From: NARUSE, Yui Date: Thu, 24 Jan 2008 09:18:53 +0000 (+0000) Subject: * change order of definitions. X-Git-Tag: v2_0_9~49 X-Git-Url: http://git.osdn.net/view?a=commitdiff_plain;h=445fdbca2c1392de250c16cc36370afa85fe2407;hp=e4cb9198d2fa87aa4ed22614a712401eecc15b7d;p=nkf%2Fnkf.git * change order of definitions. --- diff --git a/nkf.c b/nkf.c index 1cd2461..8100da6 100644 --- a/nkf.c +++ b/nkf.c @@ -30,7 +30,7 @@ * 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 1.168 2008/01/23 23:24:30 naruse Exp $" +#define NKF_IDENT "$Id: nkf.c,v 1.169 2008/01/24 00:18:53 naruse Exp $" #define NKF_VERSION "2.0.8" #define NKF_RELEASE_DATE "2008-01-23" #define COPY_RIGHT \ @@ -296,7 +296,6 @@ struct { #define DEFAULT_R 'B' -#define RANGE_NUM_MAX 18 #define GETA1 0x22 #define GETA2 0x2e @@ -324,9 +323,6 @@ static nkf_encoding *output_encoding = NULL; static nkf_char kanji_convert(FILE *f); static nkf_char h_conv(FILE *f,nkf_char c2,nkf_char c1); -static nkf_char push_hold_buf(nkf_char c2); -static void set_iconv(nkf_char f, nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)); -static 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) /* UCS Mapping * 0: Shift_JIS, eucJP-ascii @@ -348,19 +344,12 @@ static int no_best_fit_chars_f = FALSE; static int input_endian = ENDIAN_BIG; static nkf_char unicode_subchar = '?'; /* the regular substitution character */ static void (*encode_fallback)(nkf_char c) = NULL; -static nkf_char unicode_to_jis_common(nkf_char c2,nkf_char c1,nkf_char c0,nkf_char *p2,nkf_char *p1); -static nkf_char w_iconv_common(nkf_char c1,nkf_char c0,const unsigned short *const *pp,nkf_char psize,nkf_char *p2,nkf_char *p1); -static void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0); -static nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0); -static nkf_char w16e_conv(nkf_char val,nkf_char *p2,nkf_char *p1); static void w_status(struct input_code *, nkf_char); #endif #ifdef UTF8_OUTPUT_ENABLE static int output_bom_f = FALSE; static int output_endian = ENDIAN_BIG; -static nkf_char e2w_conv(nkf_char c2,nkf_char c1); #endif -static nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1); static void fold_conv(nkf_char c2,nkf_char c1); static void eol_conv(nkf_char c2,nkf_char c1); static void z_conv(nkf_char c2,nkf_char c1); @@ -385,8 +374,6 @@ static void mime_putc(nkf_char c); static unsigned char stdibuf[IOBUF_SIZE]; static unsigned char stdobuf[IOBUF_SIZE]; #endif -static unsigned char hold_buf[HOLD_SIZE*2]; -static int hold_count = 0; /* flags */ static int unbuf_f = FALSE; @@ -444,9 +431,6 @@ 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 */ -#if !defined PERL_XS -static void print_guessed_code(char *filename); -#endif static void set_input_codename(char *codename); #ifdef EXEC_IO @@ -644,6 +628,7 @@ static const unsigned char fv[] = { +static int option_mode = 0; static int file_out_f = FALSE; #ifdef OVERWRITE static int overwrite_f = FALSE; @@ -1342,554 +1327,240 @@ static void set_output_encoding(nkf_encoding *enc) } } -static int option_mode = 0; - -void options(unsigned char *cp) +struct input_code * find_inputcode_byfunc(nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)) { - nkf_char i, j; - unsigned char *p; - unsigned char *cp_back = NULL; - char codeset[32]; - nkf_encoding *enc; - - if (option_mode==1) - return; - while(*cp && *cp++!='-'); - while (*cp || cp_back) { - if(!*cp){ - cp = cp_back; - cp_back = NULL; - continue; - } - p = 0; - switch (*cp++) { - case '-': /* literal options */ - if (!*cp || *cp == SP) { /* ignore the rest of arguments */ - option_mode = 1; - return; - } - for (i=0;iname){ + if (iconv_func == p->iconv_func){ + return p; } - if (p == 0) { - fprintf(stderr, "unknown long option: --%s\n", cp); - return; - } - while(*cp && *cp != SP && cp++); - if (long_option[i].alias[0]){ - cp_back = cp; - cp = (unsigned char *)long_option[i].alias; - }else{ - if (strcmp(long_option[i].name, "ic=") == 0){ - nkf_str_upcase((char *)p, codeset, 32); - enc = nkf_enc_find(codeset); - if (!enc) continue; - input_encoding = enc; - continue; - } - if (strcmp(long_option[i].name, "oc=") == 0){ - nkf_str_upcase((char *)p, codeset, 32); - enc = nkf_enc_find(codeset); - if (enc <= 0) continue; - output_encoding = enc; - continue; - } - if (strcmp(long_option[i].name, "guess=") == 0){ - if (p[0] == '0' || p[0] == '1') { - guess_f = 1; - } else { - guess_f = 2; - } - continue; - } -#ifdef OVERWRITE - if (strcmp(long_option[i].name, "overwrite") == 0){ - file_out_f = TRUE; - overwrite_f = TRUE; - preserve_time_f = TRUE; - continue; - } - if (strcmp(long_option[i].name, "overwrite=") == 0){ - file_out_f = TRUE; - overwrite_f = TRUE; - preserve_time_f = TRUE; - backup_f = TRUE; - backup_suffix = malloc(strlen((char *) p) + 1); - strcpy(backup_suffix, (char *) p); - continue; - } - if (strcmp(long_option[i].name, "in-place") == 0){ - file_out_f = TRUE; - overwrite_f = TRUE; - preserve_time_f = FALSE; - continue; - } - if (strcmp(long_option[i].name, "in-place=") == 0){ - file_out_f = TRUE; - overwrite_f = TRUE; - preserve_time_f = FALSE; - backup_f = TRUE; - backup_suffix = malloc(strlen((char *) p) + 1); - strcpy(backup_suffix, (char *) p); - continue; - } -#endif -#ifdef INPUT_OPTION - if (strcmp(long_option[i].name, "cap-input") == 0){ - cap_f = TRUE; - continue; - } - if (strcmp(long_option[i].name, "url-input") == 0){ - url_f = TRUE; - continue; - } + p++; + } + } + return 0; +} + +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 -#ifdef NUMCHAR_OPTION - if (strcmp(long_option[i].name, "numchar-input") == 0){ - numchar_f = TRUE; - continue; - } + if (estab_f != f){ + estab_f = f; + } + + if (iconv_func +#ifdef INPUT_CODE_FIX + && (f == -TRUE || !input_encoding) /* -TRUE means "FORCE" */ #endif + ){ + iconv = iconv_func; + } #ifdef CHECK_OPTION - if (strcmp(long_option[i].name, "no-output") == 0){ - noout_f = TRUE; - continue; - } - if (strcmp(long_option[i].name, "debug") == 0){ - debug_f = TRUE; - continue; - } -#endif - if (strcmp(long_option[i].name, "cp932") == 0){ -#ifdef SHIFTJIS_CP932 - cp51932_f = TRUE; - cp932inv_f = -TRUE; -#endif -#ifdef UTF8_OUTPUT_ENABLE - ms_ucs_map_f = UCS_MAP_CP932; -#endif - continue; - } - if (strcmp(long_option[i].name, "no-cp932") == 0){ -#ifdef SHIFTJIS_CP932 - cp51932_f = FALSE; - cp932inv_f = FALSE; -#endif -#ifdef UTF8_OUTPUT_ENABLE - ms_ucs_map_f = UCS_MAP_ASCII; -#endif - continue; - } -#ifdef SHIFTJIS_CP932 - if (strcmp(long_option[i].name, "cp932inv") == 0){ - cp932inv_f = -TRUE; - continue; - } + 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; + } #endif +} #ifdef X0212_ENABLE - if (strcmp(long_option[i].name, "x0212") == 0){ - x0212_f = TRUE; - continue; - } -#endif +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); + } + }else{ + if (0x75 <= c && c <= 0x7f){ + ret = c + (0x113 - 0x75); + } + } + return ret; +} -#ifdef EXEC_IO - if (strcmp(long_option[i].name, "exec-in") == 0){ - exec_f = 1; - return; - } - if (strcmp(long_option[i].name, "exec-out") == 0){ - exec_f = -1; - return; - } -#endif -#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE) - if (strcmp(long_option[i].name, "no-cp932ext") == 0){ - no_cp932ext_f = TRUE; - continue; - } - if (strcmp(long_option[i].name, "no-best-fit-chars") == 0){ - no_best_fit_chars_f = TRUE; - continue; - } - if (strcmp(long_option[i].name, "fb-skip") == 0){ - encode_fallback = NULL; - continue; - } - if (strcmp(long_option[i].name, "fb-html") == 0){ - encode_fallback = encode_fallback_html; - continue; - } - if (strcmp(long_option[i].name, "fb-xml") == 0){ - encode_fallback = encode_fallback_xml; - continue; - } - if (strcmp(long_option[i].name, "fb-java") == 0){ - encode_fallback = encode_fallback_java; - continue; - } - if (strcmp(long_option[i].name, "fb-perl") == 0){ - encode_fallback = encode_fallback_perl; - continue; - } - if (strcmp(long_option[i].name, "fb-subchar") == 0){ - encode_fallback = encode_fallback_subchar; - continue; - } - if (strcmp(long_option[i].name, "fb-subchar=") == 0){ - encode_fallback = encode_fallback_subchar; - unicode_subchar = 0; - if (p[0] != '0'){ - /* decimal number */ - for (i = 0; i < 7 && nkf_isdigit(p[i]); i++){ - unicode_subchar *= 10; - unicode_subchar += hex2bin(p[i]); - } - }else if(p[1] == 'x' || p[1] == 'X'){ - /* hexadecimal number */ - for (i = 2; i < 8 && nkf_isxdigit(p[i]); i++){ - unicode_subchar <<= 4; - unicode_subchar |= hex2bin(p[i]); - } - }else{ - /* octal number */ - for (i = 1; i < 8 && nkf_isoctal(p[i]); i++){ - unicode_subchar *= 8; - unicode_subchar += hex2bin(p[i]); - } - } - w16e_conv(unicode_subchar, &i, &j); - unicode_subchar = i<<8 | j; - continue; - } -#endif -#ifdef UTF8_OUTPUT_ENABLE - if (strcmp(long_option[i].name, "ms-ucs-map") == 0){ - ms_ucs_map_f = UCS_MAP_MS; - continue; - } -#endif -#ifdef UNICODE_NORMALIZATION - if (strcmp(long_option[i].name, "utf8mac-input") == 0){ - nfc_f = TRUE; - continue; - } + +nkf_char x0212_unshift(nkf_char c) +{ + nkf_char ret = c; + if (0x7f <= c && c <= 0x88){ + ret = c + (0x75 - 0x7f); + }else if (0x89 <= c && c <= 0x92){ + 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) +{ + nkf_char ndx; + if (is_eucg3(c2)){ + ndx = c2 & 0x7f; + if (x0213_f){ + 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); + return 0; + }else if(0x6E <= ndx && ndx <= 0x7E){ + if (p2) *p2 = ((ndx - 1) >> 1) + 0xbe; + if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e); + return 0; + } + return 1; + } +#ifdef X0212_ENABLE + else if(nkf_isgraph(ndx)){ + nkf_char val = 0; + const unsigned short *ptr; + 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(0x7F < c2) 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; +} + +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 - if (strcmp(long_option[i].name, "prefix=") == 0){ - if (nkf_isgraph(p[0])){ - for (i = 1; nkf_isgraph(p[i]); i++){ - prefix_table[p[i]] = p[0]; - } - } - continue; - } + static const char shift_jisx0213_s1a3_table[5][2] ={ { 1, 8}, { 3, 4}, { 5,12}, {13,14}, {15, 0} }; +#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 + && 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 */ +#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; } - continue; - case 'b': /* buffered mode */ - unbuf_f = FALSE; - continue; - case 'u': /* non bufferd mode */ - unbuf_f = TRUE; - continue; - case 't': /* transparent mode */ - if (*cp=='1') { - /* alias of -t */ - cp++; - nop_f = TRUE; - } else if (*cp=='2') { - /* - * -t with put/get - * - * nkf -t2MB hoge.bin | nkf -t2mB | diff -s - hoge.bin - * - */ - cp++; - nop_f = 2; - } else - nop_f = TRUE; - continue; - case 'j': /* JIS output */ - case 'n': - output_encoding = nkf_enc_from_index(ISO_2022_JP); - continue; - case 'e': /* AT&T EUC output */ - output_encoding = nkf_enc_from_index(EUC_JP); - continue; - case 's': /* SJIS output */ - output_encoding = nkf_enc_from_index(WINDOWS_31J); - continue; - case 'l': /* ISO8859 Latin-1 support, no conversion */ - iso8859_f = TRUE; /* Only compatible with ISO-2022-JP */ - input_encoding = nkf_enc_from_index(ISO_8859_1); - continue; - case 'i': /* Kanji IN ESC-$-@/B */ - if (*cp=='@'||*cp=='B') - kanji_intro = *cp++; - continue; - case 'o': /* ASCII IN ESC-(-J/B */ - if (*cp=='J'||*cp=='B'||*cp=='H') - ascii_intro = *cp++; - continue; - case 'h': - /* - bit:1 katakana->hiragana - bit:2 hiragana->katakana - */ - if ('9'>= *cp && *cp>='0') - hira_f |= (*cp++ -'0'); - else - hira_f |= 1; - continue; - case 'r': - rot_f = TRUE; - continue; -#if defined(MSDOS) || defined(__OS2__) - case 'T': - binmode_f = FALSE; - continue; -#endif -#ifndef PERL_XS - case 'V': - show_configuration(); - exit(1); - break; - case 'v': - usage(); - exit(1); - break; + if (p2) *p2 = c2; + if (p1) *p1 = c1; + return 0; + } + } #endif -#ifdef UTF8_OUTPUT_ENABLE - case 'w': /* UTF-8 output */ - if (cp[0] == '8') { - cp++; - if (cp[0] == '0'){ - cp++; - output_encoding = nkf_enc_from_index(UTF_8N); - } else { - output_bom_f = TRUE; - output_encoding = nkf_enc_from_index(UTF_8_BOM); - } - } else { - int enc_idx; - if ('1'== cp[0] && '6'==cp[1]) { - cp += 2; - enc_idx = UTF_16; - } else if ('3'== cp[0] && '2'==cp[1]) { - cp += 2; - enc_idx = UTF_32; - } else { - output_encoding = nkf_enc_from_index(UTF_8); - continue; - } - if (cp[0]=='L') { - cp++; - output_endian = ENDIAN_LITTLE; - } else if (cp[0] == 'B') { - cp++; - } else { - output_encoding = nkf_enc_from_index(enc_idx); - continue; - } - if (cp[0] == '0'){ - cp++; - enc_idx = enc_idx == UTF_16 - ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE) - : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE); - } else { - output_bom_f = TRUE; - enc_idx = enc_idx == UTF_16 - ? (output_endian == ENDIAN_LITTLE ? UTF_16LE_BOM : UTF_16BE_BOM) - : (output_endian == ENDIAN_LITTLE ? UTF_32LE_BOM : UTF_32BE_BOM); - } - output_encoding = nkf_enc_from_index(enc_idx); + if(c2 >= 0x80){ + if(x0213_f && c2 >= 0xF0){ + if(c2 <= 0xF3 || (c2 == 0xF4 && c1 < 0x9F)){ /* k=1, 3<=k<=5, k=8, 12<=k<=15 */ + c2 = PREFIX_EUCG3 | 0x20 | shift_jisx0213_s1a3_table[c2 - 0xF0][0x9E < c1]; + }else{ /* 78<=k<=94 */ + c2 = PREFIX_EUCG3 | (c2 * 2 - 0x17B); + if (0x9E < c1) c2++; } - continue; -#endif -#ifdef UTF8_INPUT_ENABLE - case 'W': /* UTF input */ - if (cp[0] == '8') { - cp++; - input_encoding = nkf_enc_from_index(UTF_8); - }else{ - int enc_idx; - if ('1'== cp[0] && '6'==cp[1]) { - cp += 2; - input_endian = ENDIAN_BIG; - enc_idx = UTF_16; - } else if ('3'== cp[0] && '2'==cp[1]) { - cp += 2; - input_endian = ENDIAN_BIG; - enc_idx = UTF_32; - } else { - input_encoding = nkf_enc_from_index(UTF_8); - continue; - } - if (cp[0]=='L') { - cp++; - input_endian = ENDIAN_LITTLE; - } else if (cp[0] == 'B') { - cp++; - input_endian = ENDIAN_BIG; - } - enc_idx = enc_idx == UTF_16 - ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE) - : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE); - input_encoding = nkf_enc_from_index(enc_idx); - } - continue; -#endif - /* Input code assumption */ - case 'J': /* ISO-2022-JP input */ - input_encoding = nkf_enc_from_index(ISO_2022_JP); - continue; - case 'E': /* EUC-JP input */ - input_encoding = nkf_enc_from_index(EUC_JP); - continue; - case 'S': /* Windows-31J input */ - input_encoding = nkf_enc_from_index(WINDOWS_31J); - continue; - case 'Z': /* Convert X0208 alphabet to asii */ - /* alpha_f - bit:0 Convert JIS X 0208 Alphabet to ASCII - bit:1 Convert Kankaku to one space - bit:2 Convert Kankaku to two spaces - bit:3 Convert HTML Entity - bit:4 Convert JIS X 0208 Katakana to JIS X 0201 Katakana - */ - while ('0'<= *cp && *cp <='9') { - alpha_f |= 1 << (*cp++ - '0'); - } - if (!alpha_f) alpha_f = 1; - continue; - case 'x': /* Convert X0201 kana to X0208 or X0201 Conversion */ - x0201_f = FALSE; /* No X0201->X0208 conversion */ - /* accept X0201 - ESC-(-I in JIS, EUC, MS Kanji - SI/SO in JIS, EUC, MS Kanji - SSO in EUC, JIS, not in MS Kanji - MS Kanji (0xa0-0xdf) - output X0201 - ESC-(-I in JIS (0x20-0x5f) - SSO in EUC (0xa0-0xdf) - 0xa0-0xd in MS Kanji (0xa0-0xdf) - */ - continue; - case 'X': /* Convert X0201 kana to X0208 */ - x0201_f = TRUE; - continue; - case 'F': /* prserve new lines */ - fold_preserve_f = TRUE; - case 'f': /* folding -f60 or -f */ - fold_f = TRUE; - fold_len = 0; - while('0'<= *cp && *cp <='9') { /* we don't use atoi here */ - fold_len *= 10; - fold_len += *cp++ - '0'; - } - if (!(0= *cp && *cp>='0') - broken_f |= 1<<(*cp++ -'0'); - else - broken_f |= TRUE; - continue; -#ifndef PERL_XS - case 'O':/* for Output file */ - file_out_f = TRUE; - continue; -#endif - case 'c':/* add cr code */ - eolmode_f = CRLF; - continue; - case 'd':/* delete cr code */ - eolmode_f = LF; - continue; - case 'I': /* ISO-2022-JP output */ - iso2022jp_f = TRUE; - continue; - case 'L': /* line mode */ - if (*cp=='u') { /* unix */ - eolmode_f = LF; cp++; - } else if (*cp=='m') { /* mac */ - eolmode_f = CR; cp++; - } else if (*cp=='w') { /* windows */ - eolmode_f = CRLF; cp++; - } else if (*cp=='0') { /* no conversion */ - eolmode_f = 0; cp++; - } - continue; -#ifndef PERL_XS - case 'g': - if ('2' <= *cp && *cp <= '9') { - guess_f = 2; - cp++; - } else if (*cp == '0' || *cp == '1') { - guess_f = 1; - cp++; - } else { - guess_f = 1; - } - continue; + }else{ +#define SJ0162 0x00e1 /* 01 - 62 ku offset */ +#define SJ6394 0x0161 /* 63 - 94 ku offset */ + c2 = c2 + c2 - ((c2 <= 0x9F) ? SJ0162 : SJ6394); + if (0x9E < c1) c2++; + } + if (c1 < 0x9F) + c1 = c1 - ((c1 > DEL) ? SP : 0x1F); + else { + c1 = c1 - 0x7E; + } + } + +#ifdef X0212_ENABLE + c2 = x0212_unshift(c2); #endif - case SP: - /* module muliple options in a string are allowed for Perl moudle */ - while(*cp && *cp++!='-'); - continue; - default: - fprintf(stderr, "unknown option: -%c\n", *(cp-1)); - /* bogus option but ignored */ - continue; - } + if (p2) *p2 = c2; + if (p1) *p1 = c1; + return 0; +} + +#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) +void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0) +{ + val &= VALUE_MASK; + if (val < 0x80){ + *p2 = val; + *p1 = 0; + *p0 = 0; + }else if (val < 0x800){ + *p2 = 0xc0 | (val >> 6); + *p1 = 0x80 | (val & 0x3f); + *p0 = 0; + } else if (val <= NKF_INT32_C(0xFFFF)) { + *p2 = 0xe0 | (val >> 12); + *p1 = 0x80 | ((val >> 6) & 0x3f); + *p0 = 0x80 | (val & 0x3f); + } else if (val <= NKF_INT32_C(0x10FFFF)) { + *p2 = 0xe0 | (val >> 16); + *p1 = 0x80 | ((val >> 12) & 0x3f); + *p0 = 0x8080 | ((val << 2) & 0x3f00)| (val & 0x3f); + } else { + *p2 = 0; + *p1 = 0; + *p0 = 0; + } +} + +nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0) +{ + nkf_char val; + if (c2 >= 0xf8) { + val = -1; + } else if (c2 >= 0xf0){ + /* c2: 1st, c1: 2nd, c0: 3rd/4th */ + val = (c2 & 0x0f) << 18; + val |= (c1 & 0x3f) << 12; + val |= (c0 & 0x3f00) >> 2; + val |= (c0 & 0x3f); + }else if (c2 >= 0xe0){ + val = (c2 & 0x0f) << 12; + val |= (c1 & 0x3f) << 6; + val |= (c0 & 0x3f); + }else if (c2 >= 0xc0){ + val = (c2 & 0x1f) << 6; + val |= (c1 & 0x3f); + }else{ + val = c2; } + return val; } +#endif #ifdef UTF8_INPUT_ENABLE nkf_char w_iconv_common(nkf_char c1, nkf_char c0, const unsigned short *const *pp, nkf_char psize, nkf_char *p2, nkf_char *p1) @@ -2081,19 +1752,64 @@ nkf_char unicode_to_jis_common(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char * return ret; } -nkf_char w2e_conv(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char *p1) +#ifdef UTF8_OUTPUT_ENABLE +nkf_char e2w_conv(nkf_char c2, nkf_char c1) { - nkf_char ret = 0; + const unsigned short *p; - if (!c1){ - *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 = CLASS_UNICODE | ww16_conv(c2, c1, c0); + if (c2 == JIS_X_0201) { + if (ms_ucs_map_f == UCS_MAP_CP10001) { + switch (c1) { + case 0x20: + return 0xA0; + case 0x7D: + return 0xA9; + } + } + 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 0){ + if (p2) *p2 = 0; + if (p1) *p1 = CLASS_UNICODE | ww16_conv(c2, c1, c0); ret = 0; } #endif @@ -2101,6 +1817,110 @@ nkf_char w2e_conv(nkf_char c2, nkf_char c1, nkf_char c0, nkf_char *p2, nkf_char return ret; } +#ifdef UTF8_INPUT_ENABLE +nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1) +{ + nkf_char c2, c1, c0; + nkf_char ret = 0; + val &= VALUE_MASK; + if (val < 0x80){ + *p2 = 0; + *p1 = val; + }else{ + w16w_conv(val, &c2, &c1, &c0); + ret = unicode_to_jis_common(c2, c1, c0, p2, p1); +#ifdef NUMCHAR_OPTION + if (ret > 0){ + *p2 = 0; + *p1 = CLASS_UNICODE | val; + ret = 0; + } +#endif + } + return ret; +} +#endif + +nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) +{ + if (c2 == JIS_X_0201) { + c1 &= 0x7f; +#ifdef X0212_ENABLE + }else if (c2 == 0x8f){ + 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 = (c1 - 0xF5) * 94 + c0 - 0xA1 + 0xE3AC + CLASS_UNICODE; + c2 = 0; + } else { + c2 = (c2 << 8) | (c1 & 0x7f); + c1 = c0 & 0x7f; +#ifdef SHIFTJIS_CP932 + if (cp51932_f){ + nkf_char s2, s1; + if (e2s_conv(c2, c1, &s2, &s1) == 0){ + s2e_conv(s2, s1, &c2, &c1); + if (c2 < 0x100){ + c1 &= 0x7f; + c2 &= 0x7f; + } + } + } +#endif /* SHIFTJIS_CP932 */ + } +#endif /* X0212_ENABLE */ + } else if (c2 == SSO){ + c2 = JIS_X_0201; + c1 &= 0x7f; + } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) { + /* 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 */ + c1 = (c2 - 0xF5) * 94 + c1 - 0xA1 + 0xE000 + CLASS_UNICODE; + c2 = 0; + } else { + c1 &= 0x7f; + c2 &= 0x7f; +#ifdef SHIFTJIS_CP932 + if (cp51932_f && 0x79 <= c2 && c2 <= 0x7c){ + nkf_char s2, s1; + if (e2s_conv(c2, c1, &s2, &s1) == 0){ + s2e_conv(s2, s1, &c2, &c1); + if (c2 < 0x100){ + c1 &= 0x7f; + c2 &= 0x7f; + } + } + } +#endif /* SHIFTJIS_CP932 */ + } + } + (*oconv)(c2, c1); + return 0; +} + +nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0) +{ + if (c2 == JIS_X_0201) { + c1 &= 0x7f; + } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) { + /* NOP */ + } else if (!x0213_f && 0xF0 <= c2 && c2 <= 0xF9 && 0x40 <= c1 && c1 <= 0xFC) { + /* CP932 UDC */ + if(c1 == 0x7F) return 0; + c1 = (c2 - 0xF0) * 188 + (c1 - 0x40 - (0x7E < c1)) + 0xE000 + CLASS_UNICODE; + c2 = 0; + } else { + nkf_char ret = s2e_conv(c2, c1, &c2, &c1); + if (ret) return ret; + } + (*oconv)(c2, c1); + return 0; +} + nkf_char w_iconv(nkf_char c2, nkf_char c1, nkf_char c0) { nkf_char ret = 0; @@ -2210,1488 +2030,1108 @@ nkf_char w_iconv32(nkf_char c2, nkf_char c1, nkf_char c0) #endif -struct input_code * find_inputcode_byfunc(nkf_char (*iconv_func)(nkf_char c2,nkf_char c1,nkf_char c0)) +void j_oconv(nkf_char c2, nkf_char c1) { - if (iconv_func){ - struct input_code *p = input_code_list; - while (p->name){ - if (iconv_func == p->iconv_func){ - return p; - } - p++; +#ifdef NUMCHAR_OPTION + if (c2 == 0 && is_unicode_capsule(c1)){ + w16e_conv(c1, &c2, &c1); + if (c2 == 0 && is_unicode_capsule(c1)){ + c2 = c1 & VALUE_MASK; + if (ms_ucs_map_f && 0xE000 <= c2 && c2 <= 0xE757) { + /* CP5022x UDC */ + c1 &= 0xFFF; + c2 = 0x7F + c1 / 94; + c1 = 0x21 + c1 % 94; + } else { + if (encode_fallback) (*encode_fallback)(c1); + return; + } } } - return 0; -} - -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 (c2 == EOF) { + if (output_mode !