OSDN Git Service

(root) / nkf / nkf.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
fix typo of 5a54b532d970b9d76862ee47b5364bcbc2f82330
[nkf/nkf.git] / nkf.c
diff --git a/nkf.c b/nkf.c
index 7797245..b58c437 100644 (file)
--- a/nkf.c
+++ b/nkf.c
@@ -1,42 +1,30 @@
-/** Network Kanji Filter. (PDS Version)
- ** -*- coding: ISO-2022-JP -*-
- ************************************************************************
- ** Copyright (C) 1987, Fujitsu LTD. (Itaru ICHIKAWA)
- ** \e$BO"Mm@h!'\e(B \e$B!J3t!KIY;NDL8&5f=j!!%=%U%H#38&!!;T@n!!;j\e(B
- ** \e$B!J\e(BE-Mail Address: ichikawa@flab.fujitsu.co.jp\e$B!K\e(B
- ** Copyright (C) 1996,1998
- ** Copyright (C) 2002
- ** \e$BO"Mm@h!'\e(B \e$BN05eBg3X>pJs9)3X2J\e(B \e$B2OLn\e(B \e$B??<#\e(B  mime/X0208 support
- ** \e$B!J\e(BE-Mail Address: kono@ie.u-ryukyu.ac.jp\e$B!K\e(B
- ** \e$BO"Mm@h!'\e(B COW for DOS & Win16 & Win32 & OS/2
- ** \e$B!J\e(BE-Mail Address: GHG00637@niftyserve.or.p\e$B!K\e(B
- **
- **    \e$B$3$N%=!<%9$N$$$+$J$kJ#<L!$2~JQ!$=$@5$b5vBz$7$^$9!#$?$@$7!"\e(B
- **    \e$B$=$N:]$K$O!"C/$,9W8%$7$?$r<($9$3$NItJ,$r;D$9$3$H!#\e(B
- **    \e$B:FG[I[$d;(;o$NIUO?$J$I$NLd$$9g$o$;$bI,MW$"$j$^$;$s!#\e(B
- **    \e$B1DMxMxMQ$b>e5-$KH?$7$J$$HO0O$G5v2D$7$^$9!#\e(B
- **    \e$B%P%$%J%j$NG[I[$N:]$K$O\e(Bversion message\e$B$rJ]B8$9$k$3$H$r>r7o$H$7$^$9!#\e(B
- **    \e$B$3$N%W%m%0%i%`$K$D$$$F$OFC$K2?$NJ]>Z$b$7$J$$!"0-$7$+$i$:!#\e(B
- **
- **    Everyone is permitted to do anything on this program
- **    including copying, modifying, improving,
- **    as long as you don't try to pretend that you wrote it.
- **    i.e., the above copyright notice has to appear in all copies.
- **    Binary distribution requires original version messages.
- **    You don't have to ask before copying, redistribution or publishing.
- **    THE AUTHOR DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE.
- ***********************************************************************/
-
-/***********************************************************************
- * \e$B8=:_!"\e(Bnkf \e$B$O\e(B SorceForge \e$B$K$F%a%s%F%J%s%9$,B3$1$i$l$F$$$^$9!#\e(B
- * http://sourceforge.jp/projects/nkf/
- ***********************************************************************/
-#define NKF_IDENT "$Id: nkf.c,v 1.188 2008/11/07 02:43:43 naruse Exp $"
-#define NKF_VERSION "2.0.8"
-#define NKF_RELEASE_DATE "2008-11-07"
+/*
+ * Copyright (c) 1987, Fujitsu LTD. (Itaru ICHIKAWA).
+ * Copyright (c) 1996-2013, The nkf Project.
+ *
+ * This software is provided 'as-is', without any express or implied
+ * warranty. In no event will the authors be held liable for any damages
+ * arising from the use of this software.
+ *
+ * Permission is granted to anyone to use this software for any purpose,
+ * including commercial applications, and to alter it and redistribute it
+ * freely, subject to the following restrictions:
+ *
+ * 1. The origin of this software must not be misrepresented; you must not
+ * claim that you wrote the original software. If you use this software
+ * in a product, an acknowledgment in the product documentation would be
+ * appreciated but is not required.
+ *
+ * 2. Altered source versions must be plainly marked as such, and must not be
+ * misrepresented as being the original software.
+ *
+ * 3. This notice may not be removed or altered from any source distribution.
+ */
+#define NKF_VERSION "2.1.4"
+#define NKF_RELEASE_DATE "2015-12-12"
 #define COPY_RIGHT \
 #define COPY_RIGHT \
-    "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW\n" \
-    "Copyright (C) 2002-2008 Kono, Furukawa, Naruse, mastodon"
+    "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa).\n" \
+    "Copyright (C) 1996-2015, The nkf Project."
 
 #include "config.h"
 #include "nkf.h"
 
 #include "config.h"
 #include "nkf.h"
@@ -50,6 +38,8 @@
 # define INCL_DOSERRORS
 # include <os2.h>
 #endif
 # define INCL_DOSERRORS
 # include <os2.h>
 #endif
+#include <assert.h>
+
 
 /* state of output_mode and input_mode
 
 
 /* state of output_mode and input_mode
 
@@ -139,7 +129,7 @@ enum nkf_encodings {
     JIS_X_0212        = 0x1159, /* D */
     /* JIS_X_0213_2000_1 = 0x1228, */ /* O */
     JIS_X_0213_2 = 0x1229, /* P */
     JIS_X_0212        = 0x1159, /* D */
     /* JIS_X_0213_2000_1 = 0x1228, */ /* O */
     JIS_X_0213_2 = 0x1229, /* P */
-    JIS_X_0213_1 = 0x1233, /* Q */
+    JIS_X_0213_1 = 0x1233 /* Q */
 };
 
 static nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0);
 };
 
 static nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0);
@@ -220,6 +210,8 @@ struct {
 } encoding_name_to_id_table[] = {
     {"US-ASCII",               ASCII},
     {"ASCII",                  ASCII},
 } encoding_name_to_id_table[] = {
     {"US-ASCII",               ASCII},
     {"ASCII",                  ASCII},
+    {"646",                    ASCII},
+    {"ROMAN8",                 ASCII},
     {"ISO-2022-JP",            ISO_2022_JP},
     {"ISO2022JP-CP932",                CP50220},
     {"CP50220",                        CP50220},
     {"ISO-2022-JP",            ISO_2022_JP},
     {"ISO2022JP-CP932",                CP50220},
     {"CP50220",                        CP50220},
@@ -231,6 +223,8 @@ struct {
     {"ISO-2022-JP-2004",       ISO_2022_JP_2004},
     {"SHIFT_JIS",              SHIFT_JIS},
     {"SJIS",                   SHIFT_JIS},
     {"ISO-2022-JP-2004",       ISO_2022_JP_2004},
     {"SHIFT_JIS",              SHIFT_JIS},
     {"SJIS",                   SHIFT_JIS},
+    {"MS_Kanji",               SHIFT_JIS},
+    {"PCK",                    SHIFT_JIS},
     {"WINDOWS-31J",            WINDOWS_31J},
     {"CSWINDOWS31J",           WINDOWS_31J},
     {"CP932",                  WINDOWS_31J},
     {"WINDOWS-31J",            WINDOWS_31J},
     {"CSWINDOWS31J",           WINDOWS_31J},
     {"CP932",                  WINDOWS_31J},
@@ -305,7 +299,7 @@ struct {
                               && (c != '(') && (c != ')') && (c != '.') && (c != 0x22)))
 
 #define is_ibmext_in_sjis(c2) (CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END)
                               && (c != '(') && (c != ')') && (c != '.') && (c != 0x22)))
 
 #define is_ibmext_in_sjis(c2) (CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END)
-#define nkf_byte_jisx0201_katakana_p(c) (SP <= c && c < (0xE0&0x7F))
+#define nkf_byte_jisx0201_katakana_p(c) (SP <= c && c <= 0x5F)
 
 #define         HOLD_SIZE       1024
 #if defined(INT_IS_SHORT)
 
 #define         HOLD_SIZE       1024
 #if defined(INT_IS_SHORT)
@@ -362,6 +356,7 @@ static  int     no_cp932ext_f = FALSE;
 /* ignore ZERO WIDTH NO-BREAK SPACE */
 static  int     no_best_fit_chars_f = FALSE;
 static  int     input_endian = ENDIAN_BIG;
 /* ignore ZERO WIDTH NO-BREAK SPACE */
 static  int     no_best_fit_chars_f = FALSE;
 static  int     input_endian = ENDIAN_BIG;
+static  int     input_bom_f = FALSE;
 static  nkf_char     unicode_subchar = '?'; /* the regular substitution character */
 static  void    (*encode_fallback)(nkf_char c) = NULL;
 static  void    w_status(struct input_code *, nkf_char);
 static  nkf_char     unicode_subchar = '?'; /* the regular substitution character */
 static  void    (*encode_fallback)(nkf_char c) = NULL;
 static  void    w_status(struct input_code *, nkf_char);
@@ -389,6 +384,8 @@ static unsigned char   stdibuf[IOBUF_SIZE];
 static unsigned char   stdobuf[IOBUF_SIZE];
 #endif
 
 static unsigned char   stdobuf[IOBUF_SIZE];
 #endif
 
+#define NKF_UNSPECIFIED (-TRUE)
+
 /* flags */
 static int             unbuf_f = FALSE;
 static int             estab_f = FALSE;
 /* flags */
 static int             unbuf_f = FALSE;
 static int             estab_f = FALSE;
@@ -403,7 +400,7 @@ 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             broken_f = FALSE;       /* convert ESC-less broken JIS */
 static int             iso8859_f = FALSE;      /* ISO8859 through */
 static int             mimeout_f = FALSE;       /* base64 mode */
-static int             x0201_f = X0201_DEFAULT; /* convert JIS X 0201 */
+static int             x0201_f = NKF_UNSPECIFIED;   /* convert JIS X 0201 */
 static int             iso2022jp_f = FALSE;    /* replace non ISO-2022-JP with GETA */
 
 #ifdef UNICODE_NORMALIZATION
 static int             iso2022jp_f = FALSE;    /* replace non ISO-2022-JP with GETA */
 
 #ifdef UNICODE_NORMALIZATION
@@ -434,6 +431,8 @@ static nkf_char (*i_uungetc)(nkf_char c ,FILE *f) = std_ungetc;
 #define nkf_char_unicode_bmp_p(c) ((c & VALUE_MASK) <= UNICODE_BMP_MAX)
 #define nkf_char_unicode_value_p(c) ((c & VALUE_MASK) <= UNICODE_MAX)
 
 #define nkf_char_unicode_bmp_p(c) ((c & VALUE_MASK) <= UNICODE_BMP_MAX)
 #define nkf_char_unicode_value_p(c) ((c & VALUE_MASK) <= UNICODE_MAX)
 
+#define UTF16_TO_UTF32(lead, trail) (((lead) << 10) + (trail) - NKF_INT32_C(0x35FDC00))
+
 #ifdef NUMCHAR_OPTION
 static int numchar_f = FALSE;
 static nkf_char (*i_ngetc)(FILE *) = std_getc; /* input of ugetc */
 #ifdef NUMCHAR_OPTION
 static int numchar_f = FALSE;
 static nkf_char (*i_ngetc)(FILE *) = std_getc; /* input of ugetc */
@@ -478,8 +477,10 @@ struct input_code input_code_list[] = {
     {"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},
     {"Shift_JIS", 0, 0, 0, {0, 0, 0}, s_status, s_iconv, 0},
 #ifdef UTF8_INPUT_ENABLE
     {"UTF-8",     0, 0, 0, {0, 0, 0}, w_status, w_iconv, 0},
+    {"UTF-16",     0, 0, 0, {0, 0, 0}, NULL, w_iconv16, 0},
+    {"UTF-32",     0, 0, 0, {0, 0, 0}, NULL, w_iconv32, 0},
 #endif
 #endif
-    {0}
+    {NULL,        0, 0, 0, {0, 0, 0}, NULL, NULL, 0}
 };
 
 static int              mimeout_mode = 0; /* 0, -1, 'Q', 'B', 1, 2 */
 };
 
 static int              mimeout_mode = 0; /* 0, -1, 'Q', 'B', 1, 2 */
@@ -508,10 +509,10 @@ static int             fold_margin  = FOLD_MARGIN;
 /* process default */
 
 static nkf_char
 /* process default */
 
 static nkf_char
-no_connection2(nkf_char c2, nkf_char c1, nkf_char c0)
+no_connection2(ARG_UNUSED nkf_char c2, ARG_UNUSED nkf_char c1, ARG_UNUSED nkf_char c0)
 {
     fprintf(stderr,"nkf internal module connection failure.\n");
 {
     fprintf(stderr,"nkf internal module connection failure.\n");
-    exit(1);
+    exit(EXIT_FAILURE);
     return 0; /* LINT */
 }
 
     return 0; /* LINT */
 }
 
@@ -622,6 +623,27 @@ static const unsigned char ev[]= {
     0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
     0x00,0x00};
 
     0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
     0x00,0x00};
 
+/* X0201 kana to X0213 conversion table for han-daguten */
+/* 90-9F A0-DF */
+static const unsigned char ev_x0213[]= {
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x25,0x77,0x25,0x78,
+    0x25,0x79,0x25,0x7a,0x25,0x7b,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x25,0x7c,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x25,0x7d,0x00,0x00,
+    0x25,0x7e,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
+    0x00,0x00};
+
 
 /* X0208 kigou conversion table */
 /* 0x8140 - 0x819e */
 
 /* X0208 kigou conversion table */
 /* 0x8140 - 0x819e */
@@ -659,9 +681,37 @@ static nkf_char prev_cr = 0; /* CR or 0 */
 static int             end_check;
 #endif /*Easy Win */
 
 static int             end_check;
 #endif /*Easy Win */
 
-#define STD_GC_BUFSIZE (256)
-nkf_char std_gc_buf[STD_GC_BUFSIZE];
-nkf_char std_gc_ndx;
+static void *
+nkf_xmalloc(size_t size)
+{
+    void *ptr;
+
+    if (size == 0) size = 1;
+
+    ptr = malloc(size);
+    if (ptr == NULL) {
+       perror("can't malloc");
+       exit(EXIT_FAILURE);
+    }
+
+    return ptr;
+}
+
+static void *
+nkf_xrealloc(void *ptr, size_t size)
+{
+    if (size == 0) size = 1;
+
+    ptr = realloc(ptr, size);
+    if (ptr == NULL) {
+       perror("can't realloc");
+       exit(EXIT_FAILURE);
+    }
+
+    return ptr;
+}
+
+#define nkf_xfree(ptr) free(ptr)
 
 static int
 nkf_str_caseeql(const char *src, const char *target)
 
 static int
 nkf_str_caseeql(const char *src, const char *target)
@@ -723,22 +773,15 @@ nkf_enc_find(const char *name)
                                nkf_enc_to_index(enc) == CP50222)
 
 #ifdef DEFAULT_CODE_LOCALE
                                nkf_enc_to_index(enc) == CP50222)
 
 #ifdef DEFAULT_CODE_LOCALE
-static char*
-nkf_locale_charmap()
+static const char*
+nkf_locale_charmap(void)
 {
 #ifdef HAVE_LANGINFO_H
     return nl_langinfo(CODESET);
 #elif defined(__WIN32__)
 {
 #ifdef HAVE_LANGINFO_H
     return nl_langinfo(CODESET);
 #elif defined(__WIN32__)
-    char buf[16];
-    char *str;
-    int len = sprintf(buf, "CP%d", GetACP());
-    if (len > 0) {
-      str = malloc(len + 1);
-      strcpy(str, buf);
-      str[len] = '\0';
-      return str;
-    }
-    else return NULL;
+    static char buf[16];
+    sprintf(buf, "CP%d", GetACP());
+    return buf;
 #elif defined(__OS2__)
 # if defined(INT_IS_SHORT)
     /* OS/2 1.x */
 #elif defined(__OS2__)
 # if defined(INT_IS_SHORT)
     /* OS/2 1.x */
@@ -754,16 +797,15 @@ nkf_locale_charmap()
         sprintf(buf, "CP%lu", ulCP[0]);
     return buf;
 # endif
         sprintf(buf, "CP%lu", ulCP[0]);
     return buf;
 # endif
-#else
-    return NULL;
 #endif
 #endif
+    return NULL;
 }
 
 static nkf_encoding*
 }
 
 static nkf_encoding*
-nkf_locale_encoding()
+nkf_locale_encoding(void)
 {
     nkf_encoding *enc = 0;
 {
     nkf_encoding *enc = 0;
-    char *encname = nkf_locale_charmap();
+    const char *encname = nkf_locale_charmap();
     if (encname)
        enc = nkf_enc_find(encname);
     return enc;
     if (encname)
        enc = nkf_enc_find(encname);
     return enc;
@@ -771,13 +813,13 @@ nkf_locale_encoding()
 #endif /* DEFAULT_CODE_LOCALE */
 
 static nkf_encoding*
 #endif /* DEFAULT_CODE_LOCALE */
 
 static nkf_encoding*
-nkf_utf8_encoding()
+nkf_utf8_encoding(void)
 {
     return &nkf_encoding_table[UTF_8];
 }
 
 static nkf_encoding*
 {
     return &nkf_encoding_table[UTF_8];
 }
 
 static nkf_encoding*
-nkf_default_encoding()
+nkf_default_encoding(void)
 {
     nkf_encoding *enc = 0;
 #ifdef DEFAULT_CODE_LOCALE
 {
     nkf_encoding *enc = 0;
 #ifdef DEFAULT_CODE_LOCALE
@@ -789,6 +831,64 @@ nkf_default_encoding()
     return enc;
 }
 
