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? */
187 * Convert string from one encoding to another, making error checking etc
188 * Slow path version - uses (slow) iconv.
190 * @param src pointer to source string (multibyte or singlebyte)
191 * @param srclen length of the source string in bytes
192 * @param dest pointer to destination string (multibyte or singlebyte)
193 * @param destlen maximal length allowed for string
194 * @param allow_bad_conv determines if a "best effort" conversion is acceptable (never returns errors)
195 * @returns the number of bytes occupied in the destination
197 * Ensure the srclen contains the terminating zero.
201 static size_t convert_string_internal(charset_t from, charset_t to,
202 void const *src, size_t srclen,
203 void *dest, size_t destlen, bool allow_bad_conv)
207 const char* inbuf = (const char*)src;
208 char* outbuf = (char*)dest;
209 smb_iconv_t descriptor;
211 lazy_initialize_conv();
213 descriptor = conv_handles[from][to];
215 if (srclen == (size_t)-1) {
216 if (from == CH_UTF16LE || from == CH_UTF16BE) {
217 srclen = (strlen_w((const smb_ucs2_t *)src)+1) * 2;
219 srclen = strlen((const char *)src)+1;
224 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
226 DEBUG(0,("convert_string_internal: Conversion not supported.\n"));
235 retval = smb_iconv(descriptor, &inbuf, &i_len, &outbuf, &o_len);
236 if(retval==(size_t)-1) {
237 const char *reason="unknown error";
240 reason="Incomplete multibyte sequence";
242 DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
247 reason="No more room";
249 if (from == CH_UNIX) {
250 DEBUG(3,("E2BIG: convert_string(%s,%s): srclen=%u destlen=%u - '%s'\n",
251 charset_name(from), charset_name(to),
252 (unsigned int)srclen, (unsigned int)destlen, (const char *)src));
254 DEBUG(3,("E2BIG: convert_string(%s,%s): srclen=%u destlen=%u\n",
255 charset_name(from), charset_name(to),
256 (unsigned int)srclen, (unsigned int)destlen));
261 reason="Illegal multibyte sequence";
263 DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
269 DEBUG(0,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
272 /* smb_panic(reason); */
274 return destlen-o_len;
279 * Conversion not supported. This is actually an error, but there are so
280 * many misconfigured iconv systems and smb.conf's out there we can't just
281 * fail. Do a very bad conversion instead.... JRA.
285 if (o_len == 0 || i_len == 0)
286 return destlen - o_len;
288 if (((from == CH_UTF16LE)||(from == CH_UTF16BE)) &&
289 ((to != CH_UTF16LE)||(to != CH_UTF16BE))) {
290 /* Can't convert from utf16 any endian to multibyte.
291 Replace with the default fail char.
294 return destlen - o_len;
296 *outbuf = lp_failed_convert_char();
305 if (o_len == 0 || i_len == 0)
306 return destlen - o_len;
308 /* Keep trying with the next char... */
311 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
312 /* Can't convert to UTF16LE - just widen by adding the
313 default fail char then zero.
316 return destlen - o_len;
318 outbuf[0] = lp_failed_convert_char();
327 if (o_len == 0 || i_len == 0)
328 return destlen - o_len;
330 /* Keep trying with the next char... */
333 } else if (from != CH_UTF16LE && from != CH_UTF16BE &&
334 to != CH_UTF16LE && to != CH_UTF16BE) {
335 /* Failed multibyte to multibyte. Just copy the default fail char and
337 outbuf[0] = lp_failed_convert_char();
345 if (o_len == 0 || i_len == 0)
346 return destlen - o_len;
348 /* Keep trying with the next char... */
352 /* Keep compiler happy.... */
353 return destlen - o_len;
359 * Convert string from one encoding to another, making error checking etc
360 * Fast path version - handles ASCII first.
362 * @param src pointer to source string (multibyte or singlebyte)
363 * @param srclen length of the source string in bytes, or -1 for nul terminated.
364 * @param dest pointer to destination string (multibyte or singlebyte)
365 * @param destlen maximal length allowed for string - *NEVER* -1.
366 * @param allow_bad_conv determines if a "best effort" conversion is acceptable (never returns errors)
367 * @returns the number of bytes occupied in the destination
369 * Ensure the srclen contains the terminating zero.
371 * This function has been hand-tuned to provide a fast path.
372 * Don't change unless you really know what you are doing. JRA.
375 size_t convert_string(charset_t from, charset_t to,
376 void const *src, size_t srclen,
377 void *dest, size_t destlen, bool allow_bad_conv)
380 * NB. We deliberately don't do a strlen here if srclen == -1.
381 * This is very expensive over millions of calls and is taken
382 * care of in the slow path in convert_string_internal. JRA.
