summaryrefslogtreecommitdiff
path: root/ext/nkf
diff options
context:
space:
mode:
Diffstat (limited to 'ext/nkf')
-rw-r--r--ext/nkf/nkf-utf8/config.h80
-rw-r--r--ext/nkf/nkf-utf8/nkf.c6616
-rwxr-xr-xext/nkf/nkf-utf8/nkf.h166
-rw-r--r--ext/nkf/nkf-utf8/utf8tbl.c7
-rw-r--r--ext/nkf/nkf-utf8/utf8tbl.h24
5 files changed, 3494 insertions, 3399 deletions
diff --git a/ext/nkf/nkf-utf8/config.h b/ext/nkf/nkf-utf8/config.h
index 1fa74afd40..473be76a8e 100644
--- a/ext/nkf/nkf-utf8/config.h
+++ b/ext/nkf/nkf-utf8/config.h
@@ -1,91 +1,51 @@
#ifndef _CONFIG_H_
#define _CONFIG_H_
-/* UTF8 $BF~=PNO(B */
+/* UTF8 input and output */
#define UTF8_INPUT_ENABLE
#define UTF8_OUTPUT_ENABLE
-/* Shift_JIS $BHO0O30$NJ8;z$r!"(BCP932 $B$GF1CM$JJ8;z$KFI$_49$($k(B */
+/* invert characters invalid in Shift_JIS to CP932 */
#define SHIFTJIS_CP932
-/* $B%*%W%7%g%s$GF~NO$r;XDj$7$?;~$K!"J8;z%3!<%I$r8GDj$9$k(B */
+/* fix input encoding when given by option */
#define INPUT_CODE_FIX
-/* --overwrite $B%*%W%7%g%s(B */
+/* --overwrite option */
/* by Satoru Takabayashi <ccsatoru@vega.aichi-u.ac.jp> */
#define OVERWRITE
-/* --cap-input, --url-input $B%*%W%7%g%s(B */
+/* --cap-input, --url-input option */
#define INPUT_OPTION
-/* --numchar-input $B%*%W%7%g%s(B */
+/* --numchar-input option */
#define NUMCHAR_OPTION
-/* --debug, --no-output $B%*%W%7%g%s(B */
+/* --debug, --no-output option */
#define CHECK_OPTION
/* JIS X0212 */
#define X0212_ENABLE
-/* --exec-in, --exec-out $B%*%W%7%g%s(B
- * pipe, fork, execvp $B$"$?$j$,L5$$$HF0$-$^$;$s!#(B
- * MS-DOS, MinGW $B$J$I$G$O(B undef $B$K$7$F$/$@$5$$(B
- * child process $B=*N;;~$N=hM}$,$$$$$+$2$s$J$N$G!"(B
- * $B%G%U%)%k%H$GL58z$K$7$F$$$^$9!#(B
+/* --exec-in, --exec-out option
+ * require pipe, fork, execvp and so on.
+ * please undef this on MS-DOS, MinGW
+ * this is still buggy arround child process
*/
/* #define EXEC_IO */
-/* SunOS $B$N(B cc $B$r;H$&$H$-$O(B undef $B$K$7$F$/$@$5$$(B */
-#define ANSI_C_PROTOTYPE
-
-/* int $B$,(B 32bit $BL$K~$N4D6-$G(B NUMCHAR_OPTION $B$r;H$&$K$O!"(B
- * $B%3%a%s%H$r30$7$F$/$@$5$$!#(B
- */
-/* #define INT_IS_SHORT */
-
-
-#if defined(INT_IS_SHORT)
-typedef long nkf_char;
-typedef unsigned char nkf_nfchar;
-#else
-typedef int nkf_char;
-typedef int nkf_nfchar;
-#endif
-
/* Unicode Normalization */
#define UNICODE_NORMALIZATION
-#ifndef WIN32DLL
-/******************************/
-/* $B%G%U%)%k%H$N=PNO%3!<%IA*Br(B */
-/* Select DEFAULT_CODE */
-/* #define DEFAULT_CODE_JIS */
-/* #define DEFAULT_CODE_SJIS */
-/* #define DEFAULT_CODE_EUC */
-#define DEFAULT_CODE_UTF8
-/******************************/
-#else
-#define DEFAULT_CODE_SJIS
-#endif
-
-#if defined(NUMCHAR_OPTION) && !defined(UTF8_INPUT_ENABLE)
-#define UTF8_INPUT_ENABLE
-#endif
-
-#ifdef UNICODE_NORMALIZATION
-#ifndef UTF8_INPUT_ENABLE
-#define UTF8_INPUT_ENABLE
-#endif
-#define NORMALIZATION_TABLE_LENGTH 942
-#define NORMALIZATION_TABLE_NFC_LENGTH 3
-#define NORMALIZATION_TABLE_NFD_LENGTH 9
-struct normalization_pair{
- const nkf_nfchar nfc[NORMALIZATION_TABLE_NFC_LENGTH];
- const nkf_nfchar nfd[NORMALIZATION_TABLE_NFD_LENGTH];
-};
-#endif
+/*
+ * Select Default Output Encoding
+ *
+ * If not defined, locale encoding is used.
+ */
-#define MIME_DECODE_DEFAULT 0
-#define X0201_DEFAULT 0
+/* #define DEFAULT_CODE_JIS */
+/* #define DEFAULT_CODE_SJIS */
+/* #define DEFAULT_CODE_EUC */
+/* #define DEFAULT_CODE_UTF8 */
#endif /* _CONFIG_H_ */
diff --git a/ext/nkf/nkf-utf8/nkf.c b/ext/nkf/nkf-utf8/nkf.c
index 54f690a954..b51ece582c 100644
--- a/ext/nkf/nkf-utf8/nkf.c
+++ b/ext/nkf/nkf-utf8/nkf.c
@@ -1,4 +1,5 @@
/** Network Kanji Filter. (PDS Version)
+** -*- coding: ISO-2022-JP -*-
************************************************************************
** Copyright (C) 1987, Fujitsu LTD. (Itaru ICHIKAWA)
** $BO"Mm@h!'(B $B!J3t!KIY;NDL8&5f=j!!%=%U%H#38&!!;T@n!!;j(B
@@ -30,153 +31,17 @@
* $B8=:_!"(Bnkf $B$O(B SorceForge $B$K$F%a%s%F%J%s%9$,B3$1$i$l$F$$$^$9!#(B
* http://sourceforge.jp/projects/nkf/
***********************************************************************/
-/* $Id$ */
+#define NKF_IDENT "$Id$"
#define NKF_VERSION "2.0.8"
-#define NKF_RELEASE_DATE "2008-01-11"
+#define NKF_RELEASE_DATE "2008-01-23"
#define COPY_RIGHT \
"Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW\n" \
"Copyright (C) 2002-2008 Kono, Furukawa, Naruse, mastodon"
#include "config.h"
+#include "nkf.h"
#include "utf8tbl.h"
-#if defined(DEFAULT_CODE_JIS)
-#elif defined(DEFAULT_CODE_SJIS)
-#elif defined(DEFAULT_CODE_EUC)
-#elif defined(DEFAULT_CODE_UTF8)
-#else
-#define DEFAULT_CODE_JIS 1
-#endif
-
-#ifndef MIME_DECODE_DEFAULT
-#define MIME_DECODE_DEFAULT STRICT_MIME
-#endif
-#ifndef X0201_DEFAULT
-#define X0201_DEFAULT TRUE
-#endif
-
-#if DEFAULT_NEWLINE == 0x0D0A
-#define PUT_NEWLINE(func) do {\
- func(0x0D);\
- func(0x0A);\
-} while (0)
-#define OCONV_NEWLINE(func) do {\
- func(0, 0x0D);\
- func(0, 0x0A);\
-} while (0)
-#elif DEFAULT_NEWLINE == 0x0D
-#define PUT_NEWLINE(func) func(0x0D)
-#define OCONV_NEWLINE(func) func(0, 0x0D)
-#else
-#define DEFAULT_NEWLINE 0x0A
-#define PUT_NEWLINE(func) func(0x0A)
-#define OCONV_NEWLINE(func) func(0, 0x0A)
-#endif
-#ifdef HELP_OUTPUT_STDERR
-#define HELP_OUTPUT stderr
-#else
-#define HELP_OUTPUT stdout
-#endif
-
-#if (defined(__TURBOC__) || defined(_MSC_VER) || defined(LSI_C) || defined(__MINGW32__) || defined(__EMX__) || defined(__MSDOS__) || defined(__WINDOWS__) || defined(__DOS__) || defined(__OS2__)) && !defined(MSDOS)
-#define MSDOS
-#if (defined(__Win32__) || defined(_WIN32)) && !defined(__WIN32__)
-#define __WIN32__
-#endif
-#endif
-
-#ifdef PERL_XS
-#undef OVERWRITE
-#endif
-
-#ifndef PERL_XS
-#include <stdio.h>
-#endif
-
-#include <stdlib.h>
-#include <string.h>
-
-#if defined(MSDOS) || defined(__OS2__)
-#include <fcntl.h>
-#include <io.h>
-#if defined(_MSC_VER) || defined(__WATCOMC__)
-#define mktemp _mktemp
-#endif
-#endif
-
-#ifdef MSDOS
-#ifdef LSI_C
-#define setbinmode(fp) fsetbin(fp)
-#elif defined(__DJGPP__)
-#include <libc/dosio.h>
-#define setbinmode(fp) djgpp_setbinmode(fp)
-#else /* Microsoft C, Turbo C */
-#define setbinmode(fp) setmode(fileno(fp), O_BINARY)
-#endif
-#else /* UNIX */
-#define setbinmode(fp)
-#endif
-
-#if defined(__DJGPP__)
-void djgpp_setbinmode(FILE *fp)
-{
- /* we do not use libc's setmode(), which changes COOKED/RAW mode in device. */
- int fd, m;
- fd = fileno(fp);
- m = (__file_handle_modes[fd] & (~O_TEXT)) | O_BINARY;
- __file_handle_set(fd, m);
-}
-#endif
-
-#ifdef _IOFBF /* SysV and MSDOS, Windows */
-#define setvbuffer(fp, buf, size) setvbuf(fp, buf, _IOFBF, size)
-#else /* BSD */
-#define setvbuffer(fp, buf, size) setbuffer(fp, buf, size)
-#endif
-
-/*Borland C++ 4.5 EasyWin*/
-#if defined(__TURBOC__) && defined(_Windows) && !defined(__WIN32__) /*Easy Win */
-#define EASYWIN
-#ifndef __WIN16__
-#define __WIN16__
-#endif
-#include <windows.h>
-#endif
-
-#ifdef OVERWRITE
-/* added by satoru@isoternet.org */
-#if defined(__EMX__)
-#include <sys/types.h>
-#endif
-#include <sys/stat.h>
-#if !defined(MSDOS) || defined(__DJGPP__) /* UNIX, djgpp */
-#include <unistd.h>
-#if defined(__WATCOMC__)
-#include <sys/utime.h>
-#else
-#include <utime.h>
-#endif
-#else /* defined(MSDOS) */
-#ifdef __WIN32__
-#ifdef __BORLANDC__ /* BCC32 */
-#include <utime.h>
-#else /* !defined(__BORLANDC__) */
-#include <sys/utime.h>
-#endif /* (__BORLANDC__) */
-#else /* !defined(__WIN32__) */
-#if defined(_MSC_VER) || defined(__MINGW32__) || defined(__WATCOMC__) || defined(__OS2__) || defined(__EMX__) || defined(__IBMC__) || defined(__IBMCPP__) /* VC++, MinGW, Watcom, emx+gcc, IBM VAC++ */
-#include <sys/utime.h>
-#elif defined(__TURBOC__) /* BCC */
-#include <utime.h>
-#elif defined(LSI_C) /* LSI C */
-#endif /* (__WIN32__) */
-#endif
-#endif
-#endif
-
-#define FALSE 0
-#define TRUE 1
-
/* state of output_mode and input_mode
c2 0 means ASCII
@@ -384,14 +249,15 @@ struct {
{"BINARY", BINARY},
{NULL, -1}
};
+
#if defined(DEFAULT_CODE_JIS)
-#define DEFAULT_ENCODING ISO_2022_JP
+#define DEFAULT_ENCIDX ISO_2022_JP
#elif defined(DEFAULT_CODE_SJIS)
-#define DEFAULT_ENCODING SHIFT_JIS
+#define DEFAULT_ENCIDX SHIFT_JIS
#elif defined(DEFAULT_CODE_EUC)
-#define DEFAULT_ENCODING EUC_JP
+#define DEFAULT_ENCIDX EUC_JP
#elif defined(DEFAULT_CODE_UTF8)
-#define DEFAULT_ENCODING UTF_8
+#define DEFAULT_ENCIDX UTF_8
#endif
@@ -418,10 +284,6 @@ struct {
((c > SP) && (c < DEL) && (c != '?') && (c != '=') && (c != '_') \
&& (c != '(') && (c != ')') && (c != '.') && (c != 0x22)))
-#define CP932_TABLE_BEGIN 0xFA
-#define CP932_TABLE_END 0xFC
-#define CP932INV_TABLE_BEGIN 0xED
-#define CP932INV_TABLE_END 0xEE
#define is_ibmext_in_sjis(c2) (CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END)
#define HOLD_SIZE 1024
@@ -434,23 +296,11 @@ struct {
#define DEFAULT_J 'B'
#define DEFAULT_R 'B'
-#define SJ0162 0x00e1 /* 01 - 62 ku offset */
-#define SJ6394 0x0161 /* 63 - 94 ku offset */
-#define RANGE_NUM_MAX 18
#define GETA1 0x22
#define GETA2 0x2e
-#if defined(UTF8_OUTPUT_ENABLE) || defined(UTF8_INPUT_ENABLE)
-#define sizeof_euc_to_utf8_1byte 94
-#define sizeof_euc_to_utf8_2bytes 94
-#define sizeof_utf8_to_euc_C2 64
-#define sizeof_utf8_to_euc_E5B8 64
-#define sizeof_utf8_to_euc_2bytes 112
-#define sizeof_utf8_to_euc_3bytes 16
-#endif
-
/* MIME preprocessor */
#ifdef EASYWIN /*Easy Win */
@@ -472,15 +322,8 @@ static char *input_codename = NULL; /* NULL: unestablished, "": BINARY */
static nkf_encoding *input_encoding = NULL;
static nkf_encoding *output_encoding = NULL;
-#if !defined(PERL_XS) && !defined(WIN32DLL)
-static nkf_char noconvert(FILE *f);
-#endif
-static void module_connection(void);
static nkf_char kanji_convert(FILE *f);
static nkf_char h_conv(FILE *f,nkf_char c2,nkf_char c1);
-static nkf_char push_hold_buf(nkf_char c2);
-static void set_iconv(nkf_char f, nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0));
-static nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1);
#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
/* UCS Mapping
* 0: Shift_JIS, eucJP-ascii
@@ -501,39 +344,19 @@ static int no_cp932ext_f = FALSE;
static int no_best_fit_chars_f = FALSE;
static int input_endian = ENDIAN_BIG;
static nkf_char unicode_subchar = '?'; /* the regular substitution character */
-static void nkf_each_char_to_hex(void (*f)(nkf_char c2,nkf_char c1), nkf_char c);
-static void encode_fallback_html(nkf_char c);
-static void encode_fallback_xml(nkf_char c);
-static void encode_fallback_java(nkf_char c);
-static void encode_fallback_perl(nkf_char c);
-static void encode_fallback_subchar(nkf_char c);
static void (*encode_fallback)(nkf_char c) = NULL;
-static nkf_char w2e_conv(nkf_char c2,nkf_char c1,nkf_char c0,nkf_char *p2,nkf_char *p1);
-static nkf_char unicode_to_jis_common(nkf_char c2,nkf_char c1,nkf_char c0,nkf_char *p2,nkf_char *p1);
-static nkf_char w_iconv_common(nkf_char c1,nkf_char c0,const unsigned short *const *pp,nkf_char psize,nkf_char *p2,nkf_char *p1);
-static void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0);
-static nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0);
-static nkf_char w16e_conv(nkf_char val,nkf_char *p2,nkf_char *p1);
static void w_status(struct input_code *, nkf_char);
#endif
#ifdef UTF8_OUTPUT_ENABLE
static int output_bom_f = FALSE;
static int output_endian = ENDIAN_BIG;
-static nkf_char e2w_conv(nkf_char c2,nkf_char c1);
#endif
-static nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1);
static void fold_conv(nkf_char c2,nkf_char c1);
-static void nl_conv(nkf_char c2,nkf_char c1);
+static void eol_conv(nkf_char c2,nkf_char c1);
static void z_conv(nkf_char c2,nkf_char c1);
static void rot_conv(nkf_char c2,nkf_char c1);
static void hira_conv(nkf_char c2,nkf_char c1);
-static void base64_conv(nkf_char c2,nkf_char c1);
static void iso2022jp_check_conv(nkf_char c2,nkf_char c1);
-static void no_connection(nkf_char c2,nkf_char c1);
-static nkf_char no_connection2(nkf_char c2,nkf_char c1,nkf_char c0);
-
-static void code_score(struct input_code *ptr);
-static void code_status(nkf_char c);
static void std_putc(nkf_char c);
static nkf_char std_getc(FILE *f);
@@ -542,31 +365,9 @@ static nkf_char std_ungetc(nkf_char c,FILE *f);
static nkf_char broken_getc(FILE *f);
static nkf_char broken_ungetc(nkf_char c,FILE *f);
-static nkf_char mime_begin(FILE *f);
static nkf_char mime_getc(FILE *f);
-static nkf_char mime_ungetc(nkf_char c,FILE *f);
-
-static void switch_mime_getc(void);
-static void unswitch_mime_getc(void);
-static nkf_char mime_begin_strict(FILE *f);
-static nkf_char mime_getc_buf(FILE *f);
-static nkf_char mime_ungetc_buf(nkf_char c,FILE *f);
-static nkf_char mime_integrity(FILE *f,const unsigned char *p);
-
-static nkf_char base64decode(nkf_char c);
-static void mime_prechar(nkf_char c2, nkf_char c1);
-static void mime_putc(nkf_char c);
-static void open_mime(nkf_char c);
-static void close_mime(void);
-static void eof_mime(void);
-static void mimeout_addchar(nkf_char c);
-#ifndef PERL_XS
-static void usage(void);
-static void version(void);
-static void show_configuration(void);
-#endif
-static void options(unsigned char *c);
-static void reinit(void);
+
+static void mime_putc(nkf_char c);
/* buffers */
@@ -574,19 +375,6 @@ static void reinit(void);
static unsigned char stdibuf[IOBUF_SIZE];
static unsigned char stdobuf[IOBUF_SIZE];
#endif
-static unsigned char hold_buf[HOLD_SIZE*2];
-static int hold_count = 0;
-
-/* MIME preprocessor fifo */
-
-#define MIME_BUF_SIZE (1024) /* 2^n ring buffer */
-#define MIME_BUF_MASK (MIME_BUF_SIZE-1)
-#define Fifo(n) mime_buf[(n)&MIME_BUF_MASK]
-static unsigned char mime_buf[MIME_BUF_SIZE];
-static unsigned int mime_top = 0;
-static unsigned int mime_last = 0; /* decoded */
-static unsigned int mime_input = 0; /* undecoded */
-static nkf_char (*mime_iconv_back)(nkf_char c2,nkf_char c1,nkf_char c0) = NULL;
/* flags */
static int unbuf_f = FALSE;
@@ -609,29 +397,18 @@ static int iso2022jp_f = FALSE; /* replace non ISO-2022-JP with G
static int nfc_f = FALSE;
static nkf_char (*i_nfc_getc)(FILE *) = std_getc; /* input of ugetc */
static nkf_char (*i_nfc_ungetc)(nkf_char c ,FILE *f) = std_ungetc;
-static nkf_char nfc_getc(FILE *f);
-static nkf_char nfc_ungetc(nkf_char c,FILE *f);
#endif
#ifdef INPUT_OPTION
static int cap_f = FALSE;
static nkf_char (*i_cgetc)(FILE *) = std_getc; /* input of cgetc */
static nkf_char (*i_cungetc)(nkf_char c ,FILE *f) = std_ungetc;
-static nkf_char cap_getc(FILE *f);
-static nkf_char cap_ungetc(nkf_char c,FILE *f);
static int url_f = FALSE;
static nkf_char (*i_ugetc)(FILE *) = std_getc; /* input of ugetc */
static nkf_char (*i_uungetc)(nkf_char c ,FILE *f) = std_ungetc;
-static nkf_char url_getc(FILE *f);
-static nkf_char url_ungetc(nkf_char c,FILE *f);
#endif
-#if defined(INT_IS_SHORT)
-#define NKF_INT32_C(n) (n##L)
-#else
-#define NKF_INT32_C(n) (n)
-#endif
#define PREFIX_EUCG3 NKF_INT32_C(0x8F00)
#define CLASS_MASK NKF_INT32_C(0xFF000000)
#define CLASS_UNICODE NKF_INT32_C(0x01000000)
@@ -644,8 +421,6 @@ static nkf_char url_ungetc(nkf_char c,FILE *f);
static int numchar_f = FALSE;
static nkf_char (*i_ngetc)(FILE *) = std_getc; /* input of ugetc */
static nkf_char (*i_nungetc)(nkf_char c ,FILE *f) = std_ungetc;
-static nkf_char numchar_getc(FILE *f);
-static nkf_char numchar_ungetc(nkf_char c,FILE *f);
#endif
#ifdef CHECK_OPTION
@@ -657,9 +432,6 @@ static nkf_char (*iconv_for_check)(nkf_char c2,nkf_char c1,nkf_char c0) = 0;
#endif
static int guess_f = 0; /* 0: OFF, 1: ON, 2: VERBOSE */
-#if !defined PERL_XS
-static void print_guessed_code(char *filename);
-#endif
static void set_input_codename(char *codename);
#ifdef EXEC_IO
@@ -678,21 +450,11 @@ static int cp932inv_f = TRUE;
#ifdef X0212_ENABLE
static int x0212_f = FALSE;
-static nkf_char x0212_shift(nkf_char c);
-static nkf_char x0212_unshift(nkf_char c);
#endif
static int x0213_f = FALSE;
static unsigned char prefix_table[256];
-static void set_code_score(struct input_code *ptr, nkf_char score);
-static void clr_code_score(struct input_code *ptr, nkf_char score);
-static void status_disable(struct input_code *ptr);
-static void status_push_ch(struct input_code *ptr, nkf_char c);
-static void status_clear(struct input_code *ptr);
-static void status_reset(struct input_code *ptr);
-static void status_reinit(struct input_code *ptr);
-static void status_check(struct input_code *ptr, nkf_char c);
static void e_status(struct input_code *, nkf_char);
static void s_status(struct input_code *, nkf_char);
@@ -730,28 +492,26 @@ static unsigned char ascii_intro = DEFAULT_R;
static int fold_margin = FOLD_MARGIN;
-/* converters */
+/* process default */
-#ifdef DEFAULT_CODE_JIS
-# define DEFAULT_CONV j_oconv
-#endif
-#ifdef DEFAULT_CODE_SJIS
-# define DEFAULT_CONV s_oconv
-#endif
-#ifdef DEFAULT_CODE_EUC
-# define DEFAULT_CONV e_oconv
-#endif
-#ifdef DEFAULT_CODE_UTF8
-# define DEFAULT_CONV w_oconv
-#endif
+nkf_char no_connection2(nkf_char c2, nkf_char c1, nkf_char c0)
+{
+ fprintf(stderr,"nkf internal module connection failure.\n");
+ exit(1);
+ return 0; /* LINT */
+}
+
+void no_connection(nkf_char c2, nkf_char c1)
+{
+ no_connection2(c2,c1,0);
+}
-/* process default */
static nkf_char (*iconv)(nkf_char c2,nkf_char c1,nkf_char c0) = no_connection2;
static void (*oconv)(nkf_char c2,nkf_char c1) = no_connection;
static void (*o_zconv)(nkf_char c2,nkf_char c1) = no_connection;
static void (*o_fconv)(nkf_char c2,nkf_char c1) = no_connection;
-static void (*o_nlconv)(nkf_char c2,nkf_char c1) = no_connection;
+static void (*o_eol_conv)(nkf_char c2,nkf_char c1) = no_connection;
static void (*o_rot_conv)(nkf_char c2,nkf_char c1) = no_connection;
static void (*o_hira_conv)(nkf_char c2,nkf_char c1) = no_connection;
static void (*o_base64conv)(nkf_char c2,nkf_char c1) = no_connection;
@@ -869,17 +629,17 @@ static const unsigned char fv[] = {
+static int option_mode = 0;
static int file_out_f = FALSE;
#ifdef OVERWRITE
static int overwrite_f = FALSE;
static int preserve_time_f = FALSE;
static int backup_f = FALSE;
static char *backup_suffix = "";
-static char *get_backup_filename(const char *suffix, const char *filename);
#endif
-static int nlmode_f = 0; /* CR, LF, CRLF */
-static int input_newline = 0; /* 0: unestablished, EOF: MIXED */
+static int eolmode_f = 0; /* CR, LF, CRLF */
+static int input_eol = 0; /* 0: unestablished, EOF: MIXED */
static nkf_char prev_cr = 0; /* CR or 0 */
#ifdef EASYWIN /*Easy Win */
static int end_check;
@@ -954,282 +714,169 @@ static nkf_encoding *nkf_enc_find(const char *name)
nkf_enc_to_index(enc) == CP50221 ||\
nkf_enc_to_index(enc) == CP50222)
-#ifdef WIN32DLL
-#include "nkf32dll.c"
-#elif defined(PERL_XS)
-#else /* WIN32DLL */
-int main(int argc, char **argv)
+#ifndef DEFAULT_ENCIDX
+static char* nkf_locale_charmap()
{
- FILE *fin;
- unsigned char *cp;
-
- char *outfname = NULL;
- char *origfname;
-
-#ifdef EASYWIN /*Easy Win */
- _BufferSize.y = 400;/*Set Scroll Buffer Size*/
+#ifdef HAVE_LANGINFO_H
+ return nl_langinfo(CODESET);
+#elif defined(__WIN32__)
+ return sprintf("CP%d", GetACP());
+#else
+ return NULL;
#endif
+}
- for (argc--,argv++; (argc > 0) && **argv == '-'; argc--, argv++) {
- cp = (unsigned char *)*argv;
- options(cp);
- if (guess_f) {
-#ifdef CHECK_OPTION
- int debug_f_back = debug_f;
-#endif
-#ifdef EXEC_IO
- int exec_f_back = exec_f;
-#endif
-#ifdef X0212_ENABLE
- int x0212_f_back = x0212_f;
-#endif
- int x0213_f_back = x0213_f;
- int guess_f_back = guess_f;
- reinit();
- guess_f = guess_f_back;
- mime_f = FALSE;
-#ifdef CHECK_OPTION
- debug_f = debug_f_back;
-#endif
-#ifdef EXEC_IO
- exec_f = exec_f_back;
-#endif
-#ifdef X0212_ENABLE
- x0212_f = x0212_f_back;
-#endif
- x0213_f = x0213_f_back;
- }
-#ifdef EXEC_IO
- if (exec_f){
- int fds[2], pid;
- if (pipe(fds) < 0 || (pid = fork()) < 0){
- abort();
- }
- if (pid == 0){
- if (exec_f > 0){
- close(fds[0]);
- dup2(fds[1], 1);
- }else{
- close(fds[1]);
- dup2(fds[0], 0);
- }
- execvp(argv[1], &argv[1]);
- }
- if (exec_f > 0){
- close(fds[1]);
- dup2(fds[0], 0);
- }else{
- close(fds[0]);
- dup2(fds[1], 1);
- }
- argc = 0;
- break;
- }
+static nkf_encoding* nkf_locale_encoding()
+{
+ nkf_encoding *enc = 0;
+ char *encname = nkf_locale_charmap();
+ if (encname)
+ enc = nkf_enc_find(encname);
+ if (enc < 0) enc = 0;
+ return enc;
+}
#endif
- }
- if (binmode_f == TRUE)
-#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
- if (freopen("","wb",stdout) == NULL)
- return (-1);
+static nkf_encoding* nkf_default_encoding()
+{
+#ifdef DEFAULT_ENCIDX
+ return nkf_enc_from_index(DEFAULT_ENCIDX);
#else
- setbinmode(stdout);
+ nkf_encoding *enc = nkf_locale_encoding();
+ if (enc <= 0) enc = nkf_enc_from_index(ISO_2022_JP);
+ return enc;
#endif
+}
- if (unbuf_f)
- setbuf(stdout, (char *) NULL);
- else
- setvbuffer(stdout, (char *) stdobuf, IOBUF_SIZE);
+#ifndef PERL_XS
+#ifdef WIN32DLL
+#define fprintf dllprintf
+#endif
- if (argc == 0) {
- if (binmode_f == TRUE)
-#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
- if (freopen("","rb",stdin) == NULL) return (-1);
-#else
- setbinmode(stdin);
+void version(void)
+{
+ fprintf(HELP_OUTPUT,"Network Kanji Filter Version " NKF_VERSION " (" NKF_RELEASE_DATE ") \n" COPY_RIGHT "\n");
+}
+
+void usage(void)
+{
+ fprintf(HELP_OUTPUT,
+ "USAGE: nkf(nkf32,wnkf,nkf2) -[flags] [in file] .. [out file for -O flag]\n"
+ "Flags:\n"
+ "b,u Output is buffered (DEFAULT),Output is unbuffered\n"
+ "j,s,e,w Output code is ISO-2022-JP, Shift JIS, EUC-JP, UTF-8N\n"
+#ifdef UTF8_OUTPUT_ENABLE
+ " After 'w' you can add more options. -w[ 8 [0], 16 [[BL] [0]] ]\n"
#endif
- setvbuffer(stdin, (char *) stdibuf, IOBUF_SIZE);
- if (nop_f)
- noconvert(stdin);
- else {
- kanji_convert(stdin);
- if (guess_f) print_guessed_code(NULL);
- }
- } else {
- int nfiles = argc;
- int is_argument_error = FALSE;
- while (argc--) {
- input_codename = NULL;
- input_newline = 0;
-#ifdef CHECK_OPTION
- iconv_for_check = 0;
+ "J,S,E,W Input assumption is JIS 7 bit , Shift JIS, EUC-JP, UTF-8\n"
+#ifdef UTF8_INPUT_ENABLE
+ " After 'W' you can add more options. -W[ 8, 16 [BL] ] \n"
+#endif
+ "t no conversion\n"
+ "i[@B] Specify the Esc Seq for JIS X 0208-1978/83 (DEFAULT B)\n"
+ "o[BJH] Specify the Esc Seq for ASCII/Roman (DEFAULT B)\n"
+ "r {de/en}crypt ROT13/47\n"
+ "h 1 katakana->hiragana, 2 hiragana->katakana, 3 both\n"
+ "m[BQSN0] MIME decode [B:base64,Q:quoted,S:strict,N:non-strict,0:no decode]\n"
+ "M[BQ] MIME encode [B:base64 Q:quoted]\n"
+ "l ISO8859-1 (Latin-1) support\n"
+ "f/F Folding: -f60 or -f or -f60-10 (fold margin 10) F preserve nl\n"
+ "Z[0-4] Default/0: Convert JISX0208 Alphabet to ASCII\n"
+ " 1: Kankaku to one space 2: to two spaces 3: HTML Entity\n"
+ " 4: JISX0208 Katakana to JISX0201 Katakana\n"
+ "X,x Assume X0201 kana in MS-Kanji, -x preserves X0201\n"
+ "B[0-2] Broken input 0: missing ESC,1: any X on ESC-[($]-X,2: ASCII on NL\n"
+#ifdef MSDOS
+ "T Text mode output\n"
+#endif
+ "O Output to File (DEFAULT 'nkf.out')\n"
+ "I Convert non ISO-2022-JP charactor to GETA\n"
+ "d,c Convert line breaks -d: LF -c: CRLF\n"
+ "-L[uwm] line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n"
+ "v, V Show this usage. V: show configuration\n"
+ "\n"
+ "Long name options\n"
+ " --ic=<input codeset> --oc=<output codeset>\n"
+ " Specify the input or output codeset\n"
+ " --fj --unix --mac --windows\n"
+ " --jis --euc --sjis --utf8 --utf16 --mime --base64\n"
+ " Convert for the system or code\n"
+ " --hiragana --katakana --katakana-hiragana\n"
+ " To Hiragana/Katakana Conversion\n"
+ " --prefix= Insert escape before troublesome characters of Shift_JIS\n"
+#ifdef INPUT_OPTION
+ " --cap-input, --url-input Convert hex after ':' or '%%'\n"
#endif
- if ((fin = fopen((origfname = *argv++), "r")) == NULL) {
- perror(*(argv-1));
- is_argument_error = TRUE;
- continue;
- } else {
-#ifdef OVERWRITE
- int fd = 0;
- int fd_backup = 0;
+#ifdef NUMCHAR_OPTION
+ " --numchar-input Convert Unicode Character Reference\n"
#endif
-
-/* reopen file for stdout */
- if (file_out_f == TRUE) {
-#ifdef OVERWRITE
- if (overwrite_f){
- outfname = malloc(strlen(origfname)
- + strlen(".nkftmpXXXXXX")
- + 1);
- if (!outfname){
- perror(origfname);
- return -1;
- }
- strcpy(outfname, origfname);
-#ifdef MSDOS
- {
- int i;
- for (i = strlen(outfname); i; --i){
- if (outfname[i - 1] == '/'
- || outfname[i - 1] == '\\'){
- break;
- }
- }
- outfname[i] = '\0';
- }
- strcat(outfname, "ntXXXXXX");
- mktemp(outfname);
- fd = open(outfname, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL,
- S_IREAD | S_IWRITE);
-#else
- strcat(outfname, ".nkftmpXXXXXX");
- fd = mkstemp(outfname);
+#ifdef UTF8_INPUT_ENABLE
+ " --fb-{skip, html, xml, perl, java, subchar}\n"
+ " Specify how nkf handles unassigned characters\n"
