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
00031 #include "regenc.h"
00032
00033 #define ENC_CP1251_TO_LOWER_CASE(c) EncCP1251_ToLowerCaseTable[c]
00034 #define ENC_IS_CP1251_CTYPE(code,ctype) \
00035 ((EncCP1251_CtypeTable[code] & CTYPE_TO_BIT(ctype)) != 0)
00036
00037 static const UChar EncCP1251_ToLowerCaseTable[256] = {
00038 '\000', '\001', '\002', '\003', '\004', '\005', '\006', '\007',
00039 '\010', '\011', '\012', '\013', '\014', '\015', '\016', '\017',
00040 '\020', '\021', '\022', '\023', '\024', '\025', '\026', '\027',
00041 '\030', '\031', '\032', '\033', '\034', '\035', '\036', '\037',
00042 '\040', '\041', '\042', '\043', '\044', '\045', '\046', '\047',
00043 '\050', '\051', '\052', '\053', '\054', '\055', '\056', '\057',
00044 '\060', '\061', '\062', '\063', '\064', '\065', '\066', '\067',
00045 '\070', '\071', '\072', '\073', '\074', '\075', '\076', '\077',
00046 '\100', '\141', '\142', '\143', '\144', '\145', '\146', '\147',
00047 '\150', '\151', '\152', '\153', '\154', '\155', '\156', '\157',
00048 '\160', '\161', '\162', '\163', '\164', '\165', '\166', '\167',
00049 '\170', '\171', '\172', '\133', '\134', '\135', '\136', '\137',
00050 '\140', '\141', '\142', '\143', '\144', '\145', '\146', '\147',
00051 '\150', '\151', '\152', '\153', '\154', '\155', '\156', '\157',
00052 '\160', '\161', '\162', '\163', '\164', '\165', '\166', '\167',
00053 '\170', '\171', '\172', '\173', '\174', '\175', '\176', '\177',
00054 '\220', '\203', '\202', '\203', '\204', '\205', '\206', '\207',
00055 '\210', '\211', '\232', '\213', '\234', '\235', '\236', '\237',
00056 '\220', '\221', '\222', '\223', '\224', '\225', '\226', '\227',
00057 '\230', '\231', '\232', '\233', '\234', '\235', '\236', '\237',
00058 '\240', '\242', '\242', '\274', '\244', '\264', '\246', '\247',
00059 '\270', '\251', '\272', '\253', '\254', '\255', '\256', '\277',
00060 '\260', '\261', '\263', '\263', '\264', '\265', '\266', '\267',
00061 '\270', '\271', '\272', '\273', '\274', '\276', '\276', '\277',
00062 '\340', '\341', '\342', '\343', '\344', '\345', '\346', '\347',
00063 '\350', '\351', '\352', '\353', '\354', '\355', '\356', '\357',
00064 '\360', '\361', '\362', '\363', '\364', '\365', '\366', '\367',
00065 '\370', '\371', '\372', '\373', '\374', '\375', '\376', '\377',
00066 '\340', '\341', '\342', '\343', '\344', '\345', '\346', '\347',
00067 '\350', '\351', '\352', '\353', '\354', '\355', '\356', '\357',
00068 '\360', '\361', '\362', '\363', '\364', '\365', '\366', '\367',
00069 '\370', '\371', '\372', '\373', '\374', '\375', '\376', '\377'
00070 };
00071
00072 static const unsigned short EncCP1251_CtypeTable[256] = {
00073 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
00074 0x4008, 0x428c, 0x4209, 0x4208, 0x4208, 0x4208, 0x4008, 0x4008,
00075 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
00076 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
00077 0x4284, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
00078 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
00079 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0,
00080 0x78b0, 0x78b0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
00081 0x41a0, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x74a2,
00082 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
00083 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
00084 0x74a2, 0x74a2, 0x74a2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x51a0,
00085 0x41a0, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x70e2,
00086 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
00087 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
00088 0x70e2, 0x70e2, 0x70e2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x4008,
00089 0x34a2, 0x34a2, 0x01a0, 0x30e2, 0x01a0, 0x01a0, 0x01a0, 0x01a0,
00090 0x0000, 0x01a0, 0x34a2, 0x01a0, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00091 0x30e2, 0x01a0, 0x01a0, 0x01a0, 0x01a0, 0x01a0, 0x01a0, 0x01a0,
00092 0x0008, 0x0000, 0x30e2, 0x01a0, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00093 0x0280, 0x34a2, 0x30e2, 0x34a2, 0x01a0, 0x34a2, 0x01a0, 0x01a0,
00094 0x34a2, 0x01a0, 0x34a2, 0x01a0, 0x01a0, 0x01a0, 0x01a0, 0x34a2,
00095 0x01a0, 0x01a0, 0x34a2, 0x30e2, 0x30e2, 0x31e2, 0x01a0, 0x01a0,
00096 0x30e2, 0x0000, 0x30e2, 0x01a0, 0x30e2, 0x34a2, 0x30e2, 0x30e2,
00097 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00098 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00099 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00100 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00101 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00102 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00103 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00104 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2
00105 };
00106
00107 static int
00108 cp1251_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED,
00109 const UChar** pp, const UChar* end ARG_UNUSED, UChar* lower, OnigEncoding enc ARG_UNUSED)
00110 {
00111 const UChar* p = *pp;
00112
00113 *lower = ENC_CP1251_TO_LOWER_CASE(*p);
00114 (*pp)++;
00115 return 1;
00116 }
00117
00118 static int
00119 cp1251_is_code_ctype(OnigCodePoint code, unsigned int ctype, OnigEncoding enc ARG_UNUSED)
00120 {
00121 if (code < 256)
00122 return ENC_IS_CP1251_CTYPE(code, ctype);
00123 else
00124 return FALSE;
00125 }
00126
00127 static const OnigPairCaseFoldCodes CaseFoldMap[] = {
00128 { 0xb8, 0xa8 },
00129
00130 { 0xe0, 0xc0 },
00131 { 0xe1, 0xc1 },
00132 { 0xe2, 0xc2 },
00133 { 0xe3, 0xc3 },
00134 { 0xe4, 0xc4 },
00135 { 0xe5, 0xc5 },
00136 { 0xe6, 0xc6 },
00137 { 0xe7, 0xc7 },
00138 { 0xe8, 0xc8 },
00139 { 0xe9, 0xc9 },
00140 { 0xea, 0xca },
00141 { 0xeb, 0xcb },
00142 { 0xec, 0xcc },
00143 { 0xed, 0xcd },
00144 { 0xee, 0xce },
00145 { 0xef, 0xcf },
00146
00147 { 0xf0, 0xd0 },
00148 { 0xf1, 0xd1 },
00149 { 0xf2, 0xd2 },
00150 { 0xf3, 0xd3 },
00151 { 0xf4, 0xd4 },
00152 { 0xf5, 0xd5 },
00153 { 0xf6, 0xd6 },
00154 { 0xf7, 0xd7 },
00155 { 0xf8, 0xd8 },
00156 { 0xf9, 0xd9 },
00157 { 0xfa, 0xda },
00158 { 0xfb, 0xdb },
00159 { 0xfc, 0xdc },
00160 { 0xfd, 0xdd },
00161 { 0xfe, 0xde },
00162 { 0xff, 0xdf }
00163 };
00164
00165 static int
00166 cp1251_apply_all_case_fold(OnigCaseFoldType flag,
00167 OnigApplyAllCaseFoldFunc f, void* arg, OnigEncoding enc ARG_UNUSED)
00168 {
00169 return onigenc_apply_all_case_fold_with_map(
00170 sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 0,
00171 flag, f, arg);
00172 }
00173
00174 static int
00175 cp1251_get_case_fold_codes_by_str(OnigCaseFoldType flag,
00176 const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem items[], OnigEncoding enc ARG_UNUSED)
00177 {
00178 return onigenc_get_case_fold_codes_by_str_with_map(
00179 sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 0,
00180 flag, p, end, items);
00181 }
00182
00183 OnigEncodingDefine(windows_1251, Windows_1251) = {
00184 onigenc_single_byte_mbc_enc_len,
00185 "Windows-1251",
00186 1,
00187 1,
00188 onigenc_is_mbc_newline_0x0a,
00189 onigenc_single_byte_mbc_to_code,
00190 onigenc_single_byte_code_to_mbclen,
00191 onigenc_single_byte_code_to_mbc,
00192 cp1251_mbc_case_fold,
00193 cp1251_apply_all_case_fold,
00194 cp1251_get_case_fold_codes_by_str,
00195 onigenc_minimum_property_name_to_ctype,
00196 cp1251_is_code_ctype,
00197 onigenc_not_support_get_ctype_code_range,
00198 onigenc_single_byte_left_adjust_char_head,
00199 onigenc_always_true_is_allowed_reverse_match
00200 };
00201
00202
00203
00204
00205
00206
00207
00208 ENC_ALIAS("CP1251", "Windows-1251")
00209