dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Fix some RecentChange phpdoc return types"
[lhc/web/wiklou.git] / includes / parser / Parser.php
1 <?php
2 /**
3 * PHP parser that converts wiki markup to HTML.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Parser
22 */
23
24 /**
25 * @defgroup Parser Parser
26 */
27
28 /**
29 * PHP Parser - Processes wiki markup (which uses a more user-friendly
30 * syntax, such as "[[link]]" for making links), and provides a one-way
31 * transformation of that wiki markup it into (X)HTML output / markup
32 * (which in turn the browser understands, and can display).
33 *
34 * There are seven main entry points into the Parser class:
35 *
36 * - Parser::parse()
37 * produces HTML output
38 * - Parser::preSaveTransform()
39 * produces altered wiki markup
40 * - Parser::preprocess()
41 * removes HTML comments and expands templates
42 * - Parser::cleanSig() and Parser::cleanSigInSig()
43 * cleans a signature before saving it to preferences
44 * - Parser::getSection()
45 * return the content of a section from an article for section editing
46 * - Parser::replaceSection()
47 * replaces a section by number inside an article
48 * - Parser::getPreloadText()
49 * removes <noinclude> sections and <includeonly> tags
50 *
51 * Globals used:
52 * object: $wgContLang
53 *
54 * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
55 *
56 * @par Settings:
57 * $wgNamespacesWithSubpages
58 *
59 * @par Settings only within ParserOptions:
60 * $wgAllowExternalImages
61 * $wgAllowSpecialInclusion
62 * $wgInterwikiMagic
63 * $wgMaxArticleSize
64 *
65 * @ingroup Parser
66 */
67 class Parser {
68 /**
69 * Update this version number when the ParserOutput format
70 * changes in an incompatible way, so the parser cache
71 * can automatically discard old data.
72 */
73 const VERSION = '1.6.4';
74
75 /**
76 * Update this version number when the output of serialiseHalfParsedText()
77 * changes in an incompatible way
78 */
79 const HALF_PARSED_VERSION = 2;
80
81 # Flags for Parser::setFunctionHook
82 const SFH_NO_HASH = 1;
83 const SFH_OBJECT_ARGS = 2;
84
85 # Constants needed for external link processing
86 # Everything except bracket, space, or control characters
87 # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
88 # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
89 const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
90 const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
91 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
92
93 # Regular expression for a non-newline space
94 const SPACE_NOT_NL = '(?:\t|&nbsp;|&\#0*160;|&\#[Xx]0*[Aa]0;|\p{Zs})';
95
96 # State constants for the definition list colon extraction
97 const COLON_STATE_TEXT = 0;
98 const COLON_STATE_TAG = 1;
99 const COLON_STATE_TAGSTART = 2;
100 const COLON_STATE_CLOSETAG = 3;
101 const COLON_STATE_TAGSLASH = 4;
102 const COLON_STATE_COMMENT = 5;
103 const COLON_STATE_COMMENTDASH = 6;
104 const COLON_STATE_COMMENTDASHDASH = 7;
105
106 # Flags for preprocessToDom
107 const PTD_FOR_INCLUSION = 1;
108
109 # Allowed values for $this->mOutputType
110 # Parameter to startExternalParse().
111 const OT_HTML = 1; # like parse()
112 const OT_WIKI = 2; # like preSaveTransform()
113 const OT_PREPROCESS = 3; # like preprocess()
114 const OT_MSG = 3;
115 const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
116
117 /**
118 * @var string Prefix and suffix for temporary replacement strings
119 * for the multipass parser.
120 *
121 * \x7f should never appear in input as it's disallowed in XML.
122 * Using it at the front also gives us a little extra robustness
123 * since it shouldn't match when butted up against identifier-like
124 * string constructs.
125 *
126 * Must not consist of all title characters, or else it will change
127 * the behavior of <nowiki> in a link.
128 */
129 const MARKER_SUFFIX = "-QINU\x7f";
130 const MARKER_PREFIX = "\x7fUNIQ-";
131
132 # Markers used for wrapping the table of contents
133 const TOC_START = '<mw:toc>';
134 const TOC_END = '</mw:toc>';
135
136 # Persistent:
137 public $mTagHooks = array();
138 public $mTransparentTagHooks = array();
139 public $mFunctionHooks = array();
140 public $mFunctionSynonyms = array( 0 => array(), 1 => array() );
141 public $mFunctionTagHooks = array();
142 public $mStripList = array();
143 public $mDefaultStripList = array();
144 public $mVarCache = array();
145 public $mImageParams = array();
146 public $mImageParamsMagicArray = array();
147 public $mMarkerIndex = 0;
148 public $mFirstCall = true;
149
150 # Initialised by initialiseVariables()
151
152 /**
153 * @var MagicWordArray
154 */
155 public $mVariables;
156
157 /**
158 * @var MagicWordArray
159 */
160 public $mSubstWords;
161 # Initialised in constructor
162 public $mConf, $mExtLinkBracketedRegex, $mUrlProtocols;
163
164 # Initialized in getPreprocessor()
165 /** @var Preprocessor */
166 public $mPreprocessor;
167
168 # Cleared with clearState():
169 /**
170 * @var ParserOutput
171 */
172 public $mOutput;
173 public $mAutonumber, $mDTopen;
174
175 /**
176 * @var StripState
177 */
178 public $mStripState;
179
180 public $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
181 /**
182 * @var LinkHolderArray
183 */
184 public $mLinkHolders;
185
186 public $mLinkID;
187 public $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
188 public $mDefaultSort;
189 public $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
190 public $mExpensiveFunctionCount; # number of expensive parser function calls
191 public $mShowToc, $mForceTocPosition;
192
193 /**
194 * @var User
195 */
196 public $mUser; # User object; only used when doing pre-save transform
197
198 # Temporary
199 # These are variables reset at least once per parse regardless of $clearState
200
201 /**
202 * @var ParserOptions
203 */
204 public $mOptions;
205
206 /**
207 * @var Title
208 */
209 public $mTitle; # Title context, used for self-link rendering and similar things
210 public $mOutputType; # Output type, one of the OT_xxx constants
211 public $ot; # Shortcut alias, see setOutputType()
212 public $mRevisionObject; # The revision object of the specified revision ID
213 public $mRevisionId; # ID to display in {{REVISIONID}} tags
214 public $mRevisionTimestamp; # The timestamp of the specified revision ID
215 public $mRevisionUser; # User to display in {{REVISIONUSER}} tag
216 public $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
217 public $mRevIdForTs; # The revision ID which was used to fetch the timestamp
218 public $mInputSize = false; # For {{PAGESIZE}} on current page.
219
220 /**
221 * @var string Deprecated accessor for the strip marker prefix.
222 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
223 **/
224 public $mUniqPrefix = Parser::MARKER_PREFIX;
225
226 /**
227 * @var array Array with the language name of each language link (i.e. the
228 * interwiki prefix) in the key, value arbitrary. Used to avoid sending
229 * duplicate language links to the ParserOutput.
230 */
231 public $mLangLinkLanguages;
232
233 /**
234 * @var MapCacheLRU|null
235 * @since 1.24
236 *
237 * A cache of the current revisions of titles. Keys are $title->getPrefixedDbKey()
238 */
239 public $currentRevisionCache;
240
241 /**
242 * @var bool Recursive call protection.
243 * This variable should be treated as if it were private.
244 */
245 public $mInParse = false;
246
247 /** @var SectionProfiler */
248 protected $mProfiler;
249
250 /**
251 * @param array $conf
252 */
253 public function __construct( $conf = array() ) {
254 $this->mConf = $conf;
255 $this->mUrlProtocols = wfUrlProtocols();
256 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
257 self::EXT_LINK_URL_CLASS . '+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
258 if ( isset( $conf['preprocessorClass'] ) ) {
259 $this->mPreprocessorClass = $conf['preprocessorClass'];
260 } elseif ( defined( 'HPHP_VERSION' ) ) {
261 # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
262 $this->mPreprocessorClass = 'Preprocessor_Hash';
263 } elseif ( extension_loaded( 'domxml' ) ) {
264 # PECL extension that conflicts with the core DOM extension (bug 13770)
265 wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
266 $this->mPreprocessorClass = 'Preprocessor_Hash';
267 } elseif ( extension_loaded( 'dom' ) ) {
268 $this->mPreprocessorClass = 'Preprocessor_DOM';
269 } else {
270 $this->mPreprocessorClass = 'Preprocessor_Hash';
271 }
272 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
273 }
274
275 /**
276 * Reduce memory usage to reduce the impact of circular references
277 */
278 public function __destruct() {
279 if ( isset( $this->mLinkHolders ) ) {
280 unset( $this->mLinkHolders );
281 }
282 foreach ( $this as $name => $value ) {
283 unset( $this->$name );
284 }
285 }
286
287 /**
288 * Allow extensions to clean up when the parser is cloned
289 */
290 public function __clone() {
291 $this->mInParse = false;
292
293 // Bug 56226: When you create a reference "to" an object field, that
294 // makes the object field itself be a reference too (until the other
295 // reference goes out of scope). When cloning, any field that's a
296 // reference is copied as a reference in the new object. Both of these
297 // are defined PHP5 behaviors, as inconvenient as it is for us when old
298 // hooks from PHP4 days are passing fields by reference.
299 foreach ( array( 'mStripState', 'mVarCache' ) as $k ) {
300 // Make a non-reference copy of the field, then rebind the field to
301 // reference the new copy.
302 $tmp = $this->$k;
303 $this->$k =& $tmp;
304 unset( $tmp );
305 }
306
307 Hooks::run( 'ParserCloned', array( $this ) );
308 }
309
310 /**
311 * Do various kinds of initialisation on the first call of the parser
312 */
313 public function firstCallInit() {
314 if ( !$this->mFirstCall ) {
315 return;
316 }
317 $this->mFirstCall = false;
318
319 CoreParserFunctions::register( $this );
320 CoreTagHooks::register( $this );
321 $this->initialiseVariables();
322
323 Hooks::run( 'ParserFirstCallInit', array( &$this ) );
324 }
325
326 /**
327 * Clear Parser state
328 *
329 * @private
330 */
331 public function clearState() {
332 if ( $this->mFirstCall ) {
333 $this->firstCallInit();
334 }
335 $this->mOutput = new ParserOutput;
336 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
337 $this->mAutonumber = 0;
338 $this->mLastSection = '';
339 $this->mDTopen = false;
340 $this->mIncludeCount = array();
341 $this->mArgStack = false;
342 $this->mInPre = false;
343 $this->mLinkHolders = new LinkHolderArray( $this );
344 $this->mLinkID = 0;
345 $this->mRevisionObject = $this->mRevisionTimestamp =
346 $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
347 $this->mVarCache = array();
348 $this->mUser = null;
349 $this->mLangLinkLanguages = array();
350 $this->currentRevisionCache = null;
351
352 $this->mStripState = new StripState;
353
354 # Clear these on every parse, bug 4549
355 $this->mTplRedirCache = $this->mTplDomCache = array();
356
357 $this->mShowToc = true;
358 $this->mForceTocPosition = false;
359 $this->mIncludeSizes = array(
360 'post-expand' => 0,
361 'arg' => 0,
362 );
363 $this->mPPNodeCount = 0;
364 $this->mGeneratedPPNodeCount = 0;
365 $this->mHighestExpansionDepth = 0;
366 $this->mDefaultSort = false;
367 $this->mHeadings = array();
368 $this->mDoubleUnderscores = array();
369 $this->mExpensiveFunctionCount = 0;
370
371 # Fix cloning
372 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
373 $this->mPreprocessor = null;
374 }
375
376 $this->mProfiler = new SectionProfiler();
377
378 Hooks::run( 'ParserClearState', array( &$this ) );
379 }
380
381 /**
382 * Convert wikitext to HTML
383 * Do not call this function recursively.
384 *
385 * @param string $text Text we want to parse
386 * @param Title $title
387 * @param ParserOptions $options
388 * @param bool $linestart
389 * @param bool $clearState
390 * @param int $revid Number to pass in {{REVISIONID}}
391 * @return ParserOutput A ParserOutput
392 */
393 public function parse( $text, Title $title, ParserOptions $options,
394 $linestart = true, $clearState = true, $revid = null
395 ) {
396 /**
397 * First pass--just handle <nowiki> sections, pass the rest off
398 * to internalParse() which does all the real work.
399 */
400
401 global $wgShowHostnames;
402
403 if ( $clearState ) {
404 // We use U+007F DELETE to construct strip markers, so we have to make
405 // sure that this character does not occur in the input text.
406 $text = strtr( $text, "\x7f", "?" );
407 $magicScopeVariable = $this->lock();
408 }
409
410 $this->startParse( $title, $options, self::OT_HTML, $clearState );
411
412 $this->currentRevisionCache = null;
413 $this->mInputSize = strlen( $text );
414 if ( $this->mOptions->getEnableLimitReport() ) {
415 $this->mOutput->resetParseStartTime();
416 }
417
418 $oldRevisionId = $this->mRevisionId;
419 $oldRevisionObject = $this->mRevisionObject;
420 $oldRevisionTimestamp = $this->mRevisionTimestamp;
421 $oldRevisionUser = $this->mRevisionUser;
422 $oldRevisionSize = $this->mRevisionSize;
423 if ( $revid !== null ) {
424 $this->mRevisionId = $revid;
425 $this->mRevisionObject = null;
426 $this->mRevisionTimestamp = null;
427 $this->mRevisionUser = null;
428 $this->mRevisionSize = null;
429 }
430
431 Hooks::run( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
432 # No more strip!
433 Hooks::run( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
434 $text = $this->internalParse( $text );
435 Hooks::run( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
436
437 $text = $this->internalParseHalfParsed( $text, true, $linestart );
438
439 /**
440 * A converted title will be provided in the output object if title and
441 * content conversion are enabled, the article text does not contain
442 * a conversion-suppressing double-underscore tag, and no
443 * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
444 * automatic link conversion.
445 */
446 if ( !( $options->getDisableTitleConversion()
447 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
448 || isset( $this->mDoubleUnderscores['notitleconvert'] )
449 || $this->mOutput->getDisplayTitle() !== false )
450 ) {
451 $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
452 if ( $convruletitle ) {
453 $this->mOutput->setTitleText( $convruletitle );
454 } else {
455 $titleText = $this->getConverterLanguage()->convertTitle( $title );
456 $this->mOutput->setTitleText( $titleText );
457 }
458 }
459
460 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
461 $this->limitationWarn( 'expensive-parserfunction',
462 $this->mExpensiveFunctionCount,
463 $this->mOptions->getExpensiveParserFunctionLimit()
464 );
465 }
466
467 # Information on include size limits, for the benefit of users who try to skirt them
468 if ( $this->mOptions->getEnableLimitReport() ) {
469 $max = $this->mOptions->getMaxIncludeSize();
470
471 $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
472 if ( $cpuTime !== null ) {
473 $this->mOutput->setLimitReportData( 'limitreport-cputime',
474 sprintf( "%.3f", $cpuTime )
475 );
476 }
477
478 $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
479 $this->mOutput->setLimitReportData( 'limitreport-walltime',
480 sprintf( "%.3f", $wallTime )
481 );
482
483 $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
484 array( $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() )
485 );
486 $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
487 array( $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() )
488 );
489 $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
490 array( $this->mIncludeSizes['post-expand'], $max )
491 );
492 $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
493 array( $this->mIncludeSizes['arg'], $max )
494 );
495 $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
496 array( $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() )
497 );
498 $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
499 array( $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() )
500 );
501 Hooks::run( 'ParserLimitReportPrepare', array( $this, $this->mOutput ) );
502
503 $limitReport = "NewPP limit report\n";
504 if ( $wgShowHostnames ) {
505 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
506 }
507 foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
508 if ( Hooks::run( 'ParserLimitReportFormat',
509 array( $key, &$value, &$limitReport, false, false )
510 ) ) {
511 $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
512 $valueMsg = wfMessage( array( "$key-value-text", "$key-value" ) )
513 ->inLanguage( 'en' )->useDatabase( false );
514 if ( !$valueMsg->exists() ) {
515 $valueMsg = new RawMessage( '$1' );
516 }
517 if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
518 $valueMsg->params( $value );
519 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
520 }
521 }
522 }
523 // Since we're not really outputting HTML, decode the entities and
524 // then re-encode the things that need hiding inside HTML comments.
525 $limitReport = htmlspecialchars_decode( $limitReport );
526 Hooks::run( 'ParserLimitReport', array( $this, &$limitReport ) );
527
528 // Sanitize for comment. Note '‐' in the replacement is U+2010,
529 // which looks much like the problematic '-'.
530 $limitReport = str_replace( array( '-', '&' ), array( '‐', '&amp;' ), $limitReport );
531 $text .= "\n<!-- \n$limitReport-->\n";
532
533 // Add on template profiling data
534 $dataByFunc = $this->mProfiler->getFunctionStats();
535 uasort( $dataByFunc, function ( $a, $b ) {
536 return $a['real'] < $b['real']; // descending order
537 } );
538 $profileReport = "Transclusion expansion time report (%,ms,calls,template)\n";
539 foreach ( array_slice( $dataByFunc, 0, 10 ) as $item ) {
540 $profileReport .= sprintf( "%6.2f%% %8.3f %6d - %s\n",
541 $item['%real'], $item['real'], $item['calls'],
542 htmlspecialchars( $item['name'] ) );
543 }
544 $text .= "\n<!-- \n$profileReport-->\n";
545
546 if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
547 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
548 $this->mTitle->getPrefixedDBkey() );
549 }
550 }
551 $this->mOutput->setText( $text );
552
553 $this->mRevisionId = $oldRevisionId;
554 $this->mRevisionObject = $oldRevisionObject;
555 $this->mRevisionTimestamp = $oldRevisionTimestamp;
556 $this->mRevisionUser = $oldRevisionUser;
557 $this->mRevisionSize = $oldRevisionSize;
558 $this->mInputSize = false;
559 $this->currentRevisionCache = null;
560
561 return $this->mOutput;
562 }
563
564 /**
565 * Half-parse wikitext to half-parsed HTML. This recursive parser entry point
566 * can be called from an extension tag hook.
567 *
568 * The output of this function IS NOT SAFE PARSED HTML; it is "half-parsed"
569 * instead, which means that lists and links have not been fully parsed yet,
570 * and strip markers are still present.
571 *
572 * Use recursiveTagParseFully() to fully parse wikitext to output-safe HTML.
573 *
574 * Use this function if you're a parser tag hook and you want to parse
575 * wikitext before or after applying additional transformations, and you
576 * intend to *return the result as hook output*, which will cause it to go
577 * through the rest of parsing process automatically.
578 *
579 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
580 * $text are not expanded
581 *
582 * @param string $text Text extension wants to have parsed
583 * @param bool|PPFrame $frame The frame to use for expanding any template variables
584 * @return string UNSAFE half-parsed HTML
585 */
586 public function recursiveTagParse( $text, $frame = false ) {
587 Hooks::run( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
588 Hooks::run( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
589 $text = $this->internalParse( $text, false, $frame );
590 return $text;
591 }
592
593 /**
594 * Fully parse wikitext to fully parsed HTML. This recursive parser entry
595 * point can be called from an extension tag hook.
596 *
597 * The output of this function is fully-parsed HTML that is safe for output.
598 * If you're a parser tag hook, you might want to use recursiveTagParse()
599 * instead.
600 *
601 * If $frame is not provided, then template variables (e.g., {{{1}}}) within
602 * $text are not expanded
603 *
604 * @since 1.25
605 *
606 * @param string $text Text extension wants to have parsed
607 * @param bool|PPFrame $frame The frame to use for expanding any template variables
608 * @return string Fully parsed HTML
609 */
610 public function recursiveTagParseFully( $text, $frame = false ) {
611 $text = $this->recursiveTagParse( $text, $frame );
612 $text = $this->internalParseHalfParsed( $text, false );
613 return $text;
614 }
615
616 /**
617 * Expand templates and variables in the text, producing valid, static wikitext.
618 * Also removes comments.
619 * Do not call this function recursively.
620 * @param string $text
621 * @param Title $title
622 * @param ParserOptions $options
623 * @param int|null $revid
624 * @param bool|PPFrame $frame
625 * @return mixed|string
626 */
627 public function preprocess( $text, Title $title = null,
628 ParserOptions $options, $revid = null, $frame = false
629 ) {
630 $magicScopeVariable = $this->lock();
631 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
632 if ( $revid !== null ) {
633 $this->mRevisionId = $revid;
634 }
635 Hooks::run( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
636 Hooks::run( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
637 $text = $this->replaceVariables( $text, $frame );
638 $text = $this->mStripState->unstripBoth( $text );
639 return $text;
640 }
641
642 /**
643 * Recursive parser entry point that can be called from an extension tag
644 * hook.
645 *
646 * @param string $text Text to be expanded
647 * @param bool|PPFrame $frame The frame to use for expanding any template variables
648 * @return string
649 * @since 1.19
650 */
651 public function recursivePreprocess( $text, $frame = false ) {
652 $text = $this->replaceVariables( $text, $frame );
653 $text = $this->mStripState->unstripBoth( $text );
654 return $text;
655 }
656
657 /**
658 * Process the wikitext for the "?preload=" feature. (bug 5210)
659 *
660 * "<noinclude>", "<includeonly>" etc. are parsed as for template
661 * transclusion, comments, templates, arguments, tags hooks and parser
662 * functions are untouched.
663 *
664 * @param string $text
665 * @param Title $title
666 * @param ParserOptions $options
667 * @param array $params
668 * @return string
669 */
670 public function getPreloadText( $text, Title $title, ParserOptions $options, $params = array() ) {
671 $msg = new RawMessage( $text );
672 $text = $msg->params( $params )->plain();
673
674 # Parser (re)initialisation
675 $magicScopeVariable = $this->lock();
676 $this->startParse( $title, $options, self::OT_PLAIN, true );
677
678 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
679 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
680 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
681 $text = $this->mStripState->unstripBoth( $text );
682 return $text;
683 }
684
685 /**
686 * Get a random string
687 *
688 * @return string
689 * @deprecated since 1.26; use wfRandomString() instead.
690 */
691 public static function getRandomString() {
692 wfDeprecated( __METHOD__, '1.26' );
693 return wfRandomString( 16 );
694 }
695
696 /**
697 * Set the current user.
698 * Should only be used when doing pre-save transform.
699 *
700 * @param User|null $user User object or null (to reset)
701 */
702 public function setUser( $user ) {
703 $this->mUser = $user;
704 }
705
706 /**
707 * Accessor for mUniqPrefix.
708 *
709 * @return string
710 * @deprecated since 1.26; use Parser::MARKER_PREFIX instead.
711 */
712 public function uniqPrefix() {
713 wfDeprecated( __METHOD__, '1.26' );
714 return self::MARKER_PREFIX;
715 }
716
717 /**
718 * Set the context title
719 *
720 * @param Title $t
721 */
722 public function setTitle( $t ) {
723 if ( !$t ) {
724 $t = Title::newFromText( 'NO TITLE' );
725 }
726
727 if ( $t->hasFragment() ) {
728 # Strip the fragment to avoid various odd effects
729 $this->mTitle = clone $t;
730 $this->mTitle->setFragment( '' );
731 } else {
732 $this->mTitle = $t;
733 }
734 }
735
736 /**
737 * Accessor for the Title object
738 *
739 * @return Title
740 */
741 public function getTitle() {
742 return $this->mTitle;
743 }
744
745 /**
746 * Accessor/mutator for the Title object
747 *
748 * @param Title $x Title object or null to just get the current one
749 * @return Title
750 */
751 public function Title( $x = null ) {
752 return wfSetVar( $this->mTitle, $x );
753 }
754
755 /**
756 * Set the output type
757 *
758 * @param int $ot New value
759 */
760 public function setOutputType( $ot ) {
761 $this->mOutputType = $ot;
762 # Shortcut alias
763 $this->ot = array(
764 'html' => $ot == self::OT_HTML,
765 'wiki' => $ot == self::OT_WIKI,
766 'pre' => $ot == self::OT_PREPROCESS,
767 'plain' => $ot == self::OT_PLAIN,
768 );
769 }
770
771 /**
772 * Accessor/mutator for the output type
773 *
774 * @param int|null $x New value or null to just get the current one
775 * @return int
776 */
777 public function OutputType( $x = null ) {
778 return wfSetVar( $this->mOutputType, $x );
779 }
780
781 /**
782 * Get the ParserOutput object
783 *
784 * @return ParserOutput
785 */
786 public function getOutput() {
787 return $this->mOutput;
788 }
789
790 /**
791 * Get the ParserOptions object
792 *
793 * @return ParserOptions
794 */
795 public function getOptions() {
796 return $this->mOptions;
797 }
798
799 /**
800 * Accessor/mutator for the ParserOptions object
801 *
802 * @param ParserOptions $x New value or null to just get the current one
803 * @return ParserOptions Current ParserOptions object
804 */
805 public function Options( $x = null ) {
806 return wfSetVar( $this->mOptions, $x );
807 }
808
809 /**
810 * @return int
811 */
812 public function nextLinkID() {
813 return $this->mLinkID++;
814 }
815
816 /**
817 * @param int $id
818 */
819 public function setLinkID( $id ) {
820 $this->mLinkID = $id;
821 }
822
823 /**
824 * Get a language object for use in parser functions such as {{FORMATNUM:}}
825 * @return Language
826 */
827 public function getFunctionLang() {
828 return $this->getTargetLanguage();
829 }
830
831 /**
832 * Get the target language for the content being parsed. This is usually the
833 * language that the content is in.
