source3/lib/iconv.c

   1 /*
   2    Unix SMB/CIFS implementation.
   3    minimal iconv implementation
   4    Copyright (C) Andrew Tridgell 2001
   5    Copyright (C) Jelmer Vernooij 2002,2003,2003,2003,2003
   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 BOOL smb_register_charset(struct charset_functions *funcs)
  81 {
  82         struct charset_functions *c = charsets;
  83
  84         DEBUG(5, ("Attempting to register new charset %s\n", funcs->name));
  85         /* Check whether we already have this charset... */
  86
  87         if (find_charset_functions(funcs->name)) {
  88                 DEBUG(2, ("Duplicate charset %s, not registering\n", funcs->name));
  89                 return False;
  90         }
  91
  92         funcs->next = funcs->prev = NULL;
  93         DEBUG(5, ("Registered charset %s\n", funcs->name));
  94         DLIST_ADD(charsets, funcs);
  95         return True;
  96 }
  97
  98 void lazy_initialize_iconv(void)
  99 {
 100         static BOOL initialized;
 101         int i;
 102
 103         if (!initialized) {
 104                 initialized = True;
 105                 for(i = 0; builtin_functions[i].name; i++)
 106                         smb_register_charset(&builtin_functions[i]);
 107                 static_init_charset;
 108         }
 109 }
 110
 111 /* if there was an error then reset the internal state,
 112    this ensures that we don't have a shift state remaining for
 113    character sets like SJIS */
 114 static size_t sys_iconv(void *cd,
 115                         char **inbuf, size_t *inbytesleft,
 116                         char **outbuf, size_t *outbytesleft)
 117 {
 118 #ifdef HAVE_NATIVE_ICONV
 119         size_t ret = iconv((iconv_t)cd,
 120                            inbuf, inbytesleft,
 121                            outbuf, outbytesleft);
 122         if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
 123         return ret;
 124 #else
 125         errno = EINVAL;
 126         return -1;
 127 #endif
 128 }
 129
 130 /**
 131  * This is a simple portable iconv() implementaion.
 132  *
 133  * It only knows about a very small number of character sets - just
 134  * enough that Samba works on systems that don't have iconv.
 135  **/
 136 size_t smb_iconv(smb_iconv_t cd,
 137                  const char **inbuf, size_t *inbytesleft,
 138                  char **outbuf, size_t *outbytesleft)
 139 {
 140         char cvtbuf[2048];
 141         char *bufp = cvtbuf;
 142         size_t bufsize;
 143
 144         /* in many cases we can go direct */
 145         if (cd->direct) {
 146                 return cd->direct(cd->cd_direct,
 147                                   (char **)inbuf, inbytesleft, outbuf, outbytesleft);
 148         }
 149
 150
 151         /* otherwise we have to do it chunks at a time */
 152         while (*inbytesleft > 0) {
 153                 bufp = cvtbuf;
 154                 bufsize = sizeof(cvtbuf);
 155
 156                 if (cd->pull(cd->cd_pull,
 157                              (char **)inbuf, inbytesleft, &bufp, &bufsize) == -1
 158                     && errno != E2BIG) return -1;
 159
 160                 bufp = cvtbuf;
 161                 bufsize = sizeof(cvtbuf) - bufsize;
 162
 163                 if (cd->push(cd->cd_push,
 164                              &bufp, &bufsize,
 165                              outbuf, outbytesleft) == -1) return -1;
 166         }
 167
 168         return 0;
 169 }
 170
 171 /*
 172   simple iconv_open() wrapper
 173  */
 174 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
 175 {
 176         smb_iconv_t ret;
 177         struct charset_functions *from, *to;
 178
 179         lazy_initialize_iconv();
 180         from = charsets;
 181         to = charsets;
 182
 183         ret = (smb_iconv_t)malloc(sizeof(*ret));
 184         if (!ret) {
 185                 errno = ENOMEM;
 186                 return (smb_iconv_t)-1;
 187         }
 188         memset(ret, 0, sizeof(*ret));
 189
 190         ret->from_name = strdup(fromcode);
 191         ret->to_name = strdup(tocode);
 192
 193         /* check for the simplest null conversion */
 194         if (strcasecmp(fromcode, tocode) == 0) {
 195                 ret->direct = iconv_copy;
