ruby/enc/trans/newline.trans

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

156 строки
3.6 KiB
Plaintext
Исходник Обычный вид История

#include "transcode_data.h"
<%
map_normalize = {}
map_normalize["{00-ff}"] = :func_so
transcode_generate_node(ActionMap.parse(map_normalize), "universal_newline")
map_crlf = {}
map_crlf["{00-09,0b-ff}"] = :nomap
map_crlf["0a"] = "0d0a"
transcode_generate_node(ActionMap.parse(map_crlf), "crlf_newline")
map_cr = {}
map_cr["{00-09,0b-ff}"] = :nomap
map_cr["0a"] = "0d"
transcode_generate_node(ActionMap.parse(map_cr), "cr_newline")
map_normalize = {}
map_normalize["{00-ff}"] = :func_so
transcode_generate_node(ActionMap.parse(map_normalize), "lf_newline")
%>
<%= transcode_generated_code %>
#define lf_newline universal_newline
#define STATE (sp[0])
* transcode_data.h (rb_transcoding): remove stateful field. add state field. (TRANSCODING_STATE): defined. (rb_transcoder): add fields: state_size, state_init_func, state_fini_func. change rb_transcoding* argument to void*. * transcode.c (transcode_restartable0): use TRANSCODING_STATE for first arguments of transcoder functions. (rb_transcoding_open_by_transcoder): initialize state field. (rb_transcoding_close): finalize state field. * tool/transcode-tblgen.rb: provide state size/init/fini. * enc/trans/newline.trans (universal_newline_init): defined. (fun_so_universal_newline): take void* as a state pointer. (rb_universal_newline): provide state size/init/fini. (rb_crlf_newline): ditto. (rb_cr_newline): ditto. * enc/trans/iso2022.trans (iso2022jp_init): defined. (fun_si_iso2022jp_to_eucjp): take void* as a state pointer. (fun_so_iso2022jp_to_eucjp): ditto. (fun_so_eucjp_to_iso2022jp): ditto. (iso2022jp_reset_sequence_size): ditto. (finish_eucjp_to_iso2022jp): ditto. (rb_ISO_2022_JP_to_EUC_JP): provide state size/init/fini. (rb_EUC_JP_to_ISO_2022_JP): ditto. * enc/trans/utf_16_32.trans (fun_so_from_utf_16be): take void* as a state pointer. (fun_so_to_utf_16be): ditto. (fun_so_from_utf_16le): ditto. (fun_so_to_utf_16le): ditto. (fun_so_from_utf_32be): ditto. (fun_so_to_utf_32be): ditto. (fun_so_from_utf_32le): ditto. (fun_so_to_utf_32le): ditto. (rb_from_UTF_16BE): provide state size/init/fini. (rb_to_UTF_16BE): ditto. (rb_from_UTF_16LE): ditto. (rb_to_UTF_16LE): ditto. (rb_from_UTF_32BE): ditto. (rb_to_UTF_32BE): ditto. (rb_from_UTF_32LE): ditto. (rb_to_UTF_32LE): ditto. * enc/trans/japanese.trans (fun_so_eucjp2sjis): take void* as a state pointer. (fun_so_sjis2eucjp): ditto. (rb_eucjp2sjis): provide state size/init/fini. (rb_sjis2eucjp): provide state size/init/fini. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19096 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-03 18:12:06 +04:00
#define NORMAL 0
#define JUST_AFTER_CR 1
/* no way to access this information, yet. */
#define NEWLINES_MET (sp[1])
#define MET_LF 0x01
#define MET_CRLF 0x02
#define MET_CR 0x04
