|
Added
Link Here
|
| 1 |
/* |
| 2 |
* Copyright (c) 2013 Damien Miller <djm@mindrot.org> |
| 3 |
* |
| 4 |
* Permission to use, copy, modify, and distribute this software for any |
| 5 |
* purpose with or without fee is hereby granted, provided that the above |
| 6 |
* copyright notice and this permission notice appear in all copies. |
| 7 |
* |
| 8 |
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES |
| 9 |
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF |
| 10 |
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR |
| 11 |
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
| 12 |
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
| 13 |
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
| 14 |
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
| 15 |
*/ |
| 16 |
|
| 17 |
/* |
| 18 |
* This is a simple RFC3454 stringprep profile to sanitise UTF-8 strings |
| 19 |
* from untrusted sources. |
| 20 |
* |
| 21 |
* It is intended to be used prior to display of untrusted strings only. |
| 22 |
* It should not be used for logging because of bi-di ambiguity. It |
| 23 |
* should also not be used in any case where lack of normalisation may |
| 24 |
* cause problems. |
| 25 |
* |
| 26 |
* This profile uses the prohibition and mapping tables from RFC3454 |
| 27 |
* (listed below) but the unassigned character table has been updated to |
| 28 |
* Unicode 6.2. It uses a local whitelist of whitespace characters (\n, |
| 29 |
* \a and \t). Unicode normalisation and bi-di testing are not used. |
| 30 |
* |
| 31 |
* XXX: implement bi-di handling (needed for logs) |
| 32 |
* XXX: implement KC normalisation (needed for passing to libs/syscalls) |
| 33 |
*/ |
| 34 |
|
| 35 |
#include <sys/types.h> |
| 36 |
#include <stdio.h> |
| 37 |
#include <stdlib.h> |
| 38 |
#include <string.h> |
| 39 |
#include <limits.h> |
| 40 |
#include <ctype.h> |
| 41 |
|
| 42 |
#include "misc.h" |
| 43 |
|
| 44 |
#define BANNED_RFC3629 0 |
| 45 |
#define UTF8_DEBUG 0 |
| 46 |
|
| 47 |
#if UTF8_DEBUG |
| 48 |
# define DBG(x) \ |
| 49 |
do { \ |
| 50 |
printf("%s:%d %s: ", __FILE__, __LINE__, __func__); \ |
| 51 |
printf x; \ |
| 52 |
printf("\n"); \ |
| 53 |
fflush(stdout); \ |
| 54 |
} while (0) |
| 55 |
#else |
| 56 |
# define DBG(x) |
| 57 |
#endif |
| 58 |
|
| 59 |
struct u32_range { |
| 60 |
u_int32_t lo, hi; /* Inclusive */ |
| 61 |
}; |
| 62 |
|
| 63 |
/* RFC3454 Table A.1 */ |
| 64 |
static const struct u32_range unassigned[] = { |
| 65 |
{ 0x0378, 0x0379 }, |
| 66 |
{ 0x037F, 0x0383 }, |
| 67 |
{ 0x038B, 0x038B }, |
| 68 |
{ 0x038D, 0x038D }, |
| 69 |
{ 0x03A2, 0x03A2 }, |
