Unix SMB/CIFS implementation.
minimal iconv implementation
Copyright (C) Andrew Tridgell 2001
+ Copyright (C) Jelmer Vernooij 2002,2003
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
#include "includes.h"
-static size_t ascii_pull(void *,char **, size_t *, char **, size_t *);
-static size_t ascii_push(void *,char **, size_t *, char **, size_t *);
-static size_t utf8_pull(void *,char **, size_t *, char **, size_t *);
-static size_t utf8_push(void *,char **, size_t *, char **, size_t *);
-static size_t weird_pull(void *,char **, size_t *, char **, size_t *);
-static size_t weird_push(void *,char **, size_t *, char **, size_t *);
-static size_t ucs2hex_pull(void *,char **, size_t *, char **, size_t *);
-static size_t ucs2hex_push(void *,char **, size_t *, char **, size_t *);
-static size_t iconv_copy(void *,char **, size_t *, char **, size_t *);
-
/*
- for each charset we have a function that pulls from that charset to
- a ucs2 buffer, and a function that pushes to a ucs2 buffer
-*/
-static struct {
- char *name;
- size_t (*pull)(void *, char **inbuf, size_t *inbytesleft,
- char **outbuf, size_t *outbytesleft);
- size_t (*push)(void *, char **inbuf, size_t *inbytesleft,
- char **outbuf, size_t *outbytesleft);
-} charsets[] = {
+ * We have to use strcasecmp here as the character conversions
+ * haven't been initialised yet. JRA.
+ */
+
+#undef strcasecmp
+
+/**
+ * @file
+ *
+ * @brief Samba wrapper/stub for iconv character set conversion.
+ *
+ * iconv is the XPG2 interface for converting between character
+ * encodings. This file provides a Samba wrapper around it, and also
+ * a simple reimplementation that is used if the system does not
+ * implement iconv.
+ *
+ * Samba only works with encodings that are supersets of ASCII: ascii
+ * characters like whitespace can be tested for directly, multibyte
+ * sequences start with a byte with the high bit set, and strings are
+ * terminated by a nul byte.
+ *
+ * Note that the only function provided by iconv is conversion between
+ * characters. It doesn't directly support operations like
+ * uppercasing or comparison. We have to convert to UCS-2 and compare
+ * there.
+ *
+ * @sa Samba Developers Guide
+ **/
+
+static size_t ascii_pull(void *,const char **, size_t *, char **, size_t *);
+static size_t ascii_push(void *,const char **, size_t *, char **, size_t *);
+static size_t latin1_push(void *,const char **, size_t *, char **, size_t *);
+static size_t utf8_pull(void *,const char **, size_t *, char **, size_t *);
+static size_t utf8_push(void *,const char **, size_t *, char **, size_t *);
+static size_t ucs2hex_pull(void *,const char **, size_t *, char **, size_t *);
+static size_t ucs2hex_push(void *,const char **, size_t *, char **, size_t *);
+static size_t iconv_copy(void *,const char **, size_t *, char **, size_t *);
+static size_t iconv_swab (void *,const char **, size_t *, char **, size_t *);
+
+static struct charset_functions builtin_functions[] = {
+ /* windows is really neither UCS-2 not UTF-16 */
{"UCS-2LE", iconv_copy, iconv_copy},
+ {"UTF-16LE", iconv_copy, iconv_copy},
+ {"UCS-2BE", iconv_swab, iconv_swab},
+ {"UTF-16BE", iconv_swab, iconv_swab},
+
+ /* we include the UTF-8 alias to cope with differing locale settings */
{"UTF8", utf8_pull, utf8_push},
+ {"UTF-8", utf8_pull, utf8_push},
{"ASCII", ascii_pull, ascii_push},
- {"WEIRD", weird_pull, weird_push},
+ {"646", ascii_pull, ascii_push},
+ {"ISO-8859-1", ascii_pull, latin1_push},
{"UCS2-HEX", ucs2hex_pull, ucs2hex_push},
{NULL, NULL, NULL}
};
+static struct charset_functions *charsets = NULL;
+
+static struct charset_functions *find_charset_functions(const char *name)
+{
+ struct charset_functions *c = charsets;
+
+ while(c) {
+ if (strcasecmp(name, c->name) == 0) {
+ return c;
+ }
+ c = c->next;
+ }
+
+ return NULL;
+}
+
+NTSTATUS smb_register_charset(struct charset_functions *funcs)
+{
+ if (!funcs) {