#endif
- if (fd < 0
- || (fd_backup = dup(fileno(stdout))) < 0
- || dup2(fd, fileno(stdout)) < 0
- ){
- perror(origfname);
- return -1;
- }
- }else
+#ifdef OVERWRITE
+ " --in-place[=SUFFIX] --overwrite[=SUFFIX]\n"
+ " Overwrite original listed files by filtered result\n"
+ " --overwrite preserves timestamp of original files\n"
#endif
- if(argc == 1) {
- outfname = *argv++;
- argc--;
- } else {
- outfname = "nkf.out";
- }
+ " -g --guess Guess the input code\n"
+ " --help --version Show this help/the version\n"
+ " For more information, see also man nkf\n"
+ "\n");
+ version();
+}
- if(freopen(outfname, "w", stdout) == NULL) {
- perror (outfname);
- return (-1);
- }
- if (binmode_f == TRUE) {
-#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
- if (freopen("","wb",stdout) == NULL)
- return (-1);
+void show_configuration(void)
+{
+ fprintf(HELP_OUTPUT,
+ "Summary of my nkf " NKF_VERSION " (" NKF_RELEASE_DATE ") configuration:\n"
+ " nkf identity:\n"
+ " " NKF_IDENT "\n"
+ " Compile-time options:\n"
+ " Compiled at: " __DATE__ " " __TIME__ "\n"
+ );
+ fprintf(HELP_OUTPUT,
+ " Default output encoding: "
+#ifdef DEFAULT_ENCIDX
+ "%s\n", nkf_enc_name(nkf_default_encoding())
#else
- setbinmode(stdout);
+ "%s (%s)\n", nkf_locale_encoding() ? "LOCALE" : "DEFAULT",
+ nkf_enc_name(nkf_default_encoding())
#endif
- }
- }
- if (binmode_f == TRUE)
-#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
- if (freopen("","rb",fin) == NULL)
- return (-1);
+ );
+ fprintf(HELP_OUTPUT,
+ " Default output end of line: "
+#if DEFAULT_NEWLINE == CR
+ "CR"
+#elif DEFAULT_NEWLINE == CRLF
+ "CRLF"
#else
- setbinmode(fin);
+ "LF"
#endif
- setvbuffer(fin, (char *) stdibuf, IOBUF_SIZE);
- if (nop_f)
- noconvert(fin);
- else {
- char *filename = NULL;
- kanji_convert(fin);
- if (nfiles > 1) filename = origfname;
- if (guess_f) print_guessed_code(filename);
- }
- fclose(fin);
-#ifdef OVERWRITE
- if (overwrite_f) {
- struct stat sb;
-#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) && !defined(__WATCOMC__) && !defined(__EMX__) && !defined(__OS2__) && !defined(__DJGPP__)
- time_t tb[2];
+ "\n"
+ " Decode MIME encoded string: "
+#if MIME_DECODE_DEFAULT
+ "ON"
#else
- struct utimbuf tb;
+ "OFF"
#endif
-
- fflush(stdout);
- close(fd);
- if (dup2(fd_backup, fileno(stdout)) < 0){
- perror("dup2");
- }
- if (stat(origfname, &sb)) {
- fprintf(stderr, "Can't stat %s\n", origfname);
- }
- /* $B%Q!<%_%C%7%g%s$rI|85(B */
- if (chmod(outfname, sb.st_mode)) {
- fprintf(stderr, "Can't set permission %s\n", outfname);
- }
-
- /* $B%?%$%`%9%?%s%W$rI|85(B */
- if(preserve_time_f){
-#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) && !defined(__WATCOMC__) && !defined(__EMX__) && !defined(__OS2__) && !defined(__DJGPP__)
- tb[0] = tb[1] = sb.st_mtime;
- if (utime(outfname, tb)) {
- fprintf(stderr, "Can't set timestamp %s\n", outfname);
- }
+ "\n"
+ " Convert JIS X 0201 Katakana: "
+#if X0201_DEFAULT
+ "ON"
#else
- tb.actime = sb.st_atime;
- tb.modtime = sb.st_mtime;
- if (utime(outfname, &tb)) {
- fprintf(stderr, "Can't set timestamp %s\n", outfname);
- }
-#endif
- }
- if(backup_f){
- char *backup_filename = get_backup_filename(backup_suffix, origfname);
-#ifdef MSDOS
- unlink(backup_filename);
-#endif
- if (rename(origfname, backup_filename)) {
- perror(backup_filename);
- fprintf(stderr, "Can't rename %s to %s\n",
- origfname, backup_filename);
- }
- }else{
-#ifdef MSDOS
- if (unlink(origfname)){
- perror(origfname);
- }
+ "OFF"
#endif
- }
- if (rename(outfname, origfname)) {
- perror(origfname);
- fprintf(stderr, "Can't rename %s to %s\n",
- outfname, origfname);
- }
- free(outfname);
- }
+ "\n"
+ " --help, --version output: "
+#if HELP_OUTPUT_HELP_OUTPUT
+ "HELP_OUTPUT"
+#else
+ "STDOUT"
#endif
- }
- }
- if (is_argument_error)
- return(-1);
- }
-#ifdef EASYWIN /*Easy Win */
- if (file_out_f == FALSE)
- scanf("%d",&end_check);
- else
- fclose(stdout);
-#else /* for Other OS */
- if (file_out_f == TRUE)
- fclose(stdout);
-#endif /*Easy Win */
- return (0);
+ "\n");
}
-#endif /* WIN32DLL */
+#endif /*PERL_XS*/
#ifdef OVERWRITE
char *get_backup_filename(const char *suffix, const char *filename)
@@ -1272,6 +919,95 @@ char *get_backup_filename(const char *suffix, const char *filename)
}
#endif
+#ifdef UTF8_INPUT_ENABLE
+void nkf_each_char_to_hex(void (*f)(nkf_char c2,nkf_char c1), nkf_char c)
+{
+ int shift = 20;
+ c &= VALUE_MASK;
+ while(shift >= 0){
+ if(c >= 1<<shift){
+ while(shift >= 0){
+ (*f)(0, bin2hex(c>>shift));
+ shift -= 4;
+ }
+ }else{
+ shift -= 4;
+ }
+ }
+ return;
+}
+
+void encode_fallback_html(nkf_char c)
+{
+ (*oconv)(0, '&');
+ (*oconv)(0, '#');
+ c &= VALUE_MASK;
+ if(c >= NKF_INT32_C(1000000))
+ (*oconv)(0, 0x30+(c/NKF_INT32_C(1000000))%10);
+ if(c >= NKF_INT32_C(100000))
+ (*oconv)(0, 0x30+(c/NKF_INT32_C(100000) )%10);
+ if(c >= 10000)
+ (*oconv)(0, 0x30+(c/10000 )%10);
+ if(c >= 1000)
+ (*oconv)(0, 0x30+(c/1000 )%10);
+ if(c >= 100)
+ (*oconv)(0, 0x30+(c/100 )%10);
+ if(c >= 10)
+ (*oconv)(0, 0x30+(c/10 )%10);
+ if(c >= 0)
+ (*oconv)(0, 0x30+ c %10);
+ (*oconv)(0, ';');
+ return;
+}
+
+void encode_fallback_xml(nkf_char c)
+{
+ (*oconv)(0, '&');
+ (*oconv)(0, '#');
+ (*oconv)(0, 'x');
+ nkf_each_char_to_hex(oconv, c);
+ (*oconv)(0, ';');
+ return;
+}
+
+void encode_fallback_java(nkf_char c)
+{
+ (*oconv)(0, '\\');
+ c &= VALUE_MASK;
+ if(!is_unicode_bmp(c)){
+ (*oconv)(0, 'U');
+ (*oconv)(0, '0');
+ (*oconv)(0, '0');
+ (*oconv)(0, bin2hex(c>>20));
+ (*oconv)(0, bin2hex(c>>16));
+ }else{
+ (*oconv)(0, 'u');
+ }
+ (*oconv)(0, bin2hex(c>>12));
+ (*oconv)(0, bin2hex(c>> 8));
+ (*oconv)(0, bin2hex(c>> 4));
+ (*oconv)(0, bin2hex(c ));
+ return;
+}
+
+void encode_fallback_perl(nkf_char c)
+{
+ (*oconv)(0, '\\');
+ (*oconv)(0, 'x');
+ (*oconv)(0, '{');
+ nkf_each_char_to_hex(oconv, c);
+ (*oconv)(0, '}');
+ return;
+}
+
+void encode_fallback_subchar(nkf_char c)
+{
+ c = unicode_subchar;
+ (*oconv)((c>>8)&0xFF, c&0xFF);
+ return;
+}
+#endif
+
static const struct {
const char *name;
const char *alias;
@@ -1352,771 +1088,243 @@ static const struct {
{"prefix=", ""},
};
-static int option_mode = 0;
-
-void options(unsigned char *cp)
+static void set_input_encoding(nkf_encoding *enc)
{
- nkf_char i, j;
- unsigned char *p;
- unsigned char *cp_back = NULL;
- char codeset[32];
- nkf_encoding *enc;
-
- if (option_mode==1)
- return;
- while(*cp && *cp++!='-');
- while (*cp || cp_back) {
- if(!*cp){
- cp = cp_back;
- cp_back = NULL;
- continue;
- }
- p = 0;
- switch (*cp++) {
- case '-': /* literal options */
- if (!*cp || *cp == SP) { /* ignore the rest of arguments */
- option_mode = 1;
- return;
- }
- for (i=0;i<sizeof(long_option)/sizeof(long_option[0]);i++) {
- p = (unsigned char *)long_option[i].name;
- for (j=0;*p && *p != '=' && *p == cp[j];p++, j++);
- if (*p == cp[j] || cp[j] == SP){
- p = &cp[j] + 1;
- break;
- }
- p = 0;
- }
- if (p == 0) {
- fprintf(stderr, "unknown long option: --%s\n", cp);
- return;
- }
- while(*cp && *cp != SP && cp++);
- if (long_option[i].alias[0]){
- cp_back = cp;
- cp = (unsigned char *)long_option[i].alias;
- }else{
- if (strcmp(long_option[i].name, "ic=") == 0){
- nkf_str_upcase((char *)p, codeset, 32);
- enc = nkf_enc_find(codeset);
- if (!enc) continue;
- input_encoding = enc;
- switch (nkf_enc_to_index(input_encoding)) {
- case CP50220:
- case CP50221:
- case CP50222:
+ switch (nkf_enc_to_index(enc)) {
+ case CP50220:
+ case CP50221:
+ case CP50222:
#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
+ cp51932_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case ISO_2022_JP_1:
+ break;
+ case ISO_2022_JP_1:
#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ x0212_f = TRUE;
#endif
- break;
- case ISO_2022_JP_3:
+ break;
+ case ISO_2022_JP_3:
#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ x0212_f = TRUE;
#endif
- x0213_f = TRUE;
- break;
- case WINDOWS_31J:
+ x0213_f = TRUE;
+ break;
+ case SHIFT_JIS:
+ break;
+ case WINDOWS_31J:
#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
+ cp51932_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case CP10001:
+ break;
+ case EUC_JP:
+ break;
+ case CP10001:
#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
+ cp51932_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP10001;
+ ms_ucs_map_f = UCS_MAP_CP10001;
#endif
- break;
- case CP51932:
+ break;
+ case CP51932:
#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
+ cp51932_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case EUCJP_MS:
+ break;
+ case EUCJP_MS:
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp51932_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_MS;
+ ms_ucs_map_f = UCS_MAP_MS;
#endif
- break;
- case EUCJP_ASCII:
+ break;
+ case EUCJP_ASCII:
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp51932_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_ASCII;
+ ms_ucs_map_f = UCS_MAP_ASCII;
#endif
- break;
- case SHIFT_JISX0213:
- case SHIFT_JIS_2004:
- x0213_f = TRUE;
+ break;
+ case SHIFT_JISX0213:
+ case SHIFT_JIS_2004:
+ x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp51932_f = FALSE;
#endif
- break;
- case EUC_JISX0213:
- case EUC_JIS_2004:
- x0213_f = TRUE;
+ break;
+ case EUC_JISX0213:
+ case EUC_JIS_2004:
+ x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp51932_f = FALSE;
#endif
- break;
+ break;
#ifdef UTF8_INPUT_ENABLE
#ifdef UNICODE_NORMALIZATION
- case UTF8_MAC:
- nfc_f = TRUE;
- break;
+ case UTF8_MAC:
+ nfc_f = TRUE;
+ break;
#endif
- case UTF_16:
- case UTF_16BE:
- case UTF_16BE_BOM:
- input_endian = ENDIAN_BIG;
- break;
- case UTF_16LE:
- case UTF_16LE_BOM:
- input_endian = ENDIAN_LITTLE;
- break;
- case UTF_32:
- case UTF_32BE:
- case UTF_32BE_BOM:
- input_endian = ENDIAN_BIG;
- break;
- case UTF_32LE:
- case UTF_32LE_BOM:
- input_endian = ENDIAN_LITTLE;
- break;
+ case UTF_16:
+ case UTF_16BE:
+ case UTF_16BE_BOM:
+ input_endian = ENDIAN_BIG;
+ break;
+ case UTF_16LE:
+ case UTF_16LE_BOM:
+ input_endian = ENDIAN_LITTLE;
+ break;
+ case UTF_32:
+ case UTF_32BE:
+ case UTF_32BE_BOM:
+ input_endian = ENDIAN_BIG;
+ break;
+ case UTF_32LE:
+ case UTF_32LE_BOM:
+ input_endian = ENDIAN_LITTLE;
+ break;
#endif
- }
- continue;
- }
- if (strcmp(long_option[i].name, "oc=") == 0){
- x0201_f = FALSE;
- nkf_str_upcase((char *)p, codeset, 32);
- enc = nkf_enc_find(codeset);
- if (enc <= 0) continue;
- output_encoding = enc;
- switch (nkf_enc_to_index(output_encoding)) {
- case CP50220:
- x0201_f = TRUE;
+ }
+}
+
+static void set_output_encoding(nkf_encoding *enc)
+{
+ switch (nkf_enc_to_index(enc)) {
+ case CP50220:
+ x0201_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case CP50221:
+ break;
+ case CP50221:
#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case ISO_2022_JP_1:
+ break;
+ case ISO_2022_JP_1:
#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ x0212_f = TRUE;
#endif
#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
- break;
- case ISO_2022_JP_3:
+ break;
+ case ISO_2022_JP_3:
#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ x0212_f = TRUE;
#endif
- x0213_f = TRUE;
+ x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
- break;
- case WINDOWS_31J:
+ break;
+ case SHIFT_JIS:
+ break;
+ case WINDOWS_31J:
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case CP10001:
+ break;
+ case CP10001:
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP10001;
+ ms_ucs_map_f = UCS_MAP_CP10001;
#endif
- break;
- case CP51932:
+ break;
+ case EUC_JP:
+ x0212_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case EUCJP_MS:
-#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ break;
+ case CP51932:
+#ifdef SHIFTJIS_CP932
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_MS;
+ ms_ucs_map_f = UCS_MAP_CP932;
#endif
- break;
- case EUCJP_ASCII:
+ break;
+ case EUCJP_MS:
#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ x0212_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_ASCII;
+ ms_ucs_map_f = UCS_MAP_MS;
#endif
- break;
- case SHIFT_JISX0213:
- case SHIFT_JIS_2004:
- x0213_f = TRUE;
-#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
-#endif
- break;
- case EUC_JISX0213:
- case EUC_JIS_2004:
+ break;
+ case EUCJP_ASCII:
#ifdef X0212_ENABLE
- x0212_f = TRUE;
-#endif
- x0213_f = TRUE;
-#ifdef SHIFTJIS_CP932
- cp932inv_f = FALSE;
-#endif
- break;
-#ifdef UTF8_OUTPUT_ENABLE
- case UTF_8_BOM:
- output_bom_f = TRUE;
- break;
- case UTF_16:
- case UTF_16BE_BOM:
- output_bom_f = TRUE;
- break;
- case UTF_16LE:
- output_endian = ENDIAN_LITTLE;
- output_bom_f = FALSE;
- break;
- case UTF_16LE_BOM:
- output_endian = ENDIAN_LITTLE;
- output_bom_f = TRUE;
- break;
- case UTF_32BE_BOM:
- output_bom_f = TRUE;
- break;
- case UTF_32LE:
- output_endian = ENDIAN_LITTLE;
- output_bom_f = FALSE;
- break;
- case UTF_32LE_BOM:
- output_endian = ENDIAN_LITTLE;
- output_bom_f = TRUE;
- break;
-#endif
- }
- continue;
- }
- if (strcmp(long_option[i].name, "guess=") == 0){
- if (p[0] == '0' || p[0] == '1') {
- guess_f = 1;
- } else {
- guess_f = 2;
- }
- continue;
- }
-#ifdef OVERWRITE
- if (strcmp(long_option[i].name, "overwrite") == 0){
- file_out_f = TRUE;
- overwrite_f = TRUE;
- preserve_time_f = TRUE;
- continue;
- }
- if (strcmp(long_option[i].name, "overwrite=") == 0){
- file_out_f = TRUE;
- overwrite_f = TRUE;
- preserve_time_f = TRUE;
- backup_f = TRUE;
- backup_suffix = malloc(strlen((char *) p) + 1);
- strcpy(backup_suffix, (char *) p);
- continue;
- }
- if (strcmp(long_option[i].name, "in-place") == 0){
- file_out_f = TRUE;
- overwrite_f = TRUE;
- preserve_time_f = FALSE;
- continue;
- }
- if (strcmp(long_option[i].name, "in-place=") == 0){
- file_out_f = TRUE;
- overwrite_f = TRUE;
- preserve_time_f = FALSE;
- backup_f = TRUE;
- backup_suffix = malloc(strlen((char *) p) + 1);
- strcpy(backup_suffix, (char *) p);
- continue;
- }
-#endif
-#ifdef INPUT_OPTION
- if (strcmp(long_option[i].name, "cap-input") == 0){
- cap_f = TRUE;
- continue;
- }
- if (strcmp(long_option[i].name, "url-input") == 0){
- url_f = TRUE;
- continue;
- }
-#endif
-#ifdef NUMCHAR_OPTION
- if (strcmp(long_option[i].name, "numchar-input") == 0){
- numchar_f = TRUE;
- continue;
- }
-#endif
-#ifdef CHECK_OPTION
- if (strcmp(long_option[i].name, "no-output") == 0){
- noout_f = TRUE;
- continue;
- }
- if (strcmp(long_option[i].name, "debug") == 0){
- debug_f = TRUE;
- continue;
- }
-#endif
- if (strcmp(long_option[i].name, "cp932") == 0){
-#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
- cp932inv_f = TRUE;
-#endif
-#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_CP932;
-#endif
- continue;
- }
- if (strcmp(long_option[i].name, "no-cp932") == 0){
-#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
- cp932inv_f = FALSE;
+ x0212_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- ms_ucs_map_f = UCS_MAP_ASCII;
+ ms_ucs_map_f = UCS_MAP_ASCII;
#endif
- continue;
- }
+ break;
+ case SHIFT_JISX0213:
+ case SHIFT_JIS_2004:
+ x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- if (strcmp(long_option[i].name, "cp932inv") == 0){
- cp932inv_f = TRUE;
- continue;
- }
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
-
+ break;
+ case EUC_JISX0213:
+ case EUC_JIS_2004:
#ifdef X0212_ENABLE
- if (strcmp(long_option[i].name, "x0212") == 0){
- x0212_f = TRUE;
- continue;
- }
+ x0212_f = TRUE;
#endif
-
-#ifdef EXEC_IO
- if (strcmp(long_option[i].name, "exec-in") == 0){
- exec_f = 1;
- return;
- }
- if (strcmp(long_option[i].name, "exec-out") == 0){
- exec_f = -1;
- return;
- }
-#endif
-#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
- if (strcmp(long_option[i].name, "no-cp932ext") == 0){
- no_cp932ext_f = TRUE;
- continue;
- }
- if (strcmp(long_option[i].name, "no-best-fit-chars") == 0){
- no_best_fit_chars_f = TRUE;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-skip") == 0){
- encode_fallback = NULL;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-html") == 0){
- encode_fallback = encode_fallback_html;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-xml") == 0){
- encode_fallback = encode_fallback_xml;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-java") == 0){
- encode_fallback = encode_fallback_java;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-perl") == 0){
- encode_fallback = encode_fallback_perl;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-subchar") == 0){
- encode_fallback = encode_fallback_subchar;
- continue;
- }
- if (strcmp(long_option[i].name, "fb-subchar=") == 0){
- encode_fallback = encode_fallback_subchar;
- unicode_subchar = 0;
- if (p[0] != '0'){
- /* decimal number */
- for (i = 0; i < 7 && nkf_isdigit(p[i]); i++){
- unicode_subchar *= 10;
- unicode_subchar += hex2bin(p[i]);
- }
- }else if(p[1] == 'x' || p[1] == 'X'){
- /* hexadecimal number */
- for (i = 2; i < 8 && nkf_isxdigit(p[i]); i++){
- unicode_subchar <<= 4;
- unicode_subchar |= hex2bin(p[i]);
- }
- }else{
- /* octal number */
- for (i = 1; i < 8 && nkf_isoctal(p[i]); i++){
- unicode_subchar *= 8;
- unicode_subchar += hex2bin(p[i]);
- }
- }
- w16e_conv(unicode_subchar, &i, &j);
- unicode_subchar = i<<8 | j;
- continue;
- }
-#endif
-#ifdef UTF8_OUTPUT_ENABLE
- if (strcmp(long_option[i].name, "ms-ucs-map") == 0){
- ms_ucs_map_f = UCS_MAP_MS;
- continue;
- }
-#endif
-#ifdef UNICODE_NORMALIZATION
- if (strcmp(long_option[i].name, "utf8mac-input") == 0){
- nfc_f = TRUE;
- continue;
- }
-#endif
- if (strcmp(long_option[i].name, "prefix=") == 0){
- if (nkf_isgraph(p[0])){
- for (i = 1; nkf_isgraph(p[i]); i++){
- prefix_table[p[i]] = p[0];
- }
- }
- continue;
- }
- }
- continue;
- case 'b': /* buffered mode */
- unbuf_f = FALSE;
- continue;
- case 'u': /* non bufferd mode */
- unbuf_f = TRUE;
- continue;
- case 't': /* transparent mode */
- if (*cp=='1') {
- /* alias of -t */
- cp++;
- nop_f = TRUE;
- } else if (*cp=='2') {
- /*
- * -t with put/get
- *
- * nkf -t2MB hoge.bin | nkf -t2mB | diff -s - hoge.bin
- *
- */
- cp++;
- nop_f = 2;
- } else
- nop_f = TRUE;
- continue;
- case 'j': /* JIS output */
- case 'n':
- output_encoding = nkf_enc_from_index(ISO_2022_JP);
- continue;
- case 'e': /* AT&T EUC output */
- cp932inv_f = FALSE;
- output_encoding = nkf_enc_from_index(EUC_JP);
- continue;
- case 's': /* SJIS output */
- output_encoding = nkf_enc_from_index(WINDOWS_31J);
- continue;
- case 'l': /* ISO8859 Latin-1 support, no conversion */
- iso8859_f = TRUE; /* Only compatible with ISO-2022-JP */
- input_encoding = nkf_enc_from_index(ISO_8859_1);
- continue;
- case 'i': /* Kanji IN ESC-$-@/B */
- if (*cp=='@'||*cp=='B')
- kanji_intro = *cp++;
- continue;
- case 'o': /* ASCII IN ESC-(-J/B */
- if (*cp=='J'||*cp=='B'||*cp=='H')
- ascii_intro = *cp++;
- continue;
- case 'h':
- /*
- bit:1 katakana->hiragana
- bit:2 hiragana->katakana
- */
- if ('9'>= *cp && *cp>='0')
- hira_f |= (*cp++ -'0');
- else
- hira_f |= 1;
- continue;
- case 'r':
- rot_f = TRUE;
- continue;
-#if defined(MSDOS) || defined(__OS2__)
- case 'T':
- binmode_f = FALSE;
- continue;
-#endif
-#ifndef PERL_XS
- case 'V':
- show_configuration();
- exit(1);
- break;
- case 'v':
- usage();
- exit(1);
- break;
+ x0213_f = TRUE;
+#ifdef SHIFTJIS_CP932
+ if (cp932inv_f == TRUE) cp932inv_f = FALSE;
#endif
+ break;
#ifdef UTF8_OUTPUT_ENABLE
- case 'w': /* UTF-8 output */
- if (cp[0] == '8') {
- cp++;
- if (cp[0] == '0'){
- cp++;
- output_encoding = nkf_enc_from_index(UTF_8N);
- } else {
- output_bom_f = TRUE;
- output_encoding = nkf_enc_from_index(UTF_8_BOM);
- }
- } else {
- int enc_idx;
- if ('1'== cp[0] && '6'==cp[1]) {
- cp += 2;
- enc_idx = UTF_16;
- } else if ('3'== cp[0] && '2'==cp[1]) {
- cp += 2;
- enc_idx = UTF_32;
- } else {
- output_encoding = nkf_enc_from_index(UTF_8);
- continue;
- }
- if (cp[0]=='L') {
- cp++;
- output_endian = ENDIAN_LITTLE;
- } else if (cp[0] == 'B') {
- cp++;
- } else {
- output_encoding = nkf_enc_from_index(enc_idx);
- continue;
- }
- if (cp[0] == '0'){
- cp++;
- enc_idx = enc_idx == UTF_16
- ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
- : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
- } else {
- output_bom_f = TRUE;
- enc_idx = enc_idx == UTF_16
- ? (output_endian == ENDIAN_LITTLE ? UTF_16LE_BOM : UTF_16BE_BOM)
- : (output_endian == ENDIAN_LITTLE ? UTF_32LE_BOM : UTF_32BE_BOM);
- }
- output_encoding = nkf_enc_from_index(enc_idx);
- }
- continue;
-#endif
-#ifdef UTF8_INPUT_ENABLE
- case 'W': /* UTF input */
- if (cp[0] == '8') {
- cp++;
- input_encoding = nkf_enc_from_index(UTF_8);
- }else{
- int enc_idx;
- if ('1'== cp[0] && '6'==cp[1]) {
- cp += 2;
- input_endian = ENDIAN_BIG;
- enc_idx = UTF_16;
- } else if ('3'== cp[0] && '2'==cp[1]) {
- cp += 2;
- input_endian = ENDIAN_BIG;
- enc_idx = UTF_32;
- } else {
- input_encoding = nkf_enc_from_index(UTF_8);
- continue;
- }
- if (cp[0]=='L') {
- cp++;
- input_endian = ENDIAN_LITTLE;
- } else if (cp[0] == 'B') {
- cp++;
- input_endian = ENDIAN_BIG;
- }
- enc_idx = enc_idx == UTF_16
- ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
- : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
- input_encoding = nkf_enc_from_index(enc_idx);
- }
- continue;
-#endif
- /* Input code assumption */
- case 'J': /* ISO-2022-JP input */
- input_encoding = nkf_enc_from_index(ISO_2022_JP);
- continue;
- case 'E': /* EUC-JP input */
- input_encoding = nkf_enc_from_index(EUC_JP);
- continue;
- case 'S': /* Windows-31J input */
- input_encoding = nkf_enc_from_index(WINDOWS_31J);
- continue;
- case 'Z': /* Convert X0208 alphabet to asii */
- /* alpha_f
- bit:0 Convert JIS X 0208 Alphabet to ASCII
- bit:1 Convert Kankaku to one space
- bit:2 Convert Kankaku to two spaces
- bit:3 Convert HTML Entity
- bit:4 Convert JIS X 0208 Katakana to JIS X 0201 Katakana
- */
- while ('0'<= *cp && *cp <='9') {
- alpha_f |= 1 << (*cp++ - '0');
- }
- if (!alpha_f) alpha_f = 1;
- continue;
- case 'x': /* Convert X0201 kana to X0208 or X0201 Conversion */
- x0201_f = FALSE; /* No X0201->X0208 conversion */
- /* accept X0201
- ESC-(-I in JIS, EUC, MS Kanji
- SI/SO in JIS, EUC, MS Kanji
- SSO in EUC, JIS, not in MS Kanji
- MS Kanji (0xa0-0xdf)
- output X0201
- ESC-(-I in JIS (0x20-0x5f)
- SSO in EUC (0xa0-0xdf)
- 0xa0-0xd in MS Kanji (0xa0-0xdf)
- */
- continue;
- case 'X': /* Convert X0201 kana to X0208 */
- x0201_f = TRUE;
- continue;
- case 'F': /* prserve new lines */
- fold_preserve_f = TRUE;
- case 'f': /* folding -f60 or -f */
- fold_f = TRUE;
- fold_len = 0;
- while('0'<= *cp && *cp <='9') { /* we don't use atoi here */
- fold_len *= 10;
- fold_len += *cp++ - '0';
- }
- if (!(0<fold_len && fold_len<BUFSIZ))
- fold_len = DEFAULT_FOLD;
- if (*cp=='-') {
- fold_margin = 0;
- cp++;
- while('0'<= *cp && *cp <='9') { /* we don't use atoi here */
- fold_margin *= 10;
- fold_margin += *cp++ - '0';
- }
- }
- continue;
- case 'm': /* MIME support */
- /* mime_decode_f = TRUE; */ /* this has too large side effects... */
- if (*cp=='B'||*cp=='Q') {
- mime_decode_mode = *cp++;
- mimebuf_f = FIXED_MIME;
- } else if (*cp=='N') {
- mime_f = TRUE; cp++;
- } else if (*cp=='S') {
- mime_f = STRICT_MIME; cp++;
- } else if (*cp=='0') {
- mime_decode_f = FALSE;
- mime_f = FALSE; cp++;
- } else {
- mime_f = STRICT_MIME;
- }
- continue;
- case 'M': /* MIME output */
- if (*cp=='B') {
- mimeout_mode = 'B';
- mimeout_f = FIXED_MIME; cp++;
- } else if (*cp=='Q') {
- mimeout_mode = 'Q';
- mimeout_f = FIXED_MIME; cp++;
- } else {
- mimeout_f = TRUE;
- }
- continue;
- case 'B': /* Broken JIS support */
- /* bit:0 no ESC JIS
- bit:1 allow any x on ESC-(-x or ESC-$-x
- bit:2 reset to ascii on NL
- */
- if ('9'>= *cp && *cp>='0')
- broken_f |= 1<<(*cp++ -'0');
- else
- broken_f |= TRUE;
- continue;
-#ifndef PERL_XS
- case 'O':/* for Output file */
- file_out_f = TRUE;
- continue;
-#endif
- case 'c':/* add cr code */
- nlmode_f = CRLF;
- continue;
- case 'd':/* delete cr code */
- nlmode_f = LF;
- continue;
- case 'I': /* ISO-2022-JP output */
- iso2022jp_f = TRUE;
- continue;
- case 'L': /* line mode */
- if (*cp=='u') { /* unix */
- nlmode_f = LF; cp++;
- } else if (*cp=='m') { /* mac */
- nlmode_f = CR; cp++;
- } else if (*cp=='w') { /* windows */
- nlmode_f = CRLF; cp++;
- } else if (*cp=='0') { /* no conversion */
- nlmode_f = 0; cp++;
- }
- continue;
-#ifndef PERL_XS
- case 'g':
- if ('2' <= *cp && *cp <= '9') {
- guess_f = 2;
- cp++;
- } else if (*cp == '0' || *cp == '1') {
- guess_f = 1;
- cp++;
- } else {
- guess_f = 1;
- }
- continue;
+ case UTF_8_BOM:
+ output_bom_f = TRUE;
+ break;
+ case UTF_16:
+ case UTF_16BE_BOM:
+ output_bom_f = TRUE;
+ break;
+ case UTF_16LE:
+ output_endian = ENDIAN_LITTLE;
+ output_bom_f = FALSE;
+ break;
+ case UTF_16LE_BOM:
+ output_endian = ENDIAN_LITTLE;
+ output_bom_f = TRUE;
+ break;
+ case UTF_32BE_BOM:
+ output_bom_f = TRUE;
+ break;
+ case UTF_32LE:
+ output_endian = ENDIAN_LITTLE;
