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