#include "config.h"
static char *CopyRight =
- "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2003 Kono, Furukawa";
+ "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2004 Kono, Furukawa";
static char *Version =
"2.0";
static char *Patchlevel =
- "3/0301/Shinji Kono";
+ "4/0401/Shinji Kono";
/*
**
** USAGE: nkf [flags] [file]
**
** Flags:
-** b Output is bufferred (DEFAULT)
-** u Output is unbufferred
+** b Output is buffered (DEFAULT)
+** u Output is unbuffered
**
** t no operation
**
**
**/
-#if (defined(__TURBOC__) || defined(_MSC_VER) || defined(LSI_C)) && !defined(MSDOS)
+#if (defined(__TURBOC__) || defined(_MSC_VER) || defined(LSI_C) || defined(__MINGW32__)) && !defined(MSDOS)
#define MSDOS
#if (defined(__Win32__) || defined(_WIN32)) && !defined(__WIN32__)
#define __WIN32__
#include <stdio.h>
#endif
-#include <stdlib.h>
-#include <string.h>
-
#if defined(MSDOS) || defined(__OS2__)
#include <fcntl.h>
#include <io.h>
#ifdef OVERWRITE
/* added by satoru@isoternet.org */
+#include <stdlib.h>
+#include <string.h>
#include <sys/stat.h>
#ifndef MSDOS /* UNIX, OS/2 */
#include <unistd.h>
#include <utime.h>
#else
-#if defined(_MSC_VER) /* VC++ */
+#if defined(_MSC_VER) || defined(__MINGW32__) /* VC++, MinGW */
#include <sys/utime.h>
#elif defined(__TURBOC__) /* BCC */
#include <utime.h>
#endif
#endif
+#ifdef INT_IS_SHORT
+#define int long
+#endif
+
#define FALSE 0
#define TRUE 1
#define UTF8 12
#define UTF8_INPUT 13
-#define UTF16_INPUT 14
+#define UTF16LE_INPUT 14
#define UTF16BE_INPUT 15
#define WISH_TRUE 15
int _file_stat;
};
+STATIC char *input_codename = "";
+
STATIC int noconvert PROTO((FILE *f));
STATIC int kanji_convert PROTO((FILE *f));
STATIC int h_conv PROTO((FILE *f,int c2,int c1));
STATIC int w_iconv PROTO((int c2,int c1,int c0));
STATIC int w_iconv16 PROTO((int c2,int c1,int c0));
STATIC int w_iconv_common PROTO((int c1,int c0,unsigned short **pp,int psize,int *p2,int *p1));
+STATIC int ww16_conv PROTO((int c2, int c1, int c0));
#endif
#ifdef UTF8_OUTPUT_ENABLE
STATIC int e2w_conv PROTO((int c2,int c1));
#endif
static int iso2022jp_f = FALSE; /* convert ISO-2022-JP */
#ifdef UTF8_OUTPUT_ENABLE
-static int w_oconv16_begin_f= 0; /* utf-16 header */
+static int unicode_bom_f= 0; /* Output Unicode BOM */
static int w_oconv16_LE = 0; /* utf-16 little endian */
+static int ms_ucs_map_f = FALSE; /* Microsoft UCS Mapping Compatible */
#endif
+#ifdef NUMCHAR_OPTION
+
+#define CLASS_MASK 0x0f000000
+#define CLASS_UTF16 0x01000000
+#endif
+
#ifdef INPUT_OPTION
static int cap_f = FALSE;
static int (*i_cgetc)PROTO((FILE *)) = std_getc; /* input of cgetc */
STATIC void debug PROTO((char *str));
#endif
+static int guess_f = FALSE;
+STATIC void print_guessed_code PROTO((char *filename));
+STATIC void set_input_codename PROTO((char *codename));
+static int is_inputcode_mixed = FALSE;
+static int is_inputcode_set = FALSE;
+
+#ifdef EXEC_IO
+static int exec_f = 0;
+#endif
+
#ifdef SHIFTJIS_CP932
-STATIC int cp932_f = FALSE;
