libexfat/utf.c

   1 /*
   2  *  utf.c
   3  *  exFAT file system implementation library.
   4  *
   5  *  Created by Andrew Nayenko on 13.09.09.
   6  *  This software is distributed under the GNU General Public License
   7  *  version 3 or any later.
   8  */
   9
  10 #include "exfat.h"
  11 #include <errno.h>
  12
  13 static char* wchar_to_utf8(char* output, wchar_t wc, size_t outsize)
  14 {
  15         if (wc <= 0x7f)
  16         {
  17                 if (outsize < 1)
  18                         return NULL;
  19                 *output++ = (char) wc;
  20         }
  21         else if (wc <= 0x7ff)
  22         {
  23                 if (outsize < 2)
  24                         return NULL;
  25                 *output++ = 0xc0 | (wc >> 6);
  26                 *output++ = 0x80 | (wc & 0x3f);
  27         }
  28         else if (wc <= 0xffff)
  29         {
  30                 if (outsize < 3)
  31                         return NULL;
  32                 *output++ = 0xe0 | (wc >> 12);
  33                 *output++ = 0x80 | ((wc >> 6) & 0x3f);
  34                 *output++ = 0x80 | (wc & 0x3f);
  35         }
  36         else if (wc <= 0x1fffff)
  37         {
  38                 if (outsize < 4)
  39                         return NULL;
  40                 *output++ = 0xf0 | (wc >> 18);
  41                 *output++ = 0x80 | ((wc >> 12) & 0x3f);
  42                 *output++ = 0x80 | ((wc >> 6) & 0x3f);
  43                 *output++ = 0x80 | (wc & 0x3f);
  44         }
  45         else if (wc <= 0x3ffffff)
  46         {
  47                 if (outsize < 5)
  48                         return NULL;
  49                 *output++ = 0xf8 | (wc >> 24);
  50                 *output++ = 0x80 | ((wc >> 18) & 0x3f);
  51                 *output++ = 0x80 | ((wc >> 12) & 0x3f);
  52                 *output++ = 0x80 | ((wc >> 6) & 0x3f);
  53                 *output++ = 0x80 | (wc & 0x3f);
  54         }
  55         else if (wc <= 0x7fffffff)
  56         {
  57                 if (outsize < 6)
  58                         return NULL;
  59                 *output++ = 0xfc | (wc >> 30);
  60                 *output++ = 0x80 | ((wc >> 24) & 0x3f);
  61                 *output++ = 0x80 | ((wc >> 18) & 0x3f);
  62                 *output++ = 0x80 | ((wc >> 12) & 0x3f);
  63                 *output++ = 0x80 | ((wc >> 6) & 0x3f);
  64                 *output++ = 0x80 | (wc & 0x3f);
  65         }
  66         else
  67                 return NULL;
  68
  69         return output;
  70 }
  71
  72 static const le16_t* utf16_to_wchar(const le16_t* input, wchar_t* wc,
  73                 size_t insize)
  74 {
  75         if ((le16_to_cpu(input[0]) & 0xfc00) == 0xd800)
  76         {
  77                 if (insize < 2 || (le16_to_cpu(input[1]) & 0xfc00) != 0xdc00)
  78                         return NULL;
  79                 *wc = ((wchar_t) (le16_to_cpu(input[0]) & 0x3ff) << 10);
  80                 *wc |= (le16_to_cpu(input[1]) & 0x3ff);
  81                 return input + 2;
  82         }
  83         else
  84         {
  85                 *wc = le16_to_cpu(*input);
  86                 return input + 1;
  87         }
  88 }
  89
  90 int utf16_to_utf8(char* output, const le16_t* input, size_t outsize,
  91                 size_t insize)
  92 {
  93         const le16_t* inp = input;
  94         char* outp = output;
  95         wchar_t wc;
  96
  97         while (inp - input < insize && le16_to_cpu(*inp))
  98         {
  99                 inp = utf16_to_wchar(inp, &wc, insize - (inp - input));
 100                 if (inp == NULL)
 101                 {
 102                         exfat_error("illegal UTF-16 sequence");
 103                         return -EILSEQ;
 104                 }
 105                 outp = wchar_to_utf8(outp, wc, outsize - (outp - output));
 106                 if (outp == NULL)
 107                 {
 108                         exfat_error("name is too long");
 109                         return -ENAMETOOLONG;
