| 1 |
/* $Header$ */ |
| 2 |
/* |
| 3 |
* regc_locale.c -- |
| 4 |
* |
| 5 |
* This file contains the Unicode locale specific regexp routines. |
| 6 |
* This file is #included by regcomp.c. |
| 7 |
* |
| 8 |
* Copyright (c) 1998 by Scriptics Corporation. |
| 9 |
* |
| 10 |
* See the file "license.terms" for information on usage and redistribution |
| 11 |
* of this file, and for a DISCLAIMER OF ALL WARRANTIES. |
| 12 |
* |
| 13 |
* RCS: @(#) $Id: regc_locale.c,v 1.1.1.1 2001/06/13 04:32:00 dtashley Exp $ |
| 14 |
*/ |
| 15 |
|
| 16 |
/* ASCII character-name table */ |
| 17 |
|
| 18 |
static struct cname { |
| 19 |
char *name; |
| 20 |
char code; |
| 21 |
} cnames[] = { |
| 22 |
{"NUL", '\0'}, |
| 23 |
{"SOH", '\001'}, |
| 24 |
{"STX", '\002'}, |
| 25 |
{"ETX", '\003'}, |
| 26 |
{"EOT", '\004'}, |
| 27 |
{"ENQ", '\005'}, |
| 28 |
{"ACK", '\006'}, |
| 29 |
{"BEL", '\007'}, |
| 30 |
{"alert", '\007'}, |
| 31 |
{"BS", '\010'}, |
| 32 |
{"backspace", '\b'}, |
| 33 |
{"HT", '\011'}, |
| 34 |
{"tab", '\t'}, |
| 35 |
{"LF", '\012'}, |
| 36 |
{"newline", '\n'}, |
| 37 |
{"VT", '\013'}, |
| 38 |
{"vertical-tab", '\v'}, |
| 39 |
{"FF", '\014'}, |
| 40 |
{"form-feed", '\f'}, |
| 41 |
{"CR", '\015'}, |
| 42 |
{"carriage-return", '\r'}, |
| 43 |
{"SO", '\016'}, |
| 44 |
{"SI", '\017'}, |
| 45 |
{"DLE", '\020'}, |
| 46 |
{"DC1", '\021'}, |
| 47 |
{"DC2", '\022'}, |
| 48 |
{"DC3", '\023'}, |
| 49 |
{"DC4", '\024'}, |
| 50 |
{"NAK", '\025'}, |
| 51 |
{"SYN", '\026'}, |
| 52 |
{"ETB", '\027'}, |
| 53 |
{"CAN", '\030'}, |
| 54 |
{"EM", '\031'}, |
| 55 |
{"SUB", '\032'}, |
| 56 |
{"ESC", '\033'}, |
| 57 |
{"IS4", '\034'}, |
| 58 |
{"FS", '\034'}, |
| 59 |
{"IS3", '\035'}, |
| 60 |
{"GS", '\035'}, |
| 61 |
{"IS2", '\036'}, |
| 62 |
{"RS", '\036'}, |
| 63 |
{"IS1", '\037'}, |
| 64 |
{"US", '\037'}, |
| 65 |
{"space", ' '}, |
| 66 |
{"exclamation-mark", '!'}, |
| 67 |
{"quotation-mark", '"'}, |
| 68 |
{"number-sign", '#'}, |
| 69 |
{"dollar-sign", '$'}, |
| 70 |
{"percent-sign", '%'}, |
| 71 |
{"ampersand", '&'}, |
| 72 |
{"apostrophe", '\''}, |
| 73 |
{"left-parenthesis", '('}, |
| 74 |
{"right-parenthesis", ')'}, |
| 75 |
{"asterisk", '*'}, |
| 76 |
{"plus-sign", '+'}, |
| 77 |
{"comma", ','}, |
| 78 |
{"hyphen", '-'}, |
| 79 |
{"hyphen-minus", '-'}, |
| 80 |
{"period", '.'}, |
| 81 |
{"full-stop", '.'}, |
| 82 |
{"slash", '/'}, |
| 83 |
{"solidus", '/'}, |
| 84 |
{"zero", '0'}, |
| 85 |
{"one", '1'}, |
| 86 |
{"two", '2'}, |
| 87 |
{"three", '3'}, |
| 88 |
{"four", '4'}, |
| 89 |
{"five", '5'}, |
| 90 |
{"six", '6'}, |
| 91 |
{"seven", '7'}, |
| 92 |
{"eight", '8'}, |
| 93 |
{"nine", '9'}, |
| 94 |
{"colon", ':'}, |
| 95 |
{"semicolon", ';'}, |
| 96 |
{"less-than-sign", '<'}, |
| 97 |
{"equals-sign", '='}, |
| 98 |
{"greater-than-sign", '>'}, |
| 99 |
{"question-mark", '?'}, |
| 100 |
{"commercial-at", '@'}, |
| 101 |
{"left-square-bracket", '['}, |
| 102 |
{"backslash", '\\'}, |
| 103 |
{"reverse-solidus", '\\'}, |
| 104 |
{"right-square-bracket", ']'}, |
| 105 |
{"circumflex", '^'}, |
| 106 |
{"circumflex-accent", '^'}, |
| 107 |
{"underscore", '_'}, |
| 108 |
{"low-line", '_'}, |
| 109 |
{"grave-accent", '`'}, |
| 110 |
{"left-brace", '{'}, |
| 111 |
{"left-curly-bracket", '{'}, |
| 112 |
{"vertical-line", '|'}, |
| 113 |
{"right-brace", '}'}, |
| 114 |
{"right-curly-bracket", '}'}, |
| 115 |
{"tilde", '~'}, |
| 116 |
{"DEL", '\177'}, |
| 117 |
{NULL, 0} |
| 118 |
}; |
| 119 |
|
| 120 |
/* Unicode character-class tables */ |
| 121 |
|
| 122 |
typedef struct crange { |
| 123 |
chr start; |
| 124 |
chr end; |
| 125 |
} crange; |
| 126 |
|
| 127 |
/* Unicode: (Alphabetic) */ |
| 128 |
|
| 129 |
static crange alphaRangeTable[] = { |
| 130 |
{0x0041, 0x005a}, {0x0061, 0x007a}, {0x00c0, 0x00d6}, {0x00d8, 0x00f6}, |
| 131 |
{0x00f8, 0x01f5}, {0x01fa, 0x0217}, {0x0250, 0x02a8}, {0x02b0, 0x02b8}, |
| 132 |
{0x02bb, 0x02c1}, {0x02e0, 0x02e4}, {0x0388, 0x038a}, {0x038e, 0x03a1}, |
| 133 |
{0x03a3, 0x03ce}, {0x03d0, 0x03d6}, {0x03e2, 0x03f3}, {0x0401, 0x040c}, |
| 134 |
{0x040e, 0x044f}, {0x0451, 0x045c}, {0x045e, 0x0481}, {0x0490, 0x04c4}, |
| 135 |
{0x04d0, 0x04eb}, {0x04ee, 0x04f5}, {0x0531, 0x0556}, {0x0561, 0x0587}, |
| 136 |
{0x05d0, 0x05ea}, {0x05f0, 0x05f2}, {0x0621, 0x063a}, {0x0640, 0x064a}, |
| 137 |
{0x0671, 0x06b7}, {0x06ba, 0x06be}, {0x06c0, 0x06ce}, {0x06d0, 0x06d3}, |
| 138 |
{0x0905, 0x0939}, {0x0958, 0x0961}, {0x0985, 0x098c}, {0x0993, 0x09a8}, |
| 139 |
{0x09aa, 0x09b0}, {0x09b6, 0x09b9}, {0x09df, 0x09e1}, {0x0a05, 0x0a0a}, |
| 140 |
{0x0a13, 0x0a28}, {0x0a2a, 0x0a30}, {0x0a59, 0x0a5c}, {0x0a72, 0x0a74}, |
| 141 |
{0x0a85, 0x0a8b}, {0x0a8f, 0x0a91}, {0x0a93, 0x0aa8}, {0x0aaa, 0x0ab0}, |
| 142 |
{0x0ab5, 0x0ab9}, {0x0b05, 0x0b0c}, {0x0b13, 0x0b28}, {0x0b2a, 0x0b30}, |
| 143 |
{0x0b36, 0x0b39}, {0x0b5f, 0x0b61}, {0x0b85, 0x0b8a}, {0x0b8e, 0x0b90}, |
| 144 |
{0x0b92, 0x0b95}, {0x0ba8, 0x0baa}, {0x0bae, 0x0bb5}, {0x0bb7, 0x0bb9}, |
| 145 |
{0x0c05, 0x0c0c}, {0x0c0e, 0x0c10}, {0x0c12, 0x0c28}, {0x0c2a, 0x0c33}, |
| 146 |
{0x0c35, 0x0c39}, {0x0c85, 0x0c8c}, {0x0c8e, 0x0c90}, {0x0c92, 0x0ca8}, |
| 147 |
{0x0caa, 0x0cb3}, {0x0cb5, 0x0cb9}, {0x0d05, 0x0d0c}, {0x0d0e, 0x0d10}, |
