| 1 | /*============================================================================= |
| 2 | Copyright (c) 2001-2011 Hartmut Kaiser |
| 3 | Copyright (c) 2001-2011 Joel de Guzman |
| 4 | |
| 5 | Distributed under the Boost Software License, Version 1.0. (See accompanying |
| 6 | file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) |
| 7 | =============================================================================*/ |
| 8 | #if !defined(BOOST_SPIRIT_ISO8859_1_APRIL_26_2006_1106PM) |
| 9 | #define BOOST_SPIRIT_ISO8859_1_APRIL_26_2006_1106PM |
| 10 | |
| 11 | #if defined(_MSC_VER) |
| 12 | #pragma once |
| 13 | #endif |
| 14 | |
| 15 | #include <climits> |
| 16 | #include <boost/assert.hpp> |
| 17 | #include <boost/cstdint.hpp> |
| 18 | |
| 19 | /////////////////////////////////////////////////////////////////////////////// |
| 20 | // constants used to classify the single characters |
| 21 | /////////////////////////////////////////////////////////////////////////////// |
| 22 | #define BOOST_CC_DIGIT 0x0001 |
| 23 | #define BOOST_CC_XDIGIT 0x0002 |
| 24 | #define BOOST_CC_ALPHA 0x0004 |
| 25 | #define BOOST_CC_CTRL 0x0008 |
| 26 | #define BOOST_CC_LOWER 0x0010 |
| 27 | #define BOOST_CC_UPPER 0x0020 |
| 28 | #define BOOST_CC_SPACE 0x0040 |
| 29 | #define BOOST_CC_PUNCT 0x0080 |
| 30 | |
| 31 | namespace boost { namespace spirit { namespace char_encoding |
| 32 | { |
| 33 | // The detection of isgraph(), isprint() and isblank() is done programmatically |
| 34 | // to keep the character type table small. Additionally, these functions are |
| 35 | // rather seldom used and the programmatic detection is very simple. |
| 36 | |
| 37 | /////////////////////////////////////////////////////////////////////////// |
| 38 | // ISO 8859-1 character classification table |
| 39 | // |
| 40 | // the comments intentionally contain non-ascii characters |
| 41 | // boostinspect:noascii |
| 42 | /////////////////////////////////////////////////////////////////////////// |
| 43 | const unsigned char iso8859_1_char_types[] = |
| 44 | { |
| 45 | /* NUL 0 0 */ BOOST_CC_CTRL, |
| 46 | /* SOH 1 1 */ BOOST_CC_CTRL, |
| 47 | /* STX 2 2 */ BOOST_CC_CTRL, |
| 48 | /* ETX 3 3 */ BOOST_CC_CTRL, |
| 49 | /* EOT 4 4 */ BOOST_CC_CTRL, |
| 50 | /* ENQ 5 5 */ BOOST_CC_CTRL, |
| 51 | /* ACK 6 6 */ BOOST_CC_CTRL, |
| 52 | /* BEL 7 7 */ BOOST_CC_CTRL, |
| 53 | /* BS 8 8 */ BOOST_CC_CTRL, |
| 54 | /* HT 9 9 */ BOOST_CC_CTRL|BOOST_CC_SPACE, |
| 55 | /* NL 10 a */ BOOST_CC_CTRL|BOOST_CC_SPACE, |
| 56 | /* VT 11 b */ BOOST_CC_CTRL|BOOST_CC_SPACE, |
| 57 | /* NP 12 c */ BOOST_CC_CTRL|BOOST_CC_SPACE, |
| 58 | /* CR 13 d */ BOOST_CC_CTRL|BOOST_CC_SPACE, |
| 59 | /* SO 14 e */ BOOST_CC_CTRL, |
| 60 | /* SI 15 f */ BOOST_CC_CTRL, |
| 61 | /* DLE 16 10 */ BOOST_CC_CTRL, |
| 62 | /* DC1 17 11 */ BOOST_CC_CTRL, |
| 63 | /* DC2 18 12 */ BOOST_CC_CTRL, |
| 64 | /* DC3 19 13 */ BOOST_CC_CTRL, |
| 65 | /* DC4 20 14 */ BOOST_CC_CTRL, |
| 66 | /* NAK 21 15 */ BOOST_CC_CTRL, |
| 67 | /* SYN 22 16 */ BOOST_CC_CTRL, |
| 68 | /* ETB 23 17 */ BOOST_CC_CTRL, |
| 69 | /* CAN 24 18 */ BOOST_CC_CTRL, |
| 70 | /* EM 25 19 */ BOOST_CC_CTRL, |
| 71 | /* SUB 26 1a */ BOOST_CC_CTRL, |
| 72 | /* ESC 27 1b */ BOOST_CC_CTRL, |
| 73 | /* FS 28 1c */ BOOST_CC_CTRL, |
| 74 | /* GS 29 1d */ BOOST_CC_CTRL, |
| 75 | /* RS 30 1e */ BOOST_CC_CTRL, |
| 76 | /* US 31 1f */ BOOST_CC_CTRL, |
| 77 | /* SP 32 20 */ BOOST_CC_SPACE, |
| 78 | /* ! 33 21 */ BOOST_CC_PUNCT, |
| 79 | /* " 34 22 */ BOOST_CC_PUNCT, |
| 80 | /* # 35 23 */ BOOST_CC_PUNCT, |
| 81 | /* $ 36 24 */ BOOST_CC_PUNCT, |
| 82 | /* % 37 25 */ BOOST_CC_PUNCT, |
| 83 | /* & 38 26 */ BOOST_CC_PUNCT, |
| 84 | /* ' 39 27 */ BOOST_CC_PUNCT, |
| 85 | /* ( 40 28 */ BOOST_CC_PUNCT, |
