** E-Mail: furukawa@tcp-ip.or.jp
** \e$B$^$G8fO"Mm$r$*4j$$$7$^$9!#\e(B
***********************************************************************/
-/* $Id: nkf.c,v 1.101 2006/06/11 09:00:51 naruse Exp $ */
-#define NKF_VERSION "2.0.7"
-#define NKF_RELEASE_DATE "2006-05-06"
+/* $Id: nkf.c,v 1.124 2007/05/28 15:12:13 naruse Exp $ */
+#define NKF_VERSION "2.0.8"
+#define NKF_RELEASE_DATE "2007-05-29"
#include "config.h"
#include "utf8tbl.h"
#define COPY_RIGHT \
"Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW\n" \
- " 2002-2006 Kono, Furukawa, Naruse, mastodon"
+ "Copyright (C) 2002-2006 Kono, Furukawa, Naruse, mastodon"
/*
**
** t no operation
**
-** j Outout code is JIS 7 bit (DEFAULT SELECT)
+** j Output code is JIS 7 bit (DEFAULT SELECT)
** s Output code is MS Kanji (DEFAULT SELECT)
** e Output code is AT&T JIS (DEFAULT SELECT)
** w Output code is AT&T JIS (DEFAULT SELECT)
/* Input Assumption */
#define JIS_INPUT 4
+#define EUC_INPUT 16
#define SJIS_INPUT 5
#define LATIN1_INPUT 6
#define FIXED_MIME 7
#define UTF8 12
#define UTF8_INPUT 13
-#define UTF16BE_INPUT 14
-#define UTF16LE_INPUT 15
+#define UTF16_INPUT 1015
+#define UTF32_INPUT 1017
+
+/* byte order */
+
+#define ENDIAN_BIG 1234
+#define ENDIAN_LITTLE 4321
+#define ENDIAN_2143 2143
+#define ENDIAN_3412 3412
#define WISH_TRUE 15
#define nkf_isspace(c) (nkf_isblank(c) || c == CR || c == NL)
#define nkf_isalpha(c) (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z'))
#define nkf_isalnum(c) (nkf_isdigit(c) || nkf_isalpha(c))
-#define hex2bin(x) ( nkf_isdigit(x) ? x - '0' : nkf_toupper(x) - 'A' + 10)
+#define nkf_isprint(c) (' '<=c && c<='~')
+#define nkf_isgraph(c) ('!'<=c && c<='~')
+#define hex2bin(c) (('0'<=c&&c<='9') ? (c-'0') : \
+ ('A'<=c&&c<='F') ? (c-'A'+10) : \
+ ('a'<=c&&c<='f') ? (c-'a'+10) : 0 )
+#define is_eucg3(c2) (((unsigned short)c2 >> 8) == SS3)
+
+#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
#if defined(INT_IS_SHORT)
* 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
/* no NEC special, NEC-selected IBM extended and IBM extended characters */
static int no_cp932ext_f = FALSE;
/* ignore ZERO WIDTH NO-BREAK SPACE */
-static int ignore_zwnbsp_f = TRUE;
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 nkf_char w2e_conv(nkf_char c2,nkf_char c1,nkf_char c0,nkf_char *p2,nkf_char *p1);
static nkf_char w_iconv(nkf_char c2,nkf_char c1,nkf_char c0);
static nkf_char w_iconv16(nkf_char c2,nkf_char c1,nkf_char c0);
+static nkf_char w_iconv32(nkf_char c2,nkf_char c1,nkf_char c0);
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 unicode_bom_f= 0; /* Output Unicode BOM */
-static int w_oconv16_LE = 0; /* utf-16 little endian */
+static int output_bom_f = FALSE;
+static int output_endian = ENDIAN_BIG;
static nkf_char e2w_conv(nkf_char c2,nkf_char c1);
static void w_oconv(nkf_char c2,nkf_char c1);
static void w_oconv16(nkf_char c2,nkf_char c1);
+static void w_oconv32(nkf_char c2,nkf_char c1);
#endif
static void e_oconv(nkf_char c2,nkf_char c1);
static nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1);
static unsigned char stdobuf[IOBUF_SIZE];
#endif
static unsigned char hold_buf[HOLD_SIZE*2];
-static int hold_count;
+static int hold_count = 0;
/* MIME preprocessor fifo */
static nkf_char url_ungetc(nkf_char c,FILE *f);
#endif
-#ifdef NUMCHAR_OPTION
-#ifdef INT_IS_SHORT
-#define CLASS_MASK 0x0f000000L
-#define CLASS_UTF16 0x01000000L
+#if defined(INT_IS_SHORT)
+#define NKF_INT32_C(n) (n##L)
#else
-#define CLASS_MASK 0x0f000000
-#define CLASS_UTF16 0x01000000
+#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)
+#define VALUE_MASK NKF_INT32_C(0x00FFFFFF)
+#define UNICODE_MAX NKF_INT32_C(0x0010FFFF)
+#define is_unicode_capsule(c) ((c & CLASS_MASK) == CLASS_UNICODE)
+#define is_unicode_bmp(c) ((c & VALUE_MASK) <= NKF_INT32_C(0xFFFF))
+
+#ifdef NUMCHAR_OPTION
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;
#ifdef SHIFTJIS_CP932
/* invert IBM extended characters to others */
-static int cp51932_f = TRUE;
-#define CP932_TABLE_BEGIN (0xfa)
-#define CP932_TABLE_END (0xfc)
+static int cp51932_f = FALSE;
/* invert NEC-selected IBM extended characters to IBM extended characters */
static int cp932inv_f = TRUE;
-#define CP932INV_TABLE_BEGIN (0xed)
-#define CP932INV_TABLE_END (0xee)
/* static nkf_char cp932_conv(nkf_char c2, nkf_char c1); */
#endif /* SHIFTJIS_CP932 */
static void e_status(struct input_code *, nkf_char);
static void s_status(struct input_code *, nkf_char);
-#ifdef UTF8_INPUT_ENABLE
-static void w_status(struct input_code *, nkf_char);
-static void w16_status(struct input_code *, nkf_char);
