X-Git-Url: http://git.sourceforge.jp/view?p=nkf%2Fnkf.git;a=blobdiff_plain;f=nkf.c;h=b58c437d3cc0da1fd87a69c19feec8d0e6fe595f;hp=6f7c7ca127a911e1c03695c011f8f0156c9cf76e;hb=9fd04d0dfbeff6ac8cb877b9853fb09af6811f33;hpb=4f2ea7528baf174776adf1363b03bdf7cfa5b5ba diff --git a/nkf.c b/nkf.c index 6f7c7ca..b58c437 100644 --- a/nkf.c +++ b/nkf.c @@ -1,54 +1,53 @@ -/** Network Kanji Filter. (PDS Version) -** -*- coding: ISO-2022-JP -*- -************************************************************************ -** Copyright (C) 1987, Fujitsu LTD. (Itaru ICHIKAWA) -** 連絡先： （株）富士通研究所　ソフト３研　市川　至 -** （E-Mail Address: ichikawa@flab.fujitsu.co.jp） -** Copyright (C) 1996,1998 -** Copyright (C) 2002 -** 連絡先： 琉球大学情報工学科 河野 真治 mime/X0208 support -** （E-Mail Address: kono@ie.u-ryukyu.ac.jp） -** 連絡先： COW for DOS & Win16 & Win32 & OS/2 -** （E-Mail Address: GHG00637@niftyserve.or.p） -** -** このソースのいかなる複写，改変，修正も許諾します。ただし、 -** その際には、誰が貢献したを示すこの部分を残すこと。 -** 再配布や雑誌の付録などの問い合わせも必要ありません。 -** 営利利用も上記に反しない範囲で許可します。 -** バイナリの配布の際にはversion messageを保存することを条件とします。 -** このプログラムについては特に何の保証もしない、悪しからず。 -** -** 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. -***********************************************************************/ - -/*********************************************************************** - * 現在、nkf は SorceForge にてメンテナンスが続けられています。 - * http://sourceforge.jp/projects/nkf/ -***********************************************************************/ -#define NKF_IDENT "$Id: nkf.c,v 1.174 2008/02/07 19:25:29 naruse Exp $" -#define NKF_VERSION "2.0.8" -#define NKF_RELEASE_DATE "2008-02-07" +/* + * 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 \ - "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 "utf8tbl.h" +#ifdef __WIN32__ +#include +#include +#endif +#if defined(__OS2__) +# define INCL_DOS +# define INCL_DOSERRORS +# include +#endif +#include + /* state of output_mode and input_mode c2 0 means ASCII - JIS_X_0201_1976_K - ISO_8859_1 - JIS_X_0208 - EOF all termination + JIS_X_0201_1976_K + ISO_8859_1 + JIS_X_0208 + EOF all termination c1 32bit data */ @@ -130,20 +129,20 @@ enum nkf_encodings { 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 */ }; -nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0); -nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0); -nkf_char w_iconv(nkf_char c2, nkf_char c1, nkf_char c0); -nkf_char w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0); -nkf_char w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0); -void j_oconv(nkf_char c2, nkf_char c1); -void s_oconv(nkf_char c2, nkf_char c1); -void e_oconv(nkf_char c2, nkf_char c1); -void w_oconv(nkf_char c2, nkf_char c1); -void w_oconv16(nkf_char c2, nkf_char c1); -void w_oconv32(nkf_char c2, nkf_char c1); +static nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0); +static nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0); +static nkf_char w_iconv(nkf_char c2, nkf_char c1, nkf_char c0); +static nkf_char w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0); +static nkf_char w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0); +static void j_oconv(nkf_char c2, nkf_char c1); +static void s_oconv(nkf_char c2, nkf_char c1); +static void e_oconv(nkf_char c2, nkf_char c1); +static void w_oconv(nkf_char c2, nkf_char c1); +static void w_oconv16(nkf_char c2, nkf_char c1); +static void w_oconv32(nkf_char c2, nkf_char c1); typedef struct { const char *name; @@ -211,16 +210,21 @@ struct { } 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}, {"CP50221", CP50221}, + {"CSISO2022JP", CP50221}, {"CP50222", CP50222}, {"ISO-2022-JP-1", ISO_2022_JP_1}, {"ISO-2022-JP-3", ISO_2022_JP_3}, {"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}, @@ -262,6 +266,8 @@ struct { #define DEFAULT_ENCIDX ISO_2022_JP #elif defined(DEFAULT_CODE_SJIS) #define DEFAULT_ENCIDX SHIFT_JIS +#elif defined(DEFAULT_CODE_WINDOWS_31J) +#define DEFAULT_ENCIDX WINDOWS_31J #elif defined(DEFAULT_CODE_EUC) #define DEFAULT_ENCIDX EUC_JP #elif defined(DEFAULT_CODE_UTF8) @@ -270,7 +276,7 @@ struct { #define is_alnum(c) \ - (('a'<=c && c<='z')||('A'<= c && c<='Z')||('0'<=c && c<='9')) + (('a'<=c && c<='z')||('A'<= c && c<='Z')||('0'<=c && c<='9')) /* I don't trust portablity of toupper */ #define nkf_toupper(c) (('a'<=c && c<='z')?(c-('a'-'A')):c) @@ -284,15 +290,16 @@ struct { #define nkf_isprint(c) (SP<=c && c<='~') #define nkf_isgraph(c) ('!'<=c && c<='~') #define hex2bin(c) (('0'<=c&&c<='9') ? (c-'0') : \ - ('A'<=c&&c<='F') ? (c-'A'+10) : \ - ('a'<=c&&c<='f') ? (c-'a'+10) : 0) + ('A'<=c&&c<='F') ? (c-'A'+10) : \ + ('a'<=c&&c<='f') ? (c-'a'+10) : 0) #define bin2hex(c) ("0123456789ABCDEF"[c&15]) #define is_eucg3(c2) (((unsigned short)c2 >> 8) == SS3) #define nkf_noescape_mime(c) ((c == CR) || (c == LF) || \ - ((c > SP) && (c < DEL) && (c != '?') && (c != '=') && (c != '_') \ - && (c != '(') && (c != ')') && (c != '.') && (c != 0x22))) + ((c > SP) && (c < DEL) && (c != '?') && (c != '=') && (c != '_') \ + && (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 <= 0x5F) #define HOLD_SIZE 1024 #if defined(INT_IS_SHORT) @@ -316,7 +323,7 @@ extern POINT _BufferSize; #endif struct input_code{ - char *name; + const char *name; nkf_char stat; nkf_char score; nkf_char index; @@ -326,11 +333,10 @@ struct input_code{ int _file_stat; }; -static char *input_codename = NULL; /* NULL: unestablished, "": BINARY */ +static const char *input_codename = NULL; /* NULL: unestablished, "": BINARY */ static nkf_encoding *input_encoding = NULL; static nkf_encoding *output_encoding = NULL; -static int kanji_convert(FILE *f); #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) /* UCS Mapping * 0: Shift_JIS, eucJP-ascii @@ -350,6 +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; +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); @@ -377,6 +384,8 @@ static unsigned char stdibuf[IOBUF_SIZE]; static unsigned char stdobuf[IOBUF_SIZE]; #endif +#define NKF_UNSPECIFIED (-TRUE) + /* flags */ static int unbuf_f = FALSE; static int estab_f = FALSE; @@ -391,7 +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 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 @@ -419,8 +428,10 @@ static nkf_char (*i_uungetc)(nkf_char c ,FILE *f) = std_ungetc; #define nkf_char_euc3_new(c) ((c) | PREFIX_EUCG3) #define nkf_char_unicode_new(c) ((c) | CLASS_UNICODE) #define nkf_char_unicode_p(c) ((c & CLASS_MASK) == CLASS_UNICODE) -#define nkf_char_unicode_bmp_p(c) ((c & VALUE_MASK) <= NKF_INT32_C(UNICODE_BMP_MAX)) -#define nkf_char_unicode_value_p(c) ((c & VALUE_MASK) <= NKF_INT32_C(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; @@ -437,7 +448,7 @@ static nkf_char (*iconv_for_check)(nkf_char c2,nkf_char c1,nkf_char c0) = 0; #endif static int guess_f = 0; /* 0: OFF, 1: ON, 2: VERBOSE */ -static void set_input_codename(char *codename); +static void set_input_codename(const char *codename); #ifdef EXEC_IO static int exec_f = 0; @@ -466,8 +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}, + {"UTF-16", 0, 0, 0, {0, 0, 0}, NULL, w_iconv16, 0}, + {"UTF-32", 0, 0, 0, {0, 0, 0}, NULL, w_iconv32, 0}, #endif - {0} + {NULL, 0, 0, 0, {0, 0, 0}, NULL, NULL, 0} }; static int mimeout_mode = 0; /* 0, -1, 'Q', 'B', 1, 2 */ @@ -495,14 +508,16 @@ static int fold_margin = FOLD_MARGIN; /* process default */ -nkf_char no_connection2(nkf_char c2, nkf_char c1, nkf_char c0) +static nkf_char +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"); - exit(1); + exit(EXIT_FAILURE); return 0; /* LINT */ } -void no_connection(nkf_char c2, nkf_char c1) +static void +no_connection(nkf_char c2, nkf_char c1) { no_connection2(c2,c1,0); } @@ -608,6 +623,27 @@ static const unsigned char ev[]= { 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 */ @@ -645,31 +681,51 @@ static nkf_char prev_cr = 0; /* CR or 0 */ 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); + } -char* nkf_strcpy(const char *str) + return ptr; +} + +static void * +nkf_xrealloc(void *ptr, size_t size) { - char* result = malloc(strlen(str) + 1); - if (!result){ - perror(str); - return ""; + if (size == 0) size = 1; + + ptr = realloc(ptr, size); + if (ptr == NULL) { + perror("can't realloc"); + exit(EXIT_FAILURE); } - strcpy(result, str); - return result; + + return ptr; } -static void nkf_str_upcase(const char *src, char *dest, size_t length) +#define nkf_xfree(ptr) free(ptr) + +static int +nkf_str_caseeql(const char *src, const char *target) { - int i = 0; - for (; i < length && src[i]; i++) { - dest[i] = nkf_toupper(src[i]); + int i; + for (i = 0; src[i] && target[i]; i++) { + if (nkf_toupper(src[i]) != nkf_toupper(target[i])) return FALSE; } - dest[i] = 0; + if (src[i] || target[i]) return FALSE; + else return TRUE; } -static nkf_encoding *nkf_enc_from_index(int idx) +static nkf_encoding* +nkf_enc_from_index(int idx) { if (idx < 0 || NKF_ENCODING_TABLE_SIZE <= idx) { return 0; @@ -677,19 +733,21 @@ static nkf_encoding *nkf_enc_from_index(int idx) return &nkf_encoding_table[idx]; } -static int nkf_enc_find_index(const char *name) +static int +nkf_enc_find_index(const char *name) { - int i, index = -1; - if (*name == 'X' && *(name+1) == '-') name += 2; + int i; + if (name[0] == 'X' && *(name+1) == '-') name += 2; for (i = 0; encoding_name_to_id_table[i].id >= 0; i++) { - if (strcmp(name, encoding_name_to_id_table[i].name) == 0) { + if (nkf_str_caseeql(encoding_name_to_id_table[i].name, name)) { return encoding_name_to_id_table[i].id; } } - return index; + return -1; } -static nkf_encoding *nkf_enc_find(const char *name) +static nkf_encoding* +nkf_enc_find(const char *name) { int idx = -1; idx = nkf_enc_find_index(name); @@ -703,135 +761,212 @@ static nkf_encoding *nkf_enc_find(const char *name) #define nkf_enc_to_iconv(enc) nkf_enc_to_base_encoding(enc)->iconv #define nkf_enc_to_oconv(enc) nkf_enc_to_base_encoding(enc)->oconv #define nkf_enc_asciicompat(enc) (\ - nkf_enc_to_base_encoding(enc) == &NkfEncodingASCII ||\ - nkf_enc_to_base_encoding(enc) == &NkfEncodingISO_2022_JP) + nkf_enc_to_base_encoding(enc) == &NkfEncodingASCII ||\ + nkf_enc_to_base_encoding(enc) == &NkfEncodingISO_2022_JP) #define nkf_enc_unicode_p(enc) (\ - nkf_enc_to_base_encoding(enc) == &NkfEncodingUTF_8 ||\ - nkf_enc_to_base_encoding(enc) == &NkfEncodingUTF_16 ||\ - nkf_enc_to_base_encoding(enc) == &NkfEncodingUTF_32) + nkf_enc_to_base_encoding(enc) == &NkfEncodingUTF_8 ||\ + nkf_enc_to_base_encoding(enc) == &NkfEncodingUTF_16 ||\ + nkf_enc_to_base_encoding(enc) == &NkfEncodingUTF_32) #define nkf_enc_cp5022x_p(enc) (\ - nkf_enc_to_index(enc) == CP50220 ||\ - nkf_enc_to_index(enc) == CP50221 ||\ - nkf_enc_to_index(enc) == CP50222) + nkf_enc_to_index(enc) == CP50220 ||\ + nkf_enc_to_index(enc) == CP50221 ||\ + 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__) - return sprintf("CP%d", GetACP()); -#else + static char buf[16]; + sprintf(buf, "CP%d", GetACP()); + return buf; +#elif defined(__OS2__) +# if defined(INT_IS_SHORT) + /* OS/2 1.x */ return NULL; +# else + /* OS/2 32bit */ + static char buf[16]; + ULONG ulCP[1], ulncp; + DosQueryCp(sizeof(ulCP), ulCP, &ulncp); + if (ulCP[0] == 932 || ulCP[0] == 943) + strcpy(buf, "Shift_JIS"); + else + sprintf(buf, "CP%lu", ulCP[0]); + return buf; +# endif #endif + return NULL; } -static nkf_encoding* nkf_locale_encoding() +static nkf_encoding* +nkf_locale_encoding(void) { nkf_encoding *enc = 0; - char *encname = nkf_locale_charmap(); + const char *encname = nkf_locale_charmap(); if (encname) enc = nkf_enc_find(encname); - if (enc < 0) enc = 0; return enc; } #endif /* DEFAULT_CODE_LOCALE */ -static nkf_encoding* nkf_default_encoding() +static nkf_encoding* +nkf_utf8_encoding(void) +{ + return &nkf_encoding_table[UTF_8]; +} + +static nkf_encoding* +nkf_default_encoding(void) { nkf_encoding *enc = 0; #ifdef DEFAULT_CODE_LOCALE enc = nkf_locale_encoding(); -#elif DEFAULT_ENCIDX +#elif defined(DEFAULT_ENCIDX) enc = nkf_enc_from_index(DEFAULT_ENCIDX); #endif + if (!enc) enc = nkf_utf8_encoding(); 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 #endif -void version(void) +static void +version(void) { fprintf(HELP_OUTPUT,"Network Kanji Filter Version " NKF_VERSION " (" NKF_RELEASE_DATE ") \n" COPY_RIGHT "\n"); } -void usage(void) +static void +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 - " 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 - "J,S,E,W Input assumption is JIS 7 bit , Shift JIS, EUC-JP, UTF-8\n" #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 - "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= --oc=\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= Specify the input encoding\n" + " --oc= Specify the output encoding\n" + " --hiragana --katakana Hiragana/Katakana Conversion\n" + " --katakana-hiragana Converts each other\n" + ); + fprintf(HELP_OUTPUT, #ifdef INPUT_OPTION - " --cap-input, --url-input Convert hex after ':' or '%%'\n" + " --{cap, url}-input Convert hex after ':' or '%%'\n" #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" - " Specify how nkf handles unassigned characters\n" + " Specify unassigned character's replacement\n" #endif + ); + fprintf(HELP_OUTPUT, #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 - " -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(); } -void show_configuration(void) +static void +show_configuration(void) { 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" ); @@ -839,10 +974,10 @@ void show_configuration(void) " Default output encoding: " #ifdef DEFAULT_CODE_LOCALE "LOCALE (%s)\n", nkf_enc_name(nkf_default_encoding()) -#elif DEFAULT_ENCIDX +#elif defined(DEFAULT_ENCIDX) "CONFIG (%s)\n", nkf_enc_name(nkf_default_encoding()) #else - "NONE\n" + "NONE\n" #endif ); fprintf(HELP_OUTPUT, @@ -880,7 +1015,8 @@ void show_configuration(void) #endif /*PERL_XS*/ #ifdef OVERWRITE -char *get_backup_filename(const char *suffix, const char *filename) +static char* +get_backup_filename(const char *suffix, const char *filename) { char *backup_filename; int asterisk_count = 0; @@ -892,12 +1028,7 @@ char *get_backup_filename(const char *suffix, const char *filename) } 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'; @@ -910,8 +1041,8 @@ char *get_backup_filename(const char *suffix, const char *filename) } backup_filename[j] = '\0'; }else{ - j = strlen(suffix) + filename_length; - backup_filename = malloc( + 1); + j = filename_length + strlen(suffix); + backup_filename = nkf_xmalloc(j + 1); strcpy(backup_filename, filename); strcat(backup_filename, suffix); backup_filename[j] = '\0'; @@ -921,12 +1052,13 @@ char *get_backup_filename(const char *suffix, const char *filename) #endif #ifdef UTF8_INPUT_ENABLE -void nkf_each_char_to_hex(void (*f)(nkf_char c2,nkf_char c1), nkf_char c) +static void +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){ - if(c >= 1<= NKF_INT32_C(1)<= 0){ (*f)(0, bin2hex(c>>shift)); shift -= 4; @@ -938,7 +1070,8 @@ void nkf_each_char_to_hex(void (*f)(nkf_char c2,nkf_char c1), nkf_char c) return; } -void encode_fallback_html(nkf_char c) +static void +encode_fallback_html(nkf_char c) { (*oconv)(0, '&'); (*oconv)(0, '#'); @@ -961,7 +1094,8 @@ void encode_fallback_html(nkf_char c) return; } -void encode_fallback_xml(nkf_char c) +static void +encode_fallback_xml(nkf_char c) { (*oconv)(0, '&'); (*oconv)(0, '#'); @@ -971,7 +1105,8 @@ void encode_fallback_xml(nkf_char c) return; } -void encode_fallback_java(nkf_char c) +static void +encode_fallback_java(nkf_char c) { (*oconv)(0, '\\'); c &= VALUE_MASK; @@ -991,7 +1126,8 @@ void encode_fallback_java(nkf_char c) return; } -void encode_fallback_perl(nkf_char c) +static void +encode_fallback_perl(nkf_char c) { (*oconv)(0, '\\'); (*oconv)(0, 'x'); @@ -1001,7 +1137,8 @@ void encode_fallback_perl(nkf_char c) return; } -void encode_fallback_subchar(nkf_char c) +static void +encode_fallback_subchar(nkf_char c) { c = unicode_subchar; (*oconv)((c>>8)&0xFF, c&0xFF); @@ -1019,7 +1156,7 @@ static const struct { {"euc","e"}, {"euc-input","E"}, {"fj","jm"}, - {"help","v"}, + {"help",""}, {"jis","j"}, {"jis-input","J"}, {"mac","sLm"}, @@ -1029,7 +1166,7 @@ static const struct { {"sjis","s"}, {"sjis-input","S"}, {"unix","eLu"}, - {"version","V"}, + {"version","v"}, {"windows","sLw"}, {"hiragana","h1"}, {"katakana","h2"}, @@ -1089,15 +1226,17 @@ static const struct { {"prefix=", ""}, }; -static void set_input_encoding(nkf_encoding *enc) +static void +set_input_encoding(nkf_encoding *enc) { switch (nkf_enc_to_index(enc)) { case ISO_8859_1: iso8859_f = TRUE; break; - case CP50220: case CP50221: case CP50222: + if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE; /* -x specified implicitly */ + case CP50220: #ifdef SHIFTJIS_CP932 cp51932_f = TRUE; #endif @@ -1119,6 +1258,7 @@ static void set_input_encoding(nkf_encoding *enc) 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 @@ -1136,10 +1276,11 @@ static void set_input_encoding(nkf_encoding *enc) #endif break; case EUC_JP: - break; + break; case EUCJP_NKF: - break; + break; case CP51932: + if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE; /* -x specified implicitly */ #ifdef SHIFTJIS_CP932 cp51932_f = TRUE; #endif @@ -1148,6 +1289,7 @@ static void set_input_encoding(nkf_encoding *enc) #endif break; case EUCJP_MS: + if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE; /* -x specified implicitly */ #ifdef SHIFTJIS_CP932 cp51932_f = FALSE; #endif @@ -1156,6 +1298,7 @@ static void set_input_encoding(nkf_encoding *enc) #endif break; case EUCJP_ASCII: + if (x0201_f == NKF_UNSPECIFIED) x0201_f = FALSE; /* -x specified implicitly */ #ifdef SHIFTJIS_CP932 cp51932_f = FALSE; #endif @@ -1168,6 +1311,7 @@ static void set_input_encoding(nkf_encoding *enc) x0213_f = TRUE; #ifdef SHIFTJIS_CP932 cp51932_f = FALSE; + if (cp932inv_f == TRUE) cp932inv_f = FALSE; #endif break; case EUC_JISX0213: @@ -1205,11 +1349,11 @@ static void set_input_encoding(nkf_encoding *enc) } } -static void set_output_encoding(nkf_encoding *enc) +static void +set_output_encoding(nkf_encoding *enc) { switch (nkf_enc_to_index(enc)) { case CP50220: - x0201_f = TRUE; #ifdef SHIFTJIS_CP932 if (cp932inv_f == TRUE) cp932inv_f = FALSE; #endif @@ -1218,6 +1362,7 @@ static void set_output_encoding(nkf_encoding *enc) #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 @@ -1225,6 +1370,11 @@ static void set_output_encoding(nkf_encoding *enc) 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 @@ -1232,6 +1382,7 @@ static void set_output_encoding(nkf_encoding *enc) #endif break; case ISO_2022_JP_3: + case ISO_2022_JP_2004: x0212_f = TRUE; x0213_f = TRUE; #ifdef SHIFTJIS_CP932 @@ -1241,6 +1392,7 @@ static void set_output_encoding(nkf_encoding *enc) 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 @@ -1269,6 +1421,7 @@ static void set_output_encoding(nkf_encoding *enc) #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 @@ -1277,12 +1430,14 @@ static void set_output_encoding(nkf_encoding *enc) #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: + 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; @@ -1319,6 +1474,7 @@ static void set_output_encoding(nkf_encoding *enc) output_endian = ENDIAN_LITTLE; output_bom_f = TRUE; break; + case UTF_32: case UTF_32BE_BOM: output_bom_f = TRUE; break; @@ -1334,84 +1490,102 @@ static void set_output_encoding(nkf_encoding *enc) } } -struct input_code * find_inputcode_byfunc(nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)) +static struct input_code* +find_inputcode_byfunc(nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)) { if (iconv_func){ - struct input_code *p = input_code_list; - while (p->name){ - if (iconv_func == p->iconv_func){ - return p; - } - p++; - } + struct input_code *p = input_code_list; + while (p->name){ + if (iconv_func == p->iconv_func){ + return p; + } + p++; + } } return 0; } -void set_iconv(nkf_char f, nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)) +static void +set_iconv(nkf_char f, nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)) { #ifdef INPUT_CODE_FIX if (f || !input_encoding) #endif - if (estab_f != f){ - estab_f = f; - } + if (estab_f != f){ + estab_f = f; + } if (iconv_func #ifdef INPUT_CODE_FIX - && (f == -TRUE || !input_encoding) /* -TRUE means "FORCE" */ + && (f == -TRUE || !input_encoding) /* -TRUE means "FORCE" */ #endif - ){ - iconv = iconv_func; + ){ + iconv = iconv_func; } #ifdef CHECK_OPTION if (estab_f && iconv_for_check != iconv){ - struct input_code *p = find_inputcode_byfunc(iconv); - if (p){ - set_input_codename(p->name); - debug(p->name); - } - iconv_for_check = iconv; + struct input_code *p = find_inputcode_byfunc(iconv); + if (p){ + set_input_codename(p->name); + debug(p->name); + } + iconv_for_check = iconv; } #endif } #ifdef X0212_ENABLE -nkf_char x0212_shift(nkf_char c) +static nkf_char +x0212_shift(nkf_char c) { nkf_char ret = c; c &= 0x7f; if (is_eucg3(ret)){ - if (0x75 <= c && c <= 0x7f){ - ret = c + (0x109 - 0x75); - } + if (0x75 <= c && c <= 0x7f){ + ret = c + (0x109 - 0x75); + } }else{ - if (0x75 <= c && c <= 0x7f){ - ret = c + (0x113 - 0x75); - } + if (0x75 <= c && c <= 0x7f){ + ret = c + (0x113 - 0x75); + } } return ret; } -nkf_char x0212_unshift(nkf_char c) +static nkf_char +x0212_unshift(nkf_char c) { nkf_char ret = c; if (0x7f <= c && c <= 0x88){ - ret = c + (0x75 - 0x7f); + ret = c + (0x75 - 0x7f); }else if (0x89 <= c && c <= 0x92){ - ret = PREFIX_EUCG3 | 0x80 | (c + (0x75 - 0x89)); + ret = PREFIX_EUCG3 | 0x80 | (c + (0x75 - 0x89)); } return ret; } #endif /* X0212_ENABLE */ -nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) +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; - 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); @@ -1448,44 +1622,46 @@ nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) return 0; } -nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) +static nkf_char +s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) { #if defined(SHIFTJIS_CP932) || defined(X0212_ENABLE) nkf_char val; #endif 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)){ - val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40]; - if (val){ - c2 = val >> 8; - c1 = val & 0xff; - } + if (!cp932inv_f && !x0213_f && is_ibmext_in_sjis(c2)){ + val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40]; + if (val){ + c2 = val >> 8; + c1 = val & 0xff; + } } if (cp932inv_f - && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){ - nkf_char c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40]; - if (c){ - c2 = c >> 8; - c1 = c & 0xff; - } + && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){ + val = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40]; + if (val){ + c2 = val >> 8; + c1 = val & 0xff; + } } #endif /* SHIFTJIS_CP932 */ #ifdef X0212_ENABLE if (!x0213_f && is_ibmext_in_sjis(c2)){ - val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40]; - if (val){ - if (val > 0x7FFF){ - c2 = PREFIX_EUCG3 | ((val >> 8) & 0x7f); - c1 = val & 0xff; - }else{ - c2 = val >> 8; - c1 = val & 0xff; - } - if (p2) *p2 = c2; - if (p1) *p1 = c1; - return 0; - } + val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40]; + if (val){ + if (val > 0x7FFF){ + c2 = PREFIX_EUCG3 | ((val >> 8) & 0x7f); + c1 = val & 0xff; + }else{ + c2 = val >> 8; + c1 = val & 0xff; + } + if (p2) *p2 = c2; + if (p1) *p1 = c1; + return 0; + } } #endif if(c2 >= 0x80){ @@ -1518,45 +1694,47 @@ nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) } #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) -void nkf_unicode_to_utf8(nkf_char val, int *p1, int *p2, int *p3, int *p4) +static void +nkf_unicode_to_utf8(nkf_char val, nkf_char *p1, nkf_char *p2, nkf_char *p3, nkf_char *p4) { val &= VALUE_MASK; if (val < 0x80){ - *p1 = val; - *p2 = 0; - *p3 = 0; - *p4 = 0; + *p1 = val; + *p2 = 0; + *p3 = 0; + *p4 = 0; }else if (val < 0x800){ *p1 = 0xc0 | (val >> 6); *p2 = 0x80 | (val & 0x3f); - *p3 = 0; - *p4 = 0; + *p3 = 0; + *p4 = 0; } else if (nkf_char_unicode_bmp_p(val)) { - *p1 = 0xe0 | (val >> 12); - *p2 = 0x80 | ((val >> 6) & 0x3f); - *p3 = 0x80 | ( val & 0x3f); - *p4 = 0; + *p1 = 0xe0 | (val >> 12); + *p2 = 0x80 | ((val >> 6) & 0x3f); + *p3 = 0x80 | ( val & 0x3f); + *p4 = 0; } else if (nkf_char_unicode_value_p(val)) { - *p1 = 0xe0 | (val >> 16); - *p2 = 0x80 | ((val >> 12) & 0x3f); - *p3 = 0x80 | ((val >> 6) & 0x3f); - *p4 = 0x80 | ( val & 0x3f); + *p1 = 0xf0 | (val >> 18); + *p2 = 0x80 | ((val >> 12) & 0x3f); + *p3 = 0x80 | ((val >> 6) & 0x3f); + *p4 = 0x80 | ( val & 0x3f); } else { - *p1 = 0; - *p2 = 0; - *p3 = 0; - *p4 = 0; + *p1 = 0; + *p2 = 0; + *p3 = 0; + *p4 = 0; } } -nkf_char nkf_utf8_to_unicode(int c1, int c2, int c3, int c4) +static nkf_char +nkf_utf8_to_unicode(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4) { nkf_char wc; if (c1 <= 0x7F) { /* single byte */ wc = c1; } - else if (c1 <= 0xC3) { + else if (c1 <= 0xC1) { /* trail byte or invalid */ return -1; } @@ -1586,9 +1764,10 @@ nkf_char nkf_utf8_to_unicode(int c1, int c2, int c3, int c4) #endif #ifdef UTF8_INPUT_ENABLE -static int unicode_to_jis_common2(nkf_char c1, nkf_char c0, - const unsigned short *const *pp, nkf_char psize, - nkf_char *p2, nkf_char *p1) +static int +unicode_to_jis_common2(nkf_char c1, nkf_char c0, + const unsigned short *const *pp, nkf_char psize, + nkf_char *p2, nkf_char *p1) { nkf_char c2; const unsigned short *p; @@ -1606,14 +1785,14 @@ static int unicode_to_jis_common2(nkf_char c1, nkf_char c0, val = p[c0]; if (val == 0) return 1; if (no_cp932ext_f && ( - (val>>8) == 0x2D || /* NEC special characters */ - val > NKF_INT32_C(0xF300) /* IBM extended characters */ - )) return 1; + (val>>8) == 0x2D || /* NEC special characters */ + val > NKF_INT32_C(0xF300) /* IBM extended characters */ + )) return 1; c2 = val >> 8; - if (val > 0x7FFF){ - c2 &= 0x7f; - c2 |= PREFIX_EUCG3; + if (val > 0x7FFF){ + c2 &= 0x7f; + c2 |= PREFIX_EUCG3; } if (c2 == SO) c2 = JIS_X_0201_1976_K; c1 = val & 0xFF; @@ -1622,7 +1801,8 @@ static int unicode_to_jis_common2(nkf_char c1, nkf_char c0, return 0; } -static nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1) +static int +unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1) { const unsigned short *const *pp; const unsigned short *const *const *ppp; @@ -1694,6 +1874,7 @@ static nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf 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){ @@ -1721,10 +1902,10 @@ static nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf case 0xE3: switch(c1){ case 0x82: - if(c0 == 0x94) return 1; + if(c0 == 0x94) return 1; break; case 0x83: - if(c0 == 0xBB) return 1; + if(c0 == 0xBB) return 1; break; } break; @@ -1761,6 +1942,7 @@ static nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf 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; @@ -1778,7 +1960,17 @@ static nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf } #ifdef UTF8_OUTPUT_ENABLE -nkf_char e2w_conv(nkf_char c2, nkf_char c1) +#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) { const unsigned short *p; @@ -1791,23 +1983,26 @@ nkf_char e2w_conv(nkf_char c2, nkf_char c1) return 0xA9; } } - p = euc_to_utf8_1byte; + p = euc_to_utf8_1byte; #ifdef X0212_ENABLE } else if (is_eucg3(c2)){ if(ms_ucs_map_f == UCS_MAP_ASCII&& c2 == NKF_INT32_C(0x8F22) && c1 == 0x43){ return 0xA6; } - c2 = (c2&0x7f) - 0x21; - if (0<=c2 && c2= 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 -nkf_char w2e_conv(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1) +static