/*
- * Copyright (c) 1997-2002, Index Data
+ * Copyright (c) 1997-2003, Index Data
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.2 2002-08-27 21:45:28 adam Exp $
+ * $Id: siconv.c,v 1.9 2003-01-06 08:20:28 adam Exp $
*/
/* mini iconv and wrapper for system iconv library (if present) */
#include <errno.h>
#include <string.h>
#include <ctype.h>
+#if HAVE_WCHAR_H
+#include <wchar.h>
+#endif
#if HAVE_ICONV_H
#include <iconv.h>
#include <yaz/yaz-util.h>
+unsigned long yaz_marc8_conv (unsigned char *inp, size_t inbytesleft,
+ size_t *no_read);
+
struct yaz_iconv_struct {
int my_errno;
int init_flag;
return x;
}
+#if HAVE_WCHAR_H
+static unsigned long yaz_read_wchar_t (yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
+{
+ unsigned long x = 0;
+
+ if (inbytesleft < sizeof(wchar_t))
+ {
+ cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */
+ *no_read = 0;
+ }
+ else
+ {
+ wchar_t wch;
+ memcpy (&wch, inp, sizeof(wch));
+ x = wch;
+ *no_read = sizeof(wch);
+ }
+ return x;
+}
+#endif
+
+static unsigned long yaz_read_marc8 (yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
+{
+ return yaz_marc8_conv(inp, inbytesleft, no_read);
+}
+
static size_t yaz_write_UTF8 (yaz_iconv_t cd, unsigned long x,
char **outbuf, size_t *outbytesleft)
{
- unsigned char *outp = *outbuf;
+ unsigned char *outp = (unsigned char *) *outbuf;
if (x <= 0x7f && *outbytesleft >= 1)
{
- *outp++ = x;
+ *outp++ = (unsigned char) x;
(*outbytesleft)--;
}
else if (x <= 0x7ff && *outbytesleft >= 2)
{
- *outp++ = (x >> 6) | 0xc0;
- *outp++ = (x & 0x3f) | 0x80;
+ *outp++ = (unsigned char) ((x >> 6) | 0xc0);
+ *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
(*outbytesleft) -= 2;
}
else if (x <= 0xffff && *outbytesleft >= 3)
{
- *outp++ = (x >> 12) | 0xe0;
- *outp++ = ((x >> 6) & 0x3f) | 0x80;
- *outp++ = (x & 0x3f) | 0x80;
+ *outp++ = (unsigned char) ((x >> 12) | 0xe0);
+ *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
(*outbytesleft) -= 3;
}
else if (x <= 0x1fffff && *outbytesleft >= 4)
{
- *outp++ = (x >> 18) | 0xf0;
- *outp++ = ((x >> 12) & 0x3f) | 0x80;
- *outp++ = ((x >> 6) & 0x3f) | 0x80;
- *outp++ = (x & 0x3f) | 0x80;
+ *outp++ = (unsigned char) ((x >> 18) | 0xf0);
+ *outp++ = (unsigned char) (((x >> 12) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
(*outbytesleft) -= 4;
}
else if (x <= 0x3ffffff && *outbytesleft >= 5)
{
- *outp++ = (x >> 24) | 0xf8;
- *outp++ = ((x >> 18) & 0x3f) | 0x80;
- *outp++ = ((x >> 12) & 0x3f) | 0x80;
- *outp++ = ((x >> 6) & 0x3f) | 0x80;
- *outp++ = (x & 0x3f) | 0x80;
+ *outp++ = (unsigned char) ((x >> 24) | 0xf8);
+ *outp++ = (unsigned char) (((x >> 18) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) (((x >> 12) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
(*outbytesleft) -= 5;
}
else if (*outbytesleft >= 6)
{
- *outp++ = (x >> 30) | 0xfc;
- *outp++ = ((x >> 24) & 0x3f) | 0x80;