+STATIC int cp932_f = TRUE;
#define CP932_TABLE_BEGIN (0xfa)
#define CP932_TABLE_END (0xfc)
+STATIC int cp932inv_f = FALSE;
+#define CP932INV_TABLE_BEGIN (0xed)
+#define CP932INV_TABLE_END (0xee)
+
#endif /* SHIFTJIS_CP932 */
+STATIC unsigned char prefix_table[256];
+
STATIC void e_status PROTO((struct input_code *, int));
STATIC void s_status PROTO((struct input_code *, int));
#ifdef UTF8_INPUT_ENABLE
STATIC void w_status PROTO((struct input_code *, int));
STATIC void w16_status PROTO((struct input_code *, int));
-static int utf16_mode = UTF16_INPUT;
+static int utf16_mode = UTF16LE_INPUT;
#endif
struct input_code input_code_list[] = {
{"EUC-JP", 0, 0, 0, {0, 0, 0}, e_status, e_iconv, 0},
{"Shift_JIS", 0, 0, 0, {0, 0, 0}, s_status, s_iconv, 0},
{"UTF-8", 0, 0, 0, {0, 0, 0}, w_status, w_iconv, 0},
- {"UTF-16", 0, 0, 0, {0, 0, 0}, w16_status, w_iconv16, 0},
+ {"UTF-16", 0, 0, 0, {0, 0, 0}, w16_status, w_iconv16, 0},
{0}
};
FILE *fin;
unsigned char *cp;
+ char *outfname;
+ char *origfname;
+
#ifdef EASYWIN /*Easy Win */
_BufferSize.y = 400;/*Set Scroll Buffer Size*/
#endif
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(x0201_f == WISH_TRUE)
x0201_f = ((!iso2022jp_f)? TRUE : NO_X0201);
setvbuffer(stdin, stdibuf, IOBUF_SIZE);
if (nop_f)
noconvert(stdin);
- else
+ else {
kanji_convert(stdin);
+ if (guess_f) print_guessed_code(NULL);
+ }
} else {
+ int nfiles = argc;
while (argc--) {
- char *outfname;
- char *origfname;
-
if ((fin = fopen((origfname = *argv++), "r")) == NULL) {
perror(*--argv);
return(-1);
setvbuffer(fin, stdibuf, IOBUF_SIZE);
if (nop_f)
noconvert(fin);
- else
+ 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) {
struct stat sb;
-#ifdef MSDOS
+#if defined(MSDOS) && !defined(__MINGW32__)
time_t tb[2];
#else
struct utimbuf tb;
if (chmod(outfname, sb.st_mode)) {
fprintf(stderr, "Can't set permission %s\n", outfname);
}
-#ifdef MSDOS
- tb[0] = tb[1] = sb.st_mtime;
+
/* \e$B%?%$%`%9%?%s%W$rI|85\e(B */
+#if defined(MSDOS) && !defined(__MINGW32__)
+ tb[0] = tb[1] = sb.st_mtime;
if (utime(outfname, tb)) {
fprintf(stderr, "Can't set timestamp %s\n", outfname);
}
-
- if (unlink(origfname)){
- perror(origfname);
- }
#else
tb.actime = sb.st_atime;
tb.modtime = sb.st_mtime;
- /* \e$B%?%$%`%9%?%s%W$rI|85\e(B */
if (utime(outfname, &tb)) {
fprintf(stderr, "Can't set timestamp %s\n", outfname);
}
#endif
+#ifdef MSDOS
+ if (unlink(origfname)){
+ perror(origfname);
+ }
+#endif
if (rename(outfname, origfname)) {
perror(origfname);
fprintf(stderr, "Can't rename %s to %s\n",
{"hiragana","h1"},
{"katakana","h2"},
{"katakana-hiragana","h3"},
+ {"guess", "g"},
#ifdef UTF8_OUTPUT_ENABLE
{"utf8", "w"},
{"utf16", "w16"},
+ {"ms-ucs-map", ""},
#endif
#ifdef UTF8_INPUT_ENABLE
{"utf8-input", "W"},
#ifdef INPUT_OPTION
{"cap-input", ""},
{"url-input", ""},
+#endif
+#ifdef NUMCHAR_OPTION
{"numchar-input", ""},
#endif
#ifdef CHECK_OPTION
{"debug", ""},
#endif
#ifdef SHIFTJIS_CP932
- {"cp932", ""},
+ {"no-cp932", ""},
+ {"cp932inv", ""},
+#endif
