Changeset 16641

Timestamp:

2008-08-04T12:45:59+12:00 (16 years ago)

Author:

kjdon

Message:

upgraded this (using unicode 4.0) to include more Chinese characters and Japanese and Korean characters

File:

• gsdl/trunk/perllib/cnseg.pm (modified) (1 diff)

gsdl/trunk/perllib/cnseg.pm

@@ -55 +55 @@
     my $space = 1; # start doesn't need a space
     foreach $c (@$uniin) {
-    if (($c >= 0x4e00 && $c <= 0x9fa5) ||
-        ($c >= 0xf900 && $c <= 0xfa2d)) {
-        # Chinese character
+    if (($c >= 0x2e80 && $c <= 0xfa6a) || # main east asian codes
+        ($c >= 0x20000 && $c <= 0x2a6d6) || # cjk unified ideographs ext B
+        ($c >= 0x2f800 && $c <= 0x2fa1d)) { #cjk compatibility ideographs supplement
+        # CJK character
         push (@$out, 0x200b) unless $space;
         push (@$out, $c);