X-Git-Url: http://git.sourceforge.jp/view?a=blobdiff_plain;f=nkf.c;h=41636b01ef9bf7cb6cc1701eb92236b1332a6e3c;hb=05baf2e0251294e1c1201216e1c80dd0975ce58e;hp=3e4c58b6a831b085cbffe9f503562eef9ea11679;hpb=4f99e6deb7b1fb6a626eae223f9b83021796d363;p=nkf%2Fnkf.git diff --git a/nkf.c b/nkf.c index 3e4c58b..41636b0 100644 --- a/nkf.c +++ b/nkf.c @@ -39,14 +39,14 @@ ** E-Mail: furukawa@tcp-ip.or.jp ** まで御連絡をお願いします。 ***********************************************************************/ +/* $Id: nkf.c,v 1.71 2005/07/09 21:46:17 naruse Exp $ */ +#define NKF_VERSION "2.0.5" +#define NKF_RELEASE_DATE "2005-07-05" #include "config.h" static char *CopyRight = - "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2003 Kono, Furukawa"; -static char *Version = - "2.0"; -static char *Patchlevel = - "3/0310/Shinji Kono"; + "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2005 Kono, Furukawa, Naruse"; + /* ** @@ -55,8 +55,8 @@ static char *Patchlevel = ** USAGE: nkf [flags] [file] ** ** Flags: -** b Output is bufferred (DEFAULT) -** u Output is unbufferred +** b Output is buffered (DEFAULT) +** u Output is unbuffered ** ** t no operation ** @@ -105,13 +105,17 @@ static char *Patchlevel = #ifdef PERL_XS #undef OVERWRITE #endif +#if defined( UTF8_OUTPUT_ENABLE ) || defined( UTF8_INPUT_ENABLE ) +#define UNICODE_ENABLE +#else +#undef UNICODE_NORMALIZATION +#endif #ifndef PERL_XS #include #endif #include -#include #if defined(MSDOS) || defined(__OS2__) #include @@ -145,16 +149,25 @@ static char *Patchlevel = #ifdef OVERWRITE /* added by satoru@isoternet.org */ +#include #include #ifndef MSDOS /* UNIX, OS/2 */ #include #include -#else +#else /* defined(MSDOS) */ +#ifdef __WIN32__ +#ifdef __BORLANDC__ /* BCC32 */ +#include +#else /* !defined(__BORLANDC__) */ +#include +#endif /* (__BORLANDC__) */ +#else /* !defined(__WIN32__) */ #if defined(_MSC_VER) || defined(__MINGW32__) /* VC++, MinGW */ #include #elif defined(__TURBOC__) /* BCC */ #include #elif defined(LSI_C) /* LSI C */ +#endif /* (__WIN32__) */ #endif #endif #endif @@ -182,6 +195,7 @@ static char *Patchlevel = #define X0201 2 #define ISO8859_1 8 #define NO_X0201 3 +#define X0212 16 /* Input Assumption */ @@ -199,14 +213,15 @@ static char *Patchlevel = #define UTF8 12 #define UTF8_INPUT 13 -#define UTF16_INPUT 14 -#define UTF16BE_INPUT 15 +#define UTF16BE_INPUT 14 +#define UTF16LE_INPUT 15 #define WISH_TRUE 15 /* ASCII CODE */ #define BS 0x08 +#define TAB 0x09 #define NL 0x0a #define CR 0x0d #define ESC 0x1b @@ -217,6 +232,7 @@ static char *Patchlevel = #define SI 0x0f #define SO 0x0e #define SSO 0x8e +#define SS3 0x8f #define is_alnum(c) \ (('a'<=c && c<='z')||('A'<= c && c<='Z')||('0'<=c && c<='9')) @@ -235,7 +251,7 @@ static char *Patchlevel = #define GETA2 0x2e -#if defined( UTF8_OUTPUT_ENABLE ) || defined( UTF8_INPUT_ENABLE ) +#ifdef UNICODE_ENABLE #define sizeof_euc_utf8 94 #define sizeof_euc_to_utf8_1byte 94 #define sizeof_euc_to_utf8_2bytes 94 @@ -273,13 +289,13 @@ struct input_code{ int _file_stat; }; -STATIC char *input_codename; +STATIC char *input_codename = ""; STATIC int noconvert PROTO((FILE *f)); STATIC int kanji_convert PROTO((FILE *f)); STATIC int h_conv PROTO((FILE *f,int c2,int c1)); STATIC int push_hold_buf PROTO((int c2)); -STATIC void set_iconv PROTO((int f, int (*iconv_func)())); +STATIC void set_iconv PROTO((int f, int (*iconv_func)(int c2,int c1,int c0))); STATIC int s_iconv PROTO((int c2,int c1,int c0)); STATIC int s2e_conv PROTO((int c2, int c1, int *p2, int *p1)); STATIC int e_iconv PROTO((int c2,int c1,int c0)); @@ -296,7 +312,7 @@ STATIC void w_oconv PROTO((int c2,int c1)); STATIC void w_oconv16 PROTO((int c2,int c1)); #endif STATIC void e_oconv PROTO((int c2,int c1)); -STATIC void e2s_conv PROTO((int c2, int c1, int *p2, int *p1)); +STATIC int e2s_conv PROTO((int c2, int c1, int *p2, int *p1)); STATIC void s_oconv PROTO((int c2,int c1)); STATIC void j_oconv PROTO((int c2,int c1)); STATIC void fold_conv PROTO((int c2,int c1)); @@ -329,13 +345,14 @@ STATIC int mime_ungetc_buf PROTO((int c,FILE *f)); STATIC int mime_integrity PROTO((FILE *f,unsigned char *p)); STATIC int base64decode PROTO((int c)); +STATIC void mime_prechar PROTO((int c2, int c1)); STATIC void mime_putc PROTO((int c)); STATIC void open_mime PROTO((int c)); STATIC void close_mime PROTO(()); STATIC void usage PROTO(()); STATIC void version PROTO(()); STATIC void options PROTO((unsigned char *c)); -#ifdef PERL_XS +#if defined(PERL_XS) || defined(WIN32DLL) STATIC void reinit PROTO(()); #endif @@ -366,6 +383,7 @@ static int hira_f = FALSE; /* hira/kata henkan */ static int input_f = FALSE; /* non fixed input code */ static int alpha_f = FALSE; /* convert JIx0208 alphbet to ASCII */ static int mime_f = STRICT_MIME; /* convert MIME B base64 or Q */ +static int mime_decode_f = FALSE; /* mime decode is explicitly on */ static int mimebuf_f = FALSE; /* MIME buffered input */ static int broken_f = FALSE; /* convert ESC-less broken JIS */ static int iso8859_f = FALSE; /* ISO8859 through */ @@ -376,16 +394,21 @@ static int x0201_f = TRUE; /* Assume JISX0201 kana */ static int x0201_f = NO_X0201; /* Assume NO JISX0201 */ #endif static int iso2022jp_f = FALSE; /* convert ISO-2022-JP */ +#ifdef UNICODE_ENABLE +static int internal_unicode_f = FALSE; /* Internal Unicode Processing */ +#endif #ifdef UTF8_OUTPUT_ENABLE -static int w_oconv16_begin_f= 0; /* utf-16 header */ +static int unicode_bom_f= 0; /* Output Unicode BOM */ static int w_oconv16_LE = 0; /* utf-16 little endian */ +static int ms_ucs_map_f = FALSE; /* Microsoft UCS Mapping Compatible */ #endif - -#ifdef NUMCHAR_OPTION - -#define CLASS_MASK 0x0f000000 -#define CLASS_UTF16 0x01000000 +#ifdef UNICODE_NORMALIZATION +static int nfc_f = FALSE; +static int (*i_nfc_getc)PROTO((FILE *)) = std_getc; /* input of ugetc */ +static int (*i_nfc_ungetc)PROTO((int c ,FILE *f)) = std_ungetc; +STATIC int nfc_getc PROTO((FILE *f)); +STATIC int nfc_ungetc PROTO((int c,FILE *f)); #endif #ifdef INPUT_OPTION @@ -400,7 +423,11 @@ static int (*i_ugetc)PROTO((FILE *)) = std_getc; /* input of ugetc */ static int (*i_uungetc)PROTO((int c ,FILE *f)) = std_ungetc; STATIC int url_getc PROTO((FILE *f)); STATIC int url_ungetc PROTO((int c,FILE *f)); +#endif +#ifdef NUMCHAR_OPTION +#define CLASS_MASK 0x0f000000 +#define CLASS_UTF16 0x01000000 static int numchar_f = FALSE; static int (*i_ngetc)PROTO((FILE *)) = std_getc; /* input of ugetc */ static int (*i_nungetc)PROTO((int c ,FILE *f)) = std_ungetc; @@ -413,8 +440,15 @@ static int noout_f = FALSE; STATIC void no_putc PROTO((int c)); static int debug_f = FALSE; STATIC void debug PROTO((char *str)); +static int (*iconv_for_check)() = 0; #endif +static int guess_f = FALSE; +STATIC void print_guessed_code PROTO((char *filename)); +STATIC void set_input_codename PROTO((char *codename)); +static int is_inputcode_mixed = FALSE; +static int is_inputcode_set = FALSE; + #ifdef EXEC_IO static int exec_f = 0; #endif @@ -424,22 +458,35 @@ STATIC int cp932_f = TRUE; #define CP932_TABLE_BEGIN (0xfa) #define CP932_TABLE_END (0xfc) +STATIC int cp932inv_f = TRUE; +#define CP932INV_TABLE_BEGIN (0xed) +#define CP932INV_TABLE_END (0xee) + +/* STATIC int cp932_conv PROTO((int c2, int c1)); */ #endif /* SHIFTJIS_CP932 */ +#ifdef X0212_ENABLE +STATIC int x0212_f = FALSE; +static int x0212_shift PROTO((int c)); +static int x0212_unshift PROTO((int c)); +#endif + +STATIC unsigned char prefix_table[256]; + STATIC void e_status PROTO((struct input_code *, int)); STATIC void s_status PROTO((struct input_code *, int)); #ifdef UTF8_INPUT_ENABLE STATIC void w_status PROTO((struct input_code *, int)); STATIC void w16_status PROTO((struct input_code *, int)); -static int utf16_mode = UTF16_INPUT; +static int utf16_mode = UTF16BE_INPUT; #endif struct input_code input_code_list[] = { {"EUC-JP", 0, 0, 0, {0, 0, 0}, e_status, e_iconv, 0}, {"Shift_JIS", 0, 0, 0, {0, 0, 0}, s_status, s_iconv, 0}, {"UTF-8", 0, 0, 0, {0, 0, 0}, w_status, w_iconv, 0}, - {"UTF-16", 0, 0, 0, {0, 0, 0}, w16_status, w_iconv16, 0}, + {"UTF-16", 0, 0, 0, {0, 0, 0}, w16_status, w_iconv16, 0}, {0} }; @@ -456,8 +503,8 @@ static int fold_f = FALSE; static int fold_len = 0; /* options */ -static unsigned char kanji_intro = DEFAULT_J, - ascii_intro = DEFAULT_R; +static unsigned char kanji_intro = DEFAULT_J; +static unsigned char ascii_intro = DEFAULT_R; /* Folding */ @@ -623,7 +670,14 @@ static int crmode_f = 0; /* CR, NL, CRLF */ static int end_check; #endif /*Easy Win */ -#ifndef PERL_XS +#define STD_GC_BUFSIZE (256) +int std_gc_buf[STD_GC_BUFSIZE]; +int std_gc_ndx; + +#ifdef WIN32DLL +#include "nkf32dll.c" +#elif defined(PERL_XS) +#else /* WIN32DLL */ int main(argc, argv) int argc; @@ -632,6 +686,9 @@ main(argc, argv) FILE *fin; unsigned char *cp; + char *outfname = NULL; + char *origfname; + #ifdef EASYWIN /*Easy Win */ _BufferSize.y = 400;/*Set Scroll Buffer Size*/ #endif @@ -693,20 +750,26 @@ main(argc, argv) setvbuffer(stdin, stdibuf, IOBUF_SIZE); if (nop_f) noconvert(stdin); - else + else { kanji_convert(stdin); + if (guess_f) print_guessed_code(NULL); + } } else { + int nfiles = argc; while (argc--) { - char *outfname; - char *origfname; - + is_inputcode_mixed = FALSE; + is_inputcode_set = FALSE; + input_codename = ""; +#ifdef CHECK_OPTION + iconv_for_check = 0; +#endif if ((fin = fopen((origfname = *argv++), "r")) == NULL) { perror(*--argv); return(-1); } else { #ifdef OVERWRITE - int fd; - int fd_backup; + int fd = 0; + int fd_backup = 0; #endif /* reopen file for stdout */ @@ -779,13 +842,17 @@ main(argc, argv) setvbuffer(fin, stdibuf, IOBUF_SIZE); if (nop_f) noconvert(fin); - else + else { + char *filename = NULL; kanji_convert(fin); + if (nfiles > 1) filename = origfname; + if (guess_f) print_guessed_code(filename); + } fclose(fin); #ifdef OVERWRITE if (overwrite) { struct stat sb; -#if defined(MSDOS) && !defined(__MINGW32__) +#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) time_t tb[2]; #else struct utimbuf tb; @@ -805,7 +872,7 @@ main(argc, argv) } /* タイムスタンプを復元 */ -#if defined(MSDOS) && !defined(__MINGW32__) +#if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__) tb[0] = tb[1] = sb.st_mtime; if (utime(outfname, tb)) { fprintf(stderr, "Can't set timestamp %s\n", outfname); @@ -841,10 +908,10 @@ main(argc, argv) #else /* for Other OS */ if (file_out == TRUE) fclose(stdout); -#endif +#endif /*Easy Win */ return (0); } -#endif +#endif /* WIN32DLL */ static struct { @@ -870,14 +937,27 @@ struct { {"hiragana","h1"}, {"katakana","h2"}, {"katakana-hiragana","h3"}, + {"guess", "g"}, + {"cp932", ""}, + {"no-cp932", ""}, +#ifdef X0212_ENABLE + {"x0212", ""}, +#endif +#ifdef UNICODE_ENABLE + {"internal-unicode", ""}, +#endif #ifdef UTF8_OUTPUT_ENABLE {"utf8", "w"}, {"utf16", "w16"}, + {"ms-ucs-map", ""}, #endif #ifdef UTF8_INPUT_ENABLE {"utf8-input", "W"}, {"utf16-input", "W16"}, #endif +#ifdef UNICODE_NORMALIZATION + {"utf8mac-input", ""}, +#endif #ifdef OVERWRITE {"overwrite", ""}, #endif @@ -893,28 +973,29 @@ struct { {"debug", ""}, #endif #ifdef SHIFTJIS_CP932 - {"no-cp932", ""}, + {"cp932inv", ""}, #endif #ifdef EXEC_IO {"exec-in", ""}, {"exec-out", ""}, #endif + {"prefix=", ""}, }; -static int option_mode; +static int option_mode = 0; void options(cp) unsigned char *cp; { int i; - unsigned char *p; + unsigned char *p = NULL; if (option_mode==1) return; - if (*cp++ != '-') - return; + while(*cp && *cp++!='-'); while (*cp) { + p = 0; switch (*cp++) { case '-': /* literal options */ if (!