2 Unix SMB/CIFS implementation.
3 Character set conversion Extensions
4 Copyright (C) Igor Vergeichik <iverg@mail.ru> 2001
5 Copyright (C) Andrew Tridgell 2001
6 Copyright (C) Simo Sorce 2001
7 Copyright (C) Martin Pool 2003
9 This program is free software; you can redistribute it and/or modify
10 it under the terms of the GNU General Public License as published by
11 the Free Software Foundation; either version 3 of the License, or
12 (at your option) any later version.
14 This program is distributed in the hope that it will be useful,
15 but WITHOUT ANY WARRANTY; without even the implied warranty of
16 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 GNU General Public License for more details.
19 You should have received a copy of the GNU General Public License
20 along with this program. If not, see <http://www.gnu.org/licenses/>.
25 /* We can parameterize this if someone complains.... JRA. */
27 char lp_failed_convert_char(void)
35 * @brief Character-set conversion routines built on our iconv.
37 * @note Samba's internal character set (at least in the 3.0 series)
38 * is always the same as the one for the Unix filesystem. It is
39 * <b>not</b> necessarily UTF-8 and may be different on machines that
40 * need i18n filenames to be compatible with Unix software. It does
41 * have to be a superset of ASCII. All multibyte sequences must start
42 * with a byte with the high bit set.
48 static smb_iconv_t conv_handles[NUM_CHARSETS][NUM_CHARSETS];
49 static bool conv_silent; /* Should we do a debug if the conversion fails ? */
52 * Return the name of a charset to give to iconv().
54 static const char *charset_name(charset_t ch)
56 const char *ret = NULL;
58 if (ch == CH_UTF16LE) ret = "UTF-16LE";
59 else if (ch == CH_UTF16BE) ret = "UTF-16BE";
60 else if (ch == CH_UNIX) ret = lp_unix_charset();
61 else if (ch == CH_DOS) ret = lp_dos_charset();
62 else if (ch == CH_DISPLAY) ret = lp_display_charset();
63 else if (ch == CH_UTF8) ret = "UTF8";
65 #if defined(HAVE_NL_LANGINFO) && defined(CODESET)
66 if (ret && !strcmp(ret, "LOCALE")) {
67 const char *ln = NULL;
70 setlocale(LC_ALL, "");
72 ln = nl_langinfo(CODESET);
74 /* Check whether the charset name is supported
76 smb_iconv_t handle = smb_iconv_open(ln,"UCS-2LE");
77 if (handle == (smb_iconv_t) -1) {
78 DEBUG(5,("Locale charset '%s' unsupported, using ASCII instead\n", ln));
81 DEBUG(5,("Substituting charset '%s' for LOCALE\n", ln));
82 smb_iconv_close(handle);
89 if (!ret || !*ret) ret = "ASCII";
93 void lazy_initialize_conv(void)
95 static int initialized = False;
105 * Destroy global objects allocated by init_iconv()
107 void gfree_charcnv(void)
111 for (c1=0;c1<NUM_CHARSETS;c1++) {
112 for (c2=0;c2<NUM_CHARSETS;c2++) {
113 if ( conv_handles[c1][c2] ) {
114 smb_iconv_close( conv_handles[c1][c2] );
115 conv_handles[c1][c2] = 0;
122 * Initialize iconv conversion descriptors.
124 * This is called the first time it is needed, and also called again
125 * every time the configuration is reloaded, because the charset or
126 * codepage might have changed.
128 void init_iconv(void)
131 bool did_reload = False;
133 /* so that charset_name() works we need to get the UNIX<->UCS2 going
135 if (!conv_handles[CH_UNIX][CH_UTF16LE])
136 conv_handles[CH_UNIX][CH_UTF16LE] = smb_iconv_open(charset_name(CH_UTF16LE), "ASCII");
138 if (!conv_handles[CH_UTF16LE][CH_UNIX])
139 conv_handles[CH_UTF16LE][CH_UNIX] = smb_iconv_open("ASCII", charset_name(CH_UTF16LE));
141 for (c1=0;c1<NUM_CHARSETS;c1++) {
142 for (c2=0;c2<NUM_CHARSETS;c2++) {
143 const char *n1 = charset_name((charset_t)c1);
144 const char *n2 = charset_name((charset_t)c2);
145 if (conv_handles[c1][c2] &&
146 strcmp(n1, conv_handles[c1][c2]->from_name) == 0 &&
147 strcmp(n2, conv_handles[c1][c2]->to_name) == 0)
152 if (conv_handles[c1][c2])
153 smb_iconv_close(conv_handles[c1][c2]);
155 conv_handles[c1][c2] = smb_iconv_open(n2,n1);
156 if (conv_handles[c1][c2] == (smb_iconv_t)-1) {
157 DEBUG(0,("init_iconv: Conversion from %s to %s not supported\n",
158 charset_name((charset_t)c1), charset_name((charset_t)c2)));
159 if (c1 != CH_UTF16LE && c1 != CH_UTF16BE) {
162 if (c2 != CH_UTF16LE && c2 != CH_UTF16BE) {
165 DEBUG(0,("init_iconv: Attempting to replace with conversion from %s to %s\n",
167 conv_handles[c1][c2] = smb_iconv_open(n2,n1);
168 if (!conv_handles[c1][c2]) {
169 DEBUG(0,("init_iconv: Conversion from %s to %s failed", n1, n2));
170 smb_panic("init_iconv: conv_handle initialization failed");
177 /* XXX: Does this really get called every time the dos
178 * codepage changes? */
179 /* XXX: Is the did_reload test too strict? */
181 init_doschar_table();
188 * Convert string from one encoding to another, making error checking etc
189 * Slow path version - uses (slow) iconv.
