Bug-fixes to WikiTokenizer (handle weird line-feed), update to newest

[DictionaryPC.git] / src / com / hughes / android / dictionary / parser / DictFileParser.java

diff --git a/src/com/hughes/android/dictionary/parser/DictFileParser.java b/src/com/hughes/android/dictionary/parser/DictFileParser.java
index 991ed8ad417b29e4941f68fb07e4a1eaf4b4248b..edc0ce0cc54ba354f16eca3347c6f8ebc896d444 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/DictFileParser.java
+++ b/src/com/hughes/android/dictionary/parser/DictFileParser.java
@@ -55,6 +55,7 @@ public class DictFileParser implements Parser {
   static final Pattern PARENTHESIZED = Pattern.compile("\\(([^)]+)\\)");
   static final Pattern CURLY_BRACED = Pattern.compile("\\{([^}]+)\\}");
   
+  // http://www.regular-expressions.info/unicode.html
   static final Pattern NON_CHAR_DASH = Pattern.compile("[^-'\\p{L}\\p{M}\\p{N}]+");
   public static final Pattern NON_CHAR = Pattern.compile("[^\\p{L}\\p{M}\\p{N}]+");