r16524 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r16523‎ \| r16524 \| r16525 >
Date:	20:08, 15 September 2006
Author:	brion
Status:	old
Tags:
Comment:	Merge from SerbianVariants branch, trunk 16500 vs SerbianVariants 16523
Modified paths:	/trunk/phase3/RELEASE-NOTES (modified) (history) /trunk/phase3/includes/Article.php (modified) (history) /trunk/phase3/includes/Parser.php (modified) (history) /trunk/phase3/includes/SearchEngine.php (modified) (history) /trunk/phase3/includes/SpecialSearch.php (modified) (history) /trunk/phase3/languages/Language.php (modified) (history) /trunk/phase3/languages/LanguageConverter.php (modified) (history) /trunk/phase3/languages/LanguageSr.php (modified) (history) /trunk/phase3/languages/MessagesSr_ec.php (modified) (history) /trunk/phase3/languages/MessagesSr_el.php (modified) (history) /trunk/phase3/maintenance/mysql5/tables.sql (modified) (history) /trunk/phase3/maintenance/tables.sql (modified) (history) /trunk/phase3/maintenance/updaters.inc (modified) (history)

Diff [purge]

Index: trunk/phase3/maintenance/updaters.inc
—	—	@@ -57,6 +57,7 @@
58	58	array( 'ipblocks', 'ipb_range_start', 'patch-ipb_range_start.sql' ),
59	59	array( 'site_stats', 'ss_images', 'patch-ss_images.sql' ),
60	60	array( 'ipblocks', 'ipb_anon_only', 'patch-ipb_anon_only.sql' ),
	61	+ array( 'page', 'page_no_title_convert','patch-page_no_title_convert.sql' ),
61	62	);
62	63
63	64	function rename_table( $from, $to, $patch ) {
Index: trunk/phase3/maintenance/mysql5/tables.sql
—	—	@@ -218,6 +218,10 @@
219	219	-- Uncompressed length in bytes of the page's current source text.
220	220	page_len int(8) unsigned NOT NULL,
221	221
	222	+ -- Set to 1 if the page contains __NOTITLECONVERT__ magic word.
	223	+ -- Used only for languages with variants to prevent title conversion
	224	+ page_no_title_convert bool NOT NULL default 0,
	225	+
222	226	PRIMARY KEY page_id (page_id),
223	227	UNIQUE INDEX name_title (page_namespace,page_title),
224	228
Index: trunk/phase3/maintenance/tables.sql
—	—	@@ -206,6 +206,10 @@
207	207	-- Uncompressed length in bytes of the page's current source text.
208	208	page_len int(8) unsigned NOT NULL,
209	209
	210	+ -- Set to 1 if the page contains __NOTITLECONVERT__ magic word.
	211	+ -- Used only for languages with variants to prevent title conversion
	212	+ page_no_title_convert bool NOT NULL default 0,
	213	+
210	214	PRIMARY KEY page_id (page_id),
211	215	UNIQUE INDEX name_title (page_namespace,page_title),
212	216
Index: trunk/phase3/includes/Article.php
—	—	@@ -30,6 +30,8 @@
31	31	var $mGoodAdjustment; //!<
32	32	var $mLatest; //!<
33	33	var $mMinorEdit; //!<
	34	+ var $mNoTitleConvert; //!<
	35	+ var $mNoTitleConvertParser; //!<
34	36	var $mOldId; //!<
35	37	var $mRedirectedFrom; //!<
36	38	var $mRedirectUrl; //!<
—	—	@@ -130,6 +132,8 @@
131	133	$this->mRevIdFetched = 0;
132	134	$this->mRedirectUrl = false;
133	135	$this->mLatest = false;
	136	+ $this->mNoTitleConvert = false;
	137	+ $this->mNoTitleConvertParser = false;
134	138	}
135	139
136	140	/**
—	—	@@ -267,7 +271,8 @@
268	272	'page_random',
269	273	'page_touched',
270	274	'page_latest',
271		~~- 'page_len' ) ;~~
	275	+ 'page_len',
	276	+ 'page_no_title_convert') ;
272	277	wfRunHooks( 'ArticlePageDataBefore', array( &$this , &$fields ) ) ;
273	278	$row = $dbr->selectRow( 'page',
274	279	$fields,
—	—	@@ -303,6 +308,8 @@
304	309	* @private
305	310	*/
306	311	function loadPageData( $data = 'fromdb' ) {
	312	+ global $wgContLang;
	313	+
307	314	if ( $data === 'fromdb' ) {
308	315	$dbr =& $this->getDB();
309	316	$data = $this->pageDataFromId( $dbr, $this->getId() );
—	—	@@ -320,6 +327,10 @@
321	328	$this->mTouched = wfTimestamp( TS_MW, $data->page_touched );
322	329	$this->mIsRedirect = $data->page_is_redirect;
323	330	$this->mLatest = $data->page_latest;
	331	+ $this->mNoTitleConvert = $data->page_no_title_convert;
	332	+
	333	+ if($this->mNoTitleConvert)
	334	+ $wgContLang->setNoTitleConvert();
324	335	} else {
325	336	if ( is_object( $this->mTitle ) ) {
326	337	$lc->addBadLinkObj( $this->mTitle );
—	—	@@ -984,6 +995,7 @@
985	996	'page_touched' => $dbw->timestamp(),
986	997	'page_latest' => 0, # Fill this in shortly...