*cp) { /* ignore the rest of arguments */ @@ -924,12 +1005,17 @@ options(cp) for (i=0;iname); + debug(input_codename); + } iconv_for_check = iconv; } #endif } -#define SCORE_KANA (1) /* いわゆる半角カナ */ +#define SCORE_L2 (1) /* 第2水準漢字 */ +#define SCORE_KANA (SCORE_L2 << 1) /* いわゆる半角カナ */ #define SCORE_DEPEND (SCORE_KANA << 1) /* 機種依存文字 */ #ifdef SHIFTJIS_CP932 #define SCORE_CP932 (SCORE_DEPEND << 1) /* CP932 による読み換え */ @@ -1241,7 +1424,11 @@ void set_iconv(f, iconv_func) #else #define SCORE_NO_EXIST (SCORE_DEPEND << 1) /* 存在しない文字 */ #endif -#define SCORE_ERROR (SCORE_NO_EXIST << 1) /* エラー */ +#define SCORE_iMIME (SCORE_NO_EXIST << 1) /* MIME による指定 */ +#define SCORE_ERROR (SCORE_iMIME << 1) /* エラー */ + +#define SCORE_INIT (SCORE_iMIME) + int score_table_A0[] = { 0, 0, 0, 0, 0, 0, 0, 0, @@ -1250,8 +1437,8 @@ int score_table_A0[] = { }; int score_table_F0[] = { - 0, 0, 0, 0, - 0, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, + SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2, + SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR, }; @@ -1260,7 +1447,18 @@ void set_code_score(ptr, score) struct input_code *ptr; int score; { - ptr->score |= score; + if (ptr){ + ptr->score |= score; + } +} + +void clr_code_score(ptr, score) + struct input_code *ptr; + int score; +{ + if (ptr){ + ptr->score &= ~score; + } } void code_score(ptr) @@ -1270,18 +1468,19 @@ void code_score(ptr) int c1 = ptr->buf[1]; if (c2 < 0){ set_code_score(ptr, SCORE_ERROR); - }else if ((c2 & 0xf0) == 0xa0){ - set_code_score(ptr, score_table_A0[c2 & 0x0f]); - }else if ((c2 & 0xf0) == 0xf0){ - set_code_score(ptr, score_table_F0[c2 & 0x0f]); }else if (c2 == SSO){ set_code_score(ptr, SCORE_KANA); - } #ifdef UTF8_OUTPUT_ENABLE - else if (!e2w_conv(c2, c1)){ + }else if (!e2w_conv(c2, c1)){ set_code_score(ptr, SCORE_NO_EXIST); - } #endif + }else if ((c2 & 0x70) == 0x20){ + set_code_score(ptr, score_table_A0[c2 & 0x0f]); + }else if ((c2 & 0x70) == 0x70){ + set_code_score(ptr, score_table_F0[c2 & 0x0f]); + }else if ((c2 & 0x70) >= 0x50){ + set_code_score(ptr, SCORE_L2); + } } void status_disable(ptr) @@ -1300,14 +1499,20 @@ void status_push_ch(ptr, c) ptr->buf[ptr->index++] = c; } -void status_reset(ptr) +void status_clear(ptr) struct input_code *ptr; { ptr->stat = 0; - ptr->score = 0; ptr->index = 0; } +void status_reset(ptr) + struct input_code *ptr; +{ + status_clear(ptr); + ptr->score = SCORE_INIT; +} + void status_reinit(ptr) struct input_code *ptr; { @@ -1343,7 +1548,7 @@ void s_status(ptr, c) status_push_ch(ptr, SSO); status_push_ch(ptr, c); code_score(ptr); - status_reset(ptr); + status_clear(ptr); }else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xef)){ ptr->stat = 1; status_push_ch(ptr, c); @@ -1353,6 +1558,11 @@ void s_status(ptr, c) ptr->stat = 2; status_push_ch(ptr, c); #endif /* SHIFTJIS_CP932 */ +#ifdef X0212_ENABLE + }else if (x0212_f && 0xf0 <= c && c <= 0xfc){ + ptr->stat = 1; + status_push_ch(ptr, c); +#endif /* X0212_ENABLE */ }else{ status_disable(ptr); } @@ -1362,24 +1572,26 @@ void s_status(ptr, c) status_push_ch(ptr, c); s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]); code_score(ptr); - status_reset(ptr); + status_clear(ptr); }else{ status_disable(ptr); } break; -#ifdef SHIFTJIS_CP932 case 2: +#ifdef SHIFTJIS_CP932 if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){ status_push_ch(ptr, c); if (s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]) == 0){ set_code_score(ptr, SCORE_CP932); - status_reset(ptr); + status_clear(ptr); break; } } +#endif /* SHIFTJIS_CP932 */ +#ifndef X0212_ENABLE status_disable(ptr); break; -#endif /* SHIFTJIS_CP932 */ +#endif } } @@ -1401,6 +1613,11 @@ void e_status(ptr, c) }else if (SSO == c || (0xa1 <= c && c <= 0xfe)){ ptr->stat = 1; status_push_ch(ptr, c); +#ifdef X0212_ENABLE + }else if (0x8f == c){ + ptr->stat = 2; + status_push_ch(ptr, c); +#endif /* X0212_ENABLE */ }else{ status_disable(ptr); } @@ -1409,11 +1626,20 @@ void e_status(ptr, c) if (0xa1 <= c && c <= 0xfe){ status_push_ch(ptr, c); code_score(ptr); - status_reset(ptr); + status_clear(ptr); }else{ status_disable(ptr); } break; +#ifdef X0212_ENABLE + case 2: + if (0xa1 <= c && c <= 0xfe){ + ptr->stat = 1; + status_push_ch(ptr, c); + }else{ + status_disable(ptr); + } +#endif /* X0212_ENABLE */ } } @@ -1449,7 +1675,7 @@ void w16_status(ptr, c) ptr->_file_stat = -1; }else{ status_push_ch(ptr, c); - status_reset(ptr); + status_clear(ptr); } break; @@ -1457,7 +1683,7 @@ void w16_status(ptr, c) case 0xff: if (ptr->stat != c && (c == 0xfe || c == 0xff)){ status_push_ch(ptr, c); - status_reset(ptr); + status_clear(ptr); }else{ status_disable(ptr); ptr->_file_stat = -1; @@ -1496,10 +1722,14 @@ void w_status(ptr, c) if (0x80 <= c && c <= 0xbf){ status_push_ch(ptr, c); if (ptr->index > ptr->stat){ + int bom = (ptr->buf[0] == 0xef && ptr->buf[1] == 0xbb + && ptr->buf[2] == 0xbf); w2e_conv(ptr->buf[0], ptr->buf[1], ptr->buf[2], &ptr->buf[0], &ptr->buf[1]); - code_score(ptr); - status_reset(ptr); + if (!bom){ + code_score(ptr); + } + status_clear(ptr); } }else{ status_disable(ptr); @@ -1543,39 +1773,31 @@ code_status(c) } } -#ifdef PERL_XS -#define STD_GC_BUFSIZE (256) -int std_gc_buf[STD_GC_BUFSIZE]; -int std_gc_ndx; -#endif - +#ifndef WIN32DLL int std_getc(f) FILE *f; { -#ifdef PERL_XS if (std_gc_ndx){ return std_gc_buf[--std_gc_ndx]; } -#endif return getc(f); } +#endif /*WIN32DLL*/ int std_ungetc(c,f) int c; FILE *f; { -#ifdef PERL_XS if (std_gc_ndx == STD_GC_BUFSIZE){ return EOF; } std_gc_buf[std_gc_ndx++] = c; return c; -#endif - return ungetc(c,f); } +#ifndef WIN32DLL void std_putc(c) int c; @@ -1583,6 +1805,7 @@ int c; if(c!