     return enc;
 }
 
+typedef struct {
+    long capa;
+    long len;
+    nkf_char *ptr;
+} nkf_buf_t;
+
+static nkf_buf_t *
+nkf_buf_new(int length)
+{
+    nkf_buf_t *buf = nkf_xmalloc(sizeof(nkf_buf_t));
+    buf->ptr = nkf_xmalloc(sizeof(nkf_char) * length);
+    buf->capa = length;
+    buf->len = 0;
+    return buf;
+}
+
+#if 0
+static void
+nkf_buf_dispose(nkf_buf_t *buf)
+{
+    nkf_xfree(buf->ptr);
+    nkf_xfree(buf);
+}
+#endif
+
+#define nkf_buf_length(buf) ((buf)->len)
+#define nkf_buf_empty_p(buf) ((buf)->len == 0)
+
+static nkf_char
+nkf_buf_at(nkf_buf_t *buf, int index)
+{
+    assert(index <= buf->len);
+    return buf->ptr[index];
+}
+
+static void
+nkf_buf_clear(nkf_buf_t *buf)
+{
+    buf->len = 0;
+}
+
+static void
+nkf_buf_push(nkf_buf_t *buf, nkf_char c)
+{
+    if (buf->capa <= buf->len) {
+       exit(EXIT_FAILURE);
+    }
+    buf->ptr[buf->len++] = c;
+}
+
+static nkf_char
+nkf_buf_pop(nkf_buf_t *buf)
+{
+    assert(!nkf_buf_empty_p(buf));
+    return buf->ptr[--buf->len];
+}
+
+/* Normalization Form C */
 #ifndef PERL_XS
 #ifdef WIN32DLL
 #define fprintf dllprintf
 #ifndef PERL_XS
 #ifdef WIN32DLL
 #define fprintf dllprintf
@@ -804,68 +904,61 @@ static void
 usage(void)
 {
     fprintf(HELP_OUTPUT,
 usage(void)
 {
     fprintf(HELP_OUTPUT,
-           "USAGE:  nkf(nkf32,wnkf,nkf2) -[flags] [in file] .. [out file for -O flag]\n"
-           "Flags:\n"
-           "b,u      Output is buffered (DEFAULT),Output is unbuffered\n"
-           "j,s,e,w  Output code is ISO-2022-JP, Shift JIS, EUC-JP, UTF-8N\n"
+           "Usage:  nkf -[flags] [--] [in file] .. [out file for -O flag]\n"
 #ifdef UTF8_OUTPUT_ENABLE
 #ifdef UTF8_OUTPUT_ENABLE
-           "         After 'w' you can add more options. -w[ 8 [0], 16 [[BL] [0]] ]\n"
+           " j/s/e/w  Specify output encoding ISO-2022-JP, Shift_JIS, EUC-JP\n"
+           "          UTF options is -w[8[0],{16,32}[{B,L}[0]]]\n"
+#else
 #endif
 #endif
-           "J,S,E,W  Input assumption is JIS 7 bit , Shift JIS, EUC-JP, UTF-8\n"
 #ifdef UTF8_INPUT_ENABLE
 #ifdef UTF8_INPUT_ENABLE
-           "         After 'W' you can add more options. -W[ 8, 16 [BL] ] \n"
-#endif
-           "t        no conversion\n"
-           "i[@B]    Specify the Esc Seq for JIS X 0208-1978/83 (DEFAULT B)\n"
-           "o[BJH]   Specify the Esc Seq for ASCII/Roman        (DEFAULT B)\n"
-           "r        {de/en}crypt ROT13/47\n"
-           "h        1 katakana->hiragana, 2 hiragana->katakana, 3 both\n"
-           "m[BQSN0] MIME decode [B:base64,Q:quoted,S:strict,N:non-strict,0:no decode]\n"
-           "M[BQ]    MIME encode [B:base64 Q:quoted]\n"
-           "l        ISO8859-1 (Latin-1) support\n"
-           "f/F      Folding: -f60 or -f or -f60-10 (fold margin 10) F preserve nl\n"
-           "Z[0-4]   Default/0: Convert JISX0208 Alphabet to ASCII\n"
-           "         1: Kankaku to one space  2: to two spaces  3: HTML Entity\n"
-           "         4: JISX0208 Katakana to JISX0201 Katakana\n"
-           "X,x      Assume X0201 kana in MS-Kanji, -x preserves X0201\n"
-           "B[0-2]   Broken input  0: missing ESC,1: any X on ESC-[($]-X,2: ASCII on NL\n"
-#ifdef MSDOS
-           "T        Text mode output\n"
+           " J/S/E/W  Specify input encoding ISO-2022-JP, Shift_JIS, EUC-JP\n"
+           "          UTF option is -W[8,[16,32][B,L]]\n"
+#else
+           " J/S/E    Specify output encoding ISO-2022-JP, Shift_JIS, EUC-JP\n"
 #endif
 #endif
-           "O        Output to File (DEFAULT 'nkf.out')\n"
-           "I        Convert non ISO-2022-JP charactor to GETA\n"
-           "d,c      Convert line breaks  -d: LF  -c: CRLF\n"
-           "-L[uwm]  line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n"
-           "v, V     Show this usage. V: show configuration\n"
-           "\n"
-           "Long name options\n"
-           " --ic=<input codeset>  --oc=<output codeset>\n"
-           "                   Specify the input or output codeset\n"
-           " --fj  --unix --mac  --windows\n"
-           " --jis  --euc  --sjis  --utf8  --utf16  --mime  --base64\n"
-           "                   Convert for the system or code\n"
-           " --hiragana  --katakana  --katakana-hiragana\n"
-           "                   To Hiragana/Katakana Conversion\n"
-           " --prefix=         Insert escape before troublesome characters of Shift_JIS\n"
+           );
+    fprintf(HELP_OUTPUT,
+           " m[BQSN0] MIME decode [B:base64,Q:quoted,S:strict,N:nonstrict,0:no decode]\n"
+           " M[BQ]    MIME encode [B:base64 Q:quoted]\n"
+           " f/F      Folding: -f60 or -f or -f60-10 (fold margin 10) F preserve nl\n"
+           );
+    fprintf(HELP_OUTPUT,
+           " Z[0-4]   Default/0: Convert JISX0208 Alphabet to ASCII\n"
+           "          1: Kankaku to one space  2: to two spaces  3: HTML Entity\n"
+           "          4: JISX0208 Katakana to JISX0201 Katakana\n"
+           " X,x      Convert Halfwidth Katakana to Fullwidth or preserve it\n"
+           );
+    fprintf(HELP_OUTPUT,
+           " O        Output to File (DEFAULT 'nkf.out')\n"
+           " L[uwm]   Line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n"
+           );
+    fprintf(HELP_OUTPUT,
+           " --ic=<encoding>        Specify the input encoding\n"
+           " --oc=<encoding>        Specify the output encoding\n"
+           " --hiragana --katakana  Hiragana/Katakana Conversion\n"
+           " --katakana-hiragana    Converts each other\n"
+           );
+    fprintf(HELP_OUTPUT,
 #ifdef INPUT_OPTION
 #ifdef INPUT_OPTION
-           " --cap-input, --url-input  Convert hex after ':' or '%%'\n"
+           " --{cap, url}-input     Convert hex after ':' or '%%'\n"
 #endif
 #ifdef NUMCHAR_OPTION
 #endif
 #ifdef NUMCHAR_OPTION
-           " --numchar-input   Convert Unicode Character Reference\n"
+           " --numchar-input        Convert Unicode Character Reference\n"
 #endif
 #ifdef UTF8_INPUT_ENABLE
            " --fb-{skip, html, xml, perl, java, subchar}\n"
 #endif
 #ifdef UTF8_INPUT_ENABLE
            " --fb-{skip, html, xml, perl, java, subchar}\n"
-           "                   Specify how nkf handles unassigned characters\n"
+           "                        Specify unassigned character's replacement\n"
 #endif
 #endif
+           );
+    fprintf(HELP_OUTPUT,
 #ifdef OVERWRITE
 #ifdef OVERWRITE
-           " --in-place[=SUFFIX]  --overwrite[=SUFFIX]\n"
-           "                   Overwrite original listed files by filtered result\n"
-           "                   --overwrite preserves timestamp of original files\n"
+           " --in-place[=SUF]       Overwrite original files\n"
+           " --overwrite[=SUF]      Preserve timestamp of original files\n"
 #endif
 #endif
-           " -g  --guess       Guess the input code\n"
-           " --help  --version Show this help/the version\n"
-           "                   For more information, see also man nkf\n"
-           "\n");
+           " -g --guess             Guess the input code\n"
+           " -v --version           Print the version\n"
+           " --help/-V              Print this help / configuration\n"
+           );
     version();
 }
 
     version();
 }
 
@@ -874,8 +967,6 @@ show_configuration(void)
 {
     fprintf(HELP_OUTPUT,
            "Summary of my nkf " NKF_VERSION " (" NKF_RELEASE_DATE ") configuration:\n"
 {
     fprintf(HELP_OUTPUT,
            "Summary of my nkf " NKF_VERSION " (" NKF_RELEASE_DATE ") configuration:\n"
-           "  nkf identity:\n"
-           "    " NKF_IDENT "\n"
            "  Compile-time options:\n"
            "    Compiled at:                 " __DATE__ " " __TIME__ "\n"
           );
            "  Compile-time options:\n"
            "    Compiled at:                 " __DATE__ " " __TIME__ "\n"
           );
@@ -937,12 +1028,7 @@ get_backup_filename(const char *suffix, const char *filename)
     }
 
     if(asterisk_count){
     }
 
     if(asterisk_count){
-       backup_filename = malloc(strlen(suffix) + (asterisk_count * (filename_length - 1)) + 1);
-       if (!backup_filename){
-           perror("Can't malloc backup filename.");
-           return NULL;
-       }
-
+       backup_filename = nkf_xmalloc(strlen(suffix) + (asterisk_count * (filename_length - 1)) + 1);
        for(i = 0, j = 0; suffix[i];){
            if(suffix[i] == '*'){
                backup_filename[j] = '\0';
        for(i = 0, j = 0; suffix[i];){
            if(suffix[i] == '*'){
                backup_filename[j] = '\0';
@@ -956,7 +1042,7 @@ get_backup_filename(const char *suffix, const char *filename)
        backup_filename[j] = '\0';
     }else{
        j = filename_length + strlen(suffix);
        backup_filename[j] = '\0';
     }else{
        j = filename_length + strlen(suffix);
-       backup_filename = malloc(j + 1);
+       backup_filename = nkf_xmalloc(j + 1);
        strcpy(backup_filename, filename);
        strcat(backup_filename, suffix);
        backup_filename[j] = '\0';
        strcpy(backup_filename, filename);
        strcat(backup_filename, suffix);
        backup_filename[j] = '\0';
@@ -972,7 +1058,7 @@ nkf_each_char_to_hex(void (*f)(nkf_char c2,nkf_char c1), nkf_char c)
     int shift = 20;
     c &= VALUE_MASK;
     while(shift >= 0){
     int shift = 20;
     c &= VALUE_MASK;
     while(shift >= 0){
-       if(c >= 1<<shift){
+       if(c >= NKF_INT32_C(1)<<shift){
            while(shift >= 0){
                (*f)(0, bin2hex(c>>shift));
                shift -= 4;
            while(shift >= 0){
                (*f)(0, bin2hex(c>>shift));
                shift -= 4;
@@ -1070,7 +1156,7 @@ static const struct {
     {"euc","e"},
     {"euc-input","E"},
     {"fj","jm"},
     {"euc","e"},
     {"euc-input","E"},
     {"fj","jm"},
-    {"help","v"},
+    {"help",""},
     {"jis","j"},
     {"jis-input","J"},
     {"mac","sLm"},
     {"jis","j"},
     {"jis-input","J"},
     {"mac","sLm"},
@@ -1080,7 +1166,7 @@ static const struct {
     {"sjis","s"},
     {"sjis-input","S"},
     {"unix","eLu"},
     {"sjis","s"},
     {"sjis-input","S"},
     {"unix","eLu"},
-    {"version","V"},
+    {"version","v"},
     {"windows","sLw"},
     {"hiragana","h1"},
     {"katakana","h2"},
     {"windows","sLw"},
     {"hiragana","h1"},
     {"katakana","h2"},
@@ -1147,9 +1233,10 @@ set_input_encoding(nkf_encoding *enc)
     case ISO_8859_1:
        iso8859_f = TRUE;
        break;
     case ISO_8859_1:
        iso8859_f = TRUE;
        break;
-    case CP50220:
     case CP50221:
     case CP50222:
     case CP50221:
     case CP50222:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
+    case CP50220:
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1171,6 +1258,7 @@ set_input_encoding(nkf_encoding *enc)
     case SHIFT_JIS:
        break;
     case WINDOWS_31J:
     case SHIFT_JIS:
        break;
     case WINDOWS_31J:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1192,6 +1280,7 @@ set_input_encoding(nkf_encoding *enc)
     case EUCJP_NKF:
        break;
     case CP51932:
     case EUCJP_NKF:
        break;
     case CP51932:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
 #ifdef SHIFTJIS_CP932
        cp51932_f = TRUE;
 #endif
@@ -1200,6 +1289,7 @@ set_input_encoding(nkf_encoding *enc)
 #endif
        break;
     case EUCJP_MS:
 #endif
        break;
     case EUCJP_MS:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
 #endif
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
 #endif
@@ -1208,6 +1298,7 @@ set_input_encoding(nkf_encoding *enc)
 #endif
        break;
     case EUCJP_ASCII:
 #endif
        break;
     case EUCJP_ASCII:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
 #endif
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
 #endif
@@ -1220,6 +1311,7 @@ set_input_encoding(nkf_encoding *enc)
        x0213_f = TRUE;
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
        x0213_f = TRUE;
 #ifdef SHIFTJIS_CP932
        cp51932_f = FALSE;
+       if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
        break;
     case EUC_JISX0213:
 #endif
        break;
     case EUC_JISX0213:
@@ -1262,7 +1354,6 @@ set_output_encoding(nkf_encoding *enc)
 {
     switch (nkf_enc_to_index(enc)) {
     case CP50220:
 {
     switch (nkf_enc_to_index(enc)) {
     case CP50220:
-       x0201_f = TRUE;
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
@@ -1271,6 +1362,7 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case CP50221:
 #endif
        break;
     case CP50221:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
@@ -1278,6 +1370,11 @@ set_output_encoding(nkf_encoding *enc)
        ms_ucs_map_f = UCS_MAP_CP932;
 #endif
        break;
        ms_ucs_map_f = UCS_MAP_CP932;
 #endif
        break;
+    case ISO_2022_JP:
+#ifdef SHIFTJIS_CP932
+       if (cp932inv_f == TRUE) cp932inv_f = FALSE;
+#endif
+       break;
     case ISO_2022_JP_1:
        x0212_f = TRUE;
 #ifdef SHIFTJIS_CP932
     case ISO_2022_JP_1:
        x0212_f = TRUE;
 #ifdef SHIFTJIS_CP932
@@ -1285,6 +1382,7 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case ISO_2022_JP_3:
 #endif
        break;
     case ISO_2022_JP_3:
+    case ISO_2022_JP_2004:
        x0212_f = TRUE;
        x0213_f = TRUE;
 #ifdef SHIFTJIS_CP932
        x0212_f = TRUE;
        x0213_f = TRUE;
 #ifdef SHIFTJIS_CP932
@@ -1294,6 +1392,7 @@ set_output_encoding(nkf_encoding *enc)
     case SHIFT_JIS:
        break;
     case WINDOWS_31J:
     case SHIFT_JIS:
        break;
     case WINDOWS_31J:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_CP932;
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_CP932;
 #endif
@@ -1322,6 +1421,7 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case CP51932:
 #endif
        break;
     case CP51932:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
 #ifdef SHIFTJIS_CP932
        if (cp932inv_f == TRUE) cp932inv_f = FALSE;
 #endif
@@ -1330,12 +1430,14 @@ set_output_encoding(nkf_encoding *enc)
 #endif
        break;
     case EUCJP_MS:
 #endif
        break;
     case EUCJP_MS:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
        x0212_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_MS;
 #endif
        break;
     case EUCJP_ASCII:
        x0212_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_MS;
 #endif
        break;
     case EUCJP_ASCII:
+       if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE;        /* -x specified implicitly */
        x0212_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_ASCII;
        x0212_f = TRUE;
 #ifdef UTF8_OUTPUT_ENABLE
        ms_ucs_map_f = UCS_MAP_ASCII;
@@ -1372,6 +1474,7 @@ set_output_encoding(nkf_encoding *enc)
        output_endian = ENDIAN_LITTLE;
        output_bom_f = TRUE;
        break;
        output_endian = ENDIAN_LITTLE;
        output_bom_f = TRUE;
        break;
+    case UTF_32:
     case UTF_32BE_BOM:
        output_bom_f = TRUE;
        break;
     case UTF_32BE_BOM:
        output_bom_f = TRUE;
        break;
@@ -1463,13 +1566,26 @@ x0212_unshift(nkf_char c)
 }
 #endif /* X0212_ENABLE */
 
 }
 #endif /* X0212_ENABLE */
 
+static int
+is_x0213_2_in_x0212(nkf_char c1)
+{
+    static const char x0213_2_table[] =
+       {0, 1, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1};
+    int ku = c1 - 0x20;
+    if (ku <= 15)
+       return x0213_2_table[ku]; /* 1, 3-5, 8, 12-15 */
+    if (78 <= ku && ku <= 94)
+       return 1;
+    return 0;
+}
+
 static nkf_char
 e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
 {
     nkf_char ndx;
     if (is_eucg3(c2)){
        ndx = c2 & 0x7f;
 static nkf_char
 e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
 {
     nkf_char ndx;
     if (is_eucg3(c2)){
        ndx = c2 & 0x7f;
-       if (x0213_f){
+       if (x0213_f && is_x0213_2_in_x0212(ndx)){
            if((0x21 <= ndx && ndx <= 0x2F)){
                if (p2) *p2 = ((ndx - 1) >> 1) + 0xec - ndx / 8 * 3;
                if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
            if((0x21 <= ndx && ndx <= 0x2F)){
                if (p2) *p2 = ((ndx - 1) >> 1) + 0xec - ndx / 8 * 3;
                if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
@@ -1515,7 +1631,7 @@ s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1)
     static const char shift_jisx0213_s1a3_table[5][2] ={ { 1, 8}, { 3, 4}, { 5,12}, {13,14}, {15, 0} };
     if (0xFC < c1) return 1;
 #ifdef SHIFTJIS_CP932
     static const char shift_jisx0213_s1a3_table[5][2] ={ { 1, 8}, { 3, 4}, { 5,12}, {13,14}, {15, 0} };
     if (0xFC < c1) return 1;
 #ifdef SHIFTJIS_CP932
-    if (!cp932inv_f && is_ibmext_in_sjis(c2)){
+    if (!cp932inv_f && !x0213_f && is_ibmext_in_sjis(c2)){
        val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
        if (val){
            c2 = val >> 8;
        val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
        if (val){
            c2 = val >> 8;
@@ -1598,7 +1714,7 @@ nkf_unicode_to_utf8(nkf_char val, nkf_char *p1, nkf_char *p2, nkf_char *p3, nkf_
        *p3 = 0x80 | ( val        & 0x3f);
        *p4 = 0;
     } else if (nkf_char_unicode_value_p(val)) {
        *p3 = 0x80 | ( val        & 0x3f);
        *p4 = 0;
     } else if (nkf_char_unicode_value_p(val)) {
-       *p1 = 0xe0 |  (val >> 16);
+       *p1 = 0xf0 |  (val >> 18);
        *p2 = 0x80 | ((val >> 12) & 0x3f);
        *p3 = 0x80 | ((val >>  6) & 0x3f);
        *p4 = 0x80 | ( val        & 0x3f);
        *p2 = 0x80 | ((val >> 12) & 0x3f);
        *p3 = 0x80 | ((val >>  6) & 0x3f);
        *p4 = 0x80 | ( val        & 0x3f);
@@ -1618,7 +1734,7 @@ nkf_utf8_to_unicode(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
        /* single byte */
        wc = c1;
     }
        /* single byte */
        wc = c1;
     }
-    else if (c1 <= 0xC3) {
+    else if (c1 <= 0xC1) {
        /* trail byte or invalid */
        return -1;
     }
        /* trail byte or invalid */
        return -1;
     }
@@ -1758,6 +1874,7 @@ unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_c
            ms_ucs_map_f == UCS_MAP_CP932 ? utf8_to_euc_2bytes_932 :
            ms_ucs_map_f == UCS_MAP_MS ? utf8_to_euc_2bytes_ms :
            ms_ucs_map_f == UCS_MAP_CP10001 ? utf8_to_euc_2bytes_mac :
            ms_ucs_map_f == UCS_MAP_CP932 ? utf8_to_euc_2bytes_932 :
            ms_ucs_map_f == UCS_MAP_MS ? utf8_to_euc_2bytes_ms :
            ms_ucs_map_f == UCS_MAP_CP10001 ? utf8_to_euc_2bytes_mac :
+           x0213_f ? utf8_to_euc_2bytes_x0213 :
            utf8_to_euc_2bytes;
        ret =  unicode_to_jis_common2(c2, c1, pp, sizeof_utf8_to_euc_2bytes, p2, p1);
     }else if(c0 < 0xF0){
            utf8_to_euc_2bytes;
        ret =  unicode_to_jis_common2(c2, c1, pp, sizeof_utf8_to_euc_2bytes, p2, p1);
     }else if(c0 < 0xF0){
@@ -1825,6 +1942,7 @@ unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_c
            ms_ucs_map_f == UCS_MAP_CP932 ? utf8_to_euc_3bytes_932 :
            ms_ucs_map_f == UCS_MAP_MS ? utf8_to_euc_3bytes_ms :
            ms_ucs_map_f == UCS_MAP_CP10001 ? utf8_to_euc_3bytes_mac :
            ms_ucs_map_f == UCS_MAP_CP932 ? utf8_to_euc_3bytes_932 :
            ms_ucs_map_f == UCS_MAP_MS ? utf8_to_euc_3bytes_ms :
            ms_ucs_map_f == UCS_MAP_CP10001 ? utf8_to_euc_3bytes_mac :
+           x0213_f ? utf8_to_euc_3bytes_x0213 :
            utf8_to_euc_3bytes;
        ret = unicode_to_jis_common2(c1, c0, ppp[c2 - 0xE0], sizeof_utf8_to_euc_C2, p2, p1);
     }else return -1;
            utf8_to_euc_3bytes;
        ret = unicode_to_jis_common2(c1, c0, ppp[c2 - 0xE0], sizeof_utf8_to_euc_C2, p2, p1);
     }else return -1;
@@ -1842,6 +1960,15 @@ unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_c
 }
 