- *outp++ = ((x >> 18) & 0x3f) | 0x80;
- *outp++ = ((x >> 12) & 0x3f) | 0x80;
- *outp++ = ((x >> 6) & 0x3f) | 0x80;
- *outp++ = (x & 0x3f) | 0x80;
+ *outp++ = (unsigned char) ((x >> 30) | 0xfc);
+ *outp++ = (unsigned char) (((x >> 24) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) (((x >> 18) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) (((x >> 12) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
+ *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
(*outbytesleft) -= 6;
}
else
cd->my_errno = YAZ_ICONV_E2BIG; /* not room for output */
return (size_t)(-1);
}
- *outbuf = outp;
+ *outbuf = (char *) outp;
return 0;
}
static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x,
char **outbuf, size_t *outbytesleft)
{
- unsigned char *outp = *outbuf;
+ unsigned char *outp = (unsigned char *) *outbuf;
if (x > 255 || x < 1)
{
cd->my_errno = YAZ_ICONV_EILSEQ;
}
else if (*outbytesleft >= 1)
{
- *outp++ = x;
+ *outp++ = (unsigned char) x;
(*outbytesleft)--;
}
else
cd->my_errno = YAZ_ICONV_E2BIG;
return (size_t)(-1);
}
- *outbuf = outp;
+ *outbuf = (char *) outp;
return 0;
}
static size_t yaz_write_UCS4 (yaz_iconv_t cd, unsigned long x,
char **outbuf, size_t *outbytesleft)
{
- unsigned char *outp = *outbuf;
+ unsigned char *outp = (unsigned char *) *outbuf;
if (*outbytesleft >= 4)
{
- *outp++ = x<<24;
- *outp++ = x<<16;
- *outp++ = x<<8;
- *outp++ = x;
+ *outp++ = (unsigned char) (x<<24);
+ *outp++ = (unsigned char) (x<<16);
+ *outp++ = (unsigned char) (x<<8);
+ *outp++ = (unsigned char) x;
(*outbytesleft) -= 4;
}
else
cd->my_errno = YAZ_ICONV_E2BIG;
return (size_t)(-1);
}
- *outbuf = outp;
+ *outbuf = (char *) outp;
return 0;
}
static size_t yaz_write_UCS4LE (yaz_iconv_t cd, unsigned long x,
char **outbuf, size_t *outbytesleft)
{
- unsigned char *outp = *outbuf;
+ unsigned char *outp = (unsigned char *) *outbuf;
if (*outbytesleft >= 4)
{
- *outp++ = x;
- *outp++ = x<<8;
- *outp++ = x<<16;
- *outp++ = x<<24;
+ *outp++ = (unsigned char) x;
+ *outp++ = (unsigned char) (x<<8);
+ *outp++ = (unsigned char) (x<<16);
+ *outp++ = (unsigned char) (x<<24);
(*outbytesleft) -= 4;
}
else
cd->my_errno = YAZ_ICONV_E2BIG;
return (size_t)(-1);
}
- *outbuf = outp;
+ *outbuf = (char *) outp;
return 0;
}
+#if HAVE_WCHAR_H
+static size_t yaz_write_wchar_t (yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft)
+{
+ unsigned char *outp = (unsigned char *) *outbuf;
+
+ if (*outbytesleft >= sizeof(wchar_t))
+ {
+ wchar_t wch = x;
+ memcpy(outp, &wch, sizeof(wch));
+ outp += sizeof(wch);
+ (*outbytesleft) -= sizeof(wch);
+ }
+ else
+ {
+ cd->my_errno = YAZ_ICONV_E2BIG;
+ return (size_t)(-1);
+ }
+ *outbuf = (char *) outp;
+ return 0;
+}
+#endif
+
+int yaz_iconv_isbuiltin(yaz_iconv_t cd)
+{
+ return cd->read_handle && cd->write_handle;
+}
+
yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode)
{
