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 2 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, write to the Free Software
19 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
25 * We have to use strcasecmp here as the character conversions
26 * haven't been initialised yet. JRA.
34 * @brief Samba wrapper/stub for iconv character set conversion.
36 * iconv is the XPG2 interface for converting between character
37 * encodings. This file provides a Samba wrapper around it, and also
38 * a simple reimplementation that is used if the system does not
41 * Samba only works with encodings that are supersets of ASCII: ascii
42 * characters like whitespace can be tested for directly, multibyte
43 * sequences start with a byte with the high bit set, and strings are
44 * terminated by a nul byte.
46 * Note that the only function provided by iconv is conversion between
47 * characters. It doesn't directly support operations like
48 * uppercasing or comparison. We have to convert to UCS-2 and compare
51 * @sa Samba Developers Guide
54 static size_t ascii_pull(void *,const char **, size_t *, char **, size_t *);
55 static size_t ascii_push(void *,const char **, size_t *, char **, size_t *);
56 static size_t latin1_push(void *,const char **, size_t *, char **, size_t *);
57 static size_t utf8_pull(void *,const char **, size_t *, char **, size_t *);
58 static size_t utf8_push(void *,const char **, size_t *, char **, size_t *);
59 static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
60 static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
61 static size_t iconv_copy(void *,const char **, size_t *, char **, size_t *);
62 static size_t iconv_swab (void *,const char **, size_t *, char **, size_t *);
64 static struct charset_functions builtin_functions[] = {
65 /* windows is really neither UCS-2 not UTF-16 */
66 {"UCS-2LE", iconv_copy, iconv_copy},
67 {"UTF-16LE", iconv_copy, iconv_copy},
68 {"UCS-2BE", iconv_swab, iconv_swab},
69 {"UTF-16BE", iconv_swab, iconv_swab},
71 /* we include the UTF-8 alias to cope with differing locale settings */
72 {"UTF8", utf8_pull, utf8_push},
73 {"UTF-8", utf8_pull, utf8_push},
74 {"ASCII", ascii_pull, ascii_push},
75 {"646", ascii_pull, ascii_push},
76 {"ISO-8859-1", ascii_pull, latin1_push},
77 {"UCS2-HEX", ucs2hex_pull, ucs2hex_push},
81 static struct charset_functions *charsets = NULL;
83 static struct charset_functions *find_charset_functions(const char *name)
85 struct charset_functions *c = charsets;
88 if (strcasecmp(name, c->name) == 0) {
97 NTSTATUS smb_register_charset(struct charset_functions *funcs)
100 return NT_STATUS_INVALID_PARAMETER;
103 DEBUG(5, ("Attempting to register new charset %s\n", funcs->name));
104 /* Check whether we already have this charset... */
105 if (find_charset_functions(funcs->name)) {
106 DEBUG(0, ("Duplicate charset %s, not registering\n", funcs->name));
107 return NT_STATUS_OBJECT_NAME_COLLISION;
110 funcs->next = funcs->prev = NULL;
111 DEBUG(5, ("Registered charset %s\n", funcs->name));
112 DLIST_ADD(charsets, funcs);
116 static void lazy_initialize_iconv(void)
118 static BOOL initialized;
123 for(i = 0; builtin_functions[i].name; i++)
124 smb_register_charset(&builtin_functions[i]);
129 /* if there was an error then reset the internal state,
130 this ensures that we don't have a shift state remaining for
131 character sets like SJIS */
132 static size_t sys_iconv(void *cd,
133 const char **inbuf, size_t *inbytesleft,
134 char **outbuf, size_t *outbytesleft)
136 #ifdef HAVE_NATIVE_ICONV
137 size_t ret = iconv((iconv_t)cd,
138 (char **)inbuf, inbytesleft,
139 outbuf, outbytesleft);
140 if (ret == (size_t)-1) {
141 int saved_errno = errno;
142 iconv(cd, NULL, NULL, NULL, NULL);
153 * This is a simple portable iconv() implementaion.
155 * It only knows about a very small number of character sets - just
156 * enough that Samba works on systems that don't have iconv.