 #ifdef UTF8_OUTPUT_ENABLE
 }
 
 #ifdef UTF8_OUTPUT_ENABLE
+#define X0213_SURROGATE_FIND(tbl, size, euc) do { \
+       int i; \
+       for (i = 0; i < size; i++) \
+           if (tbl[i][0] == euc) { \
+               low = tbl[i][2]; \
+               break; \
+           } \
+    } while (0)
+
 static nkf_char
 e2w_conv(nkf_char c2, nkf_char c1)
 {
 static nkf_char
 e2w_conv(nkf_char c2, nkf_char c1)
 {
@@ -1864,7 +1991,9 @@ e2w_conv(nkf_char c2, nkf_char c1)
        }
        c2 = (c2&0x7f) - 0x21;
        if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
        }
        c2 = (c2&0x7f) - 0x21;
        if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
-           p = x0212_to_utf8_2bytes[c2];
+           p =
+               x0213_f ? x0212_to_utf8_2bytes_x0213[c2] :
+               x0212_to_utf8_2bytes[c2];
        else
            return 0;
 #endif
        else
            return 0;
 #endif
@@ -1873,6 +2002,7 @@ e2w_conv(nkf_char c2, nkf_char c1)
        c2 = (c2&0x7f) - 0x21;
        if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
            p =
        c2 = (c2&0x7f) - 0x21;
        if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
            p =
+               x0213_f ? euc_to_utf8_2bytes_x0213[c2] :
                ms_ucs_map_f == UCS_MAP_ASCII ? euc_to_utf8_2bytes[c2] :
                ms_ucs_map_f == UCS_MAP_CP10001 ? euc_to_utf8_2bytes_mac[c2] :
                euc_to_utf8_2bytes_ms[c2];
                ms_ucs_map_f == UCS_MAP_ASCII ? euc_to_utf8_2bytes[c2] :
                ms_ucs_map_f == UCS_MAP_CP10001 ? euc_to_utf8_2bytes_mac[c2] :
                euc_to_utf8_2bytes_ms[c2];
@@ -1881,8 +2011,39 @@ e2w_conv(nkf_char c2, nkf_char c1)
     }
     if (!p) return 0;
     c1 = (c1 & 0x7f) - 0x21;
     }
     if (!p) return 0;
     c1 = (c1 & 0x7f) - 0x21;
-    if (0<=c1 && c1<sizeof_euc_to_utf8_1byte)
-       return p[c1];
+    if (0<=c1 && c1<sizeof_euc_to_utf8_1byte) {
+       nkf_char val = p[c1];
+       if (x0213_f && 0xD800<=val && val<=0xDBFF) {
+           nkf_char euc = (c2+0x21)<<8 | (c1+0x21);
+           nkf_char low = 0;
+           if (p==x0212_to_utf8_2bytes_x0213[c2]) {
+               X0213_SURROGATE_FIND(x0213_2_surrogate_table, sizeof_x0213_2_surrogate_table, euc);
+           } else {
+               X0213_SURROGATE_FIND(x0213_1_surrogate_table, sizeof_x0213_1_surrogate_table, euc);
+           }
+           if (!low) return 0;
+           return UTF16_TO_UTF32(val, low);
+       } else {
+           return val;
+       }
+    }
+    return 0;
+}
+
+static nkf_char
+e2w_combining(nkf_char comb, nkf_char c2, nkf_char c1)
+{
+    nkf_char euc;
+    int i;
+    for (i = 0; i < sizeof_x0213_combining_chars; i++)
+       if (x0213_combining_chars[i] == comb)
+           break;
+    if (i >= sizeof_x0213_combining_chars)
+       return 0;
+    euc = (c2&0x7f)<<8 | (c1&0x7f);
+    for (i = 0; i < sizeof_x0213_combining_table; i++)
+       if (x0213_combining_table[i][0] == euc)
+           return x0213_combining_table[i][1];
     return 0;
 }
 #endif
     return 0;
 }
 #endif
@@ -1929,6 +2090,25 @@ w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1)
        }
     }
     else {
        }
     }
     else {
+       int i;
+       if (x0213_f) {
+           c1 = (val >> 10) + NKF_INT32_C(0xD7C0);   /* high surrogate */
+           c2 = (val & 0x3FF) + NKF_INT32_C(0xDC00); /* low surrogate */
+           for (i = 0; i < sizeof_x0213_1_surrogate_table; i++)
+               if (x0213_1_surrogate_table[i][1] == c1 && x0213_1_surrogate_table[i][2] == c2) {
+                   val = x0213_1_surrogate_table[i][0];
+                   *p2 = val >> 8;
+                   *p1 = val & 0xFF;
+                   return 0;
+               }
+           for (i = 0; i < sizeof_x0213_2_surrogate_table; i++)
+               if (x0213_2_surrogate_table[i][1] == c1 && x0213_2_surrogate_table[i][2] == c2) {
+                   val = x0213_2_surrogate_table[i][0];
+                   *p2 = PREFIX_EUCG3 | (val >> 8);
+                   *p1 = val & 0xFF;
+                   return 0;
+               }
+       }
        *p2 = 0;
        *p1 = nkf_char_unicode_new(val);
     }
        *p2 = 0;
        *p1 = nkf_char_unicode_new(val);
     }
@@ -2001,7 +2181,7 @@ e_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
 }
 
 static nkf_char
 }
 
 static nkf_char
-s_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
+s_iconv(ARG_UNUSED nkf_char c2, nkf_char c1, ARG_UNUSED nkf_char c0)
 {
     if (c2 == JIS_X_0201_1976_K || (0xA1 <= c2 && c2 <= 0xDF)) {
        if (iso2022jp_f && !x0201_f) {
 {
     if (c2 == JIS_X_0201_1976_K || (0xA1 <= c2 && c2 <= 0xDF)) {
        if (iso2022jp_f && !x0201_f) {
@@ -2024,6 +2204,30 @@ s_iconv(nkf_char c2, nkf_char c1, nkf_char c0)
     return 0;
 }
 
     return 0;
 }
 
+static int
+x0213_wait_combining_p(nkf_char wc)
+{
+    int i;
+    for (i = 0; i < sizeof_x0213_combining_table; i++) {
+       if (x0213_combining_table[i][1] == wc) {
+           return TRUE;
+       }
+    }
+    return FALSE;
+}
+
+static int
+x0213_combining_p(nkf_char wc)
+{
+    int i;
+    for (i = 0; i < sizeof_x0213_combining_chars; i++) {
+       if (x0213_combining_chars[i] == wc) {
+           return TRUE;
+       }
+    }
+    return FALSE;
+}
+
 static nkf_char
 w_iconv(nkf_char c1, nkf_char c2, nkf_char c3)
 {
 static nkf_char
 w_iconv(nkf_char c1, nkf_char c2, nkf_char c3)
 {
@@ -2091,6 +2295,8 @@ w_iconv(nkf_char c1, nkf_char c2, nkf_char c3)
        c2 = nkf_char_unicode_new(nkf_utf8_to_unicode(c1, c2, c3, c4));
        c1 = 0;
     } else {
        c2 = nkf_char_unicode_new(nkf_utf8_to_unicode(c1, c2, c3, c4));
        c1 = 0;
     } else {
+       if (x0213_f && x0213_wait_combining_p(nkf_utf8_to_unicode(c1, c2, c3, c4)))
+           return -3;
        ret = w2e_conv(c1, c2, c3, &c1, &c2);
     }
     if (ret == 0){
        ret = w2e_conv(c1, c2, c3, &c1, &c2);
     }
     if (ret == 0){
@@ -2099,9 +2305,22 @@ w_iconv(nkf_char c1, nkf_char c2, nkf_char c3)
     return ret;
 }
 
     return ret;
 }
 
+static nkf_char
+w_iconv_nocombine(nkf_char c1, nkf_char c2, nkf_char c3)
+{
+    /* continue from the line below 'return -3;' in w_iconv() */
+    nkf_char ret = w2e_conv(c1, c2, c3, &c1, &c2);
+    if (ret == 0){
+       (*oconv)(c1, c2);
+    }
+    return ret;
+}
+
 #define NKF_ICONV_INVALID_CODE_RANGE -13
 #define NKF_ICONV_INVALID_CODE_RANGE -13
+#define NKF_ICONV_WAIT_COMBINING_CHAR -14
+#define NKF_ICONV_NOT_COMBINED -15
 static size_t
 static size_t
-unicode_iconv(nkf_char wc)
+unicode_iconv(nkf_char wc, int nocombine)
 {
     nkf_char c1, c2;
     int ret = 0;
 {
     nkf_char c1, c2;
     int ret = 0;
@@ -2113,6 +2332,8 @@ unicode_iconv(nkf_char wc)
        /* unpaired surrogate */
        return NKF_ICONV_INVALID_CODE_RANGE;
     }else if (wc < 0xFFFF) {
        /* unpaired surrogate */
        return NKF_ICONV_INVALID_CODE_RANGE;
     }else if (wc < 0xFFFF) {
+       if (!nocombine && x0213_f && x0213_wait_combining_p(wc))
+           return NKF_ICONV_WAIT_COMBINING_CHAR;
        ret = w16e_conv(wc, &c2, &c1);
        if (ret) return ret;
     }else if (wc < 0x10FFFF) {
        ret = w16e_conv(wc, &c2, &c1);
        if (ret) return ret;
     }else if (wc < 0x10FFFF) {
@@ -2125,9 +2346,50 @@ unicode_iconv(nkf_char wc)
     return 0;
 }
 
     return 0;
 }
 
-#define NKF_ICONV_NEED_ONE_MORE_BYTE -1
-#define NKF_ICONV_NEED_TWO_MORE_BYTES -2
-#define UTF16_TO_UTF32(lead, trail) (((lead) << 10) + (trail) - NKF_INT32_C(0x35FDC00))
+static nkf_char
+unicode_iconv_combine(nkf_char wc, nkf_char wc2)
+{
+    nkf_char c1, c2;
+    int i;
+
+    if (wc2 < 0x80) {
+       return NKF_ICONV_NOT_COMBINED;
+    }else if ((wc2>>11) == 27) {
+       /* unpaired surrogate */
+       return NKF_ICONV_INVALID_CODE_RANGE;
+    }else if (wc2 < 0xFFFF) {
+       if (!x0213_combining_p(wc2))
+           return NKF_ICONV_NOT_COMBINED;
+       for (i = 0; i < sizeof_x0213_combining_table; i++) {
+           if (x0213_combining_table[i][1] == wc &&
+               x0213_combining_table[i][2] == wc2) {
+               c2 = x0213_combining_table[i][0] >> 8;
+               c1 = x0213_combining_table[i][0] & 0x7f;
+               (*oconv)(c2, c1);
+               return 0;
+           }
+       }
+    }else if (wc2 < 0x10FFFF) {
+       return NKF_ICONV_NOT_COMBINED;
+    } else {
+       return NKF_ICONV_INVALID_CODE_RANGE;
+    }
+    return NKF_ICONV_NOT_COMBINED;
+}
+
+static nkf_char
+w_iconv_combine(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4, nkf_char c5, nkf_char c6)
+{
+    nkf_char wc, wc2;
+    wc = nkf_utf8_to_unicode(c1, c2, c3, 0);
+    wc2 = nkf_utf8_to_unicode(c4, c5, c6, 0);
+    if (wc2 < 0)
+       return wc2;
+    return unicode_iconv_combine(wc, wc2);
+}
+
+#define NKF_ICONV_NEED_ONE_MORE_BYTE (size_t)-1
+#define NKF_ICONV_NEED_TWO_MORE_BYTES (size_t)-2
 static size_t
 nkf_iconv_utf_16(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
 {
 static size_t
 nkf_iconv_utf_16(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
 {
@@ -2156,31 +2418,63 @@ nkf_iconv_utf_16(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
        }
     }
 
        }
     }
 
-    return (*unicode_iconv)(wc);
+    return (*unicode_iconv)(wc, FALSE);
+}
+
+static size_t
+nkf_iconv_utf_16_combine(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
+{
+    nkf_char wc, wc2;
+
+    if (input_endian == ENDIAN_BIG) {
+       if (0xD8 <= c3 && c3 <= 0xDB) {
+           return NKF_ICONV_NOT_COMBINED;
+       } else {
+           wc = c1 << 8 | c2;
+           wc2 = c3 << 8 | c4;
+       }
+    } else {
+       if (0xD8 <= c2 && c2 <= 0xDB) {
+           return NKF_ICONV_NOT_COMBINED;
+       } else {
+           wc = c2 << 8 | c1;
+           wc2 = c4 << 8 | c3;
+       }
+    }
+
+    return unicode_iconv_combine(wc, wc2);
+}
+
+static size_t
+nkf_iconv_utf_16_nocombine(nkf_char c1, nkf_char c2)
+{
+    nkf_char wc;
+    if (input_endian == ENDIAN_BIG)
+       wc = c1 << 8 | c2;
+    else
+       wc = c2 << 8 | c1;
+    return (*unicode_iconv)(wc, TRUE);
 }
 
 static nkf_char
 }
 
 static nkf_char
-w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0)
+w_iconv16(nkf_char c2, nkf_char c1, ARG_UNUSED nkf_char c0)
 {
 {
-    return 0;
+    (*oconv)(c2, c1);
+    return 16; /* different from w_iconv32 */
 }
 
 static nkf_char
 }
 
 static nkf_char
-w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0)
+w_iconv32(nkf_char c2, nkf_char c1, ARG_UNUSED nkf_char c0)
 {
 {
-    return 0;
+    (*oconv)(c2, c1);
+    return 32; /* different from w_iconv16 */
 }
 
 }
 
-static size_t
-nkf_iconv_utf_32(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
+static nkf_char
+utf32_to_nkf_char(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
 {
     nkf_char wc;
 
 {
     nkf_char wc;
 
-    if (c1 == EOF) {
-       (*oconv)(EOF, 0);
-       return 0;
-    }
-
     switch(input_endian){
     case ENDIAN_BIG:
        wc = c2 << 16 | c3 << 8 | c4;
     switch(input_endian){
     case ENDIAN_BIG:
        wc = c2 << 16 | c3 << 8 | c4;
@@ -2197,8 +2491,48 @@ nkf_iconv_utf_32(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
     default:
        return NKF_ICONV_INVALID_CODE_RANGE;
     }
     default:
        return NKF_ICONV_INVALID_CODE_RANGE;
     }
+    return wc;
+}
+
+static size_t
+nkf_iconv_utf_32(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
+{
+    nkf_char wc;
+
+    if (c1 == EOF) {
+       (*oconv)(EOF, 0);
+       return 0;
+    }
+
+    wc = utf32_to_nkf_char(c1, c2, c3, c4);
+    if (wc < 0)
+       return wc;
+
+    return (*unicode_iconv)(wc, FALSE);
+}
+
+static nkf_char
+nkf_iconv_utf_32_combine(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4, nkf_char c5, nkf_char c6, nkf_char c7, nkf_char c8)
+{
+    nkf_char wc, wc2;
+
+    wc = utf32_to_nkf_char(c1, c2, c3, c4);
+    if (wc < 0)
+       return wc;
+    wc2 = utf32_to_nkf_char(c5, c6, c7, c8);
+    if (wc2 < 0)
+       return wc2;
+
+    return unicode_iconv_combine(wc, wc2);
+}
 
 
-    return (*unicode_iconv)(wc);
+static size_t
+nkf_iconv_utf_32_nocombine(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4)
+{
+    nkf_char wc;
+
+    wc = utf32_to_nkf_char(c1, c2, c3, c4);
+    return (*unicode_iconv)(wc, TRUE);
 }
 #endif
 
 }
 #endif
 
@@ -2453,11 +2787,19 @@ s_oconv(nkf_char c2, nkf_char c1)
 }
 
 #ifdef UTF8_OUTPUT_ENABLE
 }
 
 #ifdef UTF8_OUTPUT_ENABLE
+#define OUTPUT_UTF8(val) do { \
+       nkf_unicode_to_utf8(val, &c1, &c2, &c3, &c4); \
+       (*o_putc)(c1); \
+       if (c2) (*o_putc)(c2); \
+       if (c3) (*o_putc)(c3); \
+       if (c4) (*o_putc)(c4); \
+    } while (0)
+
 static void
 w_oconv(nkf_char c2, nkf_char c1)
 {
     nkf_char c3, c4;
 static void
 w_oconv(nkf_char c2, nkf_char c1)
 {
     nkf_char c3, c4;
-    nkf_char val;
+    nkf_char val, val2;
 
     if (output_bom_f) {
        output_bom_f = FALSE;
 
     if (output_bom_f) {
        output_bom_f = FALSE;
@@ -2473,11 +2815,7 @@ w_oconv(nkf_char c2, nkf_char c1)
 
     if (c2 == 0 && nkf_char_unicode_p(c1)){
        val = c1 & VALUE_MASK;
 
     if (c2 == 0 && nkf_char_unicode_p(c1)){
        val = c1 & VALUE_MASK;
-       nkf_unicode_to_utf8(val, &c1, &c2, &c3, &c4);
-       (*o_putc)(c1);
-       if (c2) (*o_putc)(c2);
-       if (c3) (*o_putc)(c3);
-       if (c4) (*o_putc)(c4);
+       OUTPUT_UTF8(val);
        return;
     }
 
        return;
     }
 
@@ -2486,27 +2824,46 @@ w_oconv(nkf_char c2, nkf_char c1)
     } else {
        val = e2w_conv(c2, c1);
        if (val){
     } else {
        val = e2w_conv(c2, c1);
        if (val){
-           nkf_unicode_to_utf8(val, &c1, &c2, &c3, &c4);
-           (*o_putc)(c1);
-           if (c2) (*o_putc)(c2);
-           if (c3) (*o_putc)(c3);
-           if (c4) (*o_putc)(c4);
+           val2 = e2w_combining(val, c2, c1);
+           if (val2)
+               OUTPUT_UTF8(val2);
+           OUTPUT_UTF8(val);
        }
     }
 }
 
