* Copyright (C) 1995-2006, Index Data ApS
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.22 2006-04-24 23:21:26 adam Exp $
+ * $Id: siconv.c,v 1.27 2006-08-28 12:34:41 adam Exp $
*/
/**
* \file siconv.c
* is used by YAZ to interface with iconv (if present).
* For systems where iconv is not present, this layer
* provides a few important conversions: UTF-8, MARC-8, Latin-1.
+ *
+ * MARC-8 reference:
+ * http://www.loc.gov/marc/specifications/speccharmarc8.html
*/
#if HAVE_CONFIG_H
#include <config.h>
#endif
+#include <assert.h>
#include <errno.h>
#include <string.h>
#include <ctype.h>
#include <iconv.h>
#endif
+
#include <yaz/yaz-util.h>
unsigned long yaz_marc8_1_conv(unsigned char *inp, size_t inbytesleft,
return x;
}
+static size_t yaz_init_marc8(yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
+{
+ cd->marc8_esc_mode = 'B';
+
+ cd->comb_offset = cd->comb_size = 0;
+ cd->compose_char = 0;
+
+ cd->write_marc8_comb_no = 0;
+ cd->write_marc8_last = 0;
+ cd->write_marc8_page_chr = "\033(B";
+
+ return 0;
+}
+
static size_t yaz_init_UTF8 (yaz_iconv_t cd, unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
return 0;
}
-static unsigned long yaz_read_UTF8 (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
+unsigned long yaz_read_UTF8_char(unsigned char *inp,
+ size_t inbytesleft, size_t *no_read,
+ int *error)
{
unsigned long x = 0;
else if (inp[0] <= 0xbf || inp[0] >= 0xfe)
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
+ *error = YAZ_ICONV_EILSEQ;
}
else if (inp[0] <= 0xdf && inbytesleft >= 2)
{
else
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
+ *error = YAZ_ICONV_EILSEQ;
}
}
else if (inp[0] <= 0xef && inbytesleft >= 3)
else
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
+ *error = YAZ_ICONV_EILSEQ;
}
}
else if (inp[0] <= 0xf7 && inbytesleft >= 4)
else
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
+ *error = YAZ_ICONV_EILSEQ;
}
}
else if (inp[0] <= 0xfb && inbytesleft >= 5)
else
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
+ *error = YAZ_ICONV_EILSEQ;
}
}
else if (inp[0] <= 0xfd && inbytesleft >= 6)
else
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
+ *error = YAZ_ICONV_EILSEQ;
}
}
else
{
*no_read = 0;
- cd->my_errno = YAZ_ICONV_EINVAL;
+ *error = YAZ_ICONV_EINVAL;
}
return x;
}
+static unsigned long yaz_read_UTF8 (yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
+{
+ return yaz_read_UTF8_char(inp, inbytesleft, no_read, &cd->my_errno);
+}
+
static unsigned long yaz_read_UCS4 (yaz_iconv_t cd, unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
}
}
-static size_t yaz_write_UTF8 (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+static size_t yaz_write_UTF8(yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft,
+ int last)
+{
+ return yaz_write_UTF8_char(x, outbuf, outbytesleft, &cd->my_errno);
+}
+
+size_t yaz_write_UTF8_char(unsigned long x,
+ char **outbuf, size_t *outbytesleft,
+ int *error)
{
unsigned char *outp = (unsigned char *) *outbuf;
}
else
{
- cd->my_errno = YAZ_ICONV_E2BIG; /* not room for output */
+ *error = YAZ_ICONV_E2BIG; /* not room for output */
return (size_t)(-1);
}
*outbuf = (char *) outp;
return 0;
}
-static size_t yaz_write_marc8(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft,
+ int last)
{
int comb = 0;
const char *page_chr = 0;
else
{
size_t r = flush_combos(cd, outbuf, outbytesleft);