+#ifdef EXEC_IO
+ {"exec-in", ""},
+ {"exec-out", ""},
#endif
+ {"prefix=", ""},
};
static int option_mode;
for (i=0;i<sizeof(long_option)/sizeof(long_option[0]);i++) {
int j;
p = (unsigned char *)long_option[i].name;
- for (j=0;*p && *p++ == cp[j];j++);
- if (! *p && !cp[j]) break;
+ for (j=0;*p && (*p != '=') && *p == cp[j];p++, j++);
+ if (*p == cp[j]){
+ p = &cp[j];
+ break;
+ }
+ p = 0;
}
- if (*p) return;
+ if (p == 0) return;
cp = (unsigned char *)long_option[i].alias;
if (!*cp){
#ifdef OVERWRITE
url_f = TRUE;
continue;
}
+#endif
+#ifdef NUMCHAR_OPTION
if (strcmp(long_option[i].name, "numchar-input") == 0){
numchar_f = TRUE;
continue;
}
#endif
#ifdef SHIFTJIS_CP932
- if (strcmp(long_option[i].name, "cp932") == 0){
- cp932_f = TRUE;
+ if (strcmp(long_option[i].name, "no-cp932") == 0){
+ cp932_f = FALSE;
+ continue;
+ }
+ if (strcmp(long_option[i].name, "cp932inv") == 0){
+ cp932inv_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
+#ifdef UTF8_OUTPUT_ENABLE
+ if (strcmp(long_option[i].name, "ms-ucs-map") == 0){
+ ms_ucs_map_f = TRUE;
continue;
}
#endif
+ if (strcmp(long_option[i].name, "prefix=") == 0){
+ if (*p == '=' && ' ' < p[1] && p[1] < 128){
+ for (i = 2; ' ' < p[i] && p[i] < 128; i++){
+ prefix_table[p[i]] = p[1];
+ }
+ }
+ continue;
+ }
}
continue;
case 'b': /* buffered mode */
if ('1'== cp[0] && '6'==cp[1]) {
output_conv = w_oconv16; cp+=2;
if (cp[0]=='L') {
- w_oconv16_begin_f=2; cp++;
+ unicode_bom_f=2; cp++;
w_oconv16_LE = 1;
if (cp[0] == '0'){
- w_oconv16_begin_f=1; cp++;
+ unicode_bom_f=1; cp++;
}
} else if (cp[0] == 'B') {
- w_oconv16_begin_f=2; cp++;
+ unicode_bom_f=2; cp++;
if (cp[0] == '0'){
- w_oconv16_begin_f=1; cp++;
+ unicode_bom_f=1; cp++;
}
- }
+ }
+ } else if (cp[0] == '8') {
+ output_conv = w_oconv; cp++;
+ unicode_bom_f=2;
+ if (cp[0] == '0'){
+ unicode_bom_f=1; cp++;
+ }
} else
output_conv = w_oconv;
continue;
#ifdef UTF8_INPUT_ENABLE
case 'W': /* UTF-8 input */
if ('1'== cp[0] && '6'==cp[1]) {
- input_f = UTF16_INPUT;
+ input_f = UTF16LE_INPUT;
+ if (cp[0]=='L') {
+ cp++;
+ } else if (cp[0] == 'B') {
+ cp++;
+ input_f = UTF16BE_INPUT;
+ }
} else
input_f = UTF8_INPUT;
continue;
crmode_f = 0; cp++;
}
continue;
+ case 'g':
+#ifndef PERL_XS
+ guess_f = TRUE;
+#endif
+ continue;
case ' ':
/* module muliple options in a string are allowed for Perl moudle */
while(*cp && *cp!='-') cp++;
}
#ifdef ANSI_C_PROTOTYPE
+struct input_code * find_inputcode_byfunc(int (*iconv_func)(int c2,int c1,int c0))
+#else
+struct input_code * find_inputcode_byfunc(iconv_func)
+ int (*iconv_func)();
+#endif
+{
+ if (iconv_func){
+ struct input_code *p = input_code_list;
+ while (p->name){
+ if (iconv_func == p->iconv_func){
+ return p;
+ }
+ p++;
+ }
+ }
+ return 0;
+}
+
+#ifdef ANSI_C_PROTOTYPE
void set_iconv(int f, int (*iconv_func)(int c2,int c1,int c0))
#else
void set_iconv(f, iconv_func)
}
#ifdef CHECK_OPTION
if (estab_f && iconv_for_check != iconv){
-#ifdef UTF8_INPUT_ENABLE
- if (iconv == w_iconv) debug("UTF-8\n");
