libatalk/unicode/iconv.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
  31 #include <stdio.h>
  32 #include <stdlib.h>
  33 #include <unistd.h>
  34 #include <string.h>
  35 #include <ctype.h>
  36 #include <errno.h>
  37 #include <sys/param.h>
  38 #include <sys/stat.h>
  39 #ifdef HAVE_USABLE_ICONV
  40 #include <iconv.h>
  41 #endif
  42
  43 #include <netatalk/endian.h>
  44 #include <atalk/unicode.h>
  45 #include <atalk/logger.h>
  46 #include "byteorder.h"
  47
  48
  49 /**
  50  * @file
  51  *
  52  * @brief Samba wrapper/stub for iconv character set conversion.
  53  *
  54  * iconv is the XPG2 interface for converting between character
  55  * encodings.  This file provides a Samba wrapper around it, and also
  56  * a simple reimplementation that is used if the system does not
  57  * implement iconv.
  58  *
  59  * Samba only works with encodings that are supersets of ASCII: ascii
  60  * characters like whitespace can be tested for directly, multibyte
  61  * sequences start with a byte with the high bit set, and strings are
  62  * terminated by a nul byte.
  63  *
  64  * Note that the only function provided by iconv is conversion between
  65  * characters.  It doesn't directly support operations like
  66  * uppercasing or comparison.  We have to convert to UCS-2 and compare
  67  * there.
  68  *
  69  * @sa Samba Developers Guide
  70  **/
  71 #define CHARSET_WIDECHAR    32
  72
  73 #ifdef HAVE_USABLE_ICONV
  74 #ifdef HAVE_UCS2INTERNAL
  75 #define UCS2ICONV "UCS-2-INTERNAL"
  76 #else /* !HAVE_UCS2INTERNAL */
  77 #if BYTE_ORDER==LITTLE_ENDIAN
  78 #define UCS2ICONV "UCS-2LE"
  79 #else /* !LITTLE_ENDIAN */
  80 #define UCS2ICONV "UCS-2BE"
  81 #endif /* BYTE_ORDER */
  82 #endif /* HAVE_UCS2INTERNAL */
  83 #else /* !HAVE_USABLE_ICONV */
  84 #define UCS2ICONV "UCS-2"
  85 #endif /* HAVE_USABLE_ICONV */
  86
  87 static size_t ascii_pull(void *,char **, size_t *, char **, size_t *);
  88 static size_t ascii_push(void *,char **, size_t *, char **, size_t *);
  89 static size_t iconv_copy(void *,char **, size_t *, char **, size_t *);
  90
  91 extern  struct charset_functions charset_mac_roman;
  92 extern  struct charset_functions charset_mac_hebrew;
  93 extern  struct charset_functions charset_mac_centraleurope;
  94 extern  struct charset_functions charset_mac_cyrillic;
  95 extern  struct charset_functions charset_mac_greek;
  96 extern  struct charset_functions charset_mac_turkish;
  97 extern  struct charset_functions charset_utf8;
  98 extern  struct charset_functions charset_utf8_mac;
  99
 100
 101 static struct charset_functions builtin_functions[] = {
 102         {"UCS-2",   0, iconv_copy, iconv_copy, CHARSET_WIDECHAR | CHARSET_PRECOMPOSED, NULL, NULL},
 103         {"ASCII",     0, ascii_pull, ascii_push, CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED, NULL, NULL},
 104         {"SHIFT_JIS", 1568, NULL, NULL, CHARSET_ICONV | CHARSET_PRECOMPOSED | CHARSET_CLIENT, NULL, NULL},
 105         {NULL, 0, NULL, NULL, 0, NULL, NULL}
 106 };
 107
 108
 109 #define DLIST_ADD(list, p) \
 110 { \
 111         if (!(list)) { \
 112                 (list) = (p); \
 113                 (p)->next = (p)->prev = NULL; \
 114         } else { \
 115                 (list)->prev = (p); \
 116                 (p)->next = (list); \
 117                 (p)->prev = NULL; \
 118                 (list) = (p); \
 119         }\
 120 }
 121
 122 static struct charset_functions *charsets = NULL;
 123
 124 struct charset_functions *find_charset_functions(const char *name)
 125 {
 126         struct charset_functions *c = charsets;
 127
 128         while(c) {
 129                 if (strcasecmp(name, c->name) == 0) {
 130                         return c;
 131                 }
 132                 c = c->next;
 133         }
 134
 135         return NULL;
 136 }
 137
 138 int atalk_register_charset(struct charset_functions *funcs)
 139 {
 140         if (!funcs) {
 141                 return -1;
 142         }
 143
 144         /* Check whether we already have this charset... */
 145         if (find_charset_functions(funcs->name)) {
 146                 LOG (log_debug, logtype_default, "Duplicate charset %s, not registering", funcs->name);
 147                 return -2;
 148         }
 149
 150         funcs->next = funcs->prev = NULL;
 151         DLIST_ADD(charsets, funcs);
 152         return 0;
 153 }
 154
 155 void lazy_initialize_iconv(void)
 156 {
 157         static int initialized = 0;
 158         int i;
 159
 160         if (!initialized) {
 161                 initialized = 1;
 162                 for(i = 0; builtin_functions[i].name; i++)