987	998	'page_len' => 0, # Fill this in shortly...
	999	+ 'page_no_title_convert' => 0,
988	1000	), __METHOD__ );
989	1001	$newid = $dbw->insertId();
990	1002
—	—	@@ -1023,12 +1035,21 @@
1024	1036	'page_is_new' => ($lastRevision === 0) ? 1 : 0,
1025	1037	'page_is_redirect' => Article::isRedirect( $text ) ? 1 : 0,
1026	1038	'page_len' => strlen( $text ),
	1039	+ 'page_no_title_convert' => ($this->mNoTitleConvertParser)? 1 : 0,
1027	1040	),
1028	1041	$conditions,
1029	1042	__METHOD__ );
1030	1043
	1044	+ $succ = $dbw->affectedRows() != 0;
	1045	+
	1046	+ // check if no title magic word has been changed
	1047	+ if($succ && $this->mNoTitleConvert != $this->mNoTitleConvertParser){
	1048	+ // Clear caches
	1049	+ Article::onArticleCreate( $this->mTitle );
	1050	+ }
	1051	+
1031	1052	wfProfileOut( __METHOD__ );
1032		~~- return ( $dbw->affectedRows() != 0 );~~
	1053	+ return $succ;
1033	1054	}
1034	1055
1035	1056	/**
—	—	@@ -1194,7 +1215,7 @@
1195	1216	* @return bool success
1196	1217	*/
1197	1218	function doEdit( $text, $summary, $flags = 0 ) {
1198		~~- global $wgUser, $wgDBtransactions;~~
	1219	+ global $wgUser, $wgDBtransactions, $wgContLang;
1199	1220
1200	1221	wfProfileIn( __METHOD__ );
1201	1222	$good = true;
—	—	@@ -1221,6 +1242,15 @@
1222	1243	$isminor = ( $flags & EDIT_MINOR ) && $wgUser->isAllowed('minoredit');
1223	1244	$bot = $wgUser->isAllowed( 'bot' ) \|\| ( $flags & EDIT_FORCE_BOT );
1224	1245
	1246	+ // process the notitleconvert magic for languages with variants
	1247	+ $this->mNoTitleConvertParser = false;
	1248	+ if(sizeof($wgContLang->getVariants())>1){
	1249	+ $mw =& MagicWord::get( 'notitleconvert' );
	1250	+ if( $mw->match( $text ) ){
	1251	+ $this->mNoTitleConvertParser = true;
	1252	+ }
	1253	+ }
	1254	+
1225	1255	$text = $this->preSaveTransform( $text );
1226	1256
1227	1257	$dbw =& wfGetDB( DB_MASTER );
Index: trunk/phase3/includes/SearchEngine.php
—	—	@@ -51,6 +51,7 @@
52	52	* @private
53	53	*/
54	54	function getNearMatch( $term ) {
	55	+ global $wgContLang;
55	56	# Exact match? No need to look further.
