[LanguageTool] SF.net SVN: languagetool:[7760] trunk/JLanguageTool/src

2012-07-27 Thread jaumeortola 
Revision: 7760
  
http://languagetool.svn.sourceforge.net/languagetool/?rev=7760view=rev
Author:   jaumeortola
Date: 2012-07-27 16:59:12 + (Fri, 27 Jul 2012)
Log Message:
---
[ca] Java rule improved: AccentuationCheckRule 

Modified Paths:
--

trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
trunk/JLanguageTool/src/resource/ca/catalan.dict
trunk/JLanguageTool/src/resource/ca/catalan_synth.dict
trunk/JLanguageTool/src/resource/ca/disambiguation.xml
trunk/JLanguageTool/src/rules/ca/verb_senseaccent_nom_ambaccent.txt

trunk/JLanguageTool/src/test/org/languagetool/rules/ca/AccentuationCheckRuleTest.java

Modified: 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
===
--- 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
   2012-07-27 11:43:33 UTC (rev 7759)
+++ 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
   2012-07-27 16:59:12 UTC (rev 7760)
@@ -58,7 +58,7 @@
   private static final Pattern ARTICLE_EL_FS = Pattern.compile(la|l'|La|L');
   private static final Pattern ARTICLE_EL_MP = Pattern.compile(els|Els);
   private static final Pattern ARTICLE_EL_FP = Pattern.compile(les|Les);
-//  private static final Pattern DETERMINANT = Pattern.compile(D[^R].*);
+  private static final Pattern DETERMINANT = Pattern.compile(D[^R].*);
   private static final Pattern DETERMINANT_MS = 
Pattern.compile(D[^R].[MC][SN].*);
   private static final Pattern DETERMINANT_FS = 
Pattern.compile(D[^R].[FC][SN].*);
   private static final Pattern DETERMINANT_MP = 
Pattern.compile(D[^R].[MC][PN].*);
@@ -72,7 +72,7 @@
   private static final Pattern ADJECTIU_MP = 
Pattern.compile(AQ.[MC][PN].*|V.P..PM|PX.MP.*);
   private static final Pattern ADJECTIU_FP = 
Pattern.compile(AQ.[FC][PN].*|V.P..PF|PX.FP.*);
   private static final Pattern INFINITIU = Pattern.compile(V.N.*);
-  private static final Pattern VERB_CONJUGAT = Pattern.compile(V.[^NGP].*);
+  private static final Pattern VERB_CONJUGAT = 
Pattern.compile(V.[^NGP].*|_GV_);
   private static final Pattern NOT_IN_PREV_TOKEN = 
Pattern.compile(VA.*|PP.*|P0.*|VSP.*);
   private static final Pattern BEFORE_ADJECTIVE_MS = 
Pattern.compile(SPS00|D[^R].[MC][SN].*|V.[^NGP].*|PX.*);
   private static final Pattern BEFORE_ADJECTIVE_FS = 
Pattern.compile(SPS00|D[^R].[FC][SN].*|V.[^NGP].*|PX.*);
@@ -146,8 +146,8 @@
   // verb without accent - noun with accent   
   if (isRelevantWord)
   {
-   //amb renuncies
-if (tokens[i-1].hasPosTag(SPS00)  
!matchPostagRegexp(tokens[i],INFINITIU) )
+   //amb renuncies DETERMINANT
+if (tokens[i-1].hasPosTag(SPS00)  
!matchPostagRegexp(tokens[i-1],DETERMINANT)  
!matchPostagRegexp(tokens[i],INFINITIU) )
{
replacement = relevantWords.get(token).getToken();
}
@@ -157,13 +157,23 @@
!token.equals(cantar) )
||(matchPostagRegexp(tokens[i-1],DETERMINANT_MP)  
matchPostagRegexp(relevantWords.get(token),NOM_MP))
||(matchPostagRegexp(tokens[i-1],DETERMINANT_FS)  
matchPostagRegexp(relevantWords.get(token),NOM_FS)
-   !token.equals(venia)  !token.equals(tenia)  
!token.equals(continua)  !token.equals(genera) )
+   !token.equals(venia)  !token.equals(tenia)  
!token.equals(continua)  !token.equals(genera)  !token.equals(faria))
||(matchPostagRegexp(tokens[i-1],DETERMINANT_FP)  
matchPostagRegexp(relevantWords.get(token),NOM_FP)) ) )
{
replacement = relevantWords.get(token).getToken();
}
+   //fumaré una faria (correct: fària)
+   else if ( i2  matchPostagRegexp(tokens[i-2],VERB_CONJUGAT) 
+((matchPostagRegexp(tokens[i-1],DETERMINANT_MS)  
matchPostagRegexp(relevantWords.get(token),NOM_MS))
+   ||(matchPostagRegexp(tokens[i-1],DETERMINANT_MP)  
matchPostagRegexp(relevantWords.get(token),NOM_MP))
+   ||(matchPostagRegexp(tokens[i-1],DETERMINANT_FS)  
matchPostagRegexp(relevantWords.get(token),NOM_FS))
+   ||(matchPostagRegexp(tokens[i-1],DETERMINANT_FP)  
matchPostagRegexp(relevantWords.get(token),NOM_FP)) ) )
+   {
+   replacement = relevantWords.get(token).getToken();
+   }
//circumstancies d'un altre caire
else if  ( !token.equals(venia)  !token.equals(venies)  
!token.equals(tenia)  !token.equals(tenies)
+   !token.equals(faria)  !token.equals(faries)
!token.equals(continua)  !token.equals(continues) 
 !token.equals(cantar)
mPreposicioDE.matches()  
!matchPostagRegexp(tokens[i-1],NOT_IN_PREV_TOKEN)
(itokens.length-2)  
!matchPostagRegexp(tokens[i+2],INFINITIU)
@@ -173,6 +183,7 @@
}
//la renuncia del president.

