Changeset 199 in lab

Timestamp:

Jan 26, 2013 2:18:36 AM (11 years ago)

Author:

mitty

Message:

• fix: convert non Shift_JIS character only
  • skip needless convert (such as U+8C37 -> U+7A40)

File:

• trunk/misc/findsimplified.pl (modified) (1 diff)

trunk/misc/findsimplified.pl

@@ -20 +20 @@
     
     $str = $utf8->decode($str);
-    my $jtext = $ck->conv_c2j($str);
-    if ($str ne $jtext) {
-        return $utf8->encode($jtext);
+    my $sjis = encode("cp932", $str, Encode::FB_HTMLCREF);
+    if ($sjis =~ /&#\d{4,};/) {
+        my $jtext = $ck->conv_c2j($str);
+        if ($str ne $jtext) {
+            return $utf8->encode($jtext);
+        }
     }