386 SMB_ASSERT(destlen != (size_t)-1);
392 if (from != CH_UTF16LE && from != CH_UTF16BE && to != CH_UTF16LE && to != CH_UTF16BE) {
393 const unsigned char *p = (const unsigned char *)src;
394 unsigned char *q = (unsigned char *)dest;
395 size_t slen = srclen;
396 size_t dlen = destlen;
397 unsigned char lastp = '\0';
400 /* If all characters are ascii, fast path here. */
401 while (slen && dlen) {
402 if ((lastp = *p) <= 0x7f) {
404 if (slen != (size_t)-1) {
412 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
415 return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
420 /* Even if we fast path we should note if we ran out of room. */
421 if (((slen != (size_t)-1) && slen) ||
422 ((slen == (size_t)-1) && lastp)) {
427 } else if (from == CH_UTF16LE && to != CH_UTF16LE) {
428 const unsigned char *p = (const unsigned char *)src;
429 unsigned char *q = (unsigned char *)dest;
431 size_t slen = srclen;
432 size_t dlen = destlen;
433 unsigned char lastp = '\0';
435 /* If all characters are ascii, fast path here. */
436 while (((slen == (size_t)-1) || (slen >= 2)) && dlen) {
437 if (((lastp = *p) <= 0x7f) && (p[1] == 0)) {
439 if (slen != (size_t)-1) {
448 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
451 return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
456 /* Even if we fast path we should note if we ran out of room. */
457 if (((slen != (size_t)-1) && slen) ||
458 ((slen == (size_t)-1) && lastp)) {
463 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
464 const unsigned char *p = (const unsigned char *)src;
465 unsigned char *q = (unsigned char *)dest;
467 size_t slen = srclen;
468 size_t dlen = destlen;
469 unsigned char lastp = '\0';
471 /* If all characters are ascii, fast path here. */
472 while (slen && (dlen >= 2)) {
473 if ((lastp = *p) <= 0x7F) {
476 if (slen != (size_t)-1) {
484 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
487 return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
492 /* Even if we fast path we should note if we ran out of room. */
493 if (((slen != (size_t)-1) && slen) ||
494 ((slen == (size_t)-1) && lastp)) {
501 #ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
504 return convert_string_internal(from, to, src, srclen, dest, destlen, allow_bad_conv);
508 * Convert between character sets, allocating a new buffer for the result.
510 * @param ctx TALLOC_CTX to use to allocate with. If NULL use malloc.
511 * (this is a bad interface and needs fixing. JRA).
512 * @param srclen length of source buffer.
513 * @param dest always set at least to NULL
514 * @note -1 is not accepted for srclen.
516 * @returns Size in bytes of the converted string; or -1 in case of error.
518 * Ensure the srclen contains the terminating zero.
520 * I hate the goto's in this function. It's embarressing.....
521 * There has to be a cleaner way to do this. JRA.
524 size_t convert_string_allocate(TALLOC_CTX *ctx, charset_t from, charset_t to,
525 void const *src, size_t srclen, void *dst, bool allow_bad_conv)
527 size_t i_len, o_len, destlen = MAX(srclen, 512);
529 const char *inbuf = (const char *)src;
530 char *outbuf = NULL, *ob = NULL;
531 smb_iconv_t descriptor;
532 void **dest = (void **)dst;
536 if (src == NULL || srclen == (size_t)-1)
541 lazy_initialize_conv();
543 descriptor = conv_handles[from][to];
545 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
547 DEBUG(0,("convert_string_allocate: Conversion not supported.\n"));
553 if ((destlen*2) < destlen) {
554 /* wrapped ! abort. */
556 DEBUG(0, ("convert_string_allocate: destlen wrapped !\n"));
561 destlen = destlen * 2;
565 ob = (char *)TALLOC_REALLOC(ctx, ob, destlen);
567 ob = (char *)SMB_REALLOC(ob, destlen);
571 DEBUG(0, ("convert_string_allocate: realloc failed!\n"));
580 retval = smb_iconv(descriptor,
583 if(retval == (size_t)-1) {
584 const char *reason="unknown error";
587 reason="Incomplete multibyte sequence";
589 DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
596 reason="Illegal multibyte sequence";
598 DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
604 DEBUG(0,("Conversion error: %s(%s)\n",reason,inbuf));
605 /* smb_panic(reason); */
616 destlen = destlen - o_len;
618 ob = (char *)TALLOC_REALLOC(ctx,ob,destlen);
620 ob = (char *)SMB_REALLOC(ob,destlen);
623 if (destlen && !ob) {
624 DEBUG(0, ("convert_string_allocate: out of memory!\n"));
634 * Conversion not supported. This is actually an error, but there are so
635 * many misconfigured iconv systems and smb.conf's out there we can't just
636 * fail. Do a very bad conversion instead.... JRA.
