source: trunk/server/source3/lib/iconv.c@ 454

Last change on this file since 454 was 454, checked in by Silvan Scherrer, 15 years ago

Samba Server 3.5: merged changes from 3.3

File size: 18.4 KB
Line 
1/*
2 Unix SMB/CIFS implementation.
3 minimal iconv implementation
4 Copyright (C) Andrew Tridgell 2001
5 Copyright (C) Jelmer Vernooij 2002,2003
6
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or
10 (at your option) any later version.
11
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>.
19*/
20
21#include "includes.h"
22
23/*
24 * We have to use strcasecmp here as the character conversions
25 * haven't been initialised yet. JRA.
26 */
27
28#undef strcasecmp
29
30/**
31 * @file
32 *
33 * @brief Samba wrapper/stub for iconv character set conversion.
34 *
35 * iconv is the XPG2 interface for converting between character
36 * encodings. This file provides a Samba wrapper around it, and also
37 * a simple reimplementation that is used if the system does not
38 * implement iconv.
39 *
40 * Samba only works with encodings that are supersets of ASCII: ascii
41 * characters like whitespace can be tested for directly, multibyte
42 * sequences start with a byte with the high bit set, and strings are
43 * terminated by a nul byte.
44 *
45 * Note that the only function provided by iconv is conversion between
46 * characters. It doesn't directly support operations like
47 * uppercasing or comparison. We have to convert to UCS-2 and compare
48 * there.
49 *
50 * @sa Samba Developers Guide
51 **/
52
53static_decl_charset;
54
55static size_t ascii_pull(void *,const char **, size_t *, char **, size_t *);
56static size_t ascii_push(void *,const char **, size_t *, char **, size_t *);
57static size_t latin1_push(void *,const char **, size_t *, char **, size_t *);
58static size_t utf8_pull(void *,const char **, size_t *, char **, size_t *);
59static size_t utf8_push(void *,const char **, size_t *, char **, size_t *);
60static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
61static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
62static size_t iconv_copy(void *,const char **, size_t *, char **, size_t *);
63static size_t iconv_swab (void *,const char **, size_t *, char **, size_t *);
64
65static struct charset_functions builtin_functions[] = {
66 /* windows is really neither UCS-2 not UTF-16 */
67 {"UCS-2LE", iconv_copy, iconv_copy},
68 {"UTF-16LE", iconv_copy, iconv_copy},
69 {"UCS-2BE", iconv_swab, iconv_swab},
70 {"UTF-16BE", iconv_swab, iconv_swab},
71
72 /* we include the UTF-8 alias to cope with differing locale settings */
73 {"UTF8", utf8_pull, utf8_push},
74 {"UTF-8", utf8_pull, utf8_push},
75 {"ASCII", ascii_pull, ascii_push},
76 {"646", ascii_pull, ascii_push},
77 {"ISO-8859-1", ascii_pull, latin1_push},
78 {"UCS2-HEX", ucs2hex_pull, ucs2hex_push},
79 {NULL, NULL, NULL}
80};
81
82static struct charset_functions *charsets = NULL;
83
84static struct charset_functions *find_charset_functions(const char *name)
85{
86 struct charset_functions *c = charsets;
87
88 while(c) {
89 if (strcasecmp(name, c->name) == 0) {
90 return c;
91 }
92 c = c->next;
93 }
94
95 return NULL;
96}
97
98NTSTATUS smb_register_charset(struct charset_functions *funcs)
99{
100 if (!funcs) {
101 return NT_STATUS_INVALID_PARAMETER;
102 }
103
104 DEBUG(5, ("Attempting to register new charset %s\n", funcs->name));
105 /* Check whether we already have this charset... */
106 if (find_charset_functions(funcs->name)) {
107 DEBUG(0, ("Duplicate charset %s, not registering\n", funcs->name));
108 return NT_STATUS_OBJECT_NAME_COLLISION;
109 }
110
111 funcs->next = funcs->prev = NULL;
112 DEBUG(5, ("Registered charset %s\n", funcs->name));
113 DLIST_ADD(charsets, funcs);
114 return NT_STATUS_OK;
115}
116
117static void lazy_initialize_iconv(void)
118{
119 static bool initialized;
120 int i;
121
122 if (!initialized) {
123 initialized = True;
124 for(i = 0; builtin_functions[i].name; i++)
125 smb_register_charset(&builtin_functions[i]);
126 static_init_charset;
127 }
128}
129
130#ifdef __OS2__
131// i could have done a static variable w/o this function. but i feel it's nicer this way.