nkf_char +w2e_conv(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1) { nkf_char ret = 0; if (!c1){ - *p2 = 0; - *p1 = c2; + *p2 = 0; + *p1 = c2; }else if (0xc0 <= c2 && c2 <= 0xef) { ret = unicode_to_jis_common(c2, c1, c0, p2, p1); #ifdef NUMCHAR_OPTION - if (ret > 0){ - if (p2) *p2 = 0; - if (p1) *p1 = nkf_char_unicode_new(nkf_utf8_to_unicode(c2, c1, c0, 0)); - ret = 0; - } + if (ret > 0){ + if (p2) *p2 = 0; + if (p1) *p1 = nkf_char_unicode_new(nkf_utf8_to_unicode(c2, c1, c0, 0)); + ret = 0; + } #endif } return ret; } #ifdef UTF8_INPUT_ENABLE -nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1) +static nkf_char +w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1) { - int c1, c2, c3, c4; + nkf_char c1, c2, c3, c4; nkf_char ret = 0; val &= VALUE_MASK; if (val < 0x80) { - *p2 = 0; - *p1 = val; + *p2 = 0; + *p1 = val; } else if (nkf_char_unicode_bmp_p(val)){ nkf_unicode_to_utf8(val, &c1, &c2, &c3, &c4); @@ -1862,6 +2090,25 @@ nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1) } } 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); } @@ -1869,7 +2116,8 @@ nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1) } #endif -nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) +static nkf_char +e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) { if (c2 == JIS_X_0201_1976_K || c2 == SS2){ if (iso2022jp_f && !x0201_f) { @@ -1880,9 +2128,9 @@ nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) } #ifdef X0212_ENABLE }else if (c2 == 0x8f){ - if (c0 == 0){ - return -1; - } + if (c0 == 0){ + return -1; + } if (!cp51932_f && !x0213_f && 0xF5 <= c1 && c1 <= 0xFE && 0xA1 <= c0 && c0 <= 0xFE) { /* encoding is eucJP-ms, so invert to Unicode Private User Area */ c1 = nkf_char_unicode_new((c1 - 0xF5) * 94 + c0 - 0xA1 + 0xE3AC); @@ -1902,10 +2150,10 @@ nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) } } #endif /* SHIFTJIS_CP932 */ - } + } #endif /* X0212_ENABLE */ } else if ((c2 == EOF) || (c2 == 0) || c2 < SP || c2 == ISO_8859_1) { - /* NOP */ + /* NOP */ } else { if (!cp51932_f && ms_ucs_map_f && 0xF5 <= c2 && c2 <= 0xFE && 0xA1 <= c1 && c1 <= 0xFE) { /* encoding is eucJP-ms, so invert to Unicode Private User Area */ @@ -1926,13 +2174,14 @@ nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) } } #endif /* SHIFTJIS_CP932 */ - } + } } (*oconv)(c2, c1); return 0; } -nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0) +static nkf_char +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) { @@ -1941,21 +2190,46 @@ nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0) c1 &= 0x7f; } } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) { - /* NOP */ + /* NOP */ } else if (!x0213_f && 0xF0 <= c2 && c2 <= 0xF9 && 0x40 <= c1 && c1 <= 0xFC) { /* CP932 UDC */ if(c1 == 0x7F) return 0; c1 = nkf_char_unicode_new((c2 - 0xF0) * 188 + (c1 - 0x40 - (0x7E < c1)) + 0xE000); c2 = 0; } else { - nkf_char ret = s2e_conv(c2, c1, &c2, &c1); - if (ret) return ret; + nkf_char ret = s2e_conv(c2, c1, &c2, &c1); + if (ret) return ret; } (*oconv)(c2, c1); return 0; } -nkf_char w_iconv(nkf_char c1, nkf_char c2, nkf_char c3) +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) { nkf_char ret = 0, c4 = 0; static const char w_iconv_utf8_1st_byte[] = @@ -1976,7 +2250,7 @@ nkf_char w_iconv(nkf_char c1, nkf_char c2, nkf_char c3) } else if ((c1 & 0xC0) == 0x80) { /* 0x80-0xbf : trail byte */ return 0; } else{ - switch (w_iconv_utf8_1st_byte[c1 - 0xC0]) { + switch (w_iconv_utf8_1st_byte[c1 - 0xC0]) { case 21: if (c2 < 0x80 || 0xBF < c2) return 0; break; @@ -2021,16 +2295,32 @@ nkf_char 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 { + 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){ - (*oconv)(c1, c2); + (*oconv)(c1, c2); + } + 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 -static size_t unicode_iconv(nkf_char wc) +#define NKF_ICONV_WAIT_COMBINING_CHAR -14 +#define NKF_ICONV_NOT_COMBINED -15 +static size_t +unicode_iconv(nkf_char wc, int nocombine) { nkf_char c1, c2; int ret = 0; @@ -2042,6 +2332,8 @@ static size_t unicode_iconv(nkf_char wc) /* 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) { @@ -2054,10 +2346,52 @@ static size_t unicode_iconv(nkf_char wc) 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)) -size_t nkf_iconv_utf_16(int c1, int c2, int c3, int c4) +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) { nkf_char wc; @@ -2084,27 +2418,62 @@ size_t nkf_iconv_utf_16(int c1, int c2, int c3, int c4) } } - return (*unicode_iconv)(wc); + return (*unicode_iconv)(wc, FALSE); } -nkf_char w_iconv16(nkf_char c2, nkf_char c1, nkf_char c0) +static size_t +nkf_iconv_utf_16_combine(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4) { - return 0; + 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); } -nkf_char w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0) +static size_t +nkf_iconv_utf_16_nocombine(nkf_char c1, nkf_char c2) { - return 0; + nkf_char wc; + if (input_endian == ENDIAN_BIG) + wc = c1 << 8 | c2; + else + wc = c2 << 8 | c1; + return (*unicode_iconv)(wc, TRUE); } -size_t nkf_iconv_utf_32(int c1, int c2, int c3, int c4) +static nkf_char +w_iconv16(nkf_char c2, nkf_char c1, ARG_UNUSED nkf_char c0) { - nkf_char wc; + (*oconv)(c2, c1); + return 16; /* different from w_iconv32 */ +} - if (c1 == EOF) { - (*oconv)(EOF, 0); - return 0; - } +static nkf_char +w_iconv32(nkf_char c2, nkf_char c1, ARG_UNUSED nkf_char c0) +{ + (*oconv)(c2, c1); + return 32; /* different from w_iconv16 */ +} + +static nkf_char +utf32_to_nkf_char(nkf_char c1, nkf_char c2, nkf_char c3, nkf_char c4) +{ + nkf_char wc; switch(input_endian){ case ENDIAN_BIG: @@ -2122,21 +2491,62 @@ size_t nkf_iconv_utf_32(int c1, int c2, int c3, int c4) 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 #define output_ascii_escape_sequence(mode) do { \ - if (output_mode != ASCII && output_mode != ISO_8859_1) { \ - (*o_putc)(ESC); \ - (*o_putc)('('); \ - (*o_putc)(ascii_intro); \ - output_mode = mode; \ - } \ -} while (0) - -void output_escape_sequence(int mode) + if (output_mode != ASCII && output_mode != ISO_8859_1) { \ + (*o_putc)(ESC); \ + (*o_putc)('('); \ + (*o_putc)(ascii_intro); \ + output_mode = mode; \ + } \ + } while (0) + +static void +output_escape_sequence(int mode) { if (output_mode == mode) return; @@ -2167,23 +2577,24 @@ void output_escape_sequence(int mode) (*o_putc)('$'); (*o_putc)('('); (*o_putc)('Q'); - break; + break; case JIS_X_0213_2: (*o_putc)(ESC); (*o_putc)('$'); (*o_putc)('('); (*o_putc)('P'); - break; + break; } output_mode = mode; } -void j_oconv(nkf_char c2, nkf_char c1) +static void +j_oconv(nkf_char c2, nkf_char c1) { #ifdef NUMCHAR_OPTION if (c2 == 0 && nkf_char_unicode_p(c1)){ - w16e_conv(c1, &c2, &c1); - if (c2 == 0 && nkf_char_unicode_p(c1)){ + w16e_conv(c1, &c2, &c1); + if (c2 == 0 && nkf_char_unicode_p(c1)){ c2 = c1 & VALUE_MASK; if (ms_ucs_map_f && 0xE000 <= c2 && c2 <= 0xE757) { /* CP5022x UDC */ @@ -2194,16 +2605,16 @@ void j_oconv(nkf_char c2, nkf_char c1) if (encode_fallback) (*encode_fallback)(c1); return; } - } + } } #endif if (c2 == 0) { output_ascii_escape_sequence(ASCII); - (*o_putc)(c1); + (*o_putc)(c1); } else if (c2 == EOF) { output_ascii_escape_sequence(ASCII); - (*o_putc)(EOF); + (*o_putc)(EOF); } else if (c2 == ISO_8859_1) { output_ascii_escape_sequence(ISO_8859_1); @@ -2211,28 +2622,29 @@ void j_oconv(nkf_char c2, nkf_char c1) } else if (c2 == JIS_X_0201_1976_K) { output_escape_sequence(JIS_X_0201_1976_K); - (*o_putc)(c1); + (*o_putc)(c1); #ifdef X0212_ENABLE } else if (is_eucg3(c2)){ output_escape_sequence(x0213_f ? JIS_X_0213_2 : JIS_X_0212); - (*o_putc)(c2 & 0x7f); - (*o_putc)(c1); + (*o_putc)(c2 & 0x7f); + (*o_putc)(c1); #endif } else { if(ms_ucs_map_f ? c2<0x20 || 0x92> 8; - c1 = c & 0xff; - } - } + if (cp932inv_f + && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){ + nkf_char c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40]; + if (c){ + c2 = c >> 8; + c1 = c & 0xff; + } + } #endif /* SHIFTJIS_CP932 */ - (*o_putc)(c2); + (*o_putc)(c2); if (prefix_table[(unsigned char)c1]){ - (*o_putc)(prefix_table[(unsigned char)c1]); + (*o_putc)(prefix_table[(unsigned char)c1]); } - (*o_putc)(c1); + (*o_putc)(c1); } } #ifdef UTF8_OUTPUT_ENABLE -void w_oconv(nkf_char c2, nkf_char c1) +#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) { - int c3, c4; - nkf_char val; + nkf_char c3, c4; + nkf_char val, val2; if (output_bom_f) { output_bom_f = FALSE; - (*o_putc)('\357'); + (*o_putc)('\357'); (*o_putc)('\273'); (*o_putc)('\277'); } if (c2 == EOF) { - (*o_putc)(EOF); - return; + (*o_putc)(EOF); + return; } 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; } if (c2 == 0) { - (*o_putc)(c1); + (*o_putc)(c1); } 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); - } + if (val){ + val2 = e2w_combining(val, c2, c1); + if (val2) + OUTPUT_UTF8(val2); + OUTPUT_UTF8(val); + } } } -void w_oconv16(nkf_char c2, nkf_char c1) +#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; - 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) { - (*o_putc)(EOF); - return; + (*o_putc)(EOF); + return; } 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) { - 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 (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); } } -void w_oconv32(nkf_char c2, nkf_char c1) +#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) { if (output_bom_f) { output_bom_f = FALSE; - if (output_endian == ENDIAN_LITTLE){ - (*o_putc)(0xFF); - (*o_putc)(0xFE); - (*o_putc)(0); - (*o_putc)(0); - }else{ - (*o_putc)(0); - (*o_putc)(0); - (*o_putc)(0xFE); - (*o_putc)(0xFF); - } + if (output_endian == ENDIAN_LITTLE){ + (*o_putc)(0xFF); + (*o_putc)(0xFE); + (*o_putc)(0); + (*o_putc)(0); + }else{ + (*o_putc)(0); + (*o_putc)(0); + (*o_putc)(0xFE); + (*o_putc)(0xFF); + } } if (c2 == EOF) { - (*o_putc)(EOF); - return; + (*o_putc)(EOF); + return; } if (c2 == ISO_8859_1) { - c1 |= 0x80; + c1 |= 0x80; } 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 -#define SCORE_L2 (1) /* 第2水準漢字 */ -#define SCORE_KANA (SCORE_L2 << 1) /* いわゆる半角カナ */ -#define SCORE_DEPEND (SCORE_KANA << 1) /* 機種依存文字 */ -#define SCORE_CP932 (SCORE_DEPEND << 1) /* CP932 による読み換え (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_NO_EXIST (SCORE_X0212 << 1) /* 存在しない文字 */ -#define SCORE_iMIME (SCORE_NO_EXIST << 1) /* MIME による指定 */ -#define SCORE_ERROR (SCORE_iMIME << 1) /* エラー */ +#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) -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, - 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_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, + SCORE_L2, SCORE_DEPEND, SCORE_X0213, SCORE_X0213, 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, }; -void set_code_score(struct input_code *ptr, nkf_char score) +static void +set_code_score(struct input_code *ptr, nkf_char score) { if (ptr){ - ptr->score |= score; + ptr->score |= score; } } -void clr_code_score(struct input_code *ptr, nkf_char score) +static void +clr_code_score(struct input_code *ptr, nkf_char score) { if (ptr){ - ptr->score &= ~score; + ptr->score &= ~score; } } -void code_score(struct input_code *ptr) +static void +code_score(struct input_code *ptr) { nkf_char c2 = ptr->buf[0]; -#ifdef UTF8_OUTPUT_ENABLE nkf_char c1 = ptr->buf[1]; -#endif if (c2 < 0){ - set_code_score(ptr, SCORE_ERROR); + set_code_score(ptr, SCORE_ERROR); }else if (c2 == SS2){ - set_code_score(ptr, SCORE_KANA); + 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); + set_code_score(ptr, SCORE_NO_EXIST); #endif }else if ((c2 & 0x70) == 0x20){ - set_code_score(ptr, score_table_A0[c2 & 0x0f]); + set_code_score(ptr, score_table_A0[c2 & 0x0f]); }else if ((c2 & 0x70) == 0x70){ - set_code_score(ptr, score_table_F0[c2 & 0x0f]); + set_code_score(ptr, score_table_F0[c2 & 0x0f]); }else if ((c2 & 0x70) >= 0x50){ - set_code_score(ptr, SCORE_L2); + set_code_score(ptr, SCORE_L2); } } -void status_disable(struct input_code *ptr) +static void +status_disable(struct input_code *ptr) { ptr->stat = -1; ptr->buf[0] = -1; @@ -2587,83 +3044,89 @@ void status_disable(struct input_code *ptr) if (iconv == ptr->iconv_func) set_iconv(FALSE, 0); } -void status_push_ch(struct input_code *ptr, nkf_char c) +static void +status_push_ch(struct input_code *ptr, nkf_char c) { ptr->buf[ptr->index++] = c; } -void status_clear(struct input_code *ptr) +static void +status_clear(struct input_code *ptr) { ptr->stat = 0; ptr->index = 0; } -void status_reset(struct input_code *ptr) +static void +status_reset(struct input_code *ptr) { status_clear(ptr); ptr->score = SCORE_INIT; } -void status_reinit(struct input_code *ptr) +static void +status_reinit(struct input_code *ptr) { status_reset(ptr); ptr->_file_stat = 0; } -void status_check(struct input_code *ptr, nkf_char c) +static void +status_check(struct input_code *ptr, nkf_char c) { if (c <= DEL && estab_f){ - status_reset(ptr); + status_reset(ptr); } } -void s_status(struct input_code *ptr, nkf_char c) +static void +s_status(struct input_code *ptr, nkf_char c) { switch(ptr->stat){ - case -1: - status_check(ptr, c); - break; - case 0: - if (c <= DEL){ - break; - }else if (nkf_char_unicode_p(c)){ - break; - }else if (0xa1 <= c && c <= 0xdf){ - status_push_ch(ptr, SS2); - status_push_ch(ptr, c); - code_score(ptr); - status_clear(ptr); - }else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xea)){ - ptr->stat = 1; - status_push_ch(ptr, c); - }else if (0xed <= c && c <= 0xee){ - ptr->stat = 3; - status_push_ch(ptr, c); + case -1: + status_check(ptr, c); + break; + case 0: + if (c <= DEL){ + break; + }else if (nkf_char_unicode_p(c)){ + break; + }else if (0xa1 <= c && c <= 0xdf){ + status_push_ch(ptr, SS2); + status_push_ch(ptr, c); + code_score(ptr); + status_clear(ptr); + }else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xea)){ + ptr->stat = 1; + status_push_ch(ptr, c); + }else if (0xed <= c && c <= 0xee){ + ptr->stat = 3; + status_push_ch(ptr, c); #ifdef SHIFTJIS_CP932 - }else if (is_ibmext_in_sjis(c)){ - ptr->stat = 2; - status_push_ch(ptr, c); + }else if (is_ibmext_in_sjis(c)){ + ptr->stat = 2; + status_push_ch(ptr, c); #endif /* SHIFTJIS_CP932 */ #ifdef X0212_ENABLE - }else if (0xf0 <= c && c <= 0xfc){ - ptr->stat = 1; - status_push_ch(ptr, c); + }else if (0xf0 <= c && c <= 0xfc){ + ptr->stat = 1; + status_push_ch(ptr, c); #endif /* X0212_ENABLE */ - }else{ - status_disable(ptr); - } - break; - case 1: - if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){ - status_push_ch(ptr, c); - s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]); - code_score(ptr); - status_clear(ptr); - }else{ - status_disable(ptr); - } - break; - case 2: + }else{ + status_disable(ptr); + } + break; + case 1: + if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){ + status_push_ch(ptr, c); + s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]); + code_score(ptr); + status_clear(ptr); + }else{ + status_disable(ptr); + } + break; + case 2: #ifdef SHIFTJIS_CP932 if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)) { status_push_ch(ptr, c); @@ -2675,202 +3138,238 @@ void s_status(struct input_code *ptr, nkf_char c) } #endif /* SHIFTJIS_CP932 */ status_disable(ptr); - break; - case 3: - if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){ - status_push_ch(ptr, c); - s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]); + break; + case 3: + if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){ + status_push_ch(ptr, c); + s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]); set_code_score(ptr, SCORE_CP932); status_clear(ptr); - }else{ - status_disable(ptr); - } - break; + }else{ + status_disable(ptr); + } + break; } } -void e_status(struct input_code *ptr, nkf_char c) +static void +e_status(struct input_code *ptr, nkf_char c) { switch (ptr->stat){ - case -1: - status_check(ptr, c); - break; - case 0: - if (c <= DEL){ - break; - }else if (nkf_char_unicode_p(c)){ - break; - }else if (SS2 == c || (0xa1 <= c && c <= 0xfe)){ - ptr->stat = 1; - status_push_ch(ptr, c); + case -1: + status_check(ptr, c); + break; + case 0: + if (c <= DEL){ + break; + }else if (nkf_char_unicode_p(c)){ + break; + }else if (SS2 == c || (0xa1 <= c && c <= 0xfe)){ + ptr->stat = 1; + status_push_ch(ptr, c); #ifdef X0212_ENABLE - }else if (0x8f == c){ - ptr->stat = 2; - status_push_ch(ptr, c); + }else if (0x8f == c){ + ptr->stat = 2; + status_push_ch(ptr, c); #endif /* X0212_ENABLE */ - }else{ - status_disable(ptr); - } - break; - case 1: - if (0xa1 <= c && c <= 0xfe){ - status_push_ch(ptr, c); - code_score(ptr); - status_clear(ptr); - }else{ - status_disable(ptr); - } - break; + }else{ + status_disable(ptr); + } + break; + case 1: + if (0xa1 <= c && c <= 0xfe){ + status_push_ch(ptr, c); + code_score(ptr); + status_clear(ptr); + }else{ + status_disable(ptr); + } + break; #ifdef X0212_ENABLE - case 2: - if (0xa1 <= c && c <= 0xfe){ - ptr->stat = 1; - status_push_ch(ptr, c); - }else{ - status_disable(ptr); - } + case 2: + if (0xa1 <= c && c <= 0xfe){ + ptr->stat = 1; + status_push_ch(ptr, c); + }else{ + status_disable(ptr); + } #endif /* X0212_ENABLE */ } } #ifdef UTF8_INPUT_ENABLE -void w_status(struct input_code *ptr, nkf_char c) +static void +w_status(struct input_code *ptr, nkf_char c) { switch (ptr->stat){ - case -1: - status_check(ptr, c); - break; - case 0: - if (c <= DEL){ - break; - }else if (nkf_char_unicode_p(c)){ - break; - }else if (0xc0 <= c && c <= 0xdf){ - ptr->stat = 1; - status_push_ch(ptr, c); - }else if (0xe0 <= c && c <= 0xef){ - ptr->stat = 2; - status_push_ch(ptr, c); - }else if (0xf0 <= c && c <= 0xf4){ - ptr->stat = 3; - status_push_ch(ptr, c); - }else{ - status_disable(ptr); - } - break; - case 1: - case 2: - if (0x80 <= c && c <= 0xbf){ - status_push_ch(ptr, c); - if (ptr->index > ptr->stat){ - int bom = (ptr->buf[0] == 0xef && ptr->buf[1] == 0xbb - && ptr->buf[2] == 0xbf); - w2e_conv(ptr->buf[0], ptr->buf[1], ptr->buf[2], - &ptr->buf[0], &ptr->buf[1]); - if (!bom){ - code_score(ptr); - } - status_clear(ptr); - } - }else{ - status_disable(ptr); - } - break; - case 3: + case -1: + status_check(ptr, c); + break; + case 0: + if (c <= DEL){ + break; + }else if (nkf_char_unicode_p(c)){ + break; + }else if (0xc0 <= c && c <= 0xdf){ + ptr->stat = 1; + status_push_ch(ptr, c); + }else if (0xe0 <= c && c <= 0xef){ + ptr->stat = 2; + status_push_ch(ptr, c); + }else if (0xf0 <= c && c <= 0xf4){ + ptr->stat = 3; + status_push_ch(ptr, c); + }else{ + status_disable(ptr); + } + break; + case 1: + case 2: + if (0x80 <= c && c <= 0xbf){ + status_push_ch(ptr, c); + if (ptr->index > ptr->stat){ + int bom = (ptr->buf[0] == 0xef && ptr->buf[1] == 0xbb + && ptr->buf[2] == 0xbf); + w2e_conv(ptr->buf[0], ptr->buf[1], ptr->buf[2], + &ptr->buf[0], &ptr->buf[1]); + if (!bom){ + code_score(ptr); + } + status_clear(ptr); + } + }else{ + status_disable(ptr); + } + break; + case 3: if (0x80 <= c && c <= 0xbf){ if (ptr->index < ptr->stat){ status_push_ch(ptr, c); } else { - status_clear(ptr); + status_clear(ptr); } - }else{ - status_disable(ptr); - } - break; + }else{ + status_disable(ptr); + } + break; } } #endif -void code_status(nkf_char c) +static void +code_status(nkf_char c) { int action_flag = 1; struct input_code *result = 0; struct input_code *p = input_code_list; while (p->name){ - if (!p->status_func) { + if (!p->status_func) { ++p; continue; } - if (!p->status_func) + if (!p->status_func) continue; - (p->status_func)(p, c); - if (p->stat > 0){ - action_flag = 0; - }else if(p->stat == 0){ - if (result){ - action_flag = 0; - }else{ - result = p; - } - } - ++p; + (p->status_func)(p, c); + if (p->stat > 0){ + action_flag = 0; + }else if(p->stat == 0){ + if (result){ + action_flag = 0; + }else{ + result = p; + } + } + ++p; } if (action_flag){ - if (result && !estab_f){ - set_iconv(TRUE, result->iconv_func); - }else if (c <= DEL){ - struct input_code *ptr = input_code_list; - while (ptr->name){ - status_reset(ptr); - ++ptr; - } - } + if (result && !estab_f){ + set_iconv(TRUE, result->iconv_func); + }else if (c <= DEL){ + struct input_code *ptr = input_code_list; + while (ptr->name){ + status_reset(ptr); + ++ptr; + } + } + } +} + +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 -nkf_char std_getc(FILE *f) +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*/ -nkf_char std_ungetc(nkf_char c, FILE *f) +static nkf_char +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; } #ifndef WIN32DLL -void std_putc(nkf_char c) +static void +std_putc(nkf_char c) { if(c!=EOF) - putchar(c); + putchar(c); } #endif /*WIN32DLL*/ -static unsigned char hold_buf[HOLD_SIZE*2]; +static nkf_char hold_buf[HOLD_SIZE*2]; static int hold_count = 0; -nkf_char push_hold_buf(nkf_char c2) +static nkf_char +push_hold_buf(nkf_char c2) { if (hold_count >= HOLD_SIZE*2) - return (EOF); - hold_buf[hold_count++] = (unsigned char)c2; + return (EOF); + hold_buf[hold_count++] = c2; return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count); } -static int h_conv(FILE *f, int c1, int c2) +static int +h_conv(FILE *f, nkf_char c1, nkf_char c2) { - int ret, c4, c3; + int ret; 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 */ @@ -2881,29 +3380,29 @@ static int h_conv(FILE *f, int c1, int c2) push_hold_buf(c2); while ((c2 = (*i_getc)(f)) != EOF) { - if (c2 == ESC){ + if (c2 == ESC){ (*i_ungetc)(c2,f); - break; - } - code_status(c2); - if (push_hold_buf(c2) == EOF || estab_f) { - break; - } + break; + } + code_status(c2); + if (push_hold_buf(c2) == EOF || estab_f) { + break; + } } if (!estab_f) { - struct input_code *p = input_code_list; - struct input_code *result = p; - if (c2 == EOF) { - code_status(c2); - } - while (p->name) { - if (p->status_func && p->score < result->score) { - result = p; - } - p++; - } - set_iconv(TRUE, result->iconv_func); + struct input_code *p = input_code_list; + struct input_code *result = p; + if (c2 == EOF) { + code_status(c2); + } + while (p->name) { + if (p->status_func && p->score < result->score) { + result = p; + } + p++; + } + set_iconv(TRUE, result->iconv_func); } @@ -2919,58 +3418,150 @@ static int h_conv(FILE *f, int c1, int c2) ret = c2; hold_index = 0; while (hold_index < hold_count){ - c1 = hold_buf[hold_index++]; - 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; - } - if (hold_index < hold_count){ - c2 = hold_buf[hold_index++]; - }else{ - c2 = (*i_getc)(f); - if (c2 == EOF){ - c4 = EOF; - break; - } - code_status(c2); - } - c3 = 0; - switch ((*iconv)(c1, c2, 0)) { /* can be EUC/SJIS/UTF-8 */ + c1 = hold_buf[hold_index++]; + 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; + } + fromhold_count = 1; + if (hold_index < hold_count){ + c2 = hold_buf[hold_index++]; + fromhold_count++; + }else{ + c2 = (*i_getc)(f); + if (c2 == EOF){ + c4 = EOF; + break; + } + code_status(c2); + } + c3 = 0; + switch ((*iconv)(c1, c2, 0)) { /* can be EUC/SJIS/UTF-8 */ case -2: /* 4 bytes UTF-8 */ - if (hold_index < hold_count){ - c3 = hold_buf[hold_index++]; - } else if ((c3 = (*i_getc)(f)) == EOF) { + if (hold_index < hold_count){ + c3 = hold_buf[hold_index++]; + } 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 { - code_status(c4); - (*iconv)(c1, c2, (c3<<8)|c4); + hold_index -= 2; } - } + } break; case -1: /* 3 bytes EUC or UTF-8 */ - if (hold_index < hold_count){ - c3 = hold_buf[hold_index++]; - } else if ((c3 = (*i_getc)(f)) == EOF) { + 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); - } - (*iconv)(c1, c2, c3); - break; + code_status(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; } @@ -2980,9 +3571,11 @@ static int h_conv(FILE *f, int c1, int c2) /* * Check and Ignore BOM */ -void check_bom(FILE *f) +static void +check_bom(FILE *f) { int c2; + input_bom_f = FALSE; switch(c2 = (*i_getc)(f)){ case 0x00: if((c2 = (*i_getc)(f)) == 0x00){ @@ -2992,6 +3585,7 @@ void check_bom(FILE *f) set_iconv(TRUE, w_iconv32); } if (iconv == w_iconv32) { + input_bom_f = TRUE; input_endian = ENDIAN_BIG; return; } @@ -3022,6 +3616,7 @@ void check_bom(FILE *f) set_iconv(TRUE, w_iconv); } if (iconv == w_iconv) { + input_bom_f = TRUE; return; } (*i_ungetc)(0xBF,f); @@ -3050,6 +3645,7 @@ void check_bom(FILE *f) } if (iconv == w_iconv16) { input_endian = ENDIAN_BIG; + input_bom_f = TRUE; return; } (*i_ungetc)(0xFF,f); @@ -3065,6 +3661,7 @@ void check_bom(FILE *f) } if (iconv == w_iconv32) { input_endian = ENDIAN_LITTLE; + input_bom_f = TRUE; return; } (*i_ungetc)(0x00,f); @@ -3076,6 +3673,7 @@ void check_bom(FILE *f) } if (iconv == w_iconv16) { input_endian = ENDIAN_LITTLE; + input_bom_f = TRUE; return; } (*i_ungetc)(0xFE,f); @@ -3088,73 +3686,55 @@ void 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]; -} - -nkf_char broken_getc(FILE *f) +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); - 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); - broken_state.status = 0; + nkf_state->broken_state = 0; 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; } - } 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); - broken_state.status = 0; + nkf_state->broken_state = 0; 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 { - broken_state.status = c; + nkf_state->broken_state = c; return c; } } -nkf_char broken_ungetc(nkf_char c, FILE *f) +static nkf_char +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; } -void eol_conv(nkf_char c2, nkf_char c1) +static void +eol_conv(nkf_char c2, nkf_char c1) { if (guess_f && input_eol != EOF) { if (c2 == 0 && c1 == LF) { @@ -3174,199 +3754,233 @@ void eol_conv(nkf_char c2, nkf_char 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() - LF add newline and output char - CR add newline and output nothing - SP space - 0 skip - 1 (or else) normal output + LF add newline and output char + CR add newline and output nothing + SP space + 0 skip + 1 (or else) normal output - fold state in prev (previous character) + fold state in prev (previous character) - >0x80 Japanese (X0208/X0201) - <0x80 ASCII - LF new line - SP space + >0x80 Japanese (X0208/X0201) + <0x80 ASCII + LF new line + SP space - This fold algorthm does not preserve heading space in a line. - This is the main difference from fmt. -*/ + This fold algorthm does not preserve heading space in a line. + This is the main difference from fmt. + */ #define char_size(c2,c1) (c2?2:1) -void fold_conv(nkf_char c2, nkf_char c1) +static void +fold_conv(nkf_char c2, nkf_char c1) { nkf_char prev0; nkf_char fold_state; if (c1== CR && !fold_preserve_f) { - fold_state=0; /* ignore cr */ + fold_state=0; /* ignore cr */ }else if (c1== LF&&f_prev==CR && fold_preserve_f) { - f_prev = LF; - fold_state=0; /* ignore cr */ + f_prev = LF; + fold_state=0; /* ignore cr */ } else if (c1== BS) { - if (f_line>0) f_line--; - fold_state = 1; + if (f_line>0) f_line--; + fold_state = 1; } else if (c2==EOF && f_line != 0) { /* close open last line */ - fold_state = LF; + fold_state = LF; } else if ((c1==LF && !fold_preserve_f) - || ((c1==CR||(c1==LF&&f_prev!=CR)) - && fold_preserve_f)) { - /* new line */ - if (fold_preserve_f) { - f_prev = c1; - f_line = 0; - fold_state = CR; - } else if ((f_prev == c1 && !fold_preserve_f) - || (f_prev == LF && fold_preserve_f) - ) { /* duplicate newline */ - if (f_line) { - f_line = 0; - fold_state = LF; /* output two newline */ - } else { - f_line = 0; - fold_state = 1; - } - } else { - if (f_prev&0x80) { /* Japanese? */ - f_prev = c1; - fold_state = 0; /* ignore given single newline */ - } else if (f_prev==SP) { - fold_state = 0; - } else { - f_prev = c1; - if (++f_line<=fold_len) - fold_state = SP; - else { - f_line = 0; - fold_state = CR; /* fold and output nothing */ - } - } - } + || ((c1==CR||(c1==LF&&f_prev!