+ output_bom_f = FALSE;
+ break;
+ case UTF_32LE_BOM:
+ output_endian = ENDIAN_LITTLE;
+ output_bom_f = TRUE;
+ break;
#endif
- case SP:
- /* module muliple options in a string are allowed for Perl moudle */
- while(*cp && *cp++!='-');
- continue;
- default:
- fprintf(stderr, "unknown option: -%c\n", *(cp-1));
- /* bogus option but ignored */
- continue;
- }
}
}
@@ -2162,1177 +1370,76 @@ void set_iconv(nkf_char f, nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_ch
#endif
}
-#define SCORE_L2 (1) /* $BBh(B2$B?e=`4A;z(B */
-#define SCORE_KANA (SCORE_L2 << 1) /* $B$$$o$f$kH>3Q%+%J(B */
-#define SCORE_DEPEND (SCORE_KANA << 1) /* $B5!<o0MB8J8;z(B */
-#define SCORE_CP932 (SCORE_DEPEND << 1) /* CP932 $B$K$h$kFI$_49$((B (IBM extended characters) */
-#define SCORE_X0212 (SCORE_CP932 << 1) /* JIS X 0212 */
-#define SCORE_NO_EXIST (SCORE_X0212 << 1) /* $BB8:_$7$J$$J8;z(B */
-#define SCORE_iMIME (SCORE_NO_EXIST << 1) /* MIME $B$K$h$k;XDj(B */
-#define SCORE_ERROR (SCORE_iMIME << 1) /* $B%(%i!<(B */
-
-#define SCORE_INIT (SCORE_iMIME)
-
-static const char score_table_A0[] = {
- 0, 0, 0, 0,
- 0, 0, 0, 0,
- 0, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
- SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_NO_EXIST,
-};
-
-static const char score_table_F0[] = {
- SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
- SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
- SCORE_DEPEND, SCORE_DEPEND, SCORE_CP932, SCORE_CP932,
- SCORE_CP932, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR,
-};
-
-void set_code_score(struct input_code *ptr, nkf_char score)
-{
- if (ptr){
- ptr->score |= score;
- }
-}
-
-void clr_code_score(struct input_code *ptr, nkf_char score)
-{
- if (ptr){
- ptr->score &= ~score;
- }
-}
-
-void code_score(struct input_code *ptr)
-{
- nkf_char c2 = ptr->buf[0];
-#ifdef UTF8_OUTPUT_ENABLE
- nkf_char c1 = ptr->buf[1];
-#endif
- if (c2 < 0){
- set_code_score(ptr, SCORE_ERROR);
- }else if (c2 == SSO){
- set_code_score(ptr, SCORE_KANA);
- }else if (c2 == 0x8f){
- set_code_score(ptr, SCORE_X0212);
-#ifdef UTF8_OUTPUT_ENABLE
- }else if (!e2w_conv(c2, c1)){
- set_code_score(ptr, SCORE_NO_EXIST);
-#endif
- }else if ((c2 & 0x70) == 0x20){
- set_code_score(ptr, score_table_A0[c2 & 0x0f]);
- }else if ((c2 & 0x70) == 0x70){
- set_code_score(ptr, score_table_F0[c2 & 0x0f]);
- }else if ((c2 & 0x70) >= 0x50){
- set_code_score(ptr, SCORE_L2);
- }
-}
-
-void status_disable(struct input_code *ptr)
-{
- ptr->stat = -1;
- ptr->buf[0] = -1;
- code_score(ptr);
- if (iconv == ptr->iconv_func) set_iconv(FALSE, 0);
-}
-
-void status_push_ch(struct input_code *ptr, nkf_char c)
-{
- ptr->buf[ptr->index++] = c;
-}
-
-void status_clear(struct input_code *ptr)
-{
- ptr->stat = 0;
- ptr->index = 0;
-}
-
-void status_reset(struct input_code *ptr)
-{
- status_clear(ptr);
- ptr->score = SCORE_INIT;
-}
-
-void status_reinit(struct input_code *ptr)
-{
- status_reset(ptr);
- ptr->_file_stat = 0;
-}
-
-void status_check(struct input_code *ptr, nkf_char c)
-{
- if (c <= DEL && estab_f){
- status_reset(ptr);
- }
-}
-
-void s_status(struct input_code *ptr, nkf_char c)
-{
- switch(ptr->stat){
- case -1:
- status_check(ptr, c);
- break;
- case 0:
- if (c <= DEL){
- break;
-#ifdef NUMCHAR_OPTION
- }else if (is_unicode_capsule(c)){
- break;
-#endif
- }else if (0xa1 <= c && c <= 0xdf){
- status_push_ch(ptr, SSO);
- status_push_ch(ptr, c);
- code_score(ptr);
- status_clear(ptr);
- }else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xea)){
- ptr->stat = 1;
- status_push_ch(ptr, c);
- }else if (0xed <= c && c <= 0xee){
- ptr->stat = 3;
- status_push_ch(ptr, c);
-#ifdef SHIFTJIS_CP932
- }else if (is_ibmext_in_sjis(c)){
- ptr->stat = 2;
- status_push_ch(ptr, c);
-#endif /* SHIFTJIS_CP932 */
-#ifdef X0212_ENABLE
- }else if (0xf0 <= c && c <= 0xfc){
- ptr->stat = 1;
- status_push_ch(ptr, c);
-#endif /* X0212_ENABLE */
- }else{
- status_disable(ptr);
- }
- break;
- case 1:
- if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
- status_push_ch(ptr, c);
- s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]);
- code_score(ptr);
- status_clear(ptr);
- }else{
- status_disable(ptr);
- }
- break;
- case 2:
-#ifdef SHIFTJIS_CP932
- if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)) {
- status_push_ch(ptr, c);
- if (s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]) == 0) {
- set_code_score(ptr, SCORE_CP932);
- status_clear(ptr);
- break;
- }
- }
-#endif /* SHIFTJIS_CP932 */
- status_disable(ptr);
- break;
- case 3:
- if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
- status_push_ch(ptr, c);
- s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]);
- set_code_score(ptr, SCORE_CP932);
- status_clear(ptr);
- }else{
- status_disable(ptr);
- }
- break;
- }
-}
-
-void e_status(struct input_code *ptr, nkf_char c)
-{
- switch (ptr->stat){
- case -1:
- status_check(ptr, c);
- break;
- case 0:
- if (c <= DEL){
- break;
-#ifdef NUMCHAR_OPTION
- }else if (is_unicode_capsule(c)){
- break;
-#endif
- }else if (SSO == c || (0xa1 <= c && c <= 0xfe)){
- ptr->stat = 1;
- status_push_ch(ptr, c);
#ifdef X0212_ENABLE
- }else if (0x8f == c){
- ptr->stat = 2;
- status_push_ch(ptr, c);
-#endif /* X0212_ENABLE */
- }else{
- status_disable(ptr);
- }
- break;
- case 1:
- if (0xa1 <= c && c <= 0xfe){
- status_push_ch(ptr, c);
- code_score(ptr);
- status_clear(ptr);
- }else{
- status_disable(ptr);
- }
- break;
-#ifdef X0212_ENABLE
- case 2:
- if (0xa1 <= c && c <= 0xfe){
- ptr->stat = 1;
- status_push_ch(ptr, c);
- }else{
- status_disable(ptr);
- }
-#endif /* X0212_ENABLE */
- }
-}
-
-#ifdef UTF8_INPUT_ENABLE
-void w_status(struct input_code *ptr, nkf_char c)
-{
- switch (ptr->stat){
- case -1:
- status_check(ptr, c);
- break;
- case 0:
- if (c <= DEL){
- break;
-#ifdef NUMCHAR_OPTION
- }else if (is_unicode_capsule(c)){
- break;
-#endif
- }else if (0xc0 <= c && c <= 0xdf){
- ptr->stat = 1;
- status_push_ch(ptr, c);
- }else if (0xe0 <= c && c <= 0xef){
- ptr->stat = 2;
- status_push_ch(ptr, c);
- }else if (0xf0 <= c && c <= 0xf4){
- ptr->stat = 3;
- status_push_ch(ptr, c);
- }else{
- status_disable(ptr);
- }
- break;
- case 1:
- case 2:
- if (0x80 <= c && c <= 0xbf){
- status_push_ch(ptr, c);
- if (ptr->index > ptr->stat){
- int bom = (ptr->buf[0] == 0xef && ptr->buf[1] == 0xbb
- && ptr->buf[2] == 0xbf);
- w2e_conv(ptr->buf[0], ptr->buf[1], ptr->buf[2],
- &ptr->buf[0], &ptr->buf[1]);
- if (!bom){
- code_score(ptr);
- }
- status_clear(ptr);
- }
- }else{
- status_disable(ptr);
- }
- break;
- case 3:
- if (0x80 <= c && c <= 0xbf){
- if (ptr->index < ptr->stat){
- status_push_ch(ptr, c);
- } else {
- status_clear(ptr);
- }
- }else{
- status_disable(ptr);
- }
- break;
- }
-}
-#endif
-
-void code_status(nkf_char c)
+nkf_char x0212_shift(nkf_char c)
{
- int action_flag = 1;
- struct input_code *result = 0;
- struct input_code *p = input_code_list;
- while (p->name){
- if (!p->status_func) {
- ++p;
- continue;
- }
- if (!p->status_func)
- continue;
- (p->status_func)(p, c);
- if (p->stat > 0){
- action_flag = 0;
- }else if(p->stat == 0){
- if (result){
- action_flag = 0;
- }else{
- result = p;
- }
+ nkf_char ret = c;
+ c &= 0x7f;
+ if (is_eucg3(ret)){
+ if (0x75 <= c && c <= 0x7f){
+ ret = c + (0x109 - 0x75);
}
- ++p;
- }
-
- if (action_flag){
- if (result && !estab_f){
- set_iconv(TRUE, result->iconv_func);
- }else if (c <= DEL){
- struct input_code *ptr = input_code_list;
- while (ptr->name){
- status_reset(ptr);
- ++ptr;
- }
+ }else{
+ if (0x75 <= c && c <= 0x7f){
+ ret = c + (0x113 - 0x75);
}
}
+ return ret;
}
-#ifndef WIN32DLL
-nkf_char std_getc(FILE *f)
-{
- if (std_gc_ndx){
- return std_gc_buf[--std_gc_ndx];
- }
- return getc(f);
-}
-#endif /*WIN32DLL*/
-
-nkf_char std_ungetc(nkf_char c, FILE *f)
-{
- if (std_gc_ndx == STD_GC_BUFSIZE){
- return EOF;
- }
- std_gc_buf[std_gc_ndx++] = c;
- return c;
-}
-
-#ifndef WIN32DLL
-void std_putc(nkf_char c)
-{
- if(c!=EOF)
- putchar(c);
-}
-#endif /*WIN32DLL*/
-
-#if !defined(PERL_XS) && !defined(WIN32DLL)
-nkf_char noconvert(FILE *f)
-{
- nkf_char c;
-
- if (nop_f == 2)
- module_connection();
- while ((c = (*i_getc)(f)) != EOF)
- (*o_putc)(c);
- (*o_putc)(EOF);
- return 1;
-}
-#endif
-void module_connection(void)
+nkf_char x0212_unshift(nkf_char c)
{
- if (!output_encoding) output_encoding = nkf_enc_from_index(DEFAULT_ENCODING);
- oconv = nkf_enc_to_oconv(output_encoding);
- o_putc = std_putc;
-
- /* replace continucation module, from output side */
-
- /* output redicrection */
-#ifdef CHECK_OPTION
- if (noout_f || guess_f){
- o_putc = no_putc;
- }
-#endif
- if (mimeout_f) {
- o_mputc = o_putc;
- o_putc = mime_putc;
- if (mimeout_f == TRUE) {
- o_base64conv = oconv; oconv = base64_conv;
- }
- /* base64_count = 0; */
- }
-
- if (nlmode_f || guess_f) {
- o_nlconv = oconv; oconv = nl_conv;
- }
- if (rot_f) {
- o_rot_conv = oconv; oconv = rot_conv;
- }
- if (iso2022jp_f) {
- o_iso2022jp_check_conv = oconv; oconv = iso2022jp_check_conv;
- }
- if (hira_f) {
- o_hira_conv = oconv; oconv = hira_conv;
- }
- if (fold_f) {
- o_fconv = oconv; oconv = fold_conv;
- f_line = 0;
- }
- if (alpha_f || x0201_f) {
- o_zconv = oconv; oconv = z_conv;
- }
-
- i_getc = std_getc;
- i_ungetc = std_ungetc;
- /* input redicrection */
-#ifdef INPUT_OPTION
- if (cap_f){
- i_cgetc = i_getc; i_getc = cap_getc;
- i_cungetc = i_ungetc; i_ungetc= cap_ungetc;
- }
- if (url_f){
- i_ugetc = i_getc; i_getc = url_getc;
- i_uungetc = i_ungetc; i_ungetc= url_ungetc;
- }
-#endif
-#ifdef NUMCHAR_OPTION
- if (numchar_f){
- i_ngetc = i_getc; i_getc = numchar_getc;
- i_nungetc = i_ungetc; i_ungetc= numchar_ungetc;
- }
-#endif
-#ifdef UNICODE_NORMALIZATION
- if (nfc_f){
- i_nfc_getc = i_getc; i_getc = nfc_getc;
- i_nfc_ungetc = i_ungetc; i_ungetc= nfc_ungetc;
- }
-#endif
- if (mime_f && mimebuf_f==FIXED_MIME) {
- i_mgetc = i_getc; i_getc = mime_getc;
- i_mungetc = i_ungetc; i_ungetc = mime_ungetc;
- }
- if (broken_f & 1) {
- i_bgetc = i_getc; i_getc = broken_getc;
- i_bungetc = i_ungetc; i_ungetc = broken_ungetc;
- }
- if (input_encoding) {
- set_iconv(-TRUE, nkf_enc_to_iconv(input_encoding));
- } else {
- set_iconv(FALSE, e_iconv);
- }
-
- {
- struct input_code *p = input_code_list;
- while (p->name){
- status_reinit(p++);
- }
+ nkf_char ret = c;
+ if (0x7f <= c && c <= 0x88){
+ ret = c + (0x75 - 0x7f);
+ }else if (0x89 <= c && c <= 0x92){
+ ret = PREFIX_EUCG3 | 0x80 | (c + (0x75 - 0x89));
}
+ return ret;
}
+#endif /* X0212_ENABLE */
-/*
- * Check and Ignore BOM
- */
-void check_bom(FILE *f)
+nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
{
- int c2;
- switch(c2 = (*i_getc)(f)){
- case 0x00:
- if((c2 = (*i_getc)(f)) == 0x00){
- if((c2 = (*i_getc)(f)) == 0xFE){
- if((c2 = (*i_getc)(f)) == 0xFF){
- if(!input_encoding){
- set_iconv(TRUE, w_iconv32);
- }
- if (iconv == w_iconv32) {
- input_endian = ENDIAN_BIG;
- return;
- }
- (*i_ungetc)(0xFF,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0xFE,f);
- }else if(c2 == 0xFF){
- if((c2 = (*i_getc)(f)) == 0xFE){
- if(!input_encoding){
- set_iconv(TRUE, w_iconv32);
- }
- if (iconv == w_iconv32) {
- input_endian = ENDIAN_2143;
- return;
- }
- (*i_ungetc)(0xFF,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0xFF,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0x00,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0x00,f);
- break;
- case 0xEF:
- if((c2 = (*i_getc)(f)) == 0xBB){
- if((c2 = (*i_getc)(f)) == 0xBF){
- if(!input_encoding){
- set_iconv(TRUE, w_iconv);
- }
- if (iconv == w_iconv) {
- return;
- }
- (*i_ungetc)(0xBF,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0xBB,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0xEF,f);
- break;
- case 0xFE:
- if((c2 = (*i_getc)(f)) == 0xFF){
- if((c2 = (*i_getc)(f)) == 0x00){
- if((c2 = (*i_getc)(f)) == 0x00){
- if(!input_encoding){
- set_iconv(TRUE, w_iconv32);
- }
- if (iconv == w_iconv32) {
- input_endian = ENDIAN_3412;
- return;
- }
- (*i_ungetc)(0x00,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0x00,f);
- }else (*i_ungetc)(c2,f);
- if(!input_encoding){
- set_iconv(TRUE, w_iconv16);
- }
- if (iconv == w_iconv16) {
- input_endian = ENDIAN_BIG;
- return;
- }
- (*i_ungetc)(0xFF,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0xFE,f);
- break;
- case 0xFF:
- if((c2 = (*i_getc)(f)) == 0xFE){
- if((c2 = (*i_getc)(f)) == 0x00){
- if((c2 = (*i_getc)(f)) == 0x00){
- if(!input_encoding){
- set_iconv(TRUE, w_iconv32);
- }
- if (iconv == w_iconv32) {
- input_endian = ENDIAN_LITTLE;
- return;
- }
- (*i_ungetc)(0x00,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0x00,f);
- }else (*i_ungetc)(c2,f);
- if(!input_encoding){
- set_iconv(TRUE, w_iconv16);
- }
- if (iconv == w_iconv16) {
- input_endian = ENDIAN_LITTLE;
- return;
+ nkf_char ndx;
+ if (is_eucg3(c2)){
+ ndx = c2 & 0x7f;
+ if (x0213_f){
+ if((0x21 <= ndx && ndx <= 0x2F)){
+ if (p2) *p2 = ((ndx - 1) >> 1) + 0xec - ndx / 8 * 3;
+ if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
+ return 0;
+ }else if(0x6E <= ndx && ndx <= 0x7E){
+ if (p2) *p2 = ((ndx - 1) >> 1) + 0xbe;
+ if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
+ return 0;
}
- (*i_ungetc)(0xFE,f);
- }else (*i_ungetc)(c2,f);
- (*i_ungetc)(0xFF,f);
- break;
- default:
- (*i_ungetc)(c2,f);
- break;
- }
-}
-
-/*
- Conversion main loop. Code detection only.
- */
-
-nkf_char kanji_convert(FILE *f)
-{
- nkf_char c3, c2=0, c1, c0=0;
- int is_8bit = FALSE;
-
- if (input_encoding && !nkf_enc_asciicompat(input_encoding)) {
- is_8bit = TRUE;
- }
-
- input_mode = ASCII;
- output_mode = ASCII;
- shift_mode = FALSE;
-
-#define NEXT continue /* no output, get next */
-#define SEND ; /* output c1 and c2, get next */
-#define LAST break /* end of loop, go closing */
-
- module_connection();
- check_bom(f);
-
- while ((c1 = (*i_getc)(f)) != EOF) {
-#ifdef INPUT_CODE_FIX
- if (!input_encoding)
-#endif
- code_status(c1);
- if (c2) {
- /* second byte */
- if (c2 > ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) {
- /* in case of 8th bit is on */
- if (!estab_f&&!mime_decode_mode) {
- /* in case of not established yet */
- /* It is still ambiguious */
- if (h_conv(f, c2, c1)==EOF)
- LAST;
- else
- c2 = 0;
- NEXT;
- } else {
- /* in case of already established */
- if (c1 < AT) {
- /* ignore bogus code and not CP5022x UCD */
- c2 = 0;
- NEXT;
- } else {
- SEND;
- }
- }
- } else
- /* second byte, 7 bit code */
- /* it might be kanji shitfted */
- if ((c1 == DEL) || (c1 <= SP)) {
- /* ignore bogus first code */
- c2 = 0;
- NEXT;
- } else
- SEND;
- } else {
- /* first byte */
-#ifdef UTF8_INPUT_ENABLE
- if (iconv == w_iconv16) {
- if (input_endian == ENDIAN_BIG) {
- c2 = c1;
- if ((c1 = (*i_getc)(f)) != EOF) {
- if (0xD8 <= c2 && c2 <= 0xDB) {
- if ((c0 = (*i_getc)(f)) != EOF) {
- c0 <<= 8;
- if ((c3 = (*i_getc)(f)) != EOF) {
- c0 |= c3;
- } else c2 = EOF;
- } else c2 = EOF;
- }
- } else c2 = EOF;
- } else {
- if ((c2 = (*i_getc)(f)) != EOF) {
- if (0xD8 <= c2 && c2 <= 0xDB) {
- if ((c3 = (*i_getc)(f)) != EOF) {
- if ((c0 = (*i_getc)(f)) != EOF) {
- c0 <<= 8;
- c0 |= c3;
- } else c2 = EOF;
- } else c2 = EOF;
- }
- } else c2 = EOF;
- }
- SEND;
- } else if(iconv == w_iconv32){
- int c3 = c1;
- if((c2 = (*i_getc)(f)) != EOF &&
- (c1 = (*i_getc)(f)) != EOF &&
- (c0 = (*i_getc)(f)) != EOF){
- switch(input_endian){
- case ENDIAN_BIG:
- c1 = (c2&0xFF)<<16 | (c1&0xFF)<<8 | (c0&0xFF);
- break;
- case ENDIAN_LITTLE:
- c1 = (c3&0xFF) | (c2&0xFF)<<8 | (c1&0xFF)<<16;
- break;
- case ENDIAN_2143:
- c1 = (c3&0xFF)<<16 | (c1&0xFF) | (c0&0xFF)<<8;
- break;
- case ENDIAN_3412:
- c1 = (c3&0xFF)<<8 | (c2&0xFF) | (c0&0xFF)<<16;
- break;
- }
- c2 = 0;
- }else{
- c2 = EOF;
- }
- SEND;
- } else
-#endif
-#ifdef NUMCHAR_OPTION
- if (is_unicode_capsule(c1)){
- SEND;
- } else
-#endif
- if (c1 > ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) {
- /* 8 bit code */
- if (!estab_f && !iso8859_f) {
- /* not established yet */
- c2 = c1;
- NEXT;
- } else { /* estab_f==TRUE */
- if (iso8859_f) {
- c2 = ISO_8859_1;
- c1 &= 0x7f;
- SEND;
- } else if (SSP<=c1 && c1<0xe0 && iconv == s_iconv) {
- /* SJIS X0201 Case... */
- if (iso2022jp_f && !x0201_f) {
- (*oconv)(GETA1, GETA2);
- NEXT;
- } else {
- c2 = JIS_X_0201;
- c1 &= 0x7f;
- SEND;
- }
- } else if (c1==SSO && iconv != s_iconv) {
- /* EUC X0201 Case */
- c1 = (*i_getc)(f); /* skip SSO */
- code_status(c1);
- if (SSP<=c1 && c1<0xe0) {
- if (iso2022jp_f && !x0201_f) {
- (*oconv)(GETA1, GETA2);
- NEXT;
- } else {
- c2 = JIS_X_0201;
- c1 &= 0x7f;
- SEND;
- }
- } else { /* bogus code, skip SSO and one byte */
- NEXT;
- }
- } else if (ms_ucs_map_f == UCS_MAP_CP10001 &&
- (c1 == 0xFD || c1 == 0xFE)) {
- /* CP10001 */
- c2 = JIS_X_0201;
- c1 &= 0x7f;
- SEND;
- } else {
- /* already established */
- c2 = c1;
- NEXT;
- }
- }
- } else if ((c1 > SP) && (c1 != DEL)) {
- /* in case of Roman characters */
- if (shift_mode) {
- /* output 1 shifted byte */
- if (iso8859_f) {
- c2 = ISO_8859_1;
- SEND;
- } else if (SP <= c1 && c1 < (0xe0&0x7f)){
- /* output 1 shifted byte */
- if (iso2022jp_f && !x0201_f) {
- (*oconv)(GETA1, GETA2);
- NEXT;
- } else {
- c2 = JIS_X_0201;
- SEND;
- }
- } else {
- /* look like bogus code */
- NEXT;
- }
- } else if (input_mode == JIS_X_0208 || input_mode == JIS_X_0212 ||
- input_mode == JIS_X_0213_1 || input_mode == JIS_X_0213_2) {
- /* in case of Kanji shifted */
- c2 = c1;
- NEXT;
- } else if (c1 == '=' && mime_f && !mime_decode_mode) {
- /* Check MIME code */
- if ((c1 = (*i_getc)(f)) == EOF) {
- (*oconv)(0, '=');
- LAST;
- } else if (c1 == '?') {
- /* =? is mime conversion start sequence */
- if(mime_f == STRICT_MIME) {
- /* check in real detail */
- if (mime_begin_strict(f) == EOF)
- LAST;
- else
- NEXT;
- } else if (mime_begin(f) == EOF)
- LAST;
- else
- NEXT;
- } else {
- (*oconv)(0, '=');
- (*i_ungetc)(c1,f);
- NEXT;
- }
- } else {
- /* normal ASCII code */
- SEND;
- }
- } else if (c1 == SI && (!is_8bit || mime_decode_mode)) {
- shift_mode = FALSE;
- NEXT;
- } else if (c1 == SO && (!is_8bit || mime_decode_mode)) {
- shift_mode = TRUE;
- NEXT;
- } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) {
- if ((c1 = (*i_getc)(f)) == EOF) {
- /* (*oconv)(0, ESC); don't send bogus code */
- LAST;
- } else if (c1 == '$') {
- if ((c1 = (*i_getc)(f)) == EOF) {
- /*
- (*oconv)(0, ESC); don't send bogus code
- (*oconv)(0, '$'); */
- LAST;
- } else if (c1 == '@'|| c1 == 'B') {
- /* This is kanji introduction */
- input_mode = JIS_X_0208;
- shift_mode = FALSE;
- set_input_codename("ISO-2022-JP");
-#ifdef CHECK_OPTION
- debug("ISO-2022-JP");
-#endif
- NEXT;
- } else if (c1 == '(') {
- if ((c1 = (*i_getc)(f)) == EOF) {
- /* don't send bogus code
- (*oconv)(0, ESC);
- (*oconv)(0, '$');
- (*oconv)(0, '(');
- */
- LAST;
- } else if (c1 == '@'|| c1 == 'B') {
- /* This is kanji introduction */
- input_mode = JIS_X_0208;
- shift_mode = FALSE;
- NEXT;
+ return 1;
+ }
#ifdef X0212_ENABLE
- } else if (c1 == 'D'){
- input_mode = JIS_X_0212;
- shift_mode = FALSE;
- NEXT;
-#endif /* X0212_ENABLE */
- } else if (c1 == 0x4F){
- input_mode = JIS_X_0213_1;
- shift_mode = FALSE;
- NEXT;
- } else if (c1 == 0x50){
- input_mode = JIS_X_0213_2;
- shift_mode = FALSE;
- NEXT;
- } else {
- /* could be some special code */
- (*oconv)(0, ESC);
- (*oconv)(0, '$');
- (*oconv)(0, '(');
- (*oconv)(0, c1);
- NEXT;
- }
- } else if (broken_f&0x2) {
- /* accept any ESC-(-x as broken code ... */
- input_mode = JIS_X_0208;
- shift_mode = FALSE;
- NEXT;
- } else {
- (*oconv)(0, ESC);
- (*oconv)(0, '$');
- (*oconv)(0, c1);
- NEXT;
- }
- } else if (c1 == '(') {
- if ((c1 = (*i_getc)(f)) == EOF) {
- /* don't send bogus code
- (*oconv)(0, ESC);
- (*oconv)(0, '('); */
- LAST;
- } else {
- if (c1 == 'I') {
- /* This is X0201 kana introduction */
- input_mode = JIS_X_0201; shift_mode = JIS_X_0201;
- NEXT;
- } else if (c1 == 'B' || c1 == 'J' || c1 == 'H') {
- /* This is X0208 kanji introduction */
- input_mode = ASCII; shift_mode = FALSE;
- NEXT;
- } else if (broken_f&0x2) {
- input_mode = ASCII; shift_mode = FALSE;
- NEXT;
- } else {
- (*oconv)(0, ESC);
- (*oconv)(0, '(');
- /* maintain various input_mode here */
- SEND;
- }
- }
- } else if ( c1 == 'N' || c1 == 'n'){
- /* SS2 */
- c3 = (*i_getc)(f); /* skip SS2 */
- if ( (SP<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){
- c1 = c3;
- c2 = JIS_X_0201;
- SEND;
- }else{
- (*i_ungetc)(c3, f);
- /* lonely ESC */
- (*oconv)(0, ESC);
- SEND;
- }
- } else {
- /* lonely ESC */
- (*oconv)(0, ESC);
- SEND;
- }
- } else if (c1 == ESC && iconv == s_iconv) {
- /* ESC in Shift_JIS */
- if ((c1 = (*i_getc)(f)) == EOF) {
- /* (*oconv)(0, ESC); don't send bogus code */
- LAST;
- } else if (c1 == '$') {
- /* J-PHONE emoji */
- if ((c1 = (*i_getc)(f)) == EOF) {
- /*
- (*oconv)(0, ESC); don't send bogus code
- (*oconv)(0, '$'); */
- LAST;
- } else {
- if (('E' <= c1 && c1 <= 'G') ||
- ('O' <= c1 && c1 <= 'Q')) {
- /*
- NUM : 0 1 2 3 4 5
- BYTE: G E F O P Q
- C%7 : 1 6 0 2 3 4
- C%7 : 0 1 2 3 4 5 6
- NUM : 2 0 3 4 5 X 1
- */
- static const char jphone_emoji_first_table[7] = {2, 0, 3, 4, 5, 0, 1};
- c0 = (jphone_emoji_first_table[c1 % 7] << 8) - SP + 0xE000 + CLASS_UNICODE;
- while ((c1 = (*i_getc)(f)) != EOF) {
- if (SP <= c1 && c1 <= 'z') {
- (*oconv)(0, c1 + c0);
- } else break; /* c1 == SO */
- }
- }
- }
- if (c1 == EOF) LAST;
- NEXT;
- } else {
- /* lonely ESC */
- (*oconv)(0, ESC);
- SEND;
- }
- } else if (c1 == LF || c1 == CR) {
- if (broken_f&4) {
- input_mode = ASCII; set_iconv(FALSE, 0);
- SEND;
- } else if (mime_decode_f && !mime_decode_mode){
- if (c1 == LF) {
- if ((c1=(*i_getc)(f))!=EOF && c1 == SP) {
- i_ungetc(SP,f);
- continue;
- } else {
- i_ungetc(c1,f);
- }
- c1 = LF;
- SEND;
- } else { /* if (c1 == CR)*/
- if ((c1=(*i_getc)(f))!=EOF) {
- if (c1==SP) {
- i_ungetc(SP,f);
- continue;
- } else if (c1 == LF && (c1=(*i_getc)(f))!=EOF && c1 == SP) {
- i_ungetc(SP,f);
- continue;
- } else {
- i_ungetc(c1,f);
- }
- i_ungetc(LF,f);
- } else {
- i_ungetc(c1,f);
- }
- c1 = CR;
- SEND;
- }
- }
- } else if (c1 == DEL && input_mode == JIS_X_0208) {
- /* CP5022x */
- c2 = c1;
- NEXT;
- } else
- SEND;
- }
- /* send: */
- switch(input_mode){
- case ASCII:
- switch ((*iconv)(c2, c1, c0)) { /* can be EUC / SJIS / UTF-8 / UTF-16 */
- case -2:
- /* 4 bytes UTF-8 */
- if ((c0 = (*i_getc)(f)) != EOF) {
- code_status(c0);
- c0 <<= 8;
- if ((c3 = (*i_getc)(f)) != EOF) {
- code_status(c3);
- (*iconv)(c2, c1, c0|c3);
- }
- }
- break;
- case -1:
- /* 3 bytes EUC or UTF-8 */
- if ((c0 = (*i_getc)(f)) != EOF) {
- code_status(c0);
- (*iconv)(c2, c1, c0);
- }
- break;
- }
- break;
- case JIS_X_0208:
- case JIS_X_0213_1:
- if (ms_ucs_map_f &&
- 0x7F <= c2 && c2 <= 0x92 &&
- 0x21 <= c1 && c1 <= 0x7E) {
- /* CP932 UDC */
- if(c1 == 0x7F) return 0;
- c1 = (c2 - 0x7F) * 94 + c1 - 0x21 + 0xE000 + CLASS_UNICODE;
- c2 = 0;
+ else if(nkf_isgraph(ndx)){
+ nkf_char val = 0;
+ const unsigned short *ptr;
+ ptr = x0212_shiftjis[ndx - 0x21];
+ if (ptr){
+ val = ptr[(c1 & 0x7f) - 0x21];
}
- (*oconv)(c2, c1); /* this is JIS, not SJIS/EUC case */
- break;
-#ifdef X0212_ENABLE
- case JIS_X_0212:
- (*oconv)(PREFIX_EUCG3 | c2, c1);
- break;
-#endif /* X0212_ENABLE */
- case JIS_X_0213_2:
- (*oconv)(PREFIX_EUCG3 | c2, c1);
- break;
- default:
- (*oconv)(input_mode, c1); /* other special case */
- }
-
- c2 = 0;
- c0 = 0;
- continue;
- /* goto next_word */
- }
-
- /* epilogue */
- (*iconv)(EOF, 0, 0);
- if (!input_codename)
- {
- if (is_8bit) {
- struct input_code *p = input_code_list;
- struct input_code *result = p;
- while (p->name){
- if (p->score < result->score) result = p;
- ++p;
+ if (val){
+ c2 = val >> 8;
+ c1 = val & 0xff;
+ if (p2) *p2 = c2;
+ if (p1) *p1 = c1;
+ return 0;
}
- set_input_codename(result->name);
-#ifdef CHECK_OPTION
- debug(result->name);
-#endif
- }
- }
- return 1;
-}
-
-nkf_char
-h_conv(FILE *f, nkf_char c2, nkf_char c1)
-{
- nkf_char ret, c3, c0;
- int hold_index;
-
-
- /** it must NOT be in the kanji shifte sequence */
- /** it must NOT be written in JIS7 */
- /** and it must be after 2 byte 8bit code */
-
- hold_count = 0;
- push_hold_buf(c2);
- push_hold_buf(c1);
-
- while ((c1 = (*i_getc)(f)) != EOF) {
- if (c1 == ESC){
- (*i_ungetc)(c1,f);
- break;
- }
- code_status(c1);
- if (push_hold_buf(c1) == EOF || estab_f){
- break;
- }
- }
-
- if (!estab_f){
- struct input_code *p = input_code_list;
- struct input_code *result = p;
- if (c1 == EOF){
- code_status(c1);
- }
- while (p->name){
- if (p->status_func && p->score < result->score){
- result = p;
- }
- ++p;
- }
- set_iconv(TRUE, result->iconv_func);
- }
-
-
- /** now,
- ** 1) EOF is detected, or
- ** 2) Code is established, or
- ** 3) Buffer is FULL (but last word is pushed)
- **
- ** in 1) and 3) cases, we continue to use
- ** Kanji codes by oconv and leave estab_f unchanged.