834 *
835 * @since 1.19
836 *
837 * @throws MWException
838 * @return Language
839 */
840 public function getTargetLanguage() {
841 $target = $this->mOptions->getTargetLanguage();
842
843 if ( $target !== null ) {
844 return $target;
845 } elseif ( $this->mOptions->getInterfaceMessage() ) {
846 return $this->mOptions->getUserLangObj();
847 } elseif ( is_null( $this->mTitle ) ) {
848 throw new MWException( __METHOD__ . ': $this->mTitle is null' );
849 }
850
851 return $this->mTitle->getPageLanguage();
852 }
853
854 /**
855 * Get the language object for language conversion
856 * @return Language|null
857 */
858 public function getConverterLanguage() {
859 return $this->getTargetLanguage();
860 }
861
862 /**
863 * Get a User object either from $this->mUser, if set, or from the
864 * ParserOptions object otherwise
865 *
866 * @return User
867 */
868 public function getUser() {
869 if ( !is_null( $this->mUser ) ) {
870 return $this->mUser;
871 }
872 return $this->mOptions->getUser();
873 }
874
875 /**
876 * Get a preprocessor object
877 *
878 * @return Preprocessor
879 */
880 public function getPreprocessor() {
881 if ( !isset( $this->mPreprocessor ) ) {
882 $class = $this->mPreprocessorClass;
883 $this->mPreprocessor = new $class( $this );
884 }
885 return $this->mPreprocessor;
886 }
887
888 /**
889 * Replaces all occurrences of HTML-style comments and the given tags
890 * in the text with a random marker and returns the next text. The output
891 * parameter $matches will be an associative array filled with data in
892 * the form:
893 *
894 * @code
895 * 'UNIQ-xxxxx' => array(
896 * 'element',
897 * 'tag content',
898 * array( 'param' => 'x' ),
899 * '<element param="x">tag content</element>' ) )
900 * @endcode
901 *
902 * @param array $elements List of element names. Comments are always extracted.
903 * @param string $text Source text string.
904 * @param array $matches Out parameter, Array: extracted tags
905 * @param string|null $uniq_prefix
906 * @return string Stripped text
907 * @since 1.26 The uniq_prefix argument is deprecated.
908 */
909 public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = null ) {
910 if ( $uniq_prefix !== null ) {
911 wfDeprecated( __METHOD__ . ' called with $prefix argument', '1.26' );
912 }
913 static $n = 1;
914 $stripped = '';
915 $matches = array();
916
917 $taglist = implode( '|', $elements );
918 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
919
920 while ( $text != '' ) {
921 $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
922 $stripped .= $p[0];
923 if ( count( $p ) < 5 ) {
924 break;
925 }
926 if ( count( $p ) > 5 ) {
927 # comment
928 $element = $p[4];
929 $attributes = '';
930 $close = '';
931 $inside = $p[5];
932 } else {
933 # tag
934 $element = $p[1];
935 $attributes = $p[2];
936 $close = $p[3];
937 $inside = $p[4];
938 }
939
940 $marker = self::MARKER_PREFIX . "-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
941 $stripped .= $marker;
942
943 if ( $close === '/>' ) {
944 # Empty element tag, <tag />
945 $content = null;
946 $text = $inside;
947 $tail = null;
948 } else {
949 if ( $element === '!--' ) {
950 $end = '/(-->)/';
951 } else {
952 $end = "/(<\\/$element\\s*>)/i";
953 }
954 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
955 $content = $q[0];
956 if ( count( $q ) < 3 ) {
957 # No end tag -- let it run out to the end of the text.
958 $tail = '';
959 $text = '';
960 } else {
961 $tail = $q[1];
962 $text = $q[2];
963 }
964 }
965
966 $matches[$marker] = array( $element,
967 $content,
968 Sanitizer::decodeTagAttributes( $attributes ),
969 "<$element$attributes$close$content$tail" );
970 }
971 return $stripped;
972 }
973
974 /**
975 * Get a list of strippable XML-like elements
976 *
977 * @return array
978 */
979 public function getStripList() {
980 return $this->mStripList;
981 }
982
983 /**
984 * Add an item to the strip state
985 * Returns the unique tag which must be inserted into the stripped text
986 * The tag will be replaced with the original text in unstrip()
987 *
988 * @param string $text
989 *
990 * @return string
991 */
992 public function insertStripItem( $text ) {
993 $marker = self::MARKER_PREFIX . "-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
994 $this->mMarkerIndex++;
995 $this->mStripState->addGeneral( $marker, $text );
996 return $marker;
997 }
998
999 /**
1000 * parse the wiki syntax used to render tables
1001 *
1002 * @private
1003 * @param string $text
1004 * @return string
1005 */
1006 public function doTableStuff( $text ) {
1007
1008 $lines = StringUtils::explode( "\n", $text );
1009 $out = '';
1010 $td_history = array(); # Is currently a td tag open?
1011 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
1012 $tr_history = array(); # Is currently a tr tag open?
1013 $tr_attributes = array(); # history of tr attributes
1014 $has_opened_tr = array(); # Did this table open a <tr> element?
1015 $indent_level = 0; # indent level of the table
1016
1017 foreach ( $lines as $outLine ) {
1018 $line = trim( $outLine );
1019
1020 if ( $line === '' ) { # empty line, go to next line
1021 $out .= $outLine . "\n";
1022 continue;
1023 }
1024
1025 $first_character = $line[0];
1026 $matches = array();
1027
1028 if ( preg_match( '/^(:*)\{\|(.*)$/', $line, $matches ) ) {
1029 # First check if we are starting a new table
1030 $indent_level = strlen( $matches[1] );
1031
1032 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1033 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1034
1035 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1036 array_push( $td_history, false );
1037 array_push( $last_tag_history, '' );
1038 array_push( $tr_history, false );
1039 array_push( $tr_attributes, '' );
1040 array_push( $has_opened_tr, false );
1041 } elseif ( count( $td_history ) == 0 ) {
1042 # Don't do any of the following
1043 $out .= $outLine . "\n";
1044 continue;
1045 } elseif ( substr( $line, 0, 2 ) === '|}' ) {
1046 # We are ending a table
1047 $line = '</table>' . substr( $line, 2 );
1048 $last_tag = array_pop( $last_tag_history );
1049
1050 if ( !array_pop( $has_opened_tr ) ) {
1051 $line = "<tr><td></td></tr>{$line}";
1052 }
1053
1054 if ( array_pop( $tr_history ) ) {
1055 $line = "</tr>{$line}";
1056 }
1057
1058 if ( array_pop( $td_history ) ) {
1059 $line = "</{$last_tag}>{$line}";
1060 }
1061 array_pop( $tr_attributes );
1062 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1063 } elseif ( substr( $line, 0, 2 ) === '|-' ) {
1064 # Now we have a table row
1065 $line = preg_replace( '#^\|-+#', '', $line );
1066
1067 # Whats after the tag is now only attributes
1068 $attributes = $this->mStripState->unstripBoth( $line );
1069 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1070 array_pop( $tr_attributes );
1071 array_push( $tr_attributes, $attributes );
1072
1073 $line = '';
1074 $last_tag = array_pop( $last_tag_history );
1075 array_pop( $has_opened_tr );
1076 array_push( $has_opened_tr, true );
1077
1078 if ( array_pop( $tr_history ) ) {
1079 $line = '</tr>';
1080 }
1081
1082 if ( array_pop( $td_history ) ) {
1083 $line = "</{$last_tag}>{$line}";
1084 }
1085
1086 $outLine = $line;
1087 array_push( $tr_history, false );
1088 array_push( $td_history, false );
1089 array_push( $last_tag_history, '' );
1090 } elseif ( $first_character === '|'
1091 || $first_character === '!'
1092 || substr( $line, 0, 2 ) === '|+'
1093 ) {
1094 # This might be cell elements, td, th or captions
1095 if ( substr( $line, 0, 2 ) === '|+' ) {
1096 $first_character = '+';
1097 $line = substr( $line, 1 );
1098 }
1099
1100 $line = substr( $line, 1 );
1101
1102 if ( $first_character === '!' ) {
1103 $line = str_replace( '!!', '||', $line );
1104 }
1105
1106 # Split up multiple cells on the same line.
1107 # FIXME : This can result in improper nesting of tags processed
1108 # by earlier parser steps, but should avoid splitting up eg
1109 # attribute values containing literal "||".
1110 $cells = StringUtils::explodeMarkup( '||', $line );
1111
1112 $outLine = '';
1113
1114 # Loop through each table cell
1115 foreach ( $cells as $cell ) {
1116 $previous = '';
1117 if ( $first_character !== '+' ) {
1118 $tr_after = array_pop( $tr_attributes );
1119 if ( !array_pop( $tr_history ) ) {
1120 $previous = "<tr{$tr_after}>\n";
1121 }
1122 array_push( $tr_history, true );
1123 array_push( $tr_attributes, '' );
1124 array_pop( $has_opened_tr );
1125 array_push( $has_opened_tr, true );
1126 }
1127
1128 $last_tag = array_pop( $last_tag_history );
1129
1130 if ( array_pop( $td_history ) ) {
1131 $previous = "</{$last_tag}>\n{$previous}";
1132 }
1133
1134 if ( $first_character === '|' ) {
1135 $last_tag = 'td';
1136 } elseif ( $first_character === '!' ) {
1137 $last_tag = 'th';
1138 } elseif ( $first_character === '+' ) {
1139 $last_tag = 'caption';
1140 } else {
1141 $last_tag = '';
1142 }
1143
1144 array_push( $last_tag_history, $last_tag );
1145
1146 # A cell could contain both parameters and data
1147 $cell_data = explode( '|', $cell, 2 );
1148
1149 # Bug 553: Note that a '|' inside an invalid link should not
1150 # be mistaken as delimiting cell parameters
1151 if ( strpos( $cell_data[0], '[[' ) !== false ) {
1152 $cell = "{$previous}<{$last_tag}>{$cell}";
1153 } elseif ( count( $cell_data ) == 1 ) {
1154 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1155 } else {
1156 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1157 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1158 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1159 }
1160
1161 $outLine .= $cell;
1162 array_push( $td_history, true );
1163 }
1164 }
1165 $out .= $outLine . "\n";
1166 }
1167
1168 # Closing open td, tr && table
1169 while ( count( $td_history ) > 0 ) {
1170 if ( array_pop( $td_history ) ) {
1171 $out .= "</td>\n";
1172 }
1173 if ( array_pop( $tr_history ) ) {
1174 $out .= "</tr>\n";
1175 }
1176 if ( !array_pop( $has_opened_tr ) ) {
1177 $out .= "<tr><td></td></tr>\n";
1178 }
1179
1180 $out .= "</table>\n";
1181 }
1182
1183 # Remove trailing line-ending (b/c)
1184 if ( substr( $out, -1 ) === "\n" ) {
1185 $out = substr( $out, 0, -1 );
1186 }
1187
1188 # special case: don't return empty table
1189 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1190 $out = '';
1191 }
1192
1193 return $out;
1194 }
1195
1196 /**
1197 * Helper function for parse() that transforms wiki markup into half-parsed
1198 * HTML. Only called for $mOutputType == self::OT_HTML.
1199 *
1200 * @private
1201 *
1202 * @param string $text
1203 * @param bool $isMain
1204 * @param PPFrame|bool $frame
1205 *
1206 * @return string
1207 */
1208 public function internalParse( $text, $isMain = true, $frame = false ) {
1209
1210 $origText = $text;
1211
1212 # Hook to suspend the parser in this state
1213 if ( !Hooks::run( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1214 return $text;
1215 }
1216
1217 # if $frame is provided, then use $frame for replacing any variables
1218 if ( $frame ) {
1219 # use frame depth to infer how include/noinclude tags should be handled
1220 # depth=0 means this is the top-level document; otherwise it's an included document
1221 if ( !$frame->depth ) {
1222 $flag = 0;
1223 } else {
1224 $flag = Parser::PTD_FOR_INCLUSION;
1225 }
1226 $dom = $this->preprocessToDom( $text, $flag );
1227 $text = $frame->expand( $dom );
1228 } else {
1229 # if $frame is not provided, then use old-style replaceVariables
1230 $text = $this->replaceVariables( $text );
1231 }
1232
1233 Hooks::run( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1234 $text = Sanitizer::removeHTMLtags(
1235 $text,
1236 array( &$this, 'attributeStripCallback' ),
1237 false,
1238 array_keys( $this->mTransparentTagHooks )
1239 );
1240 Hooks::run( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1241
1242 # Tables need to come after variable replacement for things to work
1243 # properly; putting them before other transformations should keep
1244 # exciting things like link expansions from showing up in surprising
1245 # places.
1246 $text = $this->doTableStuff( $text );
1247
1248 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1249
1250 $text = $this->doDoubleUnderscore( $text );
1251
1252 $text = $this->doHeadings( $text );
1253 $text = $this->replaceInternalLinks( $text );
1254 $text = $this->doAllQuotes( $text );
1255 $text = $this->replaceExternalLinks( $text );
1256
1257 # replaceInternalLinks may sometimes leave behind
1258 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1259 $text = str_replace( self::MARKER_PREFIX . 'NOPARSE', '', $text );
1260
1261 $text = $this->doMagicLinks( $text );
1262 $text = $this->formatHeadings( $text, $origText, $isMain );
1263
1264 return $text;
1265 }
1266
1267 /**
1268 * Helper function for parse() that transforms half-parsed HTML into fully
1269 * parsed HTML.
1270 *
1271 * @param string $text
1272 * @param bool $isMain
1273 * @param bool $linestart
1274 * @return string
1275 */
1276 private function internalParseHalfParsed( $text, $isMain = true, $linestart = true ) {
1277 global $wgUseTidy, $wgAlwaysUseTidy;
1278
1279 $text = $this->mStripState->unstripGeneral( $text );
1280
1281 if ( $isMain ) {
1282 Hooks::run( 'ParserAfterUnstrip', array( &$this, &$text ) );
1283 }
1284
1285 # Clean up special characters, only run once, next-to-last before doBlockLevels
1286 $fixtags = array(
1287 # french spaces, last one Guillemet-left
1288 # only if there is something before the space
1289 '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
1290 # french spaces, Guillemet-right
1291 '/(\\302\\253) /' => '\\1&#160;',
1292 '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
1293 );
1294 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
1295
1296 $text = $this->doBlockLevels( $text, $linestart );
1297
1298 $this->replaceLinkHolders( $text );
1299
1300 /**
1301 * The input doesn't get language converted if
1302 * a) It's disabled
1303 * b) Content isn't converted
1304 * c) It's a conversion table
1305 * d) it is an interface message (which is in the user language)
1306 */
1307 if ( !( $this->mOptions->getDisableContentConversion()
1308 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) )
1309 ) {
1310 if ( !$this->mOptions->getInterfaceMessage() ) {
1311 # The position of the convert() call should not be changed. it
1312 # assumes that the links are all replaced and the only thing left
1313 # is the <nowiki> mark.
1314 $text = $this->getConverterLanguage()->convert( $text );
1315 }
1316 }
1317
1318 $text = $this->mStripState->unstripNoWiki( $text );
1319
1320 if ( $isMain ) {
1321 Hooks::run( 'ParserBeforeTidy', array( &$this, &$text ) );
1322 }
1323
1324 $text = $this->replaceTransparentTags( $text );
1325 $text = $this->mStripState->unstripGeneral( $text );
1326
1327 $text = Sanitizer::normalizeCharReferences( $text );
1328
1329 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
1330 $text = MWTidy::tidy( $text );
1331 } else {
1332 # attempt to sanitize at least some nesting problems
1333 # (bug #2702 and quite a few others)
1334 $tidyregs = array(
1335 # ''Something [http://www.cool.com cool''] -->
1336 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
1337 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
1338 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
1339 # fix up an anchor inside another anchor, only
1340 # at least for a single single nested link (bug 3695)
1341 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
1342 '\\1\\2</a>\\3</a>\\1\\4</a>',
1343 # fix div inside inline elements- doBlockLevels won't wrap a line which
1344 # contains a div, so fix it up here; replace
1345 # div with escaped text
1346 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
1347 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
1348 # remove empty italic or bold tag pairs, some
1349 # introduced by rules above
1350 '/<([bi])><\/\\1>/' => '',
1351 );
1352
1353 $text = preg_replace(
1354 array_keys( $tidyregs ),
1355 array_values( $tidyregs ),
1356 $text );
1357 }
1358
1359 if ( $isMain ) {
1360 Hooks::run( 'ParserAfterTidy', array( &$this, &$text ) );
1361 }
1362
1363 return $text;
1364 }
1365
1366 /**
1367 * Replace special strings like "ISBN xxx" and "RFC xxx" with
1368 * magic external links.
1369 *
1370 * DML
1371 * @private
1372 *
1373 * @param string $text
1374 *
1375 * @return string
1376 */
1377 public function doMagicLinks( $text ) {
1378 $prots = wfUrlProtocolsWithoutProtRel();
1379 $urlChar = self::EXT_LINK_URL_CLASS;
1380 $space = self::SPACE_NOT_NL; # non-newline space
1381 $spdash = "(?:-|$space)"; # a dash or a non-newline space
1382 $spaces = "$space++"; # possessive match of 1 or more spaces
1383 $text = preg_replace_callback(
1384 '!(?: # Start cases
1385 (<a[ \t\r\n>].*?</a>) | # m[1]: Skip link text
1386 (<.*?>) | # m[2]: Skip stuff inside HTML elements' . "
1387 (\b(?i:$prots)$urlChar+) | # m[3]: Free external links
1388 \b(?:RFC|PMID) $spaces # m[4]: RFC or PMID, capture number
1389 ([0-9]+)\b |
1390 \bISBN $spaces ( # m[5]: ISBN, capture number
1391 (?: 97[89] $spdash? )? # optional 13-digit ISBN prefix
1392 (?: [0-9] $spdash? ){9} # 9 digits with opt. delimiters
1393 [0-9Xx] # check digit
1394 )\b
1395 )!xu", array( &$this, 'magicLinkCallback' ), $text );
1396 return $text;
1397 }
1398
1399 /**
1400 * @throws MWException
1401 * @param array $m
1402 * @return HTML|string
1403 */
1404 public function magicLinkCallback( $m ) {
1405 if ( isset( $m[1] ) && $m[1] !== '' ) {
1406 # Skip anchor
1407 return $m[0];
1408 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1409 # Skip HTML element
1410 return $m[0];
1411 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1412 # Free external link
1413 return $this->makeFreeExternalLink( $m[0] );
1414 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1415 # RFC or PMID
1416 if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1417 $keyword = 'RFC';
1418 $urlmsg = 'rfcurl';
1419 $cssClass = 'mw-magiclink-rfc';
1420 $id = $m[4];
1421 } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1422 $keyword = 'PMID';
1423 $urlmsg = 'pubmedurl';
1424 $cssClass = 'mw-magiclink-pmid';
1425 $id = $m[4];
1426 } else {
1427 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1428 substr( $m[0], 0, 20 ) . '"' );
1429 }
1430 $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1431 return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $cssClass );
1432 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1433 # ISBN
1434 $isbn = $m[5];
1435 $space = self::SPACE_NOT_NL; # non-newline space
1436 $isbn = preg_replace( "/$space/", ' ', $isbn );
1437 $num = strtr( $isbn, array(
1438 '-' => '',
1439 ' ' => '',
1440 'x' => 'X',
1441 ));
1442 $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1443 return '<a href="' .
1444 htmlspecialchars( $titleObj->getLocalURL() ) .
1445 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1446 } else {
1447 return $m[0];
1448 }
1449 }
1450
1451 /**
1452 * Make a free external link, given a user-supplied URL
1453 *
1454 * @param string $url
1455 *
1456 * @return string HTML
1457 * @private
1458 */
1459 public function makeFreeExternalLink( $url ) {
1460
1461 $trail = '';
1462
1463 # The characters '<' and '>' (which were escaped by
1464 # removeHTMLtags()) should not be included in
1465 # URLs, per RFC 2396.
1466 $m2 = array();
1467 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1468 $trail = substr( $url, $m2[0][1] ) . $trail;
1469 $url = substr( $url, 0, $m2[0][1] );
1470 }
1471
1472 # Move trailing punctuation to $trail
1473 $sep = ',;\.:!?';
1474 # If there is no left bracket, then consider right brackets fair game too
1475 if ( strpos( $url, '(' ) === false ) {
1476 $sep .= ')';
1477 }
1478
1479 $urlRev = strrev( $url );
1480 $numSepChars = strspn( $urlRev, $sep );
1481 # Don't break a trailing HTML entity by moving the ; into $trail
1482 # This is in hot code, so use substr_compare to avoid having to
1483 # create a new string object for the comparison
1484 if ( $numSepChars && substr_compare( $url, ";", -$numSepChars, 1 ) === 0) {
1485 # more optimization: instead of running preg_match with a $
1486 # anchor, which can be slow, do the match on the reversed
1487 # string starting at the desired offset.
1488 # un-reversed regexp is: /&([a-z]+|#x[\da-f]+|#\d+)$/i
1489 if ( preg_match( '/\G([a-z]+|[\da-f]+x#|\d+#)&/i', $urlRev, $m2, 0, $numSepChars ) ) {
1490 $numSepChars--;
1491 }
1492 }
1493 if ( $numSepChars ) {
1494 $trail = substr( $url, -$numSepChars ) . $trail;
1495 $url = substr( $url, 0, -$numSepChars );
1496 }
1497
1498 $url = Sanitizer::cleanUrl( $url );
1499
1500 # Is this an external image?
1501 $text = $this->maybeMakeExternalImage( $url );
1502 if ( $text === false ) {
1503 # Not an image, make a link
1504 $text = Linker::makeExternalLink( $url,
1505 $this->getConverterLanguage()->markNoConversion( $url, true ),
1506 true, 'free',
1507 $this->getExternalLinkAttribs( $url ) );
1508 # Register it in the output object...
1509 # Replace unnecessary URL escape codes with their equivalent characters
1510 $pasteurized = self::normalizeLinkUrl( $url );
1511 $this->mOutput->addExternalLink( $pasteurized );
1512 }
1513 return $text . $trail;
1514 }
1515
1516 /**
1517 * Parse headers and return html
1518 *
1519 * @private
1520 *
1521 * @param string $text
1522 *
1523 * @return string
1524 */
1525 public function doHeadings( $text ) {
1526 for ( $i = 6; $i >= 1; --$i ) {
1527 $h = str_repeat( '=', $i );
1528 $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1529 }
1530 return $text;
1531 }
1532
1533 /**
1534 * Replace single quotes with HTML markup
1535 * @private
1536 *
1537 * @param string $text
1538 *
1539 * @return string The altered text
1540 */
1541 public function doAllQuotes( $text ) {
1542 $outtext = '';
1543 $lines = StringUtils::explode( "\n", $text );
1544 foreach ( $lines as $line ) {
1545 $outtext .= $this->doQuotes( $line ) . "\n";
1546 }
1547 $outtext = substr( $outtext, 0, -1 );
1548 return $outtext;
1549 }
1550
1551 /**
1552 * Helper function for doAllQuotes()
1553 *
1554 * @param string $text
1555 *
1556 * @return string
1557 */
1558 public function doQuotes( $text ) {
1559 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1560 $countarr = count( $arr );
1561 if ( $countarr == 1 ) {
1562 return $text;
1563 }
1564
1565 // First, do some preliminary work. This may shift some apostrophes from
1566 // being mark-up to being text. It also counts the number of occurrences
1567 // of bold and italics mark-ups.
1568 $numbold = 0;
1569 $numitalics = 0;
1570 for ( $i = 1; $i < $countarr; $i += 2 ) {
1571 $thislen = strlen( $arr[$i] );
1572 // If there are ever four apostrophes, assume the first is supposed to
1573 // be text, and the remaining three constitute mark-up for bold text.
1574 // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1575 if ( $thislen == 4 ) {
1576 $arr[$i - 1] .= "'";
1577 $arr[$i] = "'''";
1578 $thislen = 3;
1579 } elseif ( $thislen > 5 ) {
1580 // If there are more than 5 apostrophes in a row, assume they're all
1581 // text except for the last 5.
1582 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1583 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1584 $arr[$i] = "'''''";
1585 $thislen = 5;
1586 }
1587 // Count the number of occurrences of bold and italics mark-ups.
1588 if ( $thislen == 2 ) {
1589 $numitalics++;
1590 } elseif ( $thislen == 3 ) {
1591 $numbold++;
1592 } elseif ( $thislen == 5 ) {
1593 $numitalics++;
1594 $numbold++;
1595 }
1596 }
1597
1598 // If there is an odd number of both bold and italics, it is likely
1599 // that one of the bold ones was meant to be an apostrophe followed
1600 // by italics. Which one we cannot know for certain, but it is more
1601 // likely to be one that has a single-letter word before it.
1602 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1603 $firstsingleletterword = -1;
1604 $firstmultiletterword = -1;
1605 $firstspace = -1;
1606 for ( $i = 1; $i < $countarr; $i += 2 ) {
1607 if ( strlen( $arr[$i] ) == 3 ) {
1608 $x1 = substr( $arr[$i - 1], -1 );
1609 $x2 = substr( $arr[$i - 1], -2, 1 );
1610 if ( $x1 === ' ' ) {
1611 if ( $firstspace == -1 ) {
1612 $firstspace = $i;
1613 }
1614 } elseif ( $x2 === ' ' ) {
1615 if ( $firstsingleletterword == -1 ) {
1616 $firstsingleletterword = $i;
1617 // if $firstsingleletterword is set, we don't
1618 // look at the other options, so we can bail early.
1619 break;
1620 }
1621 } else {
1622 if ( $firstmultiletterword == -1 ) {
1623 $firstmultiletterword = $i;
1624 }
1625 }
1626 }
1627 }
1628
1629 // If there is a single-letter word, use it!
1630 if ( $firstsingleletterword > -1 ) {
1631 $arr[$firstsingleletterword] = "''";
1632 $arr[$firstsingleletterword - 1] .= "'";
1633 } elseif ( $firstmultiletterword > -1 ) {
1634 // If not, but there's a multi-letter word, use that one.