        }
     }
 }
 
+#define OUTPUT_UTF16_BYTES(c1, c2) do { \
+       if (output_endian == ENDIAN_LITTLE){ \
+           (*o_putc)(c1); \
+           (*o_putc)(c2); \
+       }else{ \
+           (*o_putc)(c2); \
+           (*o_putc)(c1); \
+       } \
+    } while (0)
+
+#define OUTPUT_UTF16(val) do { \
+       if (nkf_char_unicode_bmp_p(val)) { \
+           c2 = (val >> 8) & 0xff; \
+           c1 = val & 0xff; \
+           OUTPUT_UTF16_BYTES(c1, c2); \
+       } else { \
+           val &= VALUE_MASK; \
+           if (val <= UNICODE_MAX) { \
+               c2 = (val >> 10) + NKF_INT32_C(0xD7C0);   /* high surrogate */ \
+               c1 = (val & 0x3FF) + NKF_INT32_C(0xDC00); /* low surrogate */ \
+               OUTPUT_UTF16_BYTES(c2 & 0xff, (c2 >> 8) & 0xff); \
+               OUTPUT_UTF16_BYTES(c1 & 0xff, (c1 >> 8) & 0xff); \
+           } \
+       } \
+    } while (0)
+
 static void
 w_oconv16(nkf_char c2, nkf_char c1)
 {
     if (output_bom_f) {
        output_bom_f = FALSE;
 static void
 w_oconv16(nkf_char c2, nkf_char c1)
 {
     if (output_bom_f) {
        output_bom_f = FALSE;
-       if (output_endian == ENDIAN_LITTLE){
-           (*o_putc)(0xFF);
-           (*o_putc)(0xFE);
-       }else{
-           (*o_putc)(0xFE);
-           (*o_putc)(0xFF);
-       }
+       OUTPUT_UTF16_BYTES(0xFF, 0xFE);
     }
 
     if (c2 == EOF) {
     }
 
     if (c2 == EOF) {
@@ -2515,44 +2872,34 @@ w_oconv16(nkf_char c2, nkf_char c1)
     }
 
     if (c2 == 0 && nkf_char_unicode_p(c1)) {
     }
 
     if (c2 == 0 && nkf_char_unicode_p(c1)) {
-       if (nkf_char_unicode_bmp_p(c1)) {
-           c2 = (c1 >> 8) & 0xff;
-           c1 &= 0xff;
-       } else {
-           c1 &= VALUE_MASK;
-           if (c1 <= UNICODE_MAX) {
-               c2 = (c1 >> 10) + NKF_INT32_C(0xD7C0);   /* high surrogate */
-               c1 = (c1 & 0x3FF) + NKF_INT32_C(0xDC00); /* low surrogate */
-               if (output_endian == ENDIAN_LITTLE){
-                   (*o_putc)(c2 & 0xff);
-                   (*o_putc)((c2 >> 8) & 0xff);
-                   (*o_putc)(c1 & 0xff);
-                   (*o_putc)((c1 >> 8) & 0xff);
-               }else{
-                   (*o_putc)((c2 >> 8) & 0xff);
-                   (*o_putc)(c2 & 0xff);
-                   (*o_putc)((c1 >> 8) & 0xff);
-                   (*o_putc)(c1 & 0xff);
-               }
-           }
-           return;
-       }
+       OUTPUT_UTF16(c1);
     } else if (c2) {
     } else if (c2) {
-       nkf_char val = e2w_conv(c2, c1);
-       c2 = (val >> 8) & 0xff;
-       c1 = val & 0xff;
+       nkf_char val, val2;
+       val = e2w_conv(c2, c1);
        if (!val) return;
        if (!val) return;
-    }
-
-    if (output_endian == ENDIAN_LITTLE){
-       (*o_putc)(c1);
-       (*o_putc)(c2);
-    }else{
-       (*o_putc)(c2);
-       (*o_putc)(c1);
+       val2 = e2w_combining(val, c2, c1);
+       if (val2)
+           OUTPUT_UTF16(val2);
+       OUTPUT_UTF16(val);
+    } else {
+       OUTPUT_UTF16_BYTES(c1, c2);
     }
 }
 
     }
 }
 
+#define OUTPUT_UTF32(c) do { \
+       if (output_endian == ENDIAN_LITTLE){ \
+           (*o_putc)( (c)        & 0xFF); \
+           (*o_putc)(((c) >>  8) & 0xFF); \
+           (*o_putc)(((c) >> 16) & 0xFF); \
+           (*o_putc)(0); \
+       }else{ \
+           (*o_putc)(0); \
+           (*o_putc)(((c) >> 16) & 0xFF); \
+           (*o_putc)(((c) >>  8) & 0xFF); \
+           (*o_putc)( (c)        & 0xFF); \
+       } \
+    } while (0)
+
 static void
 w_oconv32(nkf_char c2, nkf_char c1)
 {
 static void
 w_oconv32(nkf_char c2, nkf_char c1)
 {
@@ -2581,46 +2928,63 @@ w_oconv32(nkf_char c2, nkf_char c1)
     } else if (c2 == 0 && nkf_char_unicode_p(c1)) {
        c1 &= VALUE_MASK;
     } else if (c2) {
     } else if (c2 == 0 && nkf_char_unicode_p(c1)) {
        c1 &= VALUE_MASK;
     } else if (c2) {
-       c1 = e2w_conv(c2, c1);
-       if (!c1) return;
-    }
-    if (output_endian == ENDIAN_LITTLE){
-       (*o_putc)( c1        & 0xFF);
-       (*o_putc)((c1 >>  8) & 0xFF);
-       (*o_putc)((c1 >> 16) & 0xFF);
-       (*o_putc)(0);
-    }else{
-       (*o_putc)(0);
-       (*o_putc)((c1 >> 16) & 0xFF);
-       (*o_putc)((c1 >>  8) & 0xFF);
-       (*o_putc)( c1        & 0xFF);
+       nkf_char val, val2;
+       val = e2w_conv(c2, c1);
+       if (!val) return;
+       val2 = e2w_combining(val, c2, c1);
+       if (val2)
+           OUTPUT_UTF32(val2);
+       c1 = val;
     }
     }
+    OUTPUT_UTF32(c1);
 }
 #endif
 
 }
 #endif
 
-#define SCORE_L2       (1)                   /* \e$BBh\e(B2\e$B?e=`4A;z\e(B */
-#define SCORE_KANA     (SCORE_L2 << 1)       /* \e$B$$$o$f$kH>3Q%+%J\e(B */
-#define SCORE_DEPEND   (SCORE_KANA << 1)     /* \e$B5!<o0MB8J8;z\e(B */
-#define SCORE_CP932    (SCORE_DEPEND << 1)   /* CP932 \e$B$K$h$kFI$_49$(\e(B (IBM extended characters) */
+#define SCORE_L2       (1)                   /* Kanji Level 2 */
+#define SCORE_KANA     (SCORE_L2 << 1)       /* Halfwidth Katakana */
+#define SCORE_DEPEND   (SCORE_KANA << 1)     /* MD Characters */
+#define SCORE_CP932    (SCORE_DEPEND << 1)   /* IBM extended characters */
 #define SCORE_X0212    (SCORE_CP932 << 1)    /* JIS X 0212 */
 #define SCORE_X0212    (SCORE_CP932 << 1)    /* JIS X 0212 */
-#define SCORE_NO_EXIST (SCORE_X0212 << 1)    /* \e$BB8:_$7$J$$J8;z\e(B */
-#define SCORE_iMIME    (SCORE_NO_EXIST << 1) /* MIME \e$B$K$h$k;XDj\e(B */
-#define SCORE_ERROR    (SCORE_iMIME << 1) /* \e$B%(%i!<\e(B */
+#define SCORE_X0213    (SCORE_X0212 << 1)    /* JIS X 0213 */
+#define SCORE_NO_EXIST (SCORE_X0213 << 1)    /* Undefined Characters */
+#define SCORE_iMIME    (SCORE_NO_EXIST << 1) /* MIME selected */
+#define SCORE_ERROR    (SCORE_iMIME << 1) /* Error */
 
 #define SCORE_INIT (SCORE_iMIME)
 
 
 #define SCORE_INIT (SCORE_iMIME)
 
-static const char score_table_A0[] = {
+static const nkf_char score_table_A0[] = {
     0, 0, 0, 0,
     0, 0, 0, 0,
     0, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
     0, 0, 0, 0,
     0, 0, 0, 0,
     0, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
-    SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_NO_EXIST,
+    SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_X0213,
 };
 
 };
 
-static const char score_table_F0[] = {
+static const nkf_char score_table_F0[] = {
     SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
     SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
-    SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
+    SCORE_L2, SCORE_DEPEND, SCORE_X0213, SCORE_X0213,
     SCORE_DEPEND, SCORE_DEPEND, SCORE_CP932, SCORE_CP932,
     SCORE_DEPEND, SCORE_DEPEND, SCORE_CP932, SCORE_CP932,
-    SCORE_CP932, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR,
+    SCORE_CP932, SCORE_X0213, SCORE_X0213, SCORE_ERROR,
+};
+
+static const nkf_char score_table_8FA0[] = {
+    0, SCORE_X0213, SCORE_X0212, SCORE_X0213,
+    SCORE_X0213, SCORE_X0213, SCORE_X0212, SCORE_X0212,
+    SCORE_X0213, SCORE_X0212, SCORE_X0212, SCORE_X0212,
+    SCORE_X0213, SCORE_X0213, SCORE_X0213, SCORE_X0213,
+};
+
+static const nkf_char score_table_8FE0[] = {
+    SCORE_X0212, SCORE_X0212, SCORE_X0212, SCORE_X0212,
+    SCORE_X0212, SCORE_X0212, SCORE_X0212, SCORE_X0212,
+    SCORE_X0212, SCORE_X0212, SCORE_X0212, SCORE_X0212,
+    SCORE_X0212, SCORE_X0212, SCORE_X0213, SCORE_X0213,
+};
+
+static const nkf_char score_table_8FF0[] = {
+    SCORE_X0213, SCORE_X0213, SCORE_X0213, SCORE_X0212,
+    SCORE_X0212, SCORE_X0213, SCORE_X0213, SCORE_X0213,
+    SCORE_X0213, SCORE_X0213, SCORE_X0213, SCORE_X0213,
+    SCORE_X0213, SCORE_X0213, SCORE_X0213, SCORE_X0213,
 };
 
 static void
 };
 
 static void
@@ -2643,15 +3007,21 @@ static void
 code_score(struct input_code *ptr)
 {
     nkf_char c2 = ptr->buf[0];
 code_score(struct input_code *ptr)
 {
     nkf_char c2 = ptr->buf[0];
-#ifdef UTF8_OUTPUT_ENABLE
     nkf_char c1 = ptr->buf[1];
     nkf_char c1 = ptr->buf[1];
-#endif
     if (c2 < 0){
        set_code_score(ptr, SCORE_ERROR);
     }else if (c2 == SS2){
        set_code_score(ptr, SCORE_KANA);
     }else if (c2 == 0x8f){
     if (c2 < 0){
        set_code_score(ptr, SCORE_ERROR);
     }else if (c2 == SS2){
        set_code_score(ptr, SCORE_KANA);
     }else if (c2 == 0x8f){
-       set_code_score(ptr, SCORE_X0212);
+       if ((c1 & 0x70) == 0x20){
+           set_code_score(ptr, score_table_8FA0[c1 & 0x0f]);
+       }else if ((c1 & 0x70) == 0x60){
+           set_code_score(ptr, score_table_8FE0[c1 & 0x0f]);
+       }else if ((c1 & 0x70) == 0x70){
+           set_code_score(ptr, score_table_8FF0[c1 & 0x0f]);
+       }else{
+           set_code_score(ptr, SCORE_X0212);
+       }
 #ifdef UTF8_OUTPUT_ENABLE
     }else if (!e2w_conv(c2, c1)){
        set_code_score(ptr, SCORE_NO_EXIST);
 #ifdef UTF8_OUTPUT_ENABLE
     }else if (!e2w_conv(c2, c1)){
        set_code_score(ptr, SCORE_NO_EXIST);
@@ -2925,24 +3295,51 @@ code_status(nkf_char c)
     }
 }
 
     }
 }
 
+typedef struct {
+    nkf_buf_t *std_gc_buf;
+    nkf_char broken_state;
+    nkf_buf_t *broken_buf;
+    nkf_char mimeout_state;
+    nkf_buf_t *nfc_buf;
+} nkf_state_t;
+
+static nkf_state_t *nkf_state = NULL;
+
+#define STD_GC_BUFSIZE (256)
+
+static void
+nkf_state_init(void)
+{
+    if (nkf_state) {
+       nkf_buf_clear(nkf_state->std_gc_buf);
+       nkf_buf_clear(nkf_state->broken_buf);
+       nkf_buf_clear(nkf_state->nfc_buf);
+    }
+    else {
+       nkf_state = nkf_xmalloc(sizeof(nkf_state_t));
+       nkf_state->std_gc_buf = nkf_buf_new(STD_GC_BUFSIZE);
+       nkf_state->broken_buf = nkf_buf_new(3);
+       nkf_state->nfc_buf = nkf_buf_new(9);
+    }
+    nkf_state->broken_state = 0;
+    nkf_state->mimeout_state = 0;
+}
+
 #ifndef WIN32DLL
 static nkf_char
 std_getc(FILE *f)
 {
 #ifndef WIN32DLL
 static nkf_char
 std_getc(FILE *f)
 {
-    if (std_gc_ndx){
-       return std_gc_buf[--std_gc_ndx];
+    if (!nkf_buf_empty_p(nkf_state->std_gc_buf)){
+       return nkf_buf_pop(nkf_state->std_gc_buf);
     }
     return getc(f);
 }
 #endif /*WIN32DLL*/
 
 static nkf_char
     }
     return getc(f);
 }
 #endif /*WIN32DLL*/
 
 static nkf_char
-std_ungetc(nkf_char c, FILE *f)
+std_ungetc(nkf_char c, ARG_UNUSED FILE *f)
 {
 {
-    if (std_gc_ndx == STD_GC_BUFSIZE){
-       return EOF;
-    }
-    std_gc_buf[std_gc_ndx++] = c;
+    nkf_buf_push(nkf_state->std_gc_buf, c);
     return c;
 }
 
     return c;
 }
 
@@ -2955,23 +3352,24 @@ std_putc(nkf_char c)
 }
 #endif /*WIN32DLL*/
 
 }
 #endif /*WIN32DLL*/
 
-static unsigned char   hold_buf[HOLD_SIZE*2];
+static nkf_char   hold_buf[HOLD_SIZE*2];
 static int             hold_count = 0;
 static nkf_char
 push_hold_buf(nkf_char c2)
 {
     if (hold_count >= HOLD_SIZE*2)
        return (EOF);
 static int             hold_count = 0;
 static nkf_char
 push_hold_buf(nkf_char c2)
 {
     if (hold_count >= HOLD_SIZE*2)
        return (EOF);
-    hold_buf[hold_count++] = (unsigned char)c2;
+    hold_buf[hold_count++] = c2;
     return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count);
 }
 
 static int
     return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count);
 }
 
 static int
-h_conv(FILE *f, int c1, int c2)
+h_conv(FILE *f, nkf_char c1, nkf_char c2)
 {
 {
-    int ret, c4, c3;
+    int ret;
     int hold_index;
     int hold_index;
-
+    int fromhold_count;
+    nkf_char c3, c4;
 
     /** it must NOT be in the kanji shifte sequence      */
     /** it must NOT be written in JIS7                   */
 