=EOF) putchar(c); } +#endif /*WIN32DLL*/ int noconvert(f) @@ -1606,7 +1829,7 @@ module_connection() /* output redicrection */ #ifdef CHECK_OPTION - if (noout_f){ + if (noout_f || guess_f){ o_putc = no_putc; } #endif @@ -1640,6 +1863,7 @@ module_connection() } i_getc = std_getc; + i_ungetc = std_ungetc; /* input redicrection */ #ifdef INPUT_OPTION if (cap_f){ @@ -1657,6 +1881,12 @@ module_connection() i_nungetc = i_ungetc; i_ungetc= numchar_ungetc; } #endif +#ifdef UNICODE_NORMALIZATION + if (nfc_f && input_f == UTF8_INPUT){ + i_nfc_getc = i_getc; i_getc = nfc_getc; + i_nfc_ungetc = i_ungetc; i_ungetc= nfc_ungetc; + } +#endif if (mime_f && mimebuf_f==FIXED_MIME) { i_mgetc = i_getc; i_getc = mime_getc; i_mungetc = i_ungetc; i_ungetc = mime_ungetc; @@ -1672,7 +1902,9 @@ module_connection() #ifdef UTF8_INPUT_ENABLE } else if (input_f == UTF8_INPUT) { set_iconv(-TRUE, w_iconv); - } else if (input_f == UTF16_INPUT) { + } else if (input_f == UTF16BE_INPUT) { + set_iconv(-TRUE, w_iconv16); + } else if (input_f == UTF16LE_INPUT) { set_iconv(-TRUE, w_iconv16); #endif } else { @@ -1696,7 +1928,8 @@ kanji_convert(f) FILE *f; { int c1, - c2; + c2, c3; + int is_8bit = FALSE; module_connection(); c2 = 0; @@ -1716,7 +1949,7 @@ kanji_convert(f) /* second byte */ if (c2 > DEL) { /* in case of 8th bit is on */ - if (!estab_f) { + if (!estab_f&&!mime_decode_mode) { /* in case of not established yet */ /* It is still ambiguious */ if (h_conv(f, c2, c1)==EOF) @@ -1761,6 +1994,7 @@ kanji_convert(f) /* 8 bit code */ if (!estab_f && !iso8859_f) { /* not established yet */ + if (!is_8bit) is_8bit = TRUE; c2 = c1; NEXT; } else { /* estab_f==TRUE */ @@ -1870,7 +2104,8 @@ kanji_convert(f) /* This is kanji introduction */ input_mode = X0208; shift_mode = FALSE; - debug(input_codename = "ISO-2022-JP"); + set_input_codename("ISO-2022-JP"); + debug(input_codename); NEXT; } else if (c1 == '(') { if ((c1 = (*i_getc)(f)) == EOF) { @@ -1885,6 +2120,12 @@ kanji_convert(f) input_mode = X0208; shift_mode = FALSE; NEXT; +#ifdef X0212_ENABLE + } else if (c1 == 'D'){ + input_mode = X0212; + shift_mode = FALSE; + NEXT; +#endif /* X0212_ENABLE */ } else { /* could be some special code */ (*oconv)(0, ESC); @@ -1931,10 +2172,16 @@ kanji_convert(f) } } else if ( c1 == 'N' || c1 == 'n' ){ /* SS2 */ - c1 = (*i_getc)(f); /* skip SS2 */ - if ( SPACE<=c1 && c1 < 0xe0 ) { + c3 = (*i_getc)(f); /* skip SS2 */ + if ( (SPACE<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){ + c1 = c3; c2 = X0201; SEND; + }else{ + (*i_ungetc)(c3, f); + /* lonely ESC */ + (*oconv)(0, ESC); + SEND; } } else { /* lonely ESC */ @@ -1944,12 +2191,42 @@ kanji_convert(f) } else if ((c1 == NL || c1 == CR) && broken_f&4) { input_mode = ASCII; set_iconv(FALSE, 0); SEND; + } else if (c1 == NL && mime_decode_f && !mime_decode_mode ) { + if ((c1=(*i_getc)(f))!=EOF && c1 == SPACE) { + i_ungetc(SPACE,f); + continue; + } else { + i_ungetc(c1,f); + } + c1 = NL; + SEND; + } else if (c1 == CR && mime_decode_f && !mime_decode_mode ) { + if ((c1=(*i_getc)(f))!=EOF) { + if (c1==SPACE) { + i_ungetc(SPACE,f); + continue; + } else if (c1 == NL && (c1=(*i_getc)(f))!=EOF && c1 == SPACE) { + i_ungetc(SPACE,f); + continue; + } else { + i_ungetc(c1,f); + } + i_ungetc(NL,f); + } else { + i_ungetc(c1,f); + } + c1 = CR; + SEND; } else SEND; } /* send: */ if (input_mode == X0208) (*oconv)(c2, c1); /* this is JIS, not SJIS/EUC case */ +#ifdef X0212_ENABLE + else if (input_mode == X0212) + (*oconv)((0x8f << 8) | c2, c1); +#endif /* X0212_ENABLE */ else if (input_mode) (*oconv)(input_mode, c1); /* other special case */ else if ((*iconv)(c2, c1, 0) < 0){ /* can be EUC/SJIS */ @@ -1967,6 +2244,18 @@ kanji_convert(f) /* epilogue */ (*iconv)(EOF, 0, 0); + if (!is_inputcode_set) + { + if (is_8bit) { + struct input_code *p = input_code_list; + struct input_code *result = p; + while (p->name){ + if (p->score < result->score) result = p; + ++p; + } + set_input_codename(result->name); + } + } return 1; } @@ -2011,7 +2300,7 @@ h_conv(f, c2, c1) } ++p; } - set_iconv(FALSE, p->iconv_func); + set_iconv(FALSE, result->iconv_func); } @@ -2084,15 +2373,35 @@ int s2e_conv(c2, c1, p2, p1) int c2, c1; int *p2, *p1; { + int val; #ifdef SHIFTJIS_CP932 - if (CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){ + if (cp932_f && CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){ extern unsigned short shiftjis_cp932[3][189]; - c1 = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40]; - if (c1 == 0) return 1; - c2 = c1 >> 8; - c1 &= 0xff; + val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40]; + if (val){ + c2 = val >> 8; + c1 = val & 0xff; + } } #endif /* SHIFTJIS_CP932 */ +#ifdef X0212_ENABLE + if (x0212_f && 0xfa <= c2 && c2 <= 0xfc){ + extern unsigned short shiftjis_x0212[3][189]; + val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40]; + if (val){ + if (val & 0x8000){ + c2 = (0x8f << 8) | (val >> 8); + c1 = val & 0xff; + }else{ + c2 = val >> 8; + c1 = val & 0xff; + } + if (p2) *p2 = c2; + if (p1) *p1 = c1; + return 0; + } + } +#endif c2 = c2 + c2 - ((c2 <= 0x9f) ? SJ0162 : SJ6394); if (c1 < 0x9f) c1 = c1 - ((c1 > DEL) ? SPACE : 0x1f); @@ -2100,6 +2409,12 @@ int s2e_conv(c2, c1, p2, p1) c1 = c1 - 0x7e; c2++; } + +#ifdef X0212_ENABLE + if (x0212_f){ + c2 = x0212_unshift(c2); + } +#endif if (p2) *p2 = c2; if (p1) *p1 = c1; return 0; @@ -2129,6 +2444,26 @@ e_iconv(c2, c1, c0) { if (c2 == X0201) { c1 &= 0x7f; +#ifdef X0212_ENABLE + }else if (c2 == 0x8f){ + if (c0 == 0){ + return -1; + } + c2 = (c2 << 8) | (c1 & 0x7f); + c1 = c0 & 0x7f; +#ifdef SHIFTJIS_CP932 + if (cp932_f){ + int s2, s1; + if (e2s_conv(c2, c1, &s2, &s1) == 0){ + s2e_conv(s2, s1, &c2, &c1); + if ((c2 & 0xff00) == 0){ + c1 &= 0x7f; + c2 &= 0x7f; + } + } + } +#endif /* SHIFTJIS_CP932 */ +#endif /* X0212_ENABLE */ } else if (c2 == SSO){ c2 = X0201; c1 &= 0x7f; @@ -2164,8 +2499,8 @@ w2e_conv(c2, c1, c0, p2, p1) } #ifdef NUMCHAR_OPTION if (ret){ - c1 = CLASS_UTF16 | ww16_conv(c2, c1, c0); - c2 = 0; + if (p2) *p2 = 0; + if (p1) *p1 = CLASS_UTF16 | ww16_conv(c2, c1, c0); ret = 0; } #endif @@ -2183,7 +2518,32 @@ w_iconv(c2, c1, c0) int c2, c1, c0; { - int ret = w2e_conv(c2, c1, c0, &c2, &c1); + int ret = 0; + unsigned short val = 0; + + if (c0 == 0){ + if (c2 < 0x80 || (c2 & 0xc0) == 0xdf) /* 0x00-0x7f 0xc0-0xdf */ + ; /* 1 or 2ytes */ + else if ((c2 & 0xf0) == 0xe0) /* 0xe0-0xef */ + return -1; /* 3bytes */ +#ifdef __COMMENT__ + else if (0xf0 <= c2) + return 0; /* 4,5,6bytes */ + else if ((c2 & 0xc0) == 0x80) /* 0x80-0xbf */ + return 0; /* trail byte */ +#endif + else return 0; + } + if (c2 == EOF); + else if (c2 == 0xef && c1 == 0xbb && c0 == 0xbf) + return 0; /* throw BOM */ + else if (internal_unicode_f && (output_conv == w_oconv || output_conv == w_oconv16)){ + val = ww16_conv(c2, c1, c0); + c2 = (val >> 8) & 0xff; + c1 = val & 0xff; + } else { + ret = w2e_conv(c2, c1, c0, &c2, &c1); + } if (ret == 0){ (*oconv)(c2, c1); } @@ -2221,7 +2581,7 @@ ww16_conv(c2, c1, c0) val |= (c0 & 0x3f); }else if (c2 >= 0xc0){ val = (c2 & 0x1f) << 6; - val |= (c1 & 0x3f) << 6; + val |= (c1 & 0x3f); }else{ val = c2; } @@ -2258,6 +2618,9 @@ w16e_conv(val, p2, p1) ret = 0; } #endif + }else{ + *p2 = 0; + *p1 = c2; } return ret; } @@ -2266,16 +2629,16 @@ int w_iconv16(c2, c1, c0) int c2, c1,c0; { - int ret; + int ret = 0; if (c2==0376 && c1==0377){ - utf16_mode = UTF16_INPUT; + utf16_mode = UTF16BE_INPUT; return 0; } else if (c2==0377 && c1==0376){ - utf16_mode = UTF16BE_INPUT; + utf16_mode = UTF16LE_INPUT; return 0; } - if (utf16_mode == UTF16BE_INPUT) { + if (c2 != EOF && utf16_mode == UTF16LE_INPUT) { int tmp; tmp=c1; c1=c2; c2=tmp; } @@ -2283,7 +2646,8 @@ w_iconv16(c2, c1, c0) (*oconv)(c2, c1); return 0; } - ret = w16e_conv(((c2<<8)&0xff00) + c1, &c2, &c1); + if (internal_unicode_f && (output_conv == w_oconv || output_conv == w_oconv16)); + else ret = w16e_conv(((c2<<8)&0xff00) + c1, &c2, &c1); if (ret) return ret; (*oconv)(c2, c1); return 0; @@ -2313,6 +2677,10 @@ w_iconv_common(c1, c0, pp, psize, p2, p1) if (val == 0) return 1; c2 = val >> 8; + if (val & 0x8000){ + c2 &= 0x7f; + c2 |= 0x8f00; + } if (c2 == SO) c2 = X0201; c1 = val & 0x7f; if (p2) *p2 = c2; @@ -2329,15 +2697,25 @@ e2w_conv(c2, c1) { extern unsigned short euc_to_utf8_1byte[]; extern unsigned short * euc_to_utf8_2bytes[]; + extern unsigned short * euc_to_utf8_2bytes_ms[]; unsigned short *p; if (c2 == X0201) { p = euc_to_utf8_1byte; +#ifdef X0212_ENABLE + } else if (c2 >> 8 == 0x8f){ + extern unsigned short * x0212_to_utf8_2bytes[]; + c2 = (c2&0x7f) - 0x21; + if (0<=c2 && c2> 8) & 0xff; c1 = val & 0xff; } @@ -2437,6 +2833,9 @@ e_oconv(c2, c1) #ifdef NUMCHAR_OPTION if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){ w16e_conv(c1, &c2, &c1); + if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){ + return; + } } #endif if (c2 == EOF) { @@ -2451,9 +2850,31 @@ e_oconv(c2, c1) } else if (c2 == ISO8859_1) { output_mode = ISO8859_1; (*o_putc)(c1 | 0x080); +#ifdef X0212_ENABLE + } else if ((c2 & 0xff00) >> 8 == 0x8f){ + output_mode = JAPANESE_EUC; +#ifdef SHIFTJIS_CP932 + if (cp932_f){ + int s2, s1; + if (e2s_conv(c2, c1, &s2, &s1) == 0){ + s2e_conv(s2, s1, &c2, &c1); + } + } +#endif + if ((c2 & 0xff00) >> 8 == 0x8f){ + if (x0212_f){ + (*o_putc)(0x8f); + (*o_putc)((c2 & 0x7f) | 0x080); + (*o_putc)(c1 | 0x080); + } + }else{ + (*o_putc)((c2 & 0x7f) | 0x080); + (*o_putc)(c1 | 0x080); + } +#endif } else { - if ((c1<0x20 || 0x7e> 1) + ((c2 <= 0x5e) ? 0x71 : 0xb1); - if (p1) *p1 = c1 + ((c2 & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e); + int ret = c; + c &= 0x7f; + if ((ret & 0xff00) == 0x8f00){ + if (0x75 <= c && c <= 0x7f){ + ret = c + (0x109 - 0x75); + } + }else{ + if (0x75 <= c && c <= 0x7f){ + ret = c + (0x113 - 0x75); + } + } + return ret; } -void -s_oconv(c2, c1) - int c2, - c1; -{ + +int x0212_unshift(c) + int c; +{ + int ret = c; + if (0x7f <= c && c <= 0x88){ + ret = c + (0x75 - 0x7f); + }else if (0x89 <= c && c <= 0x92){ + ret = (0x8f << 8) | 0x80 | (c + (0x75 - 0x89)); + } + return ret; +} +#endif /* X0212_ENABLE */ + +int +e2s_conv(c2, c1, p2, p1) + int c2, c1, *p2, *p1; +{ + int val = 0; + unsigned short *ptr; + int ndx; + extern unsigned short *x0212_shiftjis[]; +#ifdef X0212_ENABLE + if ((c2 & 0xff00) == 0x8f00){ + ndx = c2 & 0x7f; + if (0x21 <= ndx && ndx <= 0x7e){ + ptr = x0212_shiftjis[ndx - 0x21]; + if (ptr){ + val = ptr[(c1 & 0x7f) - 0x21]; + } + if (val){ + c2 = val >> 8; + c1 = val & 0xff; + if (p2) *p2 = c2; + if (p1) *p1 = c1; + return 0; + } + } + c2 = x0212_shift(c2); + } +#endif /* X0212_ENABLE */ + if ((c2 & 0xff00) == 0x8f00){ + return 1; + } + if (p2) *p2 = ((c2 - 1) >> 1) + ((c2 <= 0x5e) ? 0x71 : 0xb1); + if (p1) *p1 = c1 + ((c2 & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e); + return 0; +} + +void +s_oconv(c2, c1) + int c2, + c1; +{ #ifdef NUMCHAR_OPTION if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){ w16e_conv(c1, &c2, &c1); @@ -2493,6 +2973,14 @@ s_oconv(c2, c1) } else if (c2 == ISO8859_1) { output_mode = ISO8859_1; (*o_putc)(c1 | 0x080); +#ifdef X0212_ENABLE + } else if ((c2 & 0xff00) >> 8 == 0x8f){ + output_mode = SHIFT_JIS; + if (e2s_conv(c2, c1, &c2, &c1) == 0){ + (*o_putc)(c2); + (*o_putc)(c1); + } +#endif } else { if ((c1<0x20 || 0x7e> 8; + c1 = c & 0xff; + } + } +#endif /* SHIFTJIS_CP932 */ + (*o_putc)(c2); + if (prefix_table[(unsigned char)c1]){ + (*o_putc)(prefix_table[(unsigned char)c1]); + } (*o_putc)(c1); } } @@ -2524,6 +3028,18 @@ j_oconv(c2, c1) output_mode = ASCII; } (*o_putc)(EOF); +#ifdef X0212_ENABLE + } else if ((c2 & 0xff00) >> 8 == 0x8f){ + if (output_mode!=X0212) { + output_mode = X0212; + (*o_putc)(ESC); + (*o_putc)('$'); + (*o_putc)('('); + (*o_putc)('D'); + } + (*o_putc)(c2 & 0x7f); + (*o_putc)(c1); +#endif } else if (c2==X0201) { if (output_mode!=X0201) { output_mode = X0201; @@ -2567,13 +3083,7 @@ base64_conv(c2, c1) int c2, c1; { - if (base64_count>50 && !mimeout_mode && c2==0 && c1==SPACE) { - (*o_putc)(NL); - } else if (base64_count>66 && mimeout_mode) { - (*o_base64conv)(EOF,0); - (*o_putc)(NL); - (*o_putc)('\t'); base64_count += 7; - } + mime_prechar(c2, c1); (*o_base64conv)(c2,c1); } @@ -3033,23 +3543,37 @@ unsigned char *mime_pattern[] = { (unsigned char *)"\075?ISO-2022-JP?Q?", #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) (unsigned char *)"\075?UTF-8?B?", + (unsigned char *)"\075?UTF-8?Q?", #endif + (unsigned char *)"\075?US-ASCII?Q?", NULL }; + +/* 該当するコードの優先度を上げるための目印 */ +int (*mime_priority_func[])PROTO((int c2, int c1, int c0)) = { + e_iconv, s_iconv, 0, 0, 0, 0, +#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) + w_iconv, w_iconv, +#endif + 0, +}; + int mime_encode[] = { JAPANESE_EUC, SHIFT_JIS,ISO8859_1, ISO8859_1, X0208, X0201, #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) - UTF8, + UTF8, UTF8, #endif + ASCII, 0 }; int mime_encode_method[] = { 'B', 'B','Q', 'B', 'B', 'Q', #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) - 'B', + 'B', 'Q', #endif + 'Q', 0 }; @@ -3060,6 +3584,10 @@ int mime_encode_method[] = { #define nkf_toupper(c) (('a'<=c && c<='z')?