X-Git-Url: http://git.sourceforge.jp/view?a=blobdiff_plain;ds=sidebyside;f=nkf.c;h=9c7483026629ed1d8037bddd46c9f8ec8f1ef7ed;hb=a1dc1c67e07f6d762cc32dd9dfc1037a3577a741;hp=f7f7795d8f5ff472b4d644104545d1ea280579c1;hpb=8825a83d27592ce905620f64a4be72e50d9b773b;p=nkf%2Fnkf.git

diff --git a/nkf.c b/nkf.c
index f7f7795..9c74830 100644
--- a/nkf.c
+++ b/nkf.c
@@ -39,14 +39,17 @@
 **        E-Mail: furukawa@tcp-ip.or.jp
 **    まで御連絡をお願いします。
 ***********************************************************************/
+/* $Id */
+#define NKF_VERSION "2.0.4"
+#define NKF_RELEASE_DATE "2004-11-06"
 #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 =
-      "3/0310/Shinji Kono";
+      "4/0410/Shinji Kono";
 
 /*
 **
@@ -55,8 +58,8 @@ static char *Patchlevel =
 ** 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
 **
@@ -110,9 +113,6 @@ static char *Patchlevel =
 #include <stdio.h>
 #endif
 
-#include <stdlib.h>
-#include <string.h>
-
 #if defined(MSDOS) || defined(__OS2__) 
 #include <fcntl.h>
 #include <io.h>
@@ -145,6 +145,8 @@ static char *Patchlevel =
 
 #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>
@@ -199,7 +201,7 @@ static char *Patchlevel =
 
 #define		UTF8           12
 #define		UTF8_INPUT     13
-#define		UTF16_INPUT    14
+#define		UTF16LE_INPUT  14
 #define		UTF16BE_INPUT  15
 
 #define         WISH_TRUE      15
@@ -377,8 +379,9 @@ static int             x0201_f = NO_X0201;     /* Assume NO JISX0201 */
 #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
 
 
@@ -415,6 +418,12 @@ static int debug_f = FALSE;
 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
@@ -424,22 +433,28 @@ 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}
 };
 
@@ -456,8 +471,8 @@ static int             fold_f  = FALSE;
 static int             fold_len  = 0;
 
 /* options */
-static unsigned char   kanji_intro = DEFAULT_J,
-                       ascii_intro = DEFAULT_R;
+static unsigned char   kanji_intro = DEFAULT_J;
+static unsigned char   ascii_intro = DEFAULT_R;
 
 /* Folding */
 
@@ -632,6 +647,9 @@ main(argc, argv)
     FILE  *fin;
     unsigned char  *cp;
 
+    char *outfname;
+    char *origfname;
+
 #ifdef EASYWIN /*Easy Win */
     _BufferSize.y = 400;/*Set Scroll Buffer Size*/
 #endif
@@ -693,13 +711,13 @@ main(argc, argv)
       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);
@@ -779,8 +797,12 @@ main(argc, argv)
               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) {
@@ -870,9 +892,11 @@ struct {
     {"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"},
@@ -894,14 +918,16 @@ struct {
 #endif
 #ifdef SHIFTJIS_CP932
     {"no-cp932", ""},
+    {"cp932inv", ""},
 #endif
 #ifdef EXEC_IO
     {"exec-in", ""},
     {"exec-out", ""},
 #endif
+    {"prefix=", ""},
 };
 
-static int option_mode;
+static int option_mode = 0;
 
