* Copyright (C) 1995-2006, Index Data ApS
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.19 2006-04-19 23:15:39 adam Exp $
+ * $Id: siconv.c,v 1.23 2006-05-09 21:37:02 adam Exp $
*/
/**
* \file siconv.c
const char *write_marc8_page_chr;
};
+static struct {
+ unsigned long x1, x2;
+ unsigned y;
+} latin1_comb[] = {
+ { 'A', 0x0300, 0xc0}, /* LATIN CAPITAL LETTER A WITH GRAVE */
+ { 'A', 0x0301, 0xc1}, /* LATIN CAPITAL LETTER A WITH ACUTE */
+ { 'A', 0x0302, 0xc2}, /* LATIN CAPITAL LETTER A WITH CIRCUMFLEX */
+ { 'A', 0x0303, 0xc3}, /* LATIN CAPITAL LETTER A WITH TILDE */
+ { 'A', 0x0308, 0xc4}, /* LATIN CAPITAL LETTER A WITH DIAERESIS */
+ { 'A', 0x030a, 0xc5}, /* LATIN CAPITAL LETTER A WITH RING ABOVE */
+ /* no need for 0xc6 LATIN CAPITAL LETTER AE */
+ { 'C', 0x0327, 0xc7}, /* LATIN CAPITAL LETTER C WITH CEDILLA */
+ { 'E', 0x0300, 0xc8}, /* LATIN CAPITAL LETTER E WITH GRAVE */
+ { 'E', 0x0301, 0xc9}, /* LATIN CAPITAL LETTER E WITH ACUTE */
+ { 'E', 0x0302, 0xca}, /* LATIN CAPITAL LETTER E WITH CIRCUMFLEX */
+ { 'E', 0x0308, 0xcb}, /* LATIN CAPITAL LETTER E WITH DIAERESIS */
+ { 'I', 0x0300, 0xcc}, /* LATIN CAPITAL LETTER I WITH GRAVE */
+ { 'I', 0x0301, 0xcd}, /* LATIN CAPITAL LETTER I WITH ACUTE */
+ { 'I', 0x0302, 0xce}, /* LATIN CAPITAL LETTER I WITH CIRCUMFLEX */
+ { 'I', 0x0308, 0xcf}, /* LATIN CAPITAL LETTER I WITH DIAERESIS */
+ { 'N', 0x0303, 0xd1}, /* LATIN CAPITAL LETTER N WITH TILDE */
+ { 'O', 0x0300, 0xd2}, /* LATIN CAPITAL LETTER O WITH GRAVE */
+ { 'O', 0x0301, 0xd3}, /* LATIN CAPITAL LETTER O WITH ACUTE */
+ { 'O', 0x0302, 0xd4}, /* LATIN CAPITAL LETTER O WITH CIRCUMFLEX */
+ { 'O', 0x0303, 0xd5}, /* LATIN CAPITAL LETTER O WITH TILDE */
+ { 'O', 0x0308, 0xd6}, /* LATIN CAPITAL LETTER O WITH DIAERESIS */
+ /* omitted: 0xd7 MULTIPLICATION SIGN */
+ /* omitted: 0xd8 LATIN CAPITAL LETTER O WITH STROKE */
+ { 'U', 0x0300, 0xd9}, /* LATIN CAPITAL LETTER U WITH GRAVE */
+ { 'U', 0x0301, 0xda}, /* LATIN CAPITAL LETTER U WITH ACUTE */
+ { 'U', 0x0302, 0xdb}, /* LATIN CAPITAL LETTER U WITH CIRCUMFLEX */
+ { 'U', 0x0308, 0xdc}, /* LATIN CAPITAL LETTER U WITH DIAERESIS */
+ { 'Y', 0x0301, 0xdd}, /* LATIN CAPITAL LETTER Y WITH ACUTE */
+ /* omitted: 0xde LATIN CAPITAL LETTER THORN */
+ /* omitted: 0xdf LATIN SMALL LETTER SHARP S */
+ { 'a', 0x0300, 0xe0}, /* LATIN SMALL LETTER A WITH GRAVE */
+ { 'a', 0x0301, 0xe1}, /* LATIN SMALL LETTER A WITH ACUTE */
+ { 'a', 0x0302, 0xe2}, /* LATIN SMALL LETTER A WITH CIRCUMFLEX */
+ { 'a', 0x0303, 0xe3}, /* LATIN SMALL LETTER A WITH TILDE */
+ { 'a', 0x0308, 0xe4}, /* LATIN SMALL LETTER A WITH DIAERESIS */
+ { 'a', 0x030a, 0xe5}, /* LATIN SMALL LETTER A WITH RING ABOVE */
