r27667 MediaWiki - Code Review archive

Repository:MediaWiki
Revision:< r27666‎ | r27667 | r27668 >
Date:10:55, 20 November 2007
Author:tstarling
Status:old
Tags:
Comment:
* Refactored the parser. See my huge entry in RELEASE-NOTES for details.
* Made it possible to configure the parser class being used, via $wgParserConf.
* Moved defines from the top of Parser.php to either class constants or Defines.php
* Added Parser_DiffTest, a differential parser class for regression testing
* Added Parser_OldPP, a parser class which operates like the parser before this commit. I made one breaking change: a bugfix to avoid losing whitespace when adding MWTEMPLATESECTION markers.
* Made internal tidy work with PHP 5
* Added the ability to supply a hook for template fetching via ParserOptions. This is handy for testing.
* Updated parserTests.txt to account for the various breaking changes I made. Removed a few parser tests that no longer test for anything useful.
Modified paths:
  • /trunk/phase3/RELEASE-NOTES (modified) (history)
  • /trunk/phase3/includes/AutoLoader.php (modified) (history)
  • /trunk/phase3/includes/DefaultSettings.php (modified) (history)
  • /trunk/phase3/includes/Defines.php (modified) (history)
  • /trunk/phase3/includes/Parser.php (modified) (history)
  • /trunk/phase3/includes/ParserOptions.php (modified) (history)
  • /trunk/phase3/includes/Parser_DiffTest.php (added) (history)
  • /trunk/phase3/includes/Parser_OldPP.php (added) (history)
  • /trunk/phase3/includes/Setup.php (modified) (history)
  • /trunk/phase3/maintenance/parserTests.txt (modified) (history)

Diff [purge]

Index: trunk/phase3/maintenance/parserTests.txt
——@@ -3878,7 +3878,7 @@
38793879 !! input
38803880 {{div style|"><script>alert(document.cookie)</script>}}
38813881 !! result
3882 -<div>Magic div</div>
 3882+<div style="float: right;">&lt;script&gt;alert(document.cookie)&lt;/script&gt;"&gt;Magic div</div>
38833883
38843884 !! end
38853885
——@@ -3887,7 +3887,7 @@
38883888 !! input
38893889 {{div style|" ><script>alert(document.cookie)</script>}}
38903890 !! result
3891 -<div style="float: right;">Magic div</div>
 3891+<div style="float: right;">&lt;script&gt;alert(document.cookie)&lt;/script&gt;"&gt;Magic div</div>
38923892
38933893 !! end
38943894
——@@ -4151,7 +4151,7 @@
41524152
41534153
41544154 !! test
4155 -Parser hook: case insensetive
 4155+Parser hook: case insensitive
41564156 !! input
41574157 <TAG>input</TAG>
41584158 !! result
——@@ -4165,7 +4165,7 @@
41664166
41674167
41684168 !! test
4169 -Parser hook: case insensetive, redux
 4169+Parser hook: case insensitive, redux
41704170 !! input
41714171 <TaG>input</TAg>
41724172 !! result
——@@ -4724,8 +4724,8 @@
47254725 |
47264726 !! result
47274727 <table>
4728 -
4729 -<u class="&#124;">} &gt;
 4728+{{{|
 4729+<u class="&#124;">}}}} &gt;
47304730 <br style="onmouseover=&#39;alert(document.cookie);&#39;" />
47314731
47324732 MOVE YOUR MOUSE CURSOR OVER THIS TEXT
——@@ -4749,8 +4749,10 @@
47504750 >
47514751 }}}blah" onmouseover="alert('hello world');" align="left"'''MOVE MOUSE CURSOR OVER HERE
47524752 !! result
 4753+<p>{{{|
 4754+</p>
47534755 <li class="&#124;&#124;">
4754 -blah" onmouseover="alert('hello world');" align="left"<b>MOVE MOUSE CURSOR OVER HERE</b>
 4756+}}}blah" onmouseover="alert('hello world');" align="left"<b>MOVE MOUSE CURSOR OVER HERE</b>
47554757
47564758 !! end
47574759
——@@ -5251,10 +5253,11 @@
52525254 section=1
52535255 !! input
52545256 ==a==
5255 -==legal== <!-- a legal section -->
 5257+==unmarked== <!-- an unmarked section -->
52565258 ==b==
52575259 !! result
52585260 ==a==
 5261+==unmarked== <!-- an unmarked section -->
52595262 !! end
52605263
52615264 !! test
——@@ -5263,10 +5266,10 @@
52645267 section=2
52655268 !! input
52665269 ==a==
5267 -==legal== <!-- a legal section -->
 5270+==unmarked== <!-- an unmarked section -->
52685271 ==b==
52695272 !! result
5270 -==legal== <!-- a legal section -->
 5273+==b==
52715274 !! end
52725275
52735276 !! test
——@@ -5295,103 +5298,80 @@
52965299 !! end
52975300
52985301
 5302+# Formerly testing for bug 2587, now resolved by the use of unmarked sections
 5303+# instead of respecting commented sections
52995304 !! test
5300 -Section extraction prefixed by comment (section 1) (bug 2587)
 5305+Section extraction prefixed by comment (section 1)
53015306 !! options
53025307 section=1
53035308 !! input
53045309 <!-- -->==sec1==
53055310 ==sec2==
53065311 !!result
5307 -<!-- -->==sec1==
 5312+==sec2==
53085313 !!end
53095314
53105315 !! test
5311 -Section extraction prefixed by comment (section 2) (bug 2587)
 5316+Section extraction prefixed by comment (section 2)
53125317 !! options
53135318 section=2
53145319 !! input
53155320 <!-- -->==sec1==
53165321 ==sec2==
53175322 !!result
5318 -==sec2==
 5323+
53195324 !!end
53205325
53215326
 5327+# Formerly testing for bug 2607, now resolved by the use of unmarked sections
 5328+# instead of respecting HTML-style headings
53225329 !! test
5323 -Section extraction, mixed wiki and html (section 1) (bug 2607)
 5330+Section extraction, mixed wiki and html (section 1)
53245331 !! options
53255332 section=1
53265333 !! input
5327 -<h2>1</h2>
 5334+<h2>unmarked</h2>
 5335+unmarked
 5336+==1==
53285337 one
53295338 ==2==
53305339 two
5331 -==3==
5332 -three
53335340 !! result
5334 -<h2>1</h2>
 5341+==1==
53355342 one
53365343 !! end
53375344
53385345 !! test
5339 -Section extraction, mixed wiki and html (section 2) (bug 2607)
 5346+Section extraction, mixed wiki and html (section 2)
53405347 !! options
53415348 section=2
53425349 !! input
5343 -<h2>1</h2>
 5350+<h2>unmarked</h2>
 5351+unmarked
 5352+==1==
53445353 one
53455354 ==2==
53465355 two
5347 -==3==
5348 -three
53495356 !! result
53505357 ==2==
53515358 two
53525359 !! end
53535360
53545361
 5362+# Formerly testing for bug 3342
53555363 !! test
5356 -Section extraction, heading surrounded by <noinclude> (bug 3342)
 5364+Section extraction, heading surrounded by <noinclude>
53575365 !! options
53585366 section=1
53595367 !! input
5360 -<noinclude>==a==</noinclude>
5361 -text
 5368+<noinclude>==unmarked==</noinclude>
 5369+==marked==
53625370 !! result
5363 -<noinclude>==a==</noinclude>
5364 -text
 5371+==marked==
53655372 !!end
53665373
53675374
53685375 !! test
5369 -Section extraction, HTML heading subsections (bug 5272)
5370 -!! options
5371 -section=1
5372 -!! input
5373 -<h2>a</h2>
5374 -<h3>aa</h3>
5375 -<h2>b</h2>
5376 -!! result
5377 -<h2>a</h2>
5378 -<h3>aa</h3>
5379 -!! end
5380 -
5381 -!! test
5382 -Section extraction, HTML headings should be ignored in extensions (bug 3476)
5383 -!! options
5384 -section=2
5385 -!! input
5386 -<h2>a</h2>
5387 -<tag>
5388 -<h2>not b</h2>
5389 -</tag>
5390 -<h2>b</h2>
5391 -!! result
5392 -<h2>b</h2>
5393 -!! end
5394 -
5395 -!! test
53965376 Section replacement test (section 0)
53975377 !! options
53985378 replace=0,"xxx"
——@@ -5723,94 +5703,6 @@
57245704
57255705
57265706 !! test
5727 -Section extraction, HTML headings not at line boundaries (section 0)
5728 -!! options
5729 -section=0
5730 -!! input
5731 -<h2>Evil</h2><i>blah blah blah</i>
5732 -
5733 -evil blah
5734 -
5735 -<h2>Nice</h2>
5736 -
5737 -nice blah
5738 -
5739 -<i>extra evil</i><h2>Extra nasty</h2>
5740 -
5741 -extra nasty
5742 -!! result
5743 -!! end
5744 -
5745 -!! test
5746 -Section extraction, HTML headings not at line boundaries (section 1)
5747 -!! options
5748 -section=1
5749 -!! input
5750 -<h2>Evil</h2><i>blah blah blah</i>
5751 -
5752 -evil blah
5753 -
5754 -<h2>Nice</h2>
5755 -
5756 -nice blah
5757 -
5758 -<i>extra evil</i><h2>Extra nasty</h2>
5759 -
5760 -extra nasty
5761 -!! result
5762 -<h2>Evil</h2><i>blah blah blah</i>
5763 -
5764 -evil blah
5765 -!! end
5766 -
5767 -!! test
5768 -Section extraction, HTML headings not at line boundaries (section 2)
5769 -!! options
5770 -section=2
5771 -!! input
5772 -<h2>Evil</h2><i>blah blah blah</i>
5773 -
5774 -evil blah
5775 -
5776 -<h2>Nice</h2>
5777 -
5778 -nice blah
5779 -
5780 -<i>extra evil</i><h2>Extra nasty</h2>
5781 -
5782 -extra nasty
5783 -!! result
5784 -<h2>Nice</h2>
5785 -
5786 -nice blah
5787 -
5788 -<i>extra evil</i>
5789 -!! end
5790 -
5791 -!! test
5792 -Section extraction, HTML headings not at line boundaries (section 3)
5793 -!! options
5794 -section=3
5795 -!! input
5796 -<h2>Evil</h2><i>blah blah blah</i>
5797 -
5798 -evil blah
5799 -
5800 -<h2>Nice</h2>
5801 -
5802 -nice blah
5803 -
5804 -<i>extra evil</i><h2>Extra nasty</h2>
5805 -
5806 -extra nasty
5807 -!! result
5808 -<h2>Extra nasty</h2>
5809 -
5810 -extra nasty
5811 -!! end
5812 -
5813 -
5814 -!! test
58155707 Section extraction, heading followed by pre with 20 spaces (bug 6398)
58165708 !! options
58175709 section=1
Index: trunk/phase3/includes/Defines.php
——@@ -263,4 +263,17 @@
264264 # Hook support constants
265265 define( 'MW_SUPPORTS_EDITFILTERMERGED', 1 );
266266
 267+# Allowed values for Parser::$mOutputType
 268+# Parameter to Parser::startExternalParse().
 269+define( 'OT_HTML', 1 );
 270+define( 'OT_WIKI', 2 );
 271+define( 'OT_MSG' , 3 );
 272+define( 'OT_PREPROCESS', 4 );
267273
 274+# Flags for Parser::setFunctionHook
 275+define( 'SFH_NO_HASH', 1 );
 276+define( 'SFH_OBJECT_ARGS', 2 );
 277+
 278+# Flags for Parser::replaceLinkHolders
 279+define( 'RLH_FOR_UPDATE', 1 );
 280+
Index: trunk/phase3/includes/Parser.php
——@@ -7,56 +7,7 @@
88 * @addtogroup Parser
99 */
1010
11 -/**
12 - * Update this version number when the ParserOutput format
13 - * changes in an incompatible way, so the parser cache
14 - * can automatically discard old data.
15 - */
16 -define( 'MW_PARSER_VERSION', '1.6.2' );
1711
18 -define( 'RLH_FOR_UPDATE', 1 );
19 -
20 -# Allowed values for $mOutputType
21 -define( 'OT_HTML', 1 );
22 -define( 'OT_WIKI', 2 );
23 -define( 'OT_MSG' , 3 );
24 -define( 'OT_PREPROCESS', 4 );
25 -
26 -# Flags for setFunctionHook
27 -define( 'SFH_NO_HASH', 1 );
28 -
29 -# string parameter for extractTags which will cause it
30 -# to strip HTML comments in addition to regular
31 -# <XML>-style tags. This should not be anything we
32 -# may want to use in wikisyntax
33 -define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
34 -
35 -# Constants needed for external link processing
36 -define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
37 -# Everything except bracket, space, or control characters
38 -define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
39 -# Including space, but excluding newlines
40 -define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
41 -define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
42 -define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
43 -define( 'EXT_LINK_BRACKETED', '/\[(\b(' . wfUrlProtocols() . ')'.
44 - EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
45 -define( 'EXT_IMAGE_REGEX',
46 - '/^('.HTTP_PROTOCOLS.')'. # Protocol
47 - '('.EXT_LINK_URL_CLASS.'+)\\/'. # Hostname and path
48 - '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
49 -);
50 -
51 -// State constants for the definition list colon extraction
52 -define( 'MW_COLON_STATE_TEXT', 0 );
53 -define( 'MW_COLON_STATE_TAG', 1 );
54 -define( 'MW_COLON_STATE_TAGSTART', 2 );
55 -define( 'MW_COLON_STATE_CLOSETAG', 3 );
56 -define( 'MW_COLON_STATE_TAGSLASH', 4 );
57 -define( 'MW_COLON_STATE_COMMENT', 5 );
58 -define( 'MW_COLON_STATE_COMMENTDASH', 6 );
59 -define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
60 -
6112 /**
6213 * PHP Parser - Processes wiki markup (which uses a more user-friendly
6314 * syntax, such as "[[link]]" for making links), and provides a one-way
——@@ -92,23 +43,51 @@
9344 */
9445 class Parser
9546 {
96 - const VERSION = MW_PARSER_VERSION;
 47+ /**
 48+ * Update this version number when the ParserOutput format
 49+ * changes in an incompatible way, so the parser cache
 50+ * can automatically discard old data.
 51+ */
 52+ const VERSION = '1.6.2';
 53+
 54+ # Flags for Parser::setFunctionHook
 55+ # Also available as global constants from Defines.php
 56+ const SFH_NO_HASH = 1;
 57+ const SFH_OBJECT_ARGS = 2;
 58+
 59+ # Constants needed for external link processing
 60+ # Everything except bracket, space, or control characters
 61+ const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
 62+ const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)\
 63+ \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/S';
 64+
 65+ // State constants for the definition list colon extraction
 66+ const COLON_STATE_TEXT = 0;
 67+ const COLON_STATE_TAG = 1;
 68+ const COLON_STATE_TAGSTART = 2;
 69+ const COLON_STATE_CLOSETAG = 3;
 70+ const COLON_STATE_TAGSLASH = 4;
 71+ const COLON_STATE_COMMENT = 5;
 72+ const COLON_STATE_COMMENTDASH = 6;
 73+ const COLON_STATE_COMMENTDASHDASH = 7;
 74+
9775 /**#@+
9876 * @private
9977 */
10078 # Persistent:
10179 var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
102 - $mImageParams, $mImageParamsMagicArray;
 80+ $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
 81+ $mExtLinkBracketedRegex;
10382
10483 # Cleared with clearState():
10584 var $mOutput, $mAutonumber, $mDTopen, $mStripState;
10685 var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
107 - var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
108 - var $mIncludeSizes, $mDefaultSort;
109 - var $mTemplates, // cache of already loaded templates, avoids
110 - // multiple SQL queries for the same string
 86+ var $mInterwikiLinkHolders, $mLinkHolders;
 87+ var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 88+ var $mTplExpandCache,// empty-frame expansion cache
11189 $mTemplatePath; // stores an unsorted hash of all the templates already loaded
11290 // in this path. Used for loop detection.
 91+ var $mTplRedirCache, $mTplDomCache, $mHeadings;
11392
11493 # Temporary
11594 # These are variables reset at least once per parse regardless of $clearState
——@@ -127,11 +106,15 @@
128107 *
129108 * @public
130109 */
131 - function Parser() {
 110+ function __construct( $conf = array() ) {
132111 $this->mTagHooks = array();
133112 $this->mTransparentTagHooks = array();
134113 $this->mFunctionHooks = array();
135114 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 115+ $this->mStripList = array( 'nowiki', 'gallery' );
 116+ $this->mMarkerSuffix = "-QINU\x7f";
 117+ $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 118+ '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
136119 $this->mFirstCall = true;
137120 }
138121
——@@ -145,7 +128,7 @@
146129
147130 wfProfileIn( __METHOD__ );
148131 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
149 -
 132+
150133 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
151134
152135 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
——@@ -204,7 +187,7 @@
205188 $this->mDTopen = false;
206189 $this->mIncludeCount = array();
207190 $this->mStripState = new StripState;
208 - $this->mArgStack = array();
 191+ $this->mArgStack = false;
209192 $this->mInPre = false;
210193 $this->mInterwikiLinkHolders = array(
211194 'texts' => array(),
——@@ -225,21 +208,26 @@
226209 * Using it at the front also gives us a little extra robustness
227210 * since it shouldn't match when butted up against identifier-like
228211 * string constructs.
 212+ *
 213+ * Must not consist of all title characters, or else it will change
 214+ * the behaviour of <nowiki> in a link.
229215 */
230 - $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 216+ #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217+ $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
231218
232219 # Clear these on every parse, bug 4549
233 - $this->mTemplates = array();
234220 $this->mTemplatePath = array();
 221+ $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
235222
236223 $this->mShowToc = true;
237224 $this->mForceTocPosition = false;
238225 $this->mIncludeSizes = array(
239 - 'pre-expand' => 0,
240226 'post-expand' => 0,
241 - 'arg' => 0
 227+ 'arg' => 0,
242228 );
 229+ $this->mPPNodeCount = 0;
243230 $this->mDefaultSort = false;
 231+ $this->mHeadings = array();
244232
245233 wfRunHooks( 'ParserClearState', array( &$this ) );
246234 wfProfileOut( __METHOD__ );
——@@ -302,7 +290,7 @@
303291 }
304292 $this->setOutputType( OT_HTML );
305293 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
306 - $text = $this->strip( $text, $this->mStripState );
 294+ # No more strip!
307295 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
308296 $text = $this->internalParse( $text );
309297 $text = $this->mStripState->unstripGeneral( $text );
——@@ -335,17 +323,17 @@
336324 //!JF Move to its own function
337325
338326 $uniq_prefix = $this->mUniqPrefix;
339 - $matches = array();
 327+ $matches = array();
340328 $elements = array_keys( $this->mTransparentTagHooks );
341 - $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 329+ $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
342330
343 - foreach( $matches as $marker => $data ) {
344 - list( $element, $content, $params, $tag ) = $data;
345 - $tagName = strtolower( $element );
346 - if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
347 - $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
348 - array( $content, $params, $this ) );
349 - } else {
 331+ foreach( $matches as $marker => $data ) {
 332+ list( $element, $content, $params, $tag ) = $data;
 333+ $tagName = strtolower( $element );
 334+ if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 335+ $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 336+ array( $content, $params, $this ) );
 337+ } else {
350338 $output = $tag;
351339 }
352340 $this->mStripState->general->setPair( $marker, $output );
——@@ -390,7 +378,7 @@
391379 if ( max( $this->mIncludeSizes ) > 1000 ) {
392380 $max = $this->mOptions->getMaxIncludeSize();
393381 $text .= "<!-- \n" .
394 - "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 382+ "Preprocessor node count: {$this->mPPNodeCount}\n" .
395383 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
396384 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
397385 "Maximum: $max bytes\n" .
——@@ -412,7 +400,6 @@
413401 function recursiveTagParse( $text ) {
414402 wfProfileIn( __METHOD__ );
415403 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
416 - $text = $this->strip( $text, $this->mStripState );
417404 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
418405 $text = $this->internalParse( $text );
419406 wfProfileOut( __METHOD__ );
——@@ -433,12 +420,11 @@
434421 $this->mRevisionId = $revid;
435422 }
436423 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
437 - $text = $this->strip( $text, $this->mStripState );
438424 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 425+ $text = $this->replaceVariables( $text );
439426 if ( $this->mOptions->getRemoveComments() ) {
440427 $text = Sanitizer::removeHTMLcomments( $text );
441428 }
442 - $text = $this->replaceVariables( $text );
443429 $text = $this->mStripState->unstripBoth( $text );
444430 wfProfileOut( __METHOD__ );
445431 return $text;
——@@ -508,7 +494,7 @@
509495 $inside = $p[4];
510496 }
511497
512 - $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 498+ $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
513499 $stripped .= $marker;
514500
515501 if ( $close === '/>' ) {
——@@ -543,125 +529,24 @@
544530 }
545531
546532 /**
547 - * Strips and renders nowiki, pre, math, hiero
548 - * If $render is set, performs necessary rendering operations on plugins
549 - * Returns the text, and fills an array with data needed in unstrip()
550 - *
551 - * @param StripState $state
552 - *
553 - * @param bool $stripcomments when set, HTML comments <!-- like this -->
554 - * will be stripped in addition to other tags. This is important
555 - * for section editing, where these comments cause confusion when
556 - * counting the sections in the wikisource
557 - *
558 - * @param array dontstrip contains tags which should not be stripped;
559 - * used to prevent stipping of <gallery> when saving (fixes bug 2700)
560 - *
561 - * @private
 533+ * Get a list of strippable XML-like elements
562534 */
563 - function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
564 - global $wgContLang;
565 - wfProfileIn( __METHOD__ );
566 - $render = ($this->mOutputType == OT_HTML);
567 -
568 - $uniq_prefix = $this->mUniqPrefix;
569 - $commentState = new ReplacementArray;
570 - $nowikiItems = array();
571 - $generalItems = array();
572 -
573 - $elements = array_merge(
574 - array( 'nowiki', 'gallery' ),
575 - array_keys( $this->mTagHooks ) );
 535+ function getStripList() {
576536 global $wgRawHtml;
 537+ $elements = $this->mStripList;
577538 if( $wgRawHtml ) {
578539 $elements[] = 'html';
579540 }
580541 if( $this->mOptions->getUseTeX() ) {
581542 $elements[] = 'math';
582543 }
 544+ return $elements;
 545+ }
583546
584 - # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
585 - foreach ( $elements AS $k => $v ) {
586 - if ( !in_array ( $v , $dontstrip ) ) continue;
587 - unset ( $elements[$k] );
588 - }
589 -
590 - $matches = array();
591 - $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
592 -
593 - foreach( $matches as $marker => $data ) {
594 - list( $element, $content, $params, $tag ) = $data;
595 - if( $render ) {
596 - $tagName = strtolower( $element );
597 - wfProfileIn( __METHOD__."-render-$tagName" );
598 - switch( $tagName ) {
599 - case '!--':
600 - // Comment
601 - if( substr( $tag, -3 ) == '-->' ) {
602 - $output = $tag;
603 - } else {
604 - // Unclosed comment in input.
605 - // Close it so later stripping can remove it
606 - $output = "$tag-->";
607 - }
608 - break;
609 - case 'html':
610 - if( $wgRawHtml ) {
611 - $output = $content;
612 - break;
613 - }
614 - // Shouldn't happen otherwise. :)
615 - case 'nowiki':
616 - $output = Xml::escapeTagsOnly( $content );
617 - break;
618 - case 'math':
619 - $output = $wgContLang->armourMath(
620 - MathRenderer::renderMath( $content, $params ) );
621 - break;
622 - case 'gallery':
623 - $output = $this->renderImageGallery( $content, $params );
624 - break;
625 - default:
626 - if( isset( $this->mTagHooks[$tagName] ) ) {
627 - $output = call_user_func_array( $this->mTagHooks[$tagName],
628 - array( $content, $params, $this ) );
629 - } else {
630 - throw new MWException( "Invalid call hook $element" );
631 - }
632 - }
633 - wfProfileOut( __METHOD__."-render-$tagName" );
634 - } else {
635 - // Just stripping tags; keep the source
636 - $output = $tag;
637 - }
638 -
639 - // Unstrip the output, to support recursive strip() calls
640 - $output = $state->unstripBoth( $output );
641 -
642 - if( !$stripcomments && $element == '!--' ) {
643 - $commentState->setPair( $marker, $output );
644 - } elseif ( $element == 'html' || $element == 'nowiki' ) {
645 - $nowikiItems[$marker] = $output;
646 - } else {
647 - $generalItems[$marker] = $output;
648 - }
649 - }
650 - # Add the new items to the state
651 - # We do this after the loop instead of during it to avoid slowing
652 - # down the recursive unstrip
653 - $state->nowiki->mergeArray( $nowikiItems );
654 - $state->general->mergeArray( $generalItems );
655 -
656 - # Unstrip comments unless explicitly told otherwise.
657 - # (The comments are always stripped prior to this point, so as to
658 - # not invoke any extension tags / parser hooks contained within
659 - # a comment.)
660 - if ( !$stripcomments ) {
661 - // Put them all back and forget them
662 - $text = $commentState->replace( $text );
663 - }
664 -
665 - wfProfileOut( __METHOD__ );
 547+ /**
 548+ * @deprecated use replaceVariables
 549+ */
 550+ function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
666551 return $text;
667552 }
668553
——@@ -700,9 +585,10 @@
701586 *
702587 * @private
703588 */
704 - function insertStripItem( $text, &$state ) {
705 - $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
706 - $state->general->setPair( $rnd, $text );
 589+ function insertStripItem( $text ) {
 590+ static $n = 0;
 591+ $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 592+ $this->mStripState->general->setPair( $rnd, $text );
707593 return $rnd;
708594 }
709595
——@@ -786,8 +672,7 @@
787673
788674 /**
789675 * Use the HTML tidy PECL extension to use the tidy library in-process,
790 - * saving the overhead of spawning a new process. Currently written to
791 - * the PHP 4.3.x version of the extension, may not work on PHP 5.
 676+ * saving the overhead of spawning a new process.
792677 *
793678 * 'pear install tidy' should be able to compile the extension module.
794679 *
——@@ -795,20 +680,19 @@
796681 * @static
797682 */
798683 function internalTidy( $text ) {
799 - global $wgTidyConf;
 684+ global $wgTidyConf, $IP;
800685 $fname = 'Parser::internalTidy';
801686 wfProfileIn( $fname );
802687
803 - tidy_load_config( $wgTidyConf );
804 - tidy_set_encoding( 'utf8' );
805 - tidy_parse_string( $text );
806 - tidy_clean_repair();
807 - if( tidy_get_status() == 2 ) {
 688+ $tidy = new tidy;
 689+ $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 690+ $tidy->cleanRepair();
 691+ if( $tidy->getStatus() == 2 ) {
808692 // 2 is magic number for fatal error
809693 // http://www.php.net/manual/en/function.tidy-get-status.php
810694 $cleansource = null;
811695 } else {
812 - $cleansource = tidy_get_output();
 696+ $cleansource = tidy_get_output( $tidy );
813697 }
814698 wfProfileOut( $fname );
815699 return $cleansource;
——@@ -1013,7 +897,6 @@
1014898 * @private
1015899 */
1016900 function internalParse( $text ) {
1017 - $args = array();
1018901 $isMain = true;
1019902 $fname = 'Parser::internalParse';
1020903 wfProfileIn( $fname );
——@@ -1029,9 +912,8 @@
1030913 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
1031914 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1032915
1033 - $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
1034 -
1035 - $text = $this->replaceVariables( $text, $args );
 916+ $text = $this->replaceVariables( $text );
 917+ $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1036918 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1037919
1038920 // Tables need to come after variable replacement for things to work
——@@ -1070,7 +952,7 @@
1071953 *
1072954 * @private
1073955 */
1074 - function &doMagicLinks( &$text ) {
 956+ function doMagicLinks( $text ) {
1075957 wfProfileIn( __METHOD__ );
1076958 $text = preg_replace_callback(
1077959 '!(?: # Start cases
——@@ -1134,8 +1016,8 @@
11351017 wfProfileIn( $fname );
11361018 for ( $i = 6; $i >= 1; --$i ) {
11371019 $h = str_repeat( '=', $i );
1138 - $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1139 - "<h{$i}>\\1</h{$i}>\\2", $text );
 1020+ $text = preg_replace( "/^$h(.+)$h\\s*$/m",
 1021+ "<h$i>\\1</h$i>", $text );
11401022 }
11411023 wfProfileOut( $fname );
11421024 return $text;
——@@ -1339,7 +1221,7 @@
13401222
13411223 $sk = $this->mOptions->getSkin();
13421224
1343 - $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 1225+ $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
13441226
13451227 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
13461228
——@@ -1433,7 +1315,7 @@
14341316 $remainder = $bits[$i++];
14351317
14361318 $m = array();
1437 - if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
 1319+ if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
14381320 # Found some characters after the protocol that look promising
14391321 $url = $protocol . $m[1];
14401322 $trail = $m[2];
——@@ -1443,7 +1325,7 @@
14441326 if(strlen($trail) == 0 &&
14451327 isset($bits[$i]) &&
14461328 preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1447 - preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
 1329+ preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
14481330 {
14491331 # add protocol, arg
14501332 $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
——@@ -1540,7 +1422,7 @@
15411423 $text = false;
15421424 if ( $this->mOptions->getAllowExternalImages()
15431425 || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1544 - if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
 1426+ if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
15451427 # Image found
15461428 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
15471429 }
——@@ -2287,7 +2169,7 @@
22882170 }
22892171
22902172 // Ugly state machine to walk through avoiding tags.
2291 - $state = MW_COLON_STATE_TEXT;
 2173+ $state = self::COLON_STATE_TEXT;
22922174 $stack = 0;
22932175 $len = strlen( $str );
22942176 for( $i = 0; $i < $len; $i++ ) {
——@@ -2295,11 +2177,11 @@
22962178
22972179 switch( $state ) {
22982180 // (Using the number is a performance hack for common cases)
2299 - case 0: // MW_COLON_STATE_TEXT:
 2181+ case 0: // self::COLON_STATE_TEXT:
23002182 switch( $c ) {
23012183 case "<":
23022184 // Could be either a <start> tag or an </end> tag
2303 - $state = MW_COLON_STATE_TAGSTART;
 2185+ $state = self::COLON_STATE_TAGSTART;
23042186 break;
23052187 case ":":
23062188 if( $stack == 0 ) {
——@@ -2336,41 +2218,41 @@
23372219 }
23382220 // Skip ahead to next tag start
23392221 $i = $lt;
2340 - $state = MW_COLON_STATE_TAGSTART;
 2222+ $state = self::COLON_STATE_TAGSTART;
23412223 }
23422224 break;
2343 - case 1: // MW_COLON_STATE_TAG:
 2225+ case 1: // self::COLON_STATE_TAG:
23442226 // In a <tag>
23452227 switch( $c ) {
23462228 case ">":
23472229 $stack++;
2348 - $state = MW_COLON_STATE_TEXT;
 2230+ $state = self::COLON_STATE_TEXT;
23492231 break;
23502232 case "/":
23512233 // Slash may be followed by >?
2352 - $state = MW_COLON_STATE_TAGSLASH;
 2234+ $state = self::COLON_STATE_TAGSLASH;
23532235 break;
23542236 default:
23552237 // ignore
23562238 }
23572239 break;
2358 - case 2: // MW_COLON_STATE_TAGSTART:
 2240+ case 2: // self::COLON_STATE_TAGSTART:
23592241 switch( $c ) {
23602242 case "/":
2361 - $state = MW_COLON_STATE_CLOSETAG;
 2243+ $state = self::COLON_STATE_CLOSETAG;
23622244 break;
23632245 case "!":
2364 - $state = MW_COLON_STATE_COMMENT;
 2246+ $state = self::COLON_STATE_COMMENT;
23652247 break;
23662248 case ">":
23672249 // Illegal early close? This shouldn't happen D:
2368 - $state = MW_COLON_STATE_TEXT;
 2250+ $state = self::COLON_STATE_TEXT;
23692251 break;
23702252 default:
2371 - $state = MW_COLON_STATE_TAG;
 2253+ $state = self::COLON_STATE_TAG;
23722254 }
23732255 break;
2374 - case 3: // MW_COLON_STATE_CLOSETAG:
 2256+ case 3: // self::COLON_STATE_CLOSETAG:
23752257 // In a </tag>
23762258 if( $c == ">" ) {
23772259 $stack--;
——@@ -2379,35 +2261,35 @@
23802262 wfProfileOut( $fname );
23812263 return false;
23822264 }
2383 - $state = MW_COLON_STATE_TEXT;
 2265+ $state = self::COLON_STATE_TEXT;
23842266 }
23852267 break;
2386 - case MW_COLON_STATE_TAGSLASH:
 2268+ case self::COLON_STATE_TAGSLASH:
23872269 if( $c == ">" ) {
23882270 // Yes, a self-closed tag <blah/>
2389 - $state = MW_COLON_STATE_TEXT;
 2271+ $state = self::COLON_STATE_TEXT;
23902272 } else {
23912273 // Probably we're jumping the gun, and this is an attribute
2392 - $state = MW_COLON_STATE_TAG;
 2274+ $state = self::COLON_STATE_TAG;
23932275 }
23942276 break;
2395 - case 5: // MW_COLON_STATE_COMMENT:
 2277+ case 5: // self::COLON_STATE_COMMENT:
23962278 if( $c == "-" ) {
2397 - $state = MW_COLON_STATE_COMMENTDASH;
 2279+ $state = self::COLON_STATE_COMMENTDASH;
23982280 }
23992281 break;
2400 - case MW_COLON_STATE_COMMENTDASH:
 2282+ case self::COLON_STATE_COMMENTDASH:
24012283 if( $c == "-" ) {
2402 - $state = MW_COLON_STATE_COMMENTDASHDASH;
 2284+ $state = self::COLON_STATE_COMMENTDASHDASH;
24032285 } else {
2404 - $state = MW_COLON_STATE_COMMENT;
 2286+ $state = self::COLON_STATE_COMMENT;
24052287 }
24062288 break;
2407 - case MW_COLON_STATE_COMMENTDASHDASH:
 2289+ case self::COLON_STATE_COMMENTDASHDASH:
24082290 if( $c == ">" ) {
2409 - $state = MW_COLON_STATE_TEXT;
 2291+ $state = self::COLON_STATE_TEXT;
24102292 } else {
2411 - $state = MW_COLON_STATE_COMMENT;
 2293+ $state = self::COLON_STATE_COMMENT;
24122294 }
24132295 break;
24142296 default:
——@@ -2592,9 +2474,9 @@
25932475 case 'numberofpages':
25942476 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
25952477 case 'numberofadmins':
2596 - return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
 2478+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
25972479 case 'numberofedits':
2598 - return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
 2480+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
25992481 case 'currenttimestamp':
26002482 return $varCache[$index] = wfTimestampNow();
26012483 case 'localtimestamp':
——@@ -2633,189 +2515,479 @@
26342516 wfProfileIn( $fname );
26352517 $variableIDs = MagicWord::getVariableIDs();
26362518
2637 - $this->mVariables = array();
2638 - foreach ( $variableIDs as $id ) {
2639 - $mw =& MagicWord::get( $id );
2640 - $mw->addToArray( $this->mVariables, $id );
2641 - }
 2519+ $this->mVariables = new MagicWordArray( $variableIDs );
26422520 wfProfileOut( $fname );
26432521 }
26442522
26452523 /**
2646 - * parse any parentheses in format ((title|part|part))
2647 - * and call callbacks to get a replacement text for any found piece
 2524+ * Parse any parentheses in format ((title|part|part)} and return the document tree
 2525+ * This is the ghost of replace_variables().