- if (iconv == w_iconv16) debug("UTF-16\n");
-#endif
- if (iconv == s_iconv) debug("Shift_JIS\n");
- if (iconv == e_iconv) debug("EUC-JP\n");
+ struct input_code *p = find_inputcode_byfunc(iconv);
+ if (p){
+ set_input_codename(p->name);
+ debug(input_codename);
+ }
iconv_for_check = iconv;
}
#endif
}
-#define SCORE_KANA (1) /* \e$B$$$o$f$kH>3Q%+%J\e(B */
+#define SCORE_L2 (1) /* \e$BBh\e(B2\e$B?e=`4A;z\e(B */
+#define SCORE_KANA (SCORE_L2 << 1) /* \e$B$$$o$f$kH>3Q%+%J\e(B */
#define SCORE_DEPEND (SCORE_KANA << 1) /* \e$B5!<o0MB8J8;z\e(B */
#ifdef SHIFTJIS_CP932
#define SCORE_CP932 (SCORE_DEPEND << 1) /* CP932 \e$B$K$h$kFI$_49$(\e(B */
#else
#define SCORE_NO_EXIST (SCORE_DEPEND << 1) /* \e$BB8:_$7$J$$J8;z\e(B */
#endif
-#define SCORE_ERROR (SCORE_NO_EXIST << 1) /* \e$B%(%i!<\e(B */
+#define SCORE_iMIME (SCORE_NO_EXIST << 1) /* MIME \e$B$K$h$k;XDj\e(B */
+#define SCORE_ERROR (SCORE_iMIME << 1) /* \e$B%(%i!<\e(B */
+
+#define SCORE_INIT (SCORE_iMIME)
+
int score_table_A0[] = {
0, 0, 0, 0,
0, 0, 0, 0,
};
int score_table_F0[] = {
- 0, 0, 0, 0,
- 0, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
+ SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
+ SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR,
};
struct input_code *ptr;
int score;
{
- ptr->score |= score;
+ if (ptr){
+ ptr->score |= score;
+ }
+}
+
+void clr_code_score(ptr, score)
+ struct input_code *ptr;
+ int score;
+{
+ if (ptr){
+ ptr->score &= ~score;
+ }
}
void code_score(ptr)
int c1 = ptr->buf[1];
if (c2 < 0){
set_code_score(ptr, SCORE_ERROR);
- }else if ((c2 & 0xf0) == 0xa0){
- set_code_score(ptr, score_table_A0[c2 & 0x0f]);
- }else if ((c2 & 0xf0) == 0xf0){
- set_code_score(ptr, score_table_F0[c2 & 0x0f]);
}else if (c2 == SSO){
set_code_score(ptr, SCORE_KANA);
- }
#ifdef UTF8_OUTPUT_ENABLE
- else if (!e2w_conv(c2, c1)){
+ }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(ptr)
ptr->buf[ptr->index++] = c;
}
-void status_reset(ptr)
+void status_clear(ptr)
struct input_code *ptr;
{
ptr->stat = 0;
- ptr->score = 0;
ptr->index = 0;
}
+void status_reset(ptr)
+ struct input_code *ptr;
+{
+ status_clear(ptr);
+ ptr->score = SCORE_INIT;
+}
+
void status_reinit(ptr)
struct input_code *ptr;
{
case 0:
if (c <= DEL){
break;
+#ifdef NUMCHAR_OPTION
+ }else if ((c & CLASS_MASK) == CLASS_UTF16){
+ break;
+#endif
}else if (0xa1 <= c && c <= 0xdf){
status_push_ch(ptr, SSO);
status_push_ch(ptr, c);
code_score(ptr);
- status_reset(ptr);
+ status_clear(ptr);
}else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xef)){
ptr->stat = 1;
status_push_ch(ptr, c);
status_push_ch(ptr, c);
s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]);
code_score(ptr);
- status_reset(ptr);
+ status_clear(ptr);
}else{
status_disable(ptr);
}
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_reset(ptr);
+ status_clear(ptr);
break;
}
}
case 0:
if (c <= DEL){
break;
+#ifdef NUMCHAR_OPTION
+ }else if ((c & CLASS_MASK) == CLASS_UTF16){
+ break;
+#endif
}else if (SSO == c || (0xa1 <= c && c <= 0xfe)){