56	57	$title = Title::newFromText( $term );
57	58	if (is_null($title))
—	—	@@ -62,33 +63,27 @@
63	64
64	65	# Now try all lower case (i.e. first letter capitalized)
65	66	#
66		~~- $title = Title::newFromText( strtolower( $term ) );~~
	67	+ $title = Title::newFromText( $wgContLang->lc( $term ) );
67	68	if ( $title->exists() ) {
68	69	return $title;
69	70	}
70	71
71	72	# Now try capitalized string
72	73	#
73		~~- $title = Title::newFromText( ucwords( strtolower( $term ) ) );~~
	74	+ $title = Title::newFromText( $wgContLang->ucwords( $term ) );
74	75	if ( $title->exists() ) {
75	76	return $title;
76	77	}
77	78
78	79	# Now try all upper case
79	80	#
80		~~- $title = Title::newFromText( strtoupper( $term ) );~~
	81	+ $title = Title::newFromText( $wgContLang->uc( $term ) );
81	82	if ( $title->exists() ) {
82	83	return $title;
83	84	}
84	85
85	86	# Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
86		~~- $title = Title::newFromText( preg_replace_callback(~~
87		~~- '/\b([\w\x80-\xff]+)\b/',~~
88		~~- create_function( '$matches', '~~
89		~~- global $wgContLang;~~
90		~~- return $wgContLang->ucfirst($matches[1]);~~
91		~~- ' ),~~
92		~~- $term ) );~~
	87	+ $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
93	88	if ( $title->exists() ) {
94	89	return $title;
95	90	}
Index: trunk/phase3/includes/Parser.php
—	—	@@ -1507,7 +1507,6 @@
1508	1508	}
1509	1509
1510	1510	$selflink = $this->mTitle->getPrefixedText();
1511		~~- $checkVariantLink = sizeof($wgContLang->getVariants())>1;~~
1512	1511	$useSubpages = $this->areSubpagesAllowed();
1513	1512	wfProfileOut( $fname.'-setup' );
1514	1513
—	—	@@ -1602,13 +1601,6 @@
1603	1602	continue;
1604	1603	}
1605	1604
1606		~~- #check other language variants of the link~~
1607		~~- #if the article does not exist~~
1608		~~- if( $checkVariantLink~~
1609		~~- && $nt->getArticleID() == 0 ) {~~
1610		~~- $wgContLang->findVariantLink($link, $nt);~~
1611		~~- }~~
1612		-
1613	1605	$ns = $nt->getNamespace();
1614	1606	$iw = $nt->getInterWiki();
1615	1607	wfProfileOut( "$fname-title" );
—	—	@@ -3897,6 +3889,7 @@
3898	3890	function replaceLinkHolders( &$text, $options = 0 ) {
3899	3891	global $wgUser;
3900	3892	global $wgOutputReplace;
	3893	+ global $wgContLang, $wgLanguageCode;
3901	3894
3902	3895	$fname = 'Parser::replaceLinkHolders';
3903	3896	wfProfileIn( $fname );
—	—	@@ -3987,6 +3980,97 @@
3988	3981	}
3989	3982	wfProfileOut( $fname.'-check' );
3990	3983
	3984	+ # Do a second query for different language variants of links (if needed)
	3985	+ if($wgContLang->hasVariants()){
	3986	+ $linkBatch = new LinkBatch();
	3987	+ $variantMap = array(); // maps $pdbkey_Variant => $pdbkey_original
	3988	+
	3989	+ // Add variants of links to link batch
	3990	+ foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
	3991	+ $title = $this->mLinkHolders['titles'][$key];
	3992	+ if ( is_null( $title ) )
	3993	+ continue;
	3994	+
	3995	+ $pdbk = $title->getPrefixedDBkey();
	3996	+
	3997	+ // add the original text into query to check for notitleconvert pages
	3998	+ $variantTitle = Title::makeTitle( $ns, $title->getText() );
	3999	+ $linkBatch->addObj( $variantTitle );
	4000	+ $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
	4001	+
	4002	+ // generate all variants of the link title text
	4003	+ $allTextVariants = $wgContLang->convertLinkToAllVariants($title->getText());
	4004	+
	4005	+ // if link was not found (in first query), add all variants to query
	4006	+ if ( !isset($colours[$pdbk]) ){
	4007	+ foreach($allTextVariants as $textVariant){
	4008	+ $variantTitle = Title::makeTitle( $ns, $textVariant );
	4009	+ if(is_null($variantTitle)) continue;
	4010	+ $linkBatch->addObj( $variantTitle );
	4011	+ $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
	4012	+ }
	4013	+ }
