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_ISO_8859_15_TO_LOWER_CASE(c) EncISO_8859_15_ToLowerCaseTable[c]
00033 #define ENC_IS_ISO_8859_15_CTYPE(code,ctype) \
00034 ((EncISO_8859_15_CtypeTable[code] & CTYPE_TO_BIT(ctype)) != 0)
00035
00036 static const UChar EncISO_8859_15_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', '\250', '\247',
00058 '\250', '\251', '\252', '\253', '\254', '\255', '\256', '\257',
00059 '\260', '\261', '\262', '\263', '\270', '\265', '\266', '\267',
00060 '\270', '\271', '\272', '\273', '\275', '\275', '\377', '\277',
00061 '\340', '\341', '\342', '\343', '\344', '\345', '\346', '\347',
00062 '\350', '\351', '\352', '\353', '\354', '\355', '\356', '\357',
00063 '\360', '\361', '\362', '\363', '\364', '\365', '\366', '\327',
00064 '\370', '\371', '\372', '\373', '\374', '\375', '\376', '\337',
00065 '\340', '\341', '\342', '\343', '\344', '\345', '\346', '\347',
00066 '\350', '\351', '\352', '\353', '\354', '\355', '\356', '\357',
00067 '\360', '\361', '\362', '\363', '\364', '\365', '\366', '\367',
00068 '\370', '\371', '\372', '\373', '\374', '\375', '\376', '\377'
00069 };
00070
00071 static const unsigned short EncISO_8859_15_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 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
00089 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
00090 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
00091 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008, 0x0008,
00092 0x0284, 0x01a0, 0x00a0, 0x00a0, 0x00a0, 0x00a0, 0x34a2, 0x00a0,
00093 0x30e2, 0x00a0, 0x30e2, 0x01a0, 0x00a0, 0x01a0, 0x00a0, 0x00a0,
00094 0x00a0, 0x00a0, 0x10a0, 0x10a0, 0x34a2, 0x30e2, 0x00a0, 0x01a0,
00095 0x30e2, 0x10a0, 0x30e2, 0x01a0, 0x34a2, 0x30e2, 0x34a2, 0x01a0,
00096 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00097 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2,
00098 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x00a0,
00099 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x34a2, 0x30e2,
00100 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00101 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2,
00102 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x00a0,
00103 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2, 0x30e2
00104 };
00105
00106 static int
00107 mbc_case_fold(OnigCaseFoldType flag,
00108 const UChar** pp, const UChar* end ARG_UNUSED,
00109 UChar* lower, OnigEncoding enc ARG_UNUSED)
00110 {
00111 const UChar* p = *pp;
00112
00113 if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
00114 *lower++ = 's';
00115 *lower = 's';
00116 (*pp)++;
00117 return 2;
00118 }
00119
00120 *lower = ENC_ISO_8859_15_TO_LOWER_CASE(*p);
00121 (*pp)++;
00122 return 1;
00123 }
00124
00125 #if 0
00126 static int
00127 is_mbc_ambiguous(OnigCaseFoldType flag,
00128 const UChar** pp, const UChar* end)
00129 {
00130 int v;
00131 const UChar* p = *pp;
00132
00133 if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
00134 (*pp)++;
00135 return TRUE;
00136 }
00137
00138 (*pp)++;
00139 v = (EncISO_8859_15_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
00140 if ((v | BIT_CTYPE_LOWER) != 0) {
00141
00142 if (*p == 0xaa || *p == 0xb5 || *p == 0xba)
00143 return FALSE;
00144 else
00145 return TRUE;
00146 }
00147
00148 return (v != 0 ? TRUE : FALSE);
00149 }
00150 #endif
00151
00152 static int
00153 is_code_ctype(OnigCodePoint code, unsigned int ctype, OnigEncoding enc ARG_UNUSED)
00154 {
00155 if (code < 256)
00156 return ENC_IS_ISO_8859_15_CTYPE(code, ctype);
00157 else
00158 return FALSE;
00159 }
00160
00161 static const OnigPairCaseFoldCodes CaseFoldMap[] = {
00162 { 0xa6, 0xa8 },
00163
00164 { 0xb4, 0xb8 },
00165 { 0xbc, 0xbd },
00166 { 0xbe, 0xff },
00167
00168 { 0xc0, 0xe0 },
00169 { 0xc1, 0xe1 },
00170 { 0xc2, 0xe2 },
00171 { 0xc3, 0xe3 },
00172 { 0xc4, 0xe4 },
00173 { 0xc5, 0xe5 },
00174 { 0xc6, 0xe6 },
00175 { 0xc7, 0xe7 },
00176 { 0xc8, 0xe8 },
00177 { 0xc9, 0xe9 },
00178 { 0xca, 0xea },
00179 { 0xcb, 0xeb },
00180 { 0xcc, 0xec },
00181 { 0xcd, 0xed },
00182 { 0xce, 0xee },
00183 { 0xcf, 0xef },
00184
00185 { 0xd0, 0xf0 },
00186 { 0xd1, 0xf1 },
00187 { 0xd2, 0xf2 },
00188 { 0xd3, 0xf3 },
00189 { 0xd4, 0xf4 },
00190 { 0xd5, 0xf5 },
00191 { 0xd6, 0xf6 },
00192 { 0xd8, 0xf8 },
00193 { 0xd9, 0xf9 },
00194 { 0xda, 0xfa },
00195 { 0xdb, 0xfb },
00196 { 0xdc, 0xfc },
00197 { 0xdd, 0xfd },
00198 { 0xde, 0xfe }
00199 };
00200
00201 static int
00202 apply_all_case_fold(OnigCaseFoldType flag,
00203 OnigApplyAllCaseFoldFunc f, void* arg,
00204 OnigEncoding enc ARG_UNUSED)
00205 {
00206 return onigenc_apply_all_case_fold_with_map(
00207 sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 1,
00208 flag, f, arg);
00209 }
00210
00211 static int
00212 get_case_fold_codes_by_str(OnigCaseFoldType flag,
00213 const OnigUChar* p, const OnigUChar* end,
00214 OnigCaseFoldCodeItem items[],
00215 OnigEncoding enc ARG_UNUSED)
00216 {
00217 return onigenc_get_case_fold_codes_by_str_with_map(
00218 sizeof(CaseFoldMap)/sizeof(OnigPairCaseFoldCodes), CaseFoldMap, 1,
00219 flag, p, end, items);
00220 }
00221
00222 OnigEncodingDefine(iso_8859_15, ISO_8859_15) = {
00223 onigenc_single_byte_mbc_enc_len,
00224 "ISO-8859-15",
00225 1,
00226 1,
00227 onigenc_is_mbc_newline_0x0a,
00228 onigenc_single_byte_mbc_to_code,
00229 onigenc_single_byte_code_to_mbclen,
00230 onigenc_single_byte_code_to_mbc,
00231 mbc_case_fold,
00232 apply_all_case_fold,
00233 get_case_fold_codes_by_str,
00234 onigenc_minimum_property_name_to_ctype,
00235 is_code_ctype,
00236 onigenc_not_support_get_ctype_code_range,
00237 onigenc_single_byte_left_adjust_char_head,
00238 onigenc_always_true_is_allowed_reverse_match
00239 };
00240 ENC_ALIAS("ISO8859-15", "ISO-8859-15")
00241