| 70 |
{ 0x0528, 0x0530 }, |
| 71 |
{ 0x0557, 0x0558 }, |
| 72 |
{ 0x0560, 0x0560 }, |
| 73 |
{ 0x0588, 0x0588 }, |
| 74 |
{ 0x058B, 0x058E }, |
| 75 |
{ 0x0590, 0x0590 }, |
| 76 |
{ 0x05C8, 0x05CF }, |
| 77 |
{ 0x05EB, 0x05EF }, |
| 78 |
{ 0x05F5, 0x05FF }, |
| 79 |
{ 0x0605, 0x0605 }, |
| 80 |
{ 0x061C, 0x061D }, |
| 81 |
{ 0x070E, 0x070E }, |
| 82 |
{ 0x074B, 0x074C }, |
| 83 |
{ 0x07B2, 0x07BF }, |
| 84 |
{ 0x07FB, 0x07FF }, |
| 85 |
{ 0x082E, 0x082F }, |
| 86 |
{ 0x083F, 0x083F }, |
| 87 |
{ 0x085C, 0x085D }, |
| 88 |
{ 0x085F, 0x089F }, |
| 89 |
{ 0x08A1, 0x08A1 }, |
| 90 |
{ 0x08AD, 0x08E3 }, |
| 91 |
{ 0x08FF, 0x08FF }, |
| 92 |
{ 0x0978, 0x0978 }, |
| 93 |
{ 0x0980, 0x0980 }, |
| 94 |
{ 0x0984, 0x0984 }, |
| 95 |
{ 0x098D, 0x098E }, |
| 96 |
{ 0x0991, 0x0992 }, |
| 97 |
{ 0x09A9, 0x09A9 }, |
| 98 |
{ 0x09B1, 0x09B1 }, |
| 99 |
{ 0x09B3, 0x09B5 }, |
| 100 |
{ 0x09BA, 0x09BB }, |
| 101 |
{ 0x09C5, 0x09C6 }, |
| 102 |
{ 0x09C9, 0x09CA }, |
| 103 |
{ 0x09CF, 0x09D6 }, |
| 104 |
{ 0x09D8, 0x09DB }, |
| 105 |
{ 0x09DE, 0x09DE }, |
| 106 |
{ 0x09E4, 0x09E5 }, |
| 107 |
{ 0x09FC, 0x0A00 }, |
| 108 |
{ 0x0A04, 0x0A04 }, |
| 109 |
{ 0x0A0B, 0x0A0E }, |
| 110 |
{ 0x0A11, 0x0A12 }, |
| 111 |
{ 0x0A29, 0x0A29 }, |
| 112 |
{ 0x0A31, 0x0A31 }, |
| 113 |
{ 0x0A34, 0x0A34 }, |
| 114 |
{ 0x0A37, 0x0A37 }, |
| 115 |
{ 0x0A3A, 0x0A3B }, |
| 116 |
{ 0x0A3D, 0x0A3D }, |
| 117 |
{ 0x0A43, 0x0A46 }, |
| 118 |
{ 0x0A49, 0x0A4A }, |
| 119 |
{ 0x0A4E, 0x0A50 }, |
| 120 |
{ 0x0A52, 0x0A58 }, |
| 121 |
{ 0x0A5D, 0x0A5D }, |
| 122 |
{ 0x0A5F, 0x0A65 }, |
| 123 |
{ 0x0A76, 0x0A80 }, |
| 124 |
{ 0x0A84, 0x0A84 }, |
| 125 |
{ 0x0A8E, 0x0A8E }, |
| 126 |
{ 0x0A92, 0x0A92 }, |
| 127 |
{ 0x0AA9, 0x0AA9 }, |
| 128 |
{ 0x0AB1, 0x0AB1 }, |
| 129 |
{ 0x0AB4, 0x0AB4 }, |
| 130 |
{ 0x0ABA, 0x0ABB }, |
| 131 |
{ 0x0AC6, 0x0AC6 }, |
| 132 |
{ 0x0ACA, 0x0ACA }, |
| 133 |
{ 0x0ACE, 0x0ACF }, |
| 134 |
{ 0x0AD1, 0x0ADF }, |
| 135 |
{ 0x0AE4, 0x0AE5 }, |
| 136 |
{ 0x0AF2, 0x0B00 }, |
| 137 |
{ 0x0B04, 0x0B04 }, |
| 138 |
{ 0x0B0D, 0x0B0E }, |
| 139 |
{ 0x0B11, 0x0B12 }, |
| 140 |
{ 0x0B29, 0x0B29 }, |
| 141 |
{ 0x0B31, 0x0B31 }, |
| 142 |
{ 0x0B34, 0x0B34 }, |
| 143 |
{ 0x0B3A, 0x0B3B }, |
| 144 |
{ 0x0B45, 0x0B46 }, |
| 145 |
{ 0x0B49, 0x0B4A }, |
| 146 |
{ 0x0B4E, 0x0B55 }, |
| 147 |
{ 0x0B58, 0x0B5B }, |
| 148 |
{ 0x0B5E, 0x0B5E }, |
| 149 |
{ 0x0B64, 0x0B65 }, |
| 150 |
{ 0x0B78, 0x0B81 }, |
| 151 |
{ 0x0B84, 0x0B84 }, |
| 152 |
{ 0x0B8B, 0x0B8D }, |
| 153 |
{ 0x0B91, 0x0B91 }, |
| 154 |
{ 0x0B96, 0x0B98 }, |
| 155 |
{ 0x0B9B, 0x0B9B }, |
| 156 |
{ 0x0B9D, 0x0B9D }, |
| 157 |
{ 0x0BA0, 0x0BA2 }, |
| 158 |