 196                 return ret;
 197         }
 198
 199         /* check if we have a builtin function for this conversion */
 200         from = find_charset_functions(fromcode);
 201         if(from)ret->pull = from->pull;
 202
 203         to = find_charset_functions(tocode);
 204         if(to)ret->push = to->push;
 205
 206         /* check if we can use iconv for this conversion */
 207 #ifdef HAVE_NATIVE_ICONV
 208         if (!ret->pull) {
 209                 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
 210                 if (ret->cd_pull != (iconv_t)-1)
 211                         ret->pull = sys_iconv;
 212         }
 213
 214         if (!ret->push) {
 215                 ret->cd_push = iconv_open(tocode, "UCS-2LE");
 216                 if (ret->cd_push != (iconv_t)-1)
 217                         ret->push = sys_iconv;
 218         }
 219 #endif
 220
 221         /* check if there is a module available that can do this conversion */
 222         if (!ret->pull && smb_probe_module("charset", fromcode)) {
 223                 if(!(from = find_charset_functions(fromcode)))
 224                         DEBUG(0, ("Module %s doesn't provide charset %s!\n", fromcode, fromcode));
 225                 else
 226                         ret->pull = from->pull;
 227         }
 228
 229         if (!ret->push && smb_probe_module("charset", tocode)) {
 230                 if(!(to = find_charset_functions(tocode)))
 231                         DEBUG(0, ("Module %s doesn't provide charset %s!\n", tocode, tocode));
 232                 else
 233                         ret->push = to->push;
 234         }
 235
 236         if (!ret->push || !ret->pull) {
 237                 SAFE_FREE(ret->from_name);
 238                 SAFE_FREE(ret->to_name);
 239                 SAFE_FREE(ret);
 240                 errno = EINVAL;
 241                 return (smb_iconv_t)-1;
 242         }
 243
 244         /* check for conversion to/from ucs2 */
 245         if (strcasecmp(fromcode, "UCS-2LE") == 0 && to) {
 246                 ret->direct = to->push;
 247                 ret->push = ret->pull = NULL;
 248                 return ret;
 249         }
 250
 251         if (strcasecmp(tocode, "UCS-2LE") == 0 && from) {
 252                 ret->direct = from->pull;
 253                 ret->push = ret->pull = NULL;
 254                 return ret;
 255         }
 256
 257         /* Check if we can do the conversion direct */
 258 #ifdef HAVE_NATIVE_ICONV
 259         if (strcasecmp(fromcode, "UCS-2LE") == 0) {
 260                 ret->direct = sys_iconv;
 261                 ret->cd_direct = ret->cd_push;
 262                 ret->cd_push = NULL;
 263                 return ret;
 264         }
 265         if (strcasecmp(tocode, "UCS-2LE") == 0) {
 266                 ret->direct = sys_iconv;
 267                 ret->cd_direct = ret->cd_pull;
 268                 ret->cd_pull = NULL;
 269                 return ret;
 270         }
 271 #endif
 272
 273         return ret;
 274 }
 275
 276 /*
 277   simple iconv_close() wrapper
 278 */
 279 int smb_iconv_close (smb_iconv_t cd)
 280 {
 281 #ifdef HAVE_NATIVE_ICONV
 282         if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
 283         if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
 284         if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
 285 #endif
 286
 287         SAFE_FREE(cd->from_name);
 288         SAFE_FREE(cd->to_name);
 289
 290         memset(cd, 0, sizeof(*cd));
 291         SAFE_FREE(cd);
 292         return 0;
 293 }
 294
 295
 296 /**********************************************************************
 297  the following functions implement the builtin character sets in Samba
 298  and also the "test" character sets that are designed to test
 299  multi-byte character set support for english users
 300 ***********************************************************************/
 301
 302 static size_t ascii_pull(void *cd, char **inbuf, size_t *inbytesleft,
 303                          char **outbuf, size_t *outbytesleft)
 304 {
 305         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 306                 (*outbuf)[0] = (*inbuf)[0];
 307                 (*outbuf)[1] = 0;
