else {
3241 # Uncached expansion
3242 $text = $newFrame->expand( $text );
3243 }
3244 }
3245 if ( $isLocalObj && $nowiki ) {
3246 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3247 $isLocalObj = false;
3248 }
3249
3250 if ( $profileSection ) {
3251 $this->mProfiler->scopedProfileOut( $profileSection );
3252 }
3253
3254 # Replace raw HTML by a placeholder
3255 if ( $isHTML ) {
3256 $text = $this->insertStripItem( $text );
3257 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3258 # Escape nowiki-style return values
3259 $text = wfEscapeWikiText( $text );
3260 } elseif ( is_string( $text )
3261 && !$piece['lineStart']
3262 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3263 ) {
3264 # Bug 529: if the template begins with a table or block-level
3265 # element, it should be treated as beginning a new line.
3266 # This behavior is somewhat controversial.
3267 $text = "\n" . $text;
3268 }
3269
3270 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3271 # Error, oversize inclusion
3272 if ( $titleText !== false ) {
3273 # Make a working, properly escaped link if possible (bug 23588)
3274 $text = "[[:$titleText]]";
3275 } else {
3276 # This will probably not be a working link, but at least it may
3277 # provide some hint of where the problem is
3278 preg_replace( '/^:/', '', $originalTitle );
3279 $text = "[[:$originalTitle]]";
3280 }
3281 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3282 . 'post-expand include size too large -->' );
3283 $this->limitationWarn( 'post-expand-template-inclusion' );
3284 }
3285
3286 if ( $isLocalObj ) {
3287 $ret = [ 'object' => $text ];
3288 } else {
3289 $ret = [ 'text' => $text ];
3290 }
3291
3292 return $ret;
3293 }
3294
3295 /**
3296 * Call a parser function and return an array with text and flags.
3297 *
3298 * The returned array will always contain a boolean 'found', indicating
3299 * whether the parser function was found or not. It may also contain the
3300 * following:
3301 * text: string|object, resulting wikitext or PP DOM object
3302 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3303 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3304 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3305 * nowiki: bool, wiki markup in $text should be escaped
3306 *
3307 * @since 1.21
3308 * @param PPFrame $frame The current frame, contains template arguments
3309 * @param string $function Function name
3310 * @param array $args Arguments to the function
3311 * @throws MWException
3312 * @return array
3313 */
3314 public function callParserFunction( $frame, $function, array $args = [] ) {
3315 global $wgContLang;
3316
3317 # Case sensitive functions
3318 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3319 $function = $this->mFunctionSynonyms[1][$function];
3320 } else {
3321 # Case insensitive functions
3322 $function = $wgContLang->lc( $function );
3323 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3324 $function = $this->mFunctionSynonyms[0][$function];
3325 } else {
3326 return [ 'found' => false ];
3327 }
3328 }
3329
3330 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3331
3332 # Workaround for PHP bug 35229 and similar
3333 if ( !is_callable( $callback ) ) {
3334 throw new MWException( "Tag hook for $function is not callable\n" );
3335 }
3336
3337 $allArgs = [ &$this ];
3338 if ( $flags & self::SFH_OBJECT_ARGS ) {
3339 # Convert arguments to PPNodes and collect for appending to $allArgs
3340 $funcArgs = [];
3341 foreach ( $args as $k => $v ) {
3342 if ( $v instanceof PPNode || $k === 0 ) {
3343 $funcArgs[] = $v;
3344 } else {
3345 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( [ $k => $v ] )->item( 0 );
3346 }
3347 }
3348
3349 # Add a frame parameter, and pass the arguments as an array
3350 $allArgs[] = $frame;
3351 $allArgs[] = $funcArgs;
3352 } else {
3353 # Convert arguments to plain text and append to $allArgs
3354 foreach ( $args as $k => $v ) {
3355 if ( $v instanceof PPNode ) {
3356 $allArgs[] = trim( $frame->expand( $v ) );
3357 } elseif ( is_int( $k ) && $k >= 0 ) {
3358 $allArgs[] = trim( $v );
3359 } else {
3360 $allArgs[] = trim( "$k=$v" );
3361 }
3362 }
3363 }
3364
3365 $result = call_user_func_array( $callback, $allArgs );
3366
3367 # The interface for function hooks allows them to return a wikitext
3368 # string or an array containing the string and any flags. This mungs
3369 # things around to match what this method should return.
3370 if ( !is_array( $result ) ) {
3371 $result =[
3372 'found' => true,
3373 'text' => $result,
3374 ];
3375 } else {
3376 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3377 $result['text'] = $result[0];
3378 }
3379 unset( $result[0] );
3380 $result += [
3381 'found' => true,
3382 ];
3383 }
3384
3385 $noparse = true;
3386 $preprocessFlags = 0;
3387 if ( isset( $result['noparse'] ) ) {
3388 $noparse = $result['noparse'];
3389 }
3390 if ( isset( $result['preprocessFlags'] ) ) {
3391 $preprocessFlags = $result['preprocessFlags'];
3392 }
3393
3394 if ( !$noparse ) {
3395 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3396 $result['isChildObj'] = true;
3397 }
3398
3399 return $result;
3400 }
3401
3402 /**
3403 * Get the semi-parsed DOM representation of a template with a given title,
3404 * and its redirect destination title. Cached.
3405 *
3406 * @param Title $title
3407 *
3408 * @return array
3409 */
3410 public function getTemplateDom( $title ) {
3411 $cacheTitle = $title;
3412 $titleText = $title->getPrefixedDBkey();
3413
3414 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3415 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3416 $title = Title::makeTitle( $ns, $dbk );
3417 $titleText = $title->getPrefixedDBkey();
3418 }
3419 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3420 return [ $this->mTplDomCache[$titleText], $title ];
3421 }
3422
3423 # Cache miss, go to the database
3424 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3425
3426 if ( $text === false ) {
3427 $this->mTplDomCache[$titleText] = false;
3428 return [ false, $title ];
3429 }
3430
3431 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3432 $this->mTplDomCache[$titleText] = $dom;
3433
3434 if ( !$title->equals( $cacheTitle ) ) {
3435 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3436 [ $title->getNamespace(), $cdb = $title->getDBkey() ];
3437 }
3438
3439 return [ $dom, $title ];
3440 }
3441
3442 /**
3443 * Fetch the current revision of a given title. Note that the revision
3444 * (and even the title) may not exist in the database, so everything
3445 * contributing to the output of the parser should use this method
3446 * where possible, rather than getting the revisions themselves. This
3447 * method also caches its results, so using it benefits performance.
3448 *
3449 * @since 1.24
3450 * @param Title $title
3451 * @return Revision
3452 */
3453 public function fetchCurrentRevisionOfTitle( $title ) {
3454 $cacheKey = $title->getPrefixedDBkey();
3455 if ( !$this->currentRevisionCache ) {
3456 $this->currentRevisionCache = new MapCacheLRU( 100 );
3457 }
3458 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3459 $this->currentRevisionCache->set( $cacheKey,
3460 // Defaults to Parser::statelessFetchRevision()
3461 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3462 );
3463 }
3464 return $this->currentRevisionCache->get( $cacheKey );
3465 }
3466
3467 /**
3468 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3469 * without passing them on to it.
3470 *
3471 * @since 1.24
3472 * @param Title $title
3473 * @param Parser|bool $parser
3474 * @return Revision|bool False if missing
3475 */
3476 public static function statelessFetchRevision( Title $title, $parser = false ) {
3477 $pageId = $title->getArticleID();
3478 $revId = $title->getLatestRevID();
3479
3480 $rev = Revision::newKnownCurrent( wfGetDB( DB_REPLICA ), $pageId, $revId );
3481 if ( $rev ) {
3482 $rev->setTitle( $title );
3483 }
3484
3485 return $rev;
3486 }
3487
3488 /**
3489 * Fetch the unparsed text of a template and register a reference to it.
3490 * @param Title $title
3491 * @return array ( string or false, Title )
3492 */
3493 public function fetchTemplateAndTitle( $title ) {
3494 // Defaults to Parser::statelessFetchTemplate()
3495 $templateCb = $this->mOptions->getTemplateCallback();
3496 $stuff = call_user_func( $templateCb, $title, $this );
3497 // We use U+007F DELETE to distinguish strip markers from regular text.
3498 $text = $stuff['text'];
3499 if ( is_string( $stuff['text'] ) ) {
3500 $text = strtr( $text, "\x7f", "?" );
3501 }
3502 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3503 if ( isset( $stuff['deps'] ) ) {
3504 foreach ( $stuff['deps'] as $dep ) {
3505 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3506 if ( $dep['title']->equals( $this->getTitle() ) ) {
3507 // If we transclude ourselves, the final result
3508 // will change based on the new version of the page
3509 $this->mOutput->setFlag( 'vary-revision' );
3510 }
3511 }
3512 }
3513 return [ $text, $finalTitle ];
3514 }
3515
3516 /**
3517 * Fetch the unparsed text of a template and register a reference to it.
3518 * @param Title $title
3519 * @return string|bool
3520 */
3521 public function fetchTemplate( $title ) {
3522 return $this->fetchTemplateAndTitle( $title )[0];
3523 }
3524
3525 /**
3526 * Static function to get a template
3527 * Can be overridden via ParserOptions::setTemplateCallback().
3528 *
3529 * @param Title $title
3530 * @param bool|Parser $parser
3531 *
3532 * @return array
3533 */
3534 public static function statelessFetchTemplate( $title, $parser = false ) {
3535 $text = $skip = false;
3536 $finalTitle = $title;
3537 $deps = [];
3538
3539 # Loop to fetch the article, with up to 1 redirect
3540 // @codingStandardsIgnoreStart Generic.CodeAnalysis.ForLoopWithTestFunctionCall.NotAllowed
3541 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3542 // @codingStandardsIgnoreEnd
3543 # Give extensions a chance to select the revision instead
3544 $id = false; # Assume current
3545 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3546 [ $parser, $title, &$skip, &$id ] );
3547
3548 if ( $skip ) {
3549 $text = false;
3550 $deps[] = [
3551 'title' => $title,
3552 'page_id' => $title->getArticleID(),
3553 'rev_id' => null
3554 ];
3555 break;
3556 }
3557 # Get the revision
3558 if ( $id ) {
3559 $rev = Revision::newFromId( $id );
3560 } elseif ( $parser ) {
3561 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3562 } else {
3563 $rev = Revision::newFromTitle( $title );
3564 }
3565 $rev_id = $rev ? $rev->getId() : 0;
3566 # If there is no current revision, there is no page
3567 if ( $id === false && !$rev ) {
3568 $linkCache = LinkCache::singleton();
3569 $linkCache->addBadLinkObj( $title );
3570 }
3571
3572 $deps[] = [
3573 'title' => $title,
3574 'page_id' => $title->getArticleID(),
3575 'rev_id' => $rev_id ];
3576 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3577 # We fetched a rev from a different title; register it too...