	4014	+ }
	4015	+
	4016	+ # construct query
	4017	+ $titleClause = $linkBatch->constructSet('page', $dbr);
	4018	+ $variantQuery = "SELECT page_id, page_namespace, page_title";
	4019	+ if ( $threshold > 0 ) {
	4020	+ $variantQuery .= ', page_len, page_is_redirect';
	4021	+ }
	4022	+ $variantQuery .= ", page_no_title_convert FROM $page WHERE $titleClause";
	4023	+ if ( $options & RLH_FOR_UPDATE ) {
	4024	+ $query .= ' FOR UPDATE';
	4025	+ }
	4026	+
	4027	+ $varRes = $dbr->query( $variantQuery, $fname );
	4028	+
	4029	+ # for each found variants, figure out link holders and replace
	4030	+ while ( $s = $dbr->fetchObject($varRes) ) {
	4031	+
	4032	+ $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
	4033	+ $varPdbk = $variantTitle->getPrefixedDBkey();
	4034	+ $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
	4035	+ $this->mOutput->addLink( $variantTitle, $s->page_id );
	4036	+
	4037	+ $noTitleConvert = $s->page_no_title_convert;
	4038	+
	4039	+ $holderKeys = $variantMap[$varPdbk];
	4040	+
	4041	+ // loop over link holders
	4042	+ foreach($holderKeys as $key){
	4043	+ $title = $this->mLinkHolders['titles'][$key];
	4044	+ if ( is_null( $title ) ) continue;
	4045	+
	4046	+ $pdbk = $title->getPrefixedDBkey();
	4047	+
	4048	+ if(!isset($colours[$pdbk]) \|\| ($noTitleConvert && $colours[$pdbk] == 1)){
	4049	+ // found link in some of the variants, replace the link holder data
	4050	+ $this->mLinkHolders['titles'][$key] = $variantTitle;
	4051	+ $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
	4052	+
	4053	+ // prevent link conversion if needed
	4054	+ if($noTitleConvert)
	4055	+ $this->mLinkHolders['texts'][$key] = $wgContLang->markNoConversion($variantTitle->getText(),true);
	4056	+
	4057	+ // set pdbk and colour
	4058	+ $pdbks[$key] = $varPdbk;
	4059	+ if ( $threshold > 0 ) {
	4060	+ $size = $s->page_len;
	4061	+ if ( $s->page_is_redirect \|\| $s->page_namespace != 0 \|\| $size >= $threshold ) {
	4062	+ $colours[$varPdbk] = 1;
	4063	+ } else {
	4064	+ $colours[$varPdbk] = 2;
	4065	+ }
	4066	+ }
	4067	+ else {
	4068	+ $colours[$varPdbk] = 1;
	4069	+ }
	4070	+ }
	4071	+ }
	4072	+ }
	4073	+ }
	4074	+
3991	4075	# Construct search and replace arrays
3992	4076	wfProfileIn( $fname.'-construct' );
3993	4077	$wgOutputReplace = array();
Index: trunk/phase3/includes/SpecialSearch.php
—	—	@@ -77,6 +77,7 @@
78	78	function goResult( $term ) {
79	79	global $wgOut;
80	80	global $wgGoToEdit;
	81	+ global $wgContLang;
81	82
82	83	$this->setupPage( $term );
83	84
—	—	@@ -96,6 +97,20 @@
97	98	return;
98	99	}
99	100
	101	+ # if language supports variants, search in all variants
	102	+ if($wgContLang->hasVariants()){
	103	+ $allTermVariants = $wgContLang->convertLinkToAllVariants($term);
	104	+
	105	+ foreach($allTermVariants as $termVariant){
	106	+ $t = SearchEngine::getNearMatch( $termVariant );
	107	+ if( !is_null( $t ) ) {
	108	+ $wgOut->redirect( $t->getFullURL() );
	109	+ wfProfileOut( $fname );
	110	+ return;
	111	+ }
	112	+ }
	113	+ }
	114	+
100	115	# No match, generate an edit URL
101	116	$t = Title::newFromText( $term );
102	117	if( is_null( $t ) ) {
Index: trunk/phase3/RELEASE-NOTES
—	—	@@ -206,6 +206,8 @@
207	207	preferences don't get stuck in proxy caches for other people
208	208	* (bug 7324) Fix error message for failure of Database::sourceFile()
209	209	* (bug 7309) Plurals: use singular form for zero in French and Brazilian Portuguese
	210	+* Add page_no_title_convert field to support language variant conversion
	211	+ for page titles which shouldn't be converted on display/linking
210	212
211	213
212	214	== Languages updated ==
Index: trunk/phase3/languages/MessagesSr_el.php
—	—	@@ -175,7 +175,7 @@
176	176	'servername' => array( 0, 'SERVERNAME', 'IMESERVERA' ),