191 * @param src pointer to source string (multibyte or singlebyte)
192 * @param srclen length of the source string in bytes
193 * @param dest pointer to destination string (multibyte or singlebyte)
194 * @param destlen maximal length allowed for string
195 * @param allow_bad_conv determines if a "best effort" conversion is acceptable (never returns errors)
196 * @returns the number of bytes occupied in the destination
198 * Ensure the srclen contains the terminating zero.
202 static size_t convert_string_internal(charset_t from, charset_t to,
203 void const *src, size_t srclen,
204 void *dest, size_t destlen, bool allow_bad_conv)
208 const char* inbuf = (const char*)src;
209 char* outbuf = (char*)dest;
210 smb_iconv_t descriptor;
212 lazy_initialize_conv();
214 descriptor = conv_handles[from][to];
216 if (srclen == (size_t)-1) {
217 if (from == CH_UTF16LE || from == CH_UTF16BE) {
218 srclen = (strlen_w((const smb_ucs2_t *)src)+1) * 2;
220 srclen = strlen((const char *)src)+1;
225 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
227 DEBUG(0,("convert_string_internal: Conversion not supported.\n"));
236 retval = smb_iconv(descriptor, &inbuf, &i_len, &outbuf, &o_len);
237 if(retval==(size_t)-1) {
238 const char *reason="unknown error";
241 reason="Incomplete multibyte sequence";
243 DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
248 reason="No more room";
250 if (from == CH_UNIX) {
251 DEBUG(3,("E2BIG: convert_string(%s,%s): srclen=%u destlen=%u - '%s'\n",
252 charset_name(from), charset_name(to),
253 (unsigned int)srclen, (unsigned int)destlen, (const char *)src));
255 DEBUG(3,("E2BIG: convert_string(%s,%s): srclen=%u destlen=%u\n",
256 charset_name(from), charset_name(to),
257 (unsigned int)srclen, (unsigned int)destlen));
262 reason="Illegal multibyte sequence";
264 DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
270 DEBUG(0,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
273 /* smb_panic(reason); */
275 return destlen-o_len;
280 * Conversion not supported. This is actually an error, but there are so
281 * many misconfigured iconv systems and smb.conf's out there we can't just
282 * fail. Do a very bad conversion instead.... JRA.
286 if (o_len == 0 || i_len == 0)
287 return destlen - o_len;
289 if (((from == CH_UTF16LE)||(from == CH_UTF16BE)) &&
290 ((to != CH_UTF16LE)||(to != CH_UTF16BE))) {
291 /* Can't convert from utf16 any endian to multibyte.
292 Replace with the default fail char.
295 return destlen - o_len;
297 *outbuf = lp_failed_convert_char();
306 if (o_len == 0 || i_len == 0)
307 return destlen - o_len;
309 /* Keep trying with the next char... */
312 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
313 /* Can't convert to UTF16LE - just widen by adding the
314 default fail char then zero.
317 return destlen - o_len;
319 outbuf[0] = lp_failed_convert_char();
328 if (o_len == 0 || i_len == 0)
329 return destlen - o_len;
331 /* Keep trying with the next char... */
334 } else if (from != CH_UTF16LE && from != CH_UTF16BE &&
335 to != CH_UTF16LE && to != CH_UTF16BE) {
336 /* Failed multibyte to multibyte. Just copy the default fail char and
338 outbuf[0] = lp_failed_convert_char();
346 if (o_len == 0 || i_len == 0)
347 return destlen - o_len;
349 /* Keep trying with the next char... */
353 /* Keep compiler happy.... */
354 return destlen - o_len;
360 * Convert string from one encoding to another, making error checking etc
361 * Fast path version - handles ASCII first.
363 * @param src pointer to source string (multibyte or singlebyte)
364 * @param srclen length of the source string in bytes, or -1 for nul terminated.
365 * @param dest pointer to destination string (multibyte or singlebyte)
366 * @param destlen maximal length allowed for string - *NEVER* -1.
367 * @param allow_bad_conv determines if a "best effort" conversion is acceptable (never returns errors)
368 * @returns the number of bytes occupied in the destination
370 * Ensure the srclen contains the terminating zero.
372 * This function has been hand-tuned to provide a fast path.
373 * Don't change unless you really know what you are doing. JRA.
376 size_t convert_string(charset_t from, charset_t to,
377 void const *src, size_t srclen,
378 void *dest, size_t destlen, bool allow_bad_conv)
381 * NB. We deliberately don't do a strlen here if srclen == -1.
382 * This is very expensive over millions of calls and is taken
383 * care of in the slow path in convert_string_internal. JRA.