{ 0x0BA5, 0x0BA7 }, |
| 159 |
{ 0x0BAB, 0x0BAD }, |
| 160 |
{ 0x0BBA, 0x0BBD }, |
| 161 |
{ 0x0BC3, 0x0BC5 }, |
| 162 |
{ 0x0BC9, 0x0BC9 }, |
| 163 |
{ 0x0BCE, 0x0BCF }, |
| 164 |
{ 0x0BD1, 0x0BD6 }, |
| 165 |
{ 0x0BD8, 0x0BE5 }, |
| 166 |
{ 0x0BFB, 0x0C00 }, |
| 167 |
{ 0x0C04, 0x0C04 }, |
| 168 |
{ 0x0C0D, 0x0C0D }, |
| 169 |
{ 0x0C11, 0x0C11 }, |
| 170 |
{ 0x0C29, 0x0C29 }, |
| 171 |
{ 0x0C34, 0x0C34 }, |
| 172 |
{ 0x0C3A, 0x0C3C }, |
| 173 |
{ 0x0C45, 0x0C45 }, |
| 174 |
{ 0x0C49, 0x0C49 }, |
| 175 |
{ 0x0C4E, 0x0C54 }, |
| 176 |
{ 0x0C57, 0x0C57 }, |
| 177 |
{ 0x0C5A, 0x0C5F }, |
| 178 |
{ 0x0C64, 0x0C65 }, |
| 179 |
{ 0x0C70, 0x0C77 }, |
| 180 |
{ 0x0C80, 0x0C81 }, |
| 181 |
{ 0x0C84, 0x0C84 }, |
| 182 |
{ 0x0C8D, 0x0C8D }, |
| 183 |
{ 0x0C91, 0x0C91 }, |
| 184 |
{ 0x0CA9, 0x0CA9 }, |
| 185 |
{ 0x0CB4, 0x0CB4 }, |
| 186 |
{ 0x0CBA, 0x0CBB }, |
| 187 |
{ 0x0CC5, 0x0CC5 }, |
| 188 |
{ 0x0CC9, 0x0CC9 }, |
| 189 |
{ 0x0CCE, 0x0CD4 }, |
| 190 |
{ 0x0CD7, 0x0CDD }, |
| 191 |
{ 0x0CDF, 0x0CDF }, |
| 192 |
{ 0x0CE4, 0x0CE5 }, |
| 193 |
{ 0x0CF0, 0x0CF0 }, |
| 194 |
{ 0x0CF3, 0x0D01 }, |
| 195 |
{ 0x0D04, 0x0D04 }, |
| 196 |
{ 0x0D0D, 0x0D0D }, |
| 197 |
{ 0x0D11, 0x0D11 }, |
| 198 |
{ 0x0D3B, 0x0D3C }, |
| 199 |
{ 0x0D45, 0x0D45 }, |
| 200 |
{ 0x0D49, 0x0D49 }, |
| 201 |
{ 0x0D4F, 0x0D56 }, |
| 202 |
{ 0x0D58, 0x0D5F }, |
| 203 |
{ 0x0D64, 0x0D65 }, |
| 204 |
{ 0x0D76, 0x0D78 }, |
| 205 |
{ 0x0D80, 0x0D81 }, |
| 206 |
{ 0x0D84, 0x0D84 }, |
| 207 |
{ 0x0D97, 0x0D99 }, |
| 208 |
{ 0x0DB2, 0x0DB2 }, |
| 209 |
{ 0x0DBC, 0x0DBC }, |
| 210 |
{ 0x0DBE, 0x0DBF }, |
| 211 |
{ 0x0DC7, 0x0DC9 }, |
| 212 |
{ 0x0DCB, 0x0DCE }, |
| 213 |
{ 0x0DD5, 0x0DD5 }, |
| 214 |
{ 0x0DD7, 0x0DD7 }, |
| 215 |
{ 0x0DE0, 0x0DF1 }, |
| 216 |
{ 0x0DF5, 0x0E00 }, |
| 217 |
{ 0x0E3B, 0x0E3E }, |
| 218 |
{ 0x0E5C, 0x0E80 }, |
| 219 |
{ 0x0E83, 0x0E83 }, |
| 220 |
{ 0x0E85, 0x0E86 }, |
| 221 |
{ 0x0E89, 0x0E89 }, |
| 222 |
{ 0x0E8B, 0x0E8C }, |
| 223 |
{ 0x0E8E, 0x0E93 }, |
| 224 |
{ 0x0E98, 0x0E98 }, |
| 225 |
{ 0x0EA0, 0x0EA0 }, |
| 226 |
{ 0x0EA4, 0x0EA4 }, |
| 227 |
{ 0x0EA6, 0x0EA6 }, |
| 228 |
{ 0x0EA8, 0x0EA9 }, |
| 229 |
{ 0x0EAC, 0x0EAC }, |
| 230 |
{ 0x0EBA, 0x0EBA }, |
| 231 |
{ 0x0EBE, 0x0EBF }, |
| 232 |
{ 0x0EC5, 0x0EC5 }, |
| 233 |
{ 0x0EC7, 0x0EC7 }, |
| 234 |
{ 0x0ECE, 0x0ECF }, |
| 235 |
{ 0x0EDA, 0x0EDB }, |
| 236 |
{ 0x0EE0, 0x0EFF }, |
| 237 |
{ 0x0F48, 0x0F48 }, |
| 238 |
{ 0x0F6D, 0x0F70 }, |
| 239 |
{ 0x0F98, 0x0F98 }, |
| 240 |
{ 0x0FBD, 0x0FBD }, |
| 241 |
{ 0x0FCD, 0x0FCD }, |
| 242 |
{ 0x0FDB, 0x0FFF }, |
| 243 |
{ 0x10C6, 0x10C6 }, |
| 244 |
{ 0x10C8, 0x10CC }, |
| 245 |
