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 <sys/param.h>
  37 #include <sys/stat.h>
  38 #include <atalk/logger.h>
  39 #include <errno.h>
  40
  41 #include <netatalk/endian.h>
  42 #include <atalk/unicode.h>
  43
  44 #ifdef HAVE_USABLE_ICONV
  45 #include <iconv.h>
  46 #endif
  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 static size_t ascii_pull(void *,char **, size_t *, char **, size_t *);
  74 static size_t ascii_push(void *,char **, size_t *, char **, size_t *);
  75 static size_t iconv_copy(void *,char **, size_t *, char **, size_t *);
  76
  77 static char hexdig[] = "0123456789abcdef";
  78 #define hextoint( c )   ( isdigit( c ) ? c - '0' : c + 10 - 'a' )
  79
  80 struct charset_functions charset_ucs2 =
  81 {
  82         "UCS-2LE",
  83         0,
  84         iconv_copy,
  85         iconv_copy,
  86         CHARSET_WIDECHAR
  87 };
  88
  89 struct charset_functions charset_ascii =
  90 {
  91         "ASCII",
  92         0,
  93         ascii_pull,
  94         ascii_push,
  95         CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED
  96 };
  97
  98 struct charset_functions charset_iconv =
  99 {
 100         NULL,
 101         0,
 102         NULL,
 103         NULL,
 104         CHARSET_ICONV | CHARSET_PRECOMPOSED
 105 };
 106
 107
 108 extern  struct charset_functions charset_mac_roman;
 109 extern  struct charset_functions charset_mac_hebrew;
 110 extern  struct charset_functions charset_mac_centraleurope;
 111 extern  struct charset_functions charset_mac_cyrillic;
 112 extern  struct charset_functions charset_mac_turkish;
 113 extern  struct charset_functions charset_utf8;
 114 extern  struct charset_functions charset_utf8_mac;
 115
 116
 117 static struct charset_functions builtin_functions[] = {
 118         {"UCS-2LE",   0, iconv_copy, iconv_copy, CHARSET_WIDECHAR},
 119         {"ASCII",     0, ascii_pull, ascii_push, CHARSET_MULTIBYTE | CHARSET_PRECOMPOSED},
 120         {NULL, 0, NULL, NULL, 0}
 121 };
 122
 123
 124 #define DLIST_ADD(list, p) \
 125 { \
 126         if (!(list)) { \
 127                 (list) = (p); \
 128                 (p)->next = (p)->prev = NULL; \
 129         } else { \
 130                 (list)->prev = (p); \
 131                 (p)->next = (list); \
 132                 (p)->prev = NULL; \
 133                 (list) = (p); \
 134         }\
 135 }
 136
 137 static struct charset_functions *charsets = NULL;
 138
 139 struct charset_functions *find_charset_functions(const char *name)
 140 {
 141         struct charset_functions *c = charsets;
 142
 143         while(c) {
 144                 if (strcasecmp(name, c->name) == 0) {
 145                         return c;
 146                 }
 147                 c = c->next;
 148         }
 149
 150         return NULL;
 151 }
 152
 153 int atalk_register_charset(struct charset_functions *funcs)
 154 {
 155         if (!funcs) {
 156                 return -1;
 157         }
 158
 159         /* Check whether we already have this charset... */
 160         if (find_charset_functions(funcs->name)) {
 161                 LOG (log_debug, logtype_default, "Duplicate charset %s, not registering", funcs->name);
 162                 return -2;
 163         }
 164
 165         funcs->next = funcs->prev = NULL;
 166         DLIST_ADD(charsets, funcs);
 167         return 0;
 168 }
 169
 170 void lazy_initialize_iconv(void)
 171 {
 172         static int initialized = 0;
 173         int i;
 174
 175         if (!initialized) {
 176                 initialized = 1;
 177                 for(i = 0; builtin_functions[i].name; i++)
 178                         atalk_register_charset(&builtin_functions[i]);
 179
 180                 /* register additional charsets */
 181                 atalk_register_charset(&charset_utf8);
 182                 atalk_register_charset(&charset_utf8_mac);
 183                 atalk_register_charset(&charset_mac_roman);
 184                 atalk_register_charset(&charset_mac_hebrew);
 185                 atalk_register_charset(&charset_mac_turkish);
 186                 atalk_register_charset(&charset_mac_centraleurope);
 187                 atalk_register_charset(&charset_mac_cyrillic);
 188         }
 189 }
 190
 191 /* if there was an error then reset the internal state,
 192    this ensures that we don't have a shift state remaining for
 193    character sets like SJIS */
 194 static size_t sys_iconv(void *cd,
 195                         char **inbuf, size_t *inbytesleft,
 196                         char **outbuf, size_t *outbytesleft)
 197 {
 198 #ifdef HAVE_USABLE_ICONV
 199         size_t ret = iconv((iconv_t)cd,
 200                            inbuf, inbytesleft,
 201                            outbuf, outbytesleft);
 202         if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
 203         return ret;
 204 #else
 205         errno = EINVAL;
 206         return -1;
 207 #endif
 208 }
 209
 210 /**
 211  * This is a simple portable iconv() implementaion.
