Moriyoshi Koizumi
moriy****@users*****
2002年 12月 21日 (土) 01:46:36 JST
moriyoshi 02/12/21 01:46:36
Modified: filters mbfilter_ascii.c mbfilter_ascii.h mbfilter_big5.c
mbfilter_big5.h mbfilter_cp1251.c mbfilter_cp1251.h
mbfilter_cp1252.c mbfilter_cp1252.h
mbfilter_cp866.c mbfilter_cp866.h mbfilter_cp932.c
mbfilter_cp932.h mbfilter_cp936.c mbfilter_cp936.h
mbfilter_euc_cn.c mbfilter_euc_cn.h
mbfilter_euc_jp.c mbfilter_euc_jp.h
mbfilter_euc_jp_win.c mbfilter_euc_jp_win.h
mbfilter_euc_kr.c mbfilter_euc_kr.h
mbfilter_euc_tw.c mbfilter_euc_tw.h mbfilter_hz.c
mbfilter_hz.h mbfilter_iso2022_kr.c
mbfilter_iso2022_kr.h mbfilter_iso8859_1.c
mbfilter_iso8859_1.h mbfilter_iso8859_10.c
mbfilter_iso8859_10.h mbfilter_iso8859_13.c
mbfilter_iso8859_13.h mbfilter_iso8859_14.c
mbfilter_iso8859_14.h mbfilter_iso8859_15.c
mbfilter_iso8859_15.h mbfilter_iso8859_2.c
mbfilter_iso8859_2.h mbfilter_iso8859_3.c
mbfilter_iso8859_3.h mbfilter_iso8859_4.c
mbfilter_iso8859_4.h mbfilter_iso8859_5.c
mbfilter_iso8859_5.h mbfilter_iso8859_6.c
mbfilter_iso8859_6.h mbfilter_iso8859_7.c
mbfilter_iso8859_7.h mbfilter_iso8859_8.c
mbfilter_iso8859_8.h mbfilter_iso8859_9.c
mbfilter_iso8859_9.h mbfilter_jis.c mbfilter_jis.h
mbfilter_koi8r.c mbfilter_koi8r.h mbfilter_sjis.c
mbfilter_sjis.h mbfilter_uhc.c mbfilter_uhc.h
mbfilter_utf7.c mbfilter_utf7.h mbfilter_utf8.c
mbfilter_utf8.h
mbfl mbfilter.c mbfilter.h
Log:
Externalised identification filter tables
Revision Changes Path
1.3 +21 -0 libmbfl/filters/mbfilter_ascii.c
Index: mbfilter_ascii.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_ascii.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_ascii.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_ascii.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -35,6 +35,8 @@
#include "mbfilter.h"
#include "mbfilter_ascii.h"
+static int mbfl_filt_ident_ascii(int c, mbfl_identify_filter *filter);
+
static const char *mbfl_encoding_ascii_aliases[] = {"ANSI_X3.4-1968", "iso-ir-6", "ANSI_X3.4-1986", "ISO_646.irv:1991", "US-ASCII", "ISO646-US", "us", "IBM367", "cp367", "csASCII", NULL};
const mbfl_encoding mbfl_encoding_ascii = {
@@ -46,6 +48,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_ascii = {
+ mbfl_no_encoding_ascii,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_ascii
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -64,4 +74,15 @@
return c;
}
+static int mbfl_filt_ident_ascii(int c, mbfl_identify_filter *filter)
+{
+ if (c >= 0x20 && c < 0x80) {
+ ;
+ } else if (c == 0x0d || c == 0x0a || c == 0x09 || c == 0) { /* CR or LF or HTAB or null */
+ ;
+ } else {
+ filter->flag = 1;
+ }
+ return c;
+}
1.2 +2 -1 libmbfl/filters/mbfilter_ascii.h
Index: mbfilter_ascii.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_ascii.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_ascii.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_ascii.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -34,7 +34,8 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_ascii;
-
+extern const struct mbfl_identify_vtbl vtbl_identify_ascii;
+
int mbfl_filt_conv_wchar_ascii(int c, mbfl_convert_filter *filter);
#endif /* MBFL_MBFILTER_ASCII_H */
1.2 +30 -0 libmbfl/filters/mbfilter_big5.c
Index: mbfilter_big5.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_big5.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_big5.c 7 Dec 2002 19:20:47 -0000 1.1
+++ mbfilter_big5.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -36,6 +36,8 @@
#include "unicode_table_big5.h"
+static int mbfl_filt_ident_big5(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_big5[] = { /* 0x81-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -66,6 +68,14 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_big5 = {
+ mbfl_no_encoding_big5,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_big5
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -178,3 +188,23 @@
return c;
}
+
+static int mbfl_filt_ident_big5(int c, mbfl_identify_filter *filter)
+{
+ if (filter->status) { /* kanji second char */
+ if (c < 0x40 || (c > 0x7e && c < 0xa1) ||c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ } else if (c >= 0 && c < 0x80) { /* latin ok */
+ ;
+ } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
+ filter->status = 1;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+
+ return c;
+}
+
+
1.2 +1 -0 libmbfl/filters/mbfilter_big5.h
Index: mbfilter_big5.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_big5.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_big5.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_big5.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_big5;
+extern const struct mbfl_identify_vtbl vtbl_identify_big5;
int mbfl_filt_conv_big5_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_big5(int c, mbfl_convert_filter *filter);
1.3 +22 -0 libmbfl/filters/mbfilter_cp1251.c
Index: mbfilter_cp1251.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp1251.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_cp1251.c 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_cp1251.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -35,6 +35,8 @@
#include "mbfilter_cp1251.h"
#include "unicode_table_cp1251.h"
+static int mbfl_filt_ident_cp1251(int c, mbfl_identify_filter *filter);
+
static const char *mbfl_encoding_cp1251_aliases[] = {"CP1251", "CP-1251", "WINDOWS-1251", NULL};
const mbfl_encoding mbfl_encoding_cp1251 = {
@@ -46,6 +48,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_cp1251 = {
+ mbfl_no_encoding_cp1251,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_cp1251
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -111,3 +121,15 @@
return c;
}
+
+/* all of this is so ugly now! */
+static int mbfl_filt_ident_cp1251(int c, mbfl_identify_filter *filter)
+{
+ if (c >= 0x80 && c < 0xff)
+ filter->flag = 0;
+ else
+ filter->flag = 1; /* not it */
+ return c;
+}
+
+
1.3 +1 -0 libmbfl/filters/mbfilter_cp1251.h
Index: mbfilter_cp1251.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp1251.h,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_cp1251.h 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_cp1251.h 20 Dec 2002 16:46:34 -0000 1.3
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_cp1251;
+extern const struct mbfl_identify_vtbl vtbl_identify_cp1251;
int mbfl_filt_conv_wchar_cp1251(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_cp1251_wchar(int c, mbfl_convert_filter *filter);
1.4 +24 -0 libmbfl/filters/mbfilter_cp1252.c
Index: mbfilter_cp1252.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp1252.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -r1.3 -r1.4
--- mbfilter_cp1252.c 7 Dec 2002 23:05:40 -0000 1.3
+++ mbfilter_cp1252.c 20 Dec 2002 16:46:34 -0000 1.4
@@ -35,6 +35,8 @@
#include "mbfilter_cp1252.h"
#include "unicode_table_cp1252.h"
+static int mbfl_filt_ident_cp1252(int c, mbfl_identify_filter *filter);
+
static const char *mbfl_encoding_cp1252_aliases[] = {"cp1252", NULL};
const mbfl_encoding mbfl_encoding_cp1252 = {
@@ -46,6 +48,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_cp1252 = {
+ mbfl_no_encoding_cp1252,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_cp1252
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -100,6 +109,21 @@
CK((*filter->output_function)(s, filter->data));
return c;
+}
+
+/* We only distinguish the MS extensions to ISO-8859-1.
+ * Actually, this is pretty much a NO-OP, since the identification
+ * system doesn't allow us to discriminate between a positive match,
+ * a possible match and a definite non-match.
+ * The problem here is that cp1252 looks like SJIS for certain chars.