=ASCII && output_mode!=ISO_8859_1) { + (*o_putc)(ESC); + (*o_putc)('('); + (*o_putc)(ascii_intro); + output_mode = ASCII; } - - if (iconv_func -#ifdef INPUT_CODE_FIX - && (f == -TRUE || !input_encoding) /* -TRUE means "FORCE" */ -#endif - ){ - 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); + (*o_putc)(EOF); +#ifdef X0212_ENABLE + } else if (is_eucg3(c2)){ + if(x0213_f){ + if(output_mode!=JIS_X_0213_2){ + output_mode = JIS_X_0213_2; + (*o_putc)(ESC); + (*o_putc)('$'); + (*o_putc)('('); + (*o_putc)(0x50); + } + }else{ + if(output_mode!=JIS_X_0212){ + output_mode = JIS_X_0212; + (*o_putc)(ESC); + (*o_putc)('$'); + (*o_putc)('('); + (*o_putc)(0x44); + } } - iconv_for_check = iconv; - } + (*o_putc)(c2 & 0x7f); + (*o_putc)(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_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_INIT (SCORE_iMIME) - -static const 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, -}; - -static const char score_table_F0[] = { - SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2, - SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, - SCORE_DEPEND, SCORE_DEPEND, SCORE_CP932, SCORE_CP932, - SCORE_CP932, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR, -}; - -void set_code_score(struct input_code *ptr, nkf_char score) -{ - if (ptr){ - ptr->score |= score; + } else if (c2==JIS_X_0201) { + if (output_mode!=JIS_X_0201) { + output_mode = JIS_X_0201; + (*o_putc)(ESC); + (*o_putc)('('); + (*o_putc)('I'); + } + (*o_putc)(c1); + } else if (c2==ISO_8859_1) { + /* iso8859 introduction, or 8th bit on */ + /* Can we convert in 7bit form using ESC-'-'-A ? + Is this popular? */ + output_mode = ISO_8859_1; + (*o_putc)(c1|0x80); + } else if (c2 == 0) { + if (output_mode !=ASCII && output_mode!=ISO_8859_1) { + (*o_putc)(ESC); + (*o_putc)('('); + (*o_putc)(ascii_intro); + output_mode = ASCII; + } + (*o_putc)(c1); + } else { + if(ms_ucs_map_f + ? c2<0x20 || 0x92score &= ~score; +#ifdef NUMCHAR_OPTION + if (c2 == 0 && is_unicode_capsule(c1)){ + w16e_conv(c1, &c2, &c1); + if (c2 == 0 && is_unicode_capsule(c1)){ + c2 = c1 & VALUE_MASK; + if (x0212_f && 0xE000 <= c2 && c2 <= 0xE757) { + /* eucJP-ms UDC */ + c1 &= 0xFFF; + c2 = c1 / 94; + c2 += c2 < 10 ? 0x75 : 0x8FEB; + c1 = 0x21 + c1 % 94; + if (is_eucg3(c2)){ + (*o_putc)(0x8f); + (*o_putc)((c2 & 0x7f) | 0x080); + (*o_putc)(c1 | 0x080); + }else{ + (*o_putc)((c2 & 0x7f) | 0x080); + (*o_putc)(c1 | 0x080); + } + return; + } else { + if (encode_fallback) (*encode_fallback)(c1); + return; + } + } } -} - -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); - }else if (c2 == SSO){ - set_code_score(ptr, SCORE_KANA); - }else if (c2 == 0x8f){ - set_code_score(ptr, SCORE_X0212); -#ifdef UTF8_OUTPUT_ENABLE - }else if (!e2w_conv(c2, c1)){ - set_code_score(ptr, SCORE_NO_EXIST); + if (c2 == EOF) { + (*o_putc)(EOF); + return; + } else if (c2 == 0) { + output_mode = ASCII; + (*o_putc)(c1); + } else if (c2 == JIS_X_0201) { + output_mode = EUC_JP; + (*o_putc)(SSO); (*o_putc)(c1|0x80); + } else if (c2 == ISO_8859_1) { + output_mode = ISO_8859_1; + (*o_putc)(c1 | 0x080); +#ifdef X0212_ENABLE + } else if (is_eucg3(c2)){ + output_mode = EUC_JP; +#ifdef SHIFTJIS_CP932 + if (!cp932inv_f){ + nkf_char s2, s1; + if (e2s_conv(c2, c1, &s2, &s1) == 0){ + s2e_conv(s2, s1, &c2, &c1); + } + } #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); + if (c2 == 0) { + output_mode = ASCII; + (*o_putc)(c1); + }else if (is_eucg3(c2)){ + 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 (!nkf_isgraph(c1) || !nkf_isgraph(c2)) { + set_iconv(FALSE, 0); + return; /* too late to rescue this char */ + } + output_mode = EUC_JP; + (*o_putc)(c2 | 0x080); + (*o_putc)(c1 | 0x080); } } -void status_disable(struct input_code *ptr) +void s_oconv(nkf_char c2, nkf_char c1) { - ptr->stat = -1; - ptr->buf[0] = -1; - code_score(ptr); - if (iconv == ptr->iconv_func) set_iconv(FALSE, 0); -} +#ifdef NUMCHAR_OPTION + if (c2 == 0 && is_unicode_capsule(c1)){ + w16e_conv(c1, &c2, &c1); + if (c2 == 0 && is_unicode_capsule(c1)){ + c2 = c1 & VALUE_MASK; + if (!x0213_f && 0xE000 <= c2 && c2 <= 0xE757) { + /* CP932 UDC */ + c1 &= 0xFFF; + c2 = c1 / 188 + (cp932inv_f ? 0xF0 : 0xEB); + c1 = c1 % 188; + c1 += 0x40 + (c1 > 0x3e); + (*o_putc)(c2); + (*o_putc)(c1); + return; + } else { + if(encode_fallback)(*encode_fallback)(c1); + return; + } + } + } +#endif + if (c2 == EOF) { + (*o_putc)(EOF); + return; + } else if (c2 == 0) { + output_mode = ASCII; + (*o_putc)(c1); + } else if (c2 == JIS_X_0201) { + output_mode = SHIFT_JIS; + (*o_putc)(c1|0x80); + } else if (c2 == ISO_8859_1) { + output_mode = ISO_8859_1; + (*o_putc)(c1 | 0x080); +#ifdef X0212_ENABLE + } else if (is_eucg3(c2)){ + output_mode = SHIFT_JIS; + if (e2s_conv(c2, c1, &c2, &c1) == 0){ + (*o_putc)(c2); + (*o_putc)(c1); + } +#endif + } else { + if (!nkf_isprint(c1) || !nkf_isprint(c2)) { + set_iconv(FALSE, 0); + return; /* too late to rescue this char */ + } + output_mode = SHIFT_JIS; + e2s_conv(c2, c1, &c2, &c1); -void status_push_ch(struct input_code *ptr, nkf_char c) -{ - ptr->buf[ptr->index++] = c; -} +#ifdef SHIFTJIS_CP932 + 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 */ -void status_clear(struct input_code *ptr) -{ - ptr->stat = 0; - ptr->index = 0; + (*o_putc)(c2); + if (prefix_table[(unsigned char)c1]){ + (*o_putc)(prefix_table[(unsigned char)c1]); + } + (*o_putc)(c1); + } } -void status_reset(struct input_code *ptr) +#ifdef UTF8_OUTPUT_ENABLE +void w_oconv(nkf_char c2, nkf_char c1) { - status_clear(ptr); - ptr->score = SCORE_INIT; -} + nkf_char c0; + nkf_char val; -void status_reinit(struct input_code *ptr) -{ - status_reset(ptr); - ptr->_file_stat = 0; -} + if (output_bom_f) { + output_bom_f = FALSE; + (*o_putc)('\357'); + (*o_putc)('\273'); + (*o_putc)('\277'); + } -void status_check(struct input_code *ptr, nkf_char c) -{ - if (c <= DEL && estab_f){ - status_reset(ptr); + if (c2 == EOF) { + (*o_putc)(EOF); + return; } -} -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; #ifdef NUMCHAR_OPTION - }else if (is_unicode_capsule(c)){ - break; + if (c2 == 0 && is_unicode_capsule(c1)){ + val = c1 & VALUE_MASK; + if (val < 0x80){ + (*o_putc)(val); + }else if (val < 0x800){ + (*o_putc)(0xC0 | (val >> 6)); + (*o_putc)(0x80 | (val & 0x3f)); + } else if (val <= NKF_INT32_C(0xFFFF)) { + (*o_putc)(0xE0 | (val >> 12)); + (*o_putc)(0x80 | ((val >> 6) & 0x3f)); + (*o_putc)(0x80 | (val & 0x3f)); + } else if (val <= NKF_INT32_C(0x10FFFF)) { + (*o_putc)(0xF0 | ( val>>18)); + (*o_putc)(0x80 | ((val>>12) & 0x3f)); + (*o_putc)(0x80 | ((val>> 6) & 0x3f)); + (*o_putc)(0x80 | ( val & 0x3f)); + } + return; + } #endif - }else if (0xa1 <= c && c <= 0xdf){ - status_push_ch(ptr, SSO); - 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); -#endif /* SHIFTJIS_CP932 */ -#ifdef X0212_ENABLE - }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: -#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_clear(ptr); - break; - } - } -#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]); - set_code_score(ptr, SCORE_CP932); - status_clear(ptr); - }else{ - status_disable(ptr); - } - break; + + if (c2 == 0) { + output_mode = ASCII; + (*o_putc)(c1); + } else if (c2 == ISO_8859_1) { + output_mode = UTF_8; + (*o_putc)(c1 | 0x080); + } else { + output_mode = UTF_8; + val = e2w_conv(c2, c1); + if (val){ + w16w_conv(val, &c2, &c1, &c0); + (*o_putc)(c2); + if (c1){ + (*o_putc)(c1); + if (c0) (*o_putc)(c0); + } + } } } -void e_status(struct input_code *ptr, nkf_char c) +void w_oconv16(nkf_char c2, nkf_char c1) { - switch (ptr->stat){ - case -1: - status_check(ptr, c); - break; - case 0: - if (c <= DEL){ - break; -#ifdef NUMCHAR_OPTION - }else if (is_unicode_capsule(c)){ - break; -#endif - }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); - } - 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); - } -#endif /* X0212_ENABLE */ + if (output_bom_f) { + output_bom_f = FALSE; + if (output_endian == ENDIAN_LITTLE){ + (*o_putc)((unsigned char)'\377'); + (*o_putc)('\376'); + }else{ + (*o_putc)('\376'); + (*o_putc)((unsigned char)'\377'); + } } -} -#ifdef UTF8_INPUT_ENABLE -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; + if (c2 == EOF) { + (*o_putc)(EOF); + return; + } + + if (c2 == ISO_8859_1) { + c2 = 0; + c1 |= 0x80; #ifdef NUMCHAR_OPTION - }else if (is_unicode_capsule(c)){ - break; + } else if (c2 == 0 && is_unicode_capsule(c1)) { + if (is_unicode_bmp(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; + } #endif - }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); - } - }else{ - status_disable(ptr); - } - break; + } else if (c2) { + nkf_char val = e2w_conv(c2, c1); + c2 = (val >> 8) & 0xff; + c1 = val & 0xff; + if (!val) return; + } + if (output_endian == ENDIAN_LITTLE){ + (*o_putc)(c1); + (*o_putc)(c2); + }else{ + (*o_putc)(c2); + (*o_putc)(c1); } } -#endif -void code_status(nkf_char c) +void w_oconv32(nkf_char c2, nkf_char c1) { - int action_flag = 1; - struct input_code *result = 0; - struct input_code *p = input_code_list; - while (p->name){ - if (!p->status_func) { - ++p; - continue; - } - 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; - } + if (output_bom_f) { + output_bom_f = FALSE; + if (output_endian == ENDIAN_LITTLE){ + (*o_putc)((unsigned char)'\377'); + (*o_putc)('\376'); + (*o_putc)('\000'); + (*o_putc)('\000'); + }else{ + (*o_putc)('\000'); + (*o_putc)('\000'); + (*o_putc)('\376'); + (*o_putc)((unsigned char)'\377'); } - ++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 (c2 == EOF) { + (*o_putc)(EOF); + return; + } + + if (c2 == ISO_8859_1) { + c1 |= 0x80; +#ifdef NUMCHAR_OPTION + } else if (c2 == 0 && is_unicode_capsule(c1)) { + c1 &= VALUE_MASK; +#endif + } else if (c2) { + c1 = e2w_conv(c2, c1); + if (!c1) return; + } + if (output_endian == ENDIAN_LITTLE){ + (*o_putc)( c1 & NKF_INT32_C(0x000000FF)); + (*o_putc)((c1 & NKF_INT32_C(0x0000FF00)) >> 8); + (*o_putc)((c1 & NKF_INT32_C(0x00FF0000)) >> 16); + (*o_putc)('\000'); + }else{ + (*o_putc)('\000'); + (*o_putc)((c1 & NKF_INT32_C(0x00FF0000)) >> 16); + (*o_putc)((c1 & NKF_INT32_C(0x0000FF00)) >> 8); + (*o_putc)( c1 & NKF_INT32_C(0x000000FF)); } } +#endif -#ifndef WIN32DLL -nkf_char std_getc(FILE *f) +#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_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_INIT (SCORE_iMIME) + +static const 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, +}; + +static const char score_table_F0[] = { + SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2, + SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, + SCORE_DEPEND, SCORE_DEPEND, SCORE_CP932, SCORE_CP932, + SCORE_CP932, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR, +}; + +void set_code_score(struct input_code *ptr, nkf_char score) { - if (std_gc_ndx){ - return std_gc_buf[--std_gc_ndx]; + if (ptr){ + ptr->score |= score; } - return getc(f); } -#endif /*WIN32DLL*/ -nkf_char std_ungetc(nkf_char c, FILE *f) +void clr_code_score(struct input_code *ptr, nkf_char score) { - if (std_gc_ndx == STD_GC_BUFSIZE){ - return EOF; + if (ptr){ + ptr->score &= ~score; } - std_gc_buf[std_gc_ndx++] = c; - return c; } -#ifndef WIN32DLL -void std_putc(nkf_char c) +void code_score(struct input_code *ptr) { - if(c!