-static int utf16_mode = UTF16BE_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},
#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}, w16_status, w_iconv16, 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}
};
}
} 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 = JIS_INPUT;
+ input_f = EUC_INPUT;
}else if(strcmp(codeset, "CP51932") == 0){
- input_f = JIS_INPUT;
- x0201_f = FALSE;
+ input_f = EUC_INPUT;
#ifdef SHIFTJIS_CP932
cp51932_f = TRUE;
#endif
}else if(strcmp(codeset, "EUC-JP-MS") == 0 ||
strcmp(codeset, "EUCJP-MS") == 0 ||
strcmp(codeset, "EUCJPMS") == 0){
- input_f = JIS_INPUT;
- x0201_f = FALSE;
+ input_f = EUC_INPUT;
#ifdef SHIFTJIS_CP932
cp51932_f = FALSE;
#endif
#endif
}else if(strcmp(codeset, "EUC-JP-ASCII") == 0 ||
strcmp(codeset, "EUCJP-ASCII") == 0){
- input_f = JIS_INPUT;
- x0201_f = FALSE;
+ input_f = EUC_INPUT;
#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 = JIS_INPUT;
- x0201_f = FALSE;
+ input_f = EUC_INPUT;
x0213_f = TRUE;
#ifdef SHIFTJIS_CP932
cp51932_f = FALSE;
- cp932inv_f = FALSE;
#endif
#ifdef UTF8_INPUT_ENABLE
}else if(strcmp(codeset, "UTF-8") == 0 ||
input_f = UTF8_INPUT;
nfc_f = TRUE;
#endif
- }else if(strcmp(codeset, "UTF-16") == 0){
- input_f = UTF16BE_INPUT;
- utf16_mode = UTF16BE_INPUT;
- }else if(strcmp(codeset, "UTF-16BE") == 0 ||
+ }else if(strcmp(codeset, "UTF-16") == 0 ||
+ strcmp(codeset, "UTF-16BE") == 0 ||
strcmp(codeset, "UTF-16BE-BOM") == 0){
- input_f = UTF16BE_INPUT;
- utf16_mode = UTF16BE_INPUT;
+ input_f = UTF16_INPUT;
+ input_endian = ENDIAN_BIG;
}else if(strcmp(codeset, "UTF-16LE") == 0 ||
strcmp(codeset, "UTF-16LE-BOM") == 0){
- input_f = UTF16LE_INPUT;
- utf16_mode = UTF16LE_INPUT;
+ input_f = UTF16_INPUT;
+ input_endian = ENDIAN_LITTLE;
+ }else if(strcmp(codeset, "UTF-32") == 0 ||
+ strcmp(codeset, "UTF-32BE") == 0 ||
+ strcmp(codeset, "UTF-32BE-BOM") == 0){
+ input_f = UTF32_INPUT;
+ input_endian = ENDIAN_BIG;
+ }else if(strcmp(codeset, "UTF-32LE") == 0 ||
+ strcmp(codeset, "UTF-32LE-BOM") == 0){
+ input_f = UTF32_INPUT;
+ input_endian = ENDIAN_LITTLE;
#endif
}
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){
output_conv = w_oconv;
}else if(strcmp(codeset, "UTF-8N") == 0){
output_conv = w_oconv;
- unicode_bom_f=1;
}else if(strcmp(codeset, "UTF-8-BOM") == 0){
output_conv = w_oconv;
- unicode_bom_f=2;
+ output_bom_f = TRUE;
}else if(strcmp(codeset, "UTF-16BE") == 0){
- output_conv = w_oconv16;
- unicode_bom_f=1;
+ output_conv = w_oconv16;
}else if(strcmp(codeset, "UTF-16") == 0 ||
strcmp(codeset, "UTF-16BE-BOM") == 0){
- output_conv = w_oconv16;
- unicode_bom_f=2;
+ output_conv = w_oconv16;
+ output_bom_f = TRUE;
}else if(strcmp(codeset, "UTF-16LE") == 0){
- output_conv = w_oconv16;
- w_oconv16_LE = 1;
- unicode_bom_f=1;
+ output_conv = w_oconv16;
+ output_endian = ENDIAN_LITTLE;
}else if(strcmp(codeset, "UTF-16LE-BOM") == 0){
- output_conv = w_oconv16;
- w_oconv16_LE = 1;
- unicode_bom_f=2;
+ output_conv = w_oconv16;
+ output_endian = ENDIAN_LITTLE;
+ output_bom_f = TRUE;
+ }else if(strcmp(codeset, "UTF-32") == 0 ||
+ strcmp(codeset, "UTF-32BE") == 0){
+ output_conv = w_oconv32;
+ }else if(strcmp(codeset, "UTF-32BE-BOM") == 0){
+ output_conv = w_oconv32;
+ output_bom_f = TRUE;
+ }else if(strcmp(codeset, "UTF-32LE") == 0){
+ output_conv = w_oconv32;
+ output_endian = ENDIAN_LITTLE;
+ }else if(strcmp(codeset, "UTF-32LE-BOM") == 0){
+ output_conv = w_oconv32;
+ output_endian = ENDIAN_LITTLE;
+ output_bom_f = TRUE;
#endif
}
continue;
}
#endif
if (strcmp(long_option[i].name, "prefix=") == 0){
- if (' ' < p[0] && p[0] < 128){
- for (i = 1; ' ' < p[i] && p[i] < 128; i++){
+ if (nkf_isgraph(p[0])){
+ for (i = 1; nkf_isgraph(p[i]); i++){
prefix_table[p[i]] = p[0];
}
}
continue;
case 'e': /* AT&T EUC output */
output_conv = e_oconv;
+ cp932inv_f = FALSE;
continue;
case 's': /* SJIS output */
output_conv = s_oconv;
#endif
#ifdef UTF8_OUTPUT_ENABLE
case 'w': /* UTF-8 output */
- if ('1'== cp[0] && '6'==cp[1]) {
- output_conv = w_oconv16; cp+=2;
+ if (cp[0] == '8') {
+ output_conv = w_oconv; cp++;
+ if (cp[0] == '0'){
+ cp++;
+ } else {
+ output_bom_f = TRUE;
+ }
+ } else {
+ if ('1'== cp[0] && '6'==cp[1]) {
+ output_conv = w_oconv16; cp+=2;
+ } else if ('3'== cp[0] && '2'==cp[1]) {
+ output_conv = w_oconv32; cp+=2;
+ } else {
+ output_conv = w_oconv;
+ continue;
+ }
if (cp[0]=='L') {
- unicode_bom_f=2; cp++;
- w_oconv16_LE = 1;
- if (cp[0] == '0'){
- unicode_bom_f=1; cp++;
- }
+ cp++;
+ output_endian = ENDIAN_LITTLE;
} else if (cp[0] == 'B') {
- unicode_bom_f=2; cp++;
- if (cp[0] == '0'){
- unicode_bom_f=1; cp++;
- }
- }
- } else if (cp[0] == '8') {
- output_conv = w_oconv; cp++;
- unicode_bom_f=2;