387 SMB_ASSERT(destlen != (size_t)-1);
393 if (from != CH_UTF16LE && from != CH_UTF16BE && to != CH_UTF16LE && to != CH_UTF16BE) {
394 const unsigned char *p = (const unsigned char *)src;
395 unsigned char *q = (unsigned char *)dest;
396 size_t slen = srclen;
397 size_t dlen = destlen;
398 unsigned char lastp = '\0';
401 /* If all characters are ascii, fast path here. */
402 while (slen && dlen) {
403 if ((lastp = *p) <= 0x7f) {
405 if (slen != (size_t)-1) {
413 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
416 return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
421 /* Even if we fast path we should note if we ran out of room. */
422 if (((slen != (size_t)-1) && slen) ||
423 ((slen == (size_t)-1) && lastp)) {
428 } else if (from == CH_UTF16LE && to != CH_UTF16LE) {
429 const unsigned char *p = (const unsigned char *)src;
430 unsigned char *q = (unsigned char *)dest;
432 size_t slen = srclen;
433 size_t dlen = destlen;
434 unsigned char lastp = '\0';
436 /* If all characters are ascii, fast path here. */
437 while (((slen == (size_t)-1) || (slen >= 2)) && dlen) {
438 if (((lastp = *p) <= 0x7f) && (p[1] == 0)) {
440 if (slen != (size_t)-1) {
449 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
452 return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
457 /* Even if we fast path we should note if we ran out of room. */
458 if (((slen != (size_t)-1) && slen) ||
459 ((slen == (size_t)-1) && lastp)) {
464 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
465 const unsigned char *p = (const unsigned char *)src;
466 unsigned char *q = (unsigned char *)dest;
468 size_t slen = srclen;
469 size_t dlen = destlen;
470 unsigned char lastp = '\0';
472 /* If all characters are ascii, fast path here. */
473 while (slen && (dlen >= 2)) {
474 if ((lastp = *p) <= 0x7F) {
477 if (slen != (size_t)-1) {
485 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
488 return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
493 /* Even if we fast path we should note if we ran out of room. */
494 if (((slen != (size_t)-1) && slen) ||
495 ((slen == (size_t)-1) && lastp)) {
502 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
505 return convert_string_internal(from, to, src, srclen, dest, destlen, allow_bad_conv);
509 * Convert between character sets, allocating a new buffer for the result.
511 * @param ctx TALLOC_CTX to use to allocate with. If NULL use malloc.
512 * (this is a bad interface and needs fixing. JRA).
513 * @param srclen length of source buffer.
514 * @param dest always set at least to NULL
515 * @note -1 is not accepted for srclen.
517 * @returns Size in bytes of the converted string; or -1 in case of error.
519 * Ensure the srclen contains the terminating zero.
521 * I hate the goto's in this function. It's embarressing.....
522 * There has to be a cleaner way to do this. JRA.
525 size_t convert_string_allocate(TALLOC_CTX *ctx, charset_t from, charset_t to,
526 void const *src, size_t srclen, void *dst, bool allow_bad_conv)
528 size_t i_len, o_len, destlen = MAX(srclen, 512);
530 const char *inbuf = (const char *)src;
531 char *outbuf = NULL, *ob = NULL;
532 smb_iconv_t descriptor;
533 void **dest = (void **)dst;
537 if (src == NULL || srclen == (size_t)-1)
542 lazy_initialize_conv();
544 descriptor = conv_handles[from][to];
546 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
548 DEBUG(0,("convert_string_allocate: Conversion not supported.\n"));
554 if ((destlen*2) < destlen) {
555 /* wrapped ! abort. */
557 DEBUG(0, ("convert_string_allocate: destlen wrapped !\n"));
562 destlen = destlen * 2;
566 ob = (char *)TALLOC_REALLOC(ctx, ob, destlen);
568 ob = (char *)SMB_REALLOC(ob, destlen);
572 DEBUG(0, ("convert_string_allocate: realloc failed!\n"));
581 retval = smb_iconv(descriptor,
584 if(retval == (size_t)-1) {
585 const char *reason="unknown error";
588 reason="Incomplete multibyte sequence";
590 DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
597 reason="Illegal multibyte sequence";
599 DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
605 DEBUG(0,("Conversion error: %s(%s)\n",reason,inbuf));
606 /* smb_panic(reason); */
617 destlen = destlen - o_len;
619 ob = (char *)TALLOC_REALLOC(ctx,ob,destlen);
621 ob = (char *)SMB_REALLOC(ob,destlen);
624 if (destlen && !ob) {
625 DEBUG(0, ("convert_string_allocate: out of memory!\n"));
635 * Conversion not supported. This is actually an error, but there are so
636 * many misconfigured iconv systems and smb.conf's out there we can't just
637 * fail. Do a very bad conversion instead.... JRA.
641 if (o_len == 0 || i_len == 0)
644 if (((from == CH_UTF16LE)||(from == CH_UTF16BE)) &&
645 ((to != CH_UTF16LE)||(to != CH_UTF16BE))) {
646 /* Can't convert from utf16 any endian to multibyte.
647 Replace with the default fail char.
654 *outbuf = lp_failed_convert_char();
663 if (o_len == 0 || i_len == 0)
666 /* Keep trying with the next char... */
669 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
670 /* Can't convert to UTF16LE - just widen by adding the
671 default fail char then zero.
676 outbuf[0] = lp_failed_convert_char();
685 if (o_len == 0 || i_len == 0)
688 /* Keep trying with the next char... */
691 } else if (from != CH_UTF16LE && from != CH_UTF16BE &&
692 to != CH_UTF16LE && to != CH_UTF16BE) {
693 /* Failed multibyte to multibyte. Just copy the default fail char and
695 outbuf[0] = lp_failed_convert_char();
703 if (o_len == 0 || i_len == 0)
706 /* Keep trying with the next char... */
710 /* Keep compiler happy.... */
717 * Convert between character sets, allocating a new buffer using talloc for the result.
719 * @param srclen length of source buffer.
720 * @param dest always set at least to NULL
721 * @note -1 is not accepted for srclen.