ptr->stat = 1;
status_push_ch(ptr, c);
if (0xa1 <= c && c <= 0xfe){
status_push_ch(ptr, c);
code_score(ptr);
- status_reset(ptr);
+ status_clear(ptr);
}else{
status_disable(ptr);
}
ptr->_file_stat = -1;
}else{
status_push_ch(ptr, c);
- status_reset(ptr);
+ status_clear(ptr);
}
break;
case 0xff:
if (ptr->stat != c && (c == 0xfe || c == 0xff)){
status_push_ch(ptr, c);
- status_reset(ptr);
+ status_clear(ptr);
}else{
status_disable(ptr);
ptr->_file_stat = -1;
case 0:
if (c <= DEL){
break;
+#ifdef NUMCHAR_OPTION
+ }else if ((c & CLASS_MASK) == CLASS_UTF16){
+ break;
+#endif
}else if (0xc0 <= c && c <= 0xdf){
ptr->stat = 1;
status_push_ch(ptr, c);
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]);
- code_score(ptr);
- status_reset(ptr);
+ if (!bom){
+ code_score(ptr);
+ }
+ status_clear(ptr);
}
}else{
status_disable(ptr);
/* output redicrection */
#ifdef CHECK_OPTION
- if (noout_f){
+ if (noout_f || guess_f){
o_putc = no_putc;
}
#endif
}
i_getc = std_getc;
+ i_ungetc = std_ungetc;
/* input redicrection */
#ifdef INPUT_OPTION
if (cap_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;
#ifdef UTF8_INPUT_ENABLE
} else if (input_f == UTF8_INPUT) {
set_iconv(-TRUE, w_iconv);
- } else if (input_f == UTF16_INPUT) {
+ } else if (input_f == UTF16LE_INPUT) {
set_iconv(-TRUE, w_iconv16);
#endif
} else {
FILE *f;
{
int c1,
- c2;
+ c2, c3;
module_connection();
c2 = 0;
c2 = c1;
c1 = (*i_getc)(f);
SEND;
+#ifdef NUMCHAR_OPTION
+ } else if ((c1 & CLASS_MASK) == CLASS_UTF16){
+ SEND;
+#endif
} else if (c1 > DEL) {
/* 8 bit code */
if (!estab_f && !iso8859_f) {
/* This is kanji introduction */
input_mode = X0208;
shift_mode = FALSE;
+ set_input_codename("ISO-2022-JP");
+ debug(input_codename);
NEXT;
} else if (c1 == '(') {
if ((c1 = (*i_getc)(f)) == EOF) {
}
} else if ( c1 == 'N' || c1 == 'n' ){
/* SS2 */
- c1 = (*i_getc)(f); /* skip SS2 */
- if ( SPACE<=c1 && c1 < 0xe0 ) {
+ c3 = (*i_getc)(f); /* skip SS2 */
+ if ( (SPACE<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){
+ c1 = c3;
c2 = X0201;
SEND;
+ }else{
+ (*i_ungetc)(c3, f);
+ /* lonely ESC */
+ (*oconv)(0, ESC);
+ SEND;
}
} else {
/* lonely ESC */
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->score < result->score){
result = p;
}
++p;
}
- set_iconv(FALSE, p->iconv_func);
+ set_iconv(FALSE, result->iconv_func);
}
** Kanji codes by oconv and leave estab_f unchanged.
**/
+ c3=c1;
wc = 0;
while (wc < hold_count){
c2 = hold_buf[wc++];
- if (c2 <= DEL){
+ if (c2 <= DEL
+#ifdef NUMCHAR_OPTION
+ || (c2 & CLASS_MASK) == CLASS_UTF16
+#endif
+ ){
(*iconv)(0, c2, 0);
continue;
}else if (iconv == s_iconv && 0xa1 <= c2 && c2 <= 0xdf){
c1 = hold_buf[wc++];
}else{
c1 = (*i_getc)(f);
- if (c1 == EOF) break;
+ if (c1 == EOF){
+ c3 = EOF;
+ break;
+ }
code_status(c1);
}
if ((*iconv)(c2, c1, 0) < 0){
c0 = hold_buf[wc++];
}else{
c0 = (*i_getc)(f);
- if (c0 == EOF) break;
+ if (c0 == EOF){
+ c3 = EOF;
+ break;
+ }
code_status(c0);
}
(*iconv)(c2, c1, c0);
c1 = c0;
}
}
- c3=c1;
