Revision: 10222
http://sourceforge.net/p/languagetool/code/10222
Author: gulp21-1
Date: 2013-06-11 14:30:07 +0000 (Tue, 11 Jun 2013)
Log Message:
-----------
[de] fix some false alarms (case and agreement rule)
Modified Paths:
--------------
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/AgreementRule.java
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/CaseRule.java
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/AgreementRuleTest.java
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/CaseRuleTest.java
Modified:
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/AgreementRule.java
===================================================================
---
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/AgreementRule.java
2013-06-10 22:49:31 UTC (rev 10221)
+++
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/AgreementRule.java
2013-06-11 14:30:07 UTC (rev 10222)
@@ -82,8 +82,18 @@
"ADJ:DAT:PLU:MAS:GRU", "ADJ:DAT:PLU:NEU:GRU", "ADJ:DAT:PLU:FEM:GRU", //
den Berliner Autos
"ADJ:AKK:PLU:MAS:GRU", "ADJ:AKK:PLU:NEU:GRU", "ADJ:AKK:PLU:FEM:GRU", //
den Berliner Bewohnern
};
-
+ /*
+ * The heuristic of maybeAddAdjectiveReadings considers every noun ending
with "er" as city name.
+ * The nouns in this list are NOT considered as city names.
+ * NOTE: Only nouns for which cutting off the final "er" produces a valid
noun must be added to this list.
+ */
+ private static final Set<String> ER_TO_BE_IGNORED = new
HashSet<String>(Arrays.asList(
+ "Alter",
+ "Kinder",
+ "Rinder"
+ ));
+
private static final Set<String> REL_PRONOUN = new HashSet<String>();
static {
REL_PRONOUN.add("der");
@@ -183,15 +193,15 @@
boolean ignore = couldBeRelativeClause(tokens, i);
if (i > 0) {
final String prevToken = tokens[i-1].getToken().toLowerCase();
- if ((prevToken.equals("der") || prevToken.equals("die") ||
prevToken.equals("das"))
+ if ((prevToken.equals("der") || prevToken.equals("die") ||
prevToken.equals("das") || prevToken.equals("des"))
&& (tokens[i].getToken().equals("eine") ||
tokens[i].getToken().equals("einen"))) {
// TODO: "der eine Polizist" -> nicht ignorieren, sondern "der
polizist" checken; "auf der einen Seite"
ignore = true;
}
}
- // avoid false alarm on "nichts Gutes":
- if (analyzedToken.getToken().equals("nichts")) {
+ // avoid false alarm on "nichts Gutes" and "alles Gute"
+ if (analyzedToken.getToken().equals("nichts") ||
analyzedToken.getToken().equals("alles")) {
ignore = true;
}
@@ -224,7 +234,7 @@
}
}
- }
+ } // for each token
return toRuleMatchArray(ruleMatches);
}
@@ -259,7 +269,7 @@
final String nextTerm = nextToken.getToken();
// Just a heuristic: nouns and proper nouns that end with "er" are
considered
// city names:
- if (nextTerm.endsWith("er") && tokens.length > tokenPos+1) {
+ if (nextTerm.endsWith("er") && tokens.length > tokenPos+1 &&
!ER_TO_BE_IGNORED.contains(nextTerm)) {
final AnalyzedGermanTokenReadings nextNextToken =
(AnalyzedGermanTokenReadings)tokens[tokenPos+1];
try {
final AnalyzedGermanTokenReadings nextATR =
tagger.lookup(nextTerm.substring(0, nextTerm.length()-2));
Modified:
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/CaseRule.java
===================================================================
---
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/CaseRule.java
2013-06-10 22:49:31 UTC (rev 10221)
+++
trunk/languagetool/languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/CaseRule.java
2013-06-11 14:30:07 UTC (rev 10222)
@@ -71,6 +71,7 @@
sentenceStartExceptions.add("“");
sentenceStartExceptions.add("«");
sentenceStartExceptions.add("»");
+ sentenceStartExceptions.add(".");
}
private static final Set<String> exceptions = new HashSet<String>();
@@ -488,7 +489,7 @@
!analyzedToken.hasReadingOfType(POSType.PROPER_NOUN) &&
!isNilReading(analyzedToken) &&
!analyzedToken.isSentenceEnd() &&
- !( (tokens[i-1].getToken().equals("]") ||
tokens[i-1].getToken().equals(")")) &&
+ !( (tokens[i-1].getToken().equals("]") ||
tokens[i-1].getToken().equals(")")) && // sentence starts with […]
( (i == 4 && tokens[i-2].getToken().equals("…")) || (i == 6 &&
tokens[i-2].getToken().equals(".")) ) ) &&
!isExceptionPhrase(i, tokens)) {
final String msg = "Außer am Satzanfang werden nur Nomen und Eigennamen
großgeschrieben";
Modified:
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/AgreementRuleTest.java
===================================================================
---
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/AgreementRuleTest.java
2013-06-10 22:49:31 UTC (rev 10221)
+++
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/AgreementRuleTest.java
2013-06-11 14:30:07 UTC (rev 10222)
@@ -114,6 +114,9 @@
assertGood("Das Gütersloher Radio.");
assertGood("Das wirklich Wichtige kommt jetzt erst.");
assertGood("Besonders wenn wir Wermut oder Absinth trinken.");
+ assertGood("Ich wünsche dir alles Gute.");
+ assertGood("Es ist nicht bekannt, mit welchem Alter Kinder diese Fähigkeit
erlernen.");
+ assertGood("Dieser ist nun in den Ortungsbereich des einen Roboters
gefahren.");
// incorrect sentences:
assertBad("Es sind die Tisch.");
Modified:
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/CaseRuleTest.java
===================================================================
---
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/CaseRuleTest.java
2013-06-10 22:49:31 UTC (rev 10221)
+++
trunk/languagetool/languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/CaseRuleTest.java
2013-06-11 14:30:07 UTC (rev 10222)
@@ -79,6 +79,9 @@
assertEquals(0, rule.match(langTool.getAnalyzedSentence("Die Schlinge
zieht sich zu.")).length);
assertEquals(0, rule.match(langTool.getAnalyzedSentence("Die Schlingen
ziehen sich zu.")).length);
+ // used to trigger error because of "abbriviation"
+ assertEquals(0, rule.match(langTool.getAnalyzedSentence("Sie fällt auf
durch ihre hilfsbereite Art. Zudem zeigt sie soziale Kompetenz.")).length);
+
// TODO: nach dem Doppelpunkt wird derzeit nicht auf groß/klein getestet:
assertEquals(0, rule.match(langTool.getAnalyzedSentence("Das ist es: kein
Satz.")).length);
assertEquals(0, rule.match(langTool.getAnalyzedSentence("Das ist es: Kein
Satz.")).length);
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
Languagetool-commits mailing list
Languagetool-commits@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-commits