* transcode_data.h (rb_transcoding): remove stateful field. add state field. (TRANSCODING_STATE): defined. (rb_transcoder): add fields: state_size, state_init_func, state_fini_func. change rb_transcoding* argument to void*. * transcode.c (transcode_restartable0): use TRANSCODING_STATE for first arguments of transcoder functions. (rb_transcoding_open_by_transcoder): initialize state field. (rb_transcoding_close): finalize state field. * tool/transcode-tblgen.rb: provide state size/init/fini. * enc/trans/newline.trans (universal_newline_init): defined. (fun_so_universal_newline): take void* as a state pointer. (rb_universal_newline): provide state size/init/fini. (rb_crlf_newline): ditto. (rb_cr_newline): ditto. * enc/trans/iso2022.trans (iso2022jp_init): defined. (fun_si_iso2022jp_to_eucjp): take void* as a state pointer. (fun_so_iso2022jp_to_eucjp): ditto. (fun_so_eucjp_to_iso2022jp): ditto. (iso2022jp_reset_sequence_size): ditto. (finish_eucjp_to_iso2022jp): ditto. (rb_ISO_2022_JP_to_EUC_JP): provide state size/init/fini. (rb_EUC_JP_to_ISO_2022_JP): ditto. * enc/trans/utf_16_32.trans (fun_so_from_utf_16be): take void* as a state pointer. (fun_so_to_utf_16be): ditto. (fun_so_from_utf_16le): ditto. (fun_so_to_utf_16le): ditto. (fun_so_from_utf_32be): ditto. (fun_so_to_utf_32be): ditto. (fun_so_from_utf_32le): ditto. (fun_so_to_utf_32le): ditto. (rb_from_UTF_16BE): provide state size/init/fini. (rb_to_UTF_16BE): ditto. (rb_from_UTF_16LE): ditto. (rb_to_UTF_16LE): ditto. (rb_from_UTF_32BE): ditto. (rb_to_UTF_32BE): ditto. (rb_from_UTF_32LE): ditto. (rb_to_UTF_32LE): ditto. * enc/trans/japanese.trans (fun_so_eucjp2sjis): take void* as a state pointer. (fun_so_sjis2eucjp): ditto. (rb_eucjp2sjis): provide state size/init/fini. (rb_sjis2eucjp): provide state size/init/fini. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19096 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-03 18:12:06 +04:00
static int
universal_newline_init(void *statep)
{
unsigned char *sp = statep;
STATE = NORMAL;
NEWLINES_MET = 0;
* transcode_data.h (rb_transcoding): remove stateful field. add state field. (TRANSCODING_STATE): defined. (rb_transcoder): add fields: state_size, state_init_func, state_fini_func. change rb_transcoding* argument to void*. * transcode.c (transcode_restartable0): use TRANSCODING_STATE for first arguments of transcoder functions. (rb_transcoding_open_by_transcoder): initialize state field. (rb_transcoding_close): finalize state field. * tool/transcode-tblgen.rb: provide state size/init/fini. * enc/trans/newline.trans (universal_newline_init): defined. (fun_so_universal_newline): take void* as a state pointer. (rb_universal_newline): provide state size/init/fini. (rb_crlf_newline): ditto. (rb_cr_newline): ditto. * enc/trans/iso2022.trans (iso2022jp_init): defined. (fun_si_iso2022jp_to_eucjp): take void* as a state pointer. (fun_so_iso2022jp_to_eucjp): ditto. (fun_so_eucjp_to_iso2022jp): ditto. (iso2022jp_reset_sequence_size): ditto. (finish_eucjp_to_iso2022jp): ditto. (rb_ISO_2022_JP_to_EUC_JP): provide state size/init/fini. (rb_EUC_JP_to_ISO_2022_JP): ditto. * enc/trans/utf_16_32.trans (fun_so_from_utf_16be): take void* as a state pointer. (fun_so_to_utf_16be): ditto. (fun_so_from_utf_16le): ditto. (fun_so_to_utf_16le): ditto. (fun_so_from_utf_32be): ditto. (fun_so_to_utf_32be): ditto. (fun_so_from_utf_32le): ditto. (fun_so_to_utf_32le): ditto. (rb_from_UTF_16BE): provide state size/init/fini. (rb_to_UTF_16BE): ditto. (rb_from_UTF_16LE): ditto. (rb_to_UTF_16LE): ditto. (rb_from_UTF_32BE): ditto. (rb_to_UTF_32BE): ditto. (rb_from_UTF_32LE): ditto. (rb_to_UTF_32LE): ditto. * enc/trans/japanese.trans (fun_so_eucjp2sjis): take void* as a state pointer. (fun_so_sjis2eucjp): ditto. (rb_eucjp2sjis): provide state size/init/fini. (rb_sjis2eucjp): provide state size/init/fini. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19096 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-03 18:12:06 +04:00