| 148 |
{0x0d12, 0x0d28}, {0x0d2a, 0x0d39}, {0x0e01, 0x0e30}, {0x0e40, 0x0e46}, |
| 149 |
{0x0e94, 0x0e97}, {0x0e99, 0x0e9f}, {0x0ea1, 0x0ea3}, {0x0ead, 0x0eb0}, |
| 150 |
{0x0ec0, 0x0ec4}, {0x0f40, 0x0f47}, {0x0f49, 0x0f69}, {0x0f88, 0x0f8b}, |
| 151 |
{0x10a0, 0x10c5}, {0x10d0, 0x10f6}, {0x1100, 0x1159}, {0x115f, 0x11a2}, |
| 152 |
{0x11a8, 0x11f9}, {0x1e00, 0x1e9b}, {0x1ea0, 0x1ef9}, {0x1f00, 0x1f15}, |
| 153 |
{0x1f18, 0x1f1d}, {0x1f20, 0x1f45}, {0x1f48, 0x1f4d}, {0x1f50, 0x1f57}, |
| 154 |
{0x1f5f, 0x1f7d}, {0x1f80, 0x1fb4}, {0x1fb6, 0x1fbc}, {0x1fc2, 0x1fc4}, |
| 155 |
{0x1fc6, 0x1fcc}, {0x1fd0, 0x1fd3}, {0x1fd6, 0x1fdb}, {0x1fe0, 0x1fec}, |
| 156 |
{0x1ff2, 0x1ff4}, {0x1ff6, 0x1ffc}, {0x210a, 0x2113}, {0x2118, 0x211d}, |
| 157 |
{0x212a, 0x2131}, {0x2133, 0x2138}, {0x3031, 0x3035}, {0x3041, 0x3094}, |
| 158 |
{0x30a1, 0x30fa}, {0x30fc, 0x30fe}, {0x3105, 0x312c}, {0x3131, 0x318e}, |
| 159 |
{0x4e00, 0x9fa5}, {0xac00, 0xd7a3}, {0xf900, 0xfa2d}, {0xfb00, 0xfb06}, |
| 160 |
{0xfb13, 0xfb17}, {0xfb1f, 0xfb28}, {0xfb2a, 0xfb36}, {0xfb38, 0xfb3c}, |
| 161 |
{0xfb46, 0xfbb1}, {0xfbd3, 0xfd3d}, {0xfd50, 0xfd8f}, {0xfd92, 0xfdc7}, |
| 162 |
{0xfdf0, 0xfdfb}, {0xfe70, 0xfe72}, {0xfe76, 0xfefc}, {0xff21, 0xff3a}, |
| 163 |
{0xff41, 0xff5a}, {0xff66, 0xffbe}, {0xffc2, 0xffc7}, {0xffca, 0xffcf}, |
| 164 |
{0xffd2, 0xffd7}, {0xffda, 0xffdc} |
| 165 |
}; |
| 166 |
|
| 167 |
#define NUM_ALPHA_RANGE (sizeof(alphaRangeTable)/sizeof(crange)) |
| 168 |
|
| 169 |
static chr alphaCharTable[] = { |
| 170 |
0x00aa, 0x00b5, 0x00ba, 0x02d0, 0x02d1, 0x037a, 0x0386, 0x038c, 0x03da, |
| 171 |
0x03dc, 0x03de, 0x03e0, 0x04c7, 0x04c8, 0x04cb, 0x04cc, 0x04f8, 0x04f9, |
| 172 |
0x0559, 0x06d5, 0x06e5, 0x06e6, 0x093d, 0x0950, 0x098f, 0x0990, 0x09b2, |
| 173 |
0x09dc, 0x09dd, 0x09f0, 0x09f1, 0x0a0f, 0x0a10, 0x0a32, 0x0a33, 0x0a35, |
| 174 |
0x0a36, 0x0a38, 0x0a39, 0x0a5e, 0x0a8d, 0x0ab2, 0x0ab3, 0x0abd, 0x0ad0, |
| 175 |
0x0ae0, 0x0b0f, 0x0b10, 0x0b32, 0x0b33, 0x0b3d, 0x0b5c, 0x0b5d, 0x0b99, |
| 176 |
0x0b9a, 0x0b9c, 0x0b9e, 0x0b9f, 0x0ba3, 0x0ba4, 0x0c60, 0x0c61, 0x0cde, |
| 177 |
0x0ce0, 0x0ce1, 0x0d60, 0x0d61, 0x0e32, 0x0e33, 0x0e81, 0x0e82, 0x0e84, |
| 178 |
0x0e87, 0x0e88, 0x0e8a, 0x0e8d, 0x0ea5, 0x0ea7, 0x0eaa, 0x0eab, 0x0eb2, |
| 179 |
0x0eb3, 0x0ebd, 0x0ec6, 0x0edc, 0x0edd, 0x0f00, 0x1f59, 0x1f5b, 0x1f5d, |
| 180 |
0x1fbe, 0x207f, 0x2102, 0x2107, 0x2115, 0x2124, 0x2126, 0x2128, 0x3005, |
| 181 |
0x3006, 0x309d, 0x309e, 0xfb3e, 0xfb40, 0xfb41, 0xfb43, 0xfb44, 0xfe74 |
| 182 |
}; |
| 183 |
|
| 184 |
#define NUM_ALPHA_CHAR (sizeof(alphaCharTable)/sizeof(chr)) |
| 185 |
|
| 186 |
/* Unicode: (Decimal digit) */ |
| 187 |
|
| 188 |
static crange digitRangeTable[] = { |
| 189 |
{0x0030, 0x0039}, {0x0660, 0x0669}, {0x06f0, 0x06f9}, {0x0966, 0x096f}, |
| 190 |
{0x09e6, 0x09ef}, {0x0a66, 0x0a6f}, {0x0ae6, 0x0aef}, {0x0b66, 0x0b6f}, |
| 191 |
{0x0be7, 0x0bef}, {0x0c66, 0x0c6f}, {0x0ce6, 0x0cef}, {0x0d66, 0x0d6f}, |
| 192 |
{0x0e50, 0x0e59}, {0x0ed0, 0x0ed9}, {0x0f20, 0x0f29}, {0xff10, 0xff19} |
| 193 |
}; |
| 194 |
|
| 195 |
#define NUM_DIGIT_RANGE (sizeof(digitRangeTable)/sizeof(crange)) |
| 196 |
|
| 197 |
/* Unicode: (Punctuation) */ |
| 198 |
|
| 199 |
static crange punctRangeTable[] = { |
| 200 |
{0x0021, 0x0023}, {0x0025, 0x002a}, {0x002c, 0x002f}, {0x005b, 0x005d}, |
| 201 |
{0x055a, 0x055f}, {0x066a, 0x066d}, {0x0f04, 0x0f12}, {0x0f3a, 0x0f3d}, |
| 202 |
{0x2010, 0x2027}, {0x2030, 0x2043}, {0x3001, 0x3003}, {0x3008, 0x3011}, |
| 203 |
{0x3014, 0x301f}, {0xfe30, 0xfe44}, {0xfe49, 0xfe52}, {0xfe54, 0xfe61}, |
| 204 |
{0xff01, 0xff03}, {0xff05, 0xff0a}, {0xff0c, 0xff0f}, {0xff3b, 0xff3d}, |
| 205 |
{0xff61, 0xff65} |
| 206 |
}; |
| 207 |
|
| 208 |
#define NUM_PUNCT_RANGE (sizeof(punctRangeTable)/sizeof(crange)) |
| 209 |
|
| 210 |
static chr punctCharTable[] = { |
| 211 |
0x003a, 0x003b, 0x003f, 0x0040, 0x005f, 0x007b, 0x007d, 0x00a1, 0x00ab, |
| 212 |
0x00ad, 0x00b7, 0x00bb, 0x00bf, 0x037e, 0x0387, 0x0589, 0x05be, 0x05c0, |
| 213 |
0x05c3, 0x05f3, 0x05f4, 0x060c, 0x061b, 0x061f, 0x06d4, 0x0964, 0x0965, |
| 214 |
0x0970, 0x0e5a, 0x0e5b, 0x0f85, 0x10fb, 0x2045, 0x2046, 0x207d, 0x207e, |
| 215 |
0x208d, 0x208e, 0x2329, 0x232a, 0x3030, 0x30fb, 0xfd3e, 0xfd3f, 0xfe63, |
| 216 |
0xfe68, 0xfe6a, 0xfe6b, 0xff1a, 0xff1b, 0xff1f, 0xff20, 0xff3f, 0xff5b, |
| 217 |
0xff5d |
| 218 |
}; |
| 219 |
|
| 220 |
#define NUM_PUNCT_CHAR (sizeof(punctCharTable)/sizeof(chr)) |
| 221 |
|
| 222 |
/* Unicode: (White space) */ |
| 223 |
|
| 224 |
static crange spaceRangeTable[] = { |
| 225 |
{0x0009, 0x000d}, {0x2000, 0x200b}, |
| 226 |
}; |
| 227 |
|
| 228 |
#define NUM_SPACE_RANGE (sizeof(spaceRangeTable)/sizeof(crange)) |
| 229 |
|
| 230 |
static chr spaceCharTable[] = { |
| 231 |
0x0020, 0x00a0, 0x2028, 0x2029, 0x3000 |
| 232 |
}; |
| 233 |
|
| 234 |
#define NUM_SPACE_CHAR (sizeof(spaceCharTable)/sizeof(chr)) |
| 235 |
|
| 236 |
/* Unicode: lowercase characters */ |
| 237 |
|
| 238 |
static crange lowerRangeTable[] = { |
| 239 |
{0x0061, 0x007a}, {0x00df, 0x00f6}, {0x00f8, 0x00ff}, {0x017e, 0x0180}, |
| 240 |
{0x0199, 0x019b}, {0x0250, 0x02a8}, {0x03ac, 0x03ce}, {0x03ef, 0x03f2}, |
| 241 |
