X-Git-Url: http://git.sourceforge.jp/view?a=blobdiff_plain;f=nkf.c;h=cafb0c4c45f3ff8eff42d34c27b2860e287facda;hb=6af6940e58650ffcae94e3a37083b6c70c3bee75;hp=9c3d9560e74f024dcb65fad558779ae9ec026782;hpb=f802477879374bdb24a77a135c9bd460499b417c;p=nkf%2Fnkf.git

diff --git a/nkf.c b/nkf.c
index 9c3d956..cafb0c4 100644
--- a/nkf.c
+++ b/nkf.c
@@ -39,13 +39,13 @@
 **        E-Mail: furukawa@tcp-ip.or.jp
 **    まで御連絡をお願いします。
 ***********************************************************************/
-/* $Id: nkf.c,v 1.46 2004/12/01 01:59:28 naruse Exp $ */
-#define NKF_VERSION "2.0.4"
-#define NKF_RELEASE_DATE __DATE__
+/* $Id: nkf.c,v 1.81 2005/11/06 20:17:01 naruse Exp $ */
+#define NKF_VERSION "2.0.5"
+#define NKF_RELEASE_DATE "2005-11-07"
 #include "config.h"
 
-static char *CopyRight =
-      "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2004 Kono, Furukawa";
+#define COPY_RIGHT \
+    "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2005 Kono, Furukawa, Naruse"
 
 
 /*
@@ -111,6 +111,7 @@ static char *CopyRight =
 #endif
 
 #include <stdlib.h>
+#include <string.h>
 
 #if defined(MSDOS) || defined(__OS2__) 
 #include <fcntl.h>
@@ -144,17 +145,24 @@ static char *CopyRight =
 
 #ifdef OVERWRITE
 /* added by satoru@isoternet.org */
-#include <string.h>
 #include <sys/stat.h>
 #ifndef MSDOS /* UNIX, OS/2 */
 #include <unistd.h>
 #include <utime.h>
-#else
+#else /* defined(MSDOS) */
+#ifdef __WIN32__
+#ifdef __BORLANDC__ /* BCC32 */
+#include <utime.h>
+#else /* !defined(__BORLANDC__) */
+#include <sys/utime.h>
+#endif /* (__BORLANDC__) */
+#else /* !defined(__WIN32__) */
 #if defined(_MSC_VER) || defined(__MINGW32__) /* VC++, MinGW */
 #include <sys/utime.h>
 #elif defined(__TURBOC__) /* BCC */
 #include <utime.h>
 #elif defined(LSI_C) /* LSI C */
+#endif /* (__WIN32__) */
 #endif
 #endif
 #endif 
@@ -182,6 +190,7 @@ static char *CopyRight =
 #define         X0201           2
 #define         ISO8859_1       8
 #define         NO_X0201        3
+#define         X0212           16
 
 /* Input Assumption */
 
@@ -199,8 +208,8 @@ static char *CopyRight =
 
 #define		UTF8           12
 #define		UTF8_INPUT     13
-#define		UTF16LE_INPUT  14
-#define		UTF16BE_INPUT  15
+#define		UTF16BE_INPUT  14
+#define		UTF16LE_INPUT  15
 
 #define         WISH_TRUE      15
 
@@ -223,6 +232,17 @@ static char *CopyRight =
 #define		is_alnum(c)  \
             (('a'<=c && c<='z')||('A'<= c && c<='Z')||('0'<=c && c<='9'))
 
+/* I don't trust portablity of toupper */
+#define nkf_toupper(c)  (('a'<=c && c<='z')?(c-('a'-'A')):c)
+#define nkf_isoctal(c)  ('0'<=c && c<='7')
+#define nkf_isdigit(c)  ('0'<=c && c<='9')
+#define nkf_isxdigit(c)  (nkf_isdigit(c) || ('a'<=c && c<='f') || ('A'<=c && c <= 'F'))
+#define nkf_isblank(c) (c == SPACE || c == TAB)
+#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         HOLD_SIZE       1024
 #define         IOBUF_SIZE      16384
 
@@ -237,7 +257,7 @@ static char *CopyRight =
 #define         GETA2   0x2e
 
 
-#if defined( UTF8_OUTPUT_ENABLE ) || defined( UTF8_INPUT_ENABLE )
+#if defined(UTF8_OUTPUT_ENABLE) || defined(UTF8_INPUT_ENABLE)
 #define sizeof_euc_utf8 94
 #define sizeof_euc_to_utf8_1byte 94
 #define sizeof_euc_to_utf8_2bytes 94
@@ -249,7 +269,6 @@ static char *CopyRight =
 
 /* MIME preprocessor */
 
-
 #ifdef EASYWIN /*Easy Win */
 extern POINT _BufferSize;
 #endif
@@ -259,9 +278,11 @@ extern POINT _BufferSize;
 #ifdef ANSI_C_PROTOTYPE
 #define PROTO(x)  x 
 #define STATIC static
+#define CONST const
 #else
 #define PROTO(x)  ()
 #define STATIC
+#define CONST
 #endif
 
 struct input_code{
@@ -277,20 +298,32 @@ struct input_code{
 
 STATIC char *input_codename = "";
 
+#ifndef PERL_XS
+STATIC const char *CopyRight = COPY_RIGHT;
+#endif
+#if !defined(PERL_XS) && !defined(WIN32DLL)
 STATIC  int     noconvert PROTO((FILE *f));
+#endif
 STATIC  int     kanji_convert PROTO((FILE *f));
 STATIC  int     h_conv PROTO((FILE *f,int c2,int c1));
 STATIC  int     push_hold_buf PROTO((int c2));
-STATIC  void    set_iconv PROTO((int f, int (*iconv_func)()));
+STATIC  void    set_iconv PROTO((int f, int (*iconv_func)(int c2,int c1,int c0)));
 STATIC  int     s_iconv PROTO((int c2,int c1,int c0));
 STATIC  int     s2e_conv PROTO((int c2, int c1, int *p2, int *p1));
 STATIC  int     e_iconv PROTO((int c2,int c1,int c0));
 #ifdef UTF8_INPUT_ENABLE
+STATIC  void    encode_fallback_html PROTO((int c));
+STATIC  void    encode_fallback_xml PROTO((int c));
+STATIC  void    encode_fallback_java PROTO((int c));
+STATIC  void    encode_fallback_perl PROTO((int c));
+STATIC  void    encode_fallback_subchar PROTO((int c));
+STATIC  void    (*encode_fallback)PROTO((int c)) = NULL;
 STATIC  int     w2e_conv PROTO((int c2,int c1,int c0,int *p2,int *p1));
 STATIC  int     w_iconv PROTO((int c2,int c1,int c0));
 STATIC  int     w_iconv16 PROTO((int c2,int c1,int c0));
-STATIC  int	w_iconv_common PROTO((int c1,int c0,unsigned short **pp,int psize,int *p2,int *p1));
+STATIC  int	w_iconv_common PROTO((int c1,int c0,const unsigned short *const *pp,int psize,int *p2,int *p1));
 STATIC  int     ww16_conv PROTO((int c2, int c1, int c0));
+STATIC  int     w16e_conv PROTO((unsigned short val,int *p2,int *p1));
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
 STATIC  int     e2w_conv PROTO((int c2,int c1));
@@ -298,7 +331,7 @@ STATIC  void    w_oconv PROTO((int c2,int c1));
 STATIC  void    w_oconv16 PROTO((int c2,int c1));
 #endif
 STATIC  void    e_oconv PROTO((int c2,int c1));
-STATIC  void    e2s_conv PROTO((int c2, int c1, int *p2, int *p1));
+STATIC  int     e2s_conv PROTO((int c2, int c1, int *p2, int *p1));
 STATIC  void    s_oconv PROTO((int c2,int c1));
 STATIC  void    j_oconv PROTO((int c2,int c1));
 STATIC  void    fold_conv PROTO((int c2,int c1));
@@ -328,117 +361,145 @@ STATIC  int     mime_ungetc PROTO((int c,FILE *f));
 STATIC  int     mime_begin_strict PROTO((FILE *f));
 STATIC  int     mime_getc_buf PROTO((FILE *f));
 STATIC  int     mime_ungetc_buf  PROTO((int c,FILE *f));
-STATIC  int     mime_integrity PROTO((FILE *f,unsigned char *p));
+STATIC  int     mime_integrity PROTO((FILE *f,const unsigned char *p));
 
 STATIC  int     base64decode PROTO((int c));
+STATIC  void    mime_prechar PROTO((int c2, int c1));
 STATIC  void    mime_putc PROTO((int c));
 STATIC  void    open_mime PROTO((int c));
 STATIC  void    close_mime PROTO(());
+#ifndef PERL_XS
 STATIC  void    usage PROTO(());
 STATIC  void    version PROTO(());
+#endif
 STATIC  void    options PROTO((unsigned char *c));
-#ifdef PERL_XS
+#if defined(PERL_XS) || defined(WIN32DLL)
 STATIC  void    reinit PROTO(());
 #endif
 
 /* buffers */
 
-static unsigned char   stdibuf[IOBUF_SIZE];
-static unsigned char   stdobuf[IOBUF_SIZE];
-static unsigned char   hold_buf[HOLD_SIZE*2];
-static int             hold_count;
+#if !defined(PERL_XS) && !defined(WIN32DLL)
+STATIC unsigned char   stdibuf[IOBUF_SIZE];
+STATIC unsigned char   stdobuf[IOBUF_SIZE];
+#endif
+STATIC unsigned char   hold_buf[HOLD_SIZE*2];
+STATIC int             hold_count;
 
 /* MIME preprocessor fifo */
 
 #define MIME_BUF_SIZE   (1024)    /* 2^n ring buffer */
 #define MIME_BUF_MASK   (MIME_BUF_SIZE-1)   
 #define Fifo(n)         mime_buf[(n)&MIME_BUF_MASK]
-static unsigned char           mime_buf[MIME_BUF_SIZE];
-static unsigned int            mime_top = 0;
-static unsigned int            mime_last = 0;  /* decoded */
-static unsigned int            mime_input = 0; /* undecoded */
+STATIC unsigned char           mime_buf[MIME_BUF_SIZE];
+STATIC unsigned int            mime_top = 0;
+STATIC unsigned int            mime_last = 0;  /* decoded */
+STATIC unsigned int            mime_input = 0; /* undecoded */
+STATIC int (*mime_iconv_back)PROTO((int c2,int c1,int c0)) = NULL;
 
 /* flags */
-static int             unbuf_f = FALSE;
-static int             estab_f = FALSE;
-static int             nop_f = FALSE;
-static int             binmode_f = TRUE;       /* binary mode */
-static int             rot_f = FALSE;          /* rot14/43 mode */
-static int             hira_f = FALSE;          /* hira/kata henkan */
-static int             input_f = FALSE;        /* non fixed input code  */
-static int             alpha_f = FALSE;        /* convert JIx0208 alphbet to ASCII */
-static int             mime_f = STRICT_MIME;   /* convert MIME B base64 or Q */
-static int             mimebuf_f = FALSE;      /* MIME buffered input */
-static int             broken_f = FALSE;       /* convert ESC-less broken JIS */
-static int             iso8859_f = FALSE;      /* ISO8859 through */
-static int             mimeout_f = FALSE;       /* base64 mode */
+STATIC int             unbuf_f = FALSE;
+STATIC int             estab_f = FALSE;
+STATIC int             nop_f = FALSE;
+STATIC int             binmode_f = TRUE;       /* binary mode */
+STATIC int             rot_f = FALSE;          /* rot14/43 mode */
+STATIC int             hira_f = FALSE;          /* hira/kata henkan */
+STATIC int             input_f = FALSE;        /* non fixed input code  */
+STATIC int             alpha_f = FALSE;        /* convert JIx0208 alphbet to ASCII */
+STATIC int             mime_f = STRICT_MIME;   /* convert MIME B base64 or Q */
+STATIC int             mime_decode_f = FALSE;  /* mime decode is explicitly on */
+STATIC int             mimebuf_f = FALSE;      /* MIME buffered input */
+STATIC int             broken_f = FALSE;       /* convert ESC-less broken JIS */
+STATIC int             iso8859_f = FALSE;      /* ISO8859 through */
+STATIC int             mimeout_f = FALSE;       /* base64 mode */
 #if defined(MSDOS) || defined(__OS2__) 
-static int             x0201_f = TRUE;         /* Assume JISX0201 kana */
+STATIC int             x0201_f = TRUE;         /* Assume JISX0201 kana */
 #else
-static int             x0201_f = NO_X0201;     /* Assume NO JISX0201 */
+STATIC int             x0201_f = NO_X0201;     /* Assume NO JISX0201 */
+#endif
+STATIC int             iso2022jp_f = FALSE;    /* convert ISO-2022-JP */
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+STATIC int             internal_unicode_f = FALSE;   /* Internal Unicode Processing */
 #endif
-static int             iso2022jp_f = FALSE;    /* convert ISO-2022-JP */
 #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             ms_ucs_map_f = FALSE;   /* Microsoft UCS Mapping Compatible */
+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 */
+STATIC int             unicode_subchar = '?'; /* the regular substitution character */
 #endif
 