177	177	'scriptpath' => array( 0, 'SCRIPTPATH', 'SKRIPTA' ),
178	178	'grammar' => array( 0, 'GRAMMAR:', 'GRAMATIKA:' ),
179		~~- 'notitleconvert' => array( 0, '__NOTITLECONVERT__', '__NOTC__', '__BEZTC__' ),~~
	179	+ 'notitleconvert' => array( 0, '__NOTITLECONVERT__', '__NOTC__', '__БЕЗКН__', '__BEZKN__' ),
180	180	'nocontentconvert' => array( 0, '__NOCONTENTCONVERT__', '__NOCC__', '__BEZCC__' ),
181	181	'currentweek' => array( 1, 'CURRENTWEEK', 'TRENUTNANEDELjA' ),
182	182	'currentdow' => array( 1, 'CURRENTDOW', 'TRENUTNIDOV' ),
Index: trunk/phase3/languages/Language.php
—	—	@@ -49,9 +49,10 @@
50	50	function findVariantLink(&$l, &$n) {}
51	51	function getExtraHashOptions() {return '';}
52	52	function getParsedTitle() {return '';}
53		~~- function markNoConversion($text) {return $text;}~~
	53	+ function markNoConversion($text, $noParse=false) {return $text;}
54	54	function convertCategoryKey( $key ) {return $key; }
55		-
	55	+ function convertLinkToAllVariants($text){ return array( $this->mLang->getCode() => $text); }
	56	+ function setNoTitleConvert(){}
56	57	}
57	58
58	59	#--------------------------------------------------------------------------
—	—	@@ -712,6 +713,34 @@
713	714	return iconv( $in, $out, $string );
714	715	}
715	716
	717	+ // callback functions for uc(), lc(), ucwords(), ucwordbreaks()
	718	+ function ucwordbreaksCallbackAscii($matches){
	719	+ return $this->ucfirst($matches[1]);
	720	+ }
	721	+
	722	+ function ucwordbreaksCallbackMB($matches){
	723	+ return mb_strtoupper($matches[0]);
	724	+ }
	725	+
	726	+ function ucCallback($matches){
	727	+ global $wikiUpperChars;
	728	+ return strtr( $matches[1] , $wikiUpperChars );
	729	+ }
	730	+
	731	+ function lcCallback($matches){
	732	+ global $wikiLowerChars;
	733	+ return strtr( $matches[1] , $wikiLowerChars );
	734	+ }
	735	+
	736	+ function ucwordsCallbackMB($matches){
	737	+ return mb_strtoupper($matches[0]);
	738	+ }
	739	+
	740	+ function ucwordsCallbackWiki($matches){
	741	+ global $wikiUpperChars;
	742	+ return strtr( $matches[0] , $wikiUpperChars );
	743	+ }
	744	+
716	745	function ucfirst( $str ) {
717	746	return self::uc( $str, true );
718	747	}
—	—	@@ -729,9 +758,9 @@
730	759	if ( self::isMultibyte( $str ) ) {
731	760	list( $wikiUpperChars ) = $this->getCaseMaps();
732	761	$x = $first ? '^' : '';
733		~~- return preg_replace(~~
734		~~- "/$x([a-z]\|[\\xc0-\\xff][\\x80-\\xbf]*)/e",~~
735		~~- "strtr( \"\$1\" , \$wikiUpperChars )",~~
	762	+ return preg_replace_callback(
	763	+ "/$x([a-z]\|[\\xc0-\\xff][\\x80-\\xbf]*)/",
	764	+ array($this,"ucCallback"),
736	765	$str
737	766	);
738	767	} else
—	—	@@ -755,9 +784,9 @@
756	785	if ( self::isMultibyte( $str ) ) {
757	786	list( , $wikiLowerChars ) = self::getCaseMaps();
758	787	$x = $first ? '^' : '';
759		~~- return preg_replace(~~
760		~~- "/$x([A-Z]\|[\\xc0-\\xff][\\x80-\\xbf]*)/e",~~
761		~~- "strtr( \"\$1\" , \$wikiLowerChars )",~~
	788	+ return preg_replace_callback(
	789	+ "/$x([A-Z]\|[\\xc0-\\xff][\\x80-\\xbf]*)/",
	790	+ array($this,"lcCallback"),
762	791	$str
763	792	);
764	793	} else
—	—	@@ -768,6 +797,66 @@
769	798	return (bool)preg_match( '/[\x80-\xff]/', $str );
770	799	}
771	800
	801	+ function ucwords($str) {
	802	+ global $wikiUpperChars;
	803	+
	804	+ if ( self::isMultibyte( $str ) ) {
	805	+ $str = self::lc($str);
	806	+
	807	+ // regexp to find first letter in each word (i.e. after each space)
	808	+ $replaceRegexp = "/^([a-z]\|[\\xc0-\\xff][\\x80-\\xbf])\| ([a-z]\|[\\xc0-\\xff][\\x80-\\xbf])/";
	809	+
	810	+ // function to use to capitalize a single char
	811	+ if ( function_exists( 'mb_strtoupper' ) )
	812	+ return preg_replace_callback(
	813	+ $replaceRegexp,
	814	+ array($this,"ucwordsCallbackMB"),
	815	+ $str
	816	+ );
	817	+ else
	818	+ return preg_replace_callback(
