/* [<][>][^][v][top][bottom][index][help] */
DEFINITIONS
This source file includes following definitions.
- charset_register_backend
- sys_iconv
- smb_iconv
- is_utf16
- smb_iconv_open_ex
- smb_iconv_open
- smb_iconv_close
- ascii_pull
- ascii_push
- latin1_push
- ucs2hex_pull
- ucs2hex_push
- iconv_swab
- iconv_copy
- utf8_pull
- utf8_push
- utf16_munged_pull
1 /*
2 Unix SMB/CIFS implementation.
3 minimal iconv implementation
4 Copyright (C) Andrew Tridgell 2001
5 Copyright (C) Jelmer Vernooij 2002
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 #include "../lib/util/dlinklist.h"
23 #include "system/iconv.h"
24 #include "system/filesys.h"
25 #undef strcasecmp
26
27
28 /**
29 * @file
30 *
31 * @brief Samba wrapper/stub for iconv character set conversion.
32 *
33 * iconv is the XPG2 interface for converting between character
34 * encodings. This file provides a Samba wrapper around it, and also
35 * a simple reimplementation that is used if the system does not
36 * implement iconv.
37 *
38 * Samba only works with encodings that are supersets of ASCII: ascii
39 * characters like whitespace can be tested for directly, multibyte
40 * sequences start with a byte with the high bit set, and strings are
41 * terminated by a nul byte.
42 *
43 * Note that the only function provided by iconv is conversion between
44 * characters. It doesn't directly support operations like
45 * uppercasing or comparison. We have to convert to UTF-16LE and
46 * compare there.
47 *
48 * @sa Samba Developers Guide
49 **/
50
51 static size_t ascii_pull (void *,const char **, size_t *, char **, size_t *);
52 static size_t ascii_push (void *,const char **, size_t *, char **, size_t *);
53 static size_t latin1_push (void *,const char **, size_t *, char **, size_t *);
54 static size_t utf8_pull (void *,const char **, size_t *, char **, size_t *);
55 static size_t utf8_push (void *,const char **, size_t *, char **, size_t *);
56 static size_t utf16_munged_pull(void *,const char **, size_t *, char **, size_t *);
57 static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
58 static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
59 static size_t iconv_copy (void *,const char **, size_t *, char **, size_t *);
60 static size_t iconv_swab (void *,const char **, size_t *, char **, size_t *);
61
62 static const struct charset_functions builtin_functions[] = {
63 /* windows is closest to UTF-16 */
64 {"UCS-2LE", iconv_copy, iconv_copy},
65 {"UTF-16LE", iconv_copy, iconv_copy},
66 {"UCS-2BE", iconv_swab, iconv_swab},
67 {"UTF-16BE", iconv_swab, iconv_swab},
68
69 /* we include the UTF-8 alias to cope with differing locale settings */
70 {"UTF8", utf8_pull, utf8_push},
71 {"UTF-8", utf8_pull, utf8_push},
72
73 /* this handles the munging needed for String2Key */
74 {"UTF16_MUNGED", utf16_munged_pull, iconv_copy},
75
76 {"ASCII", ascii_pull, ascii_push},
77 {"646", ascii_pull, ascii_push},
78 {"ISO-8859-1", ascii_pull, latin1_push},
79 {"UCS2-HEX", ucs2hex_pull, ucs2hex_push}
80 };
81
82 static struct charset_functions *charsets = NULL;
83
84 bool charset_register_backend(const void *_funcs)
/* [<][>][^][v][top][bottom][index][help] */
85 {
86 struct charset_functions *funcs = (struct charset_functions *)memdup(_funcs,sizeof(struct charset_functions));
87 struct charset_functions *c;
88
89 /* Check whether we already have this charset... */
90 for (c = charsets; c != NULL; c = c->next) {
91 if(!strcasecmp(c->name, funcs->name)) {
92 DEBUG(2, ("Duplicate charset %s, not registering\n", funcs->name));
93 return false;
94 }
95 }
96
97 funcs->next = funcs->prev = NULL;
98 DLIST_ADD(charsets, funcs);
99 return true;
100 }
101
102 #ifdef HAVE_NATIVE_ICONV
103 /* if there was an error then reset the internal state,
104 this ensures that we don't have a shift state remaining for
105 character sets like SJIS */
106 static size_t sys_iconv(void *cd,
/* [<][>][^][v][top][bottom][index][help] */
107 const char **inbuf, size_t *inbytesleft,
108 char **outbuf, size_t *outbytesleft)
109 {
110 size_t ret = iconv((iconv_t)cd,
111 discard_const_p(char *, inbuf), inbytesleft,
112 outbuf, outbytesleft);
113 if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
114 return ret;
115 }
116 #endif
117
118 /**
119 * This is a simple portable iconv() implementaion.