=EOF) - putchar(c); + 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); + }else if (c2 == SSO){ + set_code_score(ptr, SCORE_KANA); + }else if (c2 == 0x8f){ + set_code_score(ptr, SCORE_X0212); +#ifdef UTF8_OUTPUT_ENABLE + }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); + } } -#endif /*WIN32DLL*/ -nkf_char -h_conv(FILE *f, nkf_char c2, nkf_char c1) +void status_disable(struct input_code *ptr) { - nkf_char ret, c3, c0; - int hold_index; + ptr->stat = -1; + ptr->buf[0] = -1; + code_score(ptr); + if (iconv == ptr->iconv_func) set_iconv(FALSE, 0); +} +void status_push_ch(struct input_code *ptr, nkf_char c) +{ + ptr->buf[ptr->index++] = c; +} - /** it must NOT be in the kanji shifte sequence */ - /** it must NOT be written in JIS7 */ - /** and it must be after 2 byte 8bit code */ +void status_clear(struct input_code *ptr) +{ + ptr->stat = 0; + ptr->index = 0; +} - hold_count = 0; - push_hold_buf(c2); - push_hold_buf(c1); +void status_reset(struct input_code *ptr) +{ + status_clear(ptr); + ptr->score = SCORE_INIT; +} - while ((c1 = (*i_getc)(f)) != EOF) { - if (c1 == ESC){ - (*i_ungetc)(c1,f); - break; - } - code_status(c1); - if (push_hold_buf(c1) == EOF || estab_f){ - break; - } - } +void status_reinit(struct input_code *ptr) +{ + status_reset(ptr); + ptr->_file_stat = 0; +} - if (!estab_f){ - struct input_code *p = input_code_list; - struct input_code *result = p; - if (c1 == EOF){ - code_status(c1); - } - while (p->name){ - if (p->status_func && p->score < result->score){ - result = p; - } - ++p; - } - set_iconv(TRUE, result->iconv_func); +void status_check(struct input_code *ptr, nkf_char c) +{ + if (c <= DEL && estab_f){ + status_reset(ptr); } +} - - /** now, - ** 1) EOF is detected, or - ** 2) Code is established, or - ** 3) Buffer is FULL (but last word is pushed) - ** - ** in 1) and 3) cases, we continue to use - ** Kanji codes by oconv and leave estab_f unchanged. - **/ - - ret = c1; - hold_index = 0; - while (hold_index < hold_count){ - c2 = hold_buf[hold_index++]; - if (c2 <= DEL +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; #ifdef NUMCHAR_OPTION - || is_unicode_capsule(c2) + }else if (is_unicode_capsule(c)){ + break; #endif - ){ - (*iconv)(0, c2, 0); - continue; - }else if (iconv == s_iconv && 0xa1 <= c2 && c2 <= 0xdf){ - (*iconv)(JIS_X_0201, c2, 0); - continue; - } - if (hold_index < hold_count){ - c1 = hold_buf[hold_index++]; - }else{ - c1 = (*i_getc)(f); - if (c1 == EOF){ - c3 = EOF; - break; - } - code_status(c1); - } - c0 = 0; - switch ((*iconv)(c2, c1, 0)) { /* can be EUC/SJIS/UTF-8 */ - case -2: - /* 4 bytes UTF-8 */ - if (hold_index < hold_count){ - c0 = hold_buf[hold_index++]; - } else if ((c0 = (*i_getc)(f)) == EOF) { - ret = EOF; - break; - } else { - code_status(c0); - c0 <<= 8; - if (hold_index < hold_count){ - c3 = hold_buf[hold_index++]; - } else if ((c3 = (*i_getc)(f)) == EOF) { - c0 = ret = EOF; - break; - } else { - code_status(c3); - (*iconv)(c2, c1, c0|c3); - } - } - break; - case -1: - /* 3 bytes EUC or UTF-8 */ - if (hold_index < hold_count){ - c0 = hold_buf[hold_index++]; - } else if ((c0 = (*i_getc)(f)) == EOF) { - ret = EOF; + }else if (0xa1 <= c && c <= 0xdf){ + status_push_ch(ptr, SSO); + 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); +#endif /* SHIFTJIS_CP932 */ +#ifdef X0212_ENABLE + }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: +#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_clear(ptr); break; - } else { - code_status(c0); - } - (*iconv)(c2, c1, c0); - break; - } - if (c0 == EOF) break; - } - return ret; -} - -/* - * Check and Ignore BOM - */ -void check_bom(FILE *f) -{ - int c2; - switch(c2 = (*i_getc)(f)){ - case 0x00: - if((c2 = (*i_getc)(f)) == 0x00){ - if((c2 = (*i_getc)(f)) == 0xFE){ - if((c2 = (*i_getc)(f)) == 0xFF){ - if(!input_encoding){ - set_iconv(TRUE, w_iconv32); - } - if (iconv == w_iconv32) { - input_endian = ENDIAN_BIG; - return; - } - (*i_ungetc)(0xFF,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0xFE,f); - }else if(c2 == 0xFF){ - if((c2 = (*i_getc)(f)) == 0xFE){ - if(!input_encoding){ - set_iconv(TRUE, w_iconv32); - } - if (iconv == w_iconv32) { - input_endian = ENDIAN_2143; - return; - } - (*i_ungetc)(0xFF,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0xFF,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0x00,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0x00,f); - break; - case 0xEF: - if((c2 = (*i_getc)(f)) == 0xBB){ - if((c2 = (*i_getc)(f)) == 0xBF){ - if(!input_encoding){ - set_iconv(TRUE, w_iconv); - } - if (iconv == w_iconv) { - return; - } - (*i_ungetc)(0xBF,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0xBB,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0xEF,f); - break; - case 0xFE: - if((c2 = (*i_getc)(f)) == 0xFF){ - if((c2 = (*i_getc)(f)) == 0x00){ - if((c2 = (*i_getc)(f)) == 0x00){ - if(!input_encoding){ - set_iconv(TRUE, w_iconv32); - } - if (iconv == w_iconv32) { - input_endian = ENDIAN_3412; - return; - } - (*i_ungetc)(0x00,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0x00,f); - }else (*i_ungetc)(c2,f); - if(!input_encoding){ - set_iconv(TRUE, w_iconv16); - } - if (iconv == w_iconv16) { - input_endian = ENDIAN_BIG; - return; - } - (*i_ungetc)(0xFF,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0xFE,f); - break; - case 0xFF: - if((c2 = (*i_getc)(f)) == 0xFE){ - if((c2 = (*i_getc)(f)) == 0x00){ - if((c2 = (*i_getc)(f)) == 0x00){ - if(!input_encoding){ - set_iconv(TRUE, w_iconv32); - } - if (iconv == w_iconv32) { - input_endian = ENDIAN_LITTLE; - return; - } - (*i_ungetc)(0x00,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0x00,f); - }else (*i_ungetc)(c2,f); - if(!input_encoding){ - set_iconv(TRUE, w_iconv16); - } - if (iconv == w_iconv16) { - input_endian = ENDIAN_LITTLE; - return; } - (*i_ungetc)(0xFE,f); - }else (*i_ungetc)(c2,f); - (*i_ungetc)(0xFF,f); - break; - default: - (*i_ungetc)(c2,f); - break; + } +#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]); + set_code_score(ptr, SCORE_CP932); + status_clear(ptr); + }else{ + status_disable(ptr); + } + break; } } -nkf_char push_hold_buf(nkf_char c2) +void e_status(struct input_code *ptr, nkf_char c) { - if (hold_count >= HOLD_SIZE*2) - return (EOF); - hold_buf[hold_count++] = (unsigned char)c2; - return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count); -} - + switch (ptr->stat){ + case -1: + status_check(ptr, c); + break; + case 0: + if (c <= DEL){ + break; +#ifdef NUMCHAR_OPTION + }else if (is_unicode_capsule(c)){ + break; +#endif + }else if (SSO == c || (0xa1 <= c && c <= 0xfe)){ + ptr->stat = 1; + status_push_ch(ptr, c); #ifdef X0212_ENABLE -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); - } - }else{ - if (0x75 <= c && c <= 0x7f){ - ret = c + (0x113 - 0x75); - } + }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; +#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 */ } - return ret; } - -nkf_char x0212_unshift(nkf_char c) +#ifdef UTF8_INPUT_ENABLE +void w_status(struct input_code *ptr, nkf_char c) { - nkf_char ret = c; - if (0x7f <= c && c <= 0x88){ - ret = c + (0x75 - 0x7f); - }else if (0x89 <= c && c <= 0x92){ - ret = PREFIX_EUCG3 | 0x80 | (c + (0x75 - 0x89)); + switch (ptr->stat){ + case -1: + status_check(ptr, c); + break; + case 0: + if (c <= DEL){ + break; +#ifdef NUMCHAR_OPTION + }else if (is_unicode_capsule(c)){ + break; +#endif + }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); + } + }else{ + status_disable(ptr); + } + break; } - return ret; } -#endif /* X0212_ENABLE */ +#endif -nkf_char s2e_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) +void code_status(nkf_char c) { -#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} }; -#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 - && 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 */ -#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; + int action_flag = 1; + struct input_code *result = 0; + struct input_code *p = input_code_list; + while (p->name){ + if (!p->status_func) { + ++p; + continue; + } + 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{ - c2 = val >> 8; - c1 = val & 0xff; + result = p; } - if (p2) *p2 = c2; - if (p1) *p1 = c1; - return 0; } - } -#endif - if(c2 >= 0x80){ - if(x0213_f && c2 >= 0xF0){ - if(c2 <= 0xF3 || (c2 == 0xF4 && c1 < 0x9F)){ /* k=1, 3<=k<=5, k=8, 12<=k<=15 */ - c2 = PREFIX_EUCG3 | 0x20 | shift_jisx0213_s1a3_table[c2 - 0xF0][0x9E < c1]; - }else{ /* 78<=k<=94 */ - c2 = PREFIX_EUCG3 | (c2 * 2 - 0x17B); - if (0x9E < c1) c2++; - } - }else{ -#define SJ0162 0x00e1 /* 01 - 62 ku offset */ -#define SJ6394 0x0161 /* 63 - 94 ku offset */ - c2 = c2 + c2 - ((c2 <= 0x9F) ? SJ0162 : SJ6394); - if (0x9E < c1) c2++; - } - if (c1 < 0x9F) - c1 = c1 - ((c1 > DEL) ? SP : 0x1F); - else { - c1 = c1 - 0x7E; - } + ++p; } -#ifdef X0212_ENABLE - c2 = x0212_unshift(c2); -#endif - if (p2) *p2 = c2; - if (p1) *p1 = c1; - return 0; + 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; + } + } + } } -nkf_char s_iconv(nkf_char c2, nkf_char c1, nkf_char c0) +#ifndef WIN32DLL +nkf_char std_getc(FILE *f) { - if (c2 == JIS_X_0201) { - c1 &= 0x7f; - } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) { - /* NOP */ - } else if (!x0213_f && 0xF0 <= c2 && c2 <= 0xF9 && 0x40 <= c1 && c1 <= 0xFC) { - /* CP932 UDC */ - if(c1 == 0x7F) return 0; - c1 = (c2 - 0xF0) * 188 + (c1 - 0x40 - (0x7E < c1)) + 0xE000 + CLASS_UNICODE; - c2 = 0; - } else { - nkf_char ret = s2e_conv(c2, c1, &c2, &c1); - if (ret) return ret; + if (std_gc_ndx){ + return std_gc_buf[--std_gc_ndx]; } - (*oconv)(c2, c1); - return 0; + return getc(f); } +#endif /*WIN32DLL*/ -nkf_char e_iconv(nkf_char c2, nkf_char c1, nkf_char c0) +nkf_char std_ungetc(nkf_char c, FILE *f) { - if (c2 == JIS_X_0201) { - c1 &= 0x7f; -#ifdef X0212_ENABLE - }else if (c2 == 0x8f){ - 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 = (c1 - 0xF5) * 94 + c0 - 0xA1 + 0xE3AC + CLASS_UNICODE; - c2 = 0; - } else { - c2 = (c2 << 8) | (c1 & 0x7f); - c1 = c0 & 0x7f; -#ifdef SHIFTJIS_CP932 - if (cp51932_f){ - nkf_char s2, s1; - if (e2s_conv(c2, c1, &s2, &s1) == 0){ - s2e_conv(s2, s1, &c2, &c1); - if (c2 < 0x100){ - c1 &= 0x7f; - c2 &= 0x7f; - } - } - } -#endif /* SHIFTJIS_CP932 */ - } -#endif /* X0212_ENABLE */ - } else if (c2 == SSO){ - c2 = JIS_X_0201; - c1 &= 0x7f; - } else if ((c2 == EOF) || (c2 == 0) || c2 < SP) { - /* 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 */ - c1 = (c2 - 0xF5) * 94 + c1 - 0xA1 + 0xE000 + CLASS_UNICODE; - c2 = 0; - } else { - c1 &= 0x7f; - c2 &= 0x7f; -#ifdef SHIFTJIS_CP932 - if (cp51932_f && 0x79 <= c2 && c2 <= 0x7c){ - nkf_char s2, s1; - if (e2s_conv(c2, c1, &s2, &s1) == 0){ - s2e_conv(s2, s1, &c2, &c1); - if (c2 < 0x100){ - c1 &= 0x7f; - c2 &= 0x7f; - } - } - } -#endif /* SHIFTJIS_CP932 */ - } + if (std_gc_ndx == STD_GC_BUFSIZE){ + return EOF; } - (*oconv)(c2, c1); - return 0; + std_gc_buf[std_gc_ndx++] = c; + return c; } -#if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE) -void w16w_conv(nkf_char val, nkf_char *p2, nkf_char *p1, nkf_char *p0) +#ifndef WIN32DLL +void std_putc(nkf_char c) { - val &= VALUE_MASK; - if (val < 0x80){ - *p2 = val; - *p1 = 0; - *p0 = 0; - }else if (val < 0x800){ - *p2 = 0xc0 | (val >> 6); - *p1 = 0x80 | (val & 0x3f); - *p0 = 0; - } else if (val <= NKF_INT32_C(0xFFFF)) { - *p2 = 0xe0 | (val >> 12); - *p1 = 0x80 | ((val >> 6) & 0x3f); - *p0 = 0x80 | (val & 0x3f); - } else if (val <= NKF_INT32_C(0x10FFFF)) { - *p2 = 0xe0 | (val >> 16); - *p1 = 0x80 | ((val >> 12) & 0x3f); - *p0 = 0x8080 | ((val << 2) & 0x3f00)| (val & 0x3f); - } else { - *p2 = 0; - *p1 = 0; - *p0 = 0; - } -} -#endif - -#ifdef UTF8_INPUT_ENABLE -nkf_char ww16_conv(nkf_char c2, nkf_char c1, nkf_char c0) -{ - nkf_char val; - if (c2 >= 0xf8) { - val = -1; - } else if (c2 >= 0xf0){ - /* c2: 1st, c1: 2nd, c0: 3rd/4th */ - val = (c2 & 0x0f) << 18; - val |= (c1 & 0x3f) << 12; - val |= (c0 & 0x3f00) >> 2; - val |= (c0 & 0x3f); - }else if (c2 >= 0xe0){ - val = (c2 & 0x0f) << 12; - val |= (c1 & 0x3f) << 6; - val |= (c0 & 0x3f); - }else if (c2 >= 0xc0){ - val = (c2 & 0x1f) << 6; - val |= (c1 & 0x3f); - }else{ - val = c2; - } - return val; + if(c!