** E-Mail: furukawa@tcp-ip.or.jp
** \e$B$^$G8fO"Mm$r$*4j$$$7$^$9!#\e(B
***********************************************************************/
-/* $Id: nkf.c,v 1.107 2006/09/15 07:23:20 naruse Exp $ */
+/* $Id: nkf.c,v 1.127 2007/07/19 20:08:29 naruse Exp $ */
#define NKF_VERSION "2.0.8"
-#define NKF_RELEASE_DATE "2006-09-15"
+#define NKF_RELEASE_DATE "2007-07-20"
#include "config.h"
#include "utf8tbl.h"
#define COPY_RIGHT \
"Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW\n" \
- "Copyright (C) 2002-2006 Kono, Furukawa, Naruse, mastodon"
+ "Copyright (C) 2002-2007 Kono, Furukawa, Naruse, mastodon"
/*
* 0: Shift_JIS, eucJP-ascii
* 1: eucJP-ms
* 2: CP932, CP51932
+ * 3: CP10001
*/
-#define UCS_MAP_ASCII 0
-#define UCS_MAP_MS 1
-#define UCS_MAP_CP932 2
+#define UCS_MAP_ASCII 0
+#define UCS_MAP_MS 1
+#define UCS_MAP_CP932 2
+#define UCS_MAP_CP10001 3
static int ms_ucs_map_f = UCS_MAP_ASCII;
#endif
#ifdef UTF8_INPUT_ENABLE
#ifdef SHIFTJIS_CP932
/* invert IBM extended characters to others */
-static int cp51932_f = TRUE;
+static int cp51932_f = FALSE;
/* invert NEC-selected IBM extended characters to IBM extended characters */
static int cp932inv_f = TRUE;
{"Shift_JIS", 0, 0, 0, {0, 0, 0}, s_status, s_iconv, 0},
#ifdef UTF8_INPUT_ENABLE
{"UTF-8", 0, 0, 0, {0, 0, 0}, w_status, w_iconv, 0},
+ {"UTF-16", 0, 0, 0, {0, 0, 0}, NULL, w_iconv16, 0},
+ {"UTF-32", 0, 0, 0, {0, 0, 0}, NULL, w_iconv32, 0},
#endif
{0}
};
#endif
static int crmode_f = 0; /* CR, NL, CRLF */
+static nkf_char prev_cr = 0;
#ifdef EASYWIN /*Easy Win */
static int end_check;
#endif /*Easy Win */
}
} else {
int nfiles = argc;
+ int is_argument_error = FALSE;
while (argc--) {
is_inputcode_mixed = FALSE;
is_inputcode_set = FALSE;
#endif
if ((fin = fopen((origfname = *argv++), "r")) == NULL) {
perror(*--argv);
- return(-1);
+ *argv++;
+ is_argument_error = TRUE;
+ continue;
} else {
#ifdef OVERWRITE
int fd = 0;
#endif
}
}
+ if (is_argument_error)
+ return(-1);
}
#ifdef EASYWIN /*Easy Win */
if (file_out_f == FALSE)
codeset[i] = nkf_toupper(p[i]);
}
codeset[i] = 0;
- if(strcmp(codeset, "ISO-2022-JP") == 0 ||
- strcmp(codeset, "X-ISO2022JP-CP932") == 0 ||
+ if(strcmp(codeset, "ISO-2022-JP") == 0){
+ input_f = JIS_INPUT;
+ }else if(strcmp(codeset, "X-ISO2022JP-CP932") == 0 ||
strcmp(codeset, "CP50220") == 0 ||
strcmp(codeset, "CP50221") == 0 ||
- strcmp(codeset, "CP50222") == 0 ||
- strcmp(codeset, "ISO-2022-JP-MS") == 0){
+ strcmp(codeset, "CP50222") == 0){
input_f = JIS_INPUT;
+#ifdef SHIFTJIS_CP932
+ cp51932_f = TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP932;
+#endif
}else if(strcmp(codeset, "ISO-2022-JP-1") == 0){
input_f = JIS_INPUT;
#ifdef X0212_ENABLE
x0213_f = TRUE;
}else if(strcmp(codeset, "SHIFT_JIS") == 0){
input_f = SJIS_INPUT;
- if (x0201_f==NO_X0201) x0201_f=TRUE;
}else if(strcmp(codeset, "WINDOWS-31J") == 0 ||
strcmp(codeset, "CSWINDOWS31J") == 0 ||
strcmp(codeset, "CP932") == 0 ||
strcmp(codeset, "MS932") == 0){
input_f = SJIS_INPUT;
- x0201_f = FALSE;
#ifdef SHIFTJIS_CP932