+ /* omitted: 0xe6 LATIN SMALL LETTER AE */
+ { 'c', 0x0327, 0xe7}, /* LATIN SMALL LETTER C WITH CEDILLA */
+ { 'e', 0x0300, 0xe8}, /* LATIN SMALL LETTER E WITH GRAVE */
+ { 'e', 0x0301, 0xe9}, /* LATIN SMALL LETTER E WITH ACUTE */
+ { 'e', 0x0302, 0xea}, /* LATIN SMALL LETTER E WITH CIRCUMFLEX */
+ { 'e', 0x0308, 0xeb}, /* LATIN SMALL LETTER E WITH DIAERESIS */
+ { 'i', 0x0300, 0xec}, /* LATIN SMALL LETTER I WITH GRAVE */
+ { 'i', 0x0301, 0xed}, /* LATIN SMALL LETTER I WITH ACUTE */
+ { 'i', 0x0302, 0xee}, /* LATIN SMALL LETTER I WITH CIRCUMFLEX */
+ { 'i', 0x0308, 0xef}, /* LATIN SMALL LETTER I WITH DIAERESIS */
+ /* omitted: 0xf0 LATIN SMALL LETTER ETH */
+ { 'n', 0x0303, 0xf1}, /* LATIN SMALL LETTER N WITH TILDE */
+ { 'o', 0x0300, 0xf2}, /* LATIN SMALL LETTER O WITH GRAVE */
+ { 'o', 0x0301, 0xf3}, /* LATIN SMALL LETTER O WITH ACUTE */
+ { 'o', 0x0302, 0xf4}, /* LATIN SMALL LETTER O WITH CIRCUMFLEX */
+ { 'o', 0x0303, 0xf5}, /* LATIN SMALL LETTER O WITH TILDE */
+ { 'o', 0x0308, 0xf6}, /* LATIN SMALL LETTER O WITH DIAERESIS */
+ /* omitted: 0xf7 DIVISION SIGN */
+ /* omitted: 0xf8 LATIN SMALL LETTER O WITH STROKE */
+ { 'u', 0x0300, 0xf9}, /* LATIN SMALL LETTER U WITH GRAVE */
+ { 'u', 0x0301, 0xfa}, /* LATIN SMALL LETTER U WITH ACUTE */
+ { 'u', 0x0302, 0xfb}, /* LATIN SMALL LETTER U WITH CIRCUMFLEX */
+ { 'u', 0x0308, 0xfc}, /* LATIN SMALL LETTER U WITH DIAERESIS */
+ { 'y', 0x0301, 0xfd}, /* LATIN SMALL LETTER Y WITH ACUTE */
+ /* omitted: 0xfe LATIN SMALL LETTER THORN */
+ { 'y', 0x0308, 0xff}, /* LATIN SMALL LETTER Y WITH DIAERESIS */
+
+ { 0, 0, 0}
+};
+
static unsigned long yaz_read_ISO8859_1 (yaz_iconv_t cd, unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
return x;
}
-static unsigned long yaz_read_marc8_comb (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read,
- int *comb)
+static unsigned long yaz_read_marc8s(yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
+{
+ unsigned long x = yaz_read_marc8(cd, inp, inbytesleft, no_read);
+ if (x && cd->comb_size == 1)
+ {
+ /* For MARC8s we try to get a Latin-1 page code out of it */
+ int i;
+ for (i = 0; latin1_comb[i].x1; i++)
+ if (cd->comb_x[0] == latin1_comb[i].x2 && x == latin1_comb[i].x1)
+ {
+ *no_read += cd->comb_no_read[0];
+ cd->comb_size = 0;
+ x = latin1_comb[i].y;
+ break;
+ }
+ }
+ return x;
+}
+
+static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read,
+ int *comb)
{
*no_read = 0;
while(inbytesleft >= 1 && inp[0] == 27)
but since MARC-8 to UTF-8 generates these composed sequence
we get a better chance of a successful MARC-8 -> ISO-8859-1
conversion */
- static struct {
- unsigned long x1, x2;
- unsigned y;
