source: branches/samba-3.3.x/source/lib/charcnv.c@ 233

Last change on this file since 233 was 223, checked in by Herwig Bauernfeind, 16 years ago

Update Samba 3.3 branch to 3.3.3
File size: 51.1 KB
Line 
1/*
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
8
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.
13
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.
18
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/>.
21
22*/
23#include "includes.h"
24
25/* We can parameterize this if someone complains.... JRA. */
26
27char lp_failed_convert_char(void)
28{
29 return '_';
30}
31
32/**
33 * @file
34 *
35 * @brief Character-set conversion routines built on our iconv.
36 *
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.
43 *
44 * @sa lib/iconv.c
45 */
46
47
48static smb_iconv_t conv_handles[NUM_CHARSETS][NUM_CHARSETS];
49static bool conv_silent; /* Should we do a debug if the conversion fails ? */
50static bool initialized;
51
52/**
53 * Return the name of a charset to give to iconv().
54 **/
55static const char *charset_name(charset_t ch)
56{
57 const char *ret = NULL;
58
59 if (ch == CH_UTF16LE) ret = "UTF-16LE";
60 else if (ch == CH_UTF16BE) ret = "UTF-16BE";
61 else if (ch == CH_UNIX) ret = lp_unix_charset();
62 else if (ch == CH_DOS) ret = lp_dos_charset();
63 else if (ch == CH_DISPLAY) ret = lp_display_charset();
64 else if (ch == CH_UTF8) ret = "UTF8";
65
66#if defined(HAVE_NL_LANGINFO) && defined(CODESET)
67 if (ret && !strcmp(ret, "LOCALE")) {
68 const char *ln = NULL;
69
70#ifdef HAVE_SETLOCALE
71 setlocale(LC_ALL, "");
72#endif
73 ln = nl_langinfo(CODESET);
74 if (ln) {
75 /* Check whether the charset name is supported
76 by iconv */
77 smb_iconv_t handle = smb_iconv_open(ln,"UCS-2LE");
78 if (handle == (smb_iconv_t) -1) {
79 DEBUG(5,("Locale charset '%s' unsupported, using ASCII instead\n", ln));
80 ln = NULL;
81 } else {
82 DEBUG(5,("Substituting charset '%s' for LOCALE\n", ln));
83 smb_iconv_close(handle);
84 }
85 }
86 ret = ln;
87 }
88#endif
89
90 if (!ret || !*ret) ret = "ASCII";
91 return ret;
92}
93
94void lazy_initialize_conv(void)
95{
96 if (!initialized) {
97 load_case_tables();
98 init_iconv();
99 initialized = true;
100 }
101}
102
103/**
104 * Destroy global objects allocated by init_iconv()
105 **/
106void gfree_charcnv(void)
107{
108 int c1, c2;
109
110 for (c1=0;c1<NUM_CHARSETS;c1++) {
111 for (c2=0;c2<NUM_CHARSETS;c2++) {
112 if ( conv_handles[c1][c2] ) {
113 smb_iconv_close( conv_handles[c1][c2] );
114 conv_handles[c1][c2] = 0;
115 }
116 }
117 }
118 initialized = false;
119}
120
121/**
122 * Initialize iconv conversion descriptors.
123 *
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.
127 **/
128void init_iconv(void)
129{
130 int c1, c2;
131 bool did_reload = False;
132
133 /* so that charset_name() works we need to get the UNIX<->UCS2 going
134 first */
135 if (!conv_handles[CH_UNIX][CH_UTF16LE])
136 conv_handles[CH_UNIX][CH_UTF16LE] = smb_iconv_open(charset_name(CH_UTF16LE), "ASCII");
137
138 if (!conv_handles[CH_UTF16LE][CH_UNIX])
139 conv_handles[CH_UTF16LE][CH_UNIX] = smb_iconv_open("ASCII", charset_name(CH_UTF16LE));
140
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)
148 continue;
149
150 did_reload = True;
151
152 if (conv_handles[c1][c2])
153 smb_iconv_close(conv_handles[c1][c2]);
154
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) {
160 n1 = "ASCII";
161 }
162 if (c2 != CH_UTF16LE && c2 != CH_UTF16BE) {
163 n2 = "ASCII";
164 }
165 DEBUG(0,("init_iconv: Attempting to replace with conversion from %s to %s\n",
166 n1, n2 ));
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");
171 }
172 }
173 }
174 }
175
176 if (did_reload) {
177 /* XXX: Does this really get called every time the dos
178 * codepage changes? */
179 /* XXX: Is the did_reload test too strict? */
180 conv_silent = True;
181 init_valid_table();
182 conv_silent = False;
183 }
184}
185
186/**
187 * Convert string from one encoding to another, making error checking etc
188 * Slow path version - uses (slow) iconv.
189 *
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
196 *
197 * Ensure the srclen contains the terminating zero.
198 *
199 **/
200
201static 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)
204{
205 size_t i_len, o_len;
206 size_t retval;
207 const char* inbuf = (const char*)src;
208 char* outbuf = (char*)dest;
209 smb_iconv_t descriptor;
210
211 lazy_initialize_conv();
212
213 descriptor = conv_handles[from][to];
214
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;
218 } else {
219 srclen = strlen((const char *)src)+1;
220 }
221 }
222
223
224 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
225 if (!conv_silent)
226 DEBUG(0,("convert_string_internal: Conversion not supported.\n"));
227 return (size_t)-1;
228 }
229
230 i_len=srclen;
231 o_len=destlen;
232
233 again:
234
235 retval = smb_iconv(descriptor, &inbuf, &i_len, &outbuf, &o_len);
236 if(retval==(size_t)-1) {
237 const char *reason="unknown error";
238 switch(errno) {
239 case EINVAL:
240 reason="Incomplete multibyte sequence";
241 if (!conv_silent)
242 DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
243 if (allow_bad_conv)
244 goto use_as_is;
245 return (size_t)-1;
246 case E2BIG:
247 reason="No more room";
248 if (!conv_silent) {
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));
253 } else {
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));
257 }
258 }
259 break;
260 case EILSEQ:
261 reason="Illegal multibyte sequence";
262 if (!conv_silent)
263 DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
264 if (allow_bad_conv)
265 goto use_as_is;
266
267 return (size_t)-1;
268 default:
269 if (!conv_silent)
270 DEBUG(0,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
271 return (size_t)-1;
272 }
273 /* smb_panic(reason); */
274 }
275 return destlen-o_len;
276
277 use_as_is:
278
279 /*
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.
283 */
284
285 {
286 if (o_len == 0 || i_len == 0)
287 return destlen - o_len;
288
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.
293 */
294 if (i_len < 2)
295 return destlen - o_len;
296 if (i_len >= 2) {
297 *outbuf = lp_failed_convert_char();
298
299 outbuf++;
300 o_len--;
301
302 inbuf += 2;
303 i_len -= 2;
304 }
305
306 if (o_len == 0 || i_len == 0)
307 return destlen - o_len;
308
309 /* Keep trying with the next char... */
310 goto again;
311
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.
315 */
316 if (o_len < 2)
317 return destlen - o_len;
318
319 outbuf[0] = lp_failed_convert_char();
320 outbuf[1] = '\0';
321
322 inbuf++;
323 i_len--;
324
325 outbuf += 2;
326 o_len -= 2;
327
328 if (o_len == 0 || i_len == 0)
329 return destlen - o_len;
330
331 /* Keep trying with the next char... */
332 goto again;
333
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
337 try again. */
338 outbuf[0] = lp_failed_convert_char();
339
340 inbuf++;
341 i_len--;
342
343 outbuf++;
344 o_len--;
345
346 if (o_len == 0 || i_len == 0)
347 return destlen - o_len;
348
349 /* Keep trying with the next char... */
350 goto again;
351
352 } else {
353 /* Keep compiler happy.... */
354 return destlen - o_len;
355 }
356 }
357}
358
359/**
360 * Convert string from one encoding to another, making error checking etc
361 * Fast path version - handles ASCII first.