3578 $deps[] = [
3579 'title' => $rev->getTitle(),
3580 'page_id' => $rev->getPage(),
3581 'rev_id' => $rev_id ];
3582 }
3583
3584 if ( $rev ) {
3585 $content = $rev->getContent();
3586 $text = $content ? $content->getWikitextForTransclusion() : null;
3587
3588 if ( $text === false || $text === null ) {
3589 $text = false;
3590 break;
3591 }
3592 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3593 global $wgContLang;
3594 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3595 if ( !$message->exists() ) {
3596 $text = false;
3597 break;
3598 }
3599 $content = $message->content();
3600 $text = $message->plain();
3601 } else {
3602 break;
3603 }
3604 if ( !$content ) {
3605 break;
3606 }
3607 # Redirect?
3608 $finalTitle = $title;
3609 $title = $content->getRedirectTarget();
3610 }
3611 return [
3612 'text' => $text,
3613 'finalTitle' => $finalTitle,
3614 'deps' => $deps ];
3615 }
3616
3617 /**
3618 * Fetch a file and its title and register a reference to it.
3619 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3620 * @param Title $title
3621 * @param array $options Array of options to RepoGroup::findFile
3622 * @return File|bool
3623 */
3624 public function fetchFile( $title, $options = [] ) {
3625 return $this->fetchFileAndTitle( $title, $options )[0];
3626 }
3627
3628 /**
3629 * Fetch a file and its title and register a reference to it.
3630 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3631 * @param Title $title
3632 * @param array $options Array of options to RepoGroup::findFile
3633 * @return array ( File or false, Title of file )
3634 */
3635 public function fetchFileAndTitle( $title, $options = [] ) {
3636 $file = $this->fetchFileNoRegister( $title, $options );
3637
3638 $time = $file ? $file->getTimestamp() : false;
3639 $sha1 = $file ? $file->getSha1() : false;
3640 # Register the file as a dependency...
3641 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3642 if ( $file && !$title->equals( $file->getTitle() ) ) {
3643 # Update fetched file title
3644 $title = $file->getTitle();
3645 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3646 }
3647 return [ $file, $title ];
3648 }
3649
3650 /**
3651 * Helper function for fetchFileAndTitle.
3652 *
3653 * Also useful if you need to fetch a file but not use it yet,
3654 * for example to get the file's handler.
3655 *
3656 * @param Title $title
3657 * @param array $options Array of options to RepoGroup::findFile
3658 * @return File|bool
3659 */
3660 protected function fetchFileNoRegister( $title, $options = [] ) {
3661 if ( isset( $options['broken'] ) ) {
3662 $file = false; // broken thumbnail forced by hook
3663 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3664 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3665 } else { // get by (name,timestamp)
3666 $file = wfFindFile( $title, $options );
3667 }
3668 return $file;
3669 }
3670
3671 /**
3672 * Transclude an interwiki link.
3673 *
3674 * @param Title $title
3675 * @param string $action
3676 *
3677 * @return string
3678 */
3679 public function interwikiTransclude( $title, $action ) {
3680 global $wgEnableScaryTranscluding;
3681
3682 if ( !$wgEnableScaryTranscluding ) {
3683 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3684 }
3685
3686 $url = $title->getFullURL( [ 'action' => $action ] );
3687
3688 if ( strlen( $url ) > 255 ) {
3689 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3690 }
3691 return $this->fetchScaryTemplateMaybeFromCache( $url );
3692 }
3693
3694 /**
3695 * @param string $url
3696 * @return mixed|string
3697 */
3698 public function fetchScaryTemplateMaybeFromCache( $url ) {
3699 global $wgTranscludeCacheExpiry;
3700 $dbr = wfGetDB( DB_REPLICA );
3701 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3702 $obj = $dbr->selectRow( 'transcache', [ 'tc_time', 'tc_contents' ],
3703 [ 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ] );
3704 if ( $obj ) {
3705 return $obj->tc_contents;
3706 }
3707
3708 $req = MWHttpRequest::factory( $url, [], __METHOD__ );
3709 $status = $req->execute(); // Status object
3710 if ( $status->isOK() ) {
3711 $text = $req->getContent();
3712 } elseif ( $req->getStatus() != 200 ) {
3713 // Though we failed to fetch the content, this status is useless.
3714 return wfMessage( 'scarytranscludefailed-httpstatus' )
3715 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3716 } else {
3717 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3718 }
3719
3720 $dbw = wfGetDB( DB_MASTER );
3721 $dbw->replace( 'transcache', [ 'tc_url' ], [
3722 'tc_url' => $url,
3723 'tc_time' => $dbw->timestamp( time() ),
3724 'tc_contents' => $text
3725 ] );
3726 return $text;
3727 }
3728
3729 /**
3730 * Triple brace replacement -- used for template arguments
3731 * @private
3732 *
3733 * @param array $piece
3734 * @param PPFrame $frame
3735 *
3736 * @return array
3737 */
3738 public function argSubstitution( $piece, $frame ) {
3739
3740 $error = false;
3741 $parts = $piece['parts'];
3742 $nameWithSpaces = $frame->expand( $piece['title'] );
3743 $argName = trim( $nameWithSpaces );
3744 $object = false;
3745 $text = $frame->getArgument( $argName );
3746 if ( $text === false && $parts->getLength() > 0
3747 && ( $this->ot['html']
3748 || $this->ot['pre']
3749 || ( $this->ot['wiki'] && $frame->isTemplate() )
3750 )
3751 ) {
3752 # No match in frame, use the supplied default
3753 $object = $parts->item( 0 )->getChildren();
3754 }
3755 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3756 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3757 $this->limitationWarn( 'post-expand-template-argument' );
3758 }
3759
3760 if ( $text === false && $object === false ) {
3761 # No match anywhere
3762 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3763 }
3764 if ( $error !== false ) {
3765 $text .= $error;
3766 }
3767 if ( $object !== false ) {
3768 $ret = [ 'object' => $object ];
3769 } else {
3770 $ret = [ 'text' => $text ];
3771 }
3772
3773 return $ret;
3774 }
3775
3776 /**
3777 * Return the text to be used for a given extension tag.
3778 * This is the ghost of strip().
3779 *
3780 * @param array $params Associative array of parameters:
3781 * name PPNode for the tag name
3782 * attr PPNode for unparsed text where tag attributes are thought to be
3783 * attributes Optional associative array of parsed attributes
3784 * inner Contents of extension element
3785 * noClose Original text did not have a close tag
3786 * @param PPFrame $frame
3787 *
3788 * @throws MWException
3789 * @return string
3790 */
3791 public function extensionSubstitution( $params, $frame ) {
3792 static $errorStr = '<span class="error">';
3793 static $errorLen = 20;
3794
3795 $name = $frame->expand( $params['name'] );
3796 if ( substr( $name, 0, $errorLen ) === $errorStr ) {
3797 // Probably expansion depth or node count exceeded. Just punt the
3798 // error up.
3799 return $name;
3800 }
3801
3802 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3803 if ( substr( $attrText, 0, $errorLen ) === $errorStr ) {
3804 // See above
3805 return $attrText;
3806 }
3807
3808 // We can't safely check if the expansion for $content resulted in an
3809 // error, because the content could happen to be the error string
3810 // (T149622).
3811 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3812
3813 $marker = self::MARKER_PREFIX . "-$name-"
3814 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3815
3816 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3817 ( $this->ot['html'] || $this->ot['pre'] );
3818 if ( $isFunctionTag ) {
3819 $markerType = 'none';
3820 } else {
3821 $markerType = 'general';
3822 }
3823 if ( $this->ot['html'] || $isFunctionTag ) {
3824 $name = strtolower( $name );
3825 $attributes = Sanitizer::decodeTagAttributes( $attrText );
3826 if ( isset( $params['attributes'] ) ) {
3827 $attributes = $attributes + $params['attributes'];
3828 }
3829
3830 if ( isset( $this->mTagHooks[$name] ) ) {
3831 # Workaround for PHP bug 35229 and similar
3832 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3833 throw new MWException( "Tag hook for $name is not callable\n" );
3834 }
3835 $output = call_user_func_array( $this->mTagHooks[$name],
3836 [ $content, $attributes, $this, $frame ] );
3837 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3838 list( $callback, ) = $this->mFunctionTagHooks[$name];
3839 if ( !is_callable( $callback ) ) {
3840 throw new MWException( "Tag hook for $name is not callable\n" );
3841 }
3842
3843 $output = call_user_func_array( $callback, [ &$this, $frame, $content, $attributes ] );
3844 } else {
3845 $output = '<span class="error">Invalid tag extension name: ' .
3846 htmlspecialchars( $name ) . '</span>';
3847 }
3848
3849 if ( is_array( $output ) ) {
3850 # Extract flags to local scope (to override $markerType)
3851 $flags = $output;
3852 $output = $flags[0];
3853 unset( $flags[0] );
3854 extract( $flags );
3855 }
3856 } else {
3857 if ( is_null( $attrText ) ) {
3858 $attrText = '';
3859 }
3860 if ( isset( $params['attributes'] ) ) {
3861 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3862 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3863 htmlspecialchars( $attrValue ) . '"';
3864 }
3865 }
3866 if ( $content === null ) {
3867 $output = "<$name$attrText/>";
3868 } else {
3869 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3870 if ( substr( $close, 0, $errorLen ) === $errorStr ) {
3871 // See above
3872 return $close;
3873 }
3874 $output = "<$name$attrText>$content$close";
3875 }
3876 }
3877
3878 if ( $markerType === 'none' ) {
3879 return $output;
3880 } elseif ( $markerType === 'nowiki' ) {
3881 $this->mStripState->addNoWiki( $marker, $output );
3882 } elseif ( $markerType === 'general' ) {
3883 $this->mStripState->addGeneral( $marker, $output );
3884 } else {
3885 throw new MWException( __METHOD__ . ': invalid marker type' );
3886 }
3887 return $marker;
3888 }
3889
3890 /**
3891 * Increment an include size counter
3892 *
3893 * @param string $type The type of expansion
3894 * @param int $size The size of the text
3895 * @return bool False if this inclusion would take it over the maximum, true otherwise
3896 */
3897 public function incrementIncludeSize( $type, $size ) {
3898 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3899 return false;
3900 } else {
3901 $this->mIncludeSizes[$type] += $size;
3902 return true;
3903 }
3904 }
3905
3906 /**
3907 * Increment the expensive function count
3908 *
3909 * @return bool False if the limit has been exceeded
3910 */
3911 public function incrementExpensiveFunctionCount() {
3912 $this->mExpensiveFunctionCount++;
3913 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
3914 }
3915
3916 /**
3917 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3918 * Fills $this->mDoubleUnderscores, returns the modified text
3919 *
3920 * @param string $text
3921 *
3922 * @return string
3923 */
3924 public function doDoubleUnderscore( $text ) {
3925
3926 # The position of __TOC__ needs to be recorded
3927 $mw = MagicWord::get( 'toc' );
3928 if ( $mw->match( $text ) ) {
3929 $this->mShowToc = true;
3930 $this->mForceTocPosition = true;
3931
3932 # Set a placeholder. At the end we'll fill it in with the TOC.
3933 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3934
3935 # Only keep the first one.
3936 $text = $mw->replace( '', $text );
3937 }
3938
3939 # Now match and remove the rest of them
3940 $mwa = MagicWord::getDoubleUnderscoreArray();
3941 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3942
3943 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3944 $this->mOutput->mNoGallery = true;
3945 }
3946 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3947 $this->mShowToc = false;
3948 }
3949 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
3950 && $this->mTitle->getNamespace() == NS_CATEGORY
3951 ) {
3952 $this->addTrackingCategory( 'hidden-category-category' );
3953 }
3954 # (bug 8068) Allow control over whether robots index a page.
