00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024
00025
00026
00027
00028
00029
00030 #include "regenc.h"
00031
00032 #define ENC_KOI8_R_TO_LOWER_CASE(c) EncKOI8_R_ToLowerCaseTable[c]
00033 #define ENC_IS_KOI8_R_CTYPE(code,ctype) \
00034 ((EncKOI8_R_CtypeTable[code] & CTYPE_TO_BIT(ctype)) != 0)
00035
00036 static const UChar EncKOI8_R_ToLowerCaseTable[256] = {
00037 '\000', '\001', '\002', '\003', '\004', '\005', '\006', '\007',
00038 '\010', '\011', '\012', '\013', '\014', '\015', '\016', '\017',
00039 '\020', '\021', '\022', '\023', '\024', '\025', '\026', '\027',
00040 '\030', '\031', '\032', '\033', '\034', '\035', '\036', '\037',
00041 '\040', '\041', '\042', '\043', '\044', '\045', '\046', '\047',
00042 '\050', '\051', '\052', '\053', '\054', '\055', '\056', '\057',
00043 '\060', '\061', '\062', '\063', '\064', '\065', '\066', '\067',
00044 '\070', '\071', '\072', '\073', '\074', '\075', '\076', '\077',
00045 '\100', '\141', '\142', '\143', '\144', '\145', '\146', '\147',
00046 '\150', '\151', '\152', '\153', '\154', '\155', '\156', '\157',
00047 '\160', '\161', '\162', '\163', '\164', '\165', '\166', '\167',
00048 '\170', '\171', '\172', '\133', '\134', '\135', '\136', '\137',
00049 '\140', '\141', '\142', '\143', '\144', '\145', '\146', '\147',
00050 '\150', '\151', '\152', '\153', '\154', '\155', '\156', '\157',
00051 '\160', '\161', '\162', '\163', '\164', '\165', '\166', '\167',
00052 '\170', '\171', '\172', '\173', '\174', '\175', '\176', '\177',
00053 '\200', '\201', '\202', '\203', '\204', '\205', '\206', '\207',
00054 '\210', '\211', '\212', '\213', '\214', '\215', '\216', '\217',
00055 '\220', '\221', '\222', '\223', '\224', '\225', '\226', '\227',
00056 '\230', '\231', '\232', '\233', '\234', '\235', '\236', '\237',
00057 '\240', '\241', '\242', '\243', '\244', '\245', '\246', '\247',
00058 '\250', '\251', '\252', '\253', '\254', '\255', '\256', '\257',
00059 '\260', '\261', '\262', '\243', '\264', '\265', '\266', '\267',
00060 '\270', '\271', '\272', '\273', '\274', '\275', '\276', '\277',
00061 '\300', '\301', '\302', '\303', '\304', '\305', '\306', '\307',
00062 '\310', '\311', '\312', '\313', '\314', '\315', '\316', '\317',
00063 '\320', '\321', '\322', '\323', '\324', '\325', '\326', '\327',
00064 '\330', '\331', '\332', '\333', '\334', '\335', '\336', '\337',
00065 '\300', '\301', '\302', '\303', '\304', '\305', '\306', '\307',
00066 '\310', '\311', '\312', '\313', '\314', '\315', '\316', '\317',
00067 '\320', '\321', '\322', '\323', '\324', '\325', '\326', '\327',
00068 '\330', '\331', '\332', '\333', '\334', '\335', '\336', '\337'
00069 };
00070
00071 static const unsigned short EncKOI8_R_CtypeTable[256] = {
00072 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
00073 0x4008, 0x420c, 0x4209, 0x4208, 0x4208, 0x4208, 0x4008, 0x4008,
00074 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
00075 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
00076 0x4284, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
00077 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
00078 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0,
00079 0x78b0, 0x78b0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
00080 0x41a0, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x74a2,
00081 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
00082 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
00083 0x74a2, 0x74a2, 0x74a2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x51a0,
00084 0x41a0, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x70e2,
00085 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
00086 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
00087 0x70e2, 0x70e2, 0x70e2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x4008,
00088 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00089 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00090 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00091 0x00a0, 0x00a0, 0x0284, 0x00a0, 0x00a0, 0x10a0, 0x01a0, 0x00a0,
