Home | History | Annotate | Download | only in util
      1 package org.bouncycastle.util;
      2 
      3 import java.io.ByteArrayOutputStream;
      4 import java.util.Vector;
      5 
      6 public final class Strings
      7 {
      8     public static String fromUTF8ByteArray(byte[] bytes)
      9     {
     10         int i = 0;
     11         int length = 0;
     12 
     13         while (i < bytes.length)
     14         {
     15             length++;
     16             if ((bytes[i] & 0xf0) == 0xf0)
     17             {
     18                 // surrogate pair
     19                 length++;
     20                 i += 4;
     21             }
     22             else if ((bytes[i] & 0xe0) == 0xe0)
     23             {
     24                 i += 3;
     25             }
     26             else if ((bytes[i] & 0xc0) == 0xc0)
     27             {
     28                 i += 2;
     29             }
     30             else
     31             {
     32                 i += 1;
     33             }
     34         }
     35 
     36         char[] cs = new char[length];
     37 
     38         i = 0;
     39         length = 0;
     40 
     41         while (i < bytes.length)
     42         {
     43             char ch;
     44 
     45             if ((bytes[i] & 0xf0) == 0xf0)
     46             {
     47                 int codePoint = ((bytes[i] & 0x03) << 18) | ((bytes[i+1] & 0x3F) << 12) | ((bytes[i+2] & 0x3F) << 6) | (bytes[i+3] & 0x3F);
     48                 int U = codePoint - 0x10000;
     49                 char W1 = (char)(0xD800 | (U >> 10));
     50                 char W2 = (char)(0xDC00 | (U & 0x3FF));
     51                 cs[length++] = W1;
     52                 ch = W2;
     53                 i += 4;
     54             }
     55             else if ((bytes[i] & 0xe0) == 0xe0)
     56             {
     57                 ch = (char)(((bytes[i] & 0x0f) << 12)
     58                         | ((bytes[i + 1] & 0x3f) << 6) | (bytes[i + 2] & 0x3f));
     59                 i += 3;
     60             }
     61             else if ((bytes[i] & 0xd0) == 0xd0)
     62             {
     63                 ch = (char)(((bytes[i] & 0x1f) << 6) | (bytes[i + 1] & 0x3f));
     64                 i += 2;
     65             }
     66             else if ((bytes[i] & 0xc0) == 0xc0)
     67             {
     68                 ch = (char)(((bytes[i] & 0x1f) << 6) | (bytes[i + 1] & 0x3f));
     69                 i += 2;
     70             }
     71             else
     72             {
     73                 ch = (char)(bytes[i] & 0xff);
     74                 i += 1;
     75             }
     76 
     77             cs[length++] = ch;
     78         }
     79 
     80         return new String(cs);
     81     }
     82 
     83     public static byte[] toUTF8ByteArray(String string)
     84     {
     85         return toUTF8ByteArray(string.toCharArray());
     86     }
     87 
     88     public static byte[] toUTF8ByteArray(char[] string)
     89     {
     90         ByteArrayOutputStream bOut = new ByteArrayOutputStream();
     91         char[] c = string;
     92         int i = 0;
     93 
     94         while (i < c.length)
     95         {
     96             char ch = c[i];
     97 
     98             if (ch < 0x0080)
     99             {
    100                 bOut.write(ch);
    101             }
    102             else if (ch < 0x0800)
    103             {
    104                 bOut.write(0xc0 | (ch >> 6));
    105                 bOut.write(0x80 | (ch & 0x3f));
    106             }
    107             // surrogate pair
    108             else if (ch >= 0xD800 && ch <= 0xDFFF)
    109             {
    110                 // in error - can only happen, if the Java String class has a
    111                 // bug.
    112                 if (i + 1 >= c.length)
    113                 {
    114                     throw new IllegalStateException("invalid UTF-16 codepoint");
    115                 }
    116                 char W1 = ch;
    117                 ch = c[++i];
    118                 char W2 = ch;
    119                 // in error - can only happen, if the Java String class has a
    120                 // bug.
    121                 if (W1 > 0xDBFF)
    122                 {
    123                     throw new IllegalStateException("invalid UTF-16 codepoint");
    124                 }
    125                 int codePoint = (((W1 & 0x03FF) << 10) | (W2 & 0x03FF)) + 0x10000;
    126                 bOut.write(0xf0 | (codePoint >> 18));
    127                 bOut.write(0x80 | ((codePoint >> 12) & 0x3F));
    128                 bOut.write(0x80 | ((codePoint >> 6) & 0x3F));
    129                 bOut.write(0x80 | (codePoint & 0x3F));
    130             }
    131             else
    132             {
    133                 bOut.write(0xe0 | (ch >> 12));
    134                 bOut.write(0x80 | ((ch >> 6) & 0x3F));
    135                 bOut.write(0x80 | (ch & 0x3F));
    136             }
    137 
    138             i++;
    139         }
    140 
    141         return bOut.toByteArray();
    142     }
    143 
    144     /**
    145      * A locale independent version of toUpperCase.
    146      *
    147      * @param string input to be converted
    148      * @return a US Ascii uppercase version
    149      */
    150     public static String toUpperCase(String string)
    151     {
    152         boolean changed = false;
    153         char[] chars = string.toCharArray();
    154 
    155         for (int i = 0; i != chars.length; i++)
    156         {
    157             char ch = chars[i];
    158             if ('a' <= ch && 'z' >= ch)
    159             {
    160                 changed = true;
    161                 chars[i] = (char)(ch - 'a' + 'A');
    162             }
    163         }
    164 
    165         if (changed)
    166         {
    167             return new String(chars);
    168         }
    169 
    170         return string;
    171     }
    172 
    173     /**
    174      * A locale independent version of toLowerCase.
    175      *
    176      * @param string input to be converted
    177      * @return a US ASCII lowercase version
    178      */
    179     public static String toLowerCase(String string)
    180     {
    181         boolean changed = false;
    182         char[] chars = string.toCharArray();
    183 
    184         for (int i = 0; i != chars.length; i++)
    185         {
    186             char ch = chars[i];
    187             if ('A' <= ch && 'Z' >= ch)
    188             {
    189                 changed = true;
    190                 chars[i] = (char)(ch - 'A' + 'a');
    191             }
    192         }
    193 
    194         if (changed)
    195         {
    196             return new String(chars);
    197         }
    198 
    199         return string;
    200     }
    201 
    202     public static byte[] toByteArray(String string)
    203     {
    204         byte[] bytes = new byte[string.length()];
    205 
    206         for (int i = 0; i != bytes.length; i++)
    207         {
    208             char ch = string.charAt(i);
    209 
    210             bytes[i] = (byte)ch;
    211         }
    212 
    213         return bytes;
    214     }
    215 
    216     public static String[] split(String input, char delimiter)
    217     {
    218         Vector           v = new Vector();
    219         boolean moreTokens = true;
    220         String subString;
    221 
    222         while (moreTokens)
    223         {
    224             int tokenLocation = input.indexOf(delimiter);
    225             if (tokenLocation > 0)
    226             {
    227                 subString = input.substring(0, tokenLocation);
    228                 v.addElement(subString);
    229                 input = input.substring(tokenLocation + 1);
    230             }
    231             else
    232             {
    233                 moreTokens = false;
    234                 v.addElement(input);
    235             }
    236         }
    237 
    238         String[] res = new String[v.size()];
    239 
    240         for (int i = 0; i != res.length; i++)
    241         {
    242             res[i] = (String)v.elementAt(i);
    243         }
    244         return res;
    245     }
    246 }
    247