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 struct charset_functions charset_ucs2 =
  89 {
  90         "UCS-2",
  91         0,
  92         iconv_copy,
  93         iconv_copy,
  94         CHARSET_WIDECHAR | CHARSET_PRECOMPOSED
  95 };
  96
  97 struct charset_functions charset_ascii =
  98 {
  99         "ASCII",
 100         0,
 101         ascii_pull,
 102         ascii_push,
 103         CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED
 104 };
 105
 106 struct charset_functions charset_iconv =
 107 {
 108         NULL,
 109         0,
 110         NULL,
 111         NULL,
 112         CHARSET_ICONV | CHARSET_PRECOMPOSED
 113 };
 114
 115
 116 extern  struct charset_functions charset_mac_roman;
 117 extern  struct charset_functions charset_mac_hebrew;
 118 extern  struct charset_functions charset_mac_centraleurope;
 119 extern  struct charset_functions charset_mac_cyrillic;
 120 extern  struct charset_functions charset_mac_turkish;
 121 extern  struct charset_functions charset_utf8;
 122 extern  struct charset_functions charset_utf8_mac;
 123
 124
 125 static struct charset_functions builtin_functions[] = {
 126         {"UCS-2",   0, iconv_copy, iconv_copy, CHARSET_WIDECHAR},
 127         {"ASCII",     0, ascii_pull, ascii_push, CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED},
 128         {NULL, 0, NULL, NULL, 0}
 129 };
 130
 131
 132 #define DLIST_ADD(list, p) \
 133 { \
 134         if (!(list)) { \
 135                 (list) = (p); \
 136                 (p)->next = (p)->prev = NULL; \
 137         } else { \
 138                 (list)->prev = (p); \
 139                 (p)->next = (list); \
 140                 (p)->prev = NULL; \
 141                 (list) = (p); \
 142         }\
 143 }
 144
 145 static struct charset_functions *charsets = NULL;
 146
 147 struct charset_functions *find_charset_functions(const char *name)
 148 {
 149         struct charset_functions *c = charsets;
 150
 151         while(c) {
 152                 if (strcasecmp(name, c->name) == 0) {
 153                         return c;
 154                 }
 155                 c = c->next;
 156         }
 157
 158         return NULL;
 159 }
 160
 161 int atalk_register_charset(struct charset_functions *funcs)
 162 {
 163         if (!funcs) {
 164                 return -1;
 165         }
 166
 167         /* Check whether we already have this charset... */
 168         if (find_charset_functions(funcs->name)) {
 169                 LOG (log_debug, logtype_default, "Duplicate charset %s, not registering", funcs->name);
 170                 return -2;
 171         }
 172
 173         funcs->next = funcs->prev = NULL;
 174         DLIST_ADD(charsets, funcs);
 175         return 0;
 176 }
 177
 178 void lazy_initialize_iconv(void)
 179 {
 180         static int initialized = 0;
 181         int i;
 182
 183         if (!initialized) {
 184                 initialized = 1;
 185                 for(i = 0; builtin_functions[i].name; i++)
 186                         atalk_register_charset(&builtin_functions[i]);
 187
 188                 /* register additional charsets */
 189                 atalk_register_charset(&charset_utf8);
 190                 atalk_register_charset(&charset_utf8_mac);
 191                 atalk_register_charset(&charset_mac_roman);
 192                 atalk_register_charset(&charset_mac_hebrew);
 193                 atalk_register_charset(&charset_mac_turkish);
 194                 atalk_register_charset(&charset_mac_centraleurope);
 195                 atalk_register_charset(&charset_mac_cyrillic);
 196         }
 197 }
 198
 199 /* if there was an error then reset the internal state,
 200    this ensures that we don't have a shift state remaining for
 201    character sets like SJIS */
 202 static size_t sys_iconv(void *cd,
 203                         char **inbuf, size_t *inbytesleft,
 204                         char **outbuf, size_t *outbytesleft)
 205 {
 206 #ifdef HAVE_USABLE_ICONV
 207         size_t ret = iconv((iconv_t)cd,
 208                            inbuf, inbytesleft,
 209                            outbuf, outbytesleft);
 210         if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
 211         return ret;
 212 #else
 213         errno = EINVAL;
 214         return -1;
 215 #endif
 216 }
 217
 218 /**
 219  * This is a simple portable iconv() implementaion.
