source4/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 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 *,const char **, size_t *, char **, size_t *);
  49 static size_t ascii_push(void *,const char **, size_t *, char **, size_t *);
  50 static size_t  utf8_pull(void *,const char **, size_t *, char **, size_t *);
  51 static size_t  utf8_push(void *,const char **, size_t *, char **, size_t *);
  52 static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
  53 static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
  54 static size_t iconv_copy(void *,const 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 static 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 #ifdef HAVE_NATIVE_ICONV
  99 /* if there was an error then reset the internal state,
 100    this ensures that we don't have a shift state remaining for
 101    character sets like SJIS */
 102 static size_t sys_iconv(void *cd,
 103                         const char **inbuf, size_t *inbytesleft,
 104                         char **outbuf, size_t *outbytesleft)
 105 {
 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 }
 112 #endif
 113
 114 /**
 115  * This is a simple portable iconv() implementaion.
 116  *
 117  * It only knows about a very small number of character sets - just
 118  * enough that Samba works on systems that don't have iconv.
 119  **/
 120 size_t smb_iconv(smb_iconv_t cd,
 121                  const char **inbuf, size_t *inbytesleft,
 122                  char **outbuf, size_t *outbytesleft)
 123 {
 124         char cvtbuf[2048];
 125         char *bufp = cvtbuf;
 126         size_t bufsize;
 127
 128         /* in many cases we can go direct */
 129         if (cd->direct) {
 130                 return cd->direct(cd->cd_direct,
 131                                   inbuf, inbytesleft, outbuf, outbytesleft);
 132         }
 133
 134
 135         /* otherwise we have to do it chunks at a time */
 136         while (*inbytesleft > 0) {
 137                 bufp = cvtbuf;
 138                 bufsize = sizeof(cvtbuf);
 139
 140                 if (cd->pull(cd->cd_pull,
 141                              inbuf, inbytesleft, &bufp, &bufsize) == -1
 142                     && errno != E2BIG) return -1;
 143
 144                 bufp = cvtbuf;
 145                 bufsize = sizeof(cvtbuf) - bufsize;
 146
 147                 if (cd->push(cd->cd_push,
 148                              &bufp, &bufsize,
 149                              outbuf, outbytesleft) == -1) return -1;
 150         }
 151
 152         return 0;
 153 }
 154
 155 /*
 156   simple iconv_open() wrapper
 157  */
 158 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
 159 {
 160         smb_iconv_t ret;
 161         struct charset_functions *from, *to;
 162
 163         lazy_initialize_iconv();
 164         from = charsets;
 165         to = charsets;
 166
 167         ret = (smb_iconv_t)malloc(sizeof(*ret));
 168         if (!ret) {
 169                 errno = ENOMEM;
 170                 return (smb_iconv_t)-1;
 171         }
 172         memset(ret, 0, sizeof(*ret));
 173
 174         ret->from_name = strdup(fromcode);
 175         ret->to_name = strdup(tocode);
 176
 177         /* check for the simplest null conversion */
 178         if (strcmp(fromcode, tocode) == 0) {
 179                 ret->direct = iconv_copy;
 180                 return ret;
 181         }
 182
 183         while (from) {
 184                 if (strcasecmp(from->name, fromcode) == 0) break;
 185                 from = from->next;
 186         }
 187
 188         while (to) {
 189                 if (strcasecmp(to->name, tocode) == 0) break;
 190                 to = to->next;
 191         }
 192
 193 #ifdef HAVE_NATIVE_ICONV
 194         if (!from) {
 195                 ret->pull = sys_iconv;
 196                 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
 197                 if (ret->cd_pull == (iconv_t)-1) goto failed;
 198         }
 199
 200         if (!to) {
 201                 ret->push = sys_iconv;
 202                 ret->cd_push = iconv_open(tocode, "UCS-2LE");
 203                 if (ret->cd_push == (iconv_t)-1) goto failed;
 204         }
 205 #else
 206         if (!from || !to) {
 207                 goto failed;
 208         }
 209 #endif
 210
 211         /* check for conversion to/from ucs2 */
 212         if (strcasecmp(fromcode, "UCS-2LE") == 0 && to) {
 213                 ret->direct = to->push;
 214                 return ret;
 215         }
 216         if (strcasecmp(tocode, "UCS-2LE") == 0 && from) {
 217                 ret->direct = from->pull;
 218                 return ret;
 219         }
 220
 221 #ifdef HAVE_NATIVE_ICONV
 222         if (strcasecmp(fromcode, "UCS-2LE") == 0) {
 223                 ret->direct = sys_iconv;
 224                 ret->cd_direct = ret->cd_push;
 225                 ret->cd_push = NULL;
 226                 return ret;
 227         }
 228         if (strcasecmp(tocode, "UCS-2LE") == 0) {
 229                 ret->direct = sys_iconv;