	819	+ $replaceRegexp,
	820	+ array($this,"ucwordsCallbackWiki"),
	821	+ $str
	822	+ );
	823	+ }
	824	+ else
	825	+ return ucwords( strtolower( $str ) );
	826	+ }
	827	+
	828	+ # capitalize words at word breaks
	829	+ function ucwordbreaks($str){
	830	+ global $wikiUpperChars;
	831	+
	832	+ if (self::isMultibyte( $str ) ) {
	833	+ $str = self::lc($str);
	834	+
	835	+ // since \b doesn't work for UTF-8, we explicitely define word break chars
	836	+ $breaks= "[ \-\}\{\.,\?!]";
	837	+
	838	+ // find first letter after word break
	839	+ $replaceRegexp = "/^([a-z]\|[\\xc0-\\xff][\\x80-\\xbf])\|$breaks([a-z]\|[\\xc0-\\xff][\\x80-\\xbf])/";
	840	+
	841	+ if ( function_exists( 'mb_strtoupper' ) )
	842	+ return preg_replace_callback(
	843	+ $replaceRegexp,
	844	+ array($this,"ucwordbreaksCallbackMB"),
	845	+ $str
	846	+ );
	847	+ else
	848	+ return preg_replace_callback(
	849	+ $replaceRegexp,
	850	+ array($this,"ucwordsCallbackWiki"),
	851	+ $str
	852	+ );
	853	+ }
	854	+ else
	855	+ return preg_replace_callback(
	856	+ '/\b([\w\x80-\xff]+)\b/',
	857	+ array($this,"ucwordbreaksCallbackAscii"),
	858	+ $str );
	859	+ }
	860	+
772	861	function checkTitleEncoding( $s ) {
773	862	if( is_array( $s ) ) {
774	863	wfDebugDieBacktrace( 'Given array to checkTitleEncoding.' );
—	—	@@ -1169,6 +1258,17 @@
1170	1259	return $this->mConverter->parserConvert( $text, $parser );
1171	1260	}
1172	1261
	1262	+ # Tell the converter that it shouldn't convert titles
	1263	+ function setNoTitleConvert(){
	1264	+ $this->mConverter->setNotitleConvert();
	1265	+ }
	1266	+
	1267	+ # Check if this is a language with variants
	1268	+ function hasVariants(){
	1269	+ return sizeof($this->getVariants())>1;
	1270	+ }
	1271	+
	1272	+
1173	1273	/**
1174	1274	* Perform output conversion on a string, and encode for safe HTML output.
1175	1275	* @param string $text
—	—	@@ -1214,6 +1314,17 @@
1215	1315	}
1216	1316
1217	1317	/**
	1318	+ * If a language supports multiple variants, converts text
	1319	+ * into an array of all possible variants of the text:
	1320	+ * 'variant' => text in that variant
	1321	+ */
	1322	+
	1323	+ function convertLinkToAllVariants($text){
	1324	+ return $this->mConverter->convertLinkToAllVariants($text);
	1325	+ }
	1326	+
	1327	+
	1328	+ /**
1218	1329	* returns language specific options used by User::getPageRenderHash()
1219	1330	* for example, the preferred language variant
1220	1331	*
—	—	@@ -1242,8 +1353,8 @@
1243	1354	* @param string $text text to be tagged for no conversion
1244	1355	* @return string the tagged text
1245	1356	*/
1246		~~- function markNoConversion( $text ) {~~
1247		~~- return $this->mConverter->markNoConversion( $text );~~
	1357	+ function markNoConversion( $text, $noParse=false ) {
	1358	+ return $this->mConverter->markNoConversion( $text, $noParse );
1248	1359	}
1249	1360
1250	1361	/**
Index: trunk/phase3/languages/MessagesSr_ec.php
—	—	@@ -173,7 +173,7 @@
174	174	'servername' => array( 0, 'SERVERNAME', 'ИМЕСЕРВЕРА' ),
175	175	'scriptpath' => array( 0, 'SCRIPTPATH', 'СКРИПТА' ),
176	176	'grammar' => array( 0, 'GRAMMAR:', 'ГРАМАТИКА:' ),
177		~~- 'notitleconvert' => array( 0, '__NOTITLECONVERT__', '__NOTC__', '__БЕЗТЦ__' ),~~
	177	+ 'notitleconvert' => array( 0, '__NOTITLECONVERT__', '__NOTC__', '__БЕЗКН__', '__BEZKN__' ),
178	178	'nocontentconvert' => array( 0, '__NOCONTENTCONVERT__', '__NOCC__', '__БЕЗЦЦ__' ),
179	179	'currentweek' => array( 1, 'CURRENTWEEK', 'ТРЕНУТНАНЕДЕЉА' ),
180	180	'currentdow' => array( 1, 'CURRENTDOW', 'ТРЕНУТНИДОВ' ),
Index: trunk/phase3/languages/LanguageConverter.php
—	—	@@ -22,6 +22,7 @@
23	23	var $mMarkup;
24	24	var $mFlags;
25	25	var $mUcfirst = false;
	26	+ var $mNoTitleConvert = false;
26	27	/**
27	28	* Constructor
28	29	*
—	—	@@ -38,6 +39,7 @@
39	40	$markup=array(),