+ cp++;
+ } else {
+ continue;
+ }
if (cp[0] == '0'){
- unicode_bom_f=1; cp++;
+ cp++;
+ } else {
+ output_bom_f = TRUE;
}
- } else
- output_conv = w_oconv;
+ }
continue;
#endif
#ifdef UTF8_INPUT_ENABLE
- case 'W': /* UTF-8 input */
- if ('1'== cp[0] && '6'==cp[1]) {
- input_f = UTF16BE_INPUT;
- utf16_mode = UTF16BE_INPUT;
- cp += 2;
+ case 'W': /* UTF input */
+ if (cp[0] == '8') {
+ cp++;
+ input_f = UTF8_INPUT;
+ }else{
+ if ('1'== cp[0] && '6'==cp[1]) {
+ cp += 2;
+ input_f = UTF16_INPUT;
+ input_endian = ENDIAN_BIG;
+ } else if ('3'== cp[0] && '2'==cp[1]) {
+ cp += 2;
+ input_f = UTF32_INPUT;
+ input_endian = ENDIAN_BIG;
+ } else {
+ input_f = UTF8_INPUT;
+ continue;
+ }
if (cp[0]=='L') {
cp++;
- input_f = UTF16LE_INPUT;
- utf16_mode = UTF16LE_INPUT;
+ input_endian = ENDIAN_LITTLE;
} else if (cp[0] == 'B') {
cp++;
- input_f = UTF16BE_INPUT;
- utf16_mode = UTF16BE_INPUT;
}
- } else if (cp[0] == '8') {
- cp++;
- input_f = UTF8_INPUT;
- } else
- input_f = UTF8_INPUT;
+ }
continue;
#endif
/* Input code assumption */
case 'J': /* JIS input */
- case 'E': /* AT&T EUC input */
input_f = JIS_INPUT;
continue;
+ case 'E': /* AT&T EUC input */
+ input_f = EUC_INPUT;
+ continue;
case 'S': /* MS Kanji input */
input_f = SJIS_INPUT;
if (x0201_f==NO_X0201) x0201_f=TRUE;
}
}
-#ifdef ANSI_C_PROTOTYPE
struct input_code * find_inputcode_byfunc(nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0))
-#else
-struct input_code * find_inputcode_byfunc(iconv_func)
- nkf_char (*iconv_func)();
-#endif
{
if (iconv_func){
struct input_code *p = input_code_list;
if (c <= DEL){
break;
#ifdef NUMCHAR_OPTION
- }else if ((c & CLASS_MASK) == CLASS_UTF16){
+ }else if (is_unicode_capsule(c)){
break;
#endif
}else if (0xa1 <= c && c <= 0xdf){
status_push_ch(ptr, c);
#ifdef SHIFTJIS_CP932
}else if (cp51932_f
- && CP932_TABLE_BEGIN <= c && c <= CP932_TABLE_END){
+ && is_ibmext_in_sjis(c)){
ptr->stat = 2;
status_push_ch(ptr, c);
#endif /* SHIFTJIS_CP932 */
if (c <= DEL){
break;
#ifdef NUMCHAR_OPTION
- }else if ((c & CLASS_MASK) == CLASS_UTF16){
+ }else if (is_unicode_capsule(c)){
break;
#endif
}else if (SSO == c || (0xa1 <= c && c <= 0xfe)){
}
#ifdef UTF8_INPUT_ENABLE
-void w16_status(struct input_code *ptr, nkf_char c)
-{
- switch (ptr->stat){
- case -1:
- break;
- case 0:
- if (ptr->_file_stat == 0){
- if (c == 0xfe || c == 0xff){
- ptr->stat = c;
- status_push_ch(ptr, c);
- ptr->_file_stat = 1;
- }else{
- status_disable(ptr);
- ptr->_file_stat = -1;
- }
- }else if (ptr->_file_stat > 0){
- ptr->stat = 1;
- status_push_ch(ptr, c);
- }else if (ptr->_file_stat < 0){
- status_disable(ptr);
- }
- break;
-
- case 1:
- if (c == EOF){
- status_disable(ptr);
- ptr->_file_stat = -1;
- }else{
- status_push_ch(ptr, c);
- status_clear(ptr);
- }
- break;
-
- case 0xfe:
- case 0xff:
- if (ptr->stat != c && (c == 0xfe || c == 0xff)){
- status_push_ch(ptr, c);
- status_clear(ptr);
- }else{
- status_disable(ptr);
- ptr->_file_stat = -1;
- }
- break;
- }
-}
-
void w_status(struct input_code *ptr, nkf_char c)
{
switch (ptr->stat){
if (c <= DEL){
break;
#ifdef NUMCHAR_OPTION
- }else if ((c & CLASS_MASK) == CLASS_UTF16){
+ }else if (is_unicode_capsule(c)){
break;
#endif
}else if (0xc0 <= c && c <= 0xdf){
}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);
}
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
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;
i_bgetc = i_getc; i_getc = broken_getc;
i_bungetc = i_ungetc; i_ungetc = broken_ungetc;
}
- if (input_f == JIS_INPUT || input_f == LATIN1_INPUT) {
+ if (input_f == JIS_INPUT || input_f == EUC_INPUT || input_f == LATIN1_INPUT) {
set_iconv(-TRUE, e_iconv);
} else if (input_f == SJIS_INPUT) {
set_iconv(-TRUE, s_iconv);
#ifdef UTF8_INPUT_ENABLE
} else if (input_f == UTF8_INPUT) {
set_iconv(-TRUE, w_iconv);
- } else if (input_f == UTF16BE_INPUT) {
- set_iconv(-TRUE, w_iconv16);
- } else if (input_f == UTF16LE_INPUT) {
+ } else if (input_f == UTF16_INPUT) {
set_iconv(-TRUE, w_iconv16);
+ } else if (input_f == UTF32_INPUT) {
+ set_iconv(-TRUE, w_iconv32);
#endif
} else {
set_iconv(FALSE, e_iconv);
}
/*
+ * 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_f){
+ 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_f){
+ 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_f){
+ 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_f){
+ 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_f){
+ 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_f){
+ 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_f){
+ 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;
+ }
+}
+
+/*
Conversion main loop. Code detection only.