 230                 ret->cd_direct = ret->cd_pull;
 231                 ret->cd_pull = NULL;
 232                 return ret;
 233         }
 234 #endif
 235
 236         /* the general case has to go via a buffer */
 237         if (!ret->pull) ret->pull = from->pull;
 238         if (!ret->push) ret->push = to->push;
 239         return ret;
 240
 241 failed:
 242         SAFE_FREE(ret);
 243         errno = EINVAL;
 244         return (smb_iconv_t)-1;
 245 }
 246
 247 /*
 248   simple iconv_close() wrapper
 249 */
 250 int smb_iconv_close (smb_iconv_t cd)
 251 {
 252 #ifdef HAVE_NATIVE_ICONV
 253         if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
 254         if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
 255         if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
 256 #endif
 257
 258         SAFE_FREE(cd->from_name);
 259         SAFE_FREE(cd->to_name);
 260
 261         memset(cd, 0, sizeof(*cd));
 262         SAFE_FREE(cd);
 263         return 0;
 264 }
 265
 266
 267 /**********************************************************************
 268  the following functions implement the builtin character sets in Samba
 269  and also the "test" character sets that are designed to test
 270  multi-byte character set support for english users
 271 ***********************************************************************/
 272 static size_t ascii_pull(void *cd, const char **inbuf, size_t *inbytesleft,
 273                          char **outbuf, size_t *outbytesleft)
 274 {
 275         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 276                 (*outbuf)[0] = (*inbuf)[0];
 277                 (*outbuf)[1] = 0;
 278                 (*inbytesleft)  -= 1;
 279                 (*outbytesleft) -= 2;
 280                 (*inbuf)  += 1;
 281                 (*outbuf) += 2;
 282         }
 283
 284         if (*inbytesleft > 0) {
 285                 errno = E2BIG;
 286                 return -1;
 287         }
 288
 289         return 0;
 290 }
 291
 292 static size_t ascii_push(void *cd, const char **inbuf, size_t *inbytesleft,
 293                          char **outbuf, size_t *outbytesleft)
 294 {
 295         int ir_count=0;
 296
 297         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 298                 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
 299                 if ((*inbuf)[1]) ir_count++;
 300                 (*inbytesleft)  -= 2;
 301                 (*outbytesleft) -= 1;
 302                 (*inbuf)  += 2;
 303                 (*outbuf) += 1;
 304         }
 305
 306         if (*inbytesleft == 1) {
 307                 errno = EINVAL;
 308                 return -1;
 309         }
 310
 311         if (*inbytesleft > 1) {
 312                 errno = E2BIG;
 313                 return -1;
 314         }
 315
 316         return ir_count;
 317 }
 318
 319
 320 static size_t ucs2hex_pull(void *cd, const char **inbuf, size_t *inbytesleft,
 321                          char **outbuf, size_t *outbytesleft)
 322 {
 323         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 324                 unsigned v;
 325
 326                 if ((*inbuf)[0] != '@') {
 327                         /* seven bit ascii case */
 328                         (*outbuf)[0] = (*inbuf)[0];
 329                         (*outbuf)[1] = 0;
 330                         (*inbytesleft)  -= 1;
 331                         (*outbytesleft) -= 2;
 332                         (*inbuf)  += 1;
 333                         (*outbuf) += 2;
 334                         continue;
 335                 }
 336                 /* it's a hex character */
 337                 if (*inbytesleft < 5) {
 338                         errno = EINVAL;
 339                         return -1;
 340                 }
 341
 342                 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
 343                         errno = EILSEQ;
 344                         return -1;
 345                 }
 346
 347                 (*outbuf)[0] = v&0xff;
 348                 (*outbuf)[1] = v>>8;
 349                 (*inbytesleft)  -= 5;
 350                 (*outbytesleft) -= 2;
 351                 (*inbuf)  += 5;
 352                 (*outbuf) += 2;
 353         }
 354
 355         if (*inbytesleft > 0) {
 356                 errno = E2BIG;
 357                 return -1;
 358         }
 359
 360         return 0;
 361 }
 362
 363 static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
 364                            char **outbuf, size_t *outbytesleft)
 365 {
 366         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 367                 char buf[6];
 368
 369                 if ((*inbuf)[1] == 0 &&
 370                     ((*inbuf)[0] & 0x80) == 0 &&
 371                     (*inbuf)[0] != '@') {
 372                         (*outbuf)[0] = (*inbuf)[0];
 373                         (*inbytesleft)  -= 2;
 374                         (*outbytesleft) -= 1;
 375                         (*inbuf)  += 2;
 376                         (*outbuf) += 1;
 377                         continue;