26482526 *
26492527 * @param string $text The text to parse
2650 - * @param array $callbacks rules in form:
2651 - * '{' => array( # opening parentheses
2652 - * 'end' => '}', # closing parentheses
2653 - * 'cb' => array(2 => callback, # replacement callback to call if {{..}} is found
2654 - * 3 => callback # replacement callback to call if {{{..}}} is found
2655 - * )
2656 - * )
2657 - * 'min' => 2, # Minimum parenthesis count in cb
2658 - * 'max' => 3, # Maximum parenthesis count in cb
26592528 * @private
26602529 */
2661 - function replace_callback ($text, $callbacks) {
 2530+ function preprocessToDom ( $text ) {
26622531 wfProfileIn( __METHOD__ );
2663 - $openingBraceStack = array(); # this array will hold a stack of parentheses which are not closed yet
2664 - $lastOpeningBrace = -1; # last not closed parentheses
 2532+ wfProfileIn( __METHOD__.'-makexml' );
26652533
2666 - $validOpeningBraces = implode( '', array_keys( $callbacks ) );
 2534+ static $msgRules, $normalRules;
 2535+ if ( !$msgRules ) {
 2536+ $msgRules = array(
 2537+ '{' => array(
 2538+ 'end' => '}',
 2539+ 'names' => array(
 2540+ 2 => 'template',
 2541+ ),
 2542+ 'min' => 2,
 2543+ 'max' => 2,
 2544+ ),
 2545+ '[' => array(
 2546+ 'end' => ']',
 2547+ 'names' => array( 2 => null ),
 2548+ 'min' => 2,
 2549+ 'max' => 2,
 2550+ )
 2551+ );
 2552+ $normalRules = array(
 2553+ '{' => array(
 2554+ 'end' => '}',
 2555+ 'names' => array(
 2556+ 2 => 'template',
 2557+ 3 => 'tplarg',
 2558+ ),
 2559+ 'min' => 2,
 2560+ 'max' => 3,
 2561+ ),
 2562+ '[' => array(
 2563+ 'end' => ']',
 2564+ 'names' => array( 2 => null ),
 2565+ 'min' => 2,
 2566+ 'max' => 2,
 2567+ )
 2568+ );
 2569+ }
 2570+ if ( $this->ot['msg'] ) {
 2571+ $rules = $msgRules;
 2572+ } else {
 2573+ $rules = $normalRules;
 2574+ }
26672575
2668 - $i = 0;
 2576+ $extElements = implode( '|', $this->getStripList() );
 2577+ // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
 2578+ $extElementsRegex = "/($extElements)(?:\s|\/>|>)|(!--)/iA";
 2579+
 2580+ $stack = array(); # Stack of unclosed parentheses
 2581+ $stackIndex = -1; # Stack read pointer
 2582+
 2583+ $searchBase = implode( '', array_keys( $rules ) ) . '<';
 2584+
 2585+ $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
 2586+ $topAccum = '<root>'; # Top level text accumulator
 2587+ $accum =& $topAccum; # Current text accumulator
 2588+ $findEquals = false; # True to find equals signs in arguments
 2589+ $findHeading = false; # True to look at LF characters for possible headings
 2590+ $findPipe = false; # True to take notice of pipe characters
 2591+ $headingIndex = 1;
 2592+ $noMoreGT = false; # True if there are no more greater-than (>) signs right of $i
 2593+
26692594 while ( $i < strlen( $text ) ) {
2670 - # Find next opening brace, closing brace or pipe
2671 - if ( $lastOpeningBrace == -1 ) {
2672 - $currentClosing = '';
2673 - $search = $validOpeningBraces;
 2595+ if ( $i == -1 ) {
 2596+ $found = 'line-start';
 2597+ $curChar = '';
26742598 } else {
2675 - $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2676 - $search = $validOpeningBraces . '|' . $currentClosing;
 2599+ # Find next opening brace, closing brace or pipe
 2600+ $search = $searchBase;
 2601+ if ( $stackIndex == -1 ) {
 2602+ $currentClosing = '';
 2603+ // Look for headings only at the top stack level
 2604+ // Among other things, this resolves the ambiguity between =
 2605+ // for headings and = for template arguments
 2606+ $search .= "\n";
 2607+ } else {
 2608+ $currentClosing = $stack[$stackIndex]['close'];
 2609+ $search .= $currentClosing;
 2610+ }
 2611+ if ( $findPipe ) {
 2612+ $search .= '|';
 2613+ }
 2614+ if ( $findEquals ) {
 2615+ $search .= '=';
 2616+ }
 2617+ $rule = null;
 2618+ # Output literal section, advance input counter
 2619+ $literalLength = strcspn( $text, $search, $i );
 2620+ if ( $literalLength > 0 ) {
 2621+ $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
 2622+ $i += $literalLength;
 2623+ }
 2624+ if ( $i >= strlen( $text ) ) {
 2625+ if ( $currentClosing == "\n" ) {
 2626+ // Do a past-the-end run to finish off the heading
 2627+ $curChar = '';
 2628+ $found = 'line-end';
 2629+ } else {
 2630+ # All done
 2631+ break;
 2632+ }
 2633+ } else {
 2634+ $curChar = $text[$i];
 2635+ if ( $curChar == '|' ) {
 2636+ $found = 'pipe';
 2637+ } elseif ( $curChar == '=' ) {
 2638+ $found = 'equals';
 2639+ } elseif ( $curChar == '<' ) {
 2640+ $found = 'angle';
 2641+ } elseif ( $curChar == "\n" ) {
 2642+ if ( $stackIndex == -1 ) {
 2643+ $found = 'line-start';
 2644+ } else {
 2645+ $found = 'line-end';
 2646+ }
 2647+ } elseif ( $curChar == $currentClosing ) {
 2648+ $found = 'close';
 2649+ } elseif ( isset( $rules[$curChar] ) ) {
 2650+ $found = 'open';
 2651+ $rule = $rules[$curChar];
 2652+ } else {
 2653+ # Some versions of PHP have a strcspn which stops on null characters
 2654+ # Ignore and continue
 2655+ ++$i;
 2656+ continue;
 2657+ }
 2658+ }
26772659 }
2678 - $rule = null;
2679 - $i += strcspn( $text, $search, $i );
2680 - if ( $i < strlen( $text ) ) {
2681 - if ( $text[$i] == '|' ) {
2682 - $found = 'pipe';
2683 - } elseif ( $text[$i] == $currentClosing ) {
2684 - $found = 'close';
2685 - } elseif ( isset( $callbacks[$text[$i]] ) ) {
2686 - $found = 'open';
2687 - $rule = $callbacks[$text[$i]];
2688 - } else {
2689 - # Some versions of PHP have a strcspn which stops on null characters
2690 - # Ignore and continue
 2660+
 2661+ if ( $found == 'angle' ) {
 2662+ $matches = false;
 2663+ // Determine element name
 2664+ if ( !preg_match( $extElementsRegex, $text, $matches, 0, $i + 1 ) ) {
 2665+ // Element name missing or not listed
 2666+ $accum .= '&lt;';
26912667 ++$i;
26922668 continue;
26932669 }
2694 - } else {
2695 - # All done
2696 - break;
 2670+ // Handle comments
 2671+ if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
 2672+ // HTML comment, scan to end
 2673+ $endpos = strpos( $text, '-->', $i + 4 );
 2674+ if ( $endpos === false ) {
 2675+ // Unclosed comment in input, runs to end
 2676+ $accum .= htmlspecialchars( substr( $text, $i ) );
 2677+ if ( $this->ot['html'] ) {
 2678+ // Close it so later stripping can remove it
 2679+ $accum .= htmlspecialchars( '-->' );
 2680+ }
 2681+ $i = strlen( $text );
 2682+ continue;
 2683+ }
 2684+ $accum .= htmlspecialchars( substr( $text, $i, $endpos - $i + 3 ) );
 2685+ #$inner = substr( $text, $i + 4, $endpos - $i - 4 );
 2686+ #$accum .= '<ext><name>!--</name><inner>' . htmlspecialchars( $inner ) . '</inner></ext>';
 2687+ $i = $endpos + 3;
 2688+ continue;
 2689+ }
 2690+ $name = $matches[1];
 2691+ $attrStart = $i + strlen( $name ) + 1;
 2692+
 2693+ // Find end of tag
 2694+ $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
 2695+ if ( $tagEndPos === false ) {
 2696+ // Infinite backtrack
 2697+ // Disable tag search to prevent worst-case O(N^2) performance
 2698+ $noMoreGT = true;
 2699+ $accum .= '&lt;';
 2700+ ++$i;
 2701+ continue;
 2702+ }
 2703+ if ( $text[$tagEndPos-1] == '/' ) {
 2704+ $attrEnd = $tagEndPos - 1;
 2705+ $inner = null;
 2706+ $i = $tagEndPos + 1;
 2707+ $close = '';
 2708+ } else {
 2709+ $attrEnd = $tagEndPos;
 2710+ // Find closing tag
 2711+ if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
 2712+ $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
 2713+ $i = $matches[0][1] + strlen( $matches[0][0] );
 2714+ $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
 2715+ } else {
 2716+ // No end tag -- let it run out to the end of the text.
 2717+ $inner = substr( $text, $tagEndPos + 1 );
 2718+ $i = strlen( $text );
 2719+ $close = '';
 2720+ }
 2721+ }
 2722+ $accum .= '<ext>';
 2723+ if ( $attrEnd <= $attrStart ) {
 2724+ $attr = '';
 2725+ } else {
 2726+ $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
 2727+ }
 2728+ $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
 2729+ // Note that the attr element contains the whitespace between name and attribute,
 2730+ // this is necessary for precise reconstruction during pre-save transform.
 2731+ '<attr>' . htmlspecialchars( $attr ) . '</attr>';
 2732+ if ( $inner !== null ) {
 2733+ $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
 2734+ }
 2735+ $accum .= $close . '</ext>';
26972736 }
26982737
2699 - if ( $found == 'open' ) {
2700 - # found opening brace, let's add it to parentheses stack
2701 - $piece = array('brace' => $text[$i],
2702 - 'braceEnd' => $rule['end'],
2703 - 'title' => '',
2704 - 'parts' => null);
 2738+ elseif ( $found == 'line-start' ) {
 2739+ // Is this the start of a heading?
 2740+ // Line break belongs before the heading element in any case
 2741+ $accum .= $curChar;
 2742+ $i++;
 2743+
 2744+ $count = strspn( $text, '=', $i, 6 );
 2745+ if ( $count > 0 ) {
 2746+ $piece = array(
 2747+ 'open' => "\n",
 2748+ 'close' => "\n",
 2749+ 'parts' => array( str_repeat( '=', $count ) ),
 2750+ 'count' => $count );
 2751+ $stack[++$stackIndex] = $piece;
 2752+ $i += $count;
 2753+ $accum =& $stack[$stackIndex]['parts'][0];
 2754+ $findPipe = false;
 2755+ }
 2756+ }
27052757
 2758+ elseif ( $found == 'line-end' ) {
 2759+ $piece = $stack[$stackIndex];
 2760+ // A heading must be open, otherwise \n wouldn't have been in the search list
 2761+ assert( $piece['open'] == "\n" );
 2762+ assert( $stackIndex == 0 );
 2763+ // Search back through the accumulator to see if it has a proper close
 2764+ // No efficient way to do this in PHP AFAICT: strrev, PCRE search with $ anchor
 2765+ // and rtrim are all O(N) in total size. Optimal would be O(N) in trailing
 2766+ // whitespace size only.
 2767+ $m = false;
 2768+ $count = $piece['count'];
 2769+ if ( preg_match( "/(={{$count}})\s*$/", $accum, $m, 0, $count ) ) {
 2770+ // Found match, output <h>
 2771+ $count = min( strlen( $m[1] ), $count );
 2772+ $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
 2773+ $headingIndex++;
 2774+ } else {
 2775+ // No match, no <h>, just pass down the inner text
 2776+ $element = $accum;
 2777+ }
 2778+ // Unwind the stack
 2779+ // Headings can only occur on the top level, so this is a bit simpler than the
 2780+ // generic stack unwind operation in the close case
 2781+ unset( $stack[$stackIndex--] );
 2782+ $accum =& $topAccum;
 2783+ $findEquals = false;
 2784+ $findPipe = false;
 2785+
 2786+ // Append the result to the enclosing accumulator
 2787+ $accum .= $element;
 2788+ // Note that we do NOT increment the input pointer.
 2789+ // This is because the closing linebreak could be the opening linebreak of
 2790+ // another heading. Infinite loops are avoided because the next iteration MUST
 2791+ // hit the heading open case above, which unconditionally increments the
 2792+ // input pointer.
 2793+ }
 2794+
 2795+ elseif ( $found == 'open' ) {
27062796 # count opening brace characters
2707 - $piece['count'] = strspn( $text, $piece['brace'], $i );
2708 - $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2709 - $i += $piece['count'];
 2797+ $count = strspn( $text, $curChar, $i );
27102798
27112799 # we need to add to stack only if opening brace count is enough for one of the rules
2712 - if ( $piece['count'] >= $rule['min'] ) {
2713 - $lastOpeningBrace ++;
2714 - $openingBraceStack[$lastOpeningBrace] = $piece;
 2800+ if ( $count >= $rule['min'] ) {
 2801+ # Add it to the stack
 2802+ $piece = array(
 2803+ 'open' => $curChar,
 2804+ 'close' => $rule['end'],
 2805+ 'count' => $count,
 2806+ 'parts' => array( '' ),
 2807+ 'eqpos' => array(),
 2808+ 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
 2809+ );
 2810+
 2811+ $stackIndex ++;
 2812+ $stack[$stackIndex] = $piece;
 2813+ $accum =& $stack[$stackIndex]['parts'][0];
 2814+ $findEquals = false;
 2815+ $findPipe = true;
 2816+ } else {
 2817+ # Add literal brace(s)
 2818+ $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
27152819 }
2716 - } elseif ( $found == 'close' ) {
2717 - # lets check if it is enough characters for closing brace
2718 - $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2719 - $count = strspn( $text, $text[$i], $i, $maxCount );
 2820+ $i += $count;
 2821+ }
27202822
 2823+ elseif ( $found == 'close' ) {
 2824+ $piece = $stack[$stackIndex];
 2825+ # lets check if there are enough characters for closing brace
 2826+ $maxCount = $piece['count'];
 2827+ $count = strspn( $text, $curChar, $i, $maxCount );
 2828+
27212829 # check for maximum matching characters (if there are 5 closing
27222830 # characters, we will probably need only 3 - depending on the rules)
27232831 $matchingCount = 0;
2724 - $matchingCallback = null;
2725 - $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2726 - if ( $count > $cbType['max'] ) {
 2832+ $rule = $rules[$piece['open']];
 2833+ if ( $count > $rule['max'] ) {
27272834 # The specified maximum exists in the callback array, unless the caller
27282835 # has made an error
2729 - $matchingCount = $cbType['max'];
 2836+ $matchingCount = $rule['max'];
27302837 } else {
27312838 # Count is less than the maximum
27322839 # Skip any gaps in the callback array to find the true largest match
27332840 # Need to use array_key_exists not isset because the callback can be null
27342841 $matchingCount = $count;
2735 - while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
 2842+ while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
27362843 --$matchingCount;
27372844 }
27382845 }
27392846
27402847 if ($matchingCount <= 0) {
 2848+ # No matching element found in callback array
 2849+ # Output a literal closing brace and continue
 2850+ $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
27412851 $i += $count;
27422852 continue;
27432853 }
2744 - $matchingCallback = $cbType['cb'][$matchingCount];
 2854+ $name = $rule['names'][$matchingCount];
 2855+ if ( $name === null ) {
 2856+ // No element, just literal text
 2857+ $element = str_repeat( $piece['open'], $matchingCount ) .
 2858+ implode( '|', $piece['parts'] ) .
 2859+ str_repeat( $rule['end'], $matchingCount );
 2860+ } else {
 2861+ # Create XML element
 2862+ # Note: $parts is already XML, does not need to be encoded further
 2863+ $parts = $piece['parts'];
 2864+ $title = $parts[0];
 2865+ unset( $parts[0] );
27452866
2746 - # let's set a title or last part (if '|' was found)
2747 - if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2748 - $openingBraceStack[$lastOpeningBrace]['title'] =
2749 - substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2750 - $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2751 - } else {
2752 - $openingBraceStack[$lastOpeningBrace]['parts'][] =
2753 - substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2754 - $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
 2867+ # The invocation is at the start of the line if lineStart is set in
 2868+ # the stack, and all opening brackets are used up.
 2869+ if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
 2870+ $attr = ' lineStart="1"';
 2871+ } else {
 2872+ $attr = '';
 2873+ }
 2874+
 2875+ $element = "<$name$attr>";
 2876+ $element .= "<title>$title</title>";
 2877+ $argIndex = 1;
 2878+ foreach ( $parts as $partIndex => $part ) {
 2879+ if ( isset( $piece['eqpos'][$partIndex] ) ) {
 2880+ $eqpos = $piece['eqpos'][$partIndex];
 2881+ list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
 2882+ list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
 2883+ $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
 2884+ } else {
 2885+ list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
 2886+ $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
 2887+ $argIndex++;
 2888+ }
 2889+ }
 2890+ $element .= "</$name>";
27552891 }
27562892
2757 - $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2758 - $pieceEnd = $i + $matchingCount;
 2893+ # Advance input pointer
 2894+ $i += $matchingCount;
27592895
2760 - if( is_callable( $matchingCallback ) ) {
2761 - $cbArgs = array (
2762 - 'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2763 - 'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2764 - 'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2765 - 'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2766 - );
2767 - # finally we can call a user callback and replace piece of text
2768 - $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2769 - $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2770 - $i = $pieceStart + strlen($replaceWith);
 2896+ # Unwind the stack
 2897+ unset( $stack[$stackIndex--] );
 2898+ if ( $stackIndex == -1 ) {
 2899+ $accum =& $topAccum;
 2900+ $findEquals = false;
 2901+ $findPipe = false;
27712902 } else {
2772 - # null value for callback means that parentheses should be parsed, but not replaced
2773 - $i += $matchingCount;
 2903+ $partCount = count( $stack[$stackIndex]['parts'] );
 2904+ $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
 2905+ $findPipe = $stack[$stackIndex]['open'] != "\n";
 2906+ $findEquals = $findPipe && $partCount > 1
 2907+ && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
27742908 }
27752909
2776 - # reset last opening parentheses, but keep it in case there are unused characters
2777 - $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2778 - 'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2779 - 'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2780 - 'title' => '',
2781 - 'parts' => null,
2782 - 'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2783 - $openingBraceStack[$lastOpeningBrace--] = null;
2784 -
 2910+ # Re-add the old stack element if it still has unmatched opening characters remaining
27852911 if ($matchingCount < $piece['count']) {
 2912+ $piece['parts'] = array( '' );
27862913 $piece['count'] -= $matchingCount;
2787 - $piece['startAt'] -= $matchingCount;
2788 - $piece['partStart'] = $piece['startAt'];
 2914+ $piece['eqpos'] = array();
27892915 # do we still qualify for any callback with remaining count?
2790 - $currentCbList = $callbacks[$piece['brace']]['cb'];
 2916+ $names = $rules[$piece['open']]['names'];
 2917+ $skippedBraces = 0;
 2918+ $enclosingAccum =& $accum;
27912919 while ( $piece['count'] ) {
2792 - if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2793 - $lastOpeningBrace++;
2794 - $openingBraceStack[$lastOpeningBrace] = $piece;
 2920+ if ( array_key_exists( $piece['count'], $names ) ) {
 2921+ $stackIndex++;
 2922+ $stack[$stackIndex] = $piece;
 2923+ $accum =& $stack[$stackIndex]['parts'][0];
 2924+ $findEquals = true;
 2925+ $findPipe = true;
27952926 break;
27962927 }
27972928 --$piece['count'];
 2929+ $skippedBraces ++;
27982930 }
 2931+ $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
27992932 }
2800 - } elseif ( $found == 'pipe' ) {
2801 - # lets set a title if it is a first separator, or next part otherwise
2802 - if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2803 - $openingBraceStack[$lastOpeningBrace]['title'] =
2804 - substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2805 - $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2806 - $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2807 - } else {
2808 - $openingBraceStack[$lastOpeningBrace]['parts'][] =
2809 - substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2810 - $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2811 - }
2812 - $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
 2933+
 2934+ # Add XML element to the enclosing accumulator
 2935+ $accum .= $element;
28132936 }
 2937+
 2938+ elseif ( $found == 'pipe' ) {
 2939+ $stack[$stackIndex]['parts'][] = '';
 2940+ $partsCount = count( $stack[$stackIndex]['parts'] );
 2941+ $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
 2942+ $findEquals = true;
 2943+ ++$i;
 2944+ }
 2945+
 2946+ elseif ( $found == 'equals' ) {
 2947+ $findEquals = false;
 2948+ $partsCount = count( $stack[$stackIndex]['parts'] );
 2949+ $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
 2950+ $accum .= '=';
 2951+ ++$i;
 2952+ }
28142953 }
28152954
 2955+ # Output any remaining unclosed brackets
 2956+ foreach ( $stack as $piece ) {
 2957+ if ( $piece['open'] == "\n" ) {
 2958+ $topAccum .= $piece['parts'][0];
 2959+ } else {
 2960+ $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
 2961+ }
 2962+ }
 2963+ $topAccum .= '</root>';
 2964+
 2965+ wfProfileOut( __METHOD__.'-makexml' );
 2966+ wfProfileIn( __METHOD__.'-loadXML' );
 2967+ $dom = new DOMDocument;
 2968+ if ( !$dom->loadXML( $topAccum ) ) {
 2969+ throw new MWException( __METHOD__.' generated invalid XML' );
 2970+ }
 2971+ wfProfileOut( __METHOD__.'-loadXML' );
28162972 wfProfileOut( __METHOD__ );
2817 - return $text;
 2973+ return $dom;
28182974 }
28192975
 2976+ /*
 2977+ * Return a three-element array: leading whitespace, string contents, trailing whitespace
 2978+ */
 2979+ public static function splitWhitespace( $s ) {
 2980+ $ltrimmed = ltrim( $s );
 2981+ $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
 2982+ $trimmed = rtrim( $ltrimmed );
 2983+ $diff = strlen( $ltrimmed ) - strlen( $trimmed );
 2984+ if ( $diff > 0 ) {
 2985+ $w2 = substr( $ltrimmed, -$diff );
 2986+ } else {
 2987+ $w2 = '';
 2988+ }
 2989+ return array( $w1, $trimmed, $w2 );
 2990+ }
 2991+
28202992 /**
28212993 * Replace magic variables, templates, and template arguments
28222994 * with the appropriate text. Templates are substituted recursively,
——@@ -2827,89 +2999,33 @@
28283000 * OT_HTML: all templates and magic variables
28293001 *
28303002 * @param string $tex The text to transform
2831 - * @param array $args Key-value pairs representing template parameters to substitute
 3003+ * @param PPFrame $frame Object describing the arguments passed to the template
28323004 * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
28333005 * @private
28343006 */
2835 - function replaceVariables( $text, $args = array(), $argsOnly = false ) {
 3007+ function replaceVariables( $text, $frame = false, $argsOnly = false ) {
28363008 # Prevent too big inclusions
28373009 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
28383010 return $text;
28393011 }
28403012
2841 - $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
 3013+ $fname = __METHOD__;
28423014 wfProfileIn( $fname );
28433015
2844 - # This function is called recursively. To keep track of arguments we need a stack:
2845 - array_push( $this->mArgStack, $args );
2846 -
2847 - $braceCallbacks = array();
2848 - if ( !$argsOnly ) {
2849 - $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
 3016+ if ( $frame === false ) {
 3017+ $frame = new PPFrame( $this );
 3018+ } elseif ( !( $frame instanceof PPFrame ) ) {
 3019+ throw new MWException( __METHOD__ . ' called using the old argument format' );
28503020 }
2851 - if ( $this->mOutputType != OT_MSG ) {
2852 - $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2853 - }
2854 - if ( $braceCallbacks ) {
2855 - $callbacks = array(
2856 - '{' => array(
2857 - 'end' => '}',
2858 - 'cb' => $braceCallbacks,
2859 - 'min' => $argsOnly ? 3 : 2,
2860 - 'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2861 - ),
2862 - '[' => array(
2863 - 'end' => ']',
2864 - 'cb' => array(2=>null),
2865 - 'min' => 2,
2866 - 'max' => 2,
2867 - )
2868 - );
2869 - $text = $this->replace_callback ($text, $callbacks);
28703021
2871 - array_pop( $this->mArgStack );
2872 - }
2873 - wfProfileOut( $fname );
2874 - return $text;
2875 - }
 3022+ $dom = $this->preprocessToDom( $text );
 3023+ $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
 3024+ $text = $frame->expand( $dom, 0, $flags );
28763025
2877 - /**
2878 - * Replace magic variables
2879 - * @private
2880 - */
2881 - function variableSubstitution( $matches ) {
2882 - global $wgContLang;
2883 - $fname = 'Parser::variableSubstitution';
2884 - $varname = $wgContLang->lc($matches[1]);
2885 - wfProfileIn( $fname );
2886 - $skip = false;
2887 - if ( $this->mOutputType == OT_WIKI ) {
2888 - # Do only magic variables prefixed by SUBST
2889 - $mwSubst =& MagicWord::get( 'subst' );
2890 - if (!$mwSubst->matchStartAndRemove( $varname ))
2891 - $skip = true;
2892 - # Note that if we don't substitute the variable below,
2893 - # we don't remove the {{subst:}} magic word, in case
2894 - # it is a template rather than a magic variable.
2895 - }
2896 - if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2897 - $id = $this->mVariables[$varname];
2898 - # Now check if we did really match, case sensitive or not
2899 - $mw =& MagicWord::get( $id );
2900 - if ($mw->match($matches[1])) {
2901 - $text = $this->getVariableValue( $id );
2902 - $this->mOutput->mContainsOldMagic = true;
2903 - } else {
2904 - $text = $matches[0];
2905 - }
2906 - } else {
2907 - $text = $matches[0];
2908 - }
29093026 wfProfileOut( $fname );
29103027 return $text;
29113028 }
29123029
2913 -
29143030 /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
29153031 static function createAssocArgs( $args ) {
29163032 $assocArgs = array();
——@@ -2941,12 +3057,13 @@
29423058 * $piece['text']: matched text
29433059 * $piece['title']: the title, i.e. the part before the |
29443060 * $piece['parts']: the parameter array
 3061+ * @param PPFrame The current frame, contains template arguments
29453062 * @return string the text of the template
29463063 * @private
29473064 */
2948 - function braceSubstitution( $piece ) {
 3065+ function braceSubstitution( $piece, $frame ) {
29493066 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2950 - $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
 3067+ $fname = __METHOD__;
29513068 wfProfileIn( $fname );
29523069 wfProfileIn( __METHOD__.'-setup' );
29533070
——@@ -2955,33 +3072,23 @@
29563073 $nowiki = false; # wiki markup in $text should be escaped
29573074 $noparse = false; # Unsafe HTML tags should not be stripped, etc.
29583075 $noargs = false; # Don't replace triple-brace arguments in $text
2959 - $replaceHeadings = false; # Make the edit section links go to the template not the article
2960 - $headingOffset = 0; # Skip headings when number, to account for those that weren't transcluded.
29613076 $isHTML = false; # $text is HTML, armour it against wikitext transformation
29623077 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
 3078+ $isDOM = false; # $text is a DOM node needing expansion
29633079
29643080 # Title object, where $text came from
29653081 $title = NULL;
29663082
2967 - $linestart = '';
 3083+ # $part1 is the bit before the first |, and must contain only title characters.
 3084+ # Various prefixes will be stripped from it later.
 3085+ $titleWithSpaces = $frame->expand( $piece['title'] );
 3086+ $part1 = trim( $titleWithSpaces );
 3087+ $titleText = false;
29683088
 3089+ # Original title text preserved for various purposes
 3090+ $originalTitle = $part1;
29693091
2970 - # $part1 is the bit before the first |, and must contain only title characters
2971 - # $args is a list of arguments, starting from index 0, not including $part1
2972 -
2973 - $titleText = $part1 = $piece['title'];
2974 - # If the third subpattern matched anything, it will start with |
2975 -
2976 - if (null == $piece['parts']) {
2977 - $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2978 - if ($replaceWith != $piece['text']) {
2979 - $text = $replaceWith;
2980 - $found = true;
2981 - $noparse = true;
2982 - $noargs = true;
2983 - }
2984 - }
2985 -
 3092+ # $args is a list of argument nodes, starting from index 0, not including $part1
29863093 $args = (null == $piece['parts']) ? array() : $piece['parts'];
29873094 wfProfileOut( __METHOD__.'-setup' );
29883095
——@@ -2994,13 +3101,25 @@
29953102 # 1) Found SUBST but not in the PST phase
29963103 # 2) Didn't find SUBST and in the PST phase
29973104 # In either case, return without further processing
2998 - $text = $piece['text'];
 3105+ $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
29993106 $found = true;
30003107 $noparse = true;
30013108 $noargs = true;
30023109 }
30033110 }
30043111
 3112+ # Variables
 3113+ if ( !$found && $args->length == 0 ) {
 3114+ $id = $this->mVariables->matchStartToEnd( $part1 );
 3115+ if ( $id !== false ) {
 3116+ $text = $this->getVariableValue( $id );
 3117+ $this->mOutput->mContainsOldMagic = true;
 3118+ $found = true;
 3119+ $noparse = true;
 3120+ $noargs = true;
 3121+ }
 3122+ }
 3123+
30053124 # MSG, MSGNW and RAW
30063125 if ( !$found ) {
30073126 # Check for MSGNW:
——@@ -3021,7 +3140,7 @@
30223141 }
30233142 wfProfileOut( __METHOD__.'-modifiers' );
30243143
3025 - //save path level before recursing into functions & templates.
 3144+ # Save path level before recursing into functions & templates.
30263145 $lastPathLevel = $this->mTemplatePath;
30273146
30283147 # Parser functions
——@@ -3044,18 +3163,35 @@
30453164 }
30463165 }
30473166 if ( $function ) {
3048 - $funcArgs = array_map( 'trim', $args );
3049 - $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3050 - $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
 3167+ list( $callback, $flags ) = $this->mFunctionHooks[$function];
 3168+ $initialArgs = array( &$this );
 3169+ $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
 3170+ if ( $flags & SFH_OBJECT_ARGS ) {
 3171+ # Add a frame parameter, and pass the arguments as an array
 3172+ $allArgs = $initialArgs;
 3173+ $allArgs[] = $frame;
 3174+ foreach ( $args as $arg ) {
 3175+ $funcArgs[] = $arg;
 3176+ }
 3177+ $allArgs[] = $funcArgs;
 3178+ } else {
 3179+ # Convert arguments to plain text
 3180+ foreach ( $args as $arg ) {
 3181+ $funcArgs[] = trim( $frame->expand( $arg ) );
 3182+ }
 3183+ $allArgs = array_merge( $initialArgs, $funcArgs );
 3184+ }
 3185+
 3186+ $result = call_user_func_array( $callback, $allArgs );
30513187 $found = true;
30523188
30533189 // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3054 - //$noargs = true;
3055 - //$noparse = true;
 3190+ $noargs = true;
 3191+ $noparse = true;
30563192
30573193 if ( is_array( $result ) ) {
30583194 if ( isset( $result[0] ) ) {
3059 - $text = $linestart . $result[0];
 3195+ $text = $result[0];
30603196 unset( $result[0] );
30613197 }
30623198
——@@ -3063,171 +3199,123 @@
30643200 // This allows callers to set flags such as nowiki, noparse, found, etc.
30653201 extract( $result );
30663202 } else {
3067 - $text = $linestart . $result;
 3203+ $text = $result;
30683204 }
30693205 }
30703206 }
30713207 wfProfileOut( __METHOD__ . '-pfunc' );
30723208 }
30733209
3074 - # Template table test
3075 -
3076 - # Did we encounter this template already? If yes, it is in the cache
3077 - # and we need to check for loops.
3078 - if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3079 - $found = true;
3080 -
3081 - # Infinite loop test
3082 - if ( isset( $this->mTemplatePath[$part1] ) ) {
3083 - $noparse = true;
3084 - $noargs = true;
3085 - $found = true;
3086 - $text = $linestart .
3087 - "[[$part1]]<!-- WARNING: template loop detected -->";
3088 - wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3089 - } else {
3090 - # set $text to cached message.
3091 - $text = $linestart . $this->mTemplates[$piece['title']];
3092 - #treat title for cached page the same as others
3093 - $ns = NS_TEMPLATE;
3094 - $subpage = '';
3095 - $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3096 - if ($subpage !== '') {
3097 - $ns = $this->mTitle->getNamespace();
3098 - }
3099 - $title = Title::newFromText( $part1, $ns );
3100 - //used by include size checking
3101 - $titleText = $title->getPrefixedText();
3102 - //used by edit section links
3103 - $replaceHeadings = true;
3104 -
3105 - }
3106 - }
3107 -
3108 - # Load from database
 3210+ # Finish mangling title and then check for loops.
 3211+ # Set $title to a Title object and $titleText to the PDBK
31093212 if ( !$found ) {
3110 - wfProfileIn( __METHOD__ . '-loadtpl' );
31113213 $ns = NS_TEMPLATE;
3112 - # declaring $subpage directly in the function call
3113 - # does not work correctly with references and breaks
3114 - # {{/subpage}}-style inclusions
 3214+ # Split the title into page and subpage
31153215 $subpage = '';
31163216 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
31173217 if ($subpage !== '') {
31183218 $ns = $this->mTitle->getNamespace();
31193219 }
31203220 $title = Title::newFromText( $part1, $ns );
3121 -
3122 -
3123 - if ( !is_null( $title ) ) {
 3221+ if ( $title ) {
31243222 $titleText = $title->getPrefixedText();
31253223 # Check for language variants if the template is not found
31263224 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
31273225 $wgContLang->findVariantLink($part1, $title);
31283226 }
 3227+ # Do infinite loop check
 3228+ if ( isset( $this->mTemplatePath[$titleText] ) ) {
 3229+ $noparse = true;
 3230+ $noargs = true;
 3231+ $found = true;
 3232+ $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
 3233+ wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
 3234+ }
 3235+ }
 3236+ }
31293237
3130 - if ( !$title->isExternal() ) {
3131 - if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3132 - $text = SpecialPage::capturePath( $title );
3133 - if ( is_string( $text ) ) {
3134 - $found = true;
3135 - $noparse = true;
3136 - $noargs = true;
3137 - $isHTML = true;
3138 - $this->disableCache();
3139 - }
3140 - } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3141 - $found = false; //access denied
3142 - wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3143 - } else {
3144 - list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3145 - if ( $articleContent !== false ) {
3146 - $found = true;
3147 - $text = $articleContent;
3148 - $replaceHeadings = true;
3149 - }
 3238+ # Load from database
 3239+ if ( !$found && $title ) {
 3240+ wfProfileIn( __METHOD__ . '-loadtpl' );
 3241+ if ( !$title->isExternal() ) {
 3242+ if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
 3243+ $text = SpecialPage::capturePath( $title );
 3244+ if ( is_string( $text ) ) {
 3245+ $found = true;
 3246+ $noparse = true;
 3247+ $noargs = true;
 3248+ $isHTML = true;
 3249+ $this->disableCache();
31503250 }
3151 -
3152 - # If the title is valid but undisplayable, make a link to it
3153 - if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3154 - $text = "[[:$titleText]]";
 3251+ } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
 3252+ $found = false; //access denied
 3253+ wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
 3254+ } else {
 3255+ list( $text, $title ) = $this->getTemplateDom( $title );
 3256+ if ( $text !== false ) {
31553257 $found = true;
 3258+ $isDOM = true;
31563259 }
3157 - } elseif ( $title->isTrans() ) {
3158 - // Interwiki transclusion
3159 - if ( $this->ot['html'] && !$forceRawInterwiki ) {
3160 - $text = $this->interwikiTransclude( $title, 'render' );
3161 - $isHTML = true;
3162 - $noparse = true;
3163 - } else {
3164 - $text = $this->interwikiTransclude( $title, 'raw' );
3165 - $replaceHeadings = true;
3166 - }
3167 - $found = true;
31683260 }
31693261
3170 - # Template cache array insertion
3171 - # Use the original $piece['title'] not the mangled $part1, so that
3172 - # modifiers such as RAW: produce separate cache entries
3173 - if( $found ) {
3174 - if( $isHTML ) {
3175 - // A special page; don't store it in the template cache.
3176 - } else {
3177 - $this->mTemplates[$piece['title']] = $text;
3178 - }
3179 - $text = $linestart . $text;
 3262+ # If the title is valid but undisplayable, make a link to it
 3263+ if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
 3264+ $text = "[[:$titleText]]";
 3265+ $found = true;
31803266 }
 3267+ } elseif ( $title->isTrans() ) {
 3268+ // Interwiki transclusion
 3269+ if ( $this->ot['html'] && !$forceRawInterwiki ) {
 3270+ $text = $this->interwikiTransclude( $title, 'render' );
 3271+ $isHTML = true;
 3272+ $noparse = true;
 3273+ } else {
 3274+ $text = $this->interwikiTransclude( $title, 'raw' );
 3275+ }
 3276+ $found = true;
31813277 }
31823278 wfProfileOut( __METHOD__ . '-loadtpl' );
31833279 }
31843280
3185 - if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3186 - # Error, oversize inclusion
3187 - $text = $linestart .
3188 - "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3189 - $noparse = true;
3190 - $noargs = true;
3191 - }
3192 -
31933281 # Recursive parsing, escaping and link table handling
31943282 # Only for HTML output
31953283 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
 3284+ if ( $isDOM ) {
 3285+ $text = $frame->expand( $text );
 3286+ }
31963287 $text = wfEscapeWikiText( $text );
31973288 } elseif ( !$this->ot['msg'] && $found ) {
31983289 if ( $noargs ) {
3199 - $assocArgs = array();
 3290+ $newFrame = $frame->newChild();
32003291 } else {
32013292 # Clean up argument array
3202 - $assocArgs = self::createAssocArgs($args);
 3293+ $newFrame = $frame->newChild( $args, $title );
32033294 # Add a new element to the templace recursion path
3204 - $this->mTemplatePath[$part1] = 1;
 3295+ $this->mTemplatePath[$titleText] = 1;
32053296 }
32063297
32073298 if ( !$noparse ) {
3208 - # If there are any <onlyinclude> tags, only include them
3209 - if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3210 - $replacer = new OnlyIncludeReplacer;
3211 - StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3212 - array( &$replacer, 'replace' ), $text );
3213 - $text = $replacer->output;
3214 - }
3215 - # Remove <noinclude> sections and <includeonly> tags
3216 - $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3217 - $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3218 -
3219 - if( $this->ot['html'] || $this->ot['pre'] ) {
3220 - # Strip <nowiki>, <pre>, etc.
3221 - $text = $this->strip( $text, $this->mStripState );
3222 - if ( $this->ot['html'] ) {
3223 - $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3224 - } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3225 - $text = Sanitizer::removeHTMLcomments( $text );
 3299+ if ( $isDOM ) {
 3300+ if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
 3301+ # Expansion is eligible for the empty-frame cache
 3302+ if ( isset( $this->mTplExpandCache[$titleText] ) ) {
 3303+ $text = $this->mTplExpandCache[$titleText];
 3304+ } else {
 3305+ $text = $newFrame->expand( $text );
 3306+ $this->mTplExpandCache[$titleText] = $text;
 3307+ }
 3308+ } else {
 3309+ $text = $newFrame->expand( $text );
32263310 }
 3311+ } else {
 3312+ $text = $this->replaceVariables( $text, $newFrame );
32273313 }
3228 - $text = $this->replaceVariables( $text, $assocArgs );
32293314
3230 - # If the template begins with a table or block-level
 3315+ # strip woz 'ere 2004-07
 3316+
 3317+ # Bug 529: if the template begins with a table or block-level
32313318 # element, it should be treated as beginning a new line.
 3319+ # This behaviour is somewhat controversial.
32323320 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
32333321 $text = "\n" . $text;
32343322 }
——@@ -3235,59 +3323,39 @@
32363324 # $noparse and !$noargs
32373325 # Just replace the arguments, not any double-brace items
32383326 # This is used for rendered interwiki transclusion
3239 - $text = $this->replaceVariables( $text, $assocArgs, true );
 3327+ if ( $isDOM ) {
 3328+ $text = $newFrame->expand( $text, 0, PPFrame::NO_TEMPLATES );
 3329+ } else {
 3330+ $text = $this->replaceVariables( $text, $newFrame, true );
 3331+ }
 3332+ } elseif ( $isDOM ) {
 3333+ $text = $frame->expand( $text );
32403334 }
 3335+ } elseif ( $isDOM ) {
 3336+ $text = $frame->expand( $text, 0, PPFrame::NO_TEMPLATES | PPFrame::NO_ARGS );
32413337 }
 3338+
32423339 # Prune lower levels off the recursion check path
32433340 $this->mTemplatePath = $lastPathLevel;
32443341
32453342 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
32463343 # Error, oversize inclusion
3247 - $text = $linestart .