+ const char *old_page_chr = cd->write_marc8_page_chr;
if (r)
return r;
- if (strcmp(page_chr, cd->write_marc8_page_chr))
+ if (strcmp(page_chr, old_page_chr))
{
- size_t plen = strlen(page_chr);
+ size_t plen = 0;
+ const char *page_out = page_chr;
- if (*outbytesleft < plen)
+ if (*outbytesleft < 8)
{
cd->my_errno = YAZ_ICONV_E2BIG;
+
return (size_t) (-1);
}
- memcpy(*outbuf, page_chr, plen);
+ cd->write_marc8_page_chr = page_chr;
+
+ if (!strcmp(old_page_chr, "\033p")
+ || !strcmp(old_page_chr, "\033g")
+ || !strcmp(old_page_chr, "\033b"))
+ {
+ /* Technique 1 leave */
+ page_out = "\033s";
+ if (strcmp(page_chr, "\033(B")) /* Not going ASCII page? */
+ {
+ /* Must leave script + enter new page */
+ plen = strlen(page_out);
+ memcpy(*outbuf, page_out, plen);
+ (*outbuf) += plen;
+ (*outbytesleft) -= plen;
+ page_out = page_chr;
+ }
+ }
+ plen = strlen(page_out);
+ memcpy(*outbuf, page_out, plen);
(*outbuf) += plen;
(*outbytesleft) -= plen;
- cd->write_marc8_page_chr = page_chr;
}
cd->write_marc8_last = y;
}
return 0;
}
+static size_t yaz_write_marc8(yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft,
+ int last)
+{
+ int i;
+ for (i = 0; latin1_comb[i].x1; i++)
+ {
+ if (x == latin1_comb[i].y)
+ {
+ size_t r ;
+ /* save the output pointers .. */
+ char *outbuf0 = *outbuf;
+ size_t outbytesleft0 = *outbytesleft;
+ int last_ch = cd->write_marc8_last;
+
+ r = yaz_write_marc8_2(cd, latin1_comb[i].x1,
+ outbuf, outbytesleft, 0);
+ if (r)
+ return r;
+ r = yaz_write_marc8_2(cd, latin1_comb[i].x2,
+ outbuf, outbytesleft, last);
+ if (r && cd->my_errno == YAZ_ICONV_E2BIG)
+ {
+ /* not enough room. reset output to original values */
+ *outbuf = outbuf0;
+ *outbytesleft = outbytesleft0;
+ cd->write_marc8_last = last_ch;
+ }
+ return r;
+ }
+ }
+ return yaz_write_marc8_2(cd, x, outbuf, outbytesleft, last);
+}
+
+
#if HAVE_WCHAR_H
static size_t yaz_write_wchar_t (yaz_iconv_t cd, unsigned long x,
char **outbuf, size_t *outbytesleft,
cd->read_handle = 0;
cd->init_handle = 0;
cd->my_errno = YAZ_ICONV_UNKNOWN;
- cd->marc8_esc_mode = 'B';
- cd->comb_offset = cd->comb_size = 0;
- cd->compose_char = 0;
-
- cd->write_marc8_comb_no = 0;
- cd->write_marc8_last = 0;
- cd->write_marc8_page_chr = "\033(B";
/* a useful hack: if fromcode has leading @,
the library not use YAZ's own conversions .. */
else if (!yaz_matchstr(fromcode, "UCS4LE"))
cd->read_handle = yaz_read_UCS4LE;
else if (!yaz_matchstr(fromcode, "MARC8"))
+ {
cd->read_handle = yaz_read_marc8;
+ cd->init_handle = yaz_init_marc8;
+ }
else if (!yaz_matchstr(fromcode, "MARC8s"))
+ {
cd->read_handle = yaz_read_marc8s;
+ cd->init_handle = yaz_init_marc8;
+ }
#if HAVE_WCHAR_H
else if (!yaz_matchstr(fromcode, "WCHAR_T"))
cd->read_handle = yaz_read_wchar_t;
else if (!yaz_matchstr(tocode, "UCS4LE"))
cd->write_handle = yaz_write_UCS4LE;
else if (!yaz_matchstr(tocode, "MARC8"))
+ {
cd->write_handle = yaz_write_marc8;
+ cd->init_handle = yaz_init_marc8;
+ }
else if (!yaz_matchstr(tocode, "MARC8s"))
+ {
cd->write_handle = yaz_write_marc8;
+ cd->init_handle = yaz_init_marc8;
+ }
#if HAVE_WCHAR_H
else if (!yaz_matchstr(tocode, "WCHAR_T"))
cd->write_handle = yaz_write_wchar_t;
{
if (cd->init_handle)
{
- size_t no_read;
+ size_t no_read = 0;
size_t r = (cd->init_handle)(cd, (unsigned char *) *inbuf,
*inbytesleft, &no_read);
if (r)