640 if (o_len == 0 || i_len == 0)
643 if (((from == CH_UTF16LE)||(from == CH_UTF16BE)) &&
644 ((to != CH_UTF16LE)||(to != CH_UTF16BE))) {
645 /* Can't convert from utf16 any endian to multibyte.
646 Replace with the default fail char.
653 *outbuf = lp_failed_convert_char();
662 if (o_len == 0 || i_len == 0)
665 /* Keep trying with the next char... */
668 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
669 /* Can't convert to UTF16LE - just widen by adding the
670 default fail char then zero.
675 outbuf[0] = lp_failed_convert_char();
684 if (o_len == 0 || i_len == 0)
687 /* Keep trying with the next char... */
690 } else if (from != CH_UTF16LE && from != CH_UTF16BE &&
691 to != CH_UTF16LE && to != CH_UTF16BE) {
692 /* Failed multibyte to multibyte. Just copy the default fail char and
694 outbuf[0] = lp_failed_convert_char();
702 if (o_len == 0 || i_len == 0)
705 /* Keep trying with the next char... */
709 /* Keep compiler happy.... */
716 * Convert between character sets, allocating a new buffer using talloc for the result.
718 * @param srclen length of source buffer.
719 * @param dest always set at least to NULL
720 * @note -1 is not accepted for srclen.
722 * @returns Size in bytes of the converted string; or -1 in case of error.
724 size_t convert_string_talloc(TALLOC_CTX *ctx, charset_t from, charset_t to,
725 void const *src, size_t srclen, void *dst,
728 void **dest = (void **)dst;
732 dest_len=convert_string_allocate(ctx, from, to, src, srclen, dest, allow_bad_conv);
733 if (dest_len == (size_t)-1)
740 size_t unix_strupper(const char *src, size_t srclen, char *dest, size_t destlen)
745 size = push_ucs2_allocate(&buffer, src);
746 if (size == (size_t)-1) {
749 if (!strupper_w(buffer) && (dest == src)) {
754 size = convert_string(CH_UTF16LE, CH_UNIX, buffer, size, dest, destlen, True);
760 strdup() a unix string to upper case.
763 char *strdup_upper(const char *s)
765 char *out_buffer = SMB_STRDUP(s);
766 const unsigned char *p = (const unsigned char *)s;
767 unsigned char *q = (unsigned char *)out_buffer;
773 /* this is quite a common operation, so we want it to be
774 fast. We optimise for the ascii case, knowing that all our
775 supported multi-byte character sets are ascii-compatible
776 (ie. they match for the first 128 chars) */
781 *q++ = toupper_ascii(*p);
788 smb_ucs2_t *buffer = NULL;
790 SAFE_FREE(out_buffer);
791 size = convert_string_allocate(NULL,
796 (void **)(void *)&buffer,
798 if (size == (size_t)-1) {
804 size = convert_string_allocate(NULL,
809 (void **)(void *)&out_buffer,
812 /* Don't need the intermediate buffer
817 if (size == (size_t)-1) {
826 talloc_strdup() a unix string to upper case.
829 char *talloc_strdup_upper(TALLOC_CTX *ctx, const char *s)
831 char *out_buffer = talloc_strdup(ctx,s);
832 const unsigned char *p = (const unsigned char *)s;
833 unsigned char *q = (unsigned char *)out_buffer;
839 /* this is quite a common operation, so we want it to be
840 fast. We optimise for the ascii case, knowing that all our
841 supported multi-byte character sets are ascii-compatible
842 (ie. they match for the first 128 chars) */
847 *q++ = toupper_ascii(*p);
854 smb_ucs2_t *ubuf = NULL;
856 /* We're not using the ascii buffer above. */
857 TALLOC_FREE(out_buffer);
859 size = convert_string_talloc(ctx, CH_UNIX, CH_UTF16LE,
863 if (size == (size_t)-1) {
869 size = convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX,
874 /* Don't need the intermediate buffer
880 if (size == (size_t)-1) {
888 size_t unix_strlower(const char *src, size_t srclen, char *dest, size_t destlen)
891 smb_ucs2_t *buffer = NULL;
893 size = convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, src, srclen,
894 (void **)(void *)&buffer, True);
895 if (size == (size_t)-1 || !buffer) {
896 smb_panic("failed to create UCS2 buffer");
898 if (!strlower_w(buffer) && (dest == src)) {
902 size = convert_string(CH_UTF16LE, CH_UNIX, buffer, size, dest, destlen, True);
908 strdup() a unix string to lower case.