return c3;
}
int *p2, *p1;
{
#ifdef SHIFTJIS_CP932
- if (CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){
+ if (cp932_f && CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){
extern unsigned short shiftjis_cp932[3][189];
c1 = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
if (c1 == 0) return 1;
{
extern unsigned short * utf8_to_euc_2bytes[];
extern unsigned short ** utf8_to_euc_3bytes[];
+ int ret = 0;
if (0xc0 <= c2 && c2 <= 0xef) {
unsigned short **pp;
if (0xe0 <= c2) {
if (c0 == 0) return -1;
pp = utf8_to_euc_3bytes[c2 - 0x80];
- return w_iconv_common(c1, c0, pp, sizeof_utf8_to_euc_C2, p2, p1);
+ ret = w_iconv_common(c1, c0, pp, sizeof_utf8_to_euc_C2, p2, p1);
} else {
- return w_iconv_common(c2, c1, utf8_to_euc_2bytes, sizeof_utf8_to_euc_2bytes, p2, p1);
+ ret = w_iconv_common(c2, c1, utf8_to_euc_2bytes, sizeof_utf8_to_euc_2bytes, p2, p1);
+ }
+#ifdef NUMCHAR_OPTION
+ if (ret){
+ if (p2) *p2 = 0;
+ if (p1) *p1 = CLASS_UTF16 | ww16_conv(c2, c1, c0);
+ ret = 0;
}
+#endif
+ return ret;
} else if (c2 == X0201) {
c1 &= 0x7f;
}
if (p2) *p2 = c2;
if (p1) *p1 = c1;
- return 0;
+ return ret;
}
int
}
int
+ww16_conv(c2, c1, c0)
+ int c2, c1, c0;
+{
+ unsigned short val;
+ if (c2 >= 0xe0){
+ val = (c2 & 0x0f) << 12;
+ val |= (c1 & 0x3f) << 6;
+ val |= (c0 & 0x3f);
+ }else if (c2 >= 0xc0){
+ val = (c2 & 0x1f) << 6;
+ val |= (c1 & 0x3f) << 6;
+ }else{
+ val = c2;
+ }
+ return val;
+}
+
+int
w16e_conv(val, p2, p1)
unsigned short val;
int *p2, *p1;
int c2, c1, c0;
unsigned short **pp;
int psize;
+ int ret = 0;
w16w_conv(val, &c2, &c1, &c0);
if (c1){
if (c0){
pp = utf8_to_euc_3bytes[c2 - 0x80];
psize = sizeof_utf8_to_euc_C2;
+ ret = w_iconv_common(c1, c0, pp, psize, p2, p1);
}else{
pp = utf8_to_euc_2bytes;
psize = sizeof_utf8_to_euc_2bytes;
+ ret = w_iconv_common(c2, c1, pp, psize, p2, p1);
}
- return w_iconv_common(c1, c0, pp, psize, p2, p1);
+#ifdef NUMCHAR_OPTION
+ if (ret){
+ *p2 = 0;
+ *p1 = CLASS_UTF16 | val;
+ ret = 0;
+ }
+#endif
}
- return val;
+ return ret;
}
int
int ret;
if (c2==0376 && c1==0377){
- utf16_mode = UTF16_INPUT;
+ utf16_mode = UTF16LE_INPUT;
return 0;
} else if (c2==0377 && c1==0376){
utf16_mode = UTF16BE_INPUT;
return 0;
}
- if (utf16_mode == UTF16BE_INPUT) {
+ if (c2 != EOF && utf16_mode == UTF16BE_INPUT) {
int tmp;
tmp=c1; c1=c2; c2=tmp;
}
- if (c2==0 || c2==EOF) {
+ if ((c2==0 && c1 < 0x80) || c2==EOF) {
(*oconv)(c2, c1);
return 0;
}
{
extern unsigned short euc_to_utf8_1byte[];
extern unsigned short * euc_to_utf8_2bytes[];
+ extern unsigned short * euc_to_utf8_2bytes_ms[];
unsigned short *p;
if (c2 == X0201) {
c2 &= 0x7f;
c2 = (c2&0x7f) - 0x21;
if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
- p = euc_to_utf8_2bytes[c2];
+ p = ms_ucs_map_f ? euc_to_utf8_2bytes_ms[c2] : euc_to_utf8_2bytes[c2];
else
return 0;
}
int c2,
c1;
{
+ int c0;
+#ifdef NUMCHAR_OPTION
+ if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+ w16w_conv(c1, &c2, &c1, &c0);
+ (*o_putc)(c2);
+ if (c1){
+ (*o_putc)(c1);