+ return NT_STATUS_INVALID_PARAMETER;
+ }
+
+ DEBUG(5, ("Attempting to register new charset %s\n", funcs->name));
+ /* Check whether we already have this charset... */
+ if (find_charset_functions(funcs->name)) {
+ DEBUG(0, ("Duplicate charset %s, not registering\n", funcs->name));
+ return NT_STATUS_OBJECT_NAME_COLLISION;
+ }
+
+ funcs->next = funcs->prev = NULL;
+ DEBUG(5, ("Registered charset %s\n", funcs->name));
+ DLIST_ADD(charsets, funcs);
+ return NT_STATUS_OK;
+}
+
+static void lazy_initialize_iconv(void)
+{
+ static BOOL initialized;
+ int i;
+
+ if (!initialized) {
+ initialized = True;
+ for(i = 0; builtin_functions[i].name; i++)
+ smb_register_charset(&builtin_functions[i]);
+ static_init_charset;
+ }
+}
/* if there was an error then reset the internal state,
this ensures that we don't have a shift state remaining for
character sets like SJIS */
static size_t sys_iconv(void *cd,
- char **inbuf, size_t *inbytesleft,
+ const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
#ifdef HAVE_NATIVE_ICONV
size_t ret = iconv((iconv_t)cd,
- inbuf, inbytesleft,
+ (char **)inbuf, inbytesleft,
outbuf, outbytesleft);
- if (ret == (size_t)-1) iconv(cd, NULL, NULL, NULL, NULL);
+ if (ret == (size_t)-1) {
+ int saved_errno = errno;
+ iconv(cd, NULL, NULL, NULL, NULL);
+ errno = saved_errno;
+ }
return ret;
#else
errno = EINVAL;
#endif
}
-/*
- this is a simple portable iconv() implementaion. It only knows about
- a very small number of character sets - just enough that Samba works
- on systems that don't have iconv
- */
+/**
+ * This is a simple portable iconv() implementaion.
+ *
+ * It only knows about a very small number of character sets - just
+ * enough that Samba works on systems that don't have iconv.
+ **/
size_t smb_iconv(smb_iconv_t cd,
const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
/* in many cases we can go direct */
if (cd->direct) {
return cd->direct(cd->cd_direct,
- (char **)inbuf, inbytesleft, outbuf, outbytesleft);
+ inbuf, inbytesleft, outbuf, outbytesleft);
}
bufsize = sizeof(cvtbuf);
if (cd->pull(cd->cd_pull,
- (char **)inbuf, inbytesleft, &bufp, &bufsize) == -1
+ inbuf, inbytesleft, &bufp, &bufsize) == -1
&& errno != E2BIG) return -1;
bufp = cvtbuf;
bufsize = sizeof(cvtbuf) - bufsize;
if (cd->push(cd->cd_push,
- &bufp, &bufsize,
+ (const char **)&bufp, &bufsize,
outbuf, outbytesleft) == -1) return -1;
}
return 0;
}
+
+static BOOL is_utf16(const char *name)
+{
+ return strcasecmp(name, "UCS-2LE") == 0 ||
+ strcasecmp(name, "UTF-16LE") == 0;
+}
+
/*
simple iconv_open() wrapper
*/
smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
{
smb_iconv_t ret;
- int from, to;
+ struct charset_functions *from, *to;
+
+ lazy_initialize_iconv();
+ from = charsets;
+ to = charsets;
- ret = (smb_iconv_t)malloc(sizeof(*ret));
+ ret = SMB_MALLOC_P(struct _smb_iconv_t);
if (!ret) {
errno = ENOMEM;
return (smb_iconv_t)-1;
}
- memset(ret, 0, sizeof(*ret));
+ memset(ret, 0, sizeof(struct _smb_iconv_t));
- ret->from_name = strdup(fromcode);
- ret->to_name = strdup(tocode);
+ ret->from_name = SMB_STRDUP(fromcode);
+ ret->to_name = SMB_STRDUP(tocode);
/* check for the simplest null conversion */
- if (strcmp(fromcode, tocode) == 0) {
+ if (strcasecmp(fromcode, tocode) == 0) {
ret->direct = iconv_copy;
return ret;
}
- for (from=0; charsets[from].name; from++) {
- if (strcasecmp(charsets[from].name, fromcode) == 0) break;
- }
- for (to=0; charsets[to].name; to++) {
- if (strcasecmp(charsets[to].name, tocode) == 0) break;
- }
+ /* check if we have a builtin function for this conversion */
+ from = find_charset_functions(fromcode);
+ if(from)ret->pull = from->pull;
+
+ to = find_charset_functions(tocode);
+ if(to)ret->push = to->push;
+ /* check if we can use iconv for this conversion */
#ifdef HAVE_NATIVE_ICONV
- if (!charsets[from].name) {
- ret->pull = sys_iconv;
- ret->cd_pull = iconv_open("UCS-2LE", fromcode);