120 *
121 * It only knows about a very small number of character sets - just
122 * enough that Samba works on systems that don't have iconv.
123 **/
124 _PUBLIC_ size_t smb_iconv(smb_iconv_t cd,
/* [<][>][^][v][top][bottom][index][help] */
125 const char **inbuf, size_t *inbytesleft,
126 char **outbuf, size_t *outbytesleft)
127 {
128 char cvtbuf[2048];
129 size_t bufsize;
130
131 /* in many cases we can go direct */
132 if (cd->direct) {
133 return cd->direct(cd->cd_direct,
134 inbuf, inbytesleft, outbuf, outbytesleft);
135 }
136
137
138 /* otherwise we have to do it chunks at a time */
139 while (*inbytesleft > 0) {
140 char *bufp1 = cvtbuf;
141 const char *bufp2 = cvtbuf;
142
143 bufsize = sizeof(cvtbuf);
144
145 if (cd->pull(cd->cd_pull,
146 inbuf, inbytesleft, &bufp1, &bufsize) == -1
147 && errno != E2BIG) return -1;
148
149 bufsize = sizeof(cvtbuf) - bufsize;
150
151 if (cd->push(cd->cd_push,
152 &bufp2, &bufsize,
153 outbuf, outbytesleft) == -1) return -1;
154 }
155
156 return 0;
157 }
158
159 static bool is_utf16(const char *name)
/* [<][>][^][v][top][bottom][index][help] */
160 {
161 return strcasecmp(name, "UCS-2LE") == 0 ||
162 strcasecmp(name, "UTF-16LE") == 0;
163 }
164
165
166
167 _PUBLIC_ smb_iconv_t smb_iconv_open_ex(TALLOC_CTX *mem_ctx, const char *tocode,
/* [<][>][^][v][top][bottom][index][help] */
168 const char *fromcode, bool native_iconv)
169 {
170 smb_iconv_t ret;
171 const struct charset_functions *from=NULL, *to=NULL;
172 int i;
173
174 ret = (smb_iconv_t)talloc_named(mem_ctx,
175 sizeof(*ret),
176 "iconv(%s,%s)", tocode, fromcode);
177 if (!ret) {
178 errno = ENOMEM;
179 return (smb_iconv_t)-1;
180 }
181 memset(ret, 0, sizeof(*ret));
182
183 /* check for the simplest null conversion */
184 if (strcmp(fromcode, tocode) == 0) {
185 ret->direct = iconv_copy;
186 return ret;
187 }
188
189 for (i=0;i<ARRAY_SIZE(builtin_functions);i++) {
190 if (strcasecmp(fromcode, builtin_functions[i].name) == 0) {
191 from = &builtin_functions[i];
192 }
193 if (strcasecmp(tocode, builtin_functions[i].name) == 0) {
194 to = &builtin_functions[i];
195 }
196 }
197
198 if (from == NULL) {
199 for (from=charsets; from; from=from->next) {
200 if (strcasecmp(from->name, fromcode) == 0) break;
201 }
202 }
203
204 if (to == NULL) {
205 for (to=charsets; to; to=to->next) {
206 if (strcasecmp(to->name, tocode) == 0) break;
207 }
208 }
209
210 #ifdef HAVE_NATIVE_ICONV
211 if ((!from || !to) && !native_iconv) {
212 goto failed;
213 }
214 if (!from) {
215 ret->pull = sys_iconv;
216 ret->cd_pull = iconv_open("UTF-16LE", fromcode);
217 if (ret->cd_pull == (iconv_t)-1)
218 ret->cd_pull = iconv_open("UCS-2LE", fromcode);
219 if (ret->cd_pull == (iconv_t)-1) goto failed;
220 }
221
222 if (!to) {
223 ret->push = sys_iconv;
224 ret->cd_push = iconv_open(tocode, "UTF-16LE");
225 if (ret->cd_push == (iconv_t)-1)
226 ret->cd_push = iconv_open(tocode, "UCS-2LE");
227 if (ret->cd_push == (iconv_t)-1) goto failed;
228 }
229 #else
230 if (!from || !to) {
231 goto failed;
232 }
233 #endif
234
235 /* check for conversion to/from ucs2 */
236 if (is_utf16(fromcode) && to) {
237 ret->direct = to->push;
238 return ret;
239 }
240 if (is_utf16(tocode) && from) {
241 ret->direct = from->pull;
242 return ret;
243 }