-
-#ifdef NUMCHAR_OPTION
-
-#define CLASS_MASK  0x0f000000
-#define CLASS_UTF16 0x01000000
+#ifdef UNICODE_NORMALIZATION
+STATIC int nfc_f = FALSE;
+STATIC int (*i_nfc_getc)PROTO((FILE *)) = std_getc; /* input of ugetc */
+STATIC int (*i_nfc_ungetc)PROTO((int c ,FILE *f)) = std_ungetc;
+STATIC int nfc_getc PROTO((FILE *f));
+STATIC int nfc_ungetc PROTO((int c,FILE *f));
 #endif
 
 #ifdef INPUT_OPTION
-static int cap_f = FALSE;
-static int (*i_cgetc)PROTO((FILE *)) = std_getc; /* input of cgetc */
-static int (*i_cungetc)PROTO((int c ,FILE *f)) = std_ungetc;
+STATIC int cap_f = FALSE;
+STATIC int (*i_cgetc)PROTO((FILE *)) = std_getc; /* input of cgetc */
+STATIC int (*i_cungetc)PROTO((int c ,FILE *f)) = std_ungetc;
 STATIC int cap_getc PROTO((FILE *f));
 STATIC int cap_ungetc PROTO((int c,FILE *f));
 
-static int url_f = FALSE;
-static int (*i_ugetc)PROTO((FILE *)) = std_getc; /* input of ugetc */
-static int (*i_uungetc)PROTO((int c ,FILE *f)) = std_ungetc;
+STATIC int url_f = FALSE;
+STATIC int (*i_ugetc)PROTO((FILE *)) = std_getc; /* input of ugetc */
+STATIC int (*i_uungetc)PROTO((int c ,FILE *f)) = std_ungetc;
 STATIC int url_getc PROTO((FILE *f));
 STATIC int url_ungetc PROTO((int c,FILE *f));
+#endif
 
-static int numchar_f = FALSE;
-static int (*i_ngetc)PROTO((FILE *)) = std_getc; /* input of ugetc */
-static int (*i_nungetc)PROTO((int c ,FILE *f)) = std_ungetc;
+#ifdef NUMCHAR_OPTION
+#define CLASS_MASK  0x0f000000
+#define CLASS_UTF16 0x01000000
+STATIC int numchar_f = FALSE;
+STATIC int (*i_ngetc)PROTO((FILE *)) = std_getc; /* input of ugetc */
+STATIC int (*i_nungetc)PROTO((int c ,FILE *f)) = std_ungetc;
 STATIC int numchar_getc PROTO((FILE *f));
 STATIC int numchar_ungetc PROTO((int c,FILE *f));
 #endif
 
 #ifdef CHECK_OPTION
-static int noout_f = FALSE;
+STATIC int noout_f = FALSE;
 STATIC void no_putc PROTO((int c));
-static int debug_f = FALSE;
-STATIC void debug PROTO((char *str));
+STATIC int debug_f = FALSE;
+STATIC void debug PROTO((const char *str));
+STATIC int (*iconv_for_check)() = 0;
 #endif
 
-static int guess_f = FALSE;
+STATIC int guess_f = FALSE;
+#if !defined PERL_XS
 STATIC  void    print_guessed_code PROTO((char *filename));
+#endif
 STATIC  void    set_input_codename PROTO((char *codename));
-static int is_inputcode_mixed = FALSE;
-static int is_inputcode_set   = FALSE;
+STATIC int is_inputcode_mixed = FALSE;
+STATIC int is_inputcode_set   = FALSE;
 
 #ifdef EXEC_IO
-static int exec_f = 0;
+STATIC int exec_f = 0;
 #endif
 
 #ifdef SHIFTJIS_CP932
-STATIC int cp932_f = TRUE;
+/* invert IBM extended characters to others and controls some UCS mapping */
+STATIC int cp51932_f = TRUE;
 #define CP932_TABLE_BEGIN (0xfa)
 #define CP932_TABLE_END   (0xfc)
 
+/* 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 int cp932_conv PROTO((int c2, int c1)); */
 #endif /* SHIFTJIS_CP932 */
 
+#ifdef X0212_ENABLE
+STATIC int x0212_f = FALSE;
+STATIC int x0212_shift PROTO((int c));
+STATIC int x0212_unshift PROTO((int c));
+#endif
+
 STATIC unsigned char prefix_table[256];
 
 STATIC void e_status PROTO((struct input_code *, int));
@@ -447,39 +508,41 @@ 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 = UTF16LE_INPUT;
+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},
+#endif
     {0}
 };
 
-static int              mimeout_mode = 0;
-static int              base64_count = 0;
+STATIC int              mimeout_mode = 0;
+STATIC int              base64_count = 0;
 
 /* X0208 -> ASCII converter */
 
 /* fold parameter */
-static int             f_line = 0;    /* chars in line */
-static int             f_prev = 0;
-static int             fold_preserve_f = FALSE; /* preserve new lines */
-static int             fold_f  = FALSE;
-static int             fold_len  = 0;
+STATIC int             f_line = 0;    /* chars in line */
+STATIC int             f_prev = 0;
+STATIC int             fold_preserve_f = FALSE; /* preserve new lines */
+STATIC int             fold_f  = FALSE;
+STATIC int             fold_len  = 0;
 
 /* options */
-static unsigned char   kanji_intro = DEFAULT_J;
-static unsigned char   ascii_intro = DEFAULT_R;
+STATIC unsigned char   kanji_intro = DEFAULT_J;
+STATIC unsigned char   ascii_intro = DEFAULT_R;
 
 /* Folding */
 
 #define FOLD_MARGIN  10
 #define DEFAULT_FOLD 60
 
-static int             fold_margin  = FOLD_MARGIN;
+STATIC int             fold_margin  = FOLD_MARGIN;
 
 /* converters */
 
@@ -497,50 +560,50 @@ static int             fold_margin  = FOLD_MARGIN;
 #endif
 
 /* process default */
-static void (*output_conv)PROTO((int c2,int c1)) = DEFAULT_CONV;   
+STATIC void (*output_conv)PROTO((int c2,int c1)) = DEFAULT_CONV;   
 
-static void (*oconv)PROTO((int c2,int c1)) = no_connection; 
+STATIC void (*oconv)PROTO((int c2,int c1)) = no_connection; 
 /* s_iconv or oconv */
-static int (*iconv)PROTO((int c2,int c1,int c0)) = no_connection2;   
+STATIC int (*iconv)PROTO((int c2,int c1,int c0)) = no_connection2;   
 
-static void (*o_zconv)PROTO((int c2,int c1)) = no_connection; 
-static void (*o_fconv)PROTO((int c2,int c1)) = no_connection; 
-static void (*o_crconv)PROTO((int c2,int c1)) = no_connection; 
-static void (*o_rot_conv)PROTO((int c2,int c1)) = no_connection; 
-static void (*o_hira_conv)PROTO((int c2,int c1)) = no_connection; 
-static void (*o_base64conv)PROTO((int c2,int c1)) = no_connection;
-static void (*o_iso2022jp_check_conv)PROTO((int c2,int c1)) = no_connection;
+STATIC void (*o_zconv)PROTO((int c2,int c1)) = no_connection; 
+STATIC void (*o_fconv)PROTO((int c2,int c1)) = no_connection; 
+STATIC void (*o_crconv)PROTO((int c2,int c1)) = no_connection; 
+STATIC void (*o_rot_conv)PROTO((int c2,int c1)) = no_connection; 
+STATIC void (*o_hira_conv)PROTO((int c2,int c1)) = no_connection; 
+STATIC void (*o_base64conv)PROTO((int c2,int c1)) = no_connection;
+STATIC void (*o_iso2022jp_check_conv)PROTO((int c2,int c1)) = no_connection;
 
-/* static redirections */
+/* STATIC redirections */
 
-static  void   (*o_putc)PROTO((int c)) = std_putc;
+STATIC  void   (*o_putc)PROTO((int c)) = std_putc;
 
-static  int    (*i_getc)PROTO((FILE *f)) = std_getc; /* general input */
-static  int    (*i_ungetc)PROTO((int c,FILE *f)) =std_ungetc;
+STATIC  int    (*i_getc)PROTO((FILE *f)) = std_getc; /* general input */
+STATIC  int    (*i_ungetc)PROTO((int c,FILE *f)) =std_ungetc;
 
-static  int    (*i_bgetc)PROTO((FILE *)) = std_getc; /* input of mgetc */
-static  int    (*i_bungetc)PROTO((int c ,FILE *f)) = std_ungetc;
+STATIC  int    (*i_bgetc)PROTO((FILE *)) = std_getc; /* input of mgetc */
+STATIC  int    (*i_bungetc)PROTO((int c ,FILE *f)) = std_ungetc;
 
-static  void   (*o_mputc)PROTO((int c)) = std_putc ; /* output of mputc */
+STATIC  void   (*o_mputc)PROTO((int c)) = std_putc ; /* output of mputc */
 
-static  int    (*i_mgetc)PROTO((FILE *)) = std_getc; /* input of mgetc */
-static  int    (*i_mungetc)PROTO((int c ,FILE *f)) = std_ungetc;
+STATIC  int    (*i_mgetc)PROTO((FILE *)) = std_getc; /* input of mgetc */
+STATIC  int    (*i_mungetc)PROTO((int c ,FILE *f)) = std_ungetc;
 
 /* for strict mime */
-static  int    (*i_mgetc_buf)PROTO((FILE *)) = std_getc; /* input of mgetc_buf */
-static  int    (*i_mungetc_buf)PROTO((int c,FILE *f)) = std_ungetc;
+STATIC  int    (*i_mgetc_buf)PROTO((FILE *)) = std_getc; /* input of mgetc_buf */
+STATIC  int    (*i_mungetc_buf)PROTO((int c,FILE *f)) = std_ungetc;
 
 /* Global states */
-static int output_mode = ASCII,    /* output kanji mode */
+STATIC int output_mode = ASCII,    /* output kanji mode */
            input_mode =  ASCII,    /* input kanji mode */
            shift_mode =  FALSE;    /* TRUE shift out, or X0201  */
-static int mime_decode_mode =   FALSE;    /* MIME mode B base64, Q hex */
+STATIC int mime_decode_mode =   FALSE;    /* MIME mode B base64, Q hex */
 
 /* X0201 / X0208 conversion tables */
 
 /* X0201 kana conversion table */
 /* 90-9F A0-DF */
-static
+STATIC const
 unsigned char cv[]= {
     0x21,0x21,0x21,0x23,0x21,0x56,0x21,0x57,
     0x21,0x22,0x21,0x26,0x25,0x72,0x25,0x21,
@@ -563,7 +626,7 @@ unsigned char cv[]= {
 
 /* X0201 kana conversion table for daguten */
 /* 90-9F A0-DF */
-static
+STATIC const
 unsigned char dv[]= { 
     0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
     0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
@@ -585,7 +648,7 @@ unsigned char dv[]= {
 
 /* X0201 kana conversion table for han-daguten */
 /* 90-9F A0-DF */
-static
+STATIC const
 unsigned char ev[]= { 
     0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
     0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
@@ -608,7 +671,7 @@ unsigned char ev[]= {
 
 /* X0208 kigou conversion table */
 /* 0x8140 - 0x819e */
-static
+STATIC const
 unsigned char fv[] = {
 