132// the purpose of this function is to save the to_name to get the korean and japanese code set working
133char * save_toname(char *toname, bool what)
134{
135 static char *to_name=NULL;
136
137 if ( what == 0 )
138 to_name = SMB_STRDUP(toname);
139
140 return to_name;
141}
142#endif
143
144#ifdef HAVE_NATIVE_ICONV
145/* if there was an error then reset the internal state,
146 this ensures that we don't have a shift state remaining for
147 character sets like SJIS */
148static size_t sys_iconv(void *cd,
149 const char **inbuf, size_t *inbytesleft,
150 char **outbuf, size_t *outbytesleft)
151{
152#ifdef __OS2__
153 uint16 *outbuf_uc = ( uint16 * )*outbuf;
154 char *to_name = save_toname(NULL, 1);
155#endif
156
157 size_t ret = iconv((iconv_t)cd,
158 (void *)inbuf, inbytesleft,
159 outbuf, outbytesleft);
160 if (ret == (size_t)-1) {
161 int saved_errno = errno;
162 iconv(cd, NULL, NULL, NULL, NULL);
163 errno = saved_errno;
164 }
165#ifdef __OS2__
166 /* Workaround for path separator on OS/2 */
167 else
168 {
169 if( (strstr(to_name, "949") != NULL) || /* Korean CP */
170 (strstr(to_name, "932") != NULL) || /* Japanese CP */
171 (strstr(to_name, "942") != NULL) || /* Japanese CP */
172 (strstr(to_name, "943") != NULL) ) /* Japanese CP */
173 {
174 while(( char * )outbuf_uc < *outbuf )
175 {
176 if( *outbuf_uc == 0x20a9 || /* Korean WON */
177 *outbuf_uc == 0x00a5 ) /* Japanese YEN */
178 *outbuf_uc = '\\';
179
180 outbuf_uc++;
181 }
182 }
183 }
184#endif
185
186 return ret;
187}
188#endif
189
190/**
191 * This is a simple portable iconv() implementaion.
192 *
193 * It only knows about a very small number of character sets - just
194 * enough that Samba works on systems that don't have iconv.
195 **/
196size_t smb_iconv(smb_iconv_t cd,
197 const char **inbuf, size_t *inbytesleft,
198 char **outbuf, size_t *outbytesleft)
199{
200 char cvtbuf[2048];
201 char *bufp = cvtbuf;
202 size_t bufsize;
203#ifdef __OS2__
204 save_toname(cd->to_name, 0);
205#endif
206
207 /* in many cases we can go direct */
208 if (cd->direct) {
209 return cd->direct(cd->cd_direct,
210 inbuf, inbytesleft, outbuf, outbytesleft);
211 }
212
213
214 /* otherwise we have to do it chunks at a time */
215 while (*inbytesleft > 0) {
216 bufp = cvtbuf;
217 bufsize = sizeof(cvtbuf);
218
219 if (cd->pull(cd->cd_pull,
220 inbuf, inbytesleft, &bufp, &bufsize) == -1
221 && errno != E2BIG) return -1;
222
223 bufp = cvtbuf;
224 bufsize = sizeof(cvtbuf) - bufsize;
225
226 if (cd->push(cd->cd_push,
227 (const char **)&bufp, &bufsize,
228 outbuf, outbytesleft) == -1) return -1;
229 }
230
231 return 0;
232}
233
234
235static bool is_utf16(const char *name)
236{
237 return strcasecmp(name, "UCS-2LE") == 0 ||
238 strcasecmp(name, "UTF-16LE") == 0;
239}
240
241/*
242 simple iconv_open() wrapper
243 */
244smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
245{
246 smb_iconv_t ret;
247 struct charset_functions *from, *to;
248
249 lazy_initialize_iconv();
250 from = charsets;
251 to = charsets;
252
253 ret = SMB_MALLOC_P(struct smb_iconv_s);
254 if (!ret) {
255 errno = ENOMEM;
256 return (smb_iconv_t)-1;
257 }