3955 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
3956 # is not desirable, the last one on the page should win.
3957 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3958 $this->mOutput->setIndexPolicy( 'noindex' );
3959 $this->addTrackingCategory( 'noindex-category' );
3960 }
3961 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3962 $this->mOutput->setIndexPolicy( 'index' );
3963 $this->addTrackingCategory( 'index-category' );
3964 }
3965
3966 # Cache all double underscores in the database
3967 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3968 $this->mOutput->setProperty( $key, '' );
3969 }
3970
3971 return $text;
3972 }
3973
3974 /**
3975 * @see ParserOutput::addTrackingCategory()
3976 * @param string $msg Message key
3977 * @return bool Whether the addition was successful
3978 */
3979 public function addTrackingCategory( $msg ) {
3980 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
3981 }
3982
3983 /**
3984 * This function accomplishes several tasks:
3985 * 1) Auto-number headings if that option is enabled
3986 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3987 * 3) Add a Table of contents on the top for users who have enabled the option
3988 * 4) Auto-anchor headings
3989 *
3990 * It loops through all headlines, collects the necessary data, then splits up the
3991 * string and re-inserts the newly formatted headlines.
3992 *
3993 * @param string $text
3994 * @param string $origText Original, untouched wikitext
3995 * @param bool $isMain
3996 * @return mixed|string
3997 * @private
3998 */
3999 public function formatHeadings( $text, $origText, $isMain = true ) {
4000 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4001
4002 # Inhibit editsection links if requested in the page
4003 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4004 $maybeShowEditLink = $showEditLink = false;
4005 } else {
4006 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4007 $showEditLink = $this->mOptions->getEditSection();
4008 }
4009 if ( $showEditLink ) {
4010 $this->mOutput->setEditSectionTokens( true );
4011 }
4012
4013 # Get all headlines for numbering them and adding funky stuff like [edit]
4014 # links - this is for later, but we need the number of headlines right now
4015 $matches = [];
4016 $numMatches = preg_match_all(
4017 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4018 $text,
4019 $matches
4020 );
4021
4022 # if there are fewer than 4 headlines in the article, do not show TOC
4023 # unless it's been explicitly enabled.
4024 $enoughToc = $this->mShowToc &&
4025 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4026
4027 # Allow user to stipulate that a page should have a "new section"
4028 # link added via __NEWSECTIONLINK__
4029 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4030 $this->mOutput->setNewSection( true );
4031 }
4032
4033 # Allow user to remove the "new section"
4034 # link via __NONEWSECTIONLINK__
4035 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4036 $this->mOutput->hideNewSection( true );
4037 }
4038
4039 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4040 # override above conditions and always show TOC above first header
4041 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4042 $this->mShowToc = true;
4043 $enoughToc = true;
4044 }
4045
4046 # headline counter
4047 $headlineCount = 0;
4048 $numVisible = 0;
4049
4050 # Ugh .. the TOC should have neat indentation levels which can be
4051 # passed to the skin functions. These are determined here
4052 $toc = '';
4053 $full = '';
4054 $head = [];
4055 $sublevelCount = [];
4056 $levelCount = [];
4057 $level = 0;
4058 $prevlevel = 0;
4059 $toclevel = 0;
4060 $prevtoclevel = 0;
4061 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4062 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4063 $oldType = $this->mOutputType;
4064 $this->setOutputType( self::OT_WIKI );
4065 $frame = $this->getPreprocessor()->newFrame();
4066 $root = $this->preprocessToDom( $origText );
4067 $node = $root->getFirstChild();
4068 $byteOffset = 0;
4069 $tocraw = [];
4070 $refers = [];
4071
4072 $headlines = $numMatches !== false ? $matches[3] : [];
4073
4074 foreach ( $headlines as $headline ) {
4075 $isTemplate = false;
4076 $titleText = false;
4077 $sectionIndex = false;
4078 $numbering = '';
4079 $markerMatches = [];
4080 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4081 $serial = $markerMatches[1];
4082 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4083 $isTemplate = ( $titleText != $baseTitleText );
4084 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4085 }
4086
4087 if ( $toclevel ) {
4088 $prevlevel = $level;
4089 }
4090 $level = $matches[1][$headlineCount];
4091
4092 if ( $level > $prevlevel ) {
4093 # Increase TOC level
4094 $toclevel++;
4095 $sublevelCount[$toclevel] = 0;
4096 if ( $toclevel < $wgMaxTocLevel ) {
4097 $prevtoclevel = $toclevel;
4098 $toc .= Linker::tocIndent();
4099 $numVisible++;
4100 }
4101 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4102 # Decrease TOC level, find level to jump to
4103
4104 for ( $i = $toclevel; $i > 0; $i-- ) {
4105 if ( $levelCount[$i] == $level ) {
4106 # Found last matching level
4107 $toclevel = $i;
4108 break;
4109 } elseif ( $levelCount[$i] < $level ) {
4110 # Found first matching level below current level
4111 $toclevel = $i + 1;
4112 break;
4113 }
4114 }
4115 if ( $i == 0 ) {
4116 $toclevel = 1;
4117 }
4118 if ( $toclevel < $wgMaxTocLevel ) {
4119 if ( $prevtoclevel < $wgMaxTocLevel ) {
4120 # Unindent only if the previous toc level was shown :p
4121 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4122 $prevtoclevel = $toclevel;
4123 } else {
4124 $toc .= Linker::tocLineEnd();
4125 }
4126 }
4127 } else {
4128 # No change in level, end TOC line
4129 if ( $toclevel < $wgMaxTocLevel ) {
4130 $toc .= Linker::tocLineEnd();
4131 }
4132 }
4133
4134 $levelCount[$toclevel] = $level;
4135
4136 # count number of headlines for each level
4137 $sublevelCount[$toclevel]++;
4138 $dot = 0;
4139 for ( $i = 1; $i <= $toclevel; $i++ ) {
4140 if ( !empty( $sublevelCount[$i] ) ) {
4141 if ( $dot ) {
4142 $numbering .= '.';
4143 }
4144 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4145 $dot = 1;
4146 }
4147 }
4148
4149 # The safe header is a version of the header text safe to use for links
4150
4151 # Remove link placeholders by the link text.
4152 # <!--LINK number-->
4153 # turns into
4154 # link text with suffix
4155 # Do this before unstrip since link text can contain strip markers
4156 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4157
4158 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4159 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4160
4161 # Strip out HTML (first regex removes any tag not allowed)
4162 # Allowed tags are:
4163 # * <sup> and <sub> (bug 8393)
4164 # * <i> (bug 26375)
4165 # * <b> (r105284)
4166 # * <bdi> (bug 72884)
4167 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4168 # * <s> and <strike> (T35715)
4169 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4170 # to allow setting directionality in toc items.
4171 $tocline = preg_replace(
4172 [
4173 '#<(?!/?(span|sup|sub|bdi|i|b|s|strike)(?: [^>]*)?>).*?>#',
4174 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b|s|strike))(?: .*?)?>#'
4175 ],
4176 [ '', '<$1>' ],
4177 $safeHeadline
4178 );
4179
4180 # Strip '<span></span>', which is the result from the above if
4181 # <span id="foo"></span> is used to produce an additional anchor
4182 # for a section.
4183 $tocline = str_replace( '<span></span>', '', $tocline );
4184
4185 $tocline = trim( $tocline );
4186
4187 # For the anchor, strip out HTML-y stuff period
4188 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4189 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4190
4191 # Save headline for section edit hint before it's escaped
4192 $headlineHint = $safeHeadline;
4193
4194 if ( $wgExperimentalHtmlIds ) {
4195 # For reverse compatibility, provide an id that's
4196 # HTML4-compatible, like we used to.
4197 # It may be worth noting, academically, that it's possible for
4198 # the legacy anchor to conflict with a non-legacy headline
4199 # anchor on the page. In this case likely the "correct" thing
4200 # would be to either drop the legacy anchors or make sure
4201 # they're numbered first. However, this would require people
4202 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4203 # manually, so let's not bother worrying about it.
4204 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4205 [ 'noninitial', 'legacy' ] );
4206 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4207
4208 if ( $legacyHeadline == $safeHeadline ) {
4209 # No reason to have both (in fact, we can't)
4210 $legacyHeadline = false;
4211 }
4212 } else {
4213 $legacyHeadline = false;
4214 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4215 'noninitial' );
4216 }
4217
4218 # HTML names must be case-insensitively unique (bug 10721).
4219 # This does not apply to Unicode characters per
4220 # https://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4221 # @todo FIXME: We may be changing them depending on the current locale.
4222 $arrayKey = strtolower( $safeHeadline );
4223 if ( $legacyHeadline === false ) {
4224 $legacyArrayKey = false;
4225 } else {
4226 $legacyArrayKey = strtolower( $legacyHeadline );
4227 }
4228
4229 # Create the anchor for linking from the TOC to the section
4230 $anchor = $safeHeadline;
4231 $legacyAnchor = $legacyHeadline;
4232 if ( isset( $refers[$arrayKey] ) ) {
4233 // @codingStandardsIgnoreStart
4234 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4235 // @codingStandardsIgnoreEnd
4236 $anchor .= "_$i";
4237 $refers["${arrayKey}_$i"] = true;
4238 } else {
4239 $refers[$arrayKey] = true;
4240 }
4241 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4242 // @codingStandardsIgnoreStart
4243 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4244 // @codingStandardsIgnoreEnd
4245 $legacyAnchor .= "_$i";
4246 $refers["${legacyArrayKey}_$i"] = true;
4247 } else {
4248 $refers[$legacyArrayKey] = true;
4249 }
4250
4251 # Don't number the heading if it is the only one (looks silly)
4252 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4253 # the two are different if the line contains a link
4254 $headline = Html::element(
4255 'span',
4256 [ 'class' => 'mw-headline-number' ],
4257 $numbering
4258 ) . ' ' . $headline;
4259 }
4260
4261 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4262 $toc .= Linker::tocLine( $anchor, $tocline,
4263 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4264 }
4265
4266 # Add the section to the section tree
4267 # Find the DOM node for this header
4268 $noOffset = ( $isTemplate || $sectionIndex === false );
4269 while ( $node && !$noOffset ) {
4270 if ( $node->getName() === 'h' ) {
4271 $bits = $node->splitHeading();
4272 if ( $bits['i'] == $sectionIndex ) {
4273 break;
4274 }
4275 }
4276 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4277 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4278 $node = $node->getNextSibling();
4279 }
4280 $tocraw[] = [
4281 'toclevel' => $toclevel,
4282 'level' => $level,
4283 'line' => $tocline,
4284 'number' => $numbering,
4285 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4286 'fromtitle' => $titleText,
4287 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4288 'anchor' => $anchor,
4289 ];
4290
4291 # give headline the correct <h#> tag
4292 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4293 // Output edit section links as markers with styles that can be customized by skins
4294 if ( $isTemplate ) {
4295 # Put a T flag in the section identifier, to indicate to extractSections()
4296 # that sections inside <includeonly> should be counted.