[LanguageTool] SF.net SVN: languagetool:[7761] trunk/JLanguageTool/src/rules/zh/grammar. xml

2012-07-27 Thread taolin2011 
Revision: 7761
  
http://languagetool.svn.sourceforge.net/languagetool/?rev=7761view=rev
Author:   taolin2011
Date: 2012-07-27 17:09:32 + (Fri, 27 Jul 2012)
Log Message:
---
add some Chinese rules

Modified Paths:
--
trunk/JLanguageTool/src/rules/zh/grammar.xml

Modified: trunk/JLanguageTool/src/rules/zh/grammar.xml
===
--- trunk/JLanguageTool/src/rules/zh/grammar.xml2012-07-27 16:59:12 UTC 
(rev 7760)
+++ trunk/JLanguageTool/src/rules/zh/grammar.xml2012-07-27 17:09:32 UTC 
(rev 7761)
@@ -20,6 +20,1066 @@
!-- 拼写： 词语错误 --
!-- 
== --
category name=词语错误
+   rulegroup id=LIAN1_LIAN2_JIAN3 name=练/炼/拣
+   rule
+   pattern
+   marker
+   token regexp=yes炼|拣/token
+   /marker
+   token postag=q|n|v|nr|vg|ag 
regexp=yes postag_regexp=yes笔|兵|达|功|武|习
+   /token
+   /pattern
+   message
+   您的意思是
+   suggestion练/suggestion
+   \2？
+   /message
+   short词语错误/short
+   example type=incorrect correction=练
+   我出差以前需要
+   marker炼/marker
+   习一下意大利语.
+   /example
+   example type=correct她指导实习护士练习注射。/example
+   /rule
+   rule
+   pattern
+   token postag=v|a|ad regexp=yes 
postag_regexp=yes训|老|熟|干
+   /token
+   marker
+   token regexp=yes炼|拣/token
+   /marker
+   /pattern
+   message
+   您的意思是\1
+   suggestion练/suggestion
+   ？
+   /message
+   short词语错误/short
+   example type=incorrect correction=练
+   实际上，她对他的干
+   marker拣/marker
+   再一次觉得很值得佩服。
+   /example
+   example 
type=correct他是一个以干练与干劲闻名的军官。/example
+   /rule
+   rule
+   pattern
+   marker
+   token regexp=yes练|拣/token
+   /marker
+   token postag=ng|n regexp=yes 
postag_regexp=yes丹|钢|铁|焦|乳|油|狱
+   /token
+   /pattern
+   message
+   您的意思是
+   suggestion炼/suggestion
+   \2？
+   /message
+   short词语错误/short
+   example type=incorrect correction=炼
+   漂亮的女人是眼睛的天堂、灵魂的地狱、钱包的
+   marker练/marker
+   狱。
+   /example
+   example 
type=correct漂亮的女人是眼睛的天堂、灵魂的地狱、钱包的炼狱。/example
+   /rule
+   rule
+   pattern
+   token冶/token
+   marker
+   token regexp=yes练|拣/token
+   /marker
+   /pattern
+   message
+   您的意思是\1
+   suggestion炼/suggestion
+   ？
+   /message
+   short词语错误/short
+   example type=incorrect correction=炼
+   将溶液倒入容器中进行冶
+   marker练/marker
+   。
+   /example
+   example