     /** it must NOT be in the kanji shifte sequence      */
     /** it must NOT be written in JIS7                   */
@@ -3021,15 +3419,21 @@ h_conv(FILE *f, int c1, int c2)
     hold_index = 0;
     while (hold_index < hold_count){
        c1 = hold_buf[hold_index++];
     hold_index = 0;
     while (hold_index < hold_count){
        c1 = hold_buf[hold_index++];
-       if (c1 <= DEL){
+       if (nkf_char_unicode_p(c1)) {
+           (*oconv)(0, c1);
+           continue;
+       }
+       else if (c1 <= DEL){
            (*iconv)(0, c1, 0);
            continue;
        }else if (iconv == s_iconv && 0xa1 <= c1 && c1 <= 0xdf){
            (*iconv)(JIS_X_0201_1976_K, c1, 0);
            continue;
        }
            (*iconv)(0, c1, 0);
            continue;
        }else if (iconv == s_iconv && 0xa1 <= c1 && c1 <= 0xdf){
            (*iconv)(JIS_X_0201_1976_K, c1, 0);
            continue;
        }
+       fromhold_count = 1;
        if (hold_index < hold_count){
            c2 = hold_buf[hold_index++];
        if (hold_index < hold_count){
            c2 = hold_buf[hold_index++];
+           fromhold_count++;
        }else{
            c2 = (*i_getc)(f);
            if (c2 == EOF){
        }else{
            c2 = (*i_getc)(f);
            if (c2 == EOF){
@@ -3047,16 +3451,47 @@ h_conv(FILE *f, int c1, int c2)
            } else if ((c3 = (*i_getc)(f)) == EOF) {
                ret = EOF;
                break;
            } else if ((c3 = (*i_getc)(f)) == EOF) {
                ret = EOF;
                break;
-           } else {
-               code_status(c3);
-               if (hold_index < hold_count){
-                   c4 = hold_buf[hold_index++];
-               } else if ((c4 = (*i_getc)(f)) == EOF) {
-                   c3 = ret = EOF;
-                   break;
+           }
+           code_status(c3);
+           if (hold_index < hold_count){
+               c4 = hold_buf[hold_index++];
+           } else if ((c4 = (*i_getc)(f)) == EOF) {
+               c3 = ret = EOF;
+               break;
+           }
+           code_status(c4);
+           (*iconv)(c1, c2, (c3<<8)|c4);
+           break;
+       case -3:
+           /* 4 bytes UTF-8 (check combining character) */
+           if (hold_index < hold_count){
+               c3 = hold_buf[hold_index++];
+               fromhold_count++;
+           } else if ((c3 = (*i_getc)(f)) == EOF) {
+               w_iconv_nocombine(c1, c2, 0);
+               break;
+           }
+           if (hold_index < hold_count){
+               c4 = hold_buf[hold_index++];
+               fromhold_count++;
+           } else if ((c4 = (*i_getc)(f)) == EOF) {
+               w_iconv_nocombine(c1, c2, 0);
+               if (fromhold_count <= 2)
+                   (*i_ungetc)(c3,f);
+               else
+                   hold_index--;
+               continue;
+           }
+           if (w_iconv_combine(c1, c2, 0, c3, c4, 0)) {
+               w_iconv_nocombine(c1, c2, 0);
+               if (fromhold_count <= 2) {
+                   (*i_ungetc)(c4,f);
+                   (*i_ungetc)(c3,f);
+               } else if (fromhold_count == 3) {
+                   (*i_ungetc)(c4,f);
+                   hold_index--;
                } else {
                } else {
-                   code_status(c4);
-                   (*iconv)(c1, c2, (c3<<8)|c4);
+                   hold_index -= 2;
                }
            }
            break;
                }
            }
            break;
@@ -3064,13 +3499,68 @@ h_conv(FILE *f, int c1, int c2)
            /* 3 bytes EUC or UTF-8 */
            if (hold_index < hold_count){
                c3 = hold_buf[hold_index++];
            /* 3 bytes EUC or UTF-8 */
            if (hold_index < hold_count){
                c3 = hold_buf[hold_index++];
+               fromhold_count++;
            } else if ((c3 = (*i_getc)(f)) == EOF) {
                ret = EOF;
                break;
            } else {
                code_status(c3);
            }
            } else if ((c3 = (*i_getc)(f)) == EOF) {
                ret = EOF;
                break;
            } else {
                code_status(c3);
            }
-           (*iconv)(c1, c2, c3);
+           if ((*iconv)(c1, c2, c3) == -3) {
+               /* 6 bytes UTF-8 (check combining character) */
+               nkf_char c5, c6;
+               if (hold_index < hold_count){
+                   c4 = hold_buf[hold_index++];
+                   fromhold_count++;
+               } else if ((c4 = (*i_getc)(f)) == EOF) {
+                   w_iconv_nocombine(c1, c2, c3);
+                   continue;
+               }
+               if (hold_index < hold_count){
+                   c5 = hold_buf[hold_index++];
+                   fromhold_count++;
+               } else if ((c5 = (*i_getc)(f)) == EOF) {
+                   w_iconv_nocombine(c1, c2, c3);
+                   if (fromhold_count == 4)
+                       hold_index--;
+                   else
+                       (*i_ungetc)(c4,f);
+                   continue;
+               }
+               if (hold_index < hold_count){
+                   c6 = hold_buf[hold_index++];
+                   fromhold_count++;
+               } else if ((c6 = (*i_getc)(f)) == EOF) {
+                   w_iconv_nocombine(c1, c2, c3);
+                   if (fromhold_count == 5) {
+                       hold_index -= 2;
+                   } else if (fromhold_count == 4) {
+                       hold_index--;
+                       (*i_ungetc)(c5,f);
+                   } else {
+                       (*i_ungetc)(c5,f);
+                       (*i_ungetc)(c4,f);
+                   }
+                   continue;
+               }
+               if (w_iconv_combine(c1, c2, c3, c4, c5, c6)) {
+                   w_iconv_nocombine(c1, c2, c3);
+                   if (fromhold_count == 6) {
+                       hold_index -= 3;
+                   } else if (fromhold_count == 5) {
+                       hold_index -= 2;
+                       (*i_ungetc)(c6,f);
+                   } else if (fromhold_count == 4) {
+                       hold_index--;
+                       (*i_ungetc)(c6,f);
+                       (*i_ungetc)(c5,f);
+                   } else {
+                       (*i_ungetc)(c6,f);
+                       (*i_ungetc)(c5,f);
+                       (*i_ungetc)(c4,f);
+                   }
+               }
+           }
            break;
        }
        if (c3 == EOF) break;
            break;
        }
        if (c3 == EOF) break;
@@ -3085,6 +3575,7 @@ static void
 check_bom(FILE *f)
 {
     int c2;
 check_bom(FILE *f)
 {
     int c2;
+    input_bom_f = FALSE;
     switch(c2 = (*i_getc)(f)){
     case 0x00:
        if((c2 = (*i_getc)(f)) == 0x00){
     switch(c2 = (*i_getc)(f)){
     case 0x00:
        if((c2 = (*i_getc)(f)) == 0x00){
@@ -3094,6 +3585,7 @@ check_bom(FILE *f)
                        set_iconv(TRUE, w_iconv32);
                    }
                    if (iconv == w_iconv32) {
                        set_iconv(TRUE, w_iconv32);
                    }
                    if (iconv == w_iconv32) {
+                       input_bom_f = TRUE;
                        input_endian = ENDIAN_BIG;
                        return;
                    }
                        input_endian = ENDIAN_BIG;
                        return;
                    }
@@ -3124,6 +3616,7 @@ check_bom(FILE *f)
                    set_iconv(TRUE, w_iconv);
                }
                if (iconv == w_iconv) {
                    set_iconv(TRUE, w_iconv);
                }
                if (iconv == w_iconv) {
+                   input_bom_f = TRUE;
                    return;
                }
                (*i_ungetc)(0xBF,f);
                    return;
                }
                (*i_ungetc)(0xBF,f);
@@ -3152,6 +3645,7 @@ check_bom(FILE *f)
            }
            if (iconv == w_iconv16) {
                input_endian = ENDIAN_BIG;
            }
            if (iconv == w_iconv16) {
                input_endian = ENDIAN_BIG;
+               input_bom_f = TRUE;
                return;
            }
            (*i_ungetc)(0xFF,f);
                return;
            }
            (*i_ungetc)(0xFF,f);
@@ -3167,6 +3661,7 @@ check_bom(FILE *f)
                    }
                    if (iconv == w_iconv32) {
                        input_endian = ENDIAN_LITTLE;
                    }
                    if (iconv == w_iconv32) {
                        input_endian = ENDIAN_LITTLE;
+                       input_bom_f = TRUE;
                        return;
                    }
                    (*i_ungetc)(0x00,f);
                        return;
                    }
                    (*i_ungetc)(0x00,f);
@@ -3178,6 +3673,7 @@ check_bom(FILE *f)
            }
            if (iconv == w_iconv16) {
                input_endian = ENDIAN_LITTLE;
            }
            if (iconv == w_iconv16) {
                input_endian = ENDIAN_LITTLE;
+               input_bom_f = TRUE;
                return;
            }
            (*i_ungetc)(0xFE,f);
                return;
            }
            (*i_ungetc)(0xFE,f);
@@ -3190,74 +3686,50 @@ check_bom(FILE *f)
     }
 }
 
     }
 }
 
-static struct {
-    int count;
-    nkf_char status;
-    nkf_char buf[3];
-} broken_state;
-
-static void
-init_broken_state(void)
-{
-    memset(&broken_state, 0, sizeof(broken_state));
-}
-
-static void
-push_broken_buf(c)
-{
-    broken_state.buf[broken_state.count++] = c;
-}
-
-static nkf_char
-pop_broken_buf(void)
-{
-    return broken_state.buf[--broken_state.count];
-}
-
 static nkf_char
 broken_getc(FILE *f)
 {
     nkf_char c, c1;
 
 static nkf_char
 broken_getc(FILE *f)
 {
     nkf_char c, c1;
 
-    if (broken_state.count > 0) {
-       return pop_broken_buf();
+    if (!nkf_buf_empty_p(nkf_state->broken_buf)) {
+       return nkf_buf_pop(nkf_state->broken_buf);
     }
     c = (*i_bgetc)(f);
     }
     c = (*i_bgetc)(f);
-    if (c=='$' && broken_state.status != ESC
+    if (c=='$' && nkf_state->broken_state != ESC
        && (input_mode == ASCII || input_mode == JIS_X_0201_1976_K)) {
        c1= (*i_bgetc)(f);
        && (input_mode == ASCII || input_mode == JIS_X_0201_1976_K)) {
        c1= (*i_bgetc)(f);
-       broken_state.status = 0;
+       nkf_state->broken_state = 0;
        if (c1=='@'|| c1=='B') {
        if (c1=='@'|| c1=='B') {
-           push_broken_buf(c1);
-           push_broken_buf(c);
+           nkf_buf_push(nkf_state->broken_buf, c1);
+           nkf_buf_push(nkf_state->broken_buf, c);
            return ESC;
        } else {
            (*i_bungetc)(c1,f);
            return c;
        }
            return ESC;
        } else {
            (*i_bungetc)(c1,f);
            return c;
        }
-    } else if (c=='(' && broken_state.status != ESC
+    } else if (c=='(' && nkf_state->broken_state != ESC
               && (input_mode == JIS_X_0208 || input_mode == JIS_X_0201_1976_K)) {
        c1= (*i_bgetc)(f);
               && (input_mode == JIS_X_0208 || input_mode == JIS_X_0201_1976_K)) {
        c1= (*i_bgetc)(f);
-       broken_state.status = 0;
+       nkf_state->broken_state = 0;
        if (c1=='J'|| c1=='B') {
        if (c1=='J'|| c1=='B') {
-           push_broken_buf(c1);
-           push_broken_buf(c);
+           nkf_buf_push(nkf_state->broken_buf, c1);
+           nkf_buf_push(nkf_state->broken_buf, c);
            return ESC;
        } else {
            (*i_bungetc)(c1,f);
            return c;
        }
     } else {
            return ESC;
        } else {
            (*i_bungetc)(c1,f);
            return c;
        }
     } else {
-       broken_state.status = c;
+       nkf_state->broken_state = c;
        return c;
     }
 }
 
 static nkf_char
        return c;
     }
 }
 
 static nkf_char
-broken_ungetc(nkf_char c, FILE *f)
+broken_ungetc(nkf_char c, ARG_UNUSED FILE *f)
 {
 {
-    if (broken_state.count < 2)
-       push_broken_buf(c);
+    if (nkf_buf_length(nkf_state->broken_buf) < 2)
+       nkf_buf_push(nkf_state->broken_buf, c);
     return c;
 }
 
     return c;
 }
 
@@ -3282,6 +3754,40 @@ eol_conv(nkf_char c2, nkf_char c1)
     else if (c2 != 0 || c1 != LF) (*o_eol_conv)(c2, c1);
 }
 
     else if (c2 != 0 || c1 != LF) (*o_eol_conv)(c2, c1);
 }
 
+static void
+put_newline(void (*func)(nkf_char))
+{
+    switch (eolmode_f ? eolmode_f : DEFAULT_NEWLINE) {
+      case CRLF:
+       (*func)(0x0D);
+       (*func)(0x0A);
+       break;
+      case CR:
+       (*func)(0x0D);
+       break;
+      case LF:
+       (*func)(0x0A);
+       break;
+    }
+}
+
+static void
+oconv_newline(void (*func)(nkf_char, nkf_char))
+{
+    switch (eolmode_f ? eolmode_f : DEFAULT_NEWLINE) {
+      case CRLF:
+       (*func)(0, 0x0D);
+       (*func)(0, 0x0A);
+       break;
+      case CR:
+       (*func)(0, 0x0D);
+       break;
+      case LF:
+       (*func)(0, 0x0A);
+       break;
+    }
+}
+
 /*
    Return value of fold_conv()
 
 /*
    Return value of fold_conv()
 
@@ -3328,8 +3834,8 @@ fold_conv(nkf_char c2, nkf_char c1)
            f_prev = c1;
            f_line = 0;
            fold_state =  CR;
            f_prev = c1;
            f_line = 0;
            fold_state =  CR;
-       } else if ((f_prev == c1 && !fold_preserve_f)
-                  || (f_prev == LF && fold_preserve_f)
+       } else if ((f_prev == c1)
+                  || (f_prev == LF)
                  ) {        /* duplicate newline */
            if (f_line) {
                f_line = 0;
                  ) {        /* duplicate newline */
            if (f_line) {
                f_line = 0;
@@ -3358,9 +3864,7 @@ fold_conv(nkf_char c2, nkf_char c1)
        f_prev = LF;
        f_line = 0;
        fold_state =  LF;            /* output newline and clear */
        f_prev = LF;
        f_line = 0;
        fold_state =  LF;            /* output newline and clear */
-    } else if ( (c2==0  && c1==SP)||
-              (c2==0  && c1==TAB)||
-              (c2=='!'&& c1=='!')) {
+    } else if ((c2==0 && nkf_isblank(c1)) || (c2 == '!' && c1 == '!')) {
        /* X0208 kankaku or ascii space */
        if (f_prev == SP) {
            fold_state = 0;         /* remove duplicate spaces */
        /* X0208 kankaku or ascii space */
        if (f_prev == SP) {
            fold_state = 0;         /* remove duplicate spaces */
@@ -3378,7 +3882,7 @@ fold_conv(nkf_char c2, nkf_char c1)
        f_prev = c1;
        if (c2 || c2 == JIS_X_0201_1976_K)
            f_prev |= 0x80;  /* this is Japanese */
        f_prev = c1;
        if (c2 || c2 == JIS_X_0201_1976_K)
            f_prev |= 0x80;  /* this is Japanese */
-       f_line += char_size(c2,c1);
+       f_line += c2 == JIS_X_0201_1976_K ? 1: char_size(c2,c1);
        if (f_line<=fold_len) {   /* normal case */
            fold_state = 1;
        } else {
        if (f_line<=fold_len) {   /* normal case */
            fold_state = 1;
        } else {
@@ -3456,13 +3960,13 @@ fold_conv(nkf_char c2, nkf_char c1)
     /* terminator process */
     switch(fold_state) {
     case LF:
     /* terminator process */
     switch(fold_state) {
     case LF:
-       OCONV_NEWLINE((*o_fconv));
+       oconv_newline(o_fconv);
        (*o_fconv)(c2,c1);
        break;
     case 0:
        return;
     case CR:
        (*o_fconv)(c2,c1);
        break;
     case 0:
        return;
     case CR:
-       OCONV_NEWLINE((*o_fconv));
+       oconv_newline(o_fconv);
        break;
     case TAB:
     case SP:
        break;
     case TAB:
     case SP:
@@ -3497,13 +4001,17 @@ z_conv(nkf_char c2, nkf_char c1)
                    z_prev2 = 0;
                    (*o_zconv)(ev[(z_prev1-SP)*2], ev[(z_prev1-SP)*2+1]);
                    return;
                    z_prev2 = 0;
                    (*o_zconv)(ev[(z_prev1-SP)*2], ev[(z_prev1-SP)*2+1]);
                    return;
+               } else if (x0213_f && c1 == (0xdf&0x7f) && ev_x0213[(z_prev1-SP)*2]) {  /* \e$BH>ByE@\e(B */
+                   z_prev2 = 0;
+                   (*o_zconv)(ev_x0213[(z_prev1-SP)*2], ev_x0213[(z_prev1-SP)*2+1]);
+                   return;
                }
            }
            z_prev2 = 0;
            (*o_zconv)(cv[(z_prev1-SP)*2], cv[(z_prev1-SP)*2+1]);
        }
        if (c2 == JIS_X_0201_1976_K) {
                }
            }
            z_prev2 = 0;
            (*o_zconv)(cv[(z_prev1-SP)*2], cv[(z_prev1-SP)*2+1]);
        }
        if (c2 == JIS_X_0201_1976_K) {
-           if (dv[(c1-SP)*2] || ev[(c1-SP)*2]) {
+           if (dv[(c1-SP)*2] || ev[(c1-SP)*2] || (x0213_f && ev_x0213[(c1-SP)*2])) {
                /* wait for \e$BByE@\e(B or \e$BH>ByE@\e(B */
                z_prev1 = c1;
                z_prev2 = c2;
                /* wait for \e$BByE@\e(B or \e$BH>ByE@\e(B */
                z_prev1 = c1;
                z_prev2 = c2;
@@ -3558,7 +4066,7 @@ z_conv(nkf_char c2, nkf_char c1)
     if (alpha_f & 16) {
        /* JIS X 0208 Katakana to JIS X 0201 Katakana */
        if (c2 == 0x21) {
     if (alpha_f & 16) {
        /* JIS X 0208 Katakana to JIS X 0201 Katakana */
        if (c2 == 0x21) {
-           char c = 0;
+           nkf_char c = 0;
            switch (c1) {
            case 0x23:
                /* U+3002 (0x8142) Ideographic Full Stop -> U+FF61 (0xA1) Halfwidth Ideographic Full Stop */
            switch (c1) {
            case 0x23:
                /* U+3002 (0x8142) Ideographic Full Stop -> U+FF61 (0xA1) Halfwidth Ideographic Full Stop */
@@ -3611,8 +4119,8 @@ z_conv(nkf_char c2, nkf_char c1)
                0x4D00, 0x4D5E, 0x4D5F, 0x4E00, 0x4E5E, 0x4E5F, 0x4F00, 0x5000,
                0x5100, 0x5200, 0x5300, 0x2C00, 0x5400, 0x2D00, 0x5500, 0x2E00,
                0x5600, 0x5700, 0x5800, 0x5900, 0x5A00, 0x5B00, 0x0000, 0x5C00,
                0x4D00, 0x4D5E, 0x4D5F, 0x4E00, 0x4E5E, 0x4E5F, 0x4F00, 0x5000,
                0x5100, 0x5200, 0x5300, 0x2C00, 0x5400, 0x2D00, 0x5500, 0x2E00,
                0x5600, 0x5700, 0x5800, 0x5900, 0x5A00, 0x5B00, 0x0000, 0x5C00,
-               0x0000, 0x0000, 0x2600, 0x5D00, 0x335E, 0x0000, 0x0000, 0x0000,
-               0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000
+               0x0000, 0x0000, 0x2600, 0x5D00, 0x335E, 0x0000, 0x0000, 0x365F,
+               0x375F, 0x385F, 0x395F, 0x3A5F, 0x3E5F, 0x425F, 0x445F, 0x0000
            };
            if (fullwidth_to_halfwidth[c1-0x20]){
                c2 = fullwidth_to_halfwidth[c1-0x20];
            };
            if (fullwidth_to_halfwidth[c1-0x20]){
                c2 = fullwidth_to_halfwidth[c1-0x20];
@@ -3622,6 +4130,10 @@ z_conv(nkf_char c2, nkf_char c1)
                }
                return;
            }
                }
                return;
            }
+       } else if (c2 == 0 && nkf_char_unicode_p(c1) &&
+           ((c1&VALUE_MASK) == 0x3099 || (c1&VALUE_MASK) == 0x309A)) { /* \e$B9g@.MQByE@!&H>ByE@\e(B */
+           (*o_zconv)(JIS_X_0201_1976_K, 0x5E + (c1&VALUE_MASK) - 0x3099);
+           return;
        }
     }
     (*o_zconv)(c2,c1);
        }
     }
     (*o_zconv)(c2,c1);
@@ -3749,6 +4261,7 @@ static const unsigned char *mime_pattern[] = {
     (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-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?B?",
     (const unsigned char *)"\075?ISO-2022-JP?Q?",
 #if defined(UTF8_INPUT_ENABLE)
     (const unsigned char *)"\075?UTF-8?B?",
     (const unsigned char *)"\075?ISO-2022-JP?Q?",
 #if defined(UTF8_INPUT_ENABLE)
     (const unsigned char *)"\075?UTF-8?B?",
@@ -3761,7 +4274,7 @@ static const unsigned char *mime_pattern[] = {
 