cp51932_f = TRUE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
ms_ucs_map_f = UCS_MAP_CP932;
#endif
+ }else if(strcmp(codeset, "CP10001") == 0){
+ input_f = SJIS_INPUT;
+#ifdef SHIFTJIS_CP932
+ cp51932_f = TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP10001;
+#endif
}else if(strcmp(codeset, "EUCJP") == 0 ||
strcmp(codeset, "EUC-JP") == 0){
input_f = EUC_INPUT;
}else if(strcmp(codeset, "CP51932") == 0){
input_f = EUC_INPUT;
- x0201_f = FALSE;
#ifdef SHIFTJIS_CP932
cp51932_f = TRUE;
#endif
strcmp(codeset, "EUCJP-MS") == 0 ||
strcmp(codeset, "EUCJPMS") == 0){
input_f = EUC_INPUT;
- x0201_f = FALSE;
#ifdef SHIFTJIS_CP932
cp51932_f = FALSE;
#endif
}else if(strcmp(codeset, "EUC-JP-ASCII") == 0 ||
strcmp(codeset, "EUCJP-ASCII") == 0){
input_f = EUC_INPUT;
- x0201_f = FALSE;
#ifdef SHIFTJIS_CP932
cp51932_f = FALSE;
#endif
x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
cp51932_f = FALSE;
- cp932inv_f = FALSE;
#endif
- if (x0201_f==NO_X0201) x0201_f=TRUE;
}else if(strcmp(codeset, "EUC-JISX0213") == 0 ||
strcmp(codeset, "EUC-JIS-2004") == 0){
input_f = EUC_INPUT;
- x0201_f = FALSE;
x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
cp51932_f = FALSE;
- cp932inv_f = FALSE;
#endif
#ifdef UTF8_INPUT_ENABLE
}else if(strcmp(codeset, "UTF-8") == 0 ||
continue;
}
if (strcmp(long_option[i].name, "oc=") == 0){
+ x0201_f = FALSE;
for (i=0; i < 16 && SPACE < p[i] && p[i] < DEL; i++){
codeset[i] = nkf_toupper(p[i]);
}
codeset[i] = 0;
- if(strcmp(codeset, "ISO-2022-JP") == 0 ||
- strcmp(codeset, "CP50220") == 0){
+ if(strcmp(codeset, "ISO-2022-JP") == 0){
output_conv = j_oconv;
}else if(strcmp(codeset, "X-ISO2022JP-CP932") == 0){
output_conv = j_oconv;
no_cp932ext_f = TRUE;
- }else if(strcmp(codeset, "CP50221") == 0 ||
- strcmp(codeset, "ISO-2022-JP-MS") == 0){
- output_conv = j_oconv;
- x0201_f = FALSE;
- }else if(strcmp(codeset, "ISO-2022-JP-1") == 0){
+#ifdef SHIFTJIS_CP932
+ cp932inv_f = FALSE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP932;
+#endif
+ }else if(strcmp(codeset, "CP50220") == 0){
output_conv = j_oconv;
-#ifdef X0212_ENABLE
- x0212_f = TRUE;
+ x0201_f = TRUE;
+#ifdef SHIFTJIS_CP932
+ cp932inv_f = FALSE;
#endif
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP932;
+#endif
+ }else if(strcmp(codeset, "CP50221") == 0){
+ output_conv = j_oconv;
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp932inv_f = FALSE;
#endif
- }else if(strcmp(codeset, "ISO-2022-JP-3") == 0){
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP932;
+#endif
+ }else if(strcmp(codeset, "ISO-2022-JP-1") == 0){
output_conv = j_oconv;
#ifdef X0212_ENABLE
x0212_f = TRUE;
#endif
- x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp932inv_f = FALSE;
#endif
- }else if(strcmp(codeset, "ISO-2022-JP-MS") == 0){
+ }else if(strcmp(codeset, "ISO-2022-JP-3") == 0){
output_conv = j_oconv;
- x0201_f = FALSE;
#ifdef X0212_ENABLE
x0212_f = TRUE;
#endif
+ x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp932inv_f = FALSE;
#endif
}else if(strcmp(codeset, "SHIFT_JIS") == 0){
output_conv = s_oconv;
strcmp(codeset, "CP932") == 0 ||