 220  *
 221  * It only knows about a very small number of character sets - just
 222  * enough that netatalk works on systems that don't have iconv.
 223  **/
 224 size_t atalk_iconv(atalk_iconv_t cd,
 225                  const char **inbuf, size_t *inbytesleft,
 226                  char **outbuf, size_t *outbytesleft)
 227 {
 228         char cvtbuf[2048];
 229         char *bufp = cvtbuf;
 230         size_t bufsize;
 231
 232         /* in many cases we can go direct */
 233         if (cd->direct) {
 234                 return cd->direct(cd->cd_direct,
 235                                   (char **)inbuf, inbytesleft, outbuf, outbytesleft);
 236         }
 237
 238
 239         /* otherwise we have to do it chunks at a time */
 240         while (*inbytesleft > 0) {
 241                 bufp = cvtbuf;
 242                 bufsize = sizeof(cvtbuf);
 243
 244                 if (cd->pull(cd->cd_pull, (char **)inbuf, inbytesleft, &bufp, &bufsize) == (size_t)-1
 245                        && errno != E2BIG) {
 246                     return -1;
 247                 }
 248
 249                 bufp = cvtbuf;
 250                 bufsize = sizeof(cvtbuf) - bufsize;
 251
 252                 if (cd->push(cd->cd_push, &bufp, &bufsize, outbuf, outbytesleft) == (size_t)-1) {
 253                     return -1;
 254                 }
 255         }
 256
 257         return 0;
 258 }
 259
 260
 261 /*
 262   simple iconv_open() wrapper
 263  */
 264 atalk_iconv_t atalk_iconv_open(const char *tocode, const char *fromcode)
 265 {
 266         atalk_iconv_t ret;
 267         struct charset_functions *from, *to;
 268
 269
 270         lazy_initialize_iconv();
 271         from = charsets;
 272         to = charsets;
 273
 274         ret = (atalk_iconv_t)malloc(sizeof(*ret));
 275         if (!ret) {
 276                 errno = ENOMEM;
 277                 return (atalk_iconv_t)-1;
 278         }
 279         memset(ret, 0, sizeof(*ret));
 280
 281         ret->from_name = strdup(fromcode);
 282         ret->to_name = strdup(tocode);
 283
 284         /* check for the simplest null conversion */
 285         if (strcasecmp(fromcode, tocode) == 0) {
 286                 ret->direct = iconv_copy;
 287                 return ret;
 288         }
 289
 290         /* check if we have a builtin function for this conversion */
 291         from = find_charset_functions(fromcode);
 292         if(from)ret->pull = from->pull;
 293
 294         to = find_charset_functions(tocode);
 295         if(to)ret->push = to->push;
 296
 297         /* check if we can use iconv for this conversion */
 298 #ifdef HAVE_USABLE_ICONV
 299         if (!ret->pull) {
 300                 ret->cd_pull = iconv_open(UCS2ICONV, fromcode);
 301                 if (ret->cd_pull != (iconv_t)-1)
 302                         ret->pull = sys_iconv;
 303         }
 304
 305         if (!ret->push) {
 306                 ret->cd_push = iconv_open(tocode, UCS2ICONV);
 307                 if (ret->cd_push != (iconv_t)-1)
 308                         ret->push = sys_iconv;
 309         }
 310 #endif
 311
 312         if (!ret->push || !ret->pull) {
 313                 SAFE_FREE(ret->from_name);
 314                 SAFE_FREE(ret->to_name);
 315                 SAFE_FREE(ret);
 316                 errno = EINVAL;
 317                 return (atalk_iconv_t)-1;
 318         }
 319
 320         /* check for conversion to/from ucs2 */
 321         if (strcasecmp(fromcode, "UCS-2") == 0 && to) {
 322                 ret->direct = to->push;
 323                 ret->push = ret->pull = NULL;
 324                 return ret;
 325         }
 326
 327         if (strcasecmp(tocode, "UCS-2") == 0 && from) {
 328                 ret->direct = from->pull;
 329                 ret->push = ret->pull = NULL;
 330                 return ret;
 331         }
 332
 333         /* Check if we can do the conversion direct */
 334 #ifdef HAVE_USABLE_ICONV
 335         if (strcasecmp(fromcode, "UCS-2") == 0) {
 336                 ret->direct = sys_iconv;
 337                 ret->cd_direct = ret->cd_push;
 338                 ret->cd_push = NULL;
 339                 return ret;
 340         }
 341         if (strcasecmp(tocode, "UCS-2") == 0) {
 342                 ret->direct = sys_iconv;
 343                 ret->cd_direct = ret->cd_pull;
 344                 ret->cd_pull = NULL;
 345                 return ret;
 346         }
 347 #endif
 348
 349         return ret;
 350 }
 351
 352 /*
 353   simple iconv_close() wrapper
 354 */
 355 int atalk_iconv_close (atalk_iconv_t cd)
 356 {
 357 #ifdef HAVE_USABLE_ICONV
 358         if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
 359         if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
 360         if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
 361 #endif
 362
 363         SAFE_FREE(cd->from_name);
 364         SAFE_FREE(cd->to_name);
 365
 366         memset(cd, 0, sizeof(*cd));
 367         SAFE_FREE(cd);
 368         return 0;
 369 }
 370
 371
 372 /************************************************************************
 373  the following functions implement the builtin character sets in Netatalk
 374 *************************************************************************/
 375
 376 static size_t ascii_pull(void *cd, char **inbuf, size_t *inbytesleft,
 377                          char **outbuf, size_t *outbytesleft)
 378 {
 379         ucs2_t curchar;
 380
 381         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 382                 if ((unsigned char)(*inbuf)[0] < 0x80) {
 383                         curchar = (ucs2_t) (*inbuf)[0];
 384                         SSVAL((*outbuf),0,curchar);
 385                 }
 386                 else {
 387                         errno = EILSEQ;
 388                         return -1;
 389                 }
 390                 (*inbytesleft)  -= 1;
 391                 (*outbytesleft) -= 2;
 392                 (*inbuf)  += 1;
 393                 (*outbuf) += 2;
 394         }
 395
 396         if (*inbytesleft > 0) {
 397                 errno = E2BIG;
 398                 return -1;
 399         }
 400
 401         return 0;
 402 }
 403
 404 static size_t ascii_push(void *cd, char **inbuf, size_t *inbytesleft,
 405                          char **outbuf, size_t *outbytesleft)
 406 {
 407         int ir_count=0;
 408         ucs2_t curchar;
 409
 410         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 411                 curchar = SVAL((*inbuf), 0);
 412                 if (curchar < 0x0080) {
 413                         (*outbuf)[0] = curchar;
 414                 }
 415                 else {
 416                         errno = EILSEQ;
 417                         return -1;
 418                 }
 419                 (*inbytesleft)  -= 2;
 420                 (*outbytesleft) -= 1;
 421                 (*inbuf)  += 2;
 422                 (*outbuf) += 1;
 423         }
 424
 425         if (*inbytesleft == 1) {
 426                 errno = EINVAL;
 427                 return -1;
 428         }
 429
 430         if (*inbytesleft > 1) {
 431                 errno = E2BIG;
 432                 return -1;
 433         }
 434
 435         return ir_count;
 436 }
 437
 438
 439 static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
 440                          char **outbuf, size_t *outbytesleft)
 441 {
 442         int n;
 443
 444         n = MIN(*inbytesleft, *outbytesleft);
 445
 446         memmove(*outbuf, *inbuf, n);
 447
 448         (*inbytesleft) -= n;
 449         (*outbytesleft) -= n;
 450         (*inbuf) += n;
 451         (*outbuf) += n;
 452
 453         if (*inbytesleft > 0) {
 454                 errno = E2BIG;
 455                 return -1;
 456         }
 457
 458         return 0;
 459 }
 460
 461 /* ------------------------ */