dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Include edit summary in permalink"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23
24 /**
25 * @defgroup Parser Parser
26 */
27
28 /**
29 * PHP Parser - Processes wiki markup (which uses a more user-friendly
30 * syntax, such as "[[link]]" for making links), and provides a one-way
31 * transformation of that wiki markup it into (X)HTML output / markup
32 * (which in turn the browser understands, and can display).
33 *
34 * There are seven main entry points into the Parser class:
35 *
36 * - Parser::parse()
37 * produces HTML output
38 * - Parser::preSaveTransform().
39 * produces altered wiki markup.
40 * - Parser::preprocess()
41 * removes HTML comments and expands templates
42 * - Parser::cleanSig() and Parser::cleanSigInSig()
43 * Cleans a signature before saving it to preferences
44 * - Parser::getSection()
45 * Return the content of a section from an article for section editing
46 * - Parser::replaceSection()
47 * Replaces a section by number inside an article
48 * - Parser::getPreloadText()
49 * Removes <noinclude> sections, and <includeonly> tags.
50 *
51 * Globals used:
52 * object: $wgContLang
53 *
54 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
55 *
56 * @par Settings:
57 * $wgNamespacesWithSubpages
58 *
59 * @par Settings only within ParserOptions:
60 * $wgAllowExternalImages
61 * $wgAllowSpecialInclusion
62 * $wgInterwikiMagic
63 * $wgMaxArticleSize
64 *
65 * @ingroup Parser
66 */
67 class Parser {
68 /**
69 * Update this version number when the ParserOutput format
70 * changes in an incompatible way, so the parser cache
71 * can automatically discard old data.
72 */
73 const VERSION = '1.6.4';
74
75 /**
76 * Update this version number when the output of serialiseHalfParsedText()
77 * changes in an incompatible way
78 */
79 const HALF_PARSED_VERSION = 2;
80
81 # Flags for Parser::setFunctionHook
82 # Also available as global constants from Defines.php
83 const SFH_NO_HASH = 1;
84 const SFH_OBJECT_ARGS = 2;
85
86 # Constants needed for external link processing
87 # Everything except bracket, space, or control characters
88 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
89 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
90 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
91 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
92 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
93
94 # State constants for the definition list colon extraction
95 const COLON_STATE_TEXT = 0;
96 const COLON_STATE_TAG = 1;
97 const COLON_STATE_TAGSTART = 2;
98 const COLON_STATE_CLOSETAG = 3;
99 const COLON_STATE_TAGSLASH = 4;
100 const COLON_STATE_COMMENT = 5;
101 const COLON_STATE_COMMENTDASH = 6;
102 const COLON_STATE_COMMENTDASHDASH = 7;
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 # Marker Suffix needs to be accessible staticly.
116 const MARKER_SUFFIX = "-QINU\x7f";
117
118 # Markers used for wrapping the table of contents
119 const TOC_START = '<mw:toc>';
120 const TOC_END = '</mw:toc>';
121
122 # Persistent:
123 var $mTagHooks = array();
124 var $mTransparentTagHooks = array();
125 var $mFunctionHooks = array();
126 var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
127 var $mFunctionTagHooks = array();
128 var $mStripList = array();
129 var $mDefaultStripList = array();
130 var $mVarCache = array();
131 var $mImageParams = array();
132 var $mImageParamsMagicArray = array();
133 var $mMarkerIndex = 0;
134 var $mFirstCall = true;
135
136 # Initialised by initialiseVariables()
137
138 /**
139 * @var MagicWordArray
140 */
141 var $mVariables;
142
143 /**
144 * @var MagicWordArray
145 */
146 var $mSubstWords;
147 var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
148
149 # Cleared with clearState():
150 /**
151 * @var ParserOutput
152 */
153 var $mOutput;
154 var $mAutonumber, $mDTopen;
155
156 /**
157 * @var StripState
158 */
159 var $mStripState;
160
161 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
162 /**
163 * @var LinkHolderArray
164 */
165 var $mLinkHolders;
166
167 var $mLinkID;
168 var $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
169 var $mDefaultSort;
170 var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
171 var $mExpensiveFunctionCount; # number of expensive parser function calls
172 var $mShowToc, $mForceTocPosition;
173
174 /**
175 * @var User
176 */
177 var $mUser; # User object; only used when doing pre-save transform
178
179 # Temporary
180 # These are variables reset at least once per parse regardless of $clearState
181
182 /**
183 * @var ParserOptions
184 */
185 var $mOptions;
186
187 /**
188 * @var Title
189 */
190 var $mTitle; # Title context, used for self-link rendering and similar things
191 var $mOutputType; # Output type, one of the OT_xxx constants
192 var $ot; # Shortcut alias, see setOutputType()
193 var $mRevisionObject; # The revision object of the specified revision ID
194 var $mRevisionId; # ID to display in {{REVISIONID}} tags
195 var $mRevisionTimestamp; # The timestamp of the specified revision ID
196 var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
197 var $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
198 var $mRevIdForTs; # The revision ID which was used to fetch the timestamp
199 var $mInputSize = false; # For {{PAGESIZE}} on current page.
200
201 /**
202 * @var string
203 */
204 var $mUniqPrefix;
205
206 /**
207 * @var array Array with the language name of each language link (i.e. the
208 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
209 * duplicate language links to the ParserOutput.
210 */
211 var $mLangLinkLanguages;
212
213 /**
214 * @var boolean Recursive call protection.
215 * This variable should be treated as if it were private.
216 */
217 public $mInParse = false;
218
219 /**
220 * @param array $conf
221 */
222 public function __construct( $conf = array() ) {
223 $this->mConf = $conf;
224 $this->mUrlProtocols = wfUrlProtocols();
225 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
226 self::EXT_LINK_URL_CLASS . '+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
227 if ( isset( $conf['preprocessorClass'] ) ) {
228 $this->mPreprocessorClass = $conf['preprocessorClass'];
229 } elseif ( defined( 'HPHP_VERSION' ) ) {
230 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
231 $this->mPreprocessorClass = 'Preprocessor_Hash';
232 } elseif ( extension_loaded( 'domxml' ) ) {
233 # PECL extension that conflicts with the core DOM extension (bug 13770)
234 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
235 $this->mPreprocessorClass = 'Preprocessor_Hash';
236 } elseif ( extension_loaded( 'dom' ) ) {
237 $this->mPreprocessorClass = 'Preprocessor_DOM';
238 } else {
239 $this->mPreprocessorClass = 'Preprocessor_Hash';
240 }
241 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
242 }
243
244 /**
245 * Reduce memory usage to reduce the impact of circular references
246 */
247 function __destruct() {
248 if ( isset( $this->mLinkHolders ) ) {
249 unset( $this->mLinkHolders );
250 }
251 foreach ( $this as $name => $value ) {
252 unset( $this->$name );
253 }
254 }
255
256 /**
257 * Allow extensions to clean up when the parser is cloned
258 */
259 function __clone() {
260 $this->mInParse = false;
261 wfRunHooks( 'ParserCloned', array( $this ) );
262 }
263
264 /**
265 * Do various kinds of initialisation on the first call of the parser
266 */
267 function firstCallInit() {
268 if ( !$this->mFirstCall ) {
269 return;
270 }
271 $this->mFirstCall = false;
272
273 wfProfileIn( __METHOD__ );
274
275 CoreParserFunctions::register( $this );
276 CoreTagHooks::register( $this );
277 $this->initialiseVariables();
278
279 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
280 wfProfileOut( __METHOD__ );
281 }
282
283 /**
284 * Clear Parser state
285 *
286 * @private
287 */
288 function clearState() {
289 wfProfileIn( __METHOD__ );
290 if ( $this->mFirstCall ) {
291 $this->firstCallInit();
292 }
293 $this->mOutput = new ParserOutput;
294 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
295 $this->mAutonumber = 0;
296 $this->mLastSection = '';
297 $this->mDTopen = false;
298 $this->mIncludeCount = array();
299 $this->mArgStack = false;
300 $this->mInPre = false;
301 $this->mLinkHolders = new LinkHolderArray( $this );
302 $this->mLinkID = 0;
303 $this->mRevisionObject = $this->mRevisionTimestamp =
304 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
305 $this->mVarCache = array();
306 $this->mUser = null;
307 $this->mLangLinkLanguages = array();
308
309 /**
310 * Prefix for temporary replacement strings for the multipass parser.
311 * \x07 should never appear in input as it's disallowed in XML.
312 * Using it at the front also gives us a little extra robustness
313 * since it shouldn't match when butted up against identifier-like
314 * string constructs.
315 *
316 * Must not consist of all title characters, or else it will change
317 * the behavior of <nowiki> in a link.
318 */
319 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
320 $this->mStripState = new StripState( $this->mUniqPrefix );
321
322 # Clear these on every parse, bug 4549
323 $this->mTplRedirCache = $this->mTplDomCache = array();
324
325 $this->mShowToc = true;
326 $this->mForceTocPosition = false;
327 $this->mIncludeSizes = array(
328 'post-expand' => 0,
329 'arg' => 0,
330 );
331 $this->mPPNodeCount = 0;
332 $this->mGeneratedPPNodeCount = 0;
333 $this->mHighestExpansionDepth = 0;
334 $this->mDefaultSort = false;
335 $this->mHeadings = array();
336 $this->mDoubleUnderscores = array();
337 $this->mExpensiveFunctionCount = 0;
338
339 # Fix cloning
340 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
341 $this->mPreprocessor = null;
342 }
343
344 wfRunHooks( 'ParserClearState', array( &$this ) );
345 wfProfileOut( __METHOD__ );
346 }
347
348 /**
349 * Convert wikitext to HTML
350 * Do not call this function recursively.
351 *
352 * @param string $text text we want to parse
353 * @param Title $title
354 * @param ParserOptions $options
355 * @param bool $linestart
356 * @param bool $clearState
357 * @param int $revid Number to pass in {{REVISIONID}}
358 * @return ParserOutput A ParserOutput
359 */
360 public function parse( $text, Title $title, ParserOptions $options,
361 $linestart = true, $clearState = true, $revid = null
362 ) {
363 /**
364 * First pass--just handle <nowiki> sections, pass the rest off
365 * to internalParse() which does all the real work.
366 */
367
368 global $wgUseTidy, $wgAlwaysUseTidy, $wgShowHostnames;
369 $fname = __METHOD__ . '-' . wfGetCaller();
370 wfProfileIn( __METHOD__ );
371 wfProfileIn( $fname );
372
373 if ( $clearState ) {
374 $magicScopeVariable = $this->lock();
375 }
376
377 $this->startParse( $title, $options, self::OT_HTML, $clearState );
378
379 $this->mInputSize = strlen( $text );
380 if ( $this->mOptions->getEnableLimitReport() ) {
381 $this->mOutput->resetParseStartTime();
382 }
383
384 # Remove the strip marker tag prefix from the input, if present.
385 if ( $clearState ) {
386 $text = str_replace( $this->mUniqPrefix, '', $text );
387 }
388
389 $oldRevisionId = $this->mRevisionId;
390 $oldRevisionObject = $this->mRevisionObject;
391 $oldRevisionTimestamp = $this->mRevisionTimestamp;
392 $oldRevisionUser = $this->mRevisionUser;
393 $oldRevisionSize = $this->mRevisionSize;
394 if ( $revid !== null ) {
395 $this->mRevisionId = $revid;
396 $this->mRevisionObject = null;
397 $this->mRevisionTimestamp = null;
398 $this->mRevisionUser = null;
399 $this->mRevisionSize = null;
400 }
401
402 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
403 # No more strip!
404 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
405 $text = $this->internalParse( $text );
406 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
407
408 $text = $this->mStripState->unstripGeneral( $text );
409
410 # Clean up special characters, only run once, next-to-last before doBlockLevels
411 $fixtags = array(
412 # french spaces, last one Guillemet-left
413 # only if there is something before the space
414 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
415 # french spaces, Guillemet-right
416 '/(\\302\\253) /' => '\\1&#160;',
417 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
418 );
419 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
420
421 $text = $this->doBlockLevels( $text, $linestart );
422
423 $this->replaceLinkHolders( $text );
424
425 /**
426 * The input doesn't get language converted if
427 * a) It's disabled
428 * b) Content isn't converted
429 * c) It's a conversion table
430 * d) it is an interface message (which is in the user language)
431 */
432 if ( !( $options->getDisableContentConversion()
433 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
434 ) {
435 if ( !$this->mOptions->getInterfaceMessage() ) {
436 # The position of the convert() call should not be changed. it
437 # assumes that the links are all replaced and the only thing left
438 # is the <nowiki> mark.
439 $text = $this->getConverterLanguage()->convert( $text );
440 }
441 }
442
443 /**
444 * A converted title will be provided in the output object if title and
445 * content conversion are enabled, the article text does not contain
446 * a conversion-suppressing double-underscore tag, and no
447 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
448 * automatic link conversion.
449 */
450 if ( !( $options->getDisableTitleConversion()
451 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
452 || isset( $this->mDoubleUnderscores['notitleconvert'] )
453 || $this->mOutput->getDisplayTitle() !== false )
454 ) {
455 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
456 if ( $convruletitle ) {
457 $this->mOutput->setTitleText( $convruletitle );
458 } else {
459 $titleText = $this->getConverterLanguage()->convertTitle( $title );
460 $this->mOutput->setTitleText( $titleText );
461 }
462 }
463
464 $text = $this->mStripState->unstripNoWiki( $text );
465
466 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
467
468 $text = $this->replaceTransparentTags( $text );
469 $text = $this->mStripState->unstripGeneral( $text );
470
471 $text = Sanitizer::normalizeCharReferences( $text );
472
473 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
474 $text = MWTidy::tidy( $text );
475 } else {
476 # attempt to sanitize at least some nesting problems
477 # (bug #2702 and quite a few others)
478 $tidyregs = array(
479 # ''Something [http://www.cool.com cool''] -->
480 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
481 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
482 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
483 # fix up an anchor inside another anchor, only
484 # at least for a single single nested link (bug 3695)
485 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
486 '\\1\\2</a>\\3</a>\\1\\4</a>',
487 # fix div inside inline elements- doBlockLevels won't wrap a line which
488 # contains a div, so fix it up here; replace
489 # div with escaped text
490 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
491 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
492 # remove empty italic or bold tag pairs, some
493 # introduced by rules above
494 '/<([bi])><\/\\1>/' => '',
495 );
496
497 $text = preg_replace(
498 array_keys( $tidyregs ),
499 array_values( $tidyregs ),
500 $text );
501 }
502
503 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
504 $this->limitationWarn( 'expensive-parserfunction',
505 $this->mExpensiveFunctionCount,
506 $this->mOptions->getExpensiveParserFunctionLimit()
507 );
508 }
509
510 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
511
512 # Information on include size limits, for the benefit of users who try to skirt them
513 if ( $this->mOptions->getEnableLimitReport() ) {
514 $max = $this->mOptions->getMaxIncludeSize();
515
516 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
517 if ( $cpuTime !== null ) {
518 $this->mOutput->setLimitReportData( 'limitreport-cputime',
519 sprintf( "%.3f", $cpuTime )
520 );
521 }
522
523 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
524 $this->mOutput->setLimitReportData( 'limitreport-walltime',
525 sprintf( "%.3f", $wallTime )
526 );
527
528 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
529 array( $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() )
530 );
531 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
532 array( $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() )
533 );
534 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
535 array( $this->mIncludeSizes['post-expand'], $max )
536 );
537 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
538 array( $this->mIncludeSizes['arg'], $max )
539 );
540 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
541 array( $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() )
542 );
543 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
544 array( $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() )
545 );
546 wfRunHooks( 'ParserLimitReportPrepare', array( $this, $this->mOutput ) );
547
548 $limitReport = "NewPP limit report\n";
549 if ( $wgShowHostnames ) {
550 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
551 }
552 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
553 if ( wfRunHooks( 'ParserLimitReportFormat',
554 array( $key, &$value, &$limitReport, false, false )
555 ) ) {
556 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
557 $valueMsg = wfMessage( array( "$key-value-text", "$key-value" ) )
558 ->inLanguage( 'en' )->useDatabase( false );
559 if ( !$valueMsg->exists() ) {
560 $valueMsg = new RawMessage( '$1' );
561 }
562 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
563 $valueMsg->params( $value );
564 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
565 }
566 }
567 }
568 // Since we're not really outputting HTML, decode the entities and
569 // then re-encode the things that need hiding inside HTML comments.
570 $limitReport = htmlspecialchars_decode( $limitReport );
571 wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
572
573 // Sanitize for comment. Note '‐' in the replacement is U+2010,
574 // which looks much like the problematic '-'.
575 $limitReport = str_replace( array( '-', '&' ), array( '‐', '&amp;' ), $limitReport );
576 $text .= "\n<!-- \n$limitReport-->\n";
577
578 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
579 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
580 $this->mTitle->getPrefixedDBkey() );
581 }
582 }
583 $this->mOutput->setText( $text );
584
585 $this->mRevisionId = $oldRevisionId;
586 $this->mRevisionObject = $oldRevisionObject;
587 $this->mRevisionTimestamp = $oldRevisionTimestamp;
588 $this->mRevisionUser = $oldRevisionUser;
589 $this->mRevisionSize = $oldRevisionSize;
590 $this->mInputSize = false;
591 wfProfileOut( $fname );
592 wfProfileOut( __METHOD__ );
593
594 return $this->mOutput;
595 }
596
597 /**
598 * Recursive parser entry point that can be called from an extension tag
599 * hook.
600 *
601 * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
602 *
603 * @param string $text Text extension wants to have parsed
604 * @param bool|PPFrame $frame The frame to use for expanding any template variables
605 *
606 * @return string
607 */
608 function recursiveTagParse( $text, $frame = false ) {
609 wfProfileIn( __METHOD__ );
610 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
611 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
612 $text = $this->internalParse( $text, false, $frame );
613 wfProfileOut( __METHOD__ );
614 return $text;
615 }
616
617 /**
618 * Expand templates and variables in the text, producing valid, static wikitext.
619 * Also removes comments.
620 * Do not call this function recursively.
621 * @param string $text
622 * @param Title $title
623 * @param ParserOptions $options
624 * @param int|null $revid
625 * @param bool|PPFrame $frame
626 * @return mixed|string
627 */
628 function preprocess( $text, Title $title = null, ParserOptions $options, $revid = null, $frame = false ) {
629 wfProfileIn( __METHOD__ );
630 $magicScopeVariable = $this->lock();
631 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
632 if ( $revid !== null ) {
633 $this->mRevisionId = $revid;
634 }
635 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
636 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
637 $text = $this->replaceVariables( $text, $frame );
638 $text = $this->mStripState->unstripBoth( $text );
639 wfProfileOut( __METHOD__ );
640 return $text;
641 }
642
643 /**
644 * Recursive parser entry point that can be called from an extension tag
645 * hook.
646 *
647 * @param string $text Text to be expanded
648 * @param bool|PPFrame $frame The frame to use for expanding any template variables
649 * @return string
650 * @since 1.19
651 */
652 public function recursivePreprocess( $text, $frame = false ) {
653 wfProfileIn( __METHOD__ );
654 $text = $this->replaceVariables( $text, $frame );
655 $text = $this->mStripState->unstripBoth( $text );
656 wfProfileOut( __METHOD__ );
657 return $text;
658 }
659
660 /**
661 * Process the wikitext for the "?preload=" feature. (bug 5210)
662 *
663 * "<noinclude>", "<includeonly>" etc. are parsed as for template
664 * transclusion, comments, templates, arguments, tags hooks and parser
665 * functions are untouched.
666 *
667 * @param string $text
668 * @param Title $title
669 * @param ParserOptions $options
670 * @param array $params
671 * @return string
672 */
673 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = array() ) {
674 $msg = new RawMessage( $text );
675 $text = $msg->params( $params )->plain();
676
677 # Parser (re)initialisation
678 $magicScopeVariable = $this->lock();
679 $this->startParse( $title, $options, self::OT_PLAIN, true );
680
681 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
682 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
683 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
684 $text = $this->mStripState->unstripBoth( $text );
685 return $text;
686 }
687
688 /**
689 * Get a random string
690 *
691 * @return string
692 */
693 public static function getRandomString() {
694 return wfRandomString( 16 );
695 }
696
697 /**
698 * Set the current user.
699 * Should only be used when doing pre-save transform.
700 *
701 * @param User|null $user User object or null (to reset)
702 */
703 function setUser( $user ) {
704 $this->mUser = $user;
705 }
706
707 /**
708 * Accessor for mUniqPrefix.
709 *
710 * @return string
711 */
712 public function uniqPrefix() {
713 if ( !isset( $this->mUniqPrefix ) ) {
714 # @todo FIXME: This is probably *horribly wrong*
715 # LanguageConverter seems to want $wgParser's uniqPrefix, however
716 # if this is called for a parser cache hit, the parser may not
717 # have ever been initialized in the first place.
718 # Not really sure what the heck is supposed to be going on here.
719 return '';
720 # throw new MWException( "Accessing uninitialized mUniqPrefix" );
721 }
722 return $this->mUniqPrefix;
723 }
724
725 /**
726 * Set the context title
727 *
728 * @param Title $t
729 */
730 function setTitle( $t ) {
731 if ( !$t ) {
732 $t = Title::newFromText( 'NO TITLE' );
733 }
734
735 if ( $t->hasFragment() ) {
736 # Strip the fragment to avoid various odd effects
737 $this->mTitle = clone $t;
738 $this->mTitle->setFragment( '' );
739 } else {
740 $this->mTitle = $t;
741 }
742 }
743
744 /**
745 * Accessor for the Title object
746 *
747 * @return Title
748 */
749 function getTitle() {
750 return $this->mTitle;
751 }
752
753 /**
754 * Accessor/mutator for the Title object
755 *
756 * @param Title $x Title object or null to just get the current one
757 * @return Title
758 */
759 function Title( $x = null ) {
760 return wfSetVar( $this->mTitle, $x );
761 }
762
763 /**
764 * Set the output type
765 *
766 * @param int $ot New value
767 */
768 function setOutputType( $ot ) {
769 $this->mOutputType = $ot;
770 # Shortcut alias
771 $this->ot = array(
772 'html' => $ot == self::OT_HTML,
773 'wiki' => $ot == self::OT_WIKI,
774 'pre' => $ot == self::OT_PREPROCESS,
775 'plain' => $ot == self::OT_PLAIN,
776 );
777 }
778
779 /**
780 * Accessor/mutator for the output type
781 *
782 * @param int|null $x New value or null to just get the current one
783 * @return int
784 */
785 function OutputType( $x = null ) {
786 return wfSetVar( $this->mOutputType, $x );
787 }
788
789 /**
790 * Get the ParserOutput object
791 *
792 * @return ParserOutput
793 */
794 function getOutput() {
795 return $this->mOutput;
796 }
797
798 /**
799 * Get the ParserOptions object
800 *
801 * @return ParserOptions object
802 */
803 function getOptions() {
804 return $this->mOptions;
805 }
806
807 /**
808 * Accessor/mutator for the ParserOptions object
809 *
810 * @param ParserOptions $x New value or null to just get the current one
811 * @return ParserOptions Current ParserOptions object
812 */
813 function Options( $x = null ) {
814 return wfSetVar( $this->mOptions, $x );
815 }
816
817 /**
818 * @return int
819 */
820 function nextLinkID() {
821 return $this->mLinkID++;
822 }
823
824 /**
825 * @param int $id
826 */
827 function setLinkID( $id ) {
828 $this->mLinkID = $id;
829 }
830
831 /**
832 * Get a language object for use in parser functions such as {{FORMATNUM:}}
833 * @return Language
834 */
835 function getFunctionLang() {
836 return $this->getTargetLanguage();
837 }
838
839 /**
840 * Get the target language for the content being parsed. This is usually the
841 * language that the content is in.
