Revision: 8252
http://languagetool.svn.sourceforge.net/languagetool/?rev=8252&view=rev
Author: jaumeortola
Date: 2012-10-22 13:30:15 +0000 (Mon, 22 Oct 2012)
Log Message:
-----------
[ca] Fix some false alarms. Disable more rules for Wikipedia check.
Modified Paths:
--------------
trunk/JLanguageTool/src/main/java/org/languagetool/rules/ca/AccentuationCheckRule.java
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ltcommunity/disabled_rules.properties
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
trunk/JLanguageTool/src/test/java/org/languagetool/rules/ca/AccentuationCheckRuleTest.java
Modified:
trunk/JLanguageTool/src/main/java/org/languagetool/rules/ca/AccentuationCheckRule.java
===================================================================
---
trunk/JLanguageTool/src/main/java/org/languagetool/rules/ca/AccentuationCheckRule.java
2012-10-21 21:04:49 UTC (rev 8251)
+++
trunk/JLanguageTool/src/main/java/org/languagetool/rules/ca/AccentuationCheckRule.java
2012-10-22 13:30:15 UTC (rev 8252)
@@ -78,6 +78,7 @@
private static final Pattern BEFORE_ADJECTIVE_FS =
Pattern.compile("SPS00|D[^R].[FC][SN].*|V.[^NGP].*|PX.*");
private static final Pattern BEFORE_ADJECTIVE_MP =
Pattern.compile("SPS00|D[^R].[MC][PN].*|V.[^NGP].*|PX.*");
private static final Pattern BEFORE_ADJECTIVE_FP =
Pattern.compile("SPS00|D[^R].[FC][PN].*|V.[^NGP].*|PX.*");
+ private static final Pattern GN = Pattern.compile("_GN_.*");
private final Map<String, AnalyzedTokenReadings> relevantWords;
private final Map<String, AnalyzedTokenReadings> relevantWords2;
@@ -104,7 +105,7 @@
public RuleMatch[] match(final AnalyzedSentence text) {
final List<RuleMatch> ruleMatches = new ArrayList<RuleMatch>();
final AnalyzedTokenReadings[] tokens = text.getTokensWithoutWhitespace();
- for (int i = 1; i < tokens.length; i++) { //ignoring token 0, i.e.,
SENT_START
+ for (int i = 1; i < tokens.length; i++) { //ignoring token 0, i.e.,
SENT_START
final String token;
if (i == 1) {
token=tokens[i].getToken().toLowerCase();
@@ -144,7 +145,7 @@
final Matcher mArticleELFP = ARTICLE_EL_FP.matcher(prevToken);
// verb without accent -> noun with accent
- if (isRelevantWord)
+ if (isRelevantWord && !matchPostagRegexp(tokens[i],GN))
{
//amb renuncies
if (tokens[i-1].hasPosTag("SPS00") &&
!matchPostagRegexp(tokens[i-1],DETERMINANT) &&
!matchPostagRegexp(tokens[i],INFINITIU) )
@@ -236,7 +237,7 @@
}
// verb without accent -> adjective with accent
- if (isRelevantWord2)
+ if (isRelevantWord2 && !matchPostagRegexp(tokens[i],GN))
{
// de manera obvia, circumstàncies extraordinaries.