[LanguageTool] SF.net SVN: languagetool:[7762] trunk/JLanguageTool/src

2012-07-27 Thread jaumeortola 
Revision: 7762
  
http://languagetool.svn.sourceforge.net/languagetool/?rev=7762view=rev
Author:   jaumeortola
Date: 2012-07-27 18:02:52 + (Fri, 27 Jul 2012)
Log Message:
---
[ca] Improved rule: DALT.

Modified Paths:
--

trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
trunk/JLanguageTool/src/rules/ca/grammar.xml

Modified: 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
===
--- 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
   2012-07-27 17:09:32 UTC (rev 7761)
+++ 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/AccentuationCheckRule.java
   2012-07-27 18:02:52 UTC (rev 7762)
@@ -146,7 +146,7 @@
   // verb without accent - noun with accent   
   if (isRelevantWord)
   {
-   //amb renuncies DETERMINANT
+   //amb renuncies
 if (tokens[i-1].hasPosTag(SPS00)  
!matchPostagRegexp(tokens[i-1],DETERMINANT)  
!matchPostagRegexp(tokens[i],INFINITIU) )
{
replacement = relevantWords.get(token).getToken();

Modified: trunk/JLanguageTool/src/rules/ca/grammar.xml
===
--- trunk/JLanguageTool/src/rules/ca/grammar.xml2012-07-27 17:09:32 UTC 
(rev 7761)
+++ trunk/JLanguageTool/src/rules/ca/grammar.xml2012-07-27 18:02:52 UTC 
(rev 7762)
@@ -2278,19 +2278,47 @@
 example type=correctVa estar a punt de morir./example
 /rule
 /rulegroup
-rule id=DALT name=d'alt/dalt
-pattern
-token regexp=yesa|fins|de/token
-marker
-tokend'/token
-tokenalt/token
-/marker
-/pattern
-messagePotser és: suggestiondalt/suggestion./message
-shortConfusió./short
-example type=incorrectVan pujar fins 
markerd'alt/marker./example
-example type=correctVan pujar fins 
markerdalt/marker./example
-/rule
+rulegroup id=DALT name=*d'alt/dalt
+rule
+pattern
+token regexp=yesa|fins|de/token
+marker
+tokend'/token
+tokenalt/token
+/marker
+/pattern
+messagePotser és: suggestiondalt/suggestion./message
+shortConfusió./short
+example type=incorrectVan pujar fins 
markerd'alt/marker./example
+example type=correctVan pujar fins 
markerdalt/marker./example
+/rule
+rule
+pattern
+marker
+tokend'/token
+tokenalt/token
+/marker
+token postag=D[AID].* postag_regexp=yes/token
+/pattern
+messagePotser és: suggestiondalt/suggestion./message
+shortConfusió./short
+example type=incorrectVan pujar markerd'alt/marker la 
terrassa./example
+example type=correctVan pujar dalt la terrassa./example
+/rule
+rule
+pattern
+marker
+tokend'/token
+tokenalt/token
+/marker
+token inflected=yesde/token
+/pattern
+messagePotser és: suggestiondalt/suggestion./message
+shortConfusió./short
+example type=incorrectVan pujar markerd'alt/marker de 
tot./example
+example type=correctVan pujar dalt de tot./example
+/rule
+/rulegroup
 rulegroup id=DARRERA name=darrera/darrere
 rule
 pattern

This was sent by the SourceForge.net collaborative development platform, the 
world's largest Open Source development site.


--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
___
Languagetool-cvs mailing list
Languagetool-cvs@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-cvs

[LanguageTool] SF.net SVN: languagetool:[7765] trunk/JLanguageTool/src

2012-07-27 Thread jaumeortola 
Revision: 7765
  
http://languagetool.svn.sourceforge.net/languagetool/?rev=7765view=rev
Author:   jaumeortola
Date: 2012-07-27 23:45:40 + (Fri, 27 Jul 2012)
Log Message:
---
[ca] Fixed some false alarms.