 212  *
 213  * It only knows about a very small number of character sets - just
 214  * enough that netatalk works on systems that don't have iconv.
 215  **/
 216 size_t atalk_iconv(atalk_iconv_t cd,
 217                  const char **inbuf, size_t *inbytesleft,
 218                  char **outbuf, size_t *outbytesleft)
 219 {
 220         char cvtbuf[2048];
 221         char *bufp = cvtbuf;
 222         size_t bufsize;
 223
 224         /* in many cases we can go direct */
 225         if (cd->direct) {
 226                 return cd->direct(cd->cd_direct,
 227                                   (char **)inbuf, inbytesleft, outbuf, outbytesleft);
 228         }
 229
 230
 231         /* otherwise we have to do it chunks at a time */
 232         while (*inbytesleft > 0) {
 233                 bufp = cvtbuf;
 234                 bufsize = sizeof(cvtbuf);
 235
 236                 if (cd->pull(cd->cd_pull, (char **)inbuf, inbytesleft, &bufp, &bufsize) == (size_t)-1
 237                        && errno != E2BIG) {
 238                     return -1;
 239                 }
 240
 241                 bufp = cvtbuf;
 242                 bufsize = sizeof(cvtbuf) - bufsize;
 243
 244                 if (cd->push(cd->cd_push, &bufp, &bufsize, outbuf, outbytesleft) == (size_t)-1) {
 245                     return -1;
 246                 }
 247         }
 248
 249         return 0;
 250 }
 251
 252
 253 size_t atalk_iconv_ignore(atalk_iconv_t cd,
 254                  const char **inbuf, size_t *inbytesleft,
 255                  char **outbuf, size_t *outbytesleft, int *ignore)
 256 {
 257         char cvtbuf[2048];
 258         char *bufp = cvtbuf;
 259         size_t bufsize;
 260         size_t outlen = *outbytesleft;
 261         char *o_save;
 262
 263         /* in many cases we can go direct */
 264         if (cd->direct) {
 265         o_save = *outbuf;
 266 direct_loop:
 267                 if ((size_t)-1 == cd->direct(cd->cd_direct, (char **)inbuf, inbytesleft, outbuf, outbytesleft)) {
 268                         if (errno == EILSEQ) {
 269                                 o_save[outlen-*outbytesleft] = '_';
 270                                 (*outbuf) = o_save + outlen-*outbytesleft+1;
 271                                 (*outbytesleft) -=1;
 272                                 (*inbuf) += 2;
 273                                 (*inbytesleft) -= 2;
 274                                 *ignore = 1;
 275                                 goto direct_loop;
 276                     }
 277                     else
 278                         return (size_t)(-1);
 279                 }
 280         }
 281
 282         /* we have to do it chunks at a time */
 283         while (*inbytesleft > 0) {
 284                 bufp = cvtbuf;
 285                 bufsize = sizeof(cvtbuf);
 286
 287                 if (cd->pull(cd->cd_pull, (char **)inbuf, inbytesleft, &bufp, &bufsize) == (size_t)-1
 288                         && errno != E2BIG) {
 289                     return -1;
 290                 }
 291
 292                 bufp = cvtbuf;
 293                 bufsize = sizeof(cvtbuf) - bufsize;
 294
 295                 o_save = *outbuf;
 296 convert_push:
 297                 if (cd->push(cd->cd_push,
 298                              &bufp, &bufsize,
 299                              outbuf, outbytesleft) == (size_t)-1) {
 300                     if (errno == EILSEQ) {
 301                         o_save[outlen-*outbytesleft] = '_';
 302                         (*outbuf) = o_save + outlen-*outbytesleft+1;
 303                         (*outbytesleft) -=1;
 304                         bufp += 2;
 305                         bufsize -= 2;
 306                         *ignore = 1;
 307                         goto convert_push;
 308                     }
 309                     else
 310                         return (size_t)(-1);
 311                 }
 312         }
 313         return 0;
 314 }
 315
 316 /*
 317   simple iconv_open() wrapper
 318  */
 319 atalk_iconv_t atalk_iconv_open(const char *tocode, const char *fromcode)
 320 {
 321         atalk_iconv_t ret;
 322         struct charset_functions *from, *to;
 323
 324
 325         lazy_initialize_iconv();
 326         from = charsets;
 327         to = charsets;
 328
 329         ret = (atalk_iconv_t)malloc(sizeof(*ret));
 330         if (!ret) {
 331                 errno = ENOMEM;
 332                 return (atalk_iconv_t)-1;
 333         }
 334         memset(ret, 0, sizeof(*ret));
 335
 336         ret->from_name = strdup(fromcode);
 337         ret->to_name = strdup(tocode);
 338
 339         /* check for the simplest null conversion */
 340         if (strcasecmp(fromcode, tocode) == 0) {
 341                 ret->direct = iconv_copy;
 342                 return ret;
 343         }
 344
 345         /* check if we have a builtin function for this conversion */
 346         from = find_charset_functions(fromcode);