842 *
843 * @since 1.19
844 *
845 * @throws MWException
846 * @return Language|null
847 */
848 public function getTargetLanguage() {
849 $target = $this->mOptions->getTargetLanguage();
850
851 if ( $target !== null ) {
852 return $target;
853 } elseif ( $this->mOptions->getInterfaceMessage() ) {
854 return $this->mOptions->getUserLangObj();
855 } elseif ( is_null( $this->mTitle ) ) {
856 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
857 }
858
859 return $this->mTitle->getPageLanguage();
860 }
861
862 /**
863 * Get the language object for language conversion
864 * @return Language|null
865 */
866 function getConverterLanguage() {
867 return $this->getTargetLanguage();
868 }
869
870 /**
871 * Get a User object either from $this->mUser, if set, or from the
872 * ParserOptions object otherwise
873 *
874 * @return User
875 */
876 function getUser() {
877 if ( !is_null( $this->mUser ) ) {
878 return $this->mUser;
879 }
880 return $this->mOptions->getUser();
881 }
882
883 /**
884 * Get a preprocessor object
885 *
886 * @return Preprocessor
887 */
888 function getPreprocessor() {
889 if ( !isset( $this->mPreprocessor ) ) {
890 $class = $this->mPreprocessorClass;
891 $this->mPreprocessor = new $class( $this );
892 }
893 return $this->mPreprocessor;
894 }
895
896 /**
897 * Replaces all occurrences of HTML-style comments and the given tags
898 * in the text with a random marker and returns the next text. The output
899 * parameter $matches will be an associative array filled with data in
900 * the form:
901 *
902 * @code
903 * 'UNIQ-xxxxx' => array(
904 * 'element',
905 * 'tag content',
906 * array( 'param' => 'x' ),
907 * '<element param="x">tag content</element>' ) )
908 * @endcode
909 *
910 * @param array $elements List of element names. Comments are always extracted.
911 * @param string $text Source text string.
912 * @param array $matches Out parameter, Array: extracted tags
913 * @param string $uniq_prefix
914 * @return string Stripped text
915 */
916 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
917 static $n = 1;
918 $stripped = '';
919 $matches = array();
920
921 $taglist = implode( '|', $elements );
922 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
923
924 while ( $text != '' ) {
925 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
926 $stripped .= $p[0];
927 if ( count( $p ) < 5 ) {
928 break;
929 }
930 if ( count( $p ) > 5 ) {
931 # comment
932 $element = $p[4];
933 $attributes = '';
934 $close = '';
935 $inside = $p[5];
936 } else {
937 # tag
938 $element = $p[1];
939 $attributes = $p[2];
940 $close = $p[3];
941 $inside = $p[4];
942 }
943
944 $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
945 $stripped .= $marker;
946
947 if ( $close === '/>' ) {
948 # Empty element tag, <tag />
949 $content = null;
950 $text = $inside;
951 $tail = null;
952 } else {
953 if ( $element === '!--' ) {
954 $end = '/(-->)/';
955 } else {
956 $end = "/(<\\/$element\\s*>)/i";
957 }
958 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
959 $content = $q[0];
960 if ( count( $q ) < 3 ) {
961 # No end tag -- let it run out to the end of the text.
962 $tail = '';
963 $text = '';
964 } else {
965 $tail = $q[1];
966 $text = $q[2];
967 }
968 }
969
970 $matches[$marker] = array( $element,
971 $content,
972 Sanitizer::decodeTagAttributes( $attributes ),
973 "<$element$attributes$close$content$tail" );
974 }
975 return $stripped;
976 }
977
978 /**
979 * Get a list of strippable XML-like elements
980 *
981 * @return array
982 */
983 function getStripList() {
984 return $this->mStripList;
985 }
986
987 /**
988 * Add an item to the strip state
989 * Returns the unique tag which must be inserted into the stripped text
990 * The tag will be replaced with the original text in unstrip()
991 *
992 * @param string $text
993 *
994 * @return string
995 */
996 function insertStripItem( $text ) {
997 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
998 $this->mMarkerIndex++;
999 $this->mStripState->addGeneral( $rnd, $text );
1000 return $rnd;
1001 }
1002
1003 /**
1004 * parse the wiki syntax used to render tables
1005 *
1006 * @private
1007 * @param string $text
1008 * @return string
1009 */
1010 function doTableStuff( $text ) {
1011 wfProfileIn( __METHOD__ );
1012
1013 $lines = StringUtils::explode( "\n", $text );
1014 $out = '';
1015 $td_history = array(); # Is currently a td tag open?
1016 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
1017 $tr_history = array(); # Is currently a tr tag open?
1018 $tr_attributes = array(); # history of tr attributes
1019 $has_opened_tr = array(); # Did this table open a <tr> element?
1020 $indent_level = 0; # indent level of the table
1021
1022 foreach ( $lines as $outLine ) {
1023 $line = trim( $outLine );
1024
1025 if ( $line === '' ) { # empty line, go to next line
1026 $out .= $outLine . "\n";
1027 continue;
1028 }
1029
1030 $first_character = $line[0];
1031 $matches = array();
1032
1033 if ( preg_match( '/^(:*)\{\|(.*)$/', $line, $matches ) ) {
1034 # First check if we are starting a new table
1035 $indent_level = strlen( $matches[1] );
1036
1037 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1038 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1039
1040 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1041 array_push( $td_history, false );
1042 array_push( $last_tag_history, '' );
1043 array_push( $tr_history, false );
1044 array_push( $tr_attributes, '' );
1045 array_push( $has_opened_tr, false );
1046 } elseif ( count( $td_history ) == 0 ) {
1047 # Don't do any of the following
1048 $out .= $outLine . "\n";
1049 continue;
1050 } elseif ( substr( $line, 0, 2 ) === '|}' ) {
1051 # We are ending a table
1052 $line = '</table>' . substr( $line, 2 );
1053 $last_tag = array_pop( $last_tag_history );
1054
1055 if ( !array_pop( $has_opened_tr ) ) {
1056 $line = "<tr><td></td></tr>{$line}";
1057 }
1058
1059 if ( array_pop( $tr_history ) ) {
1060 $line = "</tr>{$line}";
1061 }
1062
1063 if ( array_pop( $td_history ) ) {
1064 $line = "</{$last_tag}>{$line}";
1065 }
1066 array_pop( $tr_attributes );
1067 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1068 } elseif ( substr( $line, 0, 2 ) === '|-' ) {
1069 # Now we have a table row
1070 $line = preg_replace( '#^\|-+#', '', $line );
1071
1072 # Whats after the tag is now only attributes
1073 $attributes = $this->mStripState->unstripBoth( $line );
1074 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1075 array_pop( $tr_attributes );
1076 array_push( $tr_attributes, $attributes );
1077
1078 $line = '';
1079 $last_tag = array_pop( $last_tag_history );
1080 array_pop( $has_opened_tr );
1081 array_push( $has_opened_tr, true );
1082
1083 if ( array_pop( $tr_history ) ) {
1084 $line = '</tr>';
1085 }
1086
1087 if ( array_pop( $td_history ) ) {
1088 $line = "</{$last_tag}>{$line}";
1089 }
1090
1091 $outLine = $line;
1092 array_push( $tr_history, false );
1093 array_push( $td_history, false );
1094 array_push( $last_tag_history, '' );
1095 } elseif ( $first_character === '|'
1096 || $first_character === '!'
1097 || substr( $line, 0, 2 ) === '|+'
1098 ) {
1099 # This might be cell elements, td, th or captions
1100 if ( substr( $line, 0, 2 ) === '|+' ) {
1101 $first_character = '+';
1102 $line = substr( $line, 1 );
1103 }
1104
1105 $line = substr( $line, 1 );
1106
1107 if ( $first_character === '!' ) {
1108 $line = str_replace( '!!', '||', $line );
1109 }
1110
1111 # Split up multiple cells on the same line.
1112 # FIXME : This can result in improper nesting of tags processed
1113 # by earlier parser steps, but should avoid splitting up eg
1114 # attribute values containing literal "||".
1115 $cells = StringUtils::explodeMarkup( '||', $line );
1116
1117 $outLine = '';
1118
1119 # Loop through each table cell
1120 foreach ( $cells as $cell ) {
1121 $previous = '';
1122 if ( $first_character !== '+' ) {
1123 $tr_after = array_pop( $tr_attributes );
1124 if ( !array_pop( $tr_history ) ) {
1125 $previous = "<tr{$tr_after}>\n";
1126 }
1127 array_push( $tr_history, true );
1128 array_push( $tr_attributes, '' );
1129 array_pop( $has_opened_tr );
1130 array_push( $has_opened_tr, true );
1131 }
1132
1133 $last_tag = array_pop( $last_tag_history );
1134
1135 if ( array_pop( $td_history ) ) {
1136 $previous = "</{$last_tag}>\n{$previous}";
1137 }
1138
1139 if ( $first_character === '|' ) {
1140 $last_tag = 'td';
1141 } elseif ( $first_character === '!' ) {
1142 $last_tag = 'th';
1143 } elseif ( $first_character === '+' ) {
1144 $last_tag = 'caption';
1145 } else {
1146 $last_tag = '';
1147 }
1148
1149 array_push( $last_tag_history, $last_tag );
1150
1151 # A cell could contain both parameters and data
1152 $cell_data = explode( '|', $cell, 2 );
1153
1154 # Bug 553: Note that a '|' inside an invalid link should not
1155 # be mistaken as delimiting cell parameters
1156 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1157 $cell = "{$previous}<{$last_tag}>{$cell}";
1158 } elseif ( count( $cell_data ) == 1 ) {
1159 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1160 } else {
1161 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1162 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1163 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1164 }
1165
1166 $outLine .= $cell;
1167 array_push( $td_history, true );
1168 }
1169 }
1170 $out .= $outLine . "\n";
1171 }
1172
1173 # Closing open td, tr && table
1174 while ( count( $td_history ) > 0 ) {
1175 if ( array_pop( $td_history ) ) {
1176 $out .= "</td>\n";
1177 }
1178 if ( array_pop( $tr_history ) ) {
1179 $out .= "</tr>\n";
1180 }
1181 if ( !array_pop( $has_opened_tr ) ) {
1182 $out .= "<tr><td></td></tr>\n";
1183 }
1184
1185 $out .= "</table>\n";
1186 }
1187
1188 # Remove trailing line-ending (b/c)
1189 if ( substr( $out, -1 ) === "\n" ) {
1190 $out = substr( $out, 0, -1 );
1191 }
1192
1193 # special case: don't return empty table
1194 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1195 $out = '';
1196 }
1197
1198 wfProfileOut( __METHOD__ );
1199
1200 return $out;
1201 }
1202
1203 /**
1204 * Helper function for parse() that transforms wiki markup into
1205 * HTML. Only called for $mOutputType == self::OT_HTML.
1206 *
1207 * @private
1208 *
1209 * @param string $text
1210 * @param bool $isMain
1211 * @param bool $frame
1212 *
1213 * @return string
1214 */
1215 function internalParse( $text, $isMain = true, $frame = false ) {
1216 wfProfileIn( __METHOD__ );
1217
1218 $origText = $text;
1219
1220 # Hook to suspend the parser in this state
1221 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1222 wfProfileOut( __METHOD__ );
1223 return $text;
1224 }
1225
1226 # if $frame is provided, then use $frame for replacing any variables
1227 if ( $frame ) {
1228 # use frame depth to infer how include/noinclude tags should be handled
1229 # depth=0 means this is the top-level document; otherwise it's an included document
1230 if ( !$frame->depth ) {
1231 $flag = 0;
1232 } else {
1233 $flag = Parser::PTD_FOR_INCLUSION;
1234 }
1235 $dom = $this->preprocessToDom( $text, $flag );
1236 $text = $frame->expand( $dom );
1237 } else {
1238 # if $frame is not provided, then use old-style replaceVariables
1239 $text = $this->replaceVariables( $text );
1240 }
1241
1242 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1243 $text = Sanitizer::removeHTMLtags(
1244 $text,
1245 array( &$this, 'attributeStripCallback' ),
1246 false,
1247 array_keys( $this->mTransparentTagHooks )
1248 );
1249 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1250
1251 # Tables need to come after variable replacement for things to work
1252 # properly; putting them before other transformations should keep
1253 # exciting things like link expansions from showing up in surprising
1254 # places.
1255 $text = $this->doTableStuff( $text );
1256
1257 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1258
1259 $text = $this->doDoubleUnderscore( $text );
1260
1261 $text = $this->doHeadings( $text );
1262 $text = $this->replaceInternalLinks( $text );
1263 $text = $this->doAllQuotes( $text );
1264 $text = $this->replaceExternalLinks( $text );
1265
1266 # replaceInternalLinks may sometimes leave behind
1267 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1268 $text = str_replace( $this->mUniqPrefix . 'NOPARSE', '', $text );
1269
1270 $text = $this->doMagicLinks( $text );
1271 $text = $this->formatHeadings( $text, $origText, $isMain );
1272
1273 wfProfileOut( __METHOD__ );
1274 return $text;
1275 }
1276
1277 /**
1278 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1279 * magic external links.
1280 *
1281 * DML
1282 * @private
1283 *
1284 * @param string $text
1285 *
1286 * @return string
1287 */
1288 function doMagicLinks( $text ) {
1289 wfProfileIn( __METHOD__ );
1290 $prots = wfUrlProtocolsWithoutProtRel();
1291 $urlChar = self::EXT_LINK_URL_CLASS;
1292 $text = preg_replace_callback(
1293 '!(?: # Start cases
1294 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1295 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1296 (\\b(?i:$prots)$urlChar+) | # m[3]: Free external links" . '
1297 (?:RFC|PMID)\s+([0-9]+) | # m[4]: RFC or PMID, capture number
1298 ISBN\s+(\b # m[5]: ISBN, capture number
1299 (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
1300 (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
1301 [0-9Xx] # check digit
1302 \b)
1303 )!xu', array( &$this, 'magicLinkCallback' ), $text );
1304 wfProfileOut( __METHOD__ );
1305 return $text;
1306 }
1307
1308 /**
1309 * @throws MWException
1310 * @param array $m
1311 * @return HTML|string
1312 */
1313 function magicLinkCallback( $m ) {
1314 if ( isset( $m[1] ) && $m[1] !== '' ) {
1315 # Skip anchor
1316 return $m[0];
1317 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1318 # Skip HTML element
1319 return $m[0];
1320 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1321 # Free external link
1322 return $this->makeFreeExternalLink( $m[0] );
1323 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1324 # RFC or PMID
1325 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1326 $keyword = 'RFC';
1327 $urlmsg = 'rfcurl';
1328 $cssClass = 'mw-magiclink-rfc';
1329 $id = $m[4];
1330 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1331 $keyword = 'PMID';
1332 $urlmsg = 'pubmedurl';
1333 $cssClass = 'mw-magiclink-pmid';
1334 $id = $m[4];
1335 } else {
1336 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1337 substr( $m[0], 0, 20 ) . '"' );
1338 }
1339 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1340 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass );
1341 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1342 # ISBN
1343 $isbn = $m[5];
1344 $num = strtr( $isbn, array(
1345 '-' => '',
1346 ' ' => '',
1347 'x' => 'X',
1348 ));
1349 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1350 return '<a href="' .
1351 htmlspecialchars( $titleObj->getLocalURL() ) .
1352 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1353 } else {
1354 return $m[0];
1355 }
1356 }
1357
1358 /**
1359 * Make a free external link, given a user-supplied URL
1360 *
1361 * @param string $url
1362 *
1363 * @return string HTML
1364 * @private
1365 */
1366 function makeFreeExternalLink( $url ) {
1367 wfProfileIn( __METHOD__ );
1368
1369 $trail = '';
1370
1371 # The characters '<' and '>' (which were escaped by
1372 # removeHTMLtags()) should not be included in
1373 # URLs, per RFC 2396.
1374 $m2 = array();
1375 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1376 $trail = substr( $url, $m2[0][1] ) . $trail;
1377 $url = substr( $url, 0, $m2[0][1] );
1378 }
1379
1380 # Move trailing punctuation to $trail
1381 $sep = ',;\.:!?';
1382 # If there is no left bracket, then consider right brackets fair game too
1383 if ( strpos( $url, '(' ) === false ) {
1384 $sep .= ')';
1385 }
1386
1387 $numSepChars = strspn( strrev( $url ), $sep );
1388 if ( $numSepChars ) {
1389 $trail = substr( $url, -$numSepChars ) . $trail;
1390 $url = substr( $url, 0, -$numSepChars );
1391 }
1392
1393 $url = Sanitizer::cleanUrl( $url );
1394
1395 # Is this an external image?
1396 $text = $this->maybeMakeExternalImage( $url );
1397 if ( $text === false ) {
1398 # Not an image, make a link
1399 $text = Linker::makeExternalLink( $url,
1400 $this->getConverterLanguage()->markNoConversion( $url, true ),
1401 true, 'free',
1402 $this->getExternalLinkAttribs( $url ) );
1403 # Register it in the output object...
1404 # Replace unnecessary URL escape codes with their equivalent characters
1405 $pasteurized = self::replaceUnusualEscapes( $url );
1406 $this->mOutput->addExternalLink( $pasteurized );
1407 }
1408 wfProfileOut( __METHOD__ );
1409 return $text . $trail;
1410 }
1411
1412 /**
1413 * Parse headers and return html
1414 *
1415 * @private
1416 *
1417 * @param string $text
1418 *
1419 * @return string
1420 */
1421 function doHeadings( $text ) {
1422 wfProfileIn( __METHOD__ );
1423 for ( $i = 6; $i >= 1; --$i ) {
1424 $h = str_repeat( '=', $i );
1425 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1426 }
1427 wfProfileOut( __METHOD__ );
1428 return $text;
1429 }
1430
1431 /**
1432 * Replace single quotes with HTML markup
1433 * @private
1434 *
1435 * @param string $text
1436 *
1437 * @return string the altered text
1438 */
1439 function doAllQuotes( $text ) {
1440 wfProfileIn( __METHOD__ );
1441 $outtext = '';
1442 $lines = StringUtils::explode( "\n", $text );
1443 foreach ( $lines as $line ) {
1444 $outtext .= $this->doQuotes( $line ) . "\n";
1445 }
1446 $outtext = substr( $outtext, 0, -1 );
1447 wfProfileOut( __METHOD__ );
1448 return $outtext;
1449 }
1450
1451 /**
1452 * Helper function for doAllQuotes()
1453 *
1454 * @param string $text
1455 *
1456 * @return string
1457 */
1458 public function doQuotes( $text ) {
1459 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1460 $countarr = count( $arr );
1461 if ( $countarr == 1 ) {
1462 return $text;
1463 }
1464
1465 // First, do some preliminary work. This may shift some apostrophes from
1466 // being mark-up to being text. It also counts the number of occurrences
1467 // of bold and italics mark-ups.
1468 $numbold = 0;
1469 $numitalics = 0;
1470 for ( $i = 1; $i < $countarr; $i += 2 ) {
1471 $thislen = strlen( $arr[$i] );
1472 // If there are ever four apostrophes, assume the first is supposed to
1473 // be text, and the remaining three constitute mark-up for bold text.
1474 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1475 if ( $thislen == 4 ) {
1476 $arr[$i - 1] .= "'";
1477 $arr[$i] = "'''";
1478 $thislen = 3;
1479 } elseif ( $thislen > 5 ) {
1480 // If there are more than 5 apostrophes in a row, assume they're all
1481 // text except for the last 5.
1482 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1483 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1484 $arr[$i] = "'''''";
1485 $thislen = 5;
1486 }
1487 // Count the number of occurrences of bold and italics mark-ups.
1488 if ( $thislen == 2 ) {
1489 $numitalics++;
1490 } elseif ( $thislen == 3 ) {
1491 $numbold++;
1492 } elseif ( $thislen == 5 ) {
1493 $numitalics++;
1494 $numbold++;
1495 }
1496 }
1497
1498 // If there is an odd number of both bold and italics, it is likely
1499 // that one of the bold ones was meant to be an apostrophe followed
1500 // by italics. Which one we cannot know for certain, but it is more
1501 // likely to be one that has a single-letter word before it.
1502 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1503 $firstsingleletterword = -1;
1504 $firstmultiletterword = -1;
1505 $firstspace = -1;
1506 for ( $i = 1; $i < $countarr; $i += 2 ) {
1507 if ( strlen( $arr[$i] ) == 3 ) {
1508 $x1 = substr( $arr[$i - 1], -1 );
1509 $x2 = substr( $arr[$i - 1], -2, 1 );
1510 if ( $x1 === ' ' ) {
1511 if ( $firstspace == -1 ) {
1512 $firstspace = $i;
1513 }
1514 } elseif ( $x2 === ' ' ) {
1515 if ( $firstsingleletterword == -1 ) {
1516 $firstsingleletterword = $i;
1517 // if $firstsingleletterword is set, we don't
1518 // look at the other options, so we can bail early.
1519 break;
1520 }
1521 } else {
1522 if ( $firstmultiletterword == -1 ) {
1523 $firstmultiletterword = $i;
1524 }
1525 }
1526 }
1527 }
1528
1529 // If there is a single-letter word, use it!
1530 if ( $firstsingleletterword > -1 ) {
1531 $arr[$firstsingleletterword] = "''";
1532 $arr[$firstsingleletterword - 1] .= "'";
1533 } elseif ( $firstmultiletterword > -1 ) {
1534 // If not, but there's a multi-letter word, use that one.
1535 $arr[$firstmultiletterword] = "''";
1536 $arr[$firstmultiletterword - 1] .= "'";
1537 } elseif ( $firstspace > -1 ) {
1538 // ... otherwise use the first one that has neither.
1539 // (notice that it is possible for all three to be -1 if, for example,
1540 // there is only one pentuple-apostrophe in the line)
1541 $arr[$firstspace] = "''";
1542 $arr[$firstspace - 1] .= "'";
1543 }
1544 }
1545
1546 // Now let's actually convert our apostrophic mush to HTML!
1547 $output = '';
1548 $buffer = '';
1549 $state = '';
1550 $i = 0;
1551 foreach ( $arr as $r ) {
1552 if ( ( $i % 2 ) == 0 ) {
1553 if ( $state === 'both' ) {
1554 $buffer .= $r;
1555 } else {
1556 $output .= $r;
1557 }
1558 } else {
1559 $thislen = strlen( $r );
1560 if ( $thislen == 2 ) {
1561 if ( $state === 'i' ) {
1562 $output .= '</i>';
1563 $state = '';
1564 } elseif ( $state === 'bi' ) {
1565 $output .= '</i>';
1566 $state = 'b';
1567 } elseif ( $state === 'ib' ) {
1568 $output .= '</b></i><b>';
1569 $state = 'b';
1570 } elseif ( $state === 'both' ) {
1571 $output .= '<b><i>' . $buffer . '</i>';
1572 $state = 'b';
1573 } else { // $state can be 'b' or ''
1574 $output .= '<i>';
1575 $state .= 'i';
1576 }
1577 } elseif ( $thislen == 3 ) {
1578 if ( $state === 'b' ) {
1579 $output .= '</b>';
1580 $state = '';
1581 } elseif ( $state === 'bi' ) {
1582 $output .= '</i></b><i>';
1583 $state = 'i';
1584 } elseif ( $state === 'ib' ) {
1585 $output .= '</b>';
1586 $state = 'i';
1587 } elseif ( $state === 'both' ) {
1588 $output .= '<i><b>' . $buffer . '</b>';
1589 $state = 'i';
1590 } else { // $state can be 'i' or ''
1591 $output .= '<b>';
1592 $state .= 'b';
1593 }
1594 } elseif ( $thislen == 5 ) {
1595 if ( $state === 'b' ) {
1596 $output .= '</b><i>';
1597 $state = 'i';
1598 } elseif ( $state === 'i' ) {
1599 $output .= '</i><b>';
1600 $state = 'b';
1601 } elseif ( $state === 'bi' ) {
1602 $output .= '</i></b>';
1603 $state = '';
1604 } elseif ( $state === 'ib' ) {
1605 $output .= '</b></i>';
1606 $state = '';
1607 } elseif ( $state === 'both' ) {
1608 $output .= '<i><b>' . $buffer . '</b></i>';
1609 $state = '';
1610 } else { // ($state == '')
1611 $buffer = '';
1612 $state = 'both';
1613 }
1614 }
1615 }
1616 $i++;
1617 }
1618 // Now close all remaining tags. Notice that the order is important.
1619 if ( $state === 'b' || $state === 'ib' ) {
1620 $output .= '</b>';
1621 }
1622 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1623 $output .= '</i>';
1624 }
1625 if ( $state === 'bi' ) {
1626 $output .= '</b>';
1627 }
1628 // There might be lonely ''''', so make sure we have a buffer
1629 if ( $state === 'both' && $buffer ) {
1630 $output .= '<b><i>' . $buffer . '</i></b>';
1631 }
1632 return $output;
1633 }
1634
1635 /**
1636 * Replace external links (REL)
1637 *
1638 * Note: this is all very hackish and the order of execution matters a lot.