3248 - "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
 3344+ $text = "[[$originalTitle]]" .
 3345+ $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
32493346 $noparse = true;
32503347 $noargs = true;
32513348 }
32523349
32533350 if ( !$found ) {
32543351 wfProfileOut( $fname );
3255 - return $piece['text'];
 3352+ return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
32563353 } else {
32573354 wfProfileIn( __METHOD__ . '-placeholders' );
32583355 if ( $isHTML ) {
32593356 # Replace raw HTML by a placeholder
32603357 # Add a blank line preceding, to prevent it from mucking up
32613358 # immediately preceding headings
3262 - $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3263 - } else {
3264 - # replace ==section headers==
3265 - # XXX this needs to go away once we have a better parser.
3266 - if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3267 - if( !is_null( $title ) )
3268 - $encodedname = base64_encode($title->getPrefixedDBkey());
3269 - else
3270 - $encodedname = base64_encode("");
3271 - $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3272 - PREG_SPLIT_DELIM_CAPTURE);
3273 - $text = '';
3274 - $nsec = $headingOffset;
3275 -
3276 - for( $i = 0; $i < count($m); $i += 2 ) {
3277 - $text .= $m[$i];
3278 - if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3279 - $hl = $m[$i + 1];
3280 - if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3281 - $text .= $hl;
3282 - continue;
3283 - }
3284 - $m2 = array();
3285 - preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3286 - $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3287 - . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3288 -
3289 - $nsec++;
3290 - }
3291 - }
 3359+ $text = "\n\n" . $this->insertStripItem( $text );
32923360 }
32933361 wfProfileOut( __METHOD__ . '-placeholders' );
32943362 }
——@@ -3297,7 +3365,7 @@
32983366
32993367 if ( !$found ) {
33003368 wfProfileOut( $fname );
3301 - return $piece['text'];
 3369+ return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
33023370 } else {
33033371 wfProfileOut( $fname );
33043372 return $text;
——@@ -3305,27 +3373,99 @@
33063374 }
33073375
33083376 /**
 3377+ * Get the semi-parsed DOM representation of a template with a given title,
 3378+ * and its redirect destination title. Cached.
 3379+ */
 3380+ function getTemplateDom( $title ) {
 3381+ $titleText = $title->getPrefixedDBkey();
 3382+
 3383+ if ( isset( $this->mTplRedirCache[$titleText] ) ) {
 3384+ list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
 3385+ $title = Title::makeTitle( $ns, $dbk );
 3386+ $titleText = $title->getPrefixedDBkey();
 3387+ }
 3388+ if ( isset( $this->mTplDomCache[$titleText] ) ) {
 3389+ return array( $this->mTplDomCache[$titleText], $title );
 3390+ }
 3391+
 3392+ // Cache miss, go to the database
 3393+ list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
 3394+
 3395+ if ( $text === false ) {
 3396+ $this->mTplDomCache[$titleText] = false;
 3397+ return array( false, $title );
 3398+ }
 3399+
 3400+ # If there are any <onlyinclude> tags, only include them
 3401+ if ( !$this->ot['msg'] ) {
 3402+ if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
 3403+ $replacer = new OnlyIncludeReplacer;
 3404+ StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
 3405+ array( &$replacer, 'replace' ), $text );
 3406+ $text = $replacer->output;
 3407+ }
 3408+ # Remove <noinclude> sections and <includeonly> tags
 3409+ $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
 3410+ $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
 3411+ }
 3412+
 3413+ $dom = $this->preprocessToDom( $text );
 3414+ $this->mTplDomCache[$titleText] = $dom;
 3415+ return array( $dom, $title );
 3416+ }
 3417+
 3418+ /**
33093419 * Fetch the unparsed text of a template and register a reference to it.
33103420 */
3311 - function fetchTemplateAndtitle( $title ) {
 3421+ function fetchTemplateAndTitle( $title ) {
 3422+ $templateCb = $this->mOptions->getTemplateCallback();
 3423+ $stuff = call_user_func( $templateCb, $title );
 3424+ $text = $stuff['text'];
 3425+ $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
 3426+ if ( isset( $stuff['deps'] ) ) {
 3427+ foreach ( $stuff['deps'] as $dep ) {
 3428+ $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
 3429+ }
 3430+ }
 3431+ return array($text,$finalTitle);
 3432+ }
 3433+
 3434+ function fetchTemplate( $title ) {
 3435+ $rv = $this->fetchTemplateAndTitle($title);
 3436+ return $rv[0];
 3437+ }
 3438+
 3439+ /**
 3440+ * Static function to get a template
 3441+ * Can be overridden via ParserOptions::setTemplateCallback().
 3442+ */
 3443+ static function statelessFetchTemplate( $title ) {
33123444 $text = $skip = false;
33133445 $finalTitle = $title;
 3446+ $deps = array();
 3447+
33143448 // Loop to fetch the article, with up to 1 redirect
33153449 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
33163450 # Give extensions a chance to select the revision instead
33173451 $id = false; // Assume current
3318 - wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( &$this, &$title, &$skip, &$id ) );
 3452+ wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
33193453
33203454 if( $skip ) {
33213455 $text = false;
3322 - $this->mOutput->addTemplate( $title, $title->getArticleID(), null );
 3456+ $deps[] = array(
 3457+ 'title' => $title,
 3458+ 'page_id' => $title->getArticleID(),
 3459+ 'rev_id' => null );
33233460 break;
33243461 }
33253462 $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
33263463 $rev_id = $rev ? $rev->getId() : 0;
3327 -
3328 - $this->mOutput->addTemplate( $title, $title->getArticleID(), $rev_id );
3329 -
 3464+
 3465+ $deps[] = array(
 3466+ 'title' => $title,
 3467+ 'page_id' => $title->getArticleID(),
 3468+ 'rev_id' => $rev_id );
 3469+
33303470 if( $rev ) {
33313471 $text = $rev->getText();
33323472 } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
——@@ -3346,14 +3486,12 @@
33473487 $finalTitle = $title;
33483488 $title = Title::newFromRedirect( $text );
33493489 }
3350 - return array($text,$finalTitle);
 3490+ return array(
 3491+ 'text' => $text,
 3492+ 'finalTitle' => $finalTitle,
 3493+ 'deps' => $deps );
33513494 }
33523495
3353 - function fetchTemplate( $title ) {
3354 - $rv = $this->fetchTemplateAndtitle($title);
3355 - return $rv[0];
3356 - }
3357 -
33583496 /**
33593497 * Transclude an interwiki link.
33603498 */
——@@ -3400,26 +3538,111 @@
34013539 * Triple brace replacement -- used for template arguments
34023540 * @private
34033541 */
3404 - function argSubstitution( $matches ) {
3405 - $arg = trim( $matches['title'] );
3406 - $text = $matches['text'];
3407 - $inputArgs = end( $this->mArgStack );
 3542+ function argSubstitution( $piece, $frame ) {
 3543+ wfProfileIn( __METHOD__ );
34083544
3409 - if ( array_key_exists( $arg, $inputArgs ) ) {
3410 - $text = $inputArgs[$arg];
3411 - } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3412 - null != $matches['parts'] && count($matches['parts']) > 0) {
3413 - $text = $matches['parts'][0];
 3545+ $text = false;
 3546+ $error = false;
 3547+ $parts = $piece['parts'];
 3548+ $argWithSpaces = $frame->expand( $piece['title'] );
 3549+ $arg = trim( $argWithSpaces );
 3550+
 3551+ if ( isset( $frame->args[$arg] ) ) {
 3552+ $text = $frame->parent->expand( $frame->args[$arg] );
 3553+ } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
 3554+ $text = $frame->expand( $parts->item( 0 ) );
34143555 }
34153556 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3416 - $text = $matches['text'] .
3417 - '<!-- WARNING: argument omitted, expansion size too large -->';
 3557+ $error = '<!-- WARNING: argument omitted, expansion size too large -->';
34183558 }
34193559
 3560+ if ( $text === false ) {
 3561+ $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
 3562+ }
 3563+ if ( $error !== false ) {
 3564+ $text .= $error;
 3565+ }
 3566+
 3567+ wfProfileOut( __METHOD__ );
34203568 return $text;
34213569 }
34223570
34233571 /**
 3572+ * Return the text to be used for a given extension tag.
 3573+ * This is the ghost of strip().
 3574+ *
 3575+ * @param array $params Associative array of parameters:
 3576+ * name DOMNode for the tag name
 3577+ * attrText DOMNode for unparsed text where tag attributes are thought to be
 3578+ * inner Contents of extension element
 3579+ * noClose Original text did not have a close tag
 3580+ * @param PPFrame $frame
 3581+ */
 3582+ function extensionSubstitution( $params, $frame ) {
 3583+ global $wgRawHtml, $wgContLang;
 3584+ static $n = 1;
 3585+
 3586+ $name = $frame->expand( $params['name'] );
 3587+ $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
 3588+ $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
 3589+
 3590+ $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 3591+
 3592+ if ( $this->ot['html'] ) {
 3593+ if ( $name == '!--' ) {
 3594+ return '';
 3595+ }
 3596+ $name = strtolower( $name );
 3597+
 3598+ $params = Sanitizer::decodeTagAttributes( $attrText );
 3599+ switch ( $name ) {
 3600+ case 'html':
 3601+ if( $wgRawHtml ) {
 3602+ $output = $content;
 3603+ break;
 3604+ } else {
 3605+ throw new MWException( '<html> extension tag encountered unexpectedly' );
 3606+ }
 3607+ case 'nowiki':
 3608+ $output = Xml::escapeTagsOnly( $content );
 3609+ break;
 3610+ case 'math':
 3611+ $output = $wgContLang->armourMath(
 3612+ MathRenderer::renderMath( $content, $params ) );
 3613+ break;
 3614+ case 'gallery':
 3615+ $output = $this->renderImageGallery( $content, $params );
 3616+ break;
 3617+ default:
 3618+ if( isset( $this->mTagHooks[$name] ) ) {
 3619+ $output = call_user_func_array( $this->mTagHooks[$name],
 3620+ array( $content, $params, $this ) );
 3621+ } else {
 3622+ throw new MWException( "Invalid call hook $name" );
 3623+ }
 3624+ }
 3625+ } else {
 3626+ if ( $name == '!--' ) {
 3627+ $output = '<!--' . $content . '-->';
 3628+ } else {
 3629+ if ( $content === null ) {
 3630+ $output = "<$name$attrText/>";
 3631+ } else {
 3632+ $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
 3633+ $output = "<$name$attrText>$content$close";
 3634+ }
 3635+ }
 3636+ }
 3637+
 3638+ if ( $name == 'html' || $name == 'nowiki' ) {
 3639+ $this->mStripState->nowiki->setPair( $marker, $output );
 3640+ } else {
 3641+ $this->mStripState->general->setPair( $marker, $output );
 3642+ }
 3643+ return $marker;
 3644+ }
 3645+
 3646+ /**
34243647 * Increment an include size counter
34253648 *
34263649 * @param string $type The type of expansion
——@@ -3427,7 +3650,7 @@
34283651 * @return boolean False if this inclusion would take it over the maximum, true otherwise
34293652 */
34303653 function incrementIncludeSize( $type, $size ) {
3431 - if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
 3654+ if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
34323655 return false;
34333656 } else {
34343657 $this->mIncludeSizes[$type] += $size;
——@@ -3535,7 +3758,6 @@
35363759
35373760 # headline counter
35383761 $headlineCount = 0;
3539 - $sectionCount = 0; # headlineCount excluding template sections
35403762 $numVisible = 0;
35413763
35423764 # Ugh .. the TOC should have neat indentation levels which can be
——@@ -3550,18 +3772,20 @@
35513773 $prevlevel = 0;
35523774 $toclevel = 0;
35533775 $prevtoclevel = 0;
 3776+ $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
 3777+ $baseTitleText = $this->mTitle->getPrefixedDBkey();
35543778
35553779 foreach( $matches[3] as $headline ) {
3556 - $istemplate = 0;
3557 - $templatetitle = '';
3558 - $templatesection = 0;
 3780+ $isTemplate = false;
 3781+ $titleText = false;
 3782+ $sectionIndex = false;
35593783 $numbering = '';
3560 - $mat = array();
3561 - if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3562 - $istemplate = 1;
3563 - $templatetitle = base64_decode($mat[1]);
3564 - $templatesection = 1 + (int)base64_decode($mat[2]);
3565 - $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
 3784+ $markerMatches = array();
 3785+ if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
 3786+ $serial = $markerMatches[1];
 3787+ list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
 3788+ $isTemplate = ($titleText != $baseTitleText);
 3789+ $headline = preg_replace("/^$markerRegex/", "", $headline);
35663790 }
35673791
35683792 if( $toclevel ) {
——@@ -3634,41 +3858,41 @@
36353859 }
36363860 }
36373861
3638 - # The canonized header is a version of the header text safe to use for links
 3862+ # The safe header is a version of the header text safe to use for links
36393863 # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3640 - $canonized_headline = $this->mStripState->unstripBoth( $headline );
 3864+ $safeHeadline = $this->mStripState->unstripBoth( $headline );
36413865
36423866 # Remove link placeholders by the link text.
36433867 # <!--LINK number-->
36443868 # turns into
36453869 # link text with suffix
3646 - $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
 3870+ $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
36473871 "\$this->mLinkHolders['texts'][\$1]",
3648 - $canonized_headline );
3649 - $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
 3872+ $safeHeadline );
 3873+ $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
36503874 "\$this->mInterwikiLinkHolders['texts'][\$1]",
3651 - $canonized_headline );
 3875+ $safeHeadline );
36523876
36533877 # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
36543878 $tocline = preg_replace(
36553879 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
36563880 array( '', '<$1>'),
3657 - $canonized_headline
 3881+ $safeHeadline
36583882 );
36593883 $tocline = trim( $tocline );
36603884
36613885 # For the anchor, strip out HTML-y stuff period
3662 - $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
3663 - $canonized_headline = trim( $canonized_headline );
 3886+ $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
 3887+ $safeHeadline = trim( $safeHeadline );
36643888
36653889 # Save headline for section edit hint before it's escaped
3666 - $headline_hint = $canonized_headline;
3667 - $canonized_headline = Sanitizer::escapeId( $canonized_headline );
3668 - $refers[$headlineCount] = $canonized_headline;
 3890+ $headlineHint = $safeHeadline;
 3891+ $safeHeadline = Sanitizer::escapeId( $safeHeadline );
 3892+ $refers[$headlineCount] = $safeHeadline;
36693893
36703894 # count how many in assoc. array so we can track dupes in anchors
3671 - isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3672 - $refcount[$headlineCount]=$refers[$canonized_headline];
 3895+ isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
 3896+ $refcount[$headlineCount] = $refers[$safeHeadline];
36733897
36743898 # Don't number the heading if it is the only one (looks silly)
36753899 if( $doNumberHeadings && count( $matches[3] ) > 1) {
——@@ -3677,7 +3901,7 @@
36783902 }
36793903
36803904 # Create the anchor for linking from the TOC to the section
3681 - $anchor = $canonized_headline;
 3905+ $anchor = $safeHeadline;
36823906 if($refcount[$headlineCount] > 1 ) {
36833907 $anchor .= '_' . $refcount[$headlineCount];
36843908 }
——@@ -3685,19 +3909,17 @@
36863910 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
36873911 }
36883912 # give headline the correct <h#> tag
3689 - if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3690 - if( $istemplate )
3691 - $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
 3913+ if( $showEditLink && $sectionIndex !== false ) {
 3914+ if( $isTemplate )
 3915+ $editlink = $sk->editSectionLinkForOther($titleText, $sectionIndex);
36923916 else
3693 - $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
 3917+ $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
36943918 } else {
36953919 $editlink = '';
36963920 }
36973921 $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
36983922
36993923 $headlineCount++;
3700 - if( !$istemplate )
3701 - $sectionCount++;
37023924 }
37033925
37043926 # Never ever show TOC if no headers
——@@ -3765,14 +3987,12 @@
37663988 $this->clearState();
37673989 }
37683990
3769 - $stripState = new StripState;
37703991 $pairs = array(
37713992 "\r\n" => "\n",
37723993 );
37733994 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3774 - $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3775 - $text = $this->pstPass2( $text, $stripState, $user );
3776 - $text = $stripState->unstripBoth( $text );
 3995+ $text = $this->pstPass2( $text, $user );
 3996+ $text = $this->mStripState->unstripBoth( $text );
37773997 return $text;
37783998 }
37793999
——@@ -3780,7 +4000,7 @@
37814001 * Pre-save transform helper function
37824002 * @private
37834003 */
3784 - function pstPass2( $text, &$stripState, $user ) {
 4004+ function pstPass2( $text, $user ) {
37854005 global $wgContLang, $wgLocaltimezone;
37864006
37874007 /* Note: This is the timestamp saved as hardcoded wikitext to
——@@ -3803,7 +4023,7 @@
38044024 $text = $this->replaceVariables( $text );
38054025
38064026 # Strip out <nowiki> etc. added via replaceVariables
3807 - $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
 4027+ #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
38084028
38094029 # Signatures
38104030 $sigText = $this->getUserSig( $user );
——@@ -3950,6 +4170,11 @@
39514171 /**
39524172 * Transform a MediaWiki message by replacing magic variables.
39534173 *
 4174+ * For some unknown reason, it also expands templates, but only to the
 4175+ * first recursion level. This is wrong and broken, probably introduced
 4176+ * accidentally during refactoring, but probably relied upon by thousands
 4177+ * of users.
 4178+ *
39544179 * @param string $text the text to transform
39554180 * @param ParserOptions $options options
39564181 * @return string the text with variables substituted
——@@ -4003,6 +4228,7 @@
40044229 $tag = strtolower( $tag );
40054230 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
40064231 $this->mTagHooks[$tag] = $callback;
 4232+ $this->mStripList[] = $tag;
40074233
40084234 return $oldVal;
40094235 }
——@@ -4040,8 +4266,8 @@
40414267 * @return The old callback function for this name, if any
40424268 */
40434269 function setFunctionHook( $id, $callback, $flags = 0 ) {
4044 - $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4045 - $this->mFunctionHooks[$id] = $callback;
 4270+ $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
 4271+ $this->mFunctionHooks[$id] = array( $callback, $flags );
40464272
40474273 # Add to function cache
40484274 $mw = MagicWord::get( $id );
——@@ -4660,12 +4886,12 @@
46614887 * Callback from the Sanitizer for expanding items found in HTML attribute
46624888 * values, so they can be safely tested and escaped.
46634889 * @param string $text
4664 - * @param array $args
 4890+ * @param PPFrame $frame
46654891 * @return string
46664892 * @private
46674893 */
4668 - function attributeStripCallback( &$text, $args ) {
4669 - $text = $this->replaceVariables( $text, $args );
 4894+ function attributeStripCallback( &$text, $frame = false ) {
 4895+ $text = $this->replaceVariables( $text, $frame );
46704896 $text = $this->mStripState->unstripBoth( $text );
46714897 return $text;
46724898 }
——@@ -4696,120 +4922,94 @@
46974923 * @param $text Page wikitext
46984924 * @param $section Numbered section. 0 pulls the text before the first
46994925 * heading; other numbers will pull the given section
4700 - * along with its lower-level subsections.
 4926+ * along with its lower-level subsections. If the section is
 4927+ * not found, $mode=get will return $newtext, and
 4928+ * $mode=replace will return $text.
47014929 * @param $mode One of "get" or "replace"
4702 - * @param $newtext Replacement text for section data.
 4930+ * @param $newText Replacement text for section data.
47034931 * @return string for "get", the extracted section text.
47044932 * for "replace", the whole page with the section replaced.
47054933 */
4706 - private function extractSections( $text, $section, $mode, $newtext='' ) {
4707 - # I.... _hope_ this is right.
4708 - # Otherwise, sometimes we don't have things initialized properly.
 4934+ private function extractSections( $text, $section, $mode, $newText='' ) {
47094935 $this->clearState();
4710 -
4711 - # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4712 - # comments to be stripped as well)
4713 - $stripState = new StripState;
4714 -
4715 - $oldOutputType = $this->mOutputType;
4716 - $oldOptions = $this->mOptions;
4717 - $this->mOptions = new ParserOptions();
 4936+ $this->mOptions = new ParserOptions;
47184937 $this->setOutputType( OT_WIKI );
 4938+ $curIndex = 0;
 4939+ $outText = '';
 4940+ $frame = new PPFrame( $this );
47194941
4720 - $striptext = $this->strip( $text, $stripState, true );
 4942+ // Preprocess the text
 4943+ $dom = $this->preprocessToDom( $text );
 4944+ $root = $dom->documentElement;
47214945
4722 - $this->setOutputType( $oldOutputType );
4723 - $this->mOptions = $oldOptions;
 4946+ // <h> nodes indicate section breaks
 4947+ // They can only occur at the top level, so we can find them by iterating the root's children
 4948+ $node = $root->firstChild;
47244949
4725 - # now that we can be sure that no pseudo-sections are in the source,
4726 - # split it up by section
4727 - $uniq = preg_quote( $this->uniqPrefix(), '/' );
4728 - $comment = "(?:$uniq-!--.*?QINU\x07)";
4729 - $secs = preg_split(
4730 - "/
4731 - (
4732 - ^
4733 - (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4734 - (=+) # Should this be limited to 6?
4735 - .+? # Section title...
4736 - \\2 # Ending = count must match start
4737 - (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4738 - $
4739 - |
4740 - <h([1-6])\b.*?>
4741 - .*?
4742 - <\/h\\3\s*>
4743 - )
4744 - /mix",
4745 - $striptext, -1,
4746 - PREG_SPLIT_DELIM_CAPTURE);
4747 -
4748 - if( $mode == "get" ) {
4749 - if( $section == 0 ) {
4750 - // "Section 0" returns the content before any other section.
4751 - $rv = $secs[0];
4752 - } else {
4753 - //track missing section, will replace if found.
4754 - $rv = $newtext;
 4950+ // Find the target section
 4951+ if ( $section == 0 ) {
 4952+ // Section zero doesn't nest, level=big
 4953+ $targetLevel = 1000;
 4954+ } else {
 4955+ while ( $node ) {
 4956+ if ( $node->nodeName == 'h' ) {
 4957+ if ( $curIndex + 1 == $section ) {
 4958+ break;
 4959+ }
 4960+ $curIndex++;
 4961+ }
 4962+ if ( $mode == 'replace' ) {
 4963+ $outText .= $frame->expand( $node );
 4964+ }
 4965+ $node = $node->nextSibling;
47554966 }
4756 - } elseif( $mode == "replace" ) {
4757 - if( $section == 0 ) {
4758 - $rv = $newtext . "\n\n";
4759 - $remainder = true;
4760 - } else {
4761 - $rv = $secs[0];
4762 - $remainder = false;
 4967+ if ( $node ) {
 4968+ $targetLevel = $node->getAttribute( 'level' );
47634969 }
47644970 }
4765 - $count = 0;
4766 - $sectionLevel = 0;
4767 - for( $index = 1; $index < count( $secs ); ) {
4768 - $headerLine = $secs[$index++];
4769 - if( $secs[$index] ) {
4770 - // A wiki header
4771 - $headerLevel = strlen( $secs[$index++] );
 4971+
 4972+ if ( !$node ) {
 4973+ // Not found
 4974+ if ( $mode == 'get' ) {
 4975+ return $newText;
47724976 } else {
4773 - // An HTML header
4774 - $index++;
4775 - $headerLevel = intval( $secs[$index++] );
 4977+ return $text;
47764978 }
4777 - $content = $secs[$index++];
 4979+ }
47784980
4779 - $count++;
4780 - if( $mode == "get" ) {
4781 - if( $count == $section ) {
4782 - $rv = $headerLine . $content;
4783 - $sectionLevel = $headerLevel;
4784 - } elseif( $count > $section ) {
4785 - if( $sectionLevel && $headerLevel > $sectionLevel ) {
4786 - $rv .= $headerLine . $content;
4787 - } else {
4788 - // Broke out to a higher-level section
4789 - break;
4790 - }
 4981+ // Find the end of the section, including nested sections
 4982+ do {
 4983+ if ( $node->nodeName == 'h' ) {
 4984+ $curIndex++;
 4985+ $curLevel = $node->getAttribute( 'level' );
 4986+ if ( $curIndex != $section && $curLevel <= $targetLevel ) {
 4987+ break;
47914988 }
4792 - } elseif( $mode == "replace" ) {
4793 - if( $count < $section ) {
4794 - $rv .= $headerLine . $content;
4795 - } elseif( $count == $section ) {
4796 - $rv .= $newtext . "\n\n";
4797 - $sectionLevel = $headerLevel;
4798 - } elseif( $count > $section ) {
4799 - if( $headerLevel <= $sectionLevel ) {
4800 - // Passed the section's sub-parts.
4801 - $remainder = true;
4802 - }
4803 - if( $remainder ) {
4804 - $rv .= $headerLine . $content;
4805 - }
4806 - }
48074989 }
 4990+ if ( $mode == 'get' ) {
 4991+ $outText .= $frame->expand( $node );
 4992+ }
 4993+ $node = $node->nextSibling;
 4994+ } while ( $node );
 4995+
 4996+ // Write out the remainder (in replace mode only)
 4997+ if ( $mode == 'replace' ) {
 4998+ // Output the replacement text
 4999+ // Add two newlines on -- trailing whitespace in $newText is conventionally
 5000+ // stripped by the editor, so we need both newlines to restore the paragraph gap
 5001+ $outText .= $newText . "\n\n";
 5002+ while ( $node ) {
 5003+ $outText .= $frame->expand( $node );
 5004+ $node = $node->nextSibling;
 5005+ }
48085006 }
4809 - if (is_string($rv))
4810 - # reinsert stripped tags
4811 - $rv = trim( $stripState->unstripBoth( $rv ) );
48125007
4813 - return $rv;
 5008+ if ( is_string( $outText ) ) {
 5009+ // Re-insert stripped tags
 5010+ $outText = trim( $this->mStripState->unstripBoth( $outText ) );
 5011+ }
 5012+
 5013+ return $outText;
48145014 }
48155015
48165016 /**
——@@ -4942,6 +5142,15 @@
49435143 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
49445144 return $text;
49455145 }
 5146+
 5147+ /**
 5148+ * strip/replaceVariables/unstrip for preprocessor regression testing
 5149+ */
 5150+ function srvus( $text ) {
 5151+ $text = $this->replaceVariables( $text );
 5152+ $text = $this->mStripState->unstripBoth( $text );
 5153+ return $text;
 5154+ }
49465155 }
49475156
49485157 /**
——@@ -4974,23 +5183,265 @@
49755184
49765185 function unstripGeneral( $text ) {
49775186 wfProfileIn( __METHOD__ );
4978 - $text = $this->general->replace( $text );
 5187+ do {
 5188+ $oldText = $text;
 5189+ $text = $this->general->replace( $text );
 5190+ } while ( $text != $oldText );
49795191 wfProfileOut( __METHOD__ );
49805192 return $text;
49815193 }
49825194
49835195 function unstripNoWiki( $text ) {
49845196 wfProfileIn( __METHOD__ );
4985 - $text = $this->nowiki->replace( $text );
 5197+ do {
 5198+ $oldText = $text;
 5199+ $text = $this->nowiki->replace( $text );
 5200+ } while ( $text != $oldText );
49865201 wfProfileOut( __METHOD__ );
49875202 return $text;
49885203 }
49895204
49905205 function unstripBoth( $text ) {
49915206 wfProfileIn( __METHOD__ );
4992 - $text = $this->general->replace( $text );
4993 - $text = $this->nowiki->replace( $text );
 5207+ do {
 5208+ $oldText = $text;
 5209+ $text = $this->general->replace( $text );
 5210+ $text = $this->nowiki->replace( $text );
 5211+ } while ( $text != $oldText );
49945212 wfProfileOut( __METHOD__ );
49955213 return $text;
49965214 }
49975215 }
 5216+
 5217+/**
 5218+ * An expansion frame, used as a context to expand the result of preprocessToDom()
 5219+ */
 5220+class PPFrame {
 5221+ var $parser, $title;
 5222+
 5223+ const NO_ARGS = 1;
 5224+ const NO_TEMPLATES = 2;
 5225+
 5226+ /**
 5227+ * Construct a new preprocessor frame.
 5228+ * @param Parser $parser The parent parser
 5229+ * @param Title $title The context title, or false if there isn't one
 5230+ */
 5231+ function __construct( $parser ) {
 5232+ $this->parser = $parser;
 5233+ $this->title = $parser->mTitle;
 5234+ }
 5235+
 5236+ /**
 5237+ * Create a new child frame
 5238+ * $args is optionally a DOMNodeList containing the template arguments
 5239+ */
 5240+ function newChild( $args = false, $title = false ) {
 5241+ $assocArgs = array();
 5242+ if ( $title === false ) {
 5243+ $title = $this->title;
 5244+ }
 5245+ if ( $args !== false ) {
 5246+ $xpath = false;
 5247+ foreach ( $args as $arg ) {
 5248+ if ( !$xpath ) {
 5249+ $xpath = new DOMXPath( $arg->ownerDocument );
 5250+ }
 5251+
 5252+ $nameNodes = $xpath->query( 'name', $arg );
 5253+ if ( $nameNodes->item( 0 )->hasAttributes() ) {
 5254+ // Numbered parameter
 5255+ $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
 5256+ } else {
 5257+ // Named parameter
 5258+ $name = $this->expand( $nameNodes->item( 0 ) );
 5259+ }
 5260+
 5261+ $value = $xpath->query( 'value', $arg );
 5262+ $assocArgs[$name] = $value->item( 0 );
 5263+ }
 5264+ }
 5265+ return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
 5266+ }
 5267+
 5268+ /**
 5269+ * Expand a DOMNode describing a preprocessed document into plain wikitext,
 5270+ * using the current context
 5271+ * @param $root the node
 5272+ */
 5273+ function expand( $root, $shallowFlags = 0, $deepFlags = 0 ) {
 5274+ if ( is_string( $root ) ) {
 5275+ return $root;
 5276+ }
 5277+
 5278+ if ( $this->parser->ot['html']
 5279+ && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
 5280+ {
 5281+ return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
 5282+ }
 5283+ $flags = $shallowFlags | $deepFlags;
 5284+
 5285+ if ( is_array( $root ) ) {
 5286+ $s = '';
 5287+ foreach ( $root as $node ) {
 5288+ $s .= $this->expand( $node, 0, $deepFlags );
 5289+ }
 5290+ } elseif ( $root instanceof DOMNodeList ) {
 5291+ $s = '';
 5292+ foreach ( $root as $node ) {
 5293+ $s .= $this->expand( $node, 0, $deepFlags );
 5294+ }
 5295+ } elseif ( $root instanceof DOMNode ) {
 5296+ if ( $root->nodeType == XML_TEXT_NODE ) {
 5297+ $s = $root->nodeValue;
 5298+ } elseif ( $root->nodeName == 'template' ) {
 5299+ # Double-brace expansion
 5300+ $xpath = new DOMXPath( $root->ownerDocument );
 5301+ $titles = $xpath->query( 'title', $root );
 5302+ $title = $titles->item( 0 );
 5303+ $parts = $xpath->query( 'part', $root );
 5304+ if ( $flags & self::NO_TEMPLATES ) {
 5305+ $s = '{{' . $this->implodeWithFlags( '|', 0, $deepFlags, $title, $parts ) . '}}';
 5306+ } else {
 5307+ $lineStart = $root->getAttribute( 'lineStart' );
 5308+ $params = array(
 5309+ 'title' => $title,
 5310+ 'parts' => $parts,
 5311+ 'lineStart' => $lineStart,
 5312+ 'text' => 'FIXME' );
 5313+ $s = $this->parser->braceSubstitution( $params, $this );
 5314+ }
 5315+ } elseif ( $root->nodeName == 'tplarg' ) {
 5316+ # Triple-brace expansion
 5317+ $xpath = new DOMXPath( $root->ownerDocument );
 5318+ $titles = $xpath->query( 'title', $root );
 5319+ $title = $titles->item( 0 );
 5320+ $parts = $xpath->query( 'part', $root );
 5321+ if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
 5322+ $s = '{{{' . $this->implode( '|', 0, $deepFlags, $title, $parts ) . '}}}';
 5323+ } else {
 5324+ $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
 5325+ $s = $this->parser->argSubstitution( $params, $this );
 5326+ }
 5327+ } elseif ( $root->nodeName == 'ext' ) {
 5328+ # Extension tag
 5329+ $xpath = new DOMXPath( $root->ownerDocument );
 5330+ $names = $xpath->query( 'name', $root );
 5331+ $attrs = $xpath->query( 'attr', $root );
 5332+ $inners = $xpath->query( 'inner', $root );
 5333+ $closes = $xpath->query( 'close', $root );
 5334+ $params = array(
 5335+ 'name' => $names->item( 0 ),
 5336+ 'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
 5337+ 'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
 5338+ 'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
 5339+ );
 5340+ $s = $this->parser->extensionSubstitution( $params, $this );
 5341+ } elseif ( $root->nodeName == 'h' ) {
 5342+ # Heading
 5343+ $s = $this->expand( $root->childNodes, 0, $deepFlags );
 5344+
 5345+ if ( $this->parser->ot['html'] ) {
 5346+ # Insert heading index marker
 5347+ $headingIndex = $root->getAttribute( 'i' );
 5348+ $titleText = $this->title->getPrefixedDBkey();
 5349+ $this->parser->mHeadings[] = array( $titleText, $headingIndex );
 5350+ $serial = count( $this->parser->mHeadings ) - 1;
 5351+ $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
 5352+ $count = $root->getAttribute( 'level' );
 5353+
 5354+ // FIXME: bug-for-bug with old parser
 5355+ // Lose whitespace for no apparent reason
 5356+ // Remove this after differential testing is done
 5357+ if ( true ) {
 5358+ // Good version
 5359+ $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
 5360+ } else {
 5361+ // Bad version
 5362+ if ( preg_match( '/^(={1,6})(.*?)(={1,6})\s*?$/', $s, $m ) ) {
 5363+ if ( $m[2] != '' ) {
 5364+ $s = $m[1] . $marker . $m[2] . $m[3];
 5365+ }
 5366+ }
 5367+ }
 5368+ $this->parser->mStripState->general->setPair( $marker, '' );
 5369+ }
 5370+ } else {
 5371+ # Generic recursive expansion
 5372+ $s = '';
 5373+ for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
 5374+ if ( $node->nodeType == XML_TEXT_NODE ) {
 5375+ $s .= $node->nodeValue;
 5376+ } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
 5377+ $s .= $this->expand( $node, 0, $deepFlags );
 5378+ }
 5379+ }
 5380+ }
 5381+ } else {
 5382+ throw new MWException( __METHOD__.': Invalid parameter type' );
 5383+ }
 5384+ return $s;
 5385+ }
 5386+
 5387+ function implodeWithFlags( $sep, $shallowFlags, $deepFlags /*, ... */ ) {
 5388+ $args = array_slice( func_get_args(), 3 );
 5389+
 5390+ $first = true;
 5391+ $s = '';
 5392+ foreach ( $args as $root ) {
 5393+ if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
 5394+ $root = array( $root );
 5395+ }
 5396+ foreach ( $root as $node ) {
 5397+ if ( $first ) {
 5398+ $first = false;
 5399+ } else {
 5400+ $s .= $sep;
 5401+ }
 5402+ $s .= $this->expand( $node, $shallowFlags, $deepFlags );
 5403+ }
 5404+ }
 5405+ return $s;
 5406+ }
 5407+
 5408+ function implode( $sep /*, ... */ ) {
 5409+ $args = func_get_args();
 5410+ $args = array_merge( array_slice( $args, 0, 1 ), array( 0, 0 ), array_slice( $args, 1 ) );
 5411+ return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
 5412+ }
 5413+
 5414+ function __toString() {
 5415+ return 'frame{}';
 5416+ }
 5417+}
 5418+
 5419+/**
 5420+ * Expansion frame with template arguments
 5421+ */
 5422+class PPTemplateFrame extends PPFrame {
 5423+ public $parser, $args, $parent, $serial;
 5424+
 5425+ function __construct( $parser, $parent = false, $args = array(), $title = false ) {
 5426+ $this->parser = $parser;
 5427+ $this->parent = $parent;
 5428+ $this->args = $args;
 5429+ $this->title = $title;
 5430+ }
 5431+
 5432+ function __toString() {
 5433+ $s = 'tplframe{';
 5434+ $first = true;
 5435+ foreach ( $this->args as $name => $value ) {
 5436+ if ( $first ) {
 5437+ $first = false;
 5438+ } else {
 5439+ $s .= ', ';
 5440+ }
 5441+ $s .= "\"$name\":\"" .
 5442+ str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
 5443+ }
 5444+ $s .= '}';
 5445+ return $s;
 5446+ }
 5447+}
 5448+
Index: trunk/phase3/includes/Setup.php
——@@ -235,7 +235,8 @@
236236 $wgUser = new StubUser;
237237 $wgLang = new StubUserLang;
238238 $wgOut = new StubObject( 'wgOut', 'OutputPage' );
239 -$wgParser = new StubObject( 'wgParser', 'Parser' );
 239+$wgParser = new StubObject( 'wgParser', $wgParserConf['class'], array( $wgParserConf ) );
 240+
240241 $wgMessageCache = new StubObject( 'wgMessageCache', 'MessageCache',
241242 array( $parserMemc, $wgUseDatabaseMessages, $wgMsgCacheExpiry, wfWikiID() ) );
242243
Index: trunk/phase3/includes/Parser_DiffTest.php
——@@ -0,0 +1,62 @@
 2+<?php
 3+
 4+class Parser_DiffTest
 5+{
 6+ var $parsers, $conf;
 7+
 8+ function __construct( $conf ) {
 9+ if ( !isset( $conf['parsers'] ) ) {
 10+ throw new MWException( __METHOD__ . ': no parsers specified' );
 11+ }
 12+ $this->conf = $conf;
 13+ }
 14+
 15+ function init() {
 16+ if ( !is_null( $this->parsers ) ) {
 17+ return;
 18+ }
 19+ foreach ( $this->conf['parsers'] as $i => $parserConf ) {
 20+ if ( !is_array( $parserConf ) ) {
 21+ $class = $parserConf;
 22+ $parserconf = array( 'class' => $parserConf );
 23+ } else {
 24+ $class = $parserConf['class'];
 25+ }
 26+ $this->parsers[$i] = new $class( $parserConf );
 27+ }
 28+ }
 29+
 30+ function __call( $name, $args ) {
 31+ $this->init();
 32+ $results = array();
 33+ $mismatch = false;
 34+ $lastResult = null;
 35+ $first = true;
 36+ foreach ( $this->parsers as $i => $parser ) {
 37+ $currentResult = call_user_func_array( array( &$this->parsers[$i], $name ), $args );
 38+ if ( $first ) {
 39+ $first = false;
 40+ } else {
 41+ if ( $lastResult !== $currentResult ) {
 42+ $mismatch = true;
 43+ }
 44+ }
 45+ $results[$i] = $currentResult;
 46+ $lastResult = $currentResult;
 47+ }
 48+ if ( $mismatch ) {
 49+ throw new MWException( "Parser_DiffTest: results mismatch on call to $name\n" .