362 *
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
369 *
370 * Ensure the srclen contains the terminating zero.
371 *
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.
374 **/
375
376size_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)
379{
380 /*
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.
384 */
385
386#ifdef DEVELOPER
387 SMB_ASSERT(destlen != (size_t)-1);
388#endif
389
390 if (srclen == 0)
391 return 0;
392
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';
399 size_t retval = 0;
400
401 /* If all characters are ascii, fast path here. */
402 while (slen && dlen) {
403 if ((lastp = *p) <= 0x7f) {
404 *q++ = *p++;
405 if (slen != (size_t)-1) {
406 slen--;
407 }
408 dlen--;
409 retval++;
410 if (!lastp)
411 break;
412 } else {
413#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
414 goto general_case;
415#else
416 size_t ret = convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
417 if (ret == (size_t)-1) {
418 return ret;
419 }
420 return retval + ret;
421#endif
422 }
423 }
424 if (!dlen) {
425 /* Even if we fast path we should note if we ran out of room. */
426 if (((slen != (size_t)-1) && slen) ||
427 ((slen == (size_t)-1) && lastp)) {
428 errno = E2BIG;
429 }
430 }
431 return retval;
432 } else if (from == CH_UTF16LE && to != CH_UTF16LE) {
433 const unsigned char *p = (const unsigned char *)src;
434 unsigned char *q = (unsigned char *)dest;
435 size_t retval = 0;
436 size_t slen = srclen;
437 size_t dlen = destlen;
438 unsigned char lastp = '\0';
439
440 /* If all characters are ascii, fast path here. */
441 while (((slen == (size_t)-1) || (slen >= 2)) && dlen) {
442 if (((lastp = *p) <= 0x7f) && (p[1] == 0)) {
443 *q++ = *p;
444 if (slen != (size_t)-1) {
445 slen -= 2;
446 }
447 p += 2;
448 dlen--;
449 retval++;
450 if (!lastp)
451 break;
452 } else {
453#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
454 goto general_case;
455#else
456 size_t ret = convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
457 if (ret == (size_t)-1) {
458 return ret;
459 }
460 return retval + ret;
461#endif
462 }
463 }
464 if (!dlen) {
465 /* Even if we fast path we should note if we ran out of room. */
466 if (((slen != (size_t)-1) && slen) ||
467 ((slen == (size_t)-1) && lastp)) {
468 errno = E2BIG;
469 }
470 }
471 return retval;
472 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
473 const unsigned char *p = (const unsigned char *)src;
474 unsigned char *q = (unsigned char *)dest;
475 size_t retval = 0;
476 size_t slen = srclen;
477 size_t dlen = destlen;
478 unsigned char lastp = '\0';
479
480 /* If all characters are ascii, fast path here. */
481 while (slen && (dlen >= 2)) {
482 if ((lastp = *p) <= 0x7F) {
483 *q++ = *p++;
484 *q++ = '\0';
485 if (slen != (size_t)-1) {
486 slen--;
487 }
488 dlen -= 2;
489 retval += 2;
490 if (!lastp)
491 break;
492 } else {
493#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
494 goto general_case;
495#else
496 size_t ret = convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
497 if (ret == (size_t)-1) {
498 return ret;
499 }
500 return retval + ret;
501#endif
502 }
503 }
504 if (!dlen) {
505 /* Even if we fast path we should note if we ran out of room. */
506 if (((slen != (size_t)-1) && slen) ||
507 ((slen == (size_t)-1) && lastp)) {
508 errno = E2BIG;
509 }
510 }
511 return retval;
512 }
513
514#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
515 general_case:
516#endif
517 return convert_string_internal(from, to, src, srclen, dest, destlen, allow_bad_conv);
518}
519
520/**
521 * Convert between character sets, allocating a new buffer for the result.
522 *
523 * @param ctx TALLOC_CTX to use to allocate with. If NULL use malloc.
524 * (this is a bad interface and needs fixing. JRA).
525 * @param srclen length of source buffer.
526 * @param dest always set at least to NULL
527 * @param converted_size set to the size of the allocated buffer on return
528 * true
529 * @note -1 is not accepted for srclen.
530 *
531 * @return true if new buffer was correctly allocated, and string was
532 * converted.
533 *
534 * Ensure the srclen contains the terminating zero.
535 *
536 * I hate the goto's in this function. It's embarressing.....
537 * There has to be a cleaner way to do this. JRA.
538 **/
539
540bool convert_string_allocate(TALLOC_CTX *ctx, charset_t from, charset_t to,
541 void const *src, size_t srclen, void *dst,
542 size_t *converted_size, bool allow_bad_conv)
543{
544 size_t i_len, o_len, destlen = (srclen * 3) / 2;
545 size_t retval;
546 const char *inbuf = (const char *)src;
547 char *outbuf = NULL, *ob = NULL;
548 smb_iconv_t descriptor;
549 void **dest = (void **)dst;
550
551 *dest = NULL;
552
553 if (!converted_size) {
554 errno = EINVAL;
555 return false;
556 }
557
558 if (src == NULL || srclen == (size_t)-1) {
559 errno = EINVAL;
560 return false;
561 }
562 if (srclen == 0) {
563 ob = ((ctx != NULL) ? talloc_strdup(ctx, "") : SMB_STRDUP(""));
564 if (ob == NULL) {
565 errno = ENOMEM;
566 return false;
567 }
568 *dest = ob;
569 *converted_size = 0;
570 return true;
571 }
572
573 lazy_initialize_conv();
574
575 descriptor = conv_handles[from][to];
576
577 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
578 if (!conv_silent)
579 DEBUG(0,("convert_string_allocate: Conversion not supported.\n"));
580 errno = EOPNOTSUPP;
581 return false;
582 }
583
584 convert:
585
586 /* +2 is for ucs2 null termination. */
587 if ((destlen*2)+2 < destlen) {
588 /* wrapped ! abort. */
589 if (!conv_silent)
590 DEBUG(0, ("convert_string_allocate: destlen wrapped !\n"));
591 if (!ctx)
592 SAFE_FREE(outbuf);
593 errno = EOPNOTSUPP;
594 return false;
595 } else {
596 destlen = destlen * 2;
597 }
598
599 /* +2 is for ucs2 null termination. */
600 if (ctx) {
601 ob = (char *)TALLOC_REALLOC(ctx, ob, destlen + 2);
602 } else {
603 ob = (char *)SMB_REALLOC(ob, destlen + 2);
604 }
605
606 if (!ob) {
607 DEBUG(0, ("convert_string_allocate: realloc failed!\n"));
608 errno = ENOMEM;
609 return false;
610 }
611 outbuf = ob;
612 i_len = srclen;
613 o_len = destlen;
614
615 again:
616
617 retval = smb_iconv(descriptor,
618 &inbuf, &i_len,
619 &outbuf, &o_len);
620 if(retval == (size_t)-1) {
621 const char *reason="unknown error";
622 switch(errno) {
623 case EINVAL:
624 reason="Incomplete multibyte sequence";
625 if (!conv_silent)
626 DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
627 if (allow_bad_conv)
628 goto use_as_is;
629 break;
630 case E2BIG:
631 goto convert;
632 case EILSEQ:
633 reason="Illegal multibyte sequence";
634 if (!conv_silent)
635 DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
636 if (allow_bad_conv)
637 goto use_as_is;
638 break;
639 }
640 if (!conv_silent)
641 DEBUG(0,("Conversion error: %s(%s)\n",reason,inbuf));
642 /* smb_panic(reason); */
643 if (ctx) {
644 TALLOC_FREE(ob);
645 } else {
646 SAFE_FREE(ob);
647 }
648 return false;
649 }
650
651 out:
652
653 destlen = destlen - o_len;
654 /* Don't shrink unless we're reclaiming a lot of
655 * space. This is in the hot codepath and these
656 * reallocs *cost*. JRA.