{0x0430, 0x044f}, {0x0451, 0x045c}, {0x0561, 0x0587}, {0x10d0, 0x10f6}, |
| 242 |
{0x1e95, 0x1e9b}, {0x1f00, 0x1f07}, {0x1f10, 0x1f15}, {0x1f20, 0x1f27}, |
| 243 |
{0x1f30, 0x1f37}, {0x1f40, 0x1f45}, {0x1f50, 0x1f57}, {0x1f60, 0x1f67}, |
| 244 |
{0x1f70, 0x1f7d}, {0x1f80, 0x1f87}, {0x1f90, 0x1f97}, {0x1fa0, 0x1fa7}, |
| 245 |
{0x1fb0, 0x1fb4}, {0x1fc2, 0x1fc4}, {0x1fd0, 0x1fd3}, {0x1fe0, 0x1fe7}, |
| 246 |
{0x1ff2, 0x1ff4}, {0xfb00, 0xfb06}, {0xfb13, 0xfb17}, {0xff41, 0xff5a} |
| 247 |
}; |
| 248 |
|
| 249 |
#define NUM_LOWER_RANGE (sizeof(lowerRangeTable)/sizeof(crange)) |
| 250 |
|
| 251 |
static chr lowerCharTable[] = { |
| 252 |
0x00aa, 0x00b5, 0x00ba, 0x0101, 0x0103, 0x0105, 0x0107, 0x0109, 0x010b, |
| 253 |
0x010d, 0x010f, 0x0111, 0x0113, 0x0115, 0x0117, 0x0119, 0x011b, 0x011d, |
| 254 |
0x011f, 0x0121, 0x0123, 0x0125, 0x0127, 0x0129, 0x012b, 0x012d, 0x012f, |
| 255 |
0x0131, 0x0133, 0x0135, 0x0137, 0x0138, 0x013a, 0x013c, 0x013e, 0x0140, |
| 256 |
0x0142, 0x0144, 0x0146, 0x0148, 0x0149, 0x014b, 0x014d, 0x014f, 0x0151, |
| 257 |
0x0153, 0x0155, 0x0157, 0x0159, 0x015b, 0x015d, 0x015f, 0x0161, 0x0163, |
| 258 |
0x0165, 0x0167, 0x0169, 0x016b, 0x016d, 0x016f, 0x0171, 0x0173, 0x0175, |
| 259 |
0x0177, 0x017a, 0x017c, 0x0183, 0x0185, 0x0188, 0x018c, 0x018d, 0x0192, |
| 260 |
0x0195, 0x019e, 0x01a1, 0x01a3, 0x01a5, 0x01a8, 0x01ab, 0x01ad, 0x01b0, |
| 261 |
0x01b4, 0x01b6, 0x01b9, 0x01ba, 0x01bd, 0x01c6, 0x01c9, 0x01cc, 0x01ce, |
| 262 |
0x01d0, 0x01d2, 0x01d4, 0x01d6, 0x01d8, 0x01da, 0x01dc, 0x01dd, 0x01df, |
| 263 |
0x01e1, 0x01e3, 0x01e5, 0x01e7, 0x01e9, 0x01eb, 0x01ed, 0x01ef, 0x01f0, |
| 264 |
0x01f3, 0x01f5, 0x01fb, 0x01fd, 0x01ff, 0x0201, 0x0203, 0x0205, 0x0207, |
| 265 |
0x0209, 0x020b, 0x020d, 0x020f, 0x0211, 0x0213, 0x0215, 0x0217, 0x0390, |
| 266 |
0x03d0, 0x03d1, 0x03d5, 0x03d6, 0x03e3, 0x03e5, 0x03e7, 0x03e9, 0x03eb, |
| 267 |
0x03ed, 0x045e, 0x045f, 0x0461, 0x0463, 0x0465, 0x0467, 0x0469, 0x046b, |
| 268 |
0x046d, 0x046f, 0x0471, 0x0473, 0x0475, 0x0477, 0x0479, 0x047b, 0x047d, |
| 269 |
0x047f, 0x0481, 0x0491, 0x0493, 0x0495, 0x0497, 0x0499, 0x049b, 0x049d, |
| 270 |
0x049f, 0x04a1, 0x04a3, 0x04a5, 0x04a7, 0x04a9, 0x04ab, 0x04ad, 0x04af, |
| 271 |
0x04b1, 0x04b3, 0x04b5, 0x04b7, 0x04b9, 0x04bb, 0x04bd, 0x04bf, 0x04c2, |
| 272 |
0x04c4, 0x04c8, 0x04cc, 0x04d1, 0x04d3, 0x04d5, 0x04d7, 0x04d9, 0x04db, |
| 273 |
0x04dd, 0x04df, 0x04e1, 0x04e3, 0x04e5, 0x04e7, 0x04e9, 0x04eb, 0x04ef, |
| 274 |
0x04f1, 0x04f3, 0x04f5, 0x04f9, 0x1e01, 0x1e03, 0x1e05, 0x1e07, 0x1e09, |
| 275 |
0x1e0b, 0x1e0d, 0x1e0f, 0x1e11, 0x1e13, 0x1e15, 0x1e17, 0x1e19, 0x1e1b, |
| 276 |
0x1e1d, 0x1e1f, 0x1e21, 0x1e23, 0x1e25, 0x1e27, 0x1e29, 0x1e2b, 0x1e2d, |
| 277 |
0x1e2f, 0x1e31, 0x1e33, 0x1e35, 0x1e37, 0x1e39, 0x1e3b, 0x1e3d, 0x1e3f, |
| 278 |
0x1e41, 0x1e43, 0x1e45, 0x1e47, 0x1e49, 0x1e4b, 0x1e4d, 0x1e4f, 0x1e51, |
| 279 |
0x1e53, 0x1e55, 0x1e57, 0x1e59, 0x1e5b, 0x1e5d, 0x1e5f, 0x1e61, 0x1e63, |
| 280 |
0x1e65, 0x1e67, 0x1e69, 0x1e6b, 0x1e6d, 0x1e6f, 0x1e71, 0x1e73, 0x1e75, |
| 281 |
0x1e77, 0x1e79, 0x1e7b, 0x1e7d, 0x1e7f, 0x1e81, 0x1e83, 0x1e85, 0x1e87, |
| 282 |
0x1e89, 0x1e8b, 0x1e8d, 0x1e8f, 0x1e91, 0x1e93, 0x1ea1, 0x1ea3, 0x1ea5, |
| 283 |
0x1ea7, 0x1ea9, 0x1eab, 0x1ead, 0x1eaf, 0x1eb1, 0x1eb3, 0x1eb5, 0x1eb7, |
| 284 |
0x1eb9, 0x1ebb, 0x1ebd, 0x1ebf, 0x1ec1, 0x1ec3, 0x1ec5, 0x1ec7, 0x1ec9, |
| 285 |
0x1ecb, 0x1ecd, 0x1ecf, 0x1ed1, 0x1ed3, 0x1ed5, 0x1ed7, 0x1ed9, 0x1edb, |
| 286 |
0x1edd, 0x1edf, 0x1ee1, 0x1ee3, 0x1ee5, 0x1ee7, 0x1ee9, 0x1eeb, 0x1eed, |
| 287 |
0x1eef, 0x1ef1, 0x1ef3, 0x1ef5, 0x1ef7, 0x1ef9, 0x1fb6, 0x1fb7, 0x1fbe, |
| 288 |
0x1fc6, 0x1fc7, 0x1fd6, 0x1fd7, 0x1ff6, 0x1ff7, 0x207f, 0x210a, 0x210e, |
| 289 |
0x210f, 0x2113, 0x2118, 0x212e, 0x212f, 0x2134 |
| 290 |
}; |
| 291 |
|
| 292 |
#define NUM_LOWER_CHAR (sizeof(lowerCharTable)/sizeof(chr)) |
| 293 |
|
| 294 |
/* Unicode: uppercase characters */ |
| 295 |
|
| 296 |
static crange upperRangeTable[] = { |
| 297 |
{0x0041, 0x005a}, {0x00c0, 0x00d6}, {0x00d8, 0x00de}, {0x0189, 0x018b}, |
| 298 |
{0x018e, 0x0191}, {0x0196, 0x0198}, {0x01b1, 0x01b3}, {0x0388, 0x038a}, |
| 299 |
{0x0391, 0x03a1}, {0x03a3, 0x03ab}, {0x03d2, 0x03d4}, {0x0401, 0x040c}, |
| 300 |
{0x040e, 0x042f}, {0x0531, 0x0556}, {0x10a0, 0x10c5}, {0x1f08, 0x1f0f}, |
| 301 |
{0x1f18, 0x1f1d}, {0x1f28, 0x1f2f}, {0x1f38, 0x1f3f}, {0x1f48, 0x1f4d}, |
| 302 |
{0x1f68, 0x1f6f}, {0x1f88, 0x1f8f}, {0x1f98, 0x1f9f}, {0x1fa8, 0x1faf}, |
| 303 |
{0x1fb8, 0x1fbc}, {0x1fc8, 0x1fcc}, {0x1fd8, 0x1fdb}, {0x1fe8, 0x1fec}, |
| 304 |
{0x1ff8, 0x1ffc}, {0x210b, 0x210d}, {0x2110, 0x2112}, {0x2119, 0x211d}, |
| 305 |
{0x212a, 0x212d}, {0xff21, 0xff3a} |
| 306 |
}; |
| 307 |
|
| 308 |
#define NUM_UPPER_RANGE (sizeof(upperRangeTable)/sizeof(crange)) |
| 309 |
|
| 310 |
static chr upperCharTable[] = { |
| 311 |
0x0100, 0x0102, 0x0104, 0x0106, 0x0108, 0x010a, 0x010c, 0x010e, 0x0110, |
| 312 |
0x0112, 0x0114, 0x0116, 0x0118, 0x011a, 0x011c, 0x011e, 0x0120, 0x0122, |
| 313 |
0x0124, 0x0126, 0x0128, 0x012a, 0x012c, 0x012e, 0x0130, 0x0132, 0x0134, |
| 314 |
0x0136, 0x0139, 0x013b, 0x013d, 0x013f, 0x0141, 0x0143, 0x0145, 0x0147, |
| 315 |
0x014a, 0x014c, 0x014e, 0x0150, 0x0152, 0x0154, 0x0156, 0x0158, 0x015a, |