| 86 | /* ) 41 29 */ BOOST_CC_PUNCT, |
| 87 | /* * 42 2a */ BOOST_CC_PUNCT, |
| 88 | /* + 43 2b */ BOOST_CC_PUNCT, |
| 89 | /* , 44 2c */ BOOST_CC_PUNCT, |
| 90 | /* - 45 2d */ BOOST_CC_PUNCT, |
| 91 | /* . 46 2e */ BOOST_CC_PUNCT, |
| 92 | /* / 47 2f */ BOOST_CC_PUNCT, |
| 93 | /* 0 48 30 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 94 | /* 1 49 31 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 95 | /* 2 50 32 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 96 | /* 3 51 33 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 97 | /* 4 52 34 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 98 | /* 5 53 35 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 99 | /* 6 54 36 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 100 | /* 7 55 37 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 101 | /* 8 56 38 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 102 | /* 9 57 39 */ BOOST_CC_DIGIT|BOOST_CC_XDIGIT, |
| 103 | /* : 58 3a */ BOOST_CC_PUNCT, |
| 104 | /* ; 59 3b */ BOOST_CC_PUNCT, |
| 105 | /* < 60 3c */ BOOST_CC_PUNCT, |
| 106 | /* = 61 3d */ BOOST_CC_PUNCT, |
| 107 | /* > 62 3e */ BOOST_CC_PUNCT, |
| 108 | /* ? 63 3f */ BOOST_CC_PUNCT, |
| 109 | /* @ 64 40 */ BOOST_CC_PUNCT, |
| 110 | /* A 65 41 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_UPPER, |
| 111 | /* B 66 42 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_UPPER, |
| 112 | /* C 67 43 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_UPPER, |
| 113 | /* D 68 44 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_UPPER, |
| 114 | /* E 69 45 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_UPPER, |
| 115 | /* F 70 46 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_UPPER, |
| 116 | /* G 71 47 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 117 | /* H 72 48 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 118 | /* I 73 49 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 119 | /* J 74 4a */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 120 | /* K 75 4b */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 121 | /* L 76 4c */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 122 | /* M 77 4d */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 123 | /* N 78 4e */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 124 | /* O 79 4f */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 125 | /* P 80 50 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 126 | /* Q 81 51 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 127 | /* R 82 52 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 128 | /* S 83 53 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 129 | /* T 84 54 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 130 | /* U 85 55 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 131 | /* V 86 56 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 132 | /* W 87 57 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 133 | /* X 88 58 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 134 | /* Y 89 59 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 135 | /* Z 90 5a */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 136 | /* [ 91 5b */ BOOST_CC_PUNCT, |
| 137 | /* \ 92 5c */ BOOST_CC_PUNCT, |
| 138 | /* ] 93 5d */ BOOST_CC_PUNCT, |
| 139 | /* ^ 94 5e */ BOOST_CC_PUNCT, |
| 140 | /* _ 95 5f */ BOOST_CC_PUNCT, |
| 141 | /* ` 96 60 */ BOOST_CC_PUNCT, |
| 142 | /* a 97 61 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_LOWER, |
| 143 | /* b 98 62 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_LOWER, |
| 144 | /* c 99 63 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_LOWER, |
| 145 | /* d 100 64 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_LOWER, |