1639 * Make sure to run tests/parserTests.php if you change this code.
1640 *
1641 * @private
1642 *
1643 * @param string $text
1644 *
1645 * @throws MWException
1646 * @return string
1647 */
1648 function replaceExternalLinks( $text ) {
1649 wfProfileIn( __METHOD__ );
1650
1651 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1652 if ( $bits === false ) {
1653 wfProfileOut( __METHOD__ );
1654 throw new MWException( "PCRE needs to be compiled with "
1655 . "--enable-unicode-properties in order for MediaWiki to function" );
1656 }
1657 $s = array_shift( $bits );
1658
1659 $i = 0;
1660 while ( $i < count( $bits ) ) {
1661 $url = $bits[$i++];
1662 $i++; // protocol
1663 $text = $bits[$i++];
1664 $trail = $bits[$i++];
1665
1666 # The characters '<' and '>' (which were escaped by
1667 # removeHTMLtags()) should not be included in
1668 # URLs, per RFC 2396.
1669 $m2 = array();
1670 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1671 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1672 $url = substr( $url, 0, $m2[0][1] );
1673 }
1674
1675 # If the link text is an image URL, replace it with an <img> tag
1676 # This happened by accident in the original parser, but some people used it extensively
1677 $img = $this->maybeMakeExternalImage( $text );
1678 if ( $img !== false ) {
1679 $text = $img;
1680 }
1681
1682 $dtrail = '';
1683
1684 # Set linktype for CSS - if URL==text, link is essentially free
1685 $linktype = ( $text === $url ) ? 'free' : 'text';
1686
1687 # No link text, e.g. [http://domain.tld/some.link]
1688 if ( $text == '' ) {
1689 # Autonumber
1690 $langObj = $this->getTargetLanguage();
1691 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1692 $linktype = 'autonumber';
1693 } else {
1694 # Have link text, e.g. [http://domain.tld/some.link text]s
1695 # Check for trail
1696 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1697 }
1698
1699 $text = $this->getConverterLanguage()->markNoConversion( $text );
1700
1701 $url = Sanitizer::cleanUrl( $url );
1702
1703 # Use the encoded URL
1704 # This means that users can paste URLs directly into the text
1705 # Funny characters like ö aren't valid in URLs anyway
1706 # This was changed in August 2004
1707 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1708 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1709
1710 # Register link in the output object.
1711 # Replace unnecessary URL escape codes with the referenced character
1712 # This prevents spammers from hiding links from the filters
1713 $pasteurized = self::replaceUnusualEscapes( $url );
1714 $this->mOutput->addExternalLink( $pasteurized );
1715 }
1716
1717 wfProfileOut( __METHOD__ );
1718 return $s;
1719 }
1720
1721 /**
1722 * Get the rel attribute for a particular external link.
1723 *
1724 * @since 1.21
1725 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1726 * nofollow if appropriate
1727 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1728 * @return string|null Rel attribute for $url
1729 */
1730 public static function getExternalLinkRel( $url = false, $title = null ) {
1731 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1732 $ns = $title ? $title->getNamespace() : false;
1733 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1734 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1735 ) {
1736 return 'nofollow';
1737 }
1738 return null;
1739 }
1740
1741 /**
1742 * Get an associative array of additional HTML attributes appropriate for a
1743 * particular external link. This currently may include rel => nofollow
1744 * (depending on configuration, namespace, and the URL's domain) and/or a
1745 * target attribute (depending on configuration).
1746 *
1747 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1748 * nofollow if appropriate
1749 * @return array Associative array of HTML attributes
1750 */
1751 function getExternalLinkAttribs( $url = false ) {
1752 $attribs = array();
1753 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1754
1755 if ( $this->mOptions->getExternalLinkTarget() ) {
1756 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1757 }
1758 return $attribs;
1759 }
1760
1761 /**
1762 * Replace unusual URL escape codes with their equivalent characters
1763 *
1764 * @param string $url
1765 * @return string
1766 *
1767 * @todo This can merge genuinely required bits in the path or query string,
1768 * breaking legit URLs. A proper fix would treat the various parts of
1769 * the URL differently; as a workaround, just use the output for
1770 * statistical records, not for actual linking/output.
1771 */
1772 static function replaceUnusualEscapes( $url ) {
1773 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1774 array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1775 }
1776
1777 /**
1778 * Callback function used in replaceUnusualEscapes().
1779 * Replaces unusual URL escape codes with their equivalent character
1780 *
1781 * @param array $matches
1782 *
1783 * @return string
1784 */
1785 private static function replaceUnusualEscapesCallback( $matches ) {
1786 $char = urldecode( $matches[0] );
1787 $ord = ord( $char );
1788 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1789 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1790 # No, shouldn't be escaped
1791 return $char;
1792 } else {
1793 # Yes, leave it escaped
1794 return $matches[0];
1795 }
1796 }
1797
1798 /**
1799 * make an image if it's allowed, either through the global
1800 * option, through the exception, or through the on-wiki whitelist
1801 *
1802 * @param string $url
1803 *
1804 * @return string
1805 */
1806 private function maybeMakeExternalImage( $url ) {
1807 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1808 $imagesexception = !empty( $imagesfrom );
1809 $text = false;
1810 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1811 if ( $imagesexception && is_array( $imagesfrom ) ) {
1812 $imagematch = false;
1813 foreach ( $imagesfrom as $match ) {
1814 if ( strpos( $url, $match ) === 0 ) {
1815 $imagematch = true;
1816 break;
1817 }
1818 }
1819 } elseif ( $imagesexception ) {
1820 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1821 } else {
1822 $imagematch = false;
1823 }
1824
1825 if ( $this->mOptions->getAllowExternalImages()
1826 || ( $imagesexception && $imagematch )
1827 ) {
1828 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1829 # Image found
1830 $text = Linker::makeExternalImage( $url );
1831 }
1832 }
1833 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1834 && preg_match( self::EXT_IMAGE_REGEX, $url )
1835 ) {
1836 $whitelist = explode(
1837 "\n",
1838 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
1839 );
1840
1841 foreach ( $whitelist as $entry ) {
1842 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1843 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1844 continue;
1845 }
1846 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1847 # Image matches a whitelist entry
1848 $text = Linker::makeExternalImage( $url );
1849 break;
1850 }
1851 }
1852 }
1853 return $text;
1854 }
1855
1856 /**
1857 * Process [[ ]] wikilinks
1858 *
1859 * @param string $s
1860 *
1861 * @return string Processed text
1862 *
1863 * @private
1864 */
1865 function replaceInternalLinks( $s ) {
1866 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1867 return $s;
1868 }
1869
1870 /**
1871 * Process [[ ]] wikilinks (RIL)
1872 * @param string $s
1873 * @throws MWException
1874 * @return LinkHolderArray
1875 *
1876 * @private
1877 */
1878 function replaceInternalLinks2( &$s ) {
1879 wfProfileIn( __METHOD__ );
1880
1881 wfProfileIn( __METHOD__ . '-setup' );
1882 static $tc = false, $e1, $e1_img;
1883 # the % is needed to support urlencoded titles as well
1884 if ( !$tc ) {
1885 $tc = Title::legalChars() . '#%';
1886 # Match a link having the form [[namespace:link|alternate]]trail
1887 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1888 # Match cases where there is no "]]", which might still be images
1889 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1890 }
1891
1892 $holders = new LinkHolderArray( $this );
1893
1894 # split the entire text string on occurrences of [[
1895 $a = StringUtils::explode( '[[', ' ' . $s );
1896 # get the first element (all text up to first [[), and remove the space we added
1897 $s = $a->current();
1898 $a->next();
1899 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1900 $s = substr( $s, 1 );
1901
1902 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1903 $e2 = null;
1904 if ( $useLinkPrefixExtension ) {
1905 # Match the end of a line for a word that's not followed by whitespace,
1906 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1907 global $wgContLang;
1908 $charset = $wgContLang->linkPrefixCharset();
1909 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
1910 }
1911
1912 if ( is_null( $this->mTitle ) ) {
1913 wfProfileOut( __METHOD__ . '-setup' );
1914 wfProfileOut( __METHOD__ );
1915 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
1916 }
1917 $nottalk = !$this->mTitle->isTalkPage();
1918
1919 if ( $useLinkPrefixExtension ) {
1920 $m = array();
1921 if ( preg_match( $e2, $s, $m ) ) {
1922 $first_prefix = $m[2];
1923 } else {
1924 $first_prefix = false;
1925 }
1926 } else {
1927 $prefix = '';
1928 }
1929
1930 $useSubpages = $this->areSubpagesAllowed();
1931 wfProfileOut( __METHOD__ . '-setup' );
1932
1933 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
1934 # Loop for each link
1935 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
1936 // @codingStandardsIgnoreStart
1937
1938 # Check for excessive memory usage
1939 if ( $holders->isBig() ) {
1940 # Too big
1941 # Do the existence check, replace the link holders and clear the array
1942 $holders->replace( $s );
1943 $holders->clear();
1944 }
1945
1946 if ( $useLinkPrefixExtension ) {
1947 wfProfileIn( __METHOD__ . '-prefixhandling' );
1948 if ( preg_match( $e2, $s, $m ) ) {
1949 $prefix = $m[2];
1950 $s = $m[1];
1951 } else {
1952 $prefix = '';
1953 }
1954 # first link
1955 if ( $first_prefix ) {
1956 $prefix = $first_prefix;
1957 $first_prefix = false;
1958 }
1959 wfProfileOut( __METHOD__ . '-prefixhandling' );
1960 }
1961
1962 $might_be_img = false;
1963
1964 wfProfileIn( __METHOD__ . "-e1" );
1965 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1966 $text = $m[2];
1967 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1968 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1969 # the real problem is with the $e1 regex
1970 # See bug 1300.
1971 #
1972 # Still some problems for cases where the ] is meant to be outside punctuation,
1973 # and no image is in sight. See bug 2095.
1974 #
1975 if ( $text !== ''
1976 && substr( $m[3], 0, 1 ) === ']'
1977 && strpos( $text, '[' ) !== false
1978 ) {
1979 $text .= ']'; # so that replaceExternalLinks($text) works later
1980 $m[3] = substr( $m[3], 1 );
1981 }
1982 # fix up urlencoded title texts
1983 if ( strpos( $m[1], '%' ) !== false ) {
1984 # Should anchors '#' also be rejected?
1985 $m[1] = str_replace( array( '<', '>' ), array( '&lt;', '&gt;' ), rawurldecode( $m[1] ) );
1986 }
1987 $trail = $m[3];
1988 } elseif ( preg_match( $e1_img, $line, $m ) ) {
1989 # Invalid, but might be an image with a link in its caption
1990 $might_be_img = true;
1991 $text = $m[2];
1992 if ( strpos( $m[1], '%' ) !== false ) {
1993 $m[1] = rawurldecode( $m[1] );
1994 }
1995 $trail = "";
1996 } else { # Invalid form; output directly
1997 $s .= $prefix . '[[' . $line;
1998 wfProfileOut( __METHOD__ . "-e1" );
1999 continue;
2000 }
2001 wfProfileOut( __METHOD__ . "-e1" );
2002 wfProfileIn( __METHOD__ . "-misc" );
2003
2004 # Don't allow internal links to pages containing
2005 # PROTO: where PROTO is a valid URL protocol; these
2006 # should be external links.
2007 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $m[1] ) ) {
2008 $s .= $prefix . '[[' . $line;
2009 wfProfileOut( __METHOD__ . "-misc" );
2010 continue;
2011 }
2012
2013 # Make subpage if necessary
2014 if ( $useSubpages ) {
2015 $link = $this->maybeDoSubpageLink( $m[1], $text );
2016 } else {
2017 $link = $m[1];
2018 }
2019
2020 $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
2021 if ( !$noforce ) {
2022 # Strip off leading ':'
2023 $link = substr( $link, 1 );
2024 }
2025
2026 wfProfileOut( __METHOD__ . "-misc" );
2027 wfProfileIn( __METHOD__ . "-title" );
2028 $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
2029 if ( $nt === null ) {
2030 $s .= $prefix . '[[' . $line;
2031 wfProfileOut( __METHOD__ . "-title" );
2032 continue;
2033 }
2034
2035 $ns = $nt->getNamespace();
2036 $iw = $nt->getInterwiki();
2037 wfProfileOut( __METHOD__ . "-title" );
2038
2039 if ( $might_be_img ) { # if this is actually an invalid link
2040 wfProfileIn( __METHOD__ . "-might_be_img" );
2041 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2042 $found = false;
2043 while ( true ) {
2044 # look at the next 'line' to see if we can close it there
2045 $a->next();
2046 $next_line = $a->current();
2047 if ( $next_line === false || $next_line === null ) {
2048 break;
2049 }
2050 $m = explode( ']]', $next_line, 3 );
2051 if ( count( $m ) == 3 ) {
2052 # the first ]] closes the inner link, the second the image
2053 $found = true;
2054 $text .= "[[{$m[0]}]]{$m[1]}";
2055 $trail = $m[2];
2056 break;
2057 } elseif ( count( $m ) == 2 ) {
2058 # if there's exactly one ]] that's fine, we'll keep looking
2059 $text .= "[[{$m[0]}]]{$m[1]}";
2060 } else {
2061 # if $next_line is invalid too, we need look no further
2062 $text .= '[[' . $next_line;
2063 break;
2064 }
2065 }
2066 if ( !$found ) {
2067 # we couldn't find the end of this imageLink, so output it raw
2068 # but don't ignore what might be perfectly normal links in the text we've examined
2069 $holders->merge( $this->replaceInternalLinks2( $text ) );
2070 $s .= "{$prefix}[[$link|$text";
2071 # note: no $trail, because without an end, there *is* no trail
2072 wfProfileOut( __METHOD__ . "-might_be_img" );
2073 continue;
2074 }
2075 } else { # it's not an image, so output it raw
2076 $s .= "{$prefix}[[$link|$text";
2077 # note: no $trail, because without an end, there *is* no trail
2078 wfProfileOut( __METHOD__ . "-might_be_img" );
2079 continue;
2080 }
2081 wfProfileOut( __METHOD__ . "-might_be_img" );
2082 }
2083
2084 $wasblank = ( $text == '' );
2085 if ( $wasblank ) {
2086 $text = $link;
2087 } else {
2088 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2089 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2090 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2091 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2092 $text = $this->doQuotes( $text );
2093 }
2094
2095 # Link not escaped by : , create the various objects
2096 if ( $noforce ) {
2097 # Interwikis
2098 wfProfileIn( __METHOD__ . "-interwiki" );
2099 if ( $iw && $this->mOptions->getInterwikiMagic()
2100 && $nottalk && Language::fetchLanguageName( $iw, null, 'mw' )
2101 ) {
2102 // XXX: the above check prevents links to sites with identifiers that are not language codes
2103
2104 # Bug 24502: filter duplicates
2105 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2106 $this->mLangLinkLanguages[$iw] = true;
2107 $this->mOutput->addLanguageLink( $nt->getFullText() );
2108 }
2109
2110 $s = rtrim( $s . $prefix );
2111 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2112 wfProfileOut( __METHOD__ . "-interwiki" );
2113 continue;
2114 }
2115 wfProfileOut( __METHOD__ . "-interwiki" );
2116
2117 if ( $ns == NS_FILE ) {
2118 wfProfileIn( __METHOD__ . "-image" );
2119 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2120 if ( $wasblank ) {
2121 # if no parameters were passed, $text
2122 # becomes something like "File:Foo.png",
2123 # which we don't want to pass on to the
2124 # image generator
2125 $text = '';
2126 } else {
2127 # recursively parse links inside the image caption
2128 # actually, this will parse them in any other parameters, too,
2129 # but it might be hard to fix that, and it doesn't matter ATM
2130 $text = $this->replaceExternalLinks( $text );
2131 $holders->merge( $this->replaceInternalLinks2( $text ) );
2132 }
2133 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2134 $s .= $prefix . $this->armorLinks(
2135 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2136 } else {
2137 $s .= $prefix . $trail;
2138 }
2139 wfProfileOut( __METHOD__ . "-image" );
2140 continue;
2141 }
2142
2143 if ( $ns == NS_CATEGORY ) {
2144 wfProfileIn( __METHOD__ . "-category" );
2145 $s = rtrim( $s . "\n" ); # bug 87
2146
2147 if ( $wasblank ) {
2148 $sortkey = $this->getDefaultSort();
2149 } else {
2150 $sortkey = $text;
2151 }
2152 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2153 $sortkey = str_replace( "\n", '', $sortkey );
2154 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2155 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2156
2157 /**
2158 * Strip the whitespace Category links produce, see bug 87
2159 */
2160 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2161
2162 wfProfileOut( __METHOD__ . "-category" );
2163 continue;
2164 }
2165 }
2166
2167 # Self-link checking. For some languages, variants of the title are checked in
2168 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2169 # for linking to a different variant.
2170 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2171 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2172 continue;
2173 }
2174
2175 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2176 # @todo FIXME: Should do batch file existence checks, see comment below
2177 if ( $ns == NS_MEDIA ) {
2178 wfProfileIn( __METHOD__ . "-media" );
2179 # Give extensions a chance to select the file revision for us
2180 $options = array();
2181 $descQuery = false;
2182 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2183 array( $this, $nt, &$options, &$descQuery ) );
2184 # Fetch and register the file (file title may be different via hooks)
2185 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2186 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2187 $s .= $prefix . $this->armorLinks(
2188 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2189 wfProfileOut( __METHOD__ . "-media" );
2190 continue;
2191 }
2192
2193 wfProfileIn( __METHOD__ . "-always_known" );
2194 # Some titles, such as valid special pages or files in foreign repos, should
2195 # be shown as bluelinks even though they're not included in the page table
2196 #
2197 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2198 # batch file existence checks for NS_FILE and NS_MEDIA
2199 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2200 $this->mOutput->addLink( $nt );
2201 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2202 } else {
2203 # Links will be added to the output link list after checking
2204 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2205 }
2206 wfProfileOut( __METHOD__ . "-always_known" );
2207 }
2208 wfProfileOut( __METHOD__ );
2209 return $holders;
2210 }
2211
2212 /**
2213 * Render a forced-blue link inline; protect against double expansion of
2214 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2215 * Since this little disaster has to split off the trail text to avoid
2216 * breaking URLs in the following text without breaking trails on the
2217 * wiki links, it's been made into a horrible function.
2218 *
2219 * @param Title $nt
2220 * @param string $text
2221 * @param array|string $query
2222 * @param string $trail
2223 * @param string $prefix
2224 * @return string HTML-wikitext mix oh yuck
2225 */
2226 function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2227 list( $inside, $trail ) = Linker::splitTrail( $trail );
2228
2229 if ( is_string( $query ) ) {
2230 $query = wfCgiToArray( $query );
2231 }
2232 if ( $text == '' ) {
2233 $text = htmlspecialchars( $nt->getPrefixedText() );
2234 }
2235
2236 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2237
2238 return $this->armorLinks( $link ) . $trail;
2239 }
2240
2241 /**
2242 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2243 * going to go through further parsing steps before inline URL expansion.
2244 *
2245 * Not needed quite as much as it used to be since free links are a bit
2246 * more sensible these days. But bracketed links are still an issue.
2247 *
2248 * @param string $text More-or-less HTML
2249 * @return string Less-or-more HTML with NOPARSE bits
2250 */
2251 function armorLinks( $text ) {
2252 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2253 "{$this->mUniqPrefix}NOPARSE$1", $text );
2254 }
2255
2256 /**
2257 * Return true if subpage links should be expanded on this page.
2258 * @return bool
2259 */
2260 function areSubpagesAllowed() {
2261 # Some namespaces don't allow subpages
2262 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2263 }
2264
2265 /**
2266 * Handle link to subpage if necessary
2267 *
2268 * @param string $target The source of the link
2269 * @param string &$text The link text, modified as necessary
2270 * @return string The full name of the link
2271 * @private
2272 */
2273 function maybeDoSubpageLink( $target, &$text ) {
2274 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2275 }
2276
2277 /**#@+
2278 * Used by doBlockLevels()
2279 * @private
2280 *
2281 * @return string
2282 */
2283 function closeParagraph() {
2284 $result = '';
2285 if ( $this->mLastSection != '' ) {
2286 $result = '</' . $this->mLastSection . ">\n";
2287 }
2288 $this->mInPre = false;
2289 $this->mLastSection = '';
2290 return $result;
2291 }
2292
2293 /**
2294 * getCommon() returns the length of the longest common substring
2295 * of both arguments, starting at the beginning of both.
2296 * @private
2297 *
2298 * @param string $st1
2299 * @param string $st2
2300 *
2301 * @return int
2302 */
2303 function getCommon( $st1, $st2 ) {
2304 $fl = strlen( $st1 );
2305 $shorter = strlen( $st2 );
2306 if ( $fl < $shorter ) {
2307 $shorter = $fl;
2308 }
2309
2310 for ( $i = 0; $i < $shorter; ++$i ) {
2311 if ( $st1[$i] != $st2[$i] ) {
2312 break;
2313 }
2314 }
2315 return $i;
2316 }
2317
2318 /**
2319 * These next three functions open, continue, and close the list
2320 * element appropriate to the prefix character passed into them.
2321 * @private
2322 *
2323 * @param string $char
2324 *
2325 * @return string
2326 */
2327 function openList( $char ) {
2328 $result = $this->closeParagraph();
2329
2330 if ( '*' === $char ) {
2331 $result .= "<ul>\n<li>";
2332 } elseif ( '#' === $char ) {
2333 $result .= "<ol>\n<li>";
2334 } elseif ( ':' === $char ) {
2335 $result .= "<dl>\n<dd>";
2336 } elseif ( ';' === $char ) {
2337 $result .= "<dl>\n<dt>";
2338 $this->mDTopen = true;
2339 } else {
2340 $result = '<!-- ERR 1 -->';
2341 }
2342
2343 return $result;
2344 }
2345
2346 /**
2347 * TODO: document
2348 * @param string $char
2349 * @private
2350 *
2351 * @return string
2352 */
2353 function nextItem( $char ) {
2354 if ( '*' === $char || '#' === $char ) {
2355 return "</li>\n<li>";
2356 } elseif ( ':' === $char || ';' === $char ) {
2357 $close = "</dd>\n";
2358 if ( $this->mDTopen ) {
2359 $close = "</dt>\n";
2360 }
2361 if ( ';' === $char ) {
2362 $this->mDTopen = true;
2363 return $close . '<dt>';
2364 } else {
2365 $this->mDTopen = false;
2366 return $close . '<dd>';
2367 }
2368 }
2369 return '<!-- ERR 2 -->';
2370 }
2371
2372 /**
2373 * @todo Document
2374 * @param string $char
2375 * @private
2376 *
2377 * @return string
2378 */
2379 function closeList( $char ) {
2380 if ( '*' === $char ) {
2381 $text = "</li>\n</ul>";
2382 } elseif ( '#' === $char ) {
2383 $text = "</li>\n</ol>";
2384 } elseif ( ':' === $char ) {
2385 if ( $this->mDTopen ) {
2386 $this->mDTopen = false;
2387 $text = "</dt>\n</dl>";
2388 } else {
2389 $text = "</dd>\n</dl>";
2390 }
2391 } else {
2392 return '<!-- ERR 3 -->';
2393 }
2394 return $text . "\n";
2395 }
2396 /**#@-*/
2397
2398 /**
2399 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2400 *
2401 * @param string $text
2402 * @param bool $linestart Whether or not this is at the start of a line.
2403 * @private
2404 * @return string The lists rendered as HTML
2405 */
2406 function doBlockLevels( $text, $linestart ) {
2407 wfProfileIn( __METHOD__ );
2408
2409 # Parsing through the text line by line. The main thing
2410 # happening here is handling of block-level elements p, pre,
2411 # and making lists from lines starting with * # : etc.
2412 #
2413 $textLines = StringUtils::explode( "\n", $text );
2414
2415 $lastPrefix = $output = '';
2416 $this->mDTopen = $inBlockElem = false;
2417 $prefixLength = 0;
2418 $paragraphStack = false;
2419 $inBlockquote = false;
2420
2421 foreach ( $textLines as $oLine ) {
2422 # Fix up $linestart
2423 if ( !$linestart ) {
2424 $output .= $oLine;
2425 $linestart = true;
2426 continue;
2427 }
2428 # * = ul
2429 # # = ol
2430 # ; = dt
2431 # : = dd
2432
2433 $lastPrefixLength = strlen( $lastPrefix );
2434 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2435 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2436 # If not in a <pre> element, scan for and figure out what prefixes are there.