| 316 |
0x015c, 0x015e, 0x0160, 0x0162, 0x0164, 0x0166, 0x0168, 0x016a, 0x016c, |
| 317 |
0x016e, 0x0170, 0x0172, 0x0174, 0x0176, 0x0178, 0x0179, 0x017b, 0x017d, |
| 318 |
0x0181, 0x0182, 0x0184, 0x0186, 0x0187, 0x0193, 0x0194, 0x019c, 0x019d, |
| 319 |
0x019f, 0x01a0, 0x01a2, 0x01a4, 0x01a6, 0x01a7, 0x01a9, 0x01ac, 0x01ae, |
| 320 |
0x01af, 0x01b5, 0x01b7, 0x01b8, 0x01bc, 0x01c4, 0x01c7, 0x01ca, 0x01cd, |
| 321 |
0x01cf, 0x01d1, 0x01d3, 0x01d5, 0x01d7, 0x01d9, 0x01db, 0x01de, 0x01e0, |
| 322 |
0x01e2, 0x01e4, 0x01e6, 0x01e8, 0x01ea, 0x01ec, 0x01ee, 0x01f1, 0x01f4, |
| 323 |
0x01fa, 0x01fc, 0x01fe, 0x0200, 0x0202, 0x0204, 0x0206, 0x0208, 0x020a, |
| 324 |
0x020c, 0x020e, 0x0210, 0x0212, 0x0214, 0x0216, 0x0386, 0x038c, 0x038e, |
| 325 |
0x038f, 0x03da, 0x03dc, 0x03de, 0x03e0, 0x03e2, 0x03e4, 0x03e6, 0x03e8, |
| 326 |
0x03ea, 0x03ec, 0x03ee, 0x0460, 0x0462, 0x0464, 0x0466, 0x0468, 0x046a, |
| 327 |
0x046c, 0x046e, 0x0470, 0x0472, 0x0474, 0x0476, 0x0478, 0x047a, 0x047c, |
| 328 |
0x047e, 0x0480, 0x0490, 0x0492, 0x0494, 0x0496, 0x0498, 0x049a, 0x049c, |
| 329 |
0x049e, 0x04a0, 0x04a2, 0x04a4, 0x04a6, 0x04a8, 0x04aa, 0x04ac, 0x04ae, |
| 330 |
0x04b0, 0x04b2, 0x04b4, 0x04b6, 0x04b8, 0x04ba, 0x04bc, 0x04be, 0x04c1, |
| 331 |
0x04c3, 0x04c7, 0x04cb, 0x04d0, 0x04d2, 0x04d4, 0x04d6, 0x04d8, 0x04da, |
| 332 |
0x04dc, 0x04de, 0x04e0, 0x04e2, 0x04e4, 0x04e6, 0x04e8, 0x04ea, 0x04ee, |
| 333 |
0x04f0, 0x04f2, 0x04f4, 0x04f8, 0x1e00, 0x1e02, 0x1e04, 0x1e06, 0x1e08, |
| 334 |
0x1e0a, 0x1e0c, 0x1e0e, 0x1e10, 0x1e12, 0x1e14, 0x1e16, 0x1e18, 0x1e1a, |
| 335 |
0x1e1c, 0x1e1e, 0x1e20, 0x1e22, 0x1e24, 0x1e26, 0x1e28, 0x1e2a, 0x1e2c, |
| 336 |
0x1e2e, 0x1e30, 0x1e32, 0x1e34, 0x1e36, 0x1e38, 0x1e3a, 0x1e3c, 0x1e3e, |
| 337 |
0x1e40, 0x1e42, 0x1e44, 0x1e46, 0x1e48, 0x1e4a, 0x1e4c, 0x1e4e, 0x1e50, |
| 338 |
0x1e52, 0x1e54, 0x1e56, 0x1e58, 0x1e5a, 0x1e5c, 0x1e5e, 0x1e60, 0x1e62, |
| 339 |
0x1e64, 0x1e66, 0x1e68, 0x1e6a, 0x1e6c, 0x1e6e, 0x1e70, 0x1e72, 0x1e74, |
| 340 |
0x1e76, 0x1e78, 0x1e7a, 0x1e7c, 0x1e7e, 0x1e80, 0x1e82, 0x1e84, 0x1e86, |
| 341 |
0x1e88, 0x1e8a, 0x1e8c, 0x1e8e, 0x1e90, 0x1e92, 0x1e94, 0x1ea0, 0x1ea2, |
| 342 |
0x1ea4, 0x1ea6, 0x1ea8, 0x1eaa, 0x1eac, 0x1eae, 0x1eb0, 0x1eb2, 0x1eb4, |
| 343 |
0x1eb6, 0x1eb8, 0x1eba, 0x1ebc, 0x1ebe, 0x1ec0, 0x1ec2, 0x1ec4, 0x1ec6, |
| 344 |
0x1ec8, 0x1eca, 0x1ecc, 0x1ece, 0x1ed0, 0x1ed2, 0x1ed4, 0x1ed6, 0x1ed8, |
| 345 |
0x1eda, 0x1edc, 0x1ede, 0x1ee0, 0x1ee2, 0x1ee4, 0x1ee6, 0x1ee8, 0x1eea, |
| 346 |
0x1eec, 0x1eee, 0x1ef0, 0x1ef2, 0x1ef4, 0x1ef6, 0x1ef8, 0x1f59, 0x1f5b, |
| 347 |
0x1f5d, 0x1f5f, 0x2102, 0x2107, 0x2115, 0x2124, 0x2126, 0x2128, 0x2130, |
| 348 |
0x2131, 0x2133 |
| 349 |
}; |
| 350 |
|
| 351 |
#define NUM_UPPER_CHAR (sizeof(upperCharTable)/sizeof(chr)) |
| 352 |
|
| 353 |
/* |
| 354 |
* The graph table includes the set of characters that are Unicode |
| 355 |
* print characters excluding space. |
| 356 |
*/ |
| 357 |
|
| 358 |
static crange graphRangeTable[] = { |
| 359 |
{0x0021, 0x007e}, {0x00a0, 0x011f}, {0x0121, 0x01f5}, {0x01fa, 0x0217}, |
| 360 |
{0x0250, 0x02a8}, {0x02b0, 0x02de}, {0x02e0, 0x02e9}, {0x0300, 0x031f}, |
| 361 |
{0x0321, 0x0345}, {0x0384, 0x038a}, {0x038e, 0x03a1}, {0x03a3, 0x03ce}, |
| 362 |
{0x03d0, 0x03d6}, {0x03e2, 0x03f3}, {0x0401, 0x040c}, {0x040e, 0x041f}, |
| 363 |
{0x0421, 0x044f}, {0x0451, 0x045c}, {0x045e, 0x0486}, {0x0490, 0x04c4}, |
| 364 |
{0x04d0, 0x04eb}, {0x04ee, 0x04f5}, {0x0531, 0x0556}, {0x0559, 0x055f}, |
| 365 |
{0x0561, 0x0587}, {0x0591, 0x05a1}, {0x05a3, 0x05b9}, {0x05bb, 0x05c4}, |
| 366 |
{0x05d0, 0x05ea}, {0x05f0, 0x05f4}, {0x0621, 0x063a}, {0x0640, 0x0652}, |
| 367 |
{0x0660, 0x066d}, {0x0670, 0x06b7}, {0x06ba, 0x06be}, {0x06c0, 0x06ce}, |
| 368 |
{0x06d0, 0x06ed}, {0x06f0, 0x06f9}, {0x0901, 0x0903}, {0x0905, 0x091f}, |
| 369 |
{0x0921, 0x0939}, {0x093c, 0x094d}, {0x0950, 0x0954}, {0x0958, 0x0970}, |
| 370 |
{0x0981, 0x0983}, {0x0985, 0x098c}, {0x0993, 0x09a8}, {0x09aa, 0x09b0}, |
| 371 |
{0x09b6, 0x09b9}, {0x09be, 0x09c4}, {0x09cb, 0x09cd}, {0x09df, 0x09e3}, |
| 372 |
{0x09e6, 0x09fa}, {0x0a05, 0x0a0a}, {0x0a13, 0x0a1f}, {0x0a21, 0x0a28}, |
| 373 |
{0x0a2a, 0x0a30}, {0x0a3e, 0x0a42}, {0x0a4b, 0x0a4d}, {0x0a59, 0x0a5c}, |
| 374 |
{0x0a66, 0x0a74}, {0x0a81, 0x0a83}, {0x0a85, 0x0a8b}, {0x0a8f, 0x0a91}, |
| 375 |
{0x0a93, 0x0aa8}, {0x0aaa, 0x0ab0}, {0x0ab5, 0x0ab9}, {0x0abc, 0x0ac5}, |
| 376 |
{0x0ac7, 0x0ac9}, {0x0acb, 0x0acd}, {0x0ae6, 0x0aef}, {0x0b01, 0x0b03}, |
| 377 |
{0x0b05, 0x0b0c}, {0x0b13, 0x0b1f}, {0x0b21, 0x0b28}, {0x0b2a, 0x0b30}, |
| 378 |
{0x0b36, 0x0b39}, {0x0b3c, 0x0b43}, {0x0b4b, 0x0b4d}, {0x0b5f, 0x0b61}, |
| 379 |
{0x0b66, 0x0b70}, {0x0b85, 0x0b8a}, {0x0b8e, 0x0b90}, {0x0b92, 0x0b95}, |
| 380 |
{0x0ba8, 0x0baa}, {0x0bae, 0x0bb5}, {0x0bb7, 0x0bb9}, {0x0bbe, 0x0bc2}, |
| 381 |
{0x0bc6, 0x0bc8}, {0x0bca, 0x0bcd}, {0x0be7, 0x0bf2}, {0x0c01, 0x0c03}, |
| 382 |
{0x0c05, 0x0c0c}, {0x0c0e, 0x0c10}, {0x0c12, 0x0c1f}, {0x0c21, 0x0c28}, |
| 383 |
{0x0c2a, 0x0c33}, {0x0c35, 0x0c39}, {0x0c3e, 0x0c44}, {0x0c46, 0x0c48}, |
| 384 |
{0x0c4a, 0x0c4d}, {0x0c66, 0x0c6f}, {0x0c85, 0x0c8c}, {0x0c8e, 0x0c90}, |
| 385 |
{0x0c92, 0x0ca8}, {0x0caa, 0x0cb3}, {0x0cb5, 0x0cb9}, {0x0cbe, 0x0cc4}, |