 110                 }
 111         }
 112         *outp = '\0';
 113         return 0;
 114 }
 115
 116 static const char* utf8_to_wchar(const char* input, wchar_t* wc,
 117                 size_t insize)
 118 {
 119         if ((input[0] & 0x80) == 0 && insize >= 1)
 120         {
 121                 *wc = (wchar_t) input[0];
 122                 return input + 1;
 123         }
 124         if ((input[0] & 0xe0) == 0xc0 && insize >= 2)
 125         {
 126                 *wc = (((wchar_t) input[0] & 0x1f) << 6) |
 127                        ((wchar_t) input[1] & 0x3f);
 128                 return input + 2;
 129         }
 130         if ((input[0] & 0xf0) == 0xe0 && insize >= 3)
 131         {
 132                 *wc = (((wchar_t) input[0] & 0x0f) << 12) |
 133                       (((wchar_t) input[1] & 0x3f) << 6) |
 134                        ((wchar_t) input[2] & 0x3f);
 135                 return input + 3;
 136         }
 137         if ((input[0] & 0xf8) == 0xf0 && insize >= 4)
 138         {
 139                 *wc = (((wchar_t) input[0] & 0x07) << 18) |
 140                       (((wchar_t) input[1] & 0x3f) << 12) |
 141                       (((wchar_t) input[2] & 0x3f) << 6) |
 142                        ((wchar_t) input[3] & 0x3f);
 143                 return input + 4;
 144         }
 145         if ((input[0] & 0xfc) == 0xf8 && insize >= 5)
 146         {
 147                 *wc = (((wchar_t) input[0] & 0x03) << 24) |
 148                       (((wchar_t) input[1] & 0x3f) << 18) |
 149                       (((wchar_t) input[2] & 0x3f) << 12) |
 150                       (((wchar_t) input[3] & 0x3f) << 6) |
 151                        ((wchar_t) input[4] & 0x3f);
 152                 return input + 5;
 153         }
 154         if ((input[0] & 0xfe) == 0xfc && insize >= 6)
 155         {
 156                 *wc = (((wchar_t) input[0] & 0x01) << 30) |
 157                       (((wchar_t) input[1] & 0x3f) << 24) |
 158                       (((wchar_t) input[2] & 0x3f) << 18) |
 159                       (((wchar_t) input[3] & 0x3f) << 12) |
 160                       (((wchar_t) input[4] & 0x3f) << 6) |
 161                        ((wchar_t) input[5] & 0x3f);
 162                 return input + 6;
 163         }
 164         return NULL;
 165 }
 166
 167 static le16_t* wchar_to_utf16(le16_t* output, wchar_t wc, size_t outsize)
 168 {
 169         if (wc <= 0xffff) /* if character is from BMP */
 170         {
 171                 if (outsize == 0)
 172                 {
 173                         exfat_error("name is too long");
 174                         return NULL;
 175                 }
 176                 output[0] = cpu_to_le16(wc);
 177                 return output + 1;
 178         }
 179         if (outsize < 2)
 180         {
 181                 exfat_error("name is too long");
 182                 return NULL;
 183         }
 184         output[0] = cpu_to_le16(0xd800 | ((wc >> 10) & 0x3ff));
 185         output[1] = cpu_to_le16(0xdc00 | (wc & 0x3ff));
 186         return output + 2;
 187 }
 188
 189 int utf8_to_utf16(le16_t* output, const char* input, size_t outsize,
 190                 size_t insize)
 191 {
 192         const char* inp = input;
 193         le16_t* outp = output;
 194         wchar_t wc;
 195
 196         while (inp - input < insize && *inp)
 197         {
 198                 inp = utf8_to_wchar(inp, &wc, insize - (inp - input));
 199                 if (inp == NULL)
 200                 {
 201                         exfat_error("illegal UTF-8 sequence");
 202                         return -EILSEQ;
 203                 }
 204                 outp = wchar_to_utf16(outp, wc, outsize - (outp - output));
 205                 if (outp == NULL)
 206                 {
 207                         exfat_error("name is too long");
 208                         return -ENAMETOOLONG;
 209                 }
 210         }
 211         *outp = cpu_to_le16(0);
 212         return 0;
 213 }