{ 0x10CE, 0x10CF }, |
| 246 |
{ 0x1249, 0x1249 }, |
| 247 |
{ 0x124E, 0x124F }, |
| 248 |
{ 0x1257, 0x1257 }, |
| 249 |
{ 0x1259, 0x1259 }, |
| 250 |
{ 0x125E, 0x125F }, |
| 251 |
{ 0x1289, 0x1289 }, |
| 252 |
{ 0x128E, 0x128F }, |
| 253 |
{ 0x12B1, 0x12B1 }, |
| 254 |
{ 0x12B6, 0x12B7 }, |
| 255 |
{ 0x12BF, 0x12BF }, |
| 256 |
{ 0x12C1, 0x12C1 }, |
| 257 |
{ 0x12C6, 0x12C7 }, |
| 258 |
{ 0x12D7, 0x12D7 }, |
| 259 |
{ 0x1311, 0x1311 }, |
| 260 |
{ 0x1316, 0x1317 }, |
| 261 |
{ 0x135B, 0x135C }, |
| 262 |
{ 0x137D, 0x137F }, |
| 263 |
{ 0x139A, 0x139F }, |
| 264 |
{ 0x13F5, 0x13FF }, |
| 265 |
{ 0x169D, 0x169F }, |
| 266 |
{ 0x16F1, 0x16FF }, |
| 267 |
{ 0x170D, 0x170D }, |
| 268 |
{ 0x1715, 0x171F }, |
| 269 |
{ 0x1737, 0x173F }, |
| 270 |
{ 0x1754, 0x175F }, |
| 271 |
{ 0x176D, 0x176D }, |
| 272 |
{ 0x1771, 0x1771 }, |
| 273 |
{ 0x1774, 0x177F }, |
| 274 |
{ 0x17DE, 0x17DF }, |
| 275 |
{ 0x17EA, 0x17EF }, |
| 276 |
{ 0x17FA, 0x17FF }, |
| 277 |
{ 0x180F, 0x180F }, |
| 278 |
{ 0x181A, 0x181F }, |
| 279 |
{ 0x1878, 0x187F }, |
| 280 |
{ 0x18AB, 0x18AF }, |
| 281 |
{ 0x18F6, 0x18FF }, |
| 282 |
{ 0x191D, 0x191F }, |
| 283 |
{ 0x192C, 0x192F }, |
| 284 |
{ 0x193C, 0x193F }, |
| 285 |
{ 0x1941, 0x1943 }, |
| 286 |
{ 0x196E, 0x196F }, |
| 287 |
{ 0x1975, 0x197F }, |
| 288 |
{ 0x19AC, 0x19AF }, |
| 289 |
{ 0x19CA, 0x19CF }, |
| 290 |
{ 0x19DB, 0x19DD }, |
| 291 |
{ 0x1A1C, 0x1A1D }, |
| 292 |
{ 0x1A5F, 0x1A5F }, |
| 293 |
{ 0x1A7D, 0x1A7E }, |
| 294 |
{ 0x1A8A, 0x1A8F }, |
| 295 |
{ 0x1A9A, 0x1A9F }, |
| 296 |
{ 0x1AAE, 0x1AFF }, |
| 297 |
{ 0x1B4C, 0x1B4F }, |
| 298 |
{ 0x1B7D, 0x1B7F }, |
| 299 |
{ 0x1BF4, 0x1BFB }, |
| 300 |
{ 0x1C38, 0x1C3A }, |
| 301 |
{ 0x1C4A, 0x1C4C }, |
| 302 |
{ 0x1C80, 0x1CBF }, |
| 303 |
{ 0x1CC8, 0x1CCF }, |
| 304 |
{ 0x1CF7, 0x1CFF }, |
| 305 |
{ 0x1DE7, 0x1DFB }, |
| 306 |
{ 0x1F16, 0x1F17 }, |
| 307 |
{ 0x1F1E, 0x1F1F }, |
| 308 |
{ 0x1F46, 0x1F47 }, |
| 309 |
{ 0x1F4E, 0x1F4F }, |
| 310 |
{ 0x1F58, 0x1F58 }, |
| 311 |
{ 0x1F5A, 0x1F5A }, |
| 312 |
{ 0x1F5C, 0x1F5C }, |
| 313 |
{ 0x1F5E, 0x1F5E }, |
| 314 |
{ 0x1F7E, 0x1F7F }, |
| 315 |
{ 0x1FB5, 0x1FB5 }, |
| 316 |
{ 0x1FC5, 0x1FC5 }, |
| 317 |
{ 0x1FD4, 0x1FD5 }, |
| 318 |
{ 0x1FDC, 0x1FDC }, |
| 319 |
{ 0x1FF0, 0x1FF1 }, |
| 320 |
{ 0x1FF5, 0x1FF5 }, |
| 321 |
{ 0x1FFF, 0x1FFF }, |
| 322 |
{ 0x2065, 0x2069 }, |
| 323 |
{ 0x2072, 0x2073 }, |
| 324 |
{ 0x208F, 0x208F }, |
| 325 |
{ 0x209D, 0x209F }, |
| 326 |
{ 0x20BB, 0x20CF }, |
| 327 |
{ 0x20F1, 0x20FF }, |
| 328 |
{ 0x218A, 0x218F }, |
| 329 |
{ 0x23F4, 0x23FF }, |
| 330 |
{ 0x2427, 0x243F }, |
| 331 |
{ 0x244B, 0x245F }, |
| 332 |