strcmp(codeset, "MS932") == 0){
output_conv = s_oconv;
- x0201_f = FALSE;
-#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
- cp932inv_f = TRUE;
-#endif
#ifdef UTF8_OUTPUT_ENABLE
ms_ucs_map_f = UCS_MAP_CP932;
#endif
+ }else if(strcmp(codeset, "CP10001") == 0){
+ output_conv = s_oconv;
+#ifdef UTF8_OUTPUT_ENABLE
+ ms_ucs_map_f = UCS_MAP_CP10001;
+#endif
}else if(strcmp(codeset, "EUCJP") == 0 ||
strcmp(codeset, "EUC-JP") == 0){
output_conv = e_oconv;
}else if(strcmp(codeset, "CP51932") == 0){
output_conv = e_oconv;
- x0201_f = FALSE;
#ifdef SHIFTJIS_CP932
- cp51932_f = TRUE;
+ cp932inv_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
ms_ucs_map_f = UCS_MAP_CP932;
strcmp(codeset, "EUCJP-MS") == 0 ||
strcmp(codeset, "EUCJPMS") == 0){
output_conv = e_oconv;
- x0201_f = FALSE;
#ifdef X0212_ENABLE
x0212_f = TRUE;
#endif
-#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
-#endif
#ifdef UTF8_OUTPUT_ENABLE
ms_ucs_map_f = UCS_MAP_MS;
#endif
}else if(strcmp(codeset, "EUC-JP-ASCII") == 0 ||
strcmp(codeset, "EUCJP-ASCII") == 0){
output_conv = e_oconv;
- x0201_f = FALSE;
#ifdef X0212_ENABLE
x0212_f = TRUE;
#endif
-#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
-#endif
#ifdef UTF8_OUTPUT_ENABLE
ms_ucs_map_f = UCS_MAP_ASCII;
#endif
#endif
x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
- cp51932_f = FALSE;
+ cp932inv_f = FALSE;
#endif
#ifdef UTF8_OUTPUT_ENABLE
}else if(strcmp(codeset, "UTF-8") == 0){
continue;
case 'e': /* AT&T EUC output */
output_conv = e_oconv;
+ cp932inv_f = FALSE;
continue;
case 's': /* SJIS output */
output_conv = s_oconv;
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(!input_f){
set_iconv(TRUE, w_iconv32);
}
- input_endian = ENDIAN_BIG;
- return;
+ 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(!input_f){
set_iconv(TRUE, w_iconv32);
}
- input_endian = ENDIAN_2143;
- return;
+ 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);
if(!input_f){
set_iconv(TRUE, w_iconv);
}
- return;
+ if (iconv == w_iconv) {
+ return;
+ }
+ (*i_ungetc)(0xBF,f);
}else (*i_ungetc)(c2,f);
(*i_ungetc)(0xBB,f);
}else (*i_ungetc)(c2,f);
if(!input_f){
set_iconv(TRUE, w_iconv32);
}
- input_endian = ENDIAN_3412;
- return;
+ 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_f){
set_iconv(TRUE, w_iconv16);
}
- input_endian = ENDIAN_BIG;
- return;
+ if (iconv == w_iconv16) {
+ input_endian = ENDIAN_BIG;
+ return;
+ }
+ (*i_ungetc)(0xFF,f);
}else (*i_ungetc)(c2,f);
(*i_ungetc)(0xFE,f);
break;
if(!input_f){
set_iconv(TRUE, w_iconv32);
}
- input_endian = ENDIAN_LITTLE;
- return;
+ 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_f){
set_iconv(TRUE, w_iconv16);
}
- input_endian = ENDIAN_LITTLE;
- return;
+ if (iconv == w_iconv16) {
+ input_endian = ENDIAN_LITTLE;
+ return;
+ }
+ (*i_ungetc)(0xFE,f);
}else (*i_ungetc)(c2,f);
(*i_ungetc)(0xFF,f);
break;
code_status(c1);
if (c2) {
/* second byte */
- if (c2 > DEL) {
+ if (c2 > ((input_f == JIS_INPUT && ms_ucs_map_f) ? 0x92 : DEL)) {
/* in case of 8th bit is on */
if (!estab_f&&!mime_decode_mode) {
/* in case of not established yet */