(c-('a'-'A')):c) #define nkf_isdigit(c) ('0'<=c && c<='9') #define nkf_isxdigit(c) (nkf_isdigit(c) || ('a'<=c && c<='f') || ('A'<=c && c <= 'F')) +#define nkf_isblank(c) (c == SPACE || c == TAB) +#define nkf_isspace(c) (nkf_isblank(c) || c == CR || c == NL) +#define nkf_isalpha(c) (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z')) +#define nkf_isalnum(c) (nkf_isdigit(c) || nkf_isalpha(c)) void switch_mime_getc() @@ -3119,6 +3647,10 @@ FILE *f; } } mime_decode_mode = p[i-2]; + + set_iconv(FALSE, mime_priority_func[j]); + clr_code_score(find_inputcode_byfunc(mime_priority_func[j]), SCORE_iMIME); + if (mime_decode_mode=='B') { mimebuf_f = unbuf_f; if (!unbuf_f) { @@ -3228,6 +3760,39 @@ void debug(str) } #endif +void +set_input_codename (codename) + char *codename; +{ + if (guess_f && + is_inputcode_set && + strcmp(codename, "") != 0 && + strcmp(codename, input_codename) != 0) + { + is_inputcode_mixed = TRUE; + } + input_codename = codename; + is_inputcode_set = TRUE; +} + +#ifndef WIN32DLL +void +print_guessed_code (filename) + char *filename; +{ + char *codename = "BINARY"; + if (!is_inputcode_mixed) { + if (strcmp(input_codename, "") == 0) { + codename = "ASCII"; + } else { + codename = input_codename; + } + } + if (filename != NULL) printf("%s:", filename); + printf("%s\n", codename); +} +#endif /*WIN32DLL*/ + int hex2bin(x) int x; @@ -3364,6 +3929,57 @@ numchar_ungetc(c, f) } #endif +#ifdef UNICODE_NORMALIZATION + +/* Normalization Form C */ +int +nfc_getc(f) + FILE *f; +{ + int (*g)() = i_nfc_getc; + int (*u)() = i_nfc_ungetc; + int i=0, j, k=1, lower, upper; + int buf[9]; + int *array = NULL; + extern struct normalization_pair normalization_table[]; + + buf[i] = (*g)(f); + while (k > 0 && ((buf[i] & 0xc0) != 0x80)){ + lower=0, upper=NORMALIZATION_TABLE_LENGTH-1; + while (upper >= lower) { + j = (lower+upper) / 2; + array = normalization_table[j].nfd; + for (k=0; k < NORMALIZATION_TABLE_NFD_LENGTH && array[k]; k++){ + if (array[k] != buf[k]){ + array[k] < buf[k] ? (lower = j + 1) : (upper = j - 1); + k = 0; + break; + } else if (k >= i) + buf[++i] = (*g)(f); + } + if (k > 0){ + array = normalization_table[j].nfc; + for (i=0; i < NORMALIZATION_TABLE_NFC_LENGTH && array[i]; i++) + buf[i] = array[i]; + i--; + break; + } + } + while (i > 0) + (*u)(buf[i--], f); + } + return buf[0]; +} + +int +nfc_ungetc(c, f) + int c; + FILE *f; +{ + return (*i_nfc_ungetc)(c, f); +} +#endif /* UNICODE_NORMALIZATION */ + int mime_getc(f) @@ -3371,6 +3987,10 @@ FILE *f; { int c1, c2, c3, c4, cc; int t1, t2, t3, t4, mode, exit_mode; + int lwsp_count; + char *lwsp_buf; + char *lwsp_buf_new; + int lwsp_size = 128; if (mime_top != mime_last) { /* Something is in FIFO */ return Fifo(mime_top++); @@ -3399,8 +4019,69 @@ restart_mime_q: if (c1=='?'&&c2=='=' && mimebuf_f != FIXED_MIME) { /* end Q encoding */ input_mode = exit_mode; - while((c1=(*i_getc)(f))!=EOF && c1==SPACE - /* && (c1==NL||c1==TAB||c1=='\r') */ ) ; + lwsp_count = 0; + lwsp_buf = malloc((lwsp_size+5)*sizeof(char)); + if (lwsp_buf==NULL) { + perror("can't malloc"); + return -1; + } + while ((c1=(*i_getc)(f))!=EOF) { + switch (c1) { + case NL: + case CR: + if (c1==NL) { + if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) { + i_ungetc(SPACE,f); + continue; + } else { + i_ungetc(c1,f); + } + c1 = NL; + } else { + if ((c1=(*i_getc)(f))!=EOF && c1 == NL) { + if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) { + i_ungetc(SPACE,f); + continue; + } else { + i_ungetc(c1,f); + } + i_ungetc(NL,f); + } else { + i_ungetc(c1,f); + } + c1 = CR; + } + break; + case SPACE: + case TAB: + lwsp_buf[lwsp_count] = c1; + if (lwsp_count++>lwsp_size){ + lwsp_size *= 2; + lwsp_buf_new = realloc(lwsp_buf, (lwsp_size+5)*sizeof(char)); + if (lwsp_buf_new==NULL) { + free(lwsp_buf); + lwsp_buf = NULL; + perror("can't realloc"); + return -1; + } + lwsp_buf = lwsp_buf_new; + } + continue; + } + break; + } + if (lwsp_count > 0) { + if (c1=='=' && (lwsp_buf[lwsp_count-1]==SPACE||lwsp_buf[lwsp_count-1]==TAB)) { + lwsp_count = 0; + } else { + i_ungetc(c1,f); + for(lwsp_count--;lwsp_count>0;lwsp_count--) + i_ungetc(lwsp_buf[lwsp_count],f); + c1 = lwsp_buf[0]; + } + } + free(lwsp_buf); + lwsp_buf = NULL; return c1; } if (c1=='='&&c2<' ') { /* this is soft wrap */ @@ -3454,8 +4135,72 @@ mime_c2_retry: } if ((c1 == '?') && (c2 == '=')) { input_mode = ASCII; - while((c1=(*i_getc)(f))!=EOF && c1==SPACE - /* && (c1==NL||c1==TAB||c1=='\r') */ ) ; + lwsp_count = 0; + lwsp_buf = malloc((lwsp_size+5)*sizeof(char)); + if (lwsp_buf==NULL) { + perror("can't malloc"); + return -1; + } + while ((c1=(*i_getc)(f))!=EOF) { + switch (c1) { + case NL: + case CR: + if (c1==NL) { + if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) { + i_ungetc(SPACE,f); + continue; + } else { + i_ungetc(c1,f); + } + c1 = NL; + } else { + if ((c1=(*i_getc)(f))!