 void
 options(cp) 
@@ -924,10 +950,14 @@ options(cp)
             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
@@ -965,7 +995,11 @@ options(cp)
 #endif
 #ifdef SHIFTJIS_CP932
                 if (strcmp(long_option[i].name, "no-cp932") == 0){
-                    cp932_f = TRUE;
+                    cp932_f = FALSE;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "cp932inv") == 0){
+                    cp932inv_f = TRUE;
                     continue;
                 }
 #endif
@@ -979,6 +1013,20 @@ options(cp)
                       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 */
@@ -1045,17 +1093,23 @@ options(cp)
             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;
@@ -1063,7 +1117,16 @@ options(cp)
 #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 if (cp[0] == '8') {
+		cp++;
+		input_f = UTF8_INPUT;
 	    } else
                 input_f = UTF8_INPUT;
             continue;
@@ -1183,6 +1246,11 @@ options(cp)
                 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++;
@@ -1196,6 +1264,25 @@ options(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)
@@ -1222,18 +1309,18 @@ 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(input_codename = "UTF-8");
-        if (iconv == w_iconv16) debug(input_codename = "UTF-16");
-#endif
-        if (iconv == s_iconv) debug(input_codename = "Shift_JIS");
-        if (iconv == e_iconv) debug(input_codename = "EUC-JP");
+        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)                   /* いわゆる半角カナ */
+#define SCORE_L2       (1)                   /* 第2水準漢字 */
+#define SCORE_KANA     (SCORE_L2 << 1)       /* いわゆる半角カナ */
 #define SCORE_DEPEND   (SCORE_KANA << 1)     /* 機種依存文字 */
 #ifdef SHIFTJIS_CP932
 #define SCORE_CP932    (SCORE_DEPEND << 1)   /* CP932 による読み換え */
@@ -1241,7 +1328,11 @@ void set_iconv(f, iconv_func)
 #else
 #define SCORE_NO_EXIST (SCORE_DEPEND << 1)   /* 存在しない文字 */
 #endif
-#define SCORE_ERROR    (SCORE_NO_EXIST << 1) /* エラー */
+#define SCORE_iMIME    (SCORE_NO_EXIST << 1) /* MIME による指定 */
+#define SCORE_ERROR    (SCORE_iMIME << 1) /* エラー */
+
+#define SCORE_INIT (SCORE_iMIME)
+
 int score_table_A0[] = {
     0, 0, 0, 0,
     0, 0, 0, 0,
@@ -1250,8 +1341,8 @@ int score_table_A0[] = {
 };
 
 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,
 };
@@ -1260,7 +1351,18 @@ void set_code_score(ptr, score)
      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)
@@ -1270,18 +1372,19 @@ 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)
@@ -1300,14 +1403,20 @@ void status_push_ch(ptr, c)
     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;
 {
@@ -1343,7 +1452,7 @@ void s_status(ptr, c)
               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);
@@ -1362,7 +1471,7 @@ void s_status(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);
           }
@@ -1373,7 +1482,7 @@ void s_status(ptr, c)
               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;
               }
           }
@@ -1409,7 +1518,7 @@ void e_status(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);
           }
@@ -1449,7 +1558,7 @@ void w16_status(ptr, c)
               ptr->_file_stat = -1;
           }else{
               status_push_ch(ptr, c);
-              status_reset(ptr);
+              status_clear(ptr);
           }
           break;
 
@@ -1457,7 +1566,7 @@ void w16_status(ptr, c)
       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;
@@ -1496,10 +1605,14 @@ void w_status(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);
@@ -1606,7 +1719,7 @@ module_connection()
 
     /* output redicrection */
 #ifdef CHECK_OPTION
-    if (noout_f){
+    if (noout_f || guess_f){
         o_putc = no_putc;
     }
 #endif
@@ -1640,6 +1753,7 @@ module_connection()
     }
 
     i_getc = std_getc;
+    i_ungetc = std_ungetc;
     /* input redicrection */
 #ifdef INPUT_OPTION
     if (cap_f){
@@ -1672,7 +1786,7 @@ module_connection()
 #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 {
@@ -1696,7 +1810,7 @@ kanji_convert(f)
     FILE  *f;
 {
     int    c1,
-                    c2;
+                    c2, c3;
 
     module_connection();
     c2 = 0;
@@ -1870,7 +1984,8 @@ kanji_convert(f)
                         /* This is kanji introduction */
                         input_mode = X0208;
                         shift_mode = FALSE;
-                        debug(input_codename = "ISO-2022-JP");
+                        set_input_codename("ISO-2022-JP");
+                        debug(input_codename);
                         NEXT;
                     } else if (c1 == '(') {
                         if ((c1 = (*i_getc)(f)) == EOF) {
@@ -1931,10 +2046,16 @@ kanji_convert(f)
                     }
                } 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  */
@@ -2011,7 +2132,7 @@ h_conv(f, c2, c1)
             }
             ++p;
         }
-        set_iconv(FALSE, p->iconv_func);
+        set_iconv(FALSE, result->iconv_func);
     }
 
 
@@ -2085,7 +2206,7 @@ int s2e_conv(c2, c1, p2, p1)
      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;
@@ -2164,8 +2285,8 @@ w2e_conv(c2, c1, c0, p2, p1)
         }
 #ifdef NUMCHAR_OPTION
         if (ret){
-            c1 = CLASS_UTF16 | ww16_conv(c2, c1, c0);
-            c2 = 0;
+            if (p2) *p2 = 0;
+            if (p1) *p1 = CLASS_UTF16 | ww16_conv(c2, c1, c0);
             ret = 0;
         }
 #endif
@@ -2269,13 +2390,13 @@ w_iconv16(c2, c1, c0)
     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;
     }
@@ -2329,6 +2450,7 @@ e2w_conv(c2, c1)
 {
     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) {
@@ -2337,7 +2459,7 @@ e2w_conv(c2, c1)
         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;
     }
@@ -2367,7 +2489,16 @@ w_oconv(c2, c1)
     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) {
@@ -2394,7 +2525,7 @@ w_oconv16(c2, c1)
         return;
     }    
 