else
c2 = 0;
NEXT;
- } else
- /* in case of already established */
- if (c1 < AT) {
- /* ignore bogus code */
- c2 = 0;
- NEXT;
- } else
- SEND;
+ } 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 */
c0 <<= 8;
if ((c3 = (*i_getc)(f)) != EOF) {
c0 |= c3;
- } else c1 = EOF;
- } else c1 = EOF;
+ } else c2 = EOF;
+ } else c2 = EOF;
}
- }
+ } else c2 = EOF;
} else {
if ((c2 = (*i_getc)(f)) != EOF) {
if (0xD8 <= c2 && c2 <= 0xDB) {
if ((c0 = (*i_getc)(f)) != EOF) {
c0 <<= 8;
c0 |= c3;
- } else c1 = EOF;
- } else c1 = EOF;
+ } else c2 = EOF;
+ } else c2 = EOF;
}
- } else c1 = EOF;
+ } else c2 = EOF;
}
SEND;
} else if(iconv == w_iconv32){
}
c2 = 0;
}else{
- c1 = EOF;
+ c2 = EOF;
}
SEND;
} else
SEND;
} else
#endif
- if (c1 > DEL) {
+ if (c1 > ((input_f == JIS_INPUT && ms_ucs_map_f) ? 0x92 : DEL)) {
/* 8 bit code */
if (!estab_f && !iso8859_f) {
/* not established yet */
} else { /* bogus code, skip SSO and one byte */
NEXT;
}
+ } else if (ms_ucs_map_f == UCS_MAP_CP10001 &&
+ (c1 == 0xFD || c1 == 0xFE)) {
+ /* CP10001 */
+ c2 = X0201;
+ c1 &= 0x7f;
+ SEND;
} else {
/* already established */
c2 = c1;
/* normal ASCII code */
SEND;
}
- } else if (!is_8bit && c1 == SI) {
+ } else if (c1 == SI && (!is_8bit || mime_decode_mode)) {\r
shift_mode = FALSE;
NEXT;
- } else if (!is_8bit && c1 == SO) {
+ } else if (c1 == SO && (!is_8bit || mime_decode_mode)) {\r
shift_mode = TRUE;
NEXT;
- } else if (!is_8bit && c1 == ESC ) {
+ } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) {\r
if ((c1 = (*i_getc)(f)) == EOF) {
/* (*oconv)(0, ESC); don't send bogus code */
LAST;
(*oconv)(0, ESC);
SEND;
}
- } else if ((c1 == NL || c1 == CR) && broken_f&4) {
- input_mode = ASCII; set_iconv(FALSE, 0);
- SEND;
- } else if (c1 == NL && mime_decode_f && !mime_decode_mode ) {
- if ((c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
- i_ungetc(SPACE,f);
- continue;
- } else {
- i_ungetc(c1,f);
- }
- c1 = NL;
- SEND;
- } else if (c1 == CR && mime_decode_f && !mime_decode_mode ) {
- if ((c1=(*i_getc)(f))!=EOF) {
- if (c1==SPACE) {
- i_ungetc(SPACE,f);
- continue;
- } else if (c1 == NL && (c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
- i_ungetc(SPACE,f);
- continue;
+ } 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 {
- i_ungetc(c1,f);
+ 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 int jphone_emoji_first_table[7] = {2, 0, 3, 4, 5, 0, 1};
+ c0 = (jphone_emoji_first_table[c1 % 7] << 8) - SPACE + 0xE000 + CLASS_UNICODE;
+ while ((c1 = (*i_getc)(f)) != EOF) {
+ if (SPACE <= c1 && c1 <= 'z') {
+ (*oconv)(0, c1 + c0);
+ } else break; /* c1 == SO */
+ }
+ }
}
- i_ungetc(NL,f);
+ if (c1 == EOF) LAST;
+ NEXT;
} else {
- i_ungetc(c1,f);
+ /* lonely ESC */
+ (*oconv)(0, ESC);
+ SEND;
}
- c1 = CR;
- SEND;
+ } else if (c1 == NL || c1 == CR) {
+ if (broken_f&4) {
+ input_mode = ASCII; set_iconv(FALSE, 0);
+ SEND;
+ } else if (mime_decode_f && !mime_decode_mode){
+ if (c1 == NL) {
+ if ((c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
+ i_ungetc(SPACE,f);
+ continue;
+ } else {
+ i_ungetc(c1,f);