4297 $editsectionPage = $titleText;
4298 $editsectionSection = "T-$sectionIndex";
4299 $editsectionContent = null;
4300 } else {
4301 $editsectionPage = $this->mTitle->getPrefixedText();
4302 $editsectionSection = $sectionIndex;
4303 $editsectionContent = $headlineHint;
4304 }
4305 // We use a bit of pesudo-xml for editsection markers. The
4306 // language converter is run later on. Using a UNIQ style marker
4307 // leads to the converter screwing up the tokens when it
4308 // converts stuff. And trying to insert strip tags fails too. At
4309 // this point all real inputted tags have already been escaped,
4310 // so we don't have to worry about a user trying to input one of
4311 // these markers directly. We use a page and section attribute
4312 // to stop the language converter from converting these
4313 // important bits of data, but put the headline hint inside a
4314 // content block because the language converter is supposed to
4315 // be able to convert that piece of data.
4316 // Gets replaced with html in ParserOutput::getText
4317 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4318 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4319 if ( $editsectionContent !== null ) {
4320 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4321 } else {
4322 $editlink .= '/>';
4323 }
4324 } else {
4325 $editlink = '';
4326 }
4327 $head[$headlineCount] = Linker::makeHeadline( $level,
4328 $matches['attrib'][$headlineCount], $anchor, $headline,
4329 $editlink, $legacyAnchor );
4330
4331 $headlineCount++;
4332 }
4333
4334 $this->setOutputType( $oldType );
4335
4336 # Never ever show TOC if no headers
4337 if ( $numVisible < 1 ) {
4338 $enoughToc = false;
4339 }
4340
4341 if ( $enoughToc ) {
4342 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4343 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4344 }
4345 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4346 $this->mOutput->setTOCHTML( $toc );
4347 $toc = self::TOC_START . $toc . self::TOC_END;
4348 $this->mOutput->addModules( 'mediawiki.toc' );
4349 }
4350
4351 if ( $isMain ) {
4352 $this->mOutput->setSections( $tocraw );
4353 }
4354
4355 # split up and insert constructed headlines
4356 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4357 $i = 0;
4358
4359 // build an array of document sections
4360 $sections = [];
4361 foreach ( $blocks as $block ) {
4362 // $head is zero-based, sections aren't.
4363 if ( empty( $head[$i - 1] ) ) {
4364 $sections[$i] = $block;
4365 } else {
4366 $sections[$i] = $head[$i - 1] . $block;
4367 }
4368
4369 /**
4370 * Send a hook, one per section.
4371 * The idea here is to be able to make section-level DIVs, but to do so in a
4372 * lower-impact, more correct way than r50769
4373 *
4374 * $this : caller
4375 * $section : the section number
4376 * &$sectionContent : ref to the content of the section
4377 * $showEditLinks : boolean describing whether this section has an edit link
4378 */
4379 Hooks::run( 'ParserSectionCreate', [ $this, $i, &$sections[$i], $showEditLink ] );
4380
4381 $i++;
4382 }
4383
4384 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4385 // append the TOC at the beginning
4386 // Top anchor now in skin
4387 $sections[0] = $sections[0] . $toc . "\n";
4388 }
4389
4390 $full .= implode( '', $sections );
4391
4392 if ( $this->mForceTocPosition ) {
4393 return str_replace( '<!--MWTOC-->', $toc, $full );
4394 } else {
4395 return $full;
4396 }
4397 }
4398
4399 /**
4400 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4401 * conversion, substituting signatures, {{subst:}} templates, etc.
4402 *
4403 * @param string $text The text to transform
4404 * @param Title $title The Title object for the current article
4405 * @param User $user The User object describing the current user
4406 * @param ParserOptions $options Parsing options
4407 * @param bool $clearState Whether to clear the parser state first
4408 * @return string The altered wiki markup
4409 */
4410 public function preSaveTransform( $text, Title $title, User $user,
4411 ParserOptions $options, $clearState = true
4412 ) {
4413 if ( $clearState ) {
4414 $magicScopeVariable = $this->lock();
4415 }
4416 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4417 $this->setUser( $user );
4418
4419 // We still normalize line endings for backwards-compatibility
4420 // with other code that just calls PST, but this should already
4421 // be handled in TextContent subclasses
4422 $text = TextContent::normalizeLineEndings( $text );
4423
4424 if ( $options->getPreSaveTransform() ) {
4425 $text = $this->pstPass2( $text, $user );
4426 }
4427 $text = $this->mStripState->unstripBoth( $text );
4428
4429 $this->setUser( null ); # Reset
4430
4431 return $text;
4432 }
4433
4434 /**
4435 * Pre-save transform helper function
4436 *
4437 * @param string $text
4438 * @param User $user
4439 *
4440 * @return string
4441 */
4442 private function pstPass2( $text, $user ) {
4443 global $wgContLang;
4444
4445 # Note: This is the timestamp saved as hardcoded wikitext to
4446 # the database, we use $wgContLang here in order to give
4447 # everyone the same signature and use the default one rather
4448 # than the one selected in each user's preferences.
4449 # (see also bug 12815)
4450 $ts = $this->mOptions->getTimestamp();
4451 $timestamp = MWTimestamp::getLocalInstance( $ts );
4452 $ts = $timestamp->format( 'YmdHis' );
4453 $tzMsg = $timestamp->getTimezoneMessage()->inContentLanguage()->text();
4454
4455 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4456
4457 # Variable replacement
4458 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4459 $text = $this->replaceVariables( $text );
4460
4461 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4462 # which may corrupt this parser instance via its wfMessage()->text() call-
4463
4464 # Signatures
4465 $sigText = $this->getUserSig( $user );
4466 $text = strtr( $text, [
4467 '~~~~~' => $d,
4468 '~~~~' => "$sigText $d",
4469 '~~~' => $sigText
4470 ] );
4471
4472 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4473 $tc = '[' . Title::legalChars() . ']';
4474 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4475
4476 // [[ns:page (context)|]]
4477 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4478 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4479 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4480 // [[ns:page (context), context|]] (using either single or double-width comma)
4481 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4482 // [[|page]] (reverse pipe trick: add context from page title)
4483 $p2 = "/\[\[\\|($tc+)]]/";
4484
4485 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4486 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4487 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4488 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4489
4490 $t = $this->mTitle->getText();
4491 $m = [];
4492 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4493 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4494 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4495 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4496 } else {
4497 # if there's no context, don't bother duplicating the title
4498 $text = preg_replace( $p2, '[[\\1]]', $text );
4499 }
4500
4501 return $text;
4502 }
4503
4504 /**
4505 * Fetch the user's signature text, if any, and normalize to
4506 * validated, ready-to-insert wikitext.
4507 * If you have pre-fetched the nickname or the fancySig option, you can
4508 * specify them here to save a database query.
4509 * Do not reuse this parser instance after calling getUserSig(),
4510 * as it may have changed if it's the $wgParser.
4511 *
4512 * @param User $user
4513 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4514 * @param bool|null $fancySig whether the nicknname is the complete signature
4515 * or null to use default value
4516 * @return string
4517 */
4518 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4519 global $wgMaxSigChars;
4520
4521 $username = $user->getName();
4522
4523 # If not given, retrieve from the user object.
4524 if ( $nickname === false ) {
4525 $nickname = $user->getOption( 'nickname' );
4526 }
4527
4528 if ( is_null( $fancySig ) ) {
4529 $fancySig = $user->getBoolOption( 'fancysig' );
4530 }
4531
4532 $nickname = $nickname == null ? $username : $nickname;
4533
4534 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4535 $nickname = $username;
4536 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4537 } elseif ( $fancySig !== false ) {
4538 # Sig. might contain markup; validate this
4539 if ( $this->validateSig( $nickname ) !== false ) {
4540 # Validated; clean up (if needed) and return it
4541 return $this->cleanSig( $nickname, true );
4542 } else {
4543 # Failed to validate; fall back to the default
4544 $nickname = $username;
4545 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4546 }
4547 }
4548
4549 # Make sure nickname doesnt get a sig in a sig
4550 $nickname = self::cleanSigInSig( $nickname );
4551
4552 # If we're still here, make it a link to the user page
4553 $userText = wfEscapeWikiText( $username );
4554 $nickText = wfEscapeWikiText( $nickname );
4555 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4556
4557 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4558 ->title( $this->getTitle() )->text();
4559 }
4560
4561 /**
4562 * Check that the user's signature contains no bad XML
4563 *
4564 * @param string $text
4565 * @return string|bool An expanded string, or false if invalid.
4566 */
4567 public function validateSig( $text ) {
4568 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4569 }
4570
4571 /**
4572 * Clean up signature text
4573 *
4574 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4575 * 2) Substitute all transclusions
4576 *
4577 * @param string $text
4578 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4579 * @return string Signature text
4580 */
4581 public function cleanSig( $text, $parsing = false ) {
4582 if ( !$parsing ) {
4583 global $wgTitle;
4584 $magicScopeVariable = $this->lock();
4585 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4586 }
4587
4588 # Option to disable this feature
4589 if ( !$this->mOptions->getCleanSignatures() ) {
4590 return $text;
4591 }
4592
4593 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4594 # => Move this logic to braceSubstitution()
4595 $substWord = MagicWord::get( 'subst' );
4596 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4597 $substText = '{{' . $substWord->getSynonym( 0 );
4598
4599 $text = preg_replace( $substRegex, $substText, $text );
4600 $text = self::cleanSigInSig( $text );
4601 $dom = $this->preprocessToDom( $text );
4602 $frame = $this->getPreprocessor()->newFrame();
4603 $text = $frame->expand( $dom );
4604
4605 if ( !$parsing ) {
4606 $text = $this->mStripState->unstripBoth( $text );
4607 }
4608
4609 return $text;
4610 }
4611
4612 /**
4613 * Strip 3, 4 or 5 tildes out of signatures.
4614 *
4615 * @param string $text
4616 * @return string Signature text with /~{3,5}/ removed
4617 */
4618 public static function cleanSigInSig( $text ) {
4619 $text = preg_replace( '/~{3,5}/', '', $text );
4620 return $text;
4621 }
4622
4623 /**
4624 * Set up some variables which are usually set up in parse()
4625 * so that an external function can call some class members with confidence
4626 *
4627 * @param Title|null $title
4628 * @param ParserOptions $options
4629 * @param int $outputType
4630 * @param bool $clearState
4631 */
4632 public function startExternalParse( Title $title = null, ParserOptions $options,
4633 $outputType, $clearState = true
4634 ) {
4635 $this->startParse( $title, $options, $outputType, $clearState );
4636 }
4637
4638 /**
4639 * @param Title|null $title
4640 * @param ParserOptions $options
4641 * @param int $outputType
4642 * @param bool $clearState
4643 */
4644 private function startParse( Title $title = null, ParserOptions $options,
4645 $outputType, $clearState = true
4646 ) {
4647 $this->setTitle( $title );
4648 $this->mOptions = $options;
4649 $this->setOutputType( $outputType );
4650 if ( $clearState ) {
4651 $this->clearState();
4652 }
4653 }
4654
4655 /**
4656 * Wrapper for preprocess()
4657 *
4658 * @param string $text The text to preprocess
4659 * @param ParserOptions $options Options
4660 * @param Title|null $title Title object or null to use $wgTitle
4661 * @return string
4662 */
4663 public function transformMsg( $text, $options, $title = null ) {
4664 static $executing = false;
4665
4666 # Guard against infinite recursion
4667 if ( $executing ) {
4668 return $text;
4669 }
4670 $executing = true;
4671
4672 if ( !$title ) {
4673 global $wgTitle;
4674 $title = $wgTitle;
4675 }
4676
4677 $text = $this->preprocess( $text, $title, $options );
4678
4679 $executing = false;
4680 return $text;
4681 }
4682
4683 /**
4684 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4685 * The callback should have the following form:
4686 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4687 *
4688 * Transform and return $text. Use $parser for any required context, e.g. use
4689 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4690 *
4691 * Hooks may return extended information by returning an array, of which the
4692 * first numbered element (index 0) must be the return string, and all other
4693 * entries are extracted into local variables within an internal function
4694 * in the Parser class.
