ruby--ruby/enc/trans/iso2022.trans

174 lines
4.1 KiB
Plaintext

#include "transcode_data.h"
<%
map = {}
map["1b2842"] = :func_so # designate US-ASCII to G0. "ESC ( B"
map["1b284a"] = :func_so # designate JIS X 0201 latin to G0. "ESC ( J"
map["1b2440"] = :func_so # designate JIS X 0208 1978 to G0. "ESC $ @"
map["1b2442"] = :func_so # designate JIS X 0208 1983 to G0. "ESC $ B"
map["{00-0d,10-1a,1c-7f}"] = :func_si
map_jisx0208_rest = {}
map_jisx0208_rest["{21-7e}"] = :func_so
transcode_generate_node(ActionMap.parse(map), "iso2022jp_to_eucjp")
transcode_generate_node(ActionMap.parse(map_jisx0208_rest), "iso2022jp_to_eucjp_jisx0208_rest")
map_eucjp = {
"{0e,0f,1b}" => :undef,
"{00-0d,10-1a,1c-7f}" => :func_so,
"{a1-fe}{a1-fe}" => :func_so,
"8e{a1-fe}" => :undef,
"8f{a1-fe}{a1-fe}" => :undef,
}
transcode_generate_node(ActionMap.parse(map_eucjp), "eucjp_to_iso2022jp")
%>
<%= transcode_generated_code %>
#define G0_ASCII 0
#define G0_JISX0208 1
static int
iso2022jp_init(void *statep)
{
unsigned char *sp = statep;
*sp = G0_ASCII;
return 0;
}
static VALUE
fun_si_iso2022jp_to_eucjp(void *statep, const unsigned char *s, size_t l)
{
unsigned char *sp = statep;
if (*sp == G0_ASCII)
return (VALUE)NOMAP;
else if (0x21 <= s[0] && s[0] <= 0x7e)
return (VALUE)iso2022jp_to_eucjp_jisx0208_rest;
else
return (VALUE)INVALID;
}
static int
fun_so_iso2022jp_to_eucjp(void *statep, const unsigned char *s, size_t l, unsigned char* o)
{
unsigned char *sp = statep;
if (s[0] == 0x1b) {
if (s[1] == '(') {
switch (s[l-1]) {
case 'B':
case 'J':
*sp = G0_ASCII;
break;
}
}
else {
switch (s[l-1]) {
case '@':
case 'B':
*sp = G0_JISX0208;
break;
}
}
return 0;
}
else {
o[0] = s[0] | 0x80;
o[1] = s[1] | 0x80;
return 2;
}
}
static const rb_transcoder
rb_ISO_2022_JP_to_EUC_JP = {
"ISO-2022-JP", "EUC-JP", iso2022jp_to_eucjp,
TRANSCODE_TABLE_INFO,
1, /* input_unit_length */
3, /* max_input */
3, /* max_output */
stateful_decoder, /* stateful_type */
1, iso2022jp_init, iso2022jp_init, /* state_size, state_init, state_fini */
NULL, fun_si_iso2022jp_to_eucjp, NULL, fun_so_iso2022jp_to_eucjp
};
static int
fun_so_eucjp_to_iso2022jp(void *statep, const unsigned char *s, size_t l, unsigned char *o)
{
unsigned char *sp = statep;
unsigned char *output0 = o;
if (*sp != (l == 1 ? G0_ASCII : G0_JISX0208)) {
if (l == 1) {
*o++ = 0x1b;
*o++ = '(';
*o++ = 'B';
*sp = G0_ASCII;
}
else {
*o++ = 0x1b;
*o++ = '$';
*o++ = 'B';
*sp = G0_JISX0208; /* JIS X 0208 1983 */
}
}
if (l == 1) {
*o++ = s[0] & 0x7f;
}
else {
*o++ = s[0] & 0x7f;
*o++ = s[1] & 0x7f;
}
return o - output0;
}
static int
iso2022jp_reset_sequence_size(void *statep)
{
unsigned char *sp = statep;
if (*sp == G0_JISX0208)
return 3;
return 0;
}
static int
finish_eucjp_to_iso2022jp(void *statep, unsigned char *o)
{
unsigned char *sp = statep;
unsigned char *output0 = o;
if (*sp == G0_ASCII)
return 0;
*o++ = 0x1b;
*o++ = '(';
*o++ = 'B';
*sp = G0_ASCII;
return o - output0;
}
static const rb_transcoder
rb_EUC_JP_to_ISO_2022_JP = {
"EUC-JP", "ISO-2022-JP", eucjp_to_iso2022jp,
TRANSCODE_TABLE_INFO,
1, /* input_unit_length */
3, /* max_input */
5, /* max_output */
stateful_encoder, /* stateful_type */
1, iso2022jp_init, iso2022jp_init, /* state_size, state_init, state_fini */
NULL, NULL, NULL, fun_so_eucjp_to_iso2022jp,
finish_eucjp_to_iso2022jp,
iso2022jp_reset_sequence_size, finish_eucjp_to_iso2022jp
};
void
Init_iso2022(void)
{
rb_register_transcoder(&rb_ISO_2022_JP_to_EUC_JP);
rb_register_transcoder(&rb_EUC_JP_to_ISO_2022_JP);
}