过滤3个字节以上的utf-8字符

/**
     * 过滤掉超过3个字节的UTF8字符
     * @param text
     * @return
     * @throws UnsupportedEncodingException
     */
    public static String filterOffUtf8Mb4(String text) throws UnsupportedEncodingException {
        byte[] bytes = text.getBytes("utf-8");
        ByteBuffer buffer = ByteBuffer.allocate(bytes.length);
        int i = 0;
        while (i < bytes.length) {
            short b = bytes[i];
            if (b > 0) {
                buffer.put(bytes[i++]);
                continue;
            }

            b += 256; // 去掉符号位

            if (((b >> 5) ^ 0x6) == 0) {
                buffer.put(bytes, i, 2);
                i += 2;
            } else if (((b >> 4) ^ 0xE) == 0) {
                buffer.put(bytes, i, 3);
                i += 3;
            } else if (((b >> 3) ^ 0x1E) == 0) {
                i += 4;
            } else if (((b >> 2) ^ 0x3E) == 0) {
                i += 5;
            } else if (((b >> 1) ^ 0x7E) == 0) {
                i += 6;
            } else {
                buffer.put(bytes[i++]);
            }
        }
        buffer.flip();
        return new String(buffer.array(), "utf-8");
    }

 

posted on 2017-09-07 15:44  little fat  阅读(470)  评论(0编辑  收藏  举报