723 * @returns Size in bytes of the converted string; or -1 in case of error.
725 size_t convert_string_talloc(TALLOC_CTX *ctx, charset_t from, charset_t to,
726 void const *src, size_t srclen, void *dst,
729 void **dest = (void **)dst;
733 dest_len=convert_string_allocate(ctx, from, to, src, srclen, dest, allow_bad_conv);
734 if (dest_len == (size_t)-1)
741 size_t unix_strupper(const char *src, size_t srclen, char *dest, size_t destlen)
746 size = push_ucs2_allocate(&buffer, src);
747 if (size == (size_t)-1) {
750 if (!strupper_w(buffer) && (dest == src)) {
755 size = convert_string(CH_UTF16LE, CH_UNIX, buffer, size, dest, destlen, True);
761 strdup() a unix string to upper case.
764 char *strdup_upper(const char *s)
766 char *out_buffer = SMB_STRDUP(s);
767 const unsigned char *p = (const unsigned char *)s;
768 unsigned char *q = (unsigned char *)out_buffer;
774 /* this is quite a common operation, so we want it to be
775 fast. We optimise for the ascii case, knowing that all our
776 supported multi-byte character sets are ascii-compatible
777 (ie. they match for the first 128 chars) */
782 *q++ = toupper_ascii(*p);
789 smb_ucs2_t *buffer = NULL;
791 SAFE_FREE(out_buffer);
792 size = convert_string_allocate(NULL,
797 (void **)(void *)&buffer,
799 if (size == (size_t)-1) {
805 size = convert_string_allocate(NULL,
810 (void **)(void *)&out_buffer,
813 /* Don't need the intermediate buffer
818 if (size == (size_t)-1) {
827 talloc_strdup() a unix string to upper case.
830 char *talloc_strdup_upper(TALLOC_CTX *ctx, const char *s)
832 char *out_buffer = talloc_strdup(ctx,s);
833 const unsigned char *p = (const unsigned char *)s;
834 unsigned char *q = (unsigned char *)out_buffer;
840 /* this is quite a common operation, so we want it to be
841 fast. We optimise for the ascii case, knowing that all our
842 supported multi-byte character sets are ascii-compatible
843 (ie. they match for the first 128 chars) */
848 *q++ = toupper_ascii(*p);
855 smb_ucs2_t *ubuf = NULL;
857 /* We're not using the ascii buffer above. */
858 TALLOC_FREE(out_buffer);
860 size = convert_string_talloc(ctx, CH_UNIX, CH_UTF16LE,
864 if (size == (size_t)-1) {
870 size = convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX,
875 /* Don't need the intermediate buffer
881 if (size == (size_t)-1) {
889 size_t unix_strlower(const char *src, size_t srclen, char *dest, size_t destlen)
892 smb_ucs2_t *buffer = NULL;
894 size = convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, src, srclen,
895 (void **)(void *)&buffer, True);
896 if (size == (size_t)-1 || !buffer) {
897 smb_panic("failed to create UCS2 buffer");
899 if (!strlower_w(buffer) && (dest == src)) {
903 size = convert_string(CH_UTF16LE, CH_UNIX, buffer, size, dest, destlen, True);
909 strdup() a unix string to lower case.
912 char *strdup_lower(const char *s)
915 smb_ucs2_t *buffer = NULL;
918 size = push_ucs2_allocate(&buffer, s);
919 if (size == -1 || !buffer) {
925 size = pull_ucs2_allocate(&out_buffer, buffer);
928 if (size == (size_t)-1) {
935 size_t ucs2_align(const void *base_ptr, const void *p, int flags)
937 if (flags & (STR_NOALIGN|STR_ASCII))
939 return PTR_DIFF(p, base_ptr) & 1;
944 * Copy a string from a char* unix src to a dos codepage string destination.
946 * @return the number of bytes occupied by the string in the destination.
948 * @param flags can include
950 * <dt>STR_TERMINATE</dt> <dd>means include the null termination</dd>
951 * <dt>STR_UPPER</dt> <dd>means uppercase in the destination</dd>
954 * @param dest_len the maximum length in bytes allowed in the
957 size_t push_ascii(void *dest, const char *src, size_t dest_len, int flags)
959 size_t src_len = strlen(src);
963 /* No longer allow a length of -1. */
964 if (dest_len == (size_t)-1) {
965 smb_panic("push_ascii - dest_len == -1");
968 if (flags & STR_UPPER) {
969 tmpbuf = SMB_STRDUP(src);
971 smb_panic("malloc fail");
977 if (flags & (STR_TERMINATE | STR_TERMINATE_ASCII)) {
981 ret = convert_string(CH_UNIX, CH_DOS, src, src_len, dest, dest_len, True);
982 if (ret == (size_t)-1 &&
983 (flags & (STR_TERMINATE | STR_TERMINATE_ASCII))
985 ((char *)dest)[0] = '\0';
991 size_t push_ascii_fstring(void *dest, const char *src)
993 return push_ascii(dest, src, sizeof(fstring), STR_TERMINATE);
996 /********************************************************************
997 Push an nstring - ensure null terminated. Written by
998 moriyama@miraclelinux.com (MORIYAMA Masayuki).