911 char *strdup_lower(const char *s)
914 smb_ucs2_t *buffer = NULL;
917 size = push_ucs2_allocate(&buffer, s);
918 if (size == -1 || !buffer) {
924 size = pull_ucs2_allocate(&out_buffer, buffer);
927 if (size == (size_t)-1) {
934 size_t ucs2_align(const void *base_ptr, const void *p, int flags)
936 if (flags & (STR_NOALIGN|STR_ASCII))
938 return PTR_DIFF(p, base_ptr) & 1;
943 * Copy a string from a char* unix src to a dos codepage string destination.
945 * @return the number of bytes occupied by the string in the destination.
947 * @param flags can include
949 * <dt>STR_TERMINATE</dt> <dd>means include the null termination</dd>
950 * <dt>STR_UPPER</dt> <dd>means uppercase in the destination</dd>
953 * @param dest_len the maximum length in bytes allowed in the
956 size_t push_ascii(void *dest, const char *src, size_t dest_len, int flags)
958 size_t src_len = strlen(src);
962 /* No longer allow a length of -1. */
963 if (dest_len == (size_t)-1) {
964 smb_panic("push_ascii - dest_len == -1");
967 if (flags & STR_UPPER) {
968 tmpbuf = SMB_STRDUP(src);
970 smb_panic("malloc fail");
976 if (flags & (STR_TERMINATE | STR_TERMINATE_ASCII)) {
980 ret = convert_string(CH_UNIX, CH_DOS, src, src_len, dest, dest_len, True);
981 if (ret == (size_t)-1 &&
982 (flags & (STR_TERMINATE | STR_TERMINATE_ASCII))
984 ((char *)dest)[0] = '\0';
990 size_t push_ascii_fstring(void *dest, const char *src)
992 return push_ascii(dest, src, sizeof(fstring), STR_TERMINATE);
995 /********************************************************************
996 Push an nstring - ensure null terminated. Written by
997 moriyama@miraclelinux.com (MORIYAMA Masayuki).
998 ********************************************************************/
1000 size_t push_ascii_nstring(void *dest, const char *src)
1002 size_t i, buffer_len, dest_len;
1006 buffer_len = push_ucs2_allocate(&buffer, src);
1007 if (buffer_len == (size_t)-1) {
1008 smb_panic("failed to create UCS2 buffer");
1011 /* We're using buffer_len below to count ucs2 characters, not bytes. */
1012 buffer_len /= sizeof(smb_ucs2_t);
1015 for (i = 0; buffer[i] != 0 && (i < buffer_len); i++) {
1016 unsigned char mb[10];
1017 /* Convert one smb_ucs2_t character at a time. */
1018 size_t mb_len = convert_string(CH_UTF16LE, CH_DOS, buffer+i, sizeof(smb_ucs2_t), mb, sizeof(mb), False);
1019 if ((mb_len != (size_t)-1) && (dest_len + mb_len <= MAX_NETBIOSNAME_LEN - 1)) {
1020 memcpy((char *)dest + dest_len, mb, mb_len);
1027 ((char *)dest)[dest_len] = '\0';
1030 conv_silent = False;
1034 /********************************************************************
1035 Push and malloc an ascii string. src and dest null terminated.
1036 ********************************************************************/
1038 size_t push_ascii_allocate(char **dest, const char *src)
1040 size_t src_len = strlen(src)+1;
1043 return convert_string_allocate(NULL, CH_UNIX, CH_DOS, src, src_len, (void **)dest, True);
1047 * Copy a string from a dos codepage source to a unix char* destination.
1049 * The resulting string in "dest" is always null terminated.
1051 * @param flags can have:
1053 * <dt>STR_TERMINATE</dt>
1054 * <dd>STR_TERMINATE means the string in @p src
1055 * is null terminated, and src_len is ignored.</dd>
1058 * @param src_len is the length of the source area in bytes.
1059 * @returns the number of bytes occupied by the string in @p src.
1061 size_t pull_ascii(char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
1065 if (dest_len == (size_t)-1) {
1066 /* No longer allow dest_len of -1. */
1067 smb_panic("pull_ascii - invalid dest_len of -1");
1070 if (flags & STR_TERMINATE) {
1071 if (src_len == (size_t)-1) {
1072 src_len = strlen((const char *)src) + 1;
1074 size_t len = strnlen((const char *)src, src_len);
1081 ret = convert_string(CH_DOS, CH_UNIX, src, src_len, dest, dest_len, True);
1082 if (ret == (size_t)-1) {
1087 if (dest_len && ret) {
1088 /* Did we already process the terminating zero ? */
1089 if (dest[MIN(ret-1, dest_len-1)] != 0) {
1090 dest[MIN(ret, dest_len-1)] = 0;
1100 * Copy a string from a dos codepage source to a unix char* destination.
1102 Uses malloc if TALLOC_CTX is NULL (this is a bad interface and
1105 * The resulting string in "dest" is always null terminated.
