lib/unistr/u8-mbtouc.c

   1 /* Look at first character in UTF-8 string.
   2    Copyright (C) 1999-2002, 2006-2007, 2009-2012 Free Software Foundation, Inc.
   3    Written by Bruno Haible <bruno@clisp.org>, 2001.
   4
   5    This program is free software: you can redistribute it and/or modify it
   6    under the terms of the GNU Lesser General Public License as published
   7    by the Free Software Foundation; either version 3 of the License, or
   8    (at your option) any later version.
   9
  10    This program is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13    Lesser General Public License for more details.
  14
  15    You should have received a copy of the GNU Lesser General Public License
  16    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  17
  18 #include <config.h>
  19
  20 #if defined IN_LIBUNISTRING
  21 /* Tell unistr.h to declare u8_mbtouc as 'extern', not 'static inline'.  */
  22 # include "unistring-notinline.h"
  23 #endif
  24
  25 /* Specification.  */
  26 #include "unistr.h"
  27
  28 #if !HAVE_INLINE
  29
  30 int
  31 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n)
  32 {
  33   uint8_t c = *s;
  34
  35   if (c < 0x80)
  36     {
  37       *puc = c;
  38       return 1;
  39     }
  40   else if (c >= 0xc2)
  41     {
  42       if (c < 0xe0)
  43         {
  44           if (n >= 2)
  45             {
  46               if ((s[1] ^ 0x80) < 0x40)
  47                 {
  48                   *puc = ((unsigned int) (c & 0x1f) << 6)
  49                          | (unsigned int) (s[1] ^ 0x80);
  50                   return 2;
  51                 }
  52               /* invalid multibyte character */
  53             }
  54           else
  55             {
  56               /* incomplete multibyte character */
  57               *puc = 0xfffd;
  58               return 1;
  59             }
  60         }
  61       else if (c < 0xf0)
  62         {
  63           if (n >= 3)
  64             {
  65               if ((s[1] ^ 0x80) < 0x40)
  66                 {
  67                   if ((s[2] ^ 0x80) < 0x40)
  68                     {
  69                       if ((c >= 0xe1 || s[1] >= 0xa0)
  70                           && (c != 0xed || s[1] < 0xa0))
  71                         {
  72                           *puc = ((unsigned int) (c & 0x0f) << 12)
  73                                  | ((unsigned int) (s[1] ^ 0x80) << 6)
  74                                  | (unsigned int) (s[2] ^ 0x80);
  75                           return 3;
  76                         }
  77                       /* invalid multibyte character */
  78                       *puc = 0xfffd;
  79                       return 3;
  80                     }
  81                   /* invalid multibyte character */
  82                   *puc = 0xfffd;
  83                   return 2;
  84                 }
  85               /* invalid multibyte character */
  86             }
  87           else
  88             {
  89               /* incomplete multibyte character */
  90               *puc = 0xfffd;
  91               if (n == 1 || (s[1] ^ 0x80) >= 0x40)
  92                 return 1;
  93               else
  94                 return 2;
  95             }
  96         }
  97       else if (c < 0xf8)
  98         {
  99           if (n >= 4)
 100             {
 101               if ((s[1] ^ 0x80) < 0x40)
 102                 {
 103                   if ((s[2] ^ 0x80) < 0x40)
 104                     {
 105                       if ((s[3] ^ 0x80) < 0x40)
 106                         {
 107                           if ((c >= 0xf1 || s[1] >= 0x90)
 108 #if 1
 109                               && (c < 0xf4 || (c == 0xf4 && s[1] < 0x90))
 110 #endif
 111                              )
 112                             {
 113                               *puc = ((unsigned int) (c & 0x07) << 18)
 114                                      | ((unsigned int) (s[1] ^ 0x80) << 12)
 115                                      | ((unsigned int) (s[2] ^ 0x80) << 6)
 116                                      | (unsigned int) (s[3] ^ 0x80);
 117                               return 4;
 118                             }
 119                           /* invalid multibyte character */
 120                           *puc = 0xfffd;
 121                           return 4;
 122                         }
 123                       /* invalid multibyte character */
 124                       *puc = 0xfffd;
 125                       return 3;
 126                     }
 127                   /* invalid multibyte character */
 128                   *puc = 0xfffd;
 129                   return 2;
 130                 }
 131               /* invalid multibyte character */
 132             }
 133           else
 134             {