-    if (w_oconv16_begin_f==2) {
+    if (unicode_bom_f==2) {
         if (w_oconv16_LE){
             (*o_putc)((unsigned char)'\377');
             (*o_putc)('\376');
@@ -2402,7 +2533,7 @@ w_oconv16(c2, c1)
             (*o_putc)('\376');
             (*o_putc)((unsigned char)'\377');
         }
-	w_oconv16_begin_f=1;
+	unicode_bom_f=1;
     }
 
     if (c2 == ISO8859_1) {
@@ -2452,8 +2583,8 @@ e_oconv(c2, c1)
 	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 */
         }
@@ -2501,7 +2632,23 @@ s_oconv(c2, c1)
         }
 	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);
     }
 }
@@ -3033,23 +3180,37 @@ unsigned char *mime_pattern[] = {
    (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
 };
 
+
+/* 該当するコードの優先度を上げるための目印 */
+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
 };
 
@@ -3119,6 +3280,9 @@ FILE *f;
         }
     }
     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) {
@@ -3228,6 +3392,37 @@ void debug(str)
 }
 #endif
 
+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;
@@ -3703,89 +3898,104 @@ mime_putc(c)
 void 
 reinit()
 {
+    {
+        struct input_code *p = input_code_list;
+        while (p->name){
+            status_reinit(p++);
+        }
+    }
     unbuf_f = FALSE;
     estab_f = FALSE;
     nop_f = FALSE;
-    binmode_f = TRUE;       
-    rot_f = FALSE;         
-    hira_f = FALSE;         
-    input_f = FALSE;      
-    alpha_f = FALSE;     
-    mime_f = STRICT_MIME; 
-    mimebuf_f = FALSE; 
-    broken_f = FALSE;  
-    iso8859_f = FALSE; 
-#if defined(MSDOS) || defined(__OS2__) 
-     x0201_f = TRUE;   
+    binmode_f = TRUE;
+    rot_f = FALSE;
+    hira_f = FALSE;
+    input_f = FALSE;
+    alpha_f = FALSE;
+    mime_f = STRICT_MIME;
+    mimebuf_f = FALSE;
+    broken_f = FALSE;
+    iso8859_f = FALSE;
+    mimeout_f = FALSE;
+#if defined(MSDOS) || defined(__OS2__)
+     x0201_f = TRUE;
 #else
      x0201_f = NO_X0201;
 #endif
     iso2022jp_f = FALSE;
-
+#ifdef UTF8_OUTPUT_ENABLE
+    unicode_bom_f = 0;
+    w_oconv16_LE = 0;
+    ms_ucs_map_f = FALSE;
+#endif
+#ifdef INPUT_OPTION
+    cap_f = FALSE;
+    url_f = FALSE;
+    numchar_f = FALSE;
+#endif
+#ifdef CHECK_OPTION
+    noout_f = FALSE;
+    debug_f = FALSE;
+#endif
+    guess_f = FALSE;
+    is_inputcode_mixed = FALSE;
+    is_inputcode_set   = FALSE;
+#ifdef EXEC_IO
+    exec_f = 0;
+#endif
+#ifdef SHIFTJIS_CP932
+    cp932_f = TRUE;
+    cp932inv_f = FALSE;
+#endif
+    {
+        int i;
+        for (i = 0; i < 256; i++){
+            prefix_table[i] = 0;
+        }
+    }
+#ifdef UTF8_INPUT_ENABLE
+    utf16_mode = UTF16LE_INPUT;
+#endif
+    mimeout_mode = 0;
+    base64_count = 0;
+    f_line = 0;
+    f_prev = 0;
+    fold_preserve_f = FALSE;
+    fold_f = FALSE;
+    fold_len = 0;
     kanji_intro = DEFAULT_J;
     ascii_intro = DEFAULT_R;
-
-    output_conv = DEFAULT_CONV; 
-    oconv = DEFAULT_CONV; 
-
-    i_mgetc  = std_getc; 
-    i_mungetc  = std_ungetc;
-    i_mgetc_buf = std_getc; 
-    i_mungetc_buf = std_ungetc;
-
-    i_getc= std_getc; 
-    i_ungetc=std_ungetc;
-
-    i_bgetc= std_getc;
-    i_bungetc= std_ungetc;
-
-    o_putc = std_putc;
-    o_mputc = std_putc;
-    o_crconv = no_connection; 
-    o_rot_conv = no_connection; 
-    o_iso2022jp_check_conv = no_connection;
-    o_hira_conv = no_connection; 
-    o_fconv = no_connection; 
+    fold_margin  = FOLD_MARGIN;
+    output_conv = DEFAULT_CONV;
+    oconv = DEFAULT_CONV;
     o_zconv = no_connection;
-
+    o_fconv = no_connection;
+    o_crconv = no_connection;
+    o_rot_conv = no_connection;
+    o_hira_conv = no_connection;
+    o_base64conv = no_connection;
+    o_iso2022jp_check_conv = no_connection;
+    o_putc = std_putc;
     i_getc = std_getc;
     i_ungetc = std_ungetc;
-    i_mgetc = std_getc; 
-    i_mungetc = std_ungetc; 
-
+    i_bgetc = std_getc;
+    i_bungetc = std_ungetc;
+    o_mputc = std_putc;
+    i_mgetc = std_getc;
+    i_mungetc  = std_ungetc;
+    i_mgetc_buf = std_getc;
+    i_mungetc_buf = std_ungetc;
     output_mode = ASCII;
     input_mode =  ASCII;
     shift_mode =  FALSE;
-    mime_decode_mode =   FALSE;
+    mime_decode_mode = FALSE;
     file_out = FALSE;
-    mimeout_mode = 0;
-    mimeout_f = FALSE;
-    base64_count = 0;
-    option_mode = 0;
     crmode_f = 0;
-
-    {
-        struct input_code *p = input_code_list;
-        while (p->name){
-            status_reinit(p++);
-        }
-    }
-#ifdef UTF8_OUTPUT_ENABLE
-    if (w_oconv16_begin_f) {
-	w_oconv16_begin_f = 2;
-    }
-#endif
-    f_line = 0;    
-    f_prev = 0;
-    fold_preserve_f = FALSE; 
-    fold_f  = FALSE;
-    fold_len  = 0;
-    fold_margin  = FOLD_MARGIN;
+    option_mode = 0;
     broken_counter = 0;
     broken_last = 0;
     z_prev2=0,z_prev1=0;
 