+ }
+ c1 = NL;
+ SEND;
+ } else { /* if (c1 == CR)*/
+ if ((c1=(*i_getc)(f))!=EOF) {
+ if (c1==SPACE) {
+ i_ungetc(SPACE,f);
+ continue;
+ } else if (c1 == NL && (c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
+ i_ungetc(SPACE,f);
+ continue;
+ } else {
+ i_ungetc(c1,f);
+ }
+ i_ungetc(NL,f);
+ } else {
+ i_ungetc(c1,f);
+ }
+ c1 = CR;
+ SEND;
+ }
+ }
+ if (!crmode_f) {
+ if (prev_cr && c1 == NL) crmode_f = CRLF;
+ else crmode_f = c1;
+ }
+ } else if (c1 == DEL && input_mode == X0208 ) {
+ /* CP5022x */
+ c2 = c1;
+ NEXT;
} else
SEND;
}
break;
case X0208:
case X0213_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
code_status(c1);
}
while (p->name){
- if (p->score < result->score){
+ if (p->status_func && p->score < result->score){
result = p;
}
++p;
}
- set_iconv(FALSE, result->iconv_func);
+ set_iconv(TRUE, result->iconv_func);
}
#endif
static const nkf_char shift_jisx0213_s1a3_table[5][2] ={ { 1, 8}, { 3, 4}, { 5,12}, {13,14}, {15, 0} };
#ifdef SHIFTJIS_CP932
- if (cp51932_f && is_ibmext_in_sjis(c2)){
-#if 0
- extern const unsigned short shiftjis_cp932[3][189];
-#endif
+ if (!cp932inv_f && is_ibmext_in_sjis(c2)){
val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
if (val){
c2 = val >> 8;
c1 = val & 0xff;
}
}
+ 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 */
#ifdef X0212_ENABLE
if (!x0213_f && is_ibmext_in_sjis(c2)){
-#if 0
- extern const unsigned short shiftjis_x0212[3][189];
-#endif
val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40];
if (val){
if (val > 0x7FFF){
- c2 = PREFIX_EUCG3 | (val >> 8);
+ c2 = PREFIX_EUCG3 | ((val >> 8) & 0x7f);
c1 = val & 0xff;
}else{
c2 = val >> 8;
c1 &= 0x7f;
} else if ((c2 == EOF) || (c2 == 0) || c2 < SPACE) {
/* 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;
if (c0 == 0){
return -1;
}
- c2 = (c2 << 8) | (c1 & 0x7f);
- c1 = c0 & 0x7f;
+ 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;
- }
- }
- }
+ 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 = X0201;
} else if ((c2 == EOF) || (c2 == 0) || c2 < SPACE) {
/* NOP */
} else {
- c1 &= 0x7f;
- c2 &= 0x7f;
+ 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 unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1)
{
-#if 0
- 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[];
- extern const unsigned short *const *const utf8_to_euc_3bytes[];
- extern const unsigned short *const *const utf8_to_euc_3bytes_ms[];
- extern const unsigned short *const *const utf8_to_euc_3bytes_932[];
-#endif
const unsigned short *const *pp;
const unsigned short *const *const *ppp;
static const int no_best_fit_chars_table_C2[] =
if(no_best_fit_chars_table_932_C3[c1&0x3F]) return 1;
break;
}
- }else if(cp51932_f){
+ }else if(!cp932inv_f){
switch(c2){
case 0xC2:
if(no_best_fit_chars_table_C2[c1&0x3F]) return 1;
}
}else if(ms_ucs_map_f == UCS_MAP_MS){
if(c2 == 0xC2 && no_best_fit_chars_table_C2_ms[c1&0x3F]) return 1;
+ }else if(ms_ucs_map_f == UCS_MAP_CP10001){
+ switch(c2){
+ case 0xC2:
+ switch(c1){
+ case 0xA2:
+ case 0xA3:
+ case 0xA5:
+ case 0xA6:
+ case 0xAC:
+ case 0xAF:
+ case 0xB8:
+ return 1;
+ }
+ break;
+ }
}
}
pp =
ms_ucs_map_f == UCS_MAP_CP932 ? utf8_to_euc_2bytes_932 :
ms_ucs_map_f == UCS_MAP_MS ? utf8_to_euc_2bytes_ms :
+ ms_ucs_map_f == UCS_MAP_CP10001 ? utf8_to_euc_2bytes_mac :
utf8_to_euc_2bytes;
ret = w_iconv_common(c2, c1, pp, sizeof_utf8_to_euc_2bytes, p2, p1);
}else if(c0 < 0xF0){
if(c1 == 0x80 || c0 == 0x9C) return 1;
break;
}
+ }else if(ms_ucs_map_f == UCS_MAP_CP10001){
+ switch(c2){
+ case 0xE3:
+ switch(c1){
+ case 0x82:
+ if(c0 == 0x94) return 1;
+ break;
+ case 0x83:
+ if(c0 == 0xBB) return 1;
+ break;
+ }
+ break;
+ }
}else{
switch(c2){
case 0xE2:
if(c0 == 0x8D) return 1;
break;
case 0xBD:
- if(c0 == 0x9E && cp51932_f) return 1;
+ if(c0 == 0x9E && !cp932inv_f) return 1;
break;
case 0xBF:
if(0xA0 <= c0 && c0 <= 0xA5) return 1;
ppp =
ms_ucs_map_f == UCS_MAP_CP932 ? utf8_to_euc_3bytes_932 :
ms_ucs_map_f == UCS_MAP_MS ? utf8_to_euc_3bytes_ms :
+ ms_ucs_map_f == UCS_MAP_CP10001 ? utf8_to_euc_3bytes_mac :
utf8_to_euc_3bytes;
ret = w_iconv_common(c1, c0, ppp[c2 - 0xE0], sizeof_utf8_to_euc_C2, p2, p1);
}else return -1;
#ifdef SHIFTJIS_CP932
- if (!ret && cp51932_f && is_eucg3(*p2)) {
+ if (!ret && !cp932inv_f && is_eucg3(*p2)) {
nkf_char s2, s1;
if (e2s_conv(*p2, *p1, &s2, &s1) == 0) {
s2e_conv(s2, s1, p2, p1);
#ifdef UTF8_OUTPUT_ENABLE
nkf_char e2w_conv(nkf_char c2, nkf_char c1)
{
-#if 0
- 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[];
- extern const unsigned short *const x0212_to_utf8_2bytes[];
-#endif
const unsigned short *p;
if (c2 == X0201) {
+ if (ms_ucs_map_f == UCS_MAP_CP10001) {
+ switch (c1) {
+ case 0x20:
+ return 0xA0;
+ case 0x7D:
+ return 0xA9;
+ }
+ }
p = euc_to_utf8_1byte;
#ifdef X0212_ENABLE
} else if (is_eucg3(c2)){
c2 &= 0x7f;
c2 = (c2&0x7f) - 0x21;
if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
- p = ms_ucs_map_f != UCS_MAP_ASCII ? euc_to_utf8_2bytes_ms[c2] : euc_to_utf8_2bytes[c2];
+ p =
+ ms_ucs_map_f == UCS_MAP_ASCII ? euc_to_utf8_2bytes[c2] :
+ ms_ucs_map_f == UCS_MAP_CP10001 ? euc_to_utf8_2bytes_mac[c2] :
+ euc_to_utf8_2bytes_ms[c2];
else
return 0;
}
nkf_char val = e2w_conv(c2, c1);
c2 = (val >> 8) & 0xff;
c1 = val & 0xff;
+ if (!val) return;
}
if (output_endian == ENDIAN_LITTLE){
(*o_putc)(c1);
#endif
} else if (c2) {
c1 = e2w_conv(c2, c1);
+ if (!c1) return;
}
if (output_endian == ENDIAN_LITTLE){
(*o_putc)( c1 & NKF_INT32_C(0x000000FF));
if (c2 == 0 && is_unicode_capsule(c1)){
w16e_conv(c1, &c2, &c1);
if (c2 == 0 && is_unicode_capsule(c1)){
- if(encode_fallback)(*encode_fallback)(c1);
- return;
+ 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
} else if (is_eucg3(c2)){
output_mode = JAPANESE_EUC;
#ifdef SHIFTJIS_CP932
- if (cp51932_f){
+ if (!cp932inv_f){
nkf_char s2, s1;
if (e2s_conv(c2, c1, &s2, &s1) == 0){
s2e_conv(s2, s1, &c2, &c1);
{
nkf_char ndx;
if (is_eucg3(c2)){
- ndx = c2 & 0xff;
+ ndx = c2 & 0x7f;
if (x0213_f){
if((0x21 <= ndx && ndx <= 0x2F)){
if (p2) *p2 = ((ndx - 1) >> 1) + 0xec - ndx / 8 * 3;
else if(nkf_isgraph(ndx)){
nkf_char val = 0;
const unsigned short *ptr;
-#if 0