1107 * @param flags can have:
1109 * <dt>STR_TERMINATE</dt>
1110 * <dd>STR_TERMINATE means the string in @p src
1111 * is null terminated, and src_len is ignored.</dd>
1114 * @param src_len is the length of the source area in bytes.
1115 * @returns the number of bytes occupied by the string in @p src.
1118 static size_t pull_ascii_base_talloc(TALLOC_CTX *ctx,
1125 size_t dest_len = 0;
1128 /* Ensure we never use the braindead "malloc" varient. */
1130 smb_panic("NULL talloc CTX in pull_ascii_base_talloc\n");
1136 if (flags & STR_TERMINATE) {
1137 if (src_len == (size_t)-1) {
1138 src_len = strlen((const char *)src) + 1;
1140 size_t len = strnlen((const char *)src, src_len);
1145 /* Ensure we don't use an insane length from the client. */
1146 if (src_len >= 1024*1024) {
1147 char *msg = talloc_asprintf(ctx,
1148 "Bad src length (%u) in "
1149 "pull_ascii_base_talloc",
1150 (unsigned int)src_len);
1155 dest_len = convert_string_allocate(ctx,
1163 if (dest_len == (size_t)-1) {
1167 if (dest_len && dest) {
1168 /* Did we already process the terminating zero ? */
1169 if (dest[dest_len-1] != 0) {
1170 dest[dest_len-1] = 0;
1180 size_t pull_ascii_fstring(char *dest, const void *src)
1182 return pull_ascii(dest, src, sizeof(fstring), -1, STR_TERMINATE);
1185 /* When pulling an nstring it can expand into a larger size (dos cp -> utf8). Cope with this. */
1187 size_t pull_ascii_nstring(char *dest, size_t dest_len, const void *src)
1189 return pull_ascii(dest, src, dest_len, sizeof(nstring)-1, STR_TERMINATE);
1193 * Copy a string from a char* src to a unicode destination.
1195 * @returns the number of bytes occupied by the string in the destination.
1197 * @param flags can have:
1200 * <dt>STR_TERMINATE <dd>means include the null termination.
1201 * <dt>STR_UPPER <dd>means uppercase in the destination.
1202 * <dt>STR_NOALIGN <dd>means don't do alignment.
1205 * @param dest_len is the maximum length allowed in the
1209 size_t push_ucs2(const void *base_ptr, void *dest, const char *src, size_t dest_len, int flags)
1215 if (dest_len == (size_t)-1) {
1216 /* No longer allow dest_len of -1. */
1217 smb_panic("push_ucs2 - invalid dest_len of -1");
1220 if (flags & STR_TERMINATE)
1221 src_len = (size_t)-1;
1223 src_len = strlen(src);
1225 if (ucs2_align(base_ptr, dest, flags)) {
1227 dest = (void *)((char *)dest + 1);
1233 /* ucs2 is always a multiple of 2 bytes */
1236 ret = convert_string(CH_UNIX, CH_UTF16LE, src, src_len, dest, dest_len, True);
1237 if (ret == (size_t)-1) {
1238 if ((flags & STR_TERMINATE) &&
1248 if (flags & STR_UPPER) {
1249 smb_ucs2_t *dest_ucs2 = (smb_ucs2_t *)dest;
1252 /* We check for i < (ret / 2) below as the dest string isn't null
1253 terminated if STR_TERMINATE isn't set. */
1255 for (i = 0; i < (ret / 2) && i < (dest_len / 2) && dest_ucs2[i]; i++) {
1256 smb_ucs2_t v = toupper_w(dest_ucs2[i]);
1257 if (v != dest_ucs2[i]) {
1268 * Copy a string from a unix char* src to a UCS2 destination,
1269 * allocating a buffer using talloc().
1271 * @param dest always set at least to NULL
1273 * @returns The number of bytes occupied by the string in the destination
1274 * or -1 in case of error.
1276 size_t push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src)
1278 size_t src_len = strlen(src)+1;
1281 return convert_string_talloc(ctx, CH_UNIX, CH_UTF16LE, src, src_len, (void **)dest, True);
1286 * Copy a string from a unix char* src to a UCS2 destination, allocating a buffer
1288 * @param dest always set at least to NULL
1290 * @returns The number of bytes occupied by the string in the destination
1291 * or -1 in case of error.
1294 size_t push_ucs2_allocate(smb_ucs2_t **dest, const char *src)
1296 size_t src_len = strlen(src)+1;
1299 return convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, src, src_len, (void **)dest, True);
1303 Copy a string from a char* src to a UTF-8 destination.
1304 Return the number of bytes occupied by the string in the destination
1306 STR_TERMINATE means include the null termination
1307 STR_UPPER means uppercase in the destination
1308 dest_len is the maximum length allowed in the destination. If dest_len
1309 is -1 then no maxiumum is used.
