source/lib/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
  22 #include "includes.h"
  23
  24
  25 /**
  26  * @file
  27  *
  28  * @brief Samba wrapper/stub for iconv character set conversion.
  29  *
  30  * iconv is the XPG2 interface for converting between character
  31  * encodings.  This file provides a Samba wrapper around it, and also
  32  * a simple reimplementation that is used if the system does not
  33  * implement iconv.
  34  *
  35  * Samba only works with encodings that are supersets of ASCII: ascii
  36  * characters like whitespace can be tested for directly, multibyte
  37  * sequences start with a byte with the high bit set, and strings are
  38  * terminated by a nul byte.
  39  *
  40  * Note that the only function provided by iconv is conversion between
  41  * characters.  It doesn't directly support operations like
  42  * uppercasing or comparison.  We have to convert to UCS-2 and compare
  43  * there.
  44  *
  45  * @sa Samba Developers Guide
  46  **/
  47
  48 static size_t ascii_pull(void *,char **, size_t *, char **, size_t *);
  49 static size_t ascii_push(void *,char **, size_t *, char **, size_t *);
  50 static size_t  utf8_pull(void *,char **, size_t *, char **, size_t *);
  51 static size_t  utf8_push(void *,char **, size_t *, char **, size_t *);
  52 static size_t ucs2hex_pull(void *,char **, size_t *, char **, size_t *);
  53 static size_t ucs2hex_push(void *,char **, size_t *, char **, size_t *);
  54 static size_t iconv_copy(void *,char **, size_t *, char **, size_t *);
  55
  56 static struct charset_functions builtin_functions[] = {
  57         {"UCS-2LE",  iconv_copy, iconv_copy},
  58         {"UTF8",   utf8_pull,  utf8_push},
  59         {"ASCII", ascii_pull, ascii_push},
  60         {"UCS2-HEX", ucs2hex_pull, ucs2hex_push},
  61         {NULL, NULL, NULL}
  62 };
  63
  64 static struct charset_functions *charsets = NULL;
  65
  66 static struct charset_functions *find_charset_functions(const char *name)
  67 {
  68         struct charset_functions *c = charsets;
  69
  70         while(c) {
  71                 if (strcasecmp(name, c->name) == 0) {
  72                         return c;
  73                 }
  74                 c = c->next;
  75         }
  76
  77         return NULL;
  78 }
  79
  80 NTSTATUS smb_register_charset(struct charset_functions *funcs)
  81 {
  82         if (!funcs) {
  83                 return NT_STATUS_INVALID_PARAMETER;
  84         }
  85
  86         DEBUG(5, ("Attempting to register new charset %s\n", funcs->name));
  87         /* Check whether we already have this charset... */
  88         if (find_charset_functions(funcs->name)) {
  89                 DEBUG(0, ("Duplicate charset %s, not registering\n", funcs->name));
  90                 return NT_STATUS_OBJECT_NAME_COLLISION;
  91         }
  92
  93         funcs->next = funcs->prev = NULL;
  94         DEBUG(5, ("Registered charset %s\n", funcs->name));
  95         DLIST_ADD(charsets, funcs);
  96         return NT_STATUS_OK;
  97 }
  98
  99 void lazy_initialize_iconv(void)
 100 {
 101         static BOOL initialized;
 102         int i;
 103
 104         if (!initialized) {
 105                 initialized = True;
 106                 for(i = 0; builtin_functions[i].name; i++)
 107                         smb_register_charset(&builtin_functions[i]);
 108                 static_init_charset;
 109         }
 110 }
 111
 112 /* if there was an error then reset the internal state,
 113    this ensures that we don't have a shift state remaining for
 114    character sets like SJIS */
 115 static size_t sys_iconv(void *cd,
 116                         char **inbuf, size_t *inbytesleft,
 117                         char **outbuf, size_t *outbytesleft)
 118 {
 119 #ifdef HAVE_NATIVE_ICONV
 120         size_t ret = iconv((iconv_t)cd,
 121                            inbuf, inbytesleft,
 122                            outbuf, outbytesleft);
 123         if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
 124         return ret;
 125 #else
 126         errno = EINVAL;
 127         return -1;
 128 #endif
 129 }
 130
 131 /**
 132  * This is a simple portable iconv() implementaion.
 133  *
 134  * It only knows about a very small number of character sets - just
 135  * enough that Samba works on systems that don't have iconv.
