Multi-byte name problem Shift_JIS(Japanese)

[yshobby]

Hi.

There are plural letter expression methods as a problem in specifications of Unicode.
Ex.
HIRAGANA LETTER BA : u3070 == u306f u3099
HIRAGANA LETTER PA : u3071 == u306f u309a

Local file system absorbs the problem. But Shift_JIS used server Problem. However, a problem happens with a server using Shift_JIS for.
Ex. UTF -> Shift_JIS

u3070 -> 0x82ce[[BR]]
u306f u3099 -> 0x82cd 0x3f(error)

I am not good at the Java language. Therefore I do not understand the most suitable method, but deal by the following methods.

ch.cyberduck.core.Path
    public static String normalize(final String path) {
              ||
              ||
        // Return the normalized path that we have completed[[BR]]
        //return normalized;[[BR]]
		String n="";[[BR]]
		char chw = 0;[[BR]]
		for(int i=0; i<normalized.length(); i++) {[[BR]]
			char ch = normalized.charAt(i);[[BR]]
			if(ch == 12441) {			//	daku[[BR]]
				chw += 1;[[BR]]
			} else if(ch == 12442) {	//	handaku[[BR]]
				chw += 2;[[BR]]
			} else {[[BR]]
				if(chw != 0) {[[BR]]
					n += chw;[[BR]]
				}[[BR]]
				chw = ch;[[BR]]
			}[[BR]]
		}[[BR]]
		n += chw;[[BR]]
        return n;[[BR]]
    }

Please examine an appropriate method.

thanks.

[dkocher]

​http://www.icu-project.org/apiref/icu4j/com/ibm/icu/text/Normalizer.html may be of use here.

[dkocher]

Note to me: ​http://java.sun.com/mailers/techtips/corejava/2007/tt0207.html

[dkocher]

Should be fixed in r3132. All filnames are now normalized.