- **/
-
- ret = c1;
- hold_index = 0;
- while (hold_index < hold_count){
- c2 = hold_buf[hold_index++];
- if (c2 <= DEL
-#ifdef NUMCHAR_OPTION
- || is_unicode_capsule(c2)
-#endif
- ){
- (*iconv)(0, c2, 0);
- continue;
- }else if (iconv == s_iconv && 0xa1 <= c2 && c2 <= 0xdf){
- (*iconv)(JIS_X_0201, c2, 0);
- continue;
- }
- if (hold_index < hold_count){
- c1 = hold_buf[hold_index++];
- }else{
- c1 = (*i_getc)(f);
- if (c1 == EOF){
- c3 = EOF;
- break;
- }
- code_status(c1);
- }
- c0 = 0;
- switch ((*iconv)(c2, c1, 0)) { /* can be EUC/SJIS/UTF-8 */
- case -2:
- /* 4 bytes UTF-8 */
- if (hold_index < hold_count){
- c0 = hold_buf[hold_index++];
- } else if ((c0 = (*i_getc)(f)) == EOF) {
- ret = EOF;
- break;
- } else {
- code_status(c0);
- c0 <<= 8;
- if (hold_index < hold_count){
- c3 = hold_buf[hold_index++];
- } else if ((c3 = (*i_getc)(f)) == EOF) {
- c0 = ret = EOF;
- break;
- } else {
- code_status(c3);
- (*iconv)(c2, c1, c0|c3);
- }
- }
- break;
- case -1:
- /* 3 bytes EUC or UTF-8 */
- if (hold_index < hold_count){
- c0 = hold_buf[hold_index++];
- } else if ((c0 = (*i_getc)(f)) == EOF) {
- ret = EOF;
- break;
- } else {
- code_status(c0);
- }
- (*iconv)(c2, c1, c0);
- break;
+ c2 = x0212_shift(c2);
}
- if (c0 == EOF) break;
+#endif /* X0212_ENABLE */
}
- return ret;
-}
-
-nkf_char push_hold_buf(nkf_char c2)
-{
- if (hold_count >= HOLD_SIZE*2)
- return (EOF);
- hold_buf[hold_count++] = (unsigned char)c2;
- return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count);
+ if(0x7F < c2) return 1;
+ if (p2) *p2 = ((c2 - 1) >> 1) + ((c2 <= 0x5e) ? 0x71 : 0xb1);
+ if (p1) *p1 = c1 + ((c2 & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
+ return 0;
}
nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
@@ -3384,6 +1491,8 @@ nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
if (0x9E < c1) c2++;
}
}else{
+#define SJ0162 0x00e1 /* 01 - 62 ku offset */
+#define SJ6394 0x0161 /* 63 - 94 ku offset */
c2 = c2 + c2 - ((c2 <= 0x9F) ? SJ0162 : SJ6394);
if (0x9E < c1) c2++;
}
@@ -3402,177 +1511,6 @@ nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
return 0;
}
-nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
-{
- if (c2 == JIS_X_0201) {
- c1 &= 0x7f;
- } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) {
- /* NOP */
- } else if (!x0213_f && 0xF0 <= c2 && c2 <= 0xF9 && 0x40 <= c1 && c1 <= 0xFC) {
- /* CP932 UDC */
- if(c1 == 0x7F) return 0;
- c1 = (c2 - 0xF0) * 188 + (c1 - 0x40 - (0x7E < c1)) + 0xE000 + CLASS_UNICODE;
- c2 = 0;
- } else {
- nkf_char ret = s2e_conv(c2, c1, &c2, &c1);
- if (ret) return ret;
- }
- (*oconv)(c2, c1);
- return 0;
-}
-
-nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
-{
- if (c2 == JIS_X_0201) {
- c1 &= 0x7f;
-#ifdef X0212_ENABLE
- }else if (c2 == 0x8f){
- if (c0 == 0){
- return -1;
- }
- if (!cp51932_f && !x0213_f && 0xF5 <= c1 && c1 <= 0xFE && 0xA1 <= c0 && c0 <= 0xFE) {
- /* encoding is eucJP-ms, so invert to Unicode Private User Area */
- c1 = (c1 - 0xF5) * 94 + c0 - 0xA1 + 0xE3AC + CLASS_UNICODE;
- c2 = 0;
- } else {
- c2 = (c2 << 8) | (c1 & 0x7f);
- c1 = c0 & 0x7f;
-#ifdef SHIFTJIS_CP932
- if (cp51932_f){
- nkf_char s2, s1;
- if (e2s_conv(c2, c1, &s2, &s1) == 0){
- s2e_conv(s2, s1, &c2, &c1);
- if (c2 < 0x100){
- c1 &= 0x7f;
- c2 &= 0x7f;
- }
- }
- }
-#endif /* SHIFTJIS_CP932 */
- }
-#endif /* X0212_ENABLE */
- } else if (c2 == SSO){
- c2 = JIS_X_0201;
- c1 &= 0x7f;
- } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) {
- /* NOP */
- } else {
- if (!cp51932_f && ms_ucs_map_f && 0xF5 <= c2 && c2 <= 0xFE && 0xA1 <= c1 && c1 <= 0xFE) {
- /* encoding is eucJP-ms, so invert to Unicode Private User Area */
- c1 = (c2 - 0xF5) * 94 + c1 - 0xA1 + 0xE000 + CLASS_UNICODE;
- c2 = 0;
- } else {
- c1 &= 0x7f;
- c2 &= 0x7f;
-#ifdef SHIFTJIS_CP932
- if (cp51932_f && 0x79 <= c2 && c2 <= 0x7c){
- nkf_char s2, s1;
- if (e2s_conv(c2, c1, &s2, &s1) == 0){
- s2e_conv(s2, s1, &c2, &c1);
- if (c2 < 0x100){
- c1 &= 0x7f;
- c2 &= 0x7f;
- }
- }
- }
-#endif /* SHIFTJIS_CP932 */
- }
- }
- (*oconv)(c2, c1);
- return 0;
-}
-
-#ifdef UTF8_INPUT_ENABLE
-nkf_char w2e_conv(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1)
-{
- nkf_char ret = 0;
-
- if (!c1){
- *p2 = 0;
- *p1 = c2;
- }else if (0xc0 <= c2 && c2 <= 0xef) {
- ret = unicode_to_jis_common(c2, c1, c0, p2, p1);
-#ifdef NUMCHAR_OPTION
- if (ret > 0){
- if (p2) *p2 = 0;
- if (p1) *p1 = CLASS_UNICODE | ww16_conv(c2, c1, c0);
- ret = 0;
- }
-#endif
- }
- return ret;
-}
-
-nkf_char w_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
-{
- nkf_char ret = 0;
- static const char w_iconv_utf8_1st_byte[] =
- { /* 0xC0 - 0xFF */
- 20, 20, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21,
- 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21,
- 30, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 32, 33, 33,
- 40, 41, 41, 41, 42, 43, 43, 43, 50, 50, 50, 50, 60, 60, 70, 70};
-
- if (c2 < 0 || 0xff < c2) {
- }else if (c2 == 0) { /* 0 : 1 byte*/
- c0 = 0;
- } else if ((c2 & 0xc0) == 0x80) { /* 0x80-0xbf : trail byte */
- return 0;
- } else{
- switch (w_iconv_utf8_1st_byte[c2 - 0xC0]) {
- case 21:
- if (c1 < 0x80 || 0xBF < c1) return 0;
- break;
- case 30:
- if (c0 == 0) return -1;
- if (c1 < 0xA0 || 0xBF < c1 || (c0 & 0xc0) != 0x80)
- return 0;
- break;
- case 31:
- case 33:
- if (c0 == 0) return -1;
- if ((c1 & 0xc0) != 0x80 || (c0 & 0xc0) != 0x80)
- return 0;
- break;
- case 32:
- if (c0 == 0) return -1;
- if (c1 < 0x80 || 0x9F < c1 || (c0 & 0xc0) != 0x80)
- return 0;
- break;
- case 40:
- if (c0 == 0) return -2;
- if (c1 < 0x90 || 0xBF < c1 || (c0 & 0xc0c0) != 0x8080)
- return 0;
- break;
- case 41:
- if (c0 == 0) return -2;
- if (c1 < 0x80 || 0xBF < c1 || (c0 & 0xc0c0) != 0x8080)
- return 0;
- break;
- case 42:
- if (c0 == 0) return -2;
- if (c1 < 0x80 || 0x8F < c1 || (c0 & 0xc0c0) != 0x8080)
- return 0;
- break;
- default:
- return 0;
- break;
- }
- }
- if (c2 == 0 || c2 == EOF){
- } else if ((c2 & 0xf8) == 0xf0) { /* 4 bytes */
- c1 = CLASS_UNICODE | ww16_conv(c2, c1, c0);
- c2 = 0;
- } else {
- ret = w2e_conv(c2, c1, c0, &c2, &c1);
- }
- if (ret == 0){
- (*oconv)(c2, c1);
- }
- return ret;
-}
-#endif
-
#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0)
{
@@ -3599,9 +1537,7 @@ void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0)
*p0 = 0;
}
}
-#endif
-#ifdef UTF8_INPUT_ENABLE
nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0)
{
nkf_char val;
@@ -3625,66 +1561,40 @@ nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0)
}
return val;
}
-
-nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1)
-{
- nkf_char c2, c1, c0;
- nkf_char ret = 0;
- val &= VALUE_MASK;
- if (val < 0x80){
- *p2 = 0;
- *p1 = val;
- }else{
- w16w_conv(val, &c2, &c1, &c0);
- ret = unicode_to_jis_common(c2, c1, c0, p2, p1);
-#ifdef NUMCHAR_OPTION
- if (ret > 0){
- *p2 = 0;
- *p1 = CLASS_UNICODE | val;
- ret = 0;
- }
-#endif
- }
- return ret;
-}
#endif
#ifdef UTF8_INPUT_ENABLE
-nkf_char w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0)
+nkf_char w_iconv_common(nkf_char c1, nkf_char c0, const unsigned short *const *pp, nkf_char psize, nkf_char *p2, nkf_char *p1)
{
- nkf_char ret = 0;
- if ((c2==0 && c1 < 0x80) || c2==EOF) {
- (*oconv)(c2, c1);
- return 0;
- }else if (0xD8 <= c2 && c2 <= 0xDB) {
- if (c0 < NKF_INT32_C(0xDC00) || NKF_INT32_C(0xDFFF) < c0)
- return -2;
- c1 = CLASS_UNICODE | ((c2 << 18) + (c1 << 10) + c0 - NKF_INT32_C(0x35FDC00));
- c2 = 0;
- }else if ((c2>>3) == 27) { /* unpaired surrogate */
- /*
- return 2;
- */
- return 1;
- }else ret = w16e_conv(((c2 & 0xff)<<8) + c1, &c2, &c1);
- if (ret) return ret;
- (*oconv)(c2, c1);
- return 0;
-}
+ nkf_char c2;
+ const unsigned short *p;
+ unsigned short val;
-nkf_char w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0)
-{
- int ret = 0;
+ if (pp == 0) return 1;
- if ((c2 == 0 && c1 < 0x80) || c2==EOF) {
- } else if (is_unicode_bmp(c1)) {
- ret = w16e_conv(c1, &c2, &c1);
- } else {
- c2 = 0;
- c1 = CLASS_UNICODE | c1;
+ c1 -= 0x80;
+ if (c1 < 0 || psize <= c1) return 1;
+ p = pp[c1];
+ if (p == 0) return 1;
+
+ c0 -= 0x80;
+ if (c0 < 0 || sizeof_utf8_to_euc_C2 <= c0) return 1;
+ val = p[c0];
+ if (val == 0) return 1;
+ if (no_cp932ext_f && (
+ (val>>8) == 0x2D || /* NEC special characters */
+ val > NKF_INT32_C(0xF300) /* IBM extended characters */
+ )) return 1;
+
+ c2 = val >> 8;
+ if (val > 0x7FFF){
+ c2 &= 0x7f;
+ c2 |= PREFIX_EUCG3;
}
- if (ret) return ret;
- (*oconv)(c2, c1);
+ if (c2 == SO) c2 = JIS_X_0201;
+ c1 = val & 0x7f;
+ if (p2) *p2 = c2;
+ if (p1) *p1 = c1;
return 0;
}
@@ -3843,128 +1753,6 @@ nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *
return ret;
}
-nkf_char w_iconv_common(nkf_char c1, nkf_char c0, const unsigned short *const *pp, nkf_char psize, nkf_char *p2, nkf_char *p1)
-{
- nkf_char c2;
- const unsigned short *p;
- unsigned short val;
-
- if (pp == 0) return 1;
-
- c1 -= 0x80;
- if (c1 < 0 || psize <= c1) return 1;
- p = pp[c1];
- if (p == 0) return 1;
-
- c0 -= 0x80;
- if (c0 < 0 || sizeof_utf8_to_euc_C2 <= c0) return 1;
- val = p[c0];
- if (val == 0) return 1;
- if (no_cp932ext_f && (
- (val>>8) == 0x2D || /* NEC special characters */
- val > NKF_INT32_C(0xF300) /* IBM extended characters */
- )) return 1;
-
- c2 = val >> 8;
- if (val > 0x7FFF){
- c2 &= 0x7f;
- c2 |= PREFIX_EUCG3;
- }
- if (c2 == SO) c2 = JIS_X_0201;
- c1 = val & 0x7f;
- if (p2) *p2 = c2;
- if (p1) *p1 = c1;
- return 0;
-}
-
-void nkf_each_char_to_hex(void (*f)(nkf_char c2,nkf_char c1), nkf_char c)
-{
- int shift = 20;
- c &= VALUE_MASK;
- while(shift >= 0){
- if(c >= 1<<shift){
- while(shift >= 0){
- (*f)(0, bin2hex(c>>shift));
- shift -= 4;
- }
- }else{
- shift -= 4;
- }
- }
- return;
-}
-
-void encode_fallback_html(nkf_char c)
-{
- (*oconv)(0, '&');
- (*oconv)(0, '#');
- c &= VALUE_MASK;
- if(c >= NKF_INT32_C(1000000))
- (*oconv)(0, 0x30+(c/NKF_INT32_C(1000000))%10);
- if(c >= NKF_INT32_C(100000))
- (*oconv)(0, 0x30+(c/NKF_INT32_C(100000) )%10);
- if(c >= 10000)
- (*oconv)(0, 0x30+(c/10000 )%10);
- if(c >= 1000)
- (*oconv)(0, 0x30+(c/1000 )%10);
- if(c >= 100)
- (*oconv)(0, 0x30+(c/100 )%10);
- if(c >= 10)
- (*oconv)(0, 0x30+(c/10 )%10);
- if(c >= 0)
- (*oconv)(0, 0x30+ c %10);
- (*oconv)(0, ';');
- return;
-}
-
-void encode_fallback_xml(nkf_char c)
-{
- (*oconv)(0, '&');
- (*oconv)(0, '#');
- (*oconv)(0, 'x');
- nkf_each_char_to_hex(oconv, c);
- (*oconv)(0, ';');
- return;
-}
-
-void encode_fallback_java(nkf_char c)
-{
- (*oconv)(0, '\\');
- c &= VALUE_MASK;
- if(!is_unicode_bmp(c)){
- (*oconv)(0, 'U');
- (*oconv)(0, '0');
- (*oconv)(0, '0');
- (*oconv)(0, bin2hex(c>>20));
- (*oconv)(0, bin2hex(c>>16));
- }else{
- (*oconv)(0, 'u');
- }
- (*oconv)(0, bin2hex(c>>12));
- (*oconv)(0, bin2hex(c>> 8));
- (*oconv)(0, bin2hex(c>> 4));
- (*oconv)(0, bin2hex(c ));
- return;
-}
-
-void encode_fallback_perl(nkf_char c)
-{
- (*oconv)(0, '\\');
- (*oconv)(0, 'x');
- (*oconv)(0, '{');
- nkf_each_char_to_hex(oconv, c);
- (*oconv)(0, '}');
- return;
-}
-
-void encode_fallback_subchar(nkf_char c)
-{
- c = unicode_subchar;
- (*oconv)((c>>8)&0xFF, c&0xFF);
- return;
-}
-#endif
-
#ifdef UTF8_OUTPUT_ENABLE
nkf_char e2w_conv(nkf_char c2, nkf_char c1)
{
@@ -4008,7 +1796,483 @@ nkf_char e2w_conv(nkf_char c2, nkf_char c1)
return p[c1];
return 0;
}
+#endif
+
+nkf_char w2e_conv(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1)
+{
+ nkf_char ret = 0;
+
+ if (!c1){
+ *p2 = 0;
+ *p1 = c2;
+ }else if (0xc0 <= c2 && c2 <= 0xef) {
+ ret = unicode_to_jis_common(c2, c1, c0, p2, p1);
+#ifdef NUMCHAR_OPTION
+ if (ret > 0){
+ if (p2) *p2 = 0;
+ if (p1) *p1 = CLASS_UNICODE | ww16_conv(c2, c1, c0);
+ ret = 0;
+ }
+#endif
+ }
+ return ret;
+}
+
+#ifdef UTF8_INPUT_ENABLE
+nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1)
+{
+ nkf_char c2, c1, c0;
+ nkf_char ret = 0;
+ val &= VALUE_MASK;
+ if (val < 0x80){
+ *p2 = 0;
+ *p1 = val;
+ }else{
+ w16w_conv(val, &c2, &c1, &c0);
+ ret = unicode_to_jis_common(c2, c1, c0, p2, p1);
+#ifdef NUMCHAR_OPTION
+ if (ret > 0){
+ *p2 = 0;
+ *p1 = CLASS_UNICODE | val;
+ ret = 0;
+ }
+#endif
+ }
+ return ret;
+}
+#endif
+
+nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
+{
+ if (c2 == JIS_X_0201) {
+ c1 &= 0x7f;
+#ifdef X0212_ENABLE
+ }else if (c2 == 0x8f){
+ if (c0 == 0){
+ return -1;
+ }
+ if (!cp51932_f && !x0213_f && 0xF5 <= c1 && c1 <= 0xFE && 0xA1 <= c0 && c0 <= 0xFE) {
+ /* encoding is eucJP-ms, so invert to Unicode Private User Area */
+ c1 = (c1 - 0xF5) * 94 + c0 - 0xA1 + 0xE3AC + CLASS_UNICODE;
+ c2 = 0;
+ } else {
+ c2 = (c2 << 8) | (c1 & 0x7f);
+ c1 = c0 & 0x7f;
+#ifdef SHIFTJIS_CP932
+ if (cp51932_f){
+ nkf_char s2, s1;
+ if (e2s_conv(c2, c1, &s2, &s1) == 0){
+ s2e_conv(s2, s1, &c2, &c1);
+ if (c2 < 0x100){
+ c1 &= 0x7f;
+ c2 &= 0x7f;
+ }
+ }
+ }
+#endif /* SHIFTJIS_CP932 */
+ }
+#endif /* X0212_ENABLE */
+ } else if (c2 == SSO){
+ c2 = JIS_X_0201;
+ c1 &= 0x7f;
+ } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) {
+ /* NOP */
+ } else {
+ if (!cp51932_f && ms_ucs_map_f && 0xF5 <= c2 && c2 <= 0xFE && 0xA1 <= c1 && c1 <= 0xFE) {
+ /* encoding is eucJP-ms, so invert to Unicode Private User Area */
+ c1 = (c2 - 0xF5) * 94 + c1 - 0xA1 + 0xE000 + CLASS_UNICODE;
+ c2 = 0;
+ } else {
+ c1 &= 0x7f;
+ c2 &= 0x7f;
+#ifdef SHIFTJIS_CP932
+ if (cp51932_f && 0x79 <= c2 && c2 <= 0x7c){
+ nkf_char s2, s1;
+ if (e2s_conv(c2, c1, &s2, &s1) == 0){
+ s2e_conv(s2, s1, &c2, &c1);
+ if (c2 < 0x100){
+ c1 &= 0x7f;
+ c2 &= 0x7f;
+ }
+ }
+ }
+#endif /* SHIFTJIS_CP932 */
+ }
+ }
+ (*oconv)(c2, c1);
+ return 0;
+}
+nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
+{
+ if (c2 == JIS_X_0201) {
+ c1 &= 0x7f;
+ } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) {
+ /* NOP */
+ } else if (!x0213_f && 0xF0 <= c2 && c2 <= 0xF9 && 0x40 <= c1 && c1 <= 0xFC) {
+ /* CP932 UDC */
+ if(c1 == 0x7F) return 0;
+ c1 = (c2 - 0xF0) * 188 + (c1 - 0x40 - (0x7E < c1)) + 0xE000 + CLASS_UNICODE;
+ c2 = 0;
+ } else {
+ nkf_char ret = s2e_conv(c2, c1, &c2, &c1);
+ if (ret) return ret;
+ }
+ (*oconv)(c2, c1);
+ return 0;
+}
+
+nkf_char w_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
+{
+ nkf_char ret = 0;
+ static const char w_iconv_utf8_1st_byte[] =
+ { /* 0xC0 - 0xFF */
+ 20, 20, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21,
+ 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21,
+ 30, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 32, 33, 33,
+ 40, 41, 41, 41, 42, 43, 43, 43, 50, 50, 50, 50, 60, 60, 70, 70};
+
+ if (c2 < 0 || 0xff < c2) {
+ }else if (c2 == 0) { /* 0 : 1 byte*/
+ c0 = 0;
+ } else if ((c2 & 0xc0) == 0x80) { /* 0x80-0xbf : trail byte */
+ return 0;
+ } else{
+ switch (w_iconv_utf8_1st_byte[c2 - 0xC0]) {
+ case 21:
+ if (c1 < 0x80 || 0xBF < c1) return 0;
+ break;
+ case 30:
+ if (c0 == 0) return -1;
+ if (c1 < 0xA0 || 0xBF < c1 || (c0 & 0xc0) != 0x80)
+ return 0;
+ break;
+ case 31:
+ case 33:
+ if (c0 == 0) return -1;
+ if ((c1 & 0xc0) != 0x80 || (c0 & 0xc0) != 0x80)
+ return 0;
+ break;
+ case 32:
+ if (c0 == 0) return -1;
+ if (c1 < 0x80 || 0x9F < c1 || (c0 & 0xc0) != 0x80)
+ return 0;
+ break;
+ case 40:
+ if (c0 == 0) return -2;
+ if (c1 < 0x90 || 0xBF < c1 || (c0 & 0xc0c0) != 0x8080)
+ return 0;
+ break;
+ case 41:
+ if (c0 == 0) return -2;
+ if (c1 < 0x80 || 0xBF < c1 || (c0 & 0xc0c0) != 0x8080)
+ return 0;
+ break;
+ case 42:
+ if (c0 == 0) return -2;
+ if (c1 < 0x80 || 0x8F < c1 || (c0 & 0xc0c0) != 0x8080)
+ return 0;
+ break;
+ default:
+ return 0;
+ break;
+ }
+ }
+ if (c2 == 0 || c2 == EOF){
+ } else if ((c2 & 0xf8) == 0xf0) { /* 4 bytes */
+ c1 = CLASS_UNICODE | ww16_conv(c2, c1, c0);
+ c2 = 0;
+ } else {
+ ret = w2e_conv(c2, c1, c0, &c2, &c1);
+ }
+ if (ret == 0){
+ (*oconv)(c2, c1);
+ }
+ return ret;
+}
+
+nkf_char w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0)
+{
+ nkf_char ret = 0;
+ if ((c2==0 && c1 < 0x80) || c2==EOF) {
+ (*oconv)(c2, c1);
+ return 0;
+ }else if (0xD8 <= c2 && c2 <= 0xDB) {
+ if (c0 < NKF_INT32_C(0xDC00) || NKF_INT32_C(0xDFFF) < c0)
+ return -2;
+ c1 = CLASS_UNICODE | ((c2 << 18) + (c1 << 10) + c0 - NKF_INT32_C(0x35FDC00));
+ c2 = 0;
+ }else if ((c2>>3) == 27) { /* unpaired surrogate */
+ /*
+ return 2;
+ */
+ return 1;
+ }else ret = w16e_conv(((c2 & 0xff)<<8) + c1, &c2, &c1);
+ if (ret) return ret;
+ (*oconv)(c2, c1);
+ return 0;
+}
+
+nkf_char w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0)
+{
+ int ret = 0;
+
+ if ((c2 == 0 && c1 < 0x80) || c2==EOF) {
+ } else if (is_unicode_bmp(c1)) {
+ ret = w16e_conv(c1, &c2, &c1);
+ } else {
+ c2 = 0;
+ c1 = CLASS_UNICODE | c1;
+ }
+ if (ret) return ret;
+ (*oconv)(c2, c1);
+ return 0;
+}
+
+#endif
+
+void j_oconv(nkf_char c2, nkf_char c1)
+{
+#ifdef NUMCHAR_OPTION
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ w16e_conv(c1, &c2, &c1);
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ c2 = c1 & VALUE_MASK;
+ if (ms_ucs_map_f && 0xE000 <= c2 && c2 <= 0xE757) {
+ /* CP5022x UDC */
+ c1 &= 0xFFF;
+ c2 = 0x7F + c1 / 94;
+ c1 = 0x21 + c1 % 94;
+ } else {
+ if (encode_fallback) (*encode_fallback)(c1);
+ return;
+ }
+ }
+ }
+#endif
+ if (c2 == EOF) {
+ if (output_mode !=ASCII && output_mode!=ISO_8859_1) {
+ (*o_putc)(ESC);
+ (*o_putc)('(');
+ (*o_putc)(ascii_intro);
+ output_mode = ASCII;
+ }
+ (*o_putc)(EOF);
+#ifdef X0212_ENABLE
+ } else if (is_eucg3(c2)){
+ if(x0213_f){
+ if(output_mode!=JIS_X_0213_2){
+ output_mode = JIS_X_0213_2;
+ (*o_putc)(ESC);
+ (*o_putc)('$');
+ (*o_putc)('(');
+ (*o_putc)(0x50);
+ }
+ }else{
+ if(output_mode!=JIS_X_0212){
+ output_mode = JIS_X_0212;
+ (*o_putc)(ESC);
+ (*o_putc)('$');
+ (*o_putc)('(');
+ (*o_putc)(0x44);
+ }
+ }
+ (*o_putc)(c2 & 0x7f);
+ (*o_putc)(c1);
+#endif
+ } else if (c2==JIS_X_0201) {
+ if (output_mode!=JIS_X_0201) {
+ output_mode = JIS_X_0201;
+ (*o_putc)(ESC);
+ (*o_putc)('(');
+ (*o_putc)('I');
+ }
+ (*o_putc)(c1);
+ } else if (c2==ISO_8859_1) {
+ /* iso8859 introduction, or 8th bit on */
+ /* Can we convert in 7bit form using ESC-'-'-A ?