=EOF) + putchar(c); } +#endif /*WIN32DLL*/ -nkf_char w16e_conv(nkf_char val, nkf_char *p2, nkf_char *p1) +static unsigned char hold_buf[HOLD_SIZE*2]; +static int hold_count = 0; +nkf_char push_hold_buf(nkf_char c2) { - nkf_char c2, c1, c0; - nkf_char ret = 0; - val &= VALUE_MASK; - if (val < 0x80){ - *p2 = 0; - *p1 = val; - }else{ - w16w_conv(val, &c2, &c1, &c0); - ret = unicode_to_jis_common(c2, c1, c0, p2, p1); -#ifdef NUMCHAR_OPTION - if (ret > 0){ - *p2 = 0; - *p1 = CLASS_UNICODE | val; - ret = 0; - } -#endif - } - return ret; + if (hold_count >= HOLD_SIZE*2) + return (EOF); + hold_buf[hold_count++] = (unsigned char)c2; + return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count); } -#endif -#ifdef UTF8_OUTPUT_ENABLE -nkf_char e2w_conv(nkf_char c2, nkf_char c1) +nkf_char +h_conv(FILE *f, nkf_char c2, nkf_char c1) { - const unsigned short *p; - - if (c2 == JIS_X_0201) { - if (ms_ucs_map_f == UCS_MAP_CP10001) { - switch (c1) { - case 0x20: - return 0xA0; - case 0x7D: - return 0xA9; - } - } - 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> 6)); - (*o_putc)(0x80 | (val & 0x3f)); - } else if (val <= NKF_INT32_C(0xFFFF)) { - (*o_putc)(0xE0 | (val >> 12)); - (*o_putc)(0x80 | ((val >> 6) & 0x3f)); - (*o_putc)(0x80 | (val & 0x3f)); - } else if (val <= NKF_INT32_C(0x10FFFF)) { - (*o_putc)(0xF0 | ( val>>18)); - (*o_putc)(0x80 | ((val>>12) & 0x3f)); - (*o_putc)(0x80 | ((val>> 6) & 0x3f)); - (*o_putc)(0x80 | ( val & 0x3f)); + while ((c1 = (*i_getc)(f)) != EOF) { + if (c1 == ESC){ + (*i_ungetc)(c1,f); + break; + } + code_status(c1); + if (push_hold_buf(c1) == EOF || estab_f){ + break; } - return; } -#endif - if (c2 == 0) { - output_mode = ASCII; - (*o_putc)(c1); - } else if (c2 == ISO_8859_1) { - output_mode = UTF_8; - (*o_putc)(c1 | 0x080); - } else { - output_mode = UTF_8; - val = e2w_conv(c2, c1); - if (val){ - w16w_conv(val, &c2, &c1, &c0); - (*o_putc)(c2); - if (c1){ - (*o_putc)(c1); - if (c0) (*o_putc)(c0); + if (!estab_f){ + struct input_code *p = input_code_list; + struct input_code *result = p; + if (c1 == EOF){ + code_status(c1); + } + while (p->name){ + if (p->status_func && p->score < result->score){ + result = p; } + ++p; } + set_iconv(TRUE, result->iconv_func); } -} -void w_oconv16(nkf_char c2, nkf_char c1) -{ - if (output_bom_f) { - output_bom_f = FALSE; - if (output_endian == ENDIAN_LITTLE){ - (*o_putc)((unsigned char)'\377'); - (*o_putc)('\376'); - }else{ - (*o_putc)('\376'); - (*o_putc)((unsigned char)'\377'); - } - } - if (c2 == EOF) { - (*o_putc)(EOF); - return; - } + /** now, + ** 1) EOF is detected, or + ** 2) Code is established, or + ** 3) Buffer is FULL (but last word is pushed) + ** + ** in 1) and 3) cases, we continue to use + ** Kanji codes by oconv and leave estab_f unchanged. + **/ - if (c2 == ISO_8859_1) { - c2 = 0; - c1 |= 0x80; + ret = c1; + hold_index = 0; + while (hold_index < hold_count){ + c2 = hold_buf[hold_index++]; + if (c2 <= DEL #ifdef NUMCHAR_OPTION - } else if (c2 == 0 && is_unicode_capsule(c1)) { - if (is_unicode_bmp(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); - } + || is_unicode_capsule(c2) +#endif + ){ + (*iconv)(0, c2, 0); + continue; + }else if (iconv == s_iconv && 0xa1 <= c2 && c2 <= 0xdf){ + (*iconv)(JIS_X_0201, c2, 0); + continue; + } + if (hold_index < hold_count){ + c1 = hold_buf[hold_index++]; + }else{ + c1 = (*i_getc)(f); + if (c1 == EOF){ + c3 = EOF; + break; } - return; + code_status(c1); } -#endif - } else if (c2) { - nkf_char val = e2w_conv(c2, c1); - c2 = (val >> 8) & 0xff; - c1 = val & 0xff; - if (!val) return; - } - if (output_endian == ENDIAN_LITTLE){ - (*o_putc)(c1); - (*o_putc)(c2); - }else{ - (*o_putc)(c2); - (*o_putc)(c1); + c0 = 0; + switch ((*iconv)(c2, c1, 0)) { /* can be EUC/SJIS/UTF-8 */ + case -2: + /* 4 bytes UTF-8 */ + if (hold_index < hold_count){ + c0 = hold_buf[hold_index++]; + } else if ((c0 = (*i_getc)(f)) == EOF) { + ret = EOF; + break; + } else { + code_status(c0); + c0 <<= 8; + if (hold_index < hold_count){ + c3 = hold_buf[hold_index++]; + } else if ((c3 = (*i_getc)(f)) == EOF) { + c0 = ret = EOF; + break; + } else { + code_status(c3); + (*iconv)(c2, c1, c0|c3); + } + } + break; + case -1: + /* 3 bytes EUC or UTF-8 */ + if (hold_index < hold_count){ + c0 = hold_buf[hold_index++]; + } else if ((c0 = (*i_getc)(f)) == EOF) { + ret = EOF; + break; + } else { + code_status(c0); + } + (*iconv)(c2, c1, c0); + break; + } + if (c0 == EOF) break; } + return ret; } -void w_oconv32(nkf_char c2, nkf_char c1) +/* + * Check and Ignore BOM + */ +void check_bom(FILE *f) { - if (output_bom_f) { - output_bom_f = FALSE; - if (output_endian == ENDIAN_LITTLE){ - (*o_putc)((unsigned char)'\377'); - (*o_putc)('\376'); - (*o_putc)('\000'); - (*o_putc)('\000'); - }else{ - (*o_putc)('\000'); - (*o_putc)('\000'); - (*o_putc)('\376'); - (*o_putc)((unsigned char)'\377'); - } + int c2; + switch(c2 = (*i_getc)(f)){ + case 0x00: + if((c2 = (*i_getc)(f)) == 0x00){ + if((c2 = (*i_getc)(f)) == 0xFE){ + if((c2 = (*i_getc)(f)) == 0xFF){ + if(!input_encoding){ + set_iconv(TRUE, w_iconv32); + } + if (iconv == w_iconv32) { + input_endian = ENDIAN_BIG; + return; + } + (*i_ungetc)(0xFF,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0xFE,f); + }else if(c2 == 0xFF){ + if((c2 = (*i_getc)(f)) == 0xFE){ + if(!input_encoding){ + set_iconv(TRUE, w_iconv32); + } + if (iconv == w_iconv32) { + input_endian = ENDIAN_2143; + return; + } + (*i_ungetc)(0xFF,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0xFF,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0x00,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0x00,f); + break; + case 0xEF: + if((c2 = (*i_getc)(f)) == 0xBB){ + if((c2 = (*i_getc)(f)) == 0xBF){ + if(!input_encoding){ + set_iconv(TRUE, w_iconv); + } + if (iconv == w_iconv) { + return; + } + (*i_ungetc)(0xBF,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0xBB,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0xEF,f); + break; + case 0xFE: + if((c2 = (*i_getc)(f)) == 0xFF){ + if((c2 = (*i_getc)(f)) == 0x00){ + if((c2 = (*i_getc)(f)) == 0x00){ + if(!input_encoding){ + set_iconv(TRUE, w_iconv32); + } + if (iconv == w_iconv32) { + input_endian = ENDIAN_3412; + return; + } + (*i_ungetc)(0x00,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0x00,f); + }else (*i_ungetc)(c2,f); + if(!input_encoding){ + set_iconv(TRUE, w_iconv16); + } + if (iconv == w_iconv16) { + input_endian = ENDIAN_BIG; + return; + } + (*i_ungetc)(0xFF,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0xFE,f); + break; + case 0xFF: + if((c2 = (*i_getc)(f)) == 0xFE){ + if((c2 = (*i_getc)(f)) == 0x00){ + if((c2 = (*i_getc)(f)) == 0x00){ + if(!input_encoding){ + set_iconv(TRUE, w_iconv32); + } + if (iconv == w_iconv32) { + input_endian = ENDIAN_LITTLE; + return; + } + (*i_ungetc)(0x00,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0x00,f); + }else (*i_ungetc)(c2,f); + if(!input_encoding){ + set_iconv(TRUE, w_iconv16); + } + if (iconv == w_iconv16) { + input_endian = ENDIAN_LITTLE; + return; + } + (*i_ungetc)(0xFE,f); + }else (*i_ungetc)(c2,f); + (*i_ungetc)(0xFF,f); + break; + default: + (*i_ungetc)(c2,f); + break; } +} - if (c2 == EOF) { - (*o_putc)(EOF); - return; - } +static struct { + int count; + nkf_char status; + nkf_char buf[3]; +} broken_state; - if (c2 == ISO_8859_1) { - c1 |= 0x80; -#ifdef NUMCHAR_OPTION - } else if (c2 == 0 && is_unicode_capsule(c1)) { - c1 &= VALUE_MASK; -#endif - } else if (c2) { - c1 = e2w_conv(c2, c1); - if (!c1) return; +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) +{ + nkf_char c, c1; + + if (broken_state.count > 0) { + return pop_broken_buf(); } - if (output_endian == ENDIAN_LITTLE){ - (*o_putc)( c1 & NKF_INT32_C(0x000000FF)); - (*o_putc)((c1 & NKF_INT32_C(0x0000FF00)) >> 8); - (*o_putc)((c1 & NKF_INT32_C(0x00FF0000)) >> 16); - (*o_putc)('\000'); - }else{ - (*o_putc)('\000'); - (*o_putc)((c1 & NKF_INT32_C(0x00FF0000)) >> 16); - (*o_putc)((c1 & NKF_INT32_C(0x0000FF00)) >> 8); - (*o_putc)( c1 & NKF_INT32_C(0x000000FF)); + c = (*i_bgetc)(f); + if (c=='$' && broken_state.status != ESC + && (input_mode==ASCII || input_mode==JIS_X_0201)) { + c1= (*i_bgetc)(f); + broken_state.status = 0; + if (c1=='@'|| c1=='B') { + push_broken_buf(c1); + push_broken_buf(c); + return ESC; + } else { + (*i_bungetc)(c1,f); + return c; + } + } else if (c=='(' && broken_state.status != ESC + && (input_mode==JIS_X_0208 || input_mode==JIS_X_0201)) { /* ) */ + c1= (*i_bgetc)(f); + broken_state.status = 0; + if (c1=='J'|| c1=='B') { + push_broken_buf(c1); + push_broken_buf(c); + return ESC; + } else { + (*i_bungetc)(c1,f); + return c; + } + } else { + broken_state.status = c; + return c; } } -#endif -void e_oconv(nkf_char c2, nkf_char c1) +nkf_char broken_ungetc(nkf_char c, FILE *f) { -#ifdef NUMCHAR_OPTION - if (c2 == 0 && is_unicode_capsule(c1)){ - w16e_conv(c1, &c2, &c1); - if (c2 == 0 && is_unicode_capsule(c1)){ - c2 = c1 & VALUE_MASK; - if (x0212_f && 0xE000 <= c2 && c2 <= 0xE757) { - /* eucJP-ms UDC */ - c1 &= 0xFFF; - c2 = c1 / 94; - c2 += c2 < 10 ? 