| 146 | /* e 101 65 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_LOWER, |
| 147 | /* f 102 66 */ BOOST_CC_ALPHA|BOOST_CC_XDIGIT|BOOST_CC_LOWER, |
| 148 | /* g 103 67 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 149 | /* h 104 68 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 150 | /* i 105 69 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 151 | /* j 106 6a */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 152 | /* k 107 6b */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 153 | /* l 108 6c */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 154 | /* m 109 6d */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 155 | /* n 110 6e */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 156 | /* o 111 6f */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 157 | /* p 112 70 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 158 | /* q 113 71 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 159 | /* r 114 72 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 160 | /* s 115 73 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 161 | /* t 116 74 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 162 | /* u 117 75 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 163 | /* v 118 76 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 164 | /* w 119 77 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 165 | /* x 120 78 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 166 | /* y 121 79 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 167 | /* z 122 7a */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 168 | /* { 123 7b */ BOOST_CC_PUNCT, |
| 169 | /* | 124 7c */ BOOST_CC_PUNCT, |
| 170 | /* } 125 7d */ BOOST_CC_PUNCT, |
| 171 | /* ~ 126 7e */ BOOST_CC_PUNCT, |
| 172 | /* DEL 127 7f */ BOOST_CC_CTRL, |
| 173 | /* -- 128 80 */ BOOST_CC_CTRL, |
| 174 | /* -- 129 81 */ BOOST_CC_CTRL, |
| 175 | /* -- 130 82 */ BOOST_CC_CTRL, |
| 176 | /* -- 131 83 */ BOOST_CC_CTRL, |
| 177 | /* -- 132 84 */ BOOST_CC_CTRL, |
| 178 | /* -- 133 85 */ BOOST_CC_CTRL, |
| 179 | /* -- 134 86 */ BOOST_CC_CTRL, |
| 180 | /* -- 135 87 */ BOOST_CC_CTRL, |
| 181 | /* -- 136 88 */ BOOST_CC_CTRL, |
| 182 | /* -- 137 89 */ BOOST_CC_CTRL, |
| 183 | /* -- 138 8a */ BOOST_CC_CTRL, |
| 184 | /* -- 139 8b */ BOOST_CC_CTRL, |
| 185 | /* -- 140 8c */ BOOST_CC_CTRL, |
| 186 | /* -- 141 8d */ BOOST_CC_CTRL, |
| 187 | /* -- 142 8e */ BOOST_CC_CTRL, |
| 188 | /* -- 143 8f */ BOOST_CC_CTRL, |
| 189 | /* -- 144 90 */ BOOST_CC_CTRL, |
| 190 | /* -- 145 91 */ BOOST_CC_CTRL, |
| 191 | /* -- 146 92 */ BOOST_CC_CTRL, |
| 192 | /* -- 147 93 */ BOOST_CC_CTRL, |
| 193 | /* -- 148 94 */ BOOST_CC_CTRL, |
| 194 | /* -- 149 95 */ BOOST_CC_CTRL, |
| 195 | /* -- 150 96 */ BOOST_CC_CTRL, |
| 196 | /* -- 151 97 */ BOOST_CC_CTRL, |
| 197 | /* -- 152 98 */ BOOST_CC_CTRL, |
| 198 | /* -- 153 99 */ BOOST_CC_CTRL, |
| 199 | /* -- 154 9a */ BOOST_CC_CTRL, |
| 200 | /* -- 155 9b */ BOOST_CC_CTRL, |
| 201 | /* -- 156 9c */ BOOST_CC_CTRL, |
| 202 | /* -- 157 9d */ BOOST_CC_CTRL, |
| 203 | /* -- 158 9e */ BOOST_CC_CTRL, |
| 204 | /* -- 159 9f */ BOOST_CC_CTRL, |
| 205 | /* 160 a0 */ BOOST_CC_SPACE, |
| 206 | /* � 161 a1 */ BOOST_CC_PUNCT, |
| 207 | /* � 162 a2 */ BOOST_CC_PUNCT, |
| 208 | /* � 163 a3 */ BOOST_CC_PUNCT, |
| 209 | /* � 164 a4 */ BOOST_CC_PUNCT, |
| 210 | /* � 165 a5 */ BOOST_CC_PUNCT, |
| 211 | /* � 166 a6 */ BOOST_CC_PUNCT, |
| 212 | /* � 167 a7 */ BOOST_CC_PUNCT, |
| 213 | /* � 168 a8 */ BOOST_CC_PUNCT, |
| 214 | /* � 169 a9 */ BOOST_CC_PUNCT, |
| 215 | /* � 170 aa */ BOOST_CC_PUNCT, |
| 216 | /* � 171 ab */ BOOST_CC_PUNCT, |
| 217 | /* � 172 ac */ BOOST_CC_PUNCT, |
| 218 | /* � 173 ad */ BOOST_CC_PUNCT, |