 308                 (*inbytesleft)  -= 1;
 309                 (*outbytesleft) -= 2;
 310                 (*inbuf)  += 1;
 311                 (*outbuf) += 2;
 312         }
 313
 314         if (*inbytesleft > 0) {
 315                 errno = E2BIG;
 316                 return -1;
 317         }
 318
 319         return 0;
 320 }
 321
 322 static size_t ascii_push(void *cd, char **inbuf, size_t *inbytesleft,
 323                          char **outbuf, size_t *outbytesleft)
 324 {
 325         int ir_count=0;
 326
 327         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 328                 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
 329                 if ((*inbuf)[1]) ir_count++;
 330                 (*inbytesleft)  -= 2;
 331                 (*outbytesleft) -= 1;
 332                 (*inbuf)  += 2;
 333                 (*outbuf) += 1;
 334         }
 335
 336         if (*inbytesleft == 1) {
 337                 errno = EINVAL;
 338                 return -1;
 339         }
 340
 341         if (*inbytesleft > 1) {
 342                 errno = E2BIG;
 343                 return -1;
 344         }
 345
 346         return ir_count;
 347 }
 348
 349
 350 static size_t ucs2hex_pull(void *cd, char **inbuf, size_t *inbytesleft,
 351                          char **outbuf, size_t *outbytesleft)
 352 {
 353         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 354                 unsigned v;
 355
 356                 if ((*inbuf)[0] != '@') {
 357                         /* seven bit ascii case */
 358                         (*outbuf)[0] = (*inbuf)[0];
 359                         (*outbuf)[1] = 0;
 360                         (*inbytesleft)  -= 1;
 361                         (*outbytesleft) -= 2;
 362                         (*inbuf)  += 1;
 363                         (*outbuf) += 2;
 364                         continue;
 365                 }
 366                 /* it's a hex character */
 367                 if (*inbytesleft < 5) {
 368                         errno = EINVAL;
 369                         return -1;
 370                 }
 371
 372                 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
 373                         errno = EILSEQ;
 374                         return -1;
 375                 }
 376
 377                 (*outbuf)[0] = v&0xff;
 378                 (*outbuf)[1] = v>>8;
 379                 (*inbytesleft)  -= 5;
 380                 (*outbytesleft) -= 2;
 381                 (*inbuf)  += 5;
 382                 (*outbuf) += 2;
 383         }
 384
 385         if (*inbytesleft > 0) {
 386                 errno = E2BIG;
 387                 return -1;
 388         }
 389
 390         return 0;
 391 }
 392
 393 static size_t ucs2hex_push(void *cd, char **inbuf, size_t *inbytesleft,
 394                            char **outbuf, size_t *outbytesleft)
 395 {
 396         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 397                 char buf[6];
 398
 399                 if ((*inbuf)[1] == 0 &&
 400                     ((*inbuf)[0] & 0x80) == 0 &&
 401                     (*inbuf)[0] != '@') {
 402                         (*outbuf)[0] = (*inbuf)[0];
 403                         (*inbytesleft)  -= 2;
 404                         (*outbytesleft) -= 1;
 405                         (*inbuf)  += 2;
 406                         (*outbuf) += 1;
 407                         continue;
 408                 }
 409                 if (*outbytesleft < 5) {
 410                         errno = E2BIG;
 411                         return -1;
 412                 }
 413                 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
 414                 memcpy(*outbuf, buf, 5);
 415                 (*inbytesleft)  -= 2;
 416                 (*outbytesleft) -= 5;
 417                 (*inbuf)  += 2;
 418                 (*outbuf) += 5;
 419         }
 420
 421         if (*inbytesleft == 1) {
 422                 errno = EINVAL;
 423                 return -1;
 424         }
 425
 426         if (*inbytesleft > 1) {
 427                 errno = E2BIG;
 428                 return -1;
 429         }
 430
 431         return 0;
 432 }
 433
 434
 435 static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
