[ca] Dealing with deprecated characters for ela geminada

languagetool-org · Dec 24, 2013 · 44538ec · 44538ec
1 parent 36e47a7
commit 44538ec
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 0 deletions.
diff --git a/...uagetool-language-modules/ca/src/main/java/org/languagetool/tagging/ca/CatalanTagger.java b/...uagetool-language-modules/ca/src/main/java/org/languagetool/tagging/ca/CatalanTagger.java
@@ -134,6 +134,15 @@ public List<AnalyzedToken> additionalTags(String word) {
       }
       return additionalTaggedTokens;
     }
+    // Interpret deprecated characters of "ela geminada"
+    // U+013F LATIN CAPITAL LETTER L WITH MIDDLE DOT
+    // U+0140 LATIN SMALL LETTER L WITH MIDDLE DOT
+    if (word.contains("\u0140") || word.contains("\u013f")) {
+    	final String lowerWord = word.toLowerCase(conversionLocale);
+        final String possibleWord = lowerWord.replaceAll("\u0140", "l·");
+        List<AnalyzedToken> taggerTokens = asAnalyzedTokenList(word,dictLookup.lookup(possibleWord));
+        return taggerTokens;
+    }
     return null;
   }
 

diff --git a/languagetool-language-modules/ca/src/main/resources/org/languagetool/rules/ca/grammar.xml b/languagetool-language-modules/ca/src/main/resources/org/languagetool/rules/ca/grammar.xml
@@ -2579,6 +2579,7 @@ $Id$
             <example type="incorrect">Quan ho <marker>haguem</marker> fet.</example>
             <example type="correct">Quan ho hàgiu portat</example>
             <example type="correct">Si ho haguéssiu portat</example>
+            <example type="correct">Hi havien instaŀlat una antena.</example> <!-- Using deprecated char for ela geminada-->
         </rule>
         <rulegroup id="HAVER_SENSE_HAC" name="Haver sense hac: *eu/feu fet">    
             <rule>
@@ -12999,6 +13000,18 @@ $Id$
             <example type="correct">en un estat sindical-laborista</example>
             <example type="correct">en un estat sindical.La classe obrera</example>
         </rule>
+        <rule id="ELA_GEMINADA_2CHAR" name="Ela geminada: evita codificació de dos caràcters" type="misspelling">
+            <pattern case_sensitive="no">
+                <marker>
+                    <token regexp="yes">.*ŀl.*</token>
+                </marker>
+            </pattern>
+            <message>Error de codificació.</message>
+            <suggestion><match no="1" regexp_match="(.+)ŀl(.+)" regexp_replace="$1l·l$2"/></suggestion>
+            <short>Error tipogràfic</short>
+            <example type="incorrect" correction="sol·licitud">La <marker>soŀlicitud</marker>.</example>
+            <example type="correct">la sol·licitud</example>
+        </rule>
         <rule id="ESPAI_DESPRES_DE_PUNT" name="Comprova que hi ha espai deprés de punt.">
             <pattern>
                 <token><exception>www</exception></token>

diff --git a/...-modules/ca/src/test/java/org/languagetool/rules/ca/MorfologikCatalanSpellerRuleTest.java b/...-modules/ca/src/test/java/org/languagetool/rules/ca/MorfologikCatalanSpellerRuleTest.java
@@ -205,9 +205,14 @@ public void testMorfologikSpeller() throws IOException {
         assertEquals(1, matches.length);
         assertEquals("para", matches[0].getSuggestedReplacements().get(0));
         assertEquals("pare", matches[0].getSuggestedReplacements().get(1));
+
+        // deprecated characters of "ela geminada"
+        assertEquals(0, rule.match(langTool.getAnalyzedSentence("S'hi havien instaŀlat.")).length);
+        assertEquals(0, rule.match(langTool.getAnalyzedSentence("S'HI HAVIEN INSTAĿLAT.")).length);
 
         assertEquals(1, rule.match(langTool.getAnalyzedSentence("aõh")).length);
         assertEquals(0, rule.match(langTool.getAnalyzedSentence("a")).length);
+
     }
 
 }