| 219 | /* � 174 ae */ BOOST_CC_PUNCT, |
| 220 | /* � 175 af */ BOOST_CC_PUNCT, |
| 221 | /* � 176 b0 */ BOOST_CC_PUNCT, |
| 222 | /* � 177 b1 */ BOOST_CC_PUNCT, |
| 223 | /* � 178 b2 */ BOOST_CC_DIGIT|BOOST_CC_PUNCT, |
| 224 | /* � 179 b3 */ BOOST_CC_DIGIT|BOOST_CC_PUNCT, |
| 225 | /* � 180 b4 */ BOOST_CC_PUNCT, |
| 226 | /* � 181 b5 */ BOOST_CC_PUNCT, |
| 227 | /* � 182 b6 */ BOOST_CC_PUNCT, |
| 228 | /* � 183 b7 */ BOOST_CC_PUNCT, |
| 229 | /* � 184 b8 */ BOOST_CC_PUNCT, |
| 230 | /* � 185 b9 */ BOOST_CC_DIGIT|BOOST_CC_PUNCT, |
| 231 | /* � 186 ba */ BOOST_CC_PUNCT, |
| 232 | /* � 187 bb */ BOOST_CC_PUNCT, |
| 233 | /* � 188 bc */ BOOST_CC_PUNCT, |
| 234 | /* � 189 bd */ BOOST_CC_PUNCT, |
| 235 | /* � 190 be */ BOOST_CC_PUNCT, |
| 236 | /* � 191 bf */ BOOST_CC_PUNCT, |
| 237 | /* � 192 c0 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 238 | /* � 193 c1 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 239 | /* � 194 c2 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 240 | /* � 195 c3 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 241 | /* � 196 c4 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 242 | /* � 197 c5 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 243 | /* � 198 c6 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 244 | /* � 199 c7 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 245 | /* � 200 c8 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 246 | /* � 201 c9 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 247 | /* � 202 ca */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 248 | /* � 203 cb */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 249 | /* � 204 cc */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 250 | /* � 205 cd */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 251 | /* � 206 ce */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 252 | /* � 207 cf */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 253 | /* � 208 d0 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 254 | /* � 209 d1 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 255 | /* � 210 d2 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 256 | /* � 211 d3 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 257 | /* � 212 d4 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 258 | /* � 213 d5 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 259 | /* � 214 d6 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 260 | /* � 215 d7 */ BOOST_CC_PUNCT, |
| 261 | /* � 216 d8 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 262 | /* � 217 d9 */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 263 | /* � 218 da */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 264 | /* � 219 db */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 265 | /* � 220 dc */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 266 | /* � 221 dd */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 267 | /* � 222 de */ BOOST_CC_ALPHA|BOOST_CC_UPPER, |
| 268 | /* � 223 df */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 269 | /* � 224 e0 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 270 | /* � 225 e1 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 271 | /* � 226 e2 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 272 | /* � 227 e3 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 273 | /* � 228 e4 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 274 | /* � 229 e5 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 275 | /* � 230 e6 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 276 | /* � 231 e7 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 277 | /* � 232 e8 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 278 | /* � 233 e9 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 279 | /* � 234 ea */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 280 | /* � 235 eb */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 281 | /* � 236 ec */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 282 | /* � 237 ed */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 283 | /* � 238 ee */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 284 | /* � 239 ef */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 285 | /* � 240 f0 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 286 | /* � 241 f1 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 287 | /* � 242 f2 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 288 | /* � 243 f3 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 289 | /* � 244 f4 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 290 | /* � 245 f5 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 291 | /* � 246 f6 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 292 | /* � 247 f7 */ BOOST_CC_PUNCT, |
| 293 | /* � 248 f8 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 294 | /* � 249 f9 */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 295 | /* � 250 fa */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 296 | /* � 251 fb */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 297 | /* � 252 fc */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 298 | /* � 253 fd */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 299 | /* � 254 fe */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 300 | /* � 255 ff */ BOOST_CC_ALPHA|BOOST_CC_LOWER, |
| 301 | }; |
| 302 | |
| 303 | /////////////////////////////////////////////////////////////////////////// |
| 304 | // ISO 8859-1 character conversion table |
| 305 | /////////////////////////////////////////////////////////////////////////// |
| 306 | const unsigned char iso8859_1_char_conversion[] = |
| 307 | { |
| 308 | /* NUL 0 0 */ '\0', |
| 309 | /* SOH 1 1 */ '\0', |
| 310 | /* STX 2 2 */ '\0', |
| 311 | /* ETX 3 3 */ '\0', |
| 312 | /* EOT 4 4 */ '\0', |
| 313 | /* ENQ 5 5 */ '\0', |
| 314 | /* ACK 6 6 */ '\0', |
| 315 | /* BEL 7 7 */ '\0', |
| 316 | /* BS 8 8 */ '\0', |
| 317 | /* HT 9 9 */ '\0', |
| 318 | /* NL 10 a */ '\0', |
| 319 | /* VT 11 b */ '\0', |
| 320 | /* NP 12 c */ '\0', |
| 321 | /* CR 13 d */ '\0', |
| 322 | /* SO 14 e */ '\0', |
| 323 | /* SI 15 f */ '\0', |
| 324 | /* DLE 16 10 */ '\0', |
| 325 | /* DC1 17 11 */ '\0', |
| 326 | /* DC2 18 12 */ '\0', |
| 327 | /* DC3 19 13 */ '\0', |
| 328 | /* DC4 20 14 */ '\0', |
| 329 | /* NAK 21 15 */ '\0', |
| 330 | /* SYN 22 16 */ '\0', |
| 331 | /* ETB 23 17 */ '\0', |
| 332 | /* CAN 24 18 */ '\0', |
| 333 | /* EM 25 19 */ '\0', |
| 334 | /* SUB 26 1a */ '\0', |
| 335 | /* ESC 27 1b */ '\0', |
| 336 | /* FS 28 1c */ '\0', |
| 337 | /* GS 29 1d */ '\0', |
| 338 | /* RS 30 1e */ '\0', |
| 339 | /* US 31 1f */ '\0', |
| 340 | /* SP 32 20 */ '\0', |
| 341 | /* ! 33 21 */ '\0', |
| 342 | /* " 34 22 */ '\0', |
| 343 | /* # 35 23 */ '\0', |
| 344 | /* $ 36 24 */ '\0', |
| 345 | /* % 37 25 */ '\0', |
| 346 | /* & 38 26 */ '\0', |
| 347 | /* ' 39 27 */ '\0', |
| 348 | /* ( 40 28 */ '\0', |
| 349 | /* ) 41 29 */ '\0', |
| 350 | /* * 42 2a */ '\0', |
| 351 | /* + 43 2b */ '\0', |
| 352 | /* , 44 2c */ '\0', |
| 353 | /* - 45 2d */ '\0', |
| 354 | /* . 46 2e */ '\0', |
| 355 | /* / 47 2f */ '\0', |
| 356 | /* 0 48 30 */ '\0', |
| 357 | /* 1 49 31 */ '\0', |
| 358 | /* 2 50 32 */ '\0', |
| 359 | /* 3 51 33 */ '\0', |
| 360 | /* 4 52 34 */ '\0', |
| 361 | /* 5 53 35 */ '\0', |
| 362 | /* 6 54 36 */ '\0', |
| 363 | /* 7 55 37 */ '\0', |
| 364 | /* 8 56 38 */ '\0', |
| 365 | /* 9 57 39 */ '\0', |