 /* \e$B3:Ev$9$k%3!<%I$NM%@hEY$r>e$2$k$?$a$NL\0u\e(B */
 nkf_char (*mime_priority_func[])(nkf_char c2, nkf_char c1, nkf_char c0) = {
 
 /* \e$B3:Ev$9$k%3!<%I$NM%@hEY$r>e$2$k$?$a$NL\0u\e(B */
 nkf_char (*mime_priority_func[])(nkf_char c2, nkf_char c1, nkf_char c0) = {
-    e_iconv, s_iconv, 0, 0, 0, 0,
+    e_iconv, s_iconv, 0, 0, 0, 0, 0,
 #if defined(UTF8_INPUT_ENABLE)
     w_iconv, w_iconv,
 #endif
 #if defined(UTF8_INPUT_ENABLE)
     w_iconv, w_iconv,
 #endif
@@ -3769,7 +4282,7 @@ nkf_char (*mime_priority_func[])(nkf_char c2, nkf_char c1, nkf_char c0) = {
 };
 
 static const nkf_char mime_encode[] = {
 };
 
 static const nkf_char mime_encode[] = {
-    EUC_JP, SHIFT_JIS, ISO_8859_1, ISO_8859_1, JIS_X_0208, JIS_X_0201_1976_K,
+    EUC_JP, SHIFT_JIS, ISO_8859_1, ISO_8859_1, JIS_X_0208, JIS_X_0201_1976_K, JIS_X_0201_1976_K,
 #if defined(UTF8_INPUT_ENABLE)
     UTF_8, UTF_8,
 #endif
 #if defined(UTF8_INPUT_ENABLE)
     UTF_8, UTF_8,
 #endif
@@ -3778,7 +4291,7 @@ static const nkf_char mime_encode[] = {
 };
 
 static const nkf_char mime_encode_method[] = {
 };
 
 static const nkf_char mime_encode_method[] = {
-    'B', 'B','Q', 'B', 'B', 'Q',
+    'B', 'B','Q', 'B', 'B', 'B', 'Q',
 #if defined(UTF8_INPUT_ENABLE)
     'B', 'Q',
 #endif
 #if defined(UTF8_INPUT_ENABLE)
     'B', 'Q',
 #endif
@@ -3809,7 +4322,7 @@ mime_input_buf_unshift(nkf_char c)
 }
 
 static nkf_char
 }
 
 static nkf_char
-mime_ungetc(nkf_char c, FILE *f)
+mime_ungetc(nkf_char c, ARG_UNUSED FILE *f)
 {
     mime_input_buf_unshift(c);
     return c;
 {
     mime_input_buf_unshift(c);
     return c;
@@ -3828,7 +4341,7 @@ mime_ungetc_buf(nkf_char c, FILE *f)
 static nkf_char
 mime_getc_buf(FILE *f)
 {
 static nkf_char
 mime_getc_buf(FILE *f)
 {
-    /* we don't keep eof of mime_input_buf, becase it contains ?= as
+    /* we don't keep eof of mime_input_buf, because it contains ?= as
        a terminator. It was checked in mime_integrity. */
     return ((mimebuf_f)?
            (*i_mgetc_buf)(f):mime_input_buf(mime_input_state.input++));
        a terminator. It was checked in mime_integrity. */
     return ((mimebuf_f)?
            (*i_mgetc_buf)(f):mime_input_buf(mime_input_state.input++));
@@ -3954,7 +4467,7 @@ mime_begin_strict(FILE *f)
 static nkf_char
 mime_begin(FILE *f)
 {
 static nkf_char
 mime_begin(FILE *f)
 {
-    nkf_char c1;
+    nkf_char c1 = 0;
     int i,k;
 
     /* In NONSTRICT mode, only =? is checked. In case of failure, we  */
     int i,k;
 
     /* In NONSTRICT mode, only =? is checked. In case of failure, we  */
@@ -4010,7 +4523,7 @@ mime_begin(FILE *f)
 
 #ifdef CHECK_OPTION
 static void
 
 #ifdef CHECK_OPTION
 static void
-no_putc(nkf_char c)
+no_putc(ARG_UNUSED nkf_char c)
 {
     ;
 }
 {
     ;
 }
@@ -4047,7 +4560,9 @@ get_guessed_code(void)
            if (p->score & (SCORE_DEPEND|SCORE_CP932))
                input_codename = "CP932";
        } else if (strcmp(input_codename, "EUC-JP") == 0) {
            if (p->score & (SCORE_DEPEND|SCORE_CP932))
                input_codename = "CP932";
        } else if (strcmp(input_codename, "EUC-JP") == 0) {
-           if (p->score & (SCORE_X0212))
+           if (p->score & SCORE_X0213)
+               input_codename = "EUC-JIS-2004";
+           else if (p->score & (SCORE_X0212))
                input_codename = "EUCJP-MS";
            else if (p->score & (SCORE_DEPEND|SCORE_CP932))
                input_codename = "CP51932";
                input_codename = "EUCJP-MS";
            else if (p->score & (SCORE_DEPEND|SCORE_CP932))
                input_codename = "CP51932";
@@ -4073,8 +4588,13 @@ print_guessed_code(char *filename)
        if (guess_f == 1) {
            printf("%s\n", input_codename);
        } else {
        if (guess_f == 1) {
            printf("%s\n", input_codename);
        } else {
-           printf("%s%s\n",
+           printf("%s%s%s%s\n",
                   input_codename,
                   input_codename,
+                  iconv != w_iconv16 && iconv != w_iconv32 ? "" :
+                  input_endian == ENDIAN_LITTLE ? " LE" :
+                  input_endian == ENDIAN_BIG ? " BE" :
+                  "[BUG]",
+                  input_bom_f ? " (BOM)" : "",
                   input_eol == CR   ? " (CR)" :
                   input_eol == LF   ? " (LF)" :
                   input_eol == CRLF ? " (CRLF)" :
                   input_eol == CR   ? " (CR)" :
                   input_eol == LF   ? " (LF)" :
                   input_eol == CRLF ? " (CRLF)" :
@@ -4142,7 +4662,7 @@ numchar_getc(FILE *f)
     nkf_char (*u)(nkf_char c ,FILE *f) = i_nungetc;
     int i = 0, j;
     nkf_char buf[12];
     nkf_char (*u)(nkf_char c ,FILE *f) = i_nungetc;
     int i = 0, j;
     nkf_char buf[12];
-    long c = -1;
+    nkf_char c = -1;
 
     buf[i] = (*g)(f);
     if (buf[i] == '&'){
 
     buf[i] = (*g)(f);
     if (buf[i] == '&'){
@@ -4198,42 +4718,56 @@ numchar_ungetc(nkf_char c, FILE *f)
 
 #ifdef UNICODE_NORMALIZATION
 
 
 #ifdef UNICODE_NORMALIZATION
 
-/* Normalization Form C */
 static nkf_char
 nfc_getc(FILE *f)
 {
     nkf_char (*g)(FILE *f) = i_nfc_getc;
     nkf_char (*u)(nkf_char c ,FILE *f) = i_nfc_ungetc;
 static nkf_char
 nfc_getc(FILE *f)
 {
     nkf_char (*g)(FILE *f) = i_nfc_getc;
     nkf_char (*u)(nkf_char c ,FILE *f) = i_nfc_ungetc;
-    int i=0, j, k=1, lower, upper;
-    nkf_char buf[9];
+    nkf_buf_t *buf = nkf_state->nfc_buf;
     const unsigned char *array;
     const unsigned char *array;
+    int lower=0, upper=NORMALIZATION_TABLE_LENGTH-1;
+    nkf_char c = (*g)(f);
 
 
-    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;
+    if (c == EOF || c > 0xFF || (c & 0xc0) == 0x80) return c;
+
+    nkf_buf_push(buf, c);
+    do {
+       while (lower <= upper) {
+           int mid = (lower+upper) / 2;
+           int len;
+           array = normalization_table[mid].nfd;
+           for (len=0; len < NORMALIZATION_TABLE_NFD_LENGTH && array[len]; len++) {
+               if (len >= nkf_buf_length(buf)) {
+                   c = (*g)(f);
+                   if (c == EOF) {
+                       len = 0;
+                       lower = 1, upper = 0;
+                       break;
+                   }
+                   nkf_buf_push(buf, c);
+               }
+               if (array[len] != nkf_buf_at(buf, len)) {
+                   if (array[len] < nkf_buf_at(buf, len)) lower = mid + 1;
+                   else  upper = mid - 1;
+                   len = 0;
                    break;
                    break;
-               } else if (k >= i)
-                   buf[++i] = (*g)(f);
+               }
            }
            }
-           if (k > 0){
-               array = normalization_table[j].nfc;
+           if (len > 0) {
+               int i;
+               array = normalization_table[mid].nfc;
+               nkf_buf_clear(buf);
                for (i=0; i < NORMALIZATION_TABLE_NFC_LENGTH && array[i]; i++)
                for (i=0; i < NORMALIZATION_TABLE_NFC_LENGTH && array[i]; i++)
-                   buf[i] = (nkf_char)(array[i]);
-               i--;
+                   nkf_buf_push(buf, array[i]);
                break;
            }
        }
                break;
            }
        }
-       while (i > 0)
-           (*u)(buf[i--], f);
-    }
-    return buf[0];
+    } while (lower <= upper);
+
+    while (nkf_buf_length(buf) > 1) (*u)(nkf_buf_pop(buf), f);
+    c = nkf_buf_pop(buf);
+
+    return c;
 }
 
 static nkf_char
 }
 
 static nkf_char
@@ -4307,17 +4841,13 @@ mime_getc(FILE *f)
            /* end Q encoding */
            input_mode = exit_mode;
            lwsp_count = 0;
            /* end Q encoding */
            input_mode = exit_mode;
            lwsp_count = 0;
-           lwsp_buf = malloc((lwsp_size+5)*sizeof(char));
-           if (lwsp_buf==NULL) {
-               perror("can't malloc");
-               return -1;
-           }
+           lwsp_buf = nkf_xmalloc((lwsp_size+5)*sizeof(char));
            while ((c1=(*i_getc)(f))!=EOF) {
                switch (c1) {
                case LF:
                case CR:
                    if (c1==LF) {
            while ((c1=(*i_getc)(f))!=EOF) {
                switch (c1) {
                case LF:
                case CR:
                    if (c1==LF) {
-                       if ((c1=(*i_getc)(f))!=EOF && (c1==SP||c1==TAB)) {
+                       if ((c1=(*i_getc)(f))!=EOF && nkf_isblank(c1)) {
                            i_ungetc(SP,f);
                            continue;
                        } else {
                            i_ungetc(SP,f);
                            continue;
                        } else {
@@ -4326,7 +4856,7 @@ mime_getc(FILE *f)
                        c1 = LF;
                    } else {
                        if ((c1=(*i_getc)(f))!=EOF && c1 == LF) {
                        c1 = LF;
                    } else {
                        if ((c1=(*i_getc)(f))!=EOF && c1 == LF) {
-                           if ((c1=(*i_getc)(f))!=EOF && (c1==SP||c1==TAB)) {
+                           if ((c1=(*i_getc)(f))!=EOF && nkf_isblank(c1)) {
                                i_ungetc(SP,f);
                                continue;
                            } else {
                                i_ungetc(SP,f);
                                continue;
                            } else {
@@ -4344,12 +4874,7 @@ mime_getc(FILE *f)
                    lwsp_buf[lwsp_count] = (unsigned char)c1;
                    if (lwsp_count++>lwsp_size){
                        lwsp_size <<= 1;
                    lwsp_buf[lwsp_count] = (unsigned char)c1;
                    if (lwsp_count++>lwsp_size){
                        lwsp_size <<= 1;
-                       lwsp_buf_new = realloc(lwsp_buf, (lwsp_size+5)*sizeof(char));
-                       if (lwsp_buf_new==NULL) {
-                           free(lwsp_buf);
-                           perror("can't realloc");
-                           return -1;
-                       }
+                       lwsp_buf_new = nkf_xrealloc(lwsp_buf, (lwsp_size+5)*sizeof(char));
                        lwsp_buf = lwsp_buf_new;
                    }
                    continue;
                        lwsp_buf = lwsp_buf_new;
                    }
                    continue;
@@ -4362,12 +4887,12 @@ mime_getc(FILE *f)
                    i_ungetc(lwsp_buf[lwsp_count],f);
                c1 = lwsp_buf[0];
            }
                    i_ungetc(lwsp_buf[lwsp_count],f);
                c1 = lwsp_buf[0];
            }
-           free(lwsp_buf);
+           nkf_xfree(lwsp_buf);
            return c1;
        }
        if (c1=='='&&c2<SP) { /* this is soft wrap */
            while((c1 =  (*i_mgetc)(f)) <=SP) {
            return c1;
        }
        if (c1=='='&&c2<SP) { /* this is soft wrap */
            while((c1 =  (*i_mgetc)(f)) <=SP) {
-               if ((c1 = (*i_mgetc)(f)) == EOF) return (EOF);
+               if (c1 == EOF) return (EOF);
            }
            mime_decode_mode = 'Q'; /* still in MIME */
            goto restart_mime_q;
            }
            mime_decode_mode = 'Q'; /* still in MIME */
            goto restart_mime_q;
@@ -4415,17 +4940,13 @@ mime_getc(FILE *f)
     if ((c1 == '?') && (c2 == '=')) {
        input_mode = ASCII;
        lwsp_count = 0;
     if ((c1 == '?') && (c2 == '=')) {
        input_mode = ASCII;
        lwsp_count = 0;
-       lwsp_buf = malloc((lwsp_size+5)*sizeof(char));
-       if (lwsp_buf==NULL) {
-           perror("can't malloc");
-           return -1;
-       }
+       lwsp_buf = nkf_xmalloc((lwsp_size+5)*sizeof(char));
        while ((c1=(*i_getc)(f))!=EOF) {
            switch (c1) {
            case LF:
            case CR:
                if (c1==LF) {
        while ((c1=(*i_getc)(f))!=EOF) {
            switch (c1) {
            case LF:
            case CR:
                if (c1==LF) {
-                   if ((c1=(*i_getc)(f))!=EOF && (c1==SP||c1==TAB)) {
+                   if ((c1=(*i_getc)(f))!=EOF && nkf_isblank(c1)) {
                        i_ungetc(SP,f);
                        continue;
                    } else {
                        i_ungetc(SP,f);
                        continue;
                    } else {
@@ -4437,7 +4958,7 @@ mime_getc(FILE *f)
                        if (c1==SP) {
                            i_ungetc(SP,f);
                            continue;
                        if (c1==SP) {
                            i_ungetc(SP,f);
                            continue;
-                       } else if ((c1=(*i_getc)(f))!=EOF && (c1==SP||c1==TAB)) {
+                       } else if ((c1=(*i_getc)(f))!=EOF && nkf_isblank(c1)) {
                            i_ungetc(SP,f);
                            continue;
                        } else {
                            i_ungetc(SP,f);
                            continue;
                        } else {
@@ -4455,12 +4976,7 @@ mime_getc(FILE *f)
                lwsp_buf[lwsp_count] = (unsigned char)c1;
                if (lwsp_count++>lwsp_size){
                    lwsp_size <<= 1;
                lwsp_buf[lwsp_count] = (unsigned char)c1;
                if (lwsp_count++>lwsp_size){
                    lwsp_size <<= 1;
-                   lwsp_buf_new = realloc(lwsp_buf, (lwsp_size+5)*sizeof(char));
-                   if (lwsp_buf_new==NULL) {
-                       free(lwsp_buf);
-                       perror("can't realloc");
-                       return -1;
-                   }
+                   lwsp_buf_new = nkf_xrealloc(lwsp_buf, (lwsp_size+5)*sizeof(char));
                    lwsp_buf = lwsp_buf_new;
                }
                continue;
                    lwsp_buf = lwsp_buf_new;
                }
                continue;
@@ -4473,7 +4989,7 @@ mime_getc(FILE *f)
                i_ungetc(lwsp_buf[lwsp_count],f);
            c1 = lwsp_buf[0];
        }
                i_ungetc(lwsp_buf[lwsp_count],f);
            c1 = lwsp_buf[0];
        }
-       free(lwsp_buf);
+       nkf_xfree(lwsp_buf);
        return c1;
     }
   mime_c3_retry:
        return c1;
     }
   mime_c3_retry:
@@ -4520,11 +5036,10 @@ mime_getc(FILE *f)
 static const char basis_64[] =
     "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
 
 static const char basis_64[] =
     "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
 
-#define MIMEOUT_BUF_LENGTH (60)
+#define MIMEOUT_BUF_LENGTH 74
 static struct {
 static struct {
-    char buf[MIMEOUT_BUF_LENGTH+1];
+    unsigned char buf[MIMEOUT_BUF_LENGTH+1];
     int count;
     int count;
-    nkf_char state;
 } mimeout_state;
 
 /*nkf_char mime_lastchar2, mime_lastchar1;*/
 } mimeout_state;
 