244
245 #ifdef HAVE_NATIVE_ICONV
246 if (is_utf16(fromcode)) {
247 ret->direct = sys_iconv;
248 ret->cd_direct = ret->cd_push;
249 ret->cd_push = NULL;
250 return ret;
251 }
252 if (is_utf16(tocode)) {
253 ret->direct = sys_iconv;
254 ret->cd_direct = ret->cd_pull;
255 ret->cd_pull = NULL;
256 return ret;
257 }
258 #endif
259
260 /* the general case has to go via a buffer */
261 if (!ret->pull) ret->pull = from->pull;
262 if (!ret->push) ret->push = to->push;
263 return ret;
264
265 failed:
266 talloc_free(ret);
267 errno = EINVAL;
268 return (smb_iconv_t)-1;
269 }
270
271 /*
272 simple iconv_open() wrapper
273 */
274 _PUBLIC_ smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
/* [<][>][^][v][top][bottom][index][help] */
275 {
276 return smb_iconv_open_ex(NULL, tocode, fromcode, true);
277 }
278
279 /*
280 simple iconv_close() wrapper
281 */
282 _PUBLIC_ int smb_iconv_close(smb_iconv_t cd)
/* [<][>][^][v][top][bottom][index][help] */
283 {
284 #ifdef HAVE_NATIVE_ICONV
285 if (cd->cd_direct) iconv_close((iconv_t)cd->cd_direct);
286 if (cd->cd_pull) iconv_close((iconv_t)cd->cd_pull);
287 if (cd->cd_push) iconv_close((iconv_t)cd->cd_push);
288 #endif
289
290 talloc_free(cd);
291 return 0;
292 }
293
294
295 /**********************************************************************
296 the following functions implement the builtin character sets in Samba
297 and also the "test" character sets that are designed to test
298 multi-byte character set support for english users
299 ***********************************************************************/
300 static size_t ascii_pull(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
301 char **outbuf, size_t *outbytesleft)
302 {
303 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
304 (*outbuf)[0] = (*inbuf)[0];
305 (*outbuf)[1] = 0;
306 (*inbytesleft) -= 1;
307 (*outbytesleft) -= 2;
308 (*inbuf) += 1;
309 (*outbuf) += 2;
310 }
311
312 if (*inbytesleft > 0) {
313 errno = E2BIG;
314 return -1;
315 }
316
317 return 0;
318 }
319
320 static size_t ascii_push(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
321 char **outbuf, size_t *outbytesleft)
322 {
323 int ir_count=0;
324
325 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
326 (*outbuf)[0] = (*inbuf)[0] & 0x7F;
327 if ((*inbuf)[1]) ir_count++;
328 (*inbytesleft) -= 2;
329 (*outbytesleft) -= 1;
330 (*inbuf) += 2;
331 (*outbuf) += 1;
332 }
333
334 if (*inbytesleft == 1) {
335 errno = EINVAL;
336 return -1;
337 }
338
339 if (*inbytesleft > 1) {
340 errno = E2BIG;
341 return -1;
342 }
343
344 return ir_count;
345 }
346
347 static size_t latin1_push(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
348 char **outbuf, size_t *outbytesleft)
349 {
350 int ir_count=0;
351
352 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
353 (*outbuf)[0] = (*inbuf)[0];
354 if ((*inbuf)[1]) ir_count++;
355 (*inbytesleft) -= 2;
356 (*outbytesleft) -= 1;
357 (*inbuf) += 2;
358 (*outbuf) += 1;
359 }
360
361 if (*inbytesleft == 1) {
362 errno = EINVAL;
363 return -1;
364 }
365
366 if (*inbytesleft > 1) {
367 errno = E2BIG;
368 return -1;
369 }
370
371 return ir_count;
372 }
373
374 static size_t ucs2hex_pull(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
375 char **outbuf, size_t *outbytesleft)