{ 0x2700, 0x2700 }, |
| 333 |
{ 0x2B4D, 0x2B4F }, |
| 334 |
{ 0x2B5A, 0x2BFF }, |
| 335 |
{ 0x2C2F, 0x2C2F }, |
| 336 |
{ 0x2C5F, 0x2C5F }, |
| 337 |
{ 0x2CF4, 0x2CF8 }, |
| 338 |
{ 0x2D26, 0x2D26 }, |
| 339 |
{ 0x2D28, 0x2D2C }, |
| 340 |
{ 0x2D2E, 0x2D2F }, |
| 341 |
{ 0x2D68, 0x2D6E }, |
| 342 |
{ 0x2D71, 0x2D7E }, |
| 343 |
{ 0x2D97, 0x2D9F }, |
| 344 |
{ 0x2DA7, 0x2DA7 }, |
| 345 |
{ 0x2DAF, 0x2DAF }, |
| 346 |
{ 0x2DB7, 0x2DB7 }, |
| 347 |
{ 0x2DBF, 0x2DBF }, |
| 348 |
{ 0x2DC7, 0x2DC7 }, |
| 349 |
{ 0x2DCF, 0x2DCF }, |
| 350 |
{ 0x2DD7, 0x2DD7 }, |
| 351 |
{ 0x2DDF, 0x2DDF }, |
| 352 |
{ 0x2E3C, 0x2E7F }, |
| 353 |
{ 0x2E9A, 0x2E9A }, |
| 354 |
{ 0x2EF4, 0x2EFF }, |
| 355 |
{ 0x2FD6, 0x2FEF }, |
| 356 |
{ 0x2FFC, 0x2FFF }, |
| 357 |
{ 0x3040, 0x3040 }, |
| 358 |
{ 0x3097, 0x3098 }, |
| 359 |
{ 0x3100, 0x3104 }, |
| 360 |
{ 0x312E, 0x3130 }, |
| 361 |
{ 0x318F, 0x318F }, |
| 362 |
{ 0x31BB, 0x31BF }, |
| 363 |
{ 0x31E4, 0x31EF }, |
| 364 |
{ 0x321F, 0x321F }, |
| 365 |
{ 0x32FF, 0x32FF }, |
| 366 |
{ 0x4DB6, 0x4DBF }, |
| 367 |
{ 0x9FA6, 0x9FCB }, |
| 368 |
{ 0x9FCD, 0x9FFF }, |
| 369 |
{ 0xA48D, 0xA48F }, |
| 370 |
{ 0xA4C7, 0xA4CF }, |
| 371 |
{ 0xA62C, 0xA63F }, |
| 372 |
{ 0xA698, 0xA69E }, |
| 373 |
{ 0xA6F8, 0xA6FF }, |
| 374 |
{ 0xA78F, 0xA78F }, |
| 375 |
{ 0xA794, 0xA79F }, |
| 376 |
{ 0xA7AB, 0xA7F7 }, |
| 377 |
{ 0xA82C, 0xA82F }, |
| 378 |
{ 0xA83A, 0xA83F }, |
| 379 |
{ 0xA878, 0xA87F }, |
| 380 |
{ 0xA8C5, 0xA8CD }, |
| 381 |
{ 0xA8DA, 0xA8DF }, |
| 382 |
{ 0xA8FC, 0xA8FF }, |
| 383 |
{ 0xA954, 0xA95E }, |
| 384 |
{ 0xA97D, 0xA97F }, |
| 385 |
{ 0xA9CE, 0xA9CE }, |
| 386 |
{ 0xA9DA, 0xA9DD }, |
| 387 |
{ 0xA9E0, 0xA9FF }, |
| 388 |
{ 0xAA37, 0xAA3F }, |
| 389 |
{ 0xAA4E, 0xAA4F }, |
| 390 |
{ 0xAA5A, 0xAA5B }, |
| 391 |
{ 0xAA7C, 0xAA7F }, |
| 392 |
{ 0xAAC3, 0xAADA }, |
| 393 |
{ 0xAAF7, 0xAB00 }, |
| 394 |
{ 0xAB07, 0xAB08 }, |
| 395 |
{ 0xAB0F, 0xAB10 }, |
| 396 |
{ 0xAB17, 0xAB1F }, |
| 397 |
{ 0xAB27, 0xAB27 }, |
| 398 |
{ 0xAB2F, 0xABBF }, |
| 399 |
{ 0xABEE, 0xABEF }, |
| 400 |
{ 0xABFA, 0xABFF }, |
| 401 |
{ 0xD7A4, 0xD7AF }, |
| 402 |
{ 0xD7C7, 0xD7CA }, |
| 403 |
{ 0xD7FC, 0xD7FF }, |
| 404 |
{ 0xFA6E, 0xFA6F }, |
| 405 |
{ 0xFADA, 0xFAFF }, |
| 406 |
{ 0xFB07, 0xFB12 }, |
| 407 |
{ 0xFB18, 0xFB1C }, |
| 408 |
{ 0xFB37, 0xFB37 }, |
| 409 |
{ 0xFB3D, 0xFB3D }, |
| 410 |
{ 0xFB3F, 0xFB3F }, |
| 411 |
{ 0xFB42, 0xFB42 }, |
| 412 |
{ 0xFB45, 0xFB45 }, |
| 413 |
{ 0xFBC2, 0xFBD2 }, |
| 414 |
{ 0xFD40, 0xFD4F }, |
| 415 |
{ 0xFD90, 0xFD91 }, |
| 416 |
{ 0xFDC8, 0xFDCF }, |
| 417 |
{ 0xFDFE, 0xFDFF }, |
| 418 |
{ 0xFE1A, 0xFE1F }, |
| 419 |