 /*nkf_char mime_lastchar2, mime_lastchar1;*/
@@ -4549,18 +5064,15 @@ open_mime(nkf_char mode)
            (*o_mputc)(mimeout_state.buf[i]);
            i++;
        }
            (*o_mputc)(mimeout_state.buf[i]);
            i++;
        }
-       PUT_NEWLINE((*o_mputc));
+       put_newline(o_mputc);
        (*o_mputc)(SP);
        base64_count = 1;
        (*o_mputc)(SP);
        base64_count = 1;
-       if (mimeout_state.count>0
-           && (mimeout_state.buf[i]==SP || mimeout_state.buf[i]==TAB
-               || mimeout_state.buf[i]==CR || mimeout_state.buf[i]==LF)) {
+       if (mimeout_state.count>0 && nkf_isspace(mimeout_state.buf[i])) {
            i++;
        }
     }
     for (;i<mimeout_state.count;i++) {
            i++;
        }
     }
     for (;i<mimeout_state.count;i++) {
-       if (mimeout_state.buf[i]==SP || mimeout_state.buf[i]==TAB
-           || mimeout_state.buf[i]==CR || mimeout_state.buf[i]==LF) {
+       if (nkf_isspace(mimeout_state.buf[i])) {
            (*o_mputc)(mimeout_state.buf[i]);
            base64_count ++;
        } else {
            (*o_mputc)(mimeout_state.buf[i]);
            base64_count ++;
        } else {
@@ -4585,14 +5097,14 @@ mime_prechar(nkf_char c2, nkf_char c1)
        if (c2 == EOF){
            if (base64_count + mimeout_state.count/3*4> 73){
                (*o_base64conv)(EOF,0);
        if (c2 == EOF){
            if (base64_count + mimeout_state.count/3*4> 73){
                (*o_base64conv)(EOF,0);
-               OCONV_NEWLINE((*o_base64conv));
+               oconv_newline(o_base64conv);
                (*o_base64conv)(0,SP);
                base64_count = 1;
            }
        } else {
                (*o_base64conv)(0,SP);
                base64_count = 1;
            }
        } else {
-           if (base64_count + mimeout_state.count/3*4> 66) {
+           if ((c2 != 0 || c1 > DEL) && base64_count + mimeout_state.count/3*4> 66) {
                (*o_base64conv)(EOF,0);
                (*o_base64conv)(EOF,0);
-               OCONV_NEWLINE((*o_base64conv));
+               oconv_newline(o_base64conv);
                (*o_base64conv)(0,SP);
                base64_count = 1;
                mimeout_mode = -1;
                (*o_base64conv)(0,SP);
                base64_count = 1;
                mimeout_mode = -1;
@@ -4603,7 +5115,7 @@ mime_prechar(nkf_char c2, nkf_char c1)
            mimeout_mode =  (output_mode==ASCII ||output_mode == ISO_8859_1) ? 'Q' : 'B';
            open_mime(output_mode);
            (*o_base64conv)(EOF,0);
            mimeout_mode =  (output_mode==ASCII ||output_mode == ISO_8859_1) ? 'Q' : 'B';
            open_mime(output_mode);
            (*o_base64conv)(EOF,0);
-           OCONV_NEWLINE((*o_base64conv));
+           oconv_newline(o_base64conv);
            (*o_base64conv)(0,SP);
            base64_count = 1;
            mimeout_mode = -1;
            (*o_base64conv)(0,SP);
            base64_count = 1;
            mimeout_mode = -1;
@@ -4628,13 +5140,13 @@ eof_mime(void)
     case 'B':
        break;
     case 2:
     case 'B':
        break;
     case 2:
-       (*o_mputc)(basis_64[((mimeout_state.state & 0x3)<< 4)]);
+       (*o_mputc)(basis_64[((nkf_state->mimeout_state & 0x3)<< 4)]);
        (*o_mputc)('=');
        (*o_mputc)('=');
        base64_count += 3;
        break;
     case 1:
        (*o_mputc)('=');
        (*o_mputc)('=');
        base64_count += 3;
        break;
     case 1:
-       (*o_mputc)(basis_64[((mimeout_state.state & 0xF) << 2)]);
+       (*o_mputc)(basis_64[((nkf_state->mimeout_state & 0xF) << 2)]);
        (*o_mputc)('=');
        base64_count += 2;
        break;
        (*o_mputc)('=');
        base64_count += 2;
        break;
@@ -4666,19 +5178,19 @@ mimeout_addchar(nkf_char c)
        }
        break;
     case 'B':
        }
        break;
     case 'B':
-       mimeout_state.state=c;
+       nkf_state->mimeout_state=c;
        (*o_mputc)(basis_64[c>>2]);
        mimeout_mode=2;
        base64_count ++;
        break;
     case 2:
        (*o_mputc)(basis_64[c>>2]);
        mimeout_mode=2;
        base64_count ++;
        break;
     case 2:
-       (*o_mputc)(basis_64[((mimeout_state.state & 0x3)<< 4) | ((c & 0xF0) >> 4)]);
-       mimeout_state.state=c;
+       (*o_mputc)(basis_64[((nkf_state->mimeout_state & 0x3)<< 4) | ((c & 0xF0) >> 4)]);
+       nkf_state->mimeout_state=c;
        mimeout_mode=1;
        base64_count ++;
        break;
     case 1:
        mimeout_mode=1;
        base64_count ++;
        break;
     case 1:
-       (*o_mputc)(basis_64[((mimeout_state.state & 0xF) << 2) | ((c & 0xC0) >>6)]);
+       (*o_mputc)(basis_64[((nkf_state->mimeout_state & 0xF) << 2) | ((c & 0xC0) >>6)]);
        (*o_mputc)(basis_64[c & 0x3F]);
        mimeout_mode='B';
        base64_count += 2;
        (*o_mputc)(basis_64[c & 0x3F]);
        mimeout_mode='B';
        base64_count += 2;
@@ -4701,14 +5213,14 @@ mime_putc(nkf_char c)
            if (base64_count > 71){
                if (c!=CR && c!=LF) {
                    (*o_mputc)('=');
            if (base64_count > 71){
                if (c!=CR && c!=LF) {
                    (*o_mputc)('=');
-                   PUT_NEWLINE((*o_mputc));
+                   put_newline(o_mputc);
                }
                base64_count = 0;
            }
        }else{
            if (base64_count > 71){
                eof_mime();
                }
                base64_count = 0;
            }
        }else{
            if (base64_count > 71){
                eof_mime();
-               PUT_NEWLINE((*o_mputc));
+               put_newline(o_mputc);
                base64_count = 0;
            }
            if (c == EOF) { /* c==EOF */
                base64_count = 0;
            }
            if (c == EOF) { /* c==EOF */
@@ -4770,7 +5282,7 @@ mime_putc(nkf_char c)
            } else if (c <= SP) {
                close_mime();
                if (base64_count > 70) {
            } else if (c <= SP) {
                close_mime();
                if (base64_count > 70) {
-                   PUT_NEWLINE((*o_mputc));
+                   put_newline(o_mputc);
                    base64_count = 0;
                }
                if (!nkf_isblank(c)) {
                    base64_count = 0;
                }
                if (!nkf_isblank(c)) {
@@ -4780,7 +5292,7 @@ mime_putc(nkf_char c)
            } else {
                if (base64_count > 70) {
                    close_mime();
            } else {
                if (base64_count > 70) {
                    close_mime();
-                   PUT_NEWLINE((*o_mputc));
+                   put_newline(o_mputc);
                    (*o_mputc)(SP);
                    base64_count = 1;
                    open_mime(output_mode);
                    (*o_mputc)(SP);
                    base64_count = 1;
                    open_mime(output_mode);
@@ -4790,14 +5302,17 @@ mime_putc(nkf_char c)
                    return;
                }
            }
                    return;
                }
            }
-           (*o_mputc)(c);
-           base64_count++;
+           if (c != 0x1B) {
+               (*o_mputc)(c);
+               base64_count++;
+               return;
+           }
        }
        }
-       return;
     }
 
     if (mimeout_mode <= 0) {
     }
 
     if (mimeout_mode <= 0) {
-       if (c <= DEL && (output_mode==ASCII ||output_mode == ISO_8859_1)) {
+       if (c <= DEL && (output_mode==ASCII || output_mode == ISO_8859_1 ||
+                   output_mode == UTF_8)) {
            if (nkf_isspace(c)) {
                int flag = 0;
                if (mimeout_mode == -1) {
            if (nkf_isspace(c)) {
                int flag = 0;
                if (mimeout_mode == -1) {
@@ -4830,11 +5345,36 @@ mime_putc(nkf_char c)
                if (base64_count > 1
                    && base64_count + mimeout_state.count > 76
                    && mimeout_state.buf[0] != CR && mimeout_state.buf[0] != LF){
                if (base64_count > 1
                    && base64_count + mimeout_state.count > 76
                    && mimeout_state.buf[0] != CR && mimeout_state.buf[0] != LF){
-                   PUT_NEWLINE((*o_mputc));
-                   base64_count = 0;
-                   if (!nkf_isspace(mimeout_state.buf[0])){
-                       (*o_mputc)(SP);
-                       base64_count++;
+                   static const char *str = "boundary=\"";
+                   static int len = 10;
+                   i = 0;
+
+                   for (; i < mimeout_state.count - len; ++i) {
+                       if (!strncmp((char *)(mimeout_state.buf+i), str, len)) {
+                           i += len - 2;
+                           break;
+                       }
+                   }
+
+                   if (i == 0 || i == mimeout_state.count - len) {
+                       put_newline(o_mputc);
+                       base64_count = 0;
+                       if (!nkf_isspace(mimeout_state.buf[0])){
+                           (*o_mputc)(SP);
+                           base64_count++;
+                       }
+                   }
+                   else {
+                       int j;
+                       for (j = 0; j <= i; ++j) {
+                           (*o_mputc)(mimeout_state.buf[j]);
+                       }
+                       put_newline(o_mputc);
+                       base64_count = 1;
+                       for (; j <= mimeout_state.count; ++j) {
+                           mimeout_state.buf[j - i] = mimeout_state.buf[j];
+                       }
+                       mimeout_state.count -= i;
                    }
                }
                mimeout_state.buf[mimeout_state.count++] = (char)c;
                    }
                }
                mimeout_state.buf[mimeout_state.count++] = (char)c;
@@ -4863,14 +5403,15 @@ mime_putc(nkf_char c)
        }
     }else{
        /* mimeout_mode == 'B', 1, 2 */
        }
     }else{
        /* mimeout_mode == 'B', 1, 2 */
-       if ( c<=DEL && (output_mode==ASCII ||output_mode == ISO_8859_1)) {
+       if (c <= DEL && (output_mode==ASCII || output_mode == ISO_8859_1 ||
+                   output_mode == UTF_8)) {
            if (lastchar == CR || lastchar == LF){
                if (nkf_isblank(c)) {
                    for (i=0;i<mimeout_state.count;i++) {
                        mimeout_addchar(mimeout_state.buf[i]);
                    }
                    mimeout_state.count = 0;
            if (lastchar == CR || lastchar == LF){
                if (nkf_isblank(c)) {
                    for (i=0;i<mimeout_state.count;i++) {
                        mimeout_addchar(mimeout_state.buf[i]);
                    }
                    mimeout_state.count = 0;
-               } else if (SP<c && c<DEL) {
+               } else {
                    eof_mime();
                    for (i=0;i<mimeout_state.count;i++) {
                        (*o_mputc)(mimeout_state.buf[i]);
                    eof_mime();
                    for (i=0;i<mimeout_state.count;i++) {
                        (*o_mputc)(mimeout_state.buf[i]);
@@ -4881,7 +5422,7 @@ mime_putc(nkf_char c)
                mimeout_state.buf[mimeout_state.count++] = (char)c;
                return;
            }
                mimeout_state.buf[mimeout_state.count++] = (char)c;
                return;
            }
-           if (c==SP || c==TAB || c==CR || c==LF) {
+           if (nkf_isspace(c)) {
                for (i=0;i<mimeout_state.count;i++) {
                    if (SP<mimeout_state.buf[i] && mimeout_state.buf[i]<DEL) {
                        eof_mime();
                for (i=0;i<mimeout_state.count;i++) {
                    if (SP<mimeout_state.buf[i] && mimeout_state.buf[i]<DEL) {
                        eof_mime();
@@ -4895,8 +5436,8 @@ mime_putc(nkf_char c)
                mimeout_state.buf[mimeout_state.count++] = (char)c;
                if (mimeout_state.count>MIMEOUT_BUF_LENGTH) {
                    eof_mime();
                mimeout_state.buf[mimeout_state.count++] = (char)c;
                if (mimeout_state.count>MIMEOUT_BUF_LENGTH) {
                    eof_mime();
-                   for (i=0;i<mimeout_state.count;i++) {
-                       (*o_mputc)(mimeout_state.buf[i]);
+                   for (j=0;j<mimeout_state.count;j++) {
+                       (*o_mputc)(mimeout_state.buf[j]);
                        base64_count++;
                    }
                    mimeout_state.count = 0;
                        base64_count++;
                    }
                    mimeout_state.count = 0;
@@ -4950,7 +5491,7 @@ typedef struct nkf_iconv_t {
     size_t input_buffer_size;
     char *output_buffer;
     size_t output_buffer_size;
     size_t input_buffer_size;
     char *output_buffer;
     size_t output_buffer_size;
-}
+};
 
 static nkf_iconv_t
 nkf_iconv_new(char *tocode, char *fromcode)
 
 static nkf_iconv_t
 nkf_iconv_new(char *tocode, char *fromcode)
@@ -4958,15 +5499,9 @@ nkf_iconv_new(char *tocode, char *fromcode)
     nkf_iconv_t converter;
 
     converter->input_buffer_size = IOBUF_SIZE;
     nkf_iconv_t converter;
 
     converter->input_buffer_size = IOBUF_SIZE;
-    converter->input_buffer = malloc(converter->input_buffer_size);
-    if (converter->input_buffer == NULL)
-       perror("can't malloc");
-
+    converter->input_buffer = nkf_xmalloc(converter->input_buffer_size);
     converter->output_buffer_size = IOBUF_SIZE * 2;
     converter->output_buffer_size = IOBUF_SIZE * 2;
-    converter->output_buffer = malloc(converter->output_buffer_size);
-    if (converter->output_buffer == NULL)
-       perror("can't malloc");
-
+    converter->output_buffer = nkf_xmalloc(converter->output_buffer_size);
     converter->cd = iconv_open(tocode, fromcode);
     if (converter->cd == (iconv_t)-1)
     {
     converter->cd = iconv_open(tocode, fromcode);
     if (converter->cd == (iconv_t)-1)
     {
@@ -5033,8 +5568,8 @@ nkf_iconv_convert(nkf_iconv_t *converter, FILE *input)
 static void
 nkf_iconv_close(nkf_iconv_t *convert)
 {
 static void
 nkf_iconv_close(nkf_iconv_t *convert)
 {
-    free(converter->inbuf);
-    free(converter->outbuf);
+    nkf_xfree(converter->inbuf);
+    nkf_xfree(converter->outbuf);
     iconv_close(converter->cd);
 }
 #endif
     iconv_close(converter->cd);
 }
 #endif
@@ -5062,7 +5597,7 @@ reinit(void)
     broken_f = FALSE;
     iso8859_f = FALSE;
     mimeout_f = FALSE;
     broken_f = FALSE;
     iso8859_f = FALSE;
     mimeout_f = FALSE;
-    x0201_f = X0201_DEFAULT;
+    x0201_f = NKF_UNSPECIFIED;
     iso2022jp_f = FALSE;
 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
     ms_ucs_map_f = UCS_MAP_ASCII;
     iso2022jp_f = FALSE;
 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
     ms_ucs_map_f = UCS_MAP_ASCII;
@@ -5145,7 +5680,6 @@ reinit(void)
     input_eol = 0;
     prev_cr = 0;
     option_mode = 0;
     input_eol = 0;
     prev_cr = 0;
     option_mode = 0;
-    init_broken_state();
     z_prev2=0,z_prev1=0;
 #ifdef CHECK_OPTION
     iconv_for_check = 0;
     z_prev2=0,z_prev1=0;
 #ifdef CHECK_OPTION
     iconv_for_check = 0;
@@ -5153,6 +5687,7 @@ reinit(void)
     input_codename = NULL;
     input_encoding = NULL;
     output_encoding = NULL;
     input_codename = NULL;
     input_encoding = NULL;
     output_encoding = NULL;
+    nkf_state_init();
 #ifdef WIN32DLL
     reinitdll();
 #endif /*WIN32DLL*/
 #ifdef WIN32DLL
     reinitdll();
 #endif /*WIN32DLL*/
@@ -5172,10 +5707,16 @@ module_connection(void)
     set_output_encoding(output_encoding);
     oconv = nkf_enc_to_oconv(output_encoding);
     o_putc = std_putc;
     set_output_encoding(output_encoding);
     oconv = nkf_enc_to_oconv(output_encoding);
     o_putc = std_putc;
+    if (nkf_enc_unicode_p(output_encoding))
+       output_mode = UTF_8;
+
+    if (x0201_f == NKF_UNSPECIFIED) {
+       x0201_f = X0201_DEFAULT;
+    }
 
 
-    /* replace continucation module, from output side */
+    /* replace continuation module, from output side */
 
 
-    /* output redicrection */
+    /* output redirection */
 #ifdef CHECK_OPTION
     if (noout_f || guess_f){
        o_putc = no_putc;
 #ifdef CHECK_OPTION
     if (noout_f || guess_f){
        o_putc = no_putc;
@@ -5212,7 +5753,7 @@ module_connection(void)
 
     i_getc = std_getc;
     i_ungetc = std_ungetc;
 
     i_getc = std_getc;
     i_ungetc = std_ungetc;
-    /* input redicrection */
+    /* input redirection */
 #ifdef INPUT_OPTION
     if (cap_f){
        i_cgetc = i_getc; i_getc = cap_getc;
 #ifdef INPUT_OPTION
     if (cap_f){
        i_cgetc = i_getc; i_getc = cap_getc;
@@ -5280,7 +5821,7 @@ noconvert(FILE *f)
 #define NEXT continue        /* no output, get next */
 #define SKIP c2=0;continue        /* no output, get next */
 #define MORE c2=c1;continue  /* need one more byte */
 #define NEXT continue        /* no output, get next */
 #define SKIP c2=0;continue        /* no output, get next */
 #define MORE c2=c1;continue  /* need one more byte */
-#define SEND ;               /* output c1 and c2, get next */
+#define SEND (void)0         /* output c1 and c2, get next */
 #define LAST break           /* end of loop, go closing  */
 #define set_input_mode(mode) do { \
     input_mode = mode; \
 #define LAST break           /* end of loop, go closing  */
 #define set_input_mode(mode) do { \
     input_mode = mode; \
@@ -5318,20 +5859,48 @@ kanji_convert(FILE *f)
               (c2 = (*i_getc)(f)) != EOF &&
               (c3 = (*i_getc)(f)) != EOF &&
               (c4 = (*i_getc)(f)) != EOF) {
               (c2 = (*i_getc)(f)) != EOF &&
               (c3 = (*i_getc)(f)) != EOF &&
               (c4 = (*i_getc)(f)) != EOF) {
-           nkf_iconv_utf_32(c1, c2, c3, c4);
+           nkf_char c5, c6, c7, c8;
+           if (nkf_iconv_utf_32(c1, c2, c3, c4) == (size_t)NKF_ICONV_WAIT_COMBINING_CHAR) {
+               if ((c5 = (*i_getc)(f)) != EOF &&
+                   (c6 = (*i_getc)(f)) != EOF &&
+                   (c7 = (*i_getc)(f)) != EOF &&
+                   (c8 = (*i_getc)(f)) != EOF) {
+                   if (nkf_iconv_utf_32_combine(c1, c2, c3, c4, c5, c6, c7, c8)) {
+                       (*i_ungetc)(c8, f);
+                       (*i_ungetc)(c7, f);
+                       (*i_ungetc)(c6, f);
+                       (*i_ungetc)(c5, f);
+                       nkf_iconv_utf_32_nocombine(c1, c2, c3, c4);
+                   }
+               } else {
+                   nkf_iconv_utf_32_nocombine(c1, c2, c3, c4);
+               }
+           }
        }
        }
-       (*i_ungetc)(EOF, f);
+       goto finished;
     }
     else if (iconv == w_iconv16) {
        while ((c1 = (*i_getc)(f)) != EOF &&
               (c2 = (*i_getc)(f)) != EOF) {
     }
     else if (iconv == w_iconv16) {
        while ((c1 = (*i_getc)(f)) != EOF &&
               (c2 = (*i_getc)(f)) != EOF) {
-           if (nkf_iconv_utf_16(c1, c2, 0, 0) == -2 &&
+           size_t ret = nkf_iconv_utf_16(c1, c2, 0, 0);
+           if (ret == NKF_ICONV_NEED_TWO_MORE_BYTES &&
                (c3 = (*i_getc)(f)) != EOF &&
                (c4 = (*i_getc)(f)) != EOF) {
                nkf_iconv_utf_16(c1, c2, c3, c4);
                (c3 = (*i_getc)(f)) != EOF &&
                (c4 = (*i_getc)(f)) != EOF) {
                nkf_iconv_utf_16(c1, c2, c3, c4);
+           } else if (ret == (size_t)NKF_ICONV_WAIT_COMBINING_CHAR) {
+               if ((c3 = (*i_getc)(f)) != EOF &&
+                   (c4 = (*i_getc)(f)) != EOF) {
+                   if (nkf_iconv_utf_16_combine(c1, c2, c3, c4)) {
+                       (*i_ungetc)(c4, f);
+                       (*i_ungetc)(c3, f);
+                       nkf_iconv_utf_16_nocombine(c1, c2);
+                   }
+               } else {
+                   nkf_iconv_utf_16_nocombine(c1, c2);
+               }
            }
        }
            }
        }
-       (*i_ungetc)(EOF, f);
+       goto finished;
     }
 #endif
 