*/
nkf_char kanji_convert(FILE *f)
{
- nkf_char c1,
- c2, c3;
+ nkf_char c3, c2=0, c1, c0=0;
int is_8bit = FALSE;
- module_connection();
- c2 = 0;
-
- if(input_f == SJIS_INPUT
+ if(input_f == SJIS_INPUT || input_f == EUC_INPUT
#ifdef UTF8_INPUT_ENABLE
- || input_f == UTF8_INPUT || input_f == UTF16BE_INPUT || input_f == UTF16LE_INPUT
+ || input_f == UTF8_INPUT || input_f == UTF16_INPUT
#endif
){
is_8bit = TRUE;
}
-
input_mode = ASCII;
output_mode = ASCII;
shift_mode = FALSE;
#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_f)
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 */
SEND;
} else {
/* first byte */
- if (
#ifdef UTF8_INPUT_ENABLE
- iconv == w_iconv16
-#else
- 0
-#endif
- ) {
- c2 = c1;
- c1 = (*i_getc)(f);
+ 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
- } else if ((c1 & CLASS_MASK) == CLASS_UTF16){
+ if (is_unicode_capsule(c1)){
SEND;
+ } else
#endif
- } else 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 */
- if (!is_8bit) is_8bit = TRUE;
c2 = c1;
NEXT;
} else { /* estab_f==TRUE */
} 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 == CR && c1 == NL) crmode_f = CRLF;
+ else crmode_f = c1;
+ } else if (c1 == DEL && input_mode == X0208 ) {
+ /* CP5022x */
+ c2 = c1;
+ NEXT;
} else
SEND;
}
/* send: */
switch(input_mode){
case ASCII:
- if ((*iconv)(c2, c1, 0) < 0){ /* can be EUC/SJIS */
- nkf_char c0 = (*i_getc)(f);
- if (c0 != EOF){
+ 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 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
case X0212:
-#if defined(INT_IS_SHORT)
- (*oconv)(0x8f00L | c2, c1);
-#else
- (*oconv)((0x8f << 8) | c2, c1);
-#endif
+ (*oconv)(PREFIX_EUCG3 | c2, c1);
break;
#endif /* X0212_ENABLE */
case X0213_2:
-#if defined(INT_IS_SHORT)
- (*oconv)(0x8f00L | c2, c1);
-#else
- (*oconv)((0x8f << 8) | c2, c1);
-#endif
+ (*oconv)(PREFIX_EUCG3 | c2, c1);
break;
default:
(*oconv)(input_mode, c1); /* other special case */
}
c2 = 0;
+ c0 = 0;
continue;
/* goto next_word */
}
nkf_char
h_conv(FILE *f, nkf_char c2, nkf_char c1)
{
- nkf_char wc,c3;
+ nkf_char ret, c3, c0;
+ int hold_index;
/** it must NOT be in the kanji shifte sequence */
** Kanji codes by oconv and leave estab_f unchanged.
**/
- c3=c1;
- wc = 0;
- while (wc < hold_count){
- c2 = hold_buf[wc++];
+ ret = c1;
+ hold_index = 0;
+ while (hold_index < hold_count){
+ c2 = hold_buf[hold_index++];
if (c2 <= DEL
#ifdef NUMCHAR_OPTION
- || (c2 & CLASS_MASK) == CLASS_UTF16
+ || is_unicode_capsule(c2)
#endif
){
(*iconv)(0, c2, 0);
(*iconv)(X0201, c2, 0);
continue;
}
- if (wc < hold_count){
- c1 = hold_buf[wc++];
+ if (hold_index < hold_count){
+ c1 = hold_buf[hold_index++];
}else{
c1 = (*i_getc)(f);
if (c1 == EOF){
}
code_status(c1);
}
- if ((*iconv)(c2, c1, 0) < 0){
- nkf_char c0;
- if (wc < hold_count){
- c0 = hold_buf[wc++];
- }else{
- c0 = (*i_getc)(f);
- if (c0 == EOF){
- c3 = EOF;
- break;
- }
+ 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 c3;
+ return ret;
}
-
-
-nkf_char
-push_hold_buf(nkf_char c2)
+nkf_char push_hold_buf(nkf_char c2)
{
if (hold_count >= HOLD_SIZE*2)
return (EOF);
#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 && CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){
-#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 && 0xfa <= c2 && c2 <= 0xfc){
-#if 0
- extern const unsigned short shiftjis_x0212[3][189];
-#endif
+ if (!x0213_f && is_ibmext_in_sjis(c2)){
val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40];
if (val){
-#if defined(INT_IS_SHORT)
- if (val & 0x8000L){
-#else
- if (val & 0x8000){
-#endif
- c2 = (0x8f << 8) | (val >> 8);
+ if (val > 0x7FFF){
+ c2 = PREFIX_EUCG3 | ((val >> 8) & 0x7f);
c1 = val & 0xff;
}else{
c2 = val >> 8;
if(c2 >= 0x80){
if(x0213_f && c2 >= 0xF0){
if(c2 <= 0xF3 || (c2 == 0xF4 && c1 < 0x9F)){ /* k=1, 3<=k<=5, k=8, 12<=k<=15 */
-#if defined(INT_IS_SHORT)
- c2 = 0x8F20L + shift_jisx0213_s1a3_table[c2 - 0xF0][0x9E < c1];
-#else
- c2 = 0x8F20 + shift_jisx0213_s1a3_table[c2 - 0xF0][0x9E < c1];
-#endif
+ c2 = PREFIX_EUCG3 | 0x20 | shift_jisx0213_s1a3_table[c2 - 0xF0][0x9E < c1];