     0x00,0x00,0x00,0x00,0x2c,0x2e,0x00,0x3a,
@@ -628,17 +691,24 @@ unsigned char fv[] = {
 
 #define    CRLF      1
 
-static int             file_out = FALSE;
+STATIC int             file_out = FALSE;
 #ifdef OVERWRITE
-static int             overwrite = FALSE;
+STATIC int             overwrite = FALSE;
 #endif
 
-static int             crmode_f = 0;   /* CR, NL, CRLF */
+STATIC int             crmode_f = 0;   /* CR, NL, CRLF */
 #ifdef EASYWIN /*Easy Win */
-static int             end_check;
+STATIC int             end_check;
 #endif /*Easy Win */
 
-#ifndef PERL_XS
+#define STD_GC_BUFSIZE (256)
+int std_gc_buf[STD_GC_BUFSIZE];
+int std_gc_ndx;
+
+#ifdef WIN32DLL
+#include "nkf32dll.c"
+#elif defined(PERL_XS)
+#else /* WIN32DLL */
 int
 main(argc, argv)
     int             argc;
@@ -718,6 +788,12 @@ main(argc, argv)
     } else {
       int nfiles = argc;
       while (argc--) {
+	    is_inputcode_mixed = FALSE;
+	    is_inputcode_set   = FALSE;
+	    input_codename = "";
+#ifdef CHECK_OPTION
+	    iconv_for_check = 0;
+#endif
           if ((fin = fopen((origfname = *argv++), "r")) == NULL) {
               perror(*--argv);
               return(-1);
@@ -807,7 +883,7 @@ main(argc, argv)
 #ifdef OVERWRITE
               if (overwrite) {
                   struct stat     sb;
-#if defined(MSDOS) && !defined(__MINGW32__)
+#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__)
                   time_t tb[2];
 #else
                   struct utimbuf  tb;
@@ -827,7 +903,7 @@ main(argc, argv)
                   }
 
                   /* タイムスタンプを復元 */
-#if defined(MSDOS) && !defined(__MINGW32__)
+#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__)
                   tb[0] = tb[1] = sb.st_mtime;
                   if (utime(outfname, tb)) {
                       fprintf(stderr, "Can't set timestamp %s\n", outfname);
@@ -863,16 +939,18 @@ main(argc, argv)
 #else /* for Other OS */
     if (file_out == TRUE) 
         fclose(stdout);
-#endif 
+#endif /*Easy Win */
     return (0);
 }
-#endif
+#endif /* WIN32DLL */
 
-static 
+STATIC const
 struct {
-    char *name;
-    char *alias;
+    const char *name;
+    const char *alias;
 } long_option[] = {
+    {"ic=", ""},
+    {"oc=", ""},
     {"base64","jMB"},
     {"euc","e"},
     {"euc-input","E"},
@@ -895,15 +973,31 @@ struct {
     {"guess", "g"},
     {"cp932", ""},
     {"no-cp932", ""},
+#ifdef X0212_ENABLE
+    {"x0212", ""},
+#endif
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+    {"internal-unicode", ""},
+#endif
 #ifdef UTF8_OUTPUT_ENABLE
     {"utf8", "w"},
     {"utf16", "w16"},
     {"ms-ucs-map", ""},
+    {"fb-skip", ""},
+    {"fb-html", ""},
+    {"fb-xml", ""},
+    {"fb-perl", ""},
+    {"fb-java", ""},
+    {"fb-subchar", ""},
+    {"fb-subchar=", ""},
 #endif
 #ifdef UTF8_INPUT_ENABLE
     {"utf8-input", "W"},
     {"utf16-input", "W16"},
 #endif
+#ifdef UNICODE_NORMALIZATION
+    {"utf8mac-input", ""},
+#endif
 #ifdef OVERWRITE
     {"overwrite", ""},
 #endif
@@ -928,43 +1022,187 @@ struct {
     {"prefix=", ""},
 };
 
-static int option_mode = 0;
+STATIC int option_mode = 0;
 
 void
 options(cp) 
      unsigned char *cp;
 {
-    int i;
+    int i, j;
     unsigned char *p = NULL;
+    unsigned char *cp_back = NULL;
+    unsigned char codeset[32];
 
     if (option_mode==1)
 	return;
-    if (*cp++ != '-') 
-	return;
-    while (*cp) {
-	if (p && !*cp) {
-	    cp = p;
-	    p = 0;
+    while(*cp && *cp++!='-');
+    while (*cp || cp_back) {
+	if(!*cp){
+	    cp = cp_back;
+	    cp_back = NULL;
+	    continue;
 	}
+	p = 0;
         switch (*cp++) {
         case '-':  /* literal options */
-	    if (!*cp) {        /* ignore the rest of arguments */
+	    if (!*cp || *cp == SPACE) {        /* ignore the rest of arguments */
 		option_mode = 1;
 		return;
 	    }
             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 != '=') && *p == cp[j];p++, j++);
-		if (!*p || *p == cp[j]){
-		    p = &cp[j];
+                for (j=0;*p && *p != '=' && *p == cp[j];p++, j++);
+		if (*p == cp[j] || cp[j] == ' '){
+		    p = &cp[j] + 1;
 		    break;
 		}
 		p = 0;
             }
 	    if (p == 0) return;
-            cp = (unsigned char *)long_option[i].alias;
-            if (!*cp){
+	    while(*cp && *cp != SPACE && cp++);
+            if (long_option[i].alias[0]){
+		cp_back = cp;
+		cp = (unsigned char *)long_option[i].alias;
+	    }else{
+                if (strcmp(long_option[i].name, "ic=") == 0){
+		    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){
+			input_f = JIS_INPUT;
+		    }else if(strcmp(codeset, "SHIFT_JIS") == 0){
+			input_f = SJIS_INPUT;
+			if (x0201_f==NO_X0201) x0201_f=TRUE;
+		    }else if(strcmp(codeset, "CP932") == 0){
+			input_f = SJIS_INPUT;
+			x0201_f = FALSE;
+#ifdef SHIFTJIS_CP932
+                    cp51932_f = TRUE;
+                    cp932inv_f = TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+                    ms_ucs_map_f = TRUE;
+#endif
+		    }else if(strcmp(codeset, "EUCJP") == 0 ||
+			     strcmp(codeset, "EUC-JP") == 0 ||
+			     strcmp(codeset, "CP51932") == 0){
+			input_f = JIS_INPUT;
+			x0201_f = FALSE;
+#ifdef SHIFTJIS_CP932
+                    cp51932_f = TRUE;
+                    cp932inv_f = TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+                    ms_ucs_map_f = TRUE;
+#endif
+		    }else if(strcmp(codeset, "EUC-JP-MS") == 0 ||
+			     strcmp(codeset, "EUCJP-MS") == 0){
+			input_f = JIS_INPUT;
+			x0201_f = FALSE;
+#ifdef SHIFTJIS_CP932
+                    cp51932_f = FALSE;
+                    cp932inv_f = TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+                    ms_ucs_map_f = TRUE;
+#endif
+#ifdef UTF8_INPUT_ENABLE
+		    }else if(strcmp(codeset, "UTF-8") == 0 ||
+			     strcmp(codeset, "UTF-8N") == 0 ||
+			     strcmp(codeset, "UTF-8-BOM") == 0){
+			input_f = UTF8_INPUT;
+#ifdef UNICODE_NORMALIZATION
+		    }else if(strcmp(codeset, "UTF8-MAC") == 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 ||
+			     strcmp(codeset, "UTF-16BE-BOM") == 0){
+			input_f = UTF16BE_INPUT;
+			utf16_mode = UTF16BE_INPUT;
+		    }else if(strcmp(codeset, "UTF-16LE") == 0 ||
+			     strcmp(codeset, "UTF-16LE-BOM") == 0){
+			input_f = UTF16LE_INPUT;
+			utf16_mode = UTF16LE_INPUT;
+#endif
+		    }
+                    continue;
+		}
+                if (strcmp(long_option[i].name, "oc=") == 0){
+		    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){
+			output_conv = j_oconv;
+		    }else if(strcmp(codeset, "SHIFT_JIS") == 0){
+			output_conv = s_oconv;
+		    }else if(strcmp(codeset, "CP932") == 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 = TRUE;
+#endif
+		    }else if(strcmp(codeset, "EUCJP") == 0 ||
+			     strcmp(codeset, "EUC-JP") == 0 ||
+			     strcmp(codeset, "CP51932") == 0){
+			output_conv = e_oconv;
+			x0201_f = FALSE;
+#ifdef SHIFTJIS_CP932
+                    cp51932_f = TRUE;
+                    cp932inv_f = TRUE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+                    ms_ucs_map_f = TRUE;
+#endif
+		    }else if(strcmp(codeset, "EUC-JP-MS") == 0 ||
+			     strcmp(codeset, "EUCJP-MS") == 0){
+			output_conv = e_oconv;
+			x0201_f = FALSE;
+			x0212_f = TRUE;
+#ifdef SHIFTJIS_CP932
+                    cp51932_f = FALSE;
+#endif
+#ifdef UTF8_OUTPUT_ENABLE
+                    ms_ucs_map_f = TRUE;
+#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;
+		    }else if(strcmp(codeset, "UTF-16") == 0){
+			output_conv = w_oconv16; 
+		    }else if(strcmp(codeset, "UTF-16BE") == 0){
+			output_conv = w_oconv16; 
+			unicode_bom_f=1;
+		    }else if(strcmp(codeset, "UTF-16BE-BOM") == 0){
+			output_conv = w_oconv16; 
+			unicode_bom_f=2;
+		    }else if(strcmp(codeset, "UTF-16LE") == 0){
+			output_conv = w_oconv16; 
+			w_oconv16_LE = 1;
+			unicode_bom_f=1;
+		    }else if(strcmp(codeset, "UTF-16LE-BOM") == 0){
+			output_conv = w_oconv16; 
+			w_oconv16_LE = 1;
+			unicode_bom_f=2;
+#endif
+		    }
+                    continue;
+		}
 #ifdef OVERWRITE
                 if (strcmp(long_option[i].name, "overwrite") == 0){
                     file_out = TRUE;
@@ -1000,7 +1238,7 @@ options(cp)
 #endif
                 if (strcmp(long_option[i].name, "cp932") == 0){
 #ifdef SHIFTJIS_CP932
-                    cp932_f = TRUE;
+                    cp51932_f = TRUE;
                     cp932inv_f = TRUE;
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
@@ -1010,7 +1248,7 @@ options(cp)
                 }
                 if (strcmp(long_option[i].name, "no-cp932") == 0){
 #ifdef SHIFTJIS_CP932
-                    cp932_f = FALSE;
+                    cp51932_f = FALSE;
                     cp932inv_f = FALSE;
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
@@ -1024,6 +1262,14 @@ options(cp)
                     continue;
                 }
 #endif
+
+#ifdef X0212_ENABLE
+                if (strcmp(long_option[i].name, "x0212") == 0){
+                    x0212_f = TRUE;
+                    continue;
+                }
+#endif
+
 #ifdef EXEC_IO
                   if (strcmp(long_option[i].name, "exec-in") == 0){
                       exec_f = 1;
@@ -1034,16 +1280,79 @@ options(cp)
                       return;
                   }
 #endif
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+                if (strcmp(long_option[i].name, "internal-unicode") == 0){
+                    internal_unicode_f = TRUE;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-skip") == 0){
+		    encode_fallback = NULL;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-html") == 0){
+		    encode_fallback = encode_fallback_html;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-xml" ) == 0){
+		    encode_fallback = encode_fallback_xml;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-java") == 0){
+		    encode_fallback = encode_fallback_java;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-perl") == 0){
+		    encode_fallback = encode_fallback_perl;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-subchar") == 0){
+		    encode_fallback = encode_fallback_subchar;
+                    continue;
+                }
+                if (strcmp(long_option[i].name, "fb-subchar=") == 0){
+		    encode_fallback = encode_fallback_subchar;
+		    unicode_subchar = 0;
+		    if (p[0] != '0'){
+			/* decimal number */
+			for (i = 0; i < 7 && nkf_isdigit(p[i]); i++){
+			    unicode_subchar *= 10;
+			    unicode_subchar += hex2bin(p[i]);
+			}
+		    }else if(p[1] == 'x' || p[1] == 'X'){
+			/* hexadecimal number */
+			for (i = 2; i < 8 && nkf_isxdigit(p[i]); i++){
+			    unicode_subchar <<= 4;
+			    unicode_subchar |= hex2bin(p[i]);
+			}
+		    }else{
+			/* octal number */
+			for (i = 1; i < 8 && nkf_isoctal(p[i]); i++){
+			    unicode_subchar *= 8;
+			    unicode_subchar += hex2bin(p[i]);
+			}
+		    }
+		    w16e_conv(unicode_subchar, &i, &j);
+		    unicode_subchar = i<<8 | j;
+                    continue;
+                }
+#endif
 #ifdef UTF8_OUTPUT_ENABLE
                 if (strcmp(long_option[i].name, "ms-ucs-map") == 0){
                     ms_ucs_map_f = TRUE;
                     continue;
                 }
 #endif
+#ifdef UNICODE_NORMALIZATION
+		if (strcmp(long_option[i].name, "utf8mac-input") == 0){
+		    input_f = UTF8_INPUT;
+		    nfc_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];
+                    if (' ' < p[0] && p[0] < 128){
+                        for (i = 1; ' ' < p[i] && p[i] < 128; i++){
+                            prefix_table[p[i]] = p[0];
                         }
                     }
                     continue;
@@ -1083,8 +1392,8 @@ options(cp)
             continue;
         case 'h':
             /*  
-                bit:1   hira -> kata
-                bit:2   kata -> hira
+                bit:1   katakana->hiragana
+                bit:2   hiragana->katakana
             */
             if ('9'>= *cp && *cp>='0') 
                 hira_f |= (*cp++ -'0');
@@ -1138,12 +1447,17 @@ options(cp)
 #ifdef UTF8_INPUT_ENABLE
         case 'W':           /* UTF-8 input */
             if ('1'== cp[0] && '6'==cp[1]) {
-		input_f = UTF16LE_INPUT;
+		input_f = UTF16BE_INPUT;
+		utf16_mode = UTF16BE_INPUT;
+		cp += 2;
 		if (cp[0]=='L') {
 		    cp++;
+		    input_f = UTF16LE_INPUT;
+		    utf16_mode = UTF16LE_INPUT;
 		} else if (cp[0] == 'B') {
 		    cp++;
 		    input_f = UTF16BE_INPUT;
+		    utf16_mode = UTF16BE_INPUT;
 		}
 	    } else if (cp[0] == '8') {
 		cp++;
@@ -1210,6 +1524,7 @@ options(cp)
 	    }
             continue;
         case 'm':   /* MIME support */
+            /* mime_decode_f = TRUE; */ /* this has too large side effects... */
             if (*cp=='B'||*cp=='Q') {
                 mime_decode_mode = *cp++;
                 mimebuf_f = FIXED_MIME;
@@ -1218,6 +1533,7 @@ options(cp)
             } else if (*cp=='S') {
                 mime_f = STRICT_MIME; cp++;
             } else if (*cp=='0') {
+                mime_decode_f = FALSE;
                 mime_f = FALSE; cp++;
             }
             continue;
@@ -1274,8 +1590,7 @@ options(cp)
             continue;
         case ' ':    
         /* module muliple options in a string are allowed for Perl moudle  */
-	    while(*cp && *cp!='-') cp++;
-            if(*cp=='-') cp++;
+	    while(*cp && *cp++!='-');
             continue;
         default:
             /* bogus option but ignored */
@@ -1311,9 +1626,6 @@ void set_iconv(f, iconv_func)
      int (*iconv_func)();
 #endif
 {
-#ifdef CHECK_OPTION
-    static int (*iconv_for_check)() = 0;
-#endif
 #ifdef INPUT_CODE_FIX
     if (f || !input_f)
 #endif
@@ -1354,14 +1666,14 @@ void set_iconv(f, iconv_func)
 