1635 $arr[$firstmultiletterword] = "''";
1636 $arr[$firstmultiletterword - 1] .= "'";
1637 } elseif ( $firstspace > -1 ) {
1638 // ... otherwise use the first one that has neither.
1639 // (notice that it is possible for all three to be -1 if, for example,
1640 // there is only one pentuple-apostrophe in the line)
1641 $arr[$firstspace] = "''";
1642 $arr[$firstspace - 1] .= "'";
1643 }
1644 }
1645
1646 // Now let's actually convert our apostrophic mush to HTML!
1647 $output = '';
1648 $buffer = '';
1649 $state = '';
1650 $i = 0;
1651 foreach ( $arr as $r ) {
1652 if ( ( $i % 2 ) == 0 ) {
1653 if ( $state === 'both' ) {
1654 $buffer .= $r;
1655 } else {
1656 $output .= $r;
1657 }
1658 } else {
1659 $thislen = strlen( $r );
1660 if ( $thislen == 2 ) {
1661 if ( $state === 'i' ) {
1662 $output .= '</i>';
1663 $state = '';
1664 } elseif ( $state === 'bi' ) {
1665 $output .= '</i>';
1666 $state = 'b';
1667 } elseif ( $state === 'ib' ) {
1668 $output .= '</b></i><b>';
1669 $state = 'b';
1670 } elseif ( $state === 'both' ) {
1671 $output .= '<b><i>' . $buffer . '</i>';
1672 $state = 'b';
1673 } else { // $state can be 'b' or ''
1674 $output .= '<i>';
1675 $state .= 'i';
1676 }
1677 } elseif ( $thislen == 3 ) {
1678 if ( $state === 'b' ) {
1679 $output .= '</b>';
1680 $state = '';
1681 } elseif ( $state === 'bi' ) {
1682 $output .= '</i></b><i>';
1683 $state = 'i';
1684 } elseif ( $state === 'ib' ) {
1685 $output .= '</b>';
1686 $state = 'i';
1687 } elseif ( $state === 'both' ) {
1688 $output .= '<i><b>' . $buffer . '</b>';
1689 $state = 'i';
1690 } else { // $state can be 'i' or ''
1691 $output .= '<b>';
1692 $state .= 'b';
1693 }
1694 } elseif ( $thislen == 5 ) {
1695 if ( $state === 'b' ) {
1696 $output .= '</b><i>';
1697 $state = 'i';
1698 } elseif ( $state === 'i' ) {
1699 $output .= '</i><b>';
1700 $state = 'b';
1701 } elseif ( $state === 'bi' ) {
1702 $output .= '</i></b>';
1703 $state = '';
1704 } elseif ( $state === 'ib' ) {
1705 $output .= '</b></i>';
1706 $state = '';
1707 } elseif ( $state === 'both' ) {
1708 $output .= '<i><b>' . $buffer . '</b></i>';
1709 $state = '';
1710 } else { // ($state == '')
1711 $buffer = '';
1712 $state = 'both';
1713 }
1714 }
1715 }
1716 $i++;
1717 }
1718 // Now close all remaining tags. Notice that the order is important.
1719 if ( $state === 'b' || $state === 'ib' ) {
1720 $output .= '</b>';
1721 }
1722 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1723 $output .= '</i>';
1724 }
1725 if ( $state === 'bi' ) {
1726 $output .= '</b>';
1727 }
1728 // There might be lonely ''''', so make sure we have a buffer
1729 if ( $state === 'both' && $buffer ) {
1730 $output .= '<b><i>' . $buffer . '</i></b>';
1731 }
1732 return $output;
1733 }
1734
1735 /**
1736 * Replace external links (REL)
1737 *
1738 * Note: this is all very hackish and the order of execution matters a lot.
1739 * Make sure to run tests/parserTests.php if you change this code.
1740 *
1741 * @private
1742 *
1743 * @param string $text
1744 *
1745 * @throws MWException
1746 * @return string
1747 */
1748 public function replaceExternalLinks( $text ) {
1749
1750 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1751 if ( $bits === false ) {
1752 throw new MWException( "PCRE needs to be compiled with "
1753 . "--enable-unicode-properties in order for MediaWiki to function" );
1754 }
1755 $s = array_shift( $bits );
1756
1757 $i = 0;
1758 while ( $i < count( $bits ) ) {
1759 $url = $bits[$i++];
1760 $i++; // protocol
1761 $text = $bits[$i++];
1762 $trail = $bits[$i++];
1763
1764 # The characters '<' and '>' (which were escaped by
1765 # removeHTMLtags()) should not be included in
1766 # URLs, per RFC 2396.
1767 $m2 = array();
1768 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1769 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1770 $url = substr( $url, 0, $m2[0][1] );
1771 }
1772
1773 # If the link text is an image URL, replace it with an <img> tag
1774 # This happened by accident in the original parser, but some people used it extensively
1775 $img = $this->maybeMakeExternalImage( $text );
1776 if ( $img !== false ) {
1777 $text = $img;
1778 }
1779
1780 $dtrail = '';
1781
1782 # Set linktype for CSS - if URL==text, link is essentially free
1783 $linktype = ( $text === $url ) ? 'free' : 'text';
1784
1785 # No link text, e.g. [http://domain.tld/some.link]
1786 if ( $text == '' ) {
1787 # Autonumber
1788 $langObj = $this->getTargetLanguage();
1789 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1790 $linktype = 'autonumber';
1791 } else {
1792 # Have link text, e.g. [http://domain.tld/some.link text]s
1793 # Check for trail
1794 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1795 }
1796
1797 $text = $this->getConverterLanguage()->markNoConversion( $text );
1798
1799 $url = Sanitizer::cleanUrl( $url );
1800
1801 # Use the encoded URL
1802 # This means that users can paste URLs directly into the text
1803 # Funny characters like ö aren't valid in URLs anyway
1804 # This was changed in August 2004
1805 $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1806 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1807
1808 # Register link in the output object.
1809 # Replace unnecessary URL escape codes with the referenced character
1810 # This prevents spammers from hiding links from the filters
1811 $pasteurized = self::normalizeLinkUrl( $url );
1812 $this->mOutput->addExternalLink( $pasteurized );
1813 }
1814
1815 return $s;
1816 }
1817
1818 /**
1819 * Get the rel attribute for a particular external link.
1820 *
1821 * @since 1.21
1822 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1823 * nofollow if appropriate
1824 * @param Title $title Optional Title, for wgNoFollowNsExceptions lookups
1825 * @return string|null Rel attribute for $url
1826 */
1827 public static function getExternalLinkRel( $url = false, $title = null ) {
1828 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1829 $ns = $title ? $title->getNamespace() : false;
1830 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions )
1831 && !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions )
1832 ) {
1833 return 'nofollow';
1834 }
1835 return null;
1836 }
1837
1838 /**
1839 * Get an associative array of additional HTML attributes appropriate for a
1840 * particular external link. This currently may include rel => nofollow
1841 * (depending on configuration, namespace, and the URL's domain) and/or a
1842 * target attribute (depending on configuration).
1843 *
1844 * @param string|bool $url Optional URL, to extract the domain from for rel =>
1845 * nofollow if appropriate
1846 * @return array Associative array of HTML attributes
1847 */
1848 public function getExternalLinkAttribs( $url = false ) {
1849 $attribs = array();
1850 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1851
1852 if ( $this->mOptions->getExternalLinkTarget() ) {
1853 $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1854 }
1855 return $attribs;
1856 }
1857
1858 /**
1859 * Replace unusual escape codes in a URL with their equivalent characters
1860 *
1861 * @deprecated since 1.24, use normalizeLinkUrl
1862 * @param string $url
1863 * @return string
1864 */
1865 public static function replaceUnusualEscapes( $url ) {
1866 wfDeprecated( __METHOD__, '1.24' );
1867 return self::normalizeLinkUrl( $url );
1868 }
1869
1870 /**
1871 * Replace unusual escape codes in a URL with their equivalent characters
1872 *
1873 * This generally follows the syntax defined in RFC 3986, with special
1874 * consideration for HTTP query strings.
1875 *
1876 * @param string $url
1877 * @return string
1878 */
1879 public static function normalizeLinkUrl( $url ) {
1880 # First, make sure unsafe characters are encoded
1881 $url = preg_replace_callback( '/[\x00-\x20"<>\[\\\\\]^`{|}\x7F-\xFF]/',
1882 function ( $m ) {
1883 return rawurlencode( $m[0] );
1884 },
1885 $url
1886 );
1887
1888 $ret = '';
1889 $end = strlen( $url );
1890
1891 # Fragment part - 'fragment'
1892 $start = strpos( $url, '#' );
1893 if ( $start !== false && $start < $end ) {
1894 $ret = self::normalizeUrlComponent(
1895 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}' ) . $ret;
1896 $end = $start;
1897 }
1898
1899 # Query part - 'query' minus &=+;
1900 $start = strpos( $url, '?' );
1901 if ( $start !== false && $start < $end ) {
1902 $ret = self::normalizeUrlComponent(
1903 substr( $url, $start, $end - $start ), '"#%<>[\]^`{|}&=+;' ) . $ret;
1904 $end = $start;
1905 }
1906
1907 # Scheme and path part - 'pchar'
1908 # (we assume no userinfo or encoded colons in the host)
1909 $ret = self::normalizeUrlComponent(
1910 substr( $url, 0, $end ), '"#%<>[\]^`{|}/?' ) . $ret;
1911
1912 return $ret;
1913 }
1914
1915 private static function normalizeUrlComponent( $component, $unsafe ) {
1916 $callback = function ( $matches ) use ( $unsafe ) {
1917 $char = urldecode( $matches[0] );
1918 $ord = ord( $char );
1919 if ( $ord > 32 && $ord < 127 && strpos( $unsafe, $char ) === false ) {
1920 # Unescape it
1921 return $char;
1922 } else {
1923 # Leave it escaped, but use uppercase for a-f
1924 return strtoupper( $matches[0] );
1925 }
1926 };
1927 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/', $callback, $component );
1928 }
1929
1930 /**
1931 * make an image if it's allowed, either through the global
1932 * option, through the exception, or through the on-wiki whitelist
1933 *
1934 * @param string $url
1935 *
1936 * @return string
1937 */
1938 private function maybeMakeExternalImage( $url ) {
1939 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1940 $imagesexception = !empty( $imagesfrom );
1941 $text = false;
1942 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1943 if ( $imagesexception && is_array( $imagesfrom ) ) {
1944 $imagematch = false;
1945 foreach ( $imagesfrom as $match ) {
1946 if ( strpos( $url, $match ) === 0 ) {
1947 $imagematch = true;
1948 break;
1949 }
1950 }
1951 } elseif ( $imagesexception ) {
1952 $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1953 } else {
1954 $imagematch = false;
1955 }
1956
1957 if ( $this->mOptions->getAllowExternalImages()
1958 || ( $imagesexception && $imagematch )
1959 ) {
1960 if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1961 # Image found
1962 $text = Linker::makeExternalImage( $url );
1963 }
1964 }
1965 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1966 && preg_match( self::EXT_IMAGE_REGEX, $url )
1967 ) {
1968 $whitelist = explode(
1969 "\n",
1970 wfMessage( 'external_image_whitelist' )->inContentLanguage()->text()
1971 );
1972
1973 foreach ( $whitelist as $entry ) {
1974 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1975 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1976 continue;
1977 }
1978 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1979 # Image matches a whitelist entry
1980 $text = Linker::makeExternalImage( $url );
1981 break;
1982 }
1983 }
1984 }
1985 return $text;
1986 }
1987
1988 /**
1989 * Process [[ ]] wikilinks
1990 *
1991 * @param string $s
1992 *
1993 * @return string Processed text
1994 *
1995 * @private
1996 */
1997 public function replaceInternalLinks( $s ) {
1998 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1999 return $s;
2000 }
2001
2002 /**
2003 * Process [[ ]] wikilinks (RIL)
2004 * @param string $s
2005 * @throws MWException
2006 * @return LinkHolderArray
2007 *
2008 * @private
2009 */
2010 public function replaceInternalLinks2( &$s ) {
2011 global $wgExtraInterlanguageLinkPrefixes;
2012
2013 static $tc = false, $e1, $e1_img;
2014 # the % is needed to support urlencoded titles as well
2015 if ( !$tc ) {
2016 $tc = Title::legalChars() . '#%';
2017 # Match a link having the form [[namespace:link|alternate]]trail
2018 $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
2019 # Match cases where there is no "]]", which might still be images
2020 $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
2021 }
2022
2023 $holders = new LinkHolderArray( $this );
2024
2025 # split the entire text string on occurrences of [[
2026 $a = StringUtils::explode( '[[', ' ' . $s );
2027 # get the first element (all text up to first [[), and remove the space we added
2028 $s = $a->current();
2029 $a->next();
2030 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
2031 $s = substr( $s, 1 );
2032
2033 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
2034 $e2 = null;
2035 if ( $useLinkPrefixExtension ) {
2036 # Match the end of a line for a word that's not followed by whitespace,
2037 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
2038 global $wgContLang;
2039 $charset = $wgContLang->linkPrefixCharset();
2040 $e2 = "/^((?>.*[^$charset]|))(.+)$/sDu";
2041 }
2042
2043 if ( is_null( $this->mTitle ) ) {
2044 throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
2045 }
2046 $nottalk = !$this->mTitle->isTalkPage();
2047
2048 if ( $useLinkPrefixExtension ) {
2049 $m = array();
2050 if ( preg_match( $e2, $s, $m ) ) {
2051 $first_prefix = $m[2];
2052 } else {
2053 $first_prefix = false;
2054 }
2055 } else {
2056 $prefix = '';
2057 }
2058
2059 $useSubpages = $this->areSubpagesAllowed();
2060
2061 // @codingStandardsIgnoreStart Squiz.WhiteSpace.SemicolonSpacing.Incorrect
2062 # Loop for each link
2063 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
2064 // @codingStandardsIgnoreStart
2065
2066 # Check for excessive memory usage
2067 if ( $holders->isBig() ) {
2068 # Too big
2069 # Do the existence check, replace the link holders and clear the array
2070 $holders->replace( $s );
2071 $holders->clear();
2072 }
2073
2074 if ( $useLinkPrefixExtension ) {
2075 if ( preg_match( $e2, $s, $m ) ) {
2076 $prefix = $m[2];
2077 $s = $m[1];
2078 } else {
2079 $prefix = '';
2080 }
2081 # first link
2082 if ( $first_prefix ) {
2083 $prefix = $first_prefix;
2084 $first_prefix = false;
2085 }
2086 }
2087
2088 $might_be_img = false;
2089
2090 if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
2091 $text = $m[2];
2092 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
2093 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
2094 # the real problem is with the $e1 regex
2095 # See bug 1300.
2096 #
2097 # Still some problems for cases where the ] is meant to be outside punctuation,
2098 # and no image is in sight. See bug 2095.
2099 #
2100 if ( $text !== ''
2101 && substr( $m[3], 0, 1 ) === ']'
2102 && strpos( $text, '[' ) !== false
2103 ) {
2104 $text .= ']'; # so that replaceExternalLinks($text) works later
2105 $m[3] = substr( $m[3], 1 );
2106 }
2107 # fix up urlencoded title texts
2108 if ( strpos( $m[1], '%' ) !== false ) {
2109 # Should anchors '#' also be rejected?
2110 $m[1] = str_replace( array( '<', '>' ), array( '&lt;', '&gt;' ), rawurldecode( $m[1] ) );
2111 }
2112 $trail = $m[3];
2113 } elseif ( preg_match( $e1_img, $line, $m ) ) {
2114 # Invalid, but might be an image with a link in its caption
2115 $might_be_img = true;
2116 $text = $m[2];
2117 if ( strpos( $m[1], '%' ) !== false ) {
2118 $m[1] = rawurldecode( $m[1] );
2119 }
2120 $trail = "";
2121 } else { # Invalid form; output directly
2122 $s .= $prefix . '[[' . $line;
2123 continue;
2124 }
2125
2126 $origLink = $m[1];
2127
2128 # Don't allow internal links to pages containing
2129 # PROTO: where PROTO is a valid URL protocol; these
2130 # should be external links.
2131 if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $origLink ) ) {
2132 $s .= $prefix . '[[' . $line;
2133 continue;
2134 }
2135
2136 # Make subpage if necessary
2137 if ( $useSubpages ) {
2138 $link = $this->maybeDoSubpageLink( $origLink, $text );
2139 } else {
2140 $link = $origLink;
2141 }
2142
2143 $noforce = ( substr( $origLink, 0, 1 ) !== ':' );
2144 if ( !$noforce ) {
2145 # Strip off leading ':'
2146 $link = substr( $link, 1 );
2147 }
2148
2149 $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
2150 if ( $nt === null ) {
2151 $s .= $prefix . '[[' . $line;
2152 continue;
2153 }
2154
2155 $ns = $nt->getNamespace();
2156 $iw = $nt->getInterwiki();
2157
2158 if ( $might_be_img ) { # if this is actually an invalid link
2159 if ( $ns == NS_FILE && $noforce ) { # but might be an image
2160 $found = false;
2161 while ( true ) {
2162 # look at the next 'line' to see if we can close it there
2163 $a->next();
2164 $next_line = $a->current();
2165 if ( $next_line === false || $next_line === null ) {
2166 break;
2167 }
2168 $m = explode( ']]', $next_line, 3 );
2169 if ( count( $m ) == 3 ) {
2170 # the first ]] closes the inner link, the second the image
2171 $found = true;
2172 $text .= "[[{$m[0]}]]{$m[1]}";
2173 $trail = $m[2];
2174 break;
2175 } elseif ( count( $m ) == 2 ) {
2176 # if there's exactly one ]] that's fine, we'll keep looking
2177 $text .= "[[{$m[0]}]]{$m[1]}";
2178 } else {
2179 # if $next_line is invalid too, we need look no further
2180 $text .= '[[' . $next_line;
2181 break;
2182 }
2183 }
2184 if ( !$found ) {
2185 # we couldn't find the end of this imageLink, so output it raw
2186 # but don't ignore what might be perfectly normal links in the text we've examined
2187 $holders->merge( $this->replaceInternalLinks2( $text ) );
2188 $s .= "{$prefix}[[$link|$text";
2189 # note: no $trail, because without an end, there *is* no trail
2190 continue;
2191 }
2192 } else { # it's not an image, so output it raw
2193 $s .= "{$prefix}[[$link|$text";
2194 # note: no $trail, because without an end, there *is* no trail
2195 continue;
2196 }
2197 }
2198
2199 $wasblank = ( $text == '' );
2200 if ( $wasblank ) {
2201 $text = $link;
2202 } else {
2203 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2204 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2205 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2206 # -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2207 $text = $this->doQuotes( $text );
2208 }
2209
2210 # Link not escaped by : , create the various objects
2211 if ( $noforce && !$nt->wasLocalInterwiki() ) {
2212 # Interwikis
2213 if (
2214 $iw && $this->mOptions->getInterwikiMagic() && $nottalk && (
2215 Language::fetchLanguageName( $iw, null, 'mw' ) ||
2216 in_array( $iw, $wgExtraInterlanguageLinkPrefixes )
2217 )
2218 ) {
2219 # Bug 24502: filter duplicates
2220 if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2221 $this->mLangLinkLanguages[$iw] = true;
2222 $this->mOutput->addLanguageLink( $nt->getFullText() );
2223 }
2224
2225 $s = rtrim( $s . $prefix );
2226 $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2227 continue;
2228 }
2229
2230 if ( $ns == NS_FILE ) {
2231 if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2232 if ( $wasblank ) {
2233 # if no parameters were passed, $text
2234 # becomes something like "File:Foo.png",
2235 # which we don't want to pass on to the
2236 # image generator
2237 $text = '';
2238 } else {
2239 # recursively parse links inside the image caption
2240 # actually, this will parse them in any other parameters, too,
2241 # but it might be hard to fix that, and it doesn't matter ATM
2242 $text = $this->replaceExternalLinks( $text );
2243 $holders->merge( $this->replaceInternalLinks2( $text ) );
2244 }
2245 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2246 $s .= $prefix . $this->armorLinks(
2247 $this->makeImage( $nt, $text, $holders ) ) . $trail;
2248 } else {
2249 $s .= $prefix . $trail;
2250 }
2251 continue;
2252 }
2253
2254 if ( $ns == NS_CATEGORY ) {
2255 $s = rtrim( $s . "\n" ); # bug 87
2256
2257 if ( $wasblank ) {
2258 $sortkey = $this->getDefaultSort();
2259 } else {
2260 $sortkey = $text;
2261 }
2262 $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2263 $sortkey = str_replace( "\n", '', $sortkey );
2264 $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2265 $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2266
2267 /**
2268 * Strip the whitespace Category links produce, see bug 87
2269 */
2270 $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2271
2272 continue;
2273 }
2274 }
2275
2276 # Self-link checking. For some languages, variants of the title are checked in
2277 # LinkHolderArray::doVariants() to allow batching the existence checks necessary
2278 # for linking to a different variant.
2279 if ( $ns != NS_SPECIAL && $nt->equals( $this->mTitle ) && !$nt->hasFragment() ) {
2280 $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2281 continue;
2282 }
2283
2284 # NS_MEDIA is a pseudo-namespace for linking directly to a file
2285 # @todo FIXME: Should do batch file existence checks, see comment below
2286 if ( $ns == NS_MEDIA ) {
2287 # Give extensions a chance to select the file revision for us
2288 $options = array();
2289 $descQuery = false;
2290 Hooks::run( 'BeforeParserFetchFileAndTitle',
2291 array( $this, $nt, &$options, &$descQuery ) );
2292 # Fetch and register the file (file title may be different via hooks)
2293 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2294 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2295 $s .= $prefix . $this->armorLinks(
2296 Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2297 continue;
2298 }
2299
2300 # Some titles, such as valid special pages or files in foreign repos, should
2301 # be shown as bluelinks even though they're not included in the page table
2302 #
2303 # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2304 # batch file existence checks for NS_FILE and NS_MEDIA
2305 if ( $iw == '' && $nt->isAlwaysKnown() ) {
2306 $this->mOutput->addLink( $nt );
2307 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2308 } else {
2309 # Links will be added to the output link list after checking
2310 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2311 }
2312 }
2313 return $holders;
2314 }
2315
2316 /**
2317 * Render a forced-blue link inline; protect against double expansion of
2318 * URLs if we're in a mode that prepends full URL prefixes to internal links.
2319 * Since this little disaster has to split off the trail text to avoid
2320 * breaking URLs in the following text without breaking trails on the
2321 * wiki links, it's been made into a horrible function.
2322 *
2323 * @param Title $nt
2324 * @param string $text
2325 * @param array|string $query
2326 * @param string $trail
2327 * @param string $prefix
2328 * @return string HTML-wikitext mix oh yuck
2329 */
2330 public function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2331 list( $inside, $trail ) = Linker::splitTrail( $trail );
2332
2333 if ( is_string( $query ) ) {
2334 $query = wfCgiToArray( $query );
2335 }
2336 if ( $text == '' ) {
2337 $text = htmlspecialchars( $nt->getPrefixedText() );
2338 }
2339
2340 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2341
2342 return $this->armorLinks( $link ) . $trail;
2343 }
2344
2345 /**
2346 * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2347 * going to go through further parsing steps before inline URL expansion.
2348 *
2349 * Not needed quite as much as it used to be since free links are a bit
2350 * more sensible these days. But bracketed links are still an issue.
2351 *
2352 * @param string $text More-or-less HTML
2353 * @return string Less-or-more HTML with NOPARSE bits
2354 */
2355 public function armorLinks( $text ) {
2356 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2357 self::MARKER_PREFIX . "NOPARSE$1", $text );
2358 }
2359
2360 /**
2361 * Return true if subpage links should be expanded on this page.
2362 * @return bool
2363 */
2364 public function areSubpagesAllowed() {
2365 # Some namespaces don't allow subpages
2366 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2367 }
2368
2369 /**
2370 * Handle link to subpage if necessary
2371 *
2372 * @param string $target The source of the link
2373 * @param string &$text The link text, modified as necessary
2374 * @return string The full name of the link
2375 * @private
2376 */
2377 public function maybeDoSubpageLink( $target, &$text ) {
2378 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2379 }
2380
2381 /**#@+
2382 * Used by doBlockLevels()
2383 * @private
2384 *
2385 * @return string
2386 */
2387 public function closeParagraph() {
2388 $result = '';
2389 if ( $this->mLastSection != '' ) {
2390 $result = '</' . $this->mLastSection . ">\n";
2391 }
2392 $this->mInPre = false;
2393 $this->mLastSection = '';
2394 return $result;
2395 }
2396
2397 /**
2398 * getCommon() returns the length of the longest common substring
2399 * of both arguments, starting at the beginning of both.
2400 * @private
2401 *
2402 * @param string $st1
2403 * @param string $st2
2404 *
2405 * @return int
2406 */
2407 public function getCommon( $st1, $st2 ) {
2408 $fl = strlen( $st1 );
2409 $shorter = strlen( $st2 );
2410 if ( $fl < $shorter ) {
2411 $shorter = $fl;
2412 }
2413
2414 for ( $i = 0; $i < $shorter; ++$i ) {
2415 if ( $st1[$i] != $st2[$i] ) {
2416 break;
2417 }
2418 }
2419 return $i;
2420 }
2421
2422 /**
2423 * These next three functions open, continue, and close the list
2424 * element appropriate to the prefix character passed into them.