1312 static size_t push_utf8(void *dest, const char *src, size_t dest_len, int flags)
1316 char *tmpbuf = NULL;
1318 if (dest_len == (size_t)-1) {
1319 /* No longer allow dest_len of -1. */
1320 smb_panic("push_utf8 - invalid dest_len of -1");
1323 if (flags & STR_UPPER) {
1324 tmpbuf = strdup_upper(src);
1329 src_len = strlen(src);
1332 src_len = strlen(src);
1333 if (flags & STR_TERMINATE) {
1337 ret = convert_string(CH_UNIX, CH_UTF8, src, src_len, dest, dest_len, True);
1342 size_t push_utf8_fstring(void *dest, const char *src)
1344 return push_utf8(dest, src, sizeof(fstring), STR_TERMINATE);
1348 * Copy a string from a unix char* src to a UTF-8 destination, allocating a buffer using talloc
1350 * @param dest always set at least to NULL
1352 * @returns The number of bytes occupied by the string in the destination
1355 size_t push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
1357 size_t src_len = strlen(src)+1;
1360 return convert_string_talloc(ctx, CH_UNIX, CH_UTF8, src, src_len, (void**)dest, True);
1364 * Copy a string from a unix char* src to a UTF-8 destination, allocating a buffer
1366 * @param dest always set at least to NULL
1368 * @returns The number of bytes occupied by the string in the destination
1371 size_t push_utf8_allocate(char **dest, const char *src)
1373 size_t src_len = strlen(src)+1;
1376 return convert_string_allocate(NULL, CH_UNIX, CH_UTF8, src, src_len, (void **)dest, True);
1380 Copy a string from a ucs2 source to a unix char* destination.
1382 STR_TERMINATE means the string in src is null terminated.
1383 STR_NOALIGN means don't try to align.
1384 if STR_TERMINATE is set then src_len is ignored if it is -1.
1385 src_len is the length of the source area in bytes
1386 Return the number of bytes occupied by the string in src.
1387 The resulting string in "dest" is always null terminated.
1390 size_t pull_ucs2(const void *base_ptr, char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
1394 if (dest_len == (size_t)-1) {
1395 /* No longer allow dest_len of -1. */
1396 smb_panic("pull_ucs2 - invalid dest_len of -1");
1400 if (dest && dest_len > 0) {
1406 if (ucs2_align(base_ptr, src, flags)) {
1407 src = (const void *)((const char *)src + 1);
1408 if (src_len != (size_t)-1)
1412 if (flags & STR_TERMINATE) {
1413 /* src_len -1 is the default for null terminated strings. */
1414 if (src_len != (size_t)-1) {
1415 size_t len = strnlen_w((const smb_ucs2_t *)src,
1417 if (len < src_len/2)
1423 /* ucs2 is always a multiple of 2 bytes */
1424 if (src_len != (size_t)-1)
1427 ret = convert_string(CH_UTF16LE, CH_UNIX, src, src_len, dest, dest_len, True);
1428 if (ret == (size_t)-1) {
1433 if (src_len == (size_t)-1)
1436 if (dest_len && ret) {
1437 /* Did we already process the terminating zero ? */
1438 if (dest[MIN(ret-1, dest_len-1)] != 0) {
1439 dest[MIN(ret, dest_len-1)] = 0;
1449 Copy a string from a ucs2 source to a unix char* destination.
1450 Talloc version with a base pointer.
1451 Uses malloc if TALLOC_CTX is NULL (this is a bad interface and
1454 STR_TERMINATE means the string in src is null terminated.
1455 STR_NOALIGN means don't try to align.
1456 if STR_TERMINATE is set then src_len is ignored if it is -1.
1457 src_len is the length of the source area in bytes
1458 Return the number of bytes occupied by the string in src.
1459 The resulting string in "dest" is always null terminated.
1462 size_t pull_ucs2_base_talloc(TALLOC_CTX *ctx,
1463 const void *base_ptr,
1475 /* Ensure we never use the braindead "malloc" varient. */
1477 smb_panic("NULL talloc CTX in pull_ucs2_base_talloc\n");
1485 if (ucs2_align(base_ptr, src, flags)) {
1486 src = (const void *)((const char *)src + 1);
1487 if (src_len != (size_t)-1)
1491 if (flags & STR_TERMINATE) {
1492 /* src_len -1 is the default for null terminated strings. */
1493 if (src_len != (size_t)-1) {
1494 size_t len = strnlen_w((const smb_ucs2_t *)src,
1496 if (len < src_len/2)
1501 * src_len == -1 - alloc interface won't take this
1502 * so we must calculate.
