source/lib/iconv.c

   1 /*
   2    Unix SMB/Netbios implementation.
   3    Version 3.0
   4    minimal iconv implementation
   5    Copyright (C) Andrew Tridgell 2001
   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
  22 #include "includes.h"
  23
  24 static size_t ascii_pull(char **, size_t *, char **, size_t *);
  25 static size_t ascii_push(char **, size_t *, char **, size_t *);
  26 static size_t weird_pull(char **, size_t *, char **, size_t *);
  27 static size_t weird_push(char **, size_t *, char **, size_t *);
  28 static size_t iconv_copy(char **, size_t *, char **, size_t *);
  29
  30 /*
  31   for each charset we have a function that pulls from that charset to
  32   a ucs2 buffer, and a function that pushes to a ucs2 buffer
  33 */
  34 static struct {
  35         char *name;
  36         size_t (*pull)(char **inbuf, size_t *inbytesleft,
  37                        char **outbuf, size_t *outbytesleft);
  38         size_t (*push)(char **inbuf, size_t *inbytesleft,
  39                        char **outbuf, size_t *outbytesleft);
  40 } charsets[] = {
  41         {"UCS2", iconv_copy, iconv_copy},
  42         {"ASCII", ascii_pull, ascii_push},
  43         {"WEIRD", weird_pull, weird_push},
  44         {NULL, NULL, NULL}
  45 };
  46
  47 /*
  48   this is a simple portable iconv() implementaion. It only knows about
  49   a very small number of character sets - just enough that Samba works
  50   on systems that don't have iconv
  51  */
  52 size_t smb_iconv(smb_iconv_t cd,
  53                  char **inbuf, size_t *inbytesleft,
  54                  char **outbuf, size_t *outbytesleft)
  55 {
  56         char cvtbuf[2048];
  57         char *bufp = cvtbuf;
  58         size_t bufsize;
  59
  60 #ifdef HAVE_NATIVE_ICONV
  61         if (cd->cd) {
  62                 return iconv(cd->cd, inbuf, inbytesleft, outbuf, outbytesleft);
  63         }
  64 #endif
  65
  66         if (!inbuf || ! *inbuf || !outbuf || ! *outbuf) return 0;
  67
  68         /* in most cases we can go direct */
  69         if (cd->direct) {
  70                 return cd->direct(inbuf, inbytesleft, outbuf, outbytesleft);
  71         }
  72
  73         /* otherwise we have to do it chunks at a time */
  74         while (*inbytesleft > 0) {
  75                 bufp = cvtbuf;
  76                 bufsize = sizeof(cvtbuf);
  77                 if (cd->pull(inbuf, inbytesleft, &bufp, &bufsize) == -1 &&
  78                     errno != E2BIG) return -1;
  79
  80                 bufp = cvtbuf;
  81                 bufsize = sizeof(cvtbuf) - bufsize;
  82                 if (cd->push(&bufp, &bufsize, outbuf, outbytesleft) == -1) return -1;
  83         }
  84
  85         return 0;
  86 }
  87
  88 /*
  89   simple iconv_open() wrapper
  90  */
  91 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
  92 {
  93         smb_iconv_t ret;
  94         int from, to;
  95 #ifdef HAVE_NATIVE_ICONV
  96         iconv_t cd = NULL;
  97 #endif
  98
  99         for (from=0; charsets[from].name; from++) {
 100                 if (strcasecmp(charsets[from].name, fromcode) == 0) break;
 101         }
 102         for (to=0; charsets[to].name; to++) {
 103                 if (strcasecmp(charsets[to].name, tocode) == 0) break;
 104         }
 105
 106         if (!charsets[from].name || !charsets[to].name) {
 107 #ifdef HAVE_NATIVE_ICONV
 108                 cd = iconv_open(tocode, fromcode);
 109                 if (!cd)
 110 #endif
 111                 {
 112                         errno = EINVAL;
 113                         return (smb_iconv_t)-1;
 114                 }
 115         }
 116
 117         ret = (smb_iconv_t)malloc(sizeof(*ret));
 118         if (!ret) {
 119                 errno = ENOMEM;
 120                 return (smb_iconv_t)-1;
 121         }
 122         memset(ret, 0, sizeof(*ret));
 123
 124 #ifdef HAVE_NATIVE_ICONV
 125         /* see if we wil be using the native iconv */
 126         if (cd) {
 127                 ret->cd = cd;
 128                 return ret;
 129         }
 130 #endif
 131
 132         /* check for the simplest null conversion */
 133         if (from == to) {
 134                 ret->direct = iconv_copy;
 135                 return ret;
 136         }
 137
 138         /* check for conversion to/from ucs2 */
 139         if (from == 0) {
 140                 ret->direct = charsets[to].push;
 141                 return ret;
 142         }
 143         if (to == 0) {
 144                 ret->direct = charsets[from].pull;
 145                 return ret;
 146         }
 147
 148         /* the general case has to go via a buffer */
 149         ret->pull = charsets[from].pull;
 150         ret->push = charsets[to].push;
 151         return ret;
 152 }
 153
 154 /*
 155   simple iconv_close() wrapper
 156 */
 157 int smb_iconv_close (smb_iconv_t cd)
 158 {
 159 #ifdef HAVE_NATIVE_ICONV
 160         if (cd->cd) {
 161                 iconv_close(cd->cd);
 162         }
 163 #endif
 164         memset(cd, 0, sizeof(*cd));
 165         free(cd);
 166         return 0;
 167 }
 168
 169
 170 /**********************************************************************
 171  the following functions implement the builtin character sets in Samba
 172  and also the "test" character sets that are designed to test
 173  multi-byte character set support for english users
 174 ***********************************************************************/
 175
 176 static size_t ascii_pull(char **inbuf, size_t *inbytesleft,
 177                          char **outbuf, size_t *outbytesleft)
 178 {
 179         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 180                 (*outbuf)[0] = (*inbuf)[0];
 181                 (*outbuf)[1] = 0;
 182                 (*inbytesleft)  -= 1;
 183                 (*outbytesleft) -= 2;
 184                 (*inbuf)  += 1;
 185                 (*outbuf) += 2;
 186         }
 187
 188         if (*inbytesleft > 0) {
 189                 errno = E2BIG;
 190                 return -1;
 191         }
 192
 193         return 0;
 194 }
 195
 196 static size_t ascii_push(char **inbuf, size_t *inbytesleft,
 197                          char **outbuf, size_t *outbytesleft)
 198 {
 199         int ir_count=0;
 200
 201         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 202                 (*outbuf)[0] = (*inbuf)[0];
 203                 if ((*inbuf)[1]) ir_count++;
 204                 (*inbytesleft)  -= 2;
 205                 (*outbytesleft) -= 1;
 206                 (*inbuf)  += 2;
 207                 (*outbuf) += 1;
 208         }
 209
 210         if (*inbytesleft == 1) {
 211                 errno = EINVAL;
 212                 return -1;
 213         }
 214
 215         if (*inbytesleft > 1) {
 216                 errno = E2BIG;
 217                 return -1;
 218         }
 219
 220         return ir_count;
 221 }
 222
 223
 224 /* the "weird" character set is very useful for testing multi-byte
 225    support and finding bugs. Don't use on a production system!
