/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2008 Index Data
+ * Copyright (C) Index Data
* See the file LICENSE for details.
*/
/**
* \file
- * \brief MARC-8 decoding
+ * \brief ISO 5426 decoding
*
* MARC-8 reference:
- * http://www.loc.gov/marc/specifications/speccharmarc8.html
+ * http://www.loc.gov/marc/specifications/specchariso8.html
+ *
+ * ISO 5426 reference (in German)
+ * Zeichenkonkordanz MAB2-Zeichensatz - ISO/IEC 10646 / Unicode
+ * http://www.d-nb.de/standardisierung/pdf/mab_unic.pdf
*/
#if HAVE_CONFIG_H
#include <assert.h>
#include <errno.h>
#include <string.h>
-#include <ctype.h>
#include <yaz/xmalloc.h>
#include "iconv-p.h"
size_t comb_no_read[8];
};
-yaz_conv_func_t yaz_marc8_42_conv;
-yaz_conv_func_t yaz_marc8_45_conv;
-yaz_conv_func_t yaz_marc8_67_conv;
-yaz_conv_func_t yaz_marc8_62_conv;
-yaz_conv_func_t yaz_marc8_70_conv;
-yaz_conv_func_t yaz_marc8_32_conv;
-yaz_conv_func_t yaz_marc8_4E_conv;
-yaz_conv_func_t yaz_marc8_51_conv;
-yaz_conv_func_t yaz_marc8_33_conv;
-yaz_conv_func_t yaz_marc8_34_conv;
-yaz_conv_func_t yaz_marc8_53_conv;
-yaz_conv_func_t yaz_marc8_31_conv;
-
-
-static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd,
+yaz_conv_func_t yaz_iso5426_42_conv;
+yaz_conv_func_t yaz_iso5426_45_conv;
+yaz_conv_func_t yaz_iso5426_67_conv;
+yaz_conv_func_t yaz_iso5426_62_conv;
+yaz_conv_func_t yaz_iso5426_70_conv;
+yaz_conv_func_t yaz_iso5426_32_conv;
+yaz_conv_func_t yaz_iso5426_4E_conv;
+yaz_conv_func_t yaz_iso5426_51_conv;
+yaz_conv_func_t yaz_iso5426_33_conv;
+yaz_conv_func_t yaz_iso5426_34_conv;
+yaz_conv_func_t yaz_iso5426_53_conv;
+yaz_conv_func_t yaz_iso5426_31_conv;
+
+
+static unsigned long yaz_read_iso5426_comb(yaz_iconv_t cd,
struct decoder_data *data,
unsigned char *inp,
size_t inbytesleft, size_t *no_read,
int *comb);
-static unsigned long read_marc8(yaz_iconv_t cd, yaz_iconv_decoder_t d,
+static unsigned long read_iso5426(yaz_iconv_t cd, yaz_iconv_decoder_t d,
unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
*no_read = data->comb_no_read[data->comb_offset];
x = data->comb_x[data->comb_offset];
- /* special case for double-diacritic combining characters,
+ /* special case for double-diacritic combining characters,
INVERTED BREVE and DOUBLE TILDE.
We'll increment the no_read counter by 1, since we want to skip over
the processing of the closing ligature character
*/
/* this code is no longer necessary.. our handlers code in
- yaz_marc8_?_conv (generated by charconv.tcl) now returns
+ yaz_iso5426_?_conv (generated by charconv.tcl) now returns
0 and no_read=1 when a sequence does not match the input.