| 366 | /* : 58 3a */ '\0', |
| 367 | /* ; 59 3b */ '\0', |
| 368 | /* < 60 3c */ '\0', |
| 369 | /* = 61 3d */ '\0', |
| 370 | /* > 62 3e */ '\0', |
| 371 | /* ? 63 3f */ '\0', |
| 372 | /* @ 64 40 */ '\0', |
| 373 | /* A 65 41 */ 'a', |
| 374 | /* B 66 42 */ 'b', |
| 375 | /* C 67 43 */ 'c', |
| 376 | /* D 68 44 */ 'd', |
| 377 | /* E 69 45 */ 'e', |
| 378 | /* F 70 46 */ 'f', |
| 379 | /* G 71 47 */ 'g', |
| 380 | /* H 72 48 */ 'h', |
| 381 | /* I 73 49 */ 'i', |
| 382 | /* J 74 4a */ 'j', |
| 383 | /* K 75 4b */ 'k', |
| 384 | /* L 76 4c */ 'l', |
| 385 | /* M 77 4d */ 'm', |
| 386 | /* N 78 4e */ 'n', |
| 387 | /* O 79 4f */ 'o', |
| 388 | /* P 80 50 */ 'p', |
| 389 | /* Q 81 51 */ 'q', |
| 390 | /* R 82 52 */ 'r', |
| 391 | /* S 83 53 */ 's', |
| 392 | /* T 84 54 */ 't', |
| 393 | /* U 85 55 */ 'u', |
| 394 | /* V 86 56 */ 'v', |
| 395 | /* W 87 57 */ 'w', |
| 396 | /* X 88 58 */ 'x', |
| 397 | /* Y 89 59 */ 'y', |
| 398 | /* Z 90 5a */ 'z', |
| 399 | /* [ 91 5b */ '\0', |
| 400 | /* \ 92 5c */ '\0', |
| 401 | /* ] 93 5d */ '\0', |
| 402 | /* ^ 94 5e */ '\0', |
| 403 | /* _ 95 5f */ '\0', |
| 404 | /* ` 96 60 */ '\0', |
| 405 | /* a 97 61 */ 'A', |
| 406 | /* b 98 62 */ 'B', |
| 407 | /* c 99 63 */ 'C', |
| 408 | /* d 100 64 */ 'D', |
| 409 | /* e 101 65 */ 'E', |
| 410 | /* f 102 66 */ 'F', |
| 411 | /* g 103 67 */ 'G', |
| 412 | /* h 104 68 */ 'H', |
| 413 | /* i 105 69 */ 'I', |
| 414 | /* j 106 6a */ 'J', |
| 415 | /* k 107 6b */ 'K', |
| 416 | /* l 108 6c */ 'L', |
| 417 | /* m 109 6d */ 'M', |
| 418 | /* n 110 6e */ 'N', |
| 419 | /* o 111 6f */ 'O', |
| 420 | /* p 112 70 */ 'P', |
| 421 | /* q 113 71 */ 'Q', |
| 422 | /* r 114 72 */ 'R', |
| 423 | /* s 115 73 */ 'S', |
| 424 | /* t 116 74 */ 'T', |
| 425 | /* u 117 75 */ 'U', |
| 426 | /* v 118 76 */ 'V', |
| 427 | /* w 119 77 */ 'W', |
| 428 | /* x 120 78 */ 'X', |
| 429 | /* y 121 79 */ 'Y', |
| 430 | /* z 122 7a */ 'Z', |
| 431 | /* { 123 7b */ '\0', |
| 432 | /* | 124 7c */ '\0', |
| 433 | /* } 125 7d */ '\0', |
| 434 | /* ~ 126 7e */ '\0', |
| 435 | /* DEL 127 7f */ '\0', |
| 436 | /* -- 128 80 */ '\0', |
| 437 | /* -- 129 81 */ '\0', |
| 438 | /* -- 130 82 */ '\0', |
| 439 | /* -- 131 83 */ '\0', |
| 440 | /* -- 132 84 */ '\0', |
| 441 | /* -- 133 85 */ '\0', |
| 442 | /* -- 134 86 */ '\0', |
| 443 | /* -- 135 87 */ '\0', |
| 444 | /* -- 136 88 */ '\0', |
| 445 | /* -- 137 89 */ '\0', |
| 446 | /* -- 138 8a */ '\0', |
| 447 | /* -- 139 8b */ '\0', |
| 448 | /* -- 140 8c */ '\0', |
| 449 | /* -- 141 8d */ '\0', |
| 450 | /* -- 142 8e */ '\0', |
| 451 | /* -- 143 8f */ '\0', |
| 452 | /* -- 144 90 */ '\0', |
| 453 | /* -- 145 91 */ '\0', |
| 454 | /* -- 146 92 */ '\0', |
| 455 | /* -- 147 93 */ '\0', |
| 456 | /* -- 148 94 */ '\0', |
| 457 | /* -- 149 95 */ '\0', |
| 458 | /* -- 150 96 */ '\0', |
| 459 | /* -- 151 97 */ '\0', |
| 460 | /* -- 152 98 */ '\0', |
| 461 | /* -- 153 99 */ '\0', |
| 462 | /* -- 154 9a */ '\0', |
| 463 | /* -- 155 9b */ '\0', |
| 464 | /* -- 156 9c */ '\0', |
| 465 | /* -- 157 9d */ '\0', |
| 466 | /* -- 158 9e */ '\0', |
| 467 | /* -- 159 9f */ '\0', |
| 468 | /* 160 a0 */ '\0', |
| 469 | /* � 161 a1 */ '\0', |
| 470 | /* � 162 a2 */ '\0', |
| 471 | /* � 163 a3 */ '\0', |
| 472 | /* � 164 a4 */ '\0', |
| 473 | /* � 165 a5 */ '\0', |
| 474 | /* � 166 a6 */ '\0', |
| 475 | /* � 167 a7 */ '\0', |
| 476 | /* � 168 a8 */ '\0', |
| 477 | /* � 169 a9 */ '\0', |
| 478 | /* � 170 aa */ '\0', |
| 479 | /* � 171 ab */ '\0', |
| 480 | /* � 172 ac */ '\0', |
| 481 | /* � 173 ad */ '\0', |
| 482 | /* � 174 ae */ '\0', |
| 483 | /* � 175 af */ '\0', |
| 484 | /* � 176 b0 */ '\0', |
| 485 | /* � 177 b1 */ '\0', |
| 486 | /* � 178 b2 */ '\0', |
| 487 | /* � 179 b3 */ '\0', |