-    input_codename = "";
 }
 #endif
 
@@ -3810,7 +4020,7 @@ usage()
 {
     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
@@ -3823,7 +4033,13 @@ usage()
 #ifdef DEFAULT_CODE_UTF8
     fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC), UTF-8 (DEFAULT)\n");
 #endif
+#ifdef UTF8_OUTPUT_ENABLE
+    fprintf(stderr,"         After 'w' you can add more options. (80?|16((B|L)0?)?) \n");
+#endif
     fprintf(stderr,"J,S,E,W  Input assumption is JIS 7 bit , Shift JIS, AT&T JIS (EUC), UTF-8\n");
+#ifdef UTF8_INPUT_ENABLE
+    fprintf(stderr,"         After 'W' you can add more options. (8|16(B|L)?) \n");
+#endif
     fprintf(stderr,"t        no conversion\n");
     fprintf(stderr,"i_/o_    Output sequence to designate JIS-kanji/ASCII (DEFAULT B)\n");
     fprintf(stderr,"r        {de/en}crypt ROT13/47\n");
@@ -3845,11 +4061,25 @@ usage()
     fprintf(stderr,"I        Convert non ISO-2022-JP charactor to GETA\n");
     fprintf(stderr,"-L[uwm]  line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n");
     fprintf(stderr,"long name options\n");
-    fprintf(stderr," --fj,--unix,--mac,--windows                convert for the system\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");
+    fprintf(stderr," --hiragana, --katakana    Hiragana/Katakana Conversion\n");
+#ifdef INPUT_OPTION
+    fprintf(stderr," --cap-input, --url-input  Convert hex after ':' or '%'\n");
+#endif
+#ifdef NUMCHAR_OPTION
+    fprintf(stderr," --numchar-input      Convert Unicode Character Reference\n");
+#endif
+#ifdef SHIFTJIS_CP932
+    fprintf(stderr," --no-cp932           Don't convert Shift_JIS FAxx-FCxx to equivalnet CP932\n");
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+    fprintf(stderr," --ms-ucs-map         Microsoft UCS Mapping Compatible\n");
+#endif
 #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();
 }