657 */
658 if (o_len > 1024) {
659 /* We're shrinking here so we know the +2 is safe from wrap. */
660 if (ctx) {
661 ob = (char *)TALLOC_REALLOC(ctx,ob,destlen + 2);
662 } else {
663 ob = (char *)SMB_REALLOC(ob,destlen + 2);
664 }
665 }
666
667 if (destlen && !ob) {
668 DEBUG(0, ("convert_string_allocate: out of memory!\n"));
669 errno = ENOMEM;
670 return false;
671 }
672
673 *dest = ob;
674
675 /* Must ucs2 null terminate in the extra space we allocated. */
676 ob[destlen] = '\0';
677 ob[destlen+1] = '\0';
678
679 *converted_size = destlen;
680 return true;
681
682 use_as_is:
683
684 /*
685 * Conversion not supported. This is actually an error, but there are so
686 * many misconfigured iconv systems and smb.conf's out there we can't just
687 * fail. Do a very bad conversion instead.... JRA.
688 */
689
690 {
691 if (o_len == 0 || i_len == 0)
692 goto out;
693
694 if (((from == CH_UTF16LE)||(from == CH_UTF16BE)) &&
695 ((to != CH_UTF16LE)||(to != CH_UTF16BE))) {
696 /* Can't convert from utf16 any endian to multibyte.
697 Replace with the default fail char.
698 */
699
700 if (i_len < 2)
701 goto out;
702
703 if (i_len >= 2) {
704 *outbuf = lp_failed_convert_char();
705
706 outbuf++;
707 o_len--;
708
709 inbuf += 2;
710 i_len -= 2;
711 }
712
713 if (o_len == 0 || i_len == 0)
714 goto out;
715
716 /* Keep trying with the next char... */
717 goto again;
718
719 } else if (from != CH_UTF16LE && from != CH_UTF16BE && to == CH_UTF16LE) {
720 /* Can't convert to UTF16LE - just widen by adding the
721 default fail char then zero.
722 */
723 if (o_len < 2)
724 goto out;
725
726 outbuf[0] = lp_failed_convert_char();
727 outbuf[1] = '\0';
728
729 inbuf++;
730 i_len--;
731
732 outbuf += 2;
733 o_len -= 2;
734
735 if (o_len == 0 || i_len == 0)
736 goto out;
737
738 /* Keep trying with the next char... */
739 goto again;
740
741 } else if (from != CH_UTF16LE && from != CH_UTF16BE &&
742 to != CH_UTF16LE && to != CH_UTF16BE) {
743 /* Failed multibyte to multibyte. Just copy the default fail char and
744 try again. */
745 outbuf[0] = lp_failed_convert_char();
746
747 inbuf++;
748 i_len--;
749
750 outbuf++;
751 o_len--;
752
753 if (o_len == 0 || i_len == 0)
754 goto out;
755
756 /* Keep trying with the next char... */
757 goto again;
758
759 } else {
760 /* Keep compiler happy.... */
761 goto out;
762 }
763 }
764}
765
766/**
767 * Convert between character sets, allocating a new buffer using talloc for the result.
768 *
769 * @param srclen length of source buffer.
770 * @param dest always set at least to NULL
771 * @parm converted_size set to the number of bytes occupied by the string in
772 * the destination on success.
773 * @note -1 is not accepted for srclen.
774 *
775 * @return true if new buffer was correctly allocated, and string was
776 * converted.
777 */
778bool convert_string_talloc(TALLOC_CTX *ctx, charset_t from, charset_t to,
779 void const *src, size_t srclen, void *dst,
780 size_t *converted_size, bool allow_bad_conv)
781{
782 void **dest = (void **)dst;
783
784 *dest = NULL;
785 return convert_string_allocate(ctx, from, to, src, srclen, dest,
786 converted_size, allow_bad_conv);
787}
788
789size_t unix_strupper(const char *src, size_t srclen, char *dest, size_t destlen)
790{
791 size_t size;
792 smb_ucs2_t *buffer;
793
794 if (!push_ucs2_allocate(&buffer, src, &size)) {
795 return (size_t)-1;
796 }
797
798 if (!strupper_w(buffer) && (dest == src)) {
799 free(buffer);
800 return srclen;
801 }
802
803 size = convert_string(CH_UTF16LE, CH_UNIX, buffer, size, dest, destlen, True);
804 free(buffer);
805 return size;
806}
807
808/**
809 strdup() a unix string to upper case.
810**/
811
812char *strdup_upper(const char *s)
813{
814 char *out_buffer = SMB_STRDUP(s);
815 const unsigned char *p = (const unsigned char *)s;
816 unsigned char *q = (unsigned char *)out_buffer;
817
818 if (!q) {
819 return NULL;
820 }
821
822 /* this is quite a common operation, so we want it to be
823 fast. We optimise for the ascii case, knowing that all our
824 supported multi-byte character sets are ascii-compatible
825 (ie. they match for the first 128 chars) */
826
827 while (*p) {
828 if (*p & 0x80)
829 break;
830 *q++ = toupper_ascii_fast(*p);
831 p++;
832 }
833
834 if (*p) {
835 /* MB case. */
836 size_t converted_size, converted_size2;
837 smb_ucs2_t *buffer = NULL;
838
839 SAFE_FREE(out_buffer);
840 if (!convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, s,
841 strlen(s) + 1,
842 (void **)(void *)&buffer,
843 &converted_size, True))
844 {
845 return NULL;
846 }
847
848 strupper_w(buffer);
849
850 if (!convert_string_allocate(NULL, CH_UTF16LE, CH_UNIX, buffer,
851 converted_size,
852 (void **)(void *)&out_buffer,
853 &converted_size2, True))
854 {
855 TALLOC_FREE(buffer);
856 return NULL;
857 }
858
859 /* Don't need the intermediate buffer
860 * anymore.
861 */
862 TALLOC_FREE(buffer);
863 }
864
865 return out_buffer;
866}
867
868/**
869 talloc_strdup() a unix string to upper case.
870**/
871
872char *talloc_strdup_upper(TALLOC_CTX *ctx, const char *s)
873{
874 char *out_buffer = talloc_strdup(ctx,s);
875 const unsigned char *p = (const unsigned char *)s;
876 unsigned char *q = (unsigned char *)out_buffer;
877
878 if (!q) {
879 return NULL;
880 }
881
882 /* this is quite a common operation, so we want it to be
883 fast. We optimise for the ascii case, knowing that all our
884 supported multi-byte character sets are ascii-compatible
885 (ie. they match for the first 128 chars) */
886
887 while (*p) {
888 if (*p & 0x80)
889 break;
890 *q++ = toupper_ascii_fast(*p);
891 p++;
892 }
893
894 if (*p) {
895 /* MB case. */
896 size_t converted_size, converted_size2;
897 smb_ucs2_t *ubuf = NULL;
898
899 /* We're not using the ascii buffer above. */
900 TALLOC_FREE(out_buffer);
901
902 if (!convert_string_talloc(ctx, CH_UNIX, CH_UTF16LE, s,
903 strlen(s)+1, (void *)&ubuf,
904 &converted_size, True))
905 {
906 return NULL;
907 }
908
909 strupper_w(ubuf);
910
911 if (!convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX, ubuf,
912 converted_size, (void *)&out_buffer,
913 &converted_size2, True))
914 {
915 TALLOC_FREE(ubuf);
916 return NULL;
917 }
918
919 /* Don't need the intermediate buffer
920 * anymore.