999 ********************************************************************/
1001 size_t push_ascii_nstring(void *dest, const char *src)
1003 size_t i, buffer_len, dest_len;
1007 buffer_len = push_ucs2_allocate(&buffer, src);
1008 if (buffer_len == (size_t)-1) {
1009 smb_panic("failed to create UCS2 buffer");
1012 /* We're using buffer_len below to count ucs2 characters, not bytes. */
1013 buffer_len /= sizeof(smb_ucs2_t);
1016 for (i = 0; buffer[i] != 0 && (i < buffer_len); i++) {
1017 unsigned char mb[10];
1018 /* Convert one smb_ucs2_t character at a time. */
1019 size_t mb_len = convert_string(CH_UTF16LE, CH_DOS, buffer+i, sizeof(smb_ucs2_t), mb, sizeof(mb), False);
1020 if ((mb_len != (size_t)-1) && (dest_len + mb_len <= MAX_NETBIOSNAME_LEN - 1)) {
1021 memcpy((char *)dest + dest_len, mb, mb_len);
1028 ((char *)dest)[dest_len] = '\0';
1031 conv_silent = False;
1035 /********************************************************************
1036 Push and malloc an ascii string. src and dest null terminated.
1037 ********************************************************************/
1039 size_t push_ascii_allocate(char **dest, const char *src)
1041 size_t src_len = strlen(src)+1;
1044 return convert_string_allocate(NULL, CH_UNIX, CH_DOS, src, src_len, (void **)dest, True);
1048 * Copy a string from a dos codepage source to a unix char* destination.
1050 * The resulting string in "dest" is always null terminated.
1052 * @param flags can have:
1054 * <dt>STR_TERMINATE</dt>
1055 * <dd>STR_TERMINATE means the string in @p src
1056 * is null terminated, and src_len is ignored.</dd>
1059 * @param src_len is the length of the source area in bytes.
1060 * @returns the number of bytes occupied by the string in @p src.
1062 size_t pull_ascii(char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
1066 if (dest_len == (size_t)-1) {
1067 /* No longer allow dest_len of -1. */
1068 smb_panic("pull_ascii - invalid dest_len of -1");
1071 if (flags & STR_TERMINATE) {
1072 if (src_len == (size_t)-1) {
1073 src_len = strlen((const char *)src) + 1;
1075 size_t len = strnlen((const char *)src, src_len);
1082 ret = convert_string(CH_DOS, CH_UNIX, src, src_len, dest, dest_len, True);
1083 if (ret == (size_t)-1) {
1088 if (dest_len && ret) {
1089 /* Did we already process the terminating zero ? */
1090 if (dest[MIN(ret-1, dest_len-1)] != 0) {
1091 dest[MIN(ret, dest_len-1)] = 0;
1101 * Copy a string from a dos codepage source to a unix char* destination.
1103 Uses malloc if TALLOC_CTX is NULL (this is a bad interface and
1106 * The resulting string in "dest" is always null terminated.
1108 * @param flags can have:
1110 * <dt>STR_TERMINATE</dt>
1111 * <dd>STR_TERMINATE means the string in @p src
1112 * is null terminated, and src_len is ignored.</dd>
1115 * @param src_len is the length of the source area in bytes.
1116 * @returns the number of bytes occupied by the string in @p src.
1119 static size_t pull_ascii_base_talloc(TALLOC_CTX *ctx,
1126 size_t dest_len = 0;
1129 /* Ensure we never use the braindead "malloc" varient. */
1131 smb_panic("NULL talloc CTX in pull_ascii_base_talloc\n");
1137 if (flags & STR_TERMINATE) {
1138 if (src_len == (size_t)-1) {
1139 src_len = strlen((const char *)src) + 1;
1141 size_t len = strnlen((const char *)src, src_len);
1146 /* Ensure we don't use an insane length from the client. */
1147 if (src_len >= 1024*1024) {
1148 char *msg = talloc_asprintf(ctx,
1149 "Bad src length (%u) in "
1150 "pull_ascii_base_talloc",
1151 (unsigned int)src_len);
1156 dest_len = convert_string_allocate(ctx,
1164 if (dest_len == (size_t)-1) {
1168 if (dest_len && dest) {
1169 /* Did we already process the terminating zero ? */
1170 if (dest[dest_len-1] != 0) {
1171 dest[dest_len-1] = 0;
1182 size_t pull_ascii_pstring(char *dest, const void *src)
1184 return pull_ascii(dest, src, sizeof(pstring), -1, STR_TERMINATE);
1187 size_t pull_ascii_fstring(char *dest, const void *src)
1189 return pull_ascii(dest, src, sizeof(fstring), -1, STR_TERMINATE);
1192 /* When pulling an nstring it can expand into a larger size (dos cp -> utf8). Cope with this. */
1194 size_t pull_ascii_nstring(char *dest, size_t dest_len, const void *src)
1196 return pull_ascii(dest, src, dest_len, sizeof(nstring)-1, STR_TERMINATE);
1200 * Copy a string from a char* src to a unicode destination.
1202 * @returns the number of bytes occupied by the string in the destination.
1204 * @param flags can have:
1207 * <dt>STR_TERMINATE <dd>means include the null termination.
1208 * <dt>STR_UPPER <dd>means uppercase in the destination.
1209 * <dt>STR_NOALIGN <dd>means don't do alignment.