}else{ /* 78<=k<=94 */
-#if defined(INT_IS_SHORT)
- c2 = 0x8F00L | (c2 * 2 - 0x17B);
-#else
- c2 = 0x8F00 | (c2 * 2 - 0x17B);
-#endif
+ c2 = PREFIX_EUCG3 | (c2 * 2 - 0x17B);
if (0x9E < c1) c2++;
}
}else{
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 defined(INT_IS_SHORT)
- if ((c2 & 0xff00L) == 0){
-#else
- if ((c2 & 0xff00) == 0){
-#endif
- 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;
#ifdef NUMCHAR_OPTION
if (ret > 0){
if (p2) *p2 = 0;
- if (p1) *p1 = CLASS_UTF16 | ww16_conv(c2, c1, c0);
+ if (p1) *p1 = CLASS_UNICODE | ww16_conv(c2, c1, c0);
ret = 0;
}
#endif
nkf_char w_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
{
nkf_char ret = 0;
+ static const int 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};
- /* throw away ZERO WIDTH NO-BREAK SPACE (U+FEFF) */
- if(ignore_zwnbsp_f){
- ignore_zwnbsp_f = FALSE;
- if(c2 == 0xef && c1 == 0xbb && c0 == 0xbf)
- return 0;
- }
-
- if (c2 == 0) /* 0x00-0x7f */
- c1 &= 0x7F; /* 1byte */
- else if (c0 == 0){
- if ((c2 & 0xe0) == 0xc0){ /* 0xc0-0xdf */
- /* 2ytes */
- if((c2 & 0xFE) == 0xC0 || c1 < 0x80 || 0xBF < c1) return 0;
- }else if ((c2 & 0xf0) == 0xe0) /* 0xe0-0xef */
- return -1; /* 3bytes */
-#ifdef __COMMENT__
- else if (0xf0 <= c2)
- return 0; /* 4,5,6bytes */
- else if ((c2 & 0xc0) == 0x80) /* 0x80-0xbf */
- return 0; /* trail byte */
-#endif
- else return 0;
- }else{
- /* must be 3bytes */
- if(c2 == 0xE0){
- if(c1 < 0xA0 || 0xBF < c1 || c0 < 0x80 || 0xBF < c0)
+ 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;
- }else if(c2 == 0xED){
- if(c1 < 0x80 || 0x9F < c1 || c0 < 0x80 || 0xBF < c0)
+ 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;
- }else if((c2 & 0xf0) == 0xe0){
- if(c1 < 0x80 || 0xBF < c1 || c0 < 0x80 || 0xBF < c0)
+ break;
+ case 42:
+ if (c0 == 0) return -2;
+ if (c1 < 0x80 || 0x8F < c1 || (c0 & 0xc0c0) != 0x8080)
return 0;
- }else 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 defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0)
{
-#if defined(INT_IS_SHORT)
- val &= 0xFFFFFFL;
-#else
- val &= 0xFFFFFF;
-#endif
+ val &= VALUE_MASK;
if (val < 0x80){
*p2 = val;
*p1 = 0;
*p2 = 0xc0 | (val >> 6);
*p1 = 0x80 | (val & 0x3f);
*p0 = 0;
- }else{
+ } else if (val <= NKF_INT32_C(0xFFFF)) {
*p2 = 0xe0 | (val >> 12);
*p1 = 0x80 | ((val >> 6) & 0x3f);
*p0 = 0x80 | (val & 0x3f);
+ } else if (val <= NKF_INT32_C(0x10FFFF)) {
+ *p2 = 0xe0 | (val >> 16);
+ *p1 = 0x80 | ((val >> 12) & 0x3f);
+ *p0 = 0x8080 | ((val << 2) & 0x3f00)| (val & 0x3f);
+ } else {
+ *p2 = 0;
+ *p1 = 0;
+ *p0 = 0;
}
}
#endif
nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0)
{
nkf_char val;
- if (c2 >= 0xf0){
+ if (c2 >= 0xf8) {
val = -1;
+ } else if (c2 >= 0xf0){
+ /* c2: 1st, c1: 2nd, c0: 3rd/4th */
+ val = (c2 & 0x0f) << 18;
+ val |= (c1 & 0x3f) << 12;
+ val |= (c0 & 0x3f00) >> 2;
+ val |= (c0 & 0x3f);
}else if (c2 >= 0xe0){
val = (c2 & 0x0f) << 12;
val |= (c1 & 0x3f) << 6;
{
nkf_char c2, c1, c0;
nkf_char ret = 0;
-#if defined(INT_IS_SHORT)
- val &= 0xFFFFFFL;
-#else
- val &= 0xFFFFFF;
-#endif
+ val &= VALUE_MASK;
if (val < 0x80){
*p2 = 0;
*p1 = val;
#ifdef NUMCHAR_OPTION
if (ret > 0){
*p2 = 0;
- *p1 = CLASS_UTF16 | val;
+ *p1 = CLASS_UNICODE | val;
ret = 0;
}
#endif
#ifdef UTF8_INPUT_ENABLE
nkf_char w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0)
{
- nkf_char ret;
-
- /* throw away ZERO WIDTH NO-BREAK SPACE (U+FEFF) */
- if(ignore_zwnbsp_f){
- ignore_zwnbsp_f = FALSE;
- if (c2==0376 && c1==0377){
- utf16_mode = UTF16BE_INPUT;
- return 0;
- }else if(c2==0377 && c1==0376){
- utf16_mode = UTF16LE_INPUT;
- return 0;
- }
- }
- if (c2 != EOF && utf16_mode == UTF16LE_INPUT) {
- nkf_char tmp;
- tmp=c1; c1=c2; c2=tmp;
- }
+ nkf_char ret = 0;
if ((c2==0 && c1 < 0x80) || c2==EOF) {
(*oconv)(c2, c1);
return 0;
- }else if((c2>>3)==27){ /* surrogate pair */
+ }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;
-#if defined(INT_IS_SHORT)
}else ret = w16e_conv(((c2 & 0xff)<<8) + c1, &c2, &c1);
-#else
- }else ret = w16e_conv(((c2<<8)&0xff00) + c1, &c2, &c1);