921 */
922 TALLOC_FREE(ubuf);
923 }
924
925 return out_buffer;
926}
927
928size_t unix_strlower(const char *src, size_t srclen, char *dest, size_t destlen)
929{
930 size_t size;
931 smb_ucs2_t *buffer = NULL;
932
933 if (!convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, src, srclen,
934 (void **)(void *)&buffer, &size,
935 True))
936 {
937 smb_panic("failed to create UCS2 buffer");
938 }
939 if (!strlower_w(buffer) && (dest == src)) {
940 SAFE_FREE(buffer);
941 return srclen;
942 }
943 size = convert_string(CH_UTF16LE, CH_UNIX, buffer, size, dest, destlen, True);
944 SAFE_FREE(buffer);
945 return size;
946}
947
948/**
949 strdup() a unix string to lower case.
950**/
951
952char *strdup_lower(const char *s)
953{
954 size_t converted_size;
955 smb_ucs2_t *buffer = NULL;
956 char *out_buffer;
957
958 if (!push_ucs2_allocate(&buffer, s, &converted_size)) {
959 return NULL;
960 }
961
962 strlower_w(buffer);
963
964 if (!pull_ucs2_allocate(&out_buffer, buffer, &converted_size)) {
965 SAFE_FREE(buffer);
966 return NULL;
967 }
968
969 SAFE_FREE(buffer);
970
971 return out_buffer;
972}
973
974char *talloc_strdup_lower(TALLOC_CTX *ctx, const char *s)
975{
976 size_t converted_size;
977 smb_ucs2_t *buffer = NULL;
978 char *out_buffer;
979
980 if (!push_ucs2_talloc(ctx, &buffer, s, &converted_size)) {
981 return NULL;
982 }
983
984 strlower_w(buffer);
985
986 if (!pull_ucs2_talloc(ctx, &out_buffer, buffer, &converted_size)) {
987 TALLOC_FREE(buffer);
988 return NULL;
989 }
990
991 TALLOC_FREE(buffer);
992
993 return out_buffer;
994}
995
996
997size_t ucs2_align(const void *base_ptr, const void *p, int flags)
998{
999 if (flags & (STR_NOALIGN|STR_ASCII))
1000 return 0;
1001 return PTR_DIFF(p, base_ptr) & 1;
1002}
1003
1004
1005/**
1006 * Copy a string from a char* unix src to a dos codepage string destination.
1007 *
1008 * @return the number of bytes occupied by the string in the destination.
1009 *
1010 * @param flags can include
1011 * <dl>
1012 * <dt>STR_TERMINATE</dt> <dd>means include the null termination</dd>
1013 * <dt>STR_UPPER</dt> <dd>means uppercase in the destination</dd>
1014 * </dl>
1015 *
1016 * @param dest_len the maximum length in bytes allowed in the
1017 * destination.
1018 **/
1019size_t push_ascii(void *dest, const char *src, size_t dest_len, int flags)
1020{
1021 size_t src_len = strlen(src);
1022 char *tmpbuf = NULL;
1023 size_t ret;
1024
1025 /* No longer allow a length of -1. */
1026 if (dest_len == (size_t)-1) {
1027 smb_panic("push_ascii - dest_len == -1");
1028 }
1029
1030 if (flags & STR_UPPER) {
1031 tmpbuf = SMB_STRDUP(src);
1032 if (!tmpbuf) {
1033 smb_panic("malloc fail");
1034 }
1035 strupper_m(tmpbuf);
1036 src = tmpbuf;
1037 }
1038
1039 if (flags & (STR_TERMINATE | STR_TERMINATE_ASCII)) {
1040 src_len++;
1041 }
1042
1043 ret = convert_string(CH_UNIX, CH_DOS, src, src_len, dest, dest_len, True);
1044 if (ret == (size_t)-1 &&
1045 (flags & (STR_TERMINATE | STR_TERMINATE_ASCII))
1046 && dest_len > 0) {
1047 ((char *)dest)[0] = '\0';
1048 }
1049 SAFE_FREE(tmpbuf);
1050 return ret;
1051}
1052
1053size_t push_ascii_fstring(void *dest, const char *src)
1054{
1055 return push_ascii(dest, src, sizeof(fstring), STR_TERMINATE);
1056}
1057
1058/********************************************************************
1059 Push an nstring - ensure null terminated. Written by
1060 moriyama@miraclelinux.com (MORIYAMA Masayuki).
1061********************************************************************/
1062
1063size_t push_ascii_nstring(void *dest, const char *src)
1064{
1065 size_t i, buffer_len, dest_len;
1066 smb_ucs2_t *buffer;
1067
1068 conv_silent = True;
1069 if (!push_ucs2_allocate(&buffer, src, &buffer_len)) {
1070 smb_panic("failed to create UCS2 buffer");
1071 }
1072
1073 /* We're using buffer_len below to count ucs2 characters, not bytes. */
1074 buffer_len /= sizeof(smb_ucs2_t);
1075
1076 dest_len = 0;
1077 for (i = 0; buffer[i] != 0 && (i < buffer_len); i++) {
1078 unsigned char mb[10];
1079 /* Convert one smb_ucs2_t character at a time. */
1080 size_t mb_len = convert_string(CH_UTF16LE, CH_DOS, buffer+i, sizeof(smb_ucs2_t), mb, sizeof(mb), False);
1081 if ((mb_len != (size_t)-1) && (dest_len + mb_len <= MAX_NETBIOSNAME_LEN - 1)) {
1082 memcpy((char *)dest + dest_len, mb, mb_len);
1083 dest_len += mb_len;
1084 } else {
1085 errno = E2BIG;
1086 break;
1087 }
1088 }
1089 ((char *)dest)[dest_len] = '\0';
1090
1091 SAFE_FREE(buffer);
1092 conv_silent = False;
1093 return dest_len;
1094}
1095
1096/********************************************************************
1097 Push and malloc an ascii string. src and dest null terminated.
1098********************************************************************/
1099
1100bool push_ascii_allocate(char **dest, const char *src, size_t *converted_size)
1101{
1102 size_t src_len = strlen(src)+1;
1103
1104 *dest = NULL;
1105 return convert_string_allocate(NULL, CH_UNIX, CH_DOS, src, src_len,
1106 (void **)dest, converted_size, True);
1107}
1108
1109/**
1110 * Copy a string from a dos codepage source to a unix char* destination.
1111 *
1112 * The resulting string in "dest" is always null terminated.
1113 *
1114 * @param flags can have:
1115 * <dl>
1116 * <dt>STR_TERMINATE</dt>
1117 * <dd>STR_TERMINATE means the string in @p src
1118 * is null terminated, and src_len is ignored.</dd>
1119 * </dl>
1120 *
1121 * @param src_len is the length of the source area in bytes.
1122 * @returns the number of bytes occupied by the string in @p src.