 50+ 'Arguments: ' . var_export( $args, true ) . "\n" .
 51+ 'Results: ' . var_export( $results, true ) . "\n" );
 52+ }
 53+ return $lastResult;
 54+ }
 55+
 56+ function setFunctionHook( $id, $callback, $flags = 0 ) {
 57+ $this->init();
 58+ foreach ( $this->parsers as $i => $parser ) {
 59+ $parser->setFunctionHook( $id, $callback, $flags );
 60+ }
 61+ }
 62+}
 63+
Property changes on: trunk/phase3/includes/Parser_DiffTest.php
___________________________________________________________________
Added: svn:eol-style
164 + native
Index: trunk/phase3/includes/Parser_OldPP.php
——@@ -0,0 +1,4918 @@
 2+<?php
 3+/**
 4+ * Parser with old preprocessor
 5+ */
 6+class Parser_OldPP
 7+{
 8+ /**
 9+ * Update this version number when the ParserOutput format
 10+ * changes in an incompatible way, so the parser cache
 11+ * can automatically discard old data.
 12+ */
 13+ const VERSION = '1.6.2';
 14+
 15+ # Flags for Parser::setFunctionHook
 16+ # Also available as global constants from Defines.php
 17+ const SFH_NO_HASH = 1;
 18+ const SFH_OBJECT_ARGS = 2;
 19+
 20+ # Constants needed for external link processing
 21+ # Everything except bracket, space, or control characters
 22+ const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
 23+ const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)\\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/S';
 24+
 25+ // State constants for the definition list colon extraction
 26+ const COLON_STATE_TEXT = 0;
 27+ const COLON_STATE_TAG = 1;
 28+ const COLON_STATE_TAGSTART = 2;
 29+ const COLON_STATE_CLOSETAG = 3;
 30+ const COLON_STATE_TAGSLASH = 4;
 31+ const COLON_STATE_COMMENT = 5;
 32+ const COLON_STATE_COMMENTDASH = 6;
 33+ const COLON_STATE_COMMENTDASHDASH = 7;
 34+
 35+ /**#@+
 36+ * @private
 37+ */
 38+ # Persistent:
 39+ var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
 40+ $mImageParams, $mImageParamsMagicArray, $mExtLinkBracketedRegex;
 41+
 42+ # Cleared with clearState():
 43+ var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 44+ var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 45+ var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 46+ var $mIncludeSizes, $mDefaultSort;
 47+ var $mTemplates, // cache of already loaded templates, avoids
 48+ // multiple SQL queries for the same string
 49+ $mTemplatePath; // stores an unsorted hash of all the templates already loaded
 50+ // in this path. Used for loop detection.
 51+
 52+ # Temporary
 53+ # These are variables reset at least once per parse regardless of $clearState
 54+ var $mOptions, // ParserOptions object
 55+ $mTitle, // Title context, used for self-link rendering and similar things
 56+ $mOutputType, // Output type, one of the OT_xxx constants
 57+ $ot, // Shortcut alias, see setOutputType()
 58+ $mRevisionId, // ID to display in {{REVISIONID}} tags
 59+ $mRevisionTimestamp, // The timestamp of the specified revision ID
 60+ $mRevIdForTs; // The revision ID which was used to fetch the timestamp
 61+
 62+ /**#@-*/
 63+
 64+ /**
 65+ * Constructor
 66+ *
 67+ * @public
 68+ */
 69+ function __construct( $conf = array() ) {
 70+ $this->mTagHooks = array();
 71+ $this->mTransparentTagHooks = array();
 72+ $this->mFunctionHooks = array();
 73+ $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 74+ $this->mFirstCall = true;
 75+ $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 76+ '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 77+ }
 78+
 79+ /**
 80+ * Do various kinds of initialisation on the first call of the parser
 81+ */
 82+ function firstCallInit() {
 83+ if ( !$this->mFirstCall ) {
 84+ return;
 85+ }
 86+
 87+ wfProfileIn( __METHOD__ );
 88+ global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 89+
 90+ $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 91+
 92+ $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 93+ $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 94+ $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 95+ $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 96+ $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 97+ $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 98+ $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 99+ $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 100+ $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 101+ $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 102+ $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 103+ $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 104+ $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 105+ $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 106+ $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 107+ $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 108+ $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 109+ $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 110+ $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 111+ $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 112+ $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 113+ $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 114+ $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 115+ $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 116+ $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 117+ $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 118+ $this->setFunctionHook( 'filepath', array( 'CoreParserFunctions', 'filepath' ), SFH_NO_HASH );
 119+
 120+ if ( $wgAllowDisplayTitle ) {
 121+ $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 122+ }
 123+ if ( $wgAllowSlowParserFunctions ) {
 124+ $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 125+ }
 126+
 127+ $this->initialiseVariables();
 128+ $this->mFirstCall = false;
 129+ wfProfileOut( __METHOD__ );
 130+ }
 131+
 132+ /**
 133+ * Clear Parser state
 134+ *
 135+ * @private
 136+ */
 137+ function clearState() {
 138+ wfProfileIn( __METHOD__ );
 139+ if ( $this->mFirstCall ) {
 140+ $this->firstCallInit();
 141+ }
 142+ $this->mOutput = new ParserOutput;
 143+ $this->mAutonumber = 0;
 144+ $this->mLastSection = '';
 145+ $this->mDTopen = false;
 146+ $this->mIncludeCount = array();
 147+ $this->mStripState = new StripState;
 148+ $this->mArgStack = array();
 149+ $this->mInPre = false;
 150+ $this->mInterwikiLinkHolders = array(
 151+ 'texts' => array(),
 152+ 'titles' => array()
 153+ );
 154+ $this->mLinkHolders = array(
 155+ 'namespaces' => array(),
 156+ 'dbkeys' => array(),
 157+ 'queries' => array(),
 158+ 'texts' => array(),
 159+ 'titles' => array()
 160+ );
 161+ $this->mRevisionTimestamp = $this->mRevisionId = null;
 162+
 163+ /**
 164+ * Prefix for temporary replacement strings for the multipass parser.
 165+ * \x07 should never appear in input as it's disallowed in XML.
 166+ * Using it at the front also gives us a little extra robustness
 167+ * since it shouldn't match when butted up against identifier-like
 168+ * string constructs.
 169+ */
 170+ $this->mUniqPrefix = "\x07UNIQ" . self::getRandomString();
 171+
 172+ # Clear these on every parse, bug 4549
 173+ $this->mTemplates = array();
 174+ $this->mTemplatePath = array();
 175+
 176+ $this->mShowToc = true;
 177+ $this->mForceTocPosition = false;
 178+ $this->mIncludeSizes = array(
 179+ 'pre-expand' => 0,
 180+ 'post-expand' => 0,
 181+ 'arg' => 0
 182+ );
 183+ $this->mDefaultSort = false;
 184+
 185+ wfRunHooks( 'ParserClearState', array( &$this ) );
 186+ wfProfileOut( __METHOD__ );
 187+ }
 188+
 189+ function setOutputType( $ot ) {
 190+ $this->mOutputType = $ot;
 191+ // Shortcut alias
 192+ $this->ot = array(
 193+ 'html' => $ot == OT_HTML,
 194+ 'wiki' => $ot == OT_WIKI,
 195+ 'msg' => $ot == OT_MSG,
 196+ 'pre' => $ot == OT_PREPROCESS,
 197+ );
 198+ }
 199+
 200+ /**
 201+ * Accessor for mUniqPrefix.
 202+ *
 203+ * @public
 204+ */
 205+ function uniqPrefix() {
 206+ return $this->mUniqPrefix;
 207+ }
 208+
 209+ /**
 210+ * Convert wikitext to HTML
 211+ * Do not call this function recursively.
 212+ *
 213+ * @param string $text Text we want to parse
 214+ * @param Title &$title A title object
 215+ * @param array $options
 216+ * @param boolean $linestart
 217+ * @param boolean $clearState
 218+ * @param int $revid number to pass in {{REVISIONID}}
 219+ * @return ParserOutput a ParserOutput
 220+ */
 221+ public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 222+ /**
 223+ * First pass--just handle <nowiki> sections, pass the rest off
 224+ * to internalParse() which does all the real work.
 225+ */
 226+
 227+ global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 228+ $fname = 'Parser::parse-' . wfGetCaller();
 229+ wfProfileIn( __METHOD__ );
 230+ wfProfileIn( $fname );
 231+
 232+ if ( $clearState ) {
 233+ $this->clearState();
 234+ }
 235+
 236+ $this->mOptions = $options;
 237+ $this->mTitle =& $title;
 238+ $oldRevisionId = $this->mRevisionId;
 239+ $oldRevisionTimestamp = $this->mRevisionTimestamp;
 240+ if( $revid !== null ) {
 241+ $this->mRevisionId = $revid;
 242+ $this->mRevisionTimestamp = null;
 243+ }
 244+ $this->setOutputType( OT_HTML );
 245+ wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 246+ $text = $this->strip( $text, $this->mStripState );
 247+ wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 248+ $text = $this->internalParse( $text );
 249+ $text = $this->mStripState->unstripGeneral( $text );
 250+
 251+ # Clean up special characters, only run once, next-to-last before doBlockLevels
 252+ $fixtags = array(
 253+ # french spaces, last one Guillemet-left
 254+ # only if there is something before the space
 255+ '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 256+ # french spaces, Guillemet-right
 257+ '/(\\302\\253) /' => '\\1&nbsp;',
 258+ );
 259+ $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 260+
 261+ # only once and last
 262+ $text = $this->doBlockLevels( $text, $linestart );
 263+
 264+ $this->replaceLinkHolders( $text );
 265+
 266+ # the position of the parserConvert() call should not be changed. it
 267+ # assumes that the links are all replaced and the only thing left
 268+ # is the <nowiki> mark.
 269+ # Side-effects: this calls $this->mOutput->setTitleText()
 270+ $text = $wgContLang->parserConvert( $text, $this );
 271+
 272+ $text = $this->mStripState->unstripNoWiki( $text );
 273+
 274+ wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 275+
 276+//!JF Move to its own function
 277+
 278+ $uniq_prefix = $this->mUniqPrefix;
 279+ $matches = array();
 280+ $elements = array_keys( $this->mTransparentTagHooks );
 281+ $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 282+
 283+ foreach( $matches as $marker => $data ) {
 284+ list( $element, $content, $params, $tag ) = $data;
 285+ $tagName = strtolower( $element );
 286+ if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 287+ $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 288+ array( $content, $params, $this ) );
 289+ } else {
 290+ $output = $tag;
 291+ }
 292+ $this->mStripState->general->setPair( $marker, $output );
 293+ }
 294+ $text = $this->mStripState->unstripGeneral( $text );
 295+
 296+ $text = Sanitizer::normalizeCharReferences( $text );
 297+
 298+ if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 299+ $text = self::tidy($text);
 300+ } else {
 301+ # attempt to sanitize at least some nesting problems
 302+ # (bug #2702 and quite a few others)
 303+ $tidyregs = array(
 304+ # ''Something [http://www.cool.com cool''] -->
 305+ # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 306+ '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 307+ '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 308+ # fix up an anchor inside another anchor, only
 309+ # at least for a single single nested link (bug 3695)
 310+ '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 311+ '\\1\\2</a>\\3</a>\\1\\4</a>',
 312+ # fix div inside inline elements- doBlockLevels won't wrap a line which
 313+ # contains a div, so fix it up here; replace
 314+ # div with escaped text
 315+ '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 316+ '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 317+ # remove empty italic or bold tag pairs, some
 318+ # introduced by rules above
 319+ '/<([bi])><\/\\1>/' => '',
 320+ );
 321+
 322+ $text = preg_replace(
 323+ array_keys( $tidyregs ),
 324+ array_values( $tidyregs ),
 325+ $text );
 326+ }
 327+
 328+ wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 329+
 330+ # Information on include size limits, for the benefit of users who try to skirt them
 331+ if ( max( $this->mIncludeSizes ) > 1000 ) {
 332+ $max = $this->mOptions->getMaxIncludeSize();
 333+ $text .= "<!-- \n" .
 334+ "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 335+ "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 336+ "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 337+ "Maximum: $max bytes\n" .
 338+ "-->\n";
 339+ }
 340+ $this->mOutput->setText( $text );
 341+ $this->mRevisionId = $oldRevisionId;
 342+ $this->mRevisionTimestamp = $oldRevisionTimestamp;
 343+ wfProfileOut( $fname );
 344+ wfProfileOut( __METHOD__ );
 345+
 346+ return $this->mOutput;
 347+ }
 348+
 349+ /**
 350+ * Recursive parser entry point that can be called from an extension tag
 351+ * hook.
 352+ */
 353+ function recursiveTagParse( $text ) {
 354+ wfProfileIn( __METHOD__ );
 355+ wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 356+ $text = $this->strip( $text, $this->mStripState );
 357+ wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 358+ $text = $this->internalParse( $text );
 359+ wfProfileOut( __METHOD__ );
 360+ return $text;
 361+ }
 362+
 363+ /**
 364+ * Expand templates and variables in the text, producing valid, static wikitext.
 365+ * Also removes comments.
 366+ */
 367+ function preprocess( $text, $title, $options, $revid = null ) {
 368+ wfProfileIn( __METHOD__ );
 369+ $this->clearState();
 370+ $this->setOutputType( OT_PREPROCESS );
 371+ $this->mOptions = $options;
 372+ $this->mTitle = $title;
 373+ if( $revid !== null ) {
 374+ $this->mRevisionId = $revid;
 375+ }
 376+ wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 377+ $text = $this->strip( $text, $this->mStripState );
 378+ wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 379+ if ( $this->mOptions->getRemoveComments() ) {
 380+ $text = Sanitizer::removeHTMLcomments( $text );
 381+ }
 382+ $text = $this->replaceVariables( $text );
 383+ $text = $this->mStripState->unstripBoth( $text );
 384+ wfProfileOut( __METHOD__ );
 385+ return $text;
 386+ }
 387+
 388+ /**
 389+ * Get a random string
 390+ *
 391+ * @private
 392+ * @static
 393+ */
 394+ function getRandomString() {
 395+ return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 396+ }
 397+
 398+ function &getTitle() { return $this->mTitle; }
 399+ function getOptions() { return $this->mOptions; }
 400+
 401+ function getFunctionLang() {
 402+ global $wgLang, $wgContLang;
 403+ return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 404+ }
 405+
 406+ /**
 407+ * Replaces all occurrences of HTML-style comments and the given tags
 408+ * in the text with a random marker and returns teh next text. The output
 409+ * parameter $matches will be an associative array filled with data in
 410+ * the form:
 411+ * 'UNIQ-xxxxx' => array(
 412+ * 'element',
 413+ * 'tag content',
 414+ * array( 'param' => 'x' ),
 415+ * '<element param="x">tag content</element>' ) )
 416+ *
 417+ * @param $elements list of element names. Comments are always extracted.
 418+ * @param $text Source text string.
 419+ * @param $uniq_prefix
 420+ *
 421+ * @public
 422+ * @static
 423+ */
 424+ function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 425+ static $n = 1;
 426+ $stripped = '';
 427+ $matches = array();
 428+
 429+ $taglist = implode( '|', $elements );
 430+ $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 431+
 432+ while ( '' != $text ) {
 433+ $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 434+ $stripped .= $p[0];
 435+ if( count( $p ) < 5 ) {
 436+ break;
 437+ }
 438+ if( count( $p ) > 5 ) {
 439+ // comment
 440+ $element = $p[4];
 441+ $attributes = '';
 442+ $close = '';
 443+ $inside = $p[5];
 444+ } else {
 445+ // tag
 446+ $element = $p[1];
 447+ $attributes = $p[2];
 448+ $close = $p[3];
 449+ $inside = $p[4];
 450+ }
 451+
 452+ $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 453+ $stripped .= $marker;
 454+
 455+ if ( $close === '/>' ) {
 456+ // Empty element tag, <tag />
 457+ $content = null;
 458+ $text = $inside;
 459+ $tail = null;
 460+ } else {
 461+ if( $element == '!--' ) {
 462+ $end = '/(-->)/';
 463+ } else {
 464+ $end = "/(<\\/$element\\s*>)/i";
 465+ }
 466+ $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 467+ $content = $q[0];
 468+ if( count( $q ) < 3 ) {
 469+ # No end tag -- let it run out to the end of the text.
 470+ $tail = '';
 471+ $text = '';
 472+ } else {
 473+ $tail = $q[1];
 474+ $text = $q[2];
 475+ }
 476+ }
 477+
 478+ $matches[$marker] = array( $element,
 479+ $content,
 480+ Sanitizer::decodeTagAttributes( $attributes ),
 481+ "<$element$attributes$close$content$tail" );
 482+ }
 483+ return $stripped;
 484+ }
 485+
 486+ /**
 487+ * Strips and renders nowiki, pre, math, hiero
 488+ * If $render is set, performs necessary rendering operations on plugins
 489+ * Returns the text, and fills an array with data needed in unstrip()
 490+ *
 491+ * @param StripState $state
 492+ *
 493+ * @param bool $stripcomments when set, HTML comments <!-- like this -->
 494+ * will be stripped in addition to other tags. This is important
 495+ * for section editing, where these comments cause confusion when
 496+ * counting the sections in the wikisource
 497+ *
 498+ * @param array dontstrip contains tags which should not be stripped;
 499+ * used to prevent stipping of <gallery> when saving (fixes bug 2700)
 500+ *
 501+ * @private
 502+ */
 503+ function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 504+ global $wgContLang;
 505+ wfProfileIn( __METHOD__ );
 506+ $render = ($this->mOutputType == OT_HTML);
 507+
 508+ $uniq_prefix = $this->mUniqPrefix;
 509+ $commentState = new ReplacementArray;
 510+ $nowikiItems = array();
 511+ $generalItems = array();
 512+
 513+ $elements = array_merge(
 514+ array( 'nowiki', 'gallery' ),
 515+ array_keys( $this->mTagHooks ) );
 516+ global $wgRawHtml;
 517+ if( $wgRawHtml ) {
 518+ $elements[] = 'html';
 519+ }
 520+ if( $this->mOptions->getUseTeX() ) {
 521+ $elements[] = 'math';
 522+ }
 523+
 524+ # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 525+ foreach ( $elements AS $k => $v ) {
 526+ if ( !in_array ( $v , $dontstrip ) ) continue;
 527+ unset ( $elements[$k] );
 528+ }
 529+
 530+ $matches = array();
 531+ $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 532+
 533+ foreach( $matches as $marker => $data ) {
 534+ list( $element, $content, $params, $tag ) = $data;
 535+ if( $render ) {
 536+ $tagName = strtolower( $element );
 537+ wfProfileIn( __METHOD__."-render-$tagName" );
 538+ switch( $tagName ) {
 539+ case '!--':
 540+ // Comment
 541+ if( substr( $tag, -3 ) == '-->' ) {
 542+ $output = $tag;
 543+ } else {
 544+ // Unclosed comment in input.
 545+ // Close it so later stripping can remove it
 546+ $output = "$tag-->";
 547+ }
 548+ break;
 549+ case 'html':
 550+ if( $wgRawHtml ) {
 551+ $output = $content;
 552+ break;
 553+ }
 554+ // Shouldn't happen otherwise. :)
 555+ case 'nowiki':
 556+ $output = Xml::escapeTagsOnly( $content );
 557+ break;
 558+ case 'math':
 559+ $output = $wgContLang->armourMath(
 560+ MathRenderer::renderMath( $content, $params ) );
 561+ break;
 562+ case 'gallery':
 563+ $output = $this->renderImageGallery( $content, $params );
 564+ break;
 565+ default:
 566+ if( isset( $this->mTagHooks[$tagName] ) ) {
 567+ $output = call_user_func_array( $this->mTagHooks[$tagName],
 568+ array( $content, $params, $this ) );
 569+ } else {
 570+ throw new MWException( "Invalid call hook $element" );
 571+ }
 572+ }
 573+ wfProfileOut( __METHOD__."-render-$tagName" );
 574+ } else {
 575+ // Just stripping tags; keep the source
 576+ $output = $tag;
 577+ }
 578+
 579+ // Unstrip the output, to support recursive strip() calls
 580+ $output = $state->unstripBoth( $output );
 581+
 582+ if( !$stripcomments && $element == '!--' ) {
 583+ $commentState->setPair( $marker, $output );
 584+ } elseif ( $element == 'html' || $element == 'nowiki' ) {
 585+ $nowikiItems[$marker] = $output;
 586+ } else {
 587+ $generalItems[$marker] = $output;
 588+ }
 589+ }
 590+ # Add the new items to the state
 591+ # We do this after the loop instead of during it to avoid slowing
 592+ # down the recursive unstrip
 593+ $state->nowiki->mergeArray( $nowikiItems );
 594+ $state->general->mergeArray( $generalItems );
 595+
 596+ # Unstrip comments unless explicitly told otherwise.
 597+ # (The comments are always stripped prior to this point, so as to
 598+ # not invoke any extension tags / parser hooks contained within
 599+ # a comment.)
 600+ if ( !$stripcomments ) {
 601+ // Put them all back and forget them
 602+ $text = $commentState->replace( $text );
 603+ }
 604+
 605+ wfProfileOut( __METHOD__ );
 606+ return $text;
 607+ }
 608+
 609+ /**
 610+ * Restores pre, math, and other extensions removed by strip()
 611+ *
 612+ * always call unstripNoWiki() after this one
 613+ * @private
 614+ * @deprecated use $this->mStripState->unstrip()
 615+ */
 616+ function unstrip( $text, $state ) {
 617+ return $state->unstripGeneral( $text );
 618+ }
 619+
 620+ /**
 621+ * Always call this after unstrip() to preserve the order
 622+ *
 623+ * @private
 624+ * @deprecated use $this->mStripState->unstrip()
 625+ */
 626+ function unstripNoWiki( $text, $state ) {
 627+ return $state->unstripNoWiki( $text );
 628+ }
 629+
 630+ /**
 631+ * @deprecated use $this->mStripState->unstripBoth()
 632+ */
 633+ function unstripForHTML( $text ) {
 634+ return $this->mStripState->unstripBoth( $text );
 635+ }
 636+
 637+ /**
 638+ * Add an item to the strip state
 639+ * Returns the unique tag which must be inserted into the stripped text
 640+ * The tag will be replaced with the original text in unstrip()
 641+ *
 642+ * @private
 643+ */
 644+ function insertStripItem( $text, &$state ) {
 645+ $rnd = $this->mUniqPrefix . '-item' . self::getRandomString();
 646+ $state->general->setPair( $rnd, $text );
 647+ return $rnd;
 648+ }
 649+
 650+ /**
 651+ * Interface with html tidy, used if $wgUseTidy = true.
 652+ * If tidy isn't able to correct the markup, the original will be
 653+ * returned in all its glory with a warning comment appended.
 654+ *
 655+ * Either the external tidy program or the in-process tidy extension
 656+ * will be used depending on availability. Override the default
 657+ * $wgTidyInternal setting to disable the internal if it's not working.
 658+ *
 659+ * @param string $text Hideous HTML input
 660+ * @return string Corrected HTML output
 661+ * @public
 662+ * @static
 663+ */
 664+ function tidy( $text ) {
 665+ global $wgTidyInternal;
 666+ $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 667+' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 668+'<head><title>test</title></head><body>'.$text.'</body></html>';
 669+ if( $wgTidyInternal ) {
 670+ $correctedtext = self::internalTidy( $wrappedtext );
 671+ } else {
 672+ $correctedtext = self::externalTidy( $wrappedtext );
 673+ }
 674+ if( is_null( $correctedtext ) ) {
 675+ wfDebug( "Tidy error detected!\n" );
 676+ return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 677+ }
 678+ return $correctedtext;
 679+ }
 680+
 681+ /**
 682+ * Spawn an external HTML tidy process and get corrected markup back from it.
 683+ *
 684+ * @private
 685+ * @static
 686+ */
 687+ function externalTidy( $text ) {
 688+ global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 689+ $fname = 'Parser::externalTidy';
 690+ wfProfileIn( $fname );
 691+
 692+ $cleansource = '';
 693+ $opts = ' -utf8';
 694+
 695+ $descriptorspec = array(
 696+ 0 => array('pipe', 'r'),
 697+ 1 => array('pipe', 'w'),
 698+ 2 => array('file', wfGetNull(), 'a')
 699+ );
 700+ $pipes = array();
 701+ $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 702+ if (is_resource($process)) {
 703+ // Theoretically, this style of communication could cause a deadlock
 704+ // here. If the stdout buffer fills up, then writes to stdin could
 705+ // block. This doesn't appear to happen with tidy, because tidy only
 706+ // writes to stdout after it's finished reading from stdin. Search
 707+ // for tidyParseStdin and tidySaveStdout in console/tidy.c
 708+ fwrite($pipes[0], $text);
 709+ fclose($pipes[0]);
 710+ while (!feof($pipes[1])) {
 711+ $cleansource .= fgets($pipes[1], 1024);
 712+ }
 713+ fclose($pipes[1]);
 714+ proc_close($process);
 715+ }
 716+
 717+ wfProfileOut( $fname );
 718+
 719+ if( $cleansource == '' && $text != '') {
 720+ // Some kind of error happened, so we couldn't get the corrected text.
 721+ // Just give up; we'll use the source text and append a warning.
 722+ return null;
 723+ } else {
 724+ return $cleansource;
 725+ }
 726+ }
 727+
 728+ /**
 729+ * Use the HTML tidy PECL extension to use the tidy library in-process,
 730+ * saving the overhead of spawning a new process.
 731+ *
 732+ * 'pear install tidy' should be able to compile the extension module.
 733+ *
 734+ * @private
 735+ * @static
 736+ */
 737+ function internalTidy( $text ) {
 738+ global $wgTidyConf, $IP;
 739+ $fname = 'Parser::internalTidy';
 740+ wfProfileIn( $fname );
 741+
 742+ $tidy = new tidy;
 743+ $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 744+ $tidy->cleanRepair();
 745+ if( $tidy->getStatus() == 2 ) {
 746+ // 2 is magic number for fatal error
 747+ // http://www.php.net/manual/en/function.tidy-get-status.php
 748+ $cleansource = null;
 749+ } else {
 750+ $cleansource = tidy_get_output( $tidy );
 751+ }
 752+ wfProfileOut( $fname );
 753+ return $cleansource;
 754+ }
 755+
 756+ /**
 757+ * parse the wiki syntax used to render tables
 758+ *
 759+ * @private
 760+ */
 761+ function doTableStuff ( $text ) {
 762+ $fname = 'Parser::doTableStuff';
 763+ wfProfileIn( $fname );
 764+
 765+ $lines = explode ( "\n" , $text );
 766+ $td_history = array (); // Is currently a td tag open?
 767+ $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 768+ $tr_history = array (); // Is currently a tr tag open?
 769+ $tr_attributes = array (); // history of tr attributes
 770+ $has_opened_tr = array(); // Did this table open a <tr> element?
 771+ $indent_level = 0; // indent level of the table
 772+ foreach ( $lines as $key => $line )
 773+ {
 774+ $line = trim ( $line );
 775+
 776+ if( $line == '' ) { // empty line, go to next line
 777+ continue;
 778+ }
 779+ $first_character = $line{0};
 780+ $matches = array();
 781+
 782+ if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 783+ // First check if we are starting a new table
 784+ $indent_level = strlen( $matches[1] );
 785+
 786+ $attributes = $this->mStripState->unstripBoth( $matches[2] );
 787+ $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 788+
 789+ $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 790+ array_push ( $td_history , false );
 791+ array_push ( $last_tag_history , '' );
 792+ array_push ( $tr_history , false );
 793+ array_push ( $tr_attributes , '' );
 794+ array_push ( $has_opened_tr , false );
 795+ } else if ( count ( $td_history ) == 0 ) {
 796+ // Don't do any of the following
 797+ continue;
 798+ } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 799+ // We are ending a table
 800+ $line = '</table>' . substr ( $line , 2 );
 801+ $last_tag = array_pop ( $last_tag_history );
 802+
 803+ if ( !array_pop ( $has_opened_tr ) ) {
 804+ $line = "<tr><td></td></tr>{$line}";
 805+ }
 806+
 807+ if ( array_pop ( $tr_history ) ) {
 808+ $line = "</tr>{$line}";
 809+ }
 810+
 811+ if ( array_pop ( $td_history ) ) {
 812+ $line = "</{$last_tag}>{$line}";
 813+ }
 814+ array_pop ( $tr_attributes );
 815+ $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 816+ } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 817+ // Now we have a table row
 818+ $line = preg_replace( '#^\|-+#', '', $line );
 819+
 820+ // Whats after the tag is now only attributes
 821+ $attributes = $this->mStripState->unstripBoth( $line );
 822+ $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 823+ array_pop ( $tr_attributes );
 824+ array_push ( $tr_attributes , $attributes );
 825+
 826+ $line = '';
 827+ $last_tag = array_pop ( $last_tag_history );
 828+ array_pop ( $has_opened_tr );
 829+ array_push ( $has_opened_tr , true );
 830+
 831+ if ( array_pop ( $tr_history ) ) {
 832+ $line = '</tr>';
 833+ }
 834+
 835+ if ( array_pop ( $td_history ) ) {
 836+ $line = "</{$last_tag}>{$line}";
 837+ }
 838+
 839+ $lines[$key] = $line;
 840+ array_push ( $tr_history , false );
 841+ array_push ( $td_history , false );
 842+ array_push ( $last_tag_history , '' );
 843+ }
 844+ else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 ) == '|+' ) {
 845+ // This might be cell elements, td, th or captions
 846+ if ( substr ( $line , 0 , 2 ) == '|+' ) {
 847+ $first_character = '+';
 848+ $line = substr ( $line , 1 );
 849+ }
 850+
 851+ $line = substr ( $line , 1 );
 852+
 853+ if ( $first_character == '!' ) {
 854+ $line = str_replace ( '!!' , '||' , $line );
 855+ }
 856+
 857+ // Split up multiple cells on the same line.
 858+ // FIXME : This can result in improper nesting of tags processed
 859+ // by earlier parser steps, but should avoid splitting up eg
 860+ // attribute values containing literal "||".
 861+ $cells = StringUtils::explodeMarkup( '||' , $line );
 862+
 863+ $lines[$key] = '';
 864+
 865+ // Loop through each table cell
 866+ foreach ( $cells as $cell )
 867+ {
 868+ $previous = '';
 869+ if ( $first_character != '+' )
 870+ {
 871+ $tr_after = array_pop ( $tr_attributes );
 872+ if ( !array_pop ( $tr_history ) ) {
 873+ $previous = "<tr{$tr_after}>\n";
 874+ }
 875+ array_push ( $tr_history , true );
 876+ array_push ( $tr_attributes , '' );
 877+ array_pop ( $has_opened_tr );
 878+ array_push ( $has_opened_tr , true );
 879+ }
 880+
 881+ $last_tag = array_pop ( $last_tag_history );
 882+
 883+ if ( array_pop ( $td_history ) ) {
 884+ $previous = "</{$last_tag}>{$previous}";
 885+ }
 886+
 887+ if ( $first_character == '|' ) {
 888+ $last_tag = 'td';
 889+ } else if ( $first_character == '!' ) {
 890+ $last_tag = 'th';
 891+ } else if ( $first_character == '+' ) {
 892+ $last_tag = 'caption';
 893+ } else {
 894+ $last_tag = '';
 895+ }
 896+
 897+ array_push ( $last_tag_history , $last_tag );
 898+
 899+ // A cell could contain both parameters and data
 900+ $cell_data = explode ( '|' , $cell , 2 );
 901+
 902+ // Bug 553: Note that a '|' inside an invalid link should not
 903+ // be mistaken as delimiting cell parameters
 904+ if ( strpos( $cell_data[0], '[[' ) !== false ) {
 905+ $cell = "{$previous}<{$last_tag}>{$cell}";
 906+ } else if ( count ( $cell_data ) == 1 )
 907+ $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 908+ else {
 909+ $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 910+ $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 911+ $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 912+ }
 913+
 914+ $lines[$key] .= $cell;
 915+ array_push ( $td_history , true );
 916+ }
 917+ }
 918+ }
 919+
 920+ // Closing open td, tr && table
 921+ while ( count ( $td_history ) > 0 )
 922+ {
 923+ if ( array_pop ( $td_history ) ) {
 924+ $lines[] = '</td>' ;
 925+ }
 926+ if ( array_pop ( $tr_history ) ) {
 927+ $lines[] = '</tr>' ;
 928+ }
 929+ if ( !array_pop ( $has_opened_tr ) ) {
 930+ $lines[] = "<tr><td></td></tr>" ;
 931+ }
 932+
 933+ $lines[] = '</table>' ;
 934+ }
 935+
 936+ $output = implode ( "\n" , $lines ) ;
 937+
 938+ // special case: don't return empty table
 939+ if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 940+ $output = '';
 941+ }
 942+
 943+ wfProfileOut( $fname );
 944+
 945+ return $output;
 946+ }
 947+
 948+ /**
 949+ * Helper function for parse() that transforms wiki markup into
 950+ * HTML. Only called for $mOutputType == OT_HTML.
 951+ *
 952+ * @private
 953+ */
 954+ function internalParse( $text ) {
 955+ $args = array();
 956+ $isMain = true;
 957+ $fname = 'Parser::internalParse';
 958+ wfProfileIn( $fname );
 959+
 960+ # Hook to suspend the parser in this state
 961+ if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 962+ wfProfileOut( $fname );
 963+ return $text ;
 964+ }
 965+
 966+ # Remove <noinclude> tags and <includeonly> sections
 967+ $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 968+ $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 969+ $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 970+
 971+ $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
 972+
 973+ $text = $this->replaceVariables( $text, $args );
 974+ wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 975+
 976+ // Tables need to come after variable replacement for things to work
 977+ // properly; putting them before other transformations should keep
 978+ // exciting things like link expansions from showing up in surprising
 979+ // places.
 980+ $text = $this->doTableStuff( $text );
 981+
 982+ $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 983+
 984+ $text = $this->stripToc( $text );
 985+ $this->stripNoGallery( $text );
 986+ $text = $this->doHeadings( $text );
 987+ if($this->mOptions->getUseDynamicDates()) {
 988+ $df =& DateFormatter::getInstance();
 989+ $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 990+ }
 991+ $text = $this->doAllQuotes( $text );
 992+ $text = $this->replaceInternalLinks( $text );
 993+ $text = $this->replaceExternalLinks( $text );
 994+
 995+ # replaceInternalLinks may sometimes leave behind
 996+ # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 997+ $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 998+
 999+ $text = $this->doMagicLinks( $text );
 1000+ $text = $this->formatHeadings( $text, $isMain );
 1001+
 1002+ wfProfileOut( $fname );
 1003+ return $text;
 1004+ }
 1005+
 1006+ /**
 1007+ * Replace special strings like "ISBN xxx" and "RFC xxx" with
 1008+ * magic external links.
 1009+ *
 1010+ * @private
 1011+ */
 1012+ function &doMagicLinks( &$text ) {
 1013+ wfProfileIn( __METHOD__ );
 1014+ $text = preg_replace_callback(
 1015+ '!(?: # Start cases
 1016+ <a.*?</a> | # Skip link text
 1017+ <.*?> | # Skip stuff inside HTML elements
 1018+ (?:RFC|PMID)\s+([0-9]+) | # RFC or PMID, capture number as m[1]
 1019+ ISBN\s+(\b # ISBN, capture number as m[2]
 1020+ (?: 97[89] [\ \-]? )? # optional 13-digit ISBN prefix
 1021+ (?: [0-9] [\ \-]? ){9} # 9 digits with opt. delimiters
 1022+ [0-9Xx] # check digit
 1023+ \b)
 1024+ )!x', array( &$this, 'magicLinkCallback' ), $text );
 1025+ wfProfileOut( __METHOD__ );
 1026+ return $text;
 1027+ }
 1028+
 1029+ function magicLinkCallback( $m ) {
 1030+ if ( substr( $m[0], 0, 1 ) == '<' ) {
 1031+ # Skip HTML element
 1032+ return $m[0];
 1033+ } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 1034+ $isbn = $m[2];
 1035+ $num = strtr( $isbn, array(
 1036+ '-' => '',
 1037+ ' ' => '',
 1038+ 'x' => 'X',
 1039+ ));
 1040+ $titleObj = SpecialPage::getTitleFor( 'Booksources' );
 1041+ $text = '<a href="' .
 1042+ $titleObj->escapeLocalUrl( "isbn=$num" ) .
 1043+ "\" class=\"internal\">ISBN $isbn</a>";
 1044+ } else {
 1045+ if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
 1046+ $keyword = 'RFC';
 1047+ $urlmsg = 'rfcurl';
 1048+ $id = $m[1];
 1049+ } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
 1050+ $keyword = 'PMID';
 1051+ $urlmsg = 'pubmedurl';
 1052+ $id = $m[1];
 1053+ } else {
 1054+ throw new MWException( __METHOD__.': unrecognised match type "' .