 136  **/
 137 size_t smb_iconv(smb_iconv_t cd,
 138                  const char **inbuf, size_t *inbytesleft,
 139                  char **outbuf, size_t *outbytesleft)
 140 {
 141         char cvtbuf[2048];
 142         char *bufp = cvtbuf;
 143         size_t bufsize;
 144
 145         /* in many cases we can go direct */
 146         if (cd->direct) {
 147                 return cd->direct(cd->cd_direct,
 148                                   (char **)inbuf, inbytesleft, outbuf, outbytesleft);
 149         }
 150
 151
 152         /* otherwise we have to do it chunks at a time */
 153         while (*inbytesleft > 0) {
 154                 bufp = cvtbuf;
 155                 bufsize = sizeof(cvtbuf);
 156
 157                 if (cd->pull(cd->cd_pull,
 158                              (char **)inbuf, inbytesleft, &bufp, &bufsize) == -1
 159                     && errno != E2BIG) return -1;
 160
 161                 bufp = cvtbuf;
 162                 bufsize = sizeof(cvtbuf) - bufsize;
 163
 164                 if (cd->push(cd->cd_push,
 165                              &bufp, &bufsize,
 166                              outbuf, outbytesleft) == -1) return -1;
 167         }
 168
 169         return 0;
 170 }
 171
 172 /*
 173   simple iconv_open() wrapper
 174  */
 175 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
 176 {
 177         smb_iconv_t ret;
 178         struct charset_functions *from, *to;
 179
 180         lazy_initialize_iconv();
 181         from = charsets;
 182         to = charsets;
 183
 184         ret = (smb_iconv_t)malloc(sizeof(*ret));
 185         if (!ret) {
 186                 errno = ENOMEM;
 187                 return (smb_iconv_t)-1;
 188         }
 189         memset(ret, 0, sizeof(*ret));
 190
 191         ret->from_name = strdup(fromcode);
 192         ret->to_name = strdup(tocode);
 193
 194         /* check for the simplest null conversion */
 195         if (strcasecmp(fromcode, tocode) == 0) {
 196                 ret->direct = iconv_copy;
 197                 return ret;
 198         }
 199
 200         /* check if we have a builtin function for this conversion */
 201         from = find_charset_functions(fromcode);
 202         if(from)ret->pull = from->pull;
 203
 204         to = find_charset_functions(tocode);
 205         if(to)ret->push = to->push;
 206
 207         /* check if we can use iconv for this conversion */
 208 #ifdef HAVE_NATIVE_ICONV
 209         if (!ret->pull) {
 210                 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
 211                 if (ret->cd_pull != (iconv_t)-1)
 212                         ret->pull = sys_iconv;
 213         }
 214
 215         if (!ret->push) {
 216                 ret->cd_push = iconv_open(tocode, "UCS-2LE");
 217                 if (ret->cd_push != (iconv_t)-1)
 218                         ret->push = sys_iconv;
 219         }
 220 #endif
 221
 222         /* check if there is a module available that can do this conversion */
 223         if (!ret->pull && NT_STATUS_IS_OK(smb_probe_module("charset", fromcode))) {
 224                 if(!(from = find_charset_functions(fromcode)))
 225                         DEBUG(0, ("Module %s doesn't provide charset %s!\n", fromcode, fromcode));
 226                 else
 227                         ret->pull = from->pull;
 228         }
 229
 230         if (!ret->push && NT_STATUS_IS_OK(smb_probe_module("charset", tocode))) {
 231                 if(!(to = find_charset_functions(tocode)))
 232                         DEBUG(0, ("Module %s doesn't provide charset %s!\n", tocode, tocode));
 233                 else
 234                         ret->push = to->push;
 235         }
 236
 237         if (!ret->push || !ret->pull) {
 238                 SAFE_FREE(ret->from_name);
 239                 SAFE_FREE(ret->to_name);
 240                 SAFE_FREE(ret);
 241                 errno = EINVAL;
 242                 return (smb_iconv_t)-1;
 243         }
 244
 245         /* check for conversion to/from ucs2 */
 246         if (strcasecmp(fromcode, "UCS-2LE") == 0 && to) {
 247                 ret->direct = to->push;
 248                 ret->push = ret->pull = NULL;
 249                 return ret;
 250         }
 251
 252         if (strcasecmp(tocode, "UCS-2LE") == 0 && from) {
 253                 ret->direct = from->pull;
 254                 ret->push = ret->pull = NULL;
 255                 return ret;
 256         }
 257
 258         /* Check if we can do the conversion direct */
 259 #ifdef HAVE_NATIVE_ICONV
 260         if (strcasecmp(fromcode, "UCS-2LE") == 0) {
 261                 ret->direct = sys_iconv;
 262                 ret->cd_direct = ret->cd_push;
 263                 ret->cd_push = NULL;
 264                 return ret;
 265         }
 266         if (strcasecmp(tocode, "UCS-2LE") == 0) {
 267                 ret->direct = sys_iconv;
 268                 ret->cd_direct = ret->cd_pull;