-#endif
+ 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;
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){
+ }else if(c0 < 0xF0){
if(no_best_fit_chars_f){
if(ms_ucs_map_f == UCS_MAP_CP932){
if(c2 == 0xE3 && c1 == 0x82 && c0 == 0x94) return 1;
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 defined(INT_IS_SHORT)
- if (!ret && cp51932_f && ((unsigned short)(*p2) >> 8) == 0x8f) {
-#else
- if (!ret && cp51932_f && (*p2 & 0xff00) >> 8 == 0x8f) {
-#endif
+ 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);
if (val == 0) return 1;
if (no_cp932ext_f && (
(val>>8) == 0x2D || /* NEC special characters */
-#if defined(INT_IS_SHORT)
- val > 0xF300L /* NEC special characters */
-#else
- val > 0xF300 /* NEC special characters */
-#endif
+ val > NKF_INT32_C(0xF300) /* IBM extended characters */
)) return 1;
c2 = val >> 8;
-#if defined(INT_IS_SHORT)
- if (val & 0x8000L){
- c2 &= 0x7f;
- c2 |= 0x8f00L;
- }
-#else
- if (val & 0x8000){
+ if (val > 0x7FFF){
c2 &= 0x7f;
- c2 |= 0x8f00;
+ c2 |= PREFIX_EUCG3;
}
-#endif
if (c2 == SO) c2 = X0201;
c1 = val & 0x7f;
if (p2) *p2 = c2;
{
const char *hex = "0123456789ABCDEF";
int shift = 20;
-#if defined(INT_IS_SHORT)
- c &= 0x00FFFFFFL;
-#else
- c &= 0x00FFFFFF;
-#endif
+ c &= VALUE_MASK;
while(shift >= 0){
if(c >= 1<<shift){
while(shift >= 0){
{
(*oconv)(0, '&');
(*oconv)(0, '#');
-#if defined(INT_IS_SHORT)
- c &= 0x00FFFFFFL;
- if(c >= 1000000L)
- (*oconv)(0, 0x30+(c/1000000L)%10);
- if(c >= 100000L)
- (*oconv)(0, 0x30+(c/100000L )%10);
-#else
- c &= 0x00FFFFFF;
- if(c >= 1000000)
- (*oconv)(0, 0x30+(c/1000000)%10);
- if(c >= 100000)
- (*oconv)(0, 0x30+(c/100000 )%10);
-#endif
+ 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)
{
const char *hex = "0123456789ABCDEF";
(*oconv)(0, '\\');
-#if defined(INT_IS_SHORT)
- if((c&0x00FFFFFFL) > 0xFFFFL){
-#else
- if((c&0x00FFFFFF) > 0xFFFF){
-#endif
+ c &= VALUE_MASK;
+ if(!is_unicode_bmp(c)){
(*oconv)(0, 'U');
(*oconv)(0, '0');
(*oconv)(0, '0');
#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 (c2 >> 8 == 0x8f){
-#if defined(INT_IS_SHORT)
- if(ms_ucs_map_f == UCS_MAP_ASCII&& c2 == 0x8F22L && c1 == 0x43){
-#else
- if(ms_ucs_map_f == UCS_MAP_ASCII&& c2 == 0x8F22 && c1 == 0x43){
-#endif
+ } else if (is_eucg3(c2)){
+ if(ms_ucs_map_f == UCS_MAP_ASCII&& c2 == NKF_INT32_C(0x8F22) && c1 == 0x43){
return 0xA6;
}
c2 = (c2&0x7f) - 0x21;
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 c0;
nkf_char val;
- if (c2 == EOF) {
- (*o_putc)(EOF);
- return;
- }
- if (unicode_bom_f==2) {
- (*o_putc)('\357');
+ if (output_bom_f) {
+ output_bom_f = FALSE;
+ (*o_putc)('\357');
(*o_putc)('\273');
(*o_putc)('\277');
- unicode_bom_f=1;
+ }
+
+ if (c2 == EOF) {
+ (*o_putc)(EOF);
+ return;
}
#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);
+ if (c2 == 0 && is_unicode_capsule(c1)){
+ val = c1 & VALUE_MASK;
+ if (val < 0x80){
+ (*o_putc)(val);
+ }else if (val < 0x800){
+ (*o_putc)(0xC0 | (val >> 6));
+ (*o_putc)(0x80 | (val & 0x3f));
+ } else if (val <= NKF_INT32_C(0xFFFF)) {
+ (*o_putc)(0xE0 | (val >> 12));
+ (*o_putc)(0x80 | ((val >> 6) & 0x3f));
+ (*o_putc)(0x80 | (val & 0x3f));
+ } else if (val <= NKF_INT32_C(0x10FFFF)) {
+ (*o_putc)(0xF0 | ( val>>18));
+ (*o_putc)(0x80 | ((val>>12) & 0x3f));
+ (*o_putc)(0x80 | ((val>> 6) & 0x3f));
+ (*o_putc)(0x80 | ( val & 0x3f));
}
return;
}
void w_oconv16(nkf_char c2, nkf_char c1)
{
- if (c2 == EOF) {
- (*o_putc)(EOF);
- return;
- }
-
- if (unicode_bom_f==2) {
- if (w_oconv16_LE){
+ if (output_bom_f) {
+ output_bom_f = FALSE;
+ if (output_endian == ENDIAN_LITTLE){
(*o_putc)((unsigned char)'\377');
(*o_putc)('\376');
}else{
(*o_putc)('\376');
(*o_putc)((unsigned char)'\377');
}
- unicode_bom_f=1;
+ }
+
+ if (c2 == EOF) {
+ (*o_putc)(EOF);
+ return;
}
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;
+ } else if (c2 == 0 && is_unicode_capsule(c1)) {
+ if (is_unicode_bmp(c1)) {
+ c2 = (c1 >> 8) & 0xff;
+ c1 &= 0xff;
+ } else {
+ c1 &= VALUE_MASK;
+ if (c1 <= UNICODE_MAX) {
