contrib/grep/lib/unistr/u8-mbtoucr.c

   1 /* Look at first character in UTF-8 string, returning an error code.
   2    Copyright (C) 1999-2002, 2006-2007, 2009-2015 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 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    General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  17
  18 #include <config.h>
  19
  20 /* Specification.  */
  21 #include "unistr.h"
  22
  23 int
  24 u8_mbtoucr (ucs4_t *puc, const uint8_t *s, size_t n)
  25 {
  26   uint8_t c = *s;
  27
  28   if (c < 0x80)
  29     {
  30       *puc = c;
  31       return 1;
  32     }
  33   else if (c >= 0xc2)
  34     {
  35       if (c < 0xe0)
  36         {
  37           if (n >= 2)
  38             {
  39               if ((s[1] ^ 0x80) < 0x40)
  40                 {
  41                   *puc = ((unsigned int) (c & 0x1f) << 6)
  42                          | (unsigned int) (s[1] ^ 0x80);
  43                   return 2;
  44                 }
  45               /* invalid multibyte character */
  46             }
  47           else
  48             {
  49               /* incomplete multibyte character */
  50               *puc = 0xfffd;
  51               return -2;
  52             }
  53         }
  54       else if (c < 0xf0)
  55         {
  56           if (n >= 2)
  57             {
  58               if ((s[1] ^ 0x80) < 0x40
  59                   && (c >= 0xe1 || s[1] >= 0xa0)
  60                   && (c != 0xed || s[1] < 0xa0))
  61                 {
  62                   if (n >= 3)
  63                     {
  64                       if ((s[2] ^ 0x80) < 0x40)
  65                         {
  66                           *puc = ((unsigned int) (c & 0x0f) << 12)
  67                                  | ((unsigned int) (s[1] ^ 0x80) << 6)
  68                                  | (unsigned int) (s[2] ^ 0x80);
  69                           return 3;
  70                         }
  71                       /* invalid multibyte character */
  72                     }
  73                   else
  74                     {
  75                       /* incomplete multibyte character */
  76                       *puc = 0xfffd;
  77                       return -2;
  78                     }
  79                 }
  80               /* invalid multibyte character */
  81             }
  82           else
  83             {
  84               /* incomplete multibyte character */
  85               *puc = 0xfffd;
  86               return -2;
  87             }
  88         }
  89       else if (c < 0xf8)
  90         {
  91           if (n >= 2)
  92             {
  93               if ((s[1] ^ 0x80) < 0x40
  94                   && (c >= 0xf1 || s[1] >= 0x90)
  95 #if 1
  96                   && (c < 0xf4 || (c == 0xf4 && s[1] < 0x90))
  97 #endif
  98                  )
  99                 {
 100                   if (n >= 3)
 101                     {
 102                       if ((s[2] ^ 0x80) < 0x40)
 103                         {
 104                           if (n >= 4)
 105                             {
 106                               if ((s[3] ^ 0x80) < 0x40)
 107                                 {
 108                                   *puc = ((unsigned int) (c & 0x07) << 18)
 109                                          | ((unsigned int) (s[1] ^ 0x80) << 12)
 110                                          | ((unsigned int) (s[2] ^ 0x80) << 6)
 111                                          | (unsigned int) (s[3] ^ 0x80);
 112                                   return 4;
 113                                 }
 114                               /* invalid multibyte character */
 115                             }
 116                           else
 117                             {
 118                               /* incomplete multibyte character */
 119                               *puc = 0xfffd;
 120                               return -2;
 121                             }
 122                         }
 123                       /* invalid multibyte character */
 124                     }
 125                   else
 126                     {
 127                       /* incomplete multibyte character */
 128                       *puc = 0xfffd;
 129                       return -2;
 130                     }
 131                 }
 132               /* invalid multibyte character */
 133             }
 134           else
 135             {
 136               /* incomplete multibyte character */
 137               *puc = 0xfffd;
 138               return -2;
 139             }
 140         }
 141 #if 0
 142       else if (c < 0xfc)
 143         {
 144           if (n >= 2)
 145             {
 146               if ((s[1] ^ 0x80) < 0x40
 147                   && (c >= 0xf9 || s[1] >= 0x88))
 148                 {
 149                   if (n >= 3)
 150                     {
 151                       if ((s[2] ^ 0x80) < 0x40)
 152                         {
 153                           if (n >= 4)
 154                             {
 155                               if ((s[3] ^ 0x80) < 0x40)
 156                                 {
 157                                   if (n >= 5)
 158                                     {
 159                                       if ((s[4] ^ 0x80) < 0x40)
 160                                         {
 161                                           *puc = ((unsigned int) (c & 0x03) << 24)
 162                                                  | ((unsigned int) (s[1] ^ 0x80) << 18)
 163                                                  | ((unsigned int) (s[2] ^ 0x80) << 12)