- } latin1_comb[] = {
- { 'A', 0x0300, 0xc0}, /* LATIN CAPITAL LETTER A WITH GRAVE */
- { 'A', 0x0301, 0xc1}, /* LATIN CAPITAL LETTER A WITH ACUTE */
- { 'A', 0x0302, 0xc2}, /* LATIN CAPITAL LETTER A WITH CIRCUMFLEX */
- { 'A', 0x0303, 0xc3}, /* LATIN CAPITAL LETTER A WITH TILDE */
- { 'A', 0x0308, 0xc4}, /* LATIN CAPITAL LETTER A WITH DIAERESIS */
- { 'A', 0x030a, 0xc5}, /* LATIN CAPITAL LETTER A WITH RING ABOVE */
- /* no need for 0xc6 LATIN CAPITAL LETTER AE */
- { 'C', 0x0327, 0xc7}, /* LATIN CAPITAL LETTER C WITH CEDILLA */
- { 'E', 0x0300, 0xc8}, /* LATIN CAPITAL LETTER E WITH GRAVE */
- { 'E', 0x0301, 0xc9}, /* LATIN CAPITAL LETTER E WITH ACUTE */
- { 'E', 0x0302, 0xca}, /* LATIN CAPITAL LETTER E WITH CIRCUMFLEX */
- { 'E', 0x0308, 0xcb}, /* LATIN CAPITAL LETTER E WITH DIAERESIS */
- { 'I', 0x0300, 0xcc}, /* LATIN CAPITAL LETTER I WITH GRAVE */
- { 'I', 0x0301, 0xcd}, /* LATIN CAPITAL LETTER I WITH ACUTE */
- { 'I', 0x0302, 0xce}, /* LATIN CAPITAL LETTER I WITH CIRCUMFLEX */
- { 'I', 0x0308, 0xcf}, /* LATIN CAPITAL LETTER I WITH DIAERESIS */
- { 'N', 0x0303, 0xd1}, /* LATIN CAPITAL LETTER N WITH TILDE */
- { 'O', 0x0300, 0xd2}, /* LATIN CAPITAL LETTER O WITH GRAVE */
- { 'O', 0x0301, 0xd3}, /* LATIN CAPITAL LETTER O WITH ACUTE */
- { 'O', 0x0302, 0xd4}, /* LATIN CAPITAL LETTER O WITH CIRCUMFLEX */
- { 'O', 0x0303, 0xd5}, /* LATIN CAPITAL LETTER O WITH TILDE */
- { 'O', 0x0308, 0xd6}, /* LATIN CAPITAL LETTER O WITH DIAERESIS */
- /* omitted: 0xd7 MULTIPLICATION SIGN */
- /* omitted: 0xd8 LATIN CAPITAL LETTER O WITH STROKE */
- { 'U', 0x0300, 0xd9}, /* LATIN CAPITAL LETTER U WITH GRAVE */
- { 'U', 0x0301, 0xda}, /* LATIN CAPITAL LETTER U WITH ACUTE */
- { 'U', 0x0302, 0xdb}, /* LATIN CAPITAL LETTER U WITH CIRCUMFLEX */
- { 'U', 0x0308, 0xdc}, /* LATIN CAPITAL LETTER U WITH DIAERESIS */
- { 'Y', 0x0301, 0xdd}, /* LATIN CAPITAL LETTER Y WITH ACUTE */
- /* omitted: 0xde LATIN CAPITAL LETTER THORN */
- /* omitted: 0xdf LATIN SMALL LETTER SHARP S */
- { 'a', 0x0300, 0xe0}, /* LATIN SMALL LETTER A WITH GRAVE */
- { 'a', 0x0301, 0xe1}, /* LATIN SMALL LETTER A WITH ACUTE */
- { 'a', 0x0302, 0xe2}, /* LATIN SMALL LETTER A WITH CIRCUMFLEX */
- { 'a', 0x0303, 0xe3}, /* LATIN SMALL LETTER A WITH TILDE */
- { 'a', 0x0308, 0xe4}, /* LATIN SMALL LETTER A WITH DIAERESIS */
- { 'a', 0x030a, 0xe5}, /* LATIN SMALL LETTER A WITH RING ABOVE */
- /* omitted: 0xe6 LATIN SMALL LETTER AE */
- { 'c', 0x0327, 0xe7}, /* LATIN SMALL LETTER C WITH CEDILLA */
- { 'e', 0x0300, 0xe8}, /* LATIN SMALL LETTER E WITH GRAVE */
- { 'e', 0x0301, 0xe9}, /* LATIN SMALL LETTER E WITH ACUTE */
- { 'e', 0x0302, 0xea}, /* LATIN SMALL LETTER E WITH CIRCUMFLEX */