 #define SCORE_INIT (SCORE_iMIME)
 
-int score_table_A0[] = {
+const int score_table_A0[] = {
     0, 0, 0, 0,
     0, 0, 0, 0,
     0, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
     SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_NO_EXIST,
 };
 
-int score_table_F0[] = {
+const int score_table_F0[] = {
     SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
     SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
     SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
@@ -1390,7 +1702,9 @@ void code_score(ptr)
      struct input_code *ptr;
 {
     int c2 = ptr->buf[0];
+#ifdef UTF8_OUTPUT_ENABLE
     int c1 = ptr->buf[1];
+#endif
     if (c2 < 0){
         set_code_score(ptr, SCORE_ERROR);
     }else if (c2 == SSO){
@@ -1478,11 +1792,16 @@ void s_status(ptr, c)
               ptr->stat = 1;
               status_push_ch(ptr, c);
 #ifdef SHIFTJIS_CP932
-          }else if (cp932_f
+          }else if (cp51932_f
                     && CP932_TABLE_BEGIN <= c && c <= CP932_TABLE_END){
               ptr->stat = 2;
               status_push_ch(ptr, c);
 #endif /* SHIFTJIS_CP932 */
+#ifdef X0212_ENABLE
+          }else if (x0212_f && 0xf0 <= c && c <= 0xfc){
+              ptr->stat = 1;
+              status_push_ch(ptr, c);
+#endif /* X0212_ENABLE */
           }else{
               status_disable(ptr);
           }
@@ -1497,8 +1816,8 @@ void s_status(ptr, c)
               status_disable(ptr);
           }
           break;
-#ifdef SHIFTJIS_CP932
       case 2:
+#ifdef SHIFTJIS_CP932
           if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
               status_push_ch(ptr, c);
               if (s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]) == 0){
@@ -1507,9 +1826,11 @@ void s_status(ptr, c)
                   break;
               }
           }
+#endif /* SHIFTJIS_CP932 */
+#ifndef X0212_ENABLE
           status_disable(ptr);
           break;
-#endif /* SHIFTJIS_CP932 */
+#endif
     }
 }
 
@@ -1531,6 +1852,11 @@ void e_status(ptr, c)
           }else if (SSO == c || (0xa1 <= c && c <= 0xfe)){
               ptr->stat = 1;
               status_push_ch(ptr, c);
+#ifdef X0212_ENABLE
+          }else if (0x8f == c){
+              ptr->stat = 2;
+              status_push_ch(ptr, c);
+#endif /* X0212_ENABLE */
           }else{
               status_disable(ptr);
           }
@@ -1544,6 +1870,15 @@ void e_status(ptr, c)
               status_disable(ptr);
           }
           break;
+#ifdef X0212_ENABLE
+      case 2:
+          if (0xa1 <= c && c <= 0xfe){
+              ptr->stat = 1;
+              status_push_ch(ptr, c);
+          }else{
+              status_disable(ptr);
+          }
+#endif /* X0212_ENABLE */
     }
 }
 
@@ -1677,10 +2012,7 @@ code_status(c)
     }
 }
 
-#define STD_GC_BUFSIZE (256)
-int std_gc_buf[STD_GC_BUFSIZE];
-int std_gc_ndx;
-
+#ifndef WIN32DLL
 int 
 std_getc(f)
 FILE *f;
@@ -1690,6 +2022,7 @@ FILE *f;
     }
     return getc(f);
 }
+#endif /*WIN32DLL*/
 
 int 
 std_ungetc(c,f)
@@ -1703,6 +2036,7 @@ FILE *f;
     return c;
 }
 
+#ifndef WIN32DLL
 void 
 std_putc(c)
 int c;
@@ -1710,7 +2044,9 @@ int c;
     if(c!=EOF)
       putchar(c);
 }
+#endif /*WIN32DLL*/
 
+#if !defined(PERL_XS) && !defined(WIN32DLL)
 int
 noconvert(f)
     FILE  *f;
@@ -1721,7 +2057,7 @@ noconvert(f)
       (*o_putc)(c);
     return 1;
 }
-
+#endif
 
 void
 module_connection()
@@ -1785,6 +2121,12 @@ module_connection()
         i_nungetc = i_ungetc; i_ungetc= numchar_ungetc;
     }
 #endif
+#ifdef UNICODE_NORMALIZATION
+    if (nfc_f && input_f == UTF8_INPUT){
+        i_nfc_getc = i_getc; i_getc = nfc_getc;
+        i_nfc_ungetc = i_ungetc; i_ungetc= nfc_ungetc;
+    }
+#endif
     if (mime_f && mimebuf_f==FIXED_MIME) {
 	i_mgetc = i_getc; i_getc = mime_getc;
 	i_mungetc = i_ungetc; i_ungetc = mime_ungetc;
@@ -1800,6 +2142,8 @@ module_connection()
 #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) {
         set_iconv(-TRUE, w_iconv16);
 #endif
@@ -1825,6 +2169,7 @@ kanji_convert(f)
 {
     int    c1,
                     c2, c3;
+    int is_8bit = FALSE;
 
     module_connection();
     c2 = 0;
@@ -1844,7 +2189,7 @@ kanji_convert(f)
             /* second byte */
             if (c2 > DEL) {
                 /* in case of 8th bit is on */
-                if (!estab_f) {
+                if (!estab_f&&!mime_decode_mode) {
                     /* in case of not established yet */
                     /* It is still ambiguious */
                     if (h_conv(f, c2, c1)==EOF) 
@@ -1889,6 +2234,7 @@ kanji_convert(f)
                 /* 8 bit code */
                 if (!estab_f && !iso8859_f) {
                     /* not established yet */
+		    if (!is_8bit) is_8bit = TRUE;
                     c2 = c1;
                     NEXT;
                 } else { /* estab_f==TRUE */
@@ -1978,13 +2324,13 @@ kanji_convert(f)
                     /* normal ASCII code */ 
                     SEND;
                 }
-            } else if (c1 == SI) {
+            } else if (!is_8bit && c1 == SI) {
                 shift_mode = FALSE; 
                 NEXT;
-            } else if (c1 == SO) {
+            } else if (!is_8bit && c1 == SO) {
                 shift_mode = TRUE; 
                 NEXT;
-            } else if (c1 == ESC ) {
+            } else if (!is_8bit && c1 == ESC ) {
                 if ((c1 = (*i_getc)(f)) == EOF) {
                     /*  (*oconv)(0, ESC); don't send bogus code */
                     LAST;
@@ -1999,7 +2345,9 @@ kanji_convert(f)
                         input_mode = X0208;
                         shift_mode = FALSE;
                         set_input_codename("ISO-2022-JP");
+#ifdef CHECK_OPTION
                         debug(input_codename);
+#endif
                         NEXT;
                     } else if (c1 == '(') {
                         if ((c1 = (*i_getc)(f)) == EOF) {
@@ -2014,6 +2362,12 @@ kanji_convert(f)
                             input_mode = X0208;
                             shift_mode = FALSE;
                             NEXT;
+#ifdef X0212_ENABLE
+                        } else if (c1 == 'D'){
+                            input_mode = X0212;
+                            shift_mode = FALSE;
+                            NEXT;
+#endif /* X0212_ENABLE */
                         } else {
                             /* could be some special code */
                             (*oconv)(0, ESC);
@@ -2079,8 +2433,7 @@ kanji_convert(f)
             } else if ((c1 == NL || c1 == CR) && broken_f&4) {
                 input_mode = ASCII; set_iconv(FALSE, 0);
                 SEND;
-	    /*
-	    } else if (c1 == NL && mime_f && !mime_decode_mode ) {
+	    } else if (c1 == NL && mime_decode_f && !mime_decode_mode ) {
 		if ((c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
 		    i_ungetc(SPACE,f);
 		    continue;
@@ -2089,7 +2442,7 @@ kanji_convert(f)
 		}
 		c1 = NL;
 		SEND;
-	    } else if (c1 == CR && mime_f && !mime_decode_mode ) {
+	    } else if (c1 == CR && mime_decode_f && !mime_decode_mode ) {
 		if ((c1=(*i_getc)(f))!=EOF) {
 		    if (c1==SPACE) {
 			i_ungetc(SPACE,f);
@@ -2106,13 +2459,16 @@ kanji_convert(f)
 		}
 		c1 = CR;
 		SEND;
-	    */
 	    } else 
                 SEND;
         }
         /* send: */
         if (input_mode == X0208) 
             (*oconv)(c2, c1);  /* this is JIS, not SJIS/EUC case */
+#ifdef X0212_ENABLE
+        else if (input_mode == X0212) 
+            (*oconv)((0x8f << 8) | c2, c1);
+#endif /* X0212_ENABLE */
         else if (input_mode) 
             (*oconv)(input_mode, c1);  /* other special case */
         else if ((*iconv)(c2, c1, 0) < 0){  /* can be EUC/SJIS */
@@ -2130,6 +2486,18 @@ kanji_convert(f)
 