2437 if ( !$this->mInPre ) {
2438 # Multiple prefixes may abut each other for nested lists.
2439 $prefixLength = strspn( $oLine, '*#:;' );
2440 $prefix = substr( $oLine, 0, $prefixLength );
2441
2442 # eh?
2443 # ; and : are both from definition-lists, so they're equivalent
2444 # for the purposes of determining whether or not we need to open/close
2445 # elements.
2446 $prefix2 = str_replace( ';', ':', $prefix );
2447 $t = substr( $oLine, $prefixLength );
2448 $this->mInPre = (bool)$preOpenMatch;
2449 } else {
2450 # Don't interpret any other prefixes in preformatted text
2451 $prefixLength = 0;
2452 $prefix = $prefix2 = '';
2453 $t = $oLine;
2454 }
2455
2456 # List generation
2457 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2458 # Same as the last item, so no need to deal with nesting or opening stuff
2459 $output .= $this->nextItem( substr( $prefix, -1 ) );
2460 $paragraphStack = false;
2461
2462 if ( substr( $prefix, -1 ) === ';' ) {
2463 # The one nasty exception: definition lists work like this:
2464 # ; title : definition text
2465 # So we check for : in the remainder text to split up the
2466 # title and definition, without b0rking links.
2467 $term = $t2 = '';
2468 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2469 $t = $t2;
2470 $output .= $term . $this->nextItem( ':' );
2471 }
2472 }
2473 } elseif ( $prefixLength || $lastPrefixLength ) {
2474 # We need to open or close prefixes, or both.
2475
2476 # Either open or close a level...
2477 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2478 $paragraphStack = false;
2479
2480 # Close all the prefixes which aren't shared.
2481 while ( $commonPrefixLength < $lastPrefixLength ) {
2482 $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
2483 --$lastPrefixLength;
2484 }
2485
2486 # Continue the current prefix if appropriate.
2487 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2488 $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
2489 }
2490
2491 # Open prefixes where appropriate.
2492 while ( $prefixLength > $commonPrefixLength ) {
2493 $char = substr( $prefix, $commonPrefixLength, 1 );
2494 $output .= $this->openList( $char );
2495
2496 if ( ';' === $char ) {
2497 # @todo FIXME: This is dupe of code above
2498 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2499 $t = $t2;
2500 $output .= $term . $this->nextItem( ':' );
2501 }
2502 }
2503 ++$commonPrefixLength;
2504 }
2505 $lastPrefix = $prefix2;
2506 }
2507
2508 # If we have no prefixes, go to paragraph mode.
2509 if ( 0 == $prefixLength ) {
2510 wfProfileIn( __METHOD__ . "-paragraph" );
2511 # No prefix (not in list)--go to paragraph mode
2512 # XXX: use a stack for nestable elements like span, table and div
2513 $openmatch = preg_match(
2514 '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|'
2515 . '<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS',
2516 $t
2517 );
2518 $closematch = preg_match(
2519 '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'
2520 . '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|<\\/mw:|'
2521 . $this->mUniqPrefix
2522 . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS',
2523 $t
2524 );
2525
2526 if ( $openmatch or $closematch ) {
2527 $paragraphStack = false;
2528 # @todo bug 5718: paragraph closed
2529 $output .= $this->closeParagraph();
2530 if ( $preOpenMatch and !$preCloseMatch ) {
2531 $this->mInPre = true;
2532 }
2533 $bqOffset = 0;
2534 while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t, $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset ) ) {
2535 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
2536 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
2537 }
2538 $inBlockElem = !$closematch;
2539 } elseif ( !$inBlockElem && !$this->mInPre ) {
2540 if ( ' ' == substr( $t, 0, 1 )
2541 && ( $this->mLastSection === 'pre' || trim( $t ) != '' )
2542 && !$inBlockquote
2543 ) {
2544 # pre
2545 if ( $this->mLastSection !== 'pre' ) {
2546 $paragraphStack = false;
2547 $output .= $this->closeParagraph() . '<pre>';
2548 $this->mLastSection = 'pre';
2549 }
2550 $t = substr( $t, 1 );
2551 } else {
2552 # paragraph
2553 if ( trim( $t ) === '' ) {
2554 if ( $paragraphStack ) {
2555 $output .= $paragraphStack . '<br />';
2556 $paragraphStack = false;
2557 $this->mLastSection = 'p';
2558 } else {
2559 if ( $this->mLastSection !== 'p' ) {
2560 $output .= $this->closeParagraph();
2561 $this->mLastSection = '';
2562 $paragraphStack = '<p>';
2563 } else {
2564 $paragraphStack = '</p><p>';
2565 }
2566 }
2567 } else {
2568 if ( $paragraphStack ) {
2569 $output .= $paragraphStack;
2570 $paragraphStack = false;
2571 $this->mLastSection = 'p';
2572 } elseif ( $this->mLastSection !== 'p' ) {
2573 $output .= $this->closeParagraph() . '<p>';
2574 $this->mLastSection = 'p';
2575 }
2576 }
2577 }
2578 }
2579 wfProfileOut( __METHOD__ . "-paragraph" );
2580 }
2581 # somewhere above we forget to get out of pre block (bug 785)
2582 if ( $preCloseMatch && $this->mInPre ) {
2583 $this->mInPre = false;
2584 }
2585 if ( $paragraphStack === false ) {
2586 $output .= $t . "\n";
2587 }
2588 }
2589 while ( $prefixLength ) {
2590 $output .= $this->closeList( $prefix2[$prefixLength - 1] );
2591 --$prefixLength;
2592 }
2593 if ( $this->mLastSection != '' ) {
2594 $output .= '</' . $this->mLastSection . '>';
2595 $this->mLastSection = '';
2596 }
2597
2598 wfProfileOut( __METHOD__ );
2599 return $output;
2600 }
2601
2602 /**
2603 * Split up a string on ':', ignoring any occurrences inside tags
2604 * to prevent illegal overlapping.
2605 *
2606 * @param string $str The string to split
2607 * @param string &$before Set to everything before the ':'
2608 * @param string &$after Set to everything after the ':'
2609 * @throws MWException
2610 * @return string The position of the ':', or false if none found
2611 */
2612 function findColonNoLinks( $str, &$before, &$after ) {
2613 wfProfileIn( __METHOD__ );
2614
2615 $pos = strpos( $str, ':' );
2616 if ( $pos === false ) {
2617 # Nothing to find!
2618 wfProfileOut( __METHOD__ );
2619 return false;
2620 }
2621
2622 $lt = strpos( $str, '<' );
2623 if ( $lt === false || $lt > $pos ) {
2624 # Easy; no tag nesting to worry about
2625 $before = substr( $str, 0, $pos );
2626 $after = substr( $str, $pos + 1 );
2627 wfProfileOut( __METHOD__ );
2628 return $pos;
2629 }
2630
2631 # Ugly state machine to walk through avoiding tags.
2632 $state = self::COLON_STATE_TEXT;
2633 $stack = 0;
2634 $len = strlen( $str );
2635 for ( $i = 0; $i < $len; $i++ ) {
2636 $c = $str[$i];
2637
2638 switch ( $state ) {
2639 # (Using the number is a performance hack for common cases)
2640 case 0: # self::COLON_STATE_TEXT:
2641 switch ( $c ) {
2642 case "<":
2643 # Could be either a <start> tag or an </end> tag
2644 $state = self::COLON_STATE_TAGSTART;
2645 break;
2646 case ":":
2647 if ( $stack == 0 ) {
2648 # We found it!
2649 $before = substr( $str, 0, $i );
2650 $after = substr( $str, $i + 1 );
2651 wfProfileOut( __METHOD__ );
2652 return $i;
2653 }
2654 # Embedded in a tag; don't break it.
2655 break;
2656 default:
2657 # Skip ahead looking for something interesting
2658 $colon = strpos( $str, ':', $i );
2659 if ( $colon === false ) {
2660 # Nothing else interesting
2661 wfProfileOut( __METHOD__ );
2662 return false;
2663 }
2664 $lt = strpos( $str, '<', $i );
2665 if ( $stack === 0 ) {
2666 if ( $lt === false || $colon < $lt ) {
2667 # We found it!
2668 $before = substr( $str, 0, $colon );
2669 $after = substr( $str, $colon + 1 );
2670 wfProfileOut( __METHOD__ );
2671 return $i;
2672 }
2673 }
2674 if ( $lt === false ) {
2675 # Nothing else interesting to find; abort!
2676 # We're nested, but there's no close tags left. Abort!
2677 break 2;
2678 }
2679 # Skip ahead to next tag start
2680 $i = $lt;
2681 $state = self::COLON_STATE_TAGSTART;
2682 }
2683 break;
2684 case 1: # self::COLON_STATE_TAG:
2685 # In a <tag>
2686 switch ( $c ) {
2687 case ">":
2688 $stack++;
2689 $state = self::COLON_STATE_TEXT;
2690 break;
2691 case "/":
2692 # Slash may be followed by >?
2693 $state = self::COLON_STATE_TAGSLASH;
2694 break;
2695 default:
2696 # ignore
2697 }
2698 break;
2699 case 2: # self::COLON_STATE_TAGSTART:
2700 switch ( $c ) {
2701 case "/":
2702 $state = self::COLON_STATE_CLOSETAG;
2703 break;
2704 case "!":
2705 $state = self::COLON_STATE_COMMENT;
2706 break;
2707 case ">":
2708 # Illegal early close? This shouldn't happen D:
2709 $state = self::COLON_STATE_TEXT;
2710 break;
2711 default:
2712 $state = self::COLON_STATE_TAG;
2713 }
2714 break;
2715 case 3: # self::COLON_STATE_CLOSETAG:
2716 # In a </tag>
2717 if ( $c === ">" ) {
2718 $stack--;
2719 if ( $stack < 0 ) {
2720 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
2721 wfProfileOut( __METHOD__ );
2722 return false;
2723 }
2724 $state = self::COLON_STATE_TEXT;
2725 }
2726 break;
2727 case self::COLON_STATE_TAGSLASH:
2728 if ( $c === ">" ) {
2729 # Yes, a self-closed tag <blah/>
2730 $state = self::COLON_STATE_TEXT;
2731 } else {
2732 # Probably we're jumping the gun, and this is an attribute
2733 $state = self::COLON_STATE_TAG;
2734 }
2735 break;
2736 case 5: # self::COLON_STATE_COMMENT:
2737 if ( $c === "-" ) {
2738 $state = self::COLON_STATE_COMMENTDASH;
2739 }
2740 break;
2741 case self::COLON_STATE_COMMENTDASH:
2742 if ( $c === "-" ) {
2743 $state = self::COLON_STATE_COMMENTDASHDASH;
2744 } else {
2745 $state = self::COLON_STATE_COMMENT;
2746 }
2747 break;
2748 case self::COLON_STATE_COMMENTDASHDASH:
2749 if ( $c === ">" ) {
2750 $state = self::COLON_STATE_TEXT;
2751 } else {
2752 $state = self::COLON_STATE_COMMENT;
2753 }
2754 break;
2755 default:
2756 wfProfileOut( __METHOD__ );
2757 throw new MWException( "State machine error in " . __METHOD__ );
2758 }
2759 }
2760 if ( $stack > 0 ) {
2761 wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2762 wfProfileOut( __METHOD__ );
2763 return false;
2764 }
2765 wfProfileOut( __METHOD__ );
2766 return false;
2767 }
2768
2769 /**
2770 * Return value of a magic variable (like PAGENAME)
2771 *
2772 * @private
2773 *
2774 * @param int $index
2775 * @param bool|PPFrame $frame
2776 *
2777 * @throws MWException
2778 * @return string
2779 */
2780 function getVariableValue( $index, $frame = false ) {
2781 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2782 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2783
2784 if ( is_null( $this->mTitle ) ) {
2785 // If no title set, bad things are going to happen
2786 // later. Title should always be set since this
2787 // should only be called in the middle of a parse
2788 // operation (but the unit-tests do funky stuff)
2789 throw new MWException( __METHOD__ . ' Should only be '
2790 . ' called while parsing (no title set)' );
2791 }
2792
2793 /**
2794 * Some of these require message or data lookups and can be
2795 * expensive to check many times.
2796 */
2797 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2798 if ( isset( $this->mVarCache[$index] ) ) {
2799 return $this->mVarCache[$index];
2800 }
2801 }
2802
2803 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2804 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2805
2806 $pageLang = $this->getFunctionLang();
2807
2808 switch ( $index ) {
2809 case 'currentmonth':
2810 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2811 break;
2812 case 'currentmonth1':
2813 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2814 break;
2815 case 'currentmonthname':
2816 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2817 break;
2818 case 'currentmonthnamegen':
2819 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2820 break;
2821 case 'currentmonthabbrev':
2822 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2823 break;
2824 case 'currentday':
2825 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2826 break;
2827 case 'currentday2':
2828 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2829 break;
2830 case 'localmonth':
2831 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2832 break;
2833 case 'localmonth1':
2834 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2835 break;
2836 case 'localmonthname':
2837 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2838 break;
2839 case 'localmonthnamegen':
2840 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2841 break;
2842 case 'localmonthabbrev':
2843 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2844 break;
2845 case 'localday':
2846 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2847 break;
2848 case 'localday2':
2849 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2850 break;
2851 case 'pagename':
2852 $value = wfEscapeWikiText( $this->mTitle->getText() );
2853 break;
2854 case 'pagenamee':
2855 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2856 break;
2857 case 'fullpagename':
2858 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2859 break;
2860 case 'fullpagenamee':
2861 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2862 break;
2863 case 'subpagename':
2864 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2865 break;
2866 case 'subpagenamee':
2867 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2868 break;
2869 case 'rootpagename':
2870 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2871 break;
2872 case 'rootpagenamee':
2873 $value = wfEscapeWikiText( wfUrlEncode( str_replace(
2874 ' ',
2875 '_',
2876 $this->mTitle->getRootText()
2877 ) ) );
2878 break;
2879 case 'basepagename':
2880 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2881 break;
2882 case 'basepagenamee':
2883 $value = wfEscapeWikiText( wfUrlEncode( str_replace(
2884 ' ',
2885 '_',
2886 $this->mTitle->getBaseText()
2887 ) ) );
2888 break;
2889 case 'talkpagename':
2890 if ( $this->mTitle->canTalk() ) {
2891 $talkPage = $this->mTitle->getTalkPage();
2892 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2893 } else {
2894 $value = '';
2895 }
2896 break;
2897 case 'talkpagenamee':
2898 if ( $this->mTitle->canTalk() ) {
2899 $talkPage = $this->mTitle->getTalkPage();
2900 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2901 } else {
2902 $value = '';
2903 }
2904 break;
2905 case 'subjectpagename':
2906 $subjPage = $this->mTitle->getSubjectPage();
2907 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2908 break;
2909 case 'subjectpagenamee':
2910 $subjPage = $this->mTitle->getSubjectPage();
2911 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2912 break;
2913 case 'pageid': // requested in bug 23427
2914 $pageid = $this->getTitle()->getArticleID();
2915 if ( $pageid == 0 ) {
2916 # 0 means the page doesn't exist in the database,
2917 # which means the user is previewing a new page.
2918 # The vary-revision flag must be set, because the magic word
2919 # will have a different value once the page is saved.
2920 $this->mOutput->setFlag( 'vary-revision' );
2921 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2922 }
2923 $value = $pageid ? $pageid : null;
2924 break;
2925 case 'revisionid':
2926 # Let the edit saving system know we should parse the page
2927 # *after* a revision ID has been assigned.
2928 $this->mOutput->setFlag( 'vary-revision' );
2929 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2930 $value = $this->mRevisionId;
2931 break;
2932 case 'revisionday':
2933 # Let the edit saving system know we should parse the page
2934 # *after* a revision ID has been assigned. This is for null edits.
2935 $this->mOutput->setFlag( 'vary-revision' );
2936 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2937 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2938 break;
2939 case 'revisionday2':
2940 # Let the edit saving system know we should parse the page
2941 # *after* a revision ID has been assigned. This is for null edits.
2942 $this->mOutput->setFlag( 'vary-revision' );
2943 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2944 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2945 break;
2946 case 'revisionmonth':
2947 # Let the edit saving system know we should parse the page
2948 # *after* a revision ID has been assigned. This is for null edits.
2949 $this->mOutput->setFlag( 'vary-revision' );
2950 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2951 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2952 break;
2953 case 'revisionmonth1':
2954 # Let the edit saving system know we should parse the page
2955 # *after* a revision ID has been assigned. This is for null edits.
2956 $this->mOutput->setFlag( 'vary-revision' );
2957 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2958 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2959 break;
2960 case 'revisionyear':
2961 # Let the edit saving system know we should parse the page
2962 # *after* a revision ID has been assigned. This is for null edits.
2963 $this->mOutput->setFlag( 'vary-revision' );
2964 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2965 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2966 break;
2967 case 'revisiontimestamp':
2968 # Let the edit saving system know we should parse the page
2969 # *after* a revision ID has been assigned. This is for null edits.
2970 $this->mOutput->setFlag( 'vary-revision' );
2971 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2972 $value = $this->getRevisionTimestamp();
2973 break;
2974 case 'revisionuser':
2975 # Let the edit saving system know we should parse the page
2976 # *after* a revision ID has been assigned. This is for null edits.
2977 $this->mOutput->setFlag( 'vary-revision' );
2978 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2979 $value = $this->getRevisionUser();
2980 break;
2981 case 'revisionsize':
2982 # Let the edit saving system know we should parse the page
2983 # *after* a revision ID has been assigned. This is for null edits.
2984 $this->mOutput->setFlag( 'vary-revision' );
2985 wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
2986 $value = $this->getRevisionSize();
2987 break;
2988 case 'namespace':
2989 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2990 break;
2991 case 'namespacee':
2992 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2993 break;
2994 case 'namespacenumber':
2995 $value = $this->mTitle->getNamespace();
2996 break;
2997 case 'talkspace':
2998 $value = $this->mTitle->canTalk()
2999 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
3000 : '';
3001 break;
3002 case 'talkspacee':
3003 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
3004 break;
3005 case 'subjectspace':
3006 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
3007 break;
3008 case 'subjectspacee':
3009 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
3010 break;
3011 case 'currentdayname':
3012 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
3013 break;
3014 case 'currentyear':
3015 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
3016 break;
3017 case 'currenttime':
3018 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
3019 break;
3020 case 'currenthour':
3021 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
3022 break;
3023 case 'currentweek':
3024 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3025 # int to remove the padding
3026 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
3027 break;
3028 case 'currentdow':
3029 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
3030 break;
3031 case 'localdayname':
3032 $value = $pageLang->getWeekdayName(
3033 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
3034 );
3035 break;
3036 case 'localyear':
3037 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
3038 break;
3039 case 'localtime':
3040 $value = $pageLang->time(
3041 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
3042 false,
3043 false
3044 );
3045 break;
3046 case 'localhour':
3047 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
3048 break;
3049 case 'localweek':
3050 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3051 # int to remove the padding
3052 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
3053 break;
3054 case 'localdow':
3055 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
3056 break;
3057 case 'numberofarticles':
3058 $value = $pageLang->formatNum( SiteStats::articles() );
3059 break;
3060 case 'numberoffiles':
3061 $value = $pageLang->formatNum( SiteStats::images() );
3062 break;
3063 case 'numberofusers':
3064 $value = $pageLang->formatNum( SiteStats::users() );
3065 break;
3066 case 'numberofactiveusers':
3067 $value = $pageLang->formatNum( SiteStats::activeUsers() );
3068 break;
3069 case 'numberofpages':
3070 $value = $pageLang->formatNum( SiteStats::pages() );
3071 break;
3072 case 'numberofadmins':
3073 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
3074 break;
3075 case 'numberofedits':
3076 $value = $pageLang->formatNum( SiteStats::edits() );
3077 break;
3078 case 'numberofviews':
3079 global $wgDisableCounters;
3080 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
3081 break;
3082 case 'currenttimestamp':
3083 $value = wfTimestamp( TS_MW, $ts );
3084 break;
3085 case 'localtimestamp':
3086 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
3087 break;
3088 case 'currentversion':
3089 $value = SpecialVersion::getVersion();
3090 break;
3091 case 'articlepath':
3092 return $wgArticlePath;
3093 case 'sitename':
3094 return $wgSitename;
3095 case 'server':
3096 return $wgServer;
3097 case 'servername':
3098 return $wgServerName;
3099 case 'scriptpath':
3100 return $wgScriptPath;
3101 case 'stylepath':
3102 return $wgStylePath;
3103 case 'directionmark':
3104 return $pageLang->getDirMark();
3105 case 'contentlanguage':
3106 global $wgLanguageCode;
3107 return $wgLanguageCode;
3108 case 'cascadingsources':
3109 $value = CoreParserFunctions::cascadingsources( $this );
3110 break;
3111 default:
3112 $ret = null;
3113 wfRunHooks(
3114 'ParserGetVariableValueSwitch',
3115 array( &$this, &$this->mVarCache, &$index, &$ret, &$frame )
3116 );
3117
3118 return $ret;
3119 }
3120
3121 if ( $index ) {
3122 $this->mVarCache[$index] = $value;
3123 }
3124
3125 return $value;
3126 }
3127
3128 /**
3129 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3130 *
3131 * @private
3132 */
3133 function initialiseVariables() {
3134 wfProfileIn( __METHOD__ );
3135 $variableIDs = MagicWord::getVariableIDs();
3136 $substIDs = MagicWord::getSubstIDs();
3137
3138 $this->mVariables = new MagicWordArray( $variableIDs );
3139 $this->mSubstWords = new MagicWordArray( $substIDs );
3140 wfProfileOut( __METHOD__ );
3141 }
3142
3143 /**
3144 * Preprocess some wikitext and return the document tree.
3145 * This is the ghost of replace_variables().
3146 *
3147 * @param string $text The text to parse
3148 * @param int $flags Bitwise combination of:
3149 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3150 * included. Default is to assume a direct page view.
3151 *
3152 * The generated DOM tree must depend only on the input text and the flags.
3153 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
3154 *
3155 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3156 * change in the DOM tree for a given text, must be passed through the section identifier
3157 * in the section edit link and thus back to extractSections().
3158 *
3159 * The output of this function is currently only cached in process memory, but a persistent
3160 * cache may be implemented at a later date which takes further advantage of these strict
3161 * dependency requirements.
3162 *
3163 * @return PPNode
3164 */
3165 function preprocessToDom( $text, $flags = 0 ) {
3166 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3167 return $dom;
3168 }
3169
3170 /**
3171 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3172 *
3173 * @param string $s
3174 *
3175 * @return array
3176 */
3177 public static function splitWhitespace( $s ) {
3178 $ltrimmed = ltrim( $s );
3179 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3180 $trimmed = rtrim( $ltrimmed );
3181 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3182 if ( $diff > 0 ) {
3183 $w2 = substr( $ltrimmed, -$diff );
3184 } else {
3185 $w2 = '';
3186 }
3187 return array( $w1, $trimmed, $w2 );
3188 }
3189
3190 /**
3191 * Replace magic variables, templates, and template arguments
3192 * with the appropriate text. Templates are substituted recursively,
3193 * taking care to avoid infinite loops.
3194 *
3195 * Note that the substitution depends on value of $mOutputType:
3196 * self::OT_WIKI: only {{subst:}} templates
3197 * self::OT_PREPROCESS: templates but not extension tags
3198 * self::OT_HTML: all templates and extension tags
3199 *
3200 * @param string $text The text to transform
3201 * @param bool|PPFrame $frame Object describing the arguments passed to the
3202 * template. Arguments may also be provided as an associative array, as
3203 * was the usual case before MW1.12. Providing arguments this way may be
3204 * useful for extensions wishing to perform variable replacement
3205 * explicitly.
3206 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3207 * double-brace expansion.
3208 * @return string
3209 */
3210 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3211 # Is there any text? Also, Prevent too big inclusions!