+ c2 = (c1 >> 10) + NKF_INT32_C(0xD7C0); /* high surrogate */
+ c1 = (c1 & 0x3FF) + NKF_INT32_C(0xDC00); /* low surrogate */
+ if (output_endian == ENDIAN_LITTLE){
+ (*o_putc)(c2 & 0xff);
+ (*o_putc)((c2 >> 8) & 0xff);
+ (*o_putc)(c1 & 0xff);
+ (*o_putc)((c1 >> 8) & 0xff);
+ }else{
+ (*o_putc)((c2 >> 8) & 0xff);
+ (*o_putc)(c2 & 0xff);
+ (*o_putc)((c1 >> 8) & 0xff);
+ (*o_putc)(c1 & 0xff);
+ }
+ }
+ return;
+ }
#endif
} else if (c2) {
nkf_char val = e2w_conv(c2, c1);
c2 = (val >> 8) & 0xff;
c1 = val & 0xff;
+ if (!val) return;
}
- if (w_oconv16_LE){
+ if (output_endian == ENDIAN_LITTLE){
(*o_putc)(c1);
(*o_putc)(c2);
}else{
}
}
+void w_oconv32(nkf_char c2, nkf_char c1)
+{
+ if (output_bom_f) {
+ output_bom_f = FALSE;
+ if (output_endian == ENDIAN_LITTLE){
+ (*o_putc)((unsigned char)'\377');
+ (*o_putc)('\376');
+ (*o_putc)('\000');
+ (*o_putc)('\000');
+ }else{
+ (*o_putc)('\000');
+ (*o_putc)('\000');
+ (*o_putc)('\376');
+ (*o_putc)((unsigned char)'\377');
+ }
+ }
+
+ if (c2 == EOF) {
+ (*o_putc)(EOF);
+ return;
+ }
+
+ if (c2 == ISO8859_1) {
+ c1 |= 0x80;
+#ifdef NUMCHAR_OPTION
+ } else if (c2 == 0 && is_unicode_capsule(c1)) {
+ c1 &= VALUE_MASK;
+#endif
+ } else if (c2) {
+ c1 = e2w_conv(c2, c1);
+ if (!c1) return;
+ }
+ if (output_endian == ENDIAN_LITTLE){
+ (*o_putc)( c1 & NKF_INT32_C(0x000000FF));
+ (*o_putc)((c1 & NKF_INT32_C(0x0000FF00)) >> 8);
+ (*o_putc)((c1 & NKF_INT32_C(0x00FF0000)) >> 16);
+ (*o_putc)('\000');
+ }else{
+ (*o_putc)('\000');
+ (*o_putc)((c1 & NKF_INT32_C(0x00FF0000)) >> 16);
+ (*o_putc)((c1 & NKF_INT32_C(0x0000FF00)) >> 8);
+ (*o_putc)( c1 & NKF_INT32_C(0x000000FF));
+ }
+}
#endif
void e_oconv(nkf_char c2, nkf_char c1)
{
#ifdef NUMCHAR_OPTION
- if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+ if (c2 == 0 && is_unicode_capsule(c1)){
w16e_conv(c1, &c2, &c1);
- if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
- if(encode_fallback)(*encode_fallback)(c1);
- return;
+ 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
output_mode = ISO8859_1;
(*o_putc)(c1 | 0x080);
#ifdef X0212_ENABLE
-#if defined(INT_IS_SHORT)
- } else if (((unsigned short)c2 >> 8) == 0x8f){
-#else
- } else if ((c2 & 0xff00) >> 8 == 0x8f){
-#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);
if (c2 == 0) {
output_mode = ASCII;
(*o_putc)(c1);
-#if defined(INT_IS_SHORT)
- }else if (((unsigned short)c2 >> 8) == 0x8f){
-#else
- }else if ((c2 & 0xff00) >> 8 == 0x8f){
-#endif
+ }else if (is_eucg3(c2)){
if (x0212_f){
(*o_putc)(0x8f);
(*o_putc)((c2 & 0x7f) | 0x080);
}
#endif
} else {
- if ((c1<0x21 || 0x7e<c1) ||
- (c2<0x21 || 0x7e<c2)) {
+ if (!nkf_isgraph(c1) || !nkf_isgraph(c2)) {
set_iconv(FALSE, 0);
return; /* too late to rescue this char */
}
{
nkf_char ret = c;
c &= 0x7f;
-#if defined(INT_IS_SHORT)
- if ((ret & 0xff00L) == 0x8f00L){
-#else
- if ((ret & 0xff00) == 0x8f00){
-#endif
+ if (is_eucg3(ret)){
if (0x75 <= c && c <= 0x7f){
ret = c + (0x109 - 0x75);
}
if (0x7f <= c && c <= 0x88){
ret = c + (0x75 - 0x7f);
}else if (0x89 <= c && c <= 0x92){
- ret = (0x8f << 8) | 0x80 | (c + (0x75 - 0x89));
+ ret = PREFIX_EUCG3 | 0x80 | (c + (0x75 - 0x89));
}
return ret;
}
nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
{
nkf_char ndx;
-#if defined(INT_IS_SHORT)
- if ((c2 & 0xff00L) == 0x8f00L){
-#else
- if ((c2 & 0xff00) == 0x8f00){
-#endif
- ndx = c2 & 0xff;
+ 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;
return 1;
}
#ifdef X0212_ENABLE
- else if(0x21 <= ndx && ndx <= 0x7e){
+ 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];
void s_oconv(nkf_char c2, nkf_char c1)
{
#ifdef NUMCHAR_OPTION
- if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+ if (c2 == 0 && is_unicode_capsule(c1)){
w16e_conv(c1, &c2, &c1);
- if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
- if(encode_fallback)(*encode_fallback)(c1);
- return;
- }
+ 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(encode_fallback)(*encode_fallback)(c1);
+ return;
+ }
+ }
}
#endif
if (c2 == EOF) {
output_mode = ISO8859_1;
(*o_putc)(c1 | 0x080);
#ifdef X0212_ENABLE