     /* epilogue */
     (*iconv)(EOF, 0, 0);
+    if (!is_inputcode_set)
+    {
+	if (is_8bit) {
+	    struct input_code *p = input_code_list;
+	    struct input_code *result = p;
+	    while (p->name){
+		if (p->score < result->score) result = p;
+		++p;
+	    }
+	    set_input_codename(result->name);
+	}
+    }
     return 1;
 }
 
@@ -2247,22 +2615,52 @@ int s2e_conv(c2, c1, p2, p1)
      int c2, c1;
      int *p2, *p1;
 {
+#if defined(SHIFTJIS_CP932) || defined(X0212_ENABLE)
+    int val;
+#endif
 #ifdef SHIFTJIS_CP932
-    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;
-        c2 = c1 >> 8;
-        c1 &= 0xff;
+    if (cp51932_f && CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){
+        extern const unsigned short shiftjis_cp932[3][189];
+        val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
+        if (val){
+            c2 = val >> 8;
+            c1 = val & 0xff;
+        }
     }
 #endif /* SHIFTJIS_CP932 */
-    c2 = c2 + c2 - ((c2 <= 0x9f) ? SJ0162 : SJ6394);
-    if (c1 < 0x9f)
-        c1 = c1 - ((c1 > DEL) ? SPACE : 0x1f);
-    else {
-        c1 = c1 - 0x7e;
-        c2++;
+#ifdef X0212_ENABLE
+    if (x0212_f && 0xfa <= c2 && c2 <= 0xfc){
+        extern const unsigned short shiftjis_x0212[3][189];
+        val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40];
+        if (val){
+            if (val & 0x8000){
+                c2 = (0x8f << 8) | (val >> 8);
+                c1 = val & 0xff;
+            }else{
+                c2 = val >> 8;
+                c1 = val & 0xff;
+            }
+            if (p2) *p2 = c2;
+            if (p1) *p1 = c1;
+            return 0;
+        }
+    }
+#endif
+    if(c2 >= 0x80){
+	c2 = c2 + c2 - ((c2 <= 0x9f) ? SJ0162 : SJ6394);
+	if (c1 < 0x9f)
+	    c1 = c1 - ((c1 > DEL) ? SPACE : 0x1f);
+	else {
+	    c1 = c1 - 0x7e;
+	    c2++;
+	}
+    }
+
+#ifdef X0212_ENABLE
+    if (x0212_f){
+        c2 = x0212_unshift(c2);
     }
+#endif
     if (p2) *p2 = c2;
     if (p1) *p1 = c1;
     return 0;
@@ -2292,6 +2690,26 @@ e_iconv(c2, c1, c0)
 {
     if (c2 == X0201) {
 	c1 &= 0x7f;
+#ifdef X0212_ENABLE
+    }else if (c2 == 0x8f){
+        if (c0 == 0){
+            return -1;
+        }
+        c2 = (c2 << 8) | (c1 & 0x7f);
+        c1 = c0 & 0x7f;
+#ifdef SHIFTJIS_CP932
+        if (cp51932_f){
+            int s2, s1;
+            if (e2s_conv(c2, c1, &s2, &s1) == 0){
+                s2e_conv(s2, s1, &c2, &c1);
+                if ((c2 & 0xff00) == 0){
+                    c1 &= 0x7f;
+                    c2 &= 0x7f;
+                }
+            }
+        }
+#endif /* SHIFTJIS_CP932 */
+#endif /* X0212_ENABLE */
     } else if (c2 == SSO){
         c2 = X0201;
         c1 &= 0x7f;
@@ -2311,12 +2729,12 @@ w2e_conv(c2, c1, c0, p2, p1)
     int    c2, c1, c0;
     int *p2, *p1;
 {
-    extern unsigned short * utf8_to_euc_2bytes[];
-    extern unsigned short ** utf8_to_euc_3bytes[];
+    extern const unsigned short *const utf8_to_euc_2bytes[];
+    extern const unsigned short *const *const utf8_to_euc_3bytes[];
     int ret = 0;
 
     if (0xc0 <= c2 && c2 <= 0xef) {
-        unsigned short **pp;
+        const unsigned short *const *pp;
 
         if (0xe0 <= c2) {
             if (c0 == 0) return -1;
@@ -2346,7 +2764,40 @@ w_iconv(c2, c1, c0)
     int    c2,
                     c1, c0;
 {
-    int ret = w2e_conv(c2, c1, c0, &c2, &c1);
+    int ret = 0;
+    
+    if (c0 == 0){
+	if (c2 == 0) /* 0x00-0x7f */
+	    ; /* 1byte */
+	else if ((c2 & 0xe0) == 0xc0) /* 0xc0-0xdf */
+	    ; /* 2ytes */
+	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;
+    }
+    if (c2 == EOF);
+    else if (c2 == 0xef && c1 == 0xbb && c0 == 0xbf) {
+	return 0; /* throw BOM */
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+    } else if (internal_unicode_f && (output_conv == w_oconv || output_conv == w_oconv16)){
+	unsigned short val = 0;
+	if(c2 == 0){
+	    c2 = c1;
+	    c1 = 0;
+	}
+	val = ww16_conv(c2, c1, c0);
+	c2 = (val >> 8) & 0xff;
+	c1 = val & 0xff;
+#endif
+    } else {
+	ret = w2e_conv(c2, c1, c0, &c2, &c1);
+    }
     if (ret == 0){
         (*oconv)(c2, c1);
     }
@@ -2384,7 +2835,7 @@ ww16_conv(c2, c1, c0)
         val |= (c0 & 0x3f);
     }else if (c2 >= 0xc0){
         val = (c2 & 0x1f) << 6;
-        val |= (c1 & 0x3f) << 6;
+        val |= (c1 & 0x3f);
     }else{
         val = c2;
     }
@@ -2396,10 +2847,10 @@ w16e_conv(val, p2, p1)
      unsigned short val;
      int *p2, *p1;
 {
-    extern unsigned short * utf8_to_euc_2bytes[];
-    extern unsigned short ** utf8_to_euc_3bytes[];
+    extern const unsigned short *const utf8_to_euc_2bytes[];
+    extern const unsigned short *const *const utf8_to_euc_3bytes[];
     int c2, c1, c0;
-    unsigned short **pp;
+    const unsigned short *const *pp;
     int psize;
     int ret = 0;
 
@@ -2421,6 +2872,9 @@ w16e_conv(val, p2, p1)
             ret = 0;
         }
 #endif
+    }else{
+        *p2 = 0;
+        *p1 = c2;
     }
     return ret;
 }
@@ -2429,16 +2883,16 @@ int
 w_iconv16(c2, c1, c0)
     int    c2, c1,c0;
 {
-    int ret;
+    int ret = 0;
 
     if (c2==0376 && c1==0377){
-	utf16_mode = UTF16LE_INPUT;
+	utf16_mode = UTF16BE_INPUT;
 	return 0;    
     } else if (c2==0377 && c1==0376){
-	utf16_mode = UTF16BE_INPUT;
+	utf16_mode = UTF16LE_INPUT;
 	return 0;    
     }
-    if (c2 != EOF && utf16_mode == UTF16BE_INPUT) {
+    if (c2 != EOF && utf16_mode == UTF16LE_INPUT) {
 	int tmp;
 	tmp=c1; c1=c2; c2=tmp;
     }
@@ -2446,7 +2900,10 @@ w_iconv16(c2, c1, c0)
 	(*oconv)(c2, c1);
 	return 0;
     }
-    ret = w16e_conv(((c2<<8)&0xff00) + c1, &c2, &c1);
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+    if (internal_unicode_f && (output_conv == w_oconv || output_conv == w_oconv16));
+#endif
+    else ret = w16e_conv(((c2<<8)&0xff00) + c1, &c2, &c1);
     if (ret) return ret;
     (*oconv)(c2, c1);
     return 0;
@@ -2455,14 +2912,21 @@ w_iconv16(c2, c1, c0)
 int
 w_iconv_common(c1, c0, pp, psize, p2, p1)
     int    c1,c0;
-    unsigned short **pp;
+    const unsigned short *const *pp;
     int psize;
     int *p2, *p1;
 {
     int c2;
-    unsigned short *p ;
+    const unsigned short *p;
     unsigned short val;
 
+    /* CP932/CP51932: U+00A6 (BROKEN BAR) -> not 0x8fa2c3, but 0x7c */
+    if (cp51932_f && c1 == 0xC2 && c0 == 0xA6){
+	if (p2) *p2 = 0;
+	if (p1) *p1 = 0x7C;
+	return 0;
+    }
+
     if (pp == 0) return 1;
 
     c1 -= 0x80;
@@ -2476,6 +2940,10 @@ w_iconv_common(c1, c0, pp, psize, p2, p1)
     if (val == 0) return 1;
 
     c2 = val >> 8;
+    if (val & 0x8000){
+        c2 &= 0x7f;
+        c2 |= 0x8f00;
+    }
     if (c2 == SO) c2 = X0201;
     c1 = val & 0x7f;
     if (p2) *p2 = c2;
@@ -2486,17 +2954,139 @@ w_iconv_common(c1, c0, pp, psize, p2, p1)
 #endif
 
 #ifdef UTF8_OUTPUT_ENABLE
+void
+nkf_each_char_to_hex(f, c)
+    void (*f)PROTO((int c2,int c1));
+    int c;
+{
+    const char *hex = "0123456789ABCDEF";
+    c &= 0x00FFFFFF;
+    int shift = 20;
+    while(shift >= 0){
+	if(c >= 1<<shift){
+	    while(shift >= 0){
+		(*f)(0, hex[(c>>shift)&0xF]);
+		shift -= 4;
+	    }
+	}else{
+	    shift -= 4;
+	}
+    }
+    return;
+}
+
+void
+encode_fallback_html(c)
+    int c;
+{
+    (*oconv)(0, '&');
+    (*oconv)(0, '#');
+    c &= 0x00FFFFFF;
+    if(c >= 1000000)
+	(*oconv)(0, 0x30+(c/1000000)%10);
+    if(c >= 100000)
+	(*oconv)(0, 0x30+(c/100000 )%10);
+    if(c >= 10000)
+	(*oconv)(0, 0x30+(c/10000  )%10);
+    if(c >= 1000)
+	(*oconv)(0, 0x30+(c/1000   )%10);
+    if(c >= 100)
+	(*oconv)(0, 0x30+(c/100    )%10);
+    if(c >= 10)
+	(*oconv)(0, 0x30+(c/10     )%10);
+    if(c >= 0)
+	(*oconv)(0, 0x30+ c         %10);
+    (*oconv)(0, ';');
+    return;
+}
+
+void
+encode_fallback_xml(c)
+    int c;
+{
+    (*oconv)(0, '&');
+    (*oconv)(0, '#');
+    (*oconv)(0, 'x');
+    nkf_each_char_to_hex(oconv, c);
+    (*oconv)(0, ';');
+    return;
+}
+
+void
+encode_fallback_java(c)
+    int c;
+{
+    const char *hex = "0123456789ABCDEF";
+    (*oconv)(0, '\\');
+    if((c&0x00FFFFFF) > 0xFFFF){
+	(*oconv)(0, 'U');
+	(*oconv)(0, '0');
+	(*oconv)(0, '0');
+	(*oconv)(0, hex[(c>>20)&0xF]);
+	(*oconv)(0, hex[(c>>16)&0xF]);
+    }else{
+	(*oconv)(0, 'u');
+    }
+    (*oconv)(0, hex[(c>>12)&0xF]);
+    (*oconv)(0, hex[(c>> 8)&0xF]);
+    (*oconv)(0, hex[(c>> 4)&0xF]);
+    (*oconv)(0, hex[ c     &0xF]);
+    return;
+}
+
+void
+encode_fallback_perl(c)
+    int c;
+{
+    (*oconv)(0, '\\');
+    (*oconv)(0, 'x');
+    (*oconv)(0, '{');
+    nkf_each_char_to_hex(oconv, c);
+    (*oconv)(0, '}');
+    return;
+}
+
+void
+encode_fallback_subchar(c)
+    int c;
+{
+    c = unicode_subchar;
+    (*oconv)((c>>8)&0xFF, c&0xFF);
+    return;
+    int shift = 16;
+    while(shift >= 0){
+	if(c >= 1<<shift){
+	    while(shift >= 0){
+		(*oconv)(0, (c>>shift)&0xFF);
+		shift -= 8;
+	    }
+	}else{
+	    shift -= 8;
+	}
+    }
+    return;
+}
+
 int
 e2w_conv(c2, c1)
     int    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;
+    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[];
+    const unsigned short *p;
 