=CR)) + && fold_preserve_f)) { + /* new line */ + if (fold_preserve_f) { + f_prev = c1; + f_line = 0; + fold_state = CR; + } else if ((f_prev == c1) + || (f_prev == LF) + ) { /* duplicate newline */ + if (f_line) { + f_line = 0; + fold_state = LF; /* output two newline */ + } else { + f_line = 0; + fold_state = 1; + } + } else { + if (f_prev&0x80) { /* Japanese? */ + f_prev = c1; + fold_state = 0; /* ignore given single newline */ + } else if (f_prev==SP) { + fold_state = 0; + } else { + f_prev = c1; + if (++f_line<=fold_len) + fold_state = SP; + else { + f_line = 0; + fold_state = CR; /* fold and output nothing */ + } + } + } } else if (c1=='\f') { - 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=='!')) { - /* X0208 kankaku or ascii space */ - if (f_prev == SP) { - fold_state = 0; /* remove duplicate spaces */ - } else { - f_prev = SP; - if (++f_line<=fold_len) - fold_state = SP; /* output ASCII space only */ - else { - f_prev = SP; f_line = 0; - fold_state = CR; /* fold and output nothing */ - } - } + f_prev = LF; + f_line = 0; + fold_state = LF; /* output newline and clear */ + } else if ((c2==0 && nkf_isblank(c1)) || (c2 == '!' && c1 == '!')) { + /* X0208 kankaku or ascii space */ + if (f_prev == SP) { + fold_state = 0; /* remove duplicate spaces */ + } else { + f_prev = SP; + if (++f_line<=fold_len) + fold_state = SP; /* output ASCII space only */ + else { + f_prev = SP; f_line = 0; + fold_state = CR; /* fold and output nothing */ + } + } } else { - prev0 = f_prev; /* we still need this one... , but almost done */ - f_prev = c1; - if (c2 || c2 == JIS_X_0201_1976_K) - f_prev |= 0x80; /* this is Japanese */ - f_line += char_size(c2,c1); - if (f_line<=fold_len) { /* normal case */ - fold_state = 1; - } else { - if (f_line>fold_len+fold_margin) { /* too many kinsoku suspension */ - f_line = char_size(c2,c1); - fold_state = LF; /* We can't wait, do fold now */ - } else if (c2 == JIS_X_0201_1976_K) { - /* simple kinsoku rules return 1 means no folding */ - if (c1==(0xde&0x7f)) fold_state = 1; /* ゛*/ - else if (c1==(0xdf&0x7f)) fold_state = 1; /* ゜*/ - else if (c1==(0xa4&0x7f)) fold_state = 1; /* 。*/ - else if (c1==(0xa3&0x7f)) fold_state = 1; /* ，*/ - else if (c1==(0xa1&0x7f)) fold_state = 1; /* 」*/ - else if (c1==(0xb0&0x7f)) fold_state = 1; /* - */ - else if (SP<=c1 && c1<=(0xdf&0x7f)) { /* X0201 */ + prev0 = f_prev; /* we still need this one... , but almost done */ + f_prev = c1; + if (c2 || c2 == JIS_X_0201_1976_K) + f_prev |= 0x80; /* this is Japanese */ + 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+fold_margin) { /* too many kinsoku suspension */ + f_line = char_size(c2,c1); + fold_state = LF; /* We can't wait, do fold now */ + } else if (c2 == JIS_X_0201_1976_K) { + /* simple kinsoku rules return 1 means no folding */ + if (c1==(0xde&0x7f)) fold_state = 1; /* ゛*/ + else if (c1==(0xdf&0x7f)) fold_state = 1; /* ゜*/ + else if (c1==(0xa4&0x7f)) fold_state = 1; /* 。*/ + else if (c1==(0xa3&0x7f)) fold_state = 1; /* ，*/ + else if (c1==(0xa1&0x7f)) fold_state = 1; /* 」*/ + else if (c1==(0xb0&0x7f)) fold_state = 1; /* - */ + else if (SP<=c1 && c1<=(0xdf&0x7f)) { /* X0201 */ f_line = 1; fold_state = LF;/* add one new f_line before this character */ } else { f_line = 1; fold_state = LF;/* add one new f_line before this character */ } - } else if (c2==0) { - /* kinsoku point in ASCII */ + } else if (c2==0) { + /* kinsoku point in ASCII */ if ( c1==')'|| /* { [ ( */ - c1==']'|| - c1=='}'|| - c1=='.'|| - c1==','|| - c1=='!'|| - c1=='?'|| - c1=='/'|| - c1==':'|| - c1==';') { + c1==']'|| + c1=='}'|| + c1=='.'|| + c1==','|| + c1=='!'|| + c1=='?'|| + c1=='/'|| + c1==':'|| + c1==';') { fold_state = 1; - /* just after special */ + /* just after special */ } else if (!is_alnum(prev0)) { f_line = char_size(c2,c1); fold_state = LF; } else if ((prev0==SP) || /* ignored new f_line */ - (prev0==LF)|| /* ignored new f_line */ - (prev0&0x80)) { /* X0208 - ASCII */ + (prev0==LF)|| /* ignored new f_line */ + (prev0&0x80)) { /* X0208 - ASCII */ f_line = char_size(c2,c1); - fold_state = LF;/* add one new f_line before this character */ - } else { - fold_state = 1; /* default no fold in ASCII */ - } - } else { - if (c2=='!') { - if (c1=='"') fold_state = 1; /* 、 */ - else if (c1=='#') fold_state = 1; /* 。 */ - else if (c1=='W') fold_state = 1; /* 」 */ - else if (c1=='K') fold_state = 1; /* ） */ - else if (c1=='$') fold_state = 1; /* ， */ - else if (c1=='%') fold_state = 1; /* ． */ - else if (c1=='\'') fold_state = 1; /* ＋ */ - else if (c1=='(') fold_state = 1; /* ； */ - else if (c1==')') fold_state = 1; /* ？ */ - else if (c1=='*') fold_state = 1; /* ！ */ - else if (c1=='+') fold_state = 1; /* ゛ */ - else if (c1==',') fold_state = 1; /* ゜ */ - /* default no fold in kinsoku */ + fold_state = LF;/* add one new f_line before this character */ + } else { + fold_state = 1; /* default no fold in ASCII */ + } + } else { + if (c2=='!') { + if (c1=='"') fold_state = 1; /* 、 */ + else if (c1=='#') fold_state = 1; /* 。 */ + else if (c1=='W') fold_state = 1; /* 」 */ + else if (c1=='K') fold_state = 1; /* ） */ + else if (c1=='$') fold_state = 1; /* ， */ + else if (c1=='%') fold_state = 1; /* ． */ + else if (c1=='\'') fold_state = 1; /* ＋ */ + else if (c1=='(') fold_state = 1; /* ； */ + else if (c1==')') fold_state = 1; /* ？ */ + else if (c1=='*') fold_state = 1; /* ！ */ + else if (c1=='+') fold_state = 1; /* ゛ */ + else if (c1==',') fold_state = 1; /* ゜ */ + /* default no fold in kinsoku */ else { fold_state = LF; f_line = char_size(c2,c1); /* add one new f_line before this character */ } - } else { + } else { f_line = char_size(c2,c1); - fold_state = LF; - /* add one new f_line before this character */ - } - } - } + fold_state = LF; + /* add one new f_line before this character */ + } + } + } } /* terminator process */ switch(fold_state) { - case LF: - OCONV_NEWLINE((*o_fconv)); - (*o_fconv)(c2,c1); - break; - case 0: - return; - case CR: - OCONV_NEWLINE((*o_fconv)); - break; - case TAB: - case SP: - (*o_fconv)(0,SP); - break; - default: - (*o_fconv)(c2,c1); + case LF: + oconv_newline(o_fconv); + (*o_fconv)(c2,c1); + break; + case 0: + return; + case CR: + oconv_newline(o_fconv); + break; + case TAB: + case SP: + (*o_fconv)(0,SP); + break; + default: + (*o_fconv)(c2,c1); } } -nkf_char z_prev2=0,z_prev1=0; +static nkf_char z_prev2=0,z_prev1=0; -void z_conv(nkf_char c2, nkf_char c1) +static void +z_conv(nkf_char c2, nkf_char c1) { /* if (c2) c1 &= 0x7f; assertion */ @@ -3387,13 +4001,17 @@ void 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; + } else if (x0213_f && c1 == (0xdf&0x7f) && ev_x0213[(z_prev1-SP)*2]) { /* 半濁点 */ + 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) { - 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 濁点 or 半濁点 */ z_prev1 = c1; z_prev2 = c2; @@ -3406,33 +4024,33 @@ void z_conv(nkf_char c2, nkf_char c1) } if (c2 == EOF) { - (*o_zconv)(c2, c1); - return; + (*o_zconv)(c2, c1); + return; } if (alpha_f&1 && c2 == 0x23) { /* JISX0208 Alphabet */ - c2 = 0; + c2 = 0; } else if (c2 == 0x21) { /* JISX0208 Kigou */ - if (0x21==c1) { - if (alpha_f&2) { - c2 = 0; - c1 = SP; - } else if (alpha_f&4) { - (*o_zconv)(0, SP); - (*o_zconv)(0, SP); - return; - } - } else if (alpha_f&1 && 0x20': entity = ">"; break; case '<': entity = "<"; break; @@ -3448,7 +4066,7 @@ void z_conv(nkf_char c2, nkf_char c1) 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 */ @@ -3501,8 +4119,8 @@ void 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, - 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]; @@ -3512,6 +4130,10 @@ void z_conv(nkf_char c2, nkf_char c1) } return; } + } else if (c2 == 0 && nkf_char_unicode_p(c1) && + ((c1&VALUE_MASK) == 0x3099 || (c1&VALUE_MASK) == 0x309A)) { /* 合成用濁点・半濁点 */ + (*o_zconv)(JIS_X_0201_1976_K, 0x5E + (c1&VALUE_MASK) - 0x3099); + return; } } (*o_zconv)(c2,c1); @@ -3519,23 +4141,24 @@ void z_conv(nkf_char c2, nkf_char c1) #define rot13(c) ( \ - ( c < 'A') ? c: \ - (c <= 'M') ? (c + 13): \ - (c <= 'Z') ? (c - 13): \ - (c < 'a') ? (c): \ - (c <= 'm') ? (c + 13): \ - (c <= 'z') ? (c - 13): \ - (c) \ -) + ( c < 'A') ? c: \ + (c <= 'M') ? (c + 13): \ + (c <= 'Z') ? (c - 13): \ + (c < 'a') ? (c): \ + (c <= 'm') ? (c + 13): \ + (c <= 'z') ? (c - 13): \ + (c) \ + ) #define rot47(c) ( \ - ( c < '!') ? c: \ - ( c <= 'O') ? (c + 47) : \ - ( c <= '~') ? (c - 47) : \ - c \ -) - -void rot_conv(nkf_char c2, nkf_char c1) + ( c < '!') ? c: \ + ( c <= 'O') ? (c + 47) : \ + ( c <= '~') ? (c - 47) : \ + c \ + ) + +static void +rot_conv(nkf_char c2, nkf_char c1) { if (c2 == 0 || c2 == JIS_X_0201_1976_K || c2 == ISO_8859_1) { c1 = rot13(c1); @@ -3546,62 +4169,64 @@ void rot_conv(nkf_char c2, nkf_char c1) (*o_rot_conv)(c2,c1); } -void hira_conv(nkf_char c2, nkf_char c1) +static void +hira_conv(nkf_char c2, nkf_char c1) { if (hira_f & 1) { - if (c2 == 0x25) { - if (0x20 < c1 && c1 < 0x74) { - c2 = 0x24; - (*o_hira_conv)(c2,c1); - return; - } else if (c1 == 0x74 && nkf_enc_unicode_p(output_encoding)) { - c2 = 0; - c1 = nkf_char_unicode_new(0x3094); - (*o_hira_conv)(c2,c1); - return; - } - } else if (c2 == 0x21 && (c1 == 0x33 || c1 == 0x34)) { - c1 += 2; - (*o_hira_conv)(c2,c1); - return; - } + if (c2 == 0x25) { + if (0x20 < c1 && c1 < 0x74) { + c2 = 0x24; + (*o_hira_conv)(c2,c1); + return; + } else if (c1 == 0x74 && nkf_enc_unicode_p(output_encoding)) { + c2 = 0; + c1 = nkf_char_unicode_new(0x3094); + (*o_hira_conv)(c2,c1); + return; + } + } else if (c2 == 0x21 && (c1 == 0x33 || c1 == 0x34)) { + c1 += 2; + (*o_hira_conv)(c2,c1); + return; + } } if (hira_f & 2) { - if (c2 == 0 && c1 == nkf_char_unicode_new(0x3094)) { - c2 = 0x25; - c1 = 0x74; - } else if (c2 == 0x24 && 0x20 < c1 && c1 < 0x74) { - c2 = 0x25; - } else if (c2 == 0x21 && (c1 == 0x35 || c1 == 0x36)) { - c1 -= 2; - } + if (c2 == 0 && c1 == nkf_char_unicode_new(0x3094)) { + c2 = 0x25; + c1 = 0x74; + } else if (c2 == 0x24 && 0x20 < c1 && c1 < 0x74) { + c2 = 0x25; + } else if (c2 == 0x21 && (c1 == 0x35 || c1 == 0x36)) { + c1 -= 2; + } } (*o_hira_conv)(c2,c1); } -void iso2022jp_check_conv(nkf_char c2, nkf_char c1) +static void +iso2022jp_check_conv(nkf_char c2, nkf_char c1) { #define RANGE_NUM_MAX 18 static const nkf_char range[RANGE_NUM_MAX][2] = { - {0x222f, 0x2239,}, - {0x2242, 0x2249,}, - {0x2251, 0x225b,}, - {0x226b, 0x2271,}, - {0x227a, 0x227d,}, - {0x2321, 0x232f,}, - {0x233a, 0x2340,}, - {0x235b, 0x2360,}, - {0x237b, 0x237e,}, - {0x2474, 0x247e,}, - {0x2577, 0x257e,}, - {0x2639, 0x2640,}, - {0x2659, 0x267e,}, - {0x2742, 0x2750,}, - {0x2772, 0x277e,}, - {0x2841, 0x287e,}, - {0x4f54, 0x4f7e,}, - {0x7425, 0x747e}, + {0x222f, 0x2239,}, + {0x2242, 0x2249,}, + {0x2251, 0x225b,}, + {0x226b, 0x2271,}, + {0x227a, 0x227d,}, + {0x2321, 0x232f,}, + {0x233a, 0x2340,}, + {0x235b, 0x2360,}, + {0x237b, 0x237e,}, + {0x2474, 0x247e,}, + {0x2577, 0x257e,}, + {0x2639, 0x2640,}, + {0x2659, 0x267e,}, + {0x2742, 0x2750,}, + {0x2772, 0x277e,}, + {0x2841, 0x287e,}, + {0x4f54, 0x4f7e,}, + {0x7425, 0x747e}, }; nkf_char i; nkf_char start, end, c; @@ -3636,6 +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-2022-JP?B?", (const unsigned char *)"\075?ISO-2022-JP?Q?", #if defined(UTF8_INPUT_ENABLE) (const unsigned char *)"\075?UTF-8?B?", @@ -3648,7 +4274,7 @@ static const unsigned char *mime_pattern[] = { /* 該当するコードの優先度を上げるための目印 */ 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 @@ -3656,7 +4282,7 @@ nkf_char (*mime_priority_func[])(nkf_char c2, nkf_char c1, nkf_char c0) = { }; 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 @@ -3665,7 +4291,7 @@ static const nkf_char mime_encode[] = { }; 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 @@ -3689,18 +4315,21 @@ static nkf_char (*mime_iconv_back)(nkf_char c2,nkf_char c1,nkf_char c0) = NULL; #define MAXRECOVER 20 -static void mime_input_buf_unshift(nkf_char c) +static void +mime_input_buf_unshift(nkf_char c) { mime_input_buf(--mime_input_state.top) = (unsigned char)c; } -nkf_char mime_ungetc(nkf_char c, FILE *f) +static nkf_char +mime_ungetc(nkf_char c, ARG_UNUSED FILE *f) { mime_input_buf_unshift(c); return c; } -nkf_char mime_ungetc_buf(nkf_char c, FILE *f) +static nkf_char +mime_ungetc_buf(nkf_char c, FILE *f) { if (mimebuf_f) (*i_mungetc_buf)(c,f); @@ -3709,15 +4338,17 @@ nkf_char mime_ungetc_buf(nkf_char c, FILE *f) return c; } -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++)); + (*i_mgetc_buf)(f):mime_input_buf(mime_input_state.input++)); } -void switch_mime_getc(void) +static void +switch_mime_getc(void) { if (i_getc!=mime_getc) { i_mgetc = i_getc; i_getc = mime_getc; @@ -3729,7 +4360,8 @@ void switch_mime_getc(void) } } -void unswitch_mime_getc(void) +static void +unswitch_mime_getc(void) { if(mime_f==STRICT_MIME) { i_mgetc = i_mgetc_buf; @@ -3741,7 +4373,8 @@ void unswitch_mime_getc(void) mime_iconv_back = NULL; } -nkf_char mime_integrity(FILE *f, const unsigned char *p) +static nkf_char +mime_integrity(FILE *f, const unsigned char *p) { nkf_char c,d; unsigned int q; @@ -3754,22 +4387,22 @@ nkf_char mime_integrity(FILE *f, const unsigned char *p) d = 0; q = mime_input_state.input; while((c=(*i_getc)(f))!=EOF) { - if (((mime_input_state.input-mime_input_state.top)&MIME_BUF_MASK)==0) { + if (((mime_input_state.input-mime_input_state.top)&MIME_BUF_MASK)==0) { break; /* buffer full */ } - if (c=='=' && d=='?') { - /* checked. skip header, start decode */ - mime_input_buf(mime_input_state.input++) = (unsigned char)c; - /* mime_last_input = mime_input_state.input; */ - mime_input_state.input = q; + if (c=='=' && d=='?') { + /* checked. skip header, start decode */ + mime_input_buf(mime_input_state.input++) = (unsigned char)c; + /* mime_last_input = mime_input_state.input; */ + mime_input_state.input = q; switch_mime_getc(); - return 1; - } - if (!