158 size_t smb_iconv(smb_iconv_t cd,
159 const char **inbuf, size_t *inbytesleft,
160 char **outbuf, size_t *outbytesleft)
166 /* in many cases we can go direct */
168 return cd->direct(cd->cd_direct,
169 inbuf, inbytesleft, outbuf, outbytesleft);
173 /* otherwise we have to do it chunks at a time */
174 while (*inbytesleft > 0) {
176 bufsize = sizeof(cvtbuf);
178 if (cd->pull(cd->cd_pull,
179 inbuf, inbytesleft, &bufp, &bufsize) == -1
180 && errno != E2BIG) return -1;
183 bufsize = sizeof(cvtbuf) - bufsize;
185 if (cd->push(cd->cd_push,
186 (const char **)&bufp, &bufsize,
187 outbuf, outbytesleft) == -1) return -1;
194 static BOOL is_utf16(const char *name)
196 return strcasecmp(name, "UCS-2LE") == 0 ||
197 strcasecmp(name, "UTF-16LE") == 0;
201 simple iconv_open() wrapper
203 smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
206 struct charset_functions *from, *to;
208 lazy_initialize_iconv();
212 ret = SMB_MALLOC_P(struct _smb_iconv_t);
215 return (smb_iconv_t)-1;
217 memset(ret, 0, sizeof(struct _smb_iconv_t));
219 ret->from_name = SMB_STRDUP(fromcode);
220 ret->to_name = SMB_STRDUP(tocode);
222 /* check for the simplest null conversion */
223 if (strcasecmp(fromcode, tocode) == 0) {
224 ret->direct = iconv_copy;
228 /* check if we have a builtin function for this conversion */
229 from = find_charset_functions(fromcode);
230 if(from)ret->pull = from->pull;
232 to = find_charset_functions(tocode);
233 if(to)ret->push = to->push;
235 /* check if we can use iconv for this conversion */
236 #ifdef HAVE_NATIVE_ICONV
238 ret->cd_pull = iconv_open("UTF-16LE", fromcode);
239 if (ret->cd_pull == (iconv_t)-1)
240 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
241 if (ret->cd_pull != (iconv_t)-1)
242 ret->pull = sys_iconv;
246 ret->cd_push = iconv_open(tocode, "UTF-16LE");
247 if (ret->cd_push == (iconv_t)-1)
248 ret->cd_push = iconv_open(tocode, "UCS-2LE");
249 if (ret->cd_push != (iconv_t)-1)
250 ret->push = sys_iconv;
254 /* check if there is a module available that can do this conversion */
255 if (!ret->pull && NT_STATUS_IS_OK(smb_probe_module("charset", fromcode))) {
256 if(!(from = find_charset_functions(fromcode)))
257 DEBUG(0, ("Module %s doesn't provide charset %s!\n", fromcode, fromcode));
259 ret->pull = from->pull;
262 if (!ret->push && NT_STATUS_IS_OK(smb_probe_module("charset", tocode))) {
263 if(!(to = find_charset_functions(tocode)))
264 DEBUG(0, ("Module %s doesn't provide charset %s!\n", tocode, tocode));
266 ret->push = to->push;
269 if (!ret->push || !ret->pull) {
270 SAFE_FREE(ret->from_name);
271 SAFE_FREE(ret->to_name);
274 return (smb_iconv_t)-1;
277 /* check for conversion to/from ucs2 */
278 if (is_utf16(fromcode) && to) {
279 ret->direct = to->push;
280 ret->push = ret->pull = NULL;
284 if (is_utf16(tocode) && from) {
285 ret->direct = from->pull;
286 ret->push = ret->pull = NULL;
290 /* Check if we can do the conversion direct */
291 #ifdef HAVE_NATIVE_ICONV
292 if (is_utf16(fromcode)) {
293 ret->direct = sys_iconv;
294 ret->cd_direct = ret->cd_push;
298 if (is_utf16(tocode)) {
299 ret->direct = sys_iconv;
300 ret->cd_direct = ret->cd_pull;
310 simple iconv_close() wrapper
312 int smb_iconv_close (smb_iconv_t cd)
314 #ifdef HAVE_NATIVE_ICONV
315 if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
316 if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
317 if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
320 SAFE_FREE(cd->from_name);
321 SAFE_FREE(cd->to_name);
323 memset(cd, 0, sizeof(*cd));
329 /**********************************************************************
330 the following functions implement the builtin character sets in Samba
331 and also the "test" character sets that are designed to test
332 multi-byte character set support for english users
333 ***********************************************************************/
335 static size_t ascii_pull(void *cd, const char **inbuf, size_t *inbytesleft,
336 char **outbuf, size_t *outbytesleft)
338 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
339 (*outbuf)[0] = (*inbuf)[0];
342 (*outbytesleft) -= 2;
347 if (*inbytesleft > 0) {
355 static size_t ascii_push(void *cd, const char **inbuf, size_t *inbytesleft,
356 char **outbuf, size_t *outbytesleft)
360 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
361 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
362 if ((*inbuf)[1]) ir_count++;
364 (*outbytesleft) -= 1;
369 if (*inbytesleft == 1) {