258 memset(ret, 0, sizeof(struct smb_iconv_s));
259
260 ret->from_name = SMB_STRDUP(fromcode);
261 ret->to_name = SMB_STRDUP(tocode);
262
263 /* check for the simplest null conversion */
264 if (strcasecmp(fromcode, tocode) == 0) {
265 ret->direct = iconv_copy;
266 return ret;
267 }
268
269 /* check if we have a builtin function for this conversion */
270 from = find_charset_functions(fromcode);
271 if(from)ret->pull = from->pull;
272
273 to = find_charset_functions(tocode);
274 if(to)ret->push = to->push;
275
276 /* check if we can use iconv for this conversion */
277#ifdef HAVE_NATIVE_ICONV
278 if (!ret->pull) {
279 ret->cd_pull = iconv_open("UTF-16LE", fromcode);
280 if (ret->cd_pull == (iconv_t)-1)
281 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
282 if (ret->cd_pull != (iconv_t)-1)
283 ret->pull = sys_iconv;
284 }
285
286 if (!ret->push) {
287 ret->cd_push = iconv_open(tocode, "UTF-16LE");
288 if (ret->cd_push == (iconv_t)-1)
289 ret->cd_push = iconv_open(tocode, "UCS-2LE");
290 if (ret->cd_push != (iconv_t)-1)
291 ret->push = sys_iconv;
292 }
293#endif
294
295 /* check if there is a module available that can do this conversion */
296 if (!ret->pull && NT_STATUS_IS_OK(smb_probe_module("charset", fromcode))) {
297 if(!(from = find_charset_functions(fromcode)))
298 DEBUG(0, ("Module %s doesn't provide charset %s!\n", fromcode, fromcode));
299 else
300 ret->pull = from->pull;
301 }
302
303 if (!ret->push && NT_STATUS_IS_OK(smb_probe_module("charset", tocode))) {
304 if(!(to = find_charset_functions(tocode)))
305 DEBUG(0, ("Module %s doesn't provide charset %s!\n", tocode, tocode));
306 else
307 ret->push = to->push;
308 }
309
310 if (!ret->push || !ret->pull) {
311 SAFE_FREE(ret->from_name);
312 SAFE_FREE(ret->to_name);
313 SAFE_FREE(ret);
314 errno = EINVAL;
315 return (smb_iconv_t)-1;
316 }
317
318 /* check for conversion to/from ucs2 */
319 if (is_utf16(fromcode) && to) {
320 ret->direct = to->push;
321 ret->push = ret->pull = NULL;
322 return ret;
323 }
324
325 if (is_utf16(tocode) && from) {
326 ret->direct = from->pull;
327 ret->push = ret->pull = NULL;
328 return ret;
329 }
330
331 /* Check if we can do the conversion direct */
332#ifdef HAVE_NATIVE_ICONV
333 if (is_utf16(fromcode)) {
334 ret->direct = sys_iconv;
335 ret->cd_direct = ret->cd_push;
336 ret->cd_push = NULL;
337 return ret;
338 }
339 if (is_utf16(tocode)) {
340 ret->direct = sys_iconv;
341 ret->cd_direct = ret->cd_pull;
342 ret->cd_pull = NULL;
343 return ret;
344 }
345#endif
346
347 return ret;
348}
349
350/*
351 simple iconv_close() wrapper
352*/
353int smb_iconv_close (smb_iconv_t cd)
354{
355#ifdef HAVE_NATIVE_ICONV
356 if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
357 if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
358 if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
359#endif
360
361 SAFE_FREE(cd->from_name);
362 SAFE_FREE(cd->to_name);
363
364 memset(cd, 0, sizeof(*cd));
365 SAFE_FREE(cd);
366 return 0;
367}
368
369
370/**********************************************************************
371 the following functions implement the builtin character sets in Samba