- extern const unsigned short *const x0212_shiftjis[];
-#endif
ptr = x0212_shiftjis[ndx - 0x21];
if (ptr){
val = ptr[(c1 & 0x7f) - 0x21];
if (c2 == 0 && is_unicode_capsule(c1)){
w16e_conv(c1, &c2, &c1);
if (c2 == 0 && is_unicode_capsule(c1)){
- if(encode_fallback)(*encode_fallback)(c1);
- return;
- }
+ 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(encode_fallback)(*encode_fallback)(c1);
+ return;
+ }
+ }
}
#endif
if (c2 == EOF) {
#ifdef SHIFTJIS_CP932
if (cp932inv_f
&& CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
-#if 0
- extern const unsigned short cp932inv[2][189];
-#endif
nkf_char c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
if (c){
c2 = c >> 8;
if (c2 == 0 && is_unicode_capsule(c1)){
w16e_conv(c1, &c2, &c1);
if (c2 == 0 && is_unicode_capsule(c1)){
- if(encode_fallback)(*encode_fallback)(c1);
- return;
+ 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
}
(*o_putc)(c1);
} else {
- if(c2<0x20 || 0x7e<c2 || c1<0x20 || 0x7e<c1) return;
+ 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!=X0213_1) {
output_mode = X0213_1;
return c;
}
-static nkf_char prev_cr = 0;
-
void cr_conv(nkf_char c2, nkf_char c1)
{
if (prev_cr) {
/* if (c2) c1 &= 0x7f; assertion */
+ if (c2 == X0201 && (c1 == 0x20 || c1 == 0x7D || c1 == 0x7E)) {
+ (*o_zconv)(c2,c1);
+ return;
+ }
if (x0201_f && z_prev2==X0201) { /* X0201 */
if (c1==(0xde&0x7f)) { /* \e$BByE@\e(B */
z_prev2=0;
void print_guessed_code(char *filename)
{
char *codename = "BINARY";
+ char *str_crmode = NULL;
if (!is_inputcode_mixed) {
if (strcmp(input_codename, "") == 0) {
codename = "ASCII";
} else {
codename = input_codename;
}
+ if (crmode_f == CR) str_crmode = "CR";
+ else if (crmode_f == NL) str_crmode = "LF";
+ else if (crmode_f == CRLF) str_crmode = "CRLF";
}
if (filename != NULL) printf("%s:", filename);
- printf("%s\n", codename);
+ if (str_crmode != NULL) printf("%s (%s)\n", codename, str_crmode);
+ else printf("%s\n", codename);
}
#endif /*WIN32DLL*/
int i=0, j, k=1, lower, upper;
nkf_char buf[9];
const nkf_nfchar *array;
-#if 0
- extern const struct normalization_pair normalization_table[];
-#endif
buf[i] = (*g)(f);
while (k > 0 && ((buf[i] & 0xc0) != 0x80)){
int i;
int j;
p = mime_pattern[0];
- for(i=0;mime_encode[i];i++) {
+ for(i=0;mime_pattern[i];i++) {
if (mode == mime_encode[i]) {
p = mime_pattern[i];
break;
if (mimeout_mode=='Q') {
if (c <= DEL && (output_mode==ASCII ||output_mode == ISO8859_1 ) ) {
- if (c <= SPACE) {
+ if (c == CR || c == NL) {
+ close_mime();
+ (*o_mputc)(c);
+ base64_count = 0;
+ return;
+ } else if (c <= SPACE) {
close_mime();
- (*o_mputc)(SPACE);
- base64_count++;
+ if (base64_count > 70) {
+ (*o_mputc)(NL);
+ base64_count = 0;
+ }
+ if (!nkf_isblank(c)) {
+ (*o_mputc)(SPACE);
+ base64_count++;
+ }
}
(*o_mputc)(c);
base64_count++;
mimeout_buf_count = 1;
}else{
if (base64_count > 1
- && base64_count + mimeout_buf_count > 76){
+ && base64_count + mimeout_buf_count > 76
+ && mimeout_buf[0] != CR && mimeout_buf[0] != NL){
(*o_mputc)(NL);
base64_count = 0;
if (!nkf_isspace(mimeout_buf[0])){