3212 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3213 return $text;
3214 }
3215 wfProfileIn( __METHOD__ );
3216
3217 if ( $frame === false ) {
3218 $frame = $this->getPreprocessor()->newFrame();
3219 } elseif ( !( $frame instanceof PPFrame ) ) {
3220 wfDebug( __METHOD__ . " called using plain parameters instead of "
3221 . "a PPFrame instance. Creating custom frame.\n" );
3222 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3223 }
3224
3225 $dom = $this->preprocessToDom( $text );
3226 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3227 $text = $frame->expand( $dom, $flags );
3228
3229 wfProfileOut( __METHOD__ );
3230 return $text;
3231 }
3232
3233 /**
3234 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3235 *
3236 * @param array $args
3237 *
3238 * @return array
3239 */
3240 static function createAssocArgs( $args ) {
3241 $assocArgs = array();
3242 $index = 1;
3243 foreach ( $args as $arg ) {
3244 $eqpos = strpos( $arg, '=' );
3245 if ( $eqpos === false ) {
3246 $assocArgs[$index++] = $arg;
3247 } else {
3248 $name = trim( substr( $arg, 0, $eqpos ) );
3249 $value = trim( substr( $arg, $eqpos + 1 ) );
3250 if ( $value === false ) {
3251 $value = '';
3252 }
3253 if ( $name !== false ) {
3254 $assocArgs[$name] = $value;
3255 }
3256 }
3257 }
3258
3259 return $assocArgs;
3260 }
3261
3262 /**
3263 * Warn the user when a parser limitation is reached
3264 * Will warn at most once the user per limitation type
3265 *
3266 * @param string $limitationType Should be one of:
3267 * 'expensive-parserfunction' (corresponding messages:
3268 * 'expensive-parserfunction-warning',
3269 * 'expensive-parserfunction-category')
3270 * 'post-expand-template-argument' (corresponding messages:
3271 * 'post-expand-template-argument-warning',
3272 * 'post-expand-template-argument-category')
3273 * 'post-expand-template-inclusion' (corresponding messages:
3274 * 'post-expand-template-inclusion-warning',
3275 * 'post-expand-template-inclusion-category')
3276 * 'node-count-exceeded' (corresponding messages:
3277 * 'node-count-exceeded-warning',
3278 * 'node-count-exceeded-category')
3279 * 'expansion-depth-exceeded' (corresponding messages:
3280 * 'expansion-depth-exceeded-warning',
3281 * 'expansion-depth-exceeded-category')
3282 * @param string|int|null $current Current value
3283 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3284 * exceeded, provide the values (optional)
3285 */
3286 function limitationWarn( $limitationType, $current = '', $max = '' ) {
3287 # does no harm if $current and $max are present but are unnecessary for the message
3288 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3289 ->inLanguage( $this->mOptions->getUserLangObj() )->text();
3290 $this->mOutput->addWarning( $warning );
3291 $this->addTrackingCategory( "$limitationType-category" );
3292 }
3293
3294 /**
3295 * Return the text of a template, after recursively
3296 * replacing any variables or templates within the template.
3297 *
3298 * @param array $piece The parts of the template
3299 * $piece['title']: the title, i.e. the part before the |
3300 * $piece['parts']: the parameter array
3301 * $piece['lineStart']: whether the brace was at the start of a line
3302 * @param PPFrame $frame The current frame, contains template arguments
3303 * @throws Exception
3304 * @return string The text of the template
3305 */
3306 public function braceSubstitution( $piece, $frame ) {
3307 wfProfileIn( __METHOD__ );
3308 wfProfileIn( __METHOD__ . '-setup' );
3309
3310 // Flags
3311
3312 // $text has been filled
3313 $found = false;
3314 // wiki markup in $text should be escaped
3315 $nowiki = false;
3316 // $text is HTML, armour it against wikitext transformation
3317 $isHTML = false;
3318 // Force interwiki transclusion to be done in raw mode not rendered
3319 $forceRawInterwiki = false;
3320 // $text is a DOM node needing expansion in a child frame
3321 $isChildObj = false;
3322 // $text is a DOM node needing expansion in the current frame
3323 $isLocalObj = false;
3324
3325 # Title object, where $text came from
3326 $title = false;
3327
3328 # $part1 is the bit before the first |, and must contain only title characters.
3329 # Various prefixes will be stripped from it later.
3330 $titleWithSpaces = $frame->expand( $piece['title'] );
3331 $part1 = trim( $titleWithSpaces );
3332 $titleText = false;
3333
3334 # Original title text preserved for various purposes
3335 $originalTitle = $part1;
3336
3337 # $args is a list of argument nodes, starting from index 0, not including $part1
3338 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3339 # below won't work b/c this $args isn't an object
3340 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3341 wfProfileOut( __METHOD__ . '-setup' );
3342
3343 $titleProfileIn = null; // profile templates
3344
3345 # SUBST
3346 wfProfileIn( __METHOD__ . '-modifiers' );
3347 if ( !$found ) {
3348
3349 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3350
3351 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3352 # Decide whether to expand template or keep wikitext as-is.
3353 if ( $this->ot['wiki'] ) {
3354 if ( $substMatch === false ) {
3355 $literal = true; # literal when in PST with no prefix
3356 } else {
3357 $literal = false; # expand when in PST with subst: or safesubst:
3358 }
3359 } else {
3360 if ( $substMatch == 'subst' ) {
3361 $literal = true; # literal when not in PST with plain subst:
3362 } else {
3363 $literal = false; # expand when not in PST with safesubst: or no prefix
3364 }
3365 }
3366 if ( $literal ) {
3367 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3368 $isLocalObj = true;
3369 $found = true;
3370 }
3371 }
3372
3373 # Variables
3374 if ( !$found && $args->getLength() == 0 ) {
3375 $id = $this->mVariables->matchStartToEnd( $part1 );
3376 if ( $id !== false ) {
3377 $text = $this->getVariableValue( $id, $frame );
3378 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3379 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3380 }
3381 $found = true;
3382 }
3383 }
3384
3385 # MSG, MSGNW and RAW
3386 if ( !$found ) {
3387 # Check for MSGNW:
3388 $mwMsgnw = MagicWord::get( 'msgnw' );
3389 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3390 $nowiki = true;
3391 } else {
3392 # Remove obsolete MSG:
3393 $mwMsg = MagicWord::get( 'msg' );
3394 $mwMsg->matchStartAndRemove( $part1 );
3395 }
3396
3397 # Check for RAW:
3398 $mwRaw = MagicWord::get( 'raw' );
3399 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3400 $forceRawInterwiki = true;
3401 }
3402 }
3403 wfProfileOut( __METHOD__ . '-modifiers' );
3404
3405 # Parser functions
3406 if ( !$found ) {
3407 wfProfileIn( __METHOD__ . '-pfunc' );
3408
3409 $colonPos = strpos( $part1, ':' );
3410 if ( $colonPos !== false ) {
3411 $func = substr( $part1, 0, $colonPos );
3412 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3413 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3414 $funcArgs[] = $args->item( $i );
3415 }
3416 try {
3417 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3418 } catch ( Exception $ex ) {
3419 wfProfileOut( __METHOD__ . '-pfunc' );
3420 wfProfileOut( __METHOD__ );
3421 throw $ex;
3422 }
3423
3424 # The interface for parser functions allows for extracting
3425 # flags into the local scope. Extract any forwarded flags
3426 # here.
3427 extract( $result );
3428 }
3429 wfProfileOut( __METHOD__ . '-pfunc' );
3430 }
3431
3432 # Finish mangling title and then check for loops.
3433 # Set $title to a Title object and $titleText to the PDBK
3434 if ( !$found ) {
3435 $ns = NS_TEMPLATE;
3436 # Split the title into page and subpage
3437 $subpage = '';
3438 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3439 if ( $part1 !== $relative ) {
3440 $part1 = $relative;
3441 $ns = $this->mTitle->getNamespace();
3442 }
3443 $title = Title::newFromText( $part1, $ns );
3444 if ( $title ) {
3445 $titleText = $title->getPrefixedText();
3446 # Check for language variants if the template is not found
3447 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3448 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3449 }
3450 # Do recursion depth check
3451 $limit = $this->mOptions->getMaxTemplateDepth();
3452 if ( $frame->depth >= $limit ) {
3453 $found = true;
3454 $text = '<span class="error">'
3455 . wfMessage( 'parser-template-recursion-depth-warning' )
3456 ->numParams( $limit )->inContentLanguage()->text()
3457 . '</span>';
3458 }
3459 }
3460 }
3461
3462 # Load from database
3463 if ( !$found && $title ) {
3464 if ( !Profiler::instance()->isPersistent() ) {
3465 # Too many unique items can kill profiling DBs/collectors
3466 $titleProfileIn = __METHOD__ . "-title-" . $title->getPrefixedDBkey();
3467 wfProfileIn( $titleProfileIn ); // template in
3468 }
3469 wfProfileIn( __METHOD__ . '-loadtpl' );
3470 if ( !$title->isExternal() ) {
3471 if ( $title->isSpecialPage()
3472 && $this->mOptions->getAllowSpecialInclusion()
3473 && $this->ot['html']
3474 ) {
3475 // Pass the template arguments as URL parameters.
3476 // "uselang" will have no effect since the Language object
3477 // is forced to the one defined in ParserOptions.
3478 $pageArgs = array();
3479 $argsLength = $args->getLength();
3480 for ( $i = 0; $i < $argsLength; $i++ ) {
3481 $bits = $args->item( $i )->splitArg();
3482 if ( strval( $bits['index'] ) === '' ) {
3483 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3484 $value = trim( $frame->expand( $bits['value'] ) );
3485 $pageArgs[$name] = $value;
3486 }
3487 }
3488
3489 // Create a new context to execute the special page
3490 $context = new RequestContext;
3491 $context->setTitle( $title );
3492 $context->setRequest( new FauxRequest( $pageArgs ) );
3493 $context->setUser( $this->getUser() );
3494 $context->setLanguage( $this->mOptions->getUserLangObj() );
3495 $ret = SpecialPageFactory::capturePath( $title, $context );
3496 if ( $ret ) {
3497 $text = $context->getOutput()->getHTML();
3498 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3499 $found = true;
3500 $isHTML = true;
3501 $this->disableCache();
3502 }
3503 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3504 $found = false; # access denied
3505 wfDebug( __METHOD__ . ": template inclusion denied for " .
3506 $title->getPrefixedDBkey() . "\n" );
3507 } else {
3508 list( $text, $title ) = $this->getTemplateDom( $title );
3509 if ( $text !== false ) {
3510 $found = true;
3511 $isChildObj = true;
3512 }
3513 }
3514
3515 # If the title is valid but undisplayable, make a link to it
3516 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3517 $text = "[[:$titleText]]";
3518 $found = true;
3519 }
3520 } elseif ( $title->isTrans() ) {
3521 # Interwiki transclusion
3522 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3523 $text = $this->interwikiTransclude( $title, 'render' );
3524 $isHTML = true;
3525 } else {
3526 $text = $this->interwikiTransclude( $title, 'raw' );
3527 # Preprocess it like a template
3528 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3529 $isChildObj = true;
3530 }
3531 $found = true;
3532 }
3533
3534 # Do infinite loop check
3535 # This has to be done after redirect resolution to avoid infinite loops via redirects
3536 if ( !$frame->loopCheck( $title ) ) {
3537 $found = true;
3538 $text = '<span class="error">'
3539 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3540 . '</span>';
3541 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3542 }
3543 wfProfileOut( __METHOD__ . '-loadtpl' );
3544 }
3545
3546 # If we haven't found text to substitute by now, we're done
3547 # Recover the source wikitext and return it
3548 if ( !$found ) {
3549 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3550 if ( $titleProfileIn ) {
3551 wfProfileOut( $titleProfileIn ); // template out
3552 }
3553 wfProfileOut( __METHOD__ );
3554 return array( 'object' => $text );
3555 }
3556
3557 # Expand DOM-style return values in a child frame
3558 if ( $isChildObj ) {
3559 # Clean up argument array
3560 $newFrame = $frame->newChild( $args, $title );
3561
3562 if ( $nowiki ) {
3563 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3564 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3565 # Expansion is eligible for the empty-frame cache
3566 $text = $newFrame->cachedExpand( $titleText, $text );
3567 } else {
3568 # Uncached expansion
3569 $text = $newFrame->expand( $text );
3570 }
3571 }
3572 if ( $isLocalObj && $nowiki ) {
3573 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3574 $isLocalObj = false;
3575 }
3576
3577 if ( $titleProfileIn ) {
3578 wfProfileOut( $titleProfileIn ); // template out
3579 }
3580
3581 # Replace raw HTML by a placeholder
3582 if ( $isHTML ) {
3583 $text = $this->insertStripItem( $text );
3584 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3585 # Escape nowiki-style return values
3586 $text = wfEscapeWikiText( $text );
3587 } elseif ( is_string( $text )
3588 && !$piece['lineStart']
3589 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3590 ) {
3591 # Bug 529: if the template begins with a table or block-level
3592 # element, it should be treated as beginning a new line.
3593 # This behavior is somewhat controversial.
3594 $text = "\n" . $text;
3595 }
3596
3597 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3598 # Error, oversize inclusion
3599 if ( $titleText !== false ) {
3600 # Make a working, properly escaped link if possible (bug 23588)
3601 $text = "[[:$titleText]]";
3602 } else {
3603 # This will probably not be a working link, but at least it may
3604 # provide some hint of where the problem is
3605 preg_replace( '/^:/', '', $originalTitle );
3606 $text = "[[:$originalTitle]]";
3607 }
3608 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3609 . 'post-expand include size too large -->' );
3610 $this->limitationWarn( 'post-expand-template-inclusion' );
3611 }
3612
3613 if ( $isLocalObj ) {
3614 $ret = array( 'object' => $text );
3615 } else {
3616 $ret = array( 'text' => $text );
3617 }
3618
3619 wfProfileOut( __METHOD__ );
3620 return $ret;
3621 }
3622
3623 /**
3624 * Call a parser function and return an array with text and flags.
3625 *
3626 * The returned array will always contain a boolean 'found', indicating
3627 * whether the parser function was found or not. It may also contain the
3628 * following:
3629 * text: string|object, resulting wikitext or PP DOM object
3630 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3631 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3632 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3633 * nowiki: bool, wiki markup in $text should be escaped
3634 *
3635 * @since 1.21
3636 * @param PPFrame $frame The current frame, contains template arguments
3637 * @param string $function Function name
3638 * @param array $args Arguments to the function
3639 * @throws MWException
3640 * @return array
3641 */
3642 public function callParserFunction( $frame, $function, array $args = array() ) {
3643 global $wgContLang;
3644
3645 wfProfileIn( __METHOD__ );
3646
3647 # Case sensitive functions
3648 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3649 $function = $this->mFunctionSynonyms[1][$function];
3650 } else {
3651 # Case insensitive functions
3652 $function = $wgContLang->lc( $function );
3653 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3654 $function = $this->mFunctionSynonyms[0][$function];
3655 } else {
3656 wfProfileOut( __METHOD__ );
3657 return array( 'found' => false );
3658 }
3659 }
3660
3661 wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3662 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3663
3664 # Workaround for PHP bug 35229 and similar
3665 if ( !is_callable( $callback ) ) {
3666 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3667 wfProfileOut( __METHOD__ );
3668 throw new MWException( "Tag hook for $function is not callable\n" );
3669 }
3670
3671 $allArgs = array( &$this );
3672 if ( $flags & SFH_OBJECT_ARGS ) {
3673 # Convert arguments to PPNodes and collect for appending to $allArgs
3674 $funcArgs = array();
3675 foreach ( $args as $k => $v ) {
3676 if ( $v instanceof PPNode || $k === 0 ) {
3677 $funcArgs[] = $v;
3678 } else {
3679 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( array( $k => $v ) )->item( 0 );
3680 }
3681 }
3682
3683 # Add a frame parameter, and pass the arguments as an array
3684 $allArgs[] = $frame;
3685 $allArgs[] = $funcArgs;
3686 } else {
3687 # Convert arguments to plain text and append to $allArgs
3688 foreach ( $args as $k => $v ) {
3689 if ( $v instanceof PPNode ) {
3690 $allArgs[] = trim( $frame->expand( $v ) );
3691 } elseif ( is_int( $k ) && $k >= 0 ) {
3692 $allArgs[] = trim( $v );
3693 } else {
3694 $allArgs[] = trim( "$k=$v" );
3695 }
3696 }
3697 }
3698
3699 $result = call_user_func_array( $callback, $allArgs );
3700
3701 # The interface for function hooks allows them to return a wikitext
3702 # string or an array containing the string and any flags. This mungs
3703 # things around to match what this method should return.
3704 if ( !is_array( $result ) ) {
3705 $result = array(
3706 'found' => true,
3707 'text' => $result,
3708 );
3709 } else {
3710 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3711 $result['text'] = $result[0];
3712 }
3713 unset( $result[0] );
3714 $result += array(
3715 'found' => true,
3716 );
3717 }
3718
3719 $noparse = true;
3720 $preprocessFlags = 0;
3721 if ( isset( $result['noparse'] ) ) {
3722 $noparse = $result['noparse'];
3723 }
3724 if ( isset( $result['preprocessFlags'] ) ) {
3725 $preprocessFlags = $result['preprocessFlags'];
3726 }
3727
3728 if ( !$noparse ) {
3729 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3730 $result['isChildObj'] = true;
3731 }
3732 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3733 wfProfileOut( __METHOD__ );
3734
3735 return $result;
3736 }
3737
3738 /**
3739 * Get the semi-parsed DOM representation of a template with a given title,
3740 * and its redirect destination title. Cached.
3741 *
3742 * @param Title $title
3743 *
3744 * @return array
3745 */
3746 function getTemplateDom( $title ) {
3747 $cacheTitle = $title;
3748 $titleText = $title->getPrefixedDBkey();
3749
3750 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3751 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3752 $title = Title::makeTitle( $ns, $dbk );
3753 $titleText = $title->getPrefixedDBkey();
3754 }
3755 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3756 return array( $this->mTplDomCache[$titleText], $title );
3757 }
3758
3759 # Cache miss, go to the database
3760 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3761
3762 if ( $text === false ) {
3763 $this->mTplDomCache[$titleText] = false;
3764 return array( false, $title );
3765 }
3766
3767 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3768 $this->mTplDomCache[$titleText] = $dom;
3769
3770 if ( !$title->equals( $cacheTitle ) ) {
3771 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3772 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3773 }
3774
3775 return array( $dom, $title );
3776 }
3777
3778 /**
3779 * Fetch the unparsed text of a template and register a reference to it.
3780 * @param Title $title
3781 * @return array ( string or false, Title )
3782 */
3783 function fetchTemplateAndTitle( $title ) {
3784 // Defaults to Parser::statelessFetchTemplate()
3785 $templateCb = $this->mOptions->getTemplateCallback();
3786 $stuff = call_user_func( $templateCb, $title, $this );
3787 $text = $stuff['text'];
3788 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3789 if ( isset( $stuff['deps'] ) ) {
3790 foreach ( $stuff['deps'] as $dep ) {
3791 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3792 if ( $dep['title']->equals( $this->getTitle() ) ) {
3793 // If we transclude ourselves, the final result
3794 // will change based on the new version of the page
3795 $this->mOutput->setFlag( 'vary-revision' );
3796 }
3797 }
3798 }
3799 return array( $text, $finalTitle );
3800 }
3801
3802 /**
3803 * Fetch the unparsed text of a template and register a reference to it.
3804 * @param Title $title
3805 * @return string|bool
3806 */
3807 function fetchTemplate( $title ) {
3808 $rv = $this->fetchTemplateAndTitle( $title );
3809 return $rv[0];
3810 }
3811
3812 /**
3813 * Static function to get a template
3814 * Can be overridden via ParserOptions::setTemplateCallback().
3815 *
3816 * @param Title $title
3817 * @param bool|Parser $parser
3818 *
3819 * @return array
3820 */
3821 static function statelessFetchTemplate( $title, $parser = false ) {
3822 $text = $skip = false;
3823 $finalTitle = $title;
3824 $deps = array();
3825
3826 # Loop to fetch the article, with up to 1 redirect
3827 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3828 # Give extensions a chance to select the revision instead
3829 $id = false; # Assume current
3830 wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3831 array( $parser, $title, &$skip, &$id ) );
3832
3833 if ( $skip ) {
3834 $text = false;
3835 $deps[] = array(
3836 'title' => $title,
3837 'page_id' => $title->getArticleID(),
3838 'rev_id' => null
3839 );
3840 break;
3841 }
3842 # Get the revision
3843 $rev = $id
3844 ? Revision::newFromId( $id )
3845 : Revision::newFromTitle( $title, false, Revision::READ_NORMAL );
3846 $rev_id = $rev ? $rev->getId() : 0;
3847 # If there is no current revision, there is no page
3848 if ( $id === false && !$rev ) {
3849 $linkCache = LinkCache::singleton();
3850 $linkCache->addBadLinkObj( $title );
3851 }
3852
3853 $deps[] = array(
3854 'title' => $title,
3855 'page_id' => $title->getArticleID(),
3856 'rev_id' => $rev_id );
3857 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3858 # We fetched a rev from a different title; register it too...
3859 $deps[] = array(
3860 'title' => $rev->getTitle(),
3861 'page_id' => $rev->getPage(),
3862 'rev_id' => $rev_id );
3863 }
3864
3865 if ( $rev ) {
3866 $content = $rev->getContent();
3867 $text = $content ? $content->getWikitextForTransclusion() : null;
3868
3869 if ( $text === false || $text === null ) {
3870 $text = false;
3871 break;
3872 }
3873 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3874 global $wgContLang;
3875 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3876 if ( !$message->exists() ) {
3877 $text = false;
3878 break;
3879 }
3880 $content = $message->content();
3881 $text = $message->plain();
3882 } else {
3883 break;
3884 }
3885 if ( !$content ) {
3886 break;
3887 }
3888 # Redirect?
3889 $finalTitle = $title;
3890 $title = $content->getRedirectTarget();
3891 }
3892 return array(
3893 'text' => $text,
3894 'finalTitle' => $finalTitle,
3895 'deps' => $deps );
3896 }
3897
3898 /**
3899 * Fetch a file and its title and register a reference to it.
3900 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3901 * @param Title $title
3902 * @param array $options Array of options to RepoGroup::findFile
3903 * @return File|bool
3904 */
3905 function fetchFile( $title, $options = array() ) {
3906 $res = $this->fetchFileAndTitle( $title, $options );
3907 return $res[0];
3908 }
3909
3910 /**
3911 * Fetch a file and its title and register a reference to it.
3912 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3913 * @param Title $title
3914 * @param array $options Array of options to RepoGroup::findFile
3915 * @return array ( File or false, Title of file )
3916 */
3917 function fetchFileAndTitle( $title, $options = array() ) {
3918 $file = $this->fetchFileNoRegister( $title, $options );
3919
3920 $time = $file ? $file->getTimestamp() : false;
3921 $sha1 = $file ? $file->getSha1() : false;
3922 # Register the file as a dependency...
3923 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3924 if ( $file && !$title->equals( $file->getTitle() ) ) {
3925 # Update fetched file title
3926 $title = $file->getTitle();
3927 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3928 }
3929 return array( $file, $title );
3930 }
3931
3932 /**
3933 * Helper function for fetchFileAndTitle.
3934 *
3935 * Also useful if you need to fetch a file but not use it yet,
3936 * for example to get the file's handler.
3937 *
3938 * @param Title $title
3939 * @param array $options Array of options to RepoGroup::findFile
3940 * @return File|bool
3941 */
3942 protected function fetchFileNoRegister( $title, $options = array() ) {
3943 if ( isset( $options['broken'] ) ) {
3944 $file = false; // broken thumbnail forced by hook
3945 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3946 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3947 } else { // get by (name,timestamp)
3948 $file = wfFindFile( $title, $options );
3949 }
3950 return $file;
3951 }
3952
3953 /**
3954 * Transclude an interwiki link.