+ if (c0) (*o_putc)(c0);
+ }
+ }
+#endif
if (c2 == EOF) {
(*o_putc)(EOF);
return;
- } else if (c2 == 0) {
+ }
+
+ if (unicode_bom_f==2) {
+ (*o_putc)('\357');
+ (*o_putc)('\273');
+ (*o_putc)('\277');
+ unicode_bom_f=1;
+ }
+
+ if (c2 == 0) {
output_mode = ASCII;
(*o_putc)(c1);
} else if (c2 == ISO8859_1) {
output_mode = ISO8859_1;
(*o_putc)(c1 | 0x080);
} else {
- unsigned short val = (unsigned short)e2w_conv(c2, c1);
output_mode = UTF8;
-
- if (0 < val && val < 0x80){
- (*o_putc)(val);
- }else if (val < 0x800){
- (*o_putc)(0xc0 | (val >> 6));
- (*o_putc)(0x80 | (val & 0x3f));
- }else{
- (*o_putc)(0xe0 | (val >> 12));
- (*o_putc)(0x80 | ((val >> 6) & 0x3f));
- (*o_putc)(0x80 | (val & 0x3f));
+ w16w_conv((unsigned short)e2w_conv(c2, c1), &c2, &c1, &c0);
+ (*o_putc)(c2);
+ if (c1){
+ (*o_putc)(c1);
+ if (c0) (*o_putc)(c0);
}
}
}
return;
}
- if (w_oconv16_begin_f==2) {
+ if (unicode_bom_f==2) {
if (w_oconv16_LE){
(*o_putc)((unsigned char)'\377');
(*o_putc)('\376');
(*o_putc)('\376');
(*o_putc)((unsigned char)'\377');
}
- w_oconv16_begin_f=1;
+ unicode_bom_f=1;
}
if (c2 == ISO8859_1) {
c2 = 0;
c1 |= 0x80;
+#ifdef NUMCHAR_OPTION
+ } else if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16) {
+ c2 = (c1 >> 8) & 0xff;
+ c1 &= 0xff;
+#endif
} else if (c2) {
unsigned short val = (unsigned short)e2w_conv(c2, c1);
c2 = (val >> 8) & 0xff;
int c2,
c1;
{
+#ifdef NUMCHAR_OPTION
+ if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+ w16e_conv(c1, &c2, &c1);
+ }
+#endif
if (c2 == EOF) {
(*o_putc)(EOF);
return;
output_mode = ISO8859_1;
(*o_putc)(c1 | 0x080);
} else {
- if ((c1<0x20 || 0x7e<c1) ||
- (c2<0x20 || 0x7e<c2)) {
+ if ((c1<0x21 || 0x7e<c1) ||
+ (c2<0x21 || 0x7e<c2)) {
set_iconv(FALSE, 0);
return; /* too late to rescue this char */
}
int c2,
c1;
{
+#ifdef NUMCHAR_OPTION
+ if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+ w16e_conv(c1, &c2, &c1);
+ }
+#endif
if (c2 == EOF) {
(*o_putc)(EOF);
return;
}
output_mode = SHIFT_JIS;
e2s_conv(c2, c1, &c2, &c1);
+
+#ifdef SHIFTJIS_CP932
+ if (cp932inv_f
+ && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
+ extern unsigned short cp932inv[2][189];
+ int 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);
}
}
int c2,
c1;
{
+#ifdef NUMCHAR_OPTION
+ if ((c1 & CLASS_MASK) == CLASS_UTF16){
+ w16e_conv(c1, &c2, &c1);
+ }
+#endif
if (c2 == EOF) {
if (output_mode !=ASCII && output_mode!=ISO8859_1) {
(*o_putc)(ESC);
(unsigned char *)"\075?ISO-2022-JP?Q?",
#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
(unsigned char *)"\075?UTF-8?B?",
+ (unsigned char *)"\075?UTF-8?Q?",
#endif
+ (unsigned char *)"\075?US-ASCII?Q?",
NULL
};
+
+/* \e$B3:Ev$9$k%3!<%I$NM%@hEY$r>e$2$k$?$a$NL\0u\e(B */
+int (*mime_priority_func[])PROTO((int c2, int c1, int c0)) = {
+ e_iconv, s_iconv, 0, 0, 0, 0,
+#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
+ w_iconv, w_iconv,
+#endif
+ 0,
+};
+
int mime_encode[] = {
JAPANESE_EUC, SHIFT_JIS,ISO8859_1, ISO8859_1, X0208, X0201,
#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