1123 **/
1124size_t pull_ascii(char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
1125{
1126 size_t ret;
1127
1128 if (dest_len == (size_t)-1) {
1129 /* No longer allow dest_len of -1. */
1130 smb_panic("pull_ascii - invalid dest_len of -1");
1131 }
1132
1133 if (flags & STR_TERMINATE) {
1134 if (src_len == (size_t)-1) {
1135 src_len = strlen((const char *)src) + 1;
1136 } else {
1137 size_t len = strnlen((const char *)src, src_len);
1138 if (len < src_len)
1139 len++;
1140 src_len = len;
1141 }
1142 }
1143
1144 ret = convert_string(CH_DOS, CH_UNIX, src, src_len, dest, dest_len, True);
1145 if (ret == (size_t)-1) {
1146 ret = 0;
1147 dest_len = 0;
1148 }
1149
1150 if (dest_len && ret) {
1151 /* Did we already process the terminating zero ? */
1152 if (dest[MIN(ret-1, dest_len-1)] != 0) {
1153 dest[MIN(ret, dest_len-1)] = 0;
1154 }
1155 } else {
1156 dest[0] = 0;
1157 }
1158
1159 return src_len;
1160}
1161
1162/**
1163 * Copy a string from a dos codepage source to a unix char* destination.
1164 Talloc version.
1165 Uses malloc if TALLOC_CTX is NULL (this is a bad interface and
1166 needs fixing. JRA).
1167 *
1168 * The resulting string in "dest" is always null terminated.
1169 *
1170 * @param flags can have:
1171 * <dl>
1172 * <dt>STR_TERMINATE</dt>
1173 * <dd>STR_TERMINATE means the string in @p src
1174 * is null terminated, and src_len is ignored.</dd>
1175 * </dl>
1176 *
1177 * @param src_len is the length of the source area in bytes.
1178 * @returns the number of bytes occupied by the string in @p src.
1179 **/
1180
1181static size_t pull_ascii_base_talloc(TALLOC_CTX *ctx,
1182 char **ppdest,
1183 const void *src,
1184 size_t src_len,
1185 int flags)
1186{
1187 char *dest = NULL;
1188 size_t dest_len;
1189
1190#ifdef DEVELOPER
1191 /* Ensure we never use the braindead "malloc" varient. */
1192 if (ctx == NULL) {
1193 smb_panic("NULL talloc CTX in pull_ascii_base_talloc\n");
1194 }
1195#endif
1196
1197 *ppdest = NULL;
1198
1199 if (!src_len) {
1200 return 0;
1201 }
1202
1203 if (flags & STR_TERMINATE) {
1204 if (src_len == (size_t)-1) {
1205 src_len = strlen((const char *)src) + 1;
1206 } else {
1207 size_t len = strnlen((const char *)src, src_len);
1208 if (len < src_len)
1209 len++;
1210 src_len = len;
1211 }
1212 /* Ensure we don't use an insane length from the client. */
1213 if (src_len >= 1024*1024) {
1214 char *msg = talloc_asprintf(ctx,
1215 "Bad src length (%u) in "
1216 "pull_ascii_base_talloc",
1217 (unsigned int)src_len);
1218 smb_panic(msg);
1219 }
1220 } else {
1221 /* Can't have an unlimited length
1222 * non STR_TERMINATE'd.
1223 */
1224 if (src_len == (size_t)-1) {
1225 errno = EINVAL;
1226 return 0;
1227 }
1228 }
1229
1230 /* src_len != -1 here. */
1231
1232 if (!convert_string_allocate(ctx, CH_DOS, CH_UNIX, src, src_len, &dest,
1233 &dest_len, True)) {
1234 dest_len = 0;
1235 }
1236
1237 if (dest_len && dest) {
1238 /* Did we already process the terminating zero ? */
1239 if (dest[dest_len-1] != 0) {
1240 size_t size = talloc_get_size(dest);
1241 /* Have we got space to append the '\0' ? */
1242 if (size <= dest_len) {
1243 /* No, realloc. */
1244 dest = TALLOC_REALLOC_ARRAY(ctx, dest, char,
1245 dest_len+1);
1246 if (!dest) {
1247 /* talloc fail. */
1248 dest_len = (size_t)-1;
1249 return 0;
1250 }
1251 }
1252 /* Yay - space ! */
1253 dest[dest_len] = '\0';
1254 dest_len++;
1255 }
1256 } else if (dest) {
1257 dest[0] = 0;
1258 }
1259
1260 *ppdest = dest;
1261 return src_len;
1262}
1263
1264size_t pull_ascii_fstring(char *dest, const void *src)
1265{
1266 return pull_ascii(dest, src, sizeof(fstring), -1, STR_TERMINATE);
1267}
1268
1269/* When pulling an nstring it can expand into a larger size (dos cp -> utf8). Cope with this. */
1270
1271size_t pull_ascii_nstring(char *dest, size_t dest_len, const void *src)
1272{
1273 return pull_ascii(dest, src, dest_len, sizeof(nstring)-1, STR_TERMINATE);
1274}
1275
1276/**
1277 * Copy a string from a char* src to a unicode destination.
1278 *
1279 * @returns the number of bytes occupied by the string in the destination.
1280 *
1281 * @param flags can have:
1282 *
1283 * <dl>
1284 * <dt>STR_TERMINATE <dd>means include the null termination.
1285 * <dt>STR_UPPER <dd>means uppercase in the destination.
1286 * <dt>STR_NOALIGN <dd>means don't do alignment.
1287 * </dl>
1288 *
1289 * @param dest_len is the maximum length allowed in the
1290 * destination.
1291 **/
1292
1293size_t push_ucs2(const void *base_ptr, void *dest, const char *src, size_t dest_len, int flags)
1294{
1295 size_t len=0;
1296 size_t src_len;
1297 size_t ret;
1298
1299 if (dest_len == (size_t)-1) {
1300 /* No longer allow dest_len of -1. */
1301 smb_panic("push_ucs2 - invalid dest_len of -1");
1302 }
1303
1304 if (flags & STR_TERMINATE)
1305 src_len = (size_t)-1;
1306 else
1307 src_len = strlen(src);
1308
1309 if (ucs2_align(base_ptr, dest, flags)) {
1310 *(char *)dest = 0;
1311 dest = (void *)((char *)dest + 1);
1312 if (dest_len)
1313 dest_len--;
1314 len++;
1315 }
1316
1317 /* ucs2 is always a multiple of 2 bytes */
1318 dest_len &= ~1;
1319
1320 ret = convert_string(CH_UNIX, CH_UTF16LE, src, src_len, dest, dest_len, True);
1321 if (ret == (size_t)-1) {
1322 if ((flags & STR_TERMINATE) &&
1323 dest &&
1324 dest_len) {
1325 *(char *)dest = 0;
1326 }
1327 return len;
1328 }
1329
1330 len += ret;
1331
1332 if (flags & STR_UPPER) {
1333 smb_ucs2_t *dest_ucs2 = (smb_ucs2_t *)dest;
1334 size_t i;
1335
1336 /* We check for i < (ret / 2) below as the dest string isn't null
1337 terminated if STR_TERMINATE isn't set. */
1338
1339 for (i = 0; i < (ret / 2) && i < (dest_len / 2) && dest_ucs2[i]; i++) {
1340 smb_ucs2_t v = toupper_w(dest_ucs2[i]);
1341 if (v != dest_ucs2[i]) {
1342 dest_ucs2[i] = v;
1343 }
1344 }
1345 }
1346
1347 return len;
1348}
1349
1350
1351/**
1352 * Copy a string from a unix char* src to a UCS2 destination,
1353 * allocating a buffer using talloc().
1354 *
1355 * @param dest always set at least to NULL
1356 * @parm converted_size set to the number of bytes occupied by the string in
1357 * the destination on success.