     if (c2 == X0201) {
         p = euc_to_utf8_1byte;
+#ifdef X0212_ENABLE
+    } else if (c2 >> 8 == 0x8f){
+        extern const unsigned short *const x0212_to_utf8_2bytes[];
+        c2 = (c2&0x7f) - 0x21;
+        if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
+	    p = x0212_to_utf8_2bytes[c2];
+        else
+            return 0;
+#endif
     } else {
         c2 &= 0x7f;
         c2 = (c2&0x7f) - 0x21;
@@ -2518,16 +3108,7 @@ w_oconv(c2, c1)
                     c1;
 {
     int c0;
-#ifdef NUMCHAR_OPTION
-    if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
-        w16w_conv(c1, &c2, &c1, &c0);
-        (*o_putc)(c2);
-        if (c1){
-            (*o_putc)(c1);
-            if (c0) (*o_putc)(c0);
-        }
-    }
-#endif
+    unsigned short val;
     if (c2 == EOF) {
         (*o_putc)(EOF);
         return;
@@ -2540,6 +3121,18 @@ w_oconv(c2, c1)
 	unicode_bom_f=1;
     }
 
+#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);
+        }
+        return;
+    }
+#endif
+
     if (c2 == 0) { 
 	output_mode = ASCII;
         (*o_putc)(c1);
@@ -2548,11 +3141,16 @@ w_oconv(c2, c1)
         (*o_putc)(c1 | 0x080);
     } else {
         output_mode = UTF8;
-        w16w_conv((unsigned short)e2w_conv(c2, c1), &c2, &c1, &c0);
-        (*o_putc)(c2);
-        if (c1){
-            (*o_putc)(c1);
-            if (c0) (*o_putc)(c0);
+	if (internal_unicode_f && (iconv == w_iconv || iconv == w_iconv16))
+	    val = ((c2<<8)&0xff00) + c1;
+	else val = e2w_conv(c2, c1);
+        if (val){
+            w16w_conv(val, &c2, &c1, &c0);
+            (*o_putc)(c2);
+            if (c1){
+                (*o_putc)(c1);
+                if (c0) (*o_putc)(c0);
+            }
         }
     }
 }
@@ -2578,7 +3176,8 @@ w_oconv16(c2, c1)
 	unicode_bom_f=1;
     }
 
-    if (c2 == ISO8859_1) {
+    if (internal_unicode_f && (iconv == w_iconv || iconv == w_iconv16)){
+    } else if (c2 == ISO8859_1) {
         c2 = 0;
         c1 |= 0x80;
 #ifdef NUMCHAR_OPTION
@@ -2587,7 +3186,7 @@ w_oconv16(c2, c1)
         c1 &= 0xff;
 #endif
     } else if (c2) {
-        unsigned short val = (unsigned short)e2w_conv(c2, c1);
+        unsigned short val = e2w_conv(c2, c1);
         c2 = (val >> 8) & 0xff;
         c1 = val & 0xff;
     }
@@ -2610,6 +3209,10 @@ e_oconv(c2, c1)
 #ifdef NUMCHAR_OPTION
     if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
         w16e_conv(c1, &c2, &c1);
+        if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+	    if(encode_fallback)(*encode_fallback)(c1);
+            return;
+        }
     }
 #endif
     if (c2 == EOF) {
@@ -2624,6 +3227,31 @@ e_oconv(c2, c1)
     } else if (c2 == ISO8859_1) {
 	output_mode = ISO8859_1;
         (*o_putc)(c1 | 0x080);
+#ifdef X0212_ENABLE
+    } else if ((c2 & 0xff00) >> 8 == 0x8f){
+	output_mode = JAPANESE_EUC;
+#ifdef SHIFTJIS_CP932
+        if (cp51932_f){
+            int s2, s1;
+            if (e2s_conv(c2, c1, &s2, &s1) == 0){
+                s2e_conv(s2, s1, &c2, &c1);
+            }
+        }
+#endif
+        if (c2 == 0) {
+	    output_mode = ASCII;
+	    (*o_putc)(c1);
+	}else if ((c2 & 0xff00) >> 8 == 0x8f){
+            if (x0212_f){
+                (*o_putc)(0x8f);
+                (*o_putc)((c2 & 0x7f) | 0x080);
+                (*o_putc)(c1 | 0x080);
+            }
+        }else{
+            (*o_putc)((c2 & 0x7f) | 0x080);
+            (*o_putc)(c1 | 0x080);
+        }
+#endif
     } else {
         if ((c1<0x21 || 0x7e<c1) ||
            (c2<0x21 || 0x7e<c2)) {
@@ -2636,12 +3264,71 @@ e_oconv(c2, c1)
     }
 }
 
-void
+#ifdef X0212_ENABLE
+int x0212_shift(c)
+     int c;
+{
+    int ret = c;
+    c &= 0x7f;
+    if ((ret & 0xff00) == 0x8f00){
+        if (0x75 <= c && c <= 0x7f){
+            ret = c + (0x109 - 0x75);
+        }
+    }else{
+        if (0x75 <= c && c <= 0x7f){
+            ret = c + (0x113 - 0x75);
+        }
+    }
+    return ret;
+}
+
+
+int x0212_unshift(c)
+     int c;
+{
+    int ret = c;
+    if (0x7f <= c && c <= 0x88){
+        ret = c + (0x75 - 0x7f);
+    }else if (0x89 <= c && c <= 0x92){
+        ret = (0x8f << 8) | 0x80 | (c + (0x75 - 0x89));
+    }
+    return ret;
+}
+#endif /* X0212_ENABLE */
+
+int
 e2s_conv(c2, c1, p2, p1)
      int c2, c1, *p2, *p1;
 {
+#ifdef X0212_ENABLE
+    int val = 0;
+    const unsigned short *ptr;
+    int ndx;
+    extern const unsigned short *const x0212_shiftjis[];
+    if ((c2 & 0xff00) == 0x8f00){
+        ndx = c2 & 0x7f;
+        if (0x21 <= ndx && ndx <= 0x7e){
+            ptr = x0212_shiftjis[ndx - 0x21];
+            if (ptr){
+                val = ptr[(c1 & 0x7f) - 0x21];
+            }
+            if (val){
+                c2 = val >> 8;
+                c1 = val & 0xff;
+                if (p2) *p2 = c2;
+                if (p1) *p1 = c1;
+                return 0;
+            }
+        }
+        c2 = x0212_shift(c2);
+    }
+#endif /* X0212_ENABLE */
+    if ((c2 & 0xff00) == 0x8f00){
+        return 1;
+    }
     if (p2) *p2 = ((c2 - 1) >> 1) + ((c2 <= 0x5e) ? 0x71 : 0xb1);
     if (p1) *p1 = c1 + ((c2 & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
+    return 0;
 }
 
 void
@@ -2652,6 +3339,10 @@ s_oconv(c2, c1)
 #ifdef NUMCHAR_OPTION
     if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
         w16e_conv(c1, &c2, &c1);
+        if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+	    if(encode_fallback)(*encode_fallback)(c1);
+            return;
+        }
     }
 #endif
     if (c2 == EOF) {
@@ -2666,6 +3357,14 @@ s_oconv(c2, c1)
     } else if (c2 == ISO8859_1) {
 	output_mode = ISO8859_1;
         (*o_putc)(c1 | 0x080);
+#ifdef X0212_ENABLE
+    } else if ((c2 & 0xff00) >> 8 == 0x8f){
+	output_mode = SHIFT_JIS;
+        if (e2s_conv(c2, c1, &c2, &c1) == 0){
+            (*o_putc)(c2);
+            (*o_putc)(c1);
+        }
+#endif
     } else {
         if ((c1<0x20 || 0x7e<c1) ||
            (c2<0x20 || 0x7e<c2)) {
@@ -2678,7 +3377,7 @@ s_oconv(c2, c1)
 #ifdef SHIFTJIS_CP932
         if (cp932inv_f
             && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
-            extern unsigned short cp932inv[2][189];
+            extern const unsigned short cp932inv[2][189];
             int c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
             if (c){
                 c2 = c >> 8;
@@ -2701,8 +3400,12 @@ j_oconv(c2, c1)
                     c1;
 {
 #ifdef NUMCHAR_OPTION
-    if ((c1 & CLASS_MASK) == CLASS_UTF16){
+    if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
         w16e_conv(c1, &c2, &c1);
+        if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
+	    if(encode_fallback)(*encode_fallback)(c1);
+            return;
+        }
     }
 #endif
     if (c2 == EOF) {
@@ -2713,6 +3416,18 @@ j_oconv(c2, c1)
 	    output_mode = ASCII;
         }
         (*o_putc)(EOF);
+#ifdef X0212_ENABLE
+    } else if ((c2 & 0xff00) >> 8 == 0x8f){
+        if (output_mode!=X0212) {
+            output_mode = X0212;
+            (*o_putc)(ESC);
+            (*o_putc)('$');
+            (*o_putc)('(');
+            (*o_putc)('D');
+        }
+        (*o_putc)(c2 & 0x7f);
+        (*o_putc)(c1);
+#endif
     } else if (c2==X0201) {
         if (output_mode!=X0201) {
             output_mode = X0201;
@@ -2756,21 +3471,14 @@ base64_conv(c2, c1)
     int    c2,
                     c1;
 {
-    if (base64_count>50 && !mimeout_mode && c2==0 && c1==SPACE) {
-	(*o_putc)(EOF);
-	(*o_putc)(NL);
-    } else if (base64_count>66 && mimeout_mode) {
-	(*o_base64conv)(EOF,0);
-	(*o_base64conv)(NL,0);
-	(*o_base64conv)(SPACE,0);
-    }
+    mime_prechar(c2, c1);
     (*o_base64conv)(c2,c1);
 }
 
 
-static int broken_buf[3];
-static int broken_counter = 0;
-static int broken_last = 0;
+STATIC int broken_buf[3];
+STATIC int broken_counter = 0;
+STATIC int broken_last = 0;
 int
 broken_getc(f)
 FILE *f;
@@ -2821,7 +3529,7 @@ FILE *f;
     return c;
 }
 
-static int prev_cr = 0;
+STATIC int prev_cr = 0;
 
 void
 cr_conv(c2,c1) 
@@ -2953,7 +3661,7 @@ int c2,c1;
         if (f_line<=fold_len) {   /* normal case */
             fold_state = 1;
         } else {
-            if (f_line>=fold_len+fold_margin) { /* too many kinsou suspension */
+            if (f_line>fold_len+fold_margin) { /* too many kinsoku suspension */
                 f_line = char_size(c2,c1);
                 fold_state =  '\n';       /* We can't wait, do fold now */
             } else if (c2==X0201) {
@@ -3167,7 +3875,7 @@ void
 iso2022jp_check_conv(c2,c1)
 int    c2, c1;
 {
-    static int range[RANGE_NUM_MAX][2] = {
+    STATIC const int range[RANGE_NUM_MAX][2] = {
         {0x222f, 0x2239,},
         {0x2242, 0x2249,},
         {0x2251, 0x225b,},
@@ -3214,19 +3922,19 @@ int    c2, c1;
 