+ Is this popular? */
+ output_mode = ISO_8859_1;
+ (*o_putc)(c1|0x80);
+ } else if (c2 == 0) {
+ if (output_mode !=ASCII && output_mode!=ISO_8859_1) {
+ (*o_putc)(ESC);
+ (*o_putc)('(');
+ (*o_putc)(ascii_intro);
+ output_mode = ASCII;
+ }
+ (*o_putc)(c1);
+ } else {
+ if(ms_ucs_map_f
+ ? c2<0x20 || 0x92<c2 || c1<0x20 || 0x7e<c1
+ : c2<0x20 || 0x7e<c2 || c1<0x20 || 0x7e<c1) return;
+ if(x0213_f){
+ if (output_mode!=JIS_X_0213_1) {
+ output_mode = JIS_X_0213_1;
+ (*o_putc)(ESC);
+ (*o_putc)('$');
+ (*o_putc)('(');
+ (*o_putc)(0x4F);
+ }
+ }else if (output_mode != JIS_X_0208) {
+ output_mode = JIS_X_0208;
+ (*o_putc)(ESC);
+ (*o_putc)('$');
+ (*o_putc)(kanji_intro);
+ }
+ (*o_putc)(c2);
+ (*o_putc)(c1);
+ }
+}
+
+void e_oconv(nkf_char c2, nkf_char c1)
+{
+#ifdef NUMCHAR_OPTION
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ w16e_conv(c1, &c2, &c1);
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ c2 = c1 & VALUE_MASK;
+ if (x0212_f && 0xE000 <= c2 && c2 <= 0xE757) {
+ /* eucJP-ms UDC */
+ c1 &= 0xFFF;
+ c2 = c1 / 94;
+ c2 += c2 < 10 ? 0x75 : 0x8FEB;
+ c1 = 0x21 + c1 % 94;
+ if (is_eucg3(c2)){
+ (*o_putc)(0x8f);
+ (*o_putc)((c2 & 0x7f) | 0x080);
+ (*o_putc)(c1 | 0x080);
+ }else{
+ (*o_putc)((c2 & 0x7f) | 0x080);
+ (*o_putc)(c1 | 0x080);
+ }
+ return;
+ } else {
+ if (encode_fallback) (*encode_fallback)(c1);
+ return;
+ }
+ }
+ }
+#endif
+ if (c2 == EOF) {
+ (*o_putc)(EOF);
+ return;
+ } else if (c2 == 0) {
+ output_mode = ASCII;
+ (*o_putc)(c1);
+ } else if (c2 == JIS_X_0201) {
+ output_mode = EUC_JP;
+ (*o_putc)(SSO); (*o_putc)(c1|0x80);
+ } else if (c2 == ISO_8859_1) {
+ output_mode = ISO_8859_1;
+ (*o_putc)(c1 | 0x080);
+#ifdef X0212_ENABLE
+ } else if (is_eucg3(c2)){
+ output_mode = EUC_JP;
+#ifdef SHIFTJIS_CP932
+ if (!cp932inv_f){
+ nkf_char s2, s1;
+ if (e2s_conv(c2, c1, &s2, &s1) == 0){
+ s2e_conv(s2, s1, &c2, &c1);
+ }
+ }
+#endif
+ if (c2 == 0) {
+ output_mode = ASCII;
+ (*o_putc)(c1);
+ }else if (is_eucg3(c2)){
+ if (x0212_f){
+ (*o_putc)(0x8f);
+ (*o_putc)((c2 & 0x7f) | 0x080);
+ (*o_putc)(c1 | 0x080);
+ }
+ }else{
+ (*o_putc)((c2 & 0x7f) | 0x080);
+ (*o_putc)(c1 | 0x080);
+ }
+#endif
+ } else {
+ if (!nkf_isgraph(c1) || !nkf_isgraph(c2)) {
+ set_iconv(FALSE, 0);
+ return; /* too late to rescue this char */
+ }
+ output_mode = EUC_JP;
+ (*o_putc)(c2 | 0x080);
+ (*o_putc)(c1 | 0x080);
+ }
+}
+
+void s_oconv(nkf_char c2, nkf_char c1)
+{
+#ifdef NUMCHAR_OPTION
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ w16e_conv(c1, &c2, &c1);
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ c2 = c1 & VALUE_MASK;
+ if (!x0213_f && 0xE000 <= c2 && c2 <= 0xE757) {
+ /* CP932 UDC */
+ c1 &= 0xFFF;
+ c2 = c1 / 188 + (cp932inv_f ? 0xF0 : 0xEB);
+ c1 = c1 % 188;
+ c1 += 0x40 + (c1 > 0x3e);
+ (*o_putc)(c2);
+ (*o_putc)(c1);
+ return;
+ } else {
+ if(encode_fallback)(*encode_fallback)(c1);
+ return;
+ }
+ }
+ }
+#endif
+ if (c2 == EOF) {
+ (*o_putc)(EOF);
+ return;
+ } else if (c2 == 0) {
+ output_mode = ASCII;
+ (*o_putc)(c1);
+ } else if (c2 == JIS_X_0201) {
+ output_mode = SHIFT_JIS;
+ (*o_putc)(c1|0x80);
+ } else if (c2 == ISO_8859_1) {
+ output_mode = ISO_8859_1;
+ (*o_putc)(c1 | 0x080);
+#ifdef X0212_ENABLE
+ } else if (is_eucg3(c2)){
+ output_mode = SHIFT_JIS;
+ if (e2s_conv(c2, c1, &c2, &c1) == 0){
+ (*o_putc)(c2);
+ (*o_putc)(c1);
+ }
+#endif
+ } else {
+ if (!nkf_isprint(c1) || !nkf_isprint(c2)) {
+ set_iconv(FALSE, 0);
+ return; /* too late to rescue this char */
+ }
+ output_mode = SHIFT_JIS;
+ e2s_conv(c2, c1, &c2, &c1);
+
+#ifdef SHIFTJIS_CP932
+ if (cp932inv_f
+ && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
+ nkf_char c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
+ if (c){
+ c2 = c >> 8;
+ c1 = c & 0xff;
+ }
+ }
+#endif /* SHIFTJIS_CP932 */
+
+ (*o_putc)(c2);
+ if (prefix_table[(unsigned char)c1]){
+ (*o_putc)(prefix_table[(unsigned char)c1]);
+ }
+ (*o_putc)(c1);
+ }
+}
+
+#ifdef UTF8_OUTPUT_ENABLE
void w_oconv(nkf_char c2, nkf_char c1)
{
nkf_char c0;
@@ -4175,392 +2439,675 @@ void w_oconv32(nkf_char c2, nkf_char c1)
}
#endif
-void e_oconv(nkf_char c2, nkf_char c1)
+#define SCORE_L2 (1) /* $BBh(B2$B?e=`4A;z(B */
+#define SCORE_KANA (SCORE_L2 << 1) /* $B$$$o$f$kH>3Q%+%J(B */
+#define SCORE_DEPEND (SCORE_KANA << 1) /* $B5!<o0MB8J8;z(B */
+#define SCORE_CP932 (SCORE_DEPEND << 1) /* CP932 $B$K$h$kFI$_49$((B (IBM extended characters) */
+#define SCORE_X0212 (SCORE_CP932 << 1) /* JIS X 0212 */
+#define SCORE_NO_EXIST (SCORE_X0212 << 1) /* $BB8:_$7$J$$J8;z(B */
+#define SCORE_iMIME (SCORE_NO_EXIST << 1) /* MIME $B$K$h$k;XDj(B */
+#define SCORE_ERROR (SCORE_iMIME << 1) /* $B%(%i!<(B */
+
+#define SCORE_INIT (SCORE_iMIME)
+
+static const char score_table_A0[] = {
+ 0, 0, 0, 0,
+ 0, 0, 0, 0,
+ 0, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
+ SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_NO_EXIST,
+};
+
+static const char score_table_F0[] = {
+ SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
+ SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
+ SCORE_DEPEND, SCORE_DEPEND, SCORE_CP932, SCORE_CP932,
+ SCORE_CP932, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR,
+};
+
+void set_code_score(struct input_code *ptr, nkf_char score)
{
-#ifdef NUMCHAR_OPTION
- if (c2 == 0 && is_unicode_capsule(c1)){
- w16e_conv(c1, &c2, &c1);
- if (c2 == 0 && is_unicode_capsule(c1)){
- c2 = c1 & VALUE_MASK;
- if (x0212_f && 0xE000 <= c2 && c2 <= 0xE757) {
- /* eucJP-ms UDC */
- c1 &= 0xFFF;
- c2 = c1 / 94;
- c2 += c2 < 10 ? 0x75 : 0x8FEB;
- c1 = 0x21 + c1 % 94;
- if (is_eucg3(c2)){
- (*o_putc)(0x8f);
- (*o_putc)((c2 & 0x7f) | 0x080);
- (*o_putc)(c1 | 0x080);
- }else{
- (*o_putc)((c2 & 0x7f) | 0x080);
- (*o_putc)(c1 | 0x080);
- }
- return;
- } else {
- if (encode_fallback) (*encode_fallback)(c1);
- return;
- }
- }
+ if (ptr){
+ ptr->score |= score;
}
+}
+
+void clr_code_score(struct input_code *ptr, nkf_char score)
+{
+ if (ptr){
+ ptr->score &= ~score;
+ }
+}
+
+void code_score(struct input_code *ptr)
+{
+ nkf_char c2 = ptr->buf[0];
+#ifdef UTF8_OUTPUT_ENABLE
+ nkf_char c1 = ptr->buf[1];
#endif
- if (c2 == EOF) {
- (*o_putc)(EOF);
- return;
- } else if (c2 == 0) {
- output_mode = ASCII;
- (*o_putc)(c1);
- } else if (c2 == JIS_X_0201) {
- output_mode = EUC_JP;
- (*o_putc)(SSO); (*o_putc)(c1|0x80);
- } else if (c2 == ISO_8859_1) {
- output_mode = ISO_8859_1;
- (*o_putc)(c1 | 0x080);
-#ifdef X0212_ENABLE
- } else if (is_eucg3(c2)){
- output_mode = EUC_JP;
-#ifdef SHIFTJIS_CP932
- if (!cp932inv_f){
- nkf_char s2, s1;
- if (e2s_conv(c2, c1, &s2, &s1) == 0){
- s2e_conv(s2, s1, &c2, &c1);
- }
- }
-#endif
- if (c2 == 0) {
- output_mode = ASCII;
- (*o_putc)(c1);
- }else if (is_eucg3(c2)){
- if (x0212_f){
- (*o_putc)(0x8f);
- (*o_putc)((c2 & 0x7f) | 0x080);
- (*o_putc)(c1 | 0x080);
- }
- }else{
- (*o_putc)((c2 & 0x7f) | 0x080);
- (*o_putc)(c1 | 0x080);
- }
+ if (c2 < 0){
+ set_code_score(ptr, SCORE_ERROR);
+ }else if (c2 == SSO){
+ set_code_score(ptr, SCORE_KANA);
+ }else if (c2 == 0x8f){
+ set_code_score(ptr, SCORE_X0212);
+#ifdef UTF8_OUTPUT_ENABLE
+ }else if (!e2w_conv(c2, c1)){
+ set_code_score(ptr, SCORE_NO_EXIST);
#endif
- } else {
- if (!nkf_isgraph(c1) || !nkf_isgraph(c2)) {
- set_iconv(FALSE, 0);
- return; /* too late to rescue this char */
- }
- output_mode = EUC_JP;
- (*o_putc)(c2 | 0x080);
- (*o_putc)(c1 | 0x080);
+ }else if ((c2 & 0x70) == 0x20){
+ set_code_score(ptr, score_table_A0[c2 & 0x0f]);
+ }else if ((c2 & 0x70) == 0x70){
+ set_code_score(ptr, score_table_F0[c2 & 0x0f]);
+ }else if ((c2 & 0x70) >= 0x50){
+ set_code_score(ptr, SCORE_L2);
}
}
-#ifdef X0212_ENABLE
-nkf_char x0212_shift(nkf_char c)
+void status_disable(struct input_code *ptr)
{
- nkf_char ret = c;
- c &= 0x7f;
- if (is_eucg3(ret)){
- if (0x75 <= c && c <= 0x7f){
- ret = c + (0x109 - 0x75);
- }
- }else{
- if (0x75 <= c && c <= 0x7f){
- ret = c + (0x113 - 0x75);
- }
- }
- return ret;
+ ptr->stat = -1;
+ ptr->buf[0] = -1;
+ code_score(ptr);
+ if (iconv == ptr->iconv_func) set_iconv(FALSE, 0);
}
+void status_push_ch(struct input_code *ptr, nkf_char c)
+{
+ ptr->buf[ptr->index++] = c;
+}
-nkf_char x0212_unshift(nkf_char c)
+void status_clear(struct input_code *ptr)
{
- nkf_char ret = c;
- if (0x7f <= c && c <= 0x88){
- ret = c + (0x75 - 0x7f);
- }else if (0x89 <= c && c <= 0x92){
- ret = PREFIX_EUCG3 | 0x80 | (c + (0x75 - 0x89));
+ ptr->stat = 0;
+ ptr->index = 0;
+}
+
+void status_reset(struct input_code *ptr)
+{
+ status_clear(ptr);
+ ptr->score = SCORE_INIT;
+}
+
+void status_reinit(struct input_code *ptr)
+{
+ status_reset(ptr);
+ ptr->_file_stat = 0;
+}
+
+void status_check(struct input_code *ptr, nkf_char c)
+{
+ if (c <= DEL && estab_f){
+ status_reset(ptr);
}
- return ret;
}
-#endif /* X0212_ENABLE */
-nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
+void s_status(struct input_code *ptr, nkf_char c)
{
- nkf_char ndx;
- if (is_eucg3(c2)){
- ndx = c2 & 0x7f;
- if (x0213_f){
- if((0x21 <= ndx && ndx <= 0x2F)){
- if (p2) *p2 = ((ndx - 1) >> 1) + 0xec - ndx / 8 * 3;
- if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
- return 0;
- }else if(0x6E <= ndx && ndx <= 0x7E){
- if (p2) *p2 = ((ndx - 1) >> 1) + 0xbe;
- if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
- return 0;
- }
- return 1;
- }
+ switch(ptr->stat){
+ case -1:
+ status_check(ptr, c);
+ break;
+ case 0:
+ if (c <= DEL){
+ break;
+#ifdef NUMCHAR_OPTION
+ }else if (is_unicode_capsule(c)){
+ break;
+#endif
+ }else if (0xa1 <= c && c <= 0xdf){
+ status_push_ch(ptr, SSO);
+ status_push_ch(ptr, c);
+ code_score(ptr);
+ status_clear(ptr);
+ }else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xea)){
+ ptr->stat = 1;
+ status_push_ch(ptr, c);
+ }else if (0xed <= c && c <= 0xee){
+ ptr->stat = 3;
+ status_push_ch(ptr, c);
+#ifdef SHIFTJIS_CP932
+ }else if (is_ibmext_in_sjis(c)){
+ ptr->stat = 2;
+ status_push_ch(ptr, c);
+#endif /* SHIFTJIS_CP932 */
#ifdef X0212_ENABLE
- else if(nkf_isgraph(ndx)){
- nkf_char val = 0;
- const unsigned short *ptr;
- ptr = x0212_shiftjis[ndx - 0x21];
- if (ptr){
- val = ptr[(c1 & 0x7f) - 0x21];
- }
- if (val){
- c2 = val >> 8;
- c1 = val & 0xff;
- if (p2) *p2 = c2;
- if (p1) *p1 = c1;
- return 0;
+ }else if (0xf0 <= c && c <= 0xfc){
+ ptr->stat = 1;
+ status_push_ch(ptr, c);
+#endif /* X0212_ENABLE */
+ }else{
+ status_disable(ptr);
+ }
+ break;
+ case 1:
+ if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
+ status_push_ch(ptr, c);
+ s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]);
+ code_score(ptr);
+ status_clear(ptr);
+ }else{
+ status_disable(ptr);
+ }
+ break;
+ case 2:
+#ifdef SHIFTJIS_CP932
+ if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)) {
+ status_push_ch(ptr, c);
+ if (s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]) == 0) {
+ set_code_score(ptr, SCORE_CP932);
+ status_clear(ptr);
+ break;
}
- c2 = x0212_shift(c2);
}
+#endif /* SHIFTJIS_CP932 */
+ status_disable(ptr);
+ break;
+ case 3:
+ if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
+ status_push_ch(ptr, c);
+ s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]);
+ set_code_score(ptr, SCORE_CP932);
+ status_clear(ptr);
+ }else{
+ status_disable(ptr);
+ }
+ break;
+ }
+}
+
+void e_status(struct input_code *ptr, nkf_char c)
+{
+ switch (ptr->stat){
+ case -1:
+ status_check(ptr, c);
+ break;
+ case 0:
+ if (c <= DEL){
+ break;
+#ifdef NUMCHAR_OPTION
+ }else if (is_unicode_capsule(c)){
+ break;
+#endif
+ }else if (SSO == c || (0xa1 <= c && c <= 0xfe)){
+ ptr->stat = 1;
+ status_push_ch(ptr, c);
+#ifdef X0212_ENABLE
+ }else if (0x8f == c){
+ ptr->stat = 2;
+ status_push_ch(ptr, c);
+#endif /* X0212_ENABLE */
+ }else{
+ status_disable(ptr);
+ }
+ break;
+ case 1:
+ if (0xa1 <= c && c <= 0xfe){
+ status_push_ch(ptr, c);
+ code_score(ptr);
+ status_clear(ptr);
+ }else{
+ status_disable(ptr);
+ }
+ break;
+#ifdef X0212_ENABLE
+ case 2:
+ if (0xa1 <= c && c <= 0xfe){
+ ptr->stat = 1;
+ status_push_ch(ptr, c);
+ }else{
+ status_disable(ptr);
+ }
#endif /* X0212_ENABLE */
}
- if(0x7F < c2) return 1;
- if (p2) *p2 = ((c2 - 1) >> 1) + ((c2 <= 0x5e) ? 0x71 : 0xb1);
- if (p1) *p1 = c1 + ((c2 & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
- return 0;
}
-void s_oconv(nkf_char c2, nkf_char c1)
+#ifdef UTF8_INPUT_ENABLE
+void w_status(struct input_code *ptr, nkf_char c)
{
+ switch (ptr->stat){
+ case -1:
+ status_check(ptr, c);
+ break;
+ case 0:
+ if (c <= DEL){
+ break;
#ifdef NUMCHAR_OPTION
- if (c2 == 0 && is_unicode_capsule(c1)){
- w16e_conv(c1, &c2, &c1);
- if (c2 == 0 && is_unicode_capsule(c1)){
- c2 = c1 & VALUE_MASK;
- if (!x0213_f && 0xE000 <= c2 && c2 <= 0xE757) {
- /* CP932 UDC */
- c1 &= 0xFFF;
- c2 = c1 / 188 + 0xF0;
- c1 = c1 % 188;
- c1 += 0x40 + (c1 > 0x3e);
- (*o_putc)(c2);
- (*o_putc)(c1);
- return;
+ }else if (is_unicode_capsule(c)){
+ break;
+#endif
+ }else if (0xc0 <= c && c <= 0xdf){
+ ptr->stat = 1;
+ status_push_ch(ptr, c);
+ }else if (0xe0 <= c && c <= 0xef){
+ ptr->stat = 2;
+ status_push_ch(ptr, c);
+ }else if (0xf0 <= c && c <= 0xf4){
+ ptr->stat = 3;
+ status_push_ch(ptr, c);
+ }else{
+ status_disable(ptr);
+ }
+ break;
+ case 1:
+ case 2:
+ if (0x80 <= c && c <= 0xbf){
+ status_push_ch(ptr, c);
+ if (ptr->index > ptr->stat){
+ int bom = (ptr->buf[0] == 0xef && ptr->buf[1] == 0xbb
+ && ptr->buf[2] == 0xbf);
+ w2e_conv(ptr->buf[0], ptr->buf[1], ptr->buf[2],
+ &ptr->buf[0], &ptr->buf[1]);
+ if (!bom){
+ code_score(ptr);
+ }
+ status_clear(ptr);
+ }
+ }else{
+ status_disable(ptr);
+ }
+ break;
+ case 3:
+ if (0x80 <= c && c <= 0xbf){
+ if (ptr->index < ptr->stat){
+ status_push_ch(ptr, c);
} else {
- if(encode_fallback)(*encode_fallback)(c1);
- return;
+ status_clear(ptr);
}
- }
+ }else{
+ status_disable(ptr);
+ }
+ break;
}
+}
#endif
- if (c2 == EOF) {
- (*o_putc)(EOF);
- return;
- } else if (c2 == 0) {
- output_mode = ASCII;
- (*o_putc)(c1);
- } else if (c2 == JIS_X_0201) {
- output_mode = SHIFT_JIS;
- (*o_putc)(c1|0x80);
- } else if (c2 == ISO_8859_1) {
- output_mode = ISO_8859_1;
- (*o_putc)(c1 | 0x080);
-#ifdef X0212_ENABLE
- } else if (is_eucg3(c2)){
- output_mode = SHIFT_JIS;
- if (e2s_conv(c2, c1, &c2, &c1) == 0){
- (*o_putc)(c2);
- (*o_putc)(c1);
- }
-#endif
- } else {
- if (!nkf_isprint(c1) || !nkf_isprint(c2)) {
- set_iconv(FALSE, 0);
- return; /* too late to rescue this char */
+
+void code_status(nkf_char c)
+{
+ int action_flag = 1;
+ struct input_code *result = 0;
+ struct input_code *p = input_code_list;
+ while (p->name){
+ if (!p->status_func) {
+ ++p;
+ continue;
+ }
+ if (!p->status_func)
+ continue;
+ (p->status_func)(p, c);
+ if (p->stat > 0){
+ action_flag = 0;
+ }else if(p->stat == 0){
+ if (result){
+ action_flag = 0;
+ }else{
+ result = p;
+ }
}
- output_mode = SHIFT_JIS;
- e2s_conv(c2, c1, &c2, &c1);
+ ++p;
+ }
-#ifdef SHIFTJIS_CP932
- if (cp932inv_f
- && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
- nkf_char c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
- if (c){
- c2 = c >> 8;
- c1 = c & 0xff;
+ if (action_flag){
+ if (result && !estab_f){
+ set_iconv(TRUE, result->iconv_func);
+ }else if (c <= DEL){
+ struct input_code *ptr = input_code_list;
+ while (ptr->name){
+ status_reset(ptr);
+ ++ptr;
}
}
-#endif /* SHIFTJIS_CP932 */
+ }
+}
- (*o_putc)(c2);
- if (prefix_table[(unsigned char)c1]){
- (*o_putc)(prefix_table[(unsigned char)c1]);
- }
- (*o_putc)(c1);
+#ifndef WIN32DLL
+nkf_char std_getc(FILE *f)
+{
+ if (std_gc_ndx){
+ return std_gc_buf[--std_gc_ndx];
}
+ return getc(f);
}
+#endif /*WIN32DLL*/
-void j_oconv(nkf_char c2, nkf_char c1)
+nkf_char std_ungetc(nkf_char c, FILE *f)
{
-#ifdef NUMCHAR_OPTION
- if (c2 == 0 && is_unicode_capsule(c1)){
- w16e_conv(c1, &c2, &c1);
- if (c2 == 0 && is_unicode_capsule(c1)){
- c2 = c1 & VALUE_MASK;
- if (ms_ucs_map_f && 0xE000 <= c2 && c2 <= 0xE757) {
- /* CP5022x UDC */
- c1 &= 0xFFF;
- c2 = 0x7F + c1 / 94;
- c1 = 0x21 + c1 % 94;
- } else {
- if (encode_fallback) (*encode_fallback)(c1);
- return;
- }
+ if (std_gc_ndx == STD_GC_BUFSIZE){
+ return EOF;
+ }
+ std_gc_buf[std_gc_ndx++] = c;
+ return c;
+}
+
+#ifndef WIN32DLL
+void std_putc(nkf_char c)
+{
+ if(c!=EOF)
+ putchar(c);
+}
+#endif /*WIN32DLL*/
+
+static unsigned char hold_buf[HOLD_SIZE*2];
+static int hold_count = 0;
+nkf_char push_hold_buf(nkf_char c2)
+{
+ if (hold_count >= HOLD_SIZE*2)
+ return (EOF);
+ hold_buf[hold_count++] = (unsigned char)c2;
+ return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count);
+}
+
+nkf_char
+h_conv(FILE *f, nkf_char c2, nkf_char c1)
+{
+ nkf_char ret, c3, c0;
+ int hold_index;
+
+
+ /** it must NOT be in the kanji shifte sequence */
+ /** it must NOT be written in JIS7 */
+ /** and it must be after 2 byte 8bit code */
+
+ hold_count = 0;
+ push_hold_buf(c2);
+ push_hold_buf(c1);
+
+ while ((c1 = (*i_getc)(f)) != EOF) {
+ if (c1 == ESC){
+ (*i_ungetc)(c1,f);
+ break;
+ }
+ code_status(c1);
+ if (push_hold_buf(c1) == EOF || estab_f){
+ break;
}
}
-#endif
- if (c2 == EOF) {
- if (output_mode !=ASCII && output_mode!=ISO_8859_1) {
- (*o_putc)(ESC);
- (*o_putc)('(');
- (*o_putc)(ascii_intro);
- output_mode = ASCII;
+
+ if (!estab_f){
+ struct input_code *p = input_code_list;
+ struct input_code *result = p;
+ if (c1 == EOF){
+ code_status(c1);
}
- (*o_putc)(EOF);
-#ifdef X0212_ENABLE
- } else if (is_eucg3(c2)){
- if(x0213_f){
- if(output_mode!=JIS_X_0213_2){
- output_mode = JIS_X_0213_2;
- (*o_putc)(ESC);
- (*o_putc)('$');
- (*o_putc)('(');
- (*o_putc)(0x50);
- }
- }else{
- if(output_mode!=JIS_X_0212){
- output_mode = JIS_X_0212;
- (*o_putc)(ESC);
- (*o_putc)('$');
- (*o_putc)('(');
- (*o_putc)(0x44);
- }
+ while (p->name){
+ if (p->status_func && p->score < result->score){
+ result = p;
+ }
+ ++p;
}
- (*o_putc)(c2 & 0x7f);
- (*o_putc)(c1);
+ set_iconv(TRUE, result->iconv_func);
+ }
+
+
+ /** now,
+ ** 1) EOF is detected, or
+ ** 2) Code is established, or
+ ** 3) Buffer is FULL (but last word is pushed)
+ **
+ ** in 1) and 3) cases, we continue to use
+ ** Kanji codes by oconv and leave estab_f unchanged.
+ **/
+
+ ret = c1;
+ hold_index = 0;
+ while (hold_index < hold_count){
+ c2 = hold_buf[hold_index++];
+ if (c2 <= DEL
+#ifdef NUMCHAR_OPTION
+ || is_unicode_capsule(c2)
#endif
- } else if (c2==JIS_X_0201) {
- if (output_mode!=JIS_X_0201) {
- output_mode = JIS_X_0201;
- (*o_putc)(ESC);
- (*o_putc)('(');
- (*o_putc)('I');
+ ){
+ (*iconv)(0, c2, 0);
+ continue;
+ }else if (iconv == s_iconv && 0xa1 <= c2 && c2 <= 0xdf){
+ (*iconv)(JIS_X_0201, c2, 0);
+ continue;
}
- (*o_putc)(c1);
- } else if (c2==ISO_8859_1) {
- /* iso8859 introduction, or 8th bit on */
- /* Can we convert in 7bit form using ESC-'-'-A ?
- Is this popular? */
- output_mode = ISO_8859_1;
- (*o_putc)(c1|0x80);
- } else if (c2 == 0) {
- if (output_mode !=ASCII && output_mode!=ISO_8859_1) {
- (*o_putc)(ESC);
- (*o_putc)('(');
- (*o_putc)(ascii_intro);
- output_mode = ASCII;
+ if (hold_index < hold_count){
+ c1 = hold_buf[hold_index++];
+ }else{
+ c1 = (*i_getc)(f);
+ if (c1 == EOF){
+ c3 = EOF;
+ break;
+ }
+ code_status(c1);
}
- (*o_putc)(c1);
- } else {
- if(ms_ucs_map_f
- ? c2<0x20 || 0x92<c2 || c1<0x20 || 0x7e<c1
- : c2<0x20 || 0x7e<c2 || c1<0x20 || 0x7e<c1) return;
- if(x0213_f){
- if (output_mode!=JIS_X_0213_1) {
- output_mode = JIS_X_0213_1;
- (*o_putc)(ESC);
- (*o_putc)('$');
- (*o_putc)('(');
- (*o_putc)(0x4F);
+ c0 = 0;
+ switch ((*iconv)(c2, c1, 0)) { /* can be EUC/SJIS/UTF-8 */
+ case -2:
+ /* 4 bytes UTF-8 */
+ if (hold_index < hold_count){
+ c0 = hold_buf[hold_index++];
+ } else if ((c0 = (*i_getc)(f)) == EOF) {
+ ret = EOF;
+ break;
+ } else {
+ code_status(c0);
+ c0 <<= 8;
+ if (hold_index < hold_count){
+ c3 = hold_buf[hold_index++];
+ } else if ((c3 = (*i_getc)(f)) == EOF) {
+ c0 = ret = EOF;
+ break;
+ } else {
+ code_status(c3);
+ (*iconv)(c2, c1, c0|c3);
+ }
+ }
+ break;
+ case -1:
+ /* 3 bytes EUC or UTF-8 */
+ if (hold_index < hold_count){
+ c0 = hold_buf[hold_index++];
+ } else if ((c0 = (*i_getc)(f)) == EOF) {
+ ret = EOF;
+ break;
+ } else {
+ code_status(c0);
+ }
+ (*iconv)(c2, c1, c0);
+ break;
+ }
+ if (c0 == EOF) break;
+ }
+ return ret;
+}
+
+/*
+ * Check and Ignore BOM
+ */
+void check_bom(FILE *f)
+{
+ int c2;
+ switch(c2 = (*i_getc)(f)){
+ case 0x00:
+ if((c2 = (*i_getc)(f)) == 0x00){
+ if((c2 = (*i_getc)(f)) == 0xFE){
+ if((c2 = (*i_getc)(f)) == 0xFF){
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv32);
+ }
+ if (iconv == w_iconv32) {
+ input_endian = ENDIAN_BIG;
+ return;
+ }
+ (*i_ungetc)(0xFF,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0xFE,f);
+ }else if(c2 == 0xFF){
+ if((c2 = (*i_getc)(f)) == 0xFE){
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv32);
+ }
+ if (iconv == w_iconv32) {
+ input_endian = ENDIAN_2143;
+ return;
+ }
+ (*i_ungetc)(0xFF,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0xFF,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0x00,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0x00,f);
+ break;
+ case 0xEF:
+ if((c2 = (*i_getc)(f)) == 0xBB){
+ if((c2 = (*i_getc)(f)) == 0xBF){
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv);
+ }
+ if (iconv == w_iconv) {
+ return;
+ }
+ (*i_ungetc)(0xBF,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0xBB,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0xEF,f);
+ break;
+ case 0xFE:
+ if((c2 = (*i_getc)(f)) == 0xFF){
+ if((c2 = (*i_getc)(f)) == 0x00){
+ if((c2 = (*i_getc)(f)) == 0x00){
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv32);
+ }
+ if (iconv == w_iconv32) {
+ input_endian = ENDIAN_3412;
+ return;
+ }
+ (*i_ungetc)(0x00,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0x00,f);
+ }else (*i_ungetc)(c2,f);
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv16);
}
- }else if (output_mode != JIS_X_0208) {
- output_mode = JIS_X_0208;
- (*o_putc)(ESC);
- (*o_putc)('$');
- (*o_putc)(kanji_intro);
- }
- (*o_putc)(c2);
- (*o_putc)(c1);
+ if (iconv == w_iconv16) {
+ input_endian = ENDIAN_BIG;
+ return;
+ }
+ (*i_ungetc)(0xFF,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0xFE,f);
+ break;
+ case 0xFF:
+ if((c2 = (*i_getc)(f)) == 0xFE){
+ if((c2 = (*i_getc)(f)) == 0x00){
+ if((c2 = (*i_getc)(f)) == 0x00){
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv32);
+ }
+ if (iconv == w_iconv32) {
+ input_endian = ENDIAN_LITTLE;
+ return;
+ }
+ (*i_ungetc)(0x00,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0x00,f);
+ }else (*i_ungetc)(c2,f);
+ if(!input_encoding){
+ set_iconv(TRUE, w_iconv16);
+ }
+ if (iconv == w_iconv16) {
+ input_endian = ENDIAN_LITTLE;
+ return;
+ }
+ (*i_ungetc)(0xFE,f);
+ }else (*i_ungetc)(c2,f);
+ (*i_ungetc)(0xFF,f);
+ break;
+ default:
+ (*i_ungetc)(c2,f);
+ break;
}
}
-void base64_conv(nkf_char c2, nkf_char c1)
+static struct {
+ int count;
+ nkf_char status;
+ nkf_char buf[3];
+} broken_state;
+
+static void init_broken_state(void)
{
- mime_prechar(c2, c1);
- (*o_base64conv)(c2,c1);
+ memset(&broken_state, 0, sizeof(broken_state));
}
+static void push_broken_buf(c)
+{
+ broken_state.buf[broken_state.count++] = c;
+}
+
+static nkf_char pop_broken_buf(void)
+{
+ return broken_state.buf[--broken_state.count];
+}
-static nkf_char broken_buf[3];
-static int broken_counter = 0;
-static int broken_last = 0;
nkf_char broken_getc(FILE *f)
{
- nkf_char c,c1;
+ nkf_char c, c1;
- if (broken_counter>0) {
- return broken_buf[--broken_counter];
+ if (broken_state.count > 0) {
+ return pop_broken_buf();
}
- c= (*i_bgetc)(f);
- if (c=='$' && broken_last != ESC
+ c = (*i_bgetc)(f);
+ if (c=='$' && broken_state.status != ESC
&& (input_mode==ASCII || input_mode==JIS_X_0201)) {
c1= (*i_bgetc)(f);
- broken_last = 0;
+ broken_state.status = 0;
if (c1=='@'|| c1=='B') {
- broken_buf[0]=c1; broken_buf[1]=c;
- broken_counter=2;
+ push_broken_buf(c1);
+ push_broken_buf(c);
return ESC;
} else {
(*i_bungetc)(c1,f);
return c;
}
- } else if (c=='(' && broken_last != ESC
+ } else if (c=='(' && broken_state.status != ESC
&& (input_mode==JIS_X_0208 || input_mode==JIS_X_0201)) { /* ) */
c1= (*i_bgetc)(f);
- broken_last = 0;
+ broken_state.status = 0;
if (c1=='J'|| c1=='B') {
- broken_buf[0]=c1; broken_buf[1]=c;
- broken_counter=2;
+ push_broken_buf(c1);
+ push_broken_buf(c);
return ESC;
} else {
(*i_bungetc)(c1,f);
return c;
}
} else {
- broken_last = c;
+ broken_state.status = c;
return c;
}
}
nkf_char broken_ungetc(nkf_char c, FILE *f)
{
- if (broken_counter<2)
- broken_buf[broken_counter++]=c;
+ if (broken_state.count < 2)
+ push_broken_buf(c);
return c;
}
-void nl_conv(nkf_char c2, nkf_char c1)
+void eol_conv(nkf_char c2, nkf_char c1)
{
- if (guess_f && input_newline != EOF) {
+ if (guess_f && input_eol != EOF) {
if (c2 == 0 && c1 == LF) {
- if (!input_newline) input_newline = prev_cr ? CRLF : LF;
- else if (input_newline != (prev_cr ? CRLF : LF)) input_newline = EOF;
- } else if (c2 == 0 && c1 == CR && input_newline == LF) input_newline = EOF;
+ if (!input_eol) input_eol = prev_cr ? CRLF : LF;
+ else if (input_eol != (prev_cr ? CRLF : LF)) input_eol = EOF;
+ } else if (c2 == 0 && c1 == CR && input_eol == LF) input_eol = EOF;
else if (!prev_cr);
- else if (!input_newline) input_newline = CR;
- else if (input_newline != CR) input_newline = EOF;
+ else if (!input_eol) input_eol = CR;
+ else if (input_eol != CR) input_eol = EOF;
}
if (prev_cr || (c2 == 0 && c1 == LF)) {
prev_cr = 0;
- if (nlmode_f != LF) (*o_nlconv)(0, CR);
- if (nlmode_f != CR) (*o_nlconv)(0, LF);
+ if (eolmode_f != LF) (*o_eol_conv)(0, CR);
+ if (eolmode_f != CR) (*o_eol_conv)(0, LF);
}
if (c2 == 0 && c1 == CR) prev_cr = CR;
- else if (c2 != 0 || c1 != LF) (*o_nlconv)(c2, c1);
+ else if (c2 != 0 || c1 != LF) (*o_eol_conv)(c2, c1);
}
/*
@@ -4971,6 +3518,7 @@ void hira_conv(nkf_char c2, nkf_char c1)
void iso2022jp_check_conv(nkf_char c2, nkf_char c1)
{
+#define RANGE_NUM_MAX 18
static const nkf_char range[RANGE_NUM_MAX][2] = {
{0x222f, 0x2239,},
{0x2242, 0x2249,},
@@ -5062,8 +3610,49 @@ static const nkf_char mime_encode_method[] = {
};
+/* MIME preprocessor fifo */
+
+#define MIME_BUF_SIZE (1024) /* 2^n ring buffer */
+#define MIME_BUF_MASK (MIME_BUF_SIZE-1)
+#define mime_input_buf(n) mime_input_state.buf[(n)&MIME_BUF_MASK]
+static struct {
+ unsigned char buf[MIME_BUF_SIZE];
+ unsigned int top;
+ unsigned int last; /* decoded */
+ unsigned int input; /* undecoded */
+} mime_input_state;
+static nkf_char (*mime_iconv_back)(nkf_char c2,nkf_char c1,nkf_char c0) = NULL;
+
#define MAXRECOVER 20
+static void mime_input_buf_unshift(nkf_char c)
+{
+ mime_input_buf(--mime_input_state.top) = (unsigned char)c;
+}
+
+nkf_char mime_ungetc(nkf_char c, FILE *f)
+{
+ mime_input_buf_unshift(c);
+ return c;
+}
+
+nkf_char mime_ungetc_buf(nkf_char c, FILE *f)
+{
+ if (mimebuf_f)
+ (*i_mungetc_buf)(c,f);
+ else
+ mime_input_buf(--mime_input_state.input) = (unsigned char)c;
+ return c;
+}
+
+nkf_char mime_getc_buf(FILE *f)
+{
+ /* we don't keep eof of mime_input_buf, becase it contains ?= as
+ a terminator. It was checked in mime_integrity. */
+ return ((mimebuf_f)?