0x75 : 0x8FEB; - c1 = 0x21 + c1 % 94; - if (is_eucg3(c2)){ - (*o_putc)(0x8f); - (*o_putc)((c2 & 0x7f) | 0x080); - (*o_putc)(c1 | 0x080); - }else{ - (*o_putc)((c2 & 0x7f) | 0x080); - (*o_putc)(c1 | 0x080); - } - return; - } else { - if (encode_fallback) (*encode_fallback)(c1); - return; - } - } - } -#endif - if (c2 == EOF) { - (*o_putc)(EOF); - return; - } else if (c2 == 0) { - output_mode = ASCII; - (*o_putc)(c1); - } else if (c2 == JIS_X_0201) { - output_mode = EUC_JP; - (*o_putc)(SSO); (*o_putc)(c1|0x80); - } else if (c2 == ISO_8859_1) { - output_mode = ISO_8859_1; - (*o_putc)(c1 | 0x080); -#ifdef X0212_ENABLE - } else if (is_eucg3(c2)){ - output_mode = EUC_JP; -#ifdef SHIFTJIS_CP932 - if (!cp932inv_f){ - nkf_char s2, s1; - if (e2s_conv(c2, c1, &s2, &s1) == 0){ - s2e_conv(s2, s1, &c2, &c1); - } - } -#endif - if (c2 == 0) { - output_mode = ASCII; - (*o_putc)(c1); - }else if (is_eucg3(c2)){ - 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 (!nkf_isgraph(c1) || !nkf_isgraph(c2)) { - set_iconv(FALSE, 0); - return; /* too late to rescue this char */ - } - output_mode = EUC_JP; - (*o_putc)(c2 | 0x080); - (*o_putc)(c1 | 0x080); - } + if (broken_state.count < 2) + push_broken_buf(c); + return c; } -nkf_char e2s_conv(nkf_char c2, nkf_char c1, nkf_char *p2, nkf_char *p1) +void eol_conv(nkf_char c2, nkf_char c1) { - nkf_char ndx; - if (is_eucg3(c2)){ - ndx = c2 & 0x7f; - if (x0213_f){ - 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); - return 0; - }else if(0x6E <= ndx && ndx <= 0x7E){ - if (p2) *p2 = ((ndx - 1) >> 1) + 0xbe; - if (p1) *p1 = c1 + ((ndx & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e); - return 0; - } - return 1; - } -#ifdef X0212_ENABLE - else if(nkf_isgraph(ndx)){ - nkf_char val = 0; - const unsigned short *ptr; - 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 (guess_f && input_eol != EOF) { + if (c2 == 0 && c1 == LF) { + if (!input_eol) input_eol = prev_cr ? CRLF : LF; + else if (input_eol != (prev_cr ? CRLF : LF)) input_eol = EOF; + } else if (c2 == 0 && c1 == CR && input_eol == LF) input_eol = EOF; + else if (!prev_cr); + else if (!input_eol) input_eol = CR; + else if (input_eol != CR) input_eol = EOF; } - if(0x7F < c2) 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; + if (prev_cr || (c2 == 0 && c1 == LF)) { + prev_cr = 0; + if (eolmode_f != LF) (*o_eol_conv)(0, CR); + if (eolmode_f != CR) (*o_eol_conv)(0, LF); + } + if (c2 == 0 && c1 == CR) prev_cr = CR; + else if (c2 != 0 || c1 != LF) (*o_eol_conv)(c2, c1); } -void s_oconv(nkf_char c2, nkf_char c1) -{ -#ifdef NUMCHAR_OPTION - if (c2 == 0 && is_unicode_capsule(c1)){ - w16e_conv(c1, &c2, &c1); - if (c2 == 0 && is_unicode_capsule(c1)){ - c2 = c1 & VALUE_MASK; - if (!x0213_f && 0xE000 <= c2 && c2 <= 0xE757) { - /* CP932 UDC */ - c1 &= 0xFFF; - c2 = c1 / 188 + (cp932inv_f ? 0xF0 : 0xEB); - c1 = c1 % 188; - c1 += 0x40 + (c1 > 0x3e); - (*o_putc)(c2); - (*o_putc)(c1); - return; - } else { - if(encode_fallback)(*encode_fallback)(c1); - return; - } - } - } -#endif - if (c2 == EOF) { - (*o_putc)(EOF); - return; - } else if (c2 == 0) { - output_mode = ASCII; - (*o_putc)(c1); - } else if (c2 == JIS_X_0201) { - output_mode = SHIFT_JIS; - (*o_putc)(c1|0x80); - } else if (c2 == ISO_8859_1) { - output_mode = ISO_8859_1; - (*o_putc)(c1 | 0x080); -#ifdef X0212_ENABLE - } else if (is_eucg3(c2)){ - output_mode = SHIFT_JIS; - if (e2s_conv(c2, c1, &c2, &c1) == 0){ - (*o_putc)(c2); - (*o_putc)(c1); - } -#endif - } else { - if (!nkf_isprint(c1) || !nkf_isprint(c2)) { - set_iconv(FALSE, 0); - return; /* too late to rescue this char */ - } - output_mode = SHIFT_JIS; - e2s_conv(c2, c1, &c2, &c1); +/* + Return value of fold_conv() -#ifdef SHIFTJIS_CP932 - 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 */ + LF add newline and output char + CR add newline and output nothing + SP space + 0 skip + 1 (or else) normal output - (*o_putc)(c2); - if (prefix_table[(unsigned char)c1]){ - (*o_putc)(prefix_table[(unsigned char)c1]); - } - (*o_putc)(c1); - } -} + fold state in prev (previous character) -void j_oconv(nkf_char c2, nkf_char c1) -{ -#ifdef NUMCHAR_OPTION - if (c2 == 0 && is_unicode_capsule(c1)){ - w16e_conv(c1, &c2, &c1); - if (c2 == 0 && is_unicode_capsule(c1)){ - c2 = c1 & VALUE_MASK; - if (ms_ucs_map_f && 0xE000 <= c2 && c2 <= 0xE757) { - /* CP5022x UDC */ - c1 &= 0xFFF; - c2 = 0x7F + c1 / 94; - c1 = 0x21 + c1 % 94; - } else { - if (encode_fallback) (*encode_fallback)(c1); - return; - } - } - } -#endif - if (c2 == EOF) { - if (output_mode !=ASCII && output_mode!=ISO_8859_1) { - (*o_putc)(ESC); - (*o_putc)('('); - (*o_putc)(ascii_intro); - output_mode = ASCII; - } - (*o_putc)(EOF); -#ifdef X0212_ENABLE - } else if (is_eucg3(c2)){ - if(x0213_f){ - if(output_mode!=JIS_X_0213_2){ - output_mode = JIS_X_0213_2; - (*o_putc)(ESC); - (*o_putc)('$'); - (*o_putc)('('); - (*o_putc)(0x50); - } - }else{ - if(output_mode!=JIS_X_0212){ - output_mode = JIS_X_0212; - (*o_putc)(ESC); - (*o_putc)('$'); - (*o_putc)('('); - (*o_putc)(0x44); - } - } - (*o_putc)(c2 & 0x7f); - (*o_putc)(c1); -#endif - } else if (c2==JIS_X_0201) { - if (output_mode!=JIS_X_0201) { - output_mode = JIS_X_0201; - (*o_putc)(ESC); - (*o_putc)('('); - (*o_putc)('I'); - } - (*o_putc)(c1); - } else if (c2==ISO_8859_1) { - /* iso8859 introduction, or 8th bit on */ - /* Can we convert in 7bit form using ESC-'-'-A ? - Is this popular? */ - output_mode = ISO_8859_1; - (*o_putc)(c1|0x80); - } else if (c2 == 0) { - if (output_mode !=ASCII && output_mode!=ISO_8859_1) { - (*o_putc)(ESC); - (*o_putc)('('); - (*o_putc)(ascii_intro); - output_mode = ASCII; - } - (*o_putc)(c1); - } else { - if(ms_ucs_map_f - ? c2<0x20 || 0x92 0) { - return pop_broken_buf(); - } - c = (*i_bgetc)(f); - if (c=='$' && broken_state.status != ESC - && (input_mode==ASCII || input_mode==JIS_X_0201)) { - c1= (*i_bgetc)(f); - broken_state.status = 0; - if (c1=='@'|| c1=='B') { - push_broken_buf(c1); - push_broken_buf(c); - return ESC; - } else { - (*i_bungetc)(c1,f); - return c; - } - } else if (c=='(' && broken_state.status != ESC - && (input_mode==JIS_X_0208 || input_mode==JIS_X_0201)) { /* ) */ - c1= (*i_bgetc)(f); - broken_state.status = 0; - if (c1=='J'|| c1=='B') { - push_broken_buf(c1); - push_broken_buf(c); - return ESC; - } else { - (*i_bungetc)(c1,f); - return c; - } - } else { - broken_state.status = c; - return c; - } -} - -nkf_char broken_ungetc(nkf_char c, FILE *f) -{ - if (broken_state.count < 2) - push_broken_buf(c); - return c; -} - -void eol_conv(nkf_char c2, nkf_char c1) -{ - if (guess_f && input_eol != EOF) { - if (c2 == 0 && c1 == LF) { - if (!input_eol) input_eol = prev_cr ? CRLF : LF; - else if (input_eol != (prev_cr ? CRLF : LF)) input_eol = EOF; - } else if (c2 == 0 && c1 == CR && input_eol == LF) input_eol = EOF; - else if (!prev_cr); - else if (!input_eol) input_eol = CR; - else if (input_eol != CR) input_eol = EOF; - } - if (prev_cr || (c2 == 0 && c1 == LF)) { - prev_cr = 0; - if (eolmode_f != LF) (*o_eol_conv)(0, CR); - if (eolmode_f != CR) (*o_eol_conv)(0, LF); - } - if (c2 == 0 && c1 == CR) prev_cr = CR; - else if (c2 != 0 || c1 != LF) (*o_eol_conv)(c2, c1); -} - -/* - 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 - - fold state in prev (previous character) - - >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. -*/ - -#define char_size(c2,c1) (c2?2:1) - -void fold_conv(nkf_char c2, nkf_char c1) + >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. +*/ + +#define char_size(c2,c1) (c2?2:1) + +void fold_conv(nkf_char c2, nkf_char c1) { nkf_char prev0; nkf_char fold_state; @@ -4077,6 +3517,7 @@ void hira_conv(nkf_char c2, nkf_char c1) 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,}, @@ -5592,511 +5033,1058 @@ void module_connection(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++); + } + } +} + +/* + Conversion main loop. Code detection only. + */ + +#if !defined(PERL_XS) && !defined(WIN32DLL) +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)(EOF); + return 1; +} +#endif + +nkf_char kanji_convert(FILE *f) +{ + nkf_char c3, c2=0, c1, c0=0; + int is_8bit = FALSE; + + if (input_encoding && !nkf_enc_asciicompat(input_encoding)) { + is_8bit = TRUE; + } + + input_mode = ASCII; + output_mode = ASCII; + shift_mode = FALSE; + +#define NEXT continue /* no output, get next */ +#define SEND ; /* output c1 and c2, get next */ +#define LAST break /* end of loop, go closing */ + + module_connection(); + check_bom(f); + + while ((c1 = (*i_getc)(f)) != EOF) { +#ifdef INPUT_CODE_FIX + if (!input_encoding) +#endif + code_status(c1); + 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 ambiguious */ + if (h_conv(f, c2, c1)==EOF) + LAST; + else + c2 = 0; + NEXT; + } else { + /* in case of already established */ + if (c1 < AT) { + /* ignore bogus code and not CP5022x UCD */ + c2 = 0; + NEXT; + } else { + SEND; + } + } + } else + /* second byte, 7 bit code */ + /* it might be kanji shitfted */ + if ((c1 == DEL) || (c1 <= SP)) { + /* ignore bogus first code */ + c2 = 0; + NEXT; + } else + SEND; + } else { + /* first byte */ +#ifdef UTF8_INPUT_ENABLE + if (iconv == w_iconv16) { + if (input_endian == ENDIAN_BIG) { + c2 = c1; + if ((c1 = (*i_getc)(f)) != EOF) { + if (0xD8 <= c2 && c2 <= 0xDB) { + if ((c0 = (*i_getc)(f)) != EOF) { + c0 <<= 8; + if ((c3 = (*i_getc)(f)) != EOF) { + c0 |= c3; + } else c2 = EOF; + } else c2 = EOF; + } + } else c2 = EOF; + } else { + if ((c2 = (*i_getc)(f)) != EOF) { + if (0xD8 <= c2 && c2 <= 0xDB) { + if ((c3 = (*i_getc)(f)) != EOF) { + if ((c0 = (*i_getc)(f)) != EOF) { + c0 <<= 8; + c0 |= c3; + } else c2 = EOF; + } else c2 = EOF; + } + } else c2 = EOF; + } + SEND; + } else if(iconv == w_iconv32){ + int c3 = c1; + if((c2 = (*i_getc)(f)) != EOF && + (c1 = (*i_getc)(f)) != EOF && + (c0 = (*i_getc)(f)) != EOF){ + switch(input_endian){ + case ENDIAN_BIG: + c1 = (c2&0xFF)<<16 | (c1&0xFF)<<8 | (c0&0xFF); + break; + case ENDIAN_LITTLE: + c1 = (c3&0xFF) | (c2&0xFF)<<8 | (c1&0xFF)<<16; + break; + case ENDIAN_2143: + c1 = (c3&0xFF)<<16 | (c1&0xFF) | (c0&0xFF)<<8; + break; + case ENDIAN_3412: + c1 = (c3&0xFF)<<8 | (c2&0xFF) | (c0&0xFF)<<16; + break; + } + c2 = 0; + }else{ + c2 = EOF; + } + SEND; + } else +#endif +#ifdef NUMCHAR_OPTION + if (is_unicode_capsule(c1)){ + SEND; + } else +#endif + if (c1 > ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) { + /* 8 bit code */ + if (!estab_f && !iso8859_f) { + /* not established yet */ + c2 = c1; + NEXT; + } else { /* estab_f==TRUE */ + if (iso8859_f) { + c2 = ISO_8859_1; + c1 &= 0x7f; + SEND; + } else if (SSP<=c1 && c1<0xe0 && iconv == s_iconv) { + /* SJIS X0201 Case... */ + if (iso2022jp_f && !x0201_f) { + (*oconv)(GETA1, GETA2); + NEXT; + } else { + c2 = JIS_X_0201; + c1 &= 0x7f; + SEND; + } + } else if (c1==SSO && iconv != s_iconv) { + /* EUC X0201 Case */ + c1 = (*i_getc)(f); /* skip SSO */ + code_status(c1); + if (SSP<=c1 && c1<0xe0) { + if (iso2022jp_f && !x0201_f) { + (*oconv)(GETA1, GETA2); + NEXT; + } else { + c2 = JIS_X_0201; + c1 &= 0x7f; + SEND; + } + } else { /* bogus code, skip SSO and one byte */ + NEXT; + } + } else if (ms_ucs_map_f == UCS_MAP_CP10001 && + (c1 == 0xFD || c1 == 0xFE)) { + /* CP10001 */ + c2 = JIS_X_0201; + c1 &= 0x7f; + SEND; + } else { + /* already established */ + c2 = c1; + NEXT; + } + } + } else if ((c1 > SP) && (c1 != DEL)) { + /* in case of Roman characters */ + if (shift_mode) { + /* output 1 shifted byte */ + if (iso8859_f) { + c2 = ISO_8859_1; + SEND; + } else if (SP <= c1 && c1 < (0xe0&0x7f)){ + /* output 1 shifted byte */ + if (iso2022jp_f && !x0201_f) { + (*oconv)(GETA1, GETA2); + NEXT; + } else { + c2 = JIS_X_0201; + SEND; + } + } else { + /* look like bogus code */ + NEXT; + } + } 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 */ + c2 = c1; + NEXT; + } 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) + LAST; + else + NEXT; + } else if (mime_begin(f) == EOF) + LAST; + else + NEXT; + } else { + (*oconv)(0, '='); + (*i_ungetc)(c1,f); + NEXT; + } + } else { + /* normal ASCII code */ + SEND; + } + } else if (c1 == SI && (!is_8bit || mime_decode_mode)) { + shift_mode = FALSE; + NEXT; + } else if (c1 == SO && (!is_8bit || mime_decode_mode)) { + shift_mode = TRUE; + NEXT; + } 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 + NEXT; + } else if (c1 == '(') { + if ((c1 = (*i_getc)(f)) == EOF) { + /* don't send bogus code + (*oconv)(0, ESC); + (*oconv)(0, '$'); + (*oconv)(0, '('); + */ + LAST; + } else if (c1 == '@'|| c1 == 'B') { + /* This is kanji introduction */ + input_mode = JIS_X_0208; + shift_mode = FALSE; + NEXT; +#ifdef X0212_ENABLE + } else if (c1 == 'D'){ + input_mode = JIS_X_0212; + shift_mode = FALSE; + NEXT; +#endif /* X0212_ENABLE */ + } else if (c1 == 0x4F){ + input_mode = JIS_X_0213_1; + shift_mode = FALSE; + NEXT; + } else if (c1 == 0x50){ + input_mode = JIS_X_0213_2; + shift_mode = FALSE; + NEXT; + } else { + /* could be some special code */ + (*oconv)(0, ESC); + (*oconv)(0, '$'); + (*oconv)(0, '('); + (*oconv)(0, c1); + NEXT; + } + } else if (broken_f&0x2) { + /* accept any ESC-(-x as broken code ... */ + input_mode = JIS_X_0208; + shift_mode = FALSE; + NEXT; + } else { + (*oconv)(0, ESC); + (*oconv)(0, '$'); + (*oconv)(0, c1); + NEXT; + } + } else if (c1 == '(') { + if ((c1 = (*i_getc)(f)) == EOF) { + /* don't send bogus code + (*oconv)(0, ESC); + (*oconv)(0, '('); */ + LAST; + } else { + if (c1 == 'I') { + /* This is X0201 kana introduction */ + input_mode = JIS_X_0201; shift_mode = JIS_X_0201; + NEXT; + } else if (c1 == 'B' || c1 == 'J' || c1 == 'H') { + /* This is X0208 kanji introduction */ + input_mode = ASCII; shift_mode = FALSE; + NEXT; + } else if (broken_f&0x2) { + input_mode = ASCII; shift_mode = FALSE; + NEXT; + } else { + (*oconv)(0, ESC); + (*oconv)(0, '('); + /* maintain various input_mode here */ + SEND; + } + } + } else if ( c1 == 'N' || c1 == 'n'){ + /* SS2 */ + c3 = (*i_getc)(f); /* skip SS2 */ + if ( (SP<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){ + c1 = c3; + c2 = JIS_X_0201; + SEND; + }else{ + (*i_ungetc)(c3, f); + /* lonely ESC */ + (*oconv)(0, ESC); + SEND; + } + } else { + /* lonely ESC */ + (*oconv)(0, ESC); + SEND; + } + } else if (c1 == ESC && iconv == s_iconv) { + /* ESC in Shift_JIS */ + if ((c1 = (*i_getc)(f)) == EOF) { + /* (*oconv)(0, ESC); don't send bogus code */ + LAST; + } else if (c1 == '$') { + /* 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}; + c0 = (jphone_emoji_first_table[c1 % 7] << 8) - SP + 0xE000 + CLASS_UNICODE; + while ((c1 = (*i_getc)(f)) != EOF) { + if (SP <= c1 && c1 <= 'z') { + (*oconv)(0, c1 + c0); + } else break; /* c1 == SO */ + } + } + } + if (c1 == EOF) LAST; + NEXT; + } else { + /* lonely ESC */ + (*oconv)(0, ESC); + SEND; + } + } else if (c1 == LF || c1 == CR) { + if (broken_f&4) { + input_mode = ASCII; set_iconv(FALSE, 0); + SEND; + } else if (mime_decode_f && !mime_decode_mode){ + if (c1 == LF) { + if ((c1=(*i_getc)(f))!=EOF && c1 == SP) { + i_ungetc(SP,f); + continue; + } else { + i_ungetc(c1,f); + } + c1 = LF; + SEND; + } else { /* if (c1 == CR)*/ + if ((c1=(*i_getc)(f))!=EOF) { + if (c1==SP) { + i_ungetc(SP,f); + continue; + } else if (c1 == LF && (c1=(*i_getc)(f))!=EOF && c1 == SP) { + i_ungetc(SP,f); + continue; + } else { + i_ungetc(c1,f); + } + i_ungetc(LF,f); + } else { + i_ungetc(c1,f); + } + c1 = CR; + SEND; + } + } + } else if (c1 == DEL && input_mode == JIS_X_0208) { + /* CP5022x */ + c2 = c1; + NEXT; + } else + SEND; + } + /* send: */ + switch(input_mode){ + case ASCII: + switch ((*iconv)(c2, c1, c0)) { /* can be EUC / SJIS / UTF-8 / UTF-16 */ + case -2: + /* 4 bytes UTF-8 */ + if ((c0 = (*i_getc)(f)) != EOF) { + code_status(c0); + c0 <<= 8; + if ((c3 = (*i_getc)(f)) != EOF) { + code_status(c3); + (*iconv)(c2, c1, c0|c3); + } + } + break; + case -1: + /* 3 bytes EUC or UTF-8 */ + if ((c0 = (*i_getc)(f)) != EOF) { + code_status(c0); + (*iconv)(c2, c1, c0); + } + break; + } + break; + case JIS_X_0208: + case JIS_X_0213_1: + if (ms_ucs_map_f && + 0x7F <= c2 && c2 <= 0x92 && + 0x21 <= c1 && c1 <= 0x7E) { + /* CP932 UDC */ + if(c1 == 0x7F) return 0; + c1 = (c2 - 0x7F) * 94 + c1 - 0x21 + 0xE000 + CLASS_UNICODE; + c2 = 0; + } + (*oconv)(c2, c1); /* this is JIS, not SJIS/EUC case */ + break; +#ifdef X0212_ENABLE + case JIS_X_0212: + (*oconv)(PREFIX_EUCG3 | c2, c1); + break; +#endif /* X0212_ENABLE */ + case JIS_X_0213_2: + (*oconv)(PREFIX_EUCG3 | c2, c1); + break; + default: + (*oconv)(input_mode, c1); /* other special case */ + } + + c2 = 0; + c0 = 0; + continue; + /* goto next_word */ + } + + /* epilogue */ + (*iconv)(EOF, 0, 0); + if (!input_codename) + { + 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); +#ifdef CHECK_OPTION + debug(result->name); +#endif + } } -} - -/* - Conversion main loop. Code detection only. - */ - -#if !defined(PERL_XS) && !defined(WIN32DLL) -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)(EOF); return 1; } -#endif -nkf_char kanji_convert(FILE *f) +void options(unsigned char *cp) { - nkf_char c3, c2=0, c1, c0=0; - int is_8bit = FALSE; - - if (input_encoding && !nkf_enc_asciicompat(input_encoding)) { - is_8bit = TRUE; - } - - input_mode = ASCII; - output_mode = ASCII; - shift_mode = FALSE; - -#define NEXT continue /* no output, get next */ -#define SEND ; /* output c1 and c2, get next */ -#define LAST break /* end of loop, go closing */ - - module_connection(); - check_bom(f); + nkf_char i, j; + unsigned char *p; + unsigned char *cp_back = NULL; + char codeset[32]; + nkf_encoding *enc; - while ((c1 = (*i_getc)(f)) != EOF) { -#ifdef INPUT_CODE_FIX - if (!input_encoding) -#endif - code_status(c1); - 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 ambiguious */ - if (h_conv(f, c2, c1)==EOF) - LAST; - else - c2 = 0; - NEXT; - } else { - /* in case of already established */ - if (c1 < AT) { - /* ignore bogus code and not CP5022x UCD */ - c2 = 0; - NEXT; - } else { - SEND; - } + if (option_mode==1) + return; + while(*cp && *cp++!='-'); + while (*cp || cp_back) { + if(!*cp){ + cp = cp_back; + cp_back = NULL; + continue; + } + p = 0; + switch (*cp++) { + case '-': /* literal options */ + if (!*cp || *cp == SP) { /* ignore the rest of arguments */ + option_mode = 1; + return; + } + for (i=0;i ((input_encoding && nkf_enc_cp5022x_p(input_encoding)) ? 0x92 : DEL)) { - /* 8 bit code */ - if (!estab_f && !iso8859_f) { - /* not established yet */ - c2 = c1; - NEXT; - } else { /* estab_f==TRUE */ - if (iso8859_f) { - c2 = ISO_8859_1; - c1 &= 0x7f; - SEND; - } else if (SSP<=c1 && c1<0xe0 && iconv == s_iconv) { - /* SJIS X0201 Case... */ - if (iso2022jp_f && !x0201_f) { - (*oconv)(GETA1, GETA2); - NEXT; - } else { - c2 = JIS_X_0201; - c1 &= 0x7f; - SEND; - } - } else if (c1==SSO && iconv != s_iconv) { - /* EUC X0201 Case */ - c1 = (*i_getc)(f); /* skip SSO */ - code_status(c1); - if (SSP<=c1 && c1<0xe0) { - if (iso2022jp_f && !x0201_f) { - (*oconv)(GETA1, GETA2); - NEXT; - } else { - c2 = JIS_X_0201; - c1 &= 0x7f; - SEND; - } - } else { /* bogus code, skip SSO and one byte */ - NEXT; - } - } else if (ms_ucs_map_f == UCS_MAP_CP10001 && - (c1 == 0xFD || c1 == 0xFE)) { - /* CP10001 */ - c2 = JIS_X_0201; - c1 &= 0x7f; - SEND; - } else { - /* already established */ - c2 = c1; - NEXT; - } + if (strcmp(long_option[i].name, "cp932") == 0){ +#ifdef SHIFTJIS_CP932 + cp51932_f = TRUE; + cp932inv_f = -TRUE; +#endif +#ifdef UTF8_OUTPUT_ENABLE + ms_ucs_map_f = UCS_MAP_CP932; +#endif + continue; } - } else if ((c1 > SP) && (c1 != DEL)) { - /* in case of Roman characters */ - if (shift_mode) { - /* output 1 shifted byte */ - if (iso8859_f) { - c2 = ISO_8859_1; - SEND; - } else if (SP <= c1 && c1 < (0xe0&0x7f)){ - /* output 1 shifted byte */ - if (iso2022jp_f && !x0201_f) { - (*oconv)(GETA1, GETA2); - NEXT; - } else { - c2 = JIS_X_0201; - SEND; - } - } else { - /* look like bogus code */ - NEXT; - } - } 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 */ - c2 = c1; - NEXT; - } 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) - LAST; - else - NEXT; - } else if (mime_begin(f) == EOF) - LAST; - else - NEXT; - } else { - (*oconv)(0, '='); - (*i_ungetc)(c1,f); - NEXT; - } - } else { - /* normal ASCII code */ - SEND; + if (strcmp(long_option[i].name, "no-cp932") == 0){ +#ifdef SHIFTJIS_CP932 + cp51932_f = FALSE; + cp932inv_f = FALSE; +#endif +#ifdef UTF8_OUTPUT_ENABLE + ms_ucs_map_f = UCS_MAP_ASCII; +#endif + continue; + } +#ifdef SHIFTJIS_CP932 + if (strcmp(long_option[i].name, "cp932inv") == 0){ + cp932inv_f = -TRUE; + continue; } - } else if (c1 == SI && (!is_8bit || mime_decode_mode)) { - shift_mode = FALSE; - NEXT; - } else if (c1 == SO && (!is_8bit || mime_decode_mode)) { - shift_mode = TRUE; - NEXT; - } 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, '$'); */ - LAST; - } else if (c1 == '@'|| c1 == 'B') { - /* This is kanji introduction */ - input_mode = JIS_X_0208; - shift_mode = FALSE; - set_input_codename("ISO-2022-JP"); -#ifdef CHECK_OPTION - debug("ISO-2022-JP"); #endif - NEXT; - } else if (c1 == '(') { - if ((c1 = (*i_getc)(f)) == EOF) { - /* don't send bogus code - (*oconv)(0, ESC); - (*oconv)(0, '$'); - (*oconv)(0, '('); - */ - LAST; - } else if (c1 == '@'|| c1 == 'B') { - /* This is kanji introduction */ - input_mode = JIS_X_0208; - shift_mode = FALSE; - NEXT; + #ifdef X0212_ENABLE - } else if (c1 == 'D'){ - input_mode = JIS_X_0212; - shift_mode = FALSE; - NEXT; -#endif /* X0212_ENABLE */ - } else if (c1 == 0x4F){ - input_mode = JIS_X_0213_1; - shift_mode = FALSE; - NEXT; - } else if (c1 == 0x50){ - input_mode = JIS_X_0213_2; - shift_mode = FALSE; - NEXT; - } else { - /* could be some special code */ - (*oconv)(0, ESC); - (*oconv)(0, '$'); - (*oconv)(0, '('); - (*oconv)(0, c1); - NEXT; - } - } else if (broken_f&0x2) { - /* accept any ESC-(-x as broken code ... */ - input_mode = JIS_X_0208; - shift_mode = FALSE; - NEXT; - } else { - (*oconv)(0, ESC); - (*oconv)(0, '$'); - (*oconv)(0, c1); - NEXT; - } - } else if (c1 == '(') { - 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; shift_mode = JIS_X_0201; - NEXT; - } else if (c1 == 'B' || c1 == 'J' || c1 == 'H') { - /* This is X0208 kanji introduction */ - input_mode = ASCII; shift_mode = FALSE; - NEXT; - } else if (broken_f&0x2) { - input_mode = ASCII; shift_mode = FALSE; - NEXT; - } else { - (*oconv)(0, ESC); - (*oconv)(0, '('); - /* maintain various input_mode here */ - SEND; - } - } - } else if ( c1 == 'N' || c1 == 'n'){ - /* SS2 */ - c3 = (*i_getc)(f); /* skip SS2 */ - if ( (SP<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){ - c1 = c3; - c2 = JIS_X_0201; - SEND; - }else{ - (*i_ungetc)(c3, f); - /* lonely ESC */ - (*oconv)(0, ESC); - SEND; - } - } else { - /* lonely ESC */ - (*oconv)(0, ESC); - SEND; + if (strcmp(long_option[i].name, "x0212") == 0){ + x0212_f = TRUE; + continue; + } +#endif + +#ifdef EXEC_IO + if (strcmp(long_option[i].name, "exec-in") == 0){ + exec_f = 1; + return; + } + if (strcmp(long_option[i].name, "exec-out") == 0){ + exec_f = -1; + return; + } +#endif +#if defined(UTF8_OUTPUT_ENABLE) && defined(UTF8_INPUT_ENABLE) + if (strcmp(long_option[i].name, "no-cp932ext") == 0){ + no_cp932ext_f = TRUE; + continue; } - } 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 */ - 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}; - c0 = (jphone_emoji_first_table[c1 % 7] << 8) - SP + 0xE000 + CLASS_UNICODE; - while ((c1 = (*i_getc)(f)) != EOF) { - if (SP <= c1 && c1 <= 'z') { - (*oconv)(0, c1 + c0); - } else break; /* c1 == SO */ - } - } - } - if (c1 == EOF) LAST; - NEXT; - } else { - /* lonely ESC */ - (*oconv)(0, ESC); - SEND; + if (strcmp(long_option[i].name, "no-best-fit-chars") == 0){ + no_best_fit_chars_f = TRUE; + continue; } - } else if (c1 == LF || c1 == CR) { - if (broken_f&4) { - input_mode = ASCII; set_iconv(FALSE, 0); - SEND; - } else if (mime_decode_f && !mime_decode_mode){ - if (c1 == LF) { - if ((c1=(*i_getc)(f))!