refactored Eastern emoticon regex.

tested via static { System.out.println(eastEmote.equals(eastEmote2)); }
brendano · Oct 22, 2012 · f281da3 · f281da3
1 parent d0988f8
commit f281da3
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/src/cmu/arktweetnlp/Twokenize.java b/src/cmu/arktweetnlp/Twokenize.java
@@ -104,7 +104,11 @@ public static String OR(String... parts) {
     static String s4 = "(?:<|&lt;|>|&gt;)[\\._-]+(?:<|&lt;|>|&gt;)";
     static String basicface = "(?:(?i)" +s1+s2+ ")|" +s3+ "|" + s4;
 
-    static String eastEmote = "[＼\\\\ƪԄ\\(（<>;ヽ\\-=~\\*]+(?:"+basicface+"|[^A-Za-z0-9\\s\\(\\)\\*:=-])+[\\-=\\);'\\u0022<>ʃ）/／ノﾉ丿╯σっµ~\\*]+";
+    static String eeLeft = "[＼\\\\ƪԄ\\(（<>;ヽ\\-=~\\*]+";
+    static String eeRight= "[\\-=\\);'\\u0022<>ʃ）/／ノﾉ丿╯σっµ~\\*]+";
+    static String eeSymbol = "[^A-Za-z0-9\\s\\(\\)\\*:=-]";
+    static String eastEmote = eeLeft + "(?:"+basicface+"|" +eeSymbol+")+" + eeRight;
+
 
     public static String emoticon = OR(
             // Standard version  :) :( :] :D :P