1504 src_len = (strlen_w((const smb_ucs2_t *)src)+1)*sizeof(smb_ucs2_t);
1506 /* Ensure we don't use an insane length from the client. */
1507 if (src_len >= 1024*1024) {
1508 smb_panic("Bad src length in pull_ucs2_base_talloc\n");
1512 /* ucs2 is always a multiple of 2 bytes */
1513 if (src_len != (size_t)-1) {
1517 dest_len = convert_string_talloc(ctx,
1524 if (dest_len == (size_t)-1) {
1528 if (src_len == (size_t)-1)
1529 src_len = dest_len*2;
1532 /* Did we already process the terminating zero ? */
1533 if (dest[dest_len-1] != 0) {
1534 size_t size = talloc_get_size(dest);
1535 /* Have we got space to append the '\0' ? */
1536 if (size <= dest_len) {
1538 dest = TALLOC_REALLOC_ARRAY(ctx, dest, char,
1542 dest_len = (size_t)-1;
1547 dest[dest_len] = '\0';
1558 size_t pull_ucs2_fstring(char *dest, const void *src)
1560 return pull_ucs2(NULL, dest, src, sizeof(fstring), -1, STR_TERMINATE);
1564 * Copy a string from a UCS2 src to a unix char * destination, allocating a buffer using talloc
1566 * @param dest always set at least to NULL
1568 * @returns The number of bytes occupied by the string in the destination
1571 size_t pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src)
1573 size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
1575 return convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX, src, src_len, (void **)dest, True);
1579 * Copy a string from a UCS2 src to a unix char * destination, allocating a buffer
1581 * @param dest always set at least to NULL
1583 * @returns The number of bytes occupied by the string in the destination
1586 size_t pull_ucs2_allocate(char **dest, const smb_ucs2_t *src)
1588 size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
1590 return convert_string_allocate(NULL, CH_UTF16LE, CH_UNIX, src, src_len, (void **)dest, True);
1594 * Copy a string from a UTF-8 src to a unix char * destination, allocating a buffer using talloc
1596 * @param dest always set at least to NULL
1598 * @returns The number of bytes occupied by the string in the destination
1601 size_t pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
1603 size_t src_len = strlen(src)+1;
1605 return convert_string_talloc(ctx, CH_UTF8, CH_UNIX, src, src_len, (void **)dest, True);
1609 * Copy a string from a UTF-8 src to a unix char * destination, allocating a buffer
1611 * @param dest always set at least to NULL
1613 * @returns The number of bytes occupied by the string in the destination
1616 size_t pull_utf8_allocate(char **dest, const char *src)
1618 size_t src_len = strlen(src)+1;
1620 return convert_string_allocate(NULL, CH_UTF8, CH_UNIX, src, src_len, (void **)dest, True);
1624 * Copy a string from a DOS src to a unix char * destination, allocating a buffer using talloc
1626 * @param dest always set at least to NULL
1628 * @returns The number of bytes occupied by the string in the destination
1631 size_t pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
1633 size_t src_len = strlen(src)+1;
1635 return convert_string_talloc(ctx, CH_DOS, CH_UNIX, src, src_len, (void **)dest, True);
1639 Copy a string from a char* src to a unicode or ascii
1640 dos codepage destination choosing unicode or ascii based on the
1641 flags in the SMB buffer starting at base_ptr.
1642 Return the number of bytes occupied by the string in the destination.
1644 STR_TERMINATE means include the null termination.
1645 STR_UPPER means uppercase in the destination.
1646 STR_ASCII use ascii even with unicode packet.
1647 STR_NOALIGN means don't do alignment.
1648 dest_len is the maximum length allowed in the destination. If dest_len
1649 is -1 then no maxiumum is used.
1652 size_t push_string_fn(const char *function, unsigned int line,
1653 const void *base_ptr, uint16 flags2,
1654 void *dest, const char *src,
1655 size_t dest_len, int flags)
1658 /* We really need to zero fill here, not clobber
1659 * region, as we want to ensure that valgrind thinks
1660 * all of the outgoing buffer has been written to
1661 * so a send() or write() won't trap an error.
1665 clobber_region(function, line, dest, dest_len);
1667 memset(dest, '\0', dest_len);
1671 if (!(flags & STR_ASCII) && \
1672 ((flags & STR_UNICODE || \
1673 (flags2 & FLAGS2_UNICODE_STRINGS)))) {
1674 return push_ucs2(base_ptr, dest, src, dest_len, flags);
1676 return push_ascii(dest, src, dest_len, flags);
1681 Copy a string from a unicode or ascii source (depending on
1682 the packet flags) to a char* destination.
