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         NULL, NULL
  51 };
  52
  53 struct charset_functions charset_utf8_mac =
  54 {
  55         "UTF8-MAC",
  56         0x08000103,
  57         utf8_pull,
  58         utf8_push,
  59         CHARSET_VOLUME | CHARSET_CLIENT | CHARSET_MULTIBYTE | CHARSET_DECOMPOSED,
  60         NULL, NULL
  61 };
  62
  63 /* ------------------------ */
  64 static size_t utf8_pull(void *cd _U_, char **inbuf, size_t *inbytesleft,
  65                          char **outbuf, size_t *outbytesleft)
  66 {
  67         ucs2_t uc = 0;
  68         int len;
  69
  70         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
  71                 unsigned char *c = (unsigned char *)*inbuf;
  72                 len = 1;
  73
  74                 if ((c[0] & 0x80) == 0) {
  75                         uc = c[0];
  76                 } else if ((c[0] & 0xf0) == 0xe0) {
  77                         if (*inbytesleft < 3) {
  78                                 LOG(log_debug, logtype_default, "short utf8 char");
  79                                 goto badseq;
  80                         }
  81                         uc = ((ucs2_t) (c[0] & 0x0f) << 12) | ((ucs2_t) (c[1] ^ 0x80) << 6) | (ucs2_t) (c[2] ^ 0x80);
  82                         len = 3;
  83                 } else if ((c[0] & 0xe0) == 0xc0) {
  84                         if (*inbytesleft < 2) {
  85                                 LOG(log_debug, logtype_default, "short utf8 char");
  86                                 goto badseq;
  87                         }
  88                         uc = ((ucs2_t) (c[0] & 0x1f) << 6) | (ucs2_t) (c[1] ^ 0x80);
  89                         len = 2;
  90                 }
  91                 else {
  92                         errno = EINVAL;
  93                         return -1;
  94                 }
  95
  96                 SSVAL(*outbuf,0,uc);
  97                 (*inbuf)  += len;
  98                 (*inbytesleft)  -= len;
  99                 (*outbytesleft) -= 2;
 100                 (*outbuf) += 2;
 101         }
 102
 103         if (*inbytesleft > 0) {
 104                 errno = E2BIG;
 105                 return -1;
 106         }
 107
 108         return 0;
 109
 110 badseq:
 111         errno = EINVAL;
 112         return -1;
 113 }
 114
 115 /* ------------------------ */
 116 static size_t utf8_push(void *cd _U_, char **inbuf, size_t *inbytesleft,
 117                          char **outbuf, size_t *outbytesleft)
 118 {
 119         ucs2_t uc=0;
 120         int len;
 121
 122         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 123                 unsigned char *c = (unsigned char *)*outbuf;
 124                 uc = SVAL((*inbuf),0);
 125                 len=1;
 126
 127                 if ( uc >= 0x800 ) {
 128                         if ( uc >= 0x202a && uc <= 0x202e ) {
 129                                 /* ignore bidi hint characters */
 130                            len = 0;
 131                         }
 132                         else {
 133                                 if (*outbytesleft < 3) {
 134                                         LOG(log_debug, logtype_default, "short utf8 write");
 135                                         goto toobig;
 136                                 }
 137                                 c[2] = 0x80 | (uc & 0x3f);
 138                                 uc = uc >> 6;
 139                                 uc |= 0x800;
 140                                 c[1] = 0x80 | (uc&0x3f);
 141                                 uc = uc >> 6;
 142                                 uc |= 0xc0;
 143                                 c[0] = uc;
 144                                 len = 3;
 145                         }
 146                 } else if (uc >= 0x80) {
 147                         if (*outbytesleft < 2) {
 148                                 LOG(log_debug, logtype_default, "short utf8 write");
 149                                 goto toobig;
 150                         }
 151                         c[1] = 0x80 | (uc&0x3f);
 152                         uc = uc >> 6;
 153                         uc |= 0xc0;
 154                         c[0] = uc;
 155                         len = 2;
 156                 } else {
 157                         c[0] = uc;
 158                 }
 159
 160                 (*inbytesleft)  -= 2;
 161                 (*outbytesleft) -= len;
 162                 (*inbuf)  += 2;
 163                 (*outbuf) += len;
 164         }
 165
 166         if (*inbytesleft == 1) {
 167                 errno = EINVAL;
 168                 return -1;
 169         }
 170
 171         if (*inbytesleft > 1) {
 172                 errno = E2BIG;
 173                 return -1;
 174         }
 175
 176         return 0;
 177
 178 toobig:
 179         errno = E2BIG;
 180         return -1;
 181 }