+ (*i_mgetc_buf)(f):mime_input_buf(mime_input_state.input++));
+}
+
void switch_mime_getc(void)
{
if (i_getc!=mime_getc) {
@@ -5088,6 +3677,44 @@ void unswitch_mime_getc(void)
mime_iconv_back = NULL;
}
+nkf_char mime_integrity(FILE *f, const unsigned char *p)
+{
+ nkf_char c,d;
+ unsigned int q;
+ /* In buffered mode, read until =? or NL or buffer full
+ */
+ mime_input_state.input = mime_input_state.top;
+ mime_input_state.last = mime_input_state.top;
+
+ while(*p) mime_input_buf(mime_input_state.input++) = *p++;
+ d = 0;
+ q = mime_input_state.input;
+ while((c=(*i_getc)(f))!=EOF) {
+ if (((mime_input_state.input-mime_input_state.top)&MIME_BUF_MASK)==0) {
+ break; /* buffer full */
+ }
+ if (c=='=' && d=='?') {
+ /* checked. skip header, start decode */
+ mime_input_buf(mime_input_state.input++) = (unsigned char)c;
+ /* mime_last_input = mime_input_state.input; */
+ mime_input_state.input = q;
+ switch_mime_getc();
+ return 1;
+ }
+ if (!( (c=='+'||c=='/'|| c=='=' || c=='?' || is_alnum(c))))
+ break;
+ /* Should we check length mod 4? */
+ mime_input_buf(mime_input_state.input++) = (unsigned char)c;
+ d=c;
+ }
+ /* In case of Incomplete MIME, no MIME decode */
+ mime_input_buf(mime_input_state.input++) = (unsigned char)c;
+ mime_input_state.last = mime_input_state.input; /* point undecoded buffer */
+ mime_decode_mode = 1; /* no decode on mime_input_buf last in mime_getc */
+ switch_mime_getc(); /* anyway we need buffered getc */
+ return 1;
+}
+
nkf_char mime_begin_strict(FILE *f)
{
nkf_char c1 = 0;
@@ -5139,23 +3766,6 @@ nkf_char mime_begin_strict(FILE *f)
return c1;
}
-nkf_char mime_getc_buf(FILE *f)
-{
- /* we don't keep eof of Fifo, becase it contains ?= as
- a terminator. It was checked in mime_integrity. */
- return ((mimebuf_f)?
- (*i_mgetc_buf)(f):Fifo(mime_input++));
-}
-
-nkf_char mime_ungetc_buf(nkf_char c, FILE *f)
-{
- if (mimebuf_f)
- (*i_mungetc_buf)(c,f);
- else
- Fifo(--mime_input) = (unsigned char)c;
- return c;
-}
-
nkf_char mime_begin(FILE *f)
{
nkf_char c1;
@@ -5165,23 +3775,23 @@ nkf_char mime_begin(FILE *f)
/* re-read and convert again from mime_buffer. */
/* =? has been checked */
- k = mime_last;
- Fifo(mime_last++)='='; Fifo(mime_last++)='?';
+ k = mime_input_state.last;
+ mime_input_buf(mime_input_state.last++)='='; mime_input_buf(mime_input_state.last++)='?';
for(i=2;i<MAXRECOVER;i++) { /* start at =? */
/* We accept any character type even if it is breaked by new lines */
- c1 = (*i_getc)(f); Fifo(mime_last++) = (unsigned char)c1;
+ c1 = (*i_getc)(f); mime_input_buf(mime_input_state.last++) = (unsigned char)c1;
if (c1==LF||c1==SP||c1==CR||
c1=='-'||c1=='_'||is_alnum(c1)) continue;
if (c1=='=') {
/* Failed. But this could be another MIME preemble */
(*i_ungetc)(c1,f);
- mime_last--;
+ mime_input_state.last--;
break;
}
if (c1!='?') break;
else {
/* c1=='?' */
- c1 = (*i_getc)(f); Fifo(mime_last++) = (unsigned char)c1;
+ c1 = (*i_getc)(f); mime_input_buf(mime_input_state.last++) = (unsigned char)c1;
if (!(++i<MAXRECOVER) || c1==EOF) break;
if (c1=='b'||c1=='B') {
mime_decode_mode = 'B';
@@ -5190,7 +3800,7 @@ nkf_char mime_begin(FILE *f)
} else {
break;
}
- c1 = (*i_getc)(f); Fifo(mime_last++) = (unsigned char)c1;
+ c1 = (*i_getc)(f); mime_input_buf(mime_input_state.last++) = (unsigned char)c1;
if (!(++i<MAXRECOVER) || c1==EOF) break;
if (c1!='?') {
mime_decode_mode = FALSE;
@@ -5207,7 +3817,7 @@ nkf_char mime_begin(FILE *f)
return c1;
}
/* discard mime preemble, and goto MIME mode */
- mime_last = k;
+ mime_input_state.last = k;
/* do no MIME integrity check */
return c1; /* used only for checking EOF */
}
@@ -5274,10 +3884,10 @@ void print_guessed_code(char *filename)
} else {
printf("%s%s\n",
input_codename,
- input_newline == CR ? " (CR)" :
- input_newline == LF ? " (LF)" :
- input_newline == CRLF ? " (CRLF)" :
- input_newline == EOF ? " (MIXED NL)" :
+ input_eol == CR ? " (CR)" :
+ input_eol == LF ? " (LF)" :
+ input_eol == CRLF ? " (CRLF)" :
+ input_eol == EOF ? " (MIXED NL)" :
"");
}
}
@@ -5397,7 +4007,7 @@ nkf_char nfc_getc(FILE *f)
nkf_char (*u)(nkf_char c ,FILE *f) = i_nfc_ungetc;
int i=0, j, k=1, lower, upper;
nkf_char buf[9];
- const nkf_nfchar *array;
+ const unsigned char *array;
buf[i] = (*g)(f);
while (k > 0 && ((buf[i] & 0xc0) != 0x80)){
@@ -5434,6 +4044,27 @@ nkf_char nfc_ungetc(nkf_char c, FILE *f)
#endif /* UNICODE_NORMALIZATION */
+static nkf_char base64decode(nkf_char c)
+{
+ int i;
+ if (c > '@') {
+ if (c < '[') {
+ i = c - 'A'; /* A..Z 0-25 */
+ } else if (c == '_') {
+ i = '?' /* 63 */ ; /* _ 63 */
+ } else {
+ i = c - 'G' /* - 'a' + 26 */ ; /* a..z 26-51 */
+ }
+ } else if (c > '/') {
+ i = c - '0' + '4' /* - '0' + 52 */ ; /* 0..9 52-61 */
+ } else if (c == '+' || c == '-') {
+ i = '>' /* 62 */ ; /* + and - 62 */
+ } else {
+ i = '?' /* 63 */ ; /* / 63 */
+ }
+ return (i);
+}
+
nkf_char
mime_getc(FILE *f)
{
@@ -5444,8 +4075,8 @@ mime_getc(FILE *f)
char *lwsp_buf_new;
nkf_char lwsp_size = 128;
- if (mime_top != mime_last) { /* Something is in FIFO */
- return Fifo(mime_top++);
+ if (mime_input_state.top != mime_input_state.last) { /* Something is in FIFO */
+ return mime_input_buf(mime_input_state.top++);
}
if (mime_decode_mode==1 ||mime_decode_mode==FALSE) {
mime_decode_mode=FALSE;
@@ -5671,94 +4302,33 @@ mime_c4_retry:
t4 = 0x3f & base64decode(c4);
cc = ((t1 << 2) & 0x0fc) | ((t2 >> 4) & 0x03);
if (c2 != '=') {
- Fifo(mime_last++) = (unsigned char)cc;
+ mime_input_buf(mime_input_state.last++) = (unsigned char)cc;
cc = ((t2 << 4) & 0x0f0) | ((t3 >> 2) & 0x0f);
if (c3 != '=') {
- Fifo(mime_last++) = (unsigned char)cc;
+ mime_input_buf(mime_input_state.last++) = (unsigned char)cc;
cc = ((t3 << 6) & 0x0c0) | (t4 & 0x3f);
if (c4 != '=')
- Fifo(mime_last++) = (unsigned char)cc;
+ mime_input_buf(mime_input_state.last++) = (unsigned char)cc;
}
} else {
return c1;
}
- return Fifo(mime_top++);
-}
-
-nkf_char mime_ungetc(nkf_char c, FILE *f)
-{
- Fifo(--mime_top) = (unsigned char)c;
- return c;
-}
-
-nkf_char mime_integrity(FILE *f, const unsigned char *p)
-{
- nkf_char c,d;
- unsigned int q;
- /* In buffered mode, read until =? or NL or buffer full
- */
- mime_input = mime_top;
- mime_last = mime_top;
-
- while(*p) Fifo(mime_input++) = *p++;
- d = 0;
- q = mime_input;
- while((c=(*i_getc)(f))!=EOF) {
- if (((mime_input-mime_top)&MIME_BUF_MASK)==0) {
- break; /* buffer full */
- }
- if (c=='=' && d=='?') {
- /* checked. skip header, start decode */
- Fifo(mime_input++) = (unsigned char)c;
- /* mime_last_input = mime_input; */
- mime_input = q;
- switch_mime_getc();
- return 1;
- }
- if (!( (c=='+'||c=='/'|| c=='=' || c=='?' || is_alnum(c))))
- break;
- /* Should we check length mod 4? */
- Fifo(mime_input++) = (unsigned char)c;
- d=c;
- }
- /* In case of Incomplete MIME, no MIME decode */
- Fifo(mime_input++) = (unsigned char)c;
- mime_last = mime_input; /* point undecoded buffer */
- mime_decode_mode = 1; /* no decode on Fifo last in mime_getc */
- switch_mime_getc(); /* anyway we need buffered getc */
- return 1;
-}
-
-nkf_char base64decode(nkf_char c)
-{
- int i;
- if (c > '@') {
- if (c < '[') {
- i = c - 'A'; /* A..Z 0-25 */
- } else if (c == '_') {
- i = '?' /* 63 */ ; /* _ 63 */
- } else {
- i = c - 'G' /* - 'a' + 26 */ ; /* a..z 26-51 */
- }
- } else if (c > '/') {
- i = c - '0' + '4' /* - '0' + 52 */ ; /* 0..9 52-61 */
- } else if (c == '+' || c == '-') {
- i = '>' /* 62 */ ; /* + and - 62 */
- } else {
- i = '?' /* 63 */ ; /* / 63 */
- }
- return (i);
+ return mime_input_buf(mime_input_state.top++);
}
static const char basis_64[] =
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
-static nkf_char b64c;
#define MIMEOUT_BUF_LENGTH (60)
-char mimeout_buf[MIMEOUT_BUF_LENGTH+1];
-int mimeout_buf_count = 0;
+static struct {
+ char buf[MIMEOUT_BUF_LENGTH+1];
+ int count;
+ nkf_char state;
+} mimeout_state;
+
+/*nkf_char mime_lastchar2, mime_lastchar1;*/
-void open_mime(nkf_char mode)
+static void open_mime(nkf_char mode)
{
const unsigned char *p;
int i;
@@ -5773,23 +4343,23 @@ void open_mime(nkf_char mode)
mimeout_mode = mime_encode_method[i];
i = 0;
if (base64_count>45) {
- if (mimeout_buf_count>0 && nkf_isblank(mimeout_buf[i])){
- (*o_mputc)(mimeout_buf[i]);
+ if (mimeout_state.count>0 && nkf_isblank(mimeout_state.buf[i])){
+ (*o_mputc)(mimeout_state.buf[i]);
i++;
}
PUT_NEWLINE((*o_mputc));
(*o_mputc)(SP);
base64_count = 1;
- if (mimeout_buf_count>0
- && (mimeout_buf[i]==SP || mimeout_buf[i]==TAB
- || mimeout_buf[i]==CR || mimeout_buf[i]==LF)) {
+ if (mimeout_state.count>0
+ && (mimeout_state.buf[i]==SP || mimeout_state.buf[i]==TAB
+ || mimeout_state.buf[i]==CR || mimeout_state.buf[i]==LF)) {
i++;
}
}
- for (;i<mimeout_buf_count;i++) {
- if (mimeout_buf[i]==SP || mimeout_buf[i]==TAB
- || mimeout_buf[i]==CR || mimeout_buf[i]==LF) {
- (*o_mputc)(mimeout_buf[i]);
+ for (;i<mimeout_state.count;i++) {
+ if (mimeout_state.buf[i]==SP || mimeout_state.buf[i]==TAB
+ || mimeout_state.buf[i]==CR || mimeout_state.buf[i]==LF) {
+ (*o_mputc)(mimeout_state.buf[i]);
base64_count ++;
} else {
break;
@@ -5799,14 +4369,46 @@ void open_mime(nkf_char mode)
(*o_mputc)(*p++);
base64_count ++;
}
- j = mimeout_buf_count;
- mimeout_buf_count = 0;
+ j = mimeout_state.count;
+ mimeout_state.count = 0;
for (;i<j;i++) {
- mime_putc(mimeout_buf[i]);
+ mime_putc(mimeout_state.buf[i]);
+ }
+}
+
+static void mime_prechar(nkf_char c2, nkf_char c1)
+{
+ if (mimeout_mode > 0){
+ if (c2 == EOF){
+ if (base64_count + mimeout_state.count/3*4> 73){
+ (*o_base64conv)(EOF,0);
+ OCONV_NEWLINE((*o_base64conv));
+ (*o_base64conv)(0,SP);
+ base64_count = 1;
+ }
+ } else {
+ if (base64_count + mimeout_state.count/3*4> 66) {
+ (*o_base64conv)(EOF,0);
+ OCONV_NEWLINE((*o_base64conv));
+ (*o_base64conv)(0,SP);
+ base64_count = 1;
+ mimeout_mode = -1;
+ }
+ }
+ } else if (c2) {
+ if (c2 != EOF && base64_count + mimeout_state.count/3*4> 60) {
+ mimeout_mode = (output_mode==ASCII ||output_mode == ISO_8859_1) ? 'Q' : 'B';
+ open_mime(output_mode);
+ (*o_base64conv)(EOF,0);
+ OCONV_NEWLINE((*o_base64conv));
+ (*o_base64conv)(0,SP);
+ base64_count = 1;
+ mimeout_mode = -1;
+ }
}
}
-void close_mime(void)
+static void close_mime(void)
{
(*o_mputc)('?');
(*o_mputc)('=');
@@ -5814,20 +4416,20 @@ void close_mime(void)
mimeout_mode = 0;
}
-void eof_mime(void)
+static void eof_mime(void)
{
switch(mimeout_mode) {
case 'Q':
case 'B':
break;
case 2:
- (*o_mputc)(basis_64[((b64c & 0x3)<< 4)]);
+ (*o_mputc)(basis_64[((mimeout_state.state & 0x3)<< 4)]);
(*o_mputc)('=');
(*o_mputc)('=');
base64_count += 3;
break;
case 1:
- (*o_mputc)(basis_64[((b64c & 0xF) << 2)]);
+ (*o_mputc)(basis_64[((mimeout_state.state & 0xF) << 2)]);
(*o_mputc)('=');
base64_count += 2;
break;
@@ -5840,7 +4442,7 @@ void eof_mime(void)
}
}
-void mimeout_addchar(nkf_char c)
+static void mimeout_addchar(nkf_char c)
{
switch(mimeout_mode) {
case 'Q':
@@ -5858,19 +4460,19 @@ void mimeout_addchar(nkf_char c)
}
break;
case 'B':
- b64c=c;
+ mimeout_state.state=c;
(*o_mputc)(basis_64[c>>2]);
mimeout_mode=2;
base64_count ++;
break;
case 2:
- (*o_mputc)(basis_64[((b64c & 0x3)<< 4) | ((c & 0xF0) >> 4)]);
- b64c=c;
+ (*o_mputc)(basis_64[((mimeout_state.state & 0x3)<< 4) | ((c & 0xF0) >> 4)]);
+ mimeout_state.state=c;
mimeout_mode=1;
base64_count ++;
break;
case 1:
- (*o_mputc)(basis_64[((b64c & 0xF) << 2) | ((c & 0xC0) >>6)]);
+ (*o_mputc)(basis_64[((mimeout_state.state & 0xF) << 2) | ((c & 0xC0) >>6)]);
(*o_mputc)(basis_64[c & 0x3F]);
mimeout_mode='B';
base64_count += 2;
@@ -5882,41 +4484,7 @@ void mimeout_addchar(nkf_char c)
}
}
-/*nkf_char mime_lastchar2, mime_lastchar1;*/
-
-void mime_prechar(nkf_char c2, nkf_char c1)
-{
- if (mimeout_mode > 0){
- if (c2 == EOF){
- if (base64_count + mimeout_buf_count/3*4> 73){
- (*o_base64conv)(EOF,0);
- OCONV_NEWLINE((*o_base64conv));
- (*o_base64conv)(0,SP);
- base64_count = 1;
- }
- } else {
- if (base64_count + mimeout_buf_count/3*4> 66) {
- (*o_base64conv)(EOF,0);
- OCONV_NEWLINE((*o_base64conv));
- (*o_base64conv)(0,SP);
- base64_count = 1;
- mimeout_mode = -1;
- }
- }
- } else if (c2) {
- if (c2 != EOF && base64_count + mimeout_buf_count/3*4> 60) {
- mimeout_mode = (output_mode==ASCII ||output_mode == ISO_8859_1) ? 'Q' : 'B';
- open_mime(output_mode);
- (*o_base64conv)(EOF,0);
- OCONV_NEWLINE((*o_base64conv));
- (*o_base64conv)(0,SP);
- base64_count = 1;
- mimeout_mode = -1;
- }
- }
-}
-
-void mime_putc(nkf_char c)
+static void mime_putc(nkf_char c)
{
int i, j;
nkf_char lastchar;
@@ -5949,38 +4517,38 @@ void mime_putc(nkf_char c)
/* mimeout_f != FIXED_MIME */
if (c == EOF) { /* c==EOF */
- if (mimeout_mode == -1 && mimeout_buf_count > 1) open_mime(output_mode);
- j = mimeout_buf_count;
- mimeout_buf_count = 0;
+ if (mimeout_mode == -1 && mimeout_state.count > 1) open_mime(output_mode);
+ j = mimeout_state.count;
+ mimeout_state.count = 0;
i = 0;
if (mimeout_mode > 0) {
- if (!nkf_isblank(mimeout_buf[j-1])) {
+ if (!nkf_isblank(mimeout_state.buf[j-1])) {
for (;i<j;i++) {
- if (nkf_isspace(mimeout_buf[i]) && base64_count < 71){
+ if (nkf_isspace(mimeout_state.buf[i]) && base64_count < 71){
break;
}
- mimeout_addchar(mimeout_buf[i]);
+ mimeout_addchar(mimeout_state.buf[i]);
}
eof_mime();
for (;i<j;i++) {
- mimeout_addchar(mimeout_buf[i]);
+ mimeout_addchar(mimeout_state.buf[i]);
}
} else {
for (;i<j;i++) {
- mimeout_addchar(mimeout_buf[i]);
+ mimeout_addchar(mimeout_state.buf[i]);
}
eof_mime();
}
} else {
for (;i<j;i++) {
- mimeout_addchar(mimeout_buf[i]);
+ mimeout_addchar(mimeout_state.buf[i]);
}
}
return;
}
- if (mimeout_buf_count > 0){
- lastchar = mimeout_buf[mimeout_buf_count - 1];
+ if (mimeout_state.count > 0){
+ lastchar = mimeout_state.buf[mimeout_state.count - 1];
}else{
lastchar = -1;
}
@@ -6036,9 +4604,9 @@ void mime_putc(nkf_char c)
base64_count = 0;
}
}
- for (i=0;i<mimeout_buf_count;i++) {
- (*o_mputc)(mimeout_buf[i]);
- if (mimeout_buf[i] == CR || mimeout_buf[i] == LF){
+ for (i=0;i<mimeout_state.count;i++) {
+ (*o_mputc)(mimeout_state.buf[i]);
+ if (mimeout_state.buf[i] == CR || mimeout_state.buf[i] == LF){
base64_count = 0;
}else{
base64_count++;
@@ -6049,40 +4617,40 @@ void mime_putc(nkf_char c)
base64_count = 0;
mimeout_mode = 0;
}
- mimeout_buf[0] = (char)c;
- mimeout_buf_count = 1;
+ mimeout_state.buf[0] = (char)c;
+ mimeout_state.count = 1;
}else{
if (base64_count > 1
- && base64_count + mimeout_buf_count > 76
- && mimeout_buf[0] != CR && mimeout_buf[0] != LF){
+ && base64_count + mimeout_state.count > 76
+ && mimeout_state.buf[0] != CR && mimeout_state.buf[0] != LF){
PUT_NEWLINE((*o_mputc));
base64_count = 0;
- if (!nkf_isspace(mimeout_buf[0])){
+ if (!nkf_isspace(mimeout_state.buf[0])){
(*o_mputc)(SP);
base64_count++;
}
}
- mimeout_buf[mimeout_buf_count++] = (char)c;
- if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
+ mimeout_state.buf[mimeout_state.count++] = (char)c;
+ if (mimeout_state.count>MIMEOUT_BUF_LENGTH) {
open_mime(output_mode);
}
}
return;
}else{
if (lastchar==CR || lastchar == LF){
- for (i=0;i<mimeout_buf_count;i++) {
- (*o_mputc)(mimeout_buf[i]);
+ for (i=0;i<mimeout_state.count;i++) {
+ (*o_mputc)(mimeout_state.buf[i]);
}
base64_count = 0;
- mimeout_buf_count = 0;
+ mimeout_state.count = 0;
}
if (lastchar==SP) {
- for (i=0;i<mimeout_buf_count-1;i++) {
- (*o_mputc)(mimeout_buf[i]);
+ for (i=0;i<mimeout_state.count-1;i++) {
+ (*o_mputc)(mimeout_state.buf[i]);
base64_count++;
}
- mimeout_buf[0] = SP;
- mimeout_buf_count = 1;
+ mimeout_state.buf[0] = SP;
+ mimeout_state.count = 1;
}
open_mime(output_mode);
}
@@ -6091,69 +4659,69 @@ void mime_putc(nkf_char c)
if ( c<=DEL && (output_mode==ASCII ||output_mode == ISO_8859_1)) {
if (lastchar == CR || lastchar == LF){
if (nkf_isblank(c)) {
- for (i=0;i<mimeout_buf_count;i++) {
- mimeout_addchar(mimeout_buf[i]);
+ for (i=0;i<mimeout_state.count;i++) {
+ mimeout_addchar(mimeout_state.buf[i]);
}
- mimeout_buf_count = 0;
+ mimeout_state.count = 0;
} else if (SP<c && c<DEL) {
eof_mime();
- for (i=0;i<mimeout_buf_count;i++) {
- (*o_mputc)(mimeout_buf[i]);
+ for (i=0;i<mimeout_state.count;i++) {
+ (*o_mputc)(mimeout_state.buf[i]);
}
base64_count = 0;
- mimeout_buf_count = 0;
+ mimeout_state.count = 0;
}
- mimeout_buf[mimeout_buf_count++] = (char)c;
+ mimeout_state.buf[mimeout_state.count++] = (char)c;
return;
}
if (c==SP || c==TAB || c==CR || c==LF) {
- for (i=0;i<mimeout_buf_count;i++) {
- if (SP<mimeout_buf[i] && mimeout_buf[i]<DEL) {
+ for (i=0;i<mimeout_state.count;i++) {
+ if (SP<mimeout_state.buf[i] && mimeout_state.buf[i]<DEL) {
eof_mime();
- for (i=0;i<mimeout_buf_count;i++) {
- (*o_mputc)(mimeout_buf[i]);
+ for (i=0;i<mimeout_state.count;i++) {
+ (*o_mputc)(mimeout_state.buf[i]);
base64_count++;
}
- mimeout_buf_count = 0;
+ mimeout_state.count = 0;
}
}
- mimeout_buf[mimeout_buf_count++] = (char)c;
- if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
+ mimeout_state.buf[mimeout_state.count++] = (char)c;
+ if (mimeout_state.count>MIMEOUT_BUF_LENGTH) {
eof_mime();
- for (i=0;i<mimeout_buf_count;i++) {
- (*o_mputc)(mimeout_buf[i]);
+ for (i=0;i<mimeout_state.count;i++) {
+ (*o_mputc)(mimeout_state.buf[i]);
base64_count++;
}
- mimeout_buf_count = 0;
+ mimeout_state.count = 0;
}
return;
}
- if (mimeout_buf_count>0 && SP<c && c!='=') {
- mimeout_buf[mimeout_buf_count++] = (char)c;
- if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
- j = mimeout_buf_count;
- mimeout_buf_count = 0;
+ if (mimeout_state.count>0 && SP<c && c!='=') {
+ mimeout_state.buf[mimeout_state.count++] = (char)c;
+ if (mimeout_state.count>MIMEOUT_BUF_LENGTH) {
+ j = mimeout_state.count;
+ mimeout_state.count = 0;
for (i=0;i<j;i++) {
- mimeout_addchar(mimeout_buf[i]);
+ mimeout_addchar(mimeout_state.buf[i]);
}
}
return;
}
}
}
- if (mimeout_buf_count>0) {
- j = mimeout_buf_count;
- mimeout_buf_count = 0;
+ if (mimeout_state.count>0) {
+ j = mimeout_state.count;
+ mimeout_state.count = 0;
for (i=0;i<j;i++) {
- if (mimeout_buf[i]==CR || mimeout_buf[i]==LF)
+ if (mimeout_state.buf[i]==CR || mimeout_state.buf[i]==LF)
break;
- mimeout_addchar(mimeout_buf[i]);
+ mimeout_addchar(mimeout_state.buf[i]);
}
if (i<j) {
eof_mime();
base64_count=0;
for (;i<j;i++) {
- (*o_mputc)(mimeout_buf[i]);
+ (*o_mputc)(mimeout_state.buf[i]);
}
open_mime(output_mode);
}
@@ -6161,6 +4729,105 @@ void mime_putc(nkf_char c)
mimeout_addchar(c);
}
+void base64_conv(nkf_char c2, nkf_char c1)
+{
+ mime_prechar(c2, c1);
+ (*o_base64conv)(c2,c1);
+}
+
+#ifdef HAVE_ICONV_H
+typedef struct nkf_iconv_t {
+ iconv_t cd;
+ char *input_buffer;
+ size_t input_buffer_size;
+ char *output_buffer;
+ size_t output_buffer_size;
+}
+
+nkf_iconv_t nkf_iconv_new(char *tocode, char *fromcode)
+{
+ nkf_iconv_t converter;
+
+ converter->input_buffer_size = IOBUF_SIZE;
+ converter->input_buffer = malloc(converter->input_buffer_size);
+ if (converter->input_buffer == NULL)
+ perror("can't malloc");
+
+ converter->output_buffer_size = IOBUF_SIZE * 2;
+ converter->output_buffer = malloc(converter->output_buffer_size);
+ if (converter->output_buffer == NULL)
+ perror("can't malloc");
+
+ converter->cd = iconv_open(tocode, fromcode);
+ if (converter->cd == (iconv_t)-1)
+ {
+ switch (errno) {
+ case EINVAL:
+ perror(fprintf("iconv doesn't support %s to %s conversion.", fromcode, tocode));
+ return -1;
+ default:
+ perror("can't iconv_open");
+ }
+ }
+}
+
+size_t nkf_iconv_convert(nkf_iconv_t *converter, FILE *input)
+{
+ size_t invalid = (size_t)0;
+ char *input_buffer = converter->input_buffer;
+ size_t input_length = (size_t)0;
+ char *output_buffer = converter->output_buffer;
+ size_t output_length = converter->output_buffer_size;
+ int c;
+
+ do {
+ if (c != EOF) {
+ while ((c = (*i_getc)(f)) != EOF) {
+ input_buffer[input_length++] = c;
+ if (input_length < converter->input_buffer_size) break;
+ }
+ }
+
+ size_t ret = iconv(converter->cd, &input_buffer, &input_length, &output_buffer, &output_length);
+ while (output_length-- > 0) {
+ (*o_putc)(output_buffer[converter->output_buffer_size-output_length]);
+ }
+ if (ret == (size_t) - 1) {
+ switch (errno) {
+ case EINVAL:
+ if (input_buffer != converter->input_buffer)
+ memmove(converter->input_buffer, input_buffer, input_length);
+ break;
+ case E2BIG:
+ converter->output_buffer_size *= 2;
+ output_buffer = realloc(converter->outbuf, converter->output_buffer_size);
+ if (output_buffer == NULL) {
+ perror("can't realloc");
+ return -1;
+ }
+ converter->output_buffer = output_buffer;
+ break;
+ default:
+ perror("can't iconv");
+ return -1;
+ }
+ } else {
+ invalid += ret;
+ }
+ } while (1);
+
+ return invalid;
+}
+
+
+void nkf_iconv_close(nkf_iconv_t *convert)
+{
+ free(converter->inbuf);
+ free(converter->outbuf);
+ iconv_close(converter->cd);
+}
+#endif
+
void reinit(void)
{
@@ -6230,7 +4897,7 @@ void reinit(void)
}
}
hold_count = 0;
- mimeout_buf_count = 0;
+ mimeout_state.count = 0;
mimeout_mode = 0;
base64_count = 0;
f_line = 0;
@@ -6241,10 +4908,9 @@ void reinit(void)
kanji_intro = DEFAULT_J;
ascii_intro = DEFAULT_R;
fold_margin = FOLD_MARGIN;
- oconv = DEFAULT_CONV;
o_zconv = no_connection;
o_fconv = no_connection;
- o_nlconv = no_connection;
+ o_eol_conv = no_connection;
o_rot_conv = no_connection;
o_hira_conv = no_connection;
o_base64conv = no_connection;
@@ -6264,167 +4930,1439 @@ void reinit(void)
shift_mode = FALSE;
mime_decode_mode = FALSE;
file_out_f = FALSE;
- nlmode_f = 0;
- input_newline = 0;
+ eolmode_f = 0;
+ input_eol = 0;
prev_cr = 0;
option_mode = 0;
- broken_counter = 0;
- broken_last = 0;
+ init_broken_state();
z_prev2=0,z_prev1=0;
#ifdef CHECK_OPTION
iconv_for_check = 0;
#endif
input_codename = NULL;
input_encoding = NULL;
- output_encoding = nkf_enc_from_index(DEFAULT_ENCODING);
+ output_encoding = NULL;
#ifdef WIN32DLL
reinitdll();
#endif /*WIN32DLL*/
}
-void no_connection(nkf_char c2, nkf_char c1)
+void module_connection(void)
{
- no_connection2(c2,c1,0);
+ if (input_encoding) set_input_encoding(input_encoding);
+ if (!output_encoding) {
+ output_encoding = nkf_default_encoding();
+ }
+ set_output_encoding(output_encoding);
+ oconv = nkf_enc_to_oconv(output_encoding);
+ o_putc = std_putc;
+
+ /* replace continucation module, from output side */
+
+ /* output redicrection */
+#ifdef CHECK_OPTION
+ if (noout_f || guess_f){
+ o_putc = no_putc;
+ }
+#endif
+ if (mimeout_f) {
+ o_mputc = o_putc;
+ o_putc = mime_putc;
+ if (mimeout_f == TRUE) {
+ o_base64conv = oconv; oconv = base64_conv;
+ }
+ /* base64_count = 0; */
+ }
+
+ if (eolmode_f || guess_f) {
+ o_eol_conv = oconv; oconv = eol_conv;
+ }
+ if (rot_f) {
+ o_rot_conv = oconv; oconv = rot_conv;
+ }
+ if (iso2022jp_f) {
+ o_iso2022jp_check_conv = oconv; oconv = iso2022jp_check_conv;
+ }
+ if (hira_f) {
+ o_hira_conv = oconv; oconv = hira_conv;
+ }
+ if (fold_f) {
+ o_fconv = oconv; oconv = fold_conv;
+ f_line = 0;
+ }
+ if (alpha_f || x0201_f) {
+ o_zconv = oconv; oconv = z_conv;
+ }
+
+ i_getc = std_getc;
+ i_ungetc = std_ungetc;
+ /* input redicrection */
+#ifdef INPUT_OPTION
+ if (cap_f){
+ i_cgetc = i_getc; i_getc = cap_getc;
+ i_cungetc = i_ungetc; i_ungetc= cap_ungetc;
+ }
+ if (url_f){
+ i_ugetc = i_getc; i_getc = url_getc;
+ i_uungetc = i_ungetc; i_ungetc= url_ungetc;
+ }
+#endif
+#ifdef NUMCHAR_OPTION
+ if (numchar_f){
+ i_ngetc = i_getc; i_getc = numchar_getc;
+ i_nungetc = i_ungetc; i_ungetc= numchar_ungetc;
+ }
+#endif
+#ifdef UNICODE_NORMALIZATION
+ if (nfc_f){
+ i_nfc_getc = i_getc; i_getc = nfc_getc;
+ i_nfc_ungetc = i_ungetc; i_ungetc= nfc_ungetc;
+ }
+#endif
+ if (mime_f && mimebuf_f==FIXED_MIME) {
+ i_mgetc = i_getc; i_getc = mime_getc;
+ i_mungetc = i_ungetc; i_ungetc = mime_ungetc;
+ }
+ if (broken_f & 1) {
+ i_bgetc = i_getc; i_getc = broken_getc;
+ i_bungetc = i_ungetc; i_ungetc = broken_ungetc;
+ }
+ if (input_encoding) {
+ set_iconv(-TRUE, nkf_enc_to_iconv(input_encoding));
+ } else {
+ set_iconv(FALSE, e_iconv);
+ }
+
+ {
+ struct input_code *p = input_code_list;
+ while (p->name){
+ status_reinit(p++);
+ }
+ }
}
-nkf_char no_connection2(nkf_char c2, nkf_char c1, nkf_char c0)
+/*
+ Conversion main loop. Code detection only.