2425 * @private
2426 *
2427 * @param string $char
2428 *
2429 * @return string
2430 */
2431 public function openList( $char ) {
2432 $result = $this->closeParagraph();
2433
2434 if ( '*' === $char ) {
2435 $result .= "<ul><li>";
2436 } elseif ( '#' === $char ) {
2437 $result .= "<ol><li>";
2438 } elseif ( ':' === $char ) {
2439 $result .= "<dl><dd>";
2440 } elseif ( ';' === $char ) {
2441 $result .= "<dl><dt>";
2442 $this->mDTopen = true;
2443 } else {
2444 $result = '<!-- ERR 1 -->';
2445 }
2446
2447 return $result;
2448 }
2449
2450 /**
2451 * TODO: document
2452 * @param string $char
2453 * @private
2454 *
2455 * @return string
2456 */
2457 public function nextItem( $char ) {
2458 if ( '*' === $char || '#' === $char ) {
2459 return "</li>\n<li>";
2460 } elseif ( ':' === $char || ';' === $char ) {
2461 $close = "</dd>\n";
2462 if ( $this->mDTopen ) {
2463 $close = "</dt>\n";
2464 }
2465 if ( ';' === $char ) {
2466 $this->mDTopen = true;
2467 return $close . '<dt>';
2468 } else {
2469 $this->mDTopen = false;
2470 return $close . '<dd>';
2471 }
2472 }
2473 return '<!-- ERR 2 -->';
2474 }
2475
2476 /**
2477 * @todo Document
2478 * @param string $char
2479 * @private
2480 *
2481 * @return string
2482 */
2483 public function closeList( $char ) {
2484 if ( '*' === $char ) {
2485 $text = "</li></ul>";
2486 } elseif ( '#' === $char ) {
2487 $text = "</li></ol>";
2488 } elseif ( ':' === $char ) {
2489 if ( $this->mDTopen ) {
2490 $this->mDTopen = false;
2491 $text = "</dt></dl>";
2492 } else {
2493 $text = "</dd></dl>";
2494 }
2495 } else {
2496 return '<!-- ERR 3 -->';
2497 }
2498 return $text;
2499 }
2500 /**#@-*/
2501
2502 /**
2503 * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2504 *
2505 * @param string $text
2506 * @param bool $linestart Whether or not this is at the start of a line.
2507 * @private
2508 * @return string The lists rendered as HTML
2509 */
2510 public function doBlockLevels( $text, $linestart ) {
2511
2512 # Parsing through the text line by line. The main thing
2513 # happening here is handling of block-level elements p, pre,
2514 # and making lists from lines starting with * # : etc.
2515 #
2516 $textLines = StringUtils::explode( "\n", $text );
2517
2518 $lastPrefix = $output = '';
2519 $this->mDTopen = $inBlockElem = false;
2520 $prefixLength = 0;
2521 $paragraphStack = false;
2522 $inBlockquote = false;
2523
2524 foreach ( $textLines as $oLine ) {
2525 # Fix up $linestart
2526 if ( !$linestart ) {
2527 $output .= $oLine;
2528 $linestart = true;
2529 continue;
2530 }
2531 # * = ul
2532 # # = ol
2533 # ; = dt
2534 # : = dd
2535
2536 $lastPrefixLength = strlen( $lastPrefix );
2537 $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2538 $preOpenMatch = preg_match( '/<pre/i', $oLine );
2539 # If not in a <pre> element, scan for and figure out what prefixes are there.
2540 if ( !$this->mInPre ) {
2541 # Multiple prefixes may abut each other for nested lists.
2542 $prefixLength = strspn( $oLine, '*#:;' );
2543 $prefix = substr( $oLine, 0, $prefixLength );
2544
2545 # eh?
2546 # ; and : are both from definition-lists, so they're equivalent
2547 # for the purposes of determining whether or not we need to open/close
2548 # elements.
2549 $prefix2 = str_replace( ';', ':', $prefix );
2550 $t = substr( $oLine, $prefixLength );
2551 $this->mInPre = (bool)$preOpenMatch;
2552 } else {
2553 # Don't interpret any other prefixes in preformatted text
2554 $prefixLength = 0;
2555 $prefix = $prefix2 = '';
2556 $t = $oLine;
2557 }
2558
2559 # List generation
2560 if ( $prefixLength && $lastPrefix === $prefix2 ) {
2561 # Same as the last item, so no need to deal with nesting or opening stuff
2562 $output .= $this->nextItem( substr( $prefix, -1 ) );
2563 $paragraphStack = false;
2564
2565 if ( substr( $prefix, -1 ) === ';' ) {
2566 # The one nasty exception: definition lists work like this:
2567 # ; title : definition text
2568 # So we check for : in the remainder text to split up the
2569 # title and definition, without b0rking links.
2570 $term = $t2 = '';
2571 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2572 $t = $t2;
2573 $output .= $term . $this->nextItem( ':' );
2574 }
2575 }
2576 } elseif ( $prefixLength || $lastPrefixLength ) {
2577 # We need to open or close prefixes, or both.
2578
2579 # Either open or close a level...
2580 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2581 $paragraphStack = false;
2582
2583 # Close all the prefixes which aren't shared.
2584 while ( $commonPrefixLength < $lastPrefixLength ) {
2585 $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
2586 --$lastPrefixLength;
2587 }
2588
2589 # Continue the current prefix if appropriate.
2590 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2591 $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
2592 }
2593
2594 # Open prefixes where appropriate.
2595 if ( $lastPrefix && $prefixLength > $commonPrefixLength ) {
2596 $output .= "\n";
2597 }
2598 while ( $prefixLength > $commonPrefixLength ) {
2599 $char = substr( $prefix, $commonPrefixLength, 1 );
2600 $output .= $this->openList( $char );
2601
2602 if ( ';' === $char ) {
2603 # @todo FIXME: This is dupe of code above
2604 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2605 $t = $t2;
2606 $output .= $term . $this->nextItem( ':' );
2607 }
2608 }
2609 ++$commonPrefixLength;
2610 }
2611 if ( !$prefixLength && $lastPrefix ) {
2612 $output .= "\n";
2613 }
2614 $lastPrefix = $prefix2;
2615 }
2616
2617 # If we have no prefixes, go to paragraph mode.
2618 if ( 0 == $prefixLength ) {
2619 # No prefix (not in list)--go to paragraph mode
2620 # XXX: use a stack for nestable elements like span, table and div
2621 $openmatch = preg_match(
2622 '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|'
2623 . '<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS',
2624 $t
2625 );
2626 $closematch = preg_match(
2627 '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'
2628 . '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|<\\/mw:|'
2629 . self::MARKER_PREFIX
2630 . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS',
2631 $t
2632 );
2633
2634 if ( $openmatch || $closematch ) {
2635 $paragraphStack = false;
2636 # @todo bug 5718: paragraph closed
2637 $output .= $this->closeParagraph();
2638 if ( $preOpenMatch && !$preCloseMatch ) {
2639 $this->mInPre = true;
2640 }
2641 $bqOffset = 0;
2642 while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t, $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset ) ) {
2643 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
2644 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
2645 }
2646 $inBlockElem = !$closematch;
2647 } elseif ( !$inBlockElem && !$this->mInPre ) {
2648 if ( ' ' == substr( $t, 0, 1 )
2649 && ( $this->mLastSection === 'pre' || trim( $t ) != '' )
2650 && !$inBlockquote
2651 ) {
2652 # pre
2653 if ( $this->mLastSection !== 'pre' ) {
2654 $paragraphStack = false;
2655 $output .= $this->closeParagraph() . '<pre>';
2656 $this->mLastSection = 'pre';
2657 }
2658 $t = substr( $t, 1 );
2659 } else {
2660 # paragraph
2661 if ( trim( $t ) === '' ) {
2662 if ( $paragraphStack ) {
2663 $output .= $paragraphStack . '<br />';
2664 $paragraphStack = false;
2665 $this->mLastSection = 'p';
2666 } else {
2667 if ( $this->mLastSection !== 'p' ) {
2668 $output .= $this->closeParagraph();
2669 $this->mLastSection = '';
2670 $paragraphStack = '<p>';
2671 } else {
2672 $paragraphStack = '</p><p>';
2673 }
2674 }
2675 } else {
2676 if ( $paragraphStack ) {
2677 $output .= $paragraphStack;
2678 $paragraphStack = false;
2679 $this->mLastSection = 'p';
2680 } elseif ( $this->mLastSection !== 'p' ) {
2681 $output .= $this->closeParagraph() . '<p>';
2682 $this->mLastSection = 'p';
2683 }
2684 }
2685 }
2686 }
2687 }
2688 # somewhere above we forget to get out of pre block (bug 785)
2689 if ( $preCloseMatch && $this->mInPre ) {
2690 $this->mInPre = false;
2691 }
2692 if ( $paragraphStack === false ) {
2693 $output .= $t;
2694 if ( $prefixLength === 0 ) {
2695 $output .= "\n";
2696 }
2697 }
2698 }
2699 while ( $prefixLength ) {
2700 $output .= $this->closeList( $prefix2[$prefixLength - 1] );
2701 --$prefixLength;
2702 if ( !$prefixLength ) {
2703 $output .= "\n";
2704 }
2705 }
2706 if ( $this->mLastSection != '' ) {
2707 $output .= '</' . $this->mLastSection . '>';
2708 $this->mLastSection = '';
2709 }
2710
2711 return $output;
2712 }
2713
2714 /**
2715 * Split up a string on ':', ignoring any occurrences inside tags
2716 * to prevent illegal overlapping.
2717 *
2718 * @param string $str The string to split
2719 * @param string &$before Set to everything before the ':'
2720 * @param string &$after Set to everything after the ':'
2721 * @throws MWException
2722 * @return string The position of the ':', or false if none found
2723 */
2724 public function findColonNoLinks( $str, &$before, &$after ) {
2725
2726 $pos = strpos( $str, ':' );
2727 if ( $pos === false ) {
2728 # Nothing to find!
2729 return false;
2730 }
2731
2732 $lt = strpos( $str, '<' );
2733 if ( $lt === false || $lt > $pos ) {
2734 # Easy; no tag nesting to worry about
2735 $before = substr( $str, 0, $pos );
2736 $after = substr( $str, $pos + 1 );
2737 return $pos;
2738 }
2739
2740 # Ugly state machine to walk through avoiding tags.
2741 $state = self::COLON_STATE_TEXT;
2742 $stack = 0;
2743 $len = strlen( $str );
2744 for ( $i = 0; $i < $len; $i++ ) {
2745 $c = $str[$i];
2746
2747 switch ( $state ) {
2748 # (Using the number is a performance hack for common cases)
2749 case 0: # self::COLON_STATE_TEXT:
2750 switch ( $c ) {
2751 case "<":
2752 # Could be either a <start> tag or an </end> tag
2753 $state = self::COLON_STATE_TAGSTART;
2754 break;
2755 case ":":
2756 if ( $stack == 0 ) {
2757 # We found it!
2758 $before = substr( $str, 0, $i );
2759 $after = substr( $str, $i + 1 );
2760 return $i;
2761 }
2762 # Embedded in a tag; don't break it.
2763 break;
2764 default:
2765 # Skip ahead looking for something interesting
2766 $colon = strpos( $str, ':', $i );
2767 if ( $colon === false ) {
2768 # Nothing else interesting
2769 return false;
2770 }
2771 $lt = strpos( $str, '<', $i );
2772 if ( $stack === 0 ) {
2773 if ( $lt === false || $colon < $lt ) {
2774 # We found it!
2775 $before = substr( $str, 0, $colon );
2776 $after = substr( $str, $colon + 1 );
2777 return $i;
2778 }
2779 }
2780 if ( $lt === false ) {
2781 # Nothing else interesting to find; abort!
2782 # We're nested, but there's no close tags left. Abort!
2783 break 2;
2784 }
2785 # Skip ahead to next tag start
2786 $i = $lt;
2787 $state = self::COLON_STATE_TAGSTART;
2788 }
2789 break;
2790 case 1: # self::COLON_STATE_TAG:
2791 # In a <tag>
2792 switch ( $c ) {
2793 case ">":
2794 $stack++;
2795 $state = self::COLON_STATE_TEXT;
2796 break;
2797 case "/":
2798 # Slash may be followed by >?
2799 $state = self::COLON_STATE_TAGSLASH;
2800 break;
2801 default:
2802 # ignore
2803 }
2804 break;
2805 case 2: # self::COLON_STATE_TAGSTART:
2806 switch ( $c ) {
2807 case "/":
2808 $state = self::COLON_STATE_CLOSETAG;
2809 break;
2810 case "!":
2811 $state = self::COLON_STATE_COMMENT;
2812 break;
2813 case ">":
2814 # Illegal early close? This shouldn't happen D:
2815 $state = self::COLON_STATE_TEXT;
2816 break;
2817 default:
2818 $state = self::COLON_STATE_TAG;
2819 }
2820 break;
2821 case 3: # self::COLON_STATE_CLOSETAG:
2822 # In a </tag>
2823 if ( $c === ">" ) {
2824 $stack--;
2825 if ( $stack < 0 ) {
2826 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
2827 return false;
2828 }
2829 $state = self::COLON_STATE_TEXT;
2830 }
2831 break;
2832 case self::COLON_STATE_TAGSLASH:
2833 if ( $c === ">" ) {
2834 # Yes, a self-closed tag <blah/>
2835 $state = self::COLON_STATE_TEXT;
2836 } else {
2837 # Probably we're jumping the gun, and this is an attribute
2838 $state = self::COLON_STATE_TAG;
2839 }
2840 break;
2841 case 5: # self::COLON_STATE_COMMENT:
2842 if ( $c === "-" ) {
2843 $state = self::COLON_STATE_COMMENTDASH;
2844 }
2845 break;
2846 case self::COLON_STATE_COMMENTDASH:
2847 if ( $c === "-" ) {
2848 $state = self::COLON_STATE_COMMENTDASHDASH;
2849 } else {
2850 $state = self::COLON_STATE_COMMENT;
2851 }
2852 break;
2853 case self::COLON_STATE_COMMENTDASHDASH:
2854 if ( $c === ">" ) {
2855 $state = self::COLON_STATE_TEXT;
2856 } else {
2857 $state = self::COLON_STATE_COMMENT;
2858 }
2859 break;
2860 default:
2861 throw new MWException( "State machine error in " . __METHOD__ );
2862 }
2863 }
2864 if ( $stack > 0 ) {
2865 wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2866 return false;
2867 }
2868 return false;
2869 }
2870
2871 /**
2872 * Return value of a magic variable (like PAGENAME)
2873 *
2874 * @private
2875 *
2876 * @param int $index
2877 * @param bool|PPFrame $frame
2878 *
2879 * @throws MWException
2880 * @return string
2881 */
2882 public function getVariableValue( $index, $frame = false ) {
2883 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2884 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2885
2886 if ( is_null( $this->mTitle ) ) {
2887 // If no title set, bad things are going to happen
2888 // later. Title should always be set since this
2889 // should only be called in the middle of a parse
2890 // operation (but the unit-tests do funky stuff)
2891 throw new MWException( __METHOD__ . ' Should only be '
2892 . ' called while parsing (no title set)' );
2893 }
2894
2895 /**
2896 * Some of these require message or data lookups and can be
2897 * expensive to check many times.
2898 */
2899 if ( Hooks::run( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2900 if ( isset( $this->mVarCache[$index] ) ) {
2901 return $this->mVarCache[$index];
2902 }
2903 }
2904
2905 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2906 Hooks::run( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2907
2908 $pageLang = $this->getFunctionLang();
2909
2910 switch ( $index ) {
2911 case '!':
2912 $value = '|';
2913 break;
2914 case 'currentmonth':
2915 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2916 break;
2917 case 'currentmonth1':
2918 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2919 break;
2920 case 'currentmonthname':
2921 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2922 break;
2923 case 'currentmonthnamegen':
2924 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2925 break;
2926 case 'currentmonthabbrev':
2927 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2928 break;
2929 case 'currentday':
2930 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2931 break;
2932 case 'currentday2':
2933 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2934 break;
2935 case 'localmonth':
2936 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2937 break;
2938 case 'localmonth1':
2939 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2940 break;
2941 case 'localmonthname':
2942 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2943 break;
2944 case 'localmonthnamegen':
2945 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2946 break;
2947 case 'localmonthabbrev':
2948 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2949 break;
2950 case 'localday':
2951 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2952 break;
2953 case 'localday2':
2954 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2955 break;
2956 case 'pagename':
2957 $value = wfEscapeWikiText( $this->mTitle->getText() );
2958 break;
2959 case 'pagenamee':
2960 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2961 break;
2962 case 'fullpagename':
2963 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2964 break;
2965 case 'fullpagenamee':
2966 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2967 break;
2968 case 'subpagename':
2969 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2970 break;
2971 case 'subpagenamee':
2972 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2973 break;
2974 case 'rootpagename':
2975 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2976 break;
2977 case 'rootpagenamee':
2978 $value = wfEscapeWikiText( wfUrlEncode( str_replace(
2979 ' ',
2980 '_',
2981 $this->mTitle->getRootText()
2982 ) ) );
2983 break;
2984 case 'basepagename':
2985 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2986 break;
2987 case 'basepagenamee':
2988 $value = wfEscapeWikiText( wfUrlEncode( str_replace(
2989 ' ',
2990 '_',
2991 $this->mTitle->getBaseText()
2992 ) ) );
2993 break;
2994 case 'talkpagename':
2995 if ( $this->mTitle->canTalk() ) {
2996 $talkPage = $this->mTitle->getTalkPage();
2997 $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2998 } else {
2999 $value = '';
3000 }
3001 break;
3002 case 'talkpagenamee':
3003 if ( $this->mTitle->canTalk() ) {
3004 $talkPage = $this->mTitle->getTalkPage();
3005 $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
3006 } else {
3007 $value = '';
3008 }
3009 break;
3010 case 'subjectpagename':
3011 $subjPage = $this->mTitle->getSubjectPage();
3012 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
3013 break;
3014 case 'subjectpagenamee':
3015 $subjPage = $this->mTitle->getSubjectPage();
3016 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
3017 break;
3018 case 'pageid': // requested in bug 23427
3019 $pageid = $this->getTitle()->getArticleID();
3020 if ( $pageid == 0 ) {
3021 # 0 means the page doesn't exist in the database,
3022 # which means the user is previewing a new page.
3023 # The vary-revision flag must be set, because the magic word
3024 # will have a different value once the page is saved.
3025 $this->mOutput->setFlag( 'vary-revision' );
3026 wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
3027 }
3028 $value = $pageid ? $pageid : null;
3029 break;
3030 case 'revisionid':
3031 # Let the edit saving system know we should parse the page
3032 # *after* a revision ID has been assigned.
3033 $this->mOutput->setFlag( 'vary-revision' );
3034 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
3035 $value = $this->mRevisionId;
3036 break;
3037 case 'revisionday':
3038 # Let the edit saving system know we should parse the page
3039 # *after* a revision ID has been assigned. This is for null edits.
3040 $this->mOutput->setFlag( 'vary-revision' );
3041 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
3042 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
3043 break;
3044 case 'revisionday2':
3045 # Let the edit saving system know we should parse the page
3046 # *after* a revision ID has been assigned. This is for null edits.
3047 $this->mOutput->setFlag( 'vary-revision' );
3048 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
3049 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
3050 break;
3051 case 'revisionmonth':
3052 # Let the edit saving system know we should parse the page
3053 # *after* a revision ID has been assigned. This is for null edits.
3054 $this->mOutput->setFlag( 'vary-revision' );
3055 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
3056 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
3057 break;
3058 case 'revisionmonth1':
3059 # Let the edit saving system know we should parse the page
3060 # *after* a revision ID has been assigned. This is for null edits.
3061 $this->mOutput->setFlag( 'vary-revision' );
3062 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
3063 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
3064 break;
3065 case 'revisionyear':
3066 # Let the edit saving system know we should parse the page
3067 # *after* a revision ID has been assigned. This is for null edits.
3068 $this->mOutput->setFlag( 'vary-revision' );
3069 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
3070 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
3071 break;
3072 case 'revisiontimestamp':
3073 # Let the edit saving system know we should parse the page
3074 # *after* a revision ID has been assigned. This is for null edits.
3075 $this->mOutput->setFlag( 'vary-revision' );
3076 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
3077 $value = $this->getRevisionTimestamp();
3078 break;
3079 case 'revisionuser':
3080 # Let the edit saving system know we should parse the page
3081 # *after* a revision ID has been assigned. This is for null edits.
3082 $this->mOutput->setFlag( 'vary-revision' );
3083 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
3084 $value = $this->getRevisionUser();
3085 break;
3086 case 'revisionsize':
3087 # Let the edit saving system know we should parse the page
3088 # *after* a revision ID has been assigned. This is for null edits.
3089 $this->mOutput->setFlag( 'vary-revision' );
3090 wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
3091 $value = $this->getRevisionSize();
3092 break;
3093 case 'namespace':
3094 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
3095 break;
3096 case 'namespacee':
3097 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
3098 break;
3099 case 'namespacenumber':
3100 $value = $this->mTitle->getNamespace();
3101 break;
3102 case 'talkspace':
3103 $value = $this->mTitle->canTalk()
3104 ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() )
3105 : '';
3106 break;
3107 case 'talkspacee':
3108 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
3109 break;
3110 case 'subjectspace':
3111 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
3112 break;
3113 case 'subjectspacee':
3114 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
3115 break;
3116 case 'currentdayname':
3117 $value = $pageLang->getWeekdayName( (int)MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
3118 break;
3119 case 'currentyear':
3120 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
3121 break;
3122 case 'currenttime':
3123 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
3124 break;
3125 case 'currenthour':
3126 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
3127 break;
3128 case 'currentweek':
3129 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3130 # int to remove the padding
3131 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
3132 break;
3133 case 'currentdow':
3134 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
3135 break;
3136 case 'localdayname':
3137 $value = $pageLang->getWeekdayName(
3138 (int)MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1
3139 );
3140 break;
3141 case 'localyear':
3142 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
3143 break;
3144 case 'localtime':
3145 $value = $pageLang->time(
3146 MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ),
3147 false,
3148 false
3149 );
3150 break;
3151 case 'localhour':
3152 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
3153 break;
3154 case 'localweek':
3155 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
3156 # int to remove the padding
3157 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
3158 break;
3159 case 'localdow':
3160 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
3161 break;
3162 case 'numberofarticles':
3163 $value = $pageLang->formatNum( SiteStats::articles() );
3164 break;
3165 case 'numberoffiles':
3166 $value = $pageLang->formatNum( SiteStats::images() );
3167 break;
3168 case 'numberofusers':
3169 $value = $pageLang->formatNum( SiteStats::users() );
3170 break;
3171 case 'numberofactiveusers':
3172 $value = $pageLang->formatNum( SiteStats::activeUsers() );
3173 break;
3174 case 'numberofpages':
3175 $value = $pageLang->formatNum( SiteStats::pages() );
3176 break;
3177 case 'numberofadmins':
3178 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
3179 break;
3180 case 'numberofedits':
3181 $value = $pageLang->formatNum( SiteStats::edits() );
3182 break;
3183 case 'currenttimestamp':
3184 $value = wfTimestamp( TS_MW, $ts );
3185 break;
3186 case 'localtimestamp':
3187 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
3188 break;
3189 case 'currentversion':
3190 $value = SpecialVersion::getVersion();
3191 break;
3192 case 'articlepath':
3193 return $wgArticlePath;
3194 case 'sitename':
3195 return $wgSitename;
3196 case 'server':
3197 return $wgServer;
3198 case 'servername':
3199 return $wgServerName;
3200 case 'scriptpath':
3201 return $wgScriptPath;
3202 case 'stylepath':
3203 return $wgStylePath;
3204 case 'directionmark':
3205 return $pageLang->getDirMark();
3206 case 'contentlanguage':
3207 global $wgLanguageCode;
3208 return $wgLanguageCode;
3209 case 'cascadingsources':
3210 $value = CoreParserFunctions::cascadingsources( $this );
3211 break;
3212 default:
3213 $ret = null;
3214 Hooks::run(
3215 'ParserGetVariableValueSwitch',
3216 array( &$this, &$this->mVarCache, &$index, &$ret, &$frame )
3217 );
3218
3219 return $ret;
3220 }
3221
3222 if ( $index ) {
3223 $this->mVarCache[$index] = $value;
3224 }
3225
3226 return $value;
3227 }
3228
3229 /**
3230 * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3231 *
3232 * @private
3233 */
3234 public function initialiseVariables() {
3235 $variableIDs = MagicWord::getVariableIDs();
3236 $substIDs = MagicWord::getSubstIDs();
3237
3238 $this->mVariables = new MagicWordArray( $variableIDs );
3239 $this->mSubstWords = new MagicWordArray( $substIDs );
3240 }
3241
3242 /**
3243 * Preprocess some wikitext and return the document tree.
3244 * This is the ghost of replace_variables().
3245 *
3246 * @param string $text The text to parse
3247 * @param int $flags Bitwise combination of:
3248 * - self::PTD_FOR_INCLUSION: Handle "<noinclude>" and "<includeonly>" as if the text is being
3249 * included. Default is to assume a direct page view.
3250 *
3251 * The generated DOM tree must depend only on the input text and the flags.
3252 * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
3253 *
3254 * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3255 * change in the DOM tree for a given text, must be passed through the section identifier
3256 * in the section edit link and thus back to extractSections().
3257 *
3258 * The output of this function is currently only cached in process memory, but a persistent
3259 * cache may be implemented at a later date which takes further advantage of these strict
3260 * dependency requirements.