3955 *
3956 * @param Title $title
3957 * @param string $action
3958 *
3959 * @return string
3960 */
3961 function interwikiTransclude( $title, $action ) {
3962 global $wgEnableScaryTranscluding;
3963
3964 if ( !$wgEnableScaryTranscluding ) {
3965 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3966 }
3967
3968 $url = $title->getFullURL( array( 'action' => $action ) );
3969
3970 if ( strlen( $url ) > 255 ) {
3971 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3972 }
3973 return $this->fetchScaryTemplateMaybeFromCache( $url );
3974 }
3975
3976 /**
3977 * @param string $url
3978 * @return mixed|string
3979 */
3980 function fetchScaryTemplateMaybeFromCache( $url ) {
3981 global $wgTranscludeCacheExpiry;
3982 $dbr = wfGetDB( DB_SLAVE );
3983 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3984 $obj = $dbr->selectRow( 'transcache', array( 'tc_time', 'tc_contents' ),
3985 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3986 if ( $obj ) {
3987 return $obj->tc_contents;
3988 }
3989
3990 $req = MWHttpRequest::factory( $url );
3991 $status = $req->execute(); // Status object
3992 if ( $status->isOK() ) {
3993 $text = $req->getContent();
3994 } elseif ( $req->getStatus() != 200 ) {
3995 // Though we failed to fetch the content, this status is useless.
3996 return wfMessage( 'scarytranscludefailed-httpstatus' )
3997 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3998 } else {
3999 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4000 }
4001
4002 $dbw = wfGetDB( DB_MASTER );
4003 $dbw->replace( 'transcache', array( 'tc_url' ), array(
4004 'tc_url' => $url,
4005 'tc_time' => $dbw->timestamp( time() ),
4006 'tc_contents' => $text
4007 ) );
4008 return $text;
4009 }
4010
4011 /**
4012 * Triple brace replacement -- used for template arguments
4013 * @private
4014 *
4015 * @param array $piece
4016 * @param PPFrame $frame
4017 *
4018 * @return array
4019 */
4020 function argSubstitution( $piece, $frame ) {
4021 wfProfileIn( __METHOD__ );
4022
4023 $error = false;
4024 $parts = $piece['parts'];
4025 $nameWithSpaces = $frame->expand( $piece['title'] );
4026 $argName = trim( $nameWithSpaces );
4027 $object = false;
4028 $text = $frame->getArgument( $argName );
4029 if ( $text === false && $parts->getLength() > 0
4030 && ( $this->ot['html']
4031 || $this->ot['pre']
4032 || ( $this->ot['wiki'] && $frame->isTemplate() )
4033 )
4034 ) {
4035 # No match in frame, use the supplied default
4036 $object = $parts->item( 0 )->getChildren();
4037 }
4038 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4039 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4040 $this->limitationWarn( 'post-expand-template-argument' );
4041 }
4042
4043 if ( $text === false && $object === false ) {
4044 # No match anywhere
4045 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4046 }
4047 if ( $error !== false ) {
4048 $text .= $error;
4049 }
4050 if ( $object !== false ) {
4051 $ret = array( 'object' => $object );
4052 } else {
4053 $ret = array( 'text' => $text );
4054 }
4055
4056 wfProfileOut( __METHOD__ );
4057 return $ret;
4058 }
4059
4060 /**
4061 * Return the text to be used for a given extension tag.
4062 * This is the ghost of strip().
4063 *
4064 * @param array $params Associative array of parameters:
4065 * name PPNode for the tag name
4066 * attr PPNode for unparsed text where tag attributes are thought to be
4067 * attributes Optional associative array of parsed attributes
4068 * inner Contents of extension element
4069 * noClose Original text did not have a close tag
4070 * @param PPFrame $frame
4071 *
4072 * @throws MWException
4073 * @return string
4074 */
4075 function extensionSubstitution( $params, $frame ) {
4076 $name = $frame->expand( $params['name'] );
4077 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
4078 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
4079 $marker = "{$this->mUniqPrefix}-$name-"
4080 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
4081
4082 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4083 ( $this->ot['html'] || $this->ot['pre'] );
4084 if ( $isFunctionTag ) {
4085 $markerType = 'none';
4086 } else {
4087 $markerType = 'general';
4088 }
4089 if ( $this->ot['html'] || $isFunctionTag ) {
4090 $name = strtolower( $name );
4091 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4092 if ( isset( $params['attributes'] ) ) {
4093 $attributes = $attributes + $params['attributes'];
4094 }
4095
4096 if ( isset( $this->mTagHooks[$name] ) ) {
4097 # Workaround for PHP bug 35229 and similar
4098 if ( !is_callable( $this->mTagHooks[$name] ) ) {
4099 throw new MWException( "Tag hook for $name is not callable\n" );
4100 }
4101 $output = call_user_func_array( $this->mTagHooks[$name],
4102 array( $content, $attributes, $this, $frame ) );
4103 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4104 list( $callback, ) = $this->mFunctionTagHooks[$name];
4105 if ( !is_callable( $callback ) ) {
4106 throw new MWException( "Tag hook for $name is not callable\n" );
4107 }
4108
4109 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
4110 } else {
4111 $output = '<span class="error">Invalid tag extension name: ' .
4112 htmlspecialchars( $name ) . '</span>';
4113 }
4114
4115 if ( is_array( $output ) ) {
4116 # Extract flags to local scope (to override $markerType)
4117 $flags = $output;
4118 $output = $flags[0];
4119 unset( $flags[0] );
4120 extract( $flags );
4121 }
4122 } else {
4123 if ( is_null( $attrText ) ) {
4124 $attrText = '';
4125 }
4126 if ( isset( $params['attributes'] ) ) {
4127 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4128 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4129 htmlspecialchars( $attrValue ) . '"';
4130 }
4131 }
4132 if ( $content === null ) {
4133 $output = "<$name$attrText/>";
4134 } else {
4135 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4136 $output = "<$name$attrText>$content$close";
4137 }
4138 }
4139
4140 if ( $markerType === 'none' ) {
4141 return $output;
4142 } elseif ( $markerType === 'nowiki' ) {
4143 $this->mStripState->addNoWiki( $marker, $output );
4144 } elseif ( $markerType === 'general' ) {
4145 $this->mStripState->addGeneral( $marker, $output );
4146 } else {
4147 throw new MWException( __METHOD__ . ': invalid marker type' );
4148 }
4149 return $marker;
4150 }
4151
4152 /**
4153 * Increment an include size counter
4154 *
4155 * @param string $type The type of expansion
4156 * @param int $size The size of the text
4157 * @return bool False if this inclusion would take it over the maximum, true otherwise
4158 */
4159 function incrementIncludeSize( $type, $size ) {
4160 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4161 return false;
4162 } else {
4163 $this->mIncludeSizes[$type] += $size;
4164 return true;
4165 }
4166 }
4167
4168 /**
4169 * Increment the expensive function count
4170 *
4171 * @return bool False if the limit has been exceeded
4172 */
4173 function incrementExpensiveFunctionCount() {
4174 $this->mExpensiveFunctionCount++;
4175 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4176 }
4177
4178 /**
4179 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4180 * Fills $this->mDoubleUnderscores, returns the modified text
4181 *
4182 * @param string $text
4183 *
4184 * @return string
4185 */
4186 function doDoubleUnderscore( $text ) {
4187 wfProfileIn( __METHOD__ );
4188
4189 # The position of __TOC__ needs to be recorded
4190 $mw = MagicWord::get( 'toc' );
4191 if ( $mw->match( $text ) ) {
4192 $this->mShowToc = true;
4193 $this->mForceTocPosition = true;
4194
4195 # Set a placeholder. At the end we'll fill it in with the TOC.
4196 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
4197
4198 # Only keep the first one.
4199 $text = $mw->replace( '', $text );
4200 }
4201
4202 # Now match and remove the rest of them
4203 $mwa = MagicWord::getDoubleUnderscoreArray();
4204 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4205
4206 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4207 $this->mOutput->mNoGallery = true;
4208 }
4209 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4210 $this->mShowToc = false;
4211 }
4212 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4213 && $this->mTitle->getNamespace() == NS_CATEGORY
4214 ) {
4215 $this->addTrackingCategory( 'hidden-category-category' );
4216 }
4217 # (bug 8068) Allow control over whether robots index a page.
4218 #
4219 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
4220 # is not desirable, the last one on the page should win.
4221 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4222 $this->mOutput->setIndexPolicy( 'noindex' );
4223 $this->addTrackingCategory( 'noindex-category' );
4224 }
4225 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4226 $this->mOutput->setIndexPolicy( 'index' );
4227 $this->addTrackingCategory( 'index-category' );
4228 }
4229
4230 # Cache all double underscores in the database
4231 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4232 $this->mOutput->setProperty( $key, '' );
4233 }
4234
4235 wfProfileOut( __METHOD__ );
4236 return $text;
4237 }
4238
4239 /**
4240 * Add a tracking category, getting the title from a system message,
4241 * or print a debug message if the title is invalid.
4242 *
4243 * Please add any message that you use with this function to
4244 * $wgTrackingCategories. That way they will be listed on
4245 * Special:TrackingCategories.
4246 *
4247 * @param string $msg Message key
4248 * @return bool Whether the addition was successful
4249 */
4250 public function addTrackingCategory( $msg ) {
4251 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
4252 wfDebug( __METHOD__ . ": Not adding tracking category $msg to special page!\n" );
4253 return false;
4254 }
4255 // Important to parse with correct title (bug 31469)
4256 $cat = wfMessage( $msg )
4257 ->title( $this->getTitle() )
4258 ->inContentLanguage()
4259 ->text();
4260
4261 # Allow tracking categories to be disabled by setting them to "-"
4262 if ( $cat === '-' ) {
4263 return false;
4264 }
4265
4266 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
4267 if ( $containerCategory ) {
4268 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
4269 return true;
4270 } else {
4271 wfDebug( __METHOD__ . ": [[MediaWiki:$msg]] is not a valid title!\n" );
4272 return false;
4273 }
4274 }
4275
4276 /**
4277 * This function accomplishes several tasks:
4278 * 1) Auto-number headings if that option is enabled
4279 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4280 * 3) Add a Table of contents on the top for users who have enabled the option
4281 * 4) Auto-anchor headings
4282 *
4283 * It loops through all headlines, collects the necessary data, then splits up the
4284 * string and re-inserts the newly formatted headlines.
4285 *
4286 * @param string $text
4287 * @param string $origText Original, untouched wikitext
4288 * @param bool $isMain
4289 * @return mixed|string
4290 * @private
4291 */
4292 function formatHeadings( $text, $origText, $isMain = true ) {
4293 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4294
4295 # Inhibit editsection links if requested in the page
4296 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4297 $maybeShowEditLink = $showEditLink = false;
4298 } else {
4299 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4300 $showEditLink = $this->mOptions->getEditSection();
4301 }
4302 if ( $showEditLink ) {
4303 $this->mOutput->setEditSectionTokens( true );
4304 }
4305
4306 # Get all headlines for numbering them and adding funky stuff like [edit]
4307 # links - this is for later, but we need the number of headlines right now
4308 $matches = array();
4309 $numMatches = preg_match_all(
4310 '/<H(?P<level>[1-6])(?P<attrib>.*?' . '>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4311 $text,
4312 $matches
4313 );
4314
4315 # if there are fewer than 4 headlines in the article, do not show TOC
4316 # unless it's been explicitly enabled.
4317 $enoughToc = $this->mShowToc &&
4318 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4319
4320 # Allow user to stipulate that a page should have a "new section"
4321 # link added via __NEWSECTIONLINK__
4322 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4323 $this->mOutput->setNewSection( true );
4324 }
4325
4326 # Allow user to remove the "new section"
4327 # link via __NONEWSECTIONLINK__
4328 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4329 $this->mOutput->hideNewSection( true );
4330 }
4331
4332 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4333 # override above conditions and always show TOC above first header
4334 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4335 $this->mShowToc = true;
4336 $enoughToc = true;
4337 }
4338
4339 # headline counter
4340 $headlineCount = 0;
4341 $numVisible = 0;
4342
4343 # Ugh .. the TOC should have neat indentation levels which can be
4344 # passed to the skin functions. These are determined here
4345 $toc = '';
4346 $full = '';
4347 $head = array();
4348 $sublevelCount = array();
4349 $levelCount = array();
4350 $level = 0;
4351 $prevlevel = 0;
4352 $toclevel = 0;
4353 $prevtoclevel = 0;
4354 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4355 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4356 $oldType = $this->mOutputType;
4357 $this->setOutputType( self::OT_WIKI );
4358 $frame = $this->getPreprocessor()->newFrame();
4359 $root = $this->preprocessToDom( $origText );
4360 $node = $root->getFirstChild();
4361 $byteOffset = 0;
4362 $tocraw = array();
4363 $refers = array();
4364
4365 foreach ( $matches[3] as $headline ) {
4366 $isTemplate = false;
4367 $titleText = false;
4368 $sectionIndex = false;
4369 $numbering = '';
4370 $markerMatches = array();
4371 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4372 $serial = $markerMatches[1];
4373 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4374 $isTemplate = ( $titleText != $baseTitleText );
4375 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4376 }
4377
4378 if ( $toclevel ) {
4379 $prevlevel = $level;
4380 }
4381 $level = $matches[1][$headlineCount];
4382
4383 if ( $level > $prevlevel ) {
4384 # Increase TOC level
4385 $toclevel++;
4386 $sublevelCount[$toclevel] = 0;
4387 if ( $toclevel < $wgMaxTocLevel ) {
4388 $prevtoclevel = $toclevel;
4389 $toc .= Linker::tocIndent();
4390 $numVisible++;
4391 }
4392 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4393 # Decrease TOC level, find level to jump to
4394
4395 for ( $i = $toclevel; $i > 0; $i-- ) {
4396 if ( $levelCount[$i] == $level ) {
4397 # Found last matching level
4398 $toclevel = $i;
4399 break;
4400 } elseif ( $levelCount[$i] < $level ) {
4401 # Found first matching level below current level
4402 $toclevel = $i + 1;
4403 break;
4404 }
4405 }
4406 if ( $i == 0 ) {
4407 $toclevel = 1;
4408 }
4409 if ( $toclevel < $wgMaxTocLevel ) {
4410 if ( $prevtoclevel < $wgMaxTocLevel ) {
4411 # Unindent only if the previous toc level was shown :p
4412 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4413 $prevtoclevel = $toclevel;
4414 } else {
4415 $toc .= Linker::tocLineEnd();
4416 }
4417 }
4418 } else {
4419 # No change in level, end TOC line
4420 if ( $toclevel < $wgMaxTocLevel ) {
4421 $toc .= Linker::tocLineEnd();
4422 }
4423 }
4424
4425 $levelCount[$toclevel] = $level;
4426
4427 # count number of headlines for each level
4428 $sublevelCount[$toclevel]++;
4429 $dot = 0;
4430 for ( $i = 1; $i <= $toclevel; $i++ ) {
4431 if ( !empty( $sublevelCount[$i] ) ) {
4432 if ( $dot ) {
4433 $numbering .= '.';
4434 }
4435 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4436 $dot = 1;
4437 }
4438 }
4439
4440 # The safe header is a version of the header text safe to use for links
4441
4442 # Remove link placeholders by the link text.
4443 # <!--LINK number-->
4444 # turns into
4445 # link text with suffix
4446 # Do this before unstrip since link text can contain strip markers
4447 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4448
4449 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4450 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4451
4452 # Strip out HTML (first regex removes any tag not allowed)
4453 # Allowed tags are:
4454 # * <sup> and <sub> (bug 8393)
4455 # * <i> (bug 26375)
4456 # * <b> (r105284)
4457 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4458 #
4459 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4460 # to allow setting directionality in toc items.
4461 $tocline = preg_replace(
4462 array(
4463 '#<(?!/?(span|sup|sub|i|b)(?: [^>]*)?>).*?' . '>#',
4464 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|i|b))(?: .*?)?' . '>#'
4465 ),
4466 array( '', '<$1>' ),
4467 $safeHeadline
4468 );
4469 $tocline = trim( $tocline );
4470
4471 # For the anchor, strip out HTML-y stuff period
4472 $safeHeadline = preg_replace( '/<.*?' . '>/', '', $safeHeadline );
4473 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4474
4475 # Save headline for section edit hint before it's escaped
4476 $headlineHint = $safeHeadline;
4477
4478 if ( $wgExperimentalHtmlIds ) {
4479 # For reverse compatibility, provide an id that's
4480 # HTML4-compatible, like we used to.
4481 #
4482 # It may be worth noting, academically, that it's possible for
4483 # the legacy anchor to conflict with a non-legacy headline
4484 # anchor on the page. In this case likely the "correct" thing
4485 # would be to either drop the legacy anchors or make sure
4486 # they're numbered first. However, this would require people
4487 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4488 # manually, so let's not bother worrying about it.
4489 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4490 array( 'noninitial', 'legacy' ) );
4491 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4492
4493 if ( $legacyHeadline == $safeHeadline ) {
4494 # No reason to have both (in fact, we can't)
4495 $legacyHeadline = false;
4496 }
4497 } else {
4498 $legacyHeadline = false;
4499 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4500 'noninitial' );
4501 }
4502
4503 # HTML names must be case-insensitively unique (bug 10721).
4504 # This does not apply to Unicode characters per
4505 # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4506 # @todo FIXME: We may be changing them depending on the current locale.
4507 $arrayKey = strtolower( $safeHeadline );
4508 if ( $legacyHeadline === false ) {
4509 $legacyArrayKey = false;
4510 } else {
4511 $legacyArrayKey = strtolower( $legacyHeadline );
4512 }
4513
4514 # count how many in assoc. array so we can track dupes in anchors
4515 if ( isset( $refers[$arrayKey] ) ) {
4516 $refers[$arrayKey]++;
4517 } else {
4518 $refers[$arrayKey] = 1;
4519 }
4520 if ( isset( $refers[$legacyArrayKey] ) ) {
4521 $refers[$legacyArrayKey]++;
4522 } else {
4523 $refers[$legacyArrayKey] = 1;
4524 }
4525
4526 # Don't number the heading if it is the only one (looks silly)
4527 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4528 # the two are different if the line contains a link
4529 $headline = Html::element(
4530 'span',
4531 array( 'class' => 'mw-headline-number' ),
4532 $numbering
4533 ) . ' ' . $headline;
4534 }
4535
4536 # Create the anchor for linking from the TOC to the section
4537 $anchor = $safeHeadline;
4538 $legacyAnchor = $legacyHeadline;
4539 if ( $refers[$arrayKey] > 1 ) {
4540 $anchor .= '_' . $refers[$arrayKey];
4541 }
4542 if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4543 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4544 }
4545 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4546 $toc .= Linker::tocLine( $anchor, $tocline,
4547 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4548 }
4549
4550 # Add the section to the section tree
4551 # Find the DOM node for this header
4552 $noOffset = ( $isTemplate || $sectionIndex === false );
4553 while ( $node && !$noOffset ) {
4554 if ( $node->getName() === 'h' ) {
4555 $bits = $node->splitHeading();
4556 if ( $bits['i'] == $sectionIndex ) {
4557 break;
4558 }
4559 }
4560 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4561 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4562 $node = $node->getNextSibling();
4563 }
4564 $tocraw[] = array(
4565 'toclevel' => $toclevel,
4566 'level' => $level,
4567 'line' => $tocline,
4568 'number' => $numbering,
4569 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4570 'fromtitle' => $titleText,
4571 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4572 'anchor' => $anchor,
4573 );
4574
4575 # give headline the correct <h#> tag
4576 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4577 // Output edit section links as markers with styles that can be customized by skins
4578 if ( $isTemplate ) {
4579 # Put a T flag in the section identifier, to indicate to extractSections()
4580 # that sections inside <includeonly> should be counted.
4581 $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4582 } else {
4583 $editlinkArgs = array(
4584 $this->mTitle->getPrefixedText(),
4585 $sectionIndex,
4586 $headlineHint
4587 );
4588 }
4589 // We use a bit of pesudo-xml for editsection markers. The
4590 // language converter is run later on. Using a UNIQ style marker
4591 // leads to the converter screwing up the tokens when it
4592 // converts stuff. And trying to insert strip tags fails too. At
4593 // this point all real inputted tags have already been escaped,
4594 // so we don't have to worry about a user trying to input one of
4595 // these markers directly. We use a page and section attribute
4596 // to stop the language converter from converting these
4597 // important bits of data, but put the headline hint inside a
4598 // content block because the language converter is supposed to
4599 // be able to convert that piece of data.
4600 $editlink = '<mw:editsection page="' . htmlspecialchars( $editlinkArgs[0] );
4601 $editlink .= '" section="' . htmlspecialchars( $editlinkArgs[1] ) . '"';
4602 if ( isset( $editlinkArgs[2] ) ) {
4603 $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4604 } else {
4605 $editlink .= '/>';
4606 }
4607 } else {
4608 $editlink = '';
4609 }
4610 $head[$headlineCount] = Linker::makeHeadline( $level,
4611 $matches['attrib'][$headlineCount], $anchor, $headline,
4612 $editlink, $legacyAnchor );
4613
4614 $headlineCount++;
4615 }
4616
4617 $this->setOutputType( $oldType );
4618
4619 # Never ever show TOC if no headers
4620 if ( $numVisible < 1 ) {
4621 $enoughToc = false;
4622 }
4623
4624 if ( $enoughToc ) {
4625 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4626 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4627 }
4628 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4629 $this->mOutput->setTOCHTML( $toc );
4630 $toc = self::TOC_START . $toc . self::TOC_END;
4631 $this->mOutput->addModules( 'mediawiki.toc' );
4632 }
4633
4634 if ( $isMain ) {
4635 $this->mOutput->setSections( $tocraw );
4636 }
4637
4638 # split up and insert constructed headlines
4639 $blocks = preg_split( '/<H[1-6].*?' . '>[\s\S]*?<\/H[1-6]>/i', $text );
4640 $i = 0;
4641
4642 // build an array of document sections
4643 $sections = array();
4644 foreach ( $blocks as $block ) {
4645 // $head is zero-based, sections aren't.
4646 if ( empty( $head[$i - 1] ) ) {
4647 $sections[$i] = $block;
4648 } else {
4649 $sections[$i] = $head[$i - 1] . $block;
4650 }
4651
4652 /**
4653 * Send a hook, one per section.
4654 * The idea here is to be able to make section-level DIVs, but to do so in a
4655 * lower-impact, more correct way than r50769
4656 *
4657 * $this : caller
4658 * $section : the section number
4659 * &$sectionContent : ref to the content of the section
4660 * $showEditLinks : boolean describing whether this section has an edit link
4661 */
4662 wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4663
4664 $i++;
4665 }
4666
4667 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4668 // append the TOC at the beginning
4669 // Top anchor now in skin
4670 $sections[0] = $sections[0] . $toc . "\n";
4671 }
4672
4673 $full .= join( '', $sections );
4674
4675 if ( $this->mForceTocPosition ) {
4676 return str_replace( '<!--MWTOC-->', $toc, $full );
4677 } else {
4678 return $full;
4679 }
4680 }
4681
4682 /**
4683 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4684 * conversion, substitting signatures, {{subst:}} templates, etc.
4685 *
4686 * @param string $text The text to transform
4687 * @param Title $title The Title object for the current article
4688 * @param User $user The User object describing the current user
4689 * @param ParserOptions $options Parsing options
4690 * @param bool $clearState Whether to clear the parser state first
4691 * @return string The altered wiki markup
4692 */
4693 public function preSaveTransform( $text, Title $title, User $user,
4694 ParserOptions $options, $clearState = true
4695 ) {
4696 if ( $clearState ) {
4697 $magicScopeVariable = $this->lock();
4698 }
4699 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4700 $this->setUser( $user );
4701
4702 $pairs = array(
4703 "\r\n" => "\n",
4704 );
4705 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4706 if ( $options->getPreSaveTransform() ) {
4707 $text = $this->pstPass2( $text, $user );
4708 }
4709 $text = $this->mStripState->unstripBoth( $text );
4710
4711 $this->setUser( null ); #Reset
4712
4713 return $text;
4714 }
4715
4716 /**
4717 * Pre-save transform helper function
4718 *
4719 * @param string $text
4720 * @param User $user
4721 *
4722 * @return string
4723 */
4724 private function pstPass2( $text, $user ) {
4725 global $wgContLang;
4726
4727 # Note: This is the timestamp saved as hardcoded wikitext to
4728 # the database, we use $wgContLang here in order to give
4729 # everyone the same signature and use the default one rather
4730 # than the one selected in each user's preferences.
4731 # (see also bug 12815)
4732 $ts = $this->mOptions->getTimestamp();
4733 $timestamp = MWTimestamp::getLocalInstance( $ts );
4734 $ts = $timestamp->format( 'YmdHis' );
4735 $tzMsg = $timestamp->format( 'T' ); # might vary on DST changeover!
