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