{ 0xFE27, 0xFE2F }, |
| 420 |
{ 0xFE53, 0xFE53 }, |
| 421 |
{ 0xFE67, 0xFE67 }, |
| 422 |
{ 0xFE6C, 0xFE6F }, |
| 423 |
{ 0xFE75, 0xFE75 }, |
| 424 |
{ 0xFEFD, 0xFEFE }, |
| 425 |
{ 0xFF00, 0xFF00 }, |
| 426 |
{ 0xFFBF, 0xFFC1 }, |
| 427 |
{ 0xFFC8, 0xFFC9 }, |
| 428 |
{ 0xFFD0, 0xFFD1 }, |
| 429 |
{ 0xFFD8, 0xFFD9 }, |
| 430 |
{ 0xFFDD, 0xFFDF }, |
| 431 |
{ 0xFFE7, 0xFFE7 }, |
| 432 |
{ 0xFFEF, 0xFFF8 }, |
| 433 |
{ 0x1000C, 0x1000C }, |
| 434 |
{ 0x10027, 0x10027 }, |
| 435 |
{ 0x1003B, 0x1003B }, |
| 436 |
{ 0x1003E, 0x1003E }, |
| 437 |
{ 0x1004E, 0x1004F }, |
| 438 |
{ 0x1005E, 0x1007F }, |
| 439 |
{ 0x100FB, 0x100FF }, |
| 440 |
{ 0x10103, 0x10106 }, |
| 441 |
{ 0x10134, 0x10136 }, |
| 442 |
{ 0x1018B, 0x1018F }, |
| 443 |
{ 0x1019C, 0x101CF }, |
| 444 |
{ 0x101FE, 0x1027F }, |
| 445 |
{ 0x1029D, 0x1029F }, |
| 446 |
{ 0x102D1, 0x102FF }, |
| 447 |
{ 0x1031F, 0x1031F }, |
| 448 |
{ 0x10324, 0x1032F }, |
| 449 |
{ 0x1034B, 0x1037F }, |
| 450 |
{ 0x1039E, 0x1039E }, |
| 451 |
{ 0x103C4, 0x103C7 }, |
| 452 |
{ 0x103D6, 0x103FF }, |
| 453 |
{ 0x1049E, 0x1049F }, |
| 454 |
{ 0x104AA, 0x107FF }, |
| 455 |
{ 0x10806, 0x10807 }, |
| 456 |
{ 0x10809, 0x10809 }, |
| 457 |
{ 0x10836, 0x10836 }, |
| 458 |
{ 0x10839, 0x1083B }, |
| 459 |
{ 0x1083D, 0x1083E }, |
| 460 |
{ 0x10856, 0x10856 }, |
| 461 |
{ 0x10860, 0x108FF }, |
| 462 |
{ 0x1091C, 0x1091E }, |
| 463 |
{ 0x1093A, 0x1093E }, |
| 464 |
{ 0x10940, 0x1097F }, |
| 465 |
{ 0x109B8, 0x109BD }, |
| 466 |
{ 0x109C0, 0x109FF }, |
| 467 |
{ 0x10A04, 0x10A04 }, |
| 468 |
{ 0x10A07, 0x10A0B }, |
| 469 |
{ 0x10A14, 0x10A14 }, |
| 470 |
{ 0x10A18, 0x10A18 }, |
| 471 |
{ 0x10A34, 0x10A37 }, |
| 472 |
{ 0x10A3B, 0x10A3E }, |
| 473 |
{ 0x10A48, 0x10A4F }, |
| 474 |
{ 0x10A59, 0x10A5F }, |
| 475 |
{ 0x10A80, 0x10AFF }, |
| 476 |
{ 0x10B36, 0x10B38 }, |
| 477 |
{ 0x10B56, 0x10B57 }, |
| 478 |
{ 0x10B73, 0x10B77 }, |
| 479 |
{ 0x10B80, 0x10BFF }, |
| 480 |
{ 0x10C49, 0x10E5F }, |
| 481 |
{ 0x10E7F, 0x10FFF }, |
| 482 |
{ 0x1104E, 0x11051 }, |
| 483 |
{ 0x11070, 0x1107F }, |
| 484 |
{ 0x110C2, 0x110CF }, |
| 485 |
{ 0x110E9, 0x110EF }, |
| 486 |
{ 0x110FA, 0x110FF }, |
| 487 |
{ 0x11135, 0x11135 }, |
| 488 |
{ 0x11144, 0x1117F }, |
| 489 |
{ 0x111C9, 0x111CF }, |
| 490 |
{ 0x111DA, 0x1167F }, |
| 491 |
{ 0x116B8, 0x116BF }, |
| 492 |
{ 0x116CA, 0x11FFF }, |
| 493 |
{ 0x1236F, 0x123FF }, |
| 494 |
{ 0x12463, 0x1246F }, |
| 495 |
{ 0x12474, 0x12FFF }, |
| 496 |
{ 0x1342F, 0x167FF }, |
| 497 |
{ 0x16A39, 0x16EFF }, |
| 498 |
{ 0x16F45, 0x16F4F }, |
| 499 |
{ 0x16F7F, 0x16F8E }, |
| 500 |
{ 0x16FA0, 0x1AFFF }, |
| 501 |
{ 0x1B002, 0x1CFFF }, |