( (c=='+'||c=='/'|| c=='=' || c=='?' || is_alnum(c)))) - break; - /* Should we check length mod 4? */ - mime_input_buf(mime_input_state.input++) = (unsigned char)c; - d=c; + return 1; + } + if (!( (c=='+'||c=='/'|| c=='=' || c=='?' || is_alnum(c)))) + break; + /* Should we check length mod 4? */ + mime_input_buf(mime_input_state.input++) = (unsigned char)c; + d=c; } /* In case of Incomplete MIME, no MIME decode */ mime_input_buf(mime_input_state.input++) = (unsigned char)c; @@ -3779,7 +4412,8 @@ nkf_char mime_integrity(FILE *f, const unsigned char *p) return 1; } -nkf_char mime_begin_strict(FILE *f) +static nkf_char +mime_begin_strict(FILE *f) { nkf_char c1 = 0; int i,j,k; @@ -3793,24 +4427,24 @@ nkf_char mime_begin_strict(FILE *f) r[0]='='; r[1]='?'; for(i=2;p[i]>SP;i++) { /* start at =? */ - if (((r[i] = c1 = (*i_getc)(f))==EOF) || nkf_toupper(c1) != p[i]) { - /* pattern fails, try next one */ - q = p; - while (mime_pattern[++j]) { + if (((r[i] = c1 = (*i_getc)(f))==EOF) || nkf_toupper(c1) != p[i]) { + /* pattern fails, try next one */ + q = p; + while (mime_pattern[++j]) { p = mime_pattern[j]; - for(k=2;k i */ - if (p[k]!=q[k]) break; - if (k==i && nkf_toupper(c1)==p[k]) break; - } + for(k=2;k i */ + if (p[k]!=q[k]) break; + if (k==i && nkf_toupper(c1)==p[k]) break; + } p = mime_pattern[j]; - if (p) continue; /* found next one, continue */ - /* all fails, output from recovery buffer */ - (*i_ungetc)(c1,f); - for(j=0;jscore & (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"; @@ -3936,7 +4577,8 @@ static char* get_guessed_code(void) } #if !defined(PERL_XS) && !defined(WIN32DLL) -void print_guessed_code(char *filename) +static void +print_guessed_code(char *filename) { if (filename != NULL) printf("%s: ", filename); if (input_codename && !*input_codename) { @@ -3946,8 +4588,13 @@ void print_guessed_code(char *filename) if (guess_f == 1) { printf("%s\n", input_codename); } else { - printf("%s%s\n", + printf("%s%s%s%s\n", 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)" : @@ -3960,103 +4607,110 @@ void print_guessed_code(char *filename) #ifdef INPUT_OPTION -nkf_char hex_getc(nkf_char ch, FILE *f, nkf_char (*g)(FILE *f), nkf_char (*u)(nkf_char c, FILE *f)) +static nkf_char +hex_getc(nkf_char ch, FILE *f, nkf_char (*g)(FILE *f), nkf_char (*u)(nkf_char c, FILE *f)) { nkf_char c1, c2, c3; c1 = (*g)(f); if (c1 != ch){ - return c1; + return c1; } c2 = (*g)(f); if (!nkf_isxdigit(c2)){ - (*u)(c2, f); - return c1; + (*u)(c2, f); + return c1; } c3 = (*g)(f); if (!nkf_isxdigit(c3)){ - (*u)(c2, f); - (*u)(c3, f); - return c1; + (*u)(c2, f); + (*u)(c3, f); + return c1; } return (hex2bin(c2) << 4) | hex2bin(c3); } -nkf_char cap_getc(FILE *f) +static nkf_char +cap_getc(FILE *f) { return hex_getc(':', f, i_cgetc, i_cungetc); } -nkf_char cap_ungetc(nkf_char c, FILE *f) +static nkf_char +cap_ungetc(nkf_char c, FILE *f) { return (*i_cungetc)(c, f); } -nkf_char url_getc(FILE *f) +static nkf_char +url_getc(FILE *f) { return hex_getc('%', f, i_ugetc, i_uungetc); } -nkf_char url_ungetc(nkf_char c, FILE *f) +static nkf_char +url_ungetc(nkf_char c, FILE *f) { return (*i_uungetc)(c, f); } #endif #ifdef NUMCHAR_OPTION -nkf_char numchar_getc(FILE *f) +static nkf_char +numchar_getc(FILE *f) { nkf_char (*g)(FILE *) = i_ngetc; nkf_char (*u)(nkf_char c ,FILE *f) = i_nungetc; int i = 0, j; - nkf_char buf[8]; - long c = -1; + nkf_char buf[12]; + nkf_char c = -1; buf[i] = (*g)(f); if (buf[i] == '&'){ - buf[++i] = (*g)(f); - if (buf[i] == '#'){ - c = 0; - buf[++i] = (*g)(f); - if (buf[i] == 'x' || buf[i] == 'X'){ - for (j = 0; j < 7; j++){ - buf[++i] = (*g)(f); - if (!nkf_isxdigit(buf[i])){ - if (buf[i] != ';'){ - c = -1; - } - break; - } - c <<= 4; - c |= hex2bin(buf[i]); - } - }else{ - for (j = 0; j < 8; j++){ - if (j){ - buf[++i] = (*g)(f); - } - if (!nkf_isdigit(buf[i])){ - if (buf[i] != ';'){ - c = -1; - } - break; - } - c *= 10; - c += hex2bin(buf[i]); - } - } - } + buf[++i] = (*g)(f); + if (buf[i] == '#'){ + c = 0; + buf[++i] = (*g)(f); + if (buf[i] == 'x' || buf[i] == 'X'){ + for (j = 0; j < 7; j++){ + buf[++i] = (*g)(f); + if (!nkf_isxdigit(buf[i])){ + if (buf[i] != ';'){ + c = -1; + } + break; + } + c <<= 4; + c |= hex2bin(buf[i]); + } + }else{ + for (j = 0; j < 8; j++){ + if (j){ + buf[++i] = (*g)(f); + } + if (!nkf_isdigit(buf[i])){ + if (buf[i] != ';'){ + c = -1; + } + break; + } + c *= 10; + c += hex2bin(buf[i]); + } + } + } } if (c != -1){ - return nkf_char_unicode_new(c); + return nkf_char_unicode_new(c); } while (i > 0){ - (*u)(buf[i], f); - --i; + (*u)(buf[i], f); + --i; } return buf[0]; } -nkf_char numchar_ungetc(nkf_char c, FILE *f) +static nkf_char +numchar_ungetc(nkf_char c, FILE *f) { return (*i_nungetc)(c, f); } @@ -4064,72 +4718,89 @@ nkf_char numchar_ungetc(nkf_char c, FILE *f) #ifdef UNICODE_NORMALIZATION -/* Normalization Form C */ -nkf_char nfc_getc(FILE *f) +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; + 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; - } 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++) - buf[i] = (nkf_char)(array[i]); - i--; + nkf_buf_push(buf, array[i]); 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; } -nkf_char nfc_ungetc(nkf_char c, FILE *f) +static nkf_char +nfc_ungetc(nkf_char c, FILE *f) { return (*i_nfc_ungetc)(c, f); } #endif /* UNICODE_NORMALIZATION */ -static nkf_char base64decode(nkf_char c) +static nkf_char +base64decode(nkf_char c) { int i; if (c > '@') { - if (c < '[') { - i = c - 'A'; /* A..Z 0-25 */ + if (c < '[') { + i = c - 'A'; /* A..Z 0-25 */ } else if (c == '_') { i = '?' /* 63 */ ; /* _ 63 */ - } else { - i = c - 'G' /* - 'a' + 26 */ ; /* a..z 26-51 */ + } else { + i = c - 'G' /* - 'a' + 26 */ ; /* a..z 26-51 */ } } else if (c > '/') { - i = c - '0' + '4' /* - '0' + 52 */ ; /* 0..9 52-61 */ + i = c - '0' + '4' /* - '0' + 52 */ ; /* 0..9 52-61 */ } else if (c == '+' || c == '-') { - i = '>' /* 62 */ ; /* + and - 62 */ + i = '>' /* 62 */ ; /* + and - 62 */ } else { - i = '?' /* 63 */ ; /* / 63 */ + i = '?' /* 63 */ ; /* / 63 */ } return (i); } -nkf_char +static nkf_char mime_getc(FILE *f) { nkf_char c1, c2, c3, c4, cc; @@ -4140,7 +4811,7 @@ mime_getc(FILE *f) nkf_char lwsp_size = 128; if (mime_input_state.top != mime_input_state.last) { /* Something is in FIFO */ - return mime_input_buf(mime_input_state.top++); + return mime_input_buf(mime_input_state.top++); } if (mime_decode_mode==1 ||mime_decode_mode==FALSE) { mime_decode_mode=FALSE; @@ -4149,38 +4820,34 @@ mime_getc(FILE *f) } if (mimebuf_f == FIXED_MIME) - exit_mode = mime_decode_mode; + exit_mode = mime_decode_mode; else - exit_mode = FALSE; + exit_mode = FALSE; if (mime_decode_mode == 'Q') { - if ((c1 = (*i_mgetc)(f)) == EOF) return (EOF); -restart_mime_q: - if (c1=='_' && mimebuf_f != FIXED_MIME) return SP; + if ((c1 = (*i_mgetc)(f)) == EOF) return (EOF); + restart_mime_q: + if (c1=='_' && mimebuf_f != FIXED_MIME) return SP; if (c1<=SP || DEL<=c1) { mime_decode_mode = exit_mode; /* prepare for quit */ return c1; } - if (c1!='=' && (c1!='?' || mimebuf_f == FIXED_MIME)) { + if (c1!='=' && (c1!='?' || mimebuf_f == FIXED_MIME)) { return c1; } - mime_decode_mode = exit_mode; /* prepare for quit */ - if ((c2 = (*i_mgetc)(f)) == EOF) return (EOF); - if (c1=='?'&&c2=='=' && mimebuf_f != FIXED_MIME) { - /* end Q encoding */ - input_mode = exit_mode; + mime_decode_mode = exit_mode; /* prepare for quit */ + if ((c2 = (*i_mgetc)(f)) == EOF) return (EOF); + if (c1=='?'&&c2=='=' && mimebuf_f != FIXED_MIME) { + /* 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) { - 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 { @@ -4189,7 +4856,7 @@ restart_mime_q: 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 { @@ -4207,12 +4874,7 @@ restart_mime_q: 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; @@ -4225,70 +4887,66 @@ restart_mime_q: i_ungetc(lwsp_buf[lwsp_count],f); c1 = lwsp_buf[0]; } - free(lwsp_buf); - return c1; - } - if (c1=='='&&c2lwsp_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; @@ -4336,24 +4989,24 @@ mime_c2_retry: i_ungetc(lwsp_buf[lwsp_count],f); c1 = lwsp_buf[0]; } - free(lwsp_buf); - return c1; + nkf_xfree(lwsp_buf); + return c1; } -mime_c3_retry: + mime_c3_retry: if ((c3 = (*i_mgetc)(f))<=SP) { - if (c3==EOF) - return (EOF); + if (c3==EOF) + return (EOF); if (mime_f != STRICT_MIME) goto mime_c3_retry; - if (mimebuf_f!=FIXED_MIME) input_mode = ASCII; - return c3; + if (mimebuf_f!=FIXED_MIME) input_mode = ASCII; + return c3; } -mime_c4_retry: + mime_c4_retry: if ((c4 = (*i_mgetc)(f))<=SP) { - if (c4==EOF) - return (EOF); + if (c4==EOF) + return (EOF); if (mime_f != STRICT_MIME) goto mime_c4_retry; - if (mimebuf_f!=FIXED_MIME) input_mode = ASCII; - return c4; + if (mimebuf_f!=FIXED_MIME) input_mode = ASCII; + return c4; } mime_decode_mode = mode; /* still in MIME sigh... */ @@ -4366,33 +5019,33 @@ mime_c4_retry: t4 = 0x3f & base64decode(c4); cc = ((t1 << 2) & 0x0fc) | ((t2 >> 4) & 0x03); if (c2 != '=') { - mime_input_buf(mime_input_state.last++) = (unsigned char)cc; - cc = ((t2 << 4) & 0x0f0) | ((t3 >> 2) & 0x0f); - if (c3 != '=') { - mime_input_buf(mime_input_state.last++) = (unsigned char)cc; - cc = ((t3 << 6) & 0x0c0) | (t4 & 0x3f); - if (c4 != '=') - mime_input_buf(mime_input_state.last++) = (unsigned char)cc; - } + mime_input_buf(mime_input_state.last++) = (unsigned char)cc; + cc = ((t2 << 4) & 0x0f0) | ((t3 >> 2) & 0x0f); + if (c3 != '=') { + mime_input_buf(mime_input_state.last++) = (unsigned char)cc; + cc = ((t3 << 6) & 0x0c0) | (t4 & 0x3f); + if (c4 != '=') + mime_input_buf(mime_input_state.last++) = (unsigned char)cc; + } } else { - return c1; + return c1; } return mime_input_buf(mime_input_state.top++); } static const char basis_64[] = - "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; + "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; -#define MIMEOUT_BUF_LENGTH (60) +#define MIMEOUT_BUF_LENGTH 74 static struct { - char buf[MIMEOUT_BUF_LENGTH+1]; + unsigned char buf[MIMEOUT_BUF_LENGTH+1]; int count; - nkf_char state; } mimeout_state; /*nkf_char mime_lastchar2, mime_lastchar1;*/ -static void open_mime(nkf_char mode) +static void +open_mime(nkf_char mode) { const unsigned char *p; int i; @@ -4408,21 +5061,18 @@ static void open_mime(nkf_char mode) i = 0; if (base64_count>45) { if (mimeout_state.count>0 && nkf_isblank(mimeout_state.buf[i])){ - (*o_mputc)(mimeout_state.buf[i]); + (*o_mputc)(mimeout_state.buf[i]); i++; } - PUT_NEWLINE((*o_mputc)); + put_newline(o_mputc); (*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 0){ - if (c2 == EOF){ - if (base64_count + mimeout_state.count/3*4> 73){ - (*o_base64conv)(EOF,0); - OCONV_NEWLINE((*o_base64conv)); - (*o_base64conv)(0,SP); - base64_count = 1; - } - } else { - if (base64_count + mimeout_state.count/3*4> 66) { - (*o_base64conv)(EOF,0); - OCONV_NEWLINE((*o_base64conv)); - (*o_base64conv)(0,SP); - base64_count = 1; - mimeout_mode = -1; - } - } + if (c2 == EOF){ + if (base64_count + mimeout_state.count/3*4> 73){ + (*o_base64conv)(EOF,0); + oconv_newline(o_base64conv); + (*o_base64conv)(0,SP); + base64_count = 1; + } + } else { + if ((c2 != 0 || c1 > DEL) && base64_count + mimeout_state.count/3*4> 66) { + (*o_base64conv)(EOF,0); + oconv_newline(o_base64conv); + (*o_base64conv)(0,SP); + base64_count = 1; + mimeout_mode = -1; + } + } } else if (c2) { if (c2 != EOF && base64_count + mimeout_state.count/3*4> 60) { 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; @@ -4472,7 +5123,8 @@ static void mime_prechar(nkf_char c2, nkf_char c1) } } -static void close_mime(void) +static void +close_mime(void) { (*o_mputc)('?'); (*o_mputc)('='); @@ -4480,20 +5132,21 @@ static void close_mime(void) mimeout_mode = 0; } -static void eof_mime(void) +static void +eof_mime(void) { switch(mimeout_mode) { case 'Q': 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)(basis_64[((mimeout_state.state & 0xF) << 2)]); + (*o_mputc)(basis_64[((nkf_state->mimeout_state & 0xF) << 2)]); (*o_mputc)('='); base64_count += 2; break; @@ -4506,7 +5159,8 @@ static void eof_mime(void) } } -static void mimeout_addchar(nkf_char c) +static void +mimeout_addchar(nkf_char c) { switch(mimeout_mode) { case 'Q': @@ -4522,60 +5176,61 @@ static void mimeout_addchar(nkf_char c) (*o_mputc)(c); base64_count++; } - break; + break; case 'B': - mimeout_state.state=c; - (*o_mputc)(basis_64[c>>2]); - mimeout_mode=2; - base64_count ++; - break; + nkf_state->mimeout_state=c; + (*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; - mimeout_mode=1; - base64_count ++; - break; + (*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: - (*o_mputc)(basis_64[((mimeout_state.state & 0xF) << 2) | ((c & 0xC0) >>6)]); - (*o_mputc)(basis_64[c & 0x3F]); - mimeout_mode='B'; - base64_count += 2; - break; + (*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; + break; default: (*o_mputc)(c); base64_count++; - break; + break; } } -static void mime_putc(nkf_char c) +static void +mime_putc(nkf_char c) { int i, j; nkf_char lastchar; if (mimeout_f == FIXED_MIME){ - if (mimeout_mode == 'Q'){ - if (base64_count > 71){ - if (c!=CR && c!=LF) { - (*o_mputc)('='); - PUT_NEWLINE((*o_mputc)); - } - base64_count = 0; - } - }else{ - if (base64_count > 71){ - eof_mime(); - PUT_NEWLINE((*o_mputc)); - base64_count = 0; - } - if (c == EOF) { /* c==EOF */ - eof_mime(); - } - } - if (c != EOF) { /* c==EOF */ - mimeout_addchar(c); - } - return; + if (mimeout_mode == 'Q'){ + if (base64_count > 71){ + if (c!