+ * */
+static int mbfl_filt_ident_cp1252(int c, mbfl_identify_filter *filter)
+{
+ if (c >= 0x80 && c < 0xa0)
+ filter->flag = 0;
+ else
+ filter->flag = 1; /* not it */
+ return c;
}
1.3 +1 -0 libmbfl/filters/mbfilter_cp1252.h
Index: mbfilter_cp1252.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp1252.h,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_cp1252.h 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_cp1252.h 20 Dec 2002 16:46:34 -0000 1.3
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_cp1252;
+extern const struct mbfl_identify_vtbl vtbl_identify_cp1252;
int mbfl_filt_conv_wchar_cp1252(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_cp1252_wchar(int c, mbfl_convert_filter *filter);
1.3 +21 -0 libmbfl/filters/mbfilter_cp866.c
Index: mbfilter_cp866.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp866.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_cp866.c 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_cp866.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -34,6 +34,8 @@
#include "mbfilter.h"
#include "unicode_table_cp866.h"
+static int mbfl_filt_ident_cp866(int c, mbfl_identify_filter *filter);
+
static const char *mbfl_encoding_cp866_aliases[] = {"CP866", "CP-866", "IBM-866", NULL};
const mbfl_encoding mbfl_encoding_cp866 = {
@@ -45,6 +47,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_cp866 = {
+ mbfl_no_encoding_cp866,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_cp866
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -110,3 +120,14 @@
return c;
}
+
+static int mbfl_filt_ident_cp866(int c, mbfl_identify_filter *filter)
+{
+ if (c >= 0x80 && c < 0xff)
+ filter->flag = 0;
+ else
+ filter->flag = 1; /* not it */
+ return c;
+}
+
+
1.3 +1 -0 libmbfl/filters/mbfilter_cp866.h
Index: mbfilter_cp866.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp866.h,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_cp866.h 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_cp866.h 20 Dec 2002 16:46:34 -0000 1.3
@@ -31,6 +31,7 @@
#define MBFL_MBFILTER_CP866_H
extern const mbfl_encoding mbfl_encoding_cp866;
+extern const struct mbfl_identify_vtbl vtbl_identify_cp866;
int mbfl_filt_conv_cp866_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_cp866(int c, mbfl_convert_filter *filter);
1.2 +27 -0 libmbfl/filters/mbfilter_cp932.c
Index: mbfilter_cp932.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp932.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_cp932.c 7 Dec 2002 19:20:47 -0000 1.1
+++ mbfilter_cp932.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -37,6 +37,8 @@
#include "unicode_table_cp932_ext.h"
#include "unicode_table_jis.h"
+static int mbfl_filt_ident_sjiswin(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_sjis[] = { /* 0x80-0x9f,0xE0-0xFF */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -67,6 +69,12 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_sjiswin = {
+ mbfl_no_encoding_sjis_win,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_sjiswin
+};
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
@@ -301,5 +309,24 @@
return c;
}
+static int mbfl_filt_ident_sjiswin(int c, mbfl_identify_filter *filter)
+{
+ if (filter->status) { /* kanji second char */
+ if (c < 0x40 || c > 0xfc || c == 0x7f) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ } else if (c >= 0 && c < 0x80) { /* latin ok */
+ ;
+ } else if (c > 0xa0 && c < 0xe0) { /* kana ok */
+ ;
+ } else if (c > 0x80 && c < 0xfd && c != 0xa0) { /* kanji first char */
+ filter->status = 1;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+
+ return c;
+}
1.2 +1 -0 libmbfl/filters/mbfilter_cp932.h
Index: mbfilter_cp932.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp932.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_cp932.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_cp932.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_sjis_win;
+extern const struct mbfl_identify_vtbl vtbl_identify_sjiswin;
int mbfl_filt_conv_sjiswin_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_sjiswin(int c, mbfl_convert_filter *filter);
1.2 +26 -0 libmbfl/filters/mbfilter_cp936.c
Index: mbfilter_cp936.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp936.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_cp936.c 7 Dec 2002 19:20:44 -0000 1.1
+++ mbfilter_cp936.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -36,6 +36,8 @@
#include "unicode_table_cp936.h"
+static int mbfl_filt_ident_cp936(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_cp936[] = { /* 0x81-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -66,6 +68,12 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_cp936 = {
+ mbfl_no_encoding_cp936,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_cp936
+};
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
@@ -175,6 +183,24 @@
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
CK(mbfl_filt_conv_illegal_output(c, filter));
}
+ }
+
+ return c;
+}
+
+static int mbfl_filt_ident_cp936(int c, mbfl_identify_filter *filter)
+{
+ if (filter->status) { /* kanji second char */
+ if (c < 0x40 || c > 0xfe || c == 0x7f) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ } else if (c >= 0 && c < 0x80) { /* latin ok */
+ ;
+ } else if (c > 0x80 && c < 0xff) { /* DBCS lead byte */
+ filter->status = 1;
+ } else { /* bad */
+ filter->flag = 1;
}
return c;
1.2 +2 -1 libmbfl/filters/mbfilter_cp936.h
Index: mbfilter_cp936.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_cp936.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_cp936.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_cp936.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,8 +33,9 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_cp936;
+extern const struct mbfl_identify_vtbl vtbl_identify_cp936;
int mbfl_filt_conv_cp936_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_cp936(int c, mbfl_convert_filter *filter);
-
+
#endif /* MBFL_MBFILTER_CP936_H */
1.2 +37 -0 libmbfl/filters/mbfilter_euc_cn.c
Index: mbfilter_euc_cn.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_cn.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_cn.c 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_cn.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -36,6 +36,8 @@
#include "unicode_table_cp936.h"
+static int mbfl_filt_ident_euccn(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_euccn[] = { /* 0xA1-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -66,6 +68,13 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_euccn = {
+ mbfl_no_encoding_euc_cn,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_euccn
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -173,6 +182,34 @@
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
CK(mbfl_filt_conv_illegal_output(c, filter));
}
+ }
+
+ return c;
+}
+
+static int mbfl_filt_ident_euccn(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status) {
+ case 0: /* latin */
+ if (c >= 0 && c < 0x80) { /* ok */
+ ;
+ } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
+ filter->status = 1;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ case 1: /* got lead byte */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ default:
+ filter->status = 0;
+ break;
}
return c;
1.2 +1 -0 libmbfl/filters/mbfilter_euc_cn.h
Index: mbfilter_euc_cn.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_cn.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_cn.h 7 Dec 2002 19:20:45 -0000 1.1
+++ mbfilter_euc_cn.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -31,6 +31,7 @@
#define MBFL_MBFILTER_EUC_CN_H
extern const mbfl_encoding mbfl_encoding_euc_cn;
+extern const struct mbfl_identify_vtbl vtbl_identify_euccn;
int mbfl_filt_conv_euccn_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_euccn(int c, mbfl_convert_filter *filter);
1.2 +61 -0 libmbfl/filters/mbfilter_euc_jp.c
Index: mbfilter_euc_jp.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_jp.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_jp.c 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_jp.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -37,6 +37,8 @@
#include "unicode_table_cp932_ext.h"
#include "unicode_table_jis.h"
+static int mbfl_filt_ident_eucjp(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_eucjp[] = { /* 0xA1-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -67,6 +69,13 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_eucjp = {
+ mbfl_no_encoding_euc_jp,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_eucjp
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -243,6 +252,58 @@
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
CK(mbfl_filt_conv_illegal_output(c, filter));
}
+ }
+
+ return c;
+}
+
+static int mbfl_filt_ident_eucjp(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status) {
+ case 0: /* latin */
+ if (c >= 0 && c < 0x80) { /* ok */
+ ;
+ } else if (c > 0xa0 && c < 0xff) { /* kanji first char */
+ filter->status = 1;
+ } else if (c == 0x8e) { /* kana first char */
+ filter->status = 2;
+ } else if (c == 0x8f) { /* X 0212 first char */
+ filter->status = 3;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ case 1: /* got first half */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ case 2: /* got 0x8e */
+ if (c < 0xa1 || c > 0xdf) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ case 3: /* got 0x8f */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status++;
+ break;
+ case 4: /* got 0x8f */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ default:
+ filter->status = 0;
+ break;
}
return c;
1.2 +1 -0 libmbfl/filters/mbfilter_euc_jp.h
Index: mbfilter_euc_jp.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_jp.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_jp.h 7 Dec 2002 19:20:49 -0000 1.1
+++ mbfilter_euc_jp.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_euc_jp;
+extern const struct mbfl_identify_vtbl vtbl_identify_eucjp;
int mbfl_filt_conv_eucjp_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_eucjp(int c, mbfl_convert_filter *filter);
1.2 +60 -0 libmbfl/filters/mbfilter_euc_jp_win.c