+ */
+
+#if !defined(PERL_XS) && !defined(WIN32DLL)
+nkf_char noconvert(FILE *f)
{
- fprintf(stderr,"nkf internal module connection failure.\n");
- exit(1);
- return 0; /* LINT */
-}
+ nkf_char c;
-#ifndef PERL_XS
-#ifdef WIN32DLL
-#define fprintf dllprintf
+ if (nop_f == 2)
+ module_connection();
+ while ((c = (*i_getc)(f)) != EOF)
+ (*o_putc)(c);
+ (*o_putc)(EOF);
+ return 1;
+}
#endif
-void usage(void)
+
+nkf_char kanji_convert(FILE *f)
{
- fprintf(HELP_OUTPUT,"USAGE: nkf(nkf32,wnkf,nkf2) -[flags] [in file] .. [out file for -O flag]\n");
- fprintf(HELP_OUTPUT,"Flags:\n");
- fprintf(HELP_OUTPUT,"b,u Output is buffered (DEFAULT),Output is unbuffered\n");
-#ifdef DEFAULT_CODE_SJIS
- fprintf(HELP_OUTPUT,"j,s,e,w Output code is JIS 7 bit, Shift_JIS (DEFAULT), EUC-JP, UTF-8N\n");
+ nkf_char c3, c2=0, c1, c0=0;
+ int is_8bit = FALSE;
+
+ if (input_encoding && !nkf_enc_asciicompat(input_encoding)) {
+ is_8bit = TRUE;
+ }
+
+ input_mode = ASCII;
+ output_mode = ASCII;
+ shift_mode = FALSE;
+
+#define NEXT continue /* no output, get next */
+#define SEND ; /* output c1 and c2, get next */
+#define LAST break /* end of loop, go closing */
+
+ module_connection();
+ check_bom(f);
+
+ while ((c1 = (*i_getc)(f)) != EOF) {
+#ifdef INPUT_CODE_FIX
+ if (!input_encoding)
#endif
-#ifdef DEFAULT_CODE_JIS
- fprintf(HELP_OUTPUT,"j,s,e,w Output code is JIS 7 bit (DEFAULT), Shift JIS, EUC-JP, UTF-8N\n");
+ code_status(c1);
+ if (c2) {
+ /* second byte */
+ if (c2 > ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) {
+ /* in case of 8th bit is on */
+ if (!estab_f&&!mime_decode_mode) {
+ /* in case of not established yet */
+ /* It is still ambiguious */
+ if (h_conv(f, c2, c1)==EOF)
+ LAST;
+ else
+ c2 = 0;
+ NEXT;
+ } else {
+ /* in case of already established */
+ if (c1 < AT) {
+ /* ignore bogus code and not CP5022x UCD */
+ c2 = 0;
+ NEXT;
+ } else {
+ SEND;
+ }
+ }
+ } else
+ /* second byte, 7 bit code */
+ /* it might be kanji shitfted */
+ if ((c1 == DEL) || (c1 <= SP)) {
+ /* ignore bogus first code */
+ c2 = 0;
+ NEXT;
+ } else
+ SEND;
+ } else {
+ /* first byte */
+#ifdef UTF8_INPUT_ENABLE
+ if (iconv == w_iconv16) {
+ if (input_endian == ENDIAN_BIG) {
+ c2 = c1;
+ if ((c1 = (*i_getc)(f)) != EOF) {
+ if (0xD8 <= c2 && c2 <= 0xDB) {
+ if ((c0 = (*i_getc)(f)) != EOF) {
+ c0 <<= 8;
+ if ((c3 = (*i_getc)(f)) != EOF) {
+ c0 |= c3;
+ } else c2 = EOF;
+ } else c2 = EOF;
+ }
+ } else c2 = EOF;
+ } else {
+ if ((c2 = (*i_getc)(f)) != EOF) {
+ if (0xD8 <= c2 && c2 <= 0xDB) {
+ if ((c3 = (*i_getc)(f)) != EOF) {
+ if ((c0 = (*i_getc)(f)) != EOF) {
+ c0 <<= 8;
+ c0 |= c3;
+ } else c2 = EOF;
+ } else c2 = EOF;
+ }
+ } else c2 = EOF;
+ }
+ SEND;
+ } else if(iconv == w_iconv32){
+ int c3 = c1;
+ if((c2 = (*i_getc)(f)) != EOF &&
+ (c1 = (*i_getc)(f)) != EOF &&
+ (c0 = (*i_getc)(f)) != EOF){
+ switch(input_endian){
+ case ENDIAN_BIG:
+ c1 = (c2&0xFF)<<16 | (c1&0xFF)<<8 | (c0&0xFF);
+ break;
+ case ENDIAN_LITTLE:
+ c1 = (c3&0xFF) | (c2&0xFF)<<8 | (c1&0xFF)<<16;
+ break;
+ case ENDIAN_2143:
+ c1 = (c3&0xFF)<<16 | (c1&0xFF) | (c0&0xFF)<<8;
+ break;
+ case ENDIAN_3412:
+ c1 = (c3&0xFF)<<8 | (c2&0xFF) | (c0&0xFF)<<16;
+ break;
+ }
+ c2 = 0;
+ }else{
+ c2 = EOF;
+ }
+ SEND;
+ } else
#endif
-#ifdef DEFAULT_CODE_EUC
- fprintf(HELP_OUTPUT,"j,s,e,w Output code is JIS 7 bit, Shift JIS, EUC-JP (DEFAULT), UTF-8N\n");
+#ifdef NUMCHAR_OPTION
+ if (is_unicode_capsule(c1)){
+ SEND;
+ } else
#endif
-#ifdef DEFAULT_CODE_UTF8
- fprintf(HELP_OUTPUT,"j,s,e,w Output code is JIS 7 bit, Shift JIS, EUC-JP, UTF-8N (DEFAULT)\n");
+ if (c1 > ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) {
+ /* 8 bit code */
+ if (!estab_f && !iso8859_f) {
+ /* not established yet */
+ c2 = c1;
+ NEXT;
+ } else { /* estab_f==TRUE */
+ if (iso8859_f) {
+ c2 = ISO_8859_1;
+ c1 &= 0x7f;
+ SEND;
+ } else if (SSP<=c1 && c1<0xe0 && iconv == s_iconv) {
+ /* SJIS X0201 Case... */
+ if (iso2022jp_f && !x0201_f) {
+ (*oconv)(GETA1, GETA2);
+ NEXT;
+ } else {
+ c2 = JIS_X_0201;
+ c1 &= 0x7f;
+ SEND;
+ }
+ } else if (c1==SSO && iconv != s_iconv) {
+ /* EUC X0201 Case */
+ c1 = (*i_getc)(f); /* skip SSO */
+ code_status(c1);
+ if (SSP<=c1 && c1<0xe0) {
+ if (iso2022jp_f && !x0201_f) {
+ (*oconv)(GETA1, GETA2);
+ NEXT;
+ } else {
+ c2 = JIS_X_0201;
+ c1 &= 0x7f;
+ SEND;
+ }
+ } else { /* bogus code, skip SSO and one byte */
+ NEXT;
+ }
+ } else if (ms_ucs_map_f == UCS_MAP_CP10001 &&
+ (c1 == 0xFD || c1 == 0xFE)) {
+ /* CP10001 */
+ c2 = JIS_X_0201;
+ c1 &= 0x7f;
+ SEND;
+ } else {
+ /* already established */
+ c2 = c1;
+ NEXT;
+ }
+ }
+ } else if ((c1 > SP) && (c1 != DEL)) {
+ /* in case of Roman characters */
+ if (shift_mode) {
+ /* output 1 shifted byte */
+ if (iso8859_f) {
+ c2 = ISO_8859_1;
+ SEND;
+ } else if (SP <= c1 && c1 < (0xe0&0x7f)){
+ /* output 1 shifted byte */
+ if (iso2022jp_f && !x0201_f) {
+ (*oconv)(GETA1, GETA2);
+ NEXT;
+ } else {
+ c2 = JIS_X_0201;
+ SEND;
+ }
+ } else {
+ /* look like bogus code */
+ NEXT;
+ }
+ } else if (input_mode == JIS_X_0208 || input_mode == JIS_X_0212 ||
+ input_mode == JIS_X_0213_1 || input_mode == JIS_X_0213_2) {
+ /* in case of Kanji shifted */
+ c2 = c1;
+ NEXT;
+ } else if (c1 == '=' && mime_f && !mime_decode_mode) {
+ /* Check MIME code */
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ (*oconv)(0, '=');
+ LAST;
+ } else if (c1 == '?') {
+ /* =? is mime conversion start sequence */
+ if(mime_f == STRICT_MIME) {
+ /* check in real detail */
+ if (mime_begin_strict(f) == EOF)
+ LAST;
+ else
+ NEXT;
+ } else if (mime_begin(f) == EOF)
+ LAST;
+ else
+ NEXT;
+ } else {
+ (*oconv)(0, '=');
+ (*i_ungetc)(c1,f);
+ NEXT;
+ }
+ } else {
+ /* normal ASCII code */
+ SEND;
+ }
+ } else if (c1 == SI && (!is_8bit || mime_decode_mode)) {
+ shift_mode = FALSE;
+ NEXT;
+ } else if (c1 == SO && (!is_8bit || mime_decode_mode)) {
+ shift_mode = TRUE;
+ NEXT;
+ } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) {
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ /* (*oconv)(0, ESC); don't send bogus code */
+ LAST;
+ } else if (c1 == '$') {
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ /*
+ (*oconv)(0, ESC); don't send bogus code
+ (*oconv)(0, '$'); */
+ LAST;
+ } else if (c1 == '@'|| c1 == 'B') {
+ /* This is kanji introduction */
+ input_mode = JIS_X_0208;
+ shift_mode = FALSE;
+ set_input_codename("ISO-2022-JP");
+#ifdef CHECK_OPTION
+ debug("ISO-2022-JP");
#endif
-#ifdef UTF8_OUTPUT_ENABLE
- fprintf(HELP_OUTPUT," After 'w' you can add more options. -w[ 8 [0], 16 [[BL] [0]] ]\n");
+ NEXT;
+ } else if (c1 == '(') {
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ /* don't send bogus code
+ (*oconv)(0, ESC);
+ (*oconv)(0, '$');
+ (*oconv)(0, '(');
+ */
+ LAST;
+ } else if (c1 == '@'|| c1 == 'B') {
+ /* This is kanji introduction */
+ input_mode = JIS_X_0208;
+ shift_mode = FALSE;
+ NEXT;
+#ifdef X0212_ENABLE
+ } else if (c1 == 'D'){
+ input_mode = JIS_X_0212;
+ shift_mode = FALSE;
+ NEXT;
+#endif /* X0212_ENABLE */
+ } else if (c1 == 0x4F){
+ input_mode = JIS_X_0213_1;
+ shift_mode = FALSE;
+ NEXT;
+ } else if (c1 == 0x50){
+ input_mode = JIS_X_0213_2;
+ shift_mode = FALSE;
+ NEXT;
+ } else {
+ /* could be some special code */
+ (*oconv)(0, ESC);
+ (*oconv)(0, '$');
+ (*oconv)(0, '(');
+ (*oconv)(0, c1);
+ NEXT;
+ }
+ } else if (broken_f&0x2) {
+ /* accept any ESC-(-x as broken code ... */
+ input_mode = JIS_X_0208;
+ shift_mode = FALSE;
+ NEXT;
+ } else {
+ (*oconv)(0, ESC);
+ (*oconv)(0, '$');
+ (*oconv)(0, c1);
+ NEXT;
+ }
+ } else if (c1 == '(') {
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ /* don't send bogus code
+ (*oconv)(0, ESC);
+ (*oconv)(0, '('); */
+ LAST;
+ } else {
+ if (c1 == 'I') {
+ /* This is X0201 kana introduction */
+ input_mode = JIS_X_0201; shift_mode = JIS_X_0201;
+ NEXT;
+ } else if (c1 == 'B' || c1 == 'J' || c1 == 'H') {
+ /* This is X0208 kanji introduction */
+ input_mode = ASCII; shift_mode = FALSE;
+ NEXT;
+ } else if (broken_f&0x2) {
+ input_mode = ASCII; shift_mode = FALSE;
+ NEXT;
+ } else {
+ (*oconv)(0, ESC);
+ (*oconv)(0, '(');
+ /* maintain various input_mode here */
+ SEND;
+ }
+ }
+ } else if ( c1 == 'N' || c1 == 'n'){
+ /* SS2 */
+ c3 = (*i_getc)(f); /* skip SS2 */
+ if ( (SP<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){
+ c1 = c3;
+ c2 = JIS_X_0201;
+ SEND;
+ }else{
+ (*i_ungetc)(c3, f);
+ /* lonely ESC */
+ (*oconv)(0, ESC);
+ SEND;
+ }
+ } else {
+ /* lonely ESC */
+ (*oconv)(0, ESC);
+ SEND;
+ }
+ } else if (c1 == ESC && iconv == s_iconv) {
+ /* ESC in Shift_JIS */
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ /* (*oconv)(0, ESC); don't send bogus code */
+ LAST;
+ } else if (c1 == '$') {
+ /* J-PHONE emoji */
+ if ((c1 = (*i_getc)(f)) == EOF) {
+ /*
+ (*oconv)(0, ESC); don't send bogus code
+ (*oconv)(0, '$'); */
+ LAST;
+ } else {
+ if (('E' <= c1 && c1 <= 'G') ||
+ ('O' <= c1 && c1 <= 'Q')) {
+ /*
+ NUM : 0 1 2 3 4 5
+ BYTE: G E F O P Q
+ C%7 : 1 6 0 2 3 4
+ C%7 : 0 1 2 3 4 5 6
+ NUM : 2 0 3 4 5 X 1
+ */
+ static const char jphone_emoji_first_table[7] = {2, 0, 3, 4, 5, 0, 1};
+ c0 = (jphone_emoji_first_table[c1 % 7] << 8) - SP + 0xE000 + CLASS_UNICODE;
+ while ((c1 = (*i_getc)(f)) != EOF) {
+ if (SP <= c1 && c1 <= 'z') {
+ (*oconv)(0, c1 + c0);
+ } else break; /* c1 == SO */
+ }
+ }
+ }
+ if (c1 == EOF) LAST;
+ NEXT;
+ } else {
+ /* lonely ESC */
+ (*oconv)(0, ESC);
+ SEND;
+ }
+ } else if (c1 == LF || c1 == CR) {
+ if (broken_f&4) {
+ input_mode = ASCII; set_iconv(FALSE, 0);
+ SEND;
+ } else if (mime_decode_f && !mime_decode_mode){
+ if (c1 == LF) {
+ if ((c1=(*i_getc)(f))!=EOF && c1 == SP) {
+ i_ungetc(SP,f);
+ continue;
+ } else {
+ i_ungetc(c1,f);
+ }
+ c1 = LF;
+ SEND;
+ } else { /* if (c1 == CR)*/
+ if ((c1=(*i_getc)(f))!=EOF) {
+ if (c1==SP) {
+ i_ungetc(SP,f);
+ continue;
+ } else if (c1 == LF && (c1=(*i_getc)(f))!=EOF && c1 == SP) {
+ i_ungetc(SP,f);
+ continue;
+ } else {
+ i_ungetc(c1,f);
+ }
+ i_ungetc(LF,f);
+ } else {
+ i_ungetc(c1,f);
+ }
+ c1 = CR;
+ SEND;
+ }
+ }
+ } else if (c1 == DEL && input_mode == JIS_X_0208) {
+ /* CP5022x */
+ c2 = c1;
+ NEXT;
+ } else
+ SEND;
+ }
+ /* send: */
+ switch(input_mode){
+ case ASCII:
+ switch ((*iconv)(c2, c1, c0)) { /* can be EUC / SJIS / UTF-8 / UTF-16 */
+ case -2:
+ /* 4 bytes UTF-8 */
+ if ((c0 = (*i_getc)(f)) != EOF) {
+ code_status(c0);
+ c0 <<= 8;
+ if ((c3 = (*i_getc)(f)) != EOF) {
+ code_status(c3);
+ (*iconv)(c2, c1, c0|c3);
+ }
+ }
+ break;
+ case -1:
+ /* 3 bytes EUC or UTF-8 */
+ if ((c0 = (*i_getc)(f)) != EOF) {
+ code_status(c0);
+ (*iconv)(c2, c1, c0);
+ }
+ break;
+ }
+ break;
+ case JIS_X_0208:
+ case JIS_X_0213_1:
+ if (ms_ucs_map_f &&
+ 0x7F <= c2 && c2 <= 0x92 &&
+ 0x21 <= c1 && c1 <= 0x7E) {
+ /* CP932 UDC */
+ if(c1 == 0x7F) return 0;
+ c1 = (c2 - 0x7F) * 94 + c1 - 0x21 + 0xE000 + CLASS_UNICODE;
+ c2 = 0;
+ }
+ (*oconv)(c2, c1); /* this is JIS, not SJIS/EUC case */
+ break;
+#ifdef X0212_ENABLE
+ case JIS_X_0212:
+ (*oconv)(PREFIX_EUCG3 | c2, c1);
+ break;
+#endif /* X0212_ENABLE */
+ case JIS_X_0213_2:
+ (*oconv)(PREFIX_EUCG3 | c2, c1);
+ break;
+ default:
+ (*oconv)(input_mode, c1); /* other special case */
+ }
+
+ c2 = 0;
+ c0 = 0;
+ continue;
+ /* goto next_word */
+ }
+
+ /* epilogue */
+ (*iconv)(EOF, 0, 0);
+ if (!input_codename)
+ {
+ if (is_8bit) {
+ struct input_code *p = input_code_list;
+ struct input_code *result = p;
+ while (p->name){
+ if (p->score < result->score) result = p;
+ ++p;
+ }
+ set_input_codename(result->name);
+#ifdef CHECK_OPTION
+ debug(result->name);
+#endif
+ }
+ }
+ return 1;
+}
+
+void options(unsigned char *cp)
+{
+ nkf_char i, j;
+ unsigned char *p;
+ unsigned char *cp_back = NULL;
+ char codeset[32];
+ nkf_encoding *enc;
+
+ if (option_mode==1)
+ return;
+ while(*cp && *cp++!='-');
+ while (*cp || cp_back) {
+ if(!*cp){
+ cp = cp_back;
+ cp_back = NULL;
+ continue;
+ }
+ p = 0;
+ switch (*cp++) {
+ case '-': /* literal options */
+ if (!*cp || *cp == SP) { /* ignore the rest of arguments */
+ option_mode = 1;
+ return;
+ }
+ for (i=0;i<sizeof(long_option)/sizeof(long_option[0]);i++) {
+ p = (unsigned char *)long_option[i].name;
+ for (j=0;*p && *p != '=' && *p == cp[j];p++, j++);
+ if (*p == cp[j] || cp[j] == SP){
+ p = &cp[j] + 1;
+ break;
+ }
+ p = 0;
+ }
+ if (p == 0) {
+ fprintf(stderr, "unknown long option: --%s\n", cp);
+ return;
+ }
+ while(*cp && *cp != SP && cp++);
+ if (long_option[i].alias[0]){
+ cp_back = cp;
+ cp = (unsigned char *)long_option[i].alias;
+ }else{
+ if (strcmp(long_option[i].name, "ic=") == 0){
+ nkf_str_upcase((char *)p, codeset, 32);
+ enc = nkf_enc_find(codeset);
+ if (!enc) continue;
+ input_encoding = enc;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "oc=") == 0){
+ nkf_str_upcase((char *)p, codeset, 32);
+ enc = nkf_enc_find(codeset);
+ if (enc <= 0) continue;
+ output_encoding = enc;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "guess=") == 0){
+ if (p[0] == '0' || p[0] == '1') {
+ guess_f = 1;
+ } else {
+ guess_f = 2;
+ }
+ continue;
+ }
+#ifdef OVERWRITE
+ if (strcmp(long_option[i].name, "overwrite") == 0){
+ file_out_f = TRUE;
+ overwrite_f = TRUE;
+ preserve_time_f = TRUE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "overwrite=") == 0){
+ file_out_f = TRUE;
+ overwrite_f = TRUE;
+ preserve_time_f = TRUE;
+ backup_f = TRUE;
+ backup_suffix = malloc(strlen((char *) p) + 1);
+ strcpy(backup_suffix, (char *) p);
+ continue;
+ }
+ if (strcmp(long_option[i].name, "in-place") == 0){
+ file_out_f = TRUE;
+ overwrite_f = TRUE;
+ preserve_time_f = FALSE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "in-place=") == 0){
+ file_out_f = TRUE;
+ overwrite_f = TRUE;
+ preserve_time_f = FALSE;
+ backup_f = TRUE;
+ backup_suffix = malloc(strlen((char *) p) + 1);
+ strcpy(backup_suffix, (char *) p);
+ continue;
+ }
#endif
- fprintf(HELP_OUTPUT,"J,S,E,W Input assumption is JIS 7 bit , Shift JIS, EUC-JP, UTF-8\n");
-#ifdef UTF8_INPUT_ENABLE
- fprintf(HELP_OUTPUT," After 'W' you can add more options. -W[ 8, 16 [BL] ] \n");
-#endif
- fprintf(HELP_OUTPUT,"t no conversion\n");
- fprintf(HELP_OUTPUT,"i[@B] Specify the Esc Seq for JIS X 0208-1978/83 (DEFAULT B)\n");
- fprintf(HELP_OUTPUT,"o[BJH] Specify the Esc Seq for ASCII/Roman (DEFAULT B)\n");
- fprintf(HELP_OUTPUT,"r {de/en}crypt ROT13/47\n");
- fprintf(HELP_OUTPUT,"h 1 katakana->hiragana, 2 hiragana->katakana, 3 both\n");
- fprintf(HELP_OUTPUT,"m[BQN0] MIME decode [B:base64,Q:quoted,N:non-strict,0:no decode]\n");
- fprintf(HELP_OUTPUT,"M[BQ] MIME encode [B:base64 Q:quoted]\n");
- fprintf(HELP_OUTPUT,"l ISO8859-1 (Latin-1) support\n");
- fprintf(HELP_OUTPUT,"f/F Folding: -f60 or -f or -f60-10 (fold margin 10) F preserve nl\n");
- fprintf(HELP_OUTPUT,"Z[0-4] Default/0: Convert JISX0208 Alphabet to ASCII\n");
- fprintf(HELP_OUTPUT," 1: Kankaku to one space 2: to two spaces 3: HTML Entity\n");
- fprintf(HELP_OUTPUT," 4: JISX0208 Katakana to JISX0201 Katakana\n");
- fprintf(HELP_OUTPUT,"X,x Assume X0201 kana in MS-Kanji, -x preserves X0201\n");
- fprintf(HELP_OUTPUT,"B[0-2] Broken input 0: missing ESC,1: any X on ESC-[($]-X,2: ASCII on NL\n");
-#ifdef MSDOS
- fprintf(HELP_OUTPUT,"T Text mode output\n");
-#endif
- fprintf(HELP_OUTPUT,"O Output to File (DEFAULT 'nkf.out')\n");
- fprintf(HELP_OUTPUT,"I Convert non ISO-2022-JP charactor to GETA\n");
- fprintf(HELP_OUTPUT,"d,c Convert line breaks -d: LF -c: CRLF\n");
- fprintf(HELP_OUTPUT,"-L[uwm] line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n");
- fprintf(HELP_OUTPUT,"v, V Show this usage. V: show configuration\n");
- fprintf(HELP_OUTPUT,"\n");
- fprintf(HELP_OUTPUT,"Long name options\n");
- fprintf(HELP_OUTPUT," --ic=<input codeset> --oc=<output codeset>\n");
- fprintf(HELP_OUTPUT," Specify the input or output codeset\n");
- fprintf(HELP_OUTPUT," --fj --unix --mac --windows\n");
- fprintf(HELP_OUTPUT," --jis --euc --sjis --utf8 --utf16 --mime --base64\n");
- fprintf(HELP_OUTPUT," Convert for the system or code\n");
- fprintf(HELP_OUTPUT," --hiragana --katakana --katakana-hiragana\n");
- fprintf(HELP_OUTPUT," To Hiragana/Katakana Conversion\n");
- fprintf(HELP_OUTPUT," --prefix= Insert escape before troublesome characters of Shift_JIS\n");
#ifdef INPUT_OPTION
- fprintf(HELP_OUTPUT," --cap-input, --url-input Convert hex after ':' or '%%'\n");
+ if (strcmp(long_option[i].name, "cap-input") == 0){
+ cap_f = TRUE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "url-input") == 0){
+ url_f = TRUE;
+ continue;
+ }
#endif
#ifdef NUMCHAR_OPTION
- fprintf(HELP_OUTPUT," --numchar-input Convert Unicode Character Reference\n");
+ if (strcmp(long_option[i].name, "numchar-input") == 0){
+ numchar_f = TRUE;
+ continue;
+ }
+#endif
+#ifdef CHECK_OPTION
+ if (strcmp(long_option[i].name, "no-output") == 0){
+ noout_f = TRUE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "debug") == 0){
+ debug_f = TRUE;
+ continue;
+ }
+#endif
+ if (strcmp(long_option[i].name, "cp932") == 0){
+#ifdef SHIFTJIS_CP932
+ cp51932_f = TRUE;
+ cp932inv_f = -TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP932;
+#endif
+ continue;
+ }
+ if (strcmp(long_option[i].name, "no-cp932") == 0){
+#ifdef SHIFTJIS_CP932
+ cp51932_f = FALSE;
+ cp932inv_f = FALSE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_ASCII;
+#endif
+ continue;
+ }
+#ifdef SHIFTJIS_CP932
+ if (strcmp(long_option[i].name, "cp932inv") == 0){
+ cp932inv_f = -TRUE;
+ continue;
+ }
+#endif
+
+#ifdef X0212_ENABLE
+ if (strcmp(long_option[i].name, "x0212") == 0){
+ x0212_f = TRUE;
+ continue;
+ }
+#endif
+
+#ifdef EXEC_IO
+ if (strcmp(long_option[i].name, "exec-in") == 0){
+ exec_f = 1;
+ return;
+ }
+ if (strcmp(long_option[i].name, "exec-out") == 0){
+ exec_f = -1;
+ return;
+ }
+#endif
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+ if (strcmp(long_option[i].name, "no-cp932ext") == 0){
+ no_cp932ext_f = TRUE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "no-best-fit-chars") == 0){
+ no_best_fit_chars_f = TRUE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-skip") == 0){
+ encode_fallback = NULL;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-html") == 0){
+ encode_fallback = encode_fallback_html;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-xml") == 0){
+ encode_fallback = encode_fallback_xml;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-java") == 0){
+ encode_fallback = encode_fallback_java;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-perl") == 0){
+ encode_fallback = encode_fallback_perl;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-subchar") == 0){
+ encode_fallback = encode_fallback_subchar;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "fb-subchar=") == 0){
+ encode_fallback = encode_fallback_subchar;
+ unicode_subchar = 0;
+ if (p[0] != '0'){
+ /* decimal number */
+ for (i = 0; i < 7 && nkf_isdigit(p[i]); i++){
+ unicode_subchar *= 10;
+ unicode_subchar += hex2bin(p[i]);
+ }
+ }else if(p[1] == 'x' || p[1] == 'X'){
+ /* hexadecimal number */
+ for (i = 2; i < 8 && nkf_isxdigit(p[i]); i++){
+ unicode_subchar <<= 4;
+ unicode_subchar |= hex2bin(p[i]);
+ }
+ }else{
+ /* octal number */
+ for (i = 1; i < 8 && nkf_isoctal(p[i]); i++){
+ unicode_subchar *= 8;
+ unicode_subchar += hex2bin(p[i]);
+ }
+ }
+ w16e_conv(unicode_subchar, &i, &j);
+ unicode_subchar = i<<8 | j;
+ continue;
+ }
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ if (strcmp(long_option[i].name, "ms-ucs-map") == 0){
+ ms_ucs_map_f = UCS_MAP_MS;
+ continue;
+ }
+#endif
+#ifdef UNICODE_NORMALIZATION
+ if (strcmp(long_option[i].name, "utf8mac-input") == 0){
+ nfc_f = TRUE;
+ continue;
+ }
+#endif
+ if (strcmp(long_option[i].name, "prefix=") == 0){
+ if (nkf_isgraph(p[0])){
+ for (i = 1; nkf_isgraph(p[i]); i++){
+ prefix_table[p[i]] = p[0];
+ }
+ }
+ continue;
+ }
+ }
+ continue;
+ case 'b': /* buffered mode */
+ unbuf_f = FALSE;
+ continue;
+ case 'u': /* non bufferd mode */
+ unbuf_f = TRUE;
+ continue;
+ case 't': /* transparent mode */
+ if (*cp=='1') {
+ /* alias of -t */
+ cp++;
+ nop_f = TRUE;
+ } else if (*cp=='2') {
+ /*
+ * -t with put/get
+ *
+ * nkf -t2MB hoge.bin | nkf -t2mB | diff -s - hoge.bin
+ *
+ */
+ cp++;
+ nop_f = 2;
+ } else
+ nop_f = TRUE;
+ continue;
+ case 'j': /* JIS output */
+ case 'n':
+ output_encoding = nkf_enc_from_index(ISO_2022_JP);