4736
4737 # Allow translation of timezones through wiki. format() can return
4738 # whatever crap the system uses, localised or not, so we cannot
4739 # ship premade translations.
4740 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4741 $msg = wfMessage( $key )->inContentLanguage();
4742 if ( $msg->exists() ) {
4743 $tzMsg = $msg->text();
4744 }
4745
4746 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4747
4748 # Variable replacement
4749 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4750 $text = $this->replaceVariables( $text );
4751
4752 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4753 # which may corrupt this parser instance via its wfMessage()->text() call-
4754
4755 # Signatures
4756 $sigText = $this->getUserSig( $user );
4757 $text = strtr( $text, array(
4758 '~~~~~' => $d,
4759 '~~~~' => "$sigText $d",
4760 '~~~' => $sigText
4761 ) );
4762
4763 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4764 $tc = '[' . Title::legalChars() . ']';
4765 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4766
4767 // [[ns:page (context)|]]
4768 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4769 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4770 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4771 // [[ns:page (context), context|]] (using either single or double-width comma)
4772 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4773 // [[|page]] (reverse pipe trick: add context from page title)
4774 $p2 = "/\[\[\\|($tc+)]]/";
4775
4776 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4777 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4778 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4779 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4780
4781 $t = $this->mTitle->getText();
4782 $m = array();
4783 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4784 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4785 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4786 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4787 } else {
4788 # if there's no context, don't bother duplicating the title
4789 $text = preg_replace( $p2, '[[\\1]]', $text );
4790 }
4791
4792 # Trim trailing whitespace
4793 $text = rtrim( $text );
4794
4795 return $text;
4796 }
4797
4798 /**
4799 * Fetch the user's signature text, if any, and normalize to
4800 * validated, ready-to-insert wikitext.
4801 * If you have pre-fetched the nickname or the fancySig option, you can
4802 * specify them here to save a database query.
4803 * Do not reuse this parser instance after calling getUserSig(),
4804 * as it may have changed if it's the $wgParser.
4805 *
4806 * @param User $user
4807 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4808 * @param bool|null $fancySig whether the nicknname is the complete signature
4809 * or null to use default value
4810 * @return string
4811 */
4812 function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4813 global $wgMaxSigChars;
4814
4815 $username = $user->getName();
4816
4817 # If not given, retrieve from the user object.
4818 if ( $nickname === false ) {
4819 $nickname = $user->getOption( 'nickname' );
4820 }
4821
4822 if ( is_null( $fancySig ) ) {
4823 $fancySig = $user->getBoolOption( 'fancysig' );
4824 }
4825
4826 $nickname = $nickname == null ? $username : $nickname;
4827
4828 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4829 $nickname = $username;
4830 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4831 } elseif ( $fancySig !== false ) {
4832 # Sig. might contain markup; validate this
4833 if ( $this->validateSig( $nickname ) !== false ) {
4834 # Validated; clean up (if needed) and return it
4835 return $this->cleanSig( $nickname, true );
4836 } else {
4837 # Failed to validate; fall back to the default
4838 $nickname = $username;
4839 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4840 }
4841 }
4842
4843 # Make sure nickname doesnt get a sig in a sig
4844 $nickname = self::cleanSigInSig( $nickname );
4845
4846 # If we're still here, make it a link to the user page
4847 $userText = wfEscapeWikiText( $username );
4848 $nickText = wfEscapeWikiText( $nickname );
4849 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4850
4851 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4852 ->title( $this->getTitle() )->text();
4853 }
4854
4855 /**
4856 * Check that the user's signature contains no bad XML
4857 *
4858 * @param string $text
4859 * @return string|bool An expanded string, or false if invalid.
4860 */
4861 function validateSig( $text ) {
4862 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4863 }
4864
4865 /**
4866 * Clean up signature text
4867 *
4868 * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4869 * 2) Substitute all transclusions
4870 *
4871 * @param string $text
4872 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4873 * @return string Signature text
4874 */
4875 public function cleanSig( $text, $parsing = false ) {
4876 if ( !$parsing ) {
4877 global $wgTitle;
4878 $magicScopeVariable = $this->lock();
4879 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4880 }
4881
4882 # Option to disable this feature
4883 if ( !$this->mOptions->getCleanSignatures() ) {
4884 return $text;
4885 }
4886
4887 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4888 # => Move this logic to braceSubstitution()
4889 $substWord = MagicWord::get( 'subst' );
4890 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4891 $substText = '{{' . $substWord->getSynonym( 0 );
4892
4893 $text = preg_replace( $substRegex, $substText, $text );
4894 $text = self::cleanSigInSig( $text );
4895 $dom = $this->preprocessToDom( $text );
4896 $frame = $this->getPreprocessor()->newFrame();
4897 $text = $frame->expand( $dom );
4898
4899 if ( !$parsing ) {
4900 $text = $this->mStripState->unstripBoth( $text );
4901 }
4902
4903 return $text;
4904 }
4905
4906 /**
4907 * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4908 *
4909 * @param string $text
4910 * @return string Signature text with /~{3,5}/ removed
4911 */
4912 public static function cleanSigInSig( $text ) {
4913 $text = preg_replace( '/~{3,5}/', '', $text );
4914 return $text;
4915 }
4916
4917 /**
4918 * Set up some variables which are usually set up in parse()
4919 * so that an external function can call some class members with confidence
4920 *
4921 * @param Title|null $title
4922 * @param ParserOptions $options
4923 * @param int $outputType
4924 * @param bool $clearState
4925 */
4926 public function startExternalParse( Title $title = null, ParserOptions $options,
4927 $outputType, $clearState = true
4928 ) {
4929 $this->startParse( $title, $options, $outputType, $clearState );
4930 }
4931
4932 /**
4933 * @param Title|null $title
4934 * @param ParserOptions $options
4935 * @param int $outputType
4936 * @param bool $clearState
4937 */
4938 private function startParse( Title $title = null, ParserOptions $options,
4939 $outputType, $clearState = true
4940 ) {
4941 $this->setTitle( $title );
4942 $this->mOptions = $options;
4943 $this->setOutputType( $outputType );
4944 if ( $clearState ) {
4945 $this->clearState();
4946 }
4947 }
4948
4949 /**
4950 * Wrapper for preprocess()
4951 *
4952 * @param string $text The text to preprocess
4953 * @param ParserOptions $options Options
4954 * @param Title|null $title Title object or null to use $wgTitle
4955 * @return string
4956 */
4957 public function transformMsg( $text, $options, $title = null ) {
4958 static $executing = false;
4959
4960 # Guard against infinite recursion
4961 if ( $executing ) {
4962 return $text;
4963 }
4964 $executing = true;
4965
4966 wfProfileIn( __METHOD__ );
4967 if ( !$title ) {
4968 global $wgTitle;
4969 $title = $wgTitle;
4970 }
4971
4972 $text = $this->preprocess( $text, $title, $options );
4973
4974 $executing = false;
4975 wfProfileOut( __METHOD__ );
4976 return $text;
4977 }
4978
4979 /**
4980 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4981 * The callback should have the following form:
4982 * function myParserHook( $text, $params, $parser, $frame ) { ... }
4983 *
4984 * Transform and return $text. Use $parser for any required context, e.g. use
4985 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4986 *
4987 * Hooks may return extended information by returning an array, of which the
4988 * first numbered element (index 0) must be the return string, and all other
4989 * entries are extracted into local variables within an internal function
4990 * in the Parser class.
4991 *
4992 * This interface (introduced r61913) appears to be undocumented, but
4993 * 'markerName' is used by some core tag hooks to override which strip
4994 * array their results are placed in. **Use great caution if attempting
4995 * this interface, as it is not documented and injudicious use could smash
4996 * private variables.**
4997 *
4998 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
4999 * @param mixed $callback The callback function (and object) to use for the tag
5000 * @throws MWException
5001 * @return mixed|null The old value of the mTagHooks array associated with the hook
5002 */
5003 public function setHook( $tag, $callback ) {
5004 $tag = strtolower( $tag );
5005 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5006 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
5007 }
5008 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
5009 $this->mTagHooks[$tag] = $callback;
5010 if ( !in_array( $tag, $this->mStripList ) ) {
5011 $this->mStripList[] = $tag;
5012 }
5013
5014 return $oldVal;
5015 }
5016
5017 /**
5018 * As setHook(), but letting the contents be parsed.
5019 *
5020 * Transparent tag hooks are like regular XML-style tag hooks, except they
5021 * operate late in the transformation sequence, on HTML instead of wikitext.
5022 *
5023 * This is probably obsoleted by things dealing with parser frames?
5024 * The only extension currently using it is geoserver.
5025 *
5026 * @since 1.10
5027 * @todo better document or deprecate this
5028 *
5029 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5030 * @param mixed $callback The callback function (and object) to use for the tag
5031 * @throws MWException
5032 * @return mixed|null The old value of the mTagHooks array associated with the hook
5033 */
5034 function setTransparentTagHook( $tag, $callback ) {
5035 $tag = strtolower( $tag );
5036 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5037 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5038 }
5039 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
5040 $this->mTransparentTagHooks[$tag] = $callback;
5041
5042 return $oldVal;
5043 }
5044
5045 /**
5046 * Remove all tag hooks
5047 */
5048 function clearTagHooks() {
5049 $this->mTagHooks = array();
5050 $this->mFunctionTagHooks = array();
5051 $this->mStripList = $this->mDefaultStripList;
5052 }
5053
5054 /**
5055 * Create a function, e.g. {{sum:1|2|3}}
5056 * The callback function should have the form:
5057 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5058 *
5059 * Or with SFH_OBJECT_ARGS:
5060 * function myParserFunction( $parser, $frame, $args ) { ... }
5061 *
5062 * The callback may either return the text result of the function, or an array with the text
5063 * in element 0, and a number of flags in the other elements. The names of the flags are
5064 * specified in the keys. Valid flags are:
5065 * found The text returned is valid, stop processing the template. This
5066 * is on by default.
5067 * nowiki Wiki markup in the return value should be escaped
5068 * isHTML The returned text is HTML, armour it against wikitext transformation
5069 *
5070 * @param string $id The magic word ID
5071 * @param mixed $callback The callback function (and object) to use
5072 * @param int $flags A combination of the following flags:
5073 * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5074 *
5075 * SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text. This
5076 * allows for conditional expansion of the parse tree, allowing you to eliminate dead
5077 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5078 * the arguments, and to control the way they are expanded.
5079 *
5080 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5081 * arguments, for instance:
5082 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5083 *
5084 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5085 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5086 * working if/when this is changed.
5087 *
5088 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5089 * expansion.
5090 *
5091 * Please read the documentation in includes/parser/Preprocessor.php for more information
5092 * about the methods available in PPFrame and PPNode.
5093 *
5094 * @throws MWException
5095 * @return string|callable The old callback function for this name, if any
5096 */
5097 public function setFunctionHook( $id, $callback, $flags = 0 ) {
5098 global $wgContLang;
5099
5100 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
5101 $this->mFunctionHooks[$id] = array( $callback, $flags );
5102
5103 # Add to function cache
5104 $mw = MagicWord::get( $id );
5105 if ( !$mw ) {
5106 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
5107 }
5108
5109 $synonyms = $mw->getSynonyms();
5110 $sensitive = intval( $mw->isCaseSensitive() );
5111
5112 foreach ( $synonyms as $syn ) {
5113 # Case
5114 if ( !$sensitive ) {
5115 $syn = $wgContLang->lc( $syn );
5116 }
5117 # Add leading hash
5118 if ( !( $flags & SFH_NO_HASH ) ) {
5119 $syn = '#' . $syn;
5120 }
5121 # Remove trailing colon
5122 if ( substr( $syn, -1, 1 ) === ':' ) {
5123 $syn = substr( $syn, 0, -1 );
5124 }
5125 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5126 }
5127 return $oldVal;
5128 }
5129
5130 /**
5131 * Get all registered function hook identifiers
5132 *
5133 * @return array
5134 */
5135 function getFunctionHooks() {
5136 return array_keys( $this->mFunctionHooks );
5137 }
5138
5139 /**
5140 * Create a tag function, e.g. "<test>some stuff</test>".
5141 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5142 * Unlike parser functions, their content is not preprocessed.
5143 * @param string $tag
5144 * @param callable $callback
5145 * @param int $flags
5146 * @throws MWException
5147 * @return null
5148 */
5149 function setFunctionTagHook( $tag, $callback, $flags ) {
5150 $tag = strtolower( $tag );
5151 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5152 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5153 }
5154 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
5155 $this->mFunctionTagHooks[$tag] : null;
5156 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
5157
5158 if ( !in_array( $tag, $this->mStripList ) ) {
5159 $this->mStripList[] = $tag;
5160 }
5161
5162 return $old;
5163 }
5164
5165 /**
5166 * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
5167 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5168 * Placeholders created in Skin::makeLinkObj()
5169 *
5170 * @param string $text
5171 * @param int $options
5172 *
5173 * @return array Array of link CSS classes, indexed by PDBK.
5174 */
5175 function replaceLinkHolders( &$text, $options = 0 ) {
5176 return $this->mLinkHolders->replace( $text );
5177 }
5178
5179 /**
5180 * Replace "<!--LINK-->" link placeholders with plain text of links
5181 * (not HTML-formatted).
5182 *
5183 * @param string $text
5184 * @return string
5185 */
5186 function replaceLinkHoldersText( $text ) {
5187 return $this->mLinkHolders->replaceText( $text );
5188 }
5189
5190 /**
5191 * Renders an image gallery from a text with one line per image.
5192 * text labels may be given by using |-style alternative text. E.g.
5193 * Image:one.jpg|The number "1"
5194 * Image:tree.jpg|A tree
5195 * given as text will return the HTML of a gallery with two images,
5196 * labeled 'The number "1"' and
5197 * 'A tree'.
5198 *
5199 * @param string $text
5200 * @param array $params
5201 * @return string HTML
5202 */
5203 function renderImageGallery( $text, $params ) {
5204 wfProfileIn( __METHOD__ );
5205
5206 $mode = false;
5207 if ( isset( $params['mode'] ) ) {
5208 $mode = $params['mode'];
5209 }
5210
5211 try {
5212 $ig = ImageGalleryBase::factory( $mode );
5213 } catch ( MWException $e ) {
5214 // If invalid type set, fallback to default.
5215 $ig = ImageGalleryBase::factory( false );
5216 }
5217
5218 $ig->setContextTitle( $this->mTitle );
5219 $ig->setShowBytes( false );
5220 $ig->setShowFilename( false );
5221 $ig->setParser( $this );
5222 $ig->setHideBadImages();
5223 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
5224
5225 if ( isset( $params['showfilename'] ) ) {
5226 $ig->setShowFilename( true );
5227 } else {
5228 $ig->setShowFilename( false );
5229 }
5230 if ( isset( $params['caption'] ) ) {
5231 $caption = $params['caption'];
5232 $caption = htmlspecialchars( $caption );
5233 $caption = $this->replaceInternalLinks( $caption );
5234 $ig->setCaptionHtml( $caption );
5235 }
5236 if ( isset( $params['perrow'] ) ) {
5237 $ig->setPerRow( $params['perrow'] );
5238 }
5239 if ( isset( $params['widths'] ) ) {
5240 $ig->setWidths( $params['widths'] );
5241 }
5242 if ( isset( $params['heights'] ) ) {
5243 $ig->setHeights( $params['heights'] );
5244 }
5245 $ig->setAdditionalOptions( $params );
5246
5247 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
5248
5249 $lines = StringUtils::explode( "\n", $text );
5250 foreach ( $lines as $line ) {
5251 # match lines like these:
5252 # Image:someimage.jpg|This is some image
5253 $matches = array();
5254 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5255 # Skip empty lines
5256 if ( count( $matches ) == 0 ) {
5257 continue;
5258 }
5259
5260 if ( strpos( $matches[0], '%' ) !== false ) {
5261 $matches[1] = rawurldecode( $matches[1] );
5262 }
5263 $title = Title::newFromText( $matches[1], NS_FILE );
5264 if ( is_null( $title ) ) {
5265 # Bogus title. Ignore these so we don't bomb out later.
5266 continue;
5267 }
5268
5269 # We need to get what handler the file uses, to figure out parameters.
5270 # Note, a hook can overide the file name, and chose an entirely different
5271 # file (which potentially could be of a different type and have different handler).
5272 $options = array();
5273 $descQuery = false;
5274 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5275 array( $this, $title, &$options, &$descQuery ) );
5276 # Don't register it now, as ImageGallery does that later.
5277 $file = $this->fetchFileNoRegister( $title, $options );
5278 $handler = $file ? $file->getHandler() : false;
5279
5280 wfProfileIn( __METHOD__ . '-getMagicWord' );
5281 $paramMap = array(
5282 'img_alt' => 'gallery-internal-alt',
5283 'img_link' => 'gallery-internal-link',
5284 );
5285 if ( $handler ) {
5286 $paramMap = $paramMap + $handler->getParamMap();
5287 // We don't want people to specify per-image widths.
5288 // Additionally the width parameter would need special casing anyhow.
5289 unset( $paramMap['img_width'] );
5290 }
5291
5292 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5293 wfProfileOut( __METHOD__ . '-getMagicWord' );
5294
5295 $label = '';
5296 $alt = '';
5297 $link = '';
5298 $handlerOptions = array();
5299 if ( isset( $matches[3] ) ) {
5300 // look for an |alt= definition while trying not to break existing
5301 // captions with multiple pipes (|) in it, until a more sensible grammar
5302 // is defined for images in galleries
5303
5304 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5305 // splitting on '|' is a bit odd, and different from makeImage.
5306 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5307 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5308
5309 foreach ( $parameterMatches as $parameterMatch ) {
5310 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5311 if ( $magicName ) {
5312 $paramName = $paramMap[$magicName];
5313
5314 switch ( $paramName ) {
5315 case 'gallery-internal-alt':
5316 $alt = $this->stripAltText( $match, false );
5317 break;
5318 case 'gallery-internal-link':
5319 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5320 $chars = self::EXT_LINK_URL_CLASS;
5321 $prots = $this->mUrlProtocols;
5322 //check to see if link matches an absolute url, if not then it must be a wiki link.
5323 if ( preg_match( "/^($prots)$chars+$/u", $linkValue ) ) {
5324 $link = $linkValue;
5325 } else {
5326 $localLinkTitle = Title::newFromText( $linkValue );
5327 if ( $localLinkTitle !== null ) {
5328 $link = $localLinkTitle->getLocalURL();
5329 }
5330 }
5331 break;
5332 default:
5333 // Must be a handler specific parameter.
5334 if ( $handler->validateParam( $paramName, $match ) ) {
5335 $handlerOptions[$paramName] = $match;
5336 } else {
5337 // Guess not. Append it to the caption.
5338 wfDebug( "$parameterMatch failed parameter validation\n" );
5339 $label .= '|' . $parameterMatch;
5340 }
5341 }
5342
5343 } else {
5344 // concatenate all other pipes
5345 $label .= '|' . $parameterMatch;
5346 }
5347 }
5348 // remove the first pipe
5349 $label = substr( $label, 1 );
5350 }
5351
5352 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5353 }
5354 $html = $ig->toHTML();
5355 wfProfileOut( __METHOD__ );
5356 return $html;
5357 }
5358
5359 /**
5360 * @param string $handler
5361 * @return array
5362 */
5363 function getImageParams( $handler ) {
5364 if ( $handler ) {
5365 $handlerClass = get_class( $handler );
5366 } else {
5367 $handlerClass = '';
5368 }
5369 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5370 # Initialise static lists
5371 static $internalParamNames = array(
5372 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
5373 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5374 'bottom', 'text-bottom' ),
5375 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
5376 'upright', 'border', 'link', 'alt', 'class' ),
5377 );
5378 static $internalParamMap;
5379 if ( !$internalParamMap ) {
5380 $internalParamMap = array();
5381 foreach ( $internalParamNames as $type => $names ) {
5382 foreach ( $names as $name ) {
5383 $magicName = str_replace( '-', '_', "img_$name" );
5384 $internalParamMap[$magicName] = array( $type, $name );
5385 }
5386 }
5387 }
5388
5389 # Add handler params
5390 $paramMap = $internalParamMap;
5391 if ( $handler ) {
5392 $handlerParamMap = $handler->getParamMap();
5393 foreach ( $handlerParamMap as $magic => $paramName ) {
5394 $paramMap[$magic] = array( 'handler', $paramName );
5395 }
5396 }
5397 $this->mImageParams[$handlerClass] = $paramMap;
5398 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5399 }
5400 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
5401 }
5402
5403 /**
5404 * Parse image options text and use it to make an image
5405 *
5406 * @param Title $title
5407 * @param string $options
5408 * @param LinkHolderArray|bool $holders
5409 * @return string HTML
5410 */
5411 function makeImage( $title, $options, $holders = false ) {
5412 # Check if the options text is of the form "options|alt text"
5413 # Options are:
5414 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5415 # * left no resizing, just left align. label is used for alt= only
5416 # * right same, but right aligned
5417 # * none same, but not aligned
5418 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5419 # * center center the image
5420 # * frame Keep original image size, no magnify-button.
5421 # * framed Same as "frame"
5422 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5423 # * upright reduce width for upright images, rounded to full __0 px
5424 # * border draw a 1px border around the image
5425 # * alt Text for HTML alt attribute (defaults to empty)
5426 # * class Set a class for img node
5427 # * link Set the target of the image link. Can be external, interwiki, or local
5428 # vertical-align values (no % or length right now):
5429 # * baseline
5430 # * sub
5431 # * super
5432 # * top
5433 # * text-top
5434 # * middle
5435 # * bottom
5436 # * text-bottom
5437
5438 $parts = StringUtils::explode( "|", $options );
5439
5440 # Give extensions a chance to select the file revision for us
5441 $options = array();
5442 $descQuery = false;
5443 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5444 array( $this, $title, &$options, &$descQuery ) );
5445 # Fetch and register the file (file title may be different via hooks)
5446 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5447
5448 # Get parameter map
5449 $handler = $file ? $file->getHandler() : false;
5450
5451 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5452
5453 if ( !$file ) {
5454 $this->addTrackingCategory( 'broken-file-category' );
5455 }
5456
5457 # Process the input parameters
5458 $caption = '';
5459 $params = array( 'frame' => array(), 'handler' => array(),
5460 'horizAlign' => array(), 'vertAlign' => array() );
5461 $seenformat = false;
5462 foreach ( $parts as $part ) {
5463 $part = trim( $part );
5464 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5465 $validated = false;
5466 if ( isset( $paramMap[$magicName] ) ) {
5467 list( $type, $paramName ) = $paramMap[$magicName];
5468
5469 # Special case; width and height come in one variable together
5470 if ( $type === 'handler' && $paramName === 'width' ) {
5471 $parsedWidthParam = $this->parseWidthParam( $value );
5472 if ( isset( $parsedWidthParam['width'] ) ) {
5473 $width = $parsedWidthParam['width'];
5474 if ( $handler->validateParam( 'width', $width ) ) {
5475 $params[$type]['width'] = $width;
5476 $validated = true;
5477 }
5478 }
5479 if ( isset( $parsedWidthParam['height'] ) ) {
5480 $height = $parsedWidthParam['height'];
5481 if ( $handler->validateParam( 'height', $height ) ) {
5482 $params[$type]['height'] = $height;
5483 $validated = true;
5484 }
5485 }
5486 # else no validation -- bug 13436
5487 } else {
5488 if ( $type === 'handler' ) {
5489 # Validate handler parameter
5490 $validated = $handler->validateParam( $paramName, $value );
5491 } else {
5492 # Validate internal parameters
5493 switch ( $paramName ) {
5494 case 'manualthumb':
5495 case 'alt':
5496 case 'class':
5497 # @todo FIXME: Possibly check validity here for
5498 # manualthumb? downstream behavior seems odd with
5499 # missing manual thumbs.