 /* This converts  =?ISO-2022-JP?B?HOGE HOGE?= */
 
-unsigned char *mime_pattern[] = {
-   (unsigned char *)"\075?EUC-JP?B?",
-   (unsigned char *)"\075?SHIFT_JIS?B?",
-   (unsigned char *)"\075?ISO-8859-1?Q?",
-   (unsigned char *)"\075?ISO-8859-1?B?",
-   (unsigned char *)"\075?ISO-2022-JP?B?",
-   (unsigned char *)"\075?ISO-2022-JP?Q?",
+const unsigned char *mime_pattern[] = {
+    (const unsigned char *)"\075?EUC-JP?B?",
+    (const unsigned char *)"\075?SHIFT_JIS?B?",
+    (const unsigned char *)"\075?ISO-8859-1?Q?",
+    (const unsigned char *)"\075?ISO-8859-1?B?",
+    (const unsigned char *)"\075?ISO-2022-JP?B?",
+    (const 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?",
+    (const unsigned char *)"\075?UTF-8?B?",
+    (const unsigned char *)"\075?UTF-8?Q?",
 #endif
-   (unsigned char *)"\075?US-ASCII?Q?",
-   NULL
+    (const unsigned char *)"\075?US-ASCII?Q?",
+    NULL
 };
 
 
@@ -3239,7 +3947,7 @@ int (*mime_priority_func[])PROTO((int c2, int c1, int c0)) = {
     0,
 };
 
-int      mime_encode[] = {
+const int mime_encode[] = {
     JAPANESE_EUC, SHIFT_JIS,ISO8859_1, ISO8859_1, X0208, X0201,
 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
     UTF8, UTF8,
@@ -3248,7 +3956,7 @@ int      mime_encode[] = {
     0
 };
 
-int      mime_encode_method[] = {
+const int mime_encode_method[] = {
     'B', 'B','Q', 'B', 'B', 'Q',
 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
     'B', 'Q',
@@ -3260,11 +3968,6 @@ int      mime_encode_method[] = {
 
 #define MAXRECOVER 20
 
-/* I don't trust portablity of toupper */
-#define nkf_toupper(c)  (('a'<=c && c<='z')?(c-('a'-'A')):c)
-#define nkf_isdigit(c)  ('0'<=c && c<='9')
-#define nkf_isxdigit(c)  (nkf_isdigit(c) || ('a'<=c && c<='f') || ('A'<=c && c <= 'F'))
-
 void
 switch_mime_getc()
 {
@@ -3287,6 +3990,8 @@ unswitch_mime_getc()
     }
     i_getc = i_mgetc;
     i_ungetc = i_mungetc;
+    if(mime_iconv_back)set_iconv(FALSE, mime_iconv_back);
+    mime_iconv_back = NULL;
 }
 
 int
@@ -3295,7 +4000,7 @@ FILE *f;
 {
     int c1 = 0;
     int i,j,k;
-    unsigned char *p,*q;
+    const unsigned char *p,*q;
     int r[MAXRECOVER];    /* recovery buffer, max mime pattern lenght */
 
     mime_decode_mode = FALSE;
@@ -3324,6 +4029,8 @@ FILE *f;
     }
     mime_decode_mode = p[i-2];
 
+    mime_iconv_back = iconv;
+    set_iconv(FALSE, mime_priority_func[j]);
     clr_code_score(find_inputcode_byfunc(mime_priority_func[j]), SCORE_iMIME);
 
     if (mime_decode_mode=='B') {
@@ -3427,7 +4134,7 @@ no_putc(c)
 }
 
 void debug(str)
-     char *str;
+    const char *str;
 {
     if (debug_f){
         fprintf(stderr, "%s\n", str);
@@ -3450,6 +4157,7 @@ set_input_codename (codename)
     is_inputcode_set = TRUE;
 }
 
+#if !defined(PERL_XS) && !defined(WIN32DLL)
 void
 print_guessed_code (filename)
     char *filename;
@@ -3465,14 +4173,7 @@ print_guessed_code (filename)
     if (filename != NULL) printf("%s:", filename);
     printf("%s\n", codename);
 }
-
-int
-hex2bin(x)
-     int x;
-{
-    if (nkf_isdigit(x)) return x - '0';
-    return nkf_toupper(x) - 'A' + 10;
-}
+#endif /*WIN32DLL*/
 
 #ifdef INPUT_OPTION 
 
@@ -3602,6 +4303,57 @@ numchar_ungetc(c, f)
 }
 #endif
 
+#ifdef UNICODE_NORMALIZATION
+
+/* Normalization Form C */
+int
+nfc_getc(f)
+     FILE *f;
+{
+    int (*g)() = i_nfc_getc;
+    int (*u)() = i_nfc_ungetc;
+    int i=0, j, k=1, lower, upper;
+    int buf[9];
+    const int *array = NULL;
+    extern const struct normalization_pair normalization_table[];
+    
+    buf[i] = (*g)(f);
+    while (k > 0 && ((buf[i] & 0xc0) != 0x80)){
+	lower=0, upper=NORMALIZATION_TABLE_LENGTH-1;
+	while (upper >= lower) {
+	    j = (lower+upper) / 2;
+	    array = normalization_table[j].nfd;
+	    for (k=0; k < NORMALIZATION_TABLE_NFD_LENGTH && array[k]; k++){
+		if (array[k] != buf[k]){
+		    array[k] < buf[k] ? (lower = j + 1) : (upper = j - 1);
+		    k = 0;
+		    break;
+		} else if (k >= i)
+		    buf[++i] = (*g)(f);
+	    }
+	    if (k > 0){
+		array = normalization_table[j].nfc;
+		for (i=0; i < NORMALIZATION_TABLE_NFC_LENGTH && array[i]; i++)
+		    buf[i] = array[i];
+		i--;
+		break;
+	    }
+	}
+	while (i > 0)
+	    (*u)(buf[i--], f);
+    }
+    return buf[0];
+}
+
+int
+nfc_ungetc(c, f)
+     int c;
+     FILE *f;
+{
+    return (*i_nfc_ungetc)(c, f);
+}
+#endif /* UNICODE_NORMALIZATION */
+
 
 int 
 mime_getc(f)
@@ -3631,12 +4383,15 @@ FILE *f;
         if ((c1 = (*i_mgetc)(f)) == EOF) return (EOF);
 restart_mime_q:
         if (c1=='_') return ' ';
+	if (c1<=' ' || DEL<=c1) {
+	    mime_decode_mode = exit_mode; /* prepare for quit */
+	    return c1;
+	}
         if (c1!='=' && c1!='?') {
 	    return c1;
 	}
                 
         mime_decode_mode = exit_mode; /* prepare for quit */
-        if (c1<=' ') return c1;
         if ((c2 = (*i_mgetc)(f)) == EOF) return (EOF);
         if (c1=='?'&&c2=='=' && mimebuf_f != FIXED_MIME) {
             /* end Q encoding */
@@ -3877,8 +4632,8 @@ FILE  *f;
 
 int
 mime_integrity(f,p)
-FILE *f;
-unsigned char *p;
+    FILE *f;
+    const unsigned char *p;
 {
     int c,d;
     unsigned int q;
@@ -3886,6 +4641,7 @@ unsigned char *p;
      */
     mime_input = mime_top;
     mime_last = mime_top;
+    
     while(*p) Fifo(mime_input++) = *p++;
     d = 0;
     q = mime_input;
@@ -3936,10 +4692,10 @@ base64decode(c)
     return (i);
 }
 
-static char basis_64[] =
+STATIC const char basis_64[] =
    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
 
-static int b64c;
+STATIC int b64c;
 #define MIMEOUT_BUF_LENGTH (60)
 char mimeout_buf[MIMEOUT_BUF_LENGTH+1];
 int mimeout_buf_count = 0;
@@ -3950,7 +4706,7 @@ void
 open_mime(mode)
 int mode;
 {
-    unsigned char *p;
+    const unsigned char *p;
     int i;
     int j;
     p  = mime_pattern[0];
@@ -3964,6 +4720,10 @@ int mode;
     
     i = 0;
     if (base64_count>45) {
+	if (mimeout_buf_count>0 && nkf_isblank(mimeout_buf[i])){
+            (*o_mputc)(mimeout_buf[i]);
+	    i++;
+	}
 	(*o_mputc)(NL);
 	(*o_mputc)(SPACE);
 	base64_count = 1;
@@ -4039,7 +4799,13 @@ mimeout_addchar(c)
 {
     switch(mimeout_mode) {
     case 'Q':
-	if(c>=DEL) {
+	if(c==SPACE){
+	    (*o_mputc)('_');
+	    base64_count++;
+	} else if (c==CR||c==NL) {
+	    (*o_mputc)(c);
+	    base64_count = 0;
+	} else if(c<SPACE||c=='='||c=='?'||c=='_'||DEL<=c) {
 	    (*o_mputc)('=');
 	    (*o_mputc)(itoh4(((c>>4)&0xf)));
 	    (*o_mputc)(itoh4((c&0xf)));
@@ -4067,109 +4833,84 @@ mimeout_addchar(c)
         mimeout_mode='B';
         base64_count += 2;
         break;
+    default:
+	(*o_mputc)(c);
+	base64_count++;
+        break;
     }
 }
 
