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