+ continue;
+ case 'e': /* AT&T EUC output */
+ output_encoding = nkf_enc_from_index(EUC_JP);
+ continue;
+ case 's': /* SJIS output */
+ output_encoding = nkf_enc_from_index(WINDOWS_31J);
+ continue;
+ case 'l': /* ISO8859 Latin-1 support, no conversion */
+ iso8859_f = TRUE; /* Only compatible with ISO-2022-JP */
+ input_encoding = nkf_enc_from_index(ISO_8859_1);
+ continue;
+ case 'i': /* Kanji IN ESC-$-@/B */
+ if (*cp=='@'||*cp=='B')
+ kanji_intro = *cp++;
+ continue;
+ case 'o': /* ASCII IN ESC-(-J/B */
+ if (*cp=='J'||*cp=='B'||*cp=='H')
+ ascii_intro = *cp++;
+ continue;
+ case 'h':
+ /*
+ bit:1 katakana->hiragana
+ bit:2 hiragana->katakana
+ */
+ if ('9'>= *cp && *cp>='0')
+ hira_f |= (*cp++ -'0');
+ else
+ hira_f |= 1;
+ continue;
+ case 'r':
+ rot_f = TRUE;
+ continue;
+#if defined(MSDOS) || defined(__OS2__)
+ case 'T':
+ binmode_f = FALSE;
+ continue;
+#endif
+#ifndef PERL_XS
+ case 'V':
+ show_configuration();
+ exit(1);
+ break;
+ case 'v':
+ usage();
+ exit(1);
+ break;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ case 'w': /* UTF-8 output */
+ if (cp[0] == '8') {
+ cp++;
+ if (cp[0] == '0'){
+ cp++;
+ output_encoding = nkf_enc_from_index(UTF_8N);
+ } else {
+ output_bom_f = TRUE;
+ output_encoding = nkf_enc_from_index(UTF_8_BOM);
+ }
+ } else {
+ int enc_idx;
+ if ('1'== cp[0] && '6'==cp[1]) {
+ cp += 2;
+ enc_idx = UTF_16;
+ } else if ('3'== cp[0] && '2'==cp[1]) {
+ cp += 2;
+ enc_idx = UTF_32;
+ } else {
+ output_encoding = nkf_enc_from_index(UTF_8);
+ continue;
+ }
+ if (cp[0]=='L') {
+ cp++;
+ output_endian = ENDIAN_LITTLE;
+ } else if (cp[0] == 'B') {
+ cp++;
+ } else {
+ output_encoding = nkf_enc_from_index(enc_idx);
+ continue;
+ }
+ if (cp[0] == '0'){
+ cp++;
+ enc_idx = enc_idx == UTF_16
+ ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
+ : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
+ } else {
+ output_bom_f = TRUE;
+ enc_idx = enc_idx == UTF_16
+ ? (output_endian == ENDIAN_LITTLE ? UTF_16LE_BOM : UTF_16BE_BOM)
+ : (output_endian == ENDIAN_LITTLE ? UTF_32LE_BOM : UTF_32BE_BOM);
+ }
+ output_encoding = nkf_enc_from_index(enc_idx);
+ }
+ continue;
#endif
#ifdef UTF8_INPUT_ENABLE
- fprintf(HELP_OUTPUT," --fb-{skip, html, xml, perl, java, subchar}\n");
- fprintf(HELP_OUTPUT," Specify how nkf handles unassigned characters\n");
+ case 'W': /* UTF input */
+ if (cp[0] == '8') {
+ cp++;
+ input_encoding = nkf_enc_from_index(UTF_8);
+ }else{
+ int enc_idx;
+ if ('1'== cp[0] && '6'==cp[1]) {
+ cp += 2;
+ input_endian = ENDIAN_BIG;
+ enc_idx = UTF_16;
+ } else if ('3'== cp[0] && '2'==cp[1]) {
+ cp += 2;
+ input_endian = ENDIAN_BIG;
+ enc_idx = UTF_32;
+ } else {
+ input_encoding = nkf_enc_from_index(UTF_8);
+ continue;
+ }
+ if (cp[0]=='L') {
+ cp++;
+ input_endian = ENDIAN_LITTLE;
+ } else if (cp[0] == 'B') {
+ cp++;
+ input_endian = ENDIAN_BIG;
+ }
+ enc_idx = enc_idx == UTF_16
+ ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
+ : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
+ input_encoding = nkf_enc_from_index(enc_idx);
+ }
+ continue;
#endif
-#ifdef OVERWRITE
- fprintf(HELP_OUTPUT," --in-place[=SUFFIX] --overwrite[=SUFFIX]\n");
- fprintf(HELP_OUTPUT," Overwrite original listed files by filtered result\n");
- fprintf(HELP_OUTPUT," --overwrite preserves timestamp of original files\n");
-#endif
- fprintf(HELP_OUTPUT," -g --guess Guess the input code\n");
- fprintf(HELP_OUTPUT," --help --version Show this help/the version\n");
- fprintf(HELP_OUTPUT," For more information, see also man nkf\n");
- fprintf(HELP_OUTPUT,"\n");
- version();
+ /* Input code assumption */
+ case 'J': /* ISO-2022-JP input */
+ input_encoding = nkf_enc_from_index(ISO_2022_JP);
+ continue;
+ case 'E': /* EUC-JP input */
+ input_encoding = nkf_enc_from_index(EUC_JP);
+ continue;
+ case 'S': /* Windows-31J input */
+ input_encoding = nkf_enc_from_index(WINDOWS_31J);
+ continue;
+ case 'Z': /* Convert X0208 alphabet to asii */
+ /* alpha_f
+ bit:0 Convert JIS X 0208 Alphabet to ASCII
+ bit:1 Convert Kankaku to one space
+ bit:2 Convert Kankaku to two spaces
+ bit:3 Convert HTML Entity
+ bit:4 Convert JIS X 0208 Katakana to JIS X 0201 Katakana
+ */
+ while ('0'<= *cp && *cp <='9') {
+ alpha_f |= 1 << (*cp++ - '0');
+ }
+ if (!alpha_f) alpha_f = 1;
+ continue;
+ case 'x': /* Convert X0201 kana to X0208 or X0201 Conversion */
+ x0201_f = FALSE; /* No X0201->X0208 conversion */
+ /* accept X0201
+ ESC-(-I in JIS, EUC, MS Kanji
+ SI/SO in JIS, EUC, MS Kanji
+ SSO in EUC, JIS, not in MS Kanji
+ MS Kanji (0xa0-0xdf)
+ output X0201
+ ESC-(-I in JIS (0x20-0x5f)
+ SSO in EUC (0xa0-0xdf)
+ 0xa0-0xd in MS Kanji (0xa0-0xdf)
+ */
+ continue;
+ case 'X': /* Convert X0201 kana to X0208 */
+ x0201_f = TRUE;
+ continue;
+ case 'F': /* prserve new lines */
+ fold_preserve_f = TRUE;
+ case 'f': /* folding -f60 or -f */
+ fold_f = TRUE;
+ fold_len = 0;
+ while('0'<= *cp && *cp <='9') { /* we don't use atoi here */
+ fold_len *= 10;
+ fold_len += *cp++ - '0';
+ }
+ if (!(0<fold_len && fold_len<BUFSIZ))
+ fold_len = DEFAULT_FOLD;
+ if (*cp=='-') {
+ fold_margin = 0;
+ cp++;
+ while('0'<= *cp && *cp <='9') { /* we don't use atoi here */
+ fold_margin *= 10;
+ fold_margin += *cp++ - '0';
+ }
+ }
+ continue;
+ case 'm': /* MIME support */
+ /* mime_decode_f = TRUE; */ /* this has too large side effects... */
+ if (*cp=='B'||*cp=='Q') {
+ mime_decode_mode = *cp++;
+ mimebuf_f = FIXED_MIME;
+ } else if (*cp=='N') {
+ mime_f = TRUE; cp++;
+ } else if (*cp=='S') {
+ mime_f = STRICT_MIME; cp++;
+ } else if (*cp=='0') {
+ mime_decode_f = FALSE;
+ mime_f = FALSE; cp++;
+ } else {
+ mime_f = STRICT_MIME;
+ }
+ continue;
+ case 'M': /* MIME output */
+ if (*cp=='B') {
+ mimeout_mode = 'B';
+ mimeout_f = FIXED_MIME; cp++;
+ } else if (*cp=='Q') {
+ mimeout_mode = 'Q';
+ mimeout_f = FIXED_MIME; cp++;
+ } else {
+ mimeout_f = TRUE;
+ }
+ continue;
+ case 'B': /* Broken JIS support */
+ /* bit:0 no ESC JIS
+ bit:1 allow any x on ESC-(-x or ESC-$-x
+ bit:2 reset to ascii on NL
+ */
+ if ('9'>= *cp && *cp>='0')
+ broken_f |= 1<<(*cp++ -'0');
+ else
+ broken_f |= TRUE;
+ continue;
+#ifndef PERL_XS
+ case 'O':/* for Output file */
+ file_out_f = TRUE;
+ continue;
+#endif
+ case 'c':/* add cr code */
+ eolmode_f = CRLF;
+ continue;
+ case 'd':/* delete cr code */
+ eolmode_f = LF;
+ continue;
+ case 'I': /* ISO-2022-JP output */
+ iso2022jp_f = TRUE;
+ continue;
+ case 'L': /* line mode */
+ if (*cp=='u') { /* unix */
+ eolmode_f = LF; cp++;
+ } else if (*cp=='m') { /* mac */
+ eolmode_f = CR; cp++;
+ } else if (*cp=='w') { /* windows */
+ eolmode_f = CRLF; cp++;
+ } else if (*cp=='0') { /* no conversion */
+ eolmode_f = 0; cp++;
+ }
+ continue;
+#ifndef PERL_XS
+ case 'g':
+ if ('2' <= *cp && *cp <= '9') {
+ guess_f = 2;
+ cp++;
+ } else if (*cp == '0' || *cp == '1') {
+ guess_f = 1;
+ cp++;
+ } else {
+ guess_f = 1;
+ }
+ continue;
+#endif
+ case SP:
+ /* module muliple options in a string are allowed for Perl moudle */
+ while(*cp && *cp++!='-');
+ continue;
+ default:
+ fprintf(stderr, "unknown option: -%c\n", *(cp-1));
+ /* bogus option but ignored */
+ continue;
+ }
+ }
}
-void show_configuration(void)
+#ifdef WIN32DLL
+#include "nkf32dll.c"
+#elif defined(PERL_XS)
+#else /* WIN32DLL */
+int main(int argc, char **argv)
{
- fprintf(HELP_OUTPUT, "Summary of my nkf " NKF_VERSION " (" NKF_RELEASE_DATE ") configuration:\n");
- fprintf(HELP_OUTPUT, " Compile-time options:\n");
- fprintf(HELP_OUTPUT, " Default output encoding: "
-#if defined(DEFAULT_CODE_JIS)
- "ISO-2022-JP"
-#elif defined(DEFAULT_CODE_SJIS)
- "Shift_JIS"
-#elif defined(DEFAULT_CODE_EUC)
- "EUC-JP"
-#elif defined(DEFAULT_CODE_UTF8)
- "UTF-8"
+ FILE *fin;
+ unsigned char *cp;
+
+ char *outfname = NULL;
+ char *origfname;
+
+#ifdef EASYWIN /*Easy Win */
+ _BufferSize.y = 400;/*Set Scroll Buffer Size*/
#endif
- "\n");
- fprintf(HELP_OUTPUT, " Default output newline: "
-#if DEFAULT_NEWLINE == CR
- "CR"
-#elif DEFAULT_NEWLINE == CRLF
- "CRLF"
+ setlocale(LC_CTYPE, "");
+
+ for (argc--,argv++; (argc > 0) && **argv == '-'; argc--, argv++) {
+ cp = (unsigned char *)*argv;
+ options(cp);
+#ifdef EXEC_IO
+ if (exec_f){
+ int fds[2], pid;
+ if (pipe(fds) < 0 || (pid = fork()) < 0){
+ abort();
+ }
+ if (pid == 0){
+ if (exec_f > 0){
+ close(fds[0]);
+ dup2(fds[1], 1);
+ }else{
+ close(fds[1]);
+ dup2(fds[0], 0);
+ }
+ execvp(argv[1], &argv[1]);
+ }
+ if (exec_f > 0){
+ close(fds[1]);
+ dup2(fds[0], 0);
+ }else{
+ close(fds[0]);
+ dup2(fds[1], 1);
+ }
+ argc = 0;
+ break;
+ }
+#endif
+ }
+
+ if (guess_f) {
+#ifdef CHECK_OPTION
+ int debug_f_back = debug_f;
+#endif
+#ifdef EXEC_IO
+ int exec_f_back = exec_f;
+#endif
+#ifdef X0212_ENABLE
+ int x0212_f_back = x0212_f;
+#endif
+ int x0213_f_back = x0213_f;
+ int guess_f_back = guess_f;
+ reinit();
+ guess_f = guess_f_back;
+ mime_f = FALSE;
+#ifdef CHECK_OPTION
+ debug_f = debug_f_back;
+#endif
+#ifdef EXEC_IO
+ exec_f = exec_f_back;
+#endif
+#ifdef X0212_ENABLE
+ x0212_f = x0212_f_back;
+#endif
+ x0213_f = x0213_f_back;
+ }
+
+ if (binmode_f == TRUE)
+#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
+ if (freopen("","wb",stdout) == NULL)
+ return (-1);
#else
- "LF"
+ setbinmode(stdout);
#endif
- "\n");
- fprintf(HELP_OUTPUT, " Decode MIME encoded string: "
-#if MIME_DECODE_DEFAULT
- "ON"
+
+ if (unbuf_f)
+ setbuf(stdout, (char *) NULL);
+ else
+ setvbuffer(stdout, (char *) stdobuf, IOBUF_SIZE);
+
+ if (argc == 0) {
+ if (binmode_f == TRUE)
+#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
+ if (freopen("","rb",stdin) == NULL) return (-1);
#else
- "OFF"
+ setbinmode(stdin);
#endif
- "\n");
- fprintf(HELP_OUTPUT, " Convert JIS X 0201 Katakana: "
-#if X0201_DEFAULT
- "ON"
+ setvbuffer(stdin, (char *) stdibuf, IOBUF_SIZE);
+ if (nop_f)
+ noconvert(stdin);
+ else {
+ kanji_convert(stdin);
+ if (guess_f) print_guessed_code(NULL);
+ }
+ } else {
+ int nfiles = argc;
+ int is_argument_error = FALSE;
+ while (argc--) {
+ input_codename = NULL;
+ input_eol = 0;
+#ifdef CHECK_OPTION
+ iconv_for_check = 0;
+#endif
+ if ((fin = fopen((origfname = *argv++), "r")) == NULL) {
+ perror(*(argv-1));
+ is_argument_error = TRUE;
+ continue;
+ } else {
+#ifdef OVERWRITE
+ int fd = 0;
+ int fd_backup = 0;
+#endif
+
+/* reopen file for stdout */
+ if (file_out_f == TRUE) {
+#ifdef OVERWRITE
+ if (overwrite_f){
+ outfname = malloc(strlen(origfname)
+ + strlen(".nkftmpXXXXXX")
+ + 1);
+ if (!outfname){
+ perror(origfname);
+ return -1;
+ }
+ strcpy(outfname, origfname);
+#ifdef MSDOS
+ {
+ int i;
+ for (i = strlen(outfname); i; --i){
+ if (outfname[i - 1] == '/'
+ || outfname[i - 1] == '\\'){
+ break;
+ }
+ }
+ outfname[i] = '\0';
+ }
+ strcat(outfname, "ntXXXXXX");
+ mktemp(outfname);
+ fd = open(outfname, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL,
+ S_IREAD | S_IWRITE);
#else
- "OFF"
+ strcat(outfname, ".nkftmpXXXXXX");
+ fd = mkstemp(outfname);
#endif
- "\n");
- fprintf(HELP_OUTPUT, " --help, --version output: "
-#if HELP_OUTPUT_HELP_OUTPUT
-"HELP_OUTPUT"
+ if (fd < 0
+ || (fd_backup = dup(fileno(stdout))) < 0
+ || dup2(fd, fileno(stdout)) < 0
+ ){
+ perror(origfname);
+ return -1;
+ }
+ }else
+#endif
+ if(argc == 1) {
+ outfname = *argv++;
+ argc--;
+ } else {
+ outfname = "nkf.out";
+ }
+
+ if(freopen(outfname, "w", stdout) == NULL) {
+ perror (outfname);
+ return (-1);
+ }
+ if (binmode_f == TRUE) {
+#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
+ if (freopen("","wb",stdout) == NULL)
+ return (-1);
#else
-"STDOUT"
+ setbinmode(stdout);
+#endif
+ }
+ }
+ if (binmode_f == TRUE)
+#if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__))
+ if (freopen("","rb",fin) == NULL)
+ return (-1);
+#else
+ setbinmode(fin);
+#endif
+ setvbuffer(fin, (char *) stdibuf, IOBUF_SIZE);
+ if (nop_f)
+ noconvert(fin);
+ else {
+ char *filename = NULL;
+ kanji_convert(fin);
+ if (nfiles > 1) filename = origfname;
+ if (guess_f) print_guessed_code(filename);
+ }
+ fclose(fin);
+#ifdef OVERWRITE
+ if (overwrite_f) {
+ struct stat sb;
+#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) && !defined(__WATCOMC__) && !defined(__EMX__) && !defined(__OS2__) && !defined(__DJGPP__)
+ time_t tb[2];
+#else
+ struct utimbuf tb;
#endif
-"\n");
-}
-void version(void)
-{
- fprintf(HELP_OUTPUT,"Network Kanji Filter Version " NKF_VERSION " (" NKF_RELEASE_DATE ") \n" COPY_RIGHT "\n");
+ fflush(stdout);
+ close(fd);
+ if (dup2(fd_backup, fileno(stdout)) < 0){
+ perror("dup2");
+ }
+ if (stat(origfname, &sb)) {
+ fprintf(stderr, "Can't stat %s\n", origfname);
+ }
+ /* $B%Q!<%_%C%7%g%s$rI|85(B */
+ if (chmod(outfname, sb.st_mode)) {
+ fprintf(stderr, "Can't set permission %s\n", outfname);
+ }
+
+ /* $B%?%$%`%9%?%s%W$rI|85(B */
+ if(preserve_time_f){
+#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) && !defined(__WATCOMC__) && !defined(__EMX__) && !defined(__OS2__) && !defined(__DJGPP__)
+ tb[0] = tb[1] = sb.st_mtime;
+ if (utime(outfname, tb)) {
+ fprintf(stderr, "Can't set timestamp %s\n", outfname);
+ }
+#else
+ tb.actime = sb.st_atime;
+ tb.modtime = sb.st_mtime;
+ if (utime(outfname, &tb)) {
+ fprintf(stderr, "Can't set timestamp %s\n", outfname);
+ }
+#endif
+ }
+ if(backup_f){
+ char *backup_filename = get_backup_filename(backup_suffix, origfname);
+#ifdef MSDOS
+ unlink(backup_filename);
+#endif
+ if (rename(origfname, backup_filename)) {
+ perror(backup_filename);
+ fprintf(stderr, "Can't rename %s to %s\n",
+ origfname, backup_filename);
+ }
+ }else{
+#ifdef MSDOS
+ if (unlink(origfname)){
+ perror(origfname);
+ }
+#endif
+ }
+ if (rename(outfname, origfname)) {
+ perror(origfname);
+ fprintf(stderr, "Can't rename %s to %s\n",
+ outfname, origfname);
+ }
+ free(outfname);
+ }
+#endif
+ }
+ }
+ if (is_argument_error)
+ return(-1);
+ }
+#ifdef EASYWIN /*Easy Win */
+ if (file_out_f == FALSE)
+ scanf("%d",&end_check);
+ else
+ fclose(stdout);
+#else /* for Other OS */
+ if (file_out_f == TRUE)
+ fclose(stdout);
+#endif /*Easy Win */
+ return (0);
}
-#endif /*PERL_XS*/
+#endif /* WIN32DLL */
diff --git a/ext/nkf/nkf-utf8/nkf.h b/ext/nkf/nkf-utf8/nkf.h
new file mode 100755
index 0000000000..30efd4b5dd
--- /dev/null
+++ b/ext/nkf/nkf-utf8/nkf.h
@@ -0,0 +1,166 @@
+/*
+ *
+ * nkf.h - Header file for nkf
+ *
+ * $Id: nkf.h,v 1.2 2008/01/23 09:10:25 naruse Exp $
+ */
+
+
+#ifndef NKF_H
+
+/* Wrapper of configurations */
+
+#ifndef MIME_DECODE_DEFAULT
+#define MIME_DECODE_DEFAULT STRICT_MIME
+#endif
+#ifndef X0201_DEFAULT
+#define X0201_DEFAULT TRUE
+#endif
+
+#if DEFAULT_NEWLINE == 0x0D0A
+#define PUT_NEWLINE(func) do {\
+ func(0x0D);\
+ func(0x0A);\
+} while (0)
+#define OCONV_NEWLINE(func) do {\
+ func(0, 0x0D);\
+ func(0, 0x0A);\
+} while (0)
+#elif DEFAULT_NEWLINE == 0x0D
+#define PUT_NEWLINE(func) func(0x0D)
+#define OCONV_NEWLINE(func) func(0, 0x0D)
+#else
+#define DEFAULT_NEWLINE 0x0A
+#define PUT_NEWLINE(func) func(0x0A)
+#define OCONV_NEWLINE(func) func(0, 0x0A)
+#endif
+#ifdef HELP_OUTPUT_STDERR
+#define HELP_OUTPUT stderr
+#else
+#define HELP_OUTPUT stdout
+#endif
+
+
+/* Compatibility definitions */
+
+#ifdef nkf_char
+#elif defined(INT_IS_SHORT)
+typedef long nkf_char;
+#define NKF_INT32_C(n) (n##L)
+#else
+typedef int nkf_char;
+#define NKF_INT32_C(n) (n)
+#endif
+
+#if (defined(__TURBOC__) || defined(_MSC_VER) || defined(LSI_C) || defined(__MINGW32__) || defined(__EMX__) || defined(__MSDOS__) || defined(__WINDOWS__) || defined(__DOS__) || defined(__OS2__)) && !defined(MSDOS)
+#define MSDOS
+#if (defined(__Win32__) || defined(_WIN32)) && !defined(__WIN32__)
+#define __WIN32__
+#endif
+#endif
+
+#ifdef PERL_XS
+#undef OVERWRITE
+#endif
+
+#ifndef PERL_XS
+#include <stdio.h>
+#endif
+
+#include <stdlib.h>
+#include <string.h>
+
+#if defined(MSDOS) || defined(__OS2__)
+#include <fcntl.h>
+#include <io.h>
+#if defined(_MSC_VER) || defined(__WATCOMC__)
+#define mktemp _mktemp
+#endif
+#endif
+
+#ifdef MSDOS
+#ifdef LSI_C
+#define setbinmode(fp) fsetbin(fp)
+#elif defined(__DJGPP__)
+#include <libc/dosio.h>
+void setbinmode(FILE *fp)
+{
+ /* we do not use libc's setmode(), which changes COOKED/RAW mode in device. */
+ int fd, m;
+ fd = fileno(fp);
+ m = (__file_handle_modes[fd] & (~O_TEXT)) | O_BINARY;
+ __file_handle_set(fd, m);
+}
+#else /* Microsoft C, Turbo C */
+#define setbinmode(fp) setmode(fileno(fp), O_BINARY)
+#endif
+#else /* UNIX */
+#define setbinmode(fp)
+#endif
+
+#ifdef _IOFBF /* SysV and MSDOS, Windows */
+#define setvbuffer(fp, buf, size) setvbuf(fp, buf, _IOFBF, size)
+#else /* BSD */
+#define setvbuffer(fp, buf, size) setbuffer(fp, buf, size)
+#endif
+
+/*Borland C++ 4.5 EasyWin*/
+#if defined(__TURBOC__) && defined(_Windows) && !defined(__WIN32__) /*Easy Win */
+#define EASYWIN
+#ifndef __WIN16__
+#define __WIN16__
+#endif
+#include <windows.h>
+#endif
+
+#ifdef OVERWRITE
+/* added by satoru@isoternet.org */
+#if defined(__EMX__)
+#include <sys/types.h>
+#endif
+#include <sys/stat.h>
+#if !defined(MSDOS) || defined(__DJGPP__) /* UNIX, djgpp */
+#include <unistd.h>
+#if defined(__WATCOMC__)
+#include <sys/utime.h>
+#else
+#include <utime.h>
+#endif
+#else /* defined(MSDOS) */
+#ifdef __WIN32__
+#ifdef __BORLANDC__ /* BCC32 */
+#include <utime.h>
+#else /* !defined(__BORLANDC__) */
+#include <sys/utime.h>
+#endif /* (__BORLANDC__) */
+#else /* !defined(__WIN32__) */
+#if defined(_MSC_VER) || defined(__MINGW32__) || defined(__WATCOMC__) || defined(__OS2__) || defined(__EMX__) || defined(__IBMC__) || defined(__IBMCPP__) /* VC++, MinGW, Watcom, emx+gcc, IBM VAC++ */
+#include <sys/utime.h>
+#elif defined(__TURBOC__) /* BCC */
+#include <utime.h>
+#elif defined(LSI_C) /* LSI C */
+#endif /* (__WIN32__) */
+#endif
+#endif
+#endif
+
+#ifndef __WIN32__ /* not win32 is posix */
+#define HAVE_LANGINFO_H
+#define HAVE_LOCALE_H
+#endif
+
+#ifdef HAVE_LANGINFO_H
+#include <langinfo.h>
+#endif
+#ifdef HAVE_LOCALE_H
+#include <locale.h>
+#endif
+
+#define FALSE 0
+#define TRUE 1
+
+#ifdef WIN32DLL
+#include "nkf32.h"
+#endif
+
+#endif
diff --git a/ext/nkf/nkf-utf8/utf8tbl.c b/ext/nkf/nkf-utf8/utf8tbl.c
index db67c5d89f..fbcd86b442 100644
--- a/ext/nkf/nkf-utf8/utf8tbl.c
+++ b/ext/nkf/nkf-utf8/utf8tbl.c
@@ -1,4 +1,11 @@
+/*
+ * utf8tbl.c - Convertion Table for nkf
+ *
+ * $Id$
+ */
+
#include "config.h"
+#include "utf8tbl.h"
#ifdef UTF8_OUTPUT_ENABLE
static const unsigned short euc_to_utf8_A1[] = {
diff --git a/ext/nkf/nkf-utf8/utf8tbl.h b/ext/nkf/nkf-utf8/utf8tbl.h
index 29413d4fac..c3d7709ef7 100644
--- a/ext/nkf/nkf-utf8/utf8tbl.h
+++ b/ext/nkf/nkf-utf8/utf8tbl.h
@@ -1,7 +1,15 @@
+/*
+ * utf8tbl.h - Header file for Convertion Table
+ *
+ * $Id$
+ */
+
#ifndef _UTF8TBL_H_
#define _UTF8TBL_H_
#ifdef UTF8_OUTPUT_ENABLE
+#define sizeof_euc_to_utf8_1byte 94
+#define sizeof_euc_to_utf8_2bytes 94
extern const unsigned short euc_to_utf8_1byte[];
extern const unsigned short *const euc_to_utf8_2bytes[];
extern const unsigned short *const euc_to_utf8_2bytes_ms[];
@@ -10,6 +18,10 @@ extern const unsigned short *const x0212_to_utf8_2bytes[];
#endif /* UTF8_OUTPUT_ENABLE */
#ifdef UTF8_INPUT_ENABLE
+#define sizeof_utf8_to_euc_C2 64
+#define sizeof_utf8_to_euc_E5B8 64
+#define sizeof_utf8_to_euc_2bytes 112
+#define sizeof_utf8_to_euc_3bytes 16
extern const unsigned short *const utf8_to_euc_2bytes[];
extern const unsigned short *const utf8_to_euc_2bytes_ms[];
extern const unsigned short *const utf8_to_euc_2bytes_932[];
@@ -21,11 +33,23 @@ extern const unsigned short *const *const utf8_to_euc_3bytes_mac[];
#endif /* UTF8_INPUT_ENABLE */
#ifdef UNICODE_NORMALIZATION
+
+#define NORMALIZATION_TABLE_LENGTH 942
+#define NORMALIZATION_TABLE_NFC_LENGTH 3
+#define NORMALIZATION_TABLE_NFD_LENGTH 9
+struct normalization_pair {
+ const unsigned char nfc[NORMALIZATION_TABLE_NFC_LENGTH];
+ const unsigned char nfd[NORMALIZATION_TABLE_NFD_LENGTH];
+};
extern const struct normalization_pair normalization_table[];
#endif
#ifdef SHIFTJIS_CP932
+#define CP932_TABLE_BEGIN 0xFA
+#define CP932_TABLE_END 0xFC
extern const unsigned short shiftjis_cp932[3][189];
+#define CP932INV_TABLE_BEGIN 0xED
+#define CP932INV_TABLE_END 0xEE
extern const unsigned short cp932inv[2][189];
#endif /* SHIFTJIS_CP932 */
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-06 12:00:29 +0000