| 502 |
{ 0x1D0F6, 0x1D0FF }, |
| 503 |
{ 0x1D127, 0x1D128 }, |
| 504 |
{ 0x1D1DE, 0x1D1FF }, |
| 505 |
{ 0x1D246, 0x1D2FF }, |
| 506 |
{ 0x1D357, 0x1D35F }, |
| 507 |
{ 0x1D372, 0x1D3FF }, |
| 508 |
{ 0x1D455, 0x1D455 }, |
| 509 |
{ 0x1D49D, 0x1D49D }, |
| 510 |
{ 0x1D4A0, 0x1D4A1 }, |
| 511 |
{ 0x1D4A3, 0x1D4A4 }, |
| 512 |
{ 0x1D4A7, 0x1D4A8 }, |
| 513 |
{ 0x1D4AD, 0x1D4AD }, |
| 514 |
{ 0x1D4BA, 0x1D4BA }, |
| 515 |
{ 0x1D4BC, 0x1D4BC }, |
| 516 |
{ 0x1D4C4, 0x1D4C4 }, |
| 517 |
{ 0x1D506, 0x1D506 }, |
| 518 |
{ 0x1D50B, 0x1D50C }, |
| 519 |
{ 0x1D515, 0x1D515 }, |
| 520 |
{ 0x1D51D, 0x1D51D }, |
| 521 |
{ 0x1D53A, 0x1D53A }, |
| 522 |
{ 0x1D53F, 0x1D53F }, |
| 523 |
{ 0x1D545, 0x1D545 }, |
| 524 |
{ 0x1D547, 0x1D549 }, |
| 525 |
{ 0x1D551, 0x1D551 }, |
| 526 |
{ 0x1D6A6, 0x1D6A7 }, |
| 527 |
{ 0x1D7CC, 0x1D7CD }, |
| 528 |
{ 0x1D800, 0x1EDFF }, |
| 529 |
{ 0x1EE04, 0x1EE04 }, |
| 530 |
{ 0x1EE20, 0x1EE20 }, |
| 531 |
{ 0x1EE23, 0x1EE23 }, |
| 532 |
{ 0x1EE25, 0x1EE26 }, |
| 533 |
{ 0x1EE28, 0x1EE28 }, |
| 534 |
{ 0x1EE33, 0x1EE33 }, |
| 535 |
{ 0x1EE38, 0x1EE38 }, |
| 536 |
{ 0x1EE3A, 0x1EE3A }, |
| 537 |
{ 0x1EE3C, 0x1EE41 }, |
| 538 |
{ 0x1EE43, 0x1EE46 }, |
| 539 |
{ 0x1EE48, 0x1EE48 }, |
| 540 |
{ 0x1EE4A, 0x1EE4A }, |
| 541 |
{ 0x1EE4C, 0x1EE4C }, |
| 542 |
{ 0x1EE50, 0x1EE50 }, |
| 543 |
{ 0x1EE53, 0x1EE53 }, |
| 544 |
{ 0x1EE55, 0x1EE56 }, |
| 545 |
{ 0x1EE58, 0x1EE58 }, |
| 546 |
{ 0x1EE5A, 0x1EE5A }, |
| 547 |
{ 0x1EE5C, 0x1EE5C }, |
| 548 |
{ 0x1EE5E, 0x1EE5E }, |
| 549 |
{ 0x1EE60, 0x1EE60 }, |
| 550 |
{ 0x1EE63, 0x1EE63 }, |
| 551 |
{ 0x1EE65, 0x1EE66 }, |
| 552 |
{ 0x1EE6B, 0x1EE6B }, |
| 553 |
{ 0x1EE73, 0x1EE73 }, |
| 554 |
{ 0x1EE78, 0x1EE78 }, |
| 555 |
{ 0x1EE7D, 0x1EE7D }, |
| 556 |
{ 0x1EE7F, 0x1EE7F }, |
| 557 |
{ 0x1EE8A, 0x1EE8A }, |
| 558 |
{ 0x1EE9C, 0x1EEA0 }, |
| 559 |
{ 0x1EEA4, 0x1EEA4 }, |
| 560 |
{ 0x1EEAA, 0x1EEAA }, |
| 561 |
{ 0x1EEBC, 0x1EEEF }, |
| 562 |
{ 0x1EEF2, 0x1EFFF }, |
| 563 |
{ 0x1F02C, 0x1F02F }, |
| 564 |
{ 0x1F094, 0x1F09F }, |
| 565 |
{ 0x1F0AF, 0x1F0B0 }, |
| 566 |
{ 0x1F0BF, 0x1F0C0 }, |
| 567 |
{ 0x1F0D0, 0x1F0D0 }, |
| 568 |
{ 0x1F0E0, 0x1F0FF }, |
| 569 |
{ 0x1F10B, 0x1F10F }, |
| 570 |
{ 0x1F12F, 0x1F12F }, |
| 571 |
{ 0x1F16C, 0x1F16F }, |
| 572 |
{ 0x1F19B, 0x1F1E5 }, |
| 573 |
{ 0x1F203, 0x1F20F }, |
| 574 |
{ 0x1F23B, 0x1F23F }, |
| 575 |
{ 0x1F249, 0x1F24F }, |
| 576 |
{ 0x1F252, 0x1F2FF }, |
| 577 |
{ 0x1F321, 0x1F32F }, |
| 578 |
{ 0x1F336, 0x1F336 }, |
| 579 |
{ 0x1F37D, 0x1F37F }, |
| 580 |
{ 0x1F394, 0x1F39F }, |
| 581 |
{ 0x1F3C5, 0x1F3C5 }, |
| 582 |
{ 0x1F3CB, 0x1F3DF }, |
| 583 |
{ 0x1F3F1, 0x1F3FF }, |
