2 Unix SMB/CIFS implementation.
3 minimal iconv implementation
4 Copyright (C) Andrew Tridgell 2001
5 Copyright (C) Jelmer Vernooij 2002,2003
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 3 of the License, or
10 (at your option) any later version.
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.
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>.
24 * We have to use strcasecmp here as the character conversions
25 * haven't been initialised yet. JRA.
33 * @brief Samba wrapper/stub for iconv character set conversion.
35 * iconv is the XPG2 interface for converting between character
36 * encodings. This file provides a Samba wrapper around it, and also
37 * a simple reimplementation that is used if the system does not
40 * Samba only works with encodings that are supersets of ASCII: ascii
41 * characters like whitespace can be tested for directly, multibyte
42 * sequences start with a byte with the high bit set, and strings are
43 * terminated by a nul byte.
45 * Note that the only function provided by iconv is conversion between
46 * characters. It doesn't directly support operations like
47 * uppercasing or comparison. We have to convert to UCS-2 and compare
50 * @sa Samba Developers Guide
55 static size_t ascii_pull(void *,const char **, size_t *, char **, size_t *);
56 static size_t ascii_push(void *,const char **, size_t *, char **, size_t *);
57 static size_t latin1_push(void *,const char **, size_t *, char **, size_t *);
58 static size_t utf8_pull(void *,const char **, size_t *, char **, size_t *);
59 static size_t utf8_push(void *,const char **, size_t *, char **, size_t *);
60 static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
61 static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
62 static size_t iconv_copy(void *,const char **, size_t *, char **, size_t *);
63 static size_t iconv_swab (void *,const char **, size_t *, char **, size_t *);
65 static struct charset_functions builtin_functions[] = {
66 /* windows is really neither UCS-2 not UTF-16 */
67 {"UCS-2LE", iconv_copy, iconv_copy},
68 {"UTF-16LE", iconv_copy, iconv_copy},
69 {"UCS-2BE", iconv_swab, iconv_swab},
70 {"UTF-16BE", iconv_swab, iconv_swab},
72 /* we include the UTF-8 alias to cope with differing locale settings */
73 {"UTF8", utf8_pull, utf8_push},
74 {"UTF-8", utf8_pull, utf8_push},
75 {"ASCII", ascii_pull, ascii_push},
76 {"646", ascii_pull, ascii_push},
77 {"ISO-8859-1", ascii_pull, latin1_push},
78 {"UCS2-HEX", ucs2hex_pull, ucs2hex_push},
82 static struct charset_functions *charsets = NULL;
84 static struct charset_functions *find_charset_functions(const char *name)
86 struct charset_functions *c = charsets;
89 if (strcasecmp(name, c->name) == 0) {
98 NTSTATUS smb_register_charset(struct charset_functions *funcs)
101 return NT_STATUS_INVALID_PARAMETER;
104 DEBUG(5, ("Attempting to register new charset %s\n", funcs->name));
105 /* Check whether we already have this charset... */
106 if (find_charset_functions(funcs->name)) {
107 DEBUG(0, ("Duplicate charset %s, not registering\n", funcs->name));
108 return NT_STATUS_OBJECT_NAME_COLLISION;
111 funcs->next = funcs->prev = NULL;
112 DEBUG(5, ("Registered charset %s\n", funcs->name));
113 DLIST_ADD(charsets, funcs);
117 static void lazy_initialize_iconv(void)
119 static bool initialized;
124 for(i = 0; builtin_functions[i].name; i++)
125 smb_register_charset(&builtin_functions[i]);
130 #ifdef HAVE_NATIVE_ICONV
131 /* if there was an error then reset the internal state,
132 this ensures that we don't have a shift state remaining for
133 character sets like SJIS */
134 static size_t sys_iconv(void *cd,
135 const char **inbuf, size_t *inbytesleft,
136 char **outbuf, size_t *outbytesleft)
138 size_t ret = iconv((iconv_t)cd,
139 (void *)inbuf, inbytesleft,
140 outbuf, outbytesleft);
141 if (ret == (size_t)-1) {
142 int saved_errno = errno;
143 iconv(cd, NULL, NULL, NULL, NULL);
151 * This is a simple portable iconv() implementaion.
153 * It only knows about a very small number of character sets - just
154 * enough that Samba works on systems that don't have iconv.
