UTF8 to Byte Array : UTF « Development « Android

Home
Android
1.2D Graphics
2.Animation
3.Core Class
4.Database
5.Date Type
6.Development
7.File
8.Game
9.Hardware
10.Media
11.Network
12.Security
13.UI
14.User Event
Android » Development » UTF 
UTF8 to Byte Array
 
//package org.bouncycastle.util;

import java.io.ByteArrayOutputStream;
import java.util.Vector;

public final class Strings
{
    public static String fromUTF8ByteArray(byte[] bytes)
    {
        int i = 0;
        int length = 0;

        while (i < bytes.length)
        {
            length++;
            if ((bytes[i0xf0== 0xf0)
            {
                // surrogate pair
                length++;
                i += 4;
            }
            else if ((bytes[i0xe0== 0xe0)
            {
                i += 3;
            }
            else if ((bytes[i0xc0== 0xc0)
            {
                i += 2;
            }
            else
            {
                i += 1;
            }
        }

        char[] cs = new char[length];

        i = 0;
        length = 0;

        while (i < bytes.length)
        {
            char ch;

            if ((bytes[i0xf0== 0xf0)
            {
                int codePoint = ((bytes[i0x03<< 18((bytes[i+10x3F<< 12((bytes[i+20x3F<< 6(bytes[i+30x3F);
                int U = codePoint - 0x10000;
                char W1 = (char)(0xD800 (U >> 10));
                char W2 = (char)(0xDC00 (U & 0x3FF));
                cs[length++= W1;
                ch = W2;
                i += 4;
            }
            else if ((bytes[i0xe0== 0xe0)
            {
                ch = (char)(((bytes[i0x0f<< 12)
                        ((bytes[i + 10x3f<< 6(bytes[i + 20x3f));
                i += 3;
            }
            else if ((bytes[i0xd0== 0xd0)
            {
                ch = (char)(((bytes[i0x1f<< 6(bytes[i + 10x3f));
                i += 2;
            }
            else if ((bytes[i0xc0== 0xc0)
            {
                ch = (char)(((bytes[i0x1f<< 6(bytes[i + 10x3f));
                i += 2;
            }
            else
            {
                ch = (char)(bytes[i0xff);
                i += 1;
            }

            cs[length++= ch;
        }

        return new String(cs);
    }
    
    public static byte[] toUTF8ByteArray(String string)
    {
        return toUTF8ByteArray(string.toCharArray());
    }

    public static byte[] toUTF8ByteArray(char[] string)
    {
        ByteArrayOutputStream bOut = new ByteArrayOutputStream();
        char[] c = string;
        int i = 0;

        while (i < c.length)
        {
            char ch = c[i];

            if (ch < 0x0080)
            {
                bOut.write(ch);
            }
            else if (ch < 0x0800)
            {
                bOut.write(0xc0 (ch >> 6));
                bOut.write(0x80 (ch & 0x3f));
            }
            // surrogate pair
            else if (ch >= 0xD800 && ch <= 0xDFFF)
            {
                // in error - can only happen, if the Java String class has a
                // bug.
                if (i + >= c.length)
                {
                    throw new IllegalStateException("invalid UTF-16 codepoint");
                }
                char W1 = ch;
                ch = c[++i];
                char W2 = ch;
                // in error - can only happen, if the Java String class has a
                // bug.
                if (W1 > 0xDBFF)
                {
                    throw new IllegalStateException("invalid UTF-16 codepoint");
                }
                int codePoint = (((W1 & 0x03FF<< 10(W2 & 0x03FF)) 0x10000;
                bOut.write(0xf0 (codePoint >> 18));
                bOut.write(0x80 ((codePoint >> 120x3F));
                bOut.write(0x80 ((codePoint >> 60x3F));
                bOut.write(0x80 (codePoint & 0x3F));
            }
            else
            {
                bOut.write(0xe0 (ch >> 12));
                bOut.write(0x80 ((ch >> 60x3F));
                bOut.write(0x80 (ch & 0x3F));
            }

            i++;
        }
        
        return bOut.toByteArray();
    }
    
    /**
     * A locale independent version of toUpperCase.
     
     @param string input to be converted
     @return a US Ascii uppercase version
     */
    public static String toUpperCase(String string)
    {
        boolean changed = false;
        char[] chars = string.toCharArray();
        
        for (int i = 0; i != chars.length; i++)
        {
            char ch = chars[i];
            if ('a' <= ch && 'z' >= ch)
            {
                changed = true;
                chars[i(char)(ch - 'a' 'A');
            }
        }
        
        if (changed)
        {
            return new String(chars);
        }
        
        return string;
    }
    
    /**
     * A locale independent version of toLowerCase.
     
     @param string input to be converted
     @return a US ASCII lowercase version
     */
    public static String toLowerCase(String string)
    {
        boolean changed = false;
        char[] chars = string.toCharArray();
        
        for (int i = 0; i != chars.length; i++)
        {
            char ch = chars[i];
            if ('A' <= ch && 'Z' >= ch)
            {
                changed = true;
                chars[i(char)(ch - 'A' 'a');
            }
        }
        
        if (changed)
        {
            return new String(chars);
        }
        
        return string;
    }

    public static byte[] toByteArray(char[] chars)
    {
        byte[] bytes = new byte[chars.length];

        for (int i = 0; i != bytes.length; i++)
        {
            bytes[i(byte)chars[i];
        }

        return bytes;
    }

    public static byte[] toByteArray(String string)
    {
        byte[] bytes = new byte[string.length()];

        for (int i = 0; i != bytes.length; i++)
        {
            char ch = string.charAt(i);

            bytes[i(byte)ch;
        }

        return bytes;
    }

    public static String[] split(String input, char delimiter)
    {
        Vector           v = new Vector();
        boolean moreTokens = true;
        String subString;

        while (moreTokens)
        {
            int tokenLocation = input.indexOf(delimiter);
            if (tokenLocation > 0)
            {
                subString = input.substring(0, tokenLocation);
                v.addElement(subString);
                input = input.substring(tokenLocation + 1);
            }
            else
            {
                moreTokens = false;
                v.addElement(input);
            }
        }

        String[] res = new String[v.size()];

        for (int i = 0; i != res.length; i++)
        {
            res[i(String)v.elementAt(i);
        }
        return res;
    }
}

   
  
Related examples in the same category
1.decode UTF8 to char
2.decode UTF8 to String
3.implements InputFilter
4.UTF 2 GBUtil
5.Load UTF8 encoded file to String
java2s.com  | Contact Us | Privacy Policy
Copyright 2009 - 12 Demo Source and Support. All rights reserved.
All other trademarks are property of their respective owners.