 1055+ substr($m[0], 0, 20 ) . '"' );
 1056+ }
 1057+
 1058+ $url = wfMsg( $urlmsg, $id);
 1059+ $sk = $this->mOptions->getSkin();
 1060+ $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
 1061+ $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 1062+ }
 1063+ return $text;
 1064+ }
 1065+
 1066+ /**
 1067+ * Parse headers and return html
 1068+ *
 1069+ * @private
 1070+ */
 1071+ function doHeadings( $text ) {
 1072+ $fname = 'Parser::doHeadings';
 1073+ wfProfileIn( $fname );
 1074+ for ( $i = 6; $i >= 1; --$i ) {
 1075+ $h = str_repeat( '=', $i );
 1076+ $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
 1077+ "<h{$i}>\\1</h{$i}>\\2", $text );
 1078+ }
 1079+ wfProfileOut( $fname );
 1080+ return $text;
 1081+ }
 1082+
 1083+ /**
 1084+ * Replace single quotes with HTML markup
 1085+ * @private
 1086+ * @return string the altered text
 1087+ */
 1088+ function doAllQuotes( $text ) {
 1089+ $fname = 'Parser::doAllQuotes';
 1090+ wfProfileIn( $fname );
 1091+ $outtext = '';
 1092+ $lines = explode( "\n", $text );
 1093+ foreach ( $lines as $line ) {
 1094+ $outtext .= $this->doQuotes ( $line ) . "\n";
 1095+ }
 1096+ $outtext = substr($outtext, 0,-1);
 1097+ wfProfileOut( $fname );
 1098+ return $outtext;
 1099+ }
 1100+
 1101+ /**
 1102+ * Helper function for doAllQuotes()
 1103+ */
 1104+ public function doQuotes( $text ) {
 1105+ $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 1106+ if ( count( $arr ) == 1 )
 1107+ return $text;
 1108+ else
 1109+ {
 1110+ # First, do some preliminary work. This may shift some apostrophes from
 1111+ # being mark-up to being text. It also counts the number of occurrences
 1112+ # of bold and italics mark-ups.
 1113+ $i = 0;
 1114+ $numbold = 0;
 1115+ $numitalics = 0;
 1116+ foreach ( $arr as $r )
 1117+ {
 1118+ if ( ( $i % 2 ) == 1 )
 1119+ {
 1120+ # If there are ever four apostrophes, assume the first is supposed to
 1121+ # be text, and the remaining three constitute mark-up for bold text.
 1122+ if ( strlen( $arr[$i] ) == 4 )
 1123+ {
 1124+ $arr[$i-1] .= "'";
 1125+ $arr[$i] = "'''";
 1126+ }
 1127+ # If there are more than 5 apostrophes in a row, assume they're all
 1128+ # text except for the last 5.
 1129+ else if ( strlen( $arr[$i] ) > 5 )
 1130+ {
 1131+ $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 1132+ $arr[$i] = "'''''";
 1133+ }
 1134+ # Count the number of occurrences of bold and italics mark-ups.
 1135+ # We are not counting sequences of five apostrophes.
 1136+ if ( strlen( $arr[$i] ) == 2 ) { $numitalics++; }
 1137+ else if ( strlen( $arr[$i] ) == 3 ) { $numbold++; }
 1138+ else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 1139+ }
 1140+ $i++;
 1141+ }
 1142+
 1143+ # If there is an odd number of both bold and italics, it is likely
 1144+ # that one of the bold ones was meant to be an apostrophe followed
 1145+ # by italics. Which one we cannot know for certain, but it is more
 1146+ # likely to be one that has a single-letter word before it.
 1147+ if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 1148+ {
 1149+ $i = 0;
 1150+ $firstsingleletterword = -1;
 1151+ $firstmultiletterword = -1;
 1152+ $firstspace = -1;
 1153+ foreach ( $arr as $r )
 1154+ {
 1155+ if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 1156+ {
 1157+ $x1 = substr ($arr[$i-1], -1);
 1158+ $x2 = substr ($arr[$i-1], -2, 1);
 1159+ if ($x1 == ' ') {
 1160+ if ($firstspace == -1) $firstspace = $i;
 1161+ } else if ($x2 == ' ') {
 1162+ if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 1163+ } else {
 1164+ if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 1165+ }
 1166+ }
 1167+ $i++;
 1168+ }
 1169+
 1170+ # If there is a single-letter word, use it!
 1171+ if ($firstsingleletterword > -1)
 1172+ {
 1173+ $arr [ $firstsingleletterword ] = "''";
 1174+ $arr [ $firstsingleletterword-1 ] .= "'";
 1175+ }
 1176+ # If not, but there's a multi-letter word, use that one.
 1177+ else if ($firstmultiletterword > -1)
 1178+ {
 1179+ $arr [ $firstmultiletterword ] = "''";
 1180+ $arr [ $firstmultiletterword-1 ] .= "'";
 1181+ }
 1182+ # ... otherwise use the first one that has neither.
 1183+ # (notice that it is possible for all three to be -1 if, for example,
 1184+ # there is only one pentuple-apostrophe in the line)
 1185+ else if ($firstspace > -1)
 1186+ {
 1187+ $arr [ $firstspace ] = "''";
 1188+ $arr [ $firstspace-1 ] .= "'";
 1189+ }
 1190+ }
 1191+
 1192+ # Now let's actually convert our apostrophic mush to HTML!
 1193+ $output = '';
 1194+ $buffer = '';
 1195+ $state = '';
 1196+ $i = 0;
 1197+ foreach ($arr as $r)
 1198+ {
 1199+ if (($i % 2) == 0)
 1200+ {
 1201+ if ($state == 'both')
 1202+ $buffer .= $r;
 1203+ else
 1204+ $output .= $r;
 1205+ }
 1206+ else
 1207+ {
 1208+ if (strlen ($r) == 2)
 1209+ {
 1210+ if ($state == 'i')
 1211+ { $output .= '</i>'; $state = ''; }
 1212+ else if ($state == 'bi')
 1213+ { $output .= '</i>'; $state = 'b'; }
 1214+ else if ($state == 'ib')
 1215+ { $output .= '</b></i><b>'; $state = 'b'; }
 1216+ else if ($state == 'both')
 1217+ { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 1218+ else # $state can be 'b' or ''
 1219+ { $output .= '<i>'; $state .= 'i'; }
 1220+ }
 1221+ else if (strlen ($r) == 3)
 1222+ {
 1223+ if ($state == 'b')
 1224+ { $output .= '</b>'; $state = ''; }
 1225+ else if ($state == 'bi')
 1226+ { $output .= '</i></b><i>'; $state = 'i'; }
 1227+ else if ($state == 'ib')
 1228+ { $output .= '</b>'; $state = 'i'; }
 1229+ else if ($state == 'both')
 1230+ { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 1231+ else # $state can be 'i' or ''
 1232+ { $output .= '<b>'; $state .= 'b'; }
 1233+ }
 1234+ else if (strlen ($r) == 5)
 1235+ {
 1236+ if ($state == 'b')
 1237+ { $output .= '</b><i>'; $state = 'i'; }
 1238+ else if ($state == 'i')
 1239+ { $output .= '</i><b>'; $state = 'b'; }
 1240+ else if ($state == 'bi')
 1241+ { $output .= '</i></b>'; $state = ''; }
 1242+ else if ($state == 'ib')
 1243+ { $output .= '</b></i>'; $state = ''; }
 1244+ else if ($state == 'both')
 1245+ { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
 1246+ else # ($state == '')
 1247+ { $buffer = ''; $state = 'both'; }
 1248+ }
 1249+ }
 1250+ $i++;
 1251+ }
 1252+ # Now close all remaining tags. Notice that the order is important.
 1253+ if ($state == 'b' || $state == 'ib')
 1254+ $output .= '</b>';
 1255+ if ($state == 'i' || $state == 'bi' || $state == 'ib')
 1256+ $output .= '</i>';
 1257+ if ($state == 'bi')
 1258+ $output .= '</b>';
 1259+ # There might be lonely ''''', so make sure we have a buffer
 1260+ if ($state == 'both' && $buffer)
 1261+ $output .= '<b><i>'.$buffer.'</i></b>';
 1262+ return $output;
 1263+ }
 1264+ }
 1265+
 1266+ /**
 1267+ * Replace external links
 1268+ *
 1269+ * Note: this is all very hackish and the order of execution matters a lot.
 1270+ * Make sure to run maintenance/parserTests.php if you change this code.
 1271+ *
 1272+ * @private
 1273+ */
 1274+ function replaceExternalLinks( $text ) {
 1275+ global $wgContLang;
 1276+ $fname = 'Parser::replaceExternalLinks';
 1277+ wfProfileIn( $fname );
 1278+
 1279+ $sk = $this->mOptions->getSkin();
 1280+
 1281+ $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 1282+
 1283+ $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
 1284+
 1285+ $i = 0;
 1286+ while ( $i<count( $bits ) ) {
 1287+ $url = $bits[$i++];
 1288+ $protocol = $bits[$i++];
 1289+ $text = $bits[$i++];
 1290+ $trail = $bits[$i++];
 1291+
 1292+ # The characters '<' and '>' (which were escaped by
 1293+ # removeHTMLtags()) should not be included in
 1294+ # URLs, per RFC 2396.
 1295+ $m2 = array();
 1296+ if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
 1297+ $text = substr($url, $m2[0][1]) . ' ' . $text;
 1298+ $url = substr($url, 0, $m2[0][1]);
 1299+ }
 1300+
 1301+ # If the link text is an image URL, replace it with an <img> tag
 1302+ # This happened by accident in the original parser, but some people used it extensively
 1303+ $img = $this->maybeMakeExternalImage( $text );
 1304+ if ( $img !== false ) {
 1305+ $text = $img;
 1306+ }
 1307+
 1308+ $dtrail = '';
 1309+
 1310+ # Set linktype for CSS - if URL==text, link is essentially free
 1311+ $linktype = ($text == $url) ? 'free' : 'text';
 1312+
 1313+ # No link text, e.g. [http://domain.tld/some.link]
 1314+ if ( $text == '' ) {
 1315+ # Autonumber if allowed. See bug #5918
 1316+ if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
 1317+ $text = '[' . ++$this->mAutonumber . ']';
 1318+ $linktype = 'autonumber';
 1319+ } else {
 1320+ # Otherwise just use the URL
 1321+ $text = htmlspecialchars( $url );
 1322+ $linktype = 'free';
 1323+ }
 1324+ } else {
 1325+ # Have link text, e.g. [http://domain.tld/some.link text]s
 1326+ # Check for trail
 1327+ list( $dtrail, $trail ) = Linker::splitTrail( $trail );
 1328+ }
 1329+
 1330+ $text = $wgContLang->markNoConversion($text);
 1331+
 1332+ $url = Sanitizer::cleanUrl( $url );
 1333+
 1334+ # Process the trail (i.e. everything after this link up until start of the next link),
 1335+ # replacing any non-bracketed links
 1336+ $trail = $this->replaceFreeExternalLinks( $trail );
 1337+
 1338+ # Use the encoded URL
 1339+ # This means that users can paste URLs directly into the text
 1340+ # Funny characters like &ouml; aren't valid in URLs anyway
 1341+ # This was changed in August 2004
 1342+ $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
 1343+
 1344+ # Register link in the output object.
 1345+ # Replace unnecessary URL escape codes with the referenced character
 1346+ # This prevents spammers from hiding links from the filters
 1347+ $pasteurized = self::replaceUnusualEscapes( $url );
 1348+ $this->mOutput->addExternalLink( $pasteurized );
 1349+ }
 1350+
 1351+ wfProfileOut( $fname );
 1352+ return $s;
 1353+ }
 1354+
 1355+ /**
 1356+ * Replace anything that looks like a URL with a link
 1357+ * @private
 1358+ */
 1359+ function replaceFreeExternalLinks( $text ) {
 1360+ global $wgContLang;
 1361+ $fname = 'Parser::replaceFreeExternalLinks';
 1362+ wfProfileIn( $fname );
 1363+
 1364+ $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 1365+ $s = array_shift( $bits );
 1366+ $i = 0;
 1367+
 1368+ $sk = $this->mOptions->getSkin();
 1369+
 1370+ while ( $i < count( $bits ) ){
 1371+ $protocol = $bits[$i++];
 1372+ $remainder = $bits[$i++];
 1373+
 1374+ $m = array();
 1375+ if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
 1376+ # Found some characters after the protocol that look promising
 1377+ $url = $protocol . $m[1];
 1378+ $trail = $m[2];
 1379+
 1380+ # special case: handle urls as url args:
 1381+ # http://www.example.com/foo?=http://www.example.com/bar
 1382+ if(strlen($trail) == 0 &&
 1383+ isset($bits[$i]) &&
 1384+ preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
 1385+ preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
 1386+ {
 1387+ # add protocol, arg
 1388+ $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
 1389+ $i += 2;
 1390+ $trail = $m[2];
 1391+ }
 1392+
 1393+ # The characters '<' and '>' (which were escaped by
 1394+ # removeHTMLtags()) should not be included in
 1395+ # URLs, per RFC 2396.
 1396+ $m2 = array();
 1397+ if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
 1398+ $trail = substr($url, $m2[0][1]) . $trail;
 1399+ $url = substr($url, 0, $m2[0][1]);
 1400+ }
 1401+
 1402+ # Move trailing punctuation to $trail
 1403+ $sep = ',;\.:!?';
 1404+ # If there is no left bracket, then consider right brackets fair game too
 1405+ if ( strpos( $url, '(' ) === false ) {
 1406+ $sep .= ')';
 1407+ }
 1408+
 1409+ $numSepChars = strspn( strrev( $url ), $sep );
 1410+ if ( $numSepChars ) {
 1411+ $trail = substr( $url, -$numSepChars ) . $trail;
 1412+ $url = substr( $url, 0, -$numSepChars );
 1413+ }
 1414+
 1415+ $url = Sanitizer::cleanUrl( $url );
 1416+
 1417+ # Is this an external image?
 1418+ $text = $this->maybeMakeExternalImage( $url );
 1419+ if ( $text === false ) {
 1420+ # Not an image, make a link
 1421+ $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
 1422+ # Register it in the output object...
 1423+ # Replace unnecessary URL escape codes with their equivalent characters
 1424+ $pasteurized = self::replaceUnusualEscapes( $url );
 1425+ $this->mOutput->addExternalLink( $pasteurized );
 1426+ }
 1427+ $s .= $text . $trail;
 1428+ } else {
 1429+ $s .= $protocol . $remainder;
 1430+ }
 1431+ }
 1432+ wfProfileOut( $fname );
 1433+ return $s;
 1434+ }
 1435+
 1436+ /**
 1437+ * Replace unusual URL escape codes with their equivalent characters
 1438+ * @param string
 1439+ * @return string
 1440+ * @static
 1441+ * @todo This can merge genuinely required bits in the path or query string,
 1442+ * breaking legit URLs. A proper fix would treat the various parts of
 1443+ * the URL differently; as a workaround, just use the output for
 1444+ * statistical records, not for actual linking/output.
 1445+ */
 1446+ static function replaceUnusualEscapes( $url ) {
 1447+ return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
 1448+ array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
 1449+ }
 1450+
 1451+ /**
 1452+ * Callback function used in replaceUnusualEscapes().
 1453+ * Replaces unusual URL escape codes with their equivalent character
 1454+ * @static
 1455+ * @private
 1456+ */
 1457+ private static function replaceUnusualEscapesCallback( $matches ) {
 1458+ $char = urldecode( $matches[0] );
 1459+ $ord = ord( $char );
 1460+ // Is it an unsafe or HTTP reserved character according to RFC 1738?
 1461+ if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
 1462+ // No, shouldn't be escaped
 1463+ return $char;
 1464+ } else {
 1465+ // Yes, leave it escaped
 1466+ return $matches[0];
 1467+ }
 1468+ }
 1469+
 1470+ /**
 1471+ * make an image if it's allowed, either through the global
 1472+ * option or through the exception
 1473+ * @private
 1474+ */
 1475+ function maybeMakeExternalImage( $url ) {
 1476+ $sk = $this->mOptions->getSkin();
 1477+ $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
 1478+ $imagesexception = !empty($imagesfrom);
 1479+ $text = false;
 1480+ if ( $this->mOptions->getAllowExternalImages()
 1481+ || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
 1482+ if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
 1483+ # Image found
 1484+ $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
 1485+ }
 1486+ }
 1487+ return $text;
 1488+ }
 1489+
 1490+ /**
 1491+ * Process [[ ]] wikilinks
 1492+ *
 1493+ * @private
 1494+ */
 1495+ function replaceInternalLinks( $s ) {
 1496+ global $wgContLang;
 1497+ static $fname = 'Parser::replaceInternalLinks' ;
 1498+
 1499+ wfProfileIn( $fname );
 1500+
 1501+ wfProfileIn( $fname.'-setup' );
 1502+ static $tc = FALSE;
 1503+ # the % is needed to support urlencoded titles as well
 1504+ if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
 1505+
 1506+ $sk = $this->mOptions->getSkin();
 1507+
 1508+ #split the entire text string on occurences of [[
 1509+ $a = explode( '[[', ' ' . $s );
 1510+ #get the first element (all text up to first [[), and remove the space we added
 1511+ $s = array_shift( $a );
 1512+ $s = substr( $s, 1 );
 1513+
 1514+ # Match a link having the form [[namespace:link|alternate]]trail
 1515+ static $e1 = FALSE;
 1516+ if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
 1517+ # Match cases where there is no "]]", which might still be images
 1518+ static $e1_img = FALSE;
 1519+ if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
 1520+ # Match the end of a line for a word that's not followed by whitespace,
 1521+ # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
 1522+ $e2 = wfMsgForContent( 'linkprefix' );
 1523+
 1524+ $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
 1525+ if( is_null( $this->mTitle ) ) {
 1526+ throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
 1527+ }
 1528+ $nottalk = !$this->mTitle->isTalkPage();
 1529+
 1530+ if ( $useLinkPrefixExtension ) {
 1531+ $m = array();
 1532+ if ( preg_match( $e2, $s, $m ) ) {
 1533+ $first_prefix = $m[2];
 1534+ } else {
 1535+ $first_prefix = false;
 1536+ }
 1537+ } else {
 1538+ $prefix = '';
 1539+ }
 1540+
 1541+ if($wgContLang->hasVariants()) {
 1542+ $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
 1543+ } else {
 1544+ $selflink = array($this->mTitle->getPrefixedText());
 1545+ }
 1546+ $useSubpages = $this->areSubpagesAllowed();
 1547+ wfProfileOut( $fname.'-setup' );
 1548+
 1549+ # Loop for each link
 1550+ for ($k = 0; isset( $a[$k] ); $k++) {
 1551+ $line = $a[$k];
 1552+ if ( $useLinkPrefixExtension ) {
 1553+ wfProfileIn( $fname.'-prefixhandling' );
 1554+ if ( preg_match( $e2, $s, $m ) ) {
 1555+ $prefix = $m[2];
 1556+ $s = $m[1];
 1557+ } else {
 1558+ $prefix='';
 1559+ }
 1560+ # first link
 1561+ if($first_prefix) {
 1562+ $prefix = $first_prefix;
 1563+ $first_prefix = false;
 1564+ }
 1565+ wfProfileOut( $fname.'-prefixhandling' );
 1566+ }
 1567+
 1568+ $might_be_img = false;
 1569+
 1570+ wfProfileIn( "$fname-e1" );
 1571+ if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
 1572+ $text = $m[2];
 1573+ # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
 1574+ # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
 1575+ # the real problem is with the $e1 regex
 1576+ # See bug 1300.
 1577+ #
 1578+ # Still some problems for cases where the ] is meant to be outside punctuation,
 1579+ # and no image is in sight. See bug 2095.
 1580+ #
 1581+ if( $text !== '' &&
 1582+ substr( $m[3], 0, 1 ) === ']' &&
 1583+ strpos($text, '[') !== false
 1584+ )
 1585+ {
 1586+ $text .= ']'; # so that replaceExternalLinks($text) works later
 1587+ $m[3] = substr( $m[3], 1 );
 1588+ }
 1589+ # fix up urlencoded title texts
 1590+ if( strpos( $m[1], '%' ) !== false ) {
 1591+ # Should anchors '#' also be rejected?
 1592+ $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
 1593+ }
 1594+ $trail = $m[3];
 1595+ } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
 1596+ $might_be_img = true;
 1597+ $text = $m[2];
 1598+ if ( strpos( $m[1], '%' ) !== false ) {
 1599+ $m[1] = urldecode($m[1]);
 1600+ }
 1601+ $trail = "";
 1602+ } else { # Invalid form; output directly
 1603+ $s .= $prefix . '[[' . $line ;
 1604+ wfProfileOut( "$fname-e1" );
 1605+ continue;
 1606+ }
 1607+ wfProfileOut( "$fname-e1" );
 1608+ wfProfileIn( "$fname-misc" );
 1609+
 1610+ # Don't allow internal links to pages containing
 1611+ # PROTO: where PROTO is a valid URL protocol; these
 1612+ # should be external links.
 1613+ if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
 1614+ $s .= $prefix . '[[' . $line ;
 1615+ continue;
 1616+ }
 1617+
 1618+ # Make subpage if necessary
 1619+ if( $useSubpages ) {
 1620+ $link = $this->maybeDoSubpageLink( $m[1], $text );
 1621+ } else {
 1622+ $link = $m[1];
 1623+ }
 1624+
 1625+ $noforce = (substr($m[1], 0, 1) != ':');
 1626+ if (!$noforce) {
 1627+ # Strip off leading ':'
 1628+ $link = substr($link, 1);
 1629+ }
 1630+
 1631+ wfProfileOut( "$fname-misc" );
 1632+ wfProfileIn( "$fname-title" );
 1633+ $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
 1634+ if( !$nt ) {
 1635+ $s .= $prefix . '[[' . $line;
 1636+ wfProfileOut( "$fname-title" );
 1637+ continue;
 1638+ }
 1639+
 1640+ $ns = $nt->getNamespace();
 1641+ $iw = $nt->getInterWiki();
 1642+ wfProfileOut( "$fname-title" );
 1643+
 1644+ if ($might_be_img) { # if this is actually an invalid link
 1645+ wfProfileIn( "$fname-might_be_img" );
 1646+ if ($ns == NS_IMAGE && $noforce) { #but might be an image
 1647+ $found = false;
 1648+ while (isset ($a[$k+1]) ) {
 1649+ #look at the next 'line' to see if we can close it there
 1650+ $spliced = array_splice( $a, $k + 1, 1 );
 1651+ $next_line = array_shift( $spliced );
 1652+ $m = explode( ']]', $next_line, 3 );
 1653+ if ( count( $m ) == 3 ) {
 1654+ # the first ]] closes the inner link, the second the image
 1655+ $found = true;
 1656+ $text .= "[[{$m[0]}]]{$m[1]}";
 1657+ $trail = $m[2];
 1658+ break;
 1659+ } elseif ( count( $m ) == 2 ) {
 1660+ #if there's exactly one ]] that's fine, we'll keep looking
 1661+ $text .= "[[{$m[0]}]]{$m[1]}";
 1662+ } else {
 1663+ #if $next_line is invalid too, we need look no further
 1664+ $text .= '[[' . $next_line;
 1665+ break;
 1666+ }
 1667+ }
 1668+ if ( !$found ) {
 1669+ # we couldn't find the end of this imageLink, so output it raw
 1670+ #but don't ignore what might be perfectly normal links in the text we've examined
 1671+ $text = $this->replaceInternalLinks($text);
 1672+ $s .= "{$prefix}[[$link|$text";
 1673+ # note: no $trail, because without an end, there *is* no trail
 1674+ wfProfileOut( "$fname-might_be_img" );
 1675+ continue;
 1676+ }
 1677+ } else { #it's not an image, so output it raw
 1678+ $s .= "{$prefix}[[$link|$text";
 1679+ # note: no $trail, because without an end, there *is* no trail
 1680+ wfProfileOut( "$fname-might_be_img" );
 1681+ continue;
 1682+ }
 1683+ wfProfileOut( "$fname-might_be_img" );
 1684+ }
 1685+
 1686+ $wasblank = ( '' == $text );
 1687+ if( $wasblank ) $text = $link;
 1688+
 1689+ # Link not escaped by : , create the various objects
 1690+ if( $noforce ) {
 1691+
 1692+ # Interwikis
 1693+ wfProfileIn( "$fname-interwiki" );
 1694+ if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
 1695+ $this->mOutput->addLanguageLink( $nt->getFullText() );
 1696+ $s = rtrim($s . $prefix);
 1697+ $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
 1698+ wfProfileOut( "$fname-interwiki" );
 1699+ continue;
 1700+ }
 1701+ wfProfileOut( "$fname-interwiki" );
 1702+
 1703+ if ( $ns == NS_IMAGE ) {
 1704+ wfProfileIn( "$fname-image" );
 1705+ if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
 1706+ # recursively parse links inside the image caption
 1707+ # actually, this will parse them in any other parameters, too,
 1708+ # but it might be hard to fix that, and it doesn't matter ATM
 1709+ $text = $this->replaceExternalLinks($text);
 1710+ $text = $this->replaceInternalLinks($text);
 1711+
 1712+ # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
 1713+ $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
 1714+ $this->mOutput->addImage( $nt->getDBkey() );
 1715+
 1716+ wfProfileOut( "$fname-image" );
 1717+ continue;
 1718+ } else {
 1719+ # We still need to record the image's presence on the page
 1720+ $this->mOutput->addImage( $nt->getDBkey() );
 1721+ }
 1722+ wfProfileOut( "$fname-image" );
 1723+
 1724+ }
 1725+
 1726+ if ( $ns == NS_CATEGORY ) {
 1727+ wfProfileIn( "$fname-category" );
 1728+ $s = rtrim($s . "\n"); # bug 87
 1729+
 1730+ if ( $wasblank ) {
 1731+ $sortkey = $this->getDefaultSort();
 1732+ } else {
 1733+ $sortkey = $text;
 1734+ }
 1735+ $sortkey = Sanitizer::decodeCharReferences( $sortkey );
 1736+ $sortkey = str_replace( "\n", '', $sortkey );
 1737+ $sortkey = $wgContLang->convertCategoryKey( $sortkey );
 1738+ $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
 1739+
 1740+ /**
 1741+ * Strip the whitespace Category links produce, see bug 87
 1742+ * @todo We might want to use trim($tmp, "\n") here.
 1743+ */
 1744+ $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
 1745+
 1746+ wfProfileOut( "$fname-category" );
 1747+ continue;
 1748+ }
 1749+ }
 1750+
 1751+ # Self-link checking
 1752+ if( $nt->getFragment() === '' ) {
 1753+ if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
 1754+ $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
 1755+ continue;
 1756+ }
 1757+ }
 1758+
 1759+ # Special and Media are pseudo-namespaces; no pages actually exist in them
 1760+ if( $ns == NS_MEDIA ) {
 1761+ $link = $sk->makeMediaLinkObj( $nt, $text );
 1762+ # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
 1763+ $s .= $prefix . $this->armorLinks( $link ) . $trail;
 1764+ $this->mOutput->addImage( $nt->getDBkey() );
 1765+ continue;
 1766+ } elseif( $ns == NS_SPECIAL ) {
 1767+ if( SpecialPage::exists( $nt->getDBkey() ) ) {
 1768+ $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
 1769+ } else {
 1770+ $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
 1771+ }
 1772+ continue;
 1773+ } elseif( $ns == NS_IMAGE ) {
 1774+ $img = wfFindFile( $nt );
 1775+ if( $img ) {
 1776+ // Force a blue link if the file exists; may be a remote
 1777+ // upload on the shared repository, and we want to see its
 1778+ // auto-generated page.
 1779+ $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
 1780+ $this->mOutput->addLink( $nt );
 1781+ continue;
 1782+ }
 1783+ }
 1784+ $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
 1785+ }
 1786+ wfProfileOut( $fname );
 1787+ return $s;
 1788+ }
 1789+
 1790+ /**
 1791+ * Make a link placeholder. The text returned can be later resolved to a real link with
 1792+ * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
 1793+ * parsing of interwiki links, and secondly to allow all existence checks and
 1794+ * article length checks (for stub links) to be bundled into a single query.
 1795+ *
 1796+ */
 1797+ function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
 1798+ wfProfileIn( __METHOD__ );
 1799+ if ( ! is_object($nt) ) {
 1800+ # Fail gracefully
 1801+ $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
 1802+ } else {
 1803+ # Separate the link trail from the rest of the link
 1804+ list( $inside, $trail ) = Linker::splitTrail( $trail );
 1805+
 1806+ if ( $nt->isExternal() ) {
 1807+ $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
 1808+ $this->mInterwikiLinkHolders['titles'][] = $nt;
 1809+ $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
 1810+ } else {
 1811+ $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
 1812+ $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
 1813+ $this->mLinkHolders['queries'][] = $query;
 1814+ $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
 1815+ $this->mLinkHolders['titles'][] = $nt;
 1816+
 1817+ $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
 1818+ }
 1819+ }
 1820+ wfProfileOut( __METHOD__ );
 1821+ return $retVal;
 1822+ }
 1823+
 1824+ /**
 1825+ * Render a forced-blue link inline; protect against double expansion of
 1826+ * URLs if we're in a mode that prepends full URL prefixes to internal links.
 1827+ * Since this little disaster has to split off the trail text to avoid
 1828+ * breaking URLs in the following text without breaking trails on the
 1829+ * wiki links, it's been made into a horrible function.
 1830+ *
 1831+ * @param Title $nt
 1832+ * @param string $text
 1833+ * @param string $query
 1834+ * @param string $trail
 1835+ * @param string $prefix
 1836+ * @return string HTML-wikitext mix oh yuck
 1837+ */
 1838+ function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
 1839+ list( $inside, $trail ) = Linker::splitTrail( $trail );
 1840+ $sk = $this->mOptions->getSkin();
 1841+ $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
 1842+ return $this->armorLinks( $link ) . $trail;
 1843+ }
 1844+
 1845+ /**
 1846+ * Insert a NOPARSE hacky thing into any inline links in a chunk that's
 1847+ * going to go through further parsing steps before inline URL expansion.
 1848+ *
 1849+ * In particular this is important when using action=render, which causes
 1850+ * full URLs to be included.
 1851+ *
 1852+ * Oh man I hate our multi-layer parser!
 1853+ *
 1854+ * @param string more-or-less HTML
 1855+ * @return string less-or-more HTML with NOPARSE bits
 1856+ */
 1857+ function armorLinks( $text ) {
 1858+ return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
 1859+ "{$this->mUniqPrefix}NOPARSE$1", $text );
 1860+ }
 1861+
 1862+ /**
 1863+ * Return true if subpage links should be expanded on this page.
 1864+ * @return bool
 1865+ */
 1866+ function areSubpagesAllowed() {
 1867+ # Some namespaces don't allow subpages
 1868+ global $wgNamespacesWithSubpages;
 1869+ return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
 1870+ }
 1871+
 1872+ /**
 1873+ * Handle link to subpage if necessary
 1874+ * @param string $target the source of the link
 1875+ * @param string &$text the link text, modified as necessary
 1876+ * @return string the full name of the link
 1877+ * @private
 1878+ */
 1879+ function maybeDoSubpageLink($target, &$text) {
 1880+ # Valid link forms:
 1881+ # Foobar -- normal
 1882+ # :Foobar -- override special treatment of prefix (images, language links)
 1883+ # /Foobar -- convert to CurrentPage/Foobar
 1884+ # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
 1885+ # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
 1886+ # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
 1887+
 1888+ $fname = 'Parser::maybeDoSubpageLink';
 1889+ wfProfileIn( $fname );
 1890+ $ret = $target; # default return value is no change
 1891+
 1892+ # Some namespaces don't allow subpages,
 1893+ # so only perform processing if subpages are allowed
 1894+ if( $this->areSubpagesAllowed() ) {
 1895+ $hash = strpos( $target, '#' );
 1896+ if( $hash !== false ) {
 1897+ $suffix = substr( $target, $hash );
 1898+ $target = substr( $target, 0, $hash );
 1899+ } else {
 1900+ $suffix = '';
 1901+ }
 1902+ # bug 7425
 1903+ $target = trim( $target );
 1904+ # Look at the first character
 1905+ if( $target != '' && $target{0} == '/' ) {
 1906+ # / at end means we don't want the slash to be shown
 1907+ $m = array();
 1908+ $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
 1909+ if( $trailingSlashes ) {
 1910+ $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
 1911+ } else {
 1912+ $noslash = substr( $target, 1 );
 1913+ }
 1914+
 1915+ $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
 1916+ if( '' === $text ) {
 1917+ $text = $target . $suffix;
 1918+ } # this might be changed for ugliness reasons
 1919+ } else {
 1920+ # check for .. subpage backlinks
 1921+ $dotdotcount = 0;
 1922+ $nodotdot = $target;
 1923+ while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
 1924+ ++$dotdotcount;
 1925+ $nodotdot = substr( $nodotdot, 3 );
 1926+ }
 1927+ if($dotdotcount > 0) {
 1928+ $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
 1929+ if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
 1930+ $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
 1931+ # / at the end means don't show full path
 1932+ if( substr( $nodotdot, -1, 1 ) == '/' ) {
 1933+ $nodotdot = substr( $nodotdot, 0, -1 );
 1934+ if( '' === $text ) {
 1935+ $text = $nodotdot . $suffix;
 1936+ }
 1937+ }
 1938+ $nodotdot = trim( $nodotdot );
 1939+ if( $nodotdot != '' ) {
 1940+ $ret .= '/' . $nodotdot;
 1941+ }
 1942+ $ret .= $suffix;
 1943+ }
 1944+ }
 1945+ }
 1946+ }
 1947+
 1948+ wfProfileOut( $fname );
 1949+ return $ret;
 1950+ }
 1951+
 1952+ /**#@+
 1953+ * Used by doBlockLevels()
 1954+ * @private
 1955+ */
 1956+ /* private */ function closeParagraph() {
 1957+ $result = '';
 1958+ if ( '' != $this->mLastSection ) {
 1959+ $result = '</' . $this->mLastSection . ">\n";
 1960+ }
 1961+ $this->mInPre = false;
 1962+ $this->mLastSection = '';
 1963+ return $result;
 1964+ }
 1965+ # getCommon() returns the length of the longest common substring
 1966+ # of both arguments, starting at the beginning of both.
 1967+ #
 1968+ /* private */ function getCommon( $st1, $st2 ) {
 1969+ $fl = strlen( $st1 );
 1970+ $shorter = strlen( $st2 );
 1971+ if ( $fl < $shorter ) { $shorter = $fl; }
 1972+
 1973+ for ( $i = 0; $i < $shorter; ++$i ) {
 1974+ if ( $st1{$i} != $st2{$i} ) { break; }
 1975+ }
 1976+ return $i;
 1977+ }
 1978+ # These next three functions open, continue, and close the list
 1979+ # element appropriate to the prefix character passed into them.
 1980+ #
 1981+ /* private */ function openList( $char ) {
 1982+ $result = $this->closeParagraph();
 1983+
 1984+ if ( '*' == $char ) { $result .= '<ul><li>'; }
 1985+ else if ( '#' == $char ) { $result .= '<ol><li>'; }
 1986+ else if ( ':' == $char ) { $result .= '<dl><dd>'; }
 1987+ else if ( ';' == $char ) {
 1988+ $result .= '<dl><dt>';
 1989+ $this->mDTopen = true;
 1990+ }
 1991+ else { $result = '<!-- ERR 1 -->'; }
 1992+
 1993+ return $result;
 1994+ }
 1995+
 1996+ /* private */ function nextItem( $char ) {
 1997+ if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
 1998+ else if ( ':' == $char || ';' == $char ) {
 1999+ $close = '</dd>';
 2000+ if ( $this->mDTopen ) { $close = '</dt>'; }
 2001+ if ( ';' == $char ) {
 2002+ $this->mDTopen = true;
 2003+ return $close . '<dt>';
 2004+ } else {
 2005+ $this->mDTopen = false;
 2006+ return $close . '<dd>';
 2007+ }
 2008+ }
 2009+ return '<!-- ERR 2 -->';
 2010+ }
 2011+
 2012+ /* private */ function closeList( $char ) {
 2013+ if ( '*' == $char ) { $text = '</li></ul>'; }
 2014+ else if ( '#' == $char ) { $text = '</li></ol>'; }
 2015+ else if ( ':' == $char ) {
 2016+ if ( $this->mDTopen ) {
 2017+ $this->mDTopen = false;
 2018+ $text = '</dt></dl>';
 2019+ } else {
 2020+ $text = '</dd></dl>';
 2021+ }
 2022+ }
 2023+ else { return '<!-- ERR 3 -->'; }
 2024+ return $text."\n";
 2025+ }
 2026+ /**#@-*/
 2027+
 2028+ /**
 2029+ * Make lists from lines starting with ':', '*', '#', etc.
 2030+ *
 2031+ * @private
 2032+ * @return string the lists rendered as HTML
 2033+ */
 2034+ function doBlockLevels( $text, $linestart ) {
 2035+ $fname = 'Parser::doBlockLevels';
 2036+ wfProfileIn( $fname );
 2037+
 2038+ # Parsing through the text line by line. The main thing
 2039+ # happening here is handling of block-level elements p, pre,
 2040+ # and making lists from lines starting with * # : etc.
 2041+ #
 2042+ $textLines = explode( "\n", $text );
 2043+
 2044+ $lastPrefix = $output = '';
 2045+ $this->mDTopen = $inBlockElem = false;
 2046+ $prefixLength = 0;
 2047+ $paragraphStack = false;
 2048+
 2049+ if ( !$linestart ) {
 2050+ $output .= array_shift( $textLines );
 2051+ }
 2052+ foreach ( $textLines as $oLine ) {
 2053+ $lastPrefixLength = strlen( $lastPrefix );
 2054+ $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
 2055+ $preOpenMatch = preg_match('/<pre/i', $oLine );
 2056+ if ( !$this->mInPre ) {
 2057+ # Multiple prefixes may abut each other for nested lists.
 2058+ $prefixLength = strspn( $oLine, '*#:;' );
 2059+ $pref = substr( $oLine, 0, $prefixLength );
 2060+
 2061+ # eh?
 2062+ $pref2 = str_replace( ';', ':', $pref );
 2063+ $t = substr( $oLine, $prefixLength );
 2064+ $this->mInPre = !empty($preOpenMatch);
 2065+ } else {
 2066+ # Don't interpret any other prefixes in preformatted text
 2067+ $prefixLength = 0;
 2068+ $pref = $pref2 = '';
 2069+ $t = $oLine;
 2070+ }
 2071+
 2072+ # List generation
 2073+ if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
 2074+ # Same as the last item, so no need to deal with nesting or opening stuff
 2075+ $output .= $this->nextItem( substr( $pref, -1 ) );
 2076+ $paragraphStack = false;
 2077+
 2078+ if ( substr( $pref, -1 ) == ';') {
 2079+ # The one nasty exception: definition lists work like this:
 2080+ # ; title : definition text
 2081+ # So we check for : in the remainder text to split up the
 2082+ # title and definition, without b0rking links.