1358 *
1359 * @return true if new buffer was correctly allocated, and string was
1360 * converted.
1361 **/
1362bool push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src,
1363 size_t *converted_size)
1364{
1365 size_t src_len = strlen(src)+1;
1366
1367 *dest = NULL;
1368 return convert_string_talloc(ctx, CH_UNIX, CH_UTF16LE, src, src_len,
1369 (void **)dest, converted_size, True);
1370}
1371
1372
1373/**
1374 * Copy a string from a unix char* src to a UCS2 destination, allocating a buffer
1375 *
1376 * @param dest always set at least to NULL
1377 * @parm converted_size set to the number of bytes occupied by the string in
1378 * the destination on success.
1379 *
1380 * @return true if new buffer was correctly allocated, and string was
1381 * converted.
1382 **/
1383
1384bool push_ucs2_allocate(smb_ucs2_t **dest, const char *src,
1385 size_t *converted_size)
1386{
1387 size_t src_len = strlen(src)+1;
1388
1389 *dest = NULL;
1390 return convert_string_allocate(NULL, CH_UNIX, CH_UTF16LE, src, src_len,
1391 (void **)dest, converted_size, True);
1392}
1393
1394/**
1395 Copy a string from a char* src to a UTF-8 destination.
1396 Return the number of bytes occupied by the string in the destination
1397 Flags can have:
1398 STR_TERMINATE means include the null termination
1399 STR_UPPER means uppercase in the destination
1400 dest_len is the maximum length allowed in the destination. If dest_len
1401 is -1 then no maxiumum is used.
1402**/
1403
1404static size_t push_utf8(void *dest, const char *src, size_t dest_len, int flags)
1405{
1406 size_t src_len = 0;
1407 size_t ret;
1408 char *tmpbuf = NULL;
1409
1410 if (dest_len == (size_t)-1) {
1411 /* No longer allow dest_len of -1. */
1412 smb_panic("push_utf8 - invalid dest_len of -1");
1413 }
1414
1415 if (flags & STR_UPPER) {
1416 tmpbuf = strdup_upper(src);
1417 if (!tmpbuf) {
1418 return (size_t)-1;
1419 }
1420 src = tmpbuf;
1421 src_len = strlen(src);
1422 }
1423
1424 src_len = strlen(src);
1425 if (flags & STR_TERMINATE) {
1426 src_len++;
1427 }
1428
1429 ret = convert_string(CH_UNIX, CH_UTF8, src, src_len, dest, dest_len, True);
1430 SAFE_FREE(tmpbuf);
1431 return ret;
1432}
1433
1434size_t push_utf8_fstring(void *dest, const char *src)
1435{
1436 return push_utf8(dest, src, sizeof(fstring), STR_TERMINATE);
1437}
1438
1439/**
1440 * Copy a string from a unix char* src to a UTF-8 destination, allocating a buffer using talloc
1441 *
1442 * @param dest always set at least to NULL
1443 * @parm converted_size set to the number of bytes occupied by the string in
1444 * the destination on success.
1445 *
1446 * @return true if new buffer was correctly allocated, and string was
1447 * converted.
1448 **/
1449
1450bool push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src,
1451 size_t *converted_size)
1452{
1453 size_t src_len = strlen(src)+1;
1454
1455 *dest = NULL;
1456 return convert_string_talloc(ctx, CH_UNIX, CH_UTF8, src, src_len,
1457 (void**)dest, converted_size, True);
1458}
1459
1460/**
1461 * Copy a string from a unix char* src to a UTF-8 destination, allocating a buffer
1462 *
1463 * @param dest always set at least to NULL
1464 * @parm converted_size set to the number of bytes occupied by the string in
1465 * the destination on success.
1466 *
1467 * @return true if new buffer was correctly allocated, and string was
1468 * converted.
1469 **/
1470
1471bool push_utf8_allocate(char **dest, const char *src, size_t *converted_size)
1472{
1473 size_t src_len = strlen(src)+1;
1474
1475 *dest = NULL;
1476 return convert_string_allocate(NULL, CH_UNIX, CH_UTF8, src, src_len,
1477 (void **)dest, converted_size, True);
1478}
1479
1480/**
1481 Copy a string from a ucs2 source to a unix char* destination.
1482 Flags can have:
1483 STR_TERMINATE means the string in src is null terminated.
1484 STR_NOALIGN means don't try to align.
1485 if STR_TERMINATE is set then src_len is ignored if it is -1.
1486 src_len is the length of the source area in bytes
1487 Return the number of bytes occupied by the string in src.
1488 The resulting string in "dest" is always null terminated.
1489**/
1490
1491size_t pull_ucs2(const void *base_ptr, char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
1492{
1493 size_t ret;
1494
1495 if (dest_len == (size_t)-1) {
1496 /* No longer allow dest_len of -1. */
1497 smb_panic("pull_ucs2 - invalid dest_len of -1");
1498 }
1499
1500 if (!src_len) {
1501 if (dest && dest_len > 0) {
1502 dest[0] = '\0';
1503 }
1504 return 0;
1505 }
1506
1507 if (ucs2_align(base_ptr, src, flags)) {
1508 src = (const void *)((const char *)src + 1);
1509 if (src_len != (size_t)-1)
1510 src_len--;
1511 }
1512
1513 if (flags & STR_TERMINATE) {
1514 /* src_len -1 is the default for null terminated strings. */
1515 if (src_len != (size_t)-1) {
1516 size_t len = strnlen_w((const smb_ucs2_t *)src,
1517 src_len/2);
1518 if (len < src_len/2)
1519 len++;
1520 src_len = len*2;
1521 }
1522 }
1523
1524 /* ucs2 is always a multiple of 2 bytes */
1525 if (src_len != (size_t)-1)
1526 src_len &= ~1;
1527
1528 ret = convert_string(CH_UTF16LE, CH_UNIX, src, src_len, dest, dest_len, True);
1529 if (ret == (size_t)-1) {
1530 ret = 0;
1531 dest_len = 0;
1532 }
1533
1534 if (src_len == (size_t)-1)
1535 src_len = ret*2;
1536
1537 if (dest_len && ret) {
1538 /* Did we already process the terminating zero ? */
1539 if (dest[MIN(ret-1, dest_len-1)] != 0) {
1540 dest[MIN(ret, dest_len-1)] = 0;
1541 }
1542 } else {
1543 dest[0] = 0;
1544 }
1545
1546 return src_len;
1547}
1548
1549/**
1550 Copy a string from a ucs2 source to a unix char* destination.
1551 Talloc version with a base pointer.
1552 Uses malloc if TALLOC_CTX is NULL (this is a bad interface and
1553 needs fixing. JRA).
1554 Flags can have:
1555 STR_TERMINATE means the string in src is null terminated.
1556 STR_NOALIGN means don't try to align.
1557 if STR_TERMINATE is set then src_len is ignored if it is -1.