40	41	$flags = array()) {
41	42	global $wgDBname;
	43	+ global $wgLegalTitleChars;
42	44	$this->mLangObj = $langobj;
43	45	$this->mMainLanguageCode = $maincode;
44	46	$this->mVariants = $variants;
—	—	@@ -155,14 +157,17 @@
156	158	$marker = "";
157	159
158	160	// this one is needed when the text is inside an html markup
159		~~- $htmlfix = '\|<[^>]+=\"[^(>=)]$\|^[^(<>=\")]\"[^>]*>';~~
	161	+ $htmlfix = '\|<[^>]+$\|^[^<>]*>';
160	162
161		~~- $reg = '/<[^>]+>\|&[a-z#][a-z0-9]+;' . $marker . $htmlfix . '/';~~
	163	+ // disable convert to variants between <code></code> tags
	164	+ $codefix = '<code>.+?<\/code>\|';
	165	+
	166	+ $reg = '/'.$codefix.'<[^>]+>\|&[a-z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
162	167
163	168	$matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
164	169
	170	+ $m = array_shift($matches);
165	171
166		~~- $m = array_shift($matches);~~
167	172	$ret = $this->translate($m[0], $toVariant);
168	173	$mstart = $m[1]+strlen($m[0]);
169	174	foreach($matches as $m) {
—	—	@@ -197,7 +202,7 @@
198	203	*
199	204	* @param string $text the text to be converted
200	205	* @return array of string
201		~~- * @private~~
	206	+ * @public
202	207	*/
203	208	function autoConvertToAllVariants($text) {
204	209	$fname="LanguageConverter::autoConvertToAllVariants";
—	—	@@ -209,18 +214,52 @@
210	215	foreach($this->mVariants as $variant) {
211	216	$ret[$variant] = $this->translate($text, $variant);
212	217	}
	218	+
213	219	wfProfileOut( $fname );
214	220	return $ret;
215	221	}
216	222
217	223	/**
	224	+ * convert link text to all supported variants
	225	+ *
	226	+ * @param string $text the text to be converted
	227	+ * @return array of string
	228	+ * @public
	229	+ */
	230	+ function convertLinkToAllVariants($text) {
	231	+ if( !$this->mTablesLoaded )
	232	+ $this->loadTables();
	233	+
	234	+ $ret = array();
	235	+ $tarray = explode($this->mMarkup['begin'], $text);
	236	+ $tfirst = array_shift($tarray);
	237	+
	238	+ foreach($this->mVariants as $variant)
	239	+ $ret[$variant] = $this->translate($tfirst,$variant);
	240	+
	241	+ foreach($tarray as $txt) {
	242	+ $marked = explode($this->mMarkup['end'], $txt, 2);
	243	+
	244	+ foreach($this->mVariants as $variant){
	245	+ $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
	246	+ if(array_key_exists(1, $marked))
	247	+ $ret[$variant] .= $this->translate($marked[1],$variant);
	248	+ }
	249	+
	250	+ }
	251	+
	252	+ return $ret;
	253	+ }
	254	+
	255	+
	256	+ /**
218	257	* Convert text using a parser object for context
219	258	*/
220	259	function parserConvert( $text, &$parser ) {
221	260	global $wgDisableLangConversion;
222	261	/* don't do anything if this is the conversion table */
223	262	if ( $parser->mTitle->getNamespace() == NS_MEDIAWIKI &&
224		~~- strpos($parser->mTitle->getText, "Conversiontable") !== false )~~
	263	+ strpos($parser->mTitle->getText(), "Conversiontable") !== false )
225	264	{
226	265	return $text;
227	266	}
—	—	@@ -264,6 +303,11 @@
265	304	return $text;
266	305
267	306	if( $isTitle ) {
	307	+ if($this->mNoTitleConvert){
	308	+ $this->mTitleDisplay = $text;
	309	+ return $text;
	310	+ }
	311	+
268	312	if( !$this->mDoTitleConvert ) {
269	313	$this->mTitleDisplay = $text;
270	314	return $text;
—	—	@@ -278,7 +322,7 @@
279	323	return $text;
280	324	}
281	325	else {
282		~~- $this->mTitleDisplay = $this->autoConvert($text);~~
	326	+ $this->mTitleDisplay = $this->convert($text);
283	327	return $this->mTitleDisplay;
284	328	}
285	329	}
—	—	@@ -315,7 +359,7 @@
316	360	else
317	361	$rules = $marked[0];
318	362
319		~~-#FIXME: may cause trouble here...~~
	363	+ //FIXME: may cause trouble here...
320	364	//strip   since it interferes with the parsing, plus,
321	365	//all spaces should be stripped in this tag anyway.