 2083+ $term = $t2 = '';
 2084+ if ($this->findColonNoLinks($t, $term, $t2) !== false) {
 2085+ $t = $t2;
 2086+ $output .= $term . $this->nextItem( ':' );
 2087+ }
 2088+ }
 2089+ } elseif( $prefixLength || $lastPrefixLength ) {
 2090+ # Either open or close a level...
 2091+ $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
 2092+ $paragraphStack = false;
 2093+
 2094+ while( $commonPrefixLength < $lastPrefixLength ) {
 2095+ $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
 2096+ --$lastPrefixLength;
 2097+ }
 2098+ if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
 2099+ $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
 2100+ }
 2101+ while ( $prefixLength > $commonPrefixLength ) {
 2102+ $char = substr( $pref, $commonPrefixLength, 1 );
 2103+ $output .= $this->openList( $char );
 2104+
 2105+ if ( ';' == $char ) {
 2106+ # FIXME: This is dupe of code above
 2107+ if ($this->findColonNoLinks($t, $term, $t2) !== false) {
 2108+ $t = $t2;
 2109+ $output .= $term . $this->nextItem( ':' );
 2110+ }
 2111+ }
 2112+ ++$commonPrefixLength;
 2113+ }
 2114+ $lastPrefix = $pref2;
 2115+ }
 2116+ if( 0 == $prefixLength ) {
 2117+ wfProfileIn( "$fname-paragraph" );
 2118+ # No prefix (not in list)--go to paragraph mode
 2119+ // XXX: use a stack for nestable elements like span, table and div
 2120+ $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
 2121+ $closematch = preg_match(
 2122+ '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
 2123+ '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
 2124+ if ( $openmatch or $closematch ) {
 2125+ $paragraphStack = false;
 2126+ # TODO bug 5718: paragraph closed
 2127+ $output .= $this->closeParagraph();
 2128+ if ( $preOpenMatch and !$preCloseMatch ) {
 2129+ $this->mInPre = true;
 2130+ }
 2131+ if ( $closematch ) {
 2132+ $inBlockElem = false;
 2133+ } else {
 2134+ $inBlockElem = true;
 2135+ }
 2136+ } else if ( !$inBlockElem && !$this->mInPre ) {
 2137+ if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
 2138+ // pre
 2139+ if ($this->mLastSection != 'pre') {
 2140+ $paragraphStack = false;
 2141+ $output .= $this->closeParagraph().'<pre>';
 2142+ $this->mLastSection = 'pre';
 2143+ }
 2144+ $t = substr( $t, 1 );
 2145+ } else {
 2146+ // paragraph
 2147+ if ( '' == trim($t) ) {
 2148+ if ( $paragraphStack ) {
 2149+ $output .= $paragraphStack.'<br />';
 2150+ $paragraphStack = false;
 2151+ $this->mLastSection = 'p';
 2152+ } else {
 2153+ if ($this->mLastSection != 'p' ) {
 2154+ $output .= $this->closeParagraph();
 2155+ $this->mLastSection = '';
 2156+ $paragraphStack = '<p>';
 2157+ } else {
 2158+ $paragraphStack = '</p><p>';
 2159+ }
 2160+ }
 2161+ } else {
 2162+ if ( $paragraphStack ) {
 2163+ $output .= $paragraphStack;
 2164+ $paragraphStack = false;
 2165+ $this->mLastSection = 'p';
 2166+ } else if ($this->mLastSection != 'p') {
 2167+ $output .= $this->closeParagraph().'<p>';
 2168+ $this->mLastSection = 'p';
 2169+ }
 2170+ }
 2171+ }
 2172+ }
 2173+ wfProfileOut( "$fname-paragraph" );
 2174+ }
 2175+ // somewhere above we forget to get out of pre block (bug 785)
 2176+ if($preCloseMatch && $this->mInPre) {
 2177+ $this->mInPre = false;
 2178+ }
 2179+ if ($paragraphStack === false) {
 2180+ $output .= $t."\n";
 2181+ }
 2182+ }
 2183+ while ( $prefixLength ) {
 2184+ $output .= $this->closeList( $pref2{$prefixLength-1} );
 2185+ --$prefixLength;
 2186+ }
 2187+ if ( '' != $this->mLastSection ) {
 2188+ $output .= '</' . $this->mLastSection . '>';
 2189+ $this->mLastSection = '';
 2190+ }
 2191+
 2192+ wfProfileOut( $fname );
 2193+ return $output;
 2194+ }
 2195+
 2196+ /**
 2197+ * Split up a string on ':', ignoring any occurences inside tags
 2198+ * to prevent illegal overlapping.
 2199+ * @param string $str the string to split
 2200+ * @param string &$before set to everything before the ':'
 2201+ * @param string &$after set to everything after the ':'
 2202+ * return string the position of the ':', or false if none found
 2203+ */
 2204+ function findColonNoLinks($str, &$before, &$after) {
 2205+ $fname = 'Parser::findColonNoLinks';
 2206+ wfProfileIn( $fname );
 2207+
 2208+ $pos = strpos( $str, ':' );
 2209+ if( $pos === false ) {
 2210+ // Nothing to find!
 2211+ wfProfileOut( $fname );
 2212+ return false;
 2213+ }
 2214+
 2215+ $lt = strpos( $str, '<' );
 2216+ if( $lt === false || $lt > $pos ) {
 2217+ // Easy; no tag nesting to worry about
 2218+ $before = substr( $str, 0, $pos );
 2219+ $after = substr( $str, $pos+1 );
 2220+ wfProfileOut( $fname );
 2221+ return $pos;
 2222+ }
 2223+
 2224+ // Ugly state machine to walk through avoiding tags.
 2225+ $state = self::COLON_STATE_TEXT;
 2226+ $stack = 0;
 2227+ $len = strlen( $str );
 2228+ for( $i = 0; $i < $len; $i++ ) {
 2229+ $c = $str{$i};
 2230+
 2231+ switch( $state ) {
 2232+ // (Using the number is a performance hack for common cases)
 2233+ case 0: // self::COLON_STATE_TEXT:
 2234+ switch( $c ) {
 2235+ case "<":
 2236+ // Could be either a <start> tag or an </end> tag
 2237+ $state = self::COLON_STATE_TAGSTART;
 2238+ break;
 2239+ case ":":
 2240+ if( $stack == 0 ) {
 2241+ // We found it!
 2242+ $before = substr( $str, 0, $i );
 2243+ $after = substr( $str, $i + 1 );
 2244+ wfProfileOut( $fname );
 2245+ return $i;
 2246+ }
 2247+ // Embedded in a tag; don't break it.
 2248+ break;
 2249+ default:
 2250+ // Skip ahead looking for something interesting
 2251+ $colon = strpos( $str, ':', $i );
 2252+ if( $colon === false ) {
 2253+ // Nothing else interesting
 2254+ wfProfileOut( $fname );
 2255+ return false;
 2256+ }
 2257+ $lt = strpos( $str, '<', $i );
 2258+ if( $stack === 0 ) {
 2259+ if( $lt === false || $colon < $lt ) {
 2260+ // We found it!
 2261+ $before = substr( $str, 0, $colon );
 2262+ $after = substr( $str, $colon + 1 );
 2263+ wfProfileOut( $fname );
 2264+ return $i;
 2265+ }
 2266+ }
 2267+ if( $lt === false ) {
 2268+ // Nothing else interesting to find; abort!
 2269+ // We're nested, but there's no close tags left. Abort!
 2270+ break 2;
 2271+ }
 2272+ // Skip ahead to next tag start
 2273+ $i = $lt;
 2274+ $state = self::COLON_STATE_TAGSTART;
 2275+ }
 2276+ break;
 2277+ case 1: // self::COLON_STATE_TAG:
 2278+ // In a <tag>
 2279+ switch( $c ) {
 2280+ case ">":
 2281+ $stack++;
 2282+ $state = self::COLON_STATE_TEXT;
 2283+ break;
 2284+ case "/":
 2285+ // Slash may be followed by >?
 2286+ $state = self::COLON_STATE_TAGSLASH;
 2287+ break;
 2288+ default:
 2289+ // ignore
 2290+ }
 2291+ break;
 2292+ case 2: // self::COLON_STATE_TAGSTART:
 2293+ switch( $c ) {
 2294+ case "/":
 2295+ $state = self::COLON_STATE_CLOSETAG;
 2296+ break;
 2297+ case "!":
 2298+ $state = self::COLON_STATE_COMMENT;
 2299+ break;
 2300+ case ">":
 2301+ // Illegal early close? This shouldn't happen D:
 2302+ $state = self::COLON_STATE_TEXT;
 2303+ break;
 2304+ default:
 2305+ $state = self::COLON_STATE_TAG;
 2306+ }
 2307+ break;
 2308+ case 3: // self::COLON_STATE_CLOSETAG:
 2309+ // In a </tag>
 2310+ if( $c == ">" ) {
 2311+ $stack--;
 2312+ if( $stack < 0 ) {
 2313+ wfDebug( "Invalid input in $fname; too many close tags\n" );
 2314+ wfProfileOut( $fname );
 2315+ return false;
 2316+ }
 2317+ $state = self::COLON_STATE_TEXT;
 2318+ }
 2319+ break;
 2320+ case self::COLON_STATE_TAGSLASH:
 2321+ if( $c == ">" ) {
 2322+ // Yes, a self-closed tag <blah/>
 2323+ $state = self::COLON_STATE_TEXT;
 2324+ } else {
 2325+ // Probably we're jumping the gun, and this is an attribute
 2326+ $state = self::COLON_STATE_TAG;
 2327+ }
 2328+ break;
 2329+ case 5: // self::COLON_STATE_COMMENT:
 2330+ if( $c == "-" ) {
 2331+ $state = self::COLON_STATE_COMMENTDASH;
 2332+ }
 2333+ break;
 2334+ case self::COLON_STATE_COMMENTDASH:
 2335+ if( $c == "-" ) {
 2336+ $state = self::COLON_STATE_COMMENTDASHDASH;
 2337+ } else {
 2338+ $state = self::COLON_STATE_COMMENT;
 2339+ }
 2340+ break;
 2341+ case self::COLON_STATE_COMMENTDASHDASH:
 2342+ if( $c == ">" ) {
 2343+ $state = self::COLON_STATE_TEXT;
 2344+ } else {
 2345+ $state = self::COLON_STATE_COMMENT;
 2346+ }
 2347+ break;
 2348+ default:
 2349+ throw new MWException( "State machine error in $fname" );
 2350+ }
 2351+ }
 2352+ if( $stack > 0 ) {
 2353+ wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
 2354+ return false;
 2355+ }
 2356+ wfProfileOut( $fname );
 2357+ return false;
 2358+ }
 2359+
 2360+ /**
 2361+ * Return value of a magic variable (like PAGENAME)
 2362+ *
 2363+ * @private
 2364+ */
 2365+ function getVariableValue( $index ) {
 2366+ global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
 2367+
 2368+ /**
 2369+ * Some of these require message or data lookups and can be
 2370+ * expensive to check many times.
 2371+ */
 2372+ static $varCache = array();
 2373+ if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
 2374+ if ( isset( $varCache[$index] ) ) {
 2375+ return $varCache[$index];
 2376+ }
 2377+ }
 2378+
 2379+ $ts = time();
 2380+ wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
 2381+
 2382+ # Use the time zone
 2383+ global $wgLocaltimezone;
 2384+ if ( isset( $wgLocaltimezone ) ) {
 2385+ $oldtz = getenv( 'TZ' );
 2386+ putenv( 'TZ='.$wgLocaltimezone );
 2387+ }
 2388+
 2389+ wfSuppressWarnings(); // E_STRICT system time bitching
 2390+ $localTimestamp = date( 'YmdHis', $ts );
 2391+ $localMonth = date( 'm', $ts );
 2392+ $localMonthName = date( 'n', $ts );
 2393+ $localDay = date( 'j', $ts );
 2394+ $localDay2 = date( 'd', $ts );
 2395+ $localDayOfWeek = date( 'w', $ts );
 2396+ $localWeek = date( 'W', $ts );
 2397+ $localYear = date( 'Y', $ts );
 2398+ $localHour = date( 'H', $ts );
 2399+ if ( isset( $wgLocaltimezone ) ) {
 2400+ putenv( 'TZ='.$oldtz );
 2401+ }
 2402+ wfRestoreWarnings();
 2403+
 2404+ switch ( $index ) {
 2405+ case 'currentmonth':
 2406+ return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
 2407+ case 'currentmonthname':
 2408+ return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
 2409+ case 'currentmonthnamegen':
 2410+ return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
 2411+ case 'currentmonthabbrev':
 2412+ return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
 2413+ case 'currentday':
 2414+ return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
 2415+ case 'currentday2':
 2416+ return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
 2417+ case 'localmonth':
 2418+ return $varCache[$index] = $wgContLang->formatNum( $localMonth );
 2419+ case 'localmonthname':
 2420+ return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
 2421+ case 'localmonthnamegen':
 2422+ return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
 2423+ case 'localmonthabbrev':
 2424+ return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
 2425+ case 'localday':
 2426+ return $varCache[$index] = $wgContLang->formatNum( $localDay );
 2427+ case 'localday2':
 2428+ return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
 2429+ case 'pagename':
 2430+ return wfEscapeWikiText( $this->mTitle->getText() );
 2431+ case 'pagenamee':
 2432+ return $this->mTitle->getPartialURL();
 2433+ case 'fullpagename':
 2434+ return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
 2435+ case 'fullpagenamee':
 2436+ return $this->mTitle->getPrefixedURL();
 2437+ case 'subpagename':
 2438+ return wfEscapeWikiText( $this->mTitle->getSubpageText() );
 2439+ case 'subpagenamee':
 2440+ return $this->mTitle->getSubpageUrlForm();
 2441+ case 'basepagename':
 2442+ return wfEscapeWikiText( $this->mTitle->getBaseText() );
 2443+ case 'basepagenamee':
 2444+ return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
 2445+ case 'talkpagename':
 2446+ if( $this->mTitle->canTalk() ) {
 2447+ $talkPage = $this->mTitle->getTalkPage();
 2448+ return wfEscapeWikiText( $talkPage->getPrefixedText() );
 2449+ } else {
 2450+ return '';
 2451+ }
 2452+ case 'talkpagenamee':
 2453+ if( $this->mTitle->canTalk() ) {
 2454+ $talkPage = $this->mTitle->getTalkPage();
 2455+ return $talkPage->getPrefixedUrl();
 2456+ } else {
 2457+ return '';
 2458+ }
 2459+ case 'subjectpagename':
 2460+ $subjPage = $this->mTitle->getSubjectPage();
 2461+ return wfEscapeWikiText( $subjPage->getPrefixedText() );
 2462+ case 'subjectpagenamee':
 2463+ $subjPage = $this->mTitle->getSubjectPage();
 2464+ return $subjPage->getPrefixedUrl();
 2465+ case 'revisionid':
 2466+ return $this->mRevisionId;
 2467+ case 'revisionday':
 2468+ return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
 2469+ case 'revisionday2':
 2470+ return substr( $this->getRevisionTimestamp(), 6, 2 );
 2471+ case 'revisionmonth':
 2472+ return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
 2473+ case 'revisionyear':
 2474+ return substr( $this->getRevisionTimestamp(), 0, 4 );
 2475+ case 'revisiontimestamp':
 2476+ return $this->getRevisionTimestamp();
 2477+ case 'namespace':
 2478+ return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
 2479+ case 'namespacee':
 2480+ return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
 2481+ case 'talkspace':
 2482+ return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
 2483+ case 'talkspacee':
 2484+ return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
 2485+ case 'subjectspace':
 2486+ return $this->mTitle->getSubjectNsText();
 2487+ case 'subjectspacee':
 2488+ return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
 2489+ case 'currentdayname':
 2490+ return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
 2491+ case 'currentyear':
 2492+ return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
 2493+ case 'currenttime':
 2494+ return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
 2495+ case 'currenthour':
 2496+ return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
 2497+ case 'currentweek':
 2498+ // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
 2499+ // int to remove the padding
 2500+ return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
 2501+ case 'currentdow':
 2502+ return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
 2503+ case 'localdayname':
 2504+ return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
 2505+ case 'localyear':
 2506+ return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
 2507+ case 'localtime':
 2508+ return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
 2509+ case 'localhour':
 2510+ return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
 2511+ case 'localweek':
 2512+ // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
 2513+ // int to remove the padding
 2514+ return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
 2515+ case 'localdow':
 2516+ return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
 2517+ case 'numberofarticles':
 2518+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
 2519+ case 'numberoffiles':
 2520+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
 2521+ case 'numberofusers':
 2522+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
 2523+ case 'numberofpages':
 2524+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
 2525+ case 'numberofadmins':
 2526+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
 2527+ case 'numberofedits':
 2528+ return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
 2529+ case 'currenttimestamp':
 2530+ return $varCache[$index] = wfTimestampNow();
 2531+ case 'localtimestamp':
 2532+ return $varCache[$index] = $localTimestamp;
 2533+ case 'currentversion':
 2534+ return $varCache[$index] = SpecialVersion::getVersion();
 2535+ case 'sitename':
 2536+ return $wgSitename;
 2537+ case 'server':
 2538+ return $wgServer;
 2539+ case 'servername':
 2540+ return $wgServerName;
 2541+ case 'scriptpath':
 2542+ return $wgScriptPath;
 2543+ case 'directionmark':
 2544+ return $wgContLang->getDirMark();
 2545+ case 'contentlanguage':
 2546+ global $wgContLanguageCode;
 2547+ return $wgContLanguageCode;
 2548+ default:
 2549+ $ret = null;
 2550+ if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
 2551+ return $ret;
 2552+ else
 2553+ return null;
 2554+ }
 2555+ }
 2556+
 2557+ /**
 2558+ * initialise the magic variables (like CURRENTMONTHNAME)
 2559+ *
 2560+ * @private
 2561+ */
 2562+ function initialiseVariables() {
 2563+ $fname = 'Parser::initialiseVariables';
 2564+ wfProfileIn( $fname );
 2565+ $variableIDs = MagicWord::getVariableIDs();
 2566+
 2567+ $this->mVariables = array();
 2568+ foreach ( $variableIDs as $id ) {
 2569+ $mw =& MagicWord::get( $id );
 2570+ $mw->addToArray( $this->mVariables, $id );
 2571+ }
 2572+ wfProfileOut( $fname );
 2573+ }
 2574+
 2575+ /**
 2576+ * parse any parentheses in format ((title|part|part))
 2577+ * and call callbacks to get a replacement text for any found piece
 2578+ *
 2579+ * @param string $text The text to parse
 2580+ * @param array $callbacks rules in form:
 2581+ * '{' => array( # opening parentheses
 2582+ * 'end' => '}', # closing parentheses
 2583+ * 'cb' => array(2 => callback, # replacement callback to call if {{..}} is found
 2584+ * 3 => callback # replacement callback to call if {{{..}}} is found
 2585+ * )
 2586+ * )
 2587+ * 'min' => 2, # Minimum parenthesis count in cb
 2588+ * 'max' => 3, # Maximum parenthesis count in cb
 2589+ * @private
 2590+ */
 2591+ function replace_callback ($text, $callbacks) {
 2592+ wfProfileIn( __METHOD__ );
 2593+ $openingBraceStack = array(); # this array will hold a stack of parentheses which are not closed yet
 2594+ $lastOpeningBrace = -1; # last not closed parentheses
 2595+
 2596+ $validOpeningBraces = implode( '', array_keys( $callbacks ) );
 2597+
 2598+ $i = 0;
 2599+ while ( $i < strlen( $text ) ) {
 2600+ # Find next opening brace, closing brace or pipe
 2601+ if ( $lastOpeningBrace == -1 ) {
 2602+ $currentClosing = '';
 2603+ $search = $validOpeningBraces;
 2604+ } else {
 2605+ $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
 2606+ $search = $validOpeningBraces . '|' . $currentClosing;
 2607+ }
 2608+ $rule = null;
 2609+ $i += strcspn( $text, $search, $i );
 2610+ if ( $i < strlen( $text ) ) {
 2611+ if ( $text[$i] == '|' ) {
 2612+ $found = 'pipe';
 2613+ } elseif ( $text[$i] == $currentClosing ) {
 2614+ $found = 'close';
 2615+ } elseif ( isset( $callbacks[$text[$i]] ) ) {
 2616+ $found = 'open';
 2617+ $rule = $callbacks[$text[$i]];
 2618+ } else {
 2619+ # Some versions of PHP have a strcspn which stops on null characters
 2620+ # Ignore and continue
 2621+ ++$i;
 2622+ continue;
 2623+ }
 2624+ } else {
 2625+ # All done
 2626+ break;
 2627+ }
 2628+
 2629+ if ( $found == 'open' ) {
 2630+ # found opening brace, let's add it to parentheses stack
 2631+ $piece = array('brace' => $text[$i],
 2632+ 'braceEnd' => $rule['end'],
 2633+ 'title' => '',
 2634+ 'parts' => null);
 2635+
 2636+ # count opening brace characters
 2637+ $piece['count'] = strspn( $text, $piece['brace'], $i );
 2638+ $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
 2639+ $i += $piece['count'];
 2640+
 2641+ # we need to add to stack only if opening brace count is enough for one of the rules
 2642+ if ( $piece['count'] >= $rule['min'] ) {
 2643+ $lastOpeningBrace ++;
 2644+ $openingBraceStack[$lastOpeningBrace] = $piece;
 2645+ }
 2646+ } elseif ( $found == 'close' ) {
 2647+ # lets check if it is enough characters for closing brace
 2648+ $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
 2649+ $count = strspn( $text, $text[$i], $i, $maxCount );
 2650+
 2651+ # check for maximum matching characters (if there are 5 closing
 2652+ # characters, we will probably need only 3 - depending on the rules)
 2653+ $matchingCount = 0;
 2654+ $matchingCallback = null;
 2655+ $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
 2656+ if ( $count > $cbType['max'] ) {
 2657+ # The specified maximum exists in the callback array, unless the caller
 2658+ # has made an error
 2659+ $matchingCount = $cbType['max'];
 2660+ } else {
 2661+ # Count is less than the maximum
 2662+ # Skip any gaps in the callback array to find the true largest match
 2663+ # Need to use array_key_exists not isset because the callback can be null
 2664+ $matchingCount = $count;
 2665+ while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
 2666+ --$matchingCount;
 2667+ }
 2668+ }
 2669+
 2670+ if ($matchingCount <= 0) {
 2671+ $i += $count;
 2672+ continue;
 2673+ }
 2674+ $matchingCallback = $cbType['cb'][$matchingCount];
 2675+
 2676+ # let's set a title or last part (if '|' was found)
 2677+ if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
 2678+ $openingBraceStack[$lastOpeningBrace]['title'] =
 2679+ substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
 2680+ $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
 2681+ } else {
 2682+ $openingBraceStack[$lastOpeningBrace]['parts'][] =
 2683+ substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
 2684+ $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
 2685+ }
 2686+
 2687+ $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
 2688+ $pieceEnd = $i + $matchingCount;
 2689+
 2690+ if( is_callable( $matchingCallback ) ) {
 2691+ $cbArgs = array (
 2692+ 'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
 2693+ 'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
 2694+ 'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
 2695+ 'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
 2696+ );
 2697+ # finally we can call a user callback and replace piece of text
 2698+ $replaceWith = call_user_func( $matchingCallback, $cbArgs );
 2699+ $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
 2700+ $i = $pieceStart + strlen($replaceWith);
 2701+ } else {
 2702+ # null value for callback means that parentheses should be parsed, but not replaced
 2703+ $i += $matchingCount;
 2704+ }
 2705+
 2706+ # reset last opening parentheses, but keep it in case there are unused characters
 2707+ $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
 2708+ 'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
 2709+ 'count' => $openingBraceStack[$lastOpeningBrace]['count'],
 2710+ 'title' => '',
 2711+ 'parts' => null,
 2712+ 'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
 2713+ $openingBraceStack[$lastOpeningBrace--] = null;
 2714+
 2715+ if ($matchingCount < $piece['count']) {
 2716+ $piece['count'] -= $matchingCount;
 2717+ $piece['startAt'] -= $matchingCount;
 2718+ $piece['partStart'] = $piece['startAt'];
 2719+ # do we still qualify for any callback with remaining count?
 2720+ $currentCbList = $callbacks[$piece['brace']]['cb'];
 2721+ while ( $piece['count'] ) {
 2722+ if ( array_key_exists( $piece['count'], $currentCbList ) ) {
 2723+ $lastOpeningBrace++;
 2724+ $openingBraceStack[$lastOpeningBrace] = $piece;
 2725+ break;
 2726+ }
 2727+ --$piece['count'];
 2728+ }
 2729+ }
 2730+ } elseif ( $found == 'pipe' ) {
 2731+ # lets set a title if it is a first separator, or next part otherwise
 2732+ if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
 2733+ $openingBraceStack[$lastOpeningBrace]['title'] =
 2734+ substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
 2735+ $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
 2736+ $openingBraceStack[$lastOpeningBrace]['parts'] = array();
 2737+ } else {
 2738+ $openingBraceStack[$lastOpeningBrace]['parts'][] =
 2739+ substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
 2740+ $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
 2741+ }
 2742+ $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
 2743+ }
 2744+ }
 2745+
 2746+ wfProfileOut( __METHOD__ );
 2747+ return $text;
 2748+ }
 2749+
 2750+ /**
 2751+ * Replace magic variables, templates, and template arguments
 2752+ * with the appropriate text. Templates are substituted recursively,
 2753+ * taking care to avoid infinite loops.
 2754+ *
 2755+ * Note that the substitution depends on value of $mOutputType:
 2756+ * OT_WIKI: only {{subst:}} templates
 2757+ * OT_MSG: only magic variables
 2758+ * OT_HTML: all templates and magic variables
 2759+ *
 2760+ * @param string $tex The text to transform
 2761+ * @param array $args Key-value pairs representing template parameters to substitute
 2762+ * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
 2763+ * @private
 2764+ */
 2765+ function replaceVariables( $text, $args = array(), $argsOnly = false ) {
 2766+ # Prevent too big inclusions
 2767+ if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
 2768+ return $text;
 2769+ }
 2770+
 2771+ $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
 2772+ wfProfileIn( $fname );
 2773+
 2774+ # This function is called recursively. To keep track of arguments we need a stack:
 2775+ array_push( $this->mArgStack, $args );
 2776+
 2777+ $braceCallbacks = array();
 2778+ if ( !$argsOnly ) {
 2779+ $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
 2780+ }
 2781+ if ( $this->mOutputType != OT_MSG ) {
 2782+ $braceCallbacks[3] = array( &$this, 'argSubstitution' );
 2783+ }
 2784+ if ( $braceCallbacks ) {
 2785+ $callbacks = array(
 2786+ '{' => array(
 2787+ 'end' => '}',
 2788+ 'cb' => $braceCallbacks,
 2789+ 'min' => $argsOnly ? 3 : 2,
 2790+ 'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
 2791+ ),
 2792+ '[' => array(
 2793+ 'end' => ']',
 2794+ 'cb' => array(2=>null),
 2795+ 'min' => 2,
 2796+ 'max' => 2,
 2797+ )
 2798+ );
 2799+ $text = $this->replace_callback ($text, $callbacks);
 2800+
 2801+ array_pop( $this->mArgStack );
 2802+ }
 2803+ wfProfileOut( $fname );
 2804+ return $text;
 2805+ }
 2806+
 2807+ /**
 2808+ * Replace magic variables
 2809+ * @private
 2810+ */
 2811+ function variableSubstitution( $matches ) {
 2812+ global $wgContLang;
 2813+ $fname = 'Parser::variableSubstitution';
 2814+ $varname = $wgContLang->lc($matches[1]);
 2815+ wfProfileIn( $fname );
 2816+ $skip = false;
 2817+ if ( $this->mOutputType == OT_WIKI ) {
 2818+ # Do only magic variables prefixed by SUBST
 2819+ $mwSubst =& MagicWord::get( 'subst' );
 2820+ if (!$mwSubst->matchStartAndRemove( $varname ))
 2821+ $skip = true;
 2822+ # Note that if we don't substitute the variable below,
 2823+ # we don't remove the {{subst:}} magic word, in case
 2824+ # it is a template rather than a magic variable.
 2825+ }
 2826+ if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
 2827+ $id = $this->mVariables[$varname];
 2828+ # Now check if we did really match, case sensitive or not
 2829+ $mw =& MagicWord::get( $id );
 2830+ if ($mw->match($matches[1])) {
 2831+ $text = $this->getVariableValue( $id );
 2832+ $this->mOutput->mContainsOldMagic = true;
 2833+ } else {
 2834+ $text = $matches[0];
 2835+ }
 2836+ } else {
 2837+ $text = $matches[0];
 2838+ }
 2839+ wfProfileOut( $fname );
 2840+ return $text;
 2841+ }
 2842+
 2843+
 2844+ /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
 2845+ static function createAssocArgs( $args ) {
 2846+ $assocArgs = array();
 2847+ $index = 1;
 2848+ foreach( $args as $arg ) {
 2849+ $eqpos = strpos( $arg, '=' );
 2850+ if ( $eqpos === false ) {
 2851+ $assocArgs[$index++] = $arg;
 2852+ } else {
 2853+ $name = trim( substr( $arg, 0, $eqpos ) );
 2854+ $value = trim( substr( $arg, $eqpos+1 ) );
 2855+ if ( $value === false ) {
 2856+ $value = '';
 2857+ }
 2858+ if ( $name !== false ) {
 2859+ $assocArgs[$name] = $value;
 2860+ }
 2861+ }
 2862+ }
 2863+
 2864+ return $assocArgs;
 2865+ }
 2866+
 2867+ /**
 2868+ * Return the text of a template, after recursively
 2869+ * replacing any variables or templates within the template.
 2870+ *
 2871+ * @param array $piece The parts of the template
 2872+ * $piece['text']: matched text
 2873+ * $piece['title']: the title, i.e. the part before the |
 2874+ * $piece['parts']: the parameter array
 2875+ * @return string the text of the template
 2876+ * @private
 2877+ */
 2878+ function braceSubstitution( $piece ) {
 2879+ global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
 2880+ $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
 2881+ wfProfileIn( $fname );
 2882+ wfProfileIn( __METHOD__.'-setup' );
 2883+
 2884+ # Flags
 2885+ $found = false; # $text has been filled
 2886+ $nowiki = false; # wiki markup in $text should be escaped
 2887+ $noparse = false; # Unsafe HTML tags should not be stripped, etc.
 2888+ $noargs = false; # Don't replace triple-brace arguments in $text
 2889+ $replaceHeadings = false; # Make the edit section links go to the template not the article
 2890+ $headingOffset = 0; # Skip headings when number, to account for those that weren't transcluded.
 2891+ $isHTML = false; # $text is HTML, armour it against wikitext transformation
 2892+ $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
 2893+
 2894+ # Title object, where $text came from
 2895+ $title = NULL;
 2896+
 2897+ $linestart = '';
 2898+
 2899+
 2900+ # $part1 is the bit before the first |, and must contain only title characters
 2901+ # $args is a list of arguments, starting from index 0, not including $part1
 2902+
 2903+ $titleText = $part1 = $piece['title'];
 2904+ # If the third subpattern matched anything, it will start with |
 2905+
 2906+ if (null == $piece['parts']) {
 2907+ $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
 2908+ if ($replaceWith != $piece['text']) {
 2909+ $text = $replaceWith;
 2910+ $found = true;
 2911+ $noparse = true;
 2912+ $noargs = true;
 2913+ }
 2914+ }
 2915+
 2916+ $args = (null == $piece['parts']) ? array() : $piece['parts'];
 2917+ wfProfileOut( __METHOD__.'-setup' );
 2918+
 2919+ # SUBST
 2920+ wfProfileIn( __METHOD__.'-modifiers' );
 2921+ if ( !$found ) {
 2922+ $mwSubst =& MagicWord::get( 'subst' );
 2923+ if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
 2924+ # One of two possibilities is true:
 2925+ # 1) Found SUBST but not in the PST phase
 2926+ # 2) Didn't find SUBST and in the PST phase
 2927+ # In either case, return without further processing
 2928+ $text = $piece['text'];
 2929+ $found = true;
 2930+ $noparse = true;
 2931+ $noargs = true;
 2932+ }
 2933+ }
 2934+
 2935+ # MSG, MSGNW and RAW
 2936+ if ( !$found ) {
 2937+ # Check for MSGNW:
 2938+ $mwMsgnw =& MagicWord::get( 'msgnw' );
 2939+ if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
 2940+ $nowiki = true;
 2941+ } else {
 2942+ # Remove obsolete MSG:
 2943+ $mwMsg =& MagicWord::get( 'msg' );
 2944+ $mwMsg->matchStartAndRemove( $part1 );
 2945+ }
 2946+
 2947+ # Check for RAW:
 2948+ $mwRaw =& MagicWord::get( 'raw' );
 2949+ if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
 2950+ $forceRawInterwiki = true;
 2951+ }
 2952+ }
 2953+ wfProfileOut( __METHOD__.'-modifiers' );
 2954+
 2955+ //save path level before recursing into functions & templates.
 2956+ $lastPathLevel = $this->mTemplatePath;
 2957+
 2958+ # Parser functions
 2959+ if ( !$found ) {
 2960+ wfProfileIn( __METHOD__ . '-pfunc' );
 2961+
 2962+ $colonPos = strpos( $part1, ':' );
 2963+ if ( $colonPos !== false ) {
 2964+ # Case sensitive functions
 2965+ $function = substr( $part1, 0, $colonPos );
 2966+ if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
 2967+ $function = $this->mFunctionSynonyms[1][$function];
 2968+ } else {
 2969+ # Case insensitive functions
 2970+ $function = strtolower( $function );
 2971+ if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
 2972+ $function = $this->mFunctionSynonyms[0][$function];
 2973+ } else {
 2974+ $function = false;
 2975+ }
 2976+ }
 2977+ if ( $function ) {
 2978+ $funcArgs = array_map( 'trim', $args );
 2979+ $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
 2980+ $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
 2981+ $found = true;
 2982+
 2983+ // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
 2984+ //$noargs = true;
 2985+ //$noparse = true;
 2986+
 2987+ if ( is_array( $result ) ) {
 2988+ if ( isset( $result[0] ) ) {
 2989+ $text = $linestart . $result[0];
 2990+ unset( $result[0] );
 2991+ }
 2992+
 2993+ // Extract flags into the local scope
 2994+ // This allows callers to set flags such as nowiki, noparse, found, etc.
 2995+ extract( $result );
 2996+ } else {
 2997+ $text = $linestart . $result;
 2998+ }
 2999+ }
 3000+ }
 3001+ wfProfileOut( __METHOD__ . '-pfunc' );
 3002+ }
 3003+
 3004+ # Template table test
 3005+
 3006+ # Did we encounter this template already? If yes, it is in the cache
 3007+ # and we need to check for loops.
 3008+ if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
 3009+ $found = true;
 3010+
 3011+ # Infinite loop test
 3012+ if ( isset( $this->mTemplatePath[$part1] ) ) {
 3013+ $noparse = true;
 3014+ $noargs = true;
 3015+ $found = true;
 3016+ $text = $linestart .
 3017+ "[[$part1]]<!-- WARNING: template loop detected -->";
 3018+ wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
 3019+ } else {
 3020+ # set $text to cached message.
 3021+ $text = $linestart . $this->mTemplates[$piece['title']];
 3022+ #treat title for cached page the same as others
 3023+ $ns = NS_TEMPLATE;
 3024+ $subpage = '';
 3025+ $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
 3026+ if ($subpage !== '') {
 3027+ $ns = $this->mTitle->getNamespace();
 3028+ }
 3029+ $title = Title::newFromText( $part1, $ns );
 3030+ //used by include size checking
 3031+ $titleText = $title->getPrefixedText();
 3032+ //used by edit section links
 3033+ $replaceHeadings = true;
 3034+
 3035+ }
 3036+ }
 3037+
 3038+ # Load from database
 3039+ if ( !$found ) {
 3040+ wfProfileIn( __METHOD__ . '-loadtpl' );
 3041+ $ns = NS_TEMPLATE;
 3042+ # declaring $subpage directly in the function call
 3043+ # does not work correctly with references and breaks
 3044+ # {{/subpage}}-style inclusions
 3045+ $subpage = '';
 3046+ $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
 3047+ if ($subpage !== '') {
 3048+ $ns = $this->mTitle->getNamespace();
 3049+ }
 3050+ $title = Title::newFromText( $part1, $ns );
 3051+
 3052+
 3053+ if ( !is_null( $title ) ) {
 3054+ $titleText = $title->getPrefixedText();
 3055+ # Check for language variants if the template is not found
 3056+ if($wgContLang->hasVariants() && $title->getArticleID() == 0){
 3057+ $wgContLang->findVariantLink($part1, $title);
 3058+ }
 3059+
 3060+ if ( !$title->isExternal() ) {
 3061+ if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
 3062+ $text = SpecialPage::capturePath( $title );
 3063+ if ( is_string( $text ) ) {
 3064+ $found = true;
 3065+ $noparse = true;
 3066+ $noargs = true;
 3067+ $isHTML = true;
 3068+ $this->disableCache();
 3069+ }
 3070+ } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
 3071+ $found = false; //access denied
 3072+ wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
 3073+ } else {
 3074+ list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
 3075+ if ( $articleContent !== false ) {
 3076+ $found = true;
 3077+ $text = $articleContent;
 3078+ $replaceHeadings = true;
 3079+ }
 3080+ }
 3081+
 3082+ # If the title is valid but undisplayable, make a link to it
 3083+ if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
 3084+ $text = "[[:$titleText]]";
 3085+ $found = true;
 3086+ }
 3087+ } elseif ( $title->isTrans() ) {
 3088+ // Interwiki transclusion
 3089+ if ( $this->ot['html'] && !$forceRawInterwiki ) {
 3090+ $text = $this->interwikiTransclude( $title, 'render' );
 3091+ $isHTML = true;
 3092+ $noparse = true;
 3093+ } else {
 3094+ $text = $this->interwikiTransclude( $title, 'raw' );
 3095+ $replaceHeadings = true;
 3096+ }
 3097+ $found = true;
 3098+ }
 3099+
 3100+ # Template cache array insertion
 3101+ # Use the original $piece['title'] not the mangled $part1, so that
 3102+ # modifiers such as RAW: produce separate cache entries
 3103+ if( $found ) {
 3104+ if( $isHTML ) {
 3105+ // A special page; don't store it in the template cache.