Index: mbfilter_euc_jp_win.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_jp_win.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_jp_win.c 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_jp_win.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -38,6 +38,8 @@
#include "unicode_table_jis.h"
#include "cp932_table.h"
+static int mbfl_filt_ident_eucjp_win(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_eucjp[] = { /* 0xA1-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -60,6 +62,13 @@
static const char *mbfl_encoding_eucjp_win_aliases[] = {"eucJP-open", NULL};
+const struct mbfl_identify_vtbl vtbl_identify_eucjpwin = {
+ mbfl_no_encoding_eucjp_win,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_eucjp_win
+};
+
const mbfl_encoding mbfl_encoding_eucjp_win = {
mbfl_no_encoding_eucjp_win,
"eucJP-win",
@@ -338,5 +347,56 @@
return c;
}
+static int mbfl_filt_ident_eucjp_win(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status) {
+ case 0: /* latin */
+ if (c >= 0 && c < 0x80) { /* ok */
+ ;
+ } else if (c > 0xa0 && c < 0xff) { /* kanji first char */
+ filter->status = 1;
+ } else if (c == 0x8e) { /* kana first char */
+ filter->status = 2;
+ } else if (c == 0x8f) { /* X 0212 first char */
+ filter->status = 3;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ case 1: /* got first half */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ case 2: /* got 0x8e */
+ if (c < 0xa1 || c > 0xdf) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ case 3: /* got 0x8f */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status++;
+ break;
+ case 4: /* got 0x8f */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
1.2 +1 -0 libmbfl/filters/mbfilter_euc_jp_win.h
Index: mbfilter_euc_jp_win.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_jp_win.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_jp_win.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_jp_win.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_eucjp_win;
+extern const struct mbfl_identify_vtbl vtbl_identify_eucjpwin;
int mbfl_filt_conv_eucjpwin_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_eucjpwin(int c, mbfl_convert_filter *filter);
1.2 +37 -0 libmbfl/filters/mbfilter_euc_kr.c
Index: mbfilter_euc_kr.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_kr.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_kr.c 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_kr.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -35,6 +35,8 @@
#include "mbfilter_euc_kr.h"
#include "unicode_table_uhc.h"
+static int mbfl_filt_ident_euckr(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_euckr[] = { /* 0xA1-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -65,6 +67,14 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_euckr = {
+ mbfl_no_encoding_euc_kr,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_euckr
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -199,3 +209,30 @@
return c;
}
+static int mbfl_filt_ident_euckr(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status) {
+ case 0: /* latin */
+ if (c >= 0 && c < 0x80) { /* ok */
+ ;
+ } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
+ filter->status = 1;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ case 1: /* got lead byte */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
1.2 +1 -0 libmbfl/filters/mbfilter_euc_kr.h
Index: mbfilter_euc_kr.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_kr.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_kr.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_kr.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_euc_kr;
+extern const struct mbfl_identify_vtbl vtbl_identify_euckr;
int mbfl_filt_conv_euckr_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_euckr(int c, mbfl_convert_filter *filter);
1.2 +63 -0 libmbfl/filters/mbfilter_euc_tw.c
Index: mbfilter_euc_tw.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_tw.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_tw.c 7 Dec 2002 19:20:47 -0000 1.1
+++ mbfilter_euc_tw.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -36,6 +36,8 @@
#include "unicode_table_cns11643.h"
+static int mbfl_filt_ident_euctw(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_euctw[] = { /* 0xA1-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -67,6 +69,14 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_euctw = {
+ mbfl_no_encoding_euc_tw,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_euctw
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -247,3 +257,56 @@
}
return c;
}
+
+static int mbfl_filt_ident_euctw(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status) {
+ case 0: /* latin */
+ if (c >= 0 && c < 0x80) { /* ok */
+ ;
+ } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
+ filter->status = 1;
+ } else if (c == 0x8e) { /* DBCS lead byte */
+ filter->status = 2;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ case 1: /* got lead byte */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ case 2: /* got lead byte */
+ if (c >= 0xa1 && c < 0xaf) { /* ok */
+ filter->status = 3;
+ } else {
+ filter->flag = 1; /* bad */
+ }
+ break;
+
+ case 3: /* got lead byte */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 4;
+ break;
+
+ case 4: /* got lead byte */
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
+
1.2 +1 -0 libmbfl/filters/mbfilter_euc_tw.h
Index: mbfilter_euc_tw.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_euc_tw.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_euc_tw.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_euc_tw.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_euc_tw;
+extern const struct mbfl_identify_vtbl vtbl_identify_euctw;
int mbfl_filt_conv_euctw_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_euctw(int c, mbfl_convert_filter *filter);
1.2 +57 -0 libmbfl/filters/mbfilter_hz.c
Index: mbfilter_hz.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_hz.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_hz.c 7 Dec 2002 19:20:45 -0000 1.1
+++ mbfilter_hz.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -36,6 +36,8 @@
#include "unicode_table_cp936.h"
+static int mbfl_filt_ident_hz(int c, mbfl_identify_filter *filter);
+
const mbfl_encoding mbfl_encoding_hz = {
mbfl_no_encoding_hz,
"HZ",
@@ -45,6 +47,14 @@
MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE
};
+const struct mbfl_identify_vtbl vtbl_identify_hz = {
+ mbfl_no_encoding_hz,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_hz
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -193,6 +203,53 @@
filter->status &= 0xff;
return 0;
}
+
+static int mbfl_filt_ident_hz(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status & 0xf) {
+/* case 0x00: ASCII */
+/* case 0x10: GB2312 */
+ case 0:
+ if (c == 0x7e) {
+ filter->status += 2;
+ } else if (filter->status == 0x10 && c > 0x20 && c < 0x7f) { /* DBCS first char */
+ filter->status += 1;
+ } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
+ ;
+ } else {
+ filter->flag = 1; /* bad */
+ }
+ break;
+
+/* case 0x11: GB2312 second char */
+ case 1:
+ filter->status &= ~0xf;
+ if (c < 0x21 || c > 0x7e) { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ case 2:
+ if (c == 0x7d) { /* '}' */
+ filter->status = 0;
+ } else if (c == 0x7b) { /* '{' */
+ filter->status = 0x10;
+ } else if (c == 0x7e) { /* '~' */
+ filter->status = 0;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ }
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
+
/*
* Local variables:
1.2 +1 -0 libmbfl/filters/mbfilter_hz.h
Index: mbfilter_hz.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_hz.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_hz.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_hz.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_hz;
+extern const struct mbfl_identify_vtbl vtbl_identify_hz;
int mbfl_filt_conv_hz_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_hz(int c, mbfl_convert_filter *filter);
1.5 +81 -0 libmbfl/filters/mbfilter_iso2022_kr.c
Index: mbfilter_iso2022_kr.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso2022_kr.c,v
retrieving revision 1.4
retrieving revision 1.5
diff -u -r1.4 -r1.5
--- mbfilter_iso2022_kr.c 7 Dec 2002 21:41:14 -0000 1.4
+++ mbfilter_iso2022_kr.c 20 Dec 2002 16:46:34 -0000 1.5
@@ -35,6 +35,8 @@
#include "mbfilter_euc_kr.h"
#include "unicode_table_uhc.h"
+static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter);
+
const mbfl_encoding mbfl_encoding_2022kr = {
mbfl_no_encoding_2022kr,
"ISO-2022-KR",
@@ -44,6 +46,13 @@
MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE
};
+const struct mbfl_identify_vtbl vtbl_identify_2022kr = {
+ mbfl_no_encoding_2022kr,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_2022kr
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -252,3 +261,75 @@
filter->status &= 0xff;
return 0;
}
+
+static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter)
+{
+retry:
+ switch (filter->status & 0xf) {
+/* case 0x00: ASCII */
+/* case 0x10: KSC5601 mode */
+/* case 0x20: KSC5601 DBCS */
+/* case 0x40: KSC5601 SBCS */
+ case 0:
+ if (!(filter->status & 0x10)) {
+ if (c == 0x1b)
+ filter->status += 2;
+ } else if (filter->status == 0x20 && c > 0x20 && c < 0x7f) { /* kanji first char */
+ filter->status += 1;
+ } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
+ ;
+ } else {
+ filter->flag = 1; /* bad */
+ }
+ break;
+
+/* case 0x21: KSC5601 second char */
+ case 1:
+ filter->status &= ~0xf;
+ if (c < 0x21 || c > 0x7e) { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ /* ESC */
+ case 2:
+ if (c == 0x24) { /* '$' */
+ filter->status++;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC $ */
+ case 3:
+ if (c == 0x29) { /* ')' */
+ filter->status++;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC $) */
+ case 5:
+ if (c == 0x43) { /* 'C' */
+ filter->status = 0x10;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
+
+
1.2 +1 -0 libmbfl/filters/mbfilter_iso2022_kr.h