 269                 ret->cd_pull = NULL;
 270                 return ret;
 271         }
 272 #endif
 273
 274         return ret;
 275 }
 276
 277 /*
 278   simple iconv_close() wrapper
 279 */
 280 int smb_iconv_close (smb_iconv_t cd)
 281 {
 282 #ifdef HAVE_NATIVE_ICONV
 283         if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
 284         if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
 285         if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
 286 #endif
 287
 288         SAFE_FREE(cd->from_name);
 289         SAFE_FREE(cd->to_name);
 290
 291         memset(cd, 0, sizeof(*cd));
 292         SAFE_FREE(cd);
 293         return 0;
 294 }
 295
 296
 297 /**********************************************************************
 298  the following functions implement the builtin character sets in Samba
 299  and also the "test" character sets that are designed to test
 300  multi-byte character set support for english users
 301 ***********************************************************************/
 302
 303 static size_t ascii_pull(void *cd, char **inbuf, size_t *inbytesleft,
 304                          char **outbuf, size_t *outbytesleft)
 305 {
 306         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 307                 (*outbuf)[0] = (*inbuf)[0];
 308                 (*outbuf)[1] = 0;
 309                 (*inbytesleft)  -= 1;
 310                 (*outbytesleft) -= 2;
 311                 (*inbuf)  += 1;
 312                 (*outbuf) += 2;
 313         }
 314
 315         if (*inbytesleft > 0) {
 316                 errno = E2BIG;
 317                 return -1;
 318         }
 319
 320         return 0;
 321 }
 322
 323 static size_t ascii_push(void *cd, char **inbuf, size_t *inbytesleft,
 324                          char **outbuf, size_t *outbytesleft)
 325 {
 326         int ir_count=0;
 327
 328         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 329                 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
 330                 if ((*inbuf)[1]) ir_count++;
 331                 (*inbytesleft)  -= 2;
 332                 (*outbytesleft) -= 1;
 333                 (*inbuf)  += 2;
 334                 (*outbuf) += 1;
 335         }
 336
 337         if (*inbytesleft == 1) {
 338                 errno = EINVAL;
 339                 return -1;
 340         }
 341
 342         if (*inbytesleft > 1) {
 343                 errno = E2BIG;
 344                 return -1;
 345         }
 346
 347         return ir_count;
 348 }
 349
 350
 351 static size_t ucs2hex_pull(void *cd, char **inbuf, size_t *inbytesleft,
 352                          char **outbuf, size_t *outbytesleft)
 353 {
 354         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 355                 unsigned v;
 356
 357                 if ((*inbuf)[0] != '@') {
 358                         /* seven bit ascii case */
 359                         (*outbuf)[0] = (*inbuf)[0];
 360                         (*outbuf)[1] = 0;
 361                         (*inbytesleft)  -= 1;
 362                         (*outbytesleft) -= 2;
 363                         (*inbuf)  += 1;
 364                         (*outbuf) += 2;
 365                         continue;
 366                 }
 367                 /* it's a hex character */
 368                 if (*inbytesleft < 5) {
 369                         errno = EINVAL;
 370                         return -1;
 371                 }
 372
 373                 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
 374                         errno = EILSEQ;
 375                         return -1;
 376                 }
 377
 378                 (*outbuf)[0] = v&0xff;
 379                 (*outbuf)[1] = v>>8;
 380                 (*inbytesleft)  -= 5;
 381                 (*outbytesleft) -= 2;
 382                 (*inbuf)  += 5;
 383                 (*outbuf) += 2;
 384         }
 385
 386         if (*inbytesleft > 0) {
 387                 errno = E2BIG;
 388                 return -1;
 389         }
 390
 391         return 0;
 392 }
 393
 394 static size_t ucs2hex_push(void *cd, char **inbuf, size_t *inbytesleft,
 395                            char **outbuf, size_t *outbytesleft)
 396 {
 397         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 398                 char buf[6];
 399
 400                 if ((*inbuf)[1] == 0 &&
 401                     ((*inbuf)[0] & 0x80) == 0 &&
 402                     (*inbuf)[0] != '@') {
 403                         (*outbuf)[0] = (*inbuf)[0];
 404                         (*inbytesleft)  -= 2;
 405                         (*outbytesleft) -= 1;
 406                         (*inbuf)  += 2;
 407                         (*outbuf) += 1;
 408                         continue;
 409                 }
 410                 if (*outbytesleft < 5) {
 411                         errno = E2BIG;
 412                         return -1;
