lib.utf.js

/* UTF-8 <=> UTF-16 convertion library.
 *
/* Copyright (C) 1999 Masanao Izumo <iz@onicos.co.jp>
 * 2007 Ma Bingyao <andot@ujn.edu.cn>
 * Version: 2.1
 * LastModified: Feb 25, 2007
 * This library is free. You can redistribute it and/or modify it.
 */

/*
 * Interfaces:
 * utf8 = utf16to8(utf16);
 * utf16 = utf16to8(utf8);
 */

function utf16to8(str) {
    if (str.match(/^[/x00-/x7f]*$/) != null) {
        return str;
    }
    var out, i, j, len, c, c2;
    out = [];
    len = str.length;
    for (i = 0, j = 0; i < len; i++, j++) {
        c = str.charCodeAt(i);
        if (c <= 0x7f) {
            out[j] = str.charAt(i);
        }
        else if (c <= 0x7ff) {
            out[j] = String.fromCharCode(0xc0 | (c >>> 6),
                 0x80 | (c & 0x3f));
        }
        else if (c < 0xd800 || c > 0xdfff) {
            out[j] = String.fromCharCode(0xe0 | (c >>> 12),
                 0x80 | ((c >>> 6) & 0x3f),
                 0x80 | (c & 0x3f));
        }
        else {
            if (++i < len) {
                c2 = str.charCodeAt(i);
                if (c <= 0xdbff && 0xdc00 <= c2 && c2 <= 0xdfff) {
                 c = ((c & 0x03ff) << 10 | (c2 & 0x03ff)) + 0x010000;
                 if (0x010000 <= c && c <= 0x10ffff) {
                 out[j] = String.fromCharCode(0xf0 | ((c >>> 18) & 0x3f),
                 0x80 | ((c >>> 12) & 0x3f),