libatalk/unicode/utf8.c

   1 /*
   2    Unix SMB/CIFS implementation.
   3    minimal iconv implementation
   4    Copyright (C) Andrew Tridgell 2001
   5    Copyright (C) Jelmer Vernooij 2002,2003
   6
   7    This program is free software; you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation; either version 2 of the License, or
  10    (at your option) any later version.
  11
  12    This program is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License
  18    along with this program; if not, write to the Free Software
  19    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  20
  21    From samba 3.0 beta and GNU libiconv-1.8
  22    It's bad but most of the time we can't use libc iconv service:
  23    - it doesn't round trip for most encoding
  24    - it doesn't know about Apple extension
  25 */
  26
  27 #ifdef HAVE_CONFIG_H
  28 #include "config.h"
  29 #endif /* HAVE_CONFIG_H */
  30 #include <stdlib.h>
  31 #include <errno.h>
  32
  33 #include <netatalk/endian.h>
  34 #include <atalk/unicode.h>
  35 #include <atalk/logger.h>
  36 #include <atalk/unicode.h>
  37 #include "byteorder.h"
  38
  39
  40 static size_t   utf8_pull(void *,char **, size_t *, char **, size_t *);
  41 static size_t   utf8_push(void *,char **, size_t *, char **, size_t *);
  42
  43 struct charset_functions charset_utf8 =
  44 {
  45         "UTF8",
  46         0x08000103,
  47         utf8_pull,
  48         utf8_push,
  49         CHARSET_VOLUME | CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED
  50 };
  51
  52 struct charset_functions charset_utf8_mac =
  53 {
  54         "UTF8-MAC",
  55         0x08000103,
  56         utf8_pull,
  57         utf8_push,
  58         CHARSET_VOLUME | CHARSET_CLIENT | CHARSET_MULTIBYTE | CHARSET_DECOMPOSED
  59 };
  60
  61 /* ------------------------ */
  62 static size_t utf8_pull(void *cd, char **inbuf, size_t *inbytesleft,
  63                          char **outbuf, size_t *outbytesleft)
  64 {
  65         ucs2_t uc = 0;
  66         int len;
  67
  68         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
  69                 unsigned char *c = (unsigned char *)*inbuf;
  70                 len = 1;
  71
  72                 if ((c[0] & 0x80) == 0) {
  73                         uc = c[0];
  74                 } else if ((c[0] & 0xf0) == 0xe0) {
  75                         if (*inbytesleft < 3) {
  76                                 LOG(log_debug, logtype_default, "short utf8 char");
  77                                 goto badseq;
  78                         }
  79                         uc = ((ucs2_t) (c[0] & 0x0f) << 12) | ((ucs2_t) (c[1] ^ 0x80) << 6) | (ucs2_t) (c[2] ^ 0x80);
  80                         len = 3;
  81                 } else if ((c[0] & 0xe0) == 0xc0) {
  82                         if (*inbytesleft < 2) {
  83                                 LOG(log_debug, logtype_default, "short utf8 char");
  84                                 goto badseq;
  85                         }
  86                         uc = ((ucs2_t) (c[0] & 0x1f) << 6) | (ucs2_t) (c[1] ^ 0x80);
  87                         len = 2;
  88                 }
  89                 else {
  90                         errno = EINVAL;
  91                         return -1;
  92                 }
  93
  94                 SSVAL(*outbuf,0,uc);
  95                 (*inbuf)  += len;
  96                 (*inbytesleft)  -= len;
  97                 (*outbytesleft) -= 2;
  98                 (*outbuf) += 2;
  99         }
 100
 101         if (*inbytesleft > 0) {
 102                 errno = E2BIG;
 103                 return -1;
 104         }
 105
 106         return 0;
 107
 108 badseq:
 109         errno = EINVAL;
 110         return -1;
 111 }
 112
 113 /* ------------------------ */
 114 static size_t utf8_push(void *cd, char **inbuf, size_t *inbytesleft,
 115                          char **outbuf, size_t *outbytesleft)
 116 {
 117         ucs2_t uc=0;
 118         int len;
 119
 120         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 121                 unsigned char *c = (unsigned char *)*outbuf;
 122                 uc = SVAL((*inbuf),0);
 123                 len=1;
 124
 125                 if ( uc >= 0x800 ) {
 126                         if (*outbytesleft < 3) {
 127                                 LOG(log_debug, logtype_default, "short utf8 write");
 128                                 goto toobig;
 129                         }
 130                         c[2] = 0x80 | (uc & 0x3f);
 131                         uc = uc >> 6;
 132                         uc |= 0x800;
 133                         c[1] = 0x80 | (uc&0x3f);
 134                         uc = uc >> 6;
 135                         uc |= 0xc0;
 136                         c[0] = uc;
 137                         len = 3;
 138                 } else if (uc >= 0x80) {
 139                         if (*outbytesleft < 2) {
 140                                 LOG(log_debug, logtype_default, "short utf8 write");
 141                                 goto toobig;
 142                         }
 143                         c[1] = 0x80 | (uc&0x3f);
 144                         uc = uc >> 6;
 145                         uc |= 0xc0;
 146                         c[0] = uc;
 147                         len = 2;
 148                 } else {
 149                         c[0] = uc;
 150                 }
 151
 152
 153                 (*inbytesleft)  -= 2;
 154                 (*outbytesleft) -= len;
 155                 (*inbuf)  += 2;
 156                 (*outbuf) += len;
 157         }
 158
 159         if (*inbytesleft == 1) {
 160                 errno = EINVAL;
 161                 return -1;
 162         }
 163
 164         if (*inbytesleft > 1) {
 165                 errno = E2BIG;
 166                 return -1;
 167         }
 168
 169         return 0;
 170
 171 toobig:
 172         errno = E2BIG;
 173         return -1;
 174 }