 226 */
 227 static struct {
 228         char from;
 229         char *to;
 230         int len;
 231 } weird_table[] = {
 232         {'q', "^q^", 3},
 233         {'Q', "^Q^", 3},
 234         {0, NULL}
 235 };
 236
 237 static size_t weird_pull(char **inbuf, size_t *inbytesleft,
 238                          char **outbuf, size_t *outbytesleft)
 239 {
 240         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 241                 int i;
 242                 int done = 0;
 243                 for (i=0;weird_table[i].from;i++) {
 244                         if (strncmp((*inbuf),
 245                                     weird_table[i].to,
 246                                     weird_table[i].len) == 0) {
 247                                 if (*inbytesleft < weird_table[i].len) {
 248                                         DEBUG(0,("ERROR: truncated weird string\n"));
 249                                         /* smb_panic("weird_pull"); */
 250
 251                                 } else {
 252                                         (*outbuf)[0] = weird_table[i].from;
 253                                         (*outbuf)[1] = 0;
 254                                         (*inbytesleft)  -= weird_table[i].len;
 255                                         (*outbytesleft) -= 2;
 256                                         (*inbuf)  += weird_table[i].len;
 257                                         (*outbuf) += 2;
 258                                         done = 1;
 259                                         break;
 260                                 }
 261                         }
 262                 }
 263                 if (done) continue;
 264                 (*outbuf)[0] = (*inbuf)[0];
 265                 (*outbuf)[1] = 0;
 266                 (*inbytesleft)  -= 1;
 267                 (*outbytesleft) -= 2;
 268                 (*inbuf)  += 1;
 269                 (*outbuf) += 2;
 270         }
 271
 272         if (*inbytesleft > 0) {
 273                 errno = E2BIG;
 274                 return -1;
 275         }
 276
 277         return 0;
 278 }
 279
 280 static size_t weird_push(char **inbuf, size_t *inbytesleft,
 281                          char **outbuf, size_t *outbytesleft)
 282 {
 283         int ir_count=0;
 284
 285         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 286                 int i;
 287                 int done=0;
 288                 for (i=0;weird_table[i].from;i++) {
 289                         if ((*inbuf)[0] == weird_table[i].from &&
 290                             (*inbuf)[1] == 0) {
 291                                 if (*outbytesleft < weird_table[i].len) {
 292                                         DEBUG(0,("No room for weird character\n"));
 293                                         /* smb_panic("weird_push"); */
 294                                 } else {
 295                                         memcpy(*outbuf, weird_table[i].to,
 296                                                weird_table[i].len);
 297                                         (*inbytesleft)  -= 2;
 298                                         (*outbytesleft) -= weird_table[i].len;
 299                                         (*inbuf)  += 2;
 300                                         (*outbuf) += weird_table[i].len;
 301                                         done = 1;
 302                                         break;
 303                                 }
 304                         }
 305                 }
 306                 if (done) continue;
 307
 308                 (*outbuf)[0] = (*inbuf)[0];
 309                 if ((*inbuf)[1]) ir_count++;
 310                 (*inbytesleft)  -= 2;
 311                 (*outbytesleft) -= 1;
 312                 (*inbuf)  += 2;
 313                 (*outbuf) += 1;
 314         }
 315
 316         if (*inbytesleft == 1) {
 317                 errno = EINVAL;
 318                 return -1;
 319         }
 320
 321         if (*inbytesleft > 1) {
 322                 errno = E2BIG;
 323                 return -1;
 324         }
 325
 326         return ir_count;
 327 }
 328
 329 static size_t iconv_copy(char **inbuf, size_t *inbytesleft,
 330                          char **outbuf, size_t *outbytesleft)
 331 {
 332         int n;
 333
 334         n = MIN(*inbytesleft, *outbytesleft);
 335
 336         memmove(*outbuf, *inbuf, n);
 337
 338         (*inbytesleft) -= n;
 339         (*outbytesleft) -= n;
 340         (*inbuf) += n;
 341         (*outbuf) += n;
 342
 343         if (*inbytesleft > 0) {
 344                 errno = E2BIG;
 345                 return -1;
 346         }
 347
 348         return 0;
 349 }