-#if defined(INT_IS_SHORT)
- } else if (((unsigned short)c2 >> 8) == 0x8f){
-#else
- } else if ((c2 & 0xff00) >> 8 == 0x8f){
-#endif
+ } else if (is_eucg3(c2)){
output_mode = SHIFT_JIS;
if (e2s_conv(c2, c1, &c2, &c1) == 0){
(*o_putc)(c2);
}
#endif
} else {
- if ((c1<0x20 || 0x7e<c1) ||
- (c2<0x20 || 0x7e<c2)) {
+ if (!nkf_isprint(c1) || !nkf_isprint(c2)) {
set_iconv(FALSE, 0);
return; /* too late to rescue this char */
}
#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;
void j_oconv(nkf_char c2, nkf_char c1)
{
#ifdef NUMCHAR_OPTION
- if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+ if (c2 == 0 && is_unicode_capsule(c1)){
w16e_conv(c1, &c2, &c1);
- if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
- if(encode_fallback)(*encode_fallback)(c1);
- return;
+ 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
}
(*o_putc)(EOF);
#ifdef X0212_ENABLE
-#if defined(INT_IS_SHORT)
- } else if (((unsigned short)c2 >> 8) == 0x8f){
-#else
- } else if ((c2 & 0xff00) >> 8 == 0x8f){
-#endif
+ } else if (is_eucg3(c2)){
if(x0213_f){
if(output_mode!=X0213_2){
output_mode = X0213_2;
}
(*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;
/* 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;
return;
} else if (c1 == 0x74 && (output_conv == w_oconv || output_conv == w_oconv16)) {
c2 = 0;
- c1 = CLASS_UTF16 | 0x3094;
+ c1 = CLASS_UNICODE | 0x3094;
(*o_hira_conv)(c2,c1);
return;
}
}
}
if (hira_f & 2) {
- if (c2 == 0 && c1 == (CLASS_UTF16 | 0x3094)) {
+ if (c2 == 0 && c1 == (CLASS_UNICODE | 0x3094)) {
c2 = 0x25;
c1 = 0x74;
} else if (c2 == 0x24 && 0x20 < c1 && c1 < 0x74) {
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*/
c = 0;
buf[++i] = (*g)(f);
if (buf[i] == 'x' || buf[i] == 'X'){
- for (j = 0; j < 5; j++){
+ for (j = 0; j < 7; j++){
buf[++i] = (*g)(f);
if (!nkf_isxdigit(buf[i])){
if (buf[i] != ';'){
c |= hex2bin(buf[i]);
}
}else{
- for (j = 0; j < 6; j++){
+ for (j = 0; j < 8; j++){
if (j){
buf[++i] = (*g)(f);
}
}
}
if (c != -1){
- return CLASS_UTF16 | c;
+ return CLASS_UNICODE | c;
}
while (i > 0){
(*u)(buf[i], f);
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)){
if ((c3 = (*i_mgetc)(f)) == EOF) return (EOF);
if (c2<=' ') return c2;
mime_decode_mode = 'Q'; /* still in MIME */
-#define hex(c) (('0'<=c&&c<='9')?(c-'0'):\
- ('A'<=c&&c<='F')?(c-'A'+10):('a'<=c&&c<='f')?(c-'a'+10):0)
- return ((hex(c2)<<4) + hex(c3));
+ return ((hex2bin(c2)<<4) + hex2bin(c3));
}
if (mime_decode_mode != 'B') {
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])){
#endif
#ifdef UTF8_INPUT_ENABLE
no_cp932ext_f = FALSE;
- ignore_zwnbsp_f = TRUE;
no_best_fit_chars_f = FALSE;
encode_fallback = NULL;
unicode_subchar = '?';
+ input_endian = ENDIAN_BIG;
#endif
#ifdef UTF8_OUTPUT_ENABLE
- unicode_bom_f = 0;
- w_oconv16_LE = 0;
+ output_bom_f = FALSE;
+ output_endian = ENDIAN_BIG;
#endif
#ifdef UNICODE_NORMALIZATION
nfc_f = FALSE;
prefix_table[i] = 0;
}
}
-#ifdef UTF8_INPUT_ENABLE
- utf16_mode = UTF16BE_INPUT;
-#endif
+ hold_count = 0;
mimeout_buf_count = 0;
mimeout_mode = 0;
base64_count = 0;
fprintf(stderr,"Flags:\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), EUC-JP, UTF-8N\n");
+ fprintf(stderr,"j,s,e,w Output code is JIS 7 bit, Shift_JIS (DEFAULT), EUC-JP, UTF-8N\n");
#endif
#ifdef DEFAULT_CODE_JIS
- fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit (DEFAULT), Shift JIS, EUC-JP, UTF-8N\n");
+ fprintf(stderr,"j,s,e,w Output code is JIS 7 bit (DEFAULT), Shift JIS, EUC-JP, UTF-8N\n");
#endif
#ifdef DEFAULT_CODE_EUC
- fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, EUC-JP (DEFAULT), UTF-8N\n");
+ fprintf(stderr,"j,s,e,w Output code is JIS 7 bit, Shift JIS, EUC-JP (DEFAULT), UTF-8N\n");
#endif
#ifdef DEFAULT_CODE_UTF8
- fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, EUC-JP, UTF-8N (DEFAULT)\n");
+ fprintf(stderr,"j,s,e,w Output code is JIS 7 bit, Shift JIS, EUC-JP, UTF-8N (DEFAULT)\n");
#endif
#ifdef UTF8_OUTPUT_ENABLE
fprintf(stderr," After 'w' you can add more options. -w[ 8 [0], 16 [[BL] [0]] ]\n");
OSDN Git Service