- yaz_iconv_t cd = xmalloc (sizeof(*cd));
+ yaz_iconv_t cd = (yaz_iconv_t) xmalloc (sizeof(*cd));
cd->write_handle = 0;
cd->read_handle = 0;
cd->init_handle = 0;
cd->my_errno = YAZ_ICONV_UNKNOWN;
- if (!yaz_matchstr(fromcode, "UTF8"))
+ /* a useful hack: if fromcode has leading @,
+ the library not use YAZ's own conversions .. */
+ if (fromcode[0] == '@')
+ fromcode++;
+ else
{
- cd->read_handle = yaz_read_UTF8;
- cd->init_handle = yaz_init_UTF8;
+ if (!yaz_matchstr(fromcode, "UTF8"))
+ {
+ cd->read_handle = yaz_read_UTF8;
+ cd->init_handle = yaz_init_UTF8;
+ }
+ else if (!yaz_matchstr(fromcode, "ISO88591"))
+ cd->read_handle = yaz_read_ISO8859_1;
+ else if (!yaz_matchstr(fromcode, "UCS4"))
+ cd->read_handle = yaz_read_UCS4;
+ else if (!yaz_matchstr(fromcode, "UCS4LE"))
+ cd->read_handle = yaz_read_UCS4LE;
+ else if (!yaz_matchstr(fromcode, "MARC8"))
+ cd->read_handle = yaz_read_marc8;
+#if HAVE_WCHAR_H
+ else if (!yaz_matchstr(fromcode, "WCHAR_T"))
+ cd->read_handle = yaz_read_wchar_t;
+#endif
+
+ if (!yaz_matchstr(tocode, "UTF8"))
+ cd->write_handle = yaz_write_UTF8;
+ else if (!yaz_matchstr(tocode, "ISO88591"))
+ cd->write_handle = yaz_write_ISO8859_1;
+ else if (!yaz_matchstr (tocode, "UCS4"))
+ cd->write_handle = yaz_write_UCS4;
+ else if (!yaz_matchstr(tocode, "UCS4LE"))
+ cd->write_handle = yaz_write_UCS4LE;
+#if HAVE_WCHAR_H
+ else if (!yaz_matchstr(tocode, "WCHAR_T"))
+ cd->write_handle = yaz_write_wchar_t;
+#endif
}
- else if (!yaz_matchstr(fromcode, "ISO88591"))
- cd->read_handle = yaz_read_ISO8859_1;
- else if (!yaz_matchstr(fromcode, "UCS4"))
- cd->read_handle = yaz_read_UCS4;
- else if (!yaz_matchstr(fromcode, "UCS4LE"))
- cd->read_handle = yaz_read_UCS4LE;
-
- if (!yaz_matchstr(tocode, "UTF8"))
- cd->write_handle = yaz_write_UTF8;
- else if (!yaz_matchstr(tocode, "ISO88591"))
- cd->write_handle = yaz_write_ISO8859_1;
- else if (!yaz_matchstr (tocode, "UCS4"))
- cd->write_handle = yaz_write_UCS4;
- else if (!yaz_matchstr(tocode, "UCS4LE"))
- cd->write_handle = yaz_write_UCS4LE;
-
#if HAVE_ICONV_H
cd->iconv_cd = 0;
if (!cd->read_handle || !cd->write_handle)
iconv(cd->iconv_cd, inbuf, inbytesleft, outbuf, outbytesleft);
if (r == (size_t)(-1))
{
- switch (errno)
+ switch (yaz_errno())
{
case E2BIG:
cd->my_errno = YAZ_ICONV_E2BIG;
if (cd->init_handle)
{
size_t no_read;
- size_t r = (cd->init_handle)(cd, *inbuf, *inbytesleft, &no_read);
+ size_t r = (cd->init_handle)(cd, (unsigned char *) *inbuf,
+ *inbytesleft, &no_read);
if (r)
{
if (cd->my_errno == YAZ_ICONV_EINVAL)
return r;
cd->init_flag = 0;
- if (cd->my_errno == YAZ_ICONV_EILSEQ)
- {
- *inbytesleft++;
- (*inbuf)++;
- }
return r;
}
*inbytesleft -= no_read;
break;
}
- x = (cd->read_handle)(cd, *inbuf, *inbytesleft, &no_read);
+ x = (cd->read_handle)(cd, (unsigned char *) *inbuf, *inbytesleft,
+ &no_read);
if (no_read == 0)
{
r = (size_t)(-1);