1212 * @param dest_len is the maximum length allowed in the
1216 size_t push_ucs2(const void *base_ptr, void *dest, const char *src, size_t dest_len, int flags)
1222 if (dest_len == (size_t)-1) {
1223 /* No longer allow dest_len of -1. */
1224 smb_panic("push_ucs2 - invalid dest_len of -1");
1227 if (flags & STR_TERMINATE)
1228 src_len = (size_t)-1;
1230 src_len = strlen(src);
1232 if (ucs2_align(base_ptr, dest, flags)) {
1234 dest = (void *)((char *)dest + 1);
1240 /* ucs2 is always a multiple of 2 bytes */
1243 ret = convert_string(CH_UNIX, CH_UTF16LE, src, src_len, dest, dest_len, True);
1244 if (ret == (size_t)-1) {
1245 if ((flags & STR_TERMINATE) &&
1255 if (flags & STR_UPPER) {
1256 smb_ucs2_t *dest_ucs2 = (smb_ucs2_t *)dest;
1259 /* We check for i < (ret / 2) below as the dest string isn't null
1260 terminated if STR_TERMINATE isn't set. */
1262 for (i = 0; i < (ret / 2) && i < (dest_len / 2) && dest_ucs2[i]; i++) {
1263 smb_ucs2_t v = toupper_w(dest_ucs2[i]);
1264 if (v != dest_ucs2[i]) {
1275 * Copy a string from a unix char* src to a UCS2 destination,
1276 * allocating a buffer using talloc().
1278 * @param dest always set at least to NULL
1280 * @returns The number of bytes occupied by the string in the destination
1281 * or -1 in case of error.
1283 size_t push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src)
1285 size_t src_len = strlen(src)+1;
1288 return convert_string_talloc(ctx, CH_UNIX, CH_UTF16LE, src, src_len, (void **)dest, True);
1293 * Copy a string from a unix char* src to a UCS2 destination, allocating a buffer
1295 * @param dest always set at least to NULL
1297 * @returns The number of bytes occupied by the string in the destination
1298 * or -1 in case of error.
1301 size_t push_ucs2_allocate(smb_ucs2_t **dest, const char *src)
1303 size_t src_len = strlen(src)+1;
1306 return convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, src, src_len, (void **)dest, True);
1310 Copy a string from a char* src to a UTF-8 destination.
1311 Return the number of bytes occupied by the string in the destination
1313 STR_TERMINATE means include the null termination
1314 STR_UPPER means uppercase in the destination
1315 dest_len is the maximum length allowed in the destination. If dest_len
1316 is -1 then no maxiumum is used.
1319 static size_t push_utf8(void *dest, const char *src, size_t dest_len, int flags)
1323 char *tmpbuf = NULL;
1325 if (dest_len == (size_t)-1) {
1326 /* No longer allow dest_len of -1. */
1327 smb_panic("push_utf8 - invalid dest_len of -1");
1330 if (flags & STR_UPPER) {
1331 tmpbuf = strdup_upper(src);
1336 src_len = strlen(src);
1339 src_len = strlen(src);
1340 if (flags & STR_TERMINATE) {
1344 ret = convert_string(CH_UNIX, CH_UTF8, src, src_len, dest, dest_len, True);
1349 size_t push_utf8_fstring(void *dest, const char *src)
1351 return push_utf8(dest, src, sizeof(fstring), STR_TERMINATE);
1355 * Copy a string from a unix char* src to a UTF-8 destination, allocating a buffer using talloc
1357 * @param dest always set at least to NULL
1359 * @returns The number of bytes occupied by the string in the destination
1362 size_t push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
1364 size_t src_len = strlen(src)+1;
1367 return convert_string_talloc(ctx, CH_UNIX, CH_UTF8, src, src_len, (void**)dest, True);
1371 * Copy a string from a unix char* src to a UTF-8 destination, allocating a buffer
1373 * @param dest always set at least to NULL
1375 * @returns The number of bytes occupied by the string in the destination
1378 size_t push_utf8_allocate(char **dest, const char *src)
1380 size_t src_len = strlen(src)+1;
1383 return convert_string_allocate(NULL, CH_UNIX, CH_UTF8, src, src_len, (void **)dest, True);
1387 Copy a string from a ucs2 source to a unix char* destination.
1389 STR_TERMINATE means the string in src is null terminated.
1390 STR_NOALIGN means don't try to align.
1391 if STR_TERMINATE is set then src_len is ignored if it is -1.
1392 src_len is the length of the source area in bytes
1393 Return the number of bytes occupied by the string in src.
1394 The resulting string in "dest" is always null terminated.
1397 size_t pull_ucs2(const void *base_ptr, char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
1401 if (dest_len == (size_t)-1) {
1402 /* No longer allow dest_len of -1. */
1403 smb_panic("pull_ucs2 - invalid dest_len of -1");
1407 if (dest && dest_len > 0) {
1413 if (ucs2_align(base_ptr, src, flags)) {
1414 src = (const void *)((const char *)src + 1);
1415 if (src_len != (size_t)-1)
1419 if (flags & STR_TERMINATE) {
1420 /* src_len -1 is the default for null terminated strings. */
1421 if (src_len != (size_t)-1) {
1422 size_t len = strnlen_w((const smb_ucs2_t *)src,
1424 if (len < src_len/2)
1430 /* ucs2 is always a multiple of 2 bytes */
1431 if (src_len != (size_t)-1)
1434 ret = convert_string(CH_UTF16LE, CH_UNIX, src, src_len, dest, dest_len, True);
1435 if (ret == (size_t)-1) {
1440 if (src_len == (size_t)-1)
1443 if (dest_len && ret) {
1444 /* Did we already process the terminating zero ? */
1445 if (dest[MIN(ret-1, dest_len-1)] != 0) {
1446 dest[MIN(ret, dest_len-1)] = 0;
1456 Copy a string from a ucs2 source to a unix char* destination.
1457 Talloc version with a base pointer.
1458 Uses malloc if TALLOC_CTX is NULL (this is a bad interface and
1461 STR_TERMINATE means the string in src is null terminated.
1462 STR_NOALIGN means don't try to align.
1463 if STR_TERMINATE is set then src_len is ignored if it is -1.