- if (ret->cd_pull == (iconv_t)-1) goto failed;
- }
- if (!charsets[to].name) {
- ret->push = sys_iconv;
- ret->cd_push = iconv_open(tocode, "UCS-2LE");
- if (ret->cd_push == (iconv_t)-1) goto failed;
+ if (!ret->pull) {
+ ret->cd_pull = iconv_open("UTF-16LE", fromcode);
+ if (ret->cd_pull == (iconv_t)-1)
+ ret->cd_pull = iconv_open("UCS-2LE", fromcode);
+ if (ret->cd_pull != (iconv_t)-1)
+ ret->pull = sys_iconv;
}
-#else
- if (!charsets[from].name || !charsets[to].name) {
- goto failed;
+
+ if (!ret->push) {
+ ret->cd_push = iconv_open(tocode, "UTF-16LE");
+ if (ret->cd_push == (iconv_t)-1)
+ ret->cd_push = iconv_open(tocode, "UCS-2LE");
+ if (ret->cd_push != (iconv_t)-1)
+ ret->push = sys_iconv;
}
#endif
+
+ /* check if there is a module available that can do this conversion */
+ if (!ret->pull && NT_STATUS_IS_OK(smb_probe_module("charset", fromcode))) {
+ if(!(from = find_charset_functions(fromcode)))
+ DEBUG(0, ("Module %s doesn't provide charset %s!\n", fromcode, fromcode));
+ else
+ ret->pull = from->pull;
+ }
+
+ if (!ret->push && NT_STATUS_IS_OK(smb_probe_module("charset", tocode))) {
+ if(!(to = find_charset_functions(tocode)))
+ DEBUG(0, ("Module %s doesn't provide charset %s!\n", tocode, tocode));
+ else
+ ret->push = to->push;
+ }
+
+ if (!ret->push || !ret->pull) {
+ SAFE_FREE(ret->from_name);
+ SAFE_FREE(ret->to_name);
+ SAFE_FREE(ret);
+ errno = EINVAL;
+ return (smb_iconv_t)-1;
+ }
/* check for conversion to/from ucs2 */
- if (from == 0 && charsets[to].name) {
- ret->direct = charsets[to].push;
+ if (is_utf16(fromcode) && to) {
+ ret->direct = to->push;
+ ret->push = ret->pull = NULL;
return ret;
}
- if (to == 0 && charsets[from].name) {
- ret->direct = charsets[from].pull;
+
+ if (is_utf16(tocode) && from) {
+ ret->direct = from->pull;
+ ret->push = ret->pull = NULL;
return ret;
}
+ /* Check if we can do the conversion direct */
#ifdef HAVE_NATIVE_ICONV
- if (from == 0) {
+ if (is_utf16(fromcode)) {
ret->direct = sys_iconv;
ret->cd_direct = ret->cd_push;
ret->cd_push = NULL;
return ret;
}
- if (to == 0) {
+ if (is_utf16(tocode)) {
ret->direct = sys_iconv;
ret->cd_direct = ret->cd_pull;
ret->cd_pull = NULL;
}
#endif
- /* the general case has to go via a buffer */
- if (!ret->pull) ret->pull = charsets[from].pull;
- if (!ret->push) ret->push = charsets[to].push;
return ret;
-
-failed:
- SAFE_FREE(ret);
- errno = EINVAL;
- return (smb_iconv_t)-1;
}
/*
multi-byte character set support for english users
***********************************************************************/
-static size_t ascii_pull(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t ascii_pull(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
while (*inbytesleft >= 1 && *outbytesleft >= 2) {
return 0;
}
-static size_t ascii_push(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t ascii_push(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
int ir_count=0;
return ir_count;
}
+static size_t latin1_push(void *cd, const char **inbuf, size_t *inbytesleft,
+ char **outbuf, size_t *outbytesleft)
+{
+ int ir_count=0;
+
+ while (*inbytesleft >= 2 && *outbytesleft >= 1) {
+ (*outbuf)[0] = (*inbuf)[0];
+ if ((*inbuf)[1]) ir_count++;
+ (*inbytesleft) -= 2;
+ (*outbytesleft) -= 1;
+ (*inbuf) += 2;
+ (*outbuf) += 1;
+ }
+
+ if (*inbytesleft == 1) {
+ errno = EINVAL;
+ return -1;
+ }
+
+ if (*inbytesleft > 1) {
+ errno = E2BIG;
+ return -1;
+ }
+
+ return ir_count;
+}
-static size_t ucs2hex_pull(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t ucs2hex_pull(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
while (*inbytesleft >= 1 && *outbytesleft >= 2) {
return 0;
}
-static size_t ucs2hex_push(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
while (*inbytesleft >= 2 && *outbytesleft >= 1) {
return 0;
}
-
-/* the "weird" character set is very useful for testing multi-byte
- support and finding bugs. Don't use on a production system!