 163                         atalk_register_charset(&builtin_functions[i]);
 164
 165                 /* register additional charsets */
 166                 atalk_register_charset(&charset_utf8);
 167                 atalk_register_charset(&charset_utf8_mac);
 168                 atalk_register_charset(&charset_mac_roman);
 169                 atalk_register_charset(&charset_mac_hebrew);
 170                 atalk_register_charset(&charset_mac_greek);
 171                 atalk_register_charset(&charset_mac_turkish);
 172                 atalk_register_charset(&charset_mac_centraleurope);
 173                 atalk_register_charset(&charset_mac_cyrillic);
 174         }
 175 }
 176
 177 /* if there was an error then reset the internal state,
 178    this ensures that we don't have a shift state remaining for
 179    character sets like SJIS */
 180 static size_t sys_iconv(void *cd,
 181                         char **inbuf, size_t *inbytesleft,
 182                         char **outbuf, size_t *outbytesleft)
 183 {
 184 #ifdef HAVE_USABLE_ICONV
 185         size_t ret = iconv((iconv_t)cd,
 186                            (ICONV_CONST char**)inbuf, inbytesleft,
 187                            outbuf, outbytesleft);
 188         if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
 189         return ret;
 190 #else
 191         errno = EINVAL;
 192         return -1;
 193 #endif
 194 }
 195
 196 /**
 197  * This is a simple portable iconv() implementaion.
 198  *
 199  * It only knows about a very small number of character sets - just
 200  * enough that netatalk works on systems that don't have iconv.
 201  **/
 202 size_t atalk_iconv(atalk_iconv_t cd,
 203                  const char **inbuf, size_t *inbytesleft,
 204                  char **outbuf, size_t *outbytesleft)
 205 {
 206         char cvtbuf[2048];
 207         char *bufp = cvtbuf;
 208         size_t bufsize;
 209
 210         /* in many cases we can go direct */
 211         if (cd->direct) {
 212                 return cd->direct(cd->cd_direct,
 213                                   (char **)inbuf, inbytesleft, outbuf, outbytesleft);
 214         }
 215
 216
 217         /* otherwise we have to do it chunks at a time */
 218         while (*inbytesleft > 0) {
 219                 bufp = cvtbuf;
 220                 bufsize = sizeof(cvtbuf);
 221
 222                 if (cd->pull(cd->cd_pull, (char **)inbuf, inbytesleft, &bufp, &bufsize) == (size_t)-1
 223                        && errno != E2BIG) {
 224                     return -1;
 225                 }
 226
 227                 bufp = cvtbuf;
 228                 bufsize = sizeof(cvtbuf) - bufsize;
 229
 230                 if (cd->push(cd->cd_push, &bufp, &bufsize, outbuf, outbytesleft) == (size_t)-1) {
 231                     return -1;
 232                 }
 233         }
 234
 235         return 0;
 236 }
 237
 238
 239 /*
 240   simple iconv_open() wrapper
 241  */
 242 atalk_iconv_t atalk_iconv_open(const char *tocode, const char *fromcode)
 243 {
 244         atalk_iconv_t ret;
 245         struct charset_functions *from, *to;
 246
 247
 248         lazy_initialize_iconv();
 249         from = charsets;
 250         to = charsets;
 251
 252         ret = (atalk_iconv_t)malloc(sizeof(*ret));
 253         if (!ret) {
 254                 errno = ENOMEM;
 255                 return (atalk_iconv_t)-1;
 256         }
 257         memset(ret, 0, sizeof(*ret));
 258
 259         ret->from_name = strdup(fromcode);
 260         ret->to_name = strdup(tocode);
 261
 262         /* check for the simplest null conversion */
 263         if (strcasecmp(fromcode, tocode) == 0) {
 264                 ret->direct = iconv_copy;
 265                 return ret;
 266         }
 267
 268         /* check if we have a builtin function for this conversion */
 269         from = find_charset_functions(fromcode);
 270         if(from && from->pull)ret->pull = from->pull;
 271
 272         to = find_charset_functions(tocode);
 273         if(to && to->push)ret->push = to->push;
 274
 275         /* check if we can use iconv for this conversion */
 276 #ifdef HAVE_USABLE_ICONV
 277         if (!ret->pull) {
 278                 ret->cd_pull = iconv_open(UCS2ICONV, fromcode);
 279                 if (ret->cd_pull != (iconv_t)-1)
 280                         ret->pull = sys_iconv;
 281         }
 282
 283         if (!ret->push) {
 284                 ret->cd_push = iconv_open(tocode, UCS2ICONV);
 285                 if (ret->cd_push != (iconv_t)-1)
 286                         ret->push = sys_iconv;
 287         }
 288 #endif
 289
 290         if (!ret->push || !ret->pull) {
 291                 SAFE_FREE(ret->from_name);