Index: mbfilter_iso2022_kr.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso2022_kr.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso2022_kr.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso2022_kr.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_2022kr;
+extern const struct mbfl_identify_vtbl vtbl_identify_2022kr;
int mbfl_filt_conv_2022kr_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_2022kr(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_1.c
Index: mbfilter_iso8859_1.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_1.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_1.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_1.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -45,6 +45,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_1 = {
+ mbfl_no_encoding_8859_1,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_1.h
Index: mbfilter_iso8859_1.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_1.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_1.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_1.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_1;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_1;
int mbfl_filt_conv_wchar_8859_1(int c, mbfl_convert_filter *filter);
1.3 +8 -0 libmbfl/filters/mbfilter_iso8859_10.c
Index: mbfilter_iso8859_10.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_10.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_10.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_10.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_10 = {
+ mbfl_no_encoding_8859_10,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_10.h
Index: mbfilter_iso8859_10.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_10.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_10.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_10.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_10;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_10;
int mbfl_filt_conv_8859_10_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_10(int c, mbfl_convert_filter *filter);
1.3 +8 -0 libmbfl/filters/mbfilter_iso8859_13.c
Index: mbfilter_iso8859_13.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_13.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_13.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_13.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_13 = {
+ mbfl_no_encoding_8859_13,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_13.h
Index: mbfilter_iso8859_13.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_13.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_13.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_13.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_13;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_13;
int mbfl_filt_conv_8859_13_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_13(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_14.c
Index: mbfilter_iso8859_14.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_14.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_14.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_14.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_14 = {
+ mbfl_no_encoding_8859_14,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +2 -0 libmbfl/filters/mbfilter_iso8859_14.h
Index: mbfilter_iso8859_14.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_14.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_14.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_14.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -16,6 +16,8 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_14;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_14;
+
int mbfl_filt_conv_8859_14_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_14(int c, mbfl_convert_filter *filter);
1.3 +8 -0 libmbfl/filters/mbfilter_iso8859_15.c
Index: mbfilter_iso8859_15.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_15.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_15.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_15.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_15 = {
+ mbfl_no_encoding_8859_15,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_15.h
Index: mbfilter_iso8859_15.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_15.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_15.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_15.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -16,6 +16,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_15;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_15;
int mbfl_filt_conv_8859_15_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_15(int c, mbfl_convert_filter *filter);
1.3 +6 -0 libmbfl/filters/mbfilter_iso8859_2.c
Index: mbfilter_iso8859_2.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_2.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_2.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_2.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,12 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_2 = {
+ mbfl_no_encoding_8859_2,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true };
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_2.h
Index: mbfilter_iso8859_2.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_2.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_2.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_2.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_2;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_2;
int mbfl_filt_conv_8859_2_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_2(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_3.c
Index: mbfilter_iso8859_3.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_3.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_3.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_3.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_3 = {
+ mbfl_no_encoding_8859_3,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true };
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_3.h
Index: mbfilter_iso8859_3.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_3.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_3.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_3.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_3;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_3;
int mbfl_filt_conv_8859_3_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_3(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_4.c
Index: mbfilter_iso8859_4.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_4.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_4.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_4.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_4 = {
+ mbfl_no_encoding_8859_4,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true };
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_4.h
Index: mbfilter_iso8859_4.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_4.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_4.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_4.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -31,6 +31,7 @@
#define MBFL_MBFILTER_ISO8859_4_H
extern const mbfl_encoding mbfl_encoding_8859_4;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_4;
int mbfl_filt_conv_8859_4_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_4(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_5.c
Index: mbfilter_iso8859_5.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_5.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_5.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_5.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_5 = {
+ mbfl_no_encoding_8859_5,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true };
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_5.h
Index: mbfilter_iso8859_5.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_5.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_5.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_5.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_5;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_5;
int mbfl_filt_conv_8859_5_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_5(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_6.c
Index: mbfilter_iso8859_6.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_6.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_6.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_6.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_6 = {
+ mbfl_no_encoding_8859_6,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_6.h
Index: mbfilter_iso8859_6.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_6.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_6.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_6.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_6;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_6;
int mbfl_filt_conv_8859_6_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_6(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_7.c
Index: mbfilter_iso8859_7.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_7.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_7.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_7.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_7 = {
+ mbfl_no_encoding_8859_7,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_7.h
Index: mbfilter_iso8859_7.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_7.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_7.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_7.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_7;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_7;
int mbfl_filt_conv_8859_7_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_7(int c, mbfl_convert_filter *filter);
1.3 +8 -0 libmbfl/filters/mbfilter_iso8859_8.c
Index: mbfilter_iso8859_8.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_8.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_8.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_8.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,14 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_8 = {
