Merge pull request #32209 from vespa-engine/bratseth/more-languages

Handle all languages already supported by OpenNlp
vespa-engine · Aug 21, 2024 · c48ca65 · c48ca65
2 parents b666cc6 + 00bf9af
commit c48ca65
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/opennlp-linguistics/src/main/java/com/yahoo/language/opennlp/OpenNlpTokenizer.java b/opennlp-linguistics/src/main/java/com/yahoo/language/opennlp/OpenNlpTokenizer.java
@@ -143,12 +143,17 @@ private Stemmer stemmerFor(Language language, StemMode stemMode) {
 
     private SnowballStemmer.ALGORITHM algorithmFor(Language language) {
         return switch (language) {
+            case ARABIC -> SnowballStemmer.ALGORITHM.ARABIC;
+            case CATALAN -> SnowballStemmer.ALGORITHM.CATALAN;
             case DANISH -> SnowballStemmer.ALGORITHM.DANISH;
             case DUTCH -> SnowballStemmer.ALGORITHM.DUTCH;
+            case ENGLISH -> SnowballStemmer.ALGORITHM.ENGLISH;
             case FINNISH -> SnowballStemmer.ALGORITHM.FINNISH;
             case FRENCH -> SnowballStemmer.ALGORITHM.FRENCH;
             case GERMAN -> SnowballStemmer.ALGORITHM.GERMAN;
+            case GREEK -> SnowballStemmer.ALGORITHM.GREEK;
             case HUNGARIAN -> SnowballStemmer.ALGORITHM.HUNGARIAN;
+            case INDONESIAN -> SnowballStemmer.ALGORITHM.INDONESIAN;
             case IRISH -> SnowballStemmer.ALGORITHM.IRISH;
             case ITALIAN -> SnowballStemmer.ALGORITHM.ITALIAN;
             case NORWEGIAN_BOKMAL -> SnowballStemmer.ALGORITHM.NORWEGIAN;
@@ -159,7 +164,6 @@ private SnowballStemmer.ALGORITHM algorithmFor(Language language) {
             case SPANISH -> SnowballStemmer.ALGORITHM.SPANISH;
             case SWEDISH -> SnowballStemmer.ALGORITHM.SWEDISH;
             case TURKISH -> SnowballStemmer.ALGORITHM.TURKISH;
-            case ENGLISH -> SnowballStemmer.ALGORITHM.ENGLISH;
             default -> null;
         };
     }