3261 *
3262 * @return PPNode
3263 */
3264 public function preprocessToDom( $text, $flags = 0 ) {
3265 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3266 return $dom;
3267 }
3268
3269 /**
3270 * Return a three-element array: leading whitespace, string contents, trailing whitespace
3271 *
3272 * @param string $s
3273 *
3274 * @return array
3275 */
3276 public static function splitWhitespace( $s ) {
3277 $ltrimmed = ltrim( $s );
3278 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3279 $trimmed = rtrim( $ltrimmed );
3280 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3281 if ( $diff > 0 ) {
3282 $w2 = substr( $ltrimmed, -$diff );
3283 } else {
3284 $w2 = '';
3285 }
3286 return array( $w1, $trimmed, $w2 );
3287 }
3288
3289 /**
3290 * Replace magic variables, templates, and template arguments
3291 * with the appropriate text. Templates are substituted recursively,
3292 * taking care to avoid infinite loops.
3293 *
3294 * Note that the substitution depends on value of $mOutputType:
3295 * self::OT_WIKI: only {{subst:}} templates
3296 * self::OT_PREPROCESS: templates but not extension tags
3297 * self::OT_HTML: all templates and extension tags
3298 *
3299 * @param string $text The text to transform
3300 * @param bool|PPFrame $frame Object describing the arguments passed to the
3301 * template. Arguments may also be provided as an associative array, as
3302 * was the usual case before MW1.12. Providing arguments this way may be
3303 * useful for extensions wishing to perform variable replacement
3304 * explicitly.
3305 * @param bool $argsOnly Only do argument (triple-brace) expansion, not
3306 * double-brace expansion.
3307 * @return string
3308 */
3309 public function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3310 # Is there any text? Also, Prevent too big inclusions!
3311 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3312 return $text;
3313 }
3314
3315 if ( $frame === false ) {
3316 $frame = $this->getPreprocessor()->newFrame();
3317 } elseif ( !( $frame instanceof PPFrame ) ) {
3318 wfDebug( __METHOD__ . " called using plain parameters instead of "
3319 . "a PPFrame instance. Creating custom frame.\n" );
3320 $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3321 }
3322
3323 $dom = $this->preprocessToDom( $text );
3324 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3325 $text = $frame->expand( $dom, $flags );
3326
3327 return $text;
3328 }
3329
3330 /**
3331 * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3332 *
3333 * @param array $args
3334 *
3335 * @return array
3336 */
3337 public static function createAssocArgs( $args ) {
3338 $assocArgs = array();
3339 $index = 1;
3340 foreach ( $args as $arg ) {
3341 $eqpos = strpos( $arg, '=' );
3342 if ( $eqpos === false ) {
3343 $assocArgs[$index++] = $arg;
3344 } else {
3345 $name = trim( substr( $arg, 0, $eqpos ) );
3346 $value = trim( substr( $arg, $eqpos + 1 ) );
3347 if ( $value === false ) {
3348 $value = '';
3349 }
3350 if ( $name !== false ) {
3351 $assocArgs[$name] = $value;
3352 }
3353 }
3354 }
3355
3356 return $assocArgs;
3357 }
3358
3359 /**
3360 * Warn the user when a parser limitation is reached
3361 * Will warn at most once the user per limitation type
3362 *
3363 * @param string $limitationType Should be one of:
3364 * 'expensive-parserfunction' (corresponding messages:
3365 * 'expensive-parserfunction-warning',
3366 * 'expensive-parserfunction-category')
3367 * 'post-expand-template-argument' (corresponding messages:
3368 * 'post-expand-template-argument-warning',
3369 * 'post-expand-template-argument-category')
3370 * 'post-expand-template-inclusion' (corresponding messages:
3371 * 'post-expand-template-inclusion-warning',
3372 * 'post-expand-template-inclusion-category')
3373 * 'node-count-exceeded' (corresponding messages:
3374 * 'node-count-exceeded-warning',
3375 * 'node-count-exceeded-category')
3376 * 'expansion-depth-exceeded' (corresponding messages:
3377 * 'expansion-depth-exceeded-warning',
3378 * 'expansion-depth-exceeded-category')
3379 * @param string|int|null $current Current value
3380 * @param string|int|null $max Maximum allowed, when an explicit limit has been
3381 * exceeded, provide the values (optional)
3382 */
3383 public function limitationWarn( $limitationType, $current = '', $max = '' ) {
3384 # does no harm if $current and $max are present but are unnecessary for the message
3385 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3386 ->inLanguage( $this->mOptions->getUserLangObj() )->text();
3387 $this->mOutput->addWarning( $warning );
3388 $this->addTrackingCategory( "$limitationType-category" );
3389 }
3390
3391 /**
3392 * Return the text of a template, after recursively
3393 * replacing any variables or templates within the template.
3394 *
3395 * @param array $piece The parts of the template
3396 * $piece['title']: the title, i.e. the part before the |
3397 * $piece['parts']: the parameter array
3398 * $piece['lineStart']: whether the brace was at the start of a line
3399 * @param PPFrame $frame The current frame, contains template arguments
3400 * @throws Exception
3401 * @return string The text of the template
3402 */
3403 public function braceSubstitution( $piece, $frame ) {
3404
3405 // Flags
3406
3407 // $text has been filled
3408 $found = false;
3409 // wiki markup in $text should be escaped
3410 $nowiki = false;
3411 // $text is HTML, armour it against wikitext transformation
3412 $isHTML = false;
3413 // Force interwiki transclusion to be done in raw mode not rendered
3414 $forceRawInterwiki = false;
3415 // $text is a DOM node needing expansion in a child frame
3416 $isChildObj = false;
3417 // $text is a DOM node needing expansion in the current frame
3418 $isLocalObj = false;
3419
3420 # Title object, where $text came from
3421 $title = false;
3422
3423 # $part1 is the bit before the first |, and must contain only title characters.
3424 # Various prefixes will be stripped from it later.
3425 $titleWithSpaces = $frame->expand( $piece['title'] );
3426 $part1 = trim( $titleWithSpaces );
3427 $titleText = false;
3428
3429 # Original title text preserved for various purposes
3430 $originalTitle = $part1;
3431
3432 # $args is a list of argument nodes, starting from index 0, not including $part1
3433 # @todo FIXME: If piece['parts'] is null then the call to getLength()
3434 # below won't work b/c this $args isn't an object
3435 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3436
3437 $profileSection = null; // profile templates
3438
3439 # SUBST
3440 if ( !$found ) {
3441
3442 $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3443
3444 # Possibilities for substMatch: "subst", "safesubst" or FALSE
3445 # Decide whether to expand template or keep wikitext as-is.
3446 if ( $this->ot['wiki'] ) {
3447 if ( $substMatch === false ) {
3448 $literal = true; # literal when in PST with no prefix
3449 } else {
3450 $literal = false; # expand when in PST with subst: or safesubst:
3451 }
3452 } else {
3453 if ( $substMatch == 'subst' ) {
3454 $literal = true; # literal when not in PST with plain subst:
3455 } else {
3456 $literal = false; # expand when not in PST with safesubst: or no prefix
3457 }
3458 }
3459 if ( $literal ) {
3460 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3461 $isLocalObj = true;
3462 $found = true;
3463 }
3464 }
3465
3466 # Variables
3467 if ( !$found && $args->getLength() == 0 ) {
3468 $id = $this->mVariables->matchStartToEnd( $part1 );
3469 if ( $id !== false ) {
3470 $text = $this->getVariableValue( $id, $frame );
3471 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3472 $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3473 }
3474 $found = true;
3475 }
3476 }
3477
3478 # MSG, MSGNW and RAW
3479 if ( !$found ) {
3480 # Check for MSGNW:
3481 $mwMsgnw = MagicWord::get( 'msgnw' );
3482 if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3483 $nowiki = true;
3484 } else {
3485 # Remove obsolete MSG:
3486 $mwMsg = MagicWord::get( 'msg' );
3487 $mwMsg->matchStartAndRemove( $part1 );
3488 }
3489
3490 # Check for RAW:
3491 $mwRaw = MagicWord::get( 'raw' );
3492 if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3493 $forceRawInterwiki = true;
3494 }
3495 }
3496
3497 # Parser functions
3498 if ( !$found ) {
3499
3500 $colonPos = strpos( $part1, ':' );
3501 if ( $colonPos !== false ) {
3502 $func = substr( $part1, 0, $colonPos );
3503 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3504 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3505 $funcArgs[] = $args->item( $i );
3506 }
3507 try {
3508 $result = $this->callParserFunction( $frame, $func, $funcArgs );
3509 } catch ( Exception $ex ) {
3510 throw $ex;
3511 }
3512
3513 # The interface for parser functions allows for extracting
3514 # flags into the local scope. Extract any forwarded flags
3515 # here.
3516 extract( $result );
3517 }
3518 }
3519
3520 # Finish mangling title and then check for loops.
3521 # Set $title to a Title object and $titleText to the PDBK
3522 if ( !$found ) {
3523 $ns = NS_TEMPLATE;
3524 # Split the title into page and subpage
3525 $subpage = '';
3526 $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3527 if ( $part1 !== $relative ) {
3528 $part1 = $relative;
3529 $ns = $this->mTitle->getNamespace();
3530 }
3531 $title = Title::newFromText( $part1, $ns );
3532 if ( $title ) {
3533 $titleText = $title->getPrefixedText();
3534 # Check for language variants if the template is not found
3535 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3536 $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3537 }
3538 # Do recursion depth check
3539 $limit = $this->mOptions->getMaxTemplateDepth();
3540 if ( $frame->depth >= $limit ) {
3541 $found = true;
3542 $text = '<span class="error">'
3543 . wfMessage( 'parser-template-recursion-depth-warning' )
3544 ->numParams( $limit )->inContentLanguage()->text()
3545 . '</span>';
3546 }
3547 }
3548 }
3549
3550 # Load from database
3551 if ( !$found && $title ) {
3552 $profileSection = $this->mProfiler->scopedProfileIn( $title->getPrefixedDBkey() );
3553 if ( !$title->isExternal() ) {
3554 if ( $title->isSpecialPage()
3555 && $this->mOptions->getAllowSpecialInclusion()
3556 && $this->ot['html']
3557 ) {
3558 // Pass the template arguments as URL parameters.
3559 // "uselang" will have no effect since the Language object
3560 // is forced to the one defined in ParserOptions.
3561 $pageArgs = array();
3562 $argsLength = $args->getLength();
3563 for ( $i = 0; $i < $argsLength; $i++ ) {
3564 $bits = $args->item( $i )->splitArg();
3565 if ( strval( $bits['index'] ) === '' ) {
3566 $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3567 $value = trim( $frame->expand( $bits['value'] ) );
3568 $pageArgs[$name] = $value;
3569 }
3570 }
3571
3572 // Create a new context to execute the special page
3573 $context = new RequestContext;
3574 $context->setTitle( $title );
3575 $context->setRequest( new FauxRequest( $pageArgs ) );
3576 $context->setUser( $this->getUser() );
3577 $context->setLanguage( $this->mOptions->getUserLangObj() );
3578 $ret = SpecialPageFactory::capturePath( $title, $context );
3579 if ( $ret ) {
3580 $text = $context->getOutput()->getHTML();
3581 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3582 $found = true;
3583 $isHTML = true;
3584 $this->disableCache();
3585 }
3586 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3587 $found = false; # access denied
3588 wfDebug( __METHOD__ . ": template inclusion denied for " .
3589 $title->getPrefixedDBkey() . "\n" );
3590 } else {
3591 list( $text, $title ) = $this->getTemplateDom( $title );
3592 if ( $text !== false ) {
3593 $found = true;
3594 $isChildObj = true;
3595 }
3596 }
3597
3598 # If the title is valid but undisplayable, make a link to it
3599 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3600 $text = "[[:$titleText]]";
3601 $found = true;
3602 }
3603 } elseif ( $title->isTrans() ) {
3604 # Interwiki transclusion
3605 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3606 $text = $this->interwikiTransclude( $title, 'render' );
3607 $isHTML = true;
3608 } else {
3609 $text = $this->interwikiTransclude( $title, 'raw' );
3610 # Preprocess it like a template
3611 $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3612 $isChildObj = true;
3613 }
3614 $found = true;
3615 }
3616
3617 # Do infinite loop check
3618 # This has to be done after redirect resolution to avoid infinite loops via redirects
3619 if ( !$frame->loopCheck( $title ) ) {
3620 $found = true;
3621 $text = '<span class="error">'
3622 . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3623 . '</span>';
3624 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3625 }
3626 }
3627
3628 # If we haven't found text to substitute by now, we're done
3629 # Recover the source wikitext and return it
3630 if ( !$found ) {
3631 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3632 if ( $profileSection ) {
3633 $this->mProfiler->scopedProfileOut( $profileSection );
3634 }
3635 return array( 'object' => $text );
3636 }
3637
3638 # Expand DOM-style return values in a child frame
3639 if ( $isChildObj ) {
3640 # Clean up argument array
3641 $newFrame = $frame->newChild( $args, $title );
3642
3643 if ( $nowiki ) {
3644 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3645 } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3646 # Expansion is eligible for the empty-frame cache
3647 $text = $newFrame->cachedExpand( $titleText, $text );
3648 } else {
3649 # Uncached expansion
3650 $text = $newFrame->expand( $text );
3651 }
3652 }
3653 if ( $isLocalObj && $nowiki ) {
3654 $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3655 $isLocalObj = false;
3656 }
3657
3658 if ( $profileSection ) {
3659 $this->mProfiler->scopedProfileOut( $profileSection );
3660 }
3661
3662 # Replace raw HTML by a placeholder
3663 if ( $isHTML ) {
3664 $text = $this->insertStripItem( $text );
3665 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3666 # Escape nowiki-style return values
3667 $text = wfEscapeWikiText( $text );
3668 } elseif ( is_string( $text )
3669 && !$piece['lineStart']
3670 && preg_match( '/^(?:{\\||:|;|#|\*)/', $text )
3671 ) {
3672 # Bug 529: if the template begins with a table or block-level
3673 # element, it should be treated as beginning a new line.
3674 # This behavior is somewhat controversial.
3675 $text = "\n" . $text;
3676 }
3677
3678 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3679 # Error, oversize inclusion
3680 if ( $titleText !== false ) {
3681 # Make a working, properly escaped link if possible (bug 23588)
3682 $text = "[[:$titleText]]";
3683 } else {
3684 # This will probably not be a working link, but at least it may
3685 # provide some hint of where the problem is
3686 preg_replace( '/^:/', '', $originalTitle );
3687 $text = "[[:$originalTitle]]";
3688 }
3689 $text .= $this->insertStripItem( '<!-- WARNING: template omitted, '
3690 . 'post-expand include size too large -->' );
3691 $this->limitationWarn( 'post-expand-template-inclusion' );
3692 }
3693
3694 if ( $isLocalObj ) {
3695 $ret = array( 'object' => $text );
3696 } else {
3697 $ret = array( 'text' => $text );
3698 }
3699
3700 return $ret;
3701 }
3702
3703 /**
3704 * Call a parser function and return an array with text and flags.
3705 *
3706 * The returned array will always contain a boolean 'found', indicating
3707 * whether the parser function was found or not. It may also contain the
3708 * following:
3709 * text: string|object, resulting wikitext or PP DOM object
3710 * isHTML: bool, $text is HTML, armour it against wikitext transformation
3711 * isChildObj: bool, $text is a DOM node needing expansion in a child frame
3712 * isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3713 * nowiki: bool, wiki markup in $text should be escaped
3714 *
3715 * @since 1.21
3716 * @param PPFrame $frame The current frame, contains template arguments
3717 * @param string $function Function name
3718 * @param array $args Arguments to the function
3719 * @throws MWException
3720 * @return array
3721 */
3722 public function callParserFunction( $frame, $function, array $args = array() ) {
3723 global $wgContLang;
3724
3725
3726 # Case sensitive functions
3727 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3728 $function = $this->mFunctionSynonyms[1][$function];
3729 } else {
3730 # Case insensitive functions
3731 $function = $wgContLang->lc( $function );
3732 if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3733 $function = $this->mFunctionSynonyms[0][$function];
3734 } else {
3735 return array( 'found' => false );
3736 }
3737 }
3738
3739 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3740
3741 # Workaround for PHP bug 35229 and similar
3742 if ( !is_callable( $callback ) ) {
3743 throw new MWException( "Tag hook for $function is not callable\n" );
3744 }
3745
3746 $allArgs = array( &$this );
3747 if ( $flags & self::SFH_OBJECT_ARGS ) {
3748 # Convert arguments to PPNodes and collect for appending to $allArgs
3749 $funcArgs = array();
3750 foreach ( $args as $k => $v ) {
3751 if ( $v instanceof PPNode || $k === 0 ) {
3752 $funcArgs[] = $v;
3753 } else {
3754 $funcArgs[] = $this->mPreprocessor->newPartNodeArray( array( $k => $v ) )->item( 0 );
3755 }
3756 }
3757
3758 # Add a frame parameter, and pass the arguments as an array
3759 $allArgs[] = $frame;
3760 $allArgs[] = $funcArgs;
3761 } else {
3762 # Convert arguments to plain text and append to $allArgs
3763 foreach ( $args as $k => $v ) {
3764 if ( $v instanceof PPNode ) {
3765 $allArgs[] = trim( $frame->expand( $v ) );
3766 } elseif ( is_int( $k ) && $k >= 0 ) {
3767 $allArgs[] = trim( $v );
3768 } else {
3769 $allArgs[] = trim( "$k=$v" );
3770 }
3771 }
3772 }
3773
3774 $result = call_user_func_array( $callback, $allArgs );
3775
3776 # The interface for function hooks allows them to return a wikitext
3777 # string or an array containing the string and any flags. This mungs
3778 # things around to match what this method should return.
3779 if ( !is_array( $result ) ) {
3780 $result = array(
3781 'found' => true,
3782 'text' => $result,
3783 );
3784 } else {
3785 if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3786 $result['text'] = $result[0];
3787 }
3788 unset( $result[0] );
3789 $result += array(
3790 'found' => true,
3791 );
3792 }
3793
3794 $noparse = true;
3795 $preprocessFlags = 0;
3796 if ( isset( $result['noparse'] ) ) {
3797 $noparse = $result['noparse'];
3798 }
3799 if ( isset( $result['preprocessFlags'] ) ) {
3800 $preprocessFlags = $result['preprocessFlags'];
3801 }
3802
3803 if ( !$noparse ) {
3804 $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3805 $result['isChildObj'] = true;
3806 }
3807
3808 return $result;
3809 }
3810
3811 /**
3812 * Get the semi-parsed DOM representation of a template with a given title,
3813 * and its redirect destination title. Cached.
3814 *
3815 * @param Title $title
3816 *
3817 * @return array
3818 */
3819 public function getTemplateDom( $title ) {
3820 $cacheTitle = $title;
3821 $titleText = $title->getPrefixedDBkey();
3822
3823 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3824 list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3825 $title = Title::makeTitle( $ns, $dbk );
3826 $titleText = $title->getPrefixedDBkey();
3827 }
3828 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3829 return array( $this->mTplDomCache[$titleText], $title );
3830 }
3831
3832 # Cache miss, go to the database
3833 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3834
3835 if ( $text === false ) {
3836 $this->mTplDomCache[$titleText] = false;
3837 return array( false, $title );
3838 }
3839
3840 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3841 $this->mTplDomCache[$titleText] = $dom;
3842
3843 if ( !$title->equals( $cacheTitle ) ) {
3844 $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3845 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3846 }
3847
3848 return array( $dom, $title );
3849 }
3850
3851 /**
3852 * Fetch the current revision of a given title. Note that the revision
3853 * (and even the title) may not exist in the database, so everything
3854 * contributing to the output of the parser should use this method
3855 * where possible, rather than getting the revisions themselves. This
3856 * method also caches its results, so using it benefits performance.
3857 *
3858 * @since 1.24
3859 * @param Title $title
3860 * @return Revision
3861 */
3862 public function fetchCurrentRevisionOfTitle( $title ) {
3863 $cacheKey = $title->getPrefixedDBkey();
3864 if ( !$this->currentRevisionCache ) {
3865 $this->currentRevisionCache = new MapCacheLRU( 100 );
3866 }
3867 if ( !$this->currentRevisionCache->has( $cacheKey ) ) {
3868 $this->currentRevisionCache->set( $cacheKey,
3869 // Defaults to Parser::statelessFetchRevision()
3870 call_user_func( $this->mOptions->getCurrentRevisionCallback(), $title, $this )
3871 );
3872 }
3873 return $this->currentRevisionCache->get( $cacheKey );
3874 }
3875
3876 /**
3877 * Wrapper around Revision::newFromTitle to allow passing additional parameters
3878 * without passing them on to it.
3879 *
3880 * @since 1.24
3881 * @param Title $title
3882 * @param Parser|bool $parser
3883 * @return Revision
3884 */
3885 public static function statelessFetchRevision( $title, $parser = false ) {
3886 return Revision::newFromTitle( $title );
3887 }
3888
3889 /**
3890 * Fetch the unparsed text of a template and register a reference to it.
3891 * @param Title $title
3892 * @return array ( string or false, Title )
3893 */
3894 public function fetchTemplateAndTitle( $title ) {
3895 // Defaults to Parser::statelessFetchTemplate()
3896 $templateCb = $this->mOptions->getTemplateCallback();
3897 $stuff = call_user_func( $templateCb, $title, $this );
3898 // We use U+007F DELETE to distinguish strip markers from regular text.
3899 $text = $stuff['text'];
3900 if ( is_string( $stuff['text'] ) ) {
3901 $text = strtr( $text, "\x7f", "?" );
3902 }
3903 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3904 if ( isset( $stuff['deps'] ) ) {
3905 foreach ( $stuff['deps'] as $dep ) {
3906 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3907 if ( $dep['title']->equals( $this->getTitle() ) ) {
3908 // If we transclude ourselves, the final result
3909 // will change based on the new version of the page
3910 $this->mOutput->setFlag( 'vary-revision' );
3911 }
3912 }
3913 }
3914 return array( $text, $finalTitle );
3915 }
3916
3917 /**
3918 * Fetch the unparsed text of a template and register a reference to it.
3919 * @param Title $title
3920 * @return string|bool
3921 */
3922 public function fetchTemplate( $title ) {
3923 $rv = $this->fetchTemplateAndTitle( $title );
3924 return $rv[0];
3925 }
3926
3927 /**
3928 * Static function to get a template
3929 * Can be overridden via ParserOptions::setTemplateCallback().
3930 *
3931 * @param Title $title
3932 * @param bool|Parser $parser
3933 *
3934 * @return array
3935 */
3936 public static function statelessFetchTemplate( $title, $parser = false ) {
3937 $text = $skip = false;
3938 $finalTitle = $title;
3939 $deps = array();
3940
3941 # Loop to fetch the article, with up to 1 redirect
3942 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3943 # Give extensions a chance to select the revision instead
3944 $id = false; # Assume current
3945 Hooks::run( 'BeforeParserFetchTemplateAndtitle',
3946 array( $parser, $title, &$skip, &$id ) );
3947
3948 if ( $skip ) {
3949 $text = false;
3950 $deps[] = array(
3951 'title' => $title,
3952 'page_id' => $title->getArticleID(),
3953 'rev_id' => null
3954 );
3955 break;
3956 }
3957 # Get the revision
3958 if ( $id ) {
3959 $rev = Revision::newFromId( $id );
3960 } elseif ( $parser ) {
3961 $rev = $parser->fetchCurrentRevisionOfTitle( $title );
3962 } else {
3963 $rev = Revision::newFromTitle( $title );
3964 }
3965 $rev_id = $rev ? $rev->getId() : 0;
3966 # If there is no current revision, there is no page
3967 if ( $id === false && !$rev ) {
3968 $linkCache = LinkCache::singleton();
3969 $linkCache->addBadLinkObj( $title );
3970 }
3971
3972 $deps[] = array(
3973 'title' => $title,
3974 'page_id' => $title->getArticleID(),
3975 'rev_id' => $rev_id );
3976 if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3977 # We fetched a rev from a different title; register it too...
3978 $deps[] = array(
3979 'title' => $rev->getTitle(),
3980 'page_id' => $rev->getPage(),
3981 'rev_id' => $rev_id );
3982 }
3983
3984 if ( $rev ) {
3985 $content = $rev->getContent();
3986 $text = $content ? $content->getWikitextForTransclusion() : null;
3987
3988 if ( $text === false || $text === null ) {
3989 $text = false;
3990 break;
3991 }
3992 } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3993 global $wgContLang;
3994 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3995 if ( !$message->exists() ) {
3996 $text = false;
3997 break;
3998 }
3999 $content = $message->content();
4000 $text = $message->plain();
4001 } else {
4002 break;
4003 }
4004 if ( !$content ) {
4005 break;
4006 }
4007 # Redirect?
4008 $finalTitle = $title;
4009 $title = $content->getRedirectTarget();
4010 }
4011 return array(
4012 'text' => $text,
4013 'finalTitle' => $finalTitle,
4014 'deps' => $deps );
4015 }
4016
4017 /**
4018 * Fetch a file and its title and register a reference to it.
4019 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
4020 * @param Title $title
4021 * @param array $options Array of options to RepoGroup::findFile
4022 * @return File|bool
4023 */
4024 public function fetchFile( $title, $options = array() ) {
4025 $res = $this->fetchFileAndTitle( $title, $options );
4026 return $res[0];
4027 }
4028
4029 /**
4030 * Fetch a file and its title and register a reference to it.
4031 * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
4032 * @param Title $title
4033 * @param array $options Array of options to RepoGroup::findFile
4034 * @return array ( File or false, Title of file )
4035 */
4036 public function fetchFileAndTitle( $title, $options = array() ) {
4037 $file = $this->fetchFileNoRegister( $title, $options );
4038
4039 $time = $file ? $file->getTimestamp() : false;
4040 $sha1 = $file ? $file->getSha1() : false;
4041 # Register the file as a dependency...
4042 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4043 if ( $file && !$title->equals( $file->getTitle() ) ) {
4044 # Update fetched file title
4045 $title = $file->getTitle();
4046 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
4047 }
4048 return array( $file, $title );
4049 }
4050
4051 /**
4052 * Helper function for fetchFileAndTitle.