+ mbfl_no_encoding_8859_8,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_8.h
Index: mbfilter_iso8859_8.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_8.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_8.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_8.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_8;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_8;
int mbfl_filt_conv_8859_8_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_8(int c, mbfl_convert_filter *filter);
1.3 +7 -0 libmbfl/filters/mbfilter_iso8859_9.c
Index: mbfilter_iso8859_9.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_9.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_iso8859_9.c 7 Dec 2002 21:41:14 -0000 1.2
+++ mbfilter_iso8859_9.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -46,6 +46,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_8859_9 = {
+ mbfl_no_encoding_8859_9,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_true
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
1.2 +1 -0 libmbfl/filters/mbfilter_iso8859_9.h
Index: mbfilter_iso8859_9.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_iso8859_9.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_iso8859_9.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_iso8859_9.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_8859_9;
+extern const struct mbfl_identify_vtbl vtbl_identify_8859_9;
int mbfl_filt_conv_8859_9_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_8859_9(int c, mbfl_convert_filter *filter);
1.2 +192 -0 libmbfl/filters/mbfilter_jis.c
Index: mbfilter_jis.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_jis.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_jis.c 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_jis.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -37,6 +37,9 @@
#include "unicode_table_cp932_ext.h"
#include "unicode_table_jis.h"
+static int mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter);
+static int mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter);
+
const mbfl_encoding mbfl_encoding_jis = {
mbfl_no_encoding_jis,
"JIS",
@@ -55,6 +58,21 @@
MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE
};
+const struct mbfl_identify_vtbl vtbl_identify_jis = {
+ mbfl_no_encoding_jis,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_jis
+};
+
+const struct mbfl_identify_vtbl vtbl_identify_2022jp = {
+ mbfl_no_encoding_2022jp,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_2022jp
+};
+
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -426,6 +444,180 @@
}
filter->status &= 0xff;
return 0;
+}
+
+static int mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter)
+{
+retry:
+ switch (filter->status & 0xf) {
+/* case 0x00: ASCII */
+/* case 0x10: X 0201 latin */
+/* case 0x20: X 0201 kana */
+/* case 0x80: X 0208 */
+/* case 0x90: X 0212 */
+ case 0:
+ if (c == 0x1b) {
+ filter->status += 2;
+ } else if (c == 0x0e) { /* "kana in" */
+ filter->status = 0x20;
+ } else if (c == 0x0f) { /* "kana out" */
+ filter->status = 0;
+ } else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) { /* kanji first char */
+ filter->status += 1;
+ } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
+ ;
+ } else {
+ filter->flag = 1; /* bad */
+ }
+ break;
+
+/* case 0x81: X 0208 second char */
+/* case 0x91: X 0212 second char */
+ case 1:
+ filter->status &= ~0xf;
+ if (c == 0x1b) {
+ goto retry;
+ } else if (c < 0x21 || c > 0x7e) { /* bad */
+ filter->flag = 1;
+ }
+ break;
+
+ /* ESC */
+ case 2:
+ if (c == 0x24) { /* '$' */
+ filter->status++;
+ } else if (c == 0x28) { /* '(' */
+ filter->status += 3;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC $ */
+ case 3:
+ if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
+ filter->status = 0x80;
+ } else if (c == 0x28) { /* '(' */
+ filter->status++;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC $ ( */
+ case 4:
+ if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
+ filter->status = 0x80;
+ } else if (c == 0x44) { /* 'D' */
+ filter->status = 0x90;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC ( */
+ case 5:
+ if (c == 0x42 || c == 0x48) { /* 'B' or 'H' */
+ filter->status = 0;
+ } else if (c == 0x4a) { /* 'J' */
+ filter->status = 0x10;
+ } else if (c == 0x49) { /* 'I' */
+ filter->status = 0x20;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
+
+static int mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter)
+{
+retry:
+ switch (filter->status & 0xf) {
+/* case 0x00: ASCII */
+/* case 0x10: X 0201 latin */
+/* case 0x80: X 0208 */
+ case 0:
+ if (c == 0x1b) {
+ filter->status += 2;
+ } else if (filter->status == 0x80 && c > 0x20 && c < 0x7f) { /* kanji first char */
+ filter->status += 1;
+ } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
+ ;
+ } else {
+ filter->flag = 1; /* bad */
+ }
+ break;
+
+/* case 0x81: X 0208 second char */
+ case 1:
+ if (c == 0x1b) {
+ filter->status++;
+ } else {
+ filter->status &= ~0xf;
+ if (c < 0x21 || c > 0x7e) { /* bad */
+ filter->flag = 1;
+ }
+ }
+ break;
+
+ /* ESC */
+ case 2:
+ if (c == 0x24) { /* '$' */
+ filter->status++;
+ } else if (c == 0x28) { /* '(' */
+ filter->status += 3;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC $ */
+ case 3:
+ if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
+ filter->status = 0x80;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ /* ESC ( */
+ case 5:
+ if (c == 0x42) { /* 'B' */
+ filter->status = 0;
+ } else if (c == 0x4a) { /* 'J' */
+ filter->status = 0x10;
+ } else {
+ filter->flag = 1; /* bad */
+ filter->status &= ~0xf;
+ goto retry;
+ }
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
}
1.2 +3 -1 libmbfl/filters/mbfilter_jis.h
Index: mbfilter_jis.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_jis.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_jis.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_jis.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -34,7 +34,9 @@
extern const mbfl_encoding mbfl_encoding_jis;
extern const mbfl_encoding mbfl_encoding_2022jp;
-
+extern const struct mbfl_identify_vtbl vtbl_identify_2022jp;
+extern const struct mbfl_identify_vtbl vtbl_identify_jis;
+
int mbfl_filt_conv_jis_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter);
1.3 +18 -0 libmbfl/filters/mbfilter_koi8r.c
Index: mbfilter_koi8r.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_koi8r.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_koi8r.c 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_koi8r.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -34,6 +34,8 @@
#include "mbfilter.h"
#include "unicode_table_koi8r.h"
+static int mbfl_filt_ident_koi8r(int c, mbfl_identify_filter *filter);
+
static const char *mbfl_encoding_koi8r_aliases[] = {"KOI8-R", "KOI8R", NULL};
const mbfl_encoding mbfl_encoding_koi8r = {
@@ -45,6 +47,13 @@
MBFL_ENCTYPE_SBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_koi8r = {
+ mbfl_no_encoding_koi8r,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_koi8r
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -109,4 +118,13 @@
}
return c;
+}
+
+static int mbfl_filt_ident_koi8r(int c, mbfl_identify_filter *filter)
+{
+ if (c >= 0x80 && c < 0xff)
+ filter->flag = 0;
+ else
+ filter->flag = 1; /* not it */
+ return c;
}
1.3 +1 -0 libmbfl/filters/mbfilter_koi8r.h
Index: mbfilter_koi8r.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_koi8r.h,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_koi8r.h 7 Dec 2002 23:05:40 -0000 1.2
+++ mbfilter_koi8r.h 20 Dec 2002 16:46:34 -0000 1.3
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_koi8r;
+extern const struct mbfl_identify_vtbl vtbl_identify_koi8r;
int mbfl_filt_conv_koi8r_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_koi8r(int c, mbfl_convert_filter *filter);
1.2 +30 -0 libmbfl/filters/mbfilter_sjis.c
Index: mbfilter_sjis.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_sjis.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_sjis.c 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_sjis.c 20 Dec 2002 16:46:34 -0000 1.2
@@ -37,6 +37,8 @@
#include "unicode_table_cp932_ext.h"
#include "unicode_table_jis.h"
+static int mbfl_filt_ident_sjis(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_sjis[] = { /* 0x80-0x9f,0xE0-0xFF */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -67,6 +69,13 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_sjis = {
+ mbfl_no_encoding_sjis,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_sjis
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
#define SJIS_ENCODE(c1,c2,s1,s2) \
@@ -245,3 +254,24 @@
return c;
}
+
+static int mbfl_filt_ident_sjis(int c, mbfl_identify_filter *filter)
+{
+ if (filter->status) { /* kanji second char */
+ if (c < 0x40 || c > 0xfc || c == 0x7f) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ } else if (c >= 0 && c < 0x80) { /* latin ok */
+ ;
+ } else if (c > 0xa0 && c < 0xe0) { /* kana ok */
+ ;
+ } else if (c > 0x80 && c < 0xf0 && c != 0xa0) { /* kanji first char */
+ filter->status = 1;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+
+ return c;
+}
+
1.2 +1 -0 libmbfl/filters/mbfilter_sjis.h
Index: mbfilter_sjis.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_sjis.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_sjis.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_sjis.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_sjis;
+extern const struct mbfl_identify_vtbl vtbl_identify_sjis;
int mbfl_filt_conv_sjis_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_sjis(int c, mbfl_convert_filter *filter);
1.6 +51 -0 libmbfl/filters/mbfilter_uhc.c
Index: mbfilter_uhc.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_uhc.c,v
retrieving revision 1.5
retrieving revision 1.6
diff -u -r1.5 -r1.6
--- mbfilter_uhc.c 11 Dec 2002 18:22:59 -0000 1.5
+++ mbfilter_uhc.c 20 Dec 2002 16:46:34 -0000 1.6
@@ -35,6 +35,8 @@
#include "mbfilter_euc_kr.h"
#include "unicode_table_uhc.h"