376 {
377 while (*inbytesleft >= 1 && *outbytesleft >= 2) {
378 uint_t v;
379
380 if ((*inbuf)[0] != '@') {
381 /* seven bit ascii case */
382 (*outbuf)[0] = (*inbuf)[0];
383 (*outbuf)[1] = 0;
384 (*inbytesleft) -= 1;
385 (*outbytesleft) -= 2;
386 (*inbuf) += 1;
387 (*outbuf) += 2;
388 continue;
389 }
390 /* it's a hex character */
391 if (*inbytesleft < 5) {
392 errno = EINVAL;
393 return -1;
394 }
395
396 if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
397 errno = EILSEQ;
398 return -1;
399 }
400
401 (*outbuf)[0] = v&0xff;
402 (*outbuf)[1] = v>>8;
403 (*inbytesleft) -= 5;
404 (*outbytesleft) -= 2;
405 (*inbuf) += 5;
406 (*outbuf) += 2;
407 }
408
409 if (*inbytesleft > 0) {
410 errno = E2BIG;
411 return -1;
412 }
413
414 return 0;
415 }
416
417 static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
418 char **outbuf, size_t *outbytesleft)
419 {
420 while (*inbytesleft >= 2 && *outbytesleft >= 1) {
421 char buf[6];
422
423 if ((*inbuf)[1] == 0 &&
424 ((*inbuf)[0] & 0x80) == 0 &&
425 (*inbuf)[0] != '@') {
426 (*outbuf)[0] = (*inbuf)[0];
427 (*inbytesleft) -= 2;
428 (*outbytesleft) -= 1;
429 (*inbuf) += 2;
430 (*outbuf) += 1;
431 continue;
432 }
433 if (*outbytesleft < 5) {
434 errno = E2BIG;
435 return -1;
436 }
437 snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
438 memcpy(*outbuf, buf, 5);
439 (*inbytesleft) -= 2;
440 (*outbytesleft) -= 5;
441 (*inbuf) += 2;
442 (*outbuf) += 5;
443 }
444
445 if (*inbytesleft == 1) {
446 errno = EINVAL;
447 return -1;
448 }
449
450 if (*inbytesleft > 1) {
451 errno = E2BIG;
452 return -1;
453 }
454
455 return 0;
456 }
457
458 static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
459 char **outbuf, size_t *outbytesleft)
460 {
461 int n;
462
463 n = MIN(*inbytesleft, *outbytesleft);
464
465 swab(*inbuf, *outbuf, (n&~1));
466 if (n&1) {
467 (*outbuf)[n-1] = 0;
468 }
469
470 (*inbytesleft) -= n;
471 (*outbytesleft) -= n;
472 (*inbuf) += n;
473 (*outbuf) += n;
474
475 if (*inbytesleft > 0) {
476 errno = E2BIG;
477 return -1;
478 }
479
480 return 0;
481 }
482
483
484 static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
485 char **outbuf, size_t *outbytesleft)
486 {
487 int n;
488
489 n = MIN(*inbytesleft, *outbytesleft);
490
491 memmove(*outbuf, *inbuf, n);
492
493 (*inbytesleft) -= n;
494 (*outbytesleft) -= n;
495 (*inbuf) += n;
496 (*outbuf) += n;
497
498 if (*inbytesleft > 0) {
499 errno = E2BIG;
500 return -1;
501 }
502
503 return 0;
504 }
505
506 /*
507 this takes a UTF8 sequence and produces a UTF16 sequence
508 */
509 static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
510 char **outbuf, size_t *outbytesleft)
511 {
512 size_t in_left=*inbytesleft, out_left=*outbytesleft;
513 const uint8_t *c = (const uint8_t *)*inbuf;
514 uint8_t *uc = (uint8_t *)*outbuf;
515
516 while (in_left >= 1 && out_left >= 2) {
517 if ((c[0] & 0x80) == 0) {
518 uc[0] = c[0];
519 uc[1] = 0;
520 c += 1;
521 in_left -= 1;
522 out_left -= 2;
523 uc += 2;
524 continue;
525 }
526
527 if ((c[0] & 0xe0) == 0xc0) {
528 if (in_left < 2 ||
529 (c[1] & 0xc0) != 0x80) {
530 errno = EILSEQ;
531 goto error;
532 }
533 uc[1] = (c[0]>>2) & 0x7;
534 uc[0] = (c[0]<<6) | (c[1]&0x3f);
535 c += 2;
536 in_left -= 2;
537 out_left -= 2;
538 uc += 2;