4053 *
4054 * Also useful if you need to fetch a file but not use it yet,
4055 * for example to get the file's handler.
4056 *
4057 * @param Title $title
4058 * @param array $options Array of options to RepoGroup::findFile
4059 * @return File|bool
4060 */
4061 protected function fetchFileNoRegister( $title, $options = array() ) {
4062 if ( isset( $options['broken'] ) ) {
4063 $file = false; // broken thumbnail forced by hook
4064 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
4065 $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
4066 } else { // get by (name,timestamp)
4067 $file = wfFindFile( $title, $options );
4068 }
4069 return $file;
4070 }
4071
4072 /**
4073 * Transclude an interwiki link.
4074 *
4075 * @param Title $title
4076 * @param string $action
4077 *
4078 * @return string
4079 */
4080 public function interwikiTransclude( $title, $action ) {
4081 global $wgEnableScaryTranscluding;
4082
4083 if ( !$wgEnableScaryTranscluding ) {
4084 return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
4085 }
4086
4087 $url = $title->getFullURL( array( 'action' => $action ) );
4088
4089 if ( strlen( $url ) > 255 ) {
4090 return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
4091 }
4092 return $this->fetchScaryTemplateMaybeFromCache( $url );
4093 }
4094
4095 /**
4096 * @param string $url
4097 * @return mixed|string
4098 */
4099 public function fetchScaryTemplateMaybeFromCache( $url ) {
4100 global $wgTranscludeCacheExpiry;
4101 $dbr = wfGetDB( DB_SLAVE );
4102 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
4103 $obj = $dbr->selectRow( 'transcache', array( 'tc_time', 'tc_contents' ),
4104 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
4105 if ( $obj ) {
4106 return $obj->tc_contents;
4107 }
4108
4109 $req = MWHttpRequest::factory( $url, array(), __METHOD__ );
4110 $status = $req->execute(); // Status object
4111 if ( $status->isOK() ) {
4112 $text = $req->getContent();
4113 } elseif ( $req->getStatus() != 200 ) {
4114 // Though we failed to fetch the content, this status is useless.
4115 return wfMessage( 'scarytranscludefailed-httpstatus' )
4116 ->params( $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
4117 } else {
4118 return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
4119 }
4120
4121 $dbw = wfGetDB( DB_MASTER );
4122 $dbw->replace( 'transcache', array( 'tc_url' ), array(
4123 'tc_url' => $url,
4124 'tc_time' => $dbw->timestamp( time() ),
4125 'tc_contents' => $text
4126 ) );
4127 return $text;
4128 }
4129
4130 /**
4131 * Triple brace replacement -- used for template arguments
4132 * @private
4133 *
4134 * @param array $piece
4135 * @param PPFrame $frame
4136 *
4137 * @return array
4138 */
4139 public function argSubstitution( $piece, $frame ) {
4140
4141 $error = false;
4142 $parts = $piece['parts'];
4143 $nameWithSpaces = $frame->expand( $piece['title'] );
4144 $argName = trim( $nameWithSpaces );
4145 $object = false;
4146 $text = $frame->getArgument( $argName );
4147 if ( $text === false && $parts->getLength() > 0
4148 && ( $this->ot['html']
4149 || $this->ot['pre']
4150 || ( $this->ot['wiki'] && $frame->isTemplate() )
4151 )
4152 ) {
4153 # No match in frame, use the supplied default
4154 $object = $parts->item( 0 )->getChildren();
4155 }
4156 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
4157 $error = '<!-- WARNING: argument omitted, expansion size too large -->';
4158 $this->limitationWarn( 'post-expand-template-argument' );
4159 }
4160
4161 if ( $text === false && $object === false ) {
4162 # No match anywhere
4163 $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
4164 }
4165 if ( $error !== false ) {
4166 $text .= $error;
4167 }
4168 if ( $object !== false ) {
4169 $ret = array( 'object' => $object );
4170 } else {
4171 $ret = array( 'text' => $text );
4172 }
4173
4174 return $ret;
4175 }
4176
4177 /**
4178 * Return the text to be used for a given extension tag.
4179 * This is the ghost of strip().
4180 *
4181 * @param array $params Associative array of parameters:
4182 * name PPNode for the tag name
4183 * attr PPNode for unparsed text where tag attributes are thought to be
4184 * attributes Optional associative array of parsed attributes
4185 * inner Contents of extension element
4186 * noClose Original text did not have a close tag
4187 * @param PPFrame $frame
4188 *
4189 * @throws MWException
4190 * @return string
4191 */
4192 public function extensionSubstitution( $params, $frame ) {
4193 $name = $frame->expand( $params['name'] );
4194 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
4195 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
4196 $marker = self::MARKER_PREFIX . "-$name-"
4197 . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
4198
4199 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4200 ( $this->ot['html'] || $this->ot['pre'] );
4201 if ( $isFunctionTag ) {
4202 $markerType = 'none';
4203 } else {
4204 $markerType = 'general';
4205 }
4206 if ( $this->ot['html'] || $isFunctionTag ) {
4207 $name = strtolower( $name );
4208 $attributes = Sanitizer::decodeTagAttributes( $attrText );
4209 if ( isset( $params['attributes'] ) ) {
4210 $attributes = $attributes + $params['attributes'];
4211 }
4212
4213 if ( isset( $this->mTagHooks[$name] ) ) {
4214 # Workaround for PHP bug 35229 and similar
4215 if ( !is_callable( $this->mTagHooks[$name] ) ) {
4216 throw new MWException( "Tag hook for $name is not callable\n" );
4217 }
4218 $output = call_user_func_array( $this->mTagHooks[$name],
4219 array( $content, $attributes, $this, $frame ) );
4220 } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4221 list( $callback, ) = $this->mFunctionTagHooks[$name];
4222 if ( !is_callable( $callback ) ) {
4223 throw new MWException( "Tag hook for $name is not callable\n" );
4224 }
4225
4226 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
4227 } else {
4228 $output = '<span class="error">Invalid tag extension name: ' .
4229 htmlspecialchars( $name ) . '</span>';
4230 }
4231
4232 if ( is_array( $output ) ) {
4233 # Extract flags to local scope (to override $markerType)
4234 $flags = $output;
4235 $output = $flags[0];
4236 unset( $flags[0] );
4237 extract( $flags );
4238 }
4239 } else {
4240 if ( is_null( $attrText ) ) {
4241 $attrText = '';
4242 }
4243 if ( isset( $params['attributes'] ) ) {
4244 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4245 $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4246 htmlspecialchars( $attrValue ) . '"';
4247 }
4248 }
4249 if ( $content === null ) {
4250 $output = "<$name$attrText/>";
4251 } else {
4252 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4253 $output = "<$name$attrText>$content$close";
4254 }
4255 }
4256
4257 if ( $markerType === 'none' ) {
4258 return $output;
4259 } elseif ( $markerType === 'nowiki' ) {
4260 $this->mStripState->addNoWiki( $marker, $output );
4261 } elseif ( $markerType === 'general' ) {
4262 $this->mStripState->addGeneral( $marker, $output );
4263 } else {
4264 throw new MWException( __METHOD__ . ': invalid marker type' );
4265 }
4266 return $marker;
4267 }
4268
4269 /**
4270 * Increment an include size counter
4271 *
4272 * @param string $type The type of expansion
4273 * @param int $size The size of the text
4274 * @return bool False if this inclusion would take it over the maximum, true otherwise
4275 */
4276 public function incrementIncludeSize( $type, $size ) {
4277 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4278 return false;
4279 } else {
4280 $this->mIncludeSizes[$type] += $size;
4281 return true;
4282 }
4283 }
4284
4285 /**
4286 * Increment the expensive function count
4287 *
4288 * @return bool False if the limit has been exceeded
4289 */
4290 public function incrementExpensiveFunctionCount() {
4291 $this->mExpensiveFunctionCount++;
4292 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4293 }
4294
4295 /**
4296 * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4297 * Fills $this->mDoubleUnderscores, returns the modified text
4298 *
4299 * @param string $text
4300 *
4301 * @return string
4302 */
4303 public function doDoubleUnderscore( $text ) {
4304
4305 # The position of __TOC__ needs to be recorded
4306 $mw = MagicWord::get( 'toc' );
4307 if ( $mw->match( $text ) ) {
4308 $this->mShowToc = true;
4309 $this->mForceTocPosition = true;
4310
4311 # Set a placeholder. At the end we'll fill it in with the TOC.
4312 $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
4313
4314 # Only keep the first one.
4315 $text = $mw->replace( '', $text );
4316 }
4317
4318 # Now match and remove the rest of them
4319 $mwa = MagicWord::getDoubleUnderscoreArray();
4320 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4321
4322 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4323 $this->mOutput->mNoGallery = true;
4324 }
4325 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4326 $this->mShowToc = false;
4327 }
4328 if ( isset( $this->mDoubleUnderscores['hiddencat'] )
4329 && $this->mTitle->getNamespace() == NS_CATEGORY
4330 ) {
4331 $this->addTrackingCategory( 'hidden-category-category' );
4332 }
4333 # (bug 8068) Allow control over whether robots index a page.
4334 #
4335 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here! This
4336 # is not desirable, the last one on the page should win.
4337 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4338 $this->mOutput->setIndexPolicy( 'noindex' );
4339 $this->addTrackingCategory( 'noindex-category' );
4340 }
4341 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4342 $this->mOutput->setIndexPolicy( 'index' );
4343 $this->addTrackingCategory( 'index-category' );
4344 }
4345
4346 # Cache all double underscores in the database
4347 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4348 $this->mOutput->setProperty( $key, '' );
4349 }
4350
4351 return $text;
4352 }
4353
4354 /**
4355 * @see ParserOutput::addTrackingCategory()
4356 * @param string $msg Message key
4357 * @return bool Whether the addition was successful
4358 */
4359 public function addTrackingCategory( $msg ) {
4360 return $this->mOutput->addTrackingCategory( $msg, $this->mTitle );
4361 }
4362
4363 /**
4364 * This function accomplishes several tasks:
4365 * 1) Auto-number headings if that option is enabled
4366 * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4367 * 3) Add a Table of contents on the top for users who have enabled the option
4368 * 4) Auto-anchor headings
4369 *
4370 * It loops through all headlines, collects the necessary data, then splits up the
4371 * string and re-inserts the newly formatted headlines.
4372 *
4373 * @param string $text
4374 * @param string $origText Original, untouched wikitext
4375 * @param bool $isMain
4376 * @return mixed|string
4377 * @private
4378 */
4379 public function formatHeadings( $text, $origText, $isMain = true ) {
4380 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4381
4382 # Inhibit editsection links if requested in the page
4383 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4384 $maybeShowEditLink = $showEditLink = false;
4385 } else {
4386 $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4387 $showEditLink = $this->mOptions->getEditSection();
4388 }
4389 if ( $showEditLink ) {
4390 $this->mOutput->setEditSectionTokens( true );
4391 }
4392
4393 # Get all headlines for numbering them and adding funky stuff like [edit]
4394 # links - this is for later, but we need the number of headlines right now
4395 $matches = array();
4396 $numMatches = preg_match_all(
4397 '/<H(?P<level>[1-6])(?P<attrib>.*?>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i',
4398 $text,
4399 $matches
4400 );
4401
4402 # if there are fewer than 4 headlines in the article, do not show TOC
4403 # unless it's been explicitly enabled.
4404 $enoughToc = $this->mShowToc &&
4405 ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4406
4407 # Allow user to stipulate that a page should have a "new section"
4408 # link added via __NEWSECTIONLINK__
4409 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4410 $this->mOutput->setNewSection( true );
4411 }
4412
4413 # Allow user to remove the "new section"
4414 # link via __NONEWSECTIONLINK__
4415 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4416 $this->mOutput->hideNewSection( true );
4417 }
4418
4419 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4420 # override above conditions and always show TOC above first header
4421 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4422 $this->mShowToc = true;
4423 $enoughToc = true;
4424 }
4425
4426 # headline counter
4427 $headlineCount = 0;
4428 $numVisible = 0;
4429
4430 # Ugh .. the TOC should have neat indentation levels which can be
4431 # passed to the skin functions. These are determined here
4432 $toc = '';
4433 $full = '';
4434 $head = array();
4435 $sublevelCount = array();
4436 $levelCount = array();
4437 $level = 0;
4438 $prevlevel = 0;
4439 $toclevel = 0;
4440 $prevtoclevel = 0;
4441 $markerRegex = self::MARKER_PREFIX . "-h-(\d+)-" . self::MARKER_SUFFIX;
4442 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4443 $oldType = $this->mOutputType;
4444 $this->setOutputType( self::OT_WIKI );
4445 $frame = $this->getPreprocessor()->newFrame();
4446 $root = $this->preprocessToDom( $origText );
4447 $node = $root->getFirstChild();
4448 $byteOffset = 0;
4449 $tocraw = array();
4450 $refers = array();
4451
4452 foreach ( $matches[3] as $headline ) {
4453 $isTemplate = false;
4454 $titleText = false;
4455 $sectionIndex = false;
4456 $numbering = '';
4457 $markerMatches = array();
4458 if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4459 $serial = $markerMatches[1];
4460 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4461 $isTemplate = ( $titleText != $baseTitleText );
4462 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4463 }
4464
4465 if ( $toclevel ) {
4466 $prevlevel = $level;
4467 }
4468 $level = $matches[1][$headlineCount];
4469
4470 if ( $level > $prevlevel ) {
4471 # Increase TOC level
4472 $toclevel++;
4473 $sublevelCount[$toclevel] = 0;
4474 if ( $toclevel < $wgMaxTocLevel ) {
4475 $prevtoclevel = $toclevel;
4476 $toc .= Linker::tocIndent();
4477 $numVisible++;
4478 }
4479 } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4480 # Decrease TOC level, find level to jump to
4481
4482 for ( $i = $toclevel; $i > 0; $i-- ) {
4483 if ( $levelCount[$i] == $level ) {
4484 # Found last matching level
4485 $toclevel = $i;
4486 break;
4487 } elseif ( $levelCount[$i] < $level ) {
4488 # Found first matching level below current level
4489 $toclevel = $i + 1;
4490 break;
4491 }
4492 }
4493 if ( $i == 0 ) {
4494 $toclevel = 1;
4495 }
4496 if ( $toclevel < $wgMaxTocLevel ) {
4497 if ( $prevtoclevel < $wgMaxTocLevel ) {
4498 # Unindent only if the previous toc level was shown :p
4499 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4500 $prevtoclevel = $toclevel;
4501 } else {
4502 $toc .= Linker::tocLineEnd();
4503 }
4504 }
4505 } else {
4506 # No change in level, end TOC line
4507 if ( $toclevel < $wgMaxTocLevel ) {
4508 $toc .= Linker::tocLineEnd();
4509 }
4510 }
4511
4512 $levelCount[$toclevel] = $level;
4513
4514 # count number of headlines for each level
4515 $sublevelCount[$toclevel]++;
4516 $dot = 0;
4517 for ( $i = 1; $i <= $toclevel; $i++ ) {
4518 if ( !empty( $sublevelCount[$i] ) ) {
4519 if ( $dot ) {
4520 $numbering .= '.';
4521 }
4522 $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4523 $dot = 1;
4524 }
4525 }
4526
4527 # The safe header is a version of the header text safe to use for links
4528
4529 # Remove link placeholders by the link text.
4530 # <!--LINK number-->
4531 # turns into
4532 # link text with suffix
4533 # Do this before unstrip since link text can contain strip markers
4534 $safeHeadline = $this->replaceLinkHoldersText( $headline );
4535
4536 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4537 $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4538
4539 # Strip out HTML (first regex removes any tag not allowed)
4540 # Allowed tags are:
4541 # * <sup> and <sub> (bug 8393)
4542 # * <i> (bug 26375)
4543 # * <b> (r105284)
4544 # * <bdi> (bug 72884)
4545 # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4546 #
4547 # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4548 # to allow setting directionality in toc items.
4549 $tocline = preg_replace(
4550 array(
4551 '#<(?!/?(span|sup|sub|bdi|i|b)(?: [^>]*)?>).*?>#',
4552 '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|bdi|i|b))(?: .*?)?>#'
4553 ),
4554 array( '', '<$1>' ),
4555 $safeHeadline
4556 );
4557 $tocline = trim( $tocline );
4558
4559 # For the anchor, strip out HTML-y stuff period
4560 $safeHeadline = preg_replace( '/<.*?>/', '', $safeHeadline );
4561 $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4562
4563 # Save headline for section edit hint before it's escaped
4564 $headlineHint = $safeHeadline;
4565
4566 if ( $wgExperimentalHtmlIds ) {
4567 # For reverse compatibility, provide an id that's
4568 # HTML4-compatible, like we used to.
4569 #
4570 # It may be worth noting, academically, that it's possible for
4571 # the legacy anchor to conflict with a non-legacy headline
4572 # anchor on the page. In this case likely the "correct" thing
4573 # would be to either drop the legacy anchors or make sure
4574 # they're numbered first. However, this would require people
4575 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4576 # manually, so let's not bother worrying about it.
4577 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4578 array( 'noninitial', 'legacy' ) );
4579 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4580
4581 if ( $legacyHeadline == $safeHeadline ) {
4582 # No reason to have both (in fact, we can't)
4583 $legacyHeadline = false;
4584 }
4585 } else {
4586 $legacyHeadline = false;
4587 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4588 'noninitial' );
4589 }
4590
4591 # HTML names must be case-insensitively unique (bug 10721).
4592 # This does not apply to Unicode characters per
4593 # http://www.w3.org/TR/html5/infrastructure.html#case-sensitivity-and-string-comparison
4594 # @todo FIXME: We may be changing them depending on the current locale.
4595 $arrayKey = strtolower( $safeHeadline );
4596 if ( $legacyHeadline === false ) {
4597 $legacyArrayKey = false;
4598 } else {
4599 $legacyArrayKey = strtolower( $legacyHeadline );
4600 }
4601
4602 # Create the anchor for linking from the TOC to the section
4603 $anchor = $safeHeadline;
4604 $legacyAnchor = $legacyHeadline;
4605 if ( isset( $refers[$arrayKey] ) ) {
4606 for ( $i = 2; isset( $refers["${arrayKey}_$i"] ); ++$i );
4607 $anchor .= "_$i";
4608 $refers["${arrayKey}_$i"] = true;
4609 } else {
4610 $refers[$arrayKey] = true;
4611 }
4612 if ( $legacyHeadline !== false && isset( $refers[$legacyArrayKey] ) ) {
4613 for ( $i = 2; isset( $refers["${legacyArrayKey}_$i"] ); ++$i );
4614 $legacyAnchor .= "_$i";
4615 $refers["${legacyArrayKey}_$i"] = true;
4616 } else {
4617 $refers[$legacyArrayKey] = true;
4618 }
4619
4620 # Don't number the heading if it is the only one (looks silly)
4621 if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4622 # the two are different if the line contains a link
4623 $headline = Html::element(
4624 'span',
4625 array( 'class' => 'mw-headline-number' ),
4626 $numbering
4627 ) . ' ' . $headline;
4628 }
4629
4630 if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4631 $toc .= Linker::tocLine( $anchor, $tocline,
4632 $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4633 }
4634
4635 # Add the section to the section tree
4636 # Find the DOM node for this header
4637 $noOffset = ( $isTemplate || $sectionIndex === false );
4638 while ( $node && !$noOffset ) {
4639 if ( $node->getName() === 'h' ) {
4640 $bits = $node->splitHeading();
4641 if ( $bits['i'] == $sectionIndex ) {
4642 break;
4643 }
4644 }
4645 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4646 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4647 $node = $node->getNextSibling();
4648 }
4649 $tocraw[] = array(
4650 'toclevel' => $toclevel,
4651 'level' => $level,
4652 'line' => $tocline,
4653 'number' => $numbering,
4654 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4655 'fromtitle' => $titleText,
4656 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4657 'anchor' => $anchor,
4658 );
4659
4660 # give headline the correct <h#> tag
4661 if ( $maybeShowEditLink && $sectionIndex !== false ) {
4662 // Output edit section links as markers with styles that can be customized by skins
4663 if ( $isTemplate ) {
4664 # Put a T flag in the section identifier, to indicate to extractSections()
4665 # that sections inside <includeonly> should be counted.
4666 $editsectionPage = $titleText;
4667 $editsectionSection = "T-$sectionIndex";
4668 $editsectionContent = null;
4669 } else {
4670 $editsectionPage = $this->mTitle->getPrefixedText();
4671 $editsectionSection = $sectionIndex;
4672 $editsectionContent = $headlineHint;
4673 }
4674 // We use a bit of pesudo-xml for editsection markers. The
4675 // language converter is run later on. Using a UNIQ style marker
4676 // leads to the converter screwing up the tokens when it
4677 // converts stuff. And trying to insert strip tags fails too. At
4678 // this point all real inputted tags have already been escaped,
4679 // so we don't have to worry about a user trying to input one of
4680 // these markers directly. We use a page and section attribute
4681 // to stop the language converter from converting these
4682 // important bits of data, but put the headline hint inside a
4683 // content block because the language converter is supposed to
4684 // be able to convert that piece of data.
4685 // Gets replaced with html in ParserOutput::getText
4686 $editlink = '<mw:editsection page="' . htmlspecialchars( $editsectionPage );
4687 $editlink .= '" section="' . htmlspecialchars( $editsectionSection ) . '"';
4688 if ( $editsectionContent !== null ) {
4689 $editlink .= '>' . $editsectionContent . '</mw:editsection>';
4690 } else {
4691 $editlink .= '/>';
4692 }
4693 } else {
4694 $editlink = '';
4695 }
4696 $head[$headlineCount] = Linker::makeHeadline( $level,
4697 $matches['attrib'][$headlineCount], $anchor, $headline,
4698 $editlink, $legacyAnchor );
4699
4700 $headlineCount++;
4701 }
4702
4703 $this->setOutputType( $oldType );
4704
4705 # Never ever show TOC if no headers
4706 if ( $numVisible < 1 ) {
4707 $enoughToc = false;
4708 }
4709
4710 if ( $enoughToc ) {
4711 if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4712 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4713 }
4714 $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4715 $this->mOutput->setTOCHTML( $toc );
4716 $toc = self::TOC_START . $toc . self::TOC_END;
4717 $this->mOutput->addModules( 'mediawiki.toc' );
4718 }
4719
4720 if ( $isMain ) {
4721 $this->mOutput->setSections( $tocraw );
4722 }
4723
4724 # split up and insert constructed headlines
4725 $blocks = preg_split( '/<H[1-6].*?>[\s\S]*?<\/H[1-6]>/i', $text );
4726 $i = 0;
4727
4728 // build an array of document sections
4729 $sections = array();
4730 foreach ( $blocks as $block ) {
4731 // $head is zero-based, sections aren't.
4732 if ( empty( $head[$i - 1] ) ) {
4733 $sections[$i] = $block;
4734 } else {
4735 $sections[$i] = $head[$i - 1] . $block;
4736 }
4737
4738 /**
4739 * Send a hook, one per section.
4740 * The idea here is to be able to make section-level DIVs, but to do so in a
4741 * lower-impact, more correct way than r50769
4742 *
4743 * $this : caller
4744 * $section : the section number
4745 * &$sectionContent : ref to the content of the section
4746 * $showEditLinks : boolean describing whether this section has an edit link
4747 */
4748 Hooks::run( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4749
4750 $i++;
4751 }
4752
4753 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4754 // append the TOC at the beginning
4755 // Top anchor now in skin
4756 $sections[0] = $sections[0] . $toc . "\n";
4757 }
4758
4759 $full .= join( '', $sections );
4760
4761 if ( $this->mForceTocPosition ) {
4762 return str_replace( '<!--MWTOC-->', $toc, $full );
4763 } else {
4764 return $full;
4765 }
4766 }
4767
4768 /**
4769 * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4770 * conversion, substituting signatures, {{subst:}} templates, etc.
4771 *
4772 * @param string $text The text to transform
4773 * @param Title $title The Title object for the current article
4774 * @param User $user The User object describing the current user
4775 * @param ParserOptions $options Parsing options
4776 * @param bool $clearState Whether to clear the parser state first
4777 * @return string The altered wiki markup
4778 */
4779 public function preSaveTransform( $text, Title $title, User $user,
4780 ParserOptions $options, $clearState = true
4781 ) {
4782 if ( $clearState ) {
4783 $magicScopeVariable = $this->lock();
4784 }
4785 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4786 $this->setUser( $user );
4787
4788 $pairs = array(
4789 "\r\n" => "\n",
4790 "\r" => "\n",
4791 );
4792 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4793 if ( $options->getPreSaveTransform() ) {
4794 $text = $this->pstPass2( $text, $user );
4795 }
4796 $text = $this->mStripState->unstripBoth( $text );
4797
4798 $this->setUser( null ); #Reset
4799
4800 return $text;
4801 }
4802
4803 /**
4804 * Pre-save transform helper function
4805 *
4806 * @param string $text
4807 * @param User $user
4808 *
4809 * @return string
4810 */
4811 private function pstPass2( $text, $user ) {
4812 global $wgContLang;
4813
4814 # Note: This is the timestamp saved as hardcoded wikitext to
4815 # the database, we use $wgContLang here in order to give
4816 # everyone the same signature and use the default one rather
4817 # than the one selected in each user's preferences.
4818 # (see also bug 12815)
4819 $ts = $this->mOptions->getTimestamp();
4820 $timestamp = MWTimestamp::getLocalInstance( $ts );
4821 $ts = $timestamp->format( 'YmdHis' );
4822 $tzMsg = $timestamp->format( 'T' ); # might vary on DST changeover!