+static int mbfl_filt_ident_uhc(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_uhc[] = { /* 0x81-0xFE */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -65,6 +67,13 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_uhc = {
+ mbfl_no_encoding_uhc,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_uhc
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -195,3 +204,45 @@
return c;
}
+
+static int mbfl_filt_ident_uhc(int c, mbfl_identify_filter *filter)
+{
+ switch (filter->status) {
+ case 0: /* latin */
+ if (c >= 0 && c < 0x80) { /* ok */
+ ;
+ } else if (c >= 0x81 && c <= 0xa0) { /* dbcs first char */
+ filter->status= 1;
+ } else if (c >= 0xa1 && c <= 0xc6) { /* dbcs first char */
+ filter->status= 2;
+ } else if (c >= 0xc7 && c <= 0xfe) { /* dbcs first char */
+ filter->status= 3;
+ } else { /* bad */
+ filter->flag = 1;
+ }
+
+ case 1:
+ case 2:
+ if (c < 0x41 || (c > 0x5a && c < 0x61)
+ || (c > 0x7a && c < 0x81) || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ case 3:
+ if (c < 0xa1 || c > 0xfe) { /* bad */
+ filter->flag = 1;
+ }
+ filter->status = 0;
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
+
+
1.2 +1 -0 libmbfl/filters/mbfilter_uhc.h
Index: mbfilter_uhc.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_uhc.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_uhc.h 7 Dec 2002 19:20:44 -0000 1.1
+++ mbfilter_uhc.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_uhc;
+extern const struct mbfl_identify_vtbl vtbl_identify_uhc;
int mbfl_filt_conv_uhc_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_uhc(int c, mbfl_convert_filter *filter);
1.3 +59 -0 libmbfl/filters/mbfilter_utf7.c
Index: mbfilter_utf7.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_utf7.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_utf7.c 7 Dec 2002 23:01:24 -0000 1.2
+++ mbfilter_utf7.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -34,6 +34,8 @@
#include "mbfilter.h"
#include "mbfilter_utf7.h"
+static int mbfl_filt_ident_utf7(int c, mbfl_identify_filter *filter);
+
static const unsigned char mbfl_base64_table[] = {
/* 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', */
0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x4b,0x4c,0x4d,
@@ -58,6 +60,13 @@
MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE
};
+const struct mbfl_identify_vtbl vtbl_identify_utf7 = {
+ mbfl_no_encoding_utf7,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_utf7
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -379,4 +388,54 @@
}
return 0;
}
+
+static int mbfl_filt_ident_utf7(int c, mbfl_identify_filter *filter)
+{
+ int n;
+
+ switch (filter->status) {
+ /* directly encoded characters */
+ case 0:
+ if (c == 0x2b) { /* '+' shift character */
+ filter->status++;
+ } else if (c == 0x5c || c == 0x7e || c < 0 || c > 0x7f) { /* illegal character */
+ filter->flag = 1; /* bad */
+ }
+ break;
+
+ /* Modified Base64 */
+ case 1:
+ case 2:
+ n = 0;
+ if (c >= 0x41 && c <= 0x5a) { /* A - Z */
+ n = 1;
+ } else if (c >= 0x61 && c <= 0x7a) { /* a - z */
+ n = 1;
+ } else if (c >= 0x30 && c <= 0x39) { /* 0 - 9 */
+ n = 1;
+ } else if (c == 0x2b) { /* '+' */
+ n = 1;
+ } else if (c == 0x2f) { /* '/' */
+ n = 1;
+ }
+ if (n <= 0) {
+ if (filter->status == 1 && c != 0x2d) {
+ filter->flag = 1; /* bad */
+ } else if (c < 0 || c > 0x7f) {
+ filter->flag = 1; /* bad */
+ }
+ filter->status = 0;
+ } else {
+ filter->status = 2;
+ }
+ break;
+
+ default:
+ filter->status = 0;
+ break;
+ }
+
+ return c;
+}
+
1.2 +1 -0 libmbfl/filters/mbfilter_utf7.h
Index: mbfilter_utf7.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_utf7.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_utf7.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_utf7.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -33,6 +33,7 @@
#include "mbfilter.h"
extern const mbfl_encoding mbfl_encoding_utf7;
+extern const struct mbfl_identify_vtbl vtbl_identify_utf7;
int mbfl_filt_conv_utf7_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_utf7(int c, mbfl_convert_filter *filter);
1.3 +67 -0 libmbfl/filters/mbfilter_utf8.c
Index: mbfilter_utf8.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_utf8.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfilter_utf8.c 7 Dec 2002 23:01:24 -0000 1.2
+++ mbfilter_utf8.c 20 Dec 2002 16:46:34 -0000 1.3
@@ -34,6 +34,8 @@
#include "mbfilter.h"
#include "mbfilter_utf8.h"
+static int mbfl_filt_ident_utf8(int c, mbfl_identify_filter *filter);
+
static const unsigned char mblen_table_utf8[] = {
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -64,6 +66,13 @@
MBFL_ENCTYPE_MBCS
};
+const struct mbfl_identify_vtbl vtbl_identify_utf8 = {
+ mbfl_no_encoding_utf8,
+ mbfl_filt_ident_common_ctor,
+ mbfl_filt_ident_common_dtor,
+ mbfl_filt_ident_utf8
+};
+
#define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
/*
@@ -177,6 +186,64 @@
} else {
if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
CK(mbfl_filt_conv_illegal_output(c, filter));
+ }
+ }
+
+ return c;
+}
+
+static int mbfl_filt_ident_utf8(int c, mbfl_identify_filter *filter)
+{
+ if (c < 0x80) {
+ if (c < 0) {
+ filter->flag = 1; /* bad */
+ } else if (c != 0 && filter->status) {
+ filter->flag = 1; /* bad */
+ }
+ filter->status = 0;
+ } else if (c < 0xc0) {
+ switch (filter->status) {
+ case 0x20: /* 3 byte code 2nd char */
+ case 0x30: /* 4 byte code 2nd char */
+ case 0x31: /* 4 byte code 3rd char */
+ case 0x40: /* 5 byte code 2nd char */
+ case 0x41: /* 5 byte code 3rd char */
+ case 0x42: /* 5 byte code 4th char */
+ case 0x50: /* 6 byte code 2nd char */
+ case 0x51: /* 6 byte code 3rd char */
+ case 0x52: /* 6 byte code 4th char */
+ case 0x53: /* 6 byte code 5th char */
+ filter->status++;
+ break;
+ case 0x10: /* 2 byte code 2nd char */
+ case 0x21: /* 3 byte code 3rd char */
+ case 0x32: /* 4 byte code 4th char */
+ case 0x43: /* 5 byte code 5th char */
+ case 0x54: /* 6 byte code 6th char */
+ filter->status = 0;
+ break;
+ default:
+ filter->flag = 1; /* bad */
+ filter->status = 0;
+ break;
+ }
+ } else {
+ if (filter->status) {
+ filter->flag = 1; /* bad */
+ }
+ filter->status = 0;
+ if (c < 0xe0) { /* 2 byte code first char */
+ filter->status = 0x10;
+ } else if (c < 0xf0) { /* 3 byte code 1st char */
+ filter->status = 0x20;
+ } else if (c < 0xf8) { /* 4 byte code 1st char */
+ filter->status = 0x30;
+ } else if (c < 0xfc) { /* 5 byte code 1st char */
+ filter->status = 0x40;
+ } else if (c < 0xfe) { /* 6 byte code 1st char */
+ filter->status = 0x50;
+ } else {
+ filter->flag = 1; /* bad */
}
}
1.2 +1 -0 libmbfl/filters/mbfilter_utf8.h
Index: mbfilter_utf8.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/filters/mbfilter_utf8.h,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- mbfilter_utf8.h 7 Dec 2002 19:20:50 -0000 1.1
+++ mbfilter_utf8.h 20 Dec 2002 16:46:34 -0000 1.2
@@ -31,6 +31,7 @@
#define MBFL_MBFILTER_UTF8_H
extern const mbfl_encoding mbfl_encoding_utf8;
+extern const struct mbfl_identify_vtbl vtbl_identify_utf8;
int mbfl_filt_conv_utf8_wchar(int c, mbfl_convert_filter *filter);
int mbfl_filt_conv_wchar_utf8(int c, mbfl_convert_filter *filter);
1.5 +11 -1013 libmbfl/mbfl/mbfilter.c
Index: mbfilter.c
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/mbfl/mbfilter.c,v
retrieving revision 1.4
retrieving revision 1.5
diff -u -r1.4 -r1.5
--- mbfilter.c 9 Dec 2002 01:07:39 -0000 1.4
+++ mbfilter.c 20 Dec 2002 16:46:36 -0000 1.5
@@ -329,43 +329,6 @@
0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0x41,0x42,0x43,0x44,0x45,0x46
};
-/* forward */
-static void mbfl_filt_conv_common_ctor(mbfl_convert_filter *filter);
-static int mbfl_filt_conv_common_flush(mbfl_convert_filter *filter);
-static void mbfl_filt_conv_common_dtor(mbfl_convert_filter *filter);
-
-static void mbfl_filt_ident_common_ctor(mbfl_identify_filter *filter);
-static void mbfl_filt_ident_common_dtor(mbfl_identify_filter *filter);
-static void mbfl_filt_ident_false_ctor(mbfl_identify_filter *filter);
-static int mbfl_filt_ident_utf8(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_utf7(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_ascii(int c, mbfl_identify_filter *filter);
-
-static int mbfl_filt_ident_eucjp(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_sjis(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_sjiswin(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter);
-
-static int mbfl_filt_ident_euccn(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_cp936(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_hz(int c, mbfl_identify_filter *filter);
-
-static int mbfl_filt_ident_euctw(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_big5(int c, mbfl_identify_filter *filter);
-
-static int mbfl_filt_ident_euckr(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_uhc(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter);
-
-static int mbfl_filt_ident_cp1251(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_cp866(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_koi8r(int c, mbfl_identify_filter *filter);
-
-static int mbfl_filt_ident_cp1252(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_false(int c, mbfl_identify_filter *filter);
-static int mbfl_filt_ident_true(int c, mbfl_identify_filter *filter);
-
/* convert filter function table */
static const struct mbfl_convert_vtbl vtbl_pass = {
mbfl_no_encoding_pass,
@@ -1383,215 +1346,13 @@
/* identify filter function table */
-static const struct mbfl_identify_vtbl vtbl_identify_ascii = {
- mbfl_no_encoding_ascii,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_ascii };
-
-static const struct mbfl_identify_vtbl vtbl_identify_utf8 = {
- mbfl_no_encoding_utf8,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_utf8 };
-
-static const struct mbfl_identify_vtbl vtbl_identify_utf7 = {
- mbfl_no_encoding_utf7,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_utf7 };
-
-static const struct mbfl_identify_vtbl vtbl_identify_eucjp = {
- mbfl_no_encoding_euc_jp,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_eucjp };
-
-static const struct mbfl_identify_vtbl vtbl_identify_eucjpwin = {