 3106+ } else {
 3107+ $this->mTemplates[$piece['title']] = $text;
 3108+ }
 3109+ $text = $linestart . $text;
 3110+ }
 3111+ }
 3112+ wfProfileOut( __METHOD__ . '-loadtpl' );
 3113+ }
 3114+
 3115+ if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
 3116+ # Error, oversize inclusion
 3117+ $text = $linestart .
 3118+ "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
 3119+ $noparse = true;
 3120+ $noargs = true;
 3121+ }
 3122+
 3123+ # Recursive parsing, escaping and link table handling
 3124+ # Only for HTML output
 3125+ if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
 3126+ $text = wfEscapeWikiText( $text );
 3127+ } elseif ( !$this->ot['msg'] && $found ) {
 3128+ if ( $noargs ) {
 3129+ $assocArgs = array();
 3130+ } else {
 3131+ # Clean up argument array
 3132+ $assocArgs = self::createAssocArgs($args);
 3133+ # Add a new element to the templace recursion path
 3134+ $this->mTemplatePath[$part1] = 1;
 3135+ }
 3136+
 3137+ if ( !$noparse ) {
 3138+ # If there are any <onlyinclude> tags, only include them
 3139+ if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
 3140+ $replacer = new OnlyIncludeReplacer;
 3141+ StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
 3142+ array( &$replacer, 'replace' ), $text );
 3143+ $text = $replacer->output;
 3144+ }
 3145+ # Remove <noinclude> sections and <includeonly> tags
 3146+ $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
 3147+ $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
 3148+
 3149+ if( $this->ot['html'] || $this->ot['pre'] ) {
 3150+ # Strip <nowiki>, <pre>, etc.
 3151+ $text = $this->strip( $text, $this->mStripState );
 3152+ if ( $this->ot['html'] ) {
 3153+ $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
 3154+ } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
 3155+ $text = Sanitizer::removeHTMLcomments( $text );
 3156+ }
 3157+ }
 3158+ $text = $this->replaceVariables( $text, $assocArgs );
 3159+
 3160+ # If the template begins with a table or block-level
 3161+ # element, it should be treated as beginning a new line.
 3162+ if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
 3163+ $text = "\n" . $text;
 3164+ }
 3165+ } elseif ( !$noargs ) {
 3166+ # $noparse and !$noargs
 3167+ # Just replace the arguments, not any double-brace items
 3168+ # This is used for rendered interwiki transclusion
 3169+ $text = $this->replaceVariables( $text, $assocArgs, true );
 3170+ }
 3171+ }
 3172+ # Prune lower levels off the recursion check path
 3173+ $this->mTemplatePath = $lastPathLevel;
 3174+
 3175+ if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
 3176+ # Error, oversize inclusion
 3177+ $text = $linestart .
 3178+ "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
 3179+ $noparse = true;
 3180+ $noargs = true;
 3181+ }
 3182+
 3183+ if ( !$found ) {
 3184+ wfProfileOut( $fname );
 3185+ return $piece['text'];
 3186+ } else {
 3187+ wfProfileIn( __METHOD__ . '-placeholders' );
 3188+ if ( $isHTML ) {
 3189+ # Replace raw HTML by a placeholder
 3190+ # Add a blank line preceding, to prevent it from mucking up
 3191+ # immediately preceding headings
 3192+ $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
 3193+ } else {
 3194+ # replace ==section headers==
 3195+ # XXX this needs to go away once we have a better parser.
 3196+ if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
 3197+ if( !is_null( $title ) )
 3198+ $encodedname = base64_encode($title->getPrefixedDBkey());
 3199+ else
 3200+ $encodedname = base64_encode("");
 3201+ $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
 3202+ PREG_SPLIT_DELIM_CAPTURE);
 3203+ $text = '';
 3204+ $nsec = $headingOffset;
 3205+
 3206+ for( $i = 0; $i < count($m); $i += 2 ) {
 3207+ $text .= $m[$i];
 3208+ if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
 3209+ $hl = $m[$i + 1];
 3210+ if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
 3211+ $text .= $hl;
 3212+ continue;
 3213+ }
 3214+ $m2 = array();
 3215+ preg_match('/^(={1,6})(.*?)(={1,6}\s*?)$/m', $hl, $m2);
 3216+ $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
 3217+ . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
 3218+
 3219+ $nsec++;
 3220+ }
 3221+ }
 3222+ }
 3223+ wfProfileOut( __METHOD__ . '-placeholders' );
 3224+ }
 3225+
 3226+ # Prune lower levels off the recursion check path
 3227+ $this->mTemplatePath = $lastPathLevel;
 3228+
 3229+ if ( !$found ) {
 3230+ wfProfileOut( $fname );
 3231+ return $piece['text'];
 3232+ } else {
 3233+ wfProfileOut( $fname );
 3234+ return $text;
 3235+ }
 3236+ }
 3237+
 3238+ /**
 3239+ * Fetch the unparsed text of a template and register a reference to it.
 3240+ */
 3241+ function fetchTemplateAndTitle( $title ) {
 3242+ $templateCb = $this->mOptions->getTemplateCallback();
 3243+ $stuff = call_user_func( $templateCb, $title );
 3244+ $text = $stuff['text'];
 3245+ $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
 3246+ if ( isset( $stuff['deps'] ) ) {
 3247+ foreach ( $stuff['deps'] as $dep ) {
 3248+ $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
 3249+ }
 3250+ }
 3251+ return array($text,$finalTitle);
 3252+ }
 3253+
 3254+ function fetchTemplate( $title ) {
 3255+ $rv = $this->fetchTemplateAndtitle($title);
 3256+ return $rv[0];
 3257+ }
 3258+
 3259+ /**
 3260+ * Static function to get a template
 3261+ * Can be overridden via ParserOptions::setTemplateCallback().
 3262+ *
 3263+ * Returns an associative array:
 3264+ * text The unparsed template text
 3265+ * finalTitle (Optional) The title after following redirects
 3266+ * deps (Optional) An array of associative array dependencies:
 3267+ * title: The dependency title, to be registered in templatelinks
 3268+ * page_id: The page_id of the title
 3269+ * rev_id: The revision ID loaded
 3270+ */
 3271+ static function statelessFetchTemplate( $title ) {
 3272+ $text = $skip = false;
 3273+ $finalTitle = $title;
 3274+ $deps = array();
 3275+
 3276+ // Loop to fetch the article, with up to 1 redirect
 3277+ for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
 3278+ # Give extensions a chance to select the revision instead
 3279+ $id = false; // Assume current
 3280+ wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
 3281+
 3282+ if( $skip ) {
 3283+ $text = false;
 3284+ $deps[] = array(
 3285+ 'title' => $title,
 3286+ 'page_id' => $title->getArticleID(),
 3287+ 'rev_id' => null );
 3288+ break;
 3289+ }
 3290+ $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
 3291+ $rev_id = $rev ? $rev->getId() : 0;
 3292+
 3293+ $deps[] = array(
 3294+ 'title' => $title,
 3295+ 'page_id' => $title->getArticleID(),
 3296+ 'rev_id' => $rev_id );
 3297+
 3298+ if( $rev ) {
 3299+ $text = $rev->getText();
 3300+ } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
 3301+ global $wgLang;
 3302+ $message = $wgLang->lcfirst( $title->getText() );
 3303+ $text = wfMsgForContentNoTrans( $message );
 3304+ if( wfEmptyMsg( $message, $text ) ) {
 3305+ $text = false;
 3306+ break;
 3307+ }
 3308+ } else {
 3309+ break;
 3310+ }
 3311+ if ( $text === false ) {
 3312+ break;
 3313+ }
 3314+ // Redirect?
 3315+ $finalTitle = $title;
 3316+ $title = Title::newFromRedirect( $text );
 3317+ }
 3318+ return array(
 3319+ 'text' => $text,
 3320+ 'finalTitle' => $finalTitle,
 3321+ 'deps' => $deps );
 3322+ }
 3323+
 3324+ /**
 3325+ * Transclude an interwiki link.
 3326+ */
 3327+ function interwikiTransclude( $title, $action ) {
 3328+ global $wgEnableScaryTranscluding;
 3329+
 3330+ if (!$wgEnableScaryTranscluding)
 3331+ return wfMsg('scarytranscludedisabled');
 3332+
 3333+ $url = $title->getFullUrl( "action=$action" );
 3334+
 3335+ if (strlen($url) > 255)
 3336+ return wfMsg('scarytranscludetoolong');
 3337+ return $this->fetchScaryTemplateMaybeFromCache($url);
 3338+ }
 3339+
 3340+ function fetchScaryTemplateMaybeFromCache($url) {
 3341+ global $wgTranscludeCacheExpiry;
 3342+ $dbr = wfGetDB(DB_SLAVE);
 3343+ $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
 3344+ array('tc_url' => $url));
 3345+ if ($obj) {
 3346+ $time = $obj->tc_time;
 3347+ $text = $obj->tc_contents;
 3348+ if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
 3349+ return $text;
 3350+ }
 3351+ }
 3352+
 3353+ $text = Http::get($url);
 3354+ if (!$text)
 3355+ return wfMsg('scarytranscludefailed', $url);
 3356+
 3357+ $dbw = wfGetDB(DB_MASTER);
 3358+ $dbw->replace('transcache', array('tc_url'), array(
 3359+ 'tc_url' => $url,
 3360+ 'tc_time' => time(),
 3361+ 'tc_contents' => $text));
 3362+ return $text;
 3363+ }
 3364+
 3365+
 3366+ /**
 3367+ * Triple brace replacement -- used for template arguments
 3368+ * @private
 3369+ */
 3370+ function argSubstitution( $matches ) {
 3371+ $arg = trim( $matches['title'] );
 3372+ $text = $matches['text'];
 3373+ $inputArgs = end( $this->mArgStack );
 3374+
 3375+ if ( array_key_exists( $arg, $inputArgs ) ) {
 3376+ $text = $inputArgs[$arg];
 3377+ } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
 3378+ null != $matches['parts'] && count($matches['parts']) > 0) {
 3379+ $text = $matches['parts'][0];
 3380+ }
 3381+ if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
 3382+ $text = $matches['text'] .
 3383+ '<!-- WARNING: argument omitted, expansion size too large -->';
 3384+ }
 3385+
 3386+ return $text;
 3387+ }
 3388+
 3389+ /**
 3390+ * Increment an include size counter
 3391+ *
 3392+ * @param string $type The type of expansion
 3393+ * @param integer $size The size of the text
 3394+ * @return boolean False if this inclusion would take it over the maximum, true otherwise
 3395+ */
 3396+ function incrementIncludeSize( $type, $size ) {
 3397+ if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
 3398+ return false;
 3399+ } else {
 3400+ $this->mIncludeSizes[$type] += $size;
 3401+ return true;
 3402+ }
 3403+ }
 3404+
 3405+ /**
 3406+ * Detect __NOGALLERY__ magic word and set a placeholder
 3407+ */
 3408+ function stripNoGallery( &$text ) {
 3409+ # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
 3410+ # do not add TOC
 3411+ $mw = MagicWord::get( 'nogallery' );
 3412+ $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
 3413+ }
 3414+
 3415+ /**
 3416+ * Find the first __TOC__ magic word and set a <!--MWTOC-->
 3417+ * placeholder that will then be replaced by the real TOC in
 3418+ * ->formatHeadings, this works because at this points real
 3419+ * comments will have already been discarded by the sanitizer.
 3420+ *
 3421+ * Any additional __TOC__ magic words left over will be discarded
 3422+ * as there can only be one TOC on the page.
 3423+ */
 3424+ function stripToc( $text ) {
 3425+ # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
 3426+ # do not add TOC
 3427+ $mw = MagicWord::get( 'notoc' );
 3428+ if( $mw->matchAndRemove( $text ) ) {
 3429+ $this->mShowToc = false;
 3430+ }
 3431+
 3432+ $mw = MagicWord::get( 'toc' );
 3433+ if( $mw->match( $text ) ) {
 3434+ $this->mShowToc = true;
 3435+ $this->mForceTocPosition = true;
 3436+
 3437+ // Set a placeholder. At the end we'll fill it in with the TOC.
 3438+ $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
 3439+
 3440+ // Only keep the first one.
 3441+ $text = $mw->replace( '', $text );
 3442+ }
 3443+ return $text;
 3444+ }
 3445+
 3446+ /**
 3447+ * This function accomplishes several tasks:
 3448+ * 1) Auto-number headings if that option is enabled
 3449+ * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
 3450+ * 3) Add a Table of contents on the top for users who have enabled the option
 3451+ * 4) Auto-anchor headings
 3452+ *
 3453+ * It loops through all headlines, collects the necessary data, then splits up the
 3454+ * string and re-inserts the newly formatted headlines.
 3455+ *
 3456+ * @param string $text
 3457+ * @param boolean $isMain
 3458+ * @private
 3459+ */
 3460+ function formatHeadings( $text, $isMain=true ) {
 3461+ global $wgMaxTocLevel, $wgContLang;
 3462+
 3463+ $doNumberHeadings = $this->mOptions->getNumberHeadings();
 3464+ if( !$this->mTitle->quickUserCan( 'edit' ) ) {
 3465+ $showEditLink = 0;
 3466+ } else {
 3467+ $showEditLink = $this->mOptions->getEditSection();
 3468+ }
 3469+
 3470+ # Inhibit editsection links if requested in the page
 3471+ $esw =& MagicWord::get( 'noeditsection' );
 3472+ if( $esw->matchAndRemove( $text ) ) {
 3473+ $showEditLink = 0;
 3474+ }
 3475+
 3476+ # Get all headlines for numbering them and adding funky stuff like [edit]
 3477+ # links - this is for later, but we need the number of headlines right now
 3478+ $matches = array();
 3479+ $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
 3480+
 3481+ # if there are fewer than 4 headlines in the article, do not show TOC
 3482+ # unless it's been explicitly enabled.
 3483+ $enoughToc = $this->mShowToc &&
 3484+ (($numMatches >= 4) || $this->mForceTocPosition);
 3485+
 3486+ # Allow user to stipulate that a page should have a "new section"
 3487+ # link added via __NEWSECTIONLINK__
 3488+ $mw =& MagicWord::get( 'newsectionlink' );
 3489+ if( $mw->matchAndRemove( $text ) )
 3490+ $this->mOutput->setNewSection( true );
 3491+
 3492+ # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
 3493+ # override above conditions and always show TOC above first header
 3494+ $mw =& MagicWord::get( 'forcetoc' );
 3495+ if ($mw->matchAndRemove( $text ) ) {
 3496+ $this->mShowToc = true;
 3497+ $enoughToc = true;
 3498+ }
 3499+
 3500+ # We need this to perform operations on the HTML
 3501+ $sk = $this->mOptions->getSkin();
 3502+
 3503+ # headline counter
 3504+ $headlineCount = 0;
 3505+ $sectionCount = 0; # headlineCount excluding template sections
 3506+ $numVisible = 0;
 3507+
 3508+ # Ugh .. the TOC should have neat indentation levels which can be
 3509+ # passed to the skin functions. These are determined here
 3510+ $toc = '';
 3511+ $full = '';
 3512+ $head = array();
 3513+ $sublevelCount = array();
 3514+ $levelCount = array();
 3515+ $toclevel = 0;
 3516+ $level = 0;
 3517+ $prevlevel = 0;
 3518+ $toclevel = 0;
 3519+ $prevtoclevel = 0;
 3520+
 3521+ foreach( $matches[3] as $headline ) {
 3522+ $istemplate = 0;
 3523+ $templatetitle = '';
 3524+ $templatesection = 0;
 3525+ $numbering = '';
 3526+ $mat = array();
 3527+ if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
 3528+ $istemplate = 1;
 3529+ $templatetitle = base64_decode($mat[1]);
 3530+ $templatesection = 1 + (int)base64_decode($mat[2]);
 3531+ $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
 3532+ }
 3533+
 3534+ if( $toclevel ) {
 3535+ $prevlevel = $level;
 3536+ $prevtoclevel = $toclevel;
 3537+ }
 3538+ $level = $matches[1][$headlineCount];
 3539+
 3540+ if( $doNumberHeadings || $enoughToc ) {
 3541+
 3542+ if ( $level > $prevlevel ) {
 3543+ # Increase TOC level
 3544+ $toclevel++;
 3545+ $sublevelCount[$toclevel] = 0;
 3546+ if( $toclevel<$wgMaxTocLevel ) {
 3547+ $prevtoclevel = $toclevel;
 3548+ $toc .= $sk->tocIndent();
 3549+ $numVisible++;
 3550+ }
 3551+ }
 3552+ elseif ( $level < $prevlevel && $toclevel > 1 ) {
 3553+ # Decrease TOC level, find level to jump to
 3554+
 3555+ if ( $toclevel == 2 && $level <= $levelCount[1] ) {
 3556+ # Can only go down to level 1
 3557+ $toclevel = 1;
 3558+ } else {
 3559+ for ($i = $toclevel; $i > 0; $i--) {
 3560+ if ( $levelCount[$i] == $level ) {
 3561+ # Found last matching level
 3562+ $toclevel = $i;
 3563+ break;
 3564+ }
 3565+ elseif ( $levelCount[$i] < $level ) {
 3566+ # Found first matching level below current level
 3567+ $toclevel = $i + 1;
 3568+ break;
 3569+ }
 3570+ }
 3571+ }
 3572+ if( $toclevel<$wgMaxTocLevel ) {
 3573+ if($prevtoclevel < $wgMaxTocLevel) {
 3574+ # Unindent only if the previous toc level was shown :p
 3575+ $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
 3576+ } else {
 3577+ $toc .= $sk->tocLineEnd();
 3578+ }
 3579+ }
 3580+ }
 3581+ else {
 3582+ # No change in level, end TOC line
 3583+ if( $toclevel<$wgMaxTocLevel ) {
 3584+ $toc .= $sk->tocLineEnd();
 3585+ }
 3586+ }
 3587+
 3588+ $levelCount[$toclevel] = $level;
 3589+
 3590+ # count number of headlines for each level
 3591+ @$sublevelCount[$toclevel]++;
 3592+ $dot = 0;
 3593+ for( $i = 1; $i <= $toclevel; $i++ ) {
 3594+ if( !empty( $sublevelCount[$i] ) ) {
 3595+ if( $dot ) {
 3596+ $numbering .= '.';
 3597+ }
 3598+ $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
 3599+ $dot = 1;
 3600+ }
 3601+ }
 3602+ }
 3603+
 3604+ # The canonized header is a version of the header text safe to use for links
 3605+ # Avoid insertion of weird stuff like <math> by expanding the relevant sections
 3606+ $canonized_headline = $this->mStripState->unstripBoth( $headline );
 3607+
 3608+ # Remove link placeholders by the link text.
 3609+ # <!--LINK number-->
 3610+ # turns into
 3611+ # link text with suffix
 3612+ $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
 3613+ "\$this->mLinkHolders['texts'][\$1]",
 3614+ $canonized_headline );
 3615+ $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
 3616+ "\$this->mInterwikiLinkHolders['texts'][\$1]",
 3617+ $canonized_headline );
 3618+
 3619+ # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
 3620+ $tocline = preg_replace(
 3621+ array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
 3622+ array( '', '<$1>'),
 3623+ $canonized_headline
 3624+ );
 3625+ $tocline = trim( $tocline );
 3626+
 3627+ # For the anchor, strip out HTML-y stuff period
 3628+ $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
 3629+ $canonized_headline = trim( $canonized_headline );
 3630+
 3631+ # Save headline for section edit hint before it's escaped
 3632+ $headline_hint = $canonized_headline;
 3633+ $canonized_headline = Sanitizer::escapeId( $canonized_headline );
 3634+ $refers[$headlineCount] = $canonized_headline;
 3635+
 3636+ # count how many in assoc. array so we can track dupes in anchors
 3637+ isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
 3638+ $refcount[$headlineCount]=$refers[$canonized_headline];
 3639+
 3640+ # Don't number the heading if it is the only one (looks silly)
 3641+ if( $doNumberHeadings && count( $matches[3] ) > 1) {
 3642+ # the two are different if the line contains a link
 3643+ $headline=$numbering . ' ' . $headline;
 3644+ }
 3645+
 3646+ # Create the anchor for linking from the TOC to the section
 3647+ $anchor = $canonized_headline;
 3648+ if($refcount[$headlineCount] > 1 ) {
 3649+ $anchor .= '_' . $refcount[$headlineCount];
 3650+ }
 3651+ if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
 3652+ $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
 3653+ }
 3654+ # give headline the correct <h#> tag
 3655+ if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
 3656+ if( $istemplate )
 3657+ $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
 3658+ else
 3659+ $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
 3660+ } else {
 3661+ $editlink = '';
 3662+ }
 3663+ $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
 3664+
 3665+ $headlineCount++;
 3666+ if( !$istemplate )
 3667+ $sectionCount++;
 3668+ }
 3669+
 3670+ # Never ever show TOC if no headers
 3671+ if( $numVisible < 1 ) {
 3672+ $enoughToc = false;
 3673+ }
 3674+
 3675+ if( $enoughToc ) {
 3676+ if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
 3677+ $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
 3678+ }
 3679+ $toc = $sk->tocList( $toc );
 3680+ }
 3681+
 3682+ # split up and insert constructed headlines
 3683+
 3684+ $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
 3685+ $i = 0;
 3686+
 3687+ foreach( $blocks as $block ) {
 3688+ if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
 3689+ # This is the [edit] link that appears for the top block of text when
 3690+ # section editing is enabled
 3691+
 3692+ # Disabled because it broke block formatting
 3693+ # For example, a bullet point in the top line
 3694+ # $full .= $sk->editSectionLink(0);
 3695+ }
 3696+ $full .= $block;
 3697+ if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
 3698+ # Top anchor now in skin
 3699+ $full = $full.$toc;
 3700+ }
 3701+
 3702+ if( !empty( $head[$i] ) ) {
 3703+ $full .= $head[$i];
 3704+ }
 3705+ $i++;
 3706+ }
 3707+ if( $this->mForceTocPosition ) {
 3708+ return str_replace( '<!--MWTOC-->', $toc, $full );
 3709+ } else {
 3710+ return $full;
 3711+ }
 3712+ }
 3713+
 3714+ /**
 3715+ * Transform wiki markup when saving a page by doing \r\n -> \n
 3716+ * conversion, substitting signatures, {{subst:}} templates, etc.
 3717+ *
 3718+ * @param string $text the text to transform
 3719+ * @param Title &$title the Title object for the current article
 3720+ * @param User &$user the User object describing the current user
 3721+ * @param ParserOptions $options parsing options
 3722+ * @param bool $clearState whether to clear the parser state first
 3723+ * @return string the altered wiki markup
 3724+ * @public
 3725+ */
 3726+ function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
 3727+ $this->mOptions = $options;
 3728+ $this->mTitle =& $title;
 3729+ $this->setOutputType( OT_WIKI );
 3730+
 3731+ if ( $clearState ) {
 3732+ $this->clearState();
 3733+ }
 3734+
 3735+ $stripState = new StripState;
 3736+ $pairs = array(
 3737+ "\r\n" => "\n",
 3738+ );
 3739+ $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
 3740+ $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
 3741+ $text = $this->pstPass2( $text, $stripState, $user );
 3742+ $text = $stripState->unstripBoth( $text );
 3743+ return $text;
 3744+ }
 3745+
 3746+ /**
 3747+ * Pre-save transform helper function
 3748+ * @private
 3749+ */
 3750+ function pstPass2( $text, &$stripState, $user ) {
 3751+ global $wgContLang, $wgLocaltimezone;
 3752+
 3753+ /* Note: This is the timestamp saved as hardcoded wikitext to
 3754+ * the database, we use $wgContLang here in order to give
 3755+ * everyone the same signature and use the default one rather
 3756+ * than the one selected in each user's preferences.
 3757+ */
 3758+ if ( isset( $wgLocaltimezone ) ) {
 3759+ $oldtz = getenv( 'TZ' );
 3760+ putenv( 'TZ='.$wgLocaltimezone );
 3761+ }
 3762+ $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
 3763+ ' (' . date( 'T' ) . ')';
 3764+ if ( isset( $wgLocaltimezone ) ) {
 3765+ putenv( 'TZ='.$oldtz );
 3766+ }
 3767+
 3768+ # Variable replacement
 3769+ # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
 3770+ $text = $this->replaceVariables( $text );
 3771+
 3772+ # Strip out <nowiki> etc. added via replaceVariables
 3773+ $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
 3774+
 3775+ # Signatures
 3776+ $sigText = $this->getUserSig( $user );
 3777+ $text = strtr( $text, array(
 3778+ '~~~~~' => $d,
 3779+ '~~~~' => "$sigText $d",
 3780+ '~~~' => $sigText
 3781+ ) );
 3782+
 3783+ # Context links: [[|name]] and [[name (context)|]]
 3784+ #
 3785+ global $wgLegalTitleChars;
 3786+ $tc = "[$wgLegalTitleChars]";
 3787+ $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
 3788+
 3789+ $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/"; # [[ns:page (context)|]]
 3790+ $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
 3791+ $p2 = "/\[\[\\|($tc+)]]/"; # [[|page]]
 3792+
 3793+ # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
 3794+ $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
 3795+ $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
 3796+
 3797+ $t = $this->mTitle->getText();
 3798+ $m = array();
 3799+ if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
 3800+ $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
 3801+ } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
 3802+ $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
 3803+ } else {
 3804+ # if there's no context, don't bother duplicating the title
 3805+ $text = preg_replace( $p2, '[[\\1]]', $text );
 3806+ }
 3807+
 3808+ # Trim trailing whitespace
 3809+ $text = rtrim( $text );
 3810+
 3811+ return $text;
 3812+ }
 3813+
 3814+ /**
 3815+ * Fetch the user's signature text, if any, and normalize to
 3816+ * validated, ready-to-insert wikitext.
 3817+ *
 3818+ * @param User $user
 3819+ * @return string
 3820+ * @private
 3821+ */
 3822+ function getUserSig( &$user ) {
 3823+ global $wgMaxSigChars;
 3824+
 3825+ $username = $user->getName();
 3826+ $nickname = $user->getOption( 'nickname' );
 3827+ $nickname = $nickname === '' ? $username : $nickname;
 3828+
 3829+ if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
 3830+ $nickname = $username;
 3831+ wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
 3832+ } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
 3833+ # Sig. might contain markup; validate this
 3834+ if( $this->validateSig( $nickname ) !== false ) {
 3835+ # Validated; clean up (if needed) and return it
 3836+ return $this->cleanSig( $nickname, true );
 3837+ } else {
 3838+ # Failed to validate; fall back to the default
 3839+ $nickname = $username;
 3840+ wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
 3841+ }
 3842+ }
 3843+
 3844+ // Make sure nickname doesnt get a sig in a sig
 3845+ $nickname = $this->cleanSigInSig( $nickname );
 3846+
 3847+ # If we're still here, make it a link to the user page
 3848+ $userpage = $user->getUserPage();
 3849+ return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
 3850+ }
 3851+
 3852+ /**
 3853+ * Check that the user's signature contains no bad XML
 3854+ *
 3855+ * @param string $text
 3856+ * @return mixed An expanded string, or false if invalid.
 3857+ */
 3858+ function validateSig( $text ) {
 3859+ return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
 3860+ }
 3861+
 3862+ /**
 3863+ * Clean up signature text
 3864+ *
 3865+ * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
 3866+ * 2) Substitute all transclusions
 3867+ *
 3868+ * @param string $text
 3869+ * @param $parsing Whether we're cleaning (preferences save) or parsing
 3870+ * @return string Signature text
 3871+ */
 3872+ function cleanSig( $text, $parsing = false ) {
 3873+ global $wgTitle;
 3874+ $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
 3875+
 3876+ $substWord = MagicWord::get( 'subst' );
 3877+ $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
 3878+ $substText = '{{' . $substWord->getSynonym( 0 );
 3879+
 3880+ $text = preg_replace( $substRegex, $substText, $text );
 3881+ $text = $this->cleanSigInSig( $text );
 3882+ $text = $this->replaceVariables( $text );
 3883+
 3884+ $this->clearState();
 3885+ return $text;
 3886+ }
 3887+
 3888+ /**
 3889+ * Strip ~~~, ~~~~ and ~~~~~ out of signatures
 3890+ * @param string $text
 3891+ * @return string Signature text with /~{3,5}/ removed
 3892+ */
 3893+ function cleanSigInSig( $text ) {
 3894+ $text = preg_replace( '/~{3,5}/', '', $text );
 3895+ return $text;
 3896+ }
 3897+
 3898+ /**
 3899+ * Set up some variables which are usually set up in parse()
 3900+ * so that an external function can call some class members with confidence
 3901+ * @public
 3902+ */
 3903+ function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
 3904+ $this->mTitle =& $title;
 3905+ $this->mOptions = $options;
 3906+ $this->setOutputType( $outputType );
 3907+ if ( $clearState ) {
 3908+ $this->clearState();
 3909+ }
 3910+ }
 3911+
 3912+ /**
 3913+ * Transform a MediaWiki message by replacing magic variables.
 3914+ *
 3915+ * @param string $text the text to transform
 3916+ * @param ParserOptions $options options
 3917+ * @return string the text with variables substituted
 3918+ * @public
 3919+ */
 3920+ function transformMsg( $text, $options ) {
 3921+ global $wgTitle;
 3922+ static $executing = false;
 3923+
 3924+ $fname = "Parser::transformMsg";
 3925+
 3926+ # Guard against infinite recursion
 3927+ if ( $executing ) {
 3928+ return $text;
 3929+ }
 3930+ $executing = true;
 3931+
 3932+ wfProfileIn($fname);
 3933+
 3934+ if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
 3935+ $this->mTitle = $wgTitle;
 3936+ } else {
 3937+ $this->mTitle = Title::newFromText('msg');
 3938+ }
 3939+ $this->mOptions = $options;
 3940+ $this->setOutputType( OT_MSG );
 3941+ $this->clearState();
 3942+ $text = $this->replaceVariables( $text );
 3943+
 3944+ $executing = false;
 3945+ wfProfileOut($fname);
 3946+ return $text;
 3947+ }
 3948+
 3949+ /**
 3950+ * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
 3951+ * The callback should have the following form:
 3952+ * function myParserHook( $text, $params, &$parser ) { ... }
 3953+ *
 3954+ * Transform and return $text. Use $parser for any required context, e.g. use
 3955+ * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
 3956+ *
 3957+ * @public
 3958+ *
 3959+ * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
 3960+ * @param mixed $callback The callback function (and object) to use for the tag
 3961+ *
 3962+ * @return The old value of the mTagHooks array associated with the hook
 3963+ */
 3964+ function setHook( $tag, $callback ) {
 3965+ $tag = strtolower( $tag );
 3966+ $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
 3967+ $this->mTagHooks[$tag] = $callback;
 3968+
 3969+ return $oldVal;
 3970+ }
 3971+
 3972+ function setTransparentTagHook( $tag, $callback ) {
 3973+ $tag = strtolower( $tag );
 3974+ $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
 3975+ $this->mTransparentTagHooks[$tag] = $callback;
 3976+
 3977+ return $oldVal;
 3978+ }
 3979+
 3980+ /**
 3981+ * Create a function, e.g. {{sum:1|2|3}}
 3982+ * The callback function should have the form:
 3983+ * function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
 3984+ *
 3985+ * The callback may either return the text result of the function, or an array with the text
 3986+ * in element 0, and a number of flags in the other elements. The names of the flags are
 3987+ * specified in the keys. Valid flags are:
 3988+ * found The text returned is valid, stop processing the template. This
 3989+ * is on by default.
 3990+ * nowiki Wiki markup in the return value should be escaped
 3991+ * noparse Unsafe HTML tags should not be stripped, etc.
 3992+ * noargs Don't replace triple-brace arguments in the return value
 3993+ * isHTML The returned text is HTML, armour it against wikitext transformation
 3994+ *
 3995+ * @public
 3996+ *
 3997+ * @param string $id The magic word ID
 3998+ * @param mixed $callback The callback function (and object) to use
 3999+ * @param integer $flags a combination of the following flags:
 4000+ * SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
 4001+ *
 4002+ * @return The old callback function for this name, if any
 4003+ */
 4004+ function setFunctionHook( $id, $callback, $flags = 0 ) {
 4005+ $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
 4006+ $this->mFunctionHooks[$id] = $callback;
 4007+
 4008+ # Add to function cache
 4009+ $mw = MagicWord::get( $id );
 4010+ if( !$mw )
 4011+ throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
 4012+
 4013+ $synonyms = $mw->getSynonyms();
 4014+ $sensitive = intval( $mw->isCaseSensitive() );
 4015+
 4016+ foreach ( $synonyms as $syn ) {
 4017+ # Case
 4018+ if ( !$sensitive ) {
 4019+ $syn = strtolower( $syn );
 4020+ }
 4021+ # Add leading hash
 4022+ if ( !( $flags & SFH_NO_HASH ) ) {
 4023+ $syn = '#' . $syn;
 4024+ }
 4025+ # Remove trailing colon
 4026+ if ( substr( $syn, -1, 1 ) == ':' ) {
 4027+ $syn = substr( $syn, 0, -1 );
 4028+ }
 4029+ $this->mFunctionSynonyms[$sensitive][$syn] = $id;
 4030+ }
 4031+ return $oldVal;
 4032+ }
 4033+
 4034+ /**
 4035+ * Get all registered function hook identifiers
 4036+ *
 4037+ * @return array
 4038+ */
 4039+ function getFunctionHooks() {
 4040+ return array_keys( $this->mFunctionHooks );
 4041+ }
 4042+
 4043+ /**
 4044+ * Replace <!--LINK--> link placeholders with actual links, in the buffer
 4045+ * Placeholders created in Skin::makeLinkObj()
 4046+ * Returns an array of links found, indexed by PDBK:
 4047+ * 0 - broken
 4048+ * 1 - normal link
 4049+ * 2 - stub
 4050+ * $options is a bit field, RLH_FOR_UPDATE to select for update
 4051+ */
 4052+ function replaceLinkHolders( &$text, $options = 0 ) {
 4053+ global $wgUser;
 4054+ global $wgContLang;
 4055+
 4056+ $fname = 'Parser::replaceLinkHolders';
 4057+ wfProfileIn( $fname );
 4058+
 4059+ $pdbks = array();
 4060+ $colours = array();
 4061+ $sk = $this->mOptions->getSkin();
 4062+ $linkCache =& LinkCache::singleton();
 4063+
 4064+ if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
 4065+ wfProfileIn( $fname.'-check' );
 4066+ $dbr = wfGetDB( DB_SLAVE );
 4067+ $page = $dbr->tableName( 'page' );
 4068+ $threshold = $wgUser->getOption('stubthreshold');
 4069+
 4070+ # Sort by namespace
 4071+ asort( $this->mLinkHolders['namespaces'] );
 4072+
 4073+ # Generate query
 4074+ $query = false;
 4075+ $current = null;
 4076+ foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
 4077+ # Make title object
 4078+ $title = $this->mLinkHolders['titles'][$key];
 4079+
 4080+ # Skip invalid entries.
 4081+ # Result will be ugly, but prevents crash.