The SECOND HALFs in codetables.xml produces a non-existant
entry in the conversion trie.. Hence when met, the input byte is
*no_read = 0;
break;
}
- x = yaz_read_marc8_comb(cd, data, inp, inbytesleft, no_read, &comb);
+ x = yaz_read_iso5426_comb(cd, data, inp, inbytesleft, no_read, &comb);
if (!comb || !x)
break;
data->comb_x[data->comb_size] = x;
return x;
}
-static unsigned long read_marc8s(yaz_iconv_t cd, yaz_iconv_decoder_t d,
- unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
-{
- struct decoder_data *data = (struct decoder_data *) d->data;
- unsigned long x = read_marc8(cd, d, inp, inbytesleft, no_read);
- if (x && data->comb_size == 1)
- {
- if (yaz_iso_8859_1_lookup_x12(x, data->comb_x[0], &x))
- {
- *no_read += data->comb_no_read[0];
- data->comb_size = 0;
- }
- }
- return x;
-}
-
-static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd,
+static unsigned long yaz_read_iso5426_comb(yaz_iconv_t cd,
struct decoder_data *data,
unsigned char *inp,
size_t inbytesleft, size_t *no_read,
{
case 'B': /* Basic ASCII */
case 's': /* ASCII */
- x = yaz_marc8_42_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
+ x = yaz_iso5426_42_conv(inp, inbytesleft, &no_read_sub, comb,
+ 127, 0);
break;
case 'E': /* ANSEL */
- x = yaz_marc8_45_conv(inp, inbytesleft, &no_read_sub, comb, 127, 128);
- break;
- case 'g': /* Greek */
- x = yaz_marc8_67_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case 'b': /* Subscripts */
- x = yaz_marc8_62_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case 'p': /* Superscripts */
- x = yaz_marc8_70_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case '2': /* Basic Hebrew */
- x = yaz_marc8_32_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case 'N': /* Basic Cyrillic */
- x = yaz_marc8_4E_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case 'Q': /* Extended Cyrillic */
- x = yaz_marc8_51_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case '3': /* Basic Arabic */
- x = yaz_marc8_33_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case '4': /* Extended Arabic */
- x = yaz_marc8_34_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case 'S': /* Greek */
- x = yaz_marc8_53_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
- break;
- case '1': /* Chinese, Japanese, Korean (EACC) */
- x = yaz_marc8_31_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
+ x = yaz_iso5426_45_conv(inp, inbytesleft, &no_read_sub, comb,
+ 127, 128);
break;
default:
*no_read = 0;
}
-static size_t init_marc8(yaz_iconv_t cd, yaz_iconv_decoder_t d,
+static size_t init_iso5426(yaz_iconv_t cd, yaz_iconv_decoder_t d,
unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
return 0;
}
-void destroy_marc8(yaz_iconv_decoder_t d)
+void destroy_iso5426(yaz_iconv_decoder_t d)
{
struct decoder_data *data = (struct decoder_data *) d->data;
xfree(data);
}
-yaz_iconv_decoder_t yaz_marc8_decoder(const char *fromcode,
+yaz_iconv_decoder_t yaz_iso5426_decoder(const char *fromcode,
yaz_iconv_decoder_t d)
{
- if (!yaz_matchstr(fromcode, "MARC8") || !yaz_matchstr(fromcode, "ANSEL"))
- d->read_handle = read_marc8;
- else if (!yaz_matchstr(fromcode, "ISO5426"))
- d->read_handle = read_marc8;
- else if (!yaz_matchstr(fromcode, "MARC8s"))
- d->read_handle = read_marc8s;
+ if (!yaz_matchstr(fromcode, "ISO5426"))
+ d->read_handle = read_iso5426;
else
return 0;
{
struct decoder_data *data = (struct decoder_data *)
xmalloc(sizeof(*data));
d->data = data;
- d->init_handle = init_marc8;
- d->destroy_handle = destroy_marc8;
+ d->init_handle = init_iso5426;
+ d->destroy_handle = destroy_iso5426;
}
return d;
}
/*
* Local variables:
* c-basic-offset: 4
+ * c-file-style: "Stroustrup"
* indent-tabs-mode: nil
* End:
* vim: shiftwidth=4 tabstop=8 expandtab
*/
+
projects / yaz-moved-to-github.git / blobdiff