| 386 |
{0x0cc6, 0x0cc8}, {0x0cca, 0x0ccd}, {0x0ce6, 0x0cef}, {0x0d05, 0x0d0c}, |
| 387 |
{0x0d0e, 0x0d10}, {0x0d12, 0x0d1f}, {0x0d21, 0x0d28}, {0x0d2a, 0x0d39}, |
| 388 |
{0x0d3e, 0x0d43}, {0x0d46, 0x0d48}, {0x0d4a, 0x0d4d}, {0x0d66, 0x0d6f}, |
| 389 |
{0x0e01, 0x0e1f}, {0x0e21, 0x0e3a}, {0x0e3f, 0x0e5b}, {0x0e94, 0x0e97}, |
| 390 |
{0x0e99, 0x0e9f}, {0x0ea1, 0x0ea3}, {0x0ead, 0x0eb9}, {0x0ebb, 0x0ebd}, |
| 391 |
{0x0ec0, 0x0ec4}, {0x0ec8, 0x0ecd}, {0x0ed0, 0x0ed9}, {0x0f00, 0x0f1f}, |
| 392 |
{0x0f21, 0x0f47}, {0x0f49, 0x0f69}, {0x0f71, 0x0f8b}, {0x0f90, 0x0f95}, |
| 393 |
{0x0f99, 0x0fad}, {0x0fb1, 0x0fb7}, {0x10a0, 0x10c5}, {0x10d0, 0x10f6}, |
| 394 |
{0x1100, 0x111f}, {0x1121, 0x1159}, {0x115f, 0x11a2}, {0x11a8, 0x11f9}, |
| 395 |
{0x1e00, 0x1e1f}, {0x1e21, 0x1e9b}, {0x1ea0, 0x1ef9}, {0x1f00, 0x1f15}, |
| 396 |
{0x1f18, 0x1f1d}, {0x1f21, 0x1f45}, {0x1f48, 0x1f4d}, {0x1f50, 0x1f57}, |
| 397 |
{0x1f5f, 0x1f7d}, {0x1f80, 0x1fb4}, {0x1fb6, 0x1fc4}, {0x1fc6, 0x1fd3}, |
| 398 |
{0x1fd6, 0x1fdb}, {0x1fdd, 0x1fef}, {0x1ff2, 0x1ff4}, {0x1ff6, 0x1ffe}, |
| 399 |
{0x2000, 0x200b}, {0x2010, 0x201f}, {0x2021, 0x2029}, {0x2030, 0x2046}, |
| 400 |
{0x2074, 0x208e}, {0x20a0, 0x20ac}, {0x20d0, 0x20e1}, {0x2100, 0x211f}, |
| 401 |
{0x2121, 0x2138}, {0x2153, 0x2182}, {0x2190, 0x21ea}, {0x2200, 0x221f}, |
| 402 |
{0x2221, 0x22f1}, {0x2302, 0x231f}, {0x2321, 0x237a}, {0x2400, 0x241f}, |
| 403 |
{0x2421, 0x2424}, {0x2440, 0x244a}, {0x2460, 0x24ea}, {0x2500, 0x251f}, |
| 404 |
{0x2521, 0x2595}, {0x25a0, 0x25ef}, {0x2600, 0x2613}, {0x261a, 0x261f}, |
| 405 |
{0x2621, 0x266f}, {0x2701, 0x2704}, {0x2706, 0x2709}, {0x270c, 0x271f}, |
| 406 |
{0x2721, 0x2727}, {0x2729, 0x274b}, {0x274f, 0x2752}, {0x2758, 0x275e}, |
| 407 |
{0x2761, 0x2767}, {0x2776, 0x2794}, {0x2798, 0x27af}, {0x27b1, 0x27be}, |
| 408 |
{0x3000, 0x301f}, {0x3021, 0x3037}, {0x3041, 0x3094}, {0x3099, 0x309e}, |
| 409 |
{0x30a1, 0x30fe}, {0x3105, 0x311f}, {0x3121, 0x312c}, {0x3131, 0x318e}, |
| 410 |
{0x3190, 0x319f}, {0x3200, 0x321c}, {0x3221, 0x3243}, {0x3260, 0x327b}, |
| 411 |
{0x327f, 0x32b0}, {0x32c0, 0x32cb}, {0x32d0, 0x32fe}, {0x3300, 0x331f}, |
| 412 |
{0x3321, 0x3376}, {0x337b, 0x33dd}, {0x33e0, 0x33fe}, {0x4e00, 0x4e1f}, |
| 413 |
{0x4e21, 0x4f1f}, {0x4f21, 0x501f}, {0x5021, 0x511f}, {0x5121, 0x521f}, |
| 414 |
{0x5221, 0x531f}, {0x5321, 0x541f}, {0x5421, 0x551f}, {0x5521, 0x561f}, |
| 415 |
{0x5621, 0x571f}, {0x5721, 0x581f}, {0x5821, 0x591f}, {0x5921, 0x5a1f}, |
| 416 |
{0x5a21, 0x5b1f}, {0x5b21, 0x5c1f}, {0x5c21, 0x5d1f}, {0x5d21, 0x5e1f}, |
| 417 |
{0x5e21, 0x5f1f}, {0x5f21, 0x601f}, {0x6021, 0x611f}, {0x6121, 0x621f}, |
| 418 |
{0x6221, 0x631f}, {0x6321, 0x641f}, {0x6421, 0x651f}, {0x6521, 0x661f}, |
| 419 |
{0x6621, 0x671f}, {0x6721, 0x681f}, {0x6821, 0x691f}, {0x6921, 0x6a1f}, |
| 420 |
{0x6a21, 0x6b1f}, {0x6b21, 0x6c1f}, {0x6c21, 0x6d1f}, {0x6d21, 0x6e1f}, |
| 421 |
{0x6e21, 0x6f1f}, {0x6f21, 0x701f}, {0x7021, 0x711f}, {0x7121, 0x721f}, |
| 422 |
{0x7221, 0x731f}, {0x7321, 0x741f}, {0x7421, 0x751f}, {0x7521, 0x761f}, |
| 423 |
{0x7621, 0x771f}, {0x7721, 0x781f}, {0x7821, 0x791f}, {0x7921, 0x7a1f}, |
| 424 |
{0x7a21, 0x7b1f}, {0x7b21, 0x7c1f}, {0x7c21, 0x7d1f}, {0x7d21, 0x7e1f}, |
| 425 |
{0x7e21, 0x7f1f}, {0x7f21, 0x801f}, {0x8021, 0x811f}, {0x8121, 0x821f}, |
| 426 |
{0x8221, 0x831f}, {0x8321, 0x841f}, {0x8421, 0x851f}, {0x8521, 0x861f}, |
| 427 |
{0x8621, 0x871f}, {0x8721, 0x881f}, {0x8821, 0x891f}, {0x8921, 0x8a1f}, |
| 428 |
{0x8a21, 0x8b1f}, {0x8b21, 0x8c1f}, {0x8c21, 0x8d1f}, {0x8d21, 0x8e1f}, |
| 429 |
{0x8e21, 0x8f1f}, {0x8f21, 0x901f}, {0x9021, 0x911f}, {0x9121, 0x921f}, |
| 430 |
{0x9221, 0x931f}, {0x9321, 0x941f}, {0x9421, 0x951f}, {0x9521, 0x961f}, |
| 431 |
{0x9621, 0x971f}, {0x9721, 0x981f}, {0x9821, 0x991f}, {0x9921, 0x9a1f}, |
| 432 |
{0x9a21, 0x9b1f}, {0x9b21, 0x9c1f}, {0x9c21, 0x9d1f}, {0x9d21, 0x9e1f}, |
| 433 |
{0x9e21, 0x9f1f}, {0x9f21, 0x9fa5}, {0xac00, 0xac1f}, {0xac21, 0xad1f}, |
| 434 |
{0xad21, 0xae1f}, {0xae21, 0xaf1f}, {0xaf21, 0xb01f}, {0xb021, 0xb11f}, |
| 435 |
{0xb121, 0xb21f}, {0xb221, 0xb31f}, {0xb321, 0xb41f}, {0xb421, 0xb51f}, |
| 436 |
{0xb521, 0xb61f}, {0xb621, 0xb71f}, {0xb721, 0xb81f}, {0xb821, 0xb91f}, |
| 437 |
{0xb921, 0xba1f}, {0xba21, 0xbb1f}, {0xbb21, 0xbc1f}, {0xbc21, 0xbd1f}, |
| 438 |
{0xbd21, 0xbe1f}, {0xbe21, 0xbf1f}, {0xbf21, 0xc01f}, {0xc021, 0xc11f}, |
| 439 |
{0xc121, 0xc21f}, {0xc221, 0xc31f}, {0xc321, 0xc41f}, {0xc421, 0xc51f}, |
| 440 |
{0xc521, 0xc61f}, {0xc621, 0xc71f}, {0xc721, 0xc81f}, {0xc821, 0xc91f}, |
| 441 |
{0xc921, 0xca1f}, {0xca21, 0xcb1f}, {0xcb21, 0xcc1f}, {0xcc21, 0xcd1f}, |
| 442 |
{0xcd21, 0xce1f}, {0xce21, 0xcf1f}, {0xcf21, 0xd01f}, {0xd021, 0xd11f}, |
| 443 |
{0xd121, 0xd21f}, {0xd221, 0xd31f}, {0xd321, 0xd41f}, {0xd421, 0xd51f}, |
| 444 |
{0xd521, 0xd61f}, {0xd621, 0xd71f}, {0xd721, 0xd7a3}, {0xf900, 0xf91f}, |
| 445 |
{0xf921, 0xfa1f}, {0xfa21, 0xfa2d}, {0xfb00, 0xfb06}, {0xfb13, 0xfb17}, |
| 446 |
{0xfb21, 0xfb36}, {0xfb38, 0xfb3c}, {0xfb46, 0xfbb1}, {0xfbd3, 0xfc1f}, |
| 447 |
{0xfc21, 0xfd1f}, {0xfd21, 0xfd3f}, {0xfd50, 0xfd8f}, {0xfd92, 0xfdc7}, |
| 448 |
{0xfdf0, 0xfdfb}, {0xfe21, 0xfe23}, {0xfe30, 0xfe44}, {0xfe49, 0xfe52}, |
| 449 |
{0xfe54, 0xfe66}, {0xfe68, 0xfe6b}, {0xfe70, 0xfe72}, {0xfe76, 0xfefc}, |
| 450 |
{0xff01, 0xff1f}, {0xff21, 0xff5e}, {0xff61, 0xffbe}, {0xffc2, 0xffc7}, |
| 451 |
{0xffca, 0xffcf}, {0xffd2, 0xffd7}, {0xffda, 0xffdc}, {0xffe0, 0xffe6}, |
| 452 |
{0xffe8, 0xffee} |
| 453 |
}; |
| 454 |
|
| 455 |
#define NUM_GRAPH_RANGE (sizeof(graphRangeTable)/sizeof(crange)) |
| 456 |
|
| 457 |
static chr graphCharTable[] = { |
| 458 |
0x0360, 0x0361, 0x0374, 0x0375, 0x037a, 0x037e, 0x038c, 0x03da, 0x03dc, |
| 459 |
0x03de, 0x03e0, 0x04c7, 0x04c8, 0x04cb, 0x04cc, 0x04f8, 0x04f9, 0x0589, |
| 460 |
0x060c, 0x061b, 0x061f, 0x098f, 0x0990, 0x09b2, 0x09bc, 0x09c7, 0x09c8, |
| 461 |
0x09d7, 0x09dc, 0x09dd, 0x0a02, 0x0a0f, 0x0a10, 0x0a32, 0x0a33, 0x0a35, |
| 462 |
0x0a36, 0x0a38, 0x0a39, 0x0a3c, 0x0a47, 0x0a48, 0x0a5e, 0x0a8d, 0x0ab2, |
| 463 |
0x0ab3, 0x0ad0, 0x0ae0, 0x0b0f, 0x0b10, 0x0b32, 0x0b33, 0x0b47, 0x0b48, |
| 464 |
0x0b56, 0x0b57, 0x0b5c, 0x0b5d, 0x0b82, 0x0b83, 0x0b99, 0x0b9a, 0x0b9c, |
| 465 |
0x0b9e, 0x0b9f, 0x0ba3, 0x0ba4, 0x0bd7, 0x0c55, 0x0c56, 0x0c60, 0x0c61, |
| 466 |
0x0c82, 0x0c83, 0x0cd5, 0x0cd6, 0x0cde, 0x0ce0, 0x0ce1, 0x0d02, 0x0d03, |
| 467 |
0x0d57, 0x0d60, 0x0d61, 0x0e81, 0x0e82, 0x0e84, 0x0e87, 0x0e88, 0x0e8a, |
| 468 |
0x0e8d, 0x0ea5, 0x0ea7, 0x0eaa, 0x0eab, 0x0ec6, 0x0edc, 0x0edd, 0x0f97, |
| 469 |
0x0fb9, 0x10fb, 0x1f59, 0x1f5b, 0x1f5d, 0x2070, 0x2300, 0x274d, 0x2756, |
| 470 |
0x303f, 0xfb1e, 0xfb1f, 0xfb3e, 0xfb40, 0xfb41, 0xfb43, 0xfb44, 0xfe74, |
| 471 |
0xfffc, 0xfffd |
| 472 |
}; |
| 473 |
|
| 474 |
#define NUM_GRAPH_CHAR (sizeof(graphCharTable)/sizeof(chr)) |
| 475 |
|
| 476 |
|
| 477 |
#define CH NOCELT |
| 478 |
|
| 479 |
/* |
| 480 |
- nmcces - how many distinct MCCEs are there? |
| 481 |
^ static int nmcces(struct vars *); |
| 482 |
*/ |
| 483 |
static int |
| 484 |
nmcces(v) |
| 485 |
struct vars *v; |
| 486 |
{ |
| 487 |
return 0; |
| 488 |
} |
| 489 |
|
| 490 |
/* |
| 491 |
- nleaders - how many chrs can be first chrs of MCCEs? |
| 492 |
^ static int nleaders(struct vars *); |
| 493 |
*/ |
| 494 |
static int |
| 495 |
nleaders(v) |
| 496 |
struct vars *v; |
| 497 |
{ |
| 498 |
return 0; |
| 499 |
} |
| 500 |
|
| 501 |
/* |
| 502 |
- allmcces - return a cvec with all the MCCEs of the locale |
| 503 |
^ static struct cvec *allmcces(struct vars *, struct cvec *); |
| 504 |
*/ |
| 505 |
static struct cvec * |
| 506 |
allmcces(v, cv) |
| 507 |
struct vars *v; |
| 508 |
struct cvec *cv; /* this is supposed to have enough room */ |
| 509 |
{ |
| 510 |
return clearcvec(cv); |
| 511 |
} |
| 512 |
|
| 513 |
/* |
| 514 |
- element - map collating-element name to celt |
| 515 |
^ static celt element(struct vars *, chr *, chr *); |
| 516 |
*/ |
| 517 |
static celt |
| 518 |
element(v, startp, endp) |
| 519 |
struct vars *v; |
| 520 |
chr *startp; /* points to start of name */ |
| 521 |
chr *endp; /* points just past end of name */ |
| 522 |
{ |
| 523 |
struct cname *cn; |
| 524 |
size_t len; |
| 525 |
Tcl_DString ds; |
| 526 |
char *np; |
| 527 |
|
| 528 |
/* generic: one-chr names stand for themselves */ |
| 529 |
assert(startp < endp); |
| 530 |
len = endp - startp; |
| 531 |
if (len == 1) |
| 532 |
return *startp; |
| 533 |
|
| 534 |
NOTE(REG_ULOCALE); |
| 535 |
|
| 536 |
/* search table */ |
| 537 |
Tcl_DStringInit(&ds); |
| 538 |
np = Tcl_UniCharToUtfDString(startp, (int)len, &ds); |
| 539 |
for (cn = cnames; cn->name != NULL; cn++) |
| 540 |
if (strlen(cn->name) == len && strncmp(cn->name, np, len) == 0) |
| 541 |
break; /* NOTE BREAK OUT */ |
| 542 |
Tcl_DStringFree(&ds); |
| 543 |
if (cn->name != NULL) |
| 544 |
return CHR(cn->code); |
| 545 |
|
| 546 |
/* couldn't find it */ |
| 547 |
ERR(REG_ECOLLATE); |
| 548 |
return 0; |
| 549 |
} |
| 550 |
|
| 551 |
/* |
| 552 |
- range - supply cvec for a range, including legality check |
| 553 |
^ static struct cvec *range(struct vars *, celt, celt, int); |
| 554 |
*/ |
| 555 |
static struct cvec * |
| 556 |
range(v, a, b, cases) |
| 557 |
struct vars *v; |
| 558 |
celt a; |
| 559 |
celt b; /* might equal a */ |
| 560 |
int cases; /* case-independent? */ |
| 561 |
{ |
| 562 |
int nchrs; |
| 563 |
struct cvec *cv; |
| 564 |
celt c, lc, uc, tc; |
| 565 |
|
| 566 |
if (a != b && !before(a, b)) { |
| 567 |
ERR(REG_ERANGE); |
| 568 |
return NULL; |
| 569 |
} |
| 570 |
|
| 571 |
if (!cases) { /* easy version */ |
| 572 |
cv = getcvec(v, 0, 1, 0); |
| 573 |
NOERRN(); |
| 574 |
addrange(cv, a, b); |
| 575 |
return cv; |
| 576 |
} |
| 577 |
|
| 578 |
/* |
| 579 |
* When case-independent, it's hard to decide when cvec ranges are |
| 580 |
* usable, so for now at least, we won't try. We allocate enough |
| 581 |
* space for two case variants plus a little extra for the two |
| 582 |
* title case variants. |
| 583 |
*/ |
| 584 |
|
| 585 |
nchrs = (b - a + 1)*2 + 4; |
| 586 |
|
| 587 |
cv = getcvec(v, nchrs, 0, 0); |
| 588 |
NOERRN(); |
| 589 |
|
| 590 |
for (c = a; c <= b; c++) { |
| 591 |
addchr(cv, c); |
| 592 |
lc = Tcl_UniCharToLower((chr)c); |
| 593 |
uc = Tcl_UniCharToUpper((chr)c); |
| 594 |
tc = Tcl_UniCharToTitle((chr)c); |
| 595 |
if (c != lc) { |
| 596 |
addchr(cv, lc); |
| 597 |
} |
| 598 |
if (c != uc) { |
| 599 |
addchr(cv, uc); |
| 600 |
} |
| 601 |
if (c != tc && tc != uc) { |
| 602 |
addchr(cv, tc); |
| 603 |
} |
| 604 |
} |
| 605 |
|
| 606 |
return cv; |
| 607 |
} |
| 608 |
|
| 609 |
/* |
| 610 |
- before - is celt x before celt y, for purposes of range legality? |
| 611 |
^ static int before(celt, celt); |
| 612 |
*/ |
| 613 |
static int /* predicate */ |
| 614 |
before(x, y) |
| 615 |
celt x; |
| 616 |
celt y; |
| 617 |
{ |
| 618 |
/* trivial because no MCCEs */ |
| 619 |
if (x < y) |
| 620 |
return 1; |
| 621 |
return 0; |
| 622 |
} |
| 623 |
|
| 624 |
/* |
| 625 |
- eclass - supply cvec for an equivalence class |
| 626 |
* Must include case counterparts on request. |
| 627 |
^ static struct cvec *eclass(struct vars *, celt, int); |
| 628 |
*/ |
| 629 |
static struct cvec * |
| 630 |
eclass(v, c, cases) |
| 631 |
struct vars *v; |
| 632 |
celt c; |
| 633 |
int cases; /* all cases? */ |
| 634 |
{ |
| 635 |
struct cvec *cv; |
| 636 |
|
| 637 |
/* crude fake equivalence class for testing */ |
| 638 |
if ((v->cflags®_FAKE) && c == 'x') { |
| 639 |
cv = getcvec(v, 4, 0, 0); |
| 640 |
addchr(cv, (chr)'x'); |
| 641 |
addchr(cv, (chr)'y'); |
| 642 |
if (cases) { |
| 643 |
addchr(cv, (chr)'X'); |
| 644 |
addchr(cv, (chr)'Y'); |
| 645 |
} |
| 646 |
return cv; |
| 647 |
} |
| 648 |
|
| 649 |
/* otherwise, none */ |
| 650 |
if (cases) |
| 651 |
return allcases(v, c); |
| 652 |
cv = getcvec(v, 1, 0, 0); |
| 653 |
assert(cv != NULL); |
| 654 |
addchr(cv, (chr)c); |
| 655 |
return cv; |
| 656 |
} |
| 657 |
|
| 658 |
/* |
| 659 |
- cclass - supply cvec for a character class |
| 660 |
* Must include case counterparts on request. |
| 661 |
^ static struct cvec *cclass(struct vars *, chr *, chr *, int); |
| 662 |
*/ |
| 663 |
static struct cvec * |
| 664 |
cclass(v, startp, endp, cases) |
| 665 |
struct vars *v; |
| 666 |
chr *startp; /* where the name starts */ |
| 667 |
chr *endp; /* just past the end of the name */ |
| 668 |
int cases; /* case-independent? */ |
| 669 |
{ |
| 670 |
size_t len; |
| 671 |
struct cvec *cv = NULL; |
| 672 |
Tcl_DString ds; |
| 673 |
char *np, **namePtr; |
| 674 |
int i, index; |
| 675 |
|
| 676 |
/* |
| 677 |
* The following arrays define the valid character class names. |
| 678 |
*/ |
| 679 |
|
| 680 |
static char *classNames[] = { |
| 681 |
"alnum", "alpha", "ascii", "blank", "cntrl", "digit", "graph", |
| 682 |
"lower", "print", "punct", "space", "upper", "xdigit", NULL |
| 683 |
}; |
| 684 |
|
| 685 |
enum classes { |
| 686 |
CC_ALNUM, CC_ALPHA, CC_ASCII, CC_BLANK, CC_CNTRL, CC_DIGIT, CC_GRAPH, |
| 687 |
CC_LOWER, CC_PRINT, CC_PUNCT, CC_SPACE, CC_UPPER, CC_XDIGIT |
| 688 |
}; |
| 689 |
|
| 690 |
|
| 691 |
/* |
| 692 |
* Extract the class name |
| 693 |
*/ |
| 694 |
|
| 695 |
len = endp - startp; |
| 696 |
Tcl_DStringInit(&ds); |
| 697 |
np = Tcl_UniCharToUtfDString(startp, (int)len, &ds); |
| 698 |
|
| 699 |
/* |
| 700 |
* Remap lower and upper to alpha if the match is case insensitive. |
| 701 |
*/ |
| 702 |
|
| 703 |
if (cases && len == 5 && (strncmp("lower", np, 5) == 0 |
| 704 |
|| strncmp("upper", np, 5) == 0)) { |
| 705 |
np = "alpha"; |
| 706 |
} |
| 707 |
|
| 708 |
/* |
| 709 |
* Map the name to the corresponding enumerated value. |
| 710 |
*/ |
| 711 |
|
| 712 |
index = -1; |
| 713 |
for (namePtr = classNames, i = 0; *namePtr != NULL; namePtr++, i++) { |
| 714 |
if ((strlen(*namePtr) == len) && (strncmp(*namePtr, np, len) == 0)) { |
| 715 |
index = i; |
| 716 |
break; |
| 717 |
} |
| 718 |
} |
| 719 |
Tcl_DStringInit(&ds); |
| 720 |
if (index == -1) { |
| 721 |
ERR(REG_ECTYPE); |
| 722 |
return NULL; |
| 723 |
} |
| 724 |
|
| 725 |
/* |
| 726 |
* Now compute the character class contents. |
| 727 |
*/ |
| 728 |
|
| 729 |
switch((enum classes) index) { |
| 730 |
case CC_PRINT: |
| 731 |
case CC_ALNUM: |
| 732 |
cv = getcvec(v, NUM_ALPHA_CHAR, |
| 733 |
NUM_DIGIT_RANGE + NUM_ALPHA_RANGE, 0); |
| 734 |
if (cv) { |
| 735 |
for (i = 0; i < NUM_ALPHA_CHAR; i++) { |
| 736 |
addchr(cv, alphaCharTable[i]); |
| 737 |
} |
| 738 |
for (i = 0; i < NUM_ALPHA_RANGE; i++) { |
| 739 |
addrange(cv, alphaRangeTable[i].start, |
| 740 |
alphaRangeTable[i].end); |
| 741 |
} |
| 742 |
for (i = 0; i < NUM_DIGIT_RANGE; i++) { |
| 743 |
addrange(cv, digitRangeTable[i].start, |
| 744 |
digitRangeTable[i].end); |
| 745 |
} |
| 746 |
} |
| 747 |
break; |
| 748 |
case CC_ALPHA: |
| 749 |
cv = getcvec(v, NUM_ALPHA_CHAR, NUM_ALPHA_RANGE, 0); |
| 750 |
if (cv) { |
| 751 |
for (i = 0; i < NUM_ALPHA_RANGE; i++) { |
| 752 |
addrange(cv, alphaRangeTable[i].start, |
| 753 |
alphaRangeTable[i].end); |
| 754 |
} |
| 755 |
for (i = 0; i < NUM_ALPHA_CHAR; i++) { |
| 756 |
addchr(cv, alphaCharTable[i]); |
| 757 |
} |
| 758 |
} |
| 759 |
break; |
| 760 |
case CC_ASCII: |
| 761 |
cv = getcvec(v, 0, 1, 0); |
| 762 |
if (cv) { |
| 763 |
addrange(cv, 0, 0x7f); |
| 764 |
} |
| 765 |
break; |
| 766 |
case CC_BLANK: |
| 767 |
cv = getcvec(v, 2, 0, 0); |
| 768 |
addchr(cv, '\t'); |
| 769 |
addchr(cv, ' '); |
| 770 |
break; |
| 771 |
case CC_CNTRL: |
| 772 |
cv = getcvec(v, 0, 2, 0); |
| 773 |
addrange(cv, 0x0, 0x1f); |
| 774 |