+int mime_lastchar2, mime_lastchar1;
+
+void mime_prechar(c2, c1)
+     int c2, c1;
+{
+    if (mimeout_mode){
+        if (c2){
+            if (base64_count + mimeout_buf_count/3*4> 66){
+                (*o_base64conv)(EOF,0);
+                (*o_base64conv)(0,NL);
+                (*o_base64conv)(0,SPACE);
+            }
+        }/*else if (mime_lastchar2){
+            if (c1 <=DEL && !nkf_isspace(c1)){
+                (*o_base64conv)(0,SPACE);
+            }
+        }*/
+    }/*else{
+        if (c2 && mime_lastchar2 == 0
+            && mime_lastchar1 && !nkf_isspace(mime_lastchar1)){
+            (*o_base64conv)(0,SPACE);
+        }
+    }*/
+    mime_lastchar2 = c2;
+    mime_lastchar1 = c1;
+}
+
 void
 mime_putc(c)
     int            c;
 {
     int i = 0;
     int j = 0;
-    
-    if (mimeout_f==FIXED_MIME && base64_count>50) {
-	eof_mime();
-	(*o_mputc)(NL);
-	base64_count=0;
-    } else if (c==CR||c==NL) {
-	base64_count=0;
-    }
-    if (c!=EOF && mimeout_f!=FIXED_MIME) {
-        if ( c<=DEL &&(output_mode==ASCII ||output_mode == ISO8859_1 ) ) {
-	    if (mimeout_mode=='Q') {
-		if (c<=SPACE) {
-		    close_mime();
-		    (*o_mputc)(SPACE);
-		    base64_count++;
-		}
-		(*o_mputc)(c);
-		base64_count++;
-		return;
-	    } else if (mimeout_mode) {
-		if (base64_count>63) {
-		    eof_mime();
-		    (*o_mputc)(NL);
-		    (*o_mputc)(SPACE);
-		    base64_count=1;
-		    mimeout_preserve_space = TRUE;
-		}
-		if (c==SPACE || c==TAB || c==CR || c==NL) {
-		    for (i=0;i<mimeout_buf_count;i++) {
-			if (SPACE<mimeout_buf[i] && mimeout_buf[i]<DEL) {
-			    eof_mime();
-			    for (i=0;i<mimeout_buf_count;i++) {
-				(*o_mputc)(mimeout_buf[i]);
-				base64_count++;
-			    }
-			    mimeout_buf_count = 0;
-			}
-		    }
-		    mimeout_buf[mimeout_buf_count++] = c;
-		    if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
-		    	eof_mime();
-			base64_count = 0;
-			for (i=0;i<mimeout_buf_count;i++) {
-			    (*o_mputc)(mimeout_buf[i]);
-			    base64_count++;
-			}
-		    }
-		    return;
-		}
-		if (mimeout_buf_count>0 && SPACE<c) {
-		    mimeout_buf[mimeout_buf_count++] = c;
-		    if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
-		    } else {
-			return;
-		    }
-		}
-	    } else if (!mimeout_mode) {
-		if (c==SPACE || c==TAB || c==CR || c==NL) {
-		    if ((c==CR || c==NL)
-			&&(mimeout_buf[mimeout_buf_count-1]==SPACE
-			   || mimeout_buf[mimeout_buf_count-1]==TAB)) {
-			mimeout_buf_count--;
-		    }
-		    for (i=0;i<mimeout_buf_count;i++) {
-			(*o_mputc)(mimeout_buf[i]);
-			base64_count++;
-		    }
-		    mimeout_buf_count = 0;
-		}
-		mimeout_buf[mimeout_buf_count++] = c;
-		if (mimeout_buf_count>75) {
-		    open_mime(output_mode);
-		}
-		return;
-	    }
-        } else if (!mimeout_mode) {
-	    if (mimeout_buf_count>0 && mimeout_buf[mimeout_buf_count-1]==SPACE) {
-		for (i=0;i<mimeout_buf_count-1;i++) {
-		    (*o_mputc)(mimeout_buf[i]);
-		    base64_count++;
-		}
-		mimeout_buf[0] = SPACE;
-		mimeout_buf_count = 1;
-	    }
-	    open_mime(output_mode);
+    int lastchar;
+
+    if (mimeout_f == FIXED_MIME){
+        if (mimeout_mode == 'Q'){
+            if (base64_count > 71){
+                if (c!=CR && c!=NL) {
+                    (*o_mputc)('=');
+                    (*o_mputc)(NL);
+                }
+                base64_count = 0;
+            }
+        }else{
+            if (base64_count > 71){
+                eof_mime();
+                (*o_mputc)(NL);
+                base64_count = 0;
+            }
+            if (c == EOF) { /* c==EOF */
+                eof_mime();
+            }
+        }
+        if (c != EOF) { /* c==EOF */
+            mimeout_addchar(c);
         }
-    } else { /* c==EOF */
+        return;
+    }
+    
+    /* mimeout_f != FIXED_MIME */
+
+    if (c == EOF) { /* c==EOF */
 	j = mimeout_buf_count;
+	mimeout_buf_count = 0;
 	i = 0;
 	for (;i<j;i++) {
-	    if (mimeout_buf[i]==SPACE || mimeout_buf[i]==TAB
-		|| mimeout_buf[i]==CR || mimeout_buf[i]==NL)
+	    /*if (nkf_isspace(mimeout_buf[i])){
 		break;
-	    (*mime_putc)(mimeout_buf[i]);
+            }*/
+	    mimeout_addchar(mimeout_buf[i]);
 	}
         eof_mime();
 	for (;i<j;i++) {
@@ -4178,19 +4919,151 @@ mime_putc(c)
 	}
         return;
     }
-    
+
+    if (mimeout_mode=='Q') {
+        if (c <= DEL && (output_mode==ASCII ||output_mode == ISO8859_1 ) ) {
+            if (c <= SPACE) {
+                close_mime();
+                (*o_mputc)(SPACE);
+                base64_count++;
+            }
+            (*o_mputc)(c);
+            base64_count++;
+        }
+        return;
+    }
+
+    if (mimeout_buf_count > 0){
+        lastchar = mimeout_buf[mimeout_buf_count - 1];
+    }else{
+        lastchar = -1;
+    }
+
+    if (!mimeout_mode) {
+        if (c <= DEL && (output_mode==ASCII ||output_mode == ISO8859_1)) {
+            if (nkf_isspace(c)) {
+                if (c==CR || c==NL) {
+                    base64_count=0;
+                }
+                for (i=0;i<mimeout_buf_count;i++) {
+                    (*o_mputc)(mimeout_buf[i]);
+                    if (mimeout_buf[i] == CR || mimeout_buf[i] == NL){
+                        base64_count = 0;
+                    }else{
+                        base64_count++;
+                    }
+                }
+                mimeout_buf[0] = c;
+                mimeout_buf_count = 1;
+            }else{
+                if (base64_count > 1
+                    && base64_count + mimeout_buf_count > 76){
+                    (*o_mputc)(NL);
+                    base64_count = 0;
+                    if (!nkf_isspace(mimeout_buf[0])){
+                        (*o_mputc)(SPACE);
+                        base64_count++;
+                    }
+                }
+                mimeout_buf[mimeout_buf_count++] = c;
+                if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
+                    open_mime(output_mode);
+                }
+            }
+            return;
+        }else{
+            if (lastchar==CR || lastchar == NL){
+                for (i=0;i<mimeout_buf_count;i++) {
+                    (*o_mputc)(mimeout_buf[i]);
+                }
+                base64_count = 0;
+                mimeout_buf_count = 0;
+            }
+            if (lastchar==SPACE) {
+                for (i=0;i<mimeout_buf_count-1;i++) {
+                    (*o_mputc)(mimeout_buf[i]);
+                    base64_count++;
+                }
+                mimeout_buf[0] = SPACE;
+                mimeout_buf_count = 1;
+            }
+            open_mime(output_mode);
+        }
+    }else{
+        /* mimeout_mode == 'B', 1, 2 */
+        if ( c<=DEL && (output_mode==ASCII ||output_mode == ISO8859_1 ) ) {
+            if (lastchar == CR || lastchar == NL){
+                if (nkf_isblank(c)) {
+                    for (i=0;i<mimeout_buf_count;i++) {
+                        mimeout_addchar(mimeout_buf[i]);
+                    }
+                    mimeout_buf_count = 0;
+                } else if (SPACE<c && c<DEL) {
+                    eof_mime();
+                    for (i=0;i<mimeout_buf_count;i++) {
+                        (*o_mputc)(mimeout_buf[i]);
+                    }
+                    base64_count = 0;
+                    mimeout_buf_count = 0;
+                }
+            }
+            if (c==SPACE || c==TAB || c==CR || c==NL) {
+                for (i=0;i<mimeout_buf_count;i++) {
+                    if (SPACE<mimeout_buf[i] && mimeout_buf[i]<DEL) {
+                        eof_mime();
+                        for (i=0;i<mimeout_buf_count;i++) {
+                            (*o_mputc)(mimeout_buf[i]);
+                            base64_count++;
+                        }
+                        mimeout_buf_count = 0;
+                    }
+                }
+                mimeout_buf[mimeout_buf_count++] = c;
+                if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
+                    eof_mime();
+                    for (i=0;i<mimeout_buf_count;i++) {
+                        (*o_mputc)(mimeout_buf[i]);
+                        base64_count++;
+                    }
+                    mimeout_buf_count = 0;
+                }
+                return;
+            }
+            if (mimeout_buf_count>0 && SPACE<c && c!='=') {
+                mimeout_buf[mimeout_buf_count++] = c;
+                if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
+                    j = mimeout_buf_count;
+                    mimeout_buf_count = 0;
+                    for (i=0;i<j;i++) {
+                        mimeout_addchar(mimeout_buf[i]);
+                    }
+                }
+                return;
+            }
+        }
+    }
     if (mimeout_buf_count>0) {
 	j = mimeout_buf_count;
 	mimeout_buf_count = 0;
 	for (i=0;i<j;i++) {
+	    if (mimeout_buf[i]==CR || mimeout_buf[i]==NL)
+		break;
 	    mimeout_addchar(mimeout_buf[i]);
 	}
+	if (i<j) {
+	    eof_mime();
+	    base64_count=0;
+	    for (;i<j;i++) {
+		(*o_mputc)(mimeout_buf[i]);
+	    }
+	    open_mime(output_mode);
+	}
     }
     mimeout_addchar(c);
 }
 
 
-#ifdef PERL_XS
+#if defined(PERL_XS) || defined(WIN32DLL)
 void 
 reinit()
 {
@@ -4209,6 +5082,7 @@ reinit()
     input_f = FALSE;
     alpha_f = FALSE;
     mime_f = STRICT_MIME;
+    mime_decode_f = FALSE;
     mimebuf_f = FALSE;
     broken_f = FALSE;
     iso8859_f = FALSE;
@@ -4219,10 +5093,18 @@ reinit()
      x0201_f = NO_X0201;
 #endif
     iso2022jp_f = FALSE;
+#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE)
+    internal_unicode_f = FALSE;
+#endif
 #ifdef UTF8_OUTPUT_ENABLE
     unicode_bom_f = 0;
     w_oconv16_LE = 0;
     ms_ucs_map_f = FALSE;
+    encode_fallback = NULL;
+    unicode_subchar  = '?';
+#endif
+#ifdef UNICODE_NORMALIZATION
+    nfc_f = FALSE;
 #endif
 #ifdef INPUT_OPTION
     cap_f = FALSE;
@@ -4240,9 +5122,12 @@ reinit()
     exec_f = 0;
 #endif
 #ifdef SHIFTJIS_CP932
-    cp932_f = TRUE;
+    cp51932_f = TRUE;
     cp932inv_f = TRUE;
 #endif
+#ifdef X0212_ENABLE
+    x0212_f = FALSE;
+#endif
     {
         int i;
         for (i = 0; i < 256; i++){
@@ -4250,7 +5135,7 @@ reinit()
         }
     }
 #ifdef UTF8_INPUT_ENABLE
-    utf16_mode = UTF16LE_INPUT;
+    utf16_mode = UTF16BE_INPUT;
 #endif
     mimeout_buf_count = 0;
     mimeout_mode = 0;
@@ -4292,7 +5177,13 @@ reinit()
     broken_counter = 0;
     broken_last = 0;
     z_prev2=0,z_prev1=0;
-
+#ifdef CHECK_OPTION
+    iconv_for_check = 0;
+#endif
+    input_codename = "";
+#ifdef WIN32DLL
+    reinitdll();
+#endif /*WIN32DLL*/
 }
 #endif
 
@@ -4309,9 +5200,13 @@ int c2,c1,c0;
 {
     fprintf(stderr,"nkf internal module connection failure.\n");
     exit(1);
+    return 0; /* LINT */
 }
 
 #ifndef PERL_XS
+#ifdef WIN32DLL
+#define fprintf dllprintf
+#endif
 void 
 usage()   
 {
@@ -4319,16 +5214,16 @@ usage()
     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), AT&T JIS (EUC), UTF-8\n");
+    fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit, Shift JIS (DEFAULT), AT&T JIS (EUC), UTF-8N\n");
 #endif
 #ifdef DEFAULT_CODE_JIS
-    fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit (DEFAULT), Shift JIS, AT&T JIS (EUC), UTF-8\n");
+    fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit (DEFAULT), Shift JIS, AT&T JIS (EUC), UTF-8N\n");
 #endif
 #ifdef DEFAULT_CODE_EUC
-    fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC) (DEFAULT), UTF-8\n");
+    fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC) (DEFAULT), UTF-8N\n");
 #endif
 #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");
+    fprintf(stderr,"j,s,e,w  Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC), UTF-8N (DEFAULT)\n");
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
     fprintf(stderr,"         After 'w' you can add more options. (80?|16((B|L)0?)?) \n");
@@ -4340,7 +5235,7 @@ usage()
     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");
-    fprintf(stderr,"h        1 hirakana->katakana, 2 katakana->hirakana,3 both\n");
+    fprintf(stderr,"h        1 katakana->hiragana, 2 hiragana->katakana, 3 both\n");
     fprintf(stderr,"v        Show this usage. V: show version\n");
     fprintf(stderr,"m[BQN0]  MIME decode [B:base64,Q:quoted,N:non-strict,0:no decode]\n");
     fprintf(stderr,"M[BQ]    MIME encode [B:base64 Q:quoted]\n");
@@ -4361,13 +5256,22 @@ usage()
     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");
-    fprintf(stderr," --cp932, --no-cp932       CP932 compatible\n");
+    fprintf(stderr," --x0212                   Convert JISX0212\n");
+    fprintf(stderr," --cp932, --no-cp932       CP932 compatibility\n");
+    fprintf(stderr," --prefix=    Insert escape before troublesome characters of Shift_JIS\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 UNICODE_NORMALIZATION
+    fprintf(stderr," --utf8mac-input   UTF-8-MAC input\n");
+#endif
+#ifdef UTF8_INPUT_ENABLE
+    fprintf(stderr," --fb-{skip, html, xml, perl, java, subchar}\n");
+    fprintf(stderr,"                   set the way nkf handles unassigned characters\n");
+#endif
 #ifdef UTF8_OUTPUT_ENABLE
     fprintf(stderr," --ms-ucs-map      Microsoft UCS Mapping Compatible\n");
 #endif
@@ -4398,7 +5302,7 @@ version()
                   ,NKF_VERSION,NKF_RELEASE_DATE);
     fprintf(stderr,"\n%s\n",CopyRight);
 }
-#endif
+#endif /*PERL_XS*/
 
 /**
  ** パッチ制作者