 292                 SAFE_FREE(ret->to_name);
 293                 SAFE_FREE(ret);
 294                 errno = EINVAL;
 295                 return (atalk_iconv_t)-1;
 296         }
 297
 298         /* check for conversion to/from ucs2 */
 299         if (strcasecmp(fromcode, "UCS-2") == 0 && to && to->push) {
 300                 ret->direct = to->push;
 301                 ret->push = ret->pull = NULL;
 302                 return ret;
 303         }
 304
 305         if (strcasecmp(tocode, "UCS-2") == 0 && from && from->pull) {
 306                 ret->direct = from->pull;
 307                 ret->push = ret->pull = NULL;
 308                 return ret;
 309         }
 310
 311         /* Check if we can do the conversion direct */
 312 #ifdef HAVE_USABLE_ICONV
 313         if (strcasecmp(fromcode, "UCS-2") == 0) {
 314                 ret->direct = sys_iconv;
 315                 ret->cd_direct = ret->cd_push;
 316                 ret->cd_push = NULL;
 317                 return ret;
 318         }
 319         if (strcasecmp(tocode, "UCS-2") == 0) {
 320                 ret->direct = sys_iconv;
 321                 ret->cd_direct = ret->cd_pull;
 322                 ret->cd_pull = NULL;
 323                 return ret;
 324         }
 325 #endif
 326
 327         return ret;
 328 }
 329
 330 /*
 331   simple iconv_close() wrapper
 332 */
 333 int atalk_iconv_close (atalk_iconv_t cd)
 334 {
 335 #ifdef HAVE_USABLE_ICONV
 336         if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
 337         if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
 338         if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
 339 #endif
 340
 341         SAFE_FREE(cd->from_name);
 342         SAFE_FREE(cd->to_name);
 343
 344         memset(cd, 0, sizeof(*cd));
 345         SAFE_FREE(cd);
 346         return 0;
 347 }
 348
 349
 350 /************************************************************************
 351  the following functions implement the builtin character sets in Netatalk
 352 *************************************************************************/
 353
 354 static size_t ascii_pull(void *cd _U_, char **inbuf, size_t *inbytesleft,
 355                          char **outbuf, size_t *outbytesleft)
 356 {
 357         ucs2_t curchar;
 358
 359         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 360                 if ((unsigned char)(*inbuf)[0] < 0x80) {
 361                         curchar = (ucs2_t) (*inbuf)[0];
 362                         SSVAL((*outbuf),0,curchar);
 363                 }
 364                 else {
 365                         errno = EILSEQ;
 366                         return -1;
 367                 }
 368                 (*inbytesleft)  -= 1;
 369                 (*outbytesleft) -= 2;
 370                 (*inbuf)  += 1;
 371                 (*outbuf) += 2;
 372         }
 373
 374         if (*inbytesleft > 0) {
 375                 errno = E2BIG;
 376                 return -1;
 377         }
 378
 379         return 0;
 380 }
 381
 382 static size_t ascii_push(void *cd _U_, char **inbuf, size_t *inbytesleft,
 383                          char **outbuf, size_t *outbytesleft)
 384 {
 385         int ir_count=0;
 386         ucs2_t curchar;
 387
 388         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 389                 curchar = SVAL((*inbuf), 0);
 390                 if (curchar < 0x0080) {
 391                         (*outbuf)[0] = curchar;
 392                 }
 393                 else {
 394                         errno = EILSEQ;
 395                         return -1;
 396                 }
 397                 (*inbytesleft)  -= 2;
 398                 (*outbytesleft) -= 1;
 399                 (*inbuf)  += 2;
 400                 (*outbuf) += 1;
 401         }
 402
 403         if (*inbytesleft == 1) {
 404                 errno = EINVAL;
 405                 return -1;
 406         }
 407
 408         if (*inbytesleft > 1) {
 409                 errno = E2BIG;
 410                 return -1;
 411         }
 412
 413         return ir_count;
 414 }
 415
 416
 417 static size_t iconv_copy(void *cd _U_, char **inbuf, size_t *inbytesleft,
 418                          char **outbuf, size_t *outbytesleft)
 419 {
 420         int n;
 421
 422         n = MIN(*inbytesleft, *outbytesleft);
 423
 424         memmove(*outbuf, *inbuf, n);
 425
 426         (*inbytesleft) -= n;
 427         (*outbytesleft) -= n;
 428         (*inbuf) += n;
 429         (*outbuf) += n;
 430
 431         if (*inbytesleft > 0) {
 432                 errno = E2BIG;
 433                 return -1;
 434         }
 435
 436         return 0;
 437 }
 438
 439 /* ------------------------ */