1464 src_len is the length of the source area in bytes
1465 Return the number of bytes occupied by the string in src.
1466 The resulting string in "dest" is always null terminated.
1469 size_t pull_ucs2_base_talloc(TALLOC_CTX *ctx,
1470 const void *base_ptr,
1482 /* Ensure we never use the braindead "malloc" varient. */
1484 smb_panic("NULL talloc CTX in pull_ucs2_base_talloc\n");
1492 if (ucs2_align(base_ptr, src, flags)) {
1493 src = (const void *)((const char *)src + 1);
1494 if (src_len != (size_t)-1)
1498 if (flags & STR_TERMINATE) {
1499 /* src_len -1 is the default for null terminated strings. */
1500 if (src_len != (size_t)-1) {
1501 size_t len = strnlen_w((const smb_ucs2_t *)src,
1503 if (len < src_len/2)
1508 * src_len == -1 - alloc interface won't take this
1509 * so we must calculate.
1511 src_len = (strlen_w((const smb_ucs2_t *)src)+1)*sizeof(smb_ucs2_t);
1513 /* Ensure we don't use an insane length from the client. */
1514 if (src_len >= 1024*1024) {
1515 smb_panic("Bad src length in pull_ucs2_base_talloc\n");
1519 /* ucs2 is always a multiple of 2 bytes */
1520 if (src_len != (size_t)-1) {
1524 dest_len = convert_string_talloc(ctx,
1531 if (dest_len == (size_t)-1) {
1535 if (src_len == (size_t)-1)
1536 src_len = dest_len*2;
1539 /* Did we already process the terminating zero ? */
1540 if (dest[dest_len-1] != 0) {
1541 size_t size = talloc_get_size(dest);
1542 /* Have we got space to append the '\0' ? */
1543 if (size <= dest_len) {
1545 dest = TALLOC_REALLOC_ARRAY(ctx, dest, char,
1549 dest_len = (size_t)-1;
1554 dest[dest_len] = '\0';
1565 size_t pull_ucs2_fstring(char *dest, const void *src)
1567 return pull_ucs2(NULL, dest, src, sizeof(fstring), -1, STR_TERMINATE);
1571 * Copy a string from a UCS2 src to a unix char * destination, allocating a buffer using talloc
1573 * @param dest always set at least to NULL
1575 * @returns The number of bytes occupied by the string in the destination
1578 size_t pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src)
1580 size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
1582 return convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX, src, src_len, (void **)dest, True);
1586 * Copy a string from a UCS2 src to a unix char * destination, allocating a buffer
1588 * @param dest always set at least to NULL
1590 * @returns The number of bytes occupied by the string in the destination
1593 size_t pull_ucs2_allocate(char **dest, const smb_ucs2_t *src)
1595 size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
1597 return convert_string_allocate(NULL, CH_UTF16LE, CH_UNIX, src, src_len, (void **)dest, True);
1601 * Copy a string from a UTF-8 src to a unix char * destination, allocating a buffer using talloc
1603 * @param dest always set at least to NULL
1605 * @returns The number of bytes occupied by the string in the destination
1608 size_t pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
1610 size_t src_len = strlen(src)+1;
1612 return convert_string_talloc(ctx, CH_UTF8, CH_UNIX, src, src_len, (void **)dest, True);
1616 * Copy a string from a UTF-8 src to a unix char * destination, allocating a buffer
1618 * @param dest always set at least to NULL
1620 * @returns The number of bytes occupied by the string in the destination
1623 size_t pull_utf8_allocate(char **dest, const char *src)
1625 size_t src_len = strlen(src)+1;
1627 return convert_string_allocate(NULL, CH_UTF8, CH_UNIX, src, src_len, (void **)dest, True);
1631 * Copy a string from a DOS src to a unix char * destination, allocating a buffer using talloc
1633 * @param dest always set at least to NULL
1635 * @returns The number of bytes occupied by the string in the destination
1638 size_t pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
1640 size_t src_len = strlen(src)+1;
1642 return convert_string_talloc(ctx, CH_DOS, CH_UNIX, src, src_len, (void **)dest, True);
1646 Copy a string from a char* src to a unicode or ascii
1647 dos codepage destination choosing unicode or ascii based on the
1648 flags in the SMB buffer starting at base_ptr.
1649 Return the number of bytes occupied by the string in the destination.
1651 STR_TERMINATE means include the null termination.
1652 STR_UPPER means uppercase in the destination.
1653 STR_ASCII use ascii even with unicode packet.
1654 STR_NOALIGN means don't do alignment.
1655 dest_len is the maximum length allowed in the destination. If dest_len
1656 is -1 then no maxiumum is used.
1659 size_t push_string_fn(const char *function, unsigned int line,
1660 const void *base_ptr, uint16 flags2,
1661 void *dest, const char *src,
1662 size_t dest_len, int flags)
1665 /* We really need to zero fill here, not clobber
1666 * region, as we want to ensure that valgrind thinks
1667 * all of the outgoing buffer has been written to
1668 * so a send() or write() won't trap an error.