4695 *
4696 * This interface (introduced r61913) appears to be undocumented, but
4697 * 'markerType' is used by some core tag hooks to override which strip
4698 * array their results are placed in. **Use great caution if attempting
4699 * this interface, as it is not documented and injudicious use could smash
4700 * private variables.**
4701 *
4702 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4703 * @param callable $callback The callback function (and object) to use for the tag
4704 * @throws MWException
4705 * @return callable|null The old value of the mTagHooks array associated with the hook
4706 */
4707 public function setHook( $tag, $callback ) {
4708 $tag = strtolower( $tag );
4709 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4710 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4711 }
4712 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4713 $this->mTagHooks[$tag] = $callback;
4714 if ( !in_array( $tag, $this->mStripList ) ) {
4715 $this->mStripList[] = $tag;
4716 }
4717
4718 return $oldVal;
4719 }
4720
4721 /**
4722 * As setHook(), but letting the contents be parsed.
4723 *
4724 * Transparent tag hooks are like regular XML-style tag hooks, except they
4725 * operate late in the transformation sequence, on HTML instead of wikitext.
4726 *
4727 * This is probably obsoleted by things dealing with parser frames?
4728 * The only extension currently using it is geoserver.
4729 *
4730 * @since 1.10
4731 * @todo better document or deprecate this
4732 *
4733 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4734 * @param callable $callback The callback function (and object) to use for the tag
4735 * @throws MWException
4736 * @return callable|null The old value of the mTagHooks array associated with the hook
4737 */
4738 public function setTransparentTagHook( $tag, $callback ) {
4739 $tag = strtolower( $tag );
4740 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4741 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4742 }
4743 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4744 $this->mTransparentTagHooks[$tag] = $callback;
4745
4746 return $oldVal;
4747 }
4748
4749 /**
4750 * Remove all tag hooks
4751 */
4752 public function clearTagHooks() {
4753 $this->mTagHooks = [];
4754 $this->mFunctionTagHooks = [];
4755 $this->mStripList = $this->mDefaultStripList;
4756 }
4757
4758 /**
4759 * Create a function, e.g. {{sum:1|2|3}}
4760 * The callback function should have the form:
4761 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4762 *
4763 * Or with Parser::SFH_OBJECT_ARGS:
4764 * function myParserFunction( $parser, $frame, $args ) { ... }
4765 *
4766 * The callback may either return the text result of the function, or an array with the text
4767 * in element 0, and a number of flags in the other elements. The names of the flags are
4768 * specified in the keys. Valid flags are:
4769 * found The text returned is valid, stop processing the template. This
4770 * is on by default.
4771 * nowiki Wiki markup in the return value should be escaped
4772 * isHTML The returned text is HTML, armour it against wikitext transformation
4773 *
4774 * @param string $id The magic word ID
4775 * @param callable $callback The callback function (and object) to use
4776 * @param int $flags A combination of the following flags:
4777 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4778 *
4779 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
4780 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
4781 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
4782 * the arguments, and to control the way they are expanded.
4783 *
4784 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4785 * arguments, for instance:
4786 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4787 *
4788 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4789 * future versions. Please call $frame->expand() on it anyway so that your code keeps
4790 * working if/when this is changed.
4791 *
4792 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4793 * expansion.
4794 *
4795 * Please read the documentation in includes/parser/Preprocessor.php for more information
4796 * about the methods available in PPFrame and PPNode.
4797 *
4798 * @throws MWException
4799 * @return string|callable The old callback function for this name, if any
4800 */
4801 public function setFunctionHook( $id, $callback, $flags = 0 ) {
4802 global $wgContLang;
4803
4804 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4805 $this->mFunctionHooks[$id] = [ $callback, $flags ];
4806
4807 # Add to function cache
4808 $mw = MagicWord::get( $id );
4809 if ( !$mw ) {
4810 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4811 }
4812
4813 $synonyms = $mw->getSynonyms();
4814 $sensitive = intval( $mw->isCaseSensitive() );
4815
4816 foreach ( $synonyms as $syn ) {
4817 # Case
4818 if ( !$sensitive ) {
4819 $syn = $wgContLang->lc( $syn );
4820 }
4821 # Add leading hash
4822 if ( !( $flags & self::SFH_NO_HASH ) ) {
4823 $syn = '#' . $syn;
4824 }
4825 # Remove trailing colon
4826 if ( substr( $syn, -1, 1 ) === ':' ) {
4827 $syn = substr( $syn, 0, -1 );
4828 }
4829 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4830 }
4831 return $oldVal;
4832 }
4833
4834 /**
4835 * Get all registered function hook identifiers
4836 *
4837 * @return array
4838 */
4839 public function getFunctionHooks() {
4840 return array_keys( $this->mFunctionHooks );
4841 }
4842
4843 /**
4844 * Create a tag function, e.g. "<test>some stuff</test>".
4845 * Unlike tag hooks, tag functions are parsed at preprocessor level.
4846 * Unlike parser functions, their content is not preprocessed.
4847 * @param string $tag
4848 * @param callable $callback
4849 * @param int $flags
4850 * @throws MWException
4851 * @return null
4852 */
4853 public function setFunctionTagHook( $tag, $callback, $flags ) {
4854 $tag = strtolower( $tag );
4855 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4856 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4857 }
4858 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4859 $this->mFunctionTagHooks[$tag] : null;
4860 $this->mFunctionTagHooks[$tag] = [ $callback, $flags ];
4861
4862 if ( !in_array( $tag, $this->mStripList ) ) {
4863 $this->mStripList[] = $tag;
4864 }
4865
4866 return $old;
4867 }
4868
4869 /**
4870 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
4871 * Placeholders created in Linker::link()
4872 *
4873 * @param string $text
4874 * @param int $options
4875 */
4876 public function replaceLinkHolders( &$text, $options = 0 ) {
4877 $this->mLinkHolders->replace( $text );
4878 }
4879
4880 /**
4881 * Replace "<!--LINK-->" link placeholders with plain text of links
4882 * (not HTML-formatted).
4883 *
4884 * @param string $text
4885 * @return string
4886 */
4887 public function replaceLinkHoldersText( $text ) {
4888 return $this->mLinkHolders->replaceText( $text );
4889 }
4890
4891 /**
4892 * Renders an image gallery from a text with one line per image.
4893 * text labels may be given by using |-style alternative text. E.g.
4894 * Image:one.jpg|The number "1"
4895 * Image:tree.jpg|A tree
4896 * given as text will return the HTML of a gallery with two images,
4897 * labeled 'The number "1"' and
4898 * 'A tree'.
4899 *
4900 * @param string $text
4901 * @param array $params
4902 * @return string HTML
4903 */
4904 public function renderImageGallery( $text, $params ) {
4905
4906 $mode = false;
4907 if ( isset( $params['mode'] ) ) {
4908 $mode = $params['mode'];
4909 }
4910
4911 try {
4912 $ig = ImageGalleryBase::factory( $mode );
4913 } catch ( Exception $e ) {
4914 // If invalid type set, fallback to default.
4915 $ig = ImageGalleryBase::factory( false );
4916 }
4917
4918 $ig->setContextTitle( $this->mTitle );
4919 $ig->setShowBytes( false );
4920 $ig->setShowFilename( false );
4921 $ig->setParser( $this );
4922 $ig->setHideBadImages();
4923 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4924
4925 if ( isset( $params['showfilename'] ) ) {
4926 $ig->setShowFilename( true );
4927 } else {
4928 $ig->setShowFilename( false );
4929 }
4930 if ( isset( $params['caption'] ) ) {
4931 $caption = $params['caption'];
4932 $caption = htmlspecialchars( $caption );
4933 $caption = $this->replaceInternalLinks( $caption );
4934 $ig->setCaptionHtml( $caption );
4935 }
4936 if ( isset( $params['perrow'] ) ) {
4937 $ig->setPerRow( $params['perrow'] );
4938 }
4939 if ( isset( $params['widths'] ) ) {
4940 $ig->setWidths( $params['widths'] );
4941 }
4942 if ( isset( $params['heights'] ) ) {
4943 $ig->setHeights( $params['heights'] );
4944 }
4945 $ig->setAdditionalOptions( $params );
4946
4947 Hooks::run( 'BeforeParserrenderImageGallery', [ &$this, &$ig ] );
4948
4949 $lines = StringUtils::explode( "\n", $text );
4950 foreach ( $lines as $line ) {
4951 # match lines like these:
4952 # Image:someimage.jpg|This is some image
4953 $matches = [];
4954 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4955 # Skip empty lines
4956 if ( count( $matches ) == 0 ) {
4957 continue;
4958 }
4959
4960 if ( strpos( $matches[0], '%' ) !== false ) {
4961 $matches[1] = rawurldecode( $matches[1] );
4962 }
4963 $title = Title::newFromText( $matches[1], NS_FILE );
4964 if ( is_null( $title ) ) {
4965 # Bogus title. Ignore these so we don't bomb out later.
4966 continue;
4967 }
4968
4969 # We need to get what handler the file uses, to figure out parameters.
4970 # Note, a hook can overide the file name, and chose an entirely different
4971 # file (which potentially could be of a different type and have different handler).
4972 $options = [];
4973 $descQuery = false;
4974 Hooks::run( 'BeforeParserFetchFileAndTitle',
4975 [ $this, $title, &$options, &$descQuery ] );
4976 # Don't register it now, as ImageGallery does that later.
4977 $file = $this->fetchFileNoRegister( $title, $options );
4978 $handler = $file ? $file->getHandler() : false;
4979
4980 $paramMap = [
4981 'img_alt' => 'gallery-internal-alt',
4982 'img_link' => 'gallery-internal-link',
4983 ];
4984 if ( $handler ) {
4985 $paramMap = $paramMap + $handler->getParamMap();
4986 // We don't want people to specify per-image widths.
4987 // Additionally the width parameter would need special casing anyhow.
4988 unset( $paramMap['img_width'] );
4989 }
4990
4991 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
4992
4993 $label = '';
4994 $alt = '';
4995 $link = '';
4996 $handlerOptions = [];
4997 if ( isset( $matches[3] ) ) {
4998 // look for an |alt= definition while trying not to break existing
4999 // captions with multiple pipes (|) in it, until a more sensible grammar
5000 // is defined for images in galleries
5001
5002 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5003 // splitting on '|' is a bit odd, and different from makeImage.