539 continue;
540 }
541
542 if ((c[0] & 0xf0) == 0xe0) {
543 if (in_left < 3 ||
544 (c[1] & 0xc0) != 0x80 ||
545 (c[2] & 0xc0) != 0x80) {
546 errno = EILSEQ;
547 goto error;
548 }
549 uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
550 uc[0] = (c[1]<<6) | (c[2]&0x3f);
551 c += 3;
552 in_left -= 3;
553 out_left -= 2;
554 uc += 2;
555 continue;
556 }
557
558 if ((c[0] & 0xf8) == 0xf0) {
559 unsigned int codepoint;
560 if (in_left < 4 ||
561 (c[1] & 0xc0) != 0x80 ||
562 (c[2] & 0xc0) != 0x80 ||
563 (c[3] & 0xc0) != 0x80) {
564 errno = EILSEQ;
565 goto error;
566 }
567 codepoint =
568 (c[3]&0x3f) |
569 ((c[2]&0x3f)<<6) |
570 ((c[1]&0x3f)<<12) |
571 ((c[0]&0x7)<<18);
572 if (codepoint < 0x10000) {
573 /* accept UTF-8 characters that are not
574 minimally packed, but pack the result */
575 uc[0] = (codepoint & 0xFF);
576 uc[1] = (codepoint >> 8);
577 c += 4;
578 in_left -= 4;
579 out_left -= 2;
580 uc += 2;
581 continue;
582 }
583
584 codepoint -= 0x10000;
585
586 if (out_left < 4) {
587 errno = E2BIG;
588 goto error;
589 }
590
591 uc[0] = (codepoint>>10) & 0xFF;
592 uc[1] = (codepoint>>18) | 0xd8;
593 uc[2] = codepoint & 0xFF;
594 uc[3] = ((codepoint>>8) & 0x3) | 0xdc;
595 c += 4;
596 in_left -= 4;
597 out_left -= 4;
598 uc += 4;
599 continue;
600 }
601
602 /* we don't handle 5 byte sequences */
603 errno = EINVAL;
604 goto error;
605 }
606
607 if (in_left > 0) {
608 errno = E2BIG;
609 goto error;
610 }
611
612 *inbytesleft = in_left;
613 *outbytesleft = out_left;
614 *inbuf = (const char *)c;
615 *outbuf = (char *)uc;
616 return 0;
617
618 error:
619 *inbytesleft = in_left;
620 *outbytesleft = out_left;
621 *inbuf = (const char *)c;
622 *outbuf = (char *)uc;
623 return -1;
624 }
625
626
627 /*
628 this takes a UTF16 sequence and produces a UTF8 sequence
629 */
630 static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
631 char **outbuf, size_t *outbytesleft)
632 {
633 size_t in_left=*inbytesleft, out_left=*outbytesleft;
634 uint8_t *c = (uint8_t *)*outbuf;
635 const uint8_t *uc = (const uint8_t *)*inbuf;
636
637 while (in_left >= 2 && out_left >= 1) {
638 unsigned int codepoint;
639
640 if (uc[1] == 0 && !(uc[0] & 0x80)) {
641 /* simplest case */
642 c[0] = uc[0];
643 in_left -= 2;
644 out_left -= 1;
645 uc += 2;
646 c += 1;
647 continue;
648 }
649
650 if ((uc[1]&0xf8) == 0) {
651 /* next simplest case */
652 if (out_left < 2) {
653 errno = E2BIG;
654 goto error;
655 }
656 c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
657 c[1] = 0x80 | (uc[0] & 0x3f);
658 in_left -= 2;
659 out_left -= 2;
660 uc += 2;
661 c += 2;
662 continue;
663 }
664
665 if ((uc[1] & 0xfc) == 0xdc) {
666 /* its the second part of a 4 byte sequence. Illegal */
667 if (in_left < 4) {
668 errno = EINVAL;
669 } else {
670 errno = EILSEQ;
671 }
672 goto error;
673 }
674
675 if ((uc[1] & 0xfc) != 0xd8) {
676 codepoint = uc[0] | (uc[1]<<8);
677 if (out_left < 3) {
678 errno = E2BIG;
679 goto error;
680 }
681 c[0] = 0xe0 | (codepoint >> 12);
682 c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
683 c[2] = 0x80 | (codepoint & 0x3f);
684
685 in_left -= 2;
686 out_left -= 3;
687 uc += 2;
688 c += 3;
689 continue;
690 }
691
692 /* its the first part of a 4 byte sequence */
693 if (in_left < 4) {
694 errno = EINVAL;