1684 STR_TERMINATE means the string in src is null terminated.
1685 STR_UNICODE means to force as unicode.
1686 STR_ASCII use ascii even with unicode packet.
1687 STR_NOALIGN means don't do alignment.
1688 if STR_TERMINATE is set then src_len is ignored is it is -1
1689 src_len is the length of the source area in bytes.
1690 Return the number of bytes occupied by the string in src.
1691 The resulting string in "dest" is always null terminated.
1694 size_t pull_string_fn(const char *function,
1696 const void *base_ptr,
1705 clobber_region(function, line, dest, dest_len);
1708 if ((base_ptr == NULL) && ((flags & (STR_ASCII|STR_UNICODE)) == 0)) {
1709 smb_panic("No base ptr to get flg2 and neither ASCII nor "
1713 if (!(flags & STR_ASCII) && \
1714 ((flags & STR_UNICODE || \
1715 (smb_flags2 & FLAGS2_UNICODE_STRINGS)))) {
1716 return pull_ucs2(base_ptr, dest, src, dest_len, src_len, flags);
1718 return pull_ascii(dest, src, dest_len, src_len, flags);
1722 Copy a string from a unicode or ascii source (depending on
1723 the packet flags) to a char* destination.
1724 Variant that uses talloc.
1726 STR_TERMINATE means the string in src is null terminated.
1727 STR_UNICODE means to force as unicode.
1728 STR_ASCII use ascii even with unicode packet.
1729 STR_NOALIGN means don't do alignment.
1730 if STR_TERMINATE is set then src_len is ignored is it is -1
1731 src_len is the length of the source area in bytes.
1732 Return the number of bytes occupied by the string in src.
1733 The resulting string in "dest" is always null terminated.
1736 size_t pull_string_talloc_fn(const char *function,
1739 const void *base_ptr,
1746 if ((base_ptr == NULL) && ((flags & (STR_ASCII|STR_UNICODE)) == 0)) {
1747 smb_panic("No base ptr to get flg2 and neither ASCII nor "
1751 if (!(flags & STR_ASCII) && \
1752 ((flags & STR_UNICODE || \
1753 (smb_flags2 & FLAGS2_UNICODE_STRINGS)))) {
1754 return pull_ucs2_base_talloc(ctx,
1761 return pull_ascii_base_talloc(ctx,
1769 size_t align_string(const void *base_ptr, const char *p, int flags)
1771 if (!(flags & STR_ASCII) && \
1772 ((flags & STR_UNICODE || \
1773 (SVAL(base_ptr, smb_flg2) & FLAGS2_UNICODE_STRINGS)))) {
1774 return ucs2_align(base_ptr, p, flags);
1780 Return the unicode codepoint for the next multi-byte CH_UNIX character
1781 in the string. The unicode codepoint (codepoint_t) is an unsinged 32 bit value.
1783 Also return the number of bytes consumed (which tells the caller
1784 how many bytes to skip to get to the next CH_UNIX character).
1786 Return INVALID_CODEPOINT if the next character cannot be converted.
1789 codepoint_t next_codepoint(const char *str, size_t *size)
1791 /* It cannot occupy more than 4 bytes in UTF16 format */
1793 smb_iconv_t descriptor;
1799 if ((str[0] & 0x80) == 0) {
1801 return (codepoint_t)str[0];
1804 /* We assume that no multi-byte character can take
1805 more than 5 bytes. This is OK as we only
1806 support codepoints up to 1M */
1808 ilen_orig = strnlen(str, 5);
1811 lazy_initialize_conv();
1813 descriptor = conv_handles[CH_UNIX][CH_UTF16LE];
1814 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
1816 return INVALID_CODEPOINT;
1819 /* This looks a little strange, but it is needed to cope
1820 with codepoints above 64k which are encoded as per RFC2781. */
1822 outbuf = (char *)buf;
1823 smb_iconv(descriptor, &str, &ilen, &outbuf, &olen);
1825 /* We failed to convert to a 2 byte character.
1826 See if we can convert to a 4 UTF16-LE byte char encoding.
1829 outbuf = (char *)buf;
1830 smb_iconv(descriptor, &str, &ilen, &outbuf, &olen);
1832 /* We didn't convert any bytes */
1834 return INVALID_CODEPOINT;
1841 *size = ilen_orig - ilen;
1844 /* 2 byte, UTF16-LE encoded value. */
1845 return (codepoint_t)SVAL(buf, 0);
1848 /* Decode a 4 byte UTF16-LE character manually.
1849 See RFC2871 for the encoding machanism.
1851 codepoint_t w1 = SVAL(buf,0) & ~0xD800;
1852 codepoint_t w2 = SVAL(buf,2) & ~0xDC00;
1854 return (codepoint_t)0x10000 +
1858 /* no other length is valid */
1859 return INVALID_CODEPOINT;