 378                 }
 379                 if (*outbytesleft < 5) {
 380                         errno = E2BIG;
 381                         return -1;
 382                 }
 383                 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
 384                 memcpy(*outbuf, buf, 5);
 385                 (*inbytesleft)  -= 2;
 386                 (*outbytesleft) -= 5;
 387                 (*inbuf)  += 2;
 388                 (*outbuf) += 5;
 389         }
 390
 391         if (*inbytesleft == 1) {
 392                 errno = EINVAL;
 393                 return -1;
 394         }
 395
 396         if (*inbytesleft > 1) {
 397                 errno = E2BIG;
 398                 return -1;
 399         }
 400
 401         return 0;
 402 }
 403
 404
 405 static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
 406                          char **outbuf, size_t *outbytesleft)
 407 {
 408         int n;
 409
 410         n = MIN(*inbytesleft, *outbytesleft);
 411
 412         memmove(*outbuf, *inbuf, n);
 413
 414         (*inbytesleft) -= n;
 415         (*outbytesleft) -= n;
 416         (*inbuf) += n;
 417         (*outbuf) += n;
 418
 419         if (*inbytesleft > 0) {
 420                 errno = E2BIG;
 421                 return -1;
 422         }
 423
 424         return 0;
 425 }
 426
 427 static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
 428                          char **outbuf, size_t *outbytesleft)
 429 {
 430         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
 431                 const unsigned char *c = (const unsigned char *)*inbuf;
 432                 unsigned char *uc = (unsigned char *)*outbuf;
 433                 int len = 1;
 434
 435                 if ((c[0] & 0x80) == 0) {
 436                         uc[0] = c[0];
 437                         uc[1] = 0;
 438                 } else if ((c[0] & 0xf0) == 0xe0) {
 439                         if (*inbytesleft < 3) {
 440                                 DEBUG(0,("short utf8 char\n"));
 441                                 goto badseq;
 442                         }
 443                         uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
 444                         uc[0] = (c[1]<<6) | (c[2]&0x3f);
 445                         len = 3;
 446                 } else if ((c[0] & 0xe0) == 0xc0) {
 447                         if (*inbytesleft < 2) {
 448                                 DEBUG(0,("short utf8 char\n"));
 449                                 goto badseq;
 450                         }
 451                         uc[1] = (c[0]>>2) & 0x7;
 452                         uc[0] = (c[0]<<6) | (c[1]&0x3f);
 453                         len = 2;
 454                 }
 455
 456                 (*inbuf)  += len;
 457                 (*inbytesleft)  -= len;
 458                 (*outbytesleft) -= 2;
 459                 (*outbuf) += 2;
 460         }
 461
 462         if (*inbytesleft > 0) {
 463                 errno = E2BIG;
 464                 return -1;
 465         }
 466
 467         return 0;
 468
 469 badseq:
 470         errno = EINVAL;
 471         return -1;
 472 }
 473
 474 static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
 475                          char **outbuf, size_t *outbytesleft)
 476 {
 477         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
 478                 unsigned char *c = (unsigned char *)*outbuf;
 479                 const unsigned char *uc = (const unsigned char *)*inbuf;
 480                 int len=1;
 481
 482                 if (uc[1] & 0xf8) {
 483                         if (*outbytesleft < 3) {
 484                                 DEBUG(0,("short utf8 write\n"));
 485                                 goto toobig;
 486                         }
 487                         c[0] = 0xe0 | (uc[1]>>4);
 488                         c[1] = 0x80 | ((uc[1]&0xF)<<2) | (uc[0]>>6);
 489                         c[2] = 0x80 | (uc[0]&0x3f);
 490                         len = 3;
 491                 } else if (uc[1] | (uc[0] & 0x80)) {
 492                         if (*outbytesleft < 2) {
 493                                 DEBUG(0,("short utf8 write\n"));
 494                                 goto toobig;
 495                         }
 496                         c[0] = 0xc0 | (uc[1]<<2) | (uc[0]>>6);
 497                         c[1] = 0x80 | (uc[0]&0x3f);
 498                         len = 2;
 499                 } else {
 500                         c[0] = uc[0];
 501                 }
 502
 503
 504                 (*inbytesleft)  -= 2;
 505                 (*outbytesleft) -= len;
 506                 (*inbuf)  += 2;
 507                 (*outbuf) += len;
 508         }
 509
 510         if (*inbytesleft == 1) {
 511                 errno = EINVAL;
 512                 return -1;
 513         }
 514
 515         if (*inbytesleft > 1) {
 516                 errno = E2BIG;
 517                 return -1;
 518         }
 519
 520         return 0;
 521
 522 toobig:
 523         errno = E2BIG;
 524         return -1;
 525 }
 526