| 488 | /* � 180 b4 */ '\0', |
| 489 | /* � 181 b5 */ '\0', |
| 490 | /* � 182 b6 */ '\0', |
| 491 | /* � 183 b7 */ '\0', |
| 492 | /* � 184 b8 */ '\0', |
| 493 | /* � 185 b9 */ '\0', |
| 494 | /* � 186 ba */ '\0', |
| 495 | /* � 187 bb */ '\0', |
| 496 | /* � 188 bc */ '\0', |
| 497 | /* � 189 bd */ '\0', |
| 498 | /* � 190 be */ '\0', |
| 499 | /* � 191 bf */ '\0', |
| 500 | /* � 192 c0 */ 0xe0, |
| 501 | /* � 193 c1 */ 0xe1, |
| 502 | /* � 194 c2 */ 0xe2, |
| 503 | /* � 195 c3 */ 0xe3, |
| 504 | /* � 196 c4 */ 0xe4, |
| 505 | /* � 197 c5 */ 0xe5, |
| 506 | /* � 198 c6 */ 0xe6, |
| 507 | /* � 199 c7 */ 0xe7, |
| 508 | /* � 200 c8 */ 0xe8, |
| 509 | /* � 201 c9 */ 0xe9, |
| 510 | /* � 202 ca */ 0xea, |
| 511 | /* � 203 cb */ 0xeb, |
| 512 | /* � 204 cc */ 0xec, |
| 513 | /* � 205 cd */ 0xed, |
| 514 | /* � 206 ce */ 0xee, |
| 515 | /* � 207 cf */ 0xef, |
| 516 | /* � 208 d0 */ 0xf0, |
| 517 | /* � 209 d1 */ 0xf1, |
| 518 | /* � 210 d2 */ 0xf2, |
| 519 | /* � 211 d3 */ 0xf3, |
| 520 | /* � 212 d4 */ 0xf4, |
| 521 | /* � 213 d5 */ 0xf5, |
| 522 | /* � 214 d6 */ 0xf6, |
| 523 | /* � 215 d7 */ '\0', |
| 524 | /* � 216 d8 */ 0xf8, |
| 525 | /* � 217 d9 */ 0xf9, |
| 526 | /* � 218 da */ 0xfa, |
| 527 | /* � 219 db */ 0xfb, |
| 528 | /* � 220 dc */ 0xfc, |
| 529 | /* � 221 dd */ 0xfd, |
| 530 | /* � 222 de */ 0xfe, |
| 531 | /* � 223 df */ '\0', |
| 532 | /* � 224 e0 */ 0xc0, |
| 533 | /* � 225 e1 */ 0xc1, |
| 534 | /* � 226 e2 */ 0xc2, |
| 535 | /* � 227 e3 */ 0xc3, |
| 536 | /* � 228 e4 */ 0xc4, |
| 537 | /* � 229 e5 */ 0xc5, |
| 538 | /* � 230 e6 */ 0xc6, |
| 539 | /* � 231 e7 */ 0xc7, |
| 540 | /* � 232 e8 */ 0xc8, |
| 541 | /* � 233 e9 */ 0xc9, |
| 542 | /* � 234 ea */ 0xca, |
| 543 | /* � 235 eb */ 0xcb, |
| 544 | /* � 236 ec */ 0xcc, |
| 545 | /* � 237 ed */ 0xcd, |
| 546 | /* � 238 ee */ 0xce, |
| 547 | /* � 239 ef */ 0xcf, |
| 548 | /* � 240 f0 */ 0xd0, |
| 549 | /* � 241 f1 */ 0xd1, |
| 550 | /* � 242 f2 */ 0xd2, |
| 551 | /* � 243 f3 */ 0xd3, |
| 552 | /* � 244 f4 */ 0xd4, |
| 553 | /* � 245 f5 */ 0xd5, |
| 554 | /* � 246 f6 */ 0xd6, |
| 555 | /* � 247 f7 */ '\0', |
| 556 | /* � 248 f8 */ 0xd8, |
| 557 | /* � 249 f9 */ 0xd9, |
| 558 | /* � 250 fa */ 0xda, |
| 559 | /* � 251 fb */ 0xdb, |
| 560 | /* � 252 fc */ 0xdc, |
| 561 | /* � 253 fd */ 0xdd, |
| 562 | /* � 254 fe */ 0xde, |
| 563 | /* � 255 ff */ '\0', |
| 564 | }; |
| 565 | |
| 566 | /////////////////////////////////////////////////////////////////////////// |
| 567 | // Test characters for specified conditions (using iso8859-1) |
| 568 | /////////////////////////////////////////////////////////////////////////// |
| 569 | struct iso8859_1 |
| 570 | { |
| 571 | typedef unsigned char char_type; |
| 572 | typedef unsigned char classify_type; |
| 573 | |
| 574 | static bool |
| 575 | isascii_(int ch) |
| 576 | { |
| 577 | return 0 == (ch & ~0x7f); |
| 578 | } |
| 579 | |
| 580 | static bool |
| 581 | ischar(int ch) |
| 582 | { |
| 583 | // iso8859.1 uses all 8 bits |
| 584 | // we have to watch out for sign extensions |
| 585 | return (0 == (ch & ~0xff) || ~0 == (ch | 0xff)) != 0; |
| 586 | } |
| 587 | |
| 588 | // *** Note on assertions: The precondition is that the calls to |
| 589 | // these functions do not violate the required range of ch (type int) |
| 590 | // which is that strict_ischar(ch) should be true. It is the |
| 591 | // responsibility of the caller to make sure this precondition is not |
| 592 | // violated. |
| 593 | |
| 594 | static bool |
| 595 | strict_ischar(int ch) |
| 596 | { |
| 597 | return ch >= 0 && ch <= 255; |
| 598 | } |
| 599 | |
| 600 | static bool |
| 601 | isalnum(int ch) |
| 602 | { |
| 603 | BOOST_ASSERT(strict_ischar(ch)); |
| 604 | return (iso8859_1_char_types[ch] & BOOST_CC_ALPHA) |
