summaryrefslogtreecommitdiffstats
path: root/source/lib/iconv.c
diff options
context:
space:
mode:
Diffstat (limited to 'source/lib/iconv.c')
-rw-r--r--source/lib/iconv.c347
1 files changed, 261 insertions, 86 deletions
diff --git a/source/lib/iconv.c b/source/lib/iconv.c
index 4c9ecf992e6..66a6e6cd8bc 100644
--- a/source/lib/iconv.c
+++ b/source/lib/iconv.c
@@ -51,18 +51,26 @@
* @sa Samba Developers Guide
**/
-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 latin1_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 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 *);
+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},
{"646", ascii_pull, ascii_push},
{"ISO-8859-1", ascii_pull, latin1_push},
@@ -122,12 +130,12 @@ static void lazy_initialize_iconv(void)
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) {
int saved_errno = errno;
@@ -148,7 +156,7 @@ static size_t sys_iconv(void *cd,
* enough that Samba works on systems that don't have iconv.
**/
size_t smb_iconv(smb_iconv_t cd,
- char **inbuf, size_t *inbytesleft,
+ const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
char cvtbuf[2048];
@@ -158,7 +166,7 @@ size_t smb_iconv(smb_iconv_t cd,
/* 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);
}
@@ -168,20 +176,27 @@ size_t smb_iconv(smb_iconv_t cd,
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
*/
@@ -220,13 +235,17 @@ smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
/* check if we can use iconv for this conversion */
#ifdef HAVE_NATIVE_ICONV
if (!ret->pull) {
- ret->cd_pull = iconv_open("UCS-2LE", fromcode);
+ 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;
}
if (!ret->push) {
- ret->cd_push = iconv_open(tocode, "UCS-2LE");
+ 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;
}
@@ -256,13 +275,13 @@ smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
}
/* check for conversion to/from ucs2 */
- if (strcasecmp(fromcode, "UCS-2LE") == 0 && to) {
+ if (is_utf16(fromcode) && to) {
ret->direct = to->push;
ret->push = ret->pull = NULL;
return ret;
}
- if (strcasecmp(tocode, "UCS-2LE") == 0 && from) {
+ if (is_utf16(tocode) && from) {
ret->direct = from->pull;
ret->push = ret->pull = NULL;
return ret;
@@ -270,13 +289,13 @@ smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode)
/* Check if we can do the conversion direct */
#ifdef HAVE_NATIVE_ICONV
- if (strcasecmp(fromcode, "UCS-2LE") == 0) {
+ if (is_utf16(fromcode)) {
ret->direct = sys_iconv;
ret->cd_direct = ret->cd_push;
ret->cd_push = NULL;
return ret;
}
- if (strcasecmp(tocode, "UCS-2LE") == 0) {
+ if (is_utf16(tocode)) {
ret->direct = sys_iconv;
ret->cd_direct = ret->cd_pull;
ret->cd_pull = NULL;
@@ -313,7 +332,7 @@ int smb_iconv_close (smb_iconv_t cd)
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) {
@@ -333,7 +352,7 @@ static size_t ascii_pull(void *cd, char **inbuf, size_t *inbytesleft,
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;
@@ -360,7 +379,7 @@ static size_t ascii_push(void *cd, char **inbuf, size_t *inbytesleft,
return ir_count;
}
-static size_t latin1_push(void *cd, char **inbuf, size_t *inbytesleft,
+static size_t latin1_push(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
int ir_count=0;
@@ -387,7 +406,7 @@ static size_t latin1_push(void *cd, char **inbuf, size_t *inbytesleft,
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) {
@@ -430,7 +449,7 @@ static size_t ucs2hex_pull(void *cd, char **inbuf, size_t *inbytesleft,
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) {
@@ -471,8 +490,32 @@ static size_t ucs2hex_push(void *cd, char **inbuf, size_t *inbytesleft,
return 0;
}
+static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
+ char **outbuf, size_t *outbytesleft)
+{
+ int n;
+
+ n = MIN(*inbytesleft, *outbytesleft);
+
+ swab(*inbuf, *outbuf, (n&~1));
+ if (n&1) {
+ (*outbuf)[n-1] = 0;
+ }
+
+ (*inbytesleft) -= n;
+ (*outbytesleft) -= n;
+ (*inbuf) += n;
+ (*outbuf) += n;
+
+ if (*inbytesleft > 0) {
+ errno = E2BIG;
+ return -1;
+ }
-static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
+ return 0;
+}
+
+static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
int n;
@@ -494,102 +537,234 @@ static size_t iconv_copy(void *cd, char **inbuf, size_t *inbytesleft,
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;
}
+
generated by cgit (git 2.34.1) at 2024-06-10 20:48:08 +0000