372 and also the "test" character sets that are designed to test
373 multi-byte character set support for english users
374***********************************************************************/
375
376static size_t ascii_pull(void *cd, const char **inbuf, size_t *inbytesleft,
377 char **outbuf, size_t *outbytesleft)
378{
379 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
380 (*outbuf)[0] = (*inbuf)[0];
381 (*outbuf)[1] = 0;
382 (*inbytesleft) -= 1;
383 (*outbytesleft) -= 2;
384 (*inbuf) += 1;
385 (*outbuf) += 2;
386 }
387
388 if (*inbytesleft > 0) {
389 errno = E2BIG;
390 return -1;
391 }
392
393 return 0;
394}
395
396static size_t ascii_push(void *cd, const char **inbuf, size_t *inbytesleft,
397 char **outbuf, size_t *outbytesleft)
398{
399 int ir_count=0;
400
401 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
402 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
403 if ((*inbuf)[1]) ir_count++;
404 (*inbytesleft) -= 2;
405 (*outbytesleft) -= 1;
406 (*inbuf) += 2;
407 (*outbuf) += 1;
408 }
409
410 if (*inbytesleft == 1) {
411 errno = EINVAL;
412 return -1;
413 }
414
415 if (*inbytesleft > 1) {
416 errno = E2BIG;
417 return -1;
418 }
419
420 return ir_count;
421}
422
423static size_t latin1_push(void *cd, const char **inbuf, size_t *inbytesleft,
424 char **outbuf, size_t *outbytesleft)
425{
426 int ir_count=0;
427
428 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
429 (*outbuf)[0] = (*inbuf)[0];
430 if ((*inbuf)[1]) ir_count++;
431 (*inbytesleft) -= 2;
432 (*outbytesleft) -= 1;
433 (*inbuf) += 2;
434 (*outbuf) += 1;
435 }
436
437 if (*inbytesleft == 1) {
438 errno = EINVAL;
439 return -1;
440 }
441
442 if (*inbytesleft > 1) {
443 errno = E2BIG;
444 return -1;
445 }
446
447 return ir_count;
448}
449
450static size_t ucs2hex_pull(void *cd, const char **inbuf, size_t *inbytesleft,
451 char **outbuf, size_t *outbytesleft)
452{
453 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
454 unsigned v;
455
456 if ((*inbuf)[0] != '@') {
457 /* seven bit ascii case */
458 (*outbuf)[0] = (*inbuf)[0];
459 (*outbuf)[1] = 0;
460 (*inbytesleft) -= 1;
461 (*outbytesleft) -= 2;
462 (*inbuf) += 1;
463 (*outbuf) += 2;
464 continue;
465 }
466 /* it's a hex character */
467 if (*inbytesleft < 5) {
468 errno = EINVAL;
469 return -1;
470 }
471
472 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
473 errno = EILSEQ;
474 return -1;
475 }
476
477 (*outbuf)[0] = v&0xff;
478 (*outbuf)[1] = v>>8;
479 (*inbytesleft) -= 5;
480 (*outbytesleft) -= 2;
481 (*inbuf) += 5;
482 (*outbuf) += 2;
483 }
484
485 if (*inbytesleft > 0) {
486 errno = E2BIG;
487 return -1;
488 }
489
490 return 0;
491}
492
493static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
494 char **outbuf, size_t *outbytesleft)
495{
496 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
497 char buf[6];
498
499 if ((*inbuf)[1] == 0 &&
500 ((*inbuf)[0] & 0x80) == 0 &&
501 (*inbuf)[0] != '@') {
502 (*outbuf)[0] = (*inbuf)[0];
503 (*inbytesleft) -= 2;
504 (*outbytesleft) -= 1;
505 (*inbuf) += 2;
506 (*outbuf) += 1;
507 continue;
508 }
509 if (*outbytesleft < 5) {
510 errno = E2BIG;
511 return -1;
512 }
513 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