 436                          char **outbuf, size_t *outbytesleft)
 437 {
 438         int n;
 439
 440         n = MIN(*inbytesleft, *outbytesleft);
 441
 442         memmove(*outbuf, *inbuf, n);
 443
 444         (*inbytesleft) -= n;
 445         (*outbytesleft) -= n;
 446         (*inbuf) += n;
 447         (*outbuf) += n;
 448
 449         if (*inbytesleft > 0) {
 450                 errno = E2BIG;
 451                 return -1;
 452         }
 453
 454         return 0;
 455 }
 456
 457 static size_t utf8_pull(void *cd, char **inbuf, size_t *inbytesleft,
 458                          char **outbuf, size_t *outbytesleft)
 459 {
 460         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 461                 unsigned char *c = (unsigned char *)*inbuf;
 462                 unsigned char *uc = (unsigned char *)*outbuf;
 463                 int len = 1;
 464
 465                 if ((c[0] & 0x80) == 0) {
 466                         uc[0] = c[0];
 467                         uc[1] = 0;
 468                 } else if ((c[0] & 0xf0) == 0xe0) {
 469                         if (*inbytesleft < 3) {
 470                                 DEBUG(0,("short utf8 char\n"));
 471                                 goto badseq;
 472                         }
 473                         uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
 474                         uc[0] = (c[1]<<6) | (c[2]&0x3f);
 475                         len = 3;
 476                 } else if ((c[0] & 0xe0) == 0xc0) {
 477                         if (*inbytesleft < 2) {
 478                                 DEBUG(0,("short utf8 char\n"));
 479                                 goto badseq;
 480                         }
 481                         uc[1] = (c[0]>>2) & 0x7;
 482                         uc[0] = (c[0]<<6) | (c[1]&0x3f);
 483                         len = 2;
 484                 }
 485
 486                 (*inbuf)  += len;
 487                 (*inbytesleft)  -= len;
 488                 (*outbytesleft) -= 2;
 489                 (*outbuf) += 2;
 490         }
 491
 492         if (*inbytesleft > 0) {
 493                 errno = E2BIG;
 494                 return -1;
 495         }
 496
 497         return 0;
 498
 499 badseq:
 500         errno = EINVAL;
 501         return -1;
 502 }
 503
 504 static size_t utf8_push(void *cd, char **inbuf, size_t *inbytesleft,
 505                          char **outbuf, size_t *outbytesleft)
 506 {
 507         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 508                 unsigned char *c = (unsigned char *)*outbuf;
 509                 unsigned char *uc = (unsigned char *)*inbuf;
 510                 int len=1;
 511
 512                 if (uc[1] & 0xf8) {
 513                         if (*outbytesleft < 3) {
 514                                 DEBUG(0,("short utf8 write\n"));
 515                                 goto toobig;
 516                         }
 517                         c[0] = 0xe0 | (uc[1]>>4);
 518                         c[1] = 0x80 | ((uc[1]&0xF)<<2) | (uc[0]>>6);
 519                         c[2] = 0x80 | (uc[0]&0x3f);
 520                         len = 3;
 521                 } else if (uc[1] | (uc[0] & 0x80)) {
 522                         if (*outbytesleft < 2) {
 523                                 DEBUG(0,("short utf8 write\n"));
 524                                 goto toobig;
 525                         }
 526                         c[0] = 0xc0 | (uc[1]<<2) | (uc[0]>>6);
 527                         c[1] = 0x80 | (uc[0]&0x3f);
 528                         len = 2;
 529                 } else {
 530                         c[0] = uc[0];
 531                 }
 532
 533
 534                 (*inbytesleft)  -= 2;
 535                 (*outbytesleft) -= len;
 536                 (*inbuf)  += 2;
 537                 (*outbuf) += len;
 538         }
 539
 540         if (*inbytesleft == 1) {
 541                 errno = EINVAL;
 542                 return -1;
 543         }
 544
 545         if (*inbytesleft > 1) {
 546                 errno = E2BIG;
 547                 return -1;
 548         }
 549
 550         return 0;
 551
 552 toobig:
 553         errno = E2BIG;
 554         return -1;
 555 }
 556