- mbfl_no_encoding_eucjp_win,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_eucjp };
-
-static const struct mbfl_identify_vtbl vtbl_identify_sjis = {
- mbfl_no_encoding_sjis,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_sjis };
-
-static const struct mbfl_identify_vtbl vtbl_identify_sjiswin = {
- mbfl_no_encoding_sjis_win,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_sjiswin };
-
-static const struct mbfl_identify_vtbl vtbl_identify_jis = {
- mbfl_no_encoding_jis,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_jis };
-
-static const struct mbfl_identify_vtbl vtbl_identify_2022jp = {
- mbfl_no_encoding_2022jp,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_2022jp };
-
-static struct mbfl_identify_vtbl vtbl_identify_euccn = {
- mbfl_no_encoding_euc_cn,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_euccn };
-
-static struct mbfl_identify_vtbl vtbl_identify_cp936 = {
- mbfl_no_encoding_cp936,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_cp936 };
-
-static struct mbfl_identify_vtbl vtbl_identify_hz = {
- mbfl_no_encoding_hz,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_hz };
-
-static struct mbfl_identify_vtbl vtbl_identify_euctw = {
- mbfl_no_encoding_euc_tw,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_euctw };
-
-static struct mbfl_identify_vtbl vtbl_identify_big5 = {
- mbfl_no_encoding_big5,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_big5 };
-
-static struct mbfl_identify_vtbl vtbl_identify_euckr = {
- mbfl_no_encoding_euc_kr,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_euckr };
-
-static struct mbfl_identify_vtbl vtbl_identify_uhc = {
- mbfl_no_encoding_uhc,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_uhc };
-
-static struct mbfl_identify_vtbl vtbl_identify_2022kr = {
- mbfl_no_encoding_2022kr,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_2022kr };
-
-static struct mbfl_identify_vtbl vtbl_identify_cp1251 = {
- mbfl_no_encoding_cp1251,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_cp1251 };
-
-static struct mbfl_identify_vtbl vtbl_identify_cp866 = {
- mbfl_no_encoding_cp866,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_cp866 };
-
-static struct mbfl_identify_vtbl vtbl_identify_koi8r = {
- mbfl_no_encoding_koi8r,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_koi8r };
-
-static const struct mbfl_identify_vtbl vtbl_identify_cp1252 = {
- mbfl_no_encoding_cp1252,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_cp1252 };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_1 = {
- mbfl_no_encoding_8859_1,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_2 = {
- mbfl_no_encoding_8859_2,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_3 = {
- mbfl_no_encoding_8859_3,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_4 = {
- mbfl_no_encoding_8859_4,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_5 = {
- mbfl_no_encoding_8859_5,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_6 = {
- mbfl_no_encoding_8859_6,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_7 = {
- mbfl_no_encoding_8859_7,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_8 = {
- mbfl_no_encoding_8859_8,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_9 = {
- mbfl_no_encoding_8859_9,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_10 = {
- mbfl_no_encoding_8859_10,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_13 = {
- mbfl_no_encoding_8859_13,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_14 = {
- mbfl_no_encoding_8859_14,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
-
-static const struct mbfl_identify_vtbl vtbl_identify_8859_15 = {
- mbfl_no_encoding_8859_15,
- mbfl_filt_ident_common_ctor,
- mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_true };
static const struct mbfl_identify_vtbl vtbl_identify_false = {
mbfl_no_encoding_pass,
mbfl_filt_ident_false_ctor,
mbfl_filt_ident_common_dtor,
- mbfl_filt_ident_false };
+ mbfl_filt_ident_false
+};
static const struct mbfl_identify_vtbl *mbfl_identify_filter_list[] = {
&vtbl_identify_utf8,
@@ -2191,812 +1952,49 @@
/*
* commonly used constructor and destructor
*/
-static void
-mbfl_filt_conv_common_ctor(mbfl_convert_filter *filter)
+void mbfl_filt_conv_common_ctor(mbfl_convert_filter *filter)
{
filter->status = 0;
filter->cache = 0;
}
-static int
-mbfl_filt_conv_common_flush(mbfl_convert_filter *filter)
+int mbfl_filt_conv_common_flush(mbfl_convert_filter *filter)
{
filter->status = 0;
filter->cache = 0;
return 0;
}
-static void
-mbfl_filt_conv_common_dtor(mbfl_convert_filter *filter)
+void mbfl_filt_conv_common_dtor(mbfl_convert_filter *filter)
{
filter->status = 0;
filter->cache = 0;
}
-/*
- *
- * identify filter functions
- *
- */
-
-static void
-mbfl_filt_ident_common_ctor(mbfl_identify_filter *filter)
+void mbfl_filt_ident_common_ctor(mbfl_identify_filter *filter)
{
filter->status = 0;
filter->flag = 0;
}
-static void
-mbfl_filt_ident_common_dtor(mbfl_identify_filter *filter)
+void mbfl_filt_ident_common_dtor(mbfl_identify_filter *filter)
{
filter->status = 0;
}
-static int
-mbfl_filt_ident_ascii(int c, mbfl_identify_filter *filter)
+int mbfl_filt_ident_false(int c, mbfl_identify_filter *filter)
{
- if (c >= 0x20 && c < 0x80) {
- ;
- } else if (c == 0x0d || c == 0x0a || c == 0x09 || c == 0) { /* CR or LF or HTAB or null */
- ;
- } else {
- filter->flag = 1;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_utf8(int c, mbfl_identify_filter *filter)
-{
- if (c < 0x80) {
- if (c < 0) {
- filter->flag = 1; /* bad */
- } else if (c != 0 && filter->status) {
- filter->flag = 1; /* bad */
- }
- filter->status = 0;
- } else if (c < 0xc0) {
- switch (filter->status) {
- case 0x20: /* 3 byte code 2nd char */
- case 0x30: /* 4 byte code 2nd char */
- case 0x31: /* 4 byte code 3rd char */
- case 0x40: /* 5 byte code 2nd char */
- case 0x41: /* 5 byte code 3rd char */
- case 0x42: /* 5 byte code 4th char */
- case 0x50: /* 6 byte code 2nd char */
- case 0x51: /* 6 byte code 3rd char */
- case 0x52: /* 6 byte code 4th char */
- case 0x53: /* 6 byte code 5th char */
- filter->status++;
- break;
- case 0x10: /* 2 byte code 2nd char */
- case 0x21: /* 3 byte code 3rd char */
- case 0x32: /* 4 byte code 4th char */
- case 0x43: /* 5 byte code 5th char */
- case 0x54: /* 6 byte code 6th char */
- filter->status = 0;
- break;
- default:
- filter->flag = 1; /* bad */
- filter->status = 0;
- break;
- }
- } else {
- if (filter->status) {
- filter->flag = 1; /* bad */
- }
- filter->status = 0;
- if (c < 0xe0) { /* 2 byte code first char */
- filter->status = 0x10;
- } else if (c < 0xf0) { /* 3 byte code 1st char */
- filter->status = 0x20;
- } else if (c < 0xf8) { /* 4 byte code 1st char */
- filter->status = 0x30;
- } else if (c < 0xfc) { /* 5 byte code 1st char */
- filter->status = 0x40;
- } else if (c < 0xfe) { /* 6 byte code 1st char */
- filter->status = 0x50;
- } else {
- filter->flag = 1; /* bad */
- }
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_utf7(int c, mbfl_identify_filter *filter)
-{
- int n;
-
- switch (filter->status) {
- /* directly encoded characters */
- case 0:
- if (c == 0x2b) { /* '+' shift character */
- filter->status++;
- } else if (c == 0x5c || c == 0x7e || c < 0 || c > 0x7f) { /* illegal character */
- filter->flag = 1; /* bad */
- }
- break;
-
- /* Modified Base64 */
- case 1:
- case 2:
- n = 0;
- if (c >= 0x41 && c <= 0x5a) { /* A - Z */
- n = 1;
- } else if (c >= 0x61 && c <= 0x7a) { /* a - z */
- n = 1;
- } else if (c >= 0x30 && c <= 0x39) { /* 0 - 9 */
- n = 1;
- } else if (c == 0x2b) { /* '+' */
- n = 1;
- } else if (c == 0x2f) { /* '/' */
- n = 1;
- }
- if (n <= 0) {
- if (filter->status == 1 && c != 0x2d) {
- filter->flag = 1; /* bad */
- } else if (c < 0 || c > 0x7f) {
- filter->flag = 1; /* bad */
- }
- filter->status = 0;
- } else {
- filter->status = 2;
- }
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_eucjp(int c, mbfl_identify_filter *filter)
-{
- switch (filter->status) {
- case 0: /* latin */
- if (c >= 0 && c < 0x80) { /* ok */
- ;
- } else if (c > 0xa0 && c < 0xff) { /* kanji first char */
- filter->status = 1;
- } else if (c == 0x8e) { /* kana first char */
- filter->status = 2;
- } else if (c == 0x8f) { /* X 0212 first char */
- filter->status = 3;
- } else { /* bad */
- filter->flag = 1;
- }
- break;
-
- case 1: /* got first half */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- case 2: /* got 0x8e */
- if (c < 0xa1 || c > 0xdf) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- case 3: /* got 0x8f */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status++;
- break;
- case 4: /* got 0x8f */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_sjis(int c, mbfl_identify_filter *filter)
-{
- if (filter->status) { /* kanji second char */
- if (c < 0x40 || c > 0xfc || c == 0x7f) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- } else if (c >= 0 && c < 0x80) { /* latin ok */
- ;
- } else if (c > 0xa0 && c < 0xe0) { /* kana ok */
- ;
- } else if (c > 0x80 && c < 0xf0 && c != 0xa0) { /* kanji first char */
- filter->status = 1;
- } else { /* bad */
- filter->flag = 1;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_sjiswin(int c, mbfl_identify_filter *filter)
-{
- if (filter->status) { /* kanji second char */
- if (c < 0x40 || c > 0xfc || c == 0x7f) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- } else if (c >= 0 && c < 0x80) { /* latin ok */
- ;
- } else if (c > 0xa0 && c < 0xe0) { /* kana ok */
- ;
- } else if (c > 0x80 && c < 0xfd && c != 0xa0) { /* kanji first char */
- filter->status = 1;
- } else { /* bad */
- filter->flag = 1;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter)
-{
-retry:
- switch (filter->status & 0xf) {
-/* case 0x00: ASCII */
-/* case 0x10: X 0201 latin */
-/* case 0x20: X 0201 kana */
-/* case 0x80: X 0208 */
-/* case 0x90: X 0212 */
- case 0:
- if (c == 0x1b) {
- filter->status += 2;
- } else if (c == 0x0e) { /* "kana in" */
- filter->status = 0x20;
- } else if (c == 0x0f) { /* "kana out" */
- filter->status = 0;
- } else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) { /* kanji first char */