5004 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5005 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5006
5007 foreach ( $parameterMatches as $parameterMatch ) {
5008 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5009 if ( $magicName ) {
5010 $paramName = $paramMap[$magicName];
5011
5012 switch ( $paramName ) {
5013 case 'gallery-internal-alt':
5014 $alt = $this->stripAltText( $match, false );
5015 break;
5016 case 'gallery-internal-link':
5017 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5018 $chars = self::EXT_LINK_URL_CLASS;
5019 $addr = self::EXT_LINK_ADDR;
5020 $prots = $this->mUrlProtocols;
5021 // check to see if link matches an absolute url, if not then it must be a wiki link.
5022 if ( preg_match( "/^($prots)$addr$chars*$/u", $linkValue ) ) {
5023 $link = $linkValue;
5024 } else {
5025 $localLinkTitle = Title::newFromText( $linkValue );
5026 if ( $localLinkTitle !== null ) {
5027 $link = $localLinkTitle->getLinkURL();
5028 }
5029 }
5030 break;
5031 default:
5032 // Must be a handler specific parameter.
5033 if ( $handler->validateParam( $paramName, $match ) ) {
5034 $handlerOptions[$paramName] = $match;
5035 } else {
5036 // Guess not, consider it as caption.
5037 wfDebug( "$parameterMatch failed parameter validation\n" );
5038 $label = '|' . $parameterMatch;
5039 }
5040 }
5041
5042 } else {
5043 // Last pipe wins.
5044 $label = '|' . $parameterMatch;
5045 }
5046 }
5047 // Remove the pipe.
5048 $label = substr( $label, 1 );
5049 }
5050
5051 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5052 }
5053 $html = $ig->toHTML();
5054 Hooks::run( 'AfterParserFetchFileAndTitle', [ $this, $ig, &$html ] );
5055 return $html;
5056 }
5057
5058 /**
5059 * @param MediaHandler $handler
5060 * @return array
5061 */
5062 public function getImageParams( $handler ) {
5063 if ( $handler ) {
5064 $handlerClass = get_class( $handler );
5065 } else {
5066 $handlerClass = '';
5067 }
5068 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5069 # Initialise static lists
5070 static $internalParamNames = [
5071 'horizAlign' => [ 'left', 'right', 'center', 'none' ],
5072 'vertAlign' => [ 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5073 'bottom', 'text-bottom' ],
5074 'frame' => [ 'thumbnail', 'manualthumb', 'framed', 'frameless',
5075 'upright', 'border', 'link', 'alt', 'class' ],
5076 ];
5077 static $internalParamMap;
5078 if ( !$internalParamMap ) {
5079 $internalParamMap = [];
5080 foreach ( $internalParamNames as $type => $names ) {
5081 foreach ( $names as $name ) {
5082 $magicName = str_replace( '-', '_', "img_$name" );
5083 $internalParamMap[$magicName] = [ $type, $name ];
5084 }
5085 }
5086 }
5087
5088 # Add handler params
5089 $paramMap = $internalParamMap;
5090 if ( $handler ) {
5091 $handlerParamMap = $handler->getParamMap();
5092 foreach ( $handlerParamMap as $magic => $paramName ) {
5093 $paramMap[$magic] = [ 'handler', $paramName ];
5094 }
5095 }
5096 $this->mImageParams[$handlerClass] = $paramMap;
5097 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5098 }
5099 return [ $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] ];
5100 }
5101
5102 /**
5103 * Parse image options text and use it to make an image
5104 *
5105 * @param Title $title
5106 * @param string $options
5107 * @param LinkHolderArray|bool $holders
5108 * @return string HTML
5109 */
5110 public function makeImage( $title, $options, $holders = false ) {
5111 # Check if the options text is of the form "options|alt text"
5112 # Options are:
5113 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5114 # * left no resizing, just left align. label is used for alt= only
5115 # * right same, but right aligned
5116 # * none same, but not aligned
5117 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5118 # * center center the image
5119 # * frame Keep original image size, no magnify-button.
5120 # * framed Same as "frame"
5121 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5122 # * upright reduce width for upright images, rounded to full __0 px
5123 # * border draw a 1px border around the image
5124 # * alt Text for HTML alt attribute (defaults to empty)
5125 # * class Set a class for img node
5126 # * link Set the target of the image link. Can be external, interwiki, or local
5127 # vertical-align values (no % or length right now):
5128 # * baseline
5129 # * sub
5130 # * super
5131 # * top
5132 # * text-top
5133 # * middle
5134 # * bottom
5135 # * text-bottom
5136
5137 $parts = StringUtils::explode( "|", $options );
5138
5139 # Give extensions a chance to select the file revision for us
5140 $options = [];
5141 $descQuery = false;
5142 Hooks::run( 'BeforeParserFetchFileAndTitle',
5143 [ $this, $title, &$options, &$descQuery ] );
5144 # Fetch and register the file (file title may be different via hooks)
5145 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5146
5147 # Get parameter map
5148 $handler = $file ? $file->getHandler() : false;
5149
5150 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5151
5152 if ( !$file ) {
5153 $this->addTrackingCategory( 'broken-file-category' );
5154 }
5155
5156 # Process the input parameters
5157 $caption = '';
5158 $params = [ 'frame' => [], 'handler' => [],
5159 'horizAlign' => [], 'vertAlign' => [] ];
5160 $seenformat = false;
5161 foreach ( $parts as $part ) {
5162 $part = trim( $part );
5163 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5164 $validated = false;
5165 if ( isset( $paramMap[$magicName] ) ) {
5166 list( $type, $paramName ) = $paramMap[$magicName];
5167
5168 # Special case; width and height come in one variable together
5169 if ( $type === 'handler' && $paramName === 'width' ) {
5170 $parsedWidthParam = $this->parseWidthParam( $value );
5171 if ( isset( $parsedWidthParam['width'] ) ) {
5172 $width = $parsedWidthParam['width'];
5173 if ( $handler->validateParam( 'width', $width ) ) {
5174 $params[$type]['width'] = $width;
5175 $validated = true;
5176 }
5177 }
5178 if ( isset( $parsedWidthParam['height'] ) ) {
5179 $height = $parsedWidthParam['height'];
5180 if ( $handler->validateParam( 'height', $height ) ) {
5181 $params[$type]['height'] = $height;
5182 $validated = true;
5183 }
5184 }
5185 # else no validation -- bug 13436
5186 } else {
5187 if ( $type === 'handler' ) {
5188 # Validate handler parameter
5189 $validated = $handler->validateParam( $paramName, $value );
5190 } else {
5191 # Validate internal parameters
5192 switch ( $paramName ) {
5193 case 'manualthumb':
5194 case 'alt':
5195 case 'class':
5196 # @todo FIXME: Possibly check validity here for
5197 # manualthumb? downstream behavior seems odd with
5198 # missing manual thumbs.
5199 $validated = true;
5200 $value = $this->stripAltText( $value, $holders );
5201 break;
5202 case 'link':
5203 $chars = self::EXT_LINK_URL_CLASS;
5204 $addr = self::EXT_LINK_ADDR;
5205 $prots = $this->mUrlProtocols;
5206 if ( $value === '' ) {
5207 $paramName = 'no-link';
5208 $value = true;
5209 $validated = true;
5210 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5211 if ( preg_match( "/^((?i)$prots)$addr$chars*$/u", $value, $m ) ) {
5212 $paramName = 'link-url';
5213 $this->mOutput->addExternalLink( $value );
5214 if ( $this->mOptions->getExternalLinkTarget() ) {
5215 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5216 }
5217 $validated = true;
5218 }
5219 } else {
5220 $linkTitle = Title::newFromText( $value );
5221 if ( $linkTitle ) {
5222 $paramName = 'link-title';
5223 $value = $linkTitle;
5224 $this->mOutput->addLink( $linkTitle );
5225 $validated = true;
5226 }
5227 }
5228 break;
5229 case 'frameless':
5230 case 'framed':
5231 case 'thumbnail':
5232 // use first appearing option, discard others.
5233 $validated = ! $seenformat;
5234 $seenformat = true;
5235 break;
5236 default:
5237 # Most other things appear to be empty or numeric...
5238 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5239 }
5240 }
5241
5242 if ( $validated ) {
5243 $params[$type][$paramName] = $value;
5244 }
5245 }
5246 }
5247 if ( !$validated ) {
5248 $caption = $part;
5249 }
5250 }
5251
5252 # Process alignment parameters
5253 if ( $params['horizAlign'] ) {
5254 $params['frame']['align'] = key( $params['horizAlign'] );
5255 }
5256 if ( $params['vertAlign'] ) {
5257 $params['frame']['valign'] = key( $params['vertAlign'] );
5258 }
5259
5260 $params['frame']['caption'] = $caption;
5261
5262 # Will the image be presented in a frame, with the caption below?
5263 $imageIsFramed = isset( $params['frame']['frame'] )
5264 || isset( $params['frame']['framed'] )
5265 || isset( $params['frame']['thumbnail'] )
5266 || isset( $params['frame']['manualthumb'] );
5267
5268 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5269 # came to also set the caption, ordinary text after the image -- which
5270 # makes no sense, because that just repeats the text multiple times in
5271 # screen readers. It *also* came to set the title attribute.
5272 # Now that we have an alt attribute, we should not set the alt text to
5273 # equal the caption: that's worse than useless, it just repeats the
5274 # text. This is the framed/thumbnail case. If there's no caption, we
5275 # use the unnamed parameter for alt text as well, just for the time be-
5276 # ing, if the unnamed param is set and the alt param is not.
5277 # For the future, we need to figure out if we want to tweak this more,
5278 # e.g., introducing a title= parameter for the title; ignoring the un-
5279 # named parameter entirely for images without a caption; adding an ex-
5280 # plicit caption= parameter and preserving the old magic unnamed para-
5281 # meter for BC; ...
5282 if ( $imageIsFramed ) { # Framed image
5283 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5284 # No caption or alt text, add the filename as the alt text so
5285 # that screen readers at least get some description of the image
5286 $params['frame']['alt'] = $title->getText();
5287 }
5288 # Do not set $params['frame']['title'] because tooltips don't make sense
5289 # for framed images
5290 } else { # Inline image
5291 if ( !isset( $params['frame']['alt'] ) ) {
5292 # No alt text, use the "caption" for the alt text
5293 if ( $caption !== '' ) {
5294 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5295 } else {
5296 # No caption, fall back to using the filename for the
5297 # alt text
5298 $params['frame']['alt'] = $title->getText();
5299 }
5300 }
5301 # Use the "caption" for the tooltip text
5302 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5303 }
5304
5305 Hooks::run( 'ParserMakeImageParams', [ $title, $file, &$params, $this ] );
5306
5307 # Linker does the rest
5308 $time = isset( $options['time'] ) ? $options['time'] : false;
5309 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5310 $time, $descQuery, $this->mOptions->getThumbSize() );
5311
5312 # Give the handler a chance to modify the parser object
5313 if ( $handler ) {
5314 $handler->parserTransformHook( $this, $file );
5315 }
5316
5317 return $ret;
5318 }
5319
5320 /**
5321 * @param string $caption
5322 * @param LinkHolderArray|bool $holders
5323 * @return mixed|string
5324 */
5325 protected function stripAltText( $caption, $holders ) {
5326 # Strip bad stuff out of the title (tooltip). We can't just use
5327 # replaceLinkHoldersText() here, because if this function is called
5328 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5329 if ( $holders ) {
5330 $tooltip = $holders->replaceText( $caption );
5331 } else {
5332 $tooltip = $this->replaceLinkHoldersText( $caption );
5333 }
5334
5335 # make sure there are no placeholders in thumbnail attributes
5336 # that are later expanded to html- so expand them now and
5337 # remove the tags
5338 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5339 $tooltip = Sanitizer::stripAllTags( $tooltip );
5340
5341 return $tooltip;
5342 }
5343
5344 /**
5345 * Set a flag in the output object indicating that the content is dynamic and
5346 * shouldn't be cached.
5347 * @deprecated since 1.28; use getOutput()->updateCacheExpiry()
5348 */
5349 public function disableCache() {
5350 wfDebug( "Parser output marked as uncacheable.\n" );
5351 if ( !$this->mOutput ) {
5352 throw new MWException( __METHOD__ .
5353 " can only be called when actually parsing something" );
5354 }
5355 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5356 }
5357
5358 /**
5359 * Callback from the Sanitizer for expanding items found in HTML attribute
5360 * values, so they can be safely tested and escaped.
5361 *
5362 * @param string $text
5363 * @param bool|PPFrame $frame
5364 * @return string
5365 */
5366 public function attributeStripCallback( &$text, $frame = false ) {
5367 $text = $this->replaceVariables( $text, $frame );
5368 $text = $this->mStripState->unstripBoth( $text );
5369 return $text;
5370 }
5371
5372 /**
5373 * Accessor
5374 *
5375 * @return array
5376 */
5377 public function getTags() {
5378 return array_merge(
5379 array_keys( $this->mTransparentTagHooks ),
5380 array_keys( $this->mTagHooks ),
5381 array_keys( $this->mFunctionTagHooks )
5382 );
5383 }
5384
5385 /**
5386 * Replace transparent tags in $text with the values given by the callbacks.
5387 *
5388 * Transparent tag hooks are like regular XML-style tag hooks, except they
5389 * operate late in the transformation sequence, on HTML instead of wikitext.
5390 *
5391 * @param string $text
5392 *
5393 * @return string
5394 */
5395 public function replaceTransparentTags( $text ) {
5396 $matches = [];
5397 $elements = array_keys( $this->mTransparentTagHooks );
5398 $text = self::extractTagsAndParams( $elements, $text, $matches );
5399 $replacements = [];
5400
5401 foreach ( $matches as $marker => $data ) {
5402 list( $element, $content, $params, $tag ) = $data;
5403 $tagName = strtolower( $element );
5404 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5405 $output = call_user_func_array(
5406 $this->mTransparentTagHooks[$tagName],
5407 [ $content, $params, $this ]
5408 );
5409 } else {
5410 $output = $tag;
5411 }
5412 $replacements[$marker] = $output;
5413 }
5414 return strtr( $text, $replacements );
5415 }
5416
5417 /**
5418 * Break wikitext input into sections, and either pull or replace
5419 * some particular section's text.
5420 *
5421 * External callers should use the getSection and replaceSection methods.
5422 *
5423 * @param string $text Page wikitext
5424 * @param string|number $sectionId A section identifier string of the form:
5425 * "<flag1> - <flag2> - ... - <section number>"
5426 *
5427 * Currently the only recognised flag is "T", which means the target section number
5428 * was derived during a template inclusion parse, in other words this is a template
5429 * section edit link. If no flags are given, it was an ordinary section edit link.
5430 * This flag is required to avoid a section numbering mismatch when a section is
5431 * enclosed by "<includeonly>" (bug 6563).
5432 *
5433 * The section number 0 pulls the text before the first heading; other numbers will
5434 * pull the given section along with its lower-level subsections. If the section is
5435 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5436 *
5437 * Section 0 is always considered to exist, even if it only contains the empty
5438 * string. If $text is the empty string and section 0 is replaced, $newText is
5439 * returned.
5440 *
5441 * @param string $mode One of "get" or "replace"
5442 * @param string $newText Replacement text for section data.
5443 * @return string For "get", the extracted section text.
5444 * for "replace", the whole page with the section replaced.
5445 */
5446 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5447 global $wgTitle; # not generally used but removes an ugly failure mode
5448
5449 $magicScopeVariable = $this->lock();
5450 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5451 $outText = '';
5452 $frame = $this->getPreprocessor()->newFrame();
5453
5454 # Process section extraction flags
5455 $flags = 0;
5456 $sectionParts = explode( '-', $sectionId );
5457 $sectionIndex = array_pop( $sectionParts );
5458 foreach ( $sectionParts as $part ) {
5459 if ( $part === 'T' ) {
5460 $flags |= self::PTD_FOR_INCLUSION;
5461 }
5462 }
5463
5464 # Check for empty input
5465 if ( strval( $text ) === '' ) {
5466 # Only sections 0 and T-0 exist in an empty document
5467 if ( $sectionIndex == 0 ) {
5468 if ( $mode === 'get' ) {
5469 return '';
5470 } else {
5471 return $newText;
5472 }
5473 } else {
5474 if ( $mode === 'get' ) {
5475 return $newText;
5476 } else {
5477 return $text;
5478 }
5479 }
5480 }
5481
5482 # Preprocess the text
5483 $root = $this->preprocessToDom( $text, $flags );
5484
5485 # <h> nodes indicate section breaks
5486 # They can only occur at the top level, so we can find them by iterating the root's children
5487 $node = $root->getFirstChild();
5488
5489 # Find the target section
5490 if ( $sectionIndex == 0 ) {
5491 # Section zero doesn't nest, level=big
5492 $targetLevel = 1000;
5493 } else {
5494 while ( $node ) {
5495 if ( $node->getName() === 'h' ) {
5496 $bits = $node->splitHeading();
5497 if ( $bits['i'] == $sectionIndex ) {
5498 $targetLevel = $bits['level'];
5499 break;
5500 }
5501 }
5502 if ( $mode === 'replace' ) {
5503 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5504 }
5505 $node = $node->getNextSibling();
5506 }
5507 }
5508
5509 if ( !$node ) {
5510 # Not found
5511 if ( $mode === 'get' ) {
5512 return $newText;
5513 } else {
5514 return $text;
5515 }
5516 }
5517
5518 # Find the end of the section, including nested sections
5519 do {
5520 if ( $node->getName() === 'h' ) {
5521 $bits = $node->splitHeading();
5522 $curLevel = $bits['level'];
5523 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5524 break;
5525 }
5526 }
5527 if ( $mode === 'get' ) {
5528 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5529 }
5530 $node = $node->getNextSibling();
5531 } while ( $node );
5532
5533 # Write out the remainder (in replace mode only)
5534 if ( $mode === 'replace' ) {
5535 # Output the replacement text
5536 # Add two newlines on -- trailing whitespace in $newText is conventionally
5537 # stripped by the editor, so we need both newlines to restore the paragraph gap
5538 # Only add trailing whitespace if there is newText
5539 if ( $newText != "" ) {
5540 $outText .= $newText . "\n\n";
5541 }
5542
5543 while ( $node ) {
5544 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5545 $node = $node->getNextSibling();
5546 }
5547 }
5548
5549 if ( is_string( $outText ) ) {
5550 # Re-insert stripped tags
5551 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5552 }
5553
5554 return $outText;
5555 }
5556
5557 /**
5558 * This function returns the text of a section, specified by a number ($section).
5559 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5560 * the first section before any such heading (section 0).
5561 *
5562 * If a section contains subsections, these are also returned.
5563 *
5564 * @param string $text Text to look in
5565 * @param string|number $sectionId Section identifier as a number or string
5566 * (e.g. 0, 1 or 'T-1').
5567 * @param string $defaultText Default to return if section is not found
5568 *
5569 * @return string Text of the requested section
5570 */
5571 public function getSection( $text, $sectionId, $defaultText = '' ) {
5572 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5573 }
5574
5575 /**
5576 * This function returns $oldtext after the content of the section
5577 * specified by $section has been replaced with $text. If the target
5578 * section does not exist, $oldtext is returned unchanged.
5579 *
5580 * @param string $oldText Former text of the article
5581 * @param string|number $sectionId Section identifier as a number or string
5582 * (e.g. 0, 1 or 'T-1').
5583 * @param string $newText Replacing text
5584 *
5585 * @return string Modified text
5586 */
5587 public function replaceSection( $oldText, $sectionId, $newText ) {
5588 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5589 }
5590
5591 /**
5592 * Get the ID of the revision we are parsing
5593 *
5594 * @return int|null
5595 */
5596 public function getRevisionId() {
5597 return $this->mRevisionId;
5598 }
5599
5600 /**
5601 * Get the revision object for $this->mRevisionId
5602 *
5603 * @return Revision|null Either a Revision object or null
5604 * @since 1.23 (public since 1.23)
5605 */
5606 public function getRevisionObject() {
5607 if ( !is_null( $this->mRevisionObject ) ) {
5608 return $this->mRevisionObject;
5609 }
5610 if ( is_null( $this->mRevisionId ) ) {
5611 return null;
5612 }
5613
5614 $rev = call_user_func(
5615 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5616 );
5617
5618 # If the parse is for a new revision, then the callback should have
5619 # already been set to force the object and should match mRevisionId.
5620 # If not, try to fetch by mRevisionId for sanity.
5621 if ( $rev && $rev->getId() != $this->mRevisionId ) {
5622 $rev = Revision::newFromId( $this->mRevisionId );
5623 }
5624
5625 $this->mRevisionObject = $rev;
5626
5627 return $this->mRevisionObject;
5628 }
5629
5630 /**
5631 * Get the timestamp associated with the current revision, adjusted for
5632 * the default server-local timestamp
5633 * @return string
5634 */
5635 public function getRevisionTimestamp() {
5636 if ( is_null( $this->mRevisionTimestamp ) ) {
5637 global $wgContLang;
5638
5639 $revObject = $this->getRevisionObject();
5640 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5641
5642 # The cryptic '' timezone parameter tells to use the site-default
5643 # timezone offset instead of the user settings.
5644 # Since this value will be saved into the parser cache, served
5645 # to other users, and potentially even used inside links and such,
5646 # it needs to be consistent for all visitors.
5647 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5648
5649 }
5650 return $this->mRevisionTimestamp;
5651 }
5652
5653 /**
5654 * Get the name of the user that edited the last revision
5655 *
5656 * @return string User name
5657 */
5658 public function getRevisionUser() {
5659 if ( is_null( $this->mRevisionUser ) ) {
5660 $revObject = $this->getRevisionObject();
5661
5662 # if this template is subst: the revision id will be blank,
5663 # so just use the current user's name
5664 if ( $revObject ) {
5665 $this->mRevisionUser = $revObject->getUserText();
5666 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5667 $this->mRevisionUser = $this->getUser()->getName();
5668 }
5669 }
5670 return $this->mRevisionUser;
5671 }
5672
5673 /**
5674 * Get the size of the revision
5675 *
5676 * @return int|null Revision size
5677 */
5678 public function getRevisionSize() {
5679 if ( is_null( $this->mRevisionSize ) ) {
5680 $revObject = $this->getRevisionObject();
5681
5682 # if this variable is subst: the revision id will be blank,
5683 # so just use the parser input size, because the own substituation
5684 # will change the size.