| 605 | || (iso8859_1_char_types[ch] & BOOST_CC_DIGIT); |
| 606 | } |
| 607 | |
| 608 | static bool |
| 609 | isalpha(int ch) |
| 610 | { |
| 611 | BOOST_ASSERT(strict_ischar(ch)); |
| 612 | return (iso8859_1_char_types[ch] & BOOST_CC_ALPHA) != 0; |
| 613 | } |
| 614 | |
| 615 | static bool |
| 616 | isdigit(int ch) |
| 617 | { |
| 618 | BOOST_ASSERT(strict_ischar(ch)); |
| 619 | return (iso8859_1_char_types[ch] & BOOST_CC_DIGIT) != 0; |
| 620 | } |
| 621 | |
| 622 | static bool |
| 623 | isxdigit(int ch) |
| 624 | { |
| 625 | BOOST_ASSERT(strict_ischar(ch)); |
| 626 | return (iso8859_1_char_types[ch] & BOOST_CC_XDIGIT) != 0; |
| 627 | } |
| 628 | |
| 629 | static bool |
| 630 | iscntrl(int ch) |
| 631 | { |
| 632 | BOOST_ASSERT(strict_ischar(ch)); |
| 633 | return (iso8859_1_char_types[ch] & BOOST_CC_CTRL) != 0; |
| 634 | } |
| 635 | |
| 636 | static bool |
| 637 | isgraph(int ch) |
| 638 | { |
| 639 | return ('\x21' <= ch && ch <= '\x7e') || ('\xa1' <= ch && ch <= '\xff'); |
| 640 | } |
| 641 | |
| 642 | static bool |
| 643 | islower(int ch) |
| 644 | { |
| 645 | BOOST_ASSERT(strict_ischar(ch)); |
| 646 | return (iso8859_1_char_types[ch] & BOOST_CC_LOWER) != 0; |
| 647 | } |
| 648 | |
| 649 | static bool |
| 650 | isprint(int ch) |
| 651 | { |
| 652 | return ('\x20' <= ch && ch <= '\x7e') || ('\xa0' <= ch && ch <= '\xff'); |
| 653 | } |
| 654 | |
| 655 | static bool |
| 656 | ispunct(int ch) |
| 657 | { |
| 658 | BOOST_ASSERT(strict_ischar(ch)); |
| 659 | return (iso8859_1_char_types[ch] & BOOST_CC_PUNCT) != 0; |
| 660 | } |
| 661 | |
| 662 | static bool |
| 663 | isspace(int ch) |
| 664 | { |
| 665 | BOOST_ASSERT(strict_ischar(ch)); |
| 666 | return (iso8859_1_char_types[ch] & BOOST_CC_SPACE) != 0; |
| 667 | } |
| 668 | |
| 669 | static bool |
| 670 | isblank BOOST_PREVENT_MACRO_SUBSTITUTION (int ch) |
| 671 | { |
| 672 | BOOST_ASSERT(strict_ischar(ch)); |
| 673 | return ('\x09' == ch || '\x20' == ch || '\xa0' == ch); |
| 674 | } |
| 675 | |
| 676 | static bool |
| 677 | isupper(int ch) |
| 678 | { |
| 679 | BOOST_ASSERT(strict_ischar(ch)); |
| 680 | return (iso8859_1_char_types[ch] & BOOST_CC_UPPER) != 0; |
| 681 | } |
| 682 | |
| 683 | /////////////////////////////////////////////////////////////////////////// |
| 684 | // Simple character conversions |
| 685 | /////////////////////////////////////////////////////////////////////////// |
| 686 | |
| 687 | static int |
| 688 | tolower(int ch) |
| 689 | { |
| 690 | BOOST_ASSERT(strict_ischar(ch)); |
| 691 | return isupper(ch) && '\0' != iso8859_1_char_conversion[ch] ? |
| 692 | iso8859_1_char_conversion[ch] : ch; |
| 693 | } |
| 694 | |
| 695 | static int |
| 696 | toupper(int ch) |
| 697 | { |
| 698 | BOOST_ASSERT(strict_ischar(ch)); |
| 699 | return islower(ch) && '\0' != iso8859_1_char_conversion[ch] ? |
| 700 | iso8859_1_char_conversion[ch] : ch; |
| 701 | } |
| 702 | |
| 703 | static ::boost::uint32_t |
| 704 | toucs4(int ch) |
| 705 | { |
| 706 | // The first 256 characters in Unicode and the UCS are |
| 707 | // identical to those in ISO/IEC-8859-1. |
| 708 | BOOST_ASSERT(strict_ischar(ch)); |
| 709 | return ch; |
| 710 | } |
| 711 | }; |
| 712 | |
| 713 | }}} |
| 714 | |
| 715 | /////////////////////////////////////////////////////////////////////////////// |
| 716 | // undefine macros |
| 717 | /////////////////////////////////////////////////////////////////////////////// |
| 718 | #undef BOOST_CC_DIGIT |
| 719 | #undef BOOST_CC_XDIGIT |
| 720 | #undef BOOST_CC_ALPHA |
| 721 | #undef BOOST_CC_CTRL |
| 722 | #undef BOOST_CC_LOWER |
| 723 | #undef BOOST_CC_UPPER |
| 724 | #undef BOOST_CC_PUNCT |
| 725 | #undef BOOST_CC_SPACE |
| 726 | |
| 727 | #endif |
| 728 | |
| 729 | |