695 goto error;
696 }
697 if ((uc[3] & 0xfc) != 0xdc) {
698 errno = EILSEQ;
699 goto error;
700 }
701 codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) |
702 (uc[0]<<10) | ((uc[1] & 0x3)<<18));
703
704 if (out_left < 4) {
705 errno = E2BIG;
706 goto error;
707 }
708 c[0] = 0xf0 | (codepoint >> 18);
709 c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
710 c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
711 c[3] = 0x80 | (codepoint & 0x3f);
712
713 in_left -= 4;
714 out_left -= 4;
715 uc += 4;
716 c += 4;
717 }
718
719 if (in_left == 1) {
720 errno = EINVAL;
721 goto error;
722 }
723
724 if (in_left > 1) {
725 errno = E2BIG;
726 goto error;
727 }
728
729 *inbytesleft = in_left;
730 *outbytesleft = out_left;
731 *inbuf = (const char *)uc;
732 *outbuf = (char *)c;
733
734 return 0;
735
736 error:
737 *inbytesleft = in_left;
738 *outbytesleft = out_left;
739 *inbuf = (const char *)uc;
740 *outbuf = (char *)c;
741 return -1;
742 }
743
744
745 /*
746 this takes a UTF16 munged sequence, modifies it according to the
747 string2key rules, and produces a UTF16 sequence
748
749 The rules are:
750
751 1) any 0x0000 characters are mapped to 0x0001
752
753 2) convert any instance of 0xD800 - 0xDBFF (high surrogate)
754 without an immediately following 0xDC00 - 0x0xDFFF (low surrogate) to
755 U+FFFD (OBJECT REPLACEMENT CHARACTER).
756
757 3) the same for any low surrogate that was not preceded by a high surrogate.
758
759 */
760 static size_t utf16_munged_pull(void *cd, const char **inbuf, size_t *inbytesleft,
/* [<][>][^][v][top][bottom][index][help] */
761 char **outbuf, size_t *outbytesleft)
762 {
763 size_t in_left=*inbytesleft, out_left=*outbytesleft;
764 uint8_t *c = (uint8_t *)*outbuf;
765 const uint8_t *uc = (const uint8_t *)*inbuf;
766
767 while (in_left >= 2 && out_left >= 2) {
768 unsigned int codepoint = uc[0] | (uc[1]<<8);
769
770 if (codepoint == 0) {
771 codepoint = 1;
772 }
773
774 if ((codepoint & 0xfc00) == 0xd800) {
775 /* a high surrogate */
776 unsigned int codepoint2;
777 if (in_left < 4) {
778 codepoint = 0xfffd;
779 goto codepoint16;
780 }
781 codepoint2 = uc[2] | (uc[3]<<8);
782 if ((codepoint2 & 0xfc00) != 0xdc00) {
783 /* high surrogate not followed by low
784 surrogate: convert to 0xfffd */
785 codepoint = 0xfffd;
786 goto codepoint16;
787 }
788 if (out_left < 4) {
789 errno = E2BIG;
790 goto error;
791 }
792 memcpy(c, uc, 4);
793 in_left -= 4;
794 out_left -= 4;
795 uc += 4;
796 c += 4;
797 continue;
798 }
799
800 if ((codepoint & 0xfc00) == 0xdc00) {
801 /* low surrogate not preceded by high
802 surrogate: convert to 0xfffd */
803 codepoint = 0xfffd;
804 }
805
806 codepoint16:
807 c[0] = codepoint & 0xFF;
808 c[1] = (codepoint>>8) & 0xFF;
809
810 in_left -= 2;
811 out_left -= 2;
812 uc += 2;
813 c += 2;
814 continue;
815 }
816
817 if (in_left == 1) {
818 errno = EINVAL;
819 goto error;
820 }
821
822 if (in_left > 1) {
823 errno = E2BIG;
824 goto error;
825 }
826
827 *inbytesleft = in_left;
828 *outbytesleft = out_left;
829 *inbuf = (const char *)uc;
830 *outbuf = (char *)c;
831
832 return 0;
833
834 error:
835 *inbytesleft = in_left;
836 *outbytesleft = out_left;
837 *inbuf = (const char *)uc;
838 *outbuf = (char *)c;
839 return -1;
840 }
841
842
843