00092 0x00a0, 0x00a0, 0x00a0, 0x30e2, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00093 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00094 0x00a0, 0x00a0, 0x00a0, 0x34a2, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00095 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0,
00096 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00097 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00098 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00099 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00100 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00101 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00102 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00103 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2
00104 };
00105
00106 static int
00107 koi8_r_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED,
00108 const UChar** pp, const UChar* end ARG_UNUSED,
00109 UChar* lower, OnigEncoding enc ARG_UNUSED)
00110 {
00111 const UChar* p = *pp;
00112
00113 *lower = ENC_KOI8_R_TO_LOWER_CASE(*p);
00114 (*pp)++;
00115 return 1;
00116 }
00117
00118 #if 0
00119 static int
00120 koi8_r_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp,
00121 const UChar* end, OnigEncoding enc ARG_UNUSED)
00122 {
00123 int v;
00124 const UChar* p = *pp;
00125
00126 (*pp)++;
00127 v = (EncKOI8_R_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
00128 return (v != 0 ? TRUE : FALSE);
00129 }
00130 #endif
00131
00132 static int
00133 koi8_r_is_code_ctype(OnigCodePoint code, unsigned int ctype,
00134 OnigEncoding enc ARG_UNUSED)
00135 {
00136 if (code < 256)
00137 return ENC_IS_KOI8_R_CTYPE(code, ctype);
00138 else
00139 return FALSE;
00140 }
00141
00142 static const OnigPairCaseFoldCodes CaseFoldMap[] = {
00143 { 0xa3, 0xb3 },
00144
00145 { 0xc0, 0xe0 },
00146 { 0xc1, 0xe1 },
00147 { 0xc2, 0xe2 },
00148 { 0xc3, 0xe3 },
00149 { 0xc4, 0xe4 },
00150 { 0xc5, 0xe5 },
00151 { 0xc6, 0xe6 },
00152 { 0xc7, 0xe7 },
00153 { 0xc8, 0xe8 },
00154 { 0xc9, 0xe9 },
00155 { 0xca, 0xea },
00156 { 0xcb, 0xeb },
00157 { 0xcc, 0xec },
00158 { 0xcd, 0xed },
00159 { 0xce, 0xee },
00160 { 0xcf, 0xef },
00161
00162 { 0xd0, 0xf0 },
00163 { 0xd1, 0xf1 },
00164 { 0xd2, 0xf2 },
00165 { 0xd3, 0xf3 },
00166 { 0xd4, 0xf4 },
00167 { 0xd5, 0xf5 },
00168 { 0xd6, 0xf6 },
00169 { 0xd7, 0xf7 },
00170 { 0xd8, 0xf8 },
00171 { 0xd9, 0xf9 },
00172 { 0xda, 0xfa },
00173 { 0xdb, 0xfb },
00174 { 0xdc, 0xfc },
00175 { 0xdd, 0xfd },
00176 { 0xde, 0xfe },
00177 { 0xdf, 0xff }
00178 };
00179
00180 static int
00181 koi8_r_apply_all_case_fold(OnigCaseFoldType flag,
00182 OnigApplyAllCaseFoldFunc f,
00183 void* arg, OnigEncoding enc ARG_UNUSED)
00184 {
00185 return onigenc_apply_all_case_fold_with_map(
00186 sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 0,
00187 flag, f, arg);
00188 }
00189
00190 static int
00191 koi8_r_get_case_fold_codes_by_str(OnigCaseFoldType flag,
00192 const OnigUChar* p, const OnigUChar* end,
00193 OnigCaseFoldCodeItem items[], OnigEncoding enc ARG_UNUSED)
00194 {
00195 return onigenc_get_case_fold_codes_by_str_with_map(
00196 sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 0,
00197 flag, p, end, items);
00198 }
00199
00200 OnigEncodingDefine(koi8_r, KOI8_R) = {
00201 onigenc_single_byte_mbc_enc_len,
00202 "KOI8-R",
00203 1,
00204 1,
00205 onigenc_is_mbc_newline_0x0a,
00206 onigenc_single_byte_mbc_to_code,
00207 onigenc_single_byte_code_to_mbclen,
00208 onigenc_single_byte_code_to_mbc,
00209 koi8_r_mbc_case_fold,
00210 koi8_r_apply_all_case_fold,
00211 koi8_r_get_case_fold_codes_by_str,
00212 onigenc_minimum_property_name_to_ctype,
00213 koi8_r_is_code_ctype,
00214 onigenc_not_support_get_ctype_code_range,
00215 onigenc_single_byte_left_adjust_char_head,
00216 onigenc_always_true_is_allowed_reverse_match
00217 };
00218 ENC_ALIAS("CP878", "KOI8-R")
00219
00220