1558 src_len is the length of the source area in bytes
1559 Return the number of bytes occupied by the string in src.
1560 The resulting string in "dest" is always null terminated.
1561**/
1562
1563size_t pull_ucs2_base_talloc(TALLOC_CTX *ctx,
1564 const void *base_ptr,
1565 char **ppdest,
1566 const void *src,
1567 size_t src_len,
1568 int flags)
1569{
1570 char *dest;
1571 size_t dest_len;
1572
1573 *ppdest = NULL;
1574
1575#ifdef DEVELOPER
1576 /* Ensure we never use the braindead "malloc" varient. */
1577 if (ctx == NULL) {
1578 smb_panic("NULL talloc CTX in pull_ucs2_base_talloc\n");
1579 }
1580#endif
1581
1582 if (!src_len) {
1583 return 0;
1584 }
1585
1586 if (ucs2_align(base_ptr, src, flags)) {
1587 src = (const void *)((const char *)src + 1);
1588 if (src_len != (size_t)-1)
1589 src_len--;
1590 }
1591
1592 if (flags & STR_TERMINATE) {
1593 /* src_len -1 is the default for null terminated strings. */
1594 if (src_len != (size_t)-1) {
1595 size_t len = strnlen_w((const smb_ucs2_t *)src,
1596 src_len/2);
1597 if (len < src_len/2)
1598 len++;
1599 src_len = len*2;
1600 } else {
1601 /*
1602 * src_len == -1 - alloc interface won't take this
1603 * so we must calculate.
1604 */
1605 src_len = (strlen_w((const smb_ucs2_t *)src)+1)*sizeof(smb_ucs2_t);
1606 }
1607 /* Ensure we don't use an insane length from the client. */
1608 if (src_len >= 1024*1024) {
1609 smb_panic("Bad src length in pull_ucs2_base_talloc\n");
1610 }
1611 } else {
1612 /* Can't have an unlimited length
1613 * non STR_TERMINATE'd.
1614 */
1615 if (src_len == (size_t)-1) {
1616 errno = EINVAL;
1617 return 0;
1618 }
1619 }
1620
1621 /* src_len != -1 here. */
1622
1623 /* ucs2 is always a multiple of 2 bytes */
1624 src_len &= ~1;
1625
1626 if (!convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX, src, src_len,
1627 (void *)&dest, &dest_len, True)) {
1628 dest_len = 0;
1629 }
1630
1631 if (dest_len) {
1632 /* Did we already process the terminating zero ? */
1633 if (dest[dest_len-1] != 0) {
1634 size_t size = talloc_get_size(dest);
1635 /* Have we got space to append the '\0' ? */
1636 if (size <= dest_len) {
1637 /* No, realloc. */
1638 dest = TALLOC_REALLOC_ARRAY(ctx, dest, char,
1639 dest_len+1);
1640 if (!dest) {
1641 /* talloc fail. */
1642 dest_len = (size_t)-1;
1643 return 0;
1644 }
1645 }
1646 /* Yay - space ! */
1647 dest[dest_len] = '\0';
1648 dest_len++;
1649 }
1650 } else if (dest) {
1651 dest[0] = 0;
1652 }
1653
1654 *ppdest = dest;
1655 return src_len;
1656}
1657
1658size_t pull_ucs2_fstring(char *dest, const void *src)
1659{
1660 return pull_ucs2(NULL, dest, src, sizeof(fstring), -1, STR_TERMINATE);
1661}
1662
1663/**
1664 * Copy a string from a UCS2 src to a unix char * destination, allocating a buffer using talloc
1665 *
1666 * @param dest always set at least to NULL
1667 * @parm converted_size set to the number of bytes occupied by the string in
1668 * the destination on success.
1669 *
1670 * @return true if new buffer was correctly allocated, and string was
1671 * converted.
1672 **/
1673
1674bool pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src,
1675 size_t *converted_size)
1676{
1677 size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
1678
1679 *dest = NULL;
1680 return convert_string_talloc(ctx, CH_UTF16LE, CH_UNIX, src, src_len,
1681 (void **)dest, converted_size, True);
1682}
1683
1684/**
1685 * Copy a string from a UCS2 src to a unix char * destination, allocating a buffer
1686 *
1687 * @param dest always set at least to NULL
1688 * @parm converted_size set to the number of bytes occupied by the string in
1689 * the destination on success.
1690 * @return true if new buffer was correctly allocated, and string was
1691 * converted.
1692 **/
1693
1694bool pull_ucs2_allocate(char **dest, const smb_ucs2_t *src,
1695 size_t *converted_size)
1696{
1697 size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
1698
1699 *dest = NULL;
1700 return convert_string_allocate(NULL, CH_UTF16LE, CH_UNIX, src, src_len,
1701 (void **)dest, converted_size, True);
1702}
1703
1704/**
1705 * Copy a string from a UTF-8 src to a unix char * destination, allocating a buffer using talloc
1706 *
1707 * @param dest always set at least to NULL
1708 * @parm converted_size set to the number of bytes occupied by the string in
1709 * the destination on success.
1710 *
1711 * @return true if new buffer was correctly allocated, and string was
1712 * converted.
1713 **/
1714
1715bool pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src,
1716 size_t *converted_size)
1717{
1718 size_t src_len = strlen(src)+1;
1719
1720 *dest = NULL;
1721 return convert_string_talloc(ctx, CH_UTF8, CH_UNIX, src, src_len,
1722 (void **)dest, converted_size, True);
1723}
1724
1725/**
1726 * Copy a string from a UTF-8 src to a unix char * destination, allocating a buffer
1727 *
1728 * @param dest always set at least to NULL
1729 * @parm converted_size set to the number of bytes occupied by the string in
1730 * the destination on success.
1731 *
1732 * @return true if new buffer was correctly allocated, and string was
1733 * converted.
1734 **/
1735
1736bool pull_utf8_allocate(char **dest, const char *src, size_t *converted_size)
1737{
1738 size_t src_len = strlen(src)+1;
1739
1740 *dest = NULL;
1741 return convert_string_allocate(NULL, CH_UTF8, CH_UNIX, src, src_len,
1742 (void **)dest, converted_size, True);
1743}
1744
1745/**
1746 * Copy a string from a DOS src to a unix char * destination, allocating a buffer using talloc
1747 *
1748 * @param dest always set at least to NULL
1749 * @parm converted_size set to the number of bytes occupied by the string in
1750 * the destination on success.
1751 *
1752 * @return true if new buffer was correctly allocated, and string was
1753 * converted.
1754 **/
1755
1756bool pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src,
1757 size_t *converted_size)
1758{
1759 size_t src_len = strlen(src)+1;
1760
1761 *dest = NULL;
1762 return convert_string_talloc(ctx, CH_DOS, CH_UNIX, src, src_len,
1763 (void **)dest, converted_size, True);
1764}
1765
1766/**
1767 Copy a string from a char* src to a unicode or ascii
1768 dos codepage destination choosing unicode or ascii based on the
1769 flags in the SMB buffer starting at base_ptr.
1770 Return the number of bytes occupied by the string in the destination.
1771 flags can have:
1772 STR_TERMINATE means include the null termination.
1773 STR_UPPER means uppercase in the destination.
1774 STR_ASCII use ascii even with unicode packet.
1775 STR_NOALIGN means don't do alignment.
1776 dest_len is the maximum length allowed in the destination. If dest_len
1777 is -1 then no maxiumum is used.
1778**/
1779
1780size_t push_string_fn(const char *function, unsigned int line,
1781 const void *base_ptr, uint16 flags2,
1782 void *dest, const char *src,
1783 size_t dest_len, int flags)
1784{
1785#ifdef DEVELOPER
1786 /* We really need to zero fill here, not clobber
1787 * region, as we want to ensure that valgrind thinks
1788 * all of the outgoing buffer has been written to
1789 * so a send() or write() won't trap an error.
1790 * JRA.
1791 */
1792#if 0
1793 clobber_region(function, line, dest, dest_len);
1794#else
1795 memset(dest, '\0', dest_len);
1796#endif
1797#endif
1798
1799 if (!(flags & STR_ASCII) && \
1800 ((flags & STR_UNICODE || \
1801 (flags2 & FLAGS2_UNICODE_STRINGS)))) {
1802 return push_ucs2(base_ptr, dest, src, dest_len, flags);
1803 }
1804 return push_ascii(dest, src, dest_len, flags);
1805}
1806
1807
1808/**
1809 Copy a string from a unicode or ascii source (depending on
1810 the packet flags) to a char* destination.