 4082+ if ( is_null( $title ) ) {
 4083+ continue;
 4084+ }
 4085+ $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
 4086+
 4087+ # Check if it's a static known link, e.g. interwiki
 4088+ if ( $title->isAlwaysKnown() ) {
 4089+ $colours[$pdbk] = 1;
 4090+ } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
 4091+ $colours[$pdbk] = 1;
 4092+ $this->mOutput->addLink( $title, $id );
 4093+ } elseif ( $linkCache->isBadLink( $pdbk ) ) {
 4094+ $colours[$pdbk] = 0;
 4095+ } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
 4096+ $colours[$pdbk] = 0;
 4097+ } else {
 4098+ # Not in the link cache, add it to the query
 4099+ if ( !isset( $current ) ) {
 4100+ $current = $ns;
 4101+ $query = "SELECT page_id, page_namespace, page_title";
 4102+ if ( $threshold > 0 ) {
 4103+ $query .= ', page_len, page_is_redirect';
 4104+ }
 4105+ $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
 4106+ } elseif ( $current != $ns ) {
 4107+ $current = $ns;
 4108+ $query .= ")) OR (page_namespace=$ns AND page_title IN(";
 4109+ } else {
 4110+ $query .= ', ';
 4111+ }
 4112+
 4113+ $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
 4114+ }
 4115+ }
 4116+ if ( $query ) {
 4117+ $query .= '))';
 4118+ if ( $options & RLH_FOR_UPDATE ) {
 4119+ $query .= ' FOR UPDATE';
 4120+ }
 4121+
 4122+ $res = $dbr->query( $query, $fname );
 4123+
 4124+ # Fetch data and form into an associative array
 4125+ # non-existent = broken
 4126+ # 1 = known
 4127+ # 2 = stub
 4128+ while ( $s = $dbr->fetchObject($res) ) {
 4129+ $title = Title::makeTitle( $s->page_namespace, $s->page_title );
 4130+ $pdbk = $title->getPrefixedDBkey();
 4131+ $linkCache->addGoodLinkObj( $s->page_id, $title );
 4132+ $this->mOutput->addLink( $title, $s->page_id );
 4133+
 4134+ $colours[$pdbk] = ( $threshold == 0 || (
 4135+ $s->page_len >= $threshold || # always true if $threshold <= 0
 4136+ $s->page_is_redirect ||
 4137+ !Namespace::isContent( $s->page_namespace ) )
 4138+ ? 1 : 2 );
 4139+ }
 4140+ }
 4141+ wfProfileOut( $fname.'-check' );
 4142+
 4143+ # Do a second query for different language variants of links and categories
 4144+ if($wgContLang->hasVariants()){
 4145+ $linkBatch = new LinkBatch();
 4146+ $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
 4147+ $categoryMap = array(); // maps $category_variant => $category (dbkeys)
 4148+ $varCategories = array(); // category replacements oldDBkey => newDBkey
 4149+
 4150+ $categories = $this->mOutput->getCategoryLinks();
 4151+
 4152+ // Add variants of links to link batch
 4153+ foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
 4154+ $title = $this->mLinkHolders['titles'][$key];
 4155+ if ( is_null( $title ) )
 4156+ continue;
 4157+
 4158+ $pdbk = $title->getPrefixedDBkey();
 4159+ $titleText = $title->getText();
 4160+
 4161+ // generate all variants of the link title text
 4162+ $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
 4163+
 4164+ // if link was not found (in first query), add all variants to query
 4165+ if ( !isset($colours[$pdbk]) ){
 4166+ foreach($allTextVariants as $textVariant){
 4167+ if($textVariant != $titleText){
 4168+ $variantTitle = Title::makeTitle( $ns, $textVariant );
 4169+ if(is_null($variantTitle)) continue;
 4170+ $linkBatch->addObj( $variantTitle );
 4171+ $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
 4172+ }
 4173+ }
 4174+ }
 4175+ }
 4176+
 4177+ // process categories, check if a category exists in some variant
 4178+ foreach( $categories as $category ){
 4179+ $variants = $wgContLang->convertLinkToAllVariants($category);
 4180+ foreach($variants as $variant){
 4181+ if($variant != $category){
 4182+ $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
 4183+ if(is_null($variantTitle)) continue;
 4184+ $linkBatch->addObj( $variantTitle );
 4185+ $categoryMap[$variant] = $category;
 4186+ }
 4187+ }
 4188+ }
 4189+
 4190+
 4191+ if(!$linkBatch->isEmpty()){
 4192+ // construct query
 4193+ $titleClause = $linkBatch->constructSet('page', $dbr);
 4194+
 4195+ $variantQuery = "SELECT page_id, page_namespace, page_title";
 4196+ if ( $threshold > 0 ) {
 4197+ $variantQuery .= ', page_len, page_is_redirect';
 4198+ }
 4199+
 4200+ $variantQuery .= " FROM $page WHERE $titleClause";
 4201+ if ( $options & RLH_FOR_UPDATE ) {
 4202+ $variantQuery .= ' FOR UPDATE';
 4203+ }
 4204+
 4205+ $varRes = $dbr->query( $variantQuery, $fname );
 4206+
 4207+ // for each found variants, figure out link holders and replace
 4208+ while ( $s = $dbr->fetchObject($varRes) ) {
 4209+
 4210+ $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
 4211+ $varPdbk = $variantTitle->getPrefixedDBkey();
 4212+ $vardbk = $variantTitle->getDBkey();
 4213+
 4214+ $holderKeys = array();
 4215+ if(isset($variantMap[$varPdbk])){
 4216+ $holderKeys = $variantMap[$varPdbk];
 4217+ $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
 4218+ $this->mOutput->addLink( $variantTitle, $s->page_id );
 4219+ }
 4220+
 4221+ // loop over link holders
 4222+ foreach($holderKeys as $key){
 4223+ $title = $this->mLinkHolders['titles'][$key];
 4224+ if ( is_null( $title ) ) continue;
 4225+
 4226+ $pdbk = $title->getPrefixedDBkey();
 4227+
 4228+ if(!isset($colours[$pdbk])){
 4229+ // found link in some of the variants, replace the link holder data
 4230+ $this->mLinkHolders['titles'][$key] = $variantTitle;
 4231+ $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
 4232+
 4233+ // set pdbk and colour
 4234+ $pdbks[$key] = $varPdbk;
 4235+ if ( $threshold > 0 ) {
 4236+ $size = $s->page_len;
 4237+ if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
 4238+ $colours[$varPdbk] = 1;
 4239+ } else {
 4240+ $colours[$varPdbk] = 2;
 4241+ }
 4242+ }
 4243+ else {
 4244+ $colours[$varPdbk] = 1;
 4245+ }
 4246+ }
 4247+ }
 4248+
 4249+ // check if the object is a variant of a category
 4250+ if(isset($categoryMap[$vardbk])){
 4251+ $oldkey = $categoryMap[$vardbk];
 4252+ if($oldkey != $vardbk)
 4253+ $varCategories[$oldkey]=$vardbk;
 4254+ }
 4255+ }
 4256+
 4257+ // rebuild the categories in original order (if there are replacements)
 4258+ if(count($varCategories)>0){
 4259+ $newCats = array();
 4260+ $originalCats = $this->mOutput->getCategories();
 4261+ foreach($originalCats as $cat => $sortkey){
 4262+ // make the replacement
 4263+ if( array_key_exists($cat,$varCategories) )
 4264+ $newCats[$varCategories[$cat]] = $sortkey;
 4265+ else $newCats[$cat] = $sortkey;
 4266+ }
 4267+ $this->mOutput->setCategoryLinks($newCats);
 4268+ }
 4269+ }
 4270+ }
 4271+
 4272+ # Construct search and replace arrays
 4273+ wfProfileIn( $fname.'-construct' );
 4274+ $replacePairs = array();
 4275+ foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
 4276+ $pdbk = $pdbks[$key];
 4277+ $searchkey = "<!--LINK $key-->";
 4278+ $title = $this->mLinkHolders['titles'][$key];
 4279+ if ( empty( $colours[$pdbk] ) ) {
 4280+ $linkCache->addBadLinkObj( $title );
 4281+ $colours[$pdbk] = 0;
 4282+ $this->mOutput->addLink( $title, 0 );
 4283+ $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
 4284+ $this->mLinkHolders['texts'][$key],
 4285+ $this->mLinkHolders['queries'][$key] );
 4286+ } elseif ( $colours[$pdbk] == 1 ) {
 4287+ $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
 4288+ $this->mLinkHolders['texts'][$key],
 4289+ $this->mLinkHolders['queries'][$key] );
 4290+ } elseif ( $colours[$pdbk] == 2 ) {
 4291+ $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
 4292+ $this->mLinkHolders['texts'][$key],
 4293+ $this->mLinkHolders['queries'][$key] );
 4294+ }
 4295+ }
 4296+ $replacer = new HashtableReplacer( $replacePairs, 1 );
 4297+ wfProfileOut( $fname.'-construct' );
 4298+
 4299+ # Do the thing
 4300+ wfProfileIn( $fname.'-replace' );
 4301+ $text = preg_replace_callback(
 4302+ '/(<!--LINK .*?-->)/',
 4303+ $replacer->cb(),
 4304+ $text);
 4305+
 4306+ wfProfileOut( $fname.'-replace' );
 4307+ }
 4308+
 4309+ # Now process interwiki link holders
 4310+ # This is quite a bit simpler than internal links
 4311+ if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
 4312+ wfProfileIn( $fname.'-interwiki' );
 4313+ # Make interwiki link HTML
 4314+ $replacePairs = array();
 4315+ foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
 4316+ $title = $this->mInterwikiLinkHolders['titles'][$key];
 4317+ $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
 4318+ }
 4319+ $replacer = new HashtableReplacer( $replacePairs, 1 );
 4320+
 4321+ $text = preg_replace_callback(
 4322+ '/<!--IWLINK (.*?)-->/',
 4323+ $replacer->cb(),
 4324+ $text );
 4325+ wfProfileOut( $fname.'-interwiki' );
 4326+ }
 4327+
 4328+ wfProfileOut( $fname );
 4329+ return $colours;
 4330+ }
 4331+
 4332+ /**
 4333+ * Replace <!--LINK--> link placeholders with plain text of links
 4334+ * (not HTML-formatted).
 4335+ * @param string $text
 4336+ * @return string
 4337+ */
 4338+ function replaceLinkHoldersText( $text ) {
 4339+ $fname = 'Parser::replaceLinkHoldersText';
 4340+ wfProfileIn( $fname );
 4341+
 4342+ $text = preg_replace_callback(
 4343+ '/<!--(LINK|IWLINK) (.*?)-->/',
 4344+ array( &$this, 'replaceLinkHoldersTextCallback' ),
 4345+ $text );
 4346+
 4347+ wfProfileOut( $fname );
 4348+ return $text;
 4349+ }
 4350+
 4351+ /**
 4352+ * @param array $matches
 4353+ * @return string
 4354+ * @private
 4355+ */
 4356+ function replaceLinkHoldersTextCallback( $matches ) {
 4357+ $type = $matches[1];
 4358+ $key = $matches[2];
 4359+ if( $type == 'LINK' ) {
 4360+ if( isset( $this->mLinkHolders['texts'][$key] ) ) {
 4361+ return $this->mLinkHolders['texts'][$key];
 4362+ }
 4363+ } elseif( $type == 'IWLINK' ) {
 4364+ if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
 4365+ return $this->mInterwikiLinkHolders['texts'][$key];
 4366+ }
 4367+ }
 4368+ return $matches[0];
 4369+ }
 4370+
 4371+ /**
 4372+ * Tag hook handler for 'pre'.
 4373+ */
 4374+ function renderPreTag( $text, $attribs ) {
 4375+ // Backwards-compatibility hack
 4376+ $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
 4377+
 4378+ $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
 4379+ return wfOpenElement( 'pre', $attribs ) .
 4380+ Xml::escapeTagsOnly( $content ) .
 4381+ '</pre>';
 4382+ }
 4383+
 4384+ /**
 4385+ * Renders an image gallery from a text with one line per image.
 4386+ * text labels may be given by using |-style alternative text. E.g.
 4387+ * Image:one.jpg|The number "1"
 4388+ * Image:tree.jpg|A tree
 4389+ * given as text will return the HTML of a gallery with two images,
 4390+ * labeled 'The number "1"' and
 4391+ * 'A tree'.
 4392+ */
 4393+ function renderImageGallery( $text, $params ) {
 4394+ $ig = new ImageGallery();
 4395+ $ig->setContextTitle( $this->mTitle );
 4396+ $ig->setShowBytes( false );
 4397+ $ig->setShowFilename( false );
 4398+ $ig->setParser( $this );
 4399+ $ig->setHideBadImages();
 4400+ $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
 4401+ $ig->useSkin( $this->mOptions->getSkin() );
 4402+ $ig->mRevisionId = $this->mRevisionId;
 4403+
 4404+ if( isset( $params['caption'] ) ) {
 4405+ $caption = $params['caption'];
 4406+ $caption = htmlspecialchars( $caption );
 4407+ $caption = $this->replaceInternalLinks( $caption );
 4408+ $ig->setCaptionHtml( $caption );
 4409+ }
 4410+ if( isset( $params['perrow'] ) ) {
 4411+ $ig->setPerRow( $params['perrow'] );
 4412+ }
 4413+ if( isset( $params['widths'] ) ) {
 4414+ $ig->setWidths( $params['widths'] );
 4415+ }
 4416+ if( isset( $params['heights'] ) ) {
 4417+ $ig->setHeights( $params['heights'] );
 4418+ }
 4419+
 4420+ wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
 4421+
 4422+ $lines = explode( "\n", $text );
 4423+ foreach ( $lines as $line ) {
 4424+ # match lines like these:
 4425+ # Image:someimage.jpg|This is some image
 4426+ $matches = array();
 4427+ preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
 4428+ # Skip empty lines
 4429+ if ( count( $matches ) == 0 ) {
 4430+ continue;
 4431+ }
 4432+ $tp = Title::newFromText( $matches[1] );
 4433+ $nt =& $tp;
 4434+ if( is_null( $nt ) ) {
 4435+ # Bogus title. Ignore these so we don't bomb out later.
 4436+ continue;
 4437+ }
 4438+ if ( isset( $matches[3] ) ) {
 4439+ $label = $matches[3];
 4440+ } else {
 4441+ $label = '';
 4442+ }
 4443+
 4444+ $pout = $this->parse( $label,
 4445+ $this->mTitle,
 4446+ $this->mOptions,
 4447+ false, // Strip whitespace...?
 4448+ false // Don't clear state!
 4449+ );
 4450+ $html = $pout->getText();
 4451+
 4452+ $ig->add( $nt, $html );
 4453+
 4454+ # Only add real images (bug #5586)
 4455+ if ( $nt->getNamespace() == NS_IMAGE ) {
 4456+ $this->mOutput->addImage( $nt->getDBkey() );
 4457+ }
 4458+ }
 4459+ return $ig->toHTML();
 4460+ }
 4461+
 4462+ function getImageParams( $handler ) {
 4463+ if ( $handler ) {
 4464+ $handlerClass = get_class( $handler );
 4465+ } else {
 4466+ $handlerClass = '';
 4467+ }
 4468+ if ( !isset( $this->mImageParams[$handlerClass] ) ) {
 4469+ // Initialise static lists
 4470+ static $internalParamNames = array(
 4471+ 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
 4472+ 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
 4473+ 'bottom', 'text-bottom' ),
 4474+ 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
 4475+ 'upright', 'border' ),
 4476+ );
 4477+ static $internalParamMap;
 4478+ if ( !$internalParamMap ) {
 4479+ $internalParamMap = array();
 4480+ foreach ( $internalParamNames as $type => $names ) {
 4481+ foreach ( $names as $name ) {
 4482+ $magicName = str_replace( '-', '_', "img_$name" );
 4483+ $internalParamMap[$magicName] = array( $type, $name );
 4484+ }
 4485+ }
 4486+ }
 4487+
 4488+ // Add handler params
 4489+ $paramMap = $internalParamMap;
 4490+ if ( $handler ) {
 4491+ $handlerParamMap = $handler->getParamMap();
 4492+ foreach ( $handlerParamMap as $magic => $paramName ) {
 4493+ $paramMap[$magic] = array( 'handler', $paramName );
 4494+ }
 4495+ }
 4496+ $this->mImageParams[$handlerClass] = $paramMap;
 4497+ $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
 4498+ }
 4499+ return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
 4500+ }
 4501+
 4502+ /**
 4503+ * Parse image options text and use it to make an image
 4504+ */
 4505+ function makeImage( $title, $options ) {
 4506+ # @TODO: let the MediaHandler specify its transform parameters
 4507+ #
 4508+ # Check if the options text is of the form "options|alt text"
 4509+ # Options are:
 4510+ # * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang
 4511+ # * left no resizing, just left align. label is used for alt= only
 4512+ # * right same, but right aligned
 4513+ # * none same, but not aligned
 4514+ # * ___px scale to ___ pixels width, no aligning. e.g. use in taxobox
 4515+ # * center center the image
 4516+ # * framed Keep original image size, no magnify-button.
 4517+ # * frameless like 'thumb' but without a frame. Keeps user preferences for width
 4518+ # * upright reduce width for upright images, rounded to full __0 px
 4519+ # * border draw a 1px border around the image
 4520+ # vertical-align values (no % or length right now):
 4521+ # * baseline
 4522+ # * sub
 4523+ # * super
 4524+ # * top
 4525+ # * text-top
 4526+ # * middle
 4527+ # * bottom
 4528+ # * text-bottom
 4529+
 4530+ $parts = array_map( 'trim', explode( '|', $options) );
 4531+ $sk = $this->mOptions->getSkin();
 4532+
 4533+ # Give extensions a chance to select the file revision for us
 4534+ $skip = $time = false;
 4535+ wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
 4536+
 4537+ if ( $skip ) {
 4538+ return $sk->makeLinkObj( $title );
 4539+ }
 4540+
 4541+ # Get parameter map
 4542+ $file = wfFindFile( $title, $time );
 4543+ $handler = $file ? $file->getHandler() : false;
 4544+
 4545+ list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
 4546+
 4547+ # Process the input parameters
 4548+ $caption = '';
 4549+ $params = array( 'frame' => array(), 'handler' => array(),
 4550+ 'horizAlign' => array(), 'vertAlign' => array() );
 4551+ foreach( $parts as $part ) {
 4552+ list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
 4553+ if ( isset( $paramMap[$magicName] ) ) {
 4554+ list( $type, $paramName ) = $paramMap[$magicName];
 4555+ $params[$type][$paramName] = $value;
 4556+
 4557+ // Special case; width and height come in one variable together
 4558+ if( $type == 'handler' && $paramName == 'width' ) {
 4559+ $m = array();
 4560+ if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
 4561+ $params[$type]['width'] = intval( $m[1] );
 4562+ $params[$type]['height'] = intval( $m[2] );
 4563+ } else {
 4564+ $params[$type]['width'] = intval( $value );
 4565+ }
 4566+ }
 4567+ } else {
 4568+ $caption = $part;
 4569+ }
 4570+ }
 4571+
 4572+ # Process alignment parameters
 4573+ if ( $params['horizAlign'] ) {
 4574+ $params['frame']['align'] = key( $params['horizAlign'] );
 4575+ }
 4576+ if ( $params['vertAlign'] ) {
 4577+ $params['frame']['valign'] = key( $params['vertAlign'] );
 4578+ }
 4579+
 4580+ # Validate the handler parameters
 4581+ if ( $handler ) {
 4582+ foreach ( $params['handler'] as $name => $value ) {
 4583+ if ( !$handler->validateParam( $name, $value ) ) {
 4584+ unset( $params['handler'][$name] );
 4585+ }
 4586+ }
 4587+ }
 4588+
 4589+ # Strip bad stuff out of the alt text
 4590+ $alt = $this->replaceLinkHoldersText( $caption );
 4591+
 4592+ # make sure there are no placeholders in thumbnail attributes
 4593+ # that are later expanded to html- so expand them now and
 4594+ # remove the tags
 4595+ $alt = $this->mStripState->unstripBoth( $alt );
 4596+ $alt = Sanitizer::stripAllTags( $alt );
 4597+
 4598+ $params['frame']['alt'] = $alt;
 4599+ $params['frame']['caption'] = $caption;
 4600+
 4601+ # Linker does the rest
 4602+ $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
 4603+
 4604+ # Give the handler a chance to modify the parser object
 4605+ if ( $handler ) {
 4606+ $handler->parserTransformHook( $this, $file );
 4607+ }
 4608+
 4609+ return $ret;
 4610+ }
 4611+
 4612+ /**
 4613+ * Set a flag in the output object indicating that the content is dynamic and
 4614+ * shouldn't be cached.
 4615+ */
 4616+ function disableCache() {
 4617+ wfDebug( "Parser output marked as uncacheable.\n" );
 4618+ $this->mOutput->mCacheTime = -1;
 4619+ }
 4620+
 4621+ /**#@+
 4622+ * Callback from the Sanitizer for expanding items found in HTML attribute
 4623+ * values, so they can be safely tested and escaped.
 4624+ * @param string $text
 4625+ * @param array $args
 4626+ * @return string
 4627+ * @private
 4628+ */
 4629+ function attributeStripCallback( &$text, $args ) {
 4630+ $text = $this->replaceVariables( $text, $args );
 4631+ $text = $this->mStripState->unstripBoth( $text );
 4632+ return $text;
 4633+ }
 4634+
 4635+ /**#@-*/
 4636+
 4637+ /**#@+
 4638+ * Accessor/mutator
 4639+ */
 4640+ function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
 4641+ function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
 4642+ function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
 4643+ /**#@-*/
 4644+
 4645+ /**#@+
 4646+ * Accessor
 4647+ */
 4648+ function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
 4649+ /**#@-*/
 4650+
 4651+
 4652+ /**
 4653+ * Break wikitext input into sections, and either pull or replace
 4654+ * some particular section's text.
 4655+ *
 4656+ * External callers should use the getSection and replaceSection methods.
 4657+ *
 4658+ * @param $text Page wikitext
 4659+ * @param $section Numbered section. 0 pulls the text before the first
 4660+ * heading; other numbers will pull the given section
 4661+ * along with its lower-level subsections.
 4662+ * @param $mode One of "get" or "replace"
 4663+ * @param $newtext Replacement text for section data.
 4664+ * @return string for "get", the extracted section text.
 4665+ * for "replace", the whole page with the section replaced.
 4666+ */
 4667+ private function extractSections( $text, $section, $mode, $newtext='' ) {
 4668+ # I.... _hope_ this is right.
 4669+ # Otherwise, sometimes we don't have things initialized properly.
 4670+ $this->clearState();
 4671+
 4672+ # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
 4673+ # comments to be stripped as well)
 4674+ $stripState = new StripState;
 4675+
 4676+ $oldOutputType = $this->mOutputType;
 4677+ $oldOptions = $this->mOptions;
 4678+ $this->mOptions = new ParserOptions();
 4679+ $this->setOutputType( OT_WIKI );
 4680+
 4681+ $striptext = $this->strip( $text, $stripState, true );
 4682+
 4683+ $this->setOutputType( $oldOutputType );
 4684+ $this->mOptions = $oldOptions;
 4685+
 4686+ # now that we can be sure that no pseudo-sections are in the source,
 4687+ # split it up by section
 4688+ $uniq = preg_quote( $this->uniqPrefix(), '/' );
 4689+ $comment = "(?:$uniq-!--.*?QINU\x07)";
 4690+ $secs = preg_split(
 4691+ "/
 4692+ (
 4693+ ^
 4694+ (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
 4695+ (=+) # Should this be limited to 6?
 4696+ .+? # Section title...
 4697+ \\2 # Ending = count must match start
 4698+ (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
 4699+ $
 4700+ |
 4701+ <h([1-6])\b.*?>
 4702+ .*?
 4703+ <\/h\\3\s*>
 4704+ )
 4705+ /mix",
 4706+ $striptext, -1,
 4707+ PREG_SPLIT_DELIM_CAPTURE);
 4708+
 4709+ if( $mode == "get" ) {
 4710+ if( $section == 0 ) {
 4711+ // "Section 0" returns the content before any other section.
 4712+ $rv = $secs[0];
 4713+ } else {
 4714+ //track missing section, will replace if found.
 4715+ $rv = $newtext;
 4716+ }
 4717+ } elseif( $mode == "replace" ) {
 4718+ if( $section == 0 ) {
 4719+ $rv = $newtext . "\n\n";
 4720+ $remainder = true;
 4721+ } else {
 4722+ $rv = $secs[0];
 4723+ $remainder = false;
 4724+ }
 4725+ }
 4726+ $count = 0;
 4727+ $sectionLevel = 0;
 4728+ for( $index = 1; $index < count( $secs ); ) {
 4729+ $headerLine = $secs[$index++];
 4730+ if( $secs[$index] ) {
 4731+ // A wiki header
 4732+ $headerLevel = strlen( $secs[$index++] );
 4733+ } else {
 4734+ // An HTML header
 4735+ $index++;
 4736+ $headerLevel = intval( $secs[$index++] );
 4737+ }
 4738+ $content = $secs[$index++];
 4739+
 4740+ $count++;
 4741+ if( $mode == "get" ) {
 4742+ if( $count == $section ) {
 4743+ $rv = $headerLine . $content;
 4744+ $sectionLevel = $headerLevel;
 4745+ } elseif( $count > $section ) {
 4746+ if( $sectionLevel && $headerLevel > $sectionLevel ) {
 4747+ $rv .= $headerLine . $content;
 4748+ } else {
 4749+ // Broke out to a higher-level section
 4750+ break;
 4751+ }
 4752+ }
 4753+ } elseif( $mode == "replace" ) {
 4754+ if( $count < $section ) {
 4755+ $rv .= $headerLine . $content;
 4756+ } elseif( $count == $section ) {
 4757+ $rv .= $newtext . "\n\n";
 4758+ $sectionLevel = $headerLevel;
 4759+ } elseif( $count > $section ) {
 4760+ if( $headerLevel <= $sectionLevel ) {
 4761+ // Passed the section's sub-parts.
 4762+ $remainder = true;
 4763+ }
 4764+ if( $remainder ) {
 4765+ $rv .= $headerLine . $content;
 4766+ }
 4767+ }
 4768+ }
 4769+ }
 4770+ if (is_string($rv))
 4771+ # reinsert stripped tags
 4772+ $rv = trim( $stripState->unstripBoth( $rv ) );
 4773+
 4774+ return $rv;
 4775+ }
 4776+
 4777+ /**
 4778+ * This function returns the text of a section, specified by a number ($section).
 4779+ * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
 4780+ * the first section before any such heading (section 0).
 4781+ *
 4782+ * If a section contains subsections, these are also returned.
 4783+ *
 4784+ * @param $text String: text to look in
 4785+ * @param $section Integer: section number
 4786+ * @param $deftext: default to return if section is not found
 4787+ * @return string text of the requested section
 4788+ */
 4789+ public function getSection( $text, $section, $deftext='' ) {
 4790+ return $this->extractSections( $text, $section, "get", $deftext );
 4791+ }
 4792+
 4793+ public function replaceSection( $oldtext, $section, $text ) {
 4794+ return $this->extractSections( $oldtext, $section, "replace", $text );
 4795+ }
 4796+
 4797+ /**
 4798+ * Get the timestamp associated with the current revision, adjusted for
 4799+ * the default server-local timestamp
 4800+ */
 4801+ function getRevisionTimestamp() {
 4802+ if ( is_null( $this->mRevisionTimestamp ) ) {
 4803+ wfProfileIn( __METHOD__ );
 4804+ global $wgContLang;
 4805+ $dbr = wfGetDB( DB_SLAVE );
 4806+ $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
 4807+ array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
 4808+
 4809+ // Normalize timestamp to internal MW format for timezone processing.
 4810+ // This has the added side-effect of replacing a null value with
 4811+ // the current time, which gives us more sensible behavior for
 4812+ // previews.
 4813+ $timestamp = wfTimestamp( TS_MW, $timestamp );
 4814+
 4815+ // The cryptic '' timezone parameter tells to use the site-default
 4816+ // timezone offset instead of the user settings.
 4817+ //
 4818+ // Since this value will be saved into the parser cache, served
 4819+ // to other users, and potentially even used inside links and such,
 4820+ // it needs to be consistent for all visitors.
 4821+ $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
 4822+
 4823+ wfProfileOut( __METHOD__ );
 4824+ }
 4825+ return $this->mRevisionTimestamp;
 4826+ }
 4827+
 4828+ /**
 4829+ * Mutator for $mDefaultSort
 4830+ *
 4831+ * @param $sort New value
 4832+ */
 4833+ public function setDefaultSort( $sort ) {
 4834+ $this->mDefaultSort = $sort;
 4835+ }
 4836+
 4837+ /**
 4838+ * Accessor for $mDefaultSort
 4839+ * Will use the title/prefixed title if none is set
 4840+ *
 4841+ * @return string
 4842+ */
 4843+ public function getDefaultSort() {
 4844+ if( $this->mDefaultSort !== false ) {
 4845+ return $this->mDefaultSort;
 4846+ } else {
 4847+ return $this->mTitle->getNamespace() == NS_CATEGORY
 4848+ ? $this->mTitle->getText()
 4849+ : $this->mTitle->getPrefixedText();
 4850+ }
 4851+ }
 4852+
 4853+ /**
 4854+ * Try to guess the section anchor name based on a wikitext fragment
 4855+ * presumably extracted from a heading, for example "Header" from
 4856+ * "== Header ==".
 4857+ */
 4858+ public function guessSectionNameFromWikiText( $text ) {
 4859+ # Strip out wikitext links(they break the anchor)
 4860+ $text = $this->stripSectionName( $text );
 4861+ $headline = Sanitizer::decodeCharReferences( $text );
 4862+ # strip out HTML
 4863+ $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
 4864+ $headline = trim( $headline );
 4865+ $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
 4866+ $replacearray = array(
 4867+ '%3A' => ':',
 4868+ '%' => '.'
 4869+ );
 4870+ return str_replace(
 4871+ array_keys( $replacearray ),
 4872+ array_values( $replacearray ),
 4873+ $sectionanchor );
 4874+ }
 4875+
 4876+ /**
 4877+ * Strips a text string of wikitext for use in a section anchor
 4878+ *
 4879+ * Accepts a text string and then removes all wikitext from the
 4880+ * string and leaves only the resultant text (i.e. the result of
 4881+ * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
 4882+ * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
 4883+ * to create valid section anchors by mimicing the output of the
 4884+ * parser when headings are parsed.
 4885+ *
 4886+ * @param $text string Text string to be stripped of wikitext
 4887+ * for use in a Section anchor
 4888+ * @return Filtered text string
 4889+ */
 4890+ public function stripSectionName( $text ) {
 4891+ # Strip internal link markup
 4892+ $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
 4893+ $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
 4894+
 4895+ # Strip external link markup (FIXME: Not Tolerant to blank link text
 4896+ # I.E. [http://www.mediawiki.org] will render as [1] or something depending
 4897+ # on how many empty links there are on the page - need to figure that out.
 4898+ $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
 4899+
 4900+ # Parse wikitext quotes (italics & bold)
 4901+ $text = $this->doQuotes($text);
 4902+
 4903+ # Strip HTML tags
 4904+ $text = StringUtils::delimiterReplace( '<', '>', '', $text );
 4905+ return $text;
 4906+ }
 4907+
 4908+ /**
 4909+ * strip/replaceVariables/unstrip for preprocessor regression testing
 4910+ */
 4911+ function srvus( $text ) {
 4912+ $text = $this->strip( $text, $this->mStripState );
 4913+ $text = $this->replaceVariables( $text );
 4914+ $text = preg_replace( '/<!--MWTEMPLATESECTION.*?-->/', '', $text );
 4915+ $text = $this->mStripState->unstripBoth( $text );
 4916+ return $text;
 4917+ }
 4918+}
 4919+
Property changes on: trunk/phase3/includes/Parser_OldPP.php
___________________________________________________________________
Added: svn:eol-style
14920 + native
Index: trunk/phase3/includes/AutoLoader.php
——@@ -7,6 +7,8 @@
88 function __autoload($className) {
99 global $wgAutoloadClasses;
1010
 11+ # Locations of core classes
 12+ # Extension classes are specified with $wgAutoloadClasses
1113 static $localClasses = array(
1214 # Includes
1315 'AjaxDispatcher' => 'includes/AjaxDispatcher.php',
——@@ -133,9 +135,11 @@
134136 'ReverseChronologicalPager' => 'includes/Pager.php',
135137 'TablePager' => 'includes/Pager.php',
136138 'Parser' => 'includes/Parser.php',
 139+ 'Parser_OldPP' => 'includes/Parser_OldPP.php',
 140+ 'Parser_DiffTest' => 'includes/Parser_DiffTest.php',
 141+ 'ParserCache' => 'includes/ParserCache.php',
137142 'ParserOutput' => 'includes/ParserOutput.php',
138143 'ParserOptions' => 'includes/ParserOptions.php',
139 - 'ParserCache' => 'includes/ParserCache.php',
140144 'PatrolLog' => 'includes/PatrolLog.php',
141145 'ProfilerSimple' => 'includes/ProfilerSimple.php',
142146 'ProfilerSimpleUDP' => 'includes/ProfilerSimpleUDP.php',
Index: trunk/phase3/includes/DefaultSettings.php
——@@ -881,6 +881,8 @@
882882 $wgMaxSigChars = 255; # Maximum number of Unicode characters in signature
883883 $wgMaxArticleSize = 2048; # Maximum article size in kilobytes
884884
 885+$wgMaxPPNodeCount = 1000000; # A complexity limit on template expansion
 886+
885887 $wgExtraSubtitle = '';
886888 $wgSiteSupportPage = ''; # A page where you users can receive donations
887889
——@@ -1875,7 +1877,7 @@
18761878 $wgTidyBin = 'tidy';
18771879 $wgTidyConf = $IP.'/includes/tidy.conf';
18781880 $wgTidyOpts = '';
1879 -$wgTidyInternal = function_exists( 'tidy_load_config' );
 1881+$wgTidyInternal = extension_loaded( 'tidy' );
18801882
18811883 /** See list of skins and their symbolic names in languages/Language.php */
18821884 $wgDefaultSkin = 'monobook';
——@@ -2782,3 +2784,19 @@
27832785 */
27842786 $wgSlaveLagWarning = 10;
27852787 $wgSlaveLagCritical = 30;
 2788+
 2789+/**
 2790+ * Parser configuration. Associative array with the following members:
 2791+ *
 2792+ * class The class name
 2793+ *
 2794+ * The entire associative array will be passed through to the constructor as
 2795+ * the first parameter. Note that only Setup.php can use this variable --
 2796+ * the configuration will change at runtime via $wgParser member functions, so
 2797+ * the contents of this variable will be out-of-date. The variable can only be
 2798+ * changed during LocalSettings.php, in particular, it can't be changed during
 2799+ * an extension setup function.
 2800+ */
 2801+$wgParserConf = array(
 2802+ 'class' => 'Parser',
 2803+);
Index: trunk/phase3/includes/ParserOptions.php
——@@ -21,7 +21,9 @@
2222 var $mTidy; # Ask for tidy cleanup
2323 var $mInterfaceMessage; # Which lang to call for PLURAL and GRAMMAR
2424 var $mMaxIncludeSize; # Maximum size of template expansions, in bytes
 25+ var $mMaxPPNodeCount; # Maximum number of nodes touched by PPFrame::expand()
2526 var $mRemoveComments; # Remove HTML comments. ONLY APPLIES TO PREPROCESS OPERATIONS
 27+ var $mTemplateCallback; # Callback for template fetching
2628
2729 var $mUser; # Stored user object, just used to initialise the skin
2830
——@@ -36,7 +38,9 @@
3739 function getTidy() { return $this->mTidy; }
3840 function getInterfaceMessage() { return $this->mInterfaceMessage; }
3941 function getMaxIncludeSize() { return $this->mMaxIncludeSize; }
 42+ function getMaxPPNodeCount() { return $this->mMaxPPNodeCount; }
4043 function getRemoveComments() { return $this->mRemoveComments; }
 44+ function getTemplateCallback() { return $this->mTemplateCallback; }
4145
4246 function getSkin() {
4347 if ( !isset( $this->mSkin ) ) {
——@@ -65,7 +69,9 @@
6670 function setSkin( $x ) { $this->mSkin = $x; }
6771 function setInterfaceMessage( $x ) { return wfSetVar( $this->mInterfaceMessage, $x); }
6872 function setMaxIncludeSize( $x ) { return wfSetVar( $this->mMaxIncludeSize, $x ); }
 73+ function setMaxPPNodeCount( $x ) { return wfSetVar( $this->mMaxPPNodeCount, $x ); }
6974 function setRemoveComments( $x ) { return wfSetVar( $this->mRemoveComments, $x ); }
 75+ function setTemplateCallback( $x ) { return wfSetVar( $this->mTemplateCallback, $x ); }
7076
7177 function __construct( $user = null ) {
7278 $this->initialiseFromUser( $user );
——@@ -83,6 +89,7 @@
8490 function initialiseFromUser( $userInput ) {
8591 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
8692 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion, $wgMaxArticleSize;
 93+ global $wgMaxPPNodeCount;
8794 $fname = 'ParserOptions::initialiseFromUser';
8895 wfProfileIn( $fname );
8996 if ( !$userInput ) {
——@@ -111,7 +118,9 @@
112119 $this->mTidy = false;
113120 $this->mInterfaceMessage = false;
114121 $this->mMaxIncludeSize = $wgMaxArticleSize * 1024;
 122+ $this->mMaxPPNodeCount = $wgMaxPPNodeCount;
115123 $this->mRemoveComments = true;
 124+ $this->mTemplateCallback = array( 'Parser', 'statelessFetchTemplate' );
116125 wfProfileOut( $fname );
117126 }
118127 }
Index: trunk/phase3/RELEASE-NOTES
——@@ -174,6 +174,71 @@
175175 * Make a better rate-limiting error message (i.e. a normal MW error,
176176 rather than an "Internal Server Error").
177177
 178+== Parser changes in 1.12 ==
 179+
 180+The parser pass order has changed from
 181+
 182+ * Extension tag strip and render
 183+ * HTML normalisation and security
 184+ * Template expansion
 185+ * Main section...
 186+
 187+to
 188+
 189+ * Template and extension tag parse to intermediate representation
 190+ * Template expansion and extension rendering
 191+ * HTML normalisation and security
 192+ * Main section...
 193+
 194+The main effect of this for the user is that the rules for uncovered syntax
 195+have changed.
 196+
 197+Uncovered main-pass syntax, such as HTML tags, are now generally valid, whereas
 198+previously in some cases they were escaped. For example, you could have "<ta" in
 199+one template, and "ble>" in another template, and put them together to make a
 200+valid <table> tag. Previously the result would have been "&lt;table&gt;".
 201+
 202+Uncovered preprocessor syntax is generally not recognised. For example, if you
 203+have "{{a" in Template:A and "b}}" in Template:B, then "{{a}}{{b}}" will be
 204+converted to a literal "{{ab}}" rather than the contents of Template:Ab. This
 205+was the case previously in HTML output mode, and is now uniformly the case in
 206+the other modes as well. HTML-style comments uncovered by template expansion
 207+will not be recognised by the preprocessor and hence will not prevent template
 208+expansion within them, but they will be stripped by the following HTML security
 209+pass.
 210+
 211+The rules for template expansion during message transformation were
 212+counterintuitive, mostly accidental and buggy. There are a few small changes in
 213+this version: for example, templates with dynamic names, as in "{{ {{a}} }}",
 214+are fully expanded as they are in HTML mode, whereas previously only the inner
 215+template was expanded. I'd like to make some larger breaking changes to message
 216+transformation, after a review of typical use cases.
 217+
 218+The header identification routines for section edit and for numbering section
 219+edit links have been merged. This removes a significant failure mode and fixes a
 220+whole category of bugs (tracked by bug #4899). Wikitext headings uncovered by
 221+template expansion or comment removal will still be rendered into a heading tag,
 222+and will get an entry in the TOC, but will not have a section edit link.
 223+HTML-style headings will also not have a section edit link. Valid wikitext
 224+headings present in the template source text will get a template section edit
 225+link. This is a major break from previous behaviour, but I believe the effects
 226+are almost entirely beneficial.
 227+
 228+The main motivation for making these changes was performance. The new two-pass
 229+preprocessor can skip "dead branches" in template expansion, such as unfollowed
 230+#switch cases and unused defaults for template arguments. This provides a
 231+significant performance improvement in template-heavy test cases taken from
 232+Wikipedia. Parser function hooks can participate in this performance improvement
 233+by using the new SFH_OBJECT_ARGS flag during registration.
 234+
 235+The pre-expand include size limit has been removed, since there's no efficient
 236+way to calculate such a figure, and it would now be meaningless for performance
 237+anyway. The "preprocessor node count" takes its place, with a generous default
 238+limit.
 239+
 240+The context in which XML-style extension tags are called has changed, so
 241+extensions which make use of the parser state may need compatibility changes.
 242+
178243 === API changes in 1.12 ===
179244
180245 Full API documentation is available at http://www.mediawiki.org/wiki/API

Follow-up revisions

RevisionCommit summaryAuthorDate
r99062Fix for bug 31374: reintroduce recursive unstrip as in r27667, somehow omitte...tstarling00:07, 6 October 2011

Status & tagging log