 164                                                  | ((unsigned int) (s[3] ^ 0x80) << 6)
 165                                                  | (unsigned int) (s[4] ^ 0x80);
 166                                           return 5;
 167                                         }
 168                                       /* invalid multibyte character */
 169                                     }
 170                                   else
 171                                     {
 172                                       /* incomplete multibyte character */
 173                                       *puc = 0xfffd;
 174                                       return -2;
 175                                     }
 176                                 }
 177                               /* invalid multibyte character */
 178                             }
 179                           else
 180                             {
 181                               /* incomplete multibyte character */
 182                               *puc = 0xfffd;
 183                               return -2;
 184                             }
 185                         }
 186                       /* invalid multibyte character */
 187                     }
 188                   else
 189                     {
 190                       /* incomplete multibyte character */
 191                       *puc = 0xfffd;
 192                       return -2;
 193                     }
 194                 }
 195               /* invalid multibyte character */
 196             }
 197           else
 198             {
 199               /* incomplete multibyte character */
 200               *puc = 0xfffd;
 201               return -2;
 202             }
 203         }
 204       else if (c < 0xfe)
 205         {
 206           if (n >= 2)
 207             {
 208               if ((s[1] ^ 0x80) < 0x40
 209                   && (c >= 0xfd || s[1] >= 0x84))
 210                 {
 211                   if (n >= 3)
 212                     {
 213                       if ((s[2] ^ 0x80) < 0x40)
 214                         {
 215                           if (n >= 4)
 216                             {
 217                               if ((s[3] ^ 0x80) < 0x40)
 218                                 {
 219                                   if (n >= 5)
 220                                     {
 221                                       if ((s[4] ^ 0x80) < 0x40)
 222                                         {
 223                                           if (n >= 6)
 224                                             {
 225                                               if ((s[5] ^ 0x80) < 0x40)
 226                                                 {
 227                                                   *puc = ((unsigned int) (c & 0x01) << 30)
 228                                                          | ((unsigned int) (s[1] ^ 0x80) << 24)
 229                                                          | ((unsigned int) (s[2] ^ 0x80) << 18)
 230                                                          | ((unsigned int) (s[3] ^ 0x80) << 12)
 231                                                          | ((unsigned int) (s[4] ^ 0x80) << 6)
 232                                                          | (unsigned int) (s[5] ^ 0x80);
 233                                                   return 6;
 234                                                 }
 235                                               /* invalid multibyte character */
 236                                             }
 237                                           else
 238                                             {
 239                                               /* incomplete multibyte character */
 240                                               *puc = 0xfffd;
 241                                               return -2;
 242                                             }
 243                                         }
 244                                       /* invalid multibyte character */
 245                                     }
 246                                   else
 247                                     {
 248                                       /* incomplete multibyte character */
 249                                       *puc = 0xfffd;
 250                                       return -2;
 251                                     }
 252                                 }
 253                               /* invalid multibyte character */
 254                             }
 255                           else
 256                             {
 257                               /* incomplete multibyte character */
 258                               *puc = 0xfffd;
 259                               return -2;
 260                             }
 261                         }
 262                       /* invalid multibyte character */
 263                     }
 264                   else
 265                     {
 266                       /* incomplete multibyte character */
 267                       *puc = 0xfffd;
 268                       return -2;
 269                     }
 270                 }
 271               /* invalid multibyte character */
 272             }
 273           else
 274             {
 275               /* incomplete multibyte character */
 276               *puc = 0xfffd;
 277               return -2;
 278             }
 279         }
 280 #endif
 281     }
 282   /* invalid multibyte character */
 283   *puc = 0xfffd;
 284   return -1;
 285 }