- UTF8,
+ UTF8, UTF8,
#endif
+ ASCII,
0
};
int mime_encode_method[] = {
'B', 'B','Q', 'B', 'B', 'Q',
#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
- 'B',
+ 'B', 'Q',
#endif
+ 'Q',
0
};
}
}
mime_decode_mode = p[i-2];
+
+ clr_code_score(find_inputcode_byfunc(mime_priority_func[j]), SCORE_iMIME);
+
if (mime_decode_mode=='B') {
mimebuf_f = unbuf_f;
if (!unbuf_f) {
char *str;
{
if (debug_f){
- fputs(str, stderr);
+ fprintf(stderr, "%s\n", str);
}
}
#endif
-#ifdef INPUT_OPTION
+void
+set_input_codename (codename)
+ char *codename;
+{
+ if (guess_f &&
+ is_inputcode_set &&
+ strcmp(codename, "") != 0 &&
+ strcmp(codename, input_codename) != 0)
+ {
+ is_inputcode_mixed = TRUE;
+ }
+ input_codename = codename;
+ is_inputcode_set = TRUE;
+}
+
+void
+print_guessed_code (filename)
+ char *filename;
+{
+ char *codename = "BINARY";
+ if (!is_inputcode_mixed) {
+ if (strcmp(input_codename, "") == 0) {
+ codename = "ASCII";
+ } else {
+ codename = input_codename;
+ }
+ }
+ if (filename != NULL) printf("%s:", filename);
+ printf("%s\n", codename);
+}
+
int
hex2bin(x)
int x;
return nkf_toupper(x) - 'A' + 10;
}
+#ifdef INPUT_OPTION
+
#ifdef ANSI_C_PROTOTYPE
int hex_getc(int ch, FILE *f, int (*g)(FILE *f), int (*u)(int c, FILE *f))
#else
{
return (*i_uungetc)(c, f);
}
+#endif
+#ifdef NUMCHAR_OPTION
int
numchar_getc(f)
FILE *f;
}
}
if (c != -1){
- int c2, c1, c0;
- if (c < 0x80){
- return c;
- }
- if (0x100 <= c){
- w16w_conv(c, &c2, &c1, &c0);
- if (iconv == w_iconv){
- if (c0){
- (*u)(c0, f);
- }
- (*u)(c1, f);
- return c2;
- }
- if (w2e_conv(c2, c1, c0, &c2, &c1) == 0){
- c2 |= 0x80;
- c1 |= 0x80;
- if (iconv == s_iconv){
- e2s_conv(c2, c1, &c2, &c1);
- }
- (*u)(c1, f);
- return c2;
- }
- }
+ return CLASS_UTF16 | c;
}
while (i > 0){
(*u)(buf[i], f);
}
}
#ifdef UTF8_OUTPUT_ENABLE
- if (w_oconv16_begin_f) {
- w_oconv16_begin_f = 2;
+ if (unicode_bom_f) {
+ unicode_bom_f = 2;
}
#endif
f_line = 0;
broken_last = 0;
z_prev2=0,z_prev1=0;
+ {
+ int i;
+ for (i = 0; i < 256; i++){
+ prefix_table[i] = 0;
+ }
+ }
+ input_codename = "";
+ is_inputcode_mixed = FALSE;
+ is_inputcode_set = FALSE;
}
#endif
{
fprintf(stderr,"USAGE: nkf(nkf32,wnkf,nkf2) -[flags] [in file] .. [out file for -O flag]\n");
fprintf(stderr,"Flags:\n");
- fprintf(stderr,"b,u Output is bufferred (DEFAULT),Output is unbufferred\n");
+ fprintf(stderr,"b,u Output is buffered (DEFAULT),Output is unbuffered\n");
#ifdef DEFAULT_CODE_SJIS
fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS (DEFAULT), AT&T JIS (EUC), UTF-8\n");
#endif
fprintf(stderr,"long name options\n");
fprintf(stderr," --fj,--unix,--mac,--windows convert for the system\n");
fprintf(stderr," --jis,--euc,--sjis,--utf8,--utf16,--mime,--base64 convert for the code\n");
+#ifdef OVERWRITE
+ fprintf(stderr," --overwrite Overwrite original listed files by filtered result\n");
+#endif
+ fprintf(stderr," -g, --guess Guess the input code\n");
fprintf(stderr," --help,--version\n");
version();
}
OSDN Git Service