-*/
-static struct {
- char from;
- char *to;
- int len;
-} weird_table[] = {
- {'q', "^q^", 3},
- {'Q', "^Q^", 3},
- {0, NULL}
-};
-
-static size_t weird_pull(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
- while (*inbytesleft >= 1 && *outbytesleft >= 2) {
- int i;
- int done = 0;
- for (i=0;weird_table[i].from;i++) {
- if (strncmp((*inbuf),
- weird_table[i].to,
- weird_table[i].len) == 0) {
- if (*inbytesleft < weird_table[i].len) {
- DEBUG(0,("ERROR: truncated weird string\n"));
- /* smb_panic("weird_pull"); */
-
- } else {
- (*outbuf)[0] = weird_table[i].from;
- (*outbuf)[1] = 0;
- (*inbytesleft) -= weird_table[i].len;
- (*outbytesleft) -= 2;
- (*inbuf) += weird_table[i].len;
- (*outbuf) += 2;
- done = 1;
- break;
- }
- }
- }
- if (done) continue;
- (*outbuf)[0] = (*inbuf)[0];
- (*outbuf)[1] = 0;
- (*inbytesleft) -= 1;
- (*outbytesleft) -= 2;
- (*inbuf) += 1;
- (*outbuf) += 2;
- }
-
- if (*inbytesleft > 0) {
- errno = E2BIG;
- return -1;
- }
-
- return 0;
-}
-
-static size_t weird_push(void *cd, char **inbuf, size_t *inbytesleft,
- char **outbuf, size_t *outbytesleft)
-{
- int ir_count=0;
+ int n;
- while (*inbytesleft >= 2 && *outbytesleft >= 1) {
- int i;
- int done=0;
- for (i=0;weird_table[i].from;i++) {
- if ((*inbuf)[0] == weird_table[i].from &&
- (*inbuf)[1] == 0) {
- if (*outbytesleft < weird_table[i].len) {
- DEBUG(0,("No room for weird character\n"));
- /* smb_panic("weird_push"); */
- } else {
- memcpy(*outbuf, weird_table[i].to,
- weird_table[i].len);
- (*inbytesleft) -= 2;
- (*outbytesleft) -= weird_table[i].len;
- (*inbuf) += 2;
- (*outbuf) += weird_table[i].len;
- done = 1;
- break;
- }
- }
- }
- if (done) continue;
+ n = MIN(*inbytesleft, *outbytesleft);
- (*outbuf)[0] = (*inbuf)[0];
- if ((*inbuf)[1]) ir_count++;
- (*inbytesleft) -= 2;
- (*outbytesleft) -= 1;
- (*inbuf) += 2;
- (*outbuf) += 1;
+ swab(*inbuf, *outbuf, (n&~1));
+ if (n&1) {
+ (*outbuf)[n-1] = 0;
}
- if (*inbytesleft == 1) {
- errno = EINVAL;
- return -1;
- }
+ (*inbytesleft) -= n;
+ (*outbytesleft) -= n;
+ (*inbuf) += n;
+ (*outbuf) += n;
- if (*inbytesleft > 1) {
+ if (*inbytesleft > 0) {
errno = E2BIG;
return -1;
}
-
- return ir_count;
+
+ return 0;
}
-static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
int n;
return 0;
}
-static size_t utf8_pull(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
- while (*inbytesleft >= 1 && *outbytesleft >= 2) {
- unsigned char *c = (unsigned char *)*inbuf;
- unsigned char *uc = (unsigned char *)*outbuf;
- int len = 1;
+ size_t in_left=*inbytesleft, out_left=*outbytesleft;
+ const uint8 *c = (const uint8 *)*inbuf;
+ uint8 *uc = (uint8 *)*outbuf;