=EOF && c1 == SP) { - i_ungetc(SP,f); - continue; - } else { - i_ungetc(c1,f); + if (strcmp(long_option[i].name, "fb-skip") == 0){ + encode_fallback = NULL; + continue; + } + if (strcmp(long_option[i].name, "fb-html") == 0){ + encode_fallback = encode_fallback_html; + continue; + } + if (strcmp(long_option[i].name, "fb-xml") == 0){ + encode_fallback = encode_fallback_xml; + continue; + } + if (strcmp(long_option[i].name, "fb-java") == 0){ + encode_fallback = encode_fallback_java; + continue; + } + if (strcmp(long_option[i].name, "fb-perl") == 0){ + encode_fallback = encode_fallback_perl; + continue; + } + if (strcmp(long_option[i].name, "fb-subchar") == 0){ + encode_fallback = encode_fallback_subchar; + continue; + } + if (strcmp(long_option[i].name, "fb-subchar=") == 0){ + encode_fallback = encode_fallback_subchar; + unicode_subchar = 0; + if (p[0] != '0'){ + /* decimal number */ + for (i = 0; i < 7 && nkf_isdigit(p[i]); i++){ + unicode_subchar *= 10; + unicode_subchar += hex2bin(p[i]); } - c1 = LF; - SEND; - } else { /* if (c1 == CR)*/ - if ((c1=(*i_getc)(f))!=EOF) { - if (c1==SP) { - i_ungetc(SP,f); - continue; - } else if (c1 == LF && (c1=(*i_getc)(f))!=EOF && c1 == SP) { - i_ungetc(SP,f); - continue; - } else { - i_ungetc(c1,f); - } - i_ungetc(LF,f); - } else { - i_ungetc(c1,f); + }else if(p[1] == 'x' || p[1] == 'X'){ + /* hexadecimal number */ + for (i = 2; i < 8 && nkf_isxdigit(p[i]); i++){ + unicode_subchar <<= 4; + unicode_subchar |= hex2bin(p[i]); + } + }else{ + /* octal number */ + for (i = 1; i < 8 && nkf_isoctal(p[i]); i++){ + unicode_subchar *= 8; + unicode_subchar += hex2bin(p[i]); } - c1 = CR; - SEND; } + w16e_conv(unicode_subchar, &i, &j); + unicode_subchar = i<<8 | j; + continue; + } +#endif +#ifdef UTF8_OUTPUT_ENABLE + if (strcmp(long_option[i].name, "ms-ucs-map") == 0){ + ms_ucs_map_f = UCS_MAP_MS; + continue; + } +#endif +#ifdef UNICODE_NORMALIZATION + if (strcmp(long_option[i].name, "utf8mac-input") == 0){ + nfc_f = TRUE; + continue; + } +#endif + if (strcmp(long_option[i].name, "prefix=") == 0){ + if (nkf_isgraph(p[0])){ + for (i = 1; nkf_isgraph(p[i]); i++){ + prefix_table[p[i]] = p[0]; + } + } + continue; + } + } + continue; + case 'b': /* buffered mode */ + unbuf_f = FALSE; + continue; + case 'u': /* non bufferd mode */ + unbuf_f = TRUE; + continue; + case 't': /* transparent mode */ + if (*cp=='1') { + /* alias of -t */ + cp++; + nop_f = TRUE; + } else if (*cp=='2') { + /* + * -t with put/get + * + * nkf -t2MB hoge.bin | nkf -t2mB | diff -s - hoge.bin + * + */ + cp++; + nop_f = 2; + } else + nop_f = TRUE; + continue; + case 'j': /* JIS output */ + case 'n': + output_encoding = nkf_enc_from_index(ISO_2022_JP); + continue; + case 'e': /* AT&T EUC output */ + output_encoding = nkf_enc_from_index(EUC_JP); + continue; + case 's': /* SJIS output */ + output_encoding = nkf_enc_from_index(WINDOWS_31J); + continue; + case 'l': /* ISO8859 Latin-1 support, no conversion */ + iso8859_f = TRUE; /* Only compatible with ISO-2022-JP */ + input_encoding = nkf_enc_from_index(ISO_8859_1); + continue; + case 'i': /* Kanji IN ESC-$-@/B */ + if (*cp=='@'||*cp=='B') + kanji_intro = *cp++; + continue; + case 'o': /* ASCII IN ESC-(-J/B */ + if (*cp=='J'||*cp=='B'||*cp=='H') + ascii_intro = *cp++; + continue; + case 'h': + /* + bit:1 katakana->hiragana + bit:2 hiragana->katakana + */ + if ('9'>= *cp && *cp>='0') + hira_f |= (*cp++ -'0'); + else + hira_f |= 1; + continue; + case 'r': + rot_f = TRUE; + continue; +#if defined(MSDOS) || defined(__OS2__) + case 'T': + binmode_f = FALSE; + continue; +#endif +#ifndef PERL_XS + case 'V': + show_configuration(); + exit(1); + break; + case 'v': + usage(); + exit(1); + break; +#endif +#ifdef UTF8_OUTPUT_ENABLE + case 'w': /* UTF-8 output */ + if (cp[0] == '8') { + cp++; + if (cp[0] == '0'){ + cp++; + output_encoding = nkf_enc_from_index(UTF_8N); + } else { + output_bom_f = TRUE; + output_encoding = nkf_enc_from_index(UTF_8_BOM); + } + } else { + int enc_idx; + if ('1'== cp[0] && '6'==cp[1]) { + cp += 2; + enc_idx = UTF_16; + } else if ('3'== cp[0] && '2'==cp[1]) { + cp += 2; + enc_idx = UTF_32; + } else { + output_encoding = nkf_enc_from_index(UTF_8); + continue; + } + if (cp[0]=='L') { + cp++; + output_endian = ENDIAN_LITTLE; + } else if (cp[0] == 'B') { + cp++; + } else { + output_encoding = nkf_enc_from_index(enc_idx); + continue; + } + if (cp[0] == '0'){ + cp++; + enc_idx = enc_idx == UTF_16 + ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE) + : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE); + } else { + output_bom_f = TRUE; + enc_idx = enc_idx == UTF_16 + ? (output_endian == ENDIAN_LITTLE ? UTF_16LE_BOM : UTF_16BE_BOM) + : (output_endian == ENDIAN_LITTLE ? UTF_32LE_BOM : UTF_32BE_BOM); } - } else if (c1 == DEL && input_mode == JIS_X_0208) { - /* CP5022x */ - c2 = c1; - NEXT; - } else - SEND; - } - /* send: */ - switch(input_mode){ - case ASCII: - switch ((*iconv)(c2, c1, c0)) { /* can be EUC / SJIS / UTF-8 / UTF-16 */ - case -2: - /* 4 bytes UTF-8 */ - if ((c0 = (*i_getc)(f)) != EOF) { - code_status(c0); - c0 <<= 8; - if ((c3 = (*i_getc)(f)) != EOF) { - code_status(c3); - (*iconv)(c2, c1, c0|c3); - } + output_encoding = nkf_enc_from_index(enc_idx); + } + continue; +#endif +#ifdef UTF8_INPUT_ENABLE + case 'W': /* UTF input */ + if (cp[0] == '8') { + cp++; + input_encoding = nkf_enc_from_index(UTF_8); + }else{ + int enc_idx; + if ('1'== cp[0] && '6'==cp[1]) { + cp += 2; + input_endian = ENDIAN_BIG; + enc_idx = UTF_16; + } else if ('3'== cp[0] && '2'==cp[1]) { + cp += 2; + input_endian = ENDIAN_BIG; + enc_idx = UTF_32; + } else { + input_encoding = nkf_enc_from_index(UTF_8); + continue; } - break; - case -1: - /* 3 bytes EUC or UTF-8 */ - if ((c0 = (*i_getc)(f)) != EOF) { - code_status(c0); - (*iconv)(c2, c1, c0); + if (cp[0]=='L') { + cp++; + input_endian = ENDIAN_LITTLE; + } else if (cp[0] == 'B') { + cp++; + input_endian = ENDIAN_BIG; } - break; + enc_idx = enc_idx == UTF_16 + ? (output_endian == ENDIAN_LITTLE ? UTF_16LE : UTF_16BE) + : (output_endian == ENDIAN_LITTLE ? UTF_32LE : UTF_32BE); + input_encoding = nkf_enc_from_index(enc_idx); } - break; - case JIS_X_0208: - case JIS_X_0213_1: - if (ms_ucs_map_f && - 0x7F <= c2 && c2 <= 0x92 && - 0x21 <= c1 && c1 <= 0x7E) { - /* CP932 UDC */ - if(c1 == 0x7F) return 0; - c1 = (c2 - 0x7F) * 94 + c1 - 0x21 + 0xE000 + CLASS_UNICODE; - c2 = 0; + continue; +#endif + /* Input code assumption */ + case 'J': /* ISO-2022-JP input */ + input_encoding = nkf_enc_from_index(ISO_2022_JP); + continue; + case 'E': /* EUC-JP input */ + input_encoding = nkf_enc_from_index(EUC_JP); + continue; + case 'S': /* Windows-31J input */ + input_encoding = nkf_enc_from_index(WINDOWS_31J); + continue; + case 'Z': /* Convert X0208 alphabet to asii */ + /* alpha_f + bit:0 Convert JIS X 0208 Alphabet to ASCII + bit:1 Convert Kankaku to one space + bit:2 Convert Kankaku to two spaces + bit:3 Convert HTML Entity + bit:4 Convert JIS X 0208 Katakana to JIS X 0201 Katakana + */ + while ('0'<= *cp && *cp <='9') { + alpha_f |= 1 << (*cp++ - '0'); } - (*oconv)(c2, c1); /* this is JIS, not SJIS/EUC case */ - break; -#ifdef X0212_ENABLE - case JIS_X_0212: - (*oconv)(PREFIX_EUCG3 | c2, c1); - break; -#endif /* X0212_ENABLE */ - case JIS_X_0213_2: - (*oconv)(PREFIX_EUCG3 | c2, c1); - break; - default: - (*oconv)(input_mode, c1); /* other special case */ - } - - c2 = 0; - c0 = 0; - continue; - /* goto next_word */ - } - - /* epilogue */ - (*iconv)(EOF, 0, 0); - if (!input_codename) - { - 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; + if (!alpha_f) alpha_f = 1; + continue; + case 'x': /* Convert X0201 kana to X0208 or X0201 Conversion */ + x0201_f = FALSE; /* No X0201->X0208 conversion */ + /* accept X0201 + ESC-(-I in JIS, EUC, MS Kanji + SI/SO in JIS, EUC, MS Kanji + SSO in EUC, JIS, not in MS Kanji + MS Kanji (0xa0-0xdf) + output X0201 + ESC-(-I in JIS (0x20-0x5f) + SSO in EUC (0xa0-0xdf) + 0xa0-0xd in MS Kanji (0xa0-0xdf) + */ + continue; + case 'X': /* Convert X0201 kana to X0208 */ + x0201_f = TRUE; + continue; + case 'F': /* prserve new lines */ + fold_preserve_f = TRUE; + case 'f': /* folding -f60 or -f */ + fold_f = TRUE; + fold_len = 0; + while('0'<= *cp && *cp <='9') { /* we don't use atoi here */ + fold_len *= 10; + fold_len += *cp++ - '0'; } - set_input_codename(result->name); -#ifdef CHECK_OPTION - debug(result->name); + if (!(0= *cp && *cp>='0') + broken_f |= 1<<(*cp++ -'0'); + else + broken_f |= TRUE; + continue; +#ifndef PERL_XS + case 'O':/* for Output file */ + file_out_f = TRUE; + continue; #endif - } + case 'c':/* add cr code */ + eolmode_f = CRLF; + continue; + case 'd':/* delete cr code */ + eolmode_f = LF; + continue; + case 'I': /* ISO-2022-JP output */ + iso2022jp_f = TRUE; + continue; + case 'L': /* line mode */ + if (*cp=='u') { /* unix */ + eolmode_f = LF; cp++; + } else if (*cp=='m') { /* mac */ + eolmode_f = CR; cp++; + } else if (*cp=='w') { /* windows */ + eolmode_f = CRLF; cp++; + } else if (*cp=='0') { /* no conversion */ + eolmode_f = 0; cp++; + } + continue; +#ifndef PERL_XS + case 'g': + if ('2' <= *cp && *cp <= '9') { + guess_f = 2; + cp++; + } else if (*cp == '0' || *cp == '1') { + guess_f = 1; + cp++; + } else { + guess_f = 1; + } + continue; +#endif + case SP: + /* module muliple options in a string are allowed for Perl moudle */ + while(*cp && *cp++!='-'); + continue; + default: + fprintf(stderr, "unknown option: -%c\n", *(cp-1)); + /* bogus option but ignored */ + continue; + } } - return 1; } #ifdef WIN32DLL