=EOF) { + if (c1==SPACE) { + i_ungetc(SPACE,f); + continue; + } else if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) { + i_ungetc(SPACE,f); + continue; + } else { + i_ungetc(c1,f); + } + i_ungetc(NL,f); + } else { + i_ungetc(c1,f); + } + c1 = CR; + } + break; + case SPACE: + case TAB: + lwsp_buf[lwsp_count] = c1; + if (lwsp_count++>lwsp_size){ + lwsp_size *= 2; + lwsp_buf_new = realloc(lwsp_buf, (lwsp_size+5)*sizeof(char)); + if (lwsp_buf_new==NULL) { + free(lwsp_buf); + lwsp_buf = NULL; + perror("can't realloc"); + return -1; + } + lwsp_buf = lwsp_buf_new; + } + continue; + } + break; + } + if (lwsp_count > 0) { + if (c1=='=' && (lwsp_buf[lwsp_count-1]==SPACE||lwsp_buf[lwsp_count-1]==TAB)) { + lwsp_count = 0; + } else { + i_ungetc(c1,f); + for(lwsp_count--;lwsp_count>0;lwsp_count--) + i_ungetc(lwsp_buf[lwsp_count],f); + c1 = lwsp_buf[0]; + } + } + free(lwsp_buf); + lwsp_buf = NULL; return c1; } mime_c3_retry: @@ -3573,6 +4318,11 @@ static char basis_64[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; static int b64c; +#define MIMEOUT_BUF_LENGTH (60) +char mimeout_buf[MIMEOUT_BUF_LENGTH+1]; +int mimeout_buf_count = 0; +int mimeout_preserve_space = 0; +#define itoh4(c) (c>=10?c+'A'-10:c+'0') void open_mime(mode) @@ -3580,20 +4330,53 @@ int mode; { unsigned char *p; int i; + int j; p = mime_pattern[0]; for(i=0;mime_encode[i];i++) { if (mode == mime_encode[i]) { p = mime_pattern[i]; - break; + break; } } mimeout_mode = mime_encode_method[i]; - - /* (*o_mputc)(' '); */ + + i = 0; + if (base64_count>45) { + if (mimeout_buf_count>0 && nkf_isblank(mimeout_buf[i])){ + (*o_mputc)(mimeout_buf[i]); + i++; + } + (*o_mputc)(NL); + (*o_mputc)(SPACE); + base64_count = 1; + if (!mimeout_preserve_space && mimeout_buf_count>0 + && (mimeout_buf[i]==SPACE || mimeout_buf[i]==TAB + || mimeout_buf[i]==CR || mimeout_buf[i]==NL )) { + i++; + } + } + if (!mimeout_preserve_space) { + for (;i=10?c+'A'-10:c+'0') +void +eof_mime() +{ + switch(mimeout_mode) { + case 'Q': + case 'B': + break; + case 2: + (*o_mputc)(basis_64[((b64c & 0x3)<< 4)]); + (*o_mputc)('='); + (*o_mputc)('='); + base64_count += 3; + break; + case 1: + (*o_mputc)(basis_64[((b64c & 0xF) << 2)]); + (*o_mputc)('='); + base64_count += 2; + break; + } + if (mimeout_mode) { + if (mimeout_f!=FIXED_MIME) { + close_mime(); + } else if (mimeout_mode != 'Q') + mimeout_mode = 'B'; + } +} void -mime_putc(c) +mimeout_addchar(c) int c; { - if (mimeout_f==FIXED_MIME) { - if (base64_count>71) { - (*o_mputc)('\n'); - base64_count=0; - } - } else if (c==NL) { - base64_count=0; - } - if (c!=EOF) { - if ( c<=DEL &&(output_mode==ASCII ||output_mode == ISO8859_1 ) - && mimeout_f!=FIXED_MIME) { - if (mimeout_mode=='Q') { - if (c<=SPACE) { - close_mime(); - } - (*o_mputc)(c); - return; - } - if (mimeout_mode!='B' || c!=SPACE) { - if (mimeout_mode) { - mime_putc(EOF); - mimeout_mode=0; - } - (*o_mputc)(c); - base64_count ++; - return; - } - } else if (!mimeout_mode && mimeout_f!=FIXED_MIME) { - open_mime(output_mode); - } - } else { /* c==EOF */ - switch(mimeout_mode) { - case 'Q': - case 'B': - break; - case 2: - (*o_mputc)(basis_64[((b64c & 0x3)<< 4)]); - (*o_mputc)('='); - (*o_mputc)('='); - base64_count += 3; - break; - case 1: - (*o_mputc)(basis_64[((b64c & 0xF) << 2)]); - (*o_mputc)('='); - base64_count += 2; - break; - } - if (mimeout_mode) { - if (mimeout_f!=FIXED_MIME) { - close_mime(); - } else if (mimeout_mode != 'Q') - mimeout_mode = 'B'; - } - return; - } switch(mimeout_mode) { case 'Q': - if(c>=DEL) { + if(c==SPACE){ + (*o_mputc)('_'); + base64_count++; + } else if (c==CR||c==NL) { + (*o_mputc)(c); + base64_count = 0; + } else if(c 66){ + (*o_base64conv)(EOF,0); + (*o_base64conv)(0,NL); + (*o_base64conv)(0,SPACE); + } + }/*else if (mime_lastchar2){ + if (c1 <=DEL && !nkf_isspace(c1)){ + (*o_base64conv)(0,SPACE); + } + }*/ + }/*else{ + if (c2 && mime_lastchar2 == 0 + && mime_lastchar1 && !nkf_isspace(mime_lastchar1)){ + (*o_base64conv)(0,SPACE); + } + }*/ + mime_lastchar2 = c2; + mime_lastchar1 = c1; +} + +void +mime_putc(c) + int c; +{ + int i = 0; + int j = 0; + int lastchar; + + if (mimeout_f == FIXED_MIME){ + if (mimeout_mode == 'Q'){ + if (base64_count > 71){ + if (c!=CR && c!=NL) { + (*o_mputc)('='); + (*o_mputc)(NL); + } + base64_count = 0; + } + }else{ + if (base64_count > 71){ + eof_mime(); + (*o_mputc)(NL); + base64_count = 0; + } + if (c == EOF) { /* c==EOF */ + eof_mime(); + } + } + if (c != EOF) { /* c==EOF */ + mimeout_addchar(c); + } + return; + } + + /* mimeout_f != FIXED_MIME */ + + if (c == EOF) { /* c==EOF */ + j = mimeout_buf_count; + mimeout_buf_count = 0; + i = 0; + for (;i 0){ + lastchar = mimeout_buf[mimeout_buf_count - 1]; + }else{ + lastchar = -1; + } + + if (!mimeout_mode) { + if (c <= DEL && (output_mode==ASCII ||output_mode == ISO8859_1)) { + if (nkf_isspace(c)) { + if (c==CR || c==NL) { + base64_count=0; + } + for (i=0;i 1 + && base64_count + mimeout_buf_count > 76){ + (*o_mputc)(NL); + base64_count = 0; + if (!nkf_isspace(mimeout_buf[0])){ + (*o_mputc)(SPACE); + base64_count++; + } + } + mimeout_buf[mimeout_buf_count++] = c; + if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) { + open_mime(output_mode); + } + } + return; + }else{ + if (lastchar==CR || lastchar == NL){ + for (i=0;iMIMEOUT_BUF_LENGTH) { + eof_mime(); + for (i=0;i0 && SPACEMIMEOUT_BUF_LENGTH) { + j = mimeout_buf_count; + mimeout_buf_count = 0; + for (i=0;i0) { + j = mimeout_buf_count; + mimeout_buf_count = 0; + for (i=0;iname){ + status_reinit(p++); + } + } unbuf_f = FALSE; estab_f = FALSE; nop_f = FALSE; - binmode_f = TRUE; - rot_f = FALSE; - hira_f = FALSE; - input_f = FALSE; - alpha_f = FALSE; - mime_f = STRICT_MIME; - mimebuf_f = FALSE; - broken_f = FALSE; - iso8859_f = FALSE; -#if defined(MSDOS) || defined(__OS2__) - x0201_f = TRUE; + binmode_f = TRUE; + rot_f = FALSE; + hira_f = FALSE; + input_f = FALSE; + alpha_f = FALSE; + mime_f = STRICT_MIME; + mime_decode_f = FALSE; + mimebuf_f = FALSE; + broken_f = FALSE; + iso8859_f = FALSE; + mimeout_f = FALSE; +#if defined(MSDOS) || defined(__OS2__) + x0201_f = TRUE; #else x0201_f = NO_X0201; #endif iso2022jp_f = FALSE; - +#ifdef UNICODE_ENABLE + internal_unicode_f = TRUE; +#endif +#ifdef UTF8_OUTPUT_ENABLE + unicode_bom_f = 0; + w_oconv16_LE = 0; + ms_ucs_map_f = FALSE; +#endif +#ifdef UNICODE_NORMALIZATION + nfc_f = FALSE; +#endif +#ifdef INPUT_OPTION + cap_f = FALSE; + url_f = FALSE; + numchar_f = FALSE; +#endif +#ifdef CHECK_OPTION + noout_f = FALSE; + debug_f = FALSE; +#endif + guess_f = FALSE; + is_inputcode_mixed = FALSE; + is_inputcode_set = FALSE; +#ifdef EXEC_IO + exec_f = 0; +#endif +#ifdef SHIFTJIS_CP932 + cp932_f = TRUE; + cp932inv_f = TRUE; +#endif + { + int i; + for (i = 0; i < 256; i++){ + prefix_table[i] = 0; + } + } +#ifdef UTF8_INPUT_ENABLE + utf16_mode = UTF16BE_INPUT; +#endif + mimeout_buf_count = 0; + mimeout_mode = 0; + base64_count = 0; + f_line = 0; + f_prev = 0; + fold_preserve_f = FALSE; + fold_f = FALSE; + fold_len = 0; kanji_intro = DEFAULT_J; ascii_intro = DEFAULT_R; - - output_conv = DEFAULT_CONV; - oconv = DEFAULT_CONV; - - i_mgetc = std_getc; - i_mungetc = std_ungetc; - i_mgetc_buf = std_getc; - i_mungetc_buf = std_ungetc; - - i_getc= std_getc; - i_ungetc=std_ungetc; - - i_bgetc= std_getc; - i_bungetc= std_ungetc; - - o_putc = std_putc; - o_mputc = std_putc; - o_crconv = no_connection; - o_rot_conv = no_connection; - o_iso2022jp_check_conv = no_connection; - o_hira_conv = no_connection; - o_fconv = no_connection; + fold_margin = FOLD_MARGIN; + output_conv = DEFAULT_CONV; + oconv = DEFAULT_CONV; o_zconv = no_connection; - + o_fconv = no_connection; + o_crconv = no_connection; + o_rot_conv = no_connection; + o_hira_conv = no_connection; + o_base64conv = no_connection; + o_iso2022jp_check_conv = no_connection; + o_putc = std_putc; i_getc = std_getc; i_ungetc = std_ungetc; - i_mgetc = std_getc; - i_mungetc = std_ungetc; - + i_bgetc = std_getc; + i_bungetc = std_ungetc; + o_mputc = std_putc; + i_mgetc = std_getc; + i_mungetc = std_ungetc; + i_mgetc_buf = std_getc; + i_mungetc_buf = std_ungetc; output_mode = ASCII; input_mode = ASCII; shift_mode = FALSE; - mime_decode_mode = FALSE; + mime_decode_mode = FALSE; file_out = FALSE; - mimeout_mode = 0; - mimeout_f = FALSE; - base64_count = 0; - option_mode = 0; crmode_f = 0; - - { - struct input_code *p = input_code_list; - while (p->name){ - status_reinit(p++); - } - } -#ifdef UTF8_OUTPUT_ENABLE - if (w_oconv16_begin_f) { - w_oconv16_begin_f = 2; - } -#endif - f_line = 0; - f_prev = 0; - fold_preserve_f = FALSE; - fold_f = FALSE; - fold_len = 0; - fold_margin = FOLD_MARGIN; + option_mode = 0; broken_counter = 0; broken_last = 0; z_prev2=0,z_prev1=0; - +#ifdef CHECK_OPTION + iconv_for_check = 0; +#endif + input_codename = ""; +#ifdef WIN32DLL + reinitdll(); +#endif /*WIN32DLL*/ } #endif @@ -3801,32 +4817,42 @@ int c2,c1,c0; { fprintf(stderr,"nkf internal module connection failure.\n"); exit(1); + return 0; /* LINT */ } #ifndef PERL_XS +#ifdef WIN32DLL +#define fprintf dllprintf +#endif void usage() { fprintf(stderr,"USAGE: nkf(nkf32,wnkf,nkf2) -[flags] [in file] .. [out file for -O flag]\n"); fprintf(stderr,"Flags:\n"); - fprintf(stderr,"b,u Output is bufferred (DEFAULT),Output is unbufferred\n"); + fprintf(stderr,"b,u Output is buffered (DEFAULT),Output is unbuffered\n"); #ifdef DEFAULT_CODE_SJIS - fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS (DEFAULT), AT&T JIS (EUC), UTF-8\n"); + fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS (DEFAULT), AT&T JIS (EUC), UTF-8N\n"); #endif #ifdef DEFAULT_CODE_JIS - fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit (DEFAULT), Shift JIS, AT&T JIS (EUC), UTF-8\n"); + fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit (DEFAULT), Shift JIS, AT&T JIS (EUC), UTF-8N\n"); #endif #ifdef DEFAULT_CODE_EUC - fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC) (DEFAULT), UTF-8\n"); + fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC) (DEFAULT), UTF-8N\n"); #endif #ifdef DEFAULT_CODE_UTF8 - fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC), UTF-8 (DEFAULT)\n"); + fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC), UTF-8N (DEFAULT)\n"); +#endif +#ifdef UTF8_OUTPUT_ENABLE + fprintf(stderr," After 'w' you can add more options. (80?|16((B|L)0?)?) \n"); #endif fprintf(stderr,"J,S,E,W Input assumption is JIS 7 bit , Shift JIS, AT&T JIS (EUC), UTF-8\n"); +#ifdef UTF8_INPUT_ENABLE + fprintf(stderr," After 'W' you can add more options. (8|16(B|L)?) \n"); +#endif fprintf(stderr,"t no conversion\n"); fprintf(stderr,"i_/o_ Output sequence to designate JIS-kanji/ASCII (DEFAULT B)\n"); fprintf(stderr,"r {de/en}crypt ROT13/47\n"); - fprintf(stderr,"h 1 hirakana->katakana, 2 katakana->hirakana,3 both\n"); + fprintf(stderr,"h 1 katakana->hiragana, 2 hiragana->katakana, 3 both\n"); fprintf(stderr,"v Show this usage. V: show version\n"); fprintf(stderr,"m[BQN0] MIME decode [B:base64,Q:quoted,N:non-strict,0:no decode]\n"); fprintf(stderr,"M[BQ] MIME encode [B:base64 Q:quoted]\n"); @@ -3844,11 +4870,28 @@ usage() fprintf(stderr,"I Convert non ISO-2022-JP charactor to GETA\n"); fprintf(stderr,"-L[uwm] line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n"); fprintf(stderr,"long name options\n"); - fprintf(stderr," --fj,--unix,--mac,--windows convert for the system\n"); + fprintf(stderr," --fj,--unix,--mac,--windows convert for the system\n"); fprintf(stderr," --jis,--euc,--sjis,--utf8,--utf16,--mime,--base64 convert for the code\n"); + fprintf(stderr," --hiragana, --katakana Hiragana/Katakana Conversion\n"); + fprintf(stderr," --x0212 Convert JISX0212\n"); + fprintf(stderr," --cp932, --no-cp932 CP932 compatibility\n"); + fprintf(stderr," --prefix= Insert escape before troublesome characters of Shift_JIS\n"); +#ifdef INPUT_OPTION + fprintf(stderr," --cap-input, --url-input Convert hex after ':' or '%%'\n"); +#endif +#ifdef NUMCHAR_OPTION + fprintf(stderr," --numchar-input Convert Unicode Character Reference\n"); +#endif +#ifdef UNICODE_NORMALIZATION + fprintf(stderr," --utf8mac-input UTF-8-MAC input\n"); +#endif +#ifdef UTF8_OUTPUT_ENABLE + fprintf(stderr," --ms-ucs-map Microsoft UCS Mapping Compatible\n"); +#endif #ifdef OVERWRITE - fprintf(stderr," --overwrite Overwrite original listed files by filtered result\n"); + fprintf(stderr," --overwrite Overwrite original listed files by filtered result\n"); #endif + fprintf(stderr," -g, --guess Guess the input code\n"); fprintf(stderr," --help,--version\n"); version(); } @@ -3869,10 +4912,10 @@ version() #ifdef __OS2__ "for OS/2" #endif - ,Version,Patchlevel); + ,NKF_VERSION,NKF_RELEASE_DATE); fprintf(stderr,"\n%s\n",CopyRight); } -#endif +#endif /*PERL_XS*/ /** ** パッチ制作者