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
  32 #include <netatalk/endian.h>
  33 #include <atalk/unicode.h>
  34
  35
  36 #include <atalk/logger.h>
  37 #include <errno.h>
  38
  39 #include <atalk/unicode.h>
  40
  41
  42 static size_t   utf8_pull(void *,char **, size_t *, char **, size_t *);
  43 static size_t   utf8_push(void *,char **, size_t *, char **, size_t *);
  44
  45 struct charset_functions charset_utf8 =
  46 {
  47         "UTF8",
  48         0x08000103,
  49         utf8_pull,
  50         utf8_push,
  51         CHARSET_VOLUME | CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED
  52 };
  53
  54 struct charset_functions charset_utf8_mac =
  55 {
  56         "UTF8-MAC",
  57         0x08000103,
  58         utf8_pull,
  59         utf8_push,
  60         CHARSET_VOLUME | CHARSET_CLIENT | CHARSET_MULTIBYTE | CHARSET_DECOMPOSED
  61 };
  62
  63 /* ------------------------ */
  64 static size_t utf8_pull(void *cd, char **inbuf, size_t *inbytesleft,
  65                          char **outbuf, size_t *outbytesleft)
  66 {
  67         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
  68                 unsigned char *c = (unsigned char *)*inbuf;
  69                 unsigned char *uc = (unsigned char *)*outbuf;
  70                 int len = 1;
  71
  72                 if ((c[0] & 0x80) == 0) {
  73                         uc[0] = c[0];
  74                         uc[1] = 0;
  75                 } else if ((c[0] & 0xf0) == 0xe0) {
  76                         if (*inbytesleft < 3) {
  77                                 LOG(log_debug, logtype_default, "short utf8 char\n");
  78                                 goto badseq;
  79                         }
  80                         uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
  81                         uc[0] = (c[1]<<6) | (c[2]&0x3f);
  82                         len = 3;
  83                 } else if ((c[0] & 0xe0) == 0xc0) {
  84                         if (*inbytesleft < 2) {
  85                                 LOG(log_debug, logtype_default, "short utf8 char\n");
  86                                 goto badseq;
  87                         }
  88                         uc[1] = (c[0]>>2) & 0x7;
  89                         uc[0] = (c[0]<<6) | (c[1]&0x3f);
  90                         len = 2;
  91                 }
  92
  93                 (*inbuf)  += len;
  94                 (*inbytesleft)  -= len;
  95                 (*outbytesleft) -= 2;
  96                 (*outbuf) += 2;
  97         }
  98
  99         if (*inbytesleft > 0) {
 100                 errno = E2BIG;
 101                 return -1;
 102         }
 103
 104         return 0;
 105
 106 badseq:
 107         errno = EINVAL;
 108         return -1;
 109 }
 110
 111 /* ------------------------ */
 112 static size_t utf8_push(void *cd, char **inbuf, size_t *inbytesleft,
 113                          char **outbuf, size_t *outbytesleft)
 114 {
 115         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 116                 unsigned char *c = (unsigned char *)*outbuf;
 117                 unsigned char *uc = (unsigned char *)*inbuf;
 118                 int len=1;
 119
 120                 if (uc[1] & 0xf8) {
 121                         if (*outbytesleft < 3) {
 122                                 LOG(log_debug, logtype_default, "short utf8 write\n");
 123                                 goto toobig;
 124                         }
 125                         c[0] = 0xe0 | (uc[1]>>4);
 126                         c[1] = 0x80 | ((uc[1]&0xF)<<2) | (uc[0]>>6);
 127                         c[2] = 0x80 | (uc[0]&0x3f);
 128                         len = 3;
 129                 } else if (uc[1] | (uc[0] & 0x80)) {
 130                         if (*outbytesleft < 2) {
 131                                 LOG(log_debug, logtype_default, "short utf8 write\n");
 132                                 goto toobig;
 133                         }
 134                         c[0] = 0xc0 | (uc[1]<<2) | (uc[0]>>6);
 135                         c[1] = 0x80 | (uc[0]&0x3f);
 136                         len = 2;
 137                 } else {
 138                         c[0] = uc[0];
 139                 }
 140
 141
 142                 (*inbytesleft)  -= 2;
 143                 (*outbytesleft) -= len;
 144                 (*inbuf)  += 2;
 145                 (*outbuf) += len;
 146         }
 147
 148         if (*inbytesleft == 1) {
 149                 errno = EINVAL;
 150                 return -1;
 151         }
 152
 153         if (*inbytesleft > 1) {
 154                 errno = E2BIG;
 155                 return -1;
 156         }
 157
 158         return 0;
 159
 160 toobig:
 161         errno = E2BIG;
 162         return -1;
 163 }