| 584 |
{ 0x1F43F, 0x1F43F }, |
| 585 |
{ 0x1F441, 0x1F441 }, |
| 586 |
{ 0x1F4F8, 0x1F4F8 }, |
| 587 |
{ 0x1F4FD, 0x1F4FF }, |
| 588 |
{ 0x1F53E, 0x1F53F }, |
| 589 |
{ 0x1F544, 0x1F54F }, |
| 590 |
{ 0x1F568, 0x1F5FA }, |
| 591 |
{ 0x1F641, 0x1F644 }, |
| 592 |
{ 0x1F650, 0x1F67F }, |
| 593 |
{ 0x1F6C6, 0x1F6FF }, |
| 594 |
{ 0x1F774, 0x1FFFD }, |
| 595 |
{ 0x2A6D7, 0x2A6FF }, |
| 596 |
{ 0x2A701, 0x2B733 }, |
| 597 |
{ 0x2B735, 0x2B73F }, |
| 598 |
{ 0x2B741, 0x2B81C }, |
| 599 |
{ 0x2B81E, 0x2F7FF }, |
| 600 |
{ 0x2FA1E, 0x2FFFD }, |
| 601 |
{ 0x30000, 0x3FFFD }, |
| 602 |
{ 0x40000, 0x4FFFD }, |
| 603 |
{ 0x50000, 0x5FFFD }, |
| 604 |
{ 0x60000, 0x6FFFD }, |
| 605 |
{ 0x70000, 0x7FFFD }, |
| 606 |
{ 0x80000, 0x8FFFD }, |
| 607 |
{ 0x90000, 0x9FFFD }, |
| 608 |
{ 0xA0000, 0xAFFFD }, |
| 609 |
{ 0xB0000, 0xBFFFD }, |
| 610 |
{ 0xC0000, 0xCFFFD }, |
| 611 |
{ 0xD0000, 0xDFFFD }, |
| 612 |
{ 0xE0000, 0xE0000 }, |
| 613 |
{ 0xE0002, 0xE001F }, |
| 614 |
{ 0xE0080, 0xE00FF }, |
| 615 |
{ 0xE01F0, 0xEFFFD }, |
| 616 |
}; |
| 617 |
|
| 618 |
/* RFC3454 Table B.1 */ |
| 619 |
static const struct u32_range map_to_nothing[] = { |
| 620 |
{ 0x00AD, 0x00AD }, |
| 621 |
{ 0x034F, 0x034F }, |
| 622 |
{ 0x1806, 0x1806 }, |
| 623 |
{ 0x180B, 0x180D }, |
| 624 |
{ 0x200B, 0x200D }, |
| 625 |
{ 0x2060, 0x2060 }, |
| 626 |
{ 0xFE00, 0xFE0F }, |
| 627 |
{ 0xFEFF, 0xFEFF }, |
| 628 |
}; |
| 629 |
|
| 630 |
/* Local: allow tab, CR and LF */ |
| 631 |
static const struct u32_range whitelist[] = { |
| 632 |
{ 0x09, 0x00 }, |
| 633 |
{ 0x0a, 0x0a }, |
| 634 |
{ 0x0d, 0x0d }, |
| 635 |
}; |
| 636 |
|
| 637 |
/* RFC3454 Tables in appendix C */ |
| 638 |
static const struct u32_range prohibited[] = { |
| 639 |
/* C.2.1 ASCII control characters */ |
| 640 |
{ 0x0000, 0x001F }, |
| 641 |
{ 0x007F, 0x007F }, |
| 642 |
/* C.2.2 Non-ASCII control characters */ |
| 643 |
{ 0x0080, 0x009F }, |
| 644 |
{ 0x06DD, 0x06DD }, |
| 645 |
{ 0x070F, 0x070F }, |
| 646 |
{ 0x180E, 0x180E }, |
| 647 |
{ 0x200C, 0x200C }, |
| 648 |
{ 0x200D, 0x200D }, |
| 649 |
{ 0x2028, 0x2028 }, |
| 650 |
{ 0x2029, 0x2029 }, |
| 651 |
{ 0x2060, 0x2060 }, |
| 652 |
{ 0x2061, 0x2061 }, |
| 653 |
{ 0x2062, 0x2062 }, |
| 654 |
{ 0x2063, 0x2063 }, |
| 655 |
{ 0x206A, 0x206F }, |
| 656 |
{ 0xFEFF, 0xFEFF }, |
| 657 |
{ 0xFFF9, 0xFFFC }, |
| 658 |
{ 0x1D173, 0x1D17A }, |
| 659 |
/* C.3 Private use */ |
| 660 |
{ 0xE000, 0xF8FF }, |
| 661 |
{ 0xF0000, 0xFFFFD }, |
| 662 |
{ 0x100000, 0x10FFFD }, |
| 663 |
/* C.4 Non-character code points */ |
| 664 |
{ 0xFDD0, 0xFDEF }, |
| 665 |
{ 0xFFFE, 0xFFFF }, |
| 666 |
{ 0x1FFFE, 0x1FFