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