514 memcpy(*outbuf, buf, 5);
515 (*inbytesleft) -= 2;
516 (*outbytesleft) -= 5;
517 (*inbuf) += 2;
518 (*outbuf) += 5;
519 }
520
521 if (*inbytesleft == 1) {
522 errno = EINVAL;
523 return -1;
524 }
525
526 if (*inbytesleft > 1) {
527 errno = E2BIG;
528 return -1;
529 }
530
531 return 0;
532}
533
534static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
535 char **outbuf, size_t *outbytesleft)
536{
537 int n;
538
539 n = MIN(*inbytesleft, *outbytesleft);
540
541 swab(*inbuf, *outbuf, (n&~1));
542 if (n&1) {
543 (*outbuf)[n-1] = 0;
544 }
545
546 (*inbytesleft) -= n;
547 (*outbytesleft) -= n;
548 (*inbuf) += n;
549 (*outbuf) += n;
550
551 if (*inbytesleft > 0) {
552 errno = E2BIG;
553 return -1;
554 }
555
556 return 0;
557}
558
559static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
560 char **outbuf, size_t *outbytesleft)
561{
562 int n;
563
564 n = MIN(*inbytesleft, *outbytesleft);
565
566 memmove(*outbuf, *inbuf, n);
567
568 (*inbytesleft) -= n;
569 (*outbytesleft) -= n;
570 (*inbuf) += n;
571 (*outbuf) += n;
572
573 if (*inbytesleft > 0) {
574 errno = E2BIG;
575 return -1;
576 }
577
578 return 0;
579}
580
581static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
582 char **outbuf, size_t *outbytesleft)
583{
584 size_t in_left=*inbytesleft, out_left=*outbytesleft;
585 const uint8 *c = (const uint8 *)*inbuf;
586 uint8 *uc = (uint8 *)*outbuf;
587
588 while (in_left >= 1 && out_left >= 2) {
589 unsigned int codepoint;
590
591 if ((c[0] & 0x80) == 0) {
592 uc[0] = c[0];
593 uc[1] = 0;
594 c += 1;
595 in_left -= 1;
596 out_left -= 2;
597 uc += 2;
598 continue;
599 }
600
601 if ((c[0] & 0xe0) == 0xc0) {
602 if (in_left < 2 ||
603 (c[1] & 0xc0) != 0x80) {
604 errno = EILSEQ;
605 goto error;
606 }
607 codepoint = (c[1]&0x3f) | ((c[0]&0x1f)<<6);
608 if (codepoint < 0x80) {
609 /* don't accept UTF-8 characters that are not minimally packed */
610 errno = EILSEQ;
611 goto error;
612 }
613 uc[1] = codepoint >> 8;
614 uc[0] = codepoint & 0xff;
615 c += 2;
616 in_left -= 2;
617 out_left -= 2;
618 uc += 2;
619 continue;
620 }
621
622 if ((c[0] & 0xf0) == 0xe0) {
623 if (in_left < 3 ||
624 (c[1] & 0xc0) != 0x80 ||
625 (c[2] & 0xc0) != 0x80) {
626 errno = EILSEQ;
627 goto error;
628 }
629 codepoint = (c[2]&0x3f) | ((c[1]&0x3f)<<6) | ((c[0]&0xf)<<12);
630 if (codepoint < 0x800) {
631 /* don't accept UTF-8 characters that are not minimally packed */
632 errno = EILSEQ;
633 goto error;
634 }
635 uc[1] = codepoint >> 8;
636 uc[0] = codepoint & 0xff;
637 c += 3;
638 in_left -= 3;
639 out_left -= 2;
640 uc += 2;
641 continue;
642 }
643
644 if ((c[0] & 0xf8) == 0xf0) {
645 if (in_left < 4 ||
646 (c[1] & 0xc0) != 0x80 ||
647 (c[2] & 0xc0) != 0x80 ||
648 (c[3] & 0xc0) != 0x80) {
649 errno = EILSEQ;
650 goto error;
651 }
652 codepoint =
653 (c[3]&0x3f) |
654 ((c[2]&0x3f)<<6) |
655 ((c[1]&0x3f)<<12) |
656 ((c[0]&0x7)<<18);
657 if (codepoint < 0x10000 || codepoint > 0x10ffff) {
658 /* don't accept UTF-8 characters that are not minimally packed */
659 errno = EILSEQ;
660 goto error;
661 }
662
663 codepoint -= 0x10000;
664
665 if (out_left < 4) {
666 errno = E2BIG;