     }
 #endif
 
@@ -5342,11 +5911,11 @@ kanji_convert(FILE *f)
            code_status(c1);
        if (c2) {
            /* second byte */
            code_status(c1);
        if (c2) {
            /* second byte */
-           if (c2 > DEL) {
+           if (c2 > ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) {
                /* in case of 8th bit is on */
                if (!estab_f&&!mime_decode_mode) {
                    /* in case of not established yet */
                /* in case of 8th bit is on */
                if (!estab_f&&!mime_decode_mode) {
                    /* in case of not established yet */
-                   /* It is still ambiguious */
+                   /* It is still ambiguous */
                    if (h_conv(f, c2, c1)==EOF) {
                        LAST;
                    }
                    if (h_conv(f, c2, c1)==EOF) {
                        LAST;
                    }
@@ -5378,6 +5947,12 @@ kanji_convert(FILE *f)
            if (input_mode == JIS_X_0208 && DEL <= c1 && c1 < 0x92) {
                /* CP5022x */
                MORE;
            if (input_mode == JIS_X_0208 && DEL <= c1 && c1 < 0x92) {
                /* CP5022x */
                MORE;
+           }else if (input_codename && input_codename[0] == 'I' &&
+                   0xA1 <= c1 && c1 <= 0xDF) {
+               /* JIS X 0201 Katakana in 8bit JIS */
+               c2 = JIS_X_0201_1976_K;
+               c1 &= 0x7f;
+               SEND;
            } else if (c1 > DEL) {
                /* 8 bit code */
                if (!estab_f && !iso8859_f) {
            } else if (c1 > DEL) {
                /* 8 bit code */
                if (!estab_f && !iso8859_f) {
@@ -5452,7 +6027,7 @@ kanji_convert(FILE *f)
                SKIP;
            } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) {
                if ((c1 = (*i_getc)(f)) == EOF) {
                SKIP;
            } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) {
                if ((c1 = (*i_getc)(f)) == EOF) {
-                   /*  (*oconv)(0, ESC); don't send bogus code */
+                   (*oconv)(0, ESC);
                    LAST;
                }
                else if (c1 == '&') {
                    LAST;
                }
                else if (c1 == '&') {
@@ -5528,6 +6103,7 @@ kanji_convert(FILE *f)
                    else if (c1 == 'I') {
                        /* JIS X 0201 Katakana */
                        set_input_mode(JIS_X_0201_1976_K);
                    else if (c1 == 'I') {
                        /* JIS X 0201 Katakana */
                        set_input_mode(JIS_X_0201_1976_K);
+                       shift_mode = 1;
                        SKIP;
                    }
                    else if (c1 == 'B' || c1 == 'J' || c1 == 'H') {
                        SKIP;
                    }
                    else if (c1 == 'B' || c1 == 'J' || c1 == 'H') {
@@ -5575,14 +6151,15 @@ kanji_convert(FILE *f)
                    }
                }
                else {
                    }
                }
                else {
+                   i_ungetc(c1,f);
                    /* lonely ESC  */
                    (*oconv)(0, ESC);
                    /* lonely ESC  */
                    (*oconv)(0, ESC);
-                   SEND;
+                   SKIP;
                }
            } else if (c1 == ESC && iconv == s_iconv) {
                /* ESC in Shift_JIS */
                if ((c1 = (*i_getc)(f)) == EOF) {
                }
            } else if (c1 == ESC && iconv == s_iconv) {
                /* ESC in Shift_JIS */
                if ((c1 = (*i_getc)(f)) == EOF) {
-                   /*  (*oconv)(0, ESC); don't send bogus code */
+                   (*oconv)(0, ESC);
                    LAST;
                } else if (c1 == '$') {
                    /* J-PHONE emoji */
                    LAST;
                } else if (c1 == '$') {
                    /* J-PHONE emoji */
@@ -5614,9 +6191,10 @@ kanji_convert(FILE *f)
                    }
                }
                else {
                    }
                }
                else {
+                   i_ungetc(c1,f);
                    /* lonely ESC  */
                    (*oconv)(0, ESC);
                    /* lonely ESC  */
                    (*oconv)(0, ESC);
-                   SEND;
+                   SKIP;
                }
            } else if (c1 == LF || c1 == CR) {
                if (broken_f&4) {
                }
            } else if (c1 == LF || c1 == CR) {
                if (broken_f&4) {
@@ -5669,11 +6247,52 @@ kanji_convert(FILE *f)
                    }
                }
                break;
                    }
                }
                break;
+           case -3:
+               /* 4 bytes UTF-8 (check combining character) */
+               if ((c3 = (*i_getc)(f)) != EOF) {
+                   if ((c4 = (*i_getc)(f)) != EOF) {
+                       if (w_iconv_combine(c2, c1, 0, c3, c4, 0)) {
+                           (*i_ungetc)(c4, f);
+                           (*i_ungetc)(c3, f);
+                           w_iconv_nocombine(c2, c1, 0);
+                       }
+                   } else {
+                       (*i_ungetc)(c3, f);
+                       w_iconv_nocombine(c2, c1, 0);
+                   }
+               } else {
+                   w_iconv_nocombine(c2, c1, 0);
+               }
+               break;
            case -1:
                /* 3 bytes EUC or UTF-8 */
                if ((c3 = (*i_getc)(f)) != EOF) {
                    code_status(c3);
            case -1:
                /* 3 bytes EUC or UTF-8 */
                if ((c3 = (*i_getc)(f)) != EOF) {
                    code_status(c3);
-                   (*iconv)(c2, c1, c3);
+                   if ((*iconv)(c2, c1, c3) == -3) {
+                       /* 6 bytes UTF-8 (check combining character) */
+                       nkf_char c5, c6;
+                       if ((c4 = (*i_getc)(f)) != EOF) {
+                           if ((c5 = (*i_getc)(f)) != EOF) {
+                               if ((c6 = (*i_getc)(f)) != EOF) {
+                                   if (w_iconv_combine(c2, c1, c3, c4, c5, c6)) {
+                                       (*i_ungetc)(c6, f);
+                                       (*i_ungetc)(c5, f);
+                                       (*i_ungetc)(c4, f);
+                                       w_iconv_nocombine(c2, c1, c3);
+                                   }
+                               } else {
+                                   (*i_ungetc)(c5, f);
+                                   (*i_ungetc)(c4, f);
+                                   w_iconv_nocombine(c2, c1, c3);
+                               }
+                           } else {
+                               (*i_ungetc)(c4, f);
+                               w_iconv_nocombine(c2, c1, c3);
+                           }
+                       } else {
+                           w_iconv_nocombine(c2, c1, c3);
+                       }
+                   }
                }
                break;
            }
                }
                break;
            }
@@ -5707,6 +6326,7 @@ kanji_convert(FILE *f)
        /* goto next_word */
     }
 
        /* goto next_word */
     }
 
+finished:
     /* epilogue */
     (*iconv)(EOF, 0, 0);
     if (!input_codename)
     /* epilogue */
     (*iconv)(EOF, 0, 0);
     if (!input_codename)
@@ -5729,7 +6349,7 @@ kanji_convert(FILE *f)
 
 /*
  * int options(unsigned char *cp)
 
 /*
  * int options(unsigned char *cp)
- * 
+ *
  * return values:
  *    0: success
  *   -1: ArgumentError
  * return values:
  *    0: success
  *   -1: ArgumentError
@@ -5758,7 +6378,7 @@ options(unsigned char *cp)
                option_mode = 1;
                return 0;
            }
                option_mode = 1;
                return 0;
            }
-           for (i=0;i<sizeof(long_option)/sizeof(long_option[0]);i++) {
+           for (i=0;i<(int)(sizeof(long_option)/sizeof(long_option[0]));i++) {
                p = (unsigned char *)long_option[i].name;
                for (j=0;*p && *p != '=' && *p == cp[j];p++, j++);
                if (*p == cp[j] || cp[j] == SP){
                p = (unsigned char *)long_option[i].name;
                for (j=0;*p && *p != '=' && *p == cp[j];p++, j++);
                if (*p == cp[j] || cp[j] == SP){
@@ -5778,6 +6398,12 @@ options(unsigned char *cp)
                cp_back = cp;
                cp = (unsigned char *)long_option[i].alias;
            }else{
                cp_back = cp;
                cp = (unsigned char *)long_option[i].alias;
            }else{
+#ifndef PERL_XS
+               if (strcmp(long_option[i].name, "help") == 0){
+                   usage();
+                   exit(EXIT_SUCCESS);
+               }
+#endif
                if (strcmp(long_option[i].name, "ic=") == 0){
                    enc = nkf_enc_find((char *)p);
                    if (!enc) continue;
                if (strcmp(long_option[i].name, "ic=") == 0){
                    enc = nkf_enc_find((char *)p);
                    if (!enc) continue;
@@ -5811,8 +6437,7 @@ options(unsigned char *cp)
                    overwrite_f = TRUE;
                    preserve_time_f = TRUE;
                    backup_f = TRUE;
                    overwrite_f = TRUE;
                    preserve_time_f = TRUE;
                    backup_f = TRUE;
-                   backup_suffix = malloc(strlen((char *) p) + 1);
-                   strcpy(backup_suffix, (char *) p);
+                   backup_suffix = (char *)p;
                    continue;
                }
                if (strcmp(long_option[i].name, "in-place") == 0){
                    continue;
                }
                if (strcmp(long_option[i].name, "in-place") == 0){
@@ -5826,8 +6451,7 @@ options(unsigned char *cp)
                    overwrite_f = TRUE;
                    preserve_time_f = FALSE;
                    backup_f = TRUE;
                    overwrite_f = TRUE;
                    preserve_time_f = FALSE;
                    backup_f = TRUE;
-                   backup_suffix = malloc(strlen((char *) p) + 1);
-                   strcpy(backup_suffix, (char *) p);
+                   backup_suffix = (char *)p;
                    continue;
                }
 #endif
                    continue;
                }
 #endif
@@ -6018,7 +6642,7 @@ options(unsigned char *cp)
            output_encoding = nkf_enc_from_index(EUCJP_NKF);
            continue;
        case 's':           /* SJIS output */
            output_encoding = nkf_enc_from_index(EUCJP_NKF);
            continue;
        case 's':           /* SJIS output */
-           output_encoding = nkf_enc_from_index(WINDOWS_31J);
+           output_encoding = nkf_enc_from_index(SHIFT_JIS);
            continue;
        case 'l':           /* ISO8859 Latin-1 support, no conversion */
            iso8859_f = TRUE;  /* Only compatible with ISO-2022-JP */
            continue;
        case 'l':           /* ISO8859 Latin-1 support, no conversion */
            iso8859_f = TRUE;  /* Only compatible with ISO-2022-JP */
@@ -6028,7 +6652,8 @@ options(unsigned char *cp)
            if (*cp=='@'||*cp=='B')
                kanji_intro = *cp++;
            continue;
            if (*cp=='@'||*cp=='B')
                kanji_intro = *cp++;
            continue;
-       case 'o':           /* ASCII IN ESC-(-J/B */
+       case 'o':           /* ASCII IN ESC-(-J/B/H */
+           /* ESC ( H was used in initial JUNET messages */
            if (*cp=='J'||*cp=='B'||*cp=='H')
                ascii_intro = *cp++;
            continue;
            if (*cp=='J'||*cp=='B'||*cp=='H')
                ascii_intro = *cp++;
            continue;
@@ -6053,15 +6678,15 @@ options(unsigned char *cp)
 #ifndef PERL_XS
        case 'V':
            show_configuration();
 #ifndef PERL_XS
        case 'V':
            show_configuration();
-           exit(1);
+           exit(EXIT_SUCCESS);
            break;
        case 'v':
            break;
        case 'v':
-           usage();
-           exit(1);
+           version();
+           exit(EXIT_SUCCESS);
            break;
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
            break;
 #endif
 #ifdef UTF8_OUTPUT_ENABLE
-       case 'w':           /* UTF-8 output */
+       case 'w':           /* UTF-{8,16,32} output */
            if (cp[0] == '8') {
                cp++;
                if (cp[0] == '0'){
            if (cp[0] == '8') {
                cp++;
                if (cp[0] == '0'){
@@ -6086,19 +6711,18 @@ options(unsigned char *cp)
                if (cp[0]=='L') {
                    cp++;
                    output_endian = ENDIAN_LITTLE;
                if (cp[0]=='L') {
                    cp++;
                    output_endian = ENDIAN_LITTLE;
+                   output_bom_f = TRUE;
                } else if (cp[0] == 'B') {
                    cp++;
                } else if (cp[0] == 'B') {
                    cp++;
-               } else {
-                   output_encoding = nkf_enc_from_index(enc_idx);
-                   continue;
+                   output_bom_f = TRUE;
                }
                if (cp[0] == '0'){
                }
                if (cp[0] == '0'){
+                   output_bom_f = FALSE;
                    cp++;
                    enc_idx = enc_idx == UTF_16
                        ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
                        : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
                } else {
                    cp++;
                    enc_idx = enc_idx == UTF_16
                        ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE)
                        : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE);
                } else {
-                   output_bom_f = TRUE;
                    enc_idx = enc_idx == UTF_16
                        ? (output_endian == ENDIAN_LITTLE ? UTF_16LE_BOM : UTF_16BE_BOM)
                        : (output_endian == ENDIAN_LITTLE ? UTF_32LE_BOM : UTF_32BE_BOM);
                    enc_idx = enc_idx == UTF_16
                        ? (output_endian == ENDIAN_LITTLE ? UTF_16LE_BOM : UTF_16BE_BOM)
                        : (output_endian == ENDIAN_LITTLE ? UTF_32LE_BOM : UTF_32BE_BOM);
@@ -6147,8 +6771,8 @@ options(unsigned char *cp)
        case 'E':   /* EUC-JP input */
            input_encoding = nkf_enc_from_index(EUCJP_NKF);
            continue;
        case 'E':   /* EUC-JP input */
            input_encoding = nkf_enc_from_index(EUCJP_NKF);
            continue;
-       case 'S':   /* Windows-31J input */
-           input_encoding = nkf_enc_from_index(WINDOWS_31J);
+       case 'S':   /* Shift_JIS input */
+           input_encoding = nkf_enc_from_index(SHIFT_JIS);
            continue;
        case 'Z':   /* Convert X0208 alphabet to asii */
            /* alpha_f
            continue;
        case 'Z':   /* Convert X0208 alphabet to asii */
            /* alpha_f
@@ -6158,10 +6782,10 @@ options(unsigned char *cp)
               bit:3   Convert HTML Entity
               bit:4   Convert JIS X 0208 Katakana to JIS X 0201 Katakana
             */
               bit:3   Convert HTML Entity
               bit:4   Convert JIS X 0208 Katakana to JIS X 0201 Katakana
             */
-           while ('0'<= *cp && *cp <='9') {
+           while ('0'<= *cp && *cp <='4') {
                alpha_f |= 1 << (*cp++ - '0');
            }
                alpha_f |= 1 << (*cp++ - '0');
            }
-           if (!alpha_f) alpha_f = 1;
+           alpha_f |= 1;
            continue;
        case 'x':   /* Convert X0201 kana to X0208 or X0201 Conversion */
            x0201_f = FALSE;    /* No X0201->X0208 conversion */
            continue;
        case 'x':   /* Convert X0201 kana to X0208 or X0201 Conversion */
            x0201_f = FALSE;    /* No X0201->X0208 conversion */
@@ -6275,7 +6899,7 @@ options(unsigned char *cp)
            continue;
 #endif
        case SP:
            continue;
 #endif
        case SP:
-           /* module muliple options in a string are allowed for Perl moudle  */
+           /* module multiple options in a string are allowed for Perl module  */
            while(*cp && *cp++!='-');
            continue;
        default:
            while(*cp && *cp++!='-');
            continue;
        default:
@@ -6308,6 +6932,8 @@ main(int argc, char **argv)
 #ifdef DEFAULT_CODE_LOCALE
     setlocale(LC_CTYPE, "");
 #endif
 #ifdef DEFAULT_CODE_LOCALE
     setlocale(LC_CTYPE, "");
 #endif
+    nkf_state_init();
+
     for (argc--,argv++; (argc > 0) && **argv == '-'; argc--, argv++) {
        cp = (unsigned char *)*argv;
        options(cp);
     for (argc--,argv++; (argc > 0) && **argv == '-'; argc--, argv++) {
        cp = (unsigned char *)*argv;
        options(cp);
@@ -6415,13 +7041,9 @@ main(int argc, char **argv)
                if (file_out_f == TRUE) {
 #ifdef OVERWRITE
                    if (overwrite_f){
                if (file_out_f == TRUE) {
 #ifdef OVERWRITE
                    if (overwrite_f){
-                       outfname = malloc(strlen(origfname)
+                       outfname = nkf_xmalloc(strlen(origfname)
                                          + strlen(".nkftmpXXXXXX")
                                          + 1);
                                          + strlen(".nkftmpXXXXXX")
                                          + 1);
-                       if (!outfname){
-                           perror(origfname);
-                           return -1;
-                       }
                        strcpy(outfname, origfname);
 #ifdef MSDOS
                        {
                        strcpy(outfname, origfname);
 #ifdef MSDOS
                        {
@@ -6535,7 +7157,7 @@ main(int argc, char **argv)
                            fprintf(stderr, "Can't rename %s to %s\n",
                                    origfname, backup_filename);
                        }
                            fprintf(stderr, "Can't rename %s to %s\n",
                                    origfname, backup_filename);
                        }
-                       free(backup_filename);
+                       nkf_xfree(backup_filename);
                    }else{
 #ifdef MSDOS
                        if (unlink(origfname)){
                    }else{
 #ifdef MSDOS
                        if (unlink(origfname)){
@@ -6548,7 +7170,7 @@ main(int argc, char **argv)
                        fprintf(stderr, "Can't rename %s to %s\n",
                                outfname, origfname);
                    }
                        fprintf(stderr, "Can't rename %s to %s\n",
                                outfname, origfname);
                    }
-                   free(outfname);
+                   nkf_xfree(outfname);
                }
 #endif
            }
                }
 #endif
            }
nkfのGitリポジトリです。
About OSDN
Find Software
Develop Software
Help