=CR && c!=LF) { + (*o_mputc)('='); + put_newline(o_mputc); + } + base64_count = 0; + } + }else{ + if (base64_count > 71){ + eof_mime(); + put_newline(o_mputc); + base64_count = 0; + } + if (c == EOF) { /* c==EOF */ + eof_mime(); + } + } + if (c != EOF) { /* c==EOF */ + mimeout_addchar(c); + } + return; } /* mimeout_f != FIXED_MIME */ @@ -4608,36 +5263,36 @@ static void mime_putc(nkf_char c) mimeout_addchar(mimeout_state.buf[i]); } } - return; + return; } if (mimeout_state.count > 0){ - lastchar = mimeout_state.buf[mimeout_state.count - 1]; + lastchar = mimeout_state.buf[mimeout_state.count - 1]; }else{ - lastchar = -1; + lastchar = -1; } if (mimeout_mode=='Q') { - if (c <= DEL && (output_mode==ASCII ||output_mode == ISO_8859_1)) { + if (c <= DEL && (output_mode==ASCII ||output_mode == ISO_8859_1)) { if (c == CR || c == LF) { close_mime(); (*o_mputc)(c); base64_count = 0; return; - } else if (c <= SP) { - close_mime(); + } 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)) { (*o_mputc)(SP); base64_count++; } - } else { + } 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); @@ -4647,131 +5302,160 @@ static void mime_putc(nkf_char c) return; } } - (*o_mputc)(c); - base64_count++; - } - return; + if (c != 0x1B) { + (*o_mputc)(c); + base64_count++; + return; + } + } } if (mimeout_mode <= 0) { - if (c <= DEL && (output_mode==ASCII ||output_mode == ISO_8859_1)) { - if (nkf_isspace(c)) { + 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) { flag = 1; } - if (c==CR || c==LF) { + if (c==CR || c==LF) { if (flag) { open_mime(output_mode); output_mode = 0; } else { base64_count = 0; } - } - for (i=0;i 1 - && base64_count + mimeout_state.count > 76 + }else{ + 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++; - } - } - mimeout_state.buf[mimeout_state.count++] = (char)c; - if (mimeout_state.count>MIMEOUT_BUF_LENGTH) { - open_mime(output_mode); - } - } - return; - }else{ - if (lastchar==CR || lastchar == LF){ - for (i=0;iMIMEOUT_BUF_LENGTH) { + open_mime(output_mode); + } + } + return; + }else{ + if (lastchar==CR || lastchar == LF){ + for (i=0;iMIMEOUT_BUF_LENGTH) { + eof_mime(); + for (j=0;j0 && SPMIMEOUT_BUF_LENGTH) { + j = mimeout_state.count; + mimeout_state.count = 0; + for (i=0;iMIMEOUT_BUF_LENGTH) { - eof_mime(); - for (i=0;i0 && SPMIMEOUT_BUF_LENGTH) { - j = mimeout_state.count; - mimeout_state.count = 0; - for (i=0;i0) { j = mimeout_state.count; @@ -4793,7 +5477,8 @@ static void mime_putc(nkf_char c) mimeout_addchar(c); } -void base64_conv(nkf_char c2, nkf_char c1) +static void +base64_conv(nkf_char c2, nkf_char c1) { mime_prechar(c2, c1); (*o_base64conv)(c2,c1); @@ -4806,36 +5491,32 @@ typedef struct nkf_iconv_t { size_t input_buffer_size; char *output_buffer; size_t output_buffer_size; -} +}; -nkf_iconv_t nkf_iconv_new(char *tocode, char *fromcode) +static nkf_iconv_t +nkf_iconv_new(char *tocode, char *fromcode) { 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 = 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) { - switch (errno) { - case EINVAL: - perror(fprintf("iconv doesn't support %s to %s conversion.", fromcode, tocode)); - return -1; - default: - perror("can't iconv_open"); - } + switch (errno) { + case EINVAL: + perror(fprintf("iconv doesn't support %s to %s conversion.", fromcode, tocode)); + return -1; + default: + perror("can't iconv_open"); + } } } -size_t nkf_iconv_convert(nkf_iconv_t *converter, FILE *input) +static size_t +nkf_iconv_convert(nkf_iconv_t *converter, FILE *input) { size_t invalid = (size_t)0; char *input_buffer = converter->input_buffer; @@ -4845,61 +5526,63 @@ size_t nkf_iconv_convert(nkf_iconv_t *converter, FILE *input) int c; do { - if (c != EOF) { - while ((c = (*i_getc)(f)) != EOF) { - input_buffer[input_length++] = c; - if (input_length < converter->input_buffer_size) break; - } - } - - size_t ret = iconv(converter->cd, &input_buffer, &input_length, &output_buffer, &output_length); - while (output_length-- > 0) { - (*o_putc)(output_buffer[converter->output_buffer_size-output_length]); - } - if (ret == (size_t) - 1) { - switch (errno) { - case EINVAL: - if (input_buffer != converter->input_buffer) - memmove(converter->input_buffer, input_buffer, input_length); - break; - case E2BIG: - converter->output_buffer_size *= 2; - output_buffer = realloc(converter->outbuf, converter->output_buffer_size); - if (output_buffer == NULL) { - perror("can't realloc"); - return -1; - } - converter->output_buffer = output_buffer; - break; - default: - perror("can't iconv"); - return -1; - } - } else { - invalid += ret; - } + if (c != EOF) { + while ((c = (*i_getc)(f)) != EOF) { + input_buffer[input_length++] = c; + if (input_length < converter->input_buffer_size) break; + } + } + + size_t ret = iconv(converter->cd, &input_buffer, &input_length, &output_buffer, &output_length); + while (output_length-- > 0) { + (*o_putc)(output_buffer[converter->output_buffer_size-output_length]); + } + if (ret == (size_t) - 1) { + switch (errno) { + case EINVAL: + if (input_buffer != converter->input_buffer) + memmove(converter->input_buffer, input_buffer, input_length); + break; + case E2BIG: + converter->output_buffer_size *= 2; + output_buffer = realloc(converter->outbuf, converter->output_buffer_size); + if (output_buffer == NULL) { + perror("can't realloc"); + return -1; + } + converter->output_buffer = output_buffer; + break; + default: + perror("can't iconv"); + return -1; + } + } else { + invalid += ret; + } } while (1); return invalid; } -void nkf_iconv_close(nkf_iconv_t *convert) +static void +nkf_iconv_close(nkf_iconv_t *convert) { - free(converter->inbuf); - free(converter->outbuf); - iconv_close(converter->cd); + nkf_xfree(converter->inbuf); + nkf_xfree(converter->outbuf); + iconv_close(converter->cd); } #endif -void reinit(void) +static void +reinit(void) { { - struct input_code *p = input_code_list; - while (p->name){ - status_reinit(p++); - } + struct input_code *p = input_code_list; + while (p->name){ + status_reinit(p++); + } } unbuf_f = FALSE; estab_f = FALSE; @@ -4914,7 +5597,7 @@ void reinit(void) 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; @@ -4955,10 +5638,10 @@ void reinit(void) x0213_f = FALSE; #endif { - int i; - for (i = 0; i < 256; i++){ - prefix_table[i] = 0; - } + int i; + for (i = 0; i < 256; i++){ + prefix_table[i] = 0; + } } hold_count = 0; mimeout_state.count = 0; @@ -4997,7 +5680,6 @@ void reinit(void) 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; @@ -5005,12 +5687,14 @@ void reinit(void) input_codename = NULL; input_encoding = NULL; output_encoding = NULL; + nkf_state_init(); #ifdef WIN32DLL reinitdll(); #endif /*WIN32DLL*/ } -int module_connection(void) +static int +module_connection(void) { if (input_encoding) set_input_encoding(input_encoding); if (!output_encoding) { @@ -5023,13 +5707,19 @@ int module_connection(void) 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; - /* replace continucation module, from output side */ + if (x0201_f == NKF_UNSPECIFIED) { + x0201_f = X0201_DEFAULT; + } + + /* replace continuation module, from output side */ - /* output redicrection */ + /* output redirection */ #ifdef CHECK_OPTION if (noout_f || guess_f){ - o_putc = no_putc; + o_putc = no_putc; } #endif if (mimeout_f) { @@ -5063,27 +5753,27 @@ int module_connection(void) 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; - i_cungetc = i_ungetc; i_ungetc= cap_ungetc; + i_cgetc = i_getc; i_getc = cap_getc; + i_cungetc = i_ungetc; i_ungetc= cap_ungetc; } if (url_f){ - i_ugetc = i_getc; i_getc = url_getc; - i_uungetc = i_ungetc; i_ungetc= url_ungetc; + i_ugetc = i_getc; i_getc = url_getc; + i_uungetc = i_ungetc; i_ungetc= url_ungetc; } #endif #ifdef NUMCHAR_OPTION if (numchar_f){ - i_ngetc = i_getc; i_getc = numchar_getc; - i_nungetc = i_ungetc; i_ungetc= numchar_ungetc; + i_ngetc = i_getc; i_getc = numchar_getc; + i_nungetc = i_ungetc; i_ungetc= numchar_ungetc; } #endif #ifdef UNICODE_NORMALIZATION if (nfc_f){ - i_nfc_getc = i_getc; i_getc = nfc_getc; - i_nfc_ungetc = i_ungetc; i_ungetc= nfc_ungetc; + i_nfc_getc = i_getc; i_getc = nfc_getc; + i_nfc_ungetc = i_ungetc; i_ungetc= nfc_ungetc; } #endif if (mime_f && mimebuf_f==FIXED_MIME) { @@ -5095,16 +5785,16 @@ int module_connection(void) i_bungetc = i_ungetc; i_ungetc = broken_ungetc; } if (input_encoding) { - set_iconv(-TRUE, nkf_enc_to_iconv(input_encoding)); + set_iconv(-TRUE, nkf_enc_to_iconv(input_encoding)); } else { - set_iconv(FALSE, e_iconv); + set_iconv(FALSE, e_iconv); } { - struct input_code *p = input_code_list; - while (p->name){ - status_reinit(p++); - } + struct input_code *p = input_code_list; + while (p->name){ + status_reinit(p++); + } } return 0; } @@ -5114,23 +5804,38 @@ int module_connection(void) */ #if !defined(PERL_XS) && !defined(WIN32DLL) -nkf_char noconvert(FILE *f) +static nkf_char +noconvert(FILE *f) { nkf_char c; if (nop_f == 2) module_connection(); while ((c = (*i_getc)(f)) != EOF) - (*o_putc)(c); + (*o_putc)(c); (*o_putc)(EOF); return 1; } #endif -int kanji_convert(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 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; \ + shift_mode = 0; \ + set_input_codename("ISO-2022-JP"); \ + debug("ISO-2022-JP"); \ +} while (0) + +static int +kanji_convert(FILE *f) { nkf_char c1=0, c2=0, c3=0, c4=0; - int shift_mode = FALSE; /* TRUE or FALSE or JIS_X_0201_1976_K */ + int shift_mode = 0; /* 0, 1, 2, 3 */ + int g2 = 0; int is_8bit = FALSE; if (input_encoding && !nkf_enc_asciicompat(input_encoding)) { @@ -5140,12 +5845,6 @@ int kanji_convert(FILE *f) input_mode = ASCII; output_mode = ASCII; -#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 LAST break /* end of loop, go closing */ - if (module_connection() < 0) { #if !defined(PERL_XS) && !defined(WIN32DLL) fprintf(stderr, "no output encoding given\n"); @@ -5160,20 +5859,48 @@ int kanji_convert(FILE *f) (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) { - if (nkf_iconv_utf_16(c1, c2, 0, 0) == -2 && - (c3 = (*i_getc)(f)) != EOF && - (c4 = (*i_getc)(f)) != EOF) { + 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); + } 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 @@ -5182,16 +5909,19 @@ int kanji_convert(FILE *f) if (!input_encoding) #endif code_status(c1); - if (c2) { - /* second byte */ - if (c2 > DEL) { - /* in case of 8th bit is on */ - if (!estab_f&&!mime_decode_mode) { - /* in case of not established yet */ - /* It is still ambiguious */ - if (h_conv(f, c2, c1)==EOF) - LAST; - SKIP; + if (c2) { + /* second byte */ + 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 */ + /* It is still ambiguous */ + if (h_conv(f, c2, c1)==EOF) { + LAST; + } + else { + SKIP; + } } else { /* in case of already established */ @@ -5203,26 +5933,38 @@ int kanji_convert(FILE *f) } } } - /* 2nd byte of 7 bit code or SJIS */ - SEND; + else { + /* 2nd byte of 7 bit code or SJIS */ + SEND; + } + } + else if (nkf_char_unicode_p(c1)) { + (*oconv)(0, c1); + NEXT; } else { /* first byte */ 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) { /* not established yet */ MORE; - } else { /* estab_f==TRUE */ - if (iso8859_f) { - c2 = ISO_8859_1; - c1 &= 0x7f; - SEND; - } - else if ((iconv == s_iconv && 0xA0 <= c1 && c1 <= 0xDF) || + } else { /* estab_f==TRUE */ + if (iso8859_f) { + c2 = ISO_8859_1; + c1 &= 0x7f; + SEND; + } + else if ((iconv == s_iconv && 0xA0 <= c1 && c1 <= 0xDF) || (ms_ucs_map_f == UCS_MAP_CP10001 && (c1 == 0xFD || c1 == 0xFE))) { /* JIS X 0201 */ c2 = JIS_X_0201_1976_K; @@ -5241,25 +5983,25 @@ int kanji_convert(FILE *f) if (iso8859_f) { c2 = ISO_8859_1; SEND; - } else if (SP <= c1 && c1 < (0xE0&0x7F)){ + } else if (nkf_byte_jisx0201_katakana_p(c1)){ /* output 1 shifted byte */ c2 = JIS_X_0201_1976_K; SEND; - } else { - /* look like bogus code */ - SKIP; - } - } else if (input_mode == JIS_X_0208 || input_mode == JIS_X_0212 || + } else { + /* look like bogus code */ + SKIP; + } + } else if (input_mode == JIS_X_0208 || input_mode == JIS_X_0212 || input_mode == JIS_X_0213_1 || input_mode == JIS_X_0213_2) { - /* in case of Kanji shifted */ - MORE; - } else if (c1 == '=' && mime_f && !mime_decode_mode) { - /* Check MIME code */ - if ((c1 = (*i_getc)(f)) == EOF) { - (*oconv)(0, '='); - LAST; - } else if (c1 == '?') { - /* =? is mime conversion start sequence */ + /* in case of Kanji shifted */ + MORE; + } else if (c1 == '=' && mime_f && !mime_decode_mode) { + /* Check MIME code */ + if ((c1 = (*i_getc)(f)) == EOF) { + (*oconv)(0, '='); + LAST; + } else if (c1 == '?') { + /* =? is mime conversion start sequence */ if(mime_f == STRICT_MIME) { /* check in real detail */ if (mime_begin_strict(f) == EOF) @@ -5278,31 +6020,37 @@ int kanji_convert(FILE *f) SEND; } } else if (c1 == SI && (!is_8bit || mime_decode_mode)) { - shift_mode = FALSE; + shift_mode = 0; SKIP; } else if (c1 == SO && (!is_8bit || mime_decode_mode)) { - shift_mode = TRUE; + shift_mode = 1; SKIP; - } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) { - if ((c1 = (*i_getc)(f)) == EOF) { - /* (*oconv)(0, ESC); don't send bogus code */ - LAST; - } else if (c1 == '$') { - if ((c1 = (*i_getc)(f)) == EOF) { - /* - (*oconv)(0, ESC); don't send bogus code - (*oconv)(0, '$'); */ - LAST; - } else if (c1 == '@'|| c1 == 'B') { - /* This is kanji introduction */ - input_mode = JIS_X_0208; - shift_mode = FALSE; - set_input_codename("ISO-2022-JP"); -#ifdef CHECK_OPTION - debug("ISO-2022-JP"); -#endif + } else if (c1 == ESC && (!is_8bit || mime_decode_mode)) { + if ((c1 = (*i_getc)(f)) == EOF) { + (*oconv)(0, ESC); + LAST; + } + else if (c1 == '&') { + /* IRR */ + if ((c1 = (*i_getc)(f)) == EOF) { + LAST; + } else { + SKIP; + } + } + else if (c1 == '$') { + /* GZDMx */ + if ((c1 = (*i_getc)(f)) == EOF) { + /* don't send bogus code + (*oconv)(0, ESC); + (*oconv)(0, '$'); */ + LAST; + } else if (c1 == '@' || c1 == 'B') { + /* JIS X 0208 */ + set_input_mode(JIS_X_0208); SKIP; } else if (c1 == '(') { + /* GZDM4 */ if ((c1 = (*i_getc)(f)) == EOF) { /* don't send bogus code (*oconv)(0, ESC); @@ -5311,23 +6059,19 @@ int kanji_convert(FILE *f) */ LAST; } else if (c1 == '@'|| c1 == 'B') { - /* This is kanji introduction */ - input_mode = JIS_X_0208; - shift_mode = FALSE; + /* JIS X 0208 */ + set_input_mode(JIS_X_0208); SKIP; #ifdef X0212_ENABLE } else if (c1 == 'D'){ - input_mode = JIS_X_0212; - shift_mode = FALSE; + set_input_mode(JIS_X_0212); SKIP; #endif /* X0212_ENABLE */ } else if (c1 == 'O' || c1 == 'Q'){ - input_mode = JIS_X_0213_1; - shift_mode = FALSE; + set_input_mode(JIS_X_0213_1); SKIP; } else if (c1 == 'P'){ - input_mode = JIS_X_0213_2; - shift_mode = FALSE; + set_input_mode(JIS_X_0213_2); SKIP; } else { /* could be some special code */ @@ -5340,7 +6084,7 @@ int kanji_convert(FILE *f) } else if (broken_f&0x2) { /* accept any ESC-(-x as broken code ... */ input_mode = JIS_X_0208; - shift_mode = FALSE; + shift_mode = 0; SKIP; } else { (*oconv)(0, ESC); @@ -5349,85 +6093,108 @@ int kanji_convert(FILE *f) SKIP; } } else if (c1 == '(') { + /* GZD4 */ if ((c1 = (*i_getc)(f)) == EOF) { /* don't send bogus code (*oconv)(0, ESC); (*oconv)(0, '('); */ LAST; - } else { - if (c1 == 'I') { - /* This is X0201 kana introduction */ - input_mode = JIS_X_0201_1976_K; shift_mode = JIS_X_0201_1976_K; - SKIP; - } else if (c1 == 'B' || c1 == 'J' || c1 == 'H') { - /* This is X0208 kanji introduction */ - input_mode = ASCII; shift_mode = FALSE; - SKIP; - } else if (broken_f&0x2) { - input_mode = ASCII; shift_mode = FALSE; - SKIP; - } else { - (*oconv)(0, ESC); - (*oconv)(0, '('); - /* maintain various input_mode here */ - SEND; - } - } - } else if ( c1 == 'N' || c1 == 'n'){ - /* SS2 */ - c4 = (*i_getc)(f); /* skip SS2 */ - if ( (SP<=c4 && c4 < 0x60) || (0xa0<=c4 && c4 < 0xe0)){ - c1 = c4; - c2 = JIS_X_0201_1976_K; - SEND; - }else{ - (*i_ungetc)(c4, f); - /* lonely ESC */ - (*oconv)(0, ESC); - SEND; - } - } else { - /* lonely ESC */ - (*oconv)(0, ESC); - SEND; - } + } + 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') { + /* ISO-646IRV:1983 or JIS X 0201 Roman or JUNET */ + set_input_mode(ASCII); + SKIP; + } + else if (broken_f&0x2) { + set_input_mode(ASCII); + SKIP; + } + else { + (*oconv)(0, ESC); + (*oconv)(0, '('); + SEND; + } + } + else if (c1 == '.') { + /* G2D6 */ + if ((c1 = (*i_getc)(f)) == EOF) { + LAST; + } + else if (c1 == 'A') { + /* ISO-8859-1 */ + g2 = ISO_8859_1; + SKIP; + } + else { + (*oconv)(0, ESC); + (*oconv)(0, '.'); + SEND; + } + } + else if (c1 == 'N') { + /* SS2 */ + c1 = (*i_getc)(f); + if (g2 == ISO_8859_1) { + c2 = ISO_8859_1; + SEND; + }else{ + (*i_ungetc)(c1, f); + /* lonely ESC */ + (*oconv)(0, ESC); + SEND; + } + } + else { + i_ungetc(c1,f); + /* lonely ESC */ + (*oconv)(0, ESC); + SKIP; + } } 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 == '$') { /* J-PHONE emoji */ if ((c1 = (*i_getc)(f)) == EOF) { - /* - (*oconv)(0, ESC); don't send bogus code - (*oconv)(0, '$'); */ LAST; - } else { - if (('E' <= c1 && c1 <= 'G') || - ('O' <= c1 && c1 <= 'Q')) { - /* - NUM : 0 1 2 3 4 5 - BYTE: G E F O P Q - C%7 : 1 6 0 2 3 4 - C%7 : 0 1 2 3 4 5 6 - NUM : 2 0 3 4 5 X 1 - */ - static const char jphone_emoji_first_table[7] = {2, 0, 3, 4, 5, 0, 1}; - c3 = nkf_char_unicode_new((jphone_emoji_first_table[c1 % 7] << 8) - SP + 0xE000); - while ((c1 = (*i_getc)(f)) != EOF) { - if (SP <= c1 && c1 <= 'z') { - (*oconv)(0, c1 + c3); - } else break; /* c1 == SO */ - } + } else if (('E' <= c1 && c1 <= 'G') || + ('O' <= c1 && c1 <= 'Q')) { + /* + NUM : 0 1 2 3 4 5 + BYTE: G E F O P Q + C%7 : 1 6 0 2 3 4 + C%7 : 0 1 2 3 4 5 6 + NUM : 2 0 3 4 5 X 1 + */ + static const nkf_char jphone_emoji_first_table[7] = + {0xE1E0, 0xDFE0, 0xE2E0, 0xE3E0, 0xE4E0, 0xDFE0, 0xE0E0}; + c3 = nkf_char_unicode_new(jphone_emoji_first_table[c1 % 7]); + if ((c1 = (*i_getc)(f)) == EOF) LAST; + while (SP <= c1 && c1 <= 'z') { + (*oconv)(0, c1 + c3); + if ((c1 = (*i_getc)(f)) == EOF) LAST; } + SKIP; } - if (c1 == EOF) LAST; - SKIP; - } else { + else { + (*oconv)(0, ESC); + (*oconv)(0, '$'); + SEND; + } + } + else { + i_ungetc(c1,f); /* lonely ESC */ (*oconv)(0, ESC); - SEND; + SKIP; } } else if (c1 == LF || c1 == CR) { if (broken_f&4) { @@ -5463,9 +6230,9 @@ int kanji_convert(FILE *f) } } } else - SEND; - } - /* send: */ + SEND; + } + /* send: */ switch(input_mode){ case ASCII: switch ((*iconv)(c2, c1, 0)) { /* can be EUC / SJIS / UTF-8 */ @@ -5480,11 +6247,52 @@ int kanji_convert(FILE *f) } } 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); - (*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; } @@ -5495,8 +6303,6 @@ int kanji_convert(FILE *f) 0x7F <= c2 && c2 <= 0x92 && 0x21 <= c1 && c1 <= 0x7E) { /* CP932 UDC */ - if(c1 == 0x7F) - SKIP; c1 = nkf_char_unicode_new((c2 - 0x7F) * 94 + c1 - 0x21 + 0xE000); c2 = 0; } @@ -5514,12 +6320,13 @@ int kanji_convert(FILE *f) (*oconv)(input_mode, c1); /* other special case */ } - c2 = 0; - c3 = 0; - continue; - /* goto next_word */ + c2 = 0; + c3 = 0; + continue; + /* goto next_word */ } +finished: /* epilogue */ (*iconv)(EOF, 0, 0); if (!input_codename) @@ -5542,17 +6349,17 @@ int kanji_convert(FILE *f) /* * int options(unsigned char *cp) - * + * * return values: * 0: success * -1: ArgumentError */ -int options(unsigned char *cp) +static int +options(unsigned char *cp) { nkf_char i, j; unsigned char *p; unsigned char *cp_back = NULL; - char codeset[32]; nkf_encoding *enc; if (option_mode==1) @@ -5571,7 +6378,7 @@ int options(unsigned char *cp) option_mode = 1; return 0; } - for (i=0;iX0208 conversion */ @@ -6089,7 +6899,7 @@ int options(unsigned char *cp) 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: @@ -6107,7 +6917,8 @@ int options(unsigned char *cp) #include "nkf32dll.c" #elif defined(PERL_XS) #else /* WIN32DLL */ -int main(int argc, char **argv) +int +main(int argc, char **argv) { FILE *fin; unsigned char *cp; @@ -6121,35 +6932,37 @@ int main(int argc, char **argv) #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); + cp = (unsigned char *)*argv; + options(cp); #ifdef EXEC_IO - if (exec_f){ - int fds[2], pid; - if (pipe(fds) < 0 || (pid = fork()) < 0){ - abort(); - } - if (pid == 0){ - if (exec_f > 0){ - close(fds[0]); - dup2(fds[1], 1); - }else{ - close(fds[1]); - dup2(fds[0], 0); - } - execvp(argv[1], &argv[1]); - } - if (exec_f > 0){ - close(fds[1]); - dup2(fds[0], 0); - }else{ - close(fds[0]); - dup2(fds[1], 1); - } - argc = 0; - break; - } + if (exec_f){ + int fds[2], pid; + if (pipe(fds) < 0 || (pid = fork()) < 0){ + abort(); + } + if (pid == 0){ + if (exec_f > 0){ + close(fds[0]); + dup2(fds[1], 1); + }else{ + close(fds[1]); + dup2(fds[0], 0); + } + execvp(argv[1], &argv[1]); + } + if (exec_f > 0){ + close(fds[1]); + dup2(fds[0], 0); + }else{ + close(fds[0]); + dup2(fds[1], 1); + } + argc = 0; + break; + } #endif } @@ -6180,150 +6993,146 @@ int main(int argc, char **argv) if (binmode_f == TRUE) #if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__)) - if (freopen("","wb",stdout) == NULL) - return (-1); + if (freopen("","wb",stdout) == NULL) + return (-1); #else setbinmode(stdout); #endif if (unbuf_f) - setbuf(stdout, (char *) NULL); + setbuf(stdout, (char *) NULL); else - setvbuffer(stdout, (char *) stdobuf, IOBUF_SIZE); + setvbuffer(stdout, (char *) stdobuf, IOBUF_SIZE); if (argc == 0) { - if (binmode_f == TRUE) + if (binmode_f == TRUE) #if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__)) - if (freopen("","rb",stdin) == NULL) return (-1); + if (freopen("","rb",stdin) == NULL) return (-1); #else - setbinmode(stdin); -#endif - setvbuffer(stdin, (char *) stdibuf, IOBUF_SIZE); - if (nop_f) - noconvert(stdin); - else { - kanji_convert(stdin); - if (guess_f) print_guessed_code(NULL); - } + setbinmode(stdin); +#endif + setvbuffer(stdin, (char *) stdibuf, IOBUF_SIZE); + if (nop_f) + noconvert(stdin); + else { + kanji_convert(stdin); + if (guess_f) print_guessed_code(NULL); + } } else { - int nfiles = argc; + int nfiles = argc; int is_argument_error = FALSE; - while (argc--) { + while (argc--) { input_codename = NULL; input_eol = 0; #ifdef CHECK_OPTION iconv_for_check = 0; #endif - if ((fin = fopen((origfname = *argv++), "r")) == NULL) { + if ((fin = fopen((origfname = *argv++), "r")) == NULL) { perror(*(argv-1)); is_argument_error = TRUE; continue; - } else { + } else { #ifdef OVERWRITE - int fd = 0; - int fd_backup = 0; + int fd = 0; + int fd_backup = 0; #endif -/* reopen file for stdout */ - if (file_out_f == TRUE) { + /* reopen file for stdout */ + if (file_out_f == TRUE) { #ifdef OVERWRITE - if (overwrite_f){ - outfname = malloc(strlen(origfname) - + strlen(".nkftmpXXXXXX") - + 1); - if (!outfname){ - perror(origfname); - return -1; - } - strcpy(outfname, origfname); + if (overwrite_f){ + outfname = nkf_xmalloc(strlen(origfname) + + strlen(".nkftmpXXXXXX") + + 1); + strcpy(outfname, origfname); #ifdef MSDOS - { - int i; - for (i = strlen(outfname); i; --i){ - if (outfname[i - 1] == '/' - || outfname[i - 1] == '\\'){ - break; - } - } - outfname[i] = '\0'; - } - strcat(outfname, "ntXXXXXX"); - mktemp(outfname); + { + int i; + for (i = strlen(outfname); i; --i){ + if (outfname[i - 1] == '/' + || outfname[i - 1] == '\\'){ + break; + } + } + outfname[i] = '\0'; + } + strcat(outfname, "ntXXXXXX"); + mktemp(outfname); fd = open(outfname, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, - S_IREAD | S_IWRITE); + S_IREAD | S_IWRITE); #else - strcat(outfname, ".nkftmpXXXXXX"); - fd = mkstemp(outfname); -#endif - if (fd < 0 - || (fd_backup = dup(fileno(stdout))) < 0 - || dup2(fd, fileno(stdout)) < 0 - ){ - perror(origfname); - return -1; - } - }else -#endif - if(argc == 1) { - outfname = *argv++; - argc--; - } else { - outfname = "nkf.out"; - } - - if(freopen(outfname, "w", stdout) == NULL) { - perror (outfname); - return (-1); - } - if (binmode_f == TRUE) { + strcat(outfname, ".nkftmpXXXXXX"); + fd = mkstemp(outfname); +#endif + if (fd < 0 + || (fd_backup = dup(fileno(stdout))) < 0 + || dup2(fd, fileno(stdout)) < 0 + ){ + perror(origfname); + return -1; + } + }else +#endif + if(argc == 1) { + outfname = *argv++; + argc--; + } else { + outfname = "nkf.out"; + } + + if(freopen(outfname, "w", stdout) == NULL) { + perror (outfname); + return (-1); + } + if (binmode_f == TRUE) { #if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__)) - if (freopen("","wb",stdout) == NULL) - return (-1); + if (freopen("","wb",stdout) == NULL) + return (-1); #else - setbinmode(stdout); + setbinmode(stdout); #endif - } - } - if (binmode_f == TRUE) + } + } + if (binmode_f == TRUE) #if defined(__OS2__) && (defined(__IBMC__) || defined(__IBMCPP__)) - if (freopen("","rb",fin) == NULL) - return (-1); + if (freopen("","rb",fin) == NULL) + return (-1); #else - setbinmode(fin); -#endif - setvbuffer(fin, (char *) stdibuf, IOBUF_SIZE); - if (nop_f) - noconvert(fin); - else { - char *filename = NULL; - kanji_convert(fin); - if (nfiles > 1) filename = origfname; - if (guess_f) print_guessed_code(filename); - } - fclose(fin); + setbinmode(fin); +#endif + setvbuffer(fin, (char *) stdibuf, IOBUF_SIZE); + if (nop_f) + noconvert(fin); + else { + char *filename = NULL; + kanji_convert(fin); + if (nfiles > 1) filename = origfname; + if (guess_f) print_guessed_code(filename); + } + fclose(fin); #ifdef OVERWRITE - if (overwrite_f) { - struct stat sb; + if (overwrite_f) { + struct stat sb; #if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) && !defined(__WATCOMC__) && !defined(__EMX__) && !defined(__OS2__) && !defined(__DJGPP__) - time_t tb[2]; + time_t tb[2]; #else - struct utimbuf tb; -#endif - - fflush(stdout); - close(fd); - if (dup2(fd_backup, fileno(stdout)) < 0){ - perror("dup2"); - } - if (stat(origfname, &sb)) { - fprintf(stderr, "Can't stat %s\n", origfname); - } - /* パーミッションを復元 */ - if (chmod(outfname, sb.st_mode)) { - fprintf(stderr, "Can't set permission %s\n", outfname); - } - - /* タイムスタンプを復元 */ + struct utimbuf tb; +#endif + + fflush(stdout); + close(fd); + if (dup2(fd_backup, fileno(stdout)) < 0){ + perror("dup2"); + } + if (stat(origfname, &sb)) { + fprintf(stderr, "Can't stat %s\n", origfname); + } + /* パーミッションを復元 */ + if (chmod(outfname, sb.st_mode)) { + fprintf(stderr, "Can't set permission %s\n", outfname); + } + + /* タイムスタンプを復元 */ if(preserve_time_f){ #if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) && !defined(__WATCOMC__) && !defined(__EMX__) && !defined(__OS2__) && !defined(__DJGPP__) tb[0] = tb[1] = sb.st_mtime; @@ -6348,6 +7157,7 @@ int main(int argc, char **argv) fprintf(stderr, "Can't rename %s to %s\n", origfname, backup_filename); } + nkf_xfree(backup_filename); }else{ #ifdef MSDOS if (unlink(origfname)){ @@ -6355,27 +7165,27 @@ int main(int argc, char **argv) } #endif } - if (rename(outfname, origfname)) { - perror(origfname); - fprintf(stderr, "Can't rename %s to %s\n", - outfname, origfname); - } - free(outfname); - } -#endif - } - } + if (rename(outfname, origfname)) { + perror(origfname); + fprintf(stderr, "Can't rename %s to %s\n", + outfname, origfname); + } + nkf_xfree(outfname); + } +#endif + } + } if (is_argument_error) return(-1); } #ifdef EASYWIN /*Easy Win */ if (file_out_f == FALSE) - scanf("%d",&end_check); + scanf("%d",&end_check); else - fclose(stdout); + fclose(stdout); #else /* for Other OS */ if (file_out_f == TRUE) - fclose(stdout); + fclose(stdout); #endif /*Easy Win */ return (0); }