156 size_t smb_iconv(smb_iconv_t cd,
157 const char **inbuf, size_t *inbytesleft,
158 char **outbuf, size_t *outbytesleft)
164 /* in many cases we can go direct */
166 return cd->direct(cd->cd_direct,
167 inbuf, inbytesleft, outbuf, outbytesleft);
171 /* otherwise we have to do it chunks at a time */
172 while (*inbytesleft > 0) {
174 bufsize = sizeof(cvtbuf);
176 if (cd->pull(cd->cd_pull,
177 inbuf, inbytesleft, &bufp, &bufsize) == -1
178 && errno != E2BIG) return -1;
181 bufsize = sizeof(cvtbuf) - bufsize;
183 if (cd->push(cd->cd_push,
184 (const char **)&bufp, &bufsize,
185 outbuf, outbytesleft) == -1) return -1;
192 static bool is_utf16(const char *name)
194 return strcasecmp(name, "UCS-2LE") == 0 ||
195 strcasecmp(name, "UTF-16LE") == 0;
199 simple iconv_open() wrapper
201 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
204 struct charset_functions *from, *to;
206 lazy_initialize_iconv();
210 ret = SMB_MALLOC_P(smb_iconv_t);
213 return (smb_iconv_t)-1;
215 memset(ret, 0, sizeof(smb_iconv_t));
217 ret->from_name = SMB_STRDUP(fromcode);
218 ret->to_name = SMB_STRDUP(tocode);
220 /* check for the simplest null conversion */
221 if (strcasecmp(fromcode, tocode) == 0) {
222 ret->direct = iconv_copy;
226 /* check if we have a builtin function for this conversion */
227 from = find_charset_functions(fromcode);
228 if(from)ret->pull = from->pull;
230 to = find_charset_functions(tocode);
231 if(to)ret->push = to->push;
233 /* check if we can use iconv for this conversion */
234 #ifdef HAVE_NATIVE_ICONV
236 ret->cd_pull = iconv_open("UTF-16LE", fromcode);
237 if (ret->cd_pull == (iconv_t)-1)
238 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
239 if (ret->cd_pull != (iconv_t)-1)
240 ret->pull = sys_iconv;
244 ret->cd_push = iconv_open(tocode, "UTF-16LE");
245 if (ret->cd_push == (iconv_t)-1)
246 ret->cd_push = iconv_open(tocode, "UCS-2LE");
247 if (ret->cd_push != (iconv_t)-1)
248 ret->push = sys_iconv;
252 /* check if there is a module available that can do this conversion */
253 if (!ret->pull && NT_STATUS_IS_OK(smb_probe_module("charset", fromcode))) {
254 if(!(from = find_charset_functions(fromcode)))
255 DEBUG(0, ("Module %s doesn't provide charset %s!\n", fromcode, fromcode));
257 ret->pull = from->pull;
260 if (!ret->push && NT_STATUS_IS_OK(smb_probe_module("charset", tocode))) {
261 if(!(to = find_charset_functions(tocode)))
262 DEBUG(0, ("Module %s doesn't provide charset %s!\n", tocode, tocode));
264 ret->push = to->push;
267 if (!ret->push || !ret->pull) {
268 SAFE_FREE(ret->from_name);
269 SAFE_FREE(ret->to_name);
272 return (smb_iconv_t)-1;
275 /* check for conversion to/from ucs2 */
276 if (is_utf16(fromcode) && to) {
277 ret->direct = to->push;
278 ret->push = ret->pull = NULL;
282 if (is_utf16(tocode) && from) {
283 ret->direct = from->pull;
284 ret->push = ret->pull = NULL;
288 /* Check if we can do the conversion direct */
289 #ifdef HAVE_NATIVE_ICONV
290 if (is_utf16(fromcode)) {
291 ret->direct = sys_iconv;
292 ret->cd_direct = ret->cd_push;
296 if (is_utf16(tocode)) {
297 ret->direct = sys_iconv;
298 ret->cd_direct = ret->cd_pull;
308 simple iconv_close() wrapper
310 int smb_iconv_close (smb_iconv_t cd)
312 #ifdef HAVE_NATIVE_ICONV
313 if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
314 if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
315 if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
318 SAFE_FREE(cd->from_name);
319 SAFE_FREE(cd->to_name);
321 memset(cd, 0, sizeof(*cd));
327 /**********************************************************************
328 the following functions implement the builtin character sets in Samba
329 and also the "test" character sets that are designed to test
330 multi-byte character set support for english users
331 ***********************************************************************/
333 static size_t ascii_pull(void *cd, const char **inbuf, size_t *inbytesleft,
334 char **outbuf, size_t *outbytesleft)
336 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
337 (*outbuf)[0] = (*inbuf)[0];
340 (*outbytesleft) -= 2;
345 if (*inbytesleft > 0) {
353 static size_t ascii_push(void *cd, const char **inbuf, size_t *inbytesleft,
354 char **outbuf, size_t *outbytesleft)
358 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
359 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
360 if ((*inbuf)[1]) ir_count++;
362 (*outbytesleft) -= 1;
367 if (*inbytesleft == 1) {
372 if (*inbytesleft > 1) {
380 static size_t latin1_push(void *cd, const char **inbuf, size_t *inbytesleft,