1811 Flags can have:
1812 STR_TERMINATE means the string in src is null terminated.
1813 STR_UNICODE means to force as unicode.
1814 STR_ASCII use ascii even with unicode packet.
1815 STR_NOALIGN means don't do alignment.
1816 if STR_TERMINATE is set then src_len is ignored is it is -1
1817 src_len is the length of the source area in bytes.
1818 Return the number of bytes occupied by the string in src.
1819 The resulting string in "dest" is always null terminated.
1820**/
1821
1822size_t pull_string_fn(const char *function,
1823 unsigned int line,
1824 const void *base_ptr,
1825 uint16 smb_flags2,
1826 char *dest,
1827 const void *src,
1828 size_t dest_len,
1829 size_t src_len,
1830 int flags)
1831{
1832#ifdef DEVELOPER
1833 clobber_region(function, line, dest, dest_len);
1834#endif
1835
1836 if ((base_ptr == NULL) && ((flags & (STR_ASCII|STR_UNICODE)) == 0)) {
1837 smb_panic("No base ptr to get flg2 and neither ASCII nor "
1838 "UNICODE defined");
1839 }
1840
1841 if (!(flags & STR_ASCII) && \
1842 ((flags & STR_UNICODE || \
1843 (smb_flags2 & FLAGS2_UNICODE_STRINGS)))) {
1844 return pull_ucs2(base_ptr, dest, src, dest_len, src_len, flags);
1845 }
1846 return pull_ascii(dest, src, dest_len, src_len, flags);
1847}
1848
1849/**
1850 Copy a string from a unicode or ascii source (depending on
1851 the packet flags) to a char* destination.
1852 Variant that uses talloc.
1853 Flags can have:
1854 STR_TERMINATE means the string in src is null terminated.
1855 STR_UNICODE means to force as unicode.
1856 STR_ASCII use ascii even with unicode packet.
1857 STR_NOALIGN means don't do alignment.
1858 if STR_TERMINATE is set then src_len is ignored is it is -1
1859 src_len is the length of the source area in bytes.
1860 Return the number of bytes occupied by the string in src.
1861 The resulting string in "dest" is always null terminated.
1862**/
1863
1864size_t pull_string_talloc_fn(const char *function,
1865 unsigned int line,
1866 TALLOC_CTX *ctx,
1867 const void *base_ptr,
1868 uint16 smb_flags2,
1869 char **ppdest,
1870 const void *src,
1871 size_t src_len,
1872 int flags)
1873{
1874 if ((base_ptr == NULL) && ((flags & (STR_ASCII|STR_UNICODE)) == 0)) {
1875 smb_panic("No base ptr to get flg2 and neither ASCII nor "
1876 "UNICODE defined");
1877 }
1878
1879 if (!(flags & STR_ASCII) && \
1880 ((flags & STR_UNICODE || \
1881 (smb_flags2 & FLAGS2_UNICODE_STRINGS)))) {
1882 return pull_ucs2_base_talloc(ctx,
1883 base_ptr,
1884 ppdest,
1885 src,
1886 src_len,
1887 flags);
1888 }
1889 return pull_ascii_base_talloc(ctx,
1890 ppdest,
1891 src,
1892 src_len,
1893 flags);
1894}
1895
1896
1897size_t align_string(const void *base_ptr, const char *p, int flags)
1898{
1899 if (!(flags & STR_ASCII) && \
1900 ((flags & STR_UNICODE || \
1901 (SVAL(base_ptr, smb_flg2) & FLAGS2_UNICODE_STRINGS)))) {
1902 return ucs2_align(base_ptr, p, flags);
1903 }
1904 return 0;
1905}
1906
1907/*
1908 Return the unicode codepoint for the next multi-byte CH_UNIX character
1909 in the string. The unicode codepoint (codepoint_t) is an unsinged 32 bit value.
1910
1911 Also return the number of bytes consumed (which tells the caller
1912 how many bytes to skip to get to the next CH_UNIX character).
1913
1914 Return INVALID_CODEPOINT if the next character cannot be converted.
1915*/
1916
1917codepoint_t next_codepoint(const char *str, size_t *size)
1918{
1919 /* It cannot occupy more than 4 bytes in UTF16 format */
1920 uint8_t buf[4];
1921 smb_iconv_t descriptor;
1922#ifdef __OS2__
1923 size_t ilen_max;
1924 size_t olen_orig;
1925 const char *inbuf;
1926#endif
1927 size_t ilen_orig;
1928 size_t ilen;
1929 size_t olen;
1930
1931 char *outbuf;
1932
1933#ifdef __OS2__
1934 *size = 1;
1935#endif
1936
1937 if ((str[0] & 0x80) == 0) {
1938#ifndef __OS2__
1939 *size = 1;
1940#endif
1941 return (codepoint_t)str[0];
1942 }
1943
1944 lazy_initialize_conv();
1945
1946 descriptor = conv_handles[CH_UNIX][CH_UTF16LE];
1947 if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
1948#ifndef __OS2__
1949 *size = 1;
1950#endif
1951 return INVALID_CODEPOINT;
1952 }
1953#ifdef __OS2__
1954 /* We assume that no multi-byte character can take
1955 more than 5 bytes. This is OK as we only
1956 support codepoints up to 1M */
1957
1958 ilen_max = strnlen( str, 5 );
1959#else
1960 *size = 1;
1961#endif
1962 ilen_orig = 1;
1963 olen_orig = 2;
1964 while( 1 )
1965 {
1966 ilen = ilen_orig;
1967 olen = olen_orig;
1968 inbuf = str;
1969 outbuf = ( char * )buf;
1970 if( smb_iconv( descriptor, &inbuf, &ilen, &outbuf, &olen ) != ( size_t )-1 )
1971 break;
1972
1973 switch( errno )
1974 {
1975 case E2BIG :
1976 if( olen_orig == 2 )
1977 olen_orig = 4;
1978 else
1979 return INVALID_CODEPOINT;
1980 break;
1981
1982
1983 case EINVAL :
1984#ifndef __OS2__
1985 /* We assume that no multi-byte character can take
1986 more than 5 bytes. This is OK as we only
1987 support codepoints up to 1M */
1988 if( ilen_orig < 5 )
1989#else
1990 if( ilen_orig < ilen_max )
1991#endif
1992 ilen_orig++;
1993 else
1994 return INVALID_CODEPOINT;
1995 break;
1996
1997 case EILSEQ :
1998 default :
1999 return INVALID_CODEPOINT;
2000 }
2001 }
2002
2003 olen = olen_orig - olen;
2004
2005 *size = ilen_orig - ilen;
2006
2007 if (olen == 2) {
2008 /* 2 byte, UTF16-LE encoded value. */
2009 return (codepoint_t)SVAL(buf, 0);
2010 }
2011 if (olen == 4) {
2012 /* Decode a 4 byte UTF16-LE character manually.
2013 See RFC2871 for the encoding machanism.
2014 */
2015 codepoint_t w1 = SVAL(buf,0) & ~0xD800;
2016 codepoint_t w2 = SVAL(buf,2) & ~0xDC00;
2017
2018 return (codepoint_t)0x10000 +
2019 (w1 << 10) + w2;
2020 }
2021
2022 /* no other length is valid */
2023 return INVALID_CODEPOINT;
2024}
Note: See TracBrowser for help on using the repository browser.