374 if (*inbytesleft > 1) {
382 static size_t latin1_push(void *cd, const char **inbuf, size_t *inbytesleft,
383 char **outbuf, size_t *outbytesleft)
387 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
388 (*outbuf)[0] = (*inbuf)[0];
389 if ((*inbuf)[1]) ir_count++;
391 (*outbytesleft) -= 1;
396 if (*inbytesleft == 1) {
401 if (*inbytesleft > 1) {
409 static size_t ucs2hex_pull(void *cd, const char **inbuf, size_t *inbytesleft,
410 char **outbuf, size_t *outbytesleft)
412 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
415 if ((*inbuf)[0] != '@') {
416 /* seven bit ascii case */
417 (*outbuf)[0] = (*inbuf)[0];
420 (*outbytesleft) -= 2;
425 /* it's a hex character */
426 if (*inbytesleft < 5) {
431 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
436 (*outbuf)[0] = v&0xff;
439 (*outbytesleft) -= 2;
444 if (*inbytesleft > 0) {
452 static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
453 char **outbuf, size_t *outbytesleft)
455 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
458 if ((*inbuf)[1] == 0 &&
459 ((*inbuf)[0] & 0x80) == 0 &&
460 (*inbuf)[0] != '@') {
461 (*outbuf)[0] = (*inbuf)[0];
463 (*outbytesleft) -= 1;
468 if (*outbytesleft < 5) {
472 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
473 memcpy(*outbuf, buf, 5);
475 (*outbytesleft) -= 5;
480 if (*inbytesleft == 1) {
485 if (*inbytesleft > 1) {
493 static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
494 char **outbuf, size_t *outbytesleft)
498 n = MIN(*inbytesleft, *outbytesleft);
500 swab(*inbuf, *outbuf, (n&~1));
506 (*outbytesleft) -= n;
510 if (*inbytesleft > 0) {
518 static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
519 char **outbuf, size_t *outbytesleft)
523 n = MIN(*inbytesleft, *outbytesleft);
525 memmove(*outbuf, *inbuf, n);
528 (*outbytesleft) -= n;
532 if (*inbytesleft > 0) {
540 static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
541 char **outbuf, size_t *outbytesleft)
543 size_t in_left=*inbytesleft, out_left=*outbytesleft;
544 const uint8 *c = (const uint8 *)*inbuf;
545 uint8 *uc = (uint8 *)*outbuf;
547 while (in_left >= 1 && out_left >= 2) {
548 if ((c[0] & 0x80) == 0) {
558 if ((c[0] & 0xe0) == 0xc0) {
560 (c[1] & 0xc0) != 0x80) {
564 uc[1] = (c[0]>>2) & 0x7;
565 uc[0] = (c[0]<<6) | (c[1]&0x3f);
573 if ((c[0] & 0xf0) == 0xe0) {
575 (c[1] & 0xc0) != 0x80 ||
576 (c[2] & 0xc0) != 0x80) {
580 uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
581 uc[0] = (c[1]<<6) | (c[2]&0x3f);
589 if ((c[0] & 0xf8) == 0xf0) {
590 unsigned int codepoint;
592 (c[1] & 0xc0) != 0x80 ||
593 (c[2] & 0xc0) != 0x80 ||
594 (c[3] & 0xc0) != 0x80) {
603 if (codepoint < 0x10000) {
604 /* accept UTF-8 characters that are not
605 minimally packed, but pack the result */
606 uc[0] = (codepoint & 0xFF);
607 uc[1] = (codepoint >> 8);
615 codepoint -= 0x10000;
622 uc[0] = (codepoint>>10) & 0xFF;
623 uc[1] = (codepoint>>18) | 0xd8;
624 uc[2] = codepoint & 0xFF;
625 uc[3] = ((codepoint>>8) & 0x3) | 0xdc;
633 /* we don't handle 5 byte sequences */
643 *inbytesleft = in_left;
644 *outbytesleft = out_left;
650 *inbytesleft = in_left;
651 *outbytesleft = out_left;
657 static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
658 char **outbuf, size_t *outbytesleft)
660 size_t in_left=*inbytesleft, out_left=*outbytesleft;
661 uint8 *c = (uint8 *)*outbuf;
662 const uint8 *uc = (const uint8 *)*inbuf;
664 while (in_left >= 2 && out_left >= 1) {
665 unsigned int codepoint;
667 if (uc[1] == 0 && !(uc[0] & 0x80)) {
677 if ((uc[1]&0xf8) == 0) {
678 /* next simplest case */
683 c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
684 c[1] = 0x80 | (uc[0] & 0x3f);
692 if ((uc[1] & 0xfc) == 0xdc) {
693 /* its the second part of a 4 byte sequence. Illegal */
702 if ((uc[1] & 0xfc) != 0xd8) {
703 codepoint = uc[0] | (uc[1]<<8);
708 c[0] = 0xe0 | (codepoint >> 12);
709 c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
710 c[2] = 0x80 | (codepoint & 0x3f);
719 /* its the first part of a 4 byte sequence */
724 if ((uc[3] & 0xfc) != 0xdc) {
728 codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) |
729 (uc[0]<<10) | ((uc[1] & 0x3)<<18));
735 c[0] = 0xf0 | (codepoint >> 18);
736 c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
737 c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
738 c[3] = 0x80 | (codepoint & 0x3f);
756 *inbytesleft = in_left;
757 *outbytesleft = out_left;
764 *inbytesleft = in_left;
765 *outbytesleft = out_left;