381 char **outbuf, size_t *outbytesleft)
385 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
386 (*outbuf)[0] = (*inbuf)[0];
387 if ((*inbuf)[1]) ir_count++;
389 (*outbytesleft) -= 1;
394 if (*inbytesleft == 1) {
399 if (*inbytesleft > 1) {
407 static size_t ucs2hex_pull(void *cd, const char **inbuf, size_t *inbytesleft,
408 char **outbuf, size_t *outbytesleft)
410 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
413 if ((*inbuf)[0] != '@') {
414 /* seven bit ascii case */
415 (*outbuf)[0] = (*inbuf)[0];
418 (*outbytesleft) -= 2;
423 /* it's a hex character */
424 if (*inbytesleft < 5) {
429 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
434 (*outbuf)[0] = v&0xff;
437 (*outbytesleft) -= 2;
442 if (*inbytesleft > 0) {
450 static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
451 char **outbuf, size_t *outbytesleft)
453 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
456 if ((*inbuf)[1] == 0 &&
457 ((*inbuf)[0] & 0x80) == 0 &&
458 (*inbuf)[0] != '@') {
459 (*outbuf)[0] = (*inbuf)[0];
461 (*outbytesleft) -= 1;
466 if (*outbytesleft < 5) {
470 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
471 memcpy(*outbuf, buf, 5);
473 (*outbytesleft) -= 5;
478 if (*inbytesleft == 1) {
483 if (*inbytesleft > 1) {
491 static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
492 char **outbuf, size_t *outbytesleft)
496 n = MIN(*inbytesleft, *outbytesleft);
498 swab(*inbuf, *outbuf, (n&~1));
504 (*outbytesleft) -= n;
508 if (*inbytesleft > 0) {
516 static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
517 char **outbuf, size_t *outbytesleft)
521 n = MIN(*inbytesleft, *outbytesleft);
523 memmove(*outbuf, *inbuf, n);
526 (*outbytesleft) -= n;
530 if (*inbytesleft > 0) {
538 static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
539 char **outbuf, size_t *outbytesleft)
541 size_t in_left=*inbytesleft, out_left=*outbytesleft;
542 const uint8 *c = (const uint8 *)*inbuf;
543 uint8 *uc = (uint8 *)*outbuf;
545 while (in_left >= 1 && out_left >= 2) {
546 unsigned int codepoint;
548 if ((c[0] & 0x80) == 0) {
558 if ((c[0] & 0xe0) == 0xc0) {
560 (c[1] & 0xc0) != 0x80) {
564 codepoint = (c[1]&0x3f) | ((c[0]&0x1f)<<6);
565 if (codepoint < 0x80) {
566 /* don't accept UTF-8 characters that are not minimally packed */
570 uc[1] = codepoint >> 8;
571 uc[0] = codepoint & 0xff;
579 if ((c[0] & 0xf0) == 0xe0) {
581 (c[1] & 0xc0) != 0x80 ||
582 (c[2] & 0xc0) != 0x80) {
586 codepoint = (c[2]&0x3f) | ((c[1]&0x3f)<<6) | ((c[0]&0xf)<<12);
587 if (codepoint < 0x800) {
588 /* don't accept UTF-8 characters that are not minimally packed */
592 uc[1] = codepoint >> 8;
593 uc[0] = codepoint & 0xff;
601 if ((c[0] & 0xf8) == 0xf0) {
603 (c[1] & 0xc0) != 0x80 ||
604 (c[2] & 0xc0) != 0x80 ||
605 (c[3] & 0xc0) != 0x80) {
614 if (codepoint < 0x10000 || codepoint > 0x10ffff) {
615 /* don't accept UTF-8 characters that are not minimally packed */
620 codepoint -= 0x10000;
627 uc[0] = (codepoint>>10) & 0xFF;
628 uc[1] = (codepoint>>18) | 0xd8;
629 uc[2] = codepoint & 0xFF;
630 uc[3] = ((codepoint>>8) & 0x3) | 0xdc;
638 /* we don't handle 5 byte sequences */
648 *inbytesleft = in_left;
649 *outbytesleft = out_left;
651 *outbuf = (char *)uc;
655 *inbytesleft = in_left;
656 *outbytesleft = out_left;
658 *outbuf = (char *)uc;
662 static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
663 char **outbuf, size_t *outbytesleft)
665 size_t in_left=*inbytesleft, out_left=*outbytesleft;
666 uint8 *c = (uint8 *)*outbuf;
667 const uint8 *uc = (const uint8 *)*inbuf;
669 while (in_left >= 2 && out_left >= 1) {
670 unsigned int codepoint;
672 if (uc[1] == 0 && !(uc[0] & 0x80)) {
682 if ((uc[1]&0xf8) == 0) {
683 /* next simplest case */
688 c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
689 c[1] = 0x80 | (uc[0] & 0x3f);
697 if ((uc[1] & 0xfc) == 0xdc) {
698 /* its the second part of a 4 byte sequence. Illegal */
707 if ((uc[1] & 0xfc) != 0xd8) {
708 codepoint = uc[0] | (uc[1]<<8);
713 c[0] = 0xe0 | (codepoint >> 12);
714 c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
715 c[2] = 0x80 | (codepoint & 0x3f);
724 /* its the first part of a 4 byte sequence */
729 if ((uc[3] & 0xfc) != 0xdc) {
733 codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) |
734 (uc[0]<<10) | ((uc[1] & 0x3)<<18));
740 c[0] = 0xf0 | (codepoint >> 18);
741 c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
742 c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
743 c[3] = 0x80 | (codepoint & 0x3f);
761 *inbytesleft = in_left;
762 *outbytesleft = out_left;
769 *inbytesleft = in_left;
770 *outbytesleft = out_left;