- filter->status += 1;
- } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
- ;
- } else {
- filter->flag = 1; /* bad */
- }
- break;
-
-/* case 0x81: X 0208 second char */
-/* case 0x91: X 0212 second char */
- case 1:
- filter->status &= ~0xf;
- if (c == 0x1b) {
- goto retry;
- } else if (c < 0x21 || c > 0x7e) { /* bad */
- filter->flag = 1;
- }
- break;
-
- /* ESC */
- case 2:
- if (c == 0x24) { /* '$' */
- filter->status++;
- } else if (c == 0x28) { /* '(' */
- filter->status += 3;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC $ */
- case 3:
- if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
- filter->status = 0x80;
- } else if (c == 0x28) { /* '(' */
- filter->status++;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC $ ( */
- case 4:
- if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
- filter->status = 0x80;
- } else if (c == 0x44) { /* 'D' */
- filter->status = 0x90;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC ( */
- case 5:
- if (c == 0x42 || c == 0x48) { /* 'B' or 'H' */
- filter->status = 0;
- } else if (c == 0x4a) { /* 'J' */
- filter->status = 0x10;
- } else if (c == 0x49) { /* 'I' */
- filter->status = 0x20;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_euccn(int c, mbfl_identify_filter *filter)
-{
- switch (filter->status) {
- case 0: /* latin */
- if (c >= 0 && c < 0x80) { /* ok */
- ;
- } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
- filter->status = 1;
- } else { /* bad */
- filter->flag = 1;
- }
- break;
-
- case 1: /* got lead byte */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_cp936(int c, mbfl_identify_filter *filter)
-{
- if (filter->status) { /* kanji second char */
- if (c < 0x40 || c > 0xfe || c == 0x7f) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- } else if (c >= 0 && c < 0x80) { /* latin ok */
- ;
- } else if (c > 0x80 && c < 0xff) { /* DBCS lead byte */
- filter->status = 1;
- } else { /* bad */
- filter->flag = 1;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_hz(int c, mbfl_identify_filter *filter)
-{
- switch (filter->status & 0xf) {
-/* case 0x00: ASCII */
-/* case 0x10: GB2312 */
- case 0:
- if (c == 0x7e) {
- filter->status += 2;
- } else if (filter->status == 0x10 && c > 0x20 && c < 0x7f) { /* DBCS first char */
- filter->status += 1;
- } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
- ;
- } else {
- filter->flag = 1; /* bad */
- }
- break;
-
-/* case 0x11: GB2312 second char */
- case 1:
- filter->status &= ~0xf;
- if (c < 0x21 || c > 0x7e) { /* bad */
- filter->flag = 1;
- }
- break;
-
- case 2:
- if (c == 0x7d) { /* '}' */
- filter->status = 0;
- } else if (c == 0x7b) { /* '{' */
- filter->status = 0x10;
- } else if (c == 0x7e) { /* '~' */
- filter->status = 0;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- }
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_euctw(int c, mbfl_identify_filter *filter)
-{
- switch (filter->status) {
- case 0: /* latin */
- if (c >= 0 && c < 0x80) { /* ok */
- ;
- } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
- filter->status = 1;
- } else if (c == 0x8e) { /* DBCS lead byte */
- filter->status = 2;
- } else { /* bad */
- filter->flag = 1;
- }
- break;
-
- case 1: /* got lead byte */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- case 2: /* got lead byte */
- if (c >= 0xa1 && c < 0xaf) { /* ok */
- filter->status = 3;
- } else {
- filter->flag = 1; /* bad */
- }
- break;
-
- case 3: /* got lead byte */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 4;
- break;
-
- case 4: /* got lead byte */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_big5(int c, mbfl_identify_filter *filter)
-{
- if (filter->status) { /* kanji second char */
- if (c < 0x40 || (c > 0x7e && c < 0xa1) ||c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- } else if (c >= 0 && c < 0x80) { /* latin ok */
- ;
- } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
- filter->status = 1;
- } else { /* bad */
- filter->flag = 1;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_euckr(int c, mbfl_identify_filter *filter)
-{
- switch (filter->status) {
- case 0: /* latin */
- if (c >= 0 && c < 0x80) { /* ok */
- ;
- } else if (c > 0xa0 && c < 0xff) { /* DBCS lead byte */
- filter->status = 1;
- } else { /* bad */
- filter->flag = 1;
- }
- break;
-
- case 1: /* got lead byte */
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_uhc(int c, mbfl_identify_filter *filter)
-{
- switch (filter->status) {
- case 0: /* latin */
- if (c >= 0 && c < 0x80) { /* ok */
- ;
- } else if (c >= 0x81 && c <= 0xa0) { /* dbcs first char */
- filter->status= 1;
- } else if (c >= 0xa1 && c <= 0xc6) { /* dbcs first char */
- filter->status= 2;
- } else if (c >= 0xc7 && c <= 0xfe) { /* dbcs first char */
- filter->status= 3;
- } else { /* bad */
- filter->flag = 1;
- }
-
- case 1:
- case 2:
- if (c < 0x41 || (c > 0x5a && c < 0x61)
- || (c > 0x7a && c < 0x81) || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- case 3:
- if (c < 0xa1 || c > 0xfe) { /* bad */
- filter->flag = 1;
- }
- filter->status = 0;
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-static int
-mbfl_filt_ident_2022kr(int c, mbfl_identify_filter *filter)
-{
-retry:
- switch (filter->status & 0xf) {
-/* case 0x00: ASCII */
-/* case 0x10: KSC5601 mode */
-/* case 0x20: KSC5601 DBCS */
-/* case 0x40: KSC5601 SBCS */
- case 0:
- if (!(filter->status & 0x10)) {
- if (c == 0x1b)
- filter->status += 2;
- } else if (filter->status == 0x20 && c > 0x20 && c < 0x7f) { /* kanji first char */
- filter->status += 1;
- } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
- ;
- } else {
- filter->flag = 1; /* bad */
- }
- break;
-
-/* case 0x21: KSC5601 second char */
- case 1:
- filter->status &= ~0xf;
- if (c < 0x21 || c > 0x7e) { /* bad */
- filter->flag = 1;
- }
- break;
-
- /* ESC */
- case 2:
- if (c == 0x24) { /* '$' */
- filter->status++;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC $ */
- case 3:
- if (c == 0x29) { /* ')' */
- filter->status++;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC $) */
- case 5:
- if (c == 0x43) { /* 'C' */
- filter->status = 0x10;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
- return c;
-}
-
-/* We only distinguish the MS extensions to ISO-8859-1.
- * Actually, this is pretty much a NO-OP, since the identification
- * system doesn't allow us to discriminate between a positive match,
- * a possible match and a definite non-match.
- * The problem here is that cp1252 looks like SJIS for certain chars.
- * */
-static int
-mbfl_filt_ident_cp1252(int c, mbfl_identify_filter *filter)
-{
- if (c >= 0x80 && c < 0xa0)
- filter->flag = 0;
- else
- filter->flag = 1; /* not it */
- return c;
-}
-
-/* all of this is so ugly now! */
-static int
-mbfl_filt_ident_cp1251(int c, mbfl_identify_filter *filter)
-{
- if (c >= 0x80 && c < 0xff)
- filter->flag = 0;
- else
- filter->flag = 1; /* not it */
- return c;
-}
-
-static int
-mbfl_filt_ident_cp866(int c, mbfl_identify_filter *filter)
-{
- if (c >= 0x80 && c < 0xff)
- filter->flag = 0;
- else
- filter->flag = 1; /* not it */
- return c;
-}
-
-static int
-mbfl_filt_ident_koi8r(int c, mbfl_identify_filter *filter)
-{
- if (c >= 0x80 && c < 0xff)
- filter->flag = 0;
- else
- filter->flag = 1; /* not it */
- return c;
-}
-
-static int
-mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter)
-{
-retry:
- switch (filter->status & 0xf) {
-/* case 0x00: ASCII */
-/* case 0x10: X 0201 latin */
-/* case 0x80: X 0208 */
- case 0:
- if (c == 0x1b) {
- filter->status += 2;
- } else if (filter->status == 0x80 && c > 0x20 && c < 0x7f) { /* kanji first char */
- filter->status += 1;
- } else if (c >= 0 && c < 0x80) { /* latin, CTLs */
- ;
- } else {
- filter->flag = 1; /* bad */
- }
- break;
-
-/* case 0x81: X 0208 second char */
- case 1:
- if (c == 0x1b) {
- filter->status++;
- } else {
- filter->status &= ~0xf;
- if (c < 0x21 || c > 0x7e) { /* bad */
- filter->flag = 1;
- }
- }
- break;
-
- /* ESC */
- case 2:
- if (c == 0x24) { /* '$' */
- filter->status++;
- } else if (c == 0x28) { /* '(' */
- filter->status += 3;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC $ */
- case 3:
- if (c == 0x40 || c == 0x42) { /* '@' or 'B' */
- filter->status = 0x80;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- /* ESC ( */
- case 5:
- if (c == 0x42) { /* 'B' */
- filter->status = 0;
- } else if (c == 0x4a) { /* 'J' */
- filter->status = 0x10;
- } else {
- filter->flag = 1; /* bad */
- filter->status &= ~0xf;
- goto retry;
- }
- break;
-
- default:
- filter->status = 0;
- break;
- }
-
+ filter->flag = 1; /* bad */
return c;
}
-static void
-mbfl_filt_ident_false_ctor(mbfl_identify_filter *filter)
+void mbfl_filt_ident_false_ctor(mbfl_identify_filter *filter)
{
filter->status = 0;
filter->flag = 1;
}
-static int
-mbfl_filt_ident_false(int c, mbfl_identify_filter *filter)
-{
- filter->flag = 1; /* bad */
- return c;
-}
-
-static int
-mbfl_filt_ident_true(int c, mbfl_identify_filter *filter)
+int mbfl_filt_ident_true(int c, mbfl_identify_filter *filter)
{
return c;
}
1.4 +13 -0 libmbfl/mbfl/mbfilter.h
Index: mbfilter.h
===================================================================
RCS file: /cvsroot/php-i18n/libmbfl/mbfl/mbfilter.h,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -r1.3 -r1.4
--- mbfilter.h 8 Dec 2002 11:00:02 -0000 1.3
+++ mbfilter.h 20 Dec 2002 16:46:36 -0000 1.4
@@ -612,4 +612,17 @@
#define mbfl_prealloc __mbfl_allocators->preallloc
#define mbfl_pfree __mbfl_allocators->pfree
+/* common filters */
+void mbfl_filt_conv_common_ctor(mbfl_convert_filter *filter);
+int mbfl_filt_conv_common_flush(mbfl_convert_filter *filter);
+void mbfl_filt_conv_common_dtor(mbfl_convert_filter *filter);
+
+void mbfl_filt_ident_common_ctor(mbfl_identify_filter *filter);
+void mbfl_filt_ident_common_dtor(mbfl_identify_filter *filter);
+void mbfl_filt_ident_false_ctor(mbfl_identify_filter *filter);
+
+int mbfl_filt_ident_false(int c, mbfl_identify_filter *filter);
+int mbfl_filt_ident_true(int c, mbfl_identify_filter *filter);
+
+
#endif /* MBFL_MBFILTER_H */