667 goto error;
668 }
669
670 uc[0] = (codepoint>>10) & 0xFF;
671 uc[1] = (codepoint>>18) | 0xd8;
672 uc[2] = codepoint & 0xFF;
673 uc[3] = ((codepoint>>8) & 0x3) | 0xdc;
674 c += 4;
675 in_left -= 4;
676 out_left -= 4;
677 uc += 4;
678 continue;
679 }
680
681 /* we don't handle 5 byte sequences */
682 errno = EINVAL;
683 goto error;
684 }
685
686 if (in_left > 0) {
687 errno = E2BIG;
688 goto error;
689 }
690
691 *inbytesleft = in_left;
692 *outbytesleft = out_left;
693 *inbuf = (char *)c;
694 *outbuf = (char *)uc;
695 return 0;
696
697error:
698 *inbytesleft = in_left;
699 *outbytesleft = out_left;
700 *inbuf = (char *)c;
701 *outbuf = (char *)uc;
702 return -1;
703}
704
705static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
706 char **outbuf, size_t *outbytesleft)
707{
708 size_t in_left=*inbytesleft, out_left=*outbytesleft;
709 uint8 *c = (uint8 *)*outbuf;
710 const uint8 *uc = (const uint8 *)*inbuf;
711
712 while (in_left >= 2 && out_left >= 1) {
713 unsigned int codepoint;
714
715 if (uc[1] == 0 && !(uc[0] & 0x80)) {
716 /* simplest case */
717 c[0] = uc[0];
718 in_left -= 2;
719 out_left -= 1;
720 uc += 2;
721 c += 1;
722 continue;
723 }
724
725 if ((uc[1]&0xf8) == 0) {
726 /* next simplest case */
727 if (out_left < 2) {
728 errno = E2BIG;
729 goto error;
730 }
731 c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
732 c[1] = 0x80 | (uc[0] & 0x3f);
733 in_left -= 2;
734 out_left -= 2;
735 uc += 2;
736 c += 2;
737 continue;
738 }
739
740 if ((uc[1] & 0xfc) == 0xdc) {
741 /* its the second part of a 4 byte sequence. Illegal */
742 if (in_left < 4) {
743 errno = EINVAL;
744 } else {
745 errno = EILSEQ;
746 }
747 goto error;
748 }
749
750 if ((uc[1] & 0xfc) != 0xd8) {
751 codepoint = uc[0] | (uc[1]<<8);
752 if (out_left < 3) {
753 errno = E2BIG;
754 goto error;
755 }
756 c[0] = 0xe0 | (codepoint >> 12);
757 c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
758 c[2] = 0x80 | (codepoint & 0x3f);
759
760 in_left -= 2;
761 out_left -= 3;
762 uc += 2;
763 c += 3;
764 continue;
765 }
766
767 /* its the first part of a 4 byte sequence */
768 if (in_left < 4) {
769 errno = EINVAL;
770 goto error;
771 }
772 if ((uc[3] & 0xfc) != 0xdc) {
773 errno = EILSEQ;
774 goto error;
775 }
776 codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) |
777 (uc[0]<<10) | ((uc[1] & 0x3)<<18));
778
779 if (out_left < 4) {
780 errno = E2BIG;
781 goto error;
782 }
783 c[0] = 0xf0 | (codepoint >> 18);
784 c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
785 c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
786 c[3] = 0x80 | (codepoint & 0x3f);
787
788 in_left -= 4;
789 out_left -= 4;
790 uc += 4;
791 c += 4;
792 }
793
794 if (in_left == 1) {
795 errno = EINVAL;
796 goto error;
797 }
798
799 if (in_left > 1) {
800 errno = E2BIG;
801 goto error;
802 }
803
804 *inbytesleft = in_left;
805 *outbytesleft = out_left;
806 *inbuf = (char *)uc;
807 *outbuf = (char *)c;
808
809 return 0;
810
811error:
812 *inbytesleft = in_left;
813 *outbytesleft = out_left;
814 *inbuf = (char *)uc;
815 *outbuf = (char *)c;
816 return -1;
817}
818
Note: See TracBrowser for help on using the repository browser.