4823
4824 # Allow translation of timezones through wiki. format() can return
4825 # whatever crap the system uses, localised or not, so we cannot
4826 # ship premade translations.
4827 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4828 $msg = wfMessage( $key )->inContentLanguage();
4829 if ( $msg->exists() ) {
4830 $tzMsg = $msg->text();
4831 }
4832
4833 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4834
4835 # Variable replacement
4836 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4837 $text = $this->replaceVariables( $text );
4838
4839 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4840 # which may corrupt this parser instance via its wfMessage()->text() call-
4841
4842 # Signatures
4843 $sigText = $this->getUserSig( $user );
4844 $text = strtr( $text, array(
4845 '~~~~~' => $d,
4846 '~~~~' => "$sigText $d",
4847 '~~~' => $sigText
4848 ) );
4849
4850 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4851 $tc = '[' . Title::legalChars() . ']';
4852 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4853
4854 // [[ns:page (context)|]]
4855 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";
4856 // [[ns:page（context）|]] (double-width brackets, added in r40257)
4857 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";
4858 // [[ns:page (context), context|]] (using either single or double-width comma)
4859 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";
4860 // [[|page]] (reverse pipe trick: add context from page title)
4861 $p2 = "/\[\[\\|($tc+)]]/";
4862
4863 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4864 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4865 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4866 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4867
4868 $t = $this->mTitle->getText();
4869 $m = array();
4870 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4871 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4872 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4873 $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4874 } else {
4875 # if there's no context, don't bother duplicating the title
4876 $text = preg_replace( $p2, '[[\\1]]', $text );
4877 }
4878
4879 # Trim trailing whitespace
4880 $text = rtrim( $text );
4881
4882 return $text;
4883 }
4884
4885 /**
4886 * Fetch the user's signature text, if any, and normalize to
4887 * validated, ready-to-insert wikitext.
4888 * If you have pre-fetched the nickname or the fancySig option, you can
4889 * specify them here to save a database query.
4890 * Do not reuse this parser instance after calling getUserSig(),
4891 * as it may have changed if it's the $wgParser.
4892 *
4893 * @param User $user
4894 * @param string|bool $nickname Nickname to use or false to use user's default nickname
4895 * @param bool|null $fancySig whether the nicknname is the complete signature
4896 * or null to use default value
4897 * @return string
4898 */
4899 public function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4900 global $wgMaxSigChars;
4901
4902 $username = $user->getName();
4903
4904 # If not given, retrieve from the user object.
4905 if ( $nickname === false ) {
4906 $nickname = $user->getOption( 'nickname' );
4907 }
4908
4909 if ( is_null( $fancySig ) ) {
4910 $fancySig = $user->getBoolOption( 'fancysig' );
4911 }
4912
4913 $nickname = $nickname == null ? $username : $nickname;
4914
4915 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4916 $nickname = $username;
4917 wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4918 } elseif ( $fancySig !== false ) {
4919 # Sig. might contain markup; validate this
4920 if ( $this->validateSig( $nickname ) !== false ) {
4921 # Validated; clean up (if needed) and return it
4922 return $this->cleanSig( $nickname, true );
4923 } else {
4924 # Failed to validate; fall back to the default
4925 $nickname = $username;
4926 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4927 }
4928 }
4929
4930 # Make sure nickname doesnt get a sig in a sig
4931 $nickname = self::cleanSigInSig( $nickname );
4932
4933 # If we're still here, make it a link to the user page
4934 $userText = wfEscapeWikiText( $username );
4935 $nickText = wfEscapeWikiText( $nickname );
4936 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4937
4938 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()
4939 ->title( $this->getTitle() )->text();
4940 }
4941
4942 /**
4943 * Check that the user's signature contains no bad XML
4944 *
4945 * @param string $text
4946 * @return string|bool An expanded string, or false if invalid.
4947 */
4948 public function validateSig( $text ) {
4949 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4950 }
4951
4952 /**
4953 * Clean up signature text
4954 *
4955 * 1) Strip 3, 4 or 5 tildes out of signatures @see cleanSigInSig
4956 * 2) Substitute all transclusions
4957 *
4958 * @param string $text
4959 * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4960 * @return string Signature text
4961 */
4962 public function cleanSig( $text, $parsing = false ) {
4963 if ( !$parsing ) {
4964 global $wgTitle;
4965 $magicScopeVariable = $this->lock();
4966 $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4967 }
4968
4969 # Option to disable this feature
4970 if ( !$this->mOptions->getCleanSignatures() ) {
4971 return $text;
4972 }
4973
4974 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4975 # => Move this logic to braceSubstitution()
4976 $substWord = MagicWord::get( 'subst' );
4977 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4978 $substText = '{{' . $substWord->getSynonym( 0 );
4979
4980 $text = preg_replace( $substRegex, $substText, $text );
4981 $text = self::cleanSigInSig( $text );
4982 $dom = $this->preprocessToDom( $text );
4983 $frame = $this->getPreprocessor()->newFrame();
4984 $text = $frame->expand( $dom );
4985
4986 if ( !$parsing ) {
4987 $text = $this->mStripState->unstripBoth( $text );
4988 }
4989
4990 return $text;
4991 }
4992
4993 /**
4994 * Strip 3, 4 or 5 tildes out of signatures.
4995 *
4996 * @param string $text
4997 * @return string Signature text with /~{3,5}/ removed
4998 */
4999 public static function cleanSigInSig( $text ) {
5000 $text = preg_replace( '/~{3,5}/', '', $text );
5001 return $text;
5002 }
5003
5004 /**
5005 * Set up some variables which are usually set up in parse()
5006 * so that an external function can call some class members with confidence
5007 *
5008 * @param Title|null $title
5009 * @param ParserOptions $options
5010 * @param int $outputType
5011 * @param bool $clearState
5012 */
5013 public function startExternalParse( Title $title = null, ParserOptions $options,
5014 $outputType, $clearState = true
5015 ) {
5016 $this->startParse( $title, $options, $outputType, $clearState );
5017 }
5018
5019 /**
5020 * @param Title|null $title
5021 * @param ParserOptions $options
5022 * @param int $outputType
5023 * @param bool $clearState
5024 */
5025 private function startParse( Title $title = null, ParserOptions $options,
5026 $outputType, $clearState = true
5027 ) {
5028 $this->setTitle( $title );
5029 $this->mOptions = $options;
5030 $this->setOutputType( $outputType );
5031 if ( $clearState ) {
5032 $this->clearState();
5033 }
5034 }
5035
5036 /**
5037 * Wrapper for preprocess()
5038 *
5039 * @param string $text The text to preprocess
5040 * @param ParserOptions $options Options
5041 * @param Title|null $title Title object or null to use $wgTitle
5042 * @return string
5043 */
5044 public function transformMsg( $text, $options, $title = null ) {
5045 static $executing = false;
5046
5047 # Guard against infinite recursion
5048 if ( $executing ) {
5049 return $text;
5050 }
5051 $executing = true;
5052
5053 if ( !$title ) {
5054 global $wgTitle;
5055 $title = $wgTitle;
5056 }
5057
5058 $text = $this->preprocess( $text, $title, $options );
5059
5060 $executing = false;
5061 return $text;
5062 }
5063
5064 /**
5065 * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
5066 * The callback should have the following form:
5067 * function myParserHook( $text, $params, $parser, $frame ) { ... }
5068 *
5069 * Transform and return $text. Use $parser for any required context, e.g. use
5070 * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
5071 *
5072 * Hooks may return extended information by returning an array, of which the
5073 * first numbered element (index 0) must be the return string, and all other
5074 * entries are extracted into local variables within an internal function
5075 * in the Parser class.
5076 *
5077 * This interface (introduced r61913) appears to be undocumented, but
5078 * 'markerName' is used by some core tag hooks to override which strip
5079 * array their results are placed in. **Use great caution if attempting
5080 * this interface, as it is not documented and injudicious use could smash
5081 * private variables.**
5082 *
5083 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5084 * @param callable $callback The callback function (and object) to use for the tag
5085 * @throws MWException
5086 * @return callable|null The old value of the mTagHooks array associated with the hook
5087 */
5088 public function setHook( $tag, $callback ) {
5089 $tag = strtolower( $tag );
5090 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5091 throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
5092 }
5093 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
5094 $this->mTagHooks[$tag] = $callback;
5095 if ( !in_array( $tag, $this->mStripList ) ) {
5096 $this->mStripList[] = $tag;
5097 }
5098
5099 return $oldVal;
5100 }
5101
5102 /**
5103 * As setHook(), but letting the contents be parsed.
5104 *
5105 * Transparent tag hooks are like regular XML-style tag hooks, except they
5106 * operate late in the transformation sequence, on HTML instead of wikitext.
5107 *
5108 * This is probably obsoleted by things dealing with parser frames?
5109 * The only extension currently using it is geoserver.
5110 *
5111 * @since 1.10
5112 * @todo better document or deprecate this
5113 *
5114 * @param string $tag The tag to use, e.g. 'hook' for "<hook>"
5115 * @param callable $callback The callback function (and object) to use for the tag
5116 * @throws MWException
5117 * @return callable|null The old value of the mTagHooks array associated with the hook
5118 */
5119 public function setTransparentTagHook( $tag, $callback ) {
5120 $tag = strtolower( $tag );
5121 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5122 throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
5123 }
5124 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
5125 $this->mTransparentTagHooks[$tag] = $callback;
5126
5127 return $oldVal;
5128 }
5129
5130 /**
5131 * Remove all tag hooks
5132 */
5133 public function clearTagHooks() {
5134 $this->mTagHooks = array();
5135 $this->mFunctionTagHooks = array();
5136 $this->mStripList = $this->mDefaultStripList;
5137 }
5138
5139 /**
5140 * Create a function, e.g. {{sum:1|2|3}}
5141 * The callback function should have the form:
5142 * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
5143 *
5144 * Or with Parser::SFH_OBJECT_ARGS:
5145 * function myParserFunction( $parser, $frame, $args ) { ... }
5146 *
5147 * The callback may either return the text result of the function, or an array with the text
5148 * in element 0, and a number of flags in the other elements. The names of the flags are
5149 * specified in the keys. Valid flags are:
5150 * found The text returned is valid, stop processing the template. This
5151 * is on by default.
5152 * nowiki Wiki markup in the return value should be escaped
5153 * isHTML The returned text is HTML, armour it against wikitext transformation
5154 *
5155 * @param string $id The magic word ID
5156 * @param callable $callback The callback function (and object) to use
5157 * @param int $flags A combination of the following flags:
5158 * Parser::SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
5159 *
5160 * Parser::SFH_OBJECT_ARGS Pass the template arguments as PPNode objects instead of text.
5161 * This allows for conditional expansion of the parse tree, allowing you to eliminate dead
5162 * branches and thus speed up parsing. It is also possible to analyse the parse tree of
5163 * the arguments, and to control the way they are expanded.
5164 *
5165 * The $frame parameter is a PPFrame. This can be used to produce expanded text from the
5166 * arguments, for instance:
5167 * $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
5168 *
5169 * For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
5170 * future versions. Please call $frame->expand() on it anyway so that your code keeps
5171 * working if/when this is changed.
5172 *
5173 * If you want whitespace to be trimmed from $args, you need to do it yourself, post-
5174 * expansion.
5175 *
5176 * Please read the documentation in includes/parser/Preprocessor.php for more information
5177 * about the methods available in PPFrame and PPNode.
5178 *
5179 * @throws MWException
5180 * @return string|callable The old callback function for this name, if any
5181 */
5182 public function setFunctionHook( $id, $callback, $flags = 0 ) {
5183 global $wgContLang;
5184
5185 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
5186 $this->mFunctionHooks[$id] = array( $callback, $flags );
5187
5188 # Add to function cache
5189 $mw = MagicWord::get( $id );
5190 if ( !$mw ) {
5191 throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
5192 }
5193
5194 $synonyms = $mw->getSynonyms();
5195 $sensitive = intval( $mw->isCaseSensitive() );
5196
5197 foreach ( $synonyms as $syn ) {
5198 # Case
5199 if ( !$sensitive ) {
5200 $syn = $wgContLang->lc( $syn );
5201 }
5202 # Add leading hash
5203 if ( !( $flags & self::SFH_NO_HASH ) ) {
5204 $syn = '#' . $syn;
5205 }
5206 # Remove trailing colon
5207 if ( substr( $syn, -1, 1 ) === ':' ) {
5208 $syn = substr( $syn, 0, -1 );
5209 }
5210 $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5211 }
5212 return $oldVal;
5213 }
5214
5215 /**
5216 * Get all registered function hook identifiers
5217 *
5218 * @return array
5219 */
5220 public function getFunctionHooks() {
5221 return array_keys( $this->mFunctionHooks );
5222 }
5223
5224 /**
5225 * Create a tag function, e.g. "<test>some stuff</test>".
5226 * Unlike tag hooks, tag functions are parsed at preprocessor level.
5227 * Unlike parser functions, their content is not preprocessed.
5228 * @param string $tag
5229 * @param callable $callback
5230 * @param int $flags
5231 * @throws MWException
5232 * @return null
5233 */
5234 public function setFunctionTagHook( $tag, $callback, $flags ) {
5235 $tag = strtolower( $tag );
5236 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5237 throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5238 }
5239 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
5240 $this->mFunctionTagHooks[$tag] : null;
5241 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
5242
5243 if ( !in_array( $tag, $this->mStripList ) ) {
5244 $this->mStripList[] = $tag;
5245 }
5246
5247 return $old;
5248 }
5249
5250 /**
5251 * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
5252 * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5253 * Placeholders created in Skin::makeLinkObj()
5254 *
5255 * @param string $text
5256 * @param int $options
5257 */
5258 public function replaceLinkHolders( &$text, $options = 0 ) {
5259 $this->mLinkHolders->replace( $text );
5260 }
5261
5262 /**
5263 * Replace "<!--LINK-->" link placeholders with plain text of links
5264 * (not HTML-formatted).
5265 *
5266 * @param string $text
5267 * @return string
5268 */
5269 public function replaceLinkHoldersText( $text ) {
5270 return $this->mLinkHolders->replaceText( $text );
5271 }
5272
5273 /**
5274 * Renders an image gallery from a text with one line per image.
5275 * text labels may be given by using |-style alternative text. E.g.
5276 * Image:one.jpg|The number "1"
5277 * Image:tree.jpg|A tree
5278 * given as text will return the HTML of a gallery with two images,
5279 * labeled 'The number "1"' and
5280 * 'A tree'.
5281 *
5282 * @param string $text
5283 * @param array $params
5284 * @return string HTML
5285 */
5286 public function renderImageGallery( $text, $params ) {
5287
5288 $mode = false;
5289 if ( isset( $params['mode'] ) ) {
5290 $mode = $params['mode'];
5291 }
5292
5293 try {
5294 $ig = ImageGalleryBase::factory( $mode );
5295 } catch ( Exception $e ) {
5296 // If invalid type set, fallback to default.
5297 $ig = ImageGalleryBase::factory( false );
5298 }
5299
5300 $ig->setContextTitle( $this->mTitle );
5301 $ig->setShowBytes( false );
5302 $ig->setShowFilename( false );
5303 $ig->setParser( $this );
5304 $ig->setHideBadImages();
5305 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
5306
5307 if ( isset( $params['showfilename'] ) ) {
5308 $ig->setShowFilename( true );
5309 } else {
5310 $ig->setShowFilename( false );
5311 }
5312 if ( isset( $params['caption'] ) ) {
5313 $caption = $params['caption'];
5314 $caption = htmlspecialchars( $caption );
5315 $caption = $this->replaceInternalLinks( $caption );
5316 $ig->setCaptionHtml( $caption );
5317 }
5318 if ( isset( $params['perrow'] ) ) {
5319 $ig->setPerRow( $params['perrow'] );
5320 }
5321 if ( isset( $params['widths'] ) ) {
5322 $ig->setWidths( $params['widths'] );
5323 }
5324 if ( isset( $params['heights'] ) ) {
5325 $ig->setHeights( $params['heights'] );
5326 }
5327 $ig->setAdditionalOptions( $params );
5328
5329 Hooks::run( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
5330
5331 $lines = StringUtils::explode( "\n", $text );
5332 foreach ( $lines as $line ) {
5333 # match lines like these:
5334 # Image:someimage.jpg|This is some image
5335 $matches = array();
5336 preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5337 # Skip empty lines
5338 if ( count( $matches ) == 0 ) {
5339 continue;
5340 }
5341
5342 if ( strpos( $matches[0], '%' ) !== false ) {
5343 $matches[1] = rawurldecode( $matches[1] );
5344 }
5345 $title = Title::newFromText( $matches[1], NS_FILE );
5346 if ( is_null( $title ) ) {
5347 # Bogus title. Ignore these so we don't bomb out later.
5348 continue;
5349 }
5350
5351 # We need to get what handler the file uses, to figure out parameters.
5352 # Note, a hook can overide the file name, and chose an entirely different
5353 # file (which potentially could be of a different type and have different handler).
5354 $options = array();
5355 $descQuery = false;
5356 Hooks::run( 'BeforeParserFetchFileAndTitle',
5357 array( $this, $title, &$options, &$descQuery ) );
5358 # Don't register it now, as ImageGallery does that later.
5359 $file = $this->fetchFileNoRegister( $title, $options );
5360 $handler = $file ? $file->getHandler() : false;
5361
5362 $paramMap = array(
5363 'img_alt' => 'gallery-internal-alt',
5364 'img_link' => 'gallery-internal-link',
5365 );
5366 if ( $handler ) {
5367 $paramMap = $paramMap + $handler->getParamMap();
5368 // We don't want people to specify per-image widths.
5369 // Additionally the width parameter would need special casing anyhow.
5370 unset( $paramMap['img_width'] );
5371 }
5372
5373 $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5374
5375 $label = '';
5376 $alt = '';
5377 $link = '';
5378 $handlerOptions = array();
5379 if ( isset( $matches[3] ) ) {
5380 // look for an |alt= definition while trying not to break existing
5381 // captions with multiple pipes (|) in it, until a more sensible grammar
5382 // is defined for images in galleries
5383
5384 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5385 // splitting on '|' is a bit odd, and different from makeImage.
5386 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5387 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5388
5389 foreach ( $parameterMatches as $parameterMatch ) {
5390 list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5391 if ( $magicName ) {
5392 $paramName = $paramMap[$magicName];
5393
5394 switch ( $paramName ) {
5395 case 'gallery-internal-alt':
5396 $alt = $this->stripAltText( $match, false );
5397 break;
5398 case 'gallery-internal-link':
5399 $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5400 $chars = self::EXT_LINK_URL_CLASS;
5401 $prots = $this->mUrlProtocols;
5402 //check to see if link matches an absolute url, if not then it must be a wiki link.
5403 if ( preg_match( "/^($prots)$chars+$/u", $linkValue ) ) {
5404 $link = $linkValue;
5405 } else {
5406 $localLinkTitle = Title::newFromText( $linkValue );
5407 if ( $localLinkTitle !== null ) {
5408 $link = $localLinkTitle->getLinkURL();
5409 }
5410 }
5411 break;
5412 default:
5413 // Must be a handler specific parameter.
5414 if ( $handler->validateParam( $paramName, $match ) ) {
5415 $handlerOptions[$paramName] = $match;
5416 } else {
5417 // Guess not. Append it to the caption.
5418 wfDebug( "$parameterMatch failed parameter validation\n" );
5419 $label .= '|' . $parameterMatch;
5420 }
5421 }
5422
5423 } else {
5424 // concatenate all other pipes
5425 $label .= '|' . $parameterMatch;
5426 }
5427 }
5428 // remove the first pipe
5429 $label = substr( $label, 1 );
5430 }
5431
5432 $ig->add( $title, $label, $alt, $link, $handlerOptions );
5433 }
5434 $html = $ig->toHTML();
5435 Hooks::run( 'AfterParserFetchFileAndTitle', array( $this, $ig, &$html ) );
5436 return $html;
5437 }
5438
5439 /**
5440 * @param MediaHandler $handler
5441 * @return array
5442 */
5443 public function getImageParams( $handler ) {
5444 if ( $handler ) {
5445 $handlerClass = get_class( $handler );
5446 } else {
5447 $handlerClass = '';
5448 }
5449 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5450 # Initialise static lists
5451 static $internalParamNames = array(
5452 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
5453 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5454 'bottom', 'text-bottom' ),
5455 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
5456 'upright', 'border', 'link', 'alt', 'class' ),
5457 );
5458 static $internalParamMap;
5459 if ( !$internalParamMap ) {
5460 $internalParamMap = array();
5461 foreach ( $internalParamNames as $type => $names ) {
5462 foreach ( $names as $name ) {
5463 $magicName = str_replace( '-', '_', "img_$name" );
5464 $internalParamMap[$magicName] = array( $type, $name );
5465 }
5466 }
5467 }
5468
5469 # Add handler params
5470 $paramMap = $internalParamMap;
5471 if ( $handler ) {
5472 $handlerParamMap = $handler->getParamMap();
5473 foreach ( $handlerParamMap as $magic => $paramName ) {
5474 $paramMap[$magic] = array( 'handler', $paramName );
5475 }
5476 }
5477 $this->mImageParams[$handlerClass] = $paramMap;
5478 $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5479 }
5480 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
5481 }
5482
5483 /**
5484 * Parse image options text and use it to make an image
5485 *
5486 * @param Title $title
5487 * @param string $options
5488 * @param LinkHolderArray|bool $holders
5489 * @return string HTML
5490 */
5491 public function makeImage( $title, $options, $holders = false ) {
5492 # Check if the options text is of the form "options|alt text"
5493 # Options are:
5494 # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
5495 # * left no resizing, just left align. label is used for alt= only
5496 # * right same, but right aligned
5497 # * none same, but not aligned
5498 # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
5499 # * center center the image
5500 # * frame Keep original image size, no magnify-button.
5501 # * framed Same as "frame"
5502 # * frameless like 'thumb' but without a frame. Keeps user preferences for width
5503 # * upright reduce width for upright images, rounded to full __0 px
5504 # * border draw a 1px border around the image
5505 # * alt Text for HTML alt attribute (defaults to empty)
5506 # * class Set a class for img node
5507 # * link Set the target of the image link. Can be external, interwiki, or local
5508 # vertical-align values (no % or length right now):
5509 # * baseline
5510 # * sub
5511 # * super
5512 # * top
5513 # * text-top
5514 # * middle
5515 # * bottom
5516 # * text-bottom
5517
5518 $parts = StringUtils::explode( "|", $options );
5519
5520 # Give extensions a chance to select the file revision for us
5521 $options = array();
5522 $descQuery = false;
5523 Hooks::run( 'BeforeParserFetchFileAndTitle',
5524 array( $this, $title, &$options, &$descQuery ) );
5525 # Fetch and register the file (file title may be different via hooks)
5526 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5527
5528 # Get parameter map
5529 $handler = $file ? $file->getHandler() : false;
5530
5531 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5532
5533 if ( !$file ) {
5534 $this->addTrackingCategory( 'broken-file-category' );
5535 }
5536
5537 # Process the input parameters
5538 $caption = '';
5539 $params = array( 'frame' => array(), 'handler' => array(),
5540 'horizAlign' => array(), 'vertAlign' => array() );
5541 $seenformat = false;
5542 foreach ( $parts as $part ) {
5543 $part = trim( $part );
5544 list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5545 $validated = false;
5546 if ( isset( $paramMap[$magicName] ) ) {
5547 list( $type, $paramName ) = $paramMap[$magicName];
5548
5549 # Special case; width and height come in one variable together
5550 if ( $type === 'handler' && $paramName === 'width' ) {
5551 $parsedWidthParam = $this->parseWidthParam( $value );
5552 if ( isset( $parsedWidthParam['width'] ) ) {
5553 $width = $parsedWidthParam['width'];
5554 if ( $handler->validateParam( 'width', $width ) ) {
5555 $params[$type]['width'] = $width;
5556 $validated = true;
5557 }
5558 }
5559 if ( isset( $parsedWidthParam['height'] ) ) {
5560 $height = $parsedWidthParam['height'];
5561 if ( $handler->validateParam( 'height', $height ) ) {
5562 $params[$type]['height'] = $height;
5563 $validated = true;
5564 }
5565 }
5566 # else no validation -- bug 13436
5567 } else {
5568 if ( $type === 'handler' ) {
5569 # Validate handler parameter
5570 $validated = $handler->validateParam( $paramName, $value );
5571 } else {
5572 # Validate internal parameters
5573 switch ( $paramName ) {
5574 case 'manualthumb':
5575 case 'alt':
5576 case 'class':
5577 # @todo FIXME: Possibly check validity here for
5578 # manualthumb? downstream behavior seems odd with
5579 # missing manual thumbs.
5580 $validated = true;
5581 $value = $this->stripAltText( $value, $holders );
5582 break;
5583 case 'link':
5584 $chars = self::EXT_LINK_URL_CLASS;
5585 $prots = $this->mUrlProtocols;
5586 if ( $value === '' ) {
5587 $paramName = 'no-link';
5588 $value = true;
5589 $validated = true;
5590 } elseif ( preg_match( "/^((?i)$prots)/", $value ) ) {
5591 if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
5592 $paramName = 'link-url';
5593 $this->mOutput->addExternalLink( $value );
5594 if ( $this->mOptions->getExternalLinkTarget() ) {
5595 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5596 }
5597 $validated = true;
5598 }
5599 } else {
5600 $linkTitle = Title::newFromText( $value );
5601 if ( $linkTitle ) {
5602 $paramName = 'link-title';
5603 $value = $linkTitle;
5604 $this->mOutput->addLink( $linkTitle );
5605 $validated = true;
5606 }
5607 }
5608 break;
5609 case 'frameless':
5610 case 'framed':
5611 case 'thumbnail':
5612 // use first appearing option, discard others.