- { 'e', 0x0308, 0xeb}, /* LATIN SMALL LETTER E WITH DIAERESIS */
- { 'i', 0x0300, 0xec}, /* LATIN SMALL LETTER I WITH GRAVE */
- { 'i', 0x0301, 0xed}, /* LATIN SMALL LETTER I WITH ACUTE */
- { 'i', 0x0302, 0xee}, /* LATIN SMALL LETTER I WITH CIRCUMFLEX */
- { 'i', 0x0308, 0xef}, /* LATIN SMALL LETTER I WITH DIAERESIS */
- /* omitted: 0xf0 LATIN SMALL LETTER ETH */
- { 'n', 0x0303, 0xf1}, /* LATIN SMALL LETTER N WITH TILDE */
- { 'o', 0x0300, 0xf2}, /* LATIN SMALL LETTER O WITH GRAVE */
- { 'o', 0x0301, 0xf3}, /* LATIN SMALL LETTER O WITH ACUTE */
- { 'o', 0x0302, 0xf4}, /* LATIN SMALL LETTER O WITH CIRCUMFLEX */
- { 'o', 0x0303, 0xf5}, /* LATIN SMALL LETTER O WITH TILDE */
- { 'o', 0x0308, 0xf6}, /* LATIN SMALL LETTER O WITH DIAERESIS */
- /* omitted: 0xf7 DIVISION SIGN */
- /* omitted: 0xf8 LATIN SMALL LETTER O WITH STROKE */
- { 'u', 0x0300, 0xf9}, /* LATIN SMALL LETTER U WITH GRAVE */
- { 'u', 0x0301, 0xfa}, /* LATIN SMALL LETTER U WITH ACUTE */
- { 'u', 0x0302, 0xfb}, /* LATIN SMALL LETTER U WITH CIRCUMFLEX */
- { 'u', 0x0308, 0xfc}, /* LATIN SMALL LETTER U WITH DIAERESIS */
- { 'y', 0x0301, 0xfd}, /* LATIN SMALL LETTER Y WITH ACUTE */
- /* omitted: 0xfe LATIN SMALL LETTER THORN */
- { 'y', 0x0308, 0xff}, /* LATIN SMALL LETTER Y WITH DIAERESIS */
-
- { 0, 0, 0}
- };
unsigned char *outp = (unsigned char *) *outbuf;
if (cd->compose_char)
if (!y)
return 0;
- byte = (y>>16) & 0xff;
+ byte = (unsigned char )((y>>16) & 0xff);
if (byte)
out_buf[out_no++] = byte;
- byte = (y>>8) & 0xff;
+ byte = (unsigned char)((y>>8) & 0xff);
if (byte)
out_buf[out_no++] = byte;
- byte = y & 0xff;
+ byte = (unsigned char )(y & 0xff);
if (byte)
out_buf[out_no++] = byte;
for (i = 0; i < cd->write_marc8_comb_no; i++)
{
- byte = cd->write_marc8_comb_ch[i];
+ /* all MARC-8 combined characters are simple bytes */
+ byte = (unsigned char )(cd->write_marc8_comb_ch[i]);
if (byte == 0xEB)
second_half = 0xEC;
else if (byte == 0xFA)
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;
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 = yaz_read_UCS4LE;
else if (!yaz_matchstr(fromcode, "MARC8"))
cd->read_handle = yaz_read_marc8;
+ else if (!yaz_matchstr(fromcode, "MARC8s"))
+ cd->read_handle = yaz_read_marc8s;
#if HAVE_WCHAR_H
else if (!yaz_matchstr(fromcode, "WCHAR_T"))
cd->read_handle = yaz_read_wchar_t;
cd->write_handle = yaz_write_UCS4LE;
else if (!yaz_matchstr(tocode, "MARC8"))
cd->write_handle = yaz_write_marc8;
+ else if (!yaz_matchstr(tocode, "MARC8s"))
+ cd->write_handle = yaz_write_marc8;
#if HAVE_WCHAR_H
else if (!yaz_matchstr(tocode, "WCHAR_T"))
cd->write_handle = yaz_write_wchar_t;
projects / yaz-moved-to-github.git / blobdiff