if ( (matchPostagRegexp(relevantWords2.get(token),ADJECTIU_MS) &&
matchPostagRegexp(tokens[i-1],NOM_MS) && !tokens[i-1].hasPosTag("_GN_FS") &&
matchPostagRegexp(tokens[i],VERB_CONJUGAT) )
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
2012-10-21 21:04:49 UTC (rev 8251)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
2012-10-22 13:30:15 UTC (rev 8252)
@@ -1018,7 +1018,7 @@
</and>
</marker>
</pattern>
- <disambig action="filter" postag="N.*|A.*|V.P.*" /> <!-- les ha
vistes estenent la roba -->
+ <disambig action="filter" postag="N.*|A.*|V.P.*|PX.*" /> <!-- les
ha vistes estenent la roba -->
</rule>
</rulegroup>
<rulegroup id="noms_aposicio" name="noms en aposició">
@@ -3982,7 +3982,7 @@
</and>
</marker>
</unify>
- <token postag="N.*" postag_regexp="yes"><exception
postag="V.[MSI].*|AQ.*|V.P.*" postag_regexp="yes"/><exception
regexp="yes">\p{Lu}.+</exception></token>
+ <token postag="N.*" postag_regexp="yes"><exception
postag="V.[MSI].*|AQ.*|V.P.*|PX.*" postag_regexp="yes"/><exception
regexp="yes">\p{Lu}.+</exception></token>
</pattern>
<disambig action="filter" postag="AQ.*"></disambig>
</rule>
@@ -5041,7 +5041,7 @@
<token postag="A.*" postag_regexp="yes"><exception
postag="_GN_.*" postag_regexp="yes"/></token>
</and>
</marker>
- <token postag="N.*" postag_regexp="yes"></token>
+ <token postag="N.*" postag_regexp="yes"><exception
postag="V.[SI].*" postag_regexp="yes"/></token>
</pattern>
<disambig action="filter" postag="V.*"></disambig>
</rule>
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ltcommunity/disabled_rules.properties
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ltcommunity/disabled_rules.properties
2012-10-21 21:04:49 UTC (rev 8251)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/resource/ltcommunity/disabled_rules.properties
2012-10-22 13:30:15 UTC (rev 8252)
@@ -9,4 +9,4 @@
sl=POMIŠLJAJ_1,POMIŠLJAJ_2,POMIŠLJAJ_3
en=EN_QUOTES,EN_UNPAIRED_BRACKETS
ru=RU_UNPAIRED_BRACKETS
-ca=GUIONET_SOLT,EVITA_ACCENTUACIO_VALENCIANA,EVITA_VERBS_VALENCIANS,EVITA_VERBS_BALEARS,EVITA_VERBS_BALEARS_VALENCIANS,EVITA_POSSESSIUS_U,EVITA_PRONOMS_VALENCIANS,DEMOSTRATIUS_ESTE_EIXE,PUNTS_SUSPENSIUS
+ca=GUIONET_SOLT,EVITA_ACCENTUACIO_VALENCIANA,EVITA_VERBS_VALENCIANS,EVITA_VERBS_BALEARS,EVITA_VERBS_BALEARS_VALENCIANS,EVITA_POSSESSIUS_U,EVITA_PRONOMS_VALENCIANS,DEMOSTRATIUS_ESTE,DEMOSTRATIUS_EIXE,PUNTS_SUSPENSIUS
Modified:
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
===================================================================
---
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
2012-10-21 21:04:49 UTC (rev 8251)
+++
trunk/JLanguageTool/src/main/resources/org/languagetool/rules/ca/grammar.xml
2012-10-22 13:30:15 UTC (rev 8252)
@@ -1342,12 +1342,13 @@
<marker>
<token postag="V.P.*" postag_regexp="yes"
inflected="yes">adormir</token>
</marker>
- <token><exception postag="_GN_.*" postag_regexp="yes"
/></token>
+ <token><exception postag="_GN_.*|NP.*" postag_regexp="yes"
/></token>
</pattern>
<message>¿Volíeu dir <suggestion><match no="3"
regexp_match="a(dorm.*)" regexp_replace="$1"/></suggestion>?</message>
<example type="incorrect">Joan ha
<marker>adormit</marker>.</example>
<example type="correct">Joan ha adormit la nena.</example>
<example type="correct">Joan l'ha adormit.</example>
+ <example type="correct">Jahvè va adormir Adam</example>
</rule>
<rule>
<pattern>
@@ -1593,7 +1594,7 @@
<rule id="CALDRE" name="Comprova: «caldre» seguit de verb en
infinitiu">
<pattern>
<marker>
- <token inflected="yes">caldre</token>
+ <token inflected="yes">caldre<exception
postag="_GN_MS"/></token>
<token postag="VM[^N].*" postag_regexp="yes"
regexp="yes">[\p{Ll}·]+<exception postag="[^V].*" postag_regexp="yes"/></token>
</marker>
</pattern>
@@ -1601,6 +1602,7 @@
<example type="incorrect">No <marker>cal
corre</marker>.</example>
<example type="correct">No cal córrer.</example>