5500 $validated = true;
5501 $value = $this->stripAltText( $value, $holders );
5502 break;
5503 case 'link':
5504 $chars = self::EXT_LINK_URL_CLASS;
5505 $prots = $this->mUrlProtocols;
5506 if ( $value === '' ) {
5507 $paramName = 'no-link';
5508 $value = true;
5509 $validated = true;
5510 } elseif ( preg_match( "/^(?i)$prots/", $value ) ) {
5511 if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
5512 $paramName = 'link-url';
5513 $this->mOutput->addExternalLink( $value );
5514 if ( $this->mOptions->getExternalLinkTarget() ) {
5515 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5516 }
5517 $validated = true;
5518 }
5519 } else {
5520 $linkTitle = Title::newFromText( $value );
5521 if ( $linkTitle ) {
5522 $paramName = 'link-title';
5523 $value = $linkTitle;
5524 $this->mOutput->addLink( $linkTitle );
5525 $validated = true;
5526 }
5527 }
5528 break;
5529 case 'frameless':
5530 case 'framed':
5531 case 'thumbnail':
5532 // use first appearing option, discard others.
5533 $validated = ! $seenformat;
5534 $seenformat = true;
5535 break;
5536 default:
5537 # Most other things appear to be empty or numeric...
5538 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5539 }
5540 }
5541
5542 if ( $validated ) {
5543 $params[$type][$paramName] = $value;
5544 }
5545 }
5546 }
5547 if ( !$validated ) {
5548 $caption = $part;
5549 }
5550 }
5551
5552 # Process alignment parameters
5553 if ( $params['horizAlign'] ) {
5554 $params['frame']['align'] = key( $params['horizAlign'] );
5555 }
5556 if ( $params['vertAlign'] ) {
5557 $params['frame']['valign'] = key( $params['vertAlign'] );
5558 }
5559
5560 $params['frame']['caption'] = $caption;
5561
5562 # Will the image be presented in a frame, with the caption below?
5563 $imageIsFramed = isset( $params['frame']['frame'] )
5564 || isset( $params['frame']['framed'] )
5565 || isset( $params['frame']['thumbnail'] )
5566 || isset( $params['frame']['manualthumb'] );
5567
5568 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5569 # came to also set the caption, ordinary text after the image -- which
5570 # makes no sense, because that just repeats the text multiple times in
5571 # screen readers. It *also* came to set the title attribute.
5572 #
5573 # Now that we have an alt attribute, we should not set the alt text to
5574 # equal the caption: that's worse than useless, it just repeats the
5575 # text. This is the framed/thumbnail case. If there's no caption, we
5576 # use the unnamed parameter for alt text as well, just for the time be-
5577 # ing, if the unnamed param is set and the alt param is not.
5578 #
5579 # For the future, we need to figure out if we want to tweak this more,
5580 # e.g., introducing a title= parameter for the title; ignoring the un-
5581 # named parameter entirely for images without a caption; adding an ex-
5582 # plicit caption= parameter and preserving the old magic unnamed para-
5583 # meter for BC; ...
5584 if ( $imageIsFramed ) { # Framed image
5585 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5586 # No caption or alt text, add the filename as the alt text so
5587 # that screen readers at least get some description of the image
5588 $params['frame']['alt'] = $title->getText();
5589 }
5590 # Do not set $params['frame']['title'] because tooltips don't make sense
5591 # for framed images
5592 } else { # Inline image
5593 if ( !isset( $params['frame']['alt'] ) ) {
5594 # No alt text, use the "caption" for the alt text
5595 if ( $caption !== '' ) {
5596 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5597 } else {
5598 # No caption, fall back to using the filename for the
5599 # alt text
5600 $params['frame']['alt'] = $title->getText();
5601 }
5602 }
5603 # Use the "caption" for the tooltip text
5604 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5605 }
5606
5607 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5608
5609 # Linker does the rest
5610 $time = isset( $options['time'] ) ? $options['time'] : false;
5611 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5612 $time, $descQuery, $this->mOptions->getThumbSize() );
5613
5614 # Give the handler a chance to modify the parser object
5615 if ( $handler ) {
5616 $handler->parserTransformHook( $this, $file );
5617 }
5618
5619 return $ret;
5620 }
5621
5622 /**
5623 * @param string $caption
5624 * @param LinkHolderArray|bool $holders
5625 * @return mixed|string
5626 */
5627 protected function stripAltText( $caption, $holders ) {
5628 # Strip bad stuff out of the title (tooltip). We can't just use
5629 # replaceLinkHoldersText() here, because if this function is called
5630 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5631 if ( $holders ) {
5632 $tooltip = $holders->replaceText( $caption );
5633 } else {
5634 $tooltip = $this->replaceLinkHoldersText( $caption );
5635 }
5636
5637 # make sure there are no placeholders in thumbnail attributes
5638 # that are later expanded to html- so expand them now and
5639 # remove the tags
5640 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5641 $tooltip = Sanitizer::stripAllTags( $tooltip );
5642
5643 return $tooltip;
5644 }
5645
5646 /**
5647 * Set a flag in the output object indicating that the content is dynamic and
5648 * shouldn't be cached.
5649 */
5650 function disableCache() {
5651 wfDebug( "Parser output marked as uncacheable.\n" );
5652 if ( !$this->mOutput ) {
5653 throw new MWException( __METHOD__ .
5654 " can only be called when actually parsing something" );
5655 }
5656 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5657 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5658 }
5659
5660 /**
5661 * Callback from the Sanitizer for expanding items found in HTML attribute
5662 * values, so they can be safely tested and escaped.
5663 *
5664 * @param string $text
5665 * @param bool|PPFrame $frame
5666 * @return string
5667 */
5668 function attributeStripCallback( &$text, $frame = false ) {
5669 $text = $this->replaceVariables( $text, $frame );
5670 $text = $this->mStripState->unstripBoth( $text );
5671 return $text;
5672 }
5673
5674 /**
5675 * Accessor
5676 *
5677 * @return array
5678 */
5679 function getTags() {
5680 return array_merge(
5681 array_keys( $this->mTransparentTagHooks ),
5682 array_keys( $this->mTagHooks ),
5683 array_keys( $this->mFunctionTagHooks )
5684 );
5685 }
5686
5687 /**
5688 * Replace transparent tags in $text with the values given by the callbacks.
5689 *
5690 * Transparent tag hooks are like regular XML-style tag hooks, except they
5691 * operate late in the transformation sequence, on HTML instead of wikitext.
5692 *
5693 * @param string $text
5694 *
5695 * @return string
5696 */
5697 function replaceTransparentTags( $text ) {
5698 $matches = array();
5699 $elements = array_keys( $this->mTransparentTagHooks );
5700 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5701 $replacements = array();
5702
5703 foreach ( $matches as $marker => $data ) {
5704 list( $element, $content, $params, $tag ) = $data;
5705 $tagName = strtolower( $element );
5706 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5707 $output = call_user_func_array(
5708 $this->mTransparentTagHooks[$tagName],
5709 array( $content, $params, $this )
5710 );
5711 } else {
5712 $output = $tag;
5713 }
5714 $replacements[$marker] = $output;
5715 }
5716 return strtr( $text, $replacements );
5717 }
5718
5719 /**
5720 * Break wikitext input into sections, and either pull or replace
5721 * some particular section's text.
5722 *
5723 * External callers should use the getSection and replaceSection methods.
5724 *
5725 * @param string $text Page wikitext
5726 * @param string $section A section identifier string of the form:
5727 * "<flag1> - <flag2> - ... - <section number>"
5728 *
5729 * Currently the only recognised flag is "T", which means the target section number
5730 * was derived during a template inclusion parse, in other words this is a template
5731 * section edit link. If no flags are given, it was an ordinary section edit link.
5732 * This flag is required to avoid a section numbering mismatch when a section is
5733 * enclosed by "<includeonly>" (bug 6563).
5734 *
5735 * The section number 0 pulls the text before the first heading; other numbers will
5736 * pull the given section along with its lower-level subsections. If the section is
5737 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5738 *
5739 * Section 0 is always considered to exist, even if it only contains the empty
5740 * string. If $text is the empty string and section 0 is replaced, $newText is
5741 * returned.
5742 *
5743 * @param string $mode One of "get" or "replace"
5744 * @param string $newText Replacement text for section data.
5745 * @return string For "get", the extracted section text.
5746 * for "replace", the whole page with the section replaced.
5747 */
5748 private function extractSections( $text, $section, $mode, $newText = '' ) {
5749 global $wgTitle; # not generally used but removes an ugly failure mode
5750
5751 $magicScopeVariable = $this->lock();
5752 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5753 $outText = '';
5754 $frame = $this->getPreprocessor()->newFrame();
5755
5756 # Process section extraction flags
5757 $flags = 0;
5758 $sectionParts = explode( '-', $section );
5759 $sectionIndex = array_pop( $sectionParts );
5760 foreach ( $sectionParts as $part ) {
5761 if ( $part === 'T' ) {
5762 $flags |= self::PTD_FOR_INCLUSION;
5763 }
5764 }
5765
5766 # Check for empty input
5767 if ( strval( $text ) === '' ) {
5768 # Only sections 0 and T-0 exist in an empty document
5769 if ( $sectionIndex == 0 ) {
5770 if ( $mode === 'get' ) {
5771 return '';
5772 } else {
5773 return $newText;
5774 }
5775 } else {
5776 if ( $mode === 'get' ) {
5777 return $newText;
5778 } else {
5779 return $text;
5780 }
5781 }
5782 }
5783
5784 # Preprocess the text
5785 $root = $this->preprocessToDom( $text, $flags );
5786
5787 # <h> nodes indicate section breaks
5788 # They can only occur at the top level, so we can find them by iterating the root's children
5789 $node = $root->getFirstChild();
5790
5791 # Find the target section
5792 if ( $sectionIndex == 0 ) {
5793 # Section zero doesn't nest, level=big
5794 $targetLevel = 1000;
5795 } else {
5796 while ( $node ) {
5797 if ( $node->getName() === 'h' ) {
5798 $bits = $node->splitHeading();
5799 if ( $bits['i'] == $sectionIndex ) {
5800 $targetLevel = $bits['level'];
5801 break;
5802 }
5803 }
5804 if ( $mode === 'replace' ) {
5805 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5806 }
5807 $node = $node->getNextSibling();
5808 }
5809 }
5810
5811 if ( !$node ) {
5812 # Not found
5813 if ( $mode === 'get' ) {
5814 return $newText;
5815 } else {
5816 return $text;
5817 }
5818 }
5819
5820 # Find the end of the section, including nested sections
5821 do {
5822 if ( $node->getName() === 'h' ) {
5823 $bits = $node->splitHeading();
5824 $curLevel = $bits['level'];
5825 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5826 break;
5827 }
5828 }
5829 if ( $mode === 'get' ) {
5830 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5831 }
5832 $node = $node->getNextSibling();
5833 } while ( $node );
5834
5835 # Write out the remainder (in replace mode only)
5836 if ( $mode === 'replace' ) {
5837 # Output the replacement text
5838 # Add two newlines on -- trailing whitespace in $newText is conventionally
5839 # stripped by the editor, so we need both newlines to restore the paragraph gap
5840 # Only add trailing whitespace if there is newText
5841 if ( $newText != "" ) {
5842 $outText .= $newText . "\n\n";
5843 }
5844
5845 while ( $node ) {
5846 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5847 $node = $node->getNextSibling();
5848 }
5849 }
5850
5851 if ( is_string( $outText ) ) {
5852 # Re-insert stripped tags
5853 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5854 }
5855
5856 return $outText;
5857 }
5858
5859 /**
5860 * This function returns the text of a section, specified by a number ($section).
5861 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5862 * the first section before any such heading (section 0).
5863 *
5864 * If a section contains subsections, these are also returned.
5865 *
5866 * @param string $text Text to look in
5867 * @param string $section Section identifier
5868 * @param string $deftext Default to return if section is not found
5869 * @return string Text of the requested section
5870 */
5871 public function getSection( $text, $section, $deftext = '' ) {
5872 return $this->extractSections( $text, $section, "get", $deftext );
5873 }
5874
5875 /**
5876 * This function returns $oldtext after the content of the section
5877 * specified by $section has been replaced with $text. If the target
5878 * section does not exist, $oldtext is returned unchanged.
5879 *
5880 * @param string $oldtext Former text of the article
5881 * @param int $section Section identifier
5882 * @param string $text Replacing text
5883 * @return string Modified text
5884 */
5885 public function replaceSection( $oldtext, $section, $text ) {
5886 return $this->extractSections( $oldtext, $section, "replace", $text );
5887 }
5888
5889 /**
5890 * Get the ID of the revision we are parsing
5891 *
5892 * @return int|null
5893 */
5894 function getRevisionId() {
5895 return $this->mRevisionId;
5896 }
5897
5898 /**
5899 * Get the revision object for $this->mRevisionId
5900 *
5901 * @return Revision|null Either a Revision object or null
5902 * @since 1.23 (public since 1.23)
5903 */
5904 public function getRevisionObject() {
5905 if ( !is_null( $this->mRevisionObject ) ) {
5906 return $this->mRevisionObject;
5907 }
5908 if ( is_null( $this->mRevisionId ) ) {
5909 return null;
5910 }
5911
5912 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5913 return $this->mRevisionObject;
5914 }
5915
5916 /**
5917 * Get the timestamp associated with the current revision, adjusted for
5918 * the default server-local timestamp
5919 * @return string
5920 */
5921 function getRevisionTimestamp() {
5922 if ( is_null( $this->mRevisionTimestamp ) ) {
5923 wfProfileIn( __METHOD__ );
5924
5925 global $wgContLang;
5926
5927 $revObject = $this->getRevisionObject();
5928 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5929
5930 # The cryptic '' timezone parameter tells to use the site-default
5931 # timezone offset instead of the user settings.
5932 #
5933 # Since this value will be saved into the parser cache, served
5934 # to other users, and potentially even used inside links and such,
5935 # it needs to be consistent for all visitors.
5936 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5937
5938 wfProfileOut( __METHOD__ );
5939 }
5940 return $this->mRevisionTimestamp;
5941 }
5942
5943 /**
5944 * Get the name of the user that edited the last revision
5945 *
5946 * @return string User name
5947 */
5948 function getRevisionUser() {
5949 if ( is_null( $this->mRevisionUser ) ) {
5950 $revObject = $this->getRevisionObject();
5951
5952 # if this template is subst: the revision id will be blank,
5953 # so just use the current user's name
5954 if ( $revObject ) {
5955 $this->mRevisionUser = $revObject->getUserText();
5956 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5957 $this->mRevisionUser = $this->getUser()->getName();
5958 }
5959 }
5960 return $this->mRevisionUser;
5961 }
5962
5963 /**
5964 * Get the size of the revision
5965 *
5966 * @return int|null Revision size
5967 */
5968 function getRevisionSize() {
5969 if ( is_null( $this->mRevisionSize ) ) {
5970 $revObject = $this->getRevisionObject();
5971
5972 # if this variable is subst: the revision id will be blank,
5973 # so just use the parser input size, because the own substituation
5974 # will change the size.
5975 if ( $revObject ) {
5976 $this->mRevisionSize = $revObject->getSize();
5977 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5978 $this->mRevisionSize = $this->mInputSize;
5979 }
5980 }
5981 return $this->mRevisionSize;
5982 }
5983
5984 /**
5985 * Mutator for $mDefaultSort
5986 *
5987 * @param string $sort New value
5988 */
5989 public function setDefaultSort( $sort ) {
5990 $this->mDefaultSort = $sort;
5991 $this->mOutput->setProperty( 'defaultsort', $sort );
5992 }
5993
5994 /**
5995 * Accessor for $mDefaultSort
5996 * Will use the empty string if none is set.
5997 *
5998 * This value is treated as a prefix, so the
5999 * empty string is equivalent to sorting by
6000 * page name.
6001 *
6002 * @return string
6003 */
6004 public function getDefaultSort() {
6005 if ( $this->mDefaultSort !== false ) {
6006 return $this->mDefaultSort;
6007 } else {
6008 return '';
6009 }
6010 }
6011
6012 /**
6013 * Accessor for $mDefaultSort
6014 * Unlike getDefaultSort(), will return false if none is set
6015 *
6016 * @return string|bool
6017 */
6018 public function getCustomDefaultSort() {
6019 return $this->mDefaultSort;
6020 }
6021
6022 /**
6023 * Try to guess the section anchor name based on a wikitext fragment
6024 * presumably extracted from a heading, for example "Header" from
6025 * "== Header ==".
6026 *
6027 * @param string $text
6028 *
6029 * @return string
6030 */
6031 public function guessSectionNameFromWikiText( $text ) {
6032 # Strip out wikitext links(they break the anchor)
6033 $text = $this->stripSectionName( $text );
6034 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6035 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
6036 }
6037
6038 /**
6039 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6040 * instead. For use in redirects, since IE6 interprets Redirect: headers
6041 * as something other than UTF-8 (apparently?), resulting in breakage.
6042 *
6043 * @param string $text The section name
6044 * @return string An anchor
6045 */
6046 public function guessLegacySectionNameFromWikiText( $text ) {
6047 # Strip out wikitext links(they break the anchor)
6048 $text = $this->stripSectionName( $text );
6049 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6050 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
6051 }
6052
6053 /**
6054 * Strips a text string of wikitext for use in a section anchor
6055 *
6056 * Accepts a text string and then removes all wikitext from the
6057 * string and leaves only the resultant text (i.e. the result of
6058 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6059 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6060 * to create valid section anchors by mimicing the output of the
6061 * parser when headings are parsed.
6062 *
6063 * @param string $text Text string to be stripped of wikitext
6064 * for use in a Section anchor
6065 * @return string Filtered text string
6066 */
6067 public function stripSectionName( $text ) {
6068 # Strip internal link markup
6069 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6070 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6071
6072 # Strip external link markup
6073 # @todo FIXME: Not tolerant to blank link text
6074 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6075 # on how many empty links there are on the page - need to figure that out.
6076 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6077
6078 # Parse wikitext quotes (italics & bold)
6079 $text = $this->doQuotes( $text );
6080
6081 # Strip HTML tags
6082 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6083 return $text;
6084 }
6085
6086 /**
6087 * strip/replaceVariables/unstrip for preprocessor regression testing
6088 *
6089 * @param string $text
6090 * @param Title $title
6091 * @param ParserOptions $options
6092 * @param int $outputType
6093 *
6094 * @return string
6095 */
6096 function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
6097 $magicScopeVariable = $this->lock();
6098 $this->startParse( $title, $options, $outputType, true );
6099
6100 $text = $this->replaceVariables( $text );
6101 $text = $this->mStripState->unstripBoth( $text );
6102 $text = Sanitizer::removeHTMLtags( $text );
6103 return $text;
6104 }
6105
6106 /**
6107 * @param string $text
6108 * @param Title $title
6109 * @param ParserOptions $options
6110 * @return string
6111 */
6112 function testPst( $text, Title $title, ParserOptions $options ) {
6113 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6114 }
6115
6116 /**
6117 * @param string $text
6118 * @param Title $title
6119 * @param ParserOptions $options
6120 * @return string
6121 */
6122 function testPreprocess( $text, Title $title, ParserOptions $options ) {
6123 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6124 }
6125
6126 /**
6127 * Call a callback function on all regions of the given text that are not
6128 * inside strip markers, and replace those regions with the return value
6129 * of the callback. For example, with input:
6130 *
6131 * aaa<MARKER>bbb
6132 *
6133 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6134 * two strings will be replaced with the value returned by the callback in
6135 * each case.
6136 *
6137 * @param string $s
6138 * @param callable $callback
6139 *
6140 * @return string
6141 */
6142 function markerSkipCallback( $s, $callback ) {
6143 $i = 0;
6144 $out = '';
6145 while ( $i < strlen( $s ) ) {
6146 $markerStart = strpos( $s, $this->mUniqPrefix, $i );
6147 if ( $markerStart === false ) {
6148 $out .= call_user_func( $callback, substr( $s, $i ) );
6149 break;
6150 } else {
6151 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6152 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6153 if ( $markerEnd === false ) {
6154 $out .= substr( $s, $markerStart );
6155 break;
6156 } else {
6157 $markerEnd += strlen( self::MARKER_SUFFIX );
6158 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6159 $i = $markerEnd;
6160 }
6161 }
6162 }
6163 return $out;
6164 }
6165
6166 /**
6167 * Remove any strip markers found in the given text.
6168 *
6169 * @param string $text Input string
6170 * @return string
6171 */
6172 function killMarkers( $text ) {
6173 return $this->mStripState->killMarkers( $text );
6174 }
6175
6176 /**
6177 * Save the parser state required to convert the given half-parsed text to
6178 * HTML. "Half-parsed" in this context means the output of
6179 * recursiveTagParse() or internalParse(). This output has strip markers
6180 * from replaceVariables (extensionSubstitution() etc.), and link
6181 * placeholders from replaceLinkHolders().
6182 *
6183 * Returns an array which can be serialized and stored persistently. This
6184 * array can later be loaded into another parser instance with
6185 * unserializeHalfParsedText(). The text can then be safely incorporated into
6186 * the return value of a parser hook.
6187 *
6188 * @param string $text
6189 *
6190 * @return array
6191 */
6192 function serializeHalfParsedText( $text ) {
6193 wfProfileIn( __METHOD__ );
6194 $data = array(
6195 'text' => $text,
6196 'version' => self::HALF_PARSED_VERSION,
6197 'stripState' => $this->mStripState->getSubState( $text ),
6198 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6199 );
6200 wfProfileOut( __METHOD__ );
6201 return $data;
6202 }
6203
6204 /**
6205 * Load the parser state given in the $data array, which is assumed to
6206 * have been generated by serializeHalfParsedText(). The text contents is
6207 * extracted from the array, and its markers are transformed into markers
6208 * appropriate for the current Parser instance. This transformed text is
6209 * returned, and can be safely included in the return value of a parser
6210 * hook.
6211 *
6212 * If the $data array has been stored persistently, the caller should first
6213 * check whether it is still valid, by calling isValidHalfParsedText().
6214 *
6215 * @param array $data Serialized data
6216 * @throws MWException
6217 * @return string
6218 */
6219 function unserializeHalfParsedText( $data ) {
6220 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6221 throw new MWException( __METHOD__ . ': invalid version' );
6222 }
6223
6224 # First, extract the strip state.
6225 $texts = array( $data['text'] );
6226 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6227
6228 # Now renumber links
6229 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6230
6231 # Should be good to go.
6232 return $texts[0];
6233 }
6234
6235 /**
6236 * Returns true if the given array, presumed to be generated by
6237 * serializeHalfParsedText(), is compatible with the current version of the
6238 * parser.
6239 *
6240 * @param array $data
6241 *
6242 * @return bool
6243 */
6244 function isValidHalfParsedText( $data ) {
6245 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6246 }
6247
6248 /**
6249 * Parsed a width param of imagelink like 300px or 200x300px
6250 *
6251 * @param string $value
6252 *
6253 * @return array
6254 * @since 1.20
6255 */
6256 public function parseWidthParam( $value ) {
6257 $parsedWidthParam = array();
6258 if ( $value === '' ) {
6259 return $parsedWidthParam;
6260 }
6261 $m = array();
6262 # (bug 13500) In both cases (width/height and width only),
6263 # permit trailing "px" for backward compatibility.
6264 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6265 $width = intval( $m[1] );
6266 $height = intval( $m[2] );
6267 $parsedWidthParam['width'] = $width;
6268 $parsedWidthParam['height'] = $height;
6269 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6270 $width = intval( $value );
6271 $parsedWidthParam['width'] = $width;
6272 }
6273 return $parsedWidthParam;
6274 }
6275
6276 /**
6277 * Lock the current instance of the parser.
6278 *
6279 * This is meant to stop someone from calling the parser
6280 * recursively and messing up all the strip state.
6281 *
6282 * @throws MWException If parser is in a parse
6283 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6284 */
6285 protected function lock() {
6286 if ( $this->mInParse ) {
6287 throw new MWException( "Parser state cleared while parsing. "
6288 . "Did you call Parser::parse recursively?" );
6289 }
6290 $this->mInParse = true;
6291
6292 $that = $this;
6293 $recursiveCheck = new ScopedCallback( function() use ( $that ) {
6294 $that->mInParse = false;
6295 } );
6296
6297 return $recursiveCheck;
6298 }
6299
6300 /**
6301 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6302 *
6303 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6304 * or if there is more than one <p/> tag in the input HTML.
6305 *
6306 * @param string $html
6307 * @return string
6308 * @since 1.24
6309 */
6310 public static function stripOuterParagraph( $html ) {
6311 $m = array();
6312 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6313 if ( strpos( $m[1], '</p>' ) === false ) {
6314 $html = $m[1];
6315 }
6316 }
6317
6318 return $html;
6319 }
6320 }
Wiklou, le Wiki du Wiklou