public class CodingUtil {
public static String decode(String query) throws Exception {
if (query == null)
return null;
String utfStr = java.net.URLDecoder.decode(query, "UTF-8");
String gbkStr = java.net.URLDecoder.decode(query, "GBK");
boolean encoded = (query.equals(utfStr));
if (encoded) {
utfStr = new String(query.getBytes("ISO-8859-1"), "UTF-8");
gbkStr = new String(query.getBytes("ISO-8859-1"), "GBK");
}
String src = null;
boolean isUTF = isUTF(utfStr.getBytes("UTF-8")); // java.util.Arrays.equals(utfStr.getBytes("UTF-8")
boolean isGBK = isGBK(gbkStr.getBytes("GBK"));// gbkStr.getBytes("GBK"));
if (isUTF && (!isGBK)) {
src = utfStr;
} else {
src = gbkStr;
}
return src;
}
static int L1 = 0x80;// ascii,UFT8可用1-6个字节编码,ASCII用一个字节
static int L2 = 0xC0;
static int L3 = 0xE0;
static int L4 = 0xF0;
static int LS = 0x80;
private static boolean isUTF(byte[] bytes) {
for (int i = 0; i < bytes.length;) {
int b = getInt(bytes[i]);
if ((L1 & b) == 0x00) {
i++;
} else if ((L4 & b) == L4 && (getInt(bytes[i + 1]) & LS) == LS
&& (getInt(bytes[i + 2]) & LS) == LS
&& (getInt(bytes[i + 3]) & LS) == LS) {
i += 4;
} else if ((L3 & b) == L3 && (getInt(bytes[i + 1]) & LS) == LS
&& (getInt(bytes[i + 2]) & LS) == LS) {
i += 3;
} else if ((L2 & b) == L2 && (getInt(bytes[i + 1]) & LS) == LS) {
i += 2;
} else {
System.out.println(" ====>" + i);
return false;
}
}
return true;
}
private static int getInt(byte b) {
return b & 0xFF;
}
static int GBK_START = 0x8140;
static int GBK_END = 0xFEFE;
private static boolean isGBK(byte[] bytes) {
for (int i = 0; i < bytes.length;) {
if (i + 2 > bytes.length)
return false;
int b = getInt(bytes[i]);
if ((L1 & b) == 0x00) {
i++;
continue;
}
int high = b << 8;
int c = high + getInt(bytes[i + 1]);
if (c < GBK_START || c > GBK_END) {
return false;
}
i += 2;
}
return true;
}
}
浏览器地址栏编码
最新推荐文章于 2024-01-15 19:25:23 发布