1672 clobber_region(function, line, dest, dest_len);
1674 memset(dest, '\0', dest_len);
1678 if (!(flags & STR_ASCII) && \
1679 ((flags & STR_UNICODE || \
1680 (flags2 & FLAGS2_UNICODE_STRINGS)))) {
1681 return push_ucs2(base_ptr, dest, src, dest_len, flags);
1683 return push_ascii(dest, src, dest_len, flags);
1688 Copy a string from a unicode or ascii source (depending on
1689 the packet flags) to a char* destination.
1691 STR_TERMINATE means the string in src is null terminated.
1692 STR_UNICODE means to force as unicode.
1693 STR_ASCII use ascii even with unicode packet.
1694 STR_NOALIGN means don't do alignment.
1695 if STR_TERMINATE is set then src_len is ignored is it is -1
1696 src_len is the length of the source area in bytes.
1697 Return the number of bytes occupied by the string in src.
1698 The resulting string in "dest" is always null terminated.
1701 size_t pull_string_fn(const char *function, unsigned int line,
1702 const void *base_ptr, uint16 smb_flags2, char *dest,
1703 const void *src, size_t dest_len, size_t src_len,
1707 clobber_region(function, line, dest, dest_len);
1710 if ((base_ptr == NULL) && ((flags & (STR_ASCII|STR_UNICODE)) == 0)) {
1711 smb_panic("No base ptr to get flg2 and neither ASCII nor "
1715 if (!(flags & STR_ASCII) && \
1716 ((flags & STR_UNICODE || \
1717 (smb_flags2 & FLAGS2_UNICODE_STRINGS)))) {
1718 return pull_ucs2(base_ptr, dest, src, dest_len, src_len, flags);
1720 return pull_ascii(dest, src, dest_len, src_len, flags);
1724 Copy a string from a unicode or ascii source (depending on
1725 the packet flags) to a char* destination.
1726 Variant that uses talloc.
1728 STR_TERMINATE means the string in src is null terminated.
1729 STR_UNICODE means to force as unicode.
1730 STR_ASCII use ascii even with unicode packet.
1731 STR_NOALIGN means don't do alignment.
1732 if STR_TERMINATE is set then src_len is ignored is it is -1
1733 src_len is the length of the source area in bytes.
1734 Return the number of bytes occupied by the string in src.
1735 The resulting string in "dest" is always null terminated.
1738 size_t pull_string_talloc_fn(const char *function,
1741 const void *base_ptr,
1748 if ((base_ptr == NULL) && ((flags & (STR_ASCII|STR_UNICODE)) == 0)) {
1749 smb_panic("No base ptr to get flg2 and neither ASCII nor "
1753 if (!(flags & STR_ASCII) && \
1754 ((flags & STR_UNICODE || \
1755 (smb_flags2 & FLAGS2_UNICODE_STRINGS)))) {
1756 return pull_ucs2_base_talloc(ctx,
1763 return pull_ascii_base_talloc(ctx,
1771 size_t align_string(const void *base_ptr, const char *p, int flags)
1773 if (!(flags & STR_ASCII) && \
1774 ((flags & STR_UNICODE || \
1775 (SVAL(base_ptr, smb_flg2) & FLAGS2_UNICODE_STRINGS)))) {
1776 return ucs2_align(base_ptr, p, flags);
1782 Return the unicode codepoint for the next multi-byte CH_UNIX character
1783 in the string. The unicode codepoint (codepoint_t) is an unsinged 32 bit value.
1785 Also return the number of bytes consumed (which tells the caller
1786 how many bytes to skip to get to the next CH_UNIX character).
1788 Return INVALID_CODEPOINT if the next character cannot be converted.
1791 codepoint_t next_codepoint(const char *str, size_t *size)
1793 /* It cannot occupy more than 4 bytes in UTF16 format */
1795 smb_iconv_t descriptor;
1801 if ((str[0] & 0x80) == 0) {
1803 return (codepoint_t)str[0];
1806 /* We assume that no multi-byte character can take
1807 more than 5 bytes. This is OK as we only
1808 support codepoints up to 1M */
1810 ilen_orig = strnlen(str, 5);
1813 lazy_initialize_conv();
1815 descriptor = conv_handles[CH_UNIX][CH_UTF16LE];
1816 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
1818 return INVALID_CODEPOINT;
1821 /* This looks a little strange, but it is needed to cope
1822 with codepoints above 64k which are encoded as per RFC2781. */
1824 outbuf = (char *)buf;
1825 smb_iconv(descriptor, &str, &ilen, &outbuf, &olen);
1827 /* We failed to convert to a 2 byte character.
1828 See if we can convert to a 4 UTF16-LE byte char encoding.
1831 outbuf = (char *)buf;
1832 smb_iconv(descriptor, &str, &ilen, &outbuf, &olen);
1834 /* We didn't convert any bytes */
1836 return INVALID_CODEPOINT;
1843 *size = ilen_orig - ilen;
1846 /* 2 byte, UTF16-LE encoded value. */
1847 return (codepoint_t)SVAL(buf, 0);
1850 /* Decode a 4 byte UTF16-LE character manually.
1851 See RFC2871 for the encoding machanism.
1853 codepoint_t w1 = SVAL(buf,0) & ~0xD800;
1854 codepoint_t w2 = SVAL(buf,2) & ~0xDC00;
1856 return (codepoint_t)0x10000 +
1860 /* no other length is valid */
1861 return INVALID_CODEPOINT;