322	366	$rules = str_replace(' ', '', $rules);
—	—	@@ -410,23 +454,16 @@
411	455	* @access public
412	456	*/
413	457	function findVariantLink( &$link, &$nt ) {
414		~~- static $count=0; //used to limit this operation~~
415		~~- static $cache=array();~~
416	458	global $wgDisableLangConversion;
417	459	$pref = $this->getPreferredVariant();
418	460	$ns=0;
419	461	if(is_object($nt))
420	462	$ns = $nt->getNamespace();
421		~~- if( $count > 50 && $ns != NS_CATEGORY )~~
422		~~- return;~~
423		~~- $count++;~~
	463	+
424	464	$variants = $this->autoConvertToAllVariants($link);
425	465	if($variants == false) //give up
426	466	return;
427	467	foreach( $variants as $v ) {
428		~~- if(isset($cache[$v]))~~
429		~~- continue;~~
430		~~- $cache[$v] = 1;~~
431	468	$varnt = Title::newFromText( $v, $ns );
432	469	if( $varnt && $varnt->getArticleID() > 0 ) {
433	470	$nt = $varnt;
—	—	@@ -655,7 +692,7 @@
656	693	* @param string $text text to be tagged for no conversion
657	694	* @return string the tagged text
658	695	*/
659		~~- function markNoConversion($text) {~~
	696	+ function markNoConversion($text, $noParse=false) {
660	697	# don't mark if already marked
661	698	if(strpos($text, $this->mMarkup['begin']) \|\|
662	699	strpos($text, $this->mMarkup['end']))
—	—	@@ -696,6 +733,11 @@
697	734	}
698	735	return true;
699	736	}
	737	+
	738	+ function setNoTitleConvert(){
	739	+ $this->mNoTitleConvert = true;
	740	+ }
	741	+
700	742	}
701	743
702	744	?>
Index: trunk/phase3/languages/LanguageSr.php
—	—	@@ -169,8 +169,8 @@
170	170	* We want our external link captions to be converted in variants,
171	171	* so we return the original text instead -{$text}-, except for URLs
172	172	*/
173		~~- function markNoConversion($text) {~~
174		~~- if(preg_match("/^https?:\/\/\|ftp:\/\/\|irc:\/\//",$text))~~
	173	+ function markNoConversion($text, $noParse=false) {
	174	+ if($noParse \|\| preg_match("/^https?:\/\/\|ftp:\/\/\|irc:\/\//",$text))
175	175	return parent::markNoConversion($text);
176	176	return $text;
177	177	}
—	—	@@ -188,22 +188,51 @@
189	189	return parent::autoConvert($text,$toVariant);
190	190	}
191	191
	192	+ /**
	193	+ * It translates text into variant, specials:
	194	+ * - ommiting roman numbers
	195	+ */
	196	+ function translate($text, $toVariant){
	197	+ $breaks = '[^\w\x80-\xff]';
192	198
	199	+ // regexp for roman numbers
	200	+ $roman = 'M{0,4}(CM\|CD\|D?C{0,3})(XC\|XL\|L?X{0,3})(IX\|IV\|V?I{0,3})';
	201	+
	202	+ $reg = '/^'.$roman.'$\|^'.$roman.$breaks.'\|'.$breaks.$roman.'$\|'.$breaks.$roman.$breaks.'/';
	203	+
	204	+ $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
	205	+
	206	+ $m = array_shift($matches);
	207	+ $ret = strtr($m[0], $this->mTables[$toVariant]);
	208	+ $mstart = $m[1]+strlen($m[0]);
	209	+ foreach($matches as $m) {
	210	+ $ret .= substr($text, $mstart, $m[1]-$mstart);
	211	+ $ret .= parent::translate($m[0], $toVariant);
	212	+ $mstart = $m[1] + strlen($m[0]);
	213	+ }
	214	+
	215	+ return $ret;
	216	+ }
	217	+
	218	+
193	219	}
194	220
195	221	class LanguageSr extends LanguageSr_ec {
196	222	function __construct() {
197	223	global $wgHooks;
	224	+
198	225	parent::__construct();
199	226
200		~~- $variants = array('sr', 'sr-ec', 'sr-jc', 'sr-el', 'sr-jl');~~
	227	+ // these variants are currently UNUSED:
	228	+ // 'sr-jc', 'sr-jl'
	229	+ $variants = array('sr', 'sr-ec', 'sr-el');
201	230	$variantfallbacks = array(
202	231	'sr' => 'sr-ec',
203		~~- 'sr-ec' => 'sr-jc',~~
204		~~- 'sr-jc' => 'sr-ec',~~
205		~~- 'sr-el' => 'sr-jl',~~
206		~~- 'sr-jl' => 'sr-el'~~
207		~~- );~~
	232	+ 'sr-ec' => 'sr-ec',
	233	+ 'sr-el' => 'sr-el',
	234	+ );
	235	+
	236	+
208	237	$marker = array();//don't mess with these, leave them as they are
209	238	$flags = array(
210	239	'S' => 'S', 'писмо' => 'S', 'pismo' => 'S',