+ while (in_left >= 1 && out_left >= 2) {
if ((c[0] & 0x80) == 0) {
uc[0] = c[0];
uc[1] = 0;
- } else if ((c[0] & 0xf0) == 0xe0) {
- if (*inbytesleft < 3) {
- DEBUG(0,("short utf8 char\n"));
- goto badseq;
+ c += 1;
+ in_left -= 1;
+ out_left -= 2;
+ uc += 2;
+ continue;
+ }
+
+ if ((c[0] & 0xe0) == 0xc0) {
+ if (in_left < 2 ||
+ (c[1] & 0xc0) != 0x80) {
+ errno = EILSEQ;
+ goto error;
+ }
+ uc[1] = (c[0]>>2) & 0x7;
+ uc[0] = (c[0]<<6) | (c[1]&0x3f);
+ c += 2;
+ in_left -= 2;
+ out_left -= 2;
+ uc += 2;
+ continue;
+ }
+
+ if ((c[0] & 0xf0) == 0xe0) {
+ if (in_left < 3 ||
+ (c[1] & 0xc0) != 0x80 ||
+ (c[2] & 0xc0) != 0x80) {
+ errno = EILSEQ;
+ goto error;
}
uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
uc[0] = (c[1]<<6) | (c[2]&0x3f);
- len = 3;
- } else if ((c[0] & 0xe0) == 0xc0) {
- if (*inbytesleft < 2) {
- DEBUG(0,("short utf8 char\n"));
- goto badseq;
+ c += 3;
+ in_left -= 3;
+ out_left -= 2;
+ uc += 2;
+ continue;
+ }
+
+ if ((c[0] & 0xf8) == 0xf0) {
+ unsigned int codepoint;
+ if (in_left < 4 ||
+ (c[1] & 0xc0) != 0x80 ||
+ (c[2] & 0xc0) != 0x80 ||
+ (c[3] & 0xc0) != 0x80) {
+ errno = EILSEQ;
+ goto error;
}
- uc[1] = (c[0]>>2) & 0x7;
- uc[0] = (c[0]<<6) | (c[1]&0x3f);
- len = 2;
+ codepoint =
+ (c[3]&0x3f) |
+ ((c[2]&0x3f)<<6) |
+ ((c[1]&0x3f)<<12) |
+ ((c[0]&0x7)<<18);
+ if (codepoint < 0x10000) {
+ /* accept UTF-8 characters that are not
+ minimally packed, but pack the result */
+ uc[0] = (codepoint & 0xFF);
+ uc[1] = (codepoint >> 8);
+ c += 4;
+ in_left -= 4;
+ out_left -= 2;
+ uc += 2;
+ continue;
+ }
+
+ codepoint -= 0x10000;
+
+ if (out_left < 4) {
+ errno = E2BIG;
+ goto error;
+ }
+
+ uc[0] = (codepoint>>10) & 0xFF;
+ uc[1] = (codepoint>>18) | 0xd8;
+ uc[2] = codepoint & 0xFF;
+ uc[3] = ((codepoint>>8) & 0x3) | 0xdc;
+ c += 4;
+ in_left -= 4;
+ out_left -= 4;
+ uc += 4;
+ continue;
}
- (*inbuf) += len;
- (*inbytesleft) -= len;
- (*outbytesleft) -= 2;
- (*outbuf) += 2;
+ /* we don't handle 5 byte sequences */
+ errno = EINVAL;
+ goto error;
}
- if (*inbytesleft > 0) {
+ if (in_left > 0) {
errno = E2BIG;
- return -1;
+ goto error;
}
-
+
+ *inbytesleft = in_left;
+ *outbytesleft = out_left;
+ *inbuf = c;
+ *outbuf = uc;
return 0;
-badseq:
- errno = EINVAL;
+error:
+ *inbytesleft = in_left;
+ *outbytesleft = out_left;
+ *inbuf = c;
+ *outbuf = uc;
return -1;
}
-static size_t utf8_push(void *cd, char **inbuf, size_t *inbytesleft,
- char **outbuf, size_t *outbytesleft)
+static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
+ char **outbuf, size_t *outbytesleft)
{
- while (*inbytesleft >= 2 && *outbytesleft >= 1) {
- unsigned char *c = (unsigned char *)*outbuf;
- unsigned char *uc = (unsigned char *)*inbuf;