<example type="correct">Anem a cal Cego.</example>
+ <example type="correct">El model estàndard del Big Bang calent
erra en l'objectiu</example>
</rule>
<rule id="CALDRE_DE" name="Caldre (*de) + infinitiu">
<pattern>
@@ -2083,8 +2085,8 @@
</rule>
<rule id="CONFUSIO_PARTICIPI_INFINITIU" name="Nom + adverbi +
(*infinitiu/*gerundi/participi)">
<pattern>
- <token postag="N.[MC][SN].*" postag_regexp="yes"><exception
postag="V.[SI].*" postag_regexp="yes"/></token>
- <token postag="RG"
regexp="yes">.+ment|molt|poc|tant|tan|aleshores|alhora|almanco|almenys|aviat|ben|encara|força|gaire|gairebé|gens|inclús|ja|llavors|mai|massa|menys|mig|millor|més|no|només|pas|pitjor|potser|prompte|prou|quasi|quelcom|sempre|sobretot|sols|suara|també|tampoc|tanmateix|tostemps|tot|tothora|àdhuc</token>
+ <token postag="N.[MC][SN].*" postag_regexp="yes"><exception
postag="V.[SI].*" postag_regexp="yes"/><exception
inflected="yes">numeral</exception></token>
+ <token postag="RG_anteposat"><exception
regexp="yes">tot|també</exception></token>
<marker>
<token postag="V.[GN].*" postag_regexp="yes"><exception
postag="A.*|V.P.*" postag_regexp="yes"/></token>
</marker>
@@ -2096,6 +2098,8 @@
<example type="correct">un país secularment humanitzat com el
nostre</example>
<example type="correct">i pot també produir</example>
<example type="correct">el mineral més dur conegut</example>
+ <example type="correct">enllaçat a tres més formant un
pla</example>
+ <example type="correct">a Colom tot indicant-li que
emetria</example>
</rule>
<rulegroup id="DONAR_FER" name="donar/fer">
<rule>
@@ -2288,7 +2292,7 @@
<category name="Determinants">
<rulegroup id="DETERMINANTS_IMPOSSIBLE" name="Combinacions impossibles
de determinants: *els tots els">
<!-- falten combinacions de dos: els aquests -->
- <rule>
+ <!-- <rule>
<pattern>
<marker>
<token postag="D.*" postag_regexp="yes"></token>
@@ -2303,7 +2307,7 @@
<example type="correct">els tres mil</example>
<example type="correct">els trenta-quatre mil
restants</example>
<example type="correct">totes les quals sumes</example>
- </rule>
+ </rule> -->
<rule>
<pattern>
<token>els</token>
@@ -13266,7 +13270,7 @@
</rule>
</category>
<category name="Sintaxi">
- <rulegroup id="RELATIUS" name="Comprova les contruccions de relatiu:
amb el que/amb el qual" default="off">
+ <rulegroup id="RELATIUS" name="Comprova les contruccions de relatiu:
amb el que/amb el qual" default="on">
<rule>
<pattern>
<token postag="NCMS.*|_GN_MS" postag_regexp="yes"/>
Modified:
trunk/JLanguageTool/src/test/java/org/languagetool/rules/ca/AccentuationCheckRuleTest.java
===================================================================
---
trunk/JLanguageTool/src/test/java/org/languagetool/rules/ca/AccentuationCheckRuleTest.java
2012-10-21 21:04:49 UTC (rev 8251)
+++
trunk/JLanguageTool/src/test/java/org/languagetool/rules/ca/AccentuationCheckRuleTest.java
2012-10-22 13:30:15 UTC (rev 8252)
@@ -43,6 +43,7 @@
public void testRule() throws IOException {
// correct sentences:
+ assertCorrect("els següents territoris externs habitats:");
assertCorrect("Cap faria una cosa així.");
assertCorrect("El cos genera suficient pressió interna.");
assertCorrect("Les seues contràries.");
@@ -82,7 +83,7 @@
assertIncorrect("De manera obvia.");
assertIncorrect("Ell fa tasques especifiques.");
assertIncorrect("Un home adulter.");
- assertIncorrect("Va deixar els nens atonits.");
+ // assertIncorrect("Va deixar els nens atonits."); del v. "atonir"
(=esbalair)
assertIncorrect("La sureda ocupa amplies extensions en la muntanya.");
assertIncorrect("Féu una magnifica digitació.");
assertIncorrect("Els habitats de la comarca.");
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct
_______________________________________________
Languagetool-commits mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/languagetool-commits