 413                 }
 414                 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
 415                 memcpy(*outbuf, buf, 5);
 416                 (*inbytesleft)  -= 2;
 417                 (*outbytesleft) -= 5;
 418                 (*inbuf)  += 2;
 419                 (*outbuf) += 5;
 420         }
 421
 422         if (*inbytesleft == 1) {
 423                 errno = EINVAL;
 424                 return -1;
 425         }
 426
 427         if (*inbytesleft > 1) {
 428                 errno = E2BIG;
 429                 return -1;
 430         }
 431
 432         return 0;
 433 }
 434
 435
 436 static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
 437                          char **outbuf, size_t *outbytesleft)
 438 {
 439         int n;
 440
 441         n = MIN(*inbytesleft, *outbytesleft);
 442
 443         memmove(*outbuf, *inbuf, n);
 444
 445         (*inbytesleft) -= n;
 446         (*outbytesleft) -= n;
 447         (*inbuf) += n;
 448         (*outbuf) += n;
 449
 450         if (*inbytesleft > 0) {
 451                 errno = E2BIG;
 452                 return -1;
 453         }
 454
 455         return 0;
 456 }
 457
 458 static size_t utf8_pull(void *cd, char **inbuf, size_t *inbytesleft,
 459                          char **outbuf, size_t *outbytesleft)
 460 {
 461         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 462                 unsigned char *c = (unsigned char *)*inbuf;
 463                 unsigned char *uc = (unsigned char *)*outbuf;
 464                 int len = 1;
 465
 466                 if ((c[0] & 0x80) == 0) {
 467                         uc[0] = c[0];
 468                         uc[1] = 0;
 469                 } else if ((c[0] & 0xf0) == 0xe0) {
 470                         if (*inbytesleft < 3) {
 471                                 DEBUG(0,("short utf8 char\n"));
 472                                 goto badseq;
 473                         }
 474                         uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
 475                         uc[0] = (c[1]<<6) | (c[2]&0x3f);
 476                         len = 3;
 477                 } else if ((c[0] & 0xe0) == 0xc0) {
 478                         if (*inbytesleft < 2) {
 479                                 DEBUG(0,("short utf8 char\n"));
 480                                 goto badseq;
 481                         }
 482                         uc[1] = (c[0]>>2) & 0x7;
 483                         uc[0] = (c[0]<<6) | (c[1]&0x3f);
 484                         len = 2;
 485                 }
 486
 487                 (*inbuf)  += len;
 488                 (*inbytesleft)  -= len;
 489                 (*outbytesleft) -= 2;
 490                 (*outbuf) += 2;
 491         }
 492
 493         if (*inbytesleft > 0) {
 494                 errno = E2BIG;
 495                 return -1;
 496         }
 497
 498         return 0;
 499
 500 badseq:
 501         errno = EINVAL;
 502         return -1;
 503 }
 504
 505 static size_t utf8_push(void *cd, char **inbuf, size_t *inbytesleft,
 506                          char **outbuf, size_t *outbytesleft)
 507 {
 508         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 509                 unsigned char *c = (unsigned char *)*outbuf;
 510                 unsigned char *uc = (unsigned char *)*inbuf;
 511                 int len=1;
 512
 513                 if (uc[1] & 0xf8) {
 514                         if (*outbytesleft < 3) {
 515                                 DEBUG(0,("short utf8 write\n"));
 516                                 goto toobig;
 517                         }
 518                         c[0] = 0xe0 | (uc[1]>>4);
 519                         c[1] = 0x80 | ((uc[1]&0xF)<<2) | (uc[0]>>6);
 520                         c[2] = 0x80 | (uc[0]&0x3f);
 521                         len = 3;
 522                 } else if (uc[1] | (uc[0] & 0x80)) {
 523                         if (*outbytesleft < 2) {
 524                                 DEBUG(0,("short utf8 write\n"));
 525                                 goto toobig;
 526                         }
 527                         c[0] = 0xc0 | (uc[1]<<2) | (uc[0]>>6);
 528                         c[1] = 0x80 | (uc[0]&0x3f);
 529                         len = 2;
 530                 } else {
 531                         c[0] = uc[0];
 532                 }
 533
 534
 535                 (*inbytesleft)  -= 2;
 536                 (*outbytesleft) -= len;
 537                 (*inbuf)  += 2;
 538                 (*outbuf) += len;
 539         }
 540
 541         if (*inbytesleft == 1) {
 542                 errno = EINVAL;
 543                 return -1;
 544         }
 545
 546         if (*inbytesleft > 1) {
 547                 errno = E2BIG;
 548                 return -1;
 549         }
 550
 551         return 0;
 552
 553 toobig:
 554         errno = E2BIG;
 555         return -1;
 556 }
 557