 135               /* incomplete multibyte character */
 136               *puc = 0xfffd;
 137               if (n == 1 || (s[1] ^ 0x80) >= 0x40)
 138                 return 1;
 139               else if (n == 2 || (s[2] ^ 0x80) >= 0x40)
 140                 return 2;
 141               else
 142                 return 3;
 143             }
 144         }
 145 #if 0
 146       else if (c < 0xfc)
 147         {
 148           if (n >= 5)
 149             {
 150               if ((s[1] ^ 0x80) < 0x40)
 151                 {
 152                   if ((s[2] ^ 0x80) < 0x40)
 153                     {
 154                       if ((s[3] ^ 0x80) < 0x40)
 155                         {
 156                           if ((s[4] ^ 0x80) < 0x40)
 157                             {
 158                               if (c >= 0xf9 || s[1] >= 0x88)
 159                                 {
 160                                   *puc = ((unsigned int) (c & 0x03) << 24)
 161                                          | ((unsigned int) (s[1] ^ 0x80) << 18)
 162                                          | ((unsigned int) (s[2] ^ 0x80) << 12)
 163                                          | ((unsigned int) (s[3] ^ 0x80) << 6)
 164                                          | (unsigned int) (s[4] ^ 0x80);
 165                                   return 5;
 166                                 }
 167                               /* invalid multibyte character */
 168                               *puc = 0xfffd;
 169                               return 5;
 170                             }
 171                           /* invalid multibyte character */
 172                           *puc = 0xfffd;
 173                           return 4;
 174                         }
 175                       /* invalid multibyte character */
 176                       *puc = 0xfffd;
 177                       return 3;
 178                     }
 179                   /* invalid multibyte character */
 180                   return 2;
 181                 }
 182               /* invalid multibyte character */
 183             }
 184           else
 185             {
 186               /* incomplete multibyte character */
 187               *puc = 0xfffd;
 188               return n;
 189             }
 190         }
 191       else if (c < 0xfe)
 192         {
 193           if (n >= 6)
 194             {
 195               if ((s[1] ^ 0x80) < 0x40)
 196                 {
 197                   if ((s[2] ^ 0x80) < 0x40)
 198                     {
 199                       if ((s[3] ^ 0x80) < 0x40)
 200                         {
 201                           if ((s[4] ^ 0x80) < 0x40)
 202                             {
 203                               if ((s[5] ^ 0x80) < 0x40)
 204                                 {
 205                                   if (c >= 0xfd || s[1] >= 0x84)
 206                                     {
 207                                       *puc = ((unsigned int) (c & 0x01) << 30)
 208                                              | ((unsigned int) (s[1] ^ 0x80) << 24)
 209                                              | ((unsigned int) (s[2] ^ 0x80) << 18)
 210                                              | ((unsigned int) (s[3] ^ 0x80) << 12)
 211                                              | ((unsigned int) (s[4] ^ 0x80) << 6)
 212                                              | (unsigned int) (s[5] ^ 0x80);
 213                                       return 6;
 214                                     }
 215                                   /* invalid multibyte character */
 216                                   *puc = 0xfffd;
 217                                   return 6;
 218                                 }
 219                               /* invalid multibyte character */
 220                               *puc = 0xfffd;
 221                               return 5;
 222                             }
 223                           /* invalid multibyte character */
 224                           *puc = 0xfffd;
 225                           return 4;
 226                         }
 227                       /* invalid multibyte character */
 228                       *puc = 0xfffd;
 229                       return 3;
 230                     }
 231                   /* invalid multibyte character */
 232                   return 2;
 233                 }
 234               /* invalid multibyte character */
 235             }
 236           else
 237             {
 238               /* incomplete multibyte character */
 239               *puc = 0xfffd;
 240               return n;
 241             }
 242         }
 243 #endif
 244     }
 245   /* invalid multibyte character */
 246   *puc = 0xfffd;
 247   return 1;
 248 }
 249
 250 #endif