5685 if ( $revObject ) {
5686 $this->mRevisionSize = $revObject->getSize();
5687 } else {
5688 $this->mRevisionSize = $this->mInputSize;
5689 }
5690 }
5691 return $this->mRevisionSize;
5692 }
5693
5694 /**
5695 * Mutator for $mDefaultSort
5696 *
5697 * @param string $sort New value
5698 */
5699 public function setDefaultSort( $sort ) {
5700 $this->mDefaultSort = $sort;
5701 $this->mOutput->setProperty( 'defaultsort', $sort );
5702 }
5703
5704 /**
5705 * Accessor for $mDefaultSort
5706 * Will use the empty string if none is set.
5707 *
5708 * This value is treated as a prefix, so the
5709 * empty string is equivalent to sorting by
5710 * page name.
5711 *
5712 * @return string
5713 */
5714 public function getDefaultSort() {
5715 if ( $this->mDefaultSort !== false ) {
5716 return $this->mDefaultSort;
5717 } else {
5718 return '';
5719 }
5720 }
5721
5722 /**
5723 * Accessor for $mDefaultSort
5724 * Unlike getDefaultSort(), will return false if none is set
5725 *
5726 * @return string|bool
5727 */
5728 public function getCustomDefaultSort() {
5729 return $this->mDefaultSort;
5730 }
5731
5732 /**
5733 * Try to guess the section anchor name based on a wikitext fragment
5734 * presumably extracted from a heading, for example "Header" from
5735 * "== Header ==".
5736 *
5737 * @param string $text
5738 *
5739 * @return string
5740 */
5741 public function guessSectionNameFromWikiText( $text ) {
5742 # Strip out wikitext links(they break the anchor)
5743 $text = $this->stripSectionName( $text );
5744 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5745 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5746 }
5747
5748 /**
5749 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5750 * instead. For use in redirects, since IE6 interprets Redirect: headers
5751 * as something other than UTF-8 (apparently?), resulting in breakage.
5752 *
5753 * @param string $text The section name
5754 * @return string An anchor
5755 */
5756 public function guessLegacySectionNameFromWikiText( $text ) {
5757 # Strip out wikitext links(they break the anchor)
5758 $text = $this->stripSectionName( $text );
5759 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5760 return '#' . Sanitizer::escapeId( $text, [ 'noninitial', 'legacy' ] );
5761 }
5762
5763 /**
5764 * Strips a text string of wikitext for use in a section anchor
5765 *
5766 * Accepts a text string and then removes all wikitext from the
5767 * string and leaves only the resultant text (i.e. the result of
5768 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5769 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5770 * to create valid section anchors by mimicing the output of the
5771 * parser when headings are parsed.
5772 *
5773 * @param string $text Text string to be stripped of wikitext
5774 * for use in a Section anchor
5775 * @return string Filtered text string
5776 */
5777 public function stripSectionName( $text ) {
5778 # Strip internal link markup
5779 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5780 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5781
5782 # Strip external link markup
5783 # @todo FIXME: Not tolerant to blank link text
5784 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
5785 # on how many empty links there are on the page - need to figure that out.
5786 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5787
5788 # Parse wikitext quotes (italics & bold)
5789 $text = $this->doQuotes( $text );
5790
5791 # Strip HTML tags
5792 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5793 return $text;
5794 }
5795
5796 /**
5797 * strip/replaceVariables/unstrip for preprocessor regression testing
5798 *
5799 * @param string $text
5800 * @param Title $title
5801 * @param ParserOptions $options
5802 * @param int $outputType
5803 *
5804 * @return string
5805 */
5806 public function testSrvus( $text, Title $title, ParserOptions $options,
5807 $outputType = self::OT_HTML
5808 ) {
5809 $magicScopeVariable = $this->lock();
5810 $this->startParse( $title, $options, $outputType, true );
5811
5812 $text = $this->replaceVariables( $text );
5813 $text = $this->mStripState->unstripBoth( $text );
5814 $text = Sanitizer::removeHTMLtags( $text );
5815 return $text;
5816 }
5817
5818 /**
5819 * @param string $text
5820 * @param Title $title
5821 * @param ParserOptions $options
5822 * @return string
5823 */
5824 public function testPst( $text, Title $title, ParserOptions $options ) {
5825 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5826 }
5827
5828 /**
5829 * @param string $text
5830 * @param Title $title
5831 * @param ParserOptions $options
5832 * @return string
5833 */
5834 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
5835 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5836 }
5837
5838 /**
5839 * Call a callback function on all regions of the given text that are not
5840 * inside strip markers, and replace those regions with the return value
5841 * of the callback. For example, with input:
5842 *
5843 * aaa<MARKER>bbb
5844 *
5845 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5846 * two strings will be replaced with the value returned by the callback in
5847 * each case.
5848 *
5849 * @param string $s
5850 * @param callable $callback
5851 *
5852 * @return string
5853 */
5854 public function markerSkipCallback( $s, $callback ) {
5855 $i = 0;
5856 $out = '';
5857 while ( $i < strlen( $s ) ) {
5858 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
5859 if ( $markerStart === false ) {
5860 $out .= call_user_func( $callback, substr( $s, $i ) );
5861 break;
5862 } else {
5863 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5864 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5865 if ( $markerEnd === false ) {
5866 $out .= substr( $s, $markerStart );
5867 break;
5868 } else {
5869 $markerEnd += strlen( self::MARKER_SUFFIX );
5870 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5871 $i = $markerEnd;
5872 }
5873 }
5874 }
5875 return $out;
5876 }
5877
5878 /**
5879 * Remove any strip markers found in the given text.
5880 *
5881 * @param string $text Input string
5882 * @return string
5883 */
5884 public function killMarkers( $text ) {
5885 return $this->mStripState->killMarkers( $text );
5886 }
5887
5888 /**
5889 * Save the parser state required to convert the given half-parsed text to
5890 * HTML. "Half-parsed" in this context means the output of
5891 * recursiveTagParse() or internalParse(). This output has strip markers
5892 * from replaceVariables (extensionSubstitution() etc.), and link
5893 * placeholders from replaceLinkHolders().
5894 *
5895 * Returns an array which can be serialized and stored persistently. This
5896 * array can later be loaded into another parser instance with
5897 * unserializeHalfParsedText(). The text can then be safely incorporated into
5898 * the return value of a parser hook.
5899 *
5900 * @param string $text
5901 *
5902 * @return array
5903 */
5904 public function serializeHalfParsedText( $text ) {
5905 $data = [
5906 'text' => $text,
5907 'version' => self::HALF_PARSED_VERSION,
5908 'stripState' => $this->mStripState->getSubState( $text ),
5909 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5910 ];
5911 return $data;
5912 }
5913
5914 /**
5915 * Load the parser state given in the $data array, which is assumed to
5916 * have been generated by serializeHalfParsedText(). The text contents is
5917 * extracted from the array, and its markers are transformed into markers
5918 * appropriate for the current Parser instance. This transformed text is
5919 * returned, and can be safely included in the return value of a parser
5920 * hook.
5921 *
5922 * If the $data array has been stored persistently, the caller should first
5923 * check whether it is still valid, by calling isValidHalfParsedText().
5924 *
5925 * @param array $data Serialized data
5926 * @throws MWException
5927 * @return string
5928 */
5929 public function unserializeHalfParsedText( $data ) {
5930 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5931 throw new MWException( __METHOD__ . ': invalid version' );
5932 }
5933
5934 # First, extract the strip state.
5935 $texts = [ $data['text'] ];
5936 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5937
5938 # Now renumber links
5939 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5940
5941 # Should be good to go.
5942 return $texts[0];
5943 }
5944
5945 /**
5946 * Returns true if the given array, presumed to be generated by
5947 * serializeHalfParsedText(), is compatible with the current version of the
5948 * parser.
5949 *
5950 * @param array $data
5951 *
5952 * @return bool
5953 */
5954 public function isValidHalfParsedText( $data ) {
5955 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5956 }
5957
5958 /**
5959 * Parsed a width param of imagelink like 300px or 200x300px
5960 *
5961 * @param string $value
5962 *
5963 * @return array
5964 * @since 1.20
5965 */
5966 public function parseWidthParam( $value ) {
5967 $parsedWidthParam = [];
5968 if ( $value === '' ) {
5969 return $parsedWidthParam;
5970 }
5971 $m = [];
5972 # (bug 13500) In both cases (width/height and width only),
5973 # permit trailing "px" for backward compatibility.
5974 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5975 $width = intval( $m[1] );
5976 $height = intval( $m[2] );
5977 $parsedWidthParam['width'] = $width;
5978 $parsedWidthParam['height'] = $height;
5979 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5980 $width = intval( $value );
5981 $parsedWidthParam['width'] = $width;
5982 }
5983 return $parsedWidthParam;
5984 }
5985
5986 /**
5987 * Lock the current instance of the parser.
5988 *
5989 * This is meant to stop someone from calling the parser
5990 * recursively and messing up all the strip state.
5991 *
5992 * @throws MWException If parser is in a parse
5993 * @return ScopedCallback The lock will be released once the return value goes out of scope.
5994 */
5995 protected function lock() {
5996 if ( $this->mInParse ) {
5997 throw new MWException( "Parser state cleared while parsing. "
5998 . "Did you call Parser::parse recursively?" );
5999 }
6000 $this->mInParse = true;
6001
6002 $recursiveCheck = new ScopedCallback( function() {
6003 $this->mInParse = false;
6004 } );
6005
6006 return $recursiveCheck;
6007 }
6008
6009 /**
6010 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6011 *
6012 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6013 * or if there is more than one <p/> tag in the input HTML.
6014 *
6015 * @param string $html
6016 * @return string
6017 * @since 1.24
6018 */
6019 public static function stripOuterParagraph( $html ) {
6020 $m = [];
6021 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6022 if ( strpos( $m[1], '</p>' ) === false ) {
6023 $html = $m[1];
6024 }
6025 }
6026
6027 return $html;
6028 }
6029
6030 /**
6031 * Return this parser if it is not doing anything, otherwise
6032 * get a fresh parser. You can use this method by doing
6033 * $myParser = $wgParser->getFreshParser(), or more simply
6034 * $wgParser->getFreshParser()->parse( ... );
6035 * if you're unsure if $wgParser is safe to use.
6036 *
6037 * @since 1.24
6038 * @return Parser A parser object that is not parsing anything
6039 */
6040 public function getFreshParser() {
6041 global $wgParserConf;
6042 if ( $this->mInParse ) {
6043 return new $wgParserConf['class']( $wgParserConf );
6044 } else {
6045 return $this;
6046 }
6047 }
6048
6049 /**
6050 * Set's up the PHP implementation of OOUI for use in this request
6051 * and instructs OutputPage to enable OOUI for itself.
6052 *
6053 * @since 1.26
6054 */
6055 public function enableOOUI() {
6056 OutputPage::setupOOUI();
6057 $this->mOutput->setEnableOOUI( true );
6058 }
6059 }
Wiklou, le Wiki du Wiklou