- int len=1;
-
- if (uc[1] & 0xf8) {
- if (*outbytesleft < 3) {
- DEBUG(0,("short utf8 write\n"));
- goto toobig;
+ size_t in_left=*inbytesleft, out_left=*outbytesleft;
+ uint8 *c = (uint8 *)*outbuf;
+ const uint8 *uc = (const uint8 *)*inbuf;
+
+ while (in_left >= 2 && out_left >= 1) {
+ unsigned int codepoint;
+
+ if (uc[1] == 0 && !(uc[0] & 0x80)) {
+ /* simplest case */
+ c[0] = uc[0];
+ in_left -= 2;
+ out_left -= 1;
+ uc += 2;
+ c += 1;
+ continue;
+ }
+
+ if ((uc[1]&0xf8) == 0) {
+ /* next simplest case */
+ if (out_left < 2) {
+ errno = E2BIG;
+ goto error;
}
- c[0] = 0xe0 | (uc[1]>>4);
- c[1] = 0x80 | ((uc[1]&0xF)<<2) | (uc[0]>>6);
- c[2] = 0x80 | (uc[0]&0x3f);
- len = 3;
- } else if (uc[1] | (uc[0] & 0x80)) {
- if (*outbytesleft < 2) {
- DEBUG(0,("short utf8 write\n"));
- goto toobig;
+ c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
+ c[1] = 0x80 | (uc[0] & 0x3f);
+ in_left -= 2;
+ out_left -= 2;
+ uc += 2;
+ c += 2;
+ continue;
+ }
+
+ if ((uc[1] & 0xfc) == 0xdc) {
+ /* its the second part of a 4 byte sequence. Illegal */
+ if (in_left < 4) {
+ errno = EINVAL;
+ } else {
+ errno = EILSEQ;
}
- c[0] = 0xc0 | (uc[1]<<2) | (uc[0]>>6);
- c[1] = 0x80 | (uc[0]&0x3f);
- len = 2;
- } else {
- c[0] = uc[0];
+ goto error;
}
+ if ((uc[1] & 0xfc) != 0xd8) {
+ codepoint = uc[0] | (uc[1]<<8);
+ if (out_left < 3) {
+ errno = E2BIG;
+ goto error;
+ }
+ c[0] = 0xe0 | (codepoint >> 12);
+ c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
+ c[2] = 0x80 | (codepoint & 0x3f);
+
+ in_left -= 2;
+ out_left -= 3;
+ uc += 2;
+ c += 3;
+ continue;
+ }
- (*inbytesleft) -= 2;
- (*outbytesleft) -= len;
- (*inbuf) += 2;
- (*outbuf) += len;
+ /* its the first part of a 4 byte sequence */
+ if (in_left < 4) {
+ errno = EINVAL;
+ goto error;
+ }
+ if ((uc[3] & 0xfc) != 0xdc) {
+ errno = EILSEQ;
+ goto error;
+ }
+ codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) |
+ (uc[0]<<10) | ((uc[1] & 0x3)<<18));
+
+ if (out_left < 4) {
+ errno = E2BIG;
+ goto error;
+ }
+ c[0] = 0xf0 | (codepoint >> 18);
+ c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
+ c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
+ c[3] = 0x80 | (codepoint & 0x3f);
+
+ in_left -= 4;
+ out_left -= 4;
+ uc += 4;
+ c += 4;
}
- if (*inbytesleft == 1) {
+ if (in_left == 1) {
errno = EINVAL;
- return -1;
+ goto error;
}
- if (*inbytesleft > 1) {
+ if (in_left > 1) {
errno = E2BIG;
- return -1;
+ goto error;
}
+
+ *inbytesleft = in_left;
+ *outbytesleft = out_left;
+ *inbuf = uc;
+ *outbuf = c;
return 0;
-toobig:
- errno = E2BIG;
+error:
+ *inbytesleft = in_left;
+ *outbytesleft = out_left;
+ *inbuf = uc;
+ *outbuf = c;
return -1;
}