 347         if(from)ret->pull = from->pull;
 348
 349         to = find_charset_functions(tocode);
 350         if(to)ret->push = to->push;
 351
 352         /* check if we can use iconv for this conversion */
 353 #ifdef HAVE_USABLE_ICONV
 354         if (!ret->pull) {
 355                 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
 356                 if (ret->cd_pull != (iconv_t)-1)
 357                         ret->pull = sys_iconv;
 358         }
 359
 360         if (!ret->push) {
 361                 ret->cd_push = iconv_open(tocode, "UCS-2LE");
 362                 if (ret->cd_push != (iconv_t)-1)
 363                         ret->push = sys_iconv;
 364         }
 365 #endif
 366
 367         if (!ret->push || !ret->pull) {
 368                 SAFE_FREE(ret->from_name);
 369                 SAFE_FREE(ret->to_name);
 370                 SAFE_FREE(ret);
 371                 errno = EINVAL;
 372                 return (atalk_iconv_t)-1;
 373         }
 374
 375         /* check for conversion to/from ucs2 */
 376         if (strcasecmp(fromcode, "UCS-2LE") == 0 && to) {
 377                 ret->direct = to->push;
 378                 ret->push = ret->pull = NULL;
 379                 return ret;
 380         }
 381
 382         if (strcasecmp(tocode, "UCS-2LE") == 0 && from) {
 383                 ret->direct = from->pull;
 384                 ret->push = ret->pull = NULL;
 385                 return ret;
 386         }
 387
 388         /* Check if we can do the conversion direct */
 389 #ifdef HAVE_USABLE_ICONV
 390         if (strcasecmp(fromcode, "UCS-2LE") == 0) {
 391                 ret->direct = sys_iconv;
 392                 ret->cd_direct = ret->cd_push;
 393                 ret->cd_push = NULL;
 394                 return ret;
 395         }
 396         if (strcasecmp(tocode, "UCS-2LE") == 0) {
 397                 ret->direct = sys_iconv;
 398                 ret->cd_direct = ret->cd_pull;
 399                 ret->cd_pull = NULL;
 400                 return ret;
 401         }
 402 #endif
 403
 404         return ret;
 405 }
 406
 407 /*
 408   simple iconv_close() wrapper
 409 */
 410 int atalk_iconv_close (atalk_iconv_t cd)
 411 {
 412 #ifdef HAVE_USABLE_ICONV
 413         if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
 414         if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
 415         if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
 416 #endif
 417
 418         SAFE_FREE(cd->from_name);
 419         SAFE_FREE(cd->to_name);
 420
 421         memset(cd, 0, sizeof(*cd));
 422         SAFE_FREE(cd);
 423         return 0;
 424 }
 425
 426
 427 /************************************************************************
 428  the following functions implement the builtin character sets in Netatalk
 429 *************************************************************************/
 430
 431 static size_t ascii_pull(void *cd, char **inbuf, size_t *inbytesleft,
 432                          char **outbuf, size_t *outbytesleft)
 433 {
 434         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 435                 (*outbuf)[0] = (*inbuf)[0];
 436                 (*outbuf)[1] = 0;
 437                 (*inbytesleft)  -= 1;
 438                 (*outbytesleft) -= 2;
 439                 (*inbuf)  += 1;
 440                 (*outbuf) += 2;
 441         }
 442
 443         if (*inbytesleft > 0) {
 444                 errno = E2BIG;
 445                 return -1;
 446         }
 447
 448         return 0;
 449 }
 450
 451 static size_t ascii_push(void *cd, char **inbuf, size_t *inbytesleft,
 452                          char **outbuf, size_t *outbytesleft)
 453 {
 454         int ir_count=0;
 455
 456         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 457                 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
 458                 if ((*inbuf)[1]) ir_count++;
 459                 (*inbytesleft)  -= 2;
 460                 (*outbytesleft) -= 1;
 461                 (*inbuf)  += 2;
 462                 (*outbuf) += 1;
 463         }
 464
 465         if (*inbytesleft == 1) {
 466                 errno = EINVAL;
 467                 return -1;
 468         }
 469
 470         if (*inbytesleft > 1) {
 471                 errno = E2BIG;
 472                 return -1;
 473         }
 474
 475         return ir_count;
 476 }
 477
 478
 479 static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
 480                          char **outbuf, size_t *outbytesleft)
 481 {
 482         int n;
 483
 484         n = MIN(*inbytesleft, *outbytesleft);
 485
 486         memmove(*outbuf, *inbuf, n);
 487
 488         (*inbytesleft) -= n;
 489         (*outbytesleft) -= n;
 490         (*inbuf) += n;
 491         (*outbuf) += n;
 492
 493         if (*inbytesleft > 0) {
 494                 errno = E2BIG;
 495                 return -1;
 496         }
 497
 498         return 0;
 499 }
 500
 501 /* ------------------------ */