return 0;
}
static ssize_t
fun_so_universal_newline(void *statep, const unsigned char *s, size_t l, unsigned char *o, size_t osize)
{
* transcode_data.h (rb_transcoding): remove stateful field. add state field. (TRANSCODING_STATE): defined. (rb_transcoder): add fields: state_size, state_init_func, state_fini_func. change rb_transcoding* argument to void*. * transcode.c (transcode_restartable0): use TRANSCODING_STATE for first arguments of transcoder functions. (rb_transcoding_open_by_transcoder): initialize state field. (rb_transcoding_close): finalize state field. * tool/transcode-tblgen.rb: provide state size/init/fini. * enc/trans/newline.trans (universal_newline_init): defined. (fun_so_universal_newline): take void* as a state pointer. (rb_universal_newline): provide state size/init/fini. (rb_crlf_newline): ditto. (rb_cr_newline): ditto. * enc/trans/iso2022.trans (iso2022jp_init): defined. (fun_si_iso2022jp_to_eucjp): take void* as a state pointer. (fun_so_iso2022jp_to_eucjp): ditto. (fun_so_eucjp_to_iso2022jp): ditto. (iso2022jp_reset_sequence_size): ditto. (finish_eucjp_to_iso2022jp): ditto. (rb_ISO_2022_JP_to_EUC_JP): provide state size/init/fini. (rb_EUC_JP_to_ISO_2022_JP): ditto. * enc/trans/utf_16_32.trans (fun_so_from_utf_16be): take void* as a state pointer. (fun_so_to_utf_16be): ditto. (fun_so_from_utf_16le): ditto. (fun_so_to_utf_16le): ditto. (fun_so_from_utf_32be): ditto. (fun_so_to_utf_32be): ditto. (fun_so_from_utf_32le): ditto. (fun_so_to_utf_32le): ditto. (rb_from_UTF_16BE): provide state size/init/fini. (rb_to_UTF_16BE): ditto. (rb_from_UTF_16LE): ditto. (rb_to_UTF_16LE): ditto. (rb_from_UTF_32BE): ditto. (rb_to_UTF_32BE): ditto. (rb_from_UTF_32LE): ditto. (rb_to_UTF_32LE): ditto. * enc/trans/japanese.trans (fun_so_eucjp2sjis): take void* as a state pointer. (fun_so_sjis2eucjp): ditto. (rb_eucjp2sjis): provide state size/init/fini. (rb_sjis2eucjp): provide state size/init/fini. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19096 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-03 18:12:06 +04:00
unsigned char *sp = statep;
int len;
if (s[0] == '\n') {
if (STATE == NORMAL) {
NEWLINES_MET |= MET_LF;
}
* transcode_data.h (rb_transcoding): remove stateful field. add state field. (TRANSCODING_STATE): defined. (rb_transcoder): add fields: state_size, state_init_func, state_fini_func. change rb_transcoding* argument to void*. * transcode.c (transcode_restartable0): use TRANSCODING_STATE for first arguments of transcoder functions. (rb_transcoding_open_by_transcoder): initialize state field. (rb_transcoding_close): finalize state field. * tool/transcode-tblgen.rb: provide state size/init/fini. * enc/trans/newline.trans (universal_newline_init): defined. (fun_so_universal_newline): take void* as a state pointer. (rb_universal_newline): provide state size/init/fini. (rb_crlf_newline): ditto. (rb_cr_newline): ditto. * enc/trans/iso2022.trans (iso2022jp_init): defined. (fun_si_iso2022jp_to_eucjp): take void* as a state pointer. (fun_so_iso2022jp_to_eucjp): ditto. (fun_so_eucjp_to_iso2022jp): ditto. (iso2022jp_reset_sequence_size): ditto. (finish_eucjp_to_iso2022jp): ditto. (rb_ISO_2022_JP_to_EUC_JP): provide state size/init/fini. (rb_EUC_JP_to_ISO_2022_JP): ditto. * enc/trans/utf_16_32.trans (fun_so_from_utf_16be): take void* as a state pointer. (fun_so_to_utf_16be): ditto. (fun_so_from_utf_16le): ditto. (fun_so_to_utf_16le): ditto. (fun_so_from_utf_32be): ditto. (fun_so_to_utf_32be): ditto. (fun_so_from_utf_32le): ditto. (fun_so_to_utf_32le): ditto. (rb_from_UTF_16BE): provide state size/init/fini. (rb_to_UTF_16BE): ditto. (rb_from_UTF_16LE): ditto. (rb_to_UTF_16LE): ditto. (rb_from_UTF_32BE): ditto. (rb_to_UTF_32BE): ditto. (rb_from_UTF_32LE): ditto. (rb_to_UTF_32LE): ditto. * enc/trans/japanese.trans (fun_so_eucjp2sjis): take void* as a state pointer. (fun_so_sjis2eucjp): ditto. (rb_eucjp2sjis): provide state size/init/fini. (rb_sjis2eucjp): provide state size/init/fini. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19096 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-03 18:12:06 +04:00
else { /* JUST_AFTER_CR */
NEWLINES_MET |= MET_CRLF;
}
o[0] = '\n';
len = 1;
STATE = NORMAL;
}
else {
len = 0;
if (STATE == JUST_AFTER_CR) {
o[0] = '\n';
len = 1;
NEWLINES_MET |= MET_CR;
}
if (s[0] == '\r') {
STATE = JUST_AFTER_CR;
}
else {
o[len++] = s[0];
STATE = NORMAL;
}
}
return len;
}
static ssize_t
universal_newline_finish(void *statep, unsigned char *o, size_t osize)
{
unsigned char *sp = statep;
int len = 0;
if (STATE == JUST_AFTER_CR) {
o[0] = '\n';
len = 1;
NEWLINES_MET |= MET_CR;
}
STATE = NORMAL;
return len;
}
static const rb_transcoder
rb_universal_newline = {
"", "universal_newline", universal_newline,
TRANSCODE_TABLE_INFO,
1, /* input_unit_length */
1, /* max_input */
2, /* max_output */
asciicompat_converter, /* asciicompat_type */
2, universal_newline_init, universal_newline_init, /* state_size, state_init, state_fini */
0, 0, 0, fun_so_universal_newline,
universal_newline_finish
};
static const rb_transcoder
rb_crlf_newline = {
"", "crlf_newline", crlf_newline,
TRANSCODE_TABLE_INFO,
1, /* input_unit_length */
1, /* max_input */
2, /* max_output */
asciicompat_converter, /* asciicompat_type */
0, 0, 0, /* state_size, state_init, state_fini */
0, 0, 0, 0
};
static const rb_transcoder
rb_cr_newline = {
"", "cr_newline", cr_newline,
TRANSCODE_TABLE_INFO,
1, /* input_unit_length */
1, /* max_input */
1, /* max_output */
asciicompat_converter, /* asciicompat_type */
0, 0, 0, /* state_size, state_init, state_fini */
0, 0, 0, 0
};
static const rb_transcoder
rb_lf_newline = {
"", "lf_newline", lf_newline,
TRANSCODE_TABLE_INFO,
1, /* input_unit_length */
1, /* max_input */
2, /* max_output */
asciicompat_converter, /* asciicompat_type */
2, universal_newline_init, universal_newline_init, /* state_size, state_init, state_fini */
0, 0, 0, fun_so_universal_newline,
universal_newline_finish
};
void
Init_newline(void)
{
rb_register_transcoder(&rb_universal_newline);
rb_register_transcoder(&rb_crlf_newline);
rb_register_transcoder(&rb_cr_newline);
rb_register_transcoder(&rb_lf_newline);
}