addrange(cv, 0x7f, 0x9f); |
| 775 |
break; |
| 776 |
case CC_DIGIT: |
| 777 |
cv = getcvec(v, 0, NUM_DIGIT_RANGE, 0); |
| 778 |
if (cv) { |
| 779 |
for (i = 0; i < NUM_DIGIT_RANGE; i++) { |
| 780 |
addrange(cv, digitRangeTable[i].start, |
| 781 |
digitRangeTable[i].end); |
| 782 |
} |
| 783 |
} |
| 784 |
break; |
| 785 |
case CC_PUNCT: |
| 786 |
cv = getcvec(v, NUM_PUNCT_CHAR, NUM_PUNCT_RANGE, 0); |
| 787 |
if (cv) { |
| 788 |
for (i = 0; i < NUM_PUNCT_RANGE; i++) { |
| 789 |
addrange(cv, punctRangeTable[i].start, |
| 790 |
punctRangeTable[i].end); |
| 791 |
} |
| 792 |
for (i = 0; i < NUM_PUNCT_CHAR; i++) { |
| 793 |
addchr(cv, punctCharTable[i]); |
| 794 |
} |
| 795 |
} |
| 796 |
break; |
| 797 |
case CC_XDIGIT: |
| 798 |
cv = getcvec(v, 0, NUM_DIGIT_RANGE+2, 0); |
| 799 |
if (cv) { |
| 800 |
addrange(cv, '0', '9'); |
| 801 |
addrange(cv, 'a', 'f'); |
| 802 |
addrange(cv, 'A', 'F'); |
| 803 |
} |
| 804 |
break; |
| 805 |
case CC_SPACE: |
| 806 |
cv = getcvec(v, NUM_SPACE_CHAR, NUM_SPACE_RANGE, 0); |
| 807 |
if (cv) { |
| 808 |
for (i = 0; i < NUM_SPACE_RANGE; i++) { |
| 809 |
addrange(cv, spaceRangeTable[i].start, |
| 810 |
spaceRangeTable[i].end); |
| 811 |
} |
| 812 |
for (i = 0; i < NUM_SPACE_CHAR; i++) { |
| 813 |
addchr(cv, spaceCharTable[i]); |
| 814 |
} |
| 815 |
} |
| 816 |
break; |
| 817 |
case CC_LOWER: |
| 818 |
cv = getcvec(v, NUM_LOWER_CHAR, NUM_LOWER_RANGE, 0); |
| 819 |
if (cv) { |
| 820 |
for (i = 0; i < NUM_LOWER_RANGE; i++) { |
| 821 |
addrange(cv, lowerRangeTable[i].start, |
| 822 |
lowerRangeTable[i].end); |
| 823 |
} |
| 824 |
for (i = 0; i < NUM_LOWER_CHAR; i++) { |
| 825 |
addchr(cv, lowerCharTable[i]); |
| 826 |
} |
| 827 |
} |
| 828 |
break; |
| 829 |
case CC_UPPER: |
| 830 |
cv = getcvec(v, NUM_UPPER_CHAR, NUM_UPPER_RANGE, 0); |
| 831 |
if (cv) { |
| 832 |
for (i = 0; i < NUM_UPPER_RANGE; i++) { |
| 833 |
addrange(cv, upperRangeTable[i].start, |
| 834 |
upperRangeTable[i].end); |
| 835 |
} |
| 836 |
for (i = 0; i < NUM_UPPER_CHAR; i++) { |
| 837 |
addchr(cv, upperCharTable[i]); |
| 838 |
} |
| 839 |
} |
| 840 |
break; |
| 841 |
case CC_GRAPH: |
| 842 |
cv = getcvec(v, NUM_GRAPH_CHAR, NUM_GRAPH_RANGE, 0); |
| 843 |
if (cv) { |
| 844 |
for (i = 0; i < NUM_GRAPH_RANGE; i++) { |
| 845 |
addrange(cv, graphRangeTable[i].start, |
| 846 |
graphRangeTable[i].end); |
| 847 |
} |
| 848 |
for (i = 0; i < NUM_GRAPH_CHAR; i++) { |
| 849 |
addchr(cv, graphCharTable[i]); |
| 850 |
} |
| 851 |
} |
| 852 |
break; |
| 853 |
} |
| 854 |
if (cv == NULL) { |
| 855 |
ERR(REG_ESPACE); |
| 856 |
} |
| 857 |
return cv; |
| 858 |
} |
| 859 |
|
| 860 |
/* |
| 861 |
- allcases - supply cvec for all case counterparts of a chr (including itself) |
| 862 |
* This is a shortcut, preferably an efficient one, for simple characters; |
| 863 |
* messy cases are done via range(). |
| 864 |
^ static struct cvec *allcases(struct vars *, pchr); |
| 865 |
*/ |
| 866 |
static struct cvec * |
| 867 |
allcases(v, pc) |
| 868 |
struct vars *v; |
| 869 |
pchr pc; |
| 870 |
{ |
| 871 |
struct cvec *cv; |
| 872 |
chr c = (chr)pc; |
| 873 |
chr lc, uc, tc; |
| 874 |
|
| 875 |
lc = Tcl_UniCharToLower((chr)c); |
| 876 |
uc = Tcl_UniCharToUpper((chr)c); |
| 877 |
tc = Tcl_UniCharToTitle((chr)c); |
| 878 |
|
| 879 |
if (tc != uc) { |
| 880 |
cv = getcvec(v, 3, 0, 0); |
| 881 |
addchr(cv, tc); |
| 882 |
} else { |
| 883 |
cv = getcvec(v, 2, 0, 0); |
| 884 |
} |
| 885 |
addchr(cv, lc); |
| 886 |
if (lc != uc) { |
| 887 |
addchr(cv, uc); |
| 888 |
} |
| 889 |
return cv; |
| 890 |
} |
| 891 |
|
| 892 |
/* |
| 893 |
- cmp - chr-substring compare |
| 894 |
* Backrefs need this. It should preferably be efficient. |
| 895 |
* Note that it does not need to report anything except equal/unequal. |
| 896 |
* Note also that the length is exact, and the comparison should not |
| 897 |
* stop at embedded NULs! |
| 898 |
^ static int cmp(CONST chr *, CONST chr *, size_t); |
| 899 |
*/ |
| 900 |
static int /* 0 for equal, nonzero for unequal */ |
| 901 |
cmp(x, y, len) |
| 902 |
CONST chr *x; |
| 903 |
CONST chr *y; |
| 904 |
size_t len; /* exact length of comparison */ |
| 905 |
{ |
| 906 |
return memcmp(VS(x), VS(y), len*sizeof(chr)); |
| 907 |
} |
| 908 |
|
| 909 |
/* |
| 910 |
- casecmp - case-independent chr-substring compare |
| 911 |
* REG_ICASE backrefs need this. It should preferably be efficient. |
| 912 |
* Note that it does not need to report anything except equal/unequal. |
| 913 |
* Note also that the length is exact, and the comparison should not |
| 914 |
* stop at embedded NULs! |
| 915 |
^ static int casecmp(CONST chr *, CONST chr *, size_t); |
| 916 |
*/ |
| 917 |
static int /* 0 for equal, nonzero for unequal */ |
| 918 |
casecmp(x, y, len) |
| 919 |
CONST chr *x; |
| 920 |
CONST chr *y; |
| 921 |
size_t len; /* exact length of comparison */ |
| 922 |
{ |
| 923 |
size_t i; |
| 924 |
CONST chr *xp; |
| 925 |
CONST chr *yp; |
| 926 |
|
| 927 |
for (xp = x, yp = y, i = len; i > 0; i--) |
| 928 |
if (Tcl_UniCharToLower(*xp++) != Tcl_UniCharToLower(*yp++)) |
| 929 |
return 1; |
| 930 |
return 0; |
| 931 |
} |
| 932 |
|
| 933 |
/* End of regc_locale.c */ |
Parent Directory
| 
Revision Log