5613 $validated = ! $seenformat;
5614 $seenformat = true;
5615 break;
5616 default:
5617 # Most other things appear to be empty or numeric...
5618 $validated = ( $value === false || is_numeric( trim( $value ) ) );
5619 }
5620 }
5621
5622 if ( $validated ) {
5623 $params[$type][$paramName] = $value;
5624 }
5625 }
5626 }
5627 if ( !$validated ) {
5628 $caption = $part;
5629 }
5630 }
5631
5632 # Process alignment parameters
5633 if ( $params['horizAlign'] ) {
5634 $params['frame']['align'] = key( $params['horizAlign'] );
5635 }
5636 if ( $params['vertAlign'] ) {
5637 $params['frame']['valign'] = key( $params['vertAlign'] );
5638 }
5639
5640 $params['frame']['caption'] = $caption;
5641
5642 # Will the image be presented in a frame, with the caption below?
5643 $imageIsFramed = isset( $params['frame']['frame'] )
5644 || isset( $params['frame']['framed'] )
5645 || isset( $params['frame']['thumbnail'] )
5646 || isset( $params['frame']['manualthumb'] );
5647
5648 # In the old days, [[Image:Foo|text...]] would set alt text. Later it
5649 # came to also set the caption, ordinary text after the image -- which
5650 # makes no sense, because that just repeats the text multiple times in
5651 # screen readers. It *also* came to set the title attribute.
5652 #
5653 # Now that we have an alt attribute, we should not set the alt text to
5654 # equal the caption: that's worse than useless, it just repeats the
5655 # text. This is the framed/thumbnail case. If there's no caption, we
5656 # use the unnamed parameter for alt text as well, just for the time be-
5657 # ing, if the unnamed param is set and the alt param is not.
5658 #
5659 # For the future, we need to figure out if we want to tweak this more,
5660 # e.g., introducing a title= parameter for the title; ignoring the un-
5661 # named parameter entirely for images without a caption; adding an ex-
5662 # plicit caption= parameter and preserving the old magic unnamed para-
5663 # meter for BC; ...
5664 if ( $imageIsFramed ) { # Framed image
5665 if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5666 # No caption or alt text, add the filename as the alt text so
5667 # that screen readers at least get some description of the image
5668 $params['frame']['alt'] = $title->getText();
5669 }
5670 # Do not set $params['frame']['title'] because tooltips don't make sense
5671 # for framed images
5672 } else { # Inline image
5673 if ( !isset( $params['frame']['alt'] ) ) {
5674 # No alt text, use the "caption" for the alt text
5675 if ( $caption !== '' ) {
5676 $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5677 } else {
5678 # No caption, fall back to using the filename for the
5679 # alt text
5680 $params['frame']['alt'] = $title->getText();
5681 }
5682 }
5683 # Use the "caption" for the tooltip text
5684 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5685 }
5686
5687 Hooks::run( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5688
5689 # Linker does the rest
5690 $time = isset( $options['time'] ) ? $options['time'] : false;
5691 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5692 $time, $descQuery, $this->mOptions->getThumbSize() );
5693
5694 # Give the handler a chance to modify the parser object
5695 if ( $handler ) {
5696 $handler->parserTransformHook( $this, $file );
5697 }
5698
5699 return $ret;
5700 }
5701
5702 /**
5703 * @param string $caption
5704 * @param LinkHolderArray|bool $holders
5705 * @return mixed|string
5706 */
5707 protected function stripAltText( $caption, $holders ) {
5708 # Strip bad stuff out of the title (tooltip). We can't just use
5709 # replaceLinkHoldersText() here, because if this function is called
5710 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5711 if ( $holders ) {
5712 $tooltip = $holders->replaceText( $caption );
5713 } else {
5714 $tooltip = $this->replaceLinkHoldersText( $caption );
5715 }
5716
5717 # make sure there are no placeholders in thumbnail attributes
5718 # that are later expanded to html- so expand them now and
5719 # remove the tags
5720 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5721 $tooltip = Sanitizer::stripAllTags( $tooltip );
5722
5723 return $tooltip;
5724 }
5725
5726 /**
5727 * Set a flag in the output object indicating that the content is dynamic and
5728 * shouldn't be cached.
5729 */
5730 public function disableCache() {
5731 wfDebug( "Parser output marked as uncacheable.\n" );
5732 if ( !$this->mOutput ) {
5733 throw new MWException( __METHOD__ .
5734 " can only be called when actually parsing something" );
5735 }
5736 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5737 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5738 }
5739
5740 /**
5741 * Callback from the Sanitizer for expanding items found in HTML attribute
5742 * values, so they can be safely tested and escaped.
5743 *
5744 * @param string $text
5745 * @param bool|PPFrame $frame
5746 * @return string
5747 */
5748 public function attributeStripCallback( &$text, $frame = false ) {
5749 $text = $this->replaceVariables( $text, $frame );
5750 $text = $this->mStripState->unstripBoth( $text );
5751 return $text;
5752 }
5753
5754 /**
5755 * Accessor
5756 *
5757 * @return array
5758 */
5759 public function getTags() {
5760 return array_merge(
5761 array_keys( $this->mTransparentTagHooks ),
5762 array_keys( $this->mTagHooks ),
5763 array_keys( $this->mFunctionTagHooks )
5764 );
5765 }
5766
5767 /**
5768 * Replace transparent tags in $text with the values given by the callbacks.
5769 *
5770 * Transparent tag hooks are like regular XML-style tag hooks, except they
5771 * operate late in the transformation sequence, on HTML instead of wikitext.
5772 *
5773 * @param string $text
5774 *
5775 * @return string
5776 */
5777 public function replaceTransparentTags( $text ) {
5778 $matches = array();
5779 $elements = array_keys( $this->mTransparentTagHooks );
5780 $text = self::extractTagsAndParams( $elements, $text, $matches );
5781 $replacements = array();
5782
5783 foreach ( $matches as $marker => $data ) {
5784 list( $element, $content, $params, $tag ) = $data;
5785 $tagName = strtolower( $element );
5786 if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5787 $output = call_user_func_array(
5788 $this->mTransparentTagHooks[$tagName],
5789 array( $content, $params, $this )
5790 );
5791 } else {
5792 $output = $tag;
5793 }
5794 $replacements[$marker] = $output;
5795 }
5796 return strtr( $text, $replacements );
5797 }
5798
5799 /**
5800 * Break wikitext input into sections, and either pull or replace
5801 * some particular section's text.
5802 *
5803 * External callers should use the getSection and replaceSection methods.
5804 *
5805 * @param string $text Page wikitext
5806 * @param string|number $sectionId A section identifier string of the form:
5807 * "<flag1> - <flag2> - ... - <section number>"
5808 *
5809 * Currently the only recognised flag is "T", which means the target section number
5810 * was derived during a template inclusion parse, in other words this is a template
5811 * section edit link. If no flags are given, it was an ordinary section edit link.
5812 * This flag is required to avoid a section numbering mismatch when a section is
5813 * enclosed by "<includeonly>" (bug 6563).
5814 *
5815 * The section number 0 pulls the text before the first heading; other numbers will
5816 * pull the given section along with its lower-level subsections. If the section is
5817 * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5818 *
5819 * Section 0 is always considered to exist, even if it only contains the empty
5820 * string. If $text is the empty string and section 0 is replaced, $newText is
5821 * returned.
5822 *
5823 * @param string $mode One of "get" or "replace"
5824 * @param string $newText Replacement text for section data.
5825 * @return string For "get", the extracted section text.
5826 * for "replace", the whole page with the section replaced.
5827 */
5828 private function extractSections( $text, $sectionId, $mode, $newText = '' ) {
5829 global $wgTitle; # not generally used but removes an ugly failure mode
5830
5831 $magicScopeVariable = $this->lock();
5832 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5833 $outText = '';
5834 $frame = $this->getPreprocessor()->newFrame();
5835
5836 # Process section extraction flags
5837 $flags = 0;
5838 $sectionParts = explode( '-', $sectionId );
5839 $sectionIndex = array_pop( $sectionParts );
5840 foreach ( $sectionParts as $part ) {
5841 if ( $part === 'T' ) {
5842 $flags |= self::PTD_FOR_INCLUSION;
5843 }
5844 }
5845
5846 # Check for empty input
5847 if ( strval( $text ) === '' ) {
5848 # Only sections 0 and T-0 exist in an empty document
5849 if ( $sectionIndex == 0 ) {
5850 if ( $mode === 'get' ) {
5851 return '';
5852 } else {
5853 return $newText;
5854 }
5855 } else {
5856 if ( $mode === 'get' ) {
5857 return $newText;
5858 } else {
5859 return $text;
5860 }
5861 }
5862 }
5863
5864 # Preprocess the text
5865 $root = $this->preprocessToDom( $text, $flags );
5866
5867 # <h> nodes indicate section breaks
5868 # They can only occur at the top level, so we can find them by iterating the root's children
5869 $node = $root->getFirstChild();
5870
5871 # Find the target section
5872 if ( $sectionIndex == 0 ) {
5873 # Section zero doesn't nest, level=big
5874 $targetLevel = 1000;
5875 } else {
5876 while ( $node ) {
5877 if ( $node->getName() === 'h' ) {
5878 $bits = $node->splitHeading();
5879 if ( $bits['i'] == $sectionIndex ) {
5880 $targetLevel = $bits['level'];
5881 break;
5882 }
5883 }
5884 if ( $mode === 'replace' ) {
5885 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5886 }
5887 $node = $node->getNextSibling();
5888 }
5889 }
5890
5891 if ( !$node ) {
5892 # Not found
5893 if ( $mode === 'get' ) {
5894 return $newText;
5895 } else {
5896 return $text;
5897 }
5898 }
5899
5900 # Find the end of the section, including nested sections
5901 do {
5902 if ( $node->getName() === 'h' ) {
5903 $bits = $node->splitHeading();
5904 $curLevel = $bits['level'];
5905 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5906 break;
5907 }
5908 }
5909 if ( $mode === 'get' ) {
5910 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5911 }
5912 $node = $node->getNextSibling();
5913 } while ( $node );
5914
5915 # Write out the remainder (in replace mode only)
5916 if ( $mode === 'replace' ) {
5917 # Output the replacement text
5918 # Add two newlines on -- trailing whitespace in $newText is conventionally
5919 # stripped by the editor, so we need both newlines to restore the paragraph gap
5920 # Only add trailing whitespace if there is newText
5921 if ( $newText != "" ) {
5922 $outText .= $newText . "\n\n";
5923 }
5924
5925 while ( $node ) {
5926 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5927 $node = $node->getNextSibling();
5928 }
5929 }
5930
5931 if ( is_string( $outText ) ) {
5932 # Re-insert stripped tags
5933 $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5934 }
5935
5936 return $outText;
5937 }
5938
5939 /**
5940 * This function returns the text of a section, specified by a number ($section).
5941 * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5942 * the first section before any such heading (section 0).
5943 *
5944 * If a section contains subsections, these are also returned.
5945 *
5946 * @param string $text Text to look in
5947 * @param string|number $sectionId Section identifier as a number or string
5948 * (e.g. 0, 1 or 'T-1').
5949 * @param string $defaultText Default to return if section is not found
5950 *
5951 * @return string Text of the requested section
5952 */
5953 public function getSection( $text, $sectionId, $defaultText = '' ) {
5954 return $this->extractSections( $text, $sectionId, 'get', $defaultText );
5955 }
5956
5957 /**
5958 * This function returns $oldtext after the content of the section
5959 * specified by $section has been replaced with $text. If the target
5960 * section does not exist, $oldtext is returned unchanged.
5961 *
5962 * @param string $oldText Former text of the article
5963 * @param string|number $sectionId Section identifier as a number or string
5964 * (e.g. 0, 1 or 'T-1').
5965 * @param string $newText Replacing text
5966 *
5967 * @return string Modified text
5968 */
5969 public function replaceSection( $oldText, $sectionId, $newText ) {
5970 return $this->extractSections( $oldText, $sectionId, 'replace', $newText );
5971 }
5972
5973 /**
5974 * Get the ID of the revision we are parsing
5975 *
5976 * @return int|null
5977 */
5978 public function getRevisionId() {
5979 return $this->mRevisionId;
5980 }
5981
5982 /**
5983 * Get the revision object for $this->mRevisionId
5984 *
5985 * @return Revision|null Either a Revision object or null
5986 * @since 1.23 (public since 1.23)
5987 */
5988 public function getRevisionObject() {
5989 if ( !is_null( $this->mRevisionObject ) ) {
5990 return $this->mRevisionObject;
5991 }
5992 if ( is_null( $this->mRevisionId ) ) {
5993 return null;
5994 }
5995
5996 $rev = call_user_func(
5997 $this->mOptions->getCurrentRevisionCallback(), $this->getTitle(), $this
5998 );
5999
6000 # If the parse is for a new revision, then the callback should have
6001 # already been set to force the object and should match mRevisionId.
6002 # If not, try to fetch by mRevisionId for sanity.
6003 if ( $rev && $rev->getId() != $this->mRevisionId ) {
6004 $rev = Revision::newFromId( $this->mRevisionId );
6005 }
6006
6007 $this->mRevisionObject = $rev;
6008
6009 return $this->mRevisionObject;
6010 }
6011
6012 /**
6013 * Get the timestamp associated with the current revision, adjusted for
6014 * the default server-local timestamp
6015 * @return string
6016 */
6017 public function getRevisionTimestamp() {
6018 if ( is_null( $this->mRevisionTimestamp ) ) {
6019 global $wgContLang;
6020
6021 $revObject = $this->getRevisionObject();
6022 $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
6023
6024 # The cryptic '' timezone parameter tells to use the site-default
6025 # timezone offset instead of the user settings.
6026 #
6027 # Since this value will be saved into the parser cache, served
6028 # to other users, and potentially even used inside links and such,
6029 # it needs to be consistent for all visitors.
6030 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
6031
6032 }
6033 return $this->mRevisionTimestamp;
6034 }
6035
6036 /**
6037 * Get the name of the user that edited the last revision
6038 *
6039 * @return string User name
6040 */
6041 public function getRevisionUser() {
6042 if ( is_null( $this->mRevisionUser ) ) {
6043 $revObject = $this->getRevisionObject();
6044
6045 # if this template is subst: the revision id will be blank,
6046 # so just use the current user's name
6047 if ( $revObject ) {
6048 $this->mRevisionUser = $revObject->getUserText();
6049 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6050 $this->mRevisionUser = $this->getUser()->getName();
6051 }
6052 }
6053 return $this->mRevisionUser;
6054 }
6055
6056 /**
6057 * Get the size of the revision
6058 *
6059 * @return int|null Revision size
6060 */
6061 public function getRevisionSize() {
6062 if ( is_null( $this->mRevisionSize ) ) {
6063 $revObject = $this->getRevisionObject();
6064
6065 # if this variable is subst: the revision id will be blank,
6066 # so just use the parser input size, because the own substituation
6067 # will change the size.
6068 if ( $revObject ) {
6069 $this->mRevisionSize = $revObject->getSize();
6070 } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
6071 $this->mRevisionSize = $this->mInputSize;
6072 }
6073 }
6074 return $this->mRevisionSize;
6075 }
6076
6077 /**
6078 * Mutator for $mDefaultSort
6079 *
6080 * @param string $sort New value
6081 */
6082 public function setDefaultSort( $sort ) {
6083 $this->mDefaultSort = $sort;
6084 $this->mOutput->setProperty( 'defaultsort', $sort );
6085 }
6086
6087 /**
6088 * Accessor for $mDefaultSort
6089 * Will use the empty string if none is set.
6090 *
6091 * This value is treated as a prefix, so the
6092 * empty string is equivalent to sorting by
6093 * page name.
6094 *
6095 * @return string
6096 */
6097 public function getDefaultSort() {
6098 if ( $this->mDefaultSort !== false ) {
6099 return $this->mDefaultSort;
6100 } else {
6101 return '';
6102 }
6103 }
6104
6105 /**
6106 * Accessor for $mDefaultSort
6107 * Unlike getDefaultSort(), will return false if none is set
6108 *
6109 * @return string|bool
6110 */
6111 public function getCustomDefaultSort() {
6112 return $this->mDefaultSort;
6113 }
6114
6115 /**
6116 * Try to guess the section anchor name based on a wikitext fragment
6117 * presumably extracted from a heading, for example "Header" from
6118 * "== Header ==".
6119 *
6120 * @param string $text
6121 *
6122 * @return string
6123 */
6124 public function guessSectionNameFromWikiText( $text ) {
6125 # Strip out wikitext links(they break the anchor)
6126 $text = $this->stripSectionName( $text );
6127 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6128 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
6129 }
6130
6131 /**
6132 * Same as guessSectionNameFromWikiText(), but produces legacy anchors
6133 * instead. For use in redirects, since IE6 interprets Redirect: headers
6134 * as something other than UTF-8 (apparently?), resulting in breakage.
6135 *
6136 * @param string $text The section name
6137 * @return string An anchor
6138 */
6139 public function guessLegacySectionNameFromWikiText( $text ) {
6140 # Strip out wikitext links(they break the anchor)
6141 $text = $this->stripSectionName( $text );
6142 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
6143 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
6144 }
6145
6146 /**
6147 * Strips a text string of wikitext for use in a section anchor
6148 *
6149 * Accepts a text string and then removes all wikitext from the
6150 * string and leaves only the resultant text (i.e. the result of
6151 * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
6152 * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
6153 * to create valid section anchors by mimicing the output of the
6154 * parser when headings are parsed.
6155 *
6156 * @param string $text Text string to be stripped of wikitext
6157 * for use in a Section anchor
6158 * @return string Filtered text string
6159 */
6160 public function stripSectionName( $text ) {
6161 # Strip internal link markup
6162 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
6163 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
6164
6165 # Strip external link markup
6166 # @todo FIXME: Not tolerant to blank link text
6167 # I.E. [https://www.mediawiki.org] will render as [1] or something depending
6168 # on how many empty links there are on the page - need to figure that out.
6169 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
6170
6171 # Parse wikitext quotes (italics & bold)
6172 $text = $this->doQuotes( $text );
6173
6174 # Strip HTML tags
6175 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
6176 return $text;
6177 }
6178
6179 /**
6180 * strip/replaceVariables/unstrip for preprocessor regression testing
6181 *
6182 * @param string $text
6183 * @param Title $title
6184 * @param ParserOptions $options
6185 * @param int $outputType
6186 *
6187 * @return string
6188 */
6189 public function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
6190 $magicScopeVariable = $this->lock();
6191 $this->startParse( $title, $options, $outputType, true );
6192
6193 $text = $this->replaceVariables( $text );
6194 $text = $this->mStripState->unstripBoth( $text );
6195 $text = Sanitizer::removeHTMLtags( $text );
6196 return $text;
6197 }
6198
6199 /**
6200 * @param string $text
6201 * @param Title $title
6202 * @param ParserOptions $options
6203 * @return string
6204 */
6205 public function testPst( $text, Title $title, ParserOptions $options ) {
6206 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
6207 }
6208
6209 /**
6210 * @param string $text
6211 * @param Title $title
6212 * @param ParserOptions $options
6213 * @return string
6214 */
6215 public function testPreprocess( $text, Title $title, ParserOptions $options ) {
6216 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
6217 }
6218
6219 /**
6220 * Call a callback function on all regions of the given text that are not
6221 * inside strip markers, and replace those regions with the return value
6222 * of the callback. For example, with input:
6223 *
6224 * aaa<MARKER>bbb
6225 *
6226 * This will call the callback function twice, with 'aaa' and 'bbb'. Those
6227 * two strings will be replaced with the value returned by the callback in
6228 * each case.
6229 *
6230 * @param string $s
6231 * @param callable $callback
6232 *
6233 * @return string
6234 */
6235 public function markerSkipCallback( $s, $callback ) {
6236 $i = 0;
6237 $out = '';
6238 while ( $i < strlen( $s ) ) {
6239 $markerStart = strpos( $s, self::MARKER_PREFIX, $i );
6240 if ( $markerStart === false ) {
6241 $out .= call_user_func( $callback, substr( $s, $i ) );
6242 break;
6243 } else {
6244 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6245 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6246 if ( $markerEnd === false ) {
6247 $out .= substr( $s, $markerStart );
6248 break;
6249 } else {
6250 $markerEnd += strlen( self::MARKER_SUFFIX );
6251 $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6252 $i = $markerEnd;
6253 }
6254 }
6255 }
6256 return $out;
6257 }
6258
6259 /**
6260 * Remove any strip markers found in the given text.
6261 *
6262 * @param string $text Input string
6263 * @return string
6264 */
6265 public function killMarkers( $text ) {
6266 return $this->mStripState->killMarkers( $text );
6267 }
6268
6269 /**
6270 * Save the parser state required to convert the given half-parsed text to
6271 * HTML. "Half-parsed" in this context means the output of
6272 * recursiveTagParse() or internalParse(). This output has strip markers
6273 * from replaceVariables (extensionSubstitution() etc.), and link
6274 * placeholders from replaceLinkHolders().
6275 *
6276 * Returns an array which can be serialized and stored persistently. This
6277 * array can later be loaded into another parser instance with
6278 * unserializeHalfParsedText(). The text can then be safely incorporated into
6279 * the return value of a parser hook.
6280 *
6281 * @param string $text
6282 *
6283 * @return array
6284 */
6285 public function serializeHalfParsedText( $text ) {
6286 $data = array(
6287 'text' => $text,
6288 'version' => self::HALF_PARSED_VERSION,
6289 'stripState' => $this->mStripState->getSubState( $text ),
6290 'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6291 );
6292 return $data;
6293 }
6294
6295 /**
6296 * Load the parser state given in the $data array, which is assumed to
6297 * have been generated by serializeHalfParsedText(). The text contents is
6298 * extracted from the array, and its markers are transformed into markers
6299 * appropriate for the current Parser instance. This transformed text is
6300 * returned, and can be safely included in the return value of a parser
6301 * hook.
6302 *
6303 * If the $data array has been stored persistently, the caller should first
6304 * check whether it is still valid, by calling isValidHalfParsedText().
6305 *
6306 * @param array $data Serialized data
6307 * @throws MWException
6308 * @return string
6309 */
6310 public function unserializeHalfParsedText( $data ) {
6311 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6312 throw new MWException( __METHOD__ . ': invalid version' );
6313 }
6314
6315 # First, extract the strip state.
6316 $texts = array( $data['text'] );
6317 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6318
6319 # Now renumber links
6320 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6321
6322 # Should be good to go.
6323 return $texts[0];
6324 }
6325
6326 /**
6327 * Returns true if the given array, presumed to be generated by
6328 * serializeHalfParsedText(), is compatible with the current version of the
6329 * parser.
6330 *
6331 * @param array $data
6332 *
6333 * @return bool
6334 */
6335 public function isValidHalfParsedText( $data ) {
6336 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6337 }
6338
6339 /**
6340 * Parsed a width param of imagelink like 300px or 200x300px
6341 *
6342 * @param string $value
6343 *
6344 * @return array
6345 * @since 1.20
6346 */
6347 public function parseWidthParam( $value ) {
6348 $parsedWidthParam = array();
6349 if ( $value === '' ) {
6350 return $parsedWidthParam;
6351 }
6352 $m = array();
6353 # (bug 13500) In both cases (width/height and width only),
6354 # permit trailing "px" for backward compatibility.
6355 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6356 $width = intval( $m[1] );
6357 $height = intval( $m[2] );
6358 $parsedWidthParam['width'] = $width;
6359 $parsedWidthParam['height'] = $height;
6360 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6361 $width = intval( $value );
6362 $parsedWidthParam['width'] = $width;
6363 }
6364 return $parsedWidthParam;
6365 }
6366
6367 /**
6368 * Lock the current instance of the parser.
6369 *
6370 * This is meant to stop someone from calling the parser
6371 * recursively and messing up all the strip state.
6372 *
6373 * @throws MWException If parser is in a parse
6374 * @return ScopedCallback The lock will be released once the return value goes out of scope.
6375 */
6376 protected function lock() {
6377 if ( $this->mInParse ) {
6378 throw new MWException( "Parser state cleared while parsing. "
6379 . "Did you call Parser::parse recursively?" );
6380 }
6381 $this->mInParse = true;
6382
6383 $that = $this;
6384 $recursiveCheck = new ScopedCallback( function() use ( $that ) {
6385 $that->mInParse = false;
6386 } );
6387
6388 return $recursiveCheck;
6389 }
6390
6391 /**
6392 * Strip outer <p></p> tag from the HTML source of a single paragraph.
6393 *
6394 * Returns original HTML if the <p/> tag has any attributes, if there's no wrapping <p/> tag,
6395 * or if there is more than one <p/> tag in the input HTML.
6396 *
6397 * @param string $html
6398 * @return string
6399 * @since 1.24
6400 */
6401 public static function stripOuterParagraph( $html ) {
6402 $m = array();
6403 if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $html, $m ) ) {
6404 if ( strpos( $m[1], '</p>' ) === false ) {
6405 $html = $m[1];
6406 }
6407 }
6408
6409 return $html;
6410 }
6411
6412 /**
6413 * Return this parser if it is not doing anything, otherwise
6414 * get a fresh parser. You can use this method by doing
6415 * $myParser = $wgParser->getFreshParser(), or more simply
6416 * $wgParser->getFreshParser()->parse( ... );
6417 * if you're unsure if $wgParser is safe to use.
6418 *
6419 * @since 1.24
6420 * @return Parser A parser object that is not parsing anything
6421 */
6422 public function getFreshParser() {
6423 global $wgParserConf;
6424 if ( $this->mInParse ) {
6425 return new $wgParserConf['class']( $wgParserConf );
6426 } else {
6427 return $this;
6428 }
6429 }
6430 }
Wiklou, le Wiki du Wiklou