Modified Paths:
--

trunk/JLanguageTool/src/java/org/languagetool/rules/ca/ComplexAdjectiveConcordanceRule.java
trunk/JLanguageTool/src/resource/ca/disambiguation.xml
trunk/JLanguageTool/src/rules/ca/grammar.xml

Modified: 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/ComplexAdjectiveConcordanceRule.java
===
--- 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/ComplexAdjectiveConcordanceRule.java
 2012-07-27 22:18:47 UTC (rev 7764)
+++ 
trunk/JLanguageTool/src/java/org/languagetool/rules/ca/ComplexAdjectiveConcordanceRule.java
 2012-07-27 23:45:40 UTC (rev 7765)
@@ -76,7 +76,7 @@
private static final Pattern COORDINACIO = Pattern.compile(,|i|o);
private static final Pattern COORDINACIO_IONI = 
Pattern.compile(i|o|ni);
private static final Pattern KEEP_COUNT = 
Pattern.compile(A.*|N.*|D[NAID].*|SPS.*|R.*|V.P.*);
-   private static final Pattern KEEP_COUNT2 = Pattern.compile(,|i|o|ni);
+   private static final Pattern KEEP_COUNT2 = 
Pattern.compile(,|i|o|ni|\\d+);
private static final Pattern PREPOSICIONS = Pattern.compile(SPS.*);
private static final Pattern VERB_AUXILIAR = Pattern.compile(V[AS].*);
private static final Pattern EXCEPCIONS_PARTICIPI = 
Pattern.compile(atès|atés|atesa|atesos|ateses|donat|donats|donada|donades);

Modified: trunk/JLanguageTool/src/resource/ca/disambiguation.xml
===
--- trunk/JLanguageTool/src/resource/ca/disambiguation.xml  2012-07-27 
22:18:47 UTC (rev 7764)
+++ trunk/JLanguageTool/src/resource/ca/disambiguation.xml  2012-07-27 
23:45:40 UTC (rev 7765)
@@ -4712,7 +4712,7 @@
 marker
 and
 token postag=A.* postag_regexp=yesexception 
postag=_GN_.* postag_regexp=yes /exception 
regexp=yes\p{Lu}+[\p{Ll}·]*/exceptionexception 
regexp=yes[\p{Ll}·]+(e|[aei]sses|[ae]res|era)/exception/token
-token postag=V.* postag_regexp=yesexception 
postag=_GV_ //token
+token postag=V.* postag_regexp=yesexception 
postag=_GV_|VMIP1S0 postag_regexp=yes //token
 /and
 /marker
 /pattern
@@ -5125,6 +5125,18 @@
 disambig action=filter postag=I /
 /rule
 /rulegroup
+rulegroup id=interjeccio name=interjeccions
+rule
+pattern
+token postag=SENT_START|CC postag_regexp=yes/token
+marker
+tokenvaja/token
+/marker
+token regexp=yes,|\.\.\.|\./token
+/pattern
+disambig action=filter postag=I /
+/rule
+/rulegroup
 rule id=PI_N name=pronom indefinit i nom
 pattern
 and

Modified: trunk/JLanguageTool/src/rules/ca/grammar.xml
===
--- trunk/JLanguageTool/src/rules/ca/grammar.xml2012-07-27 22:18:47 UTC 
(rev 7764)
+++ trunk/JLanguageTool/src/rules/ca/grammar.xml2012-07-27 23:45:40 UTC 
(rev 7765)
@@ -3051,10 +3051,8 @@
 /pattern
 messageCal dir suggestionel match 
no=2/match/suggestion./message
 shortError ortogràfic/short
-example type=incorrectmarkerl'iogurt/marker
-/example
-example type=correctmarkerel iogurt/marker
-/example
+example type=incorrectmarkerl'iogurt/marker/example
+example type=correctmarkerel iogurt/marker/example
 /rule
 rule
 pattern
@@ -3063,12 +3061,9 @@
 /pattern
 messageCal dir suggestionla match 
no=2/match/suggestion./message
 shortError ortogràfic/short
-example type=incorrectmarkerl'universitat/marker
-/example
-example type=correctmarkerla universitat/marker
-/example
+example 
type=incorrectmarkerl'universitat/marker/example
+example type=correctmarkerla 
universitat/marker/example
 example type=correctl'intel·lectual valencià/example
-example type=correctla intel·lectual valenciana/example
 /rule
 rule
 pattern
@@ -3079,12 +3074,14 @@
 token postag=_GN_MS /
 /and
 /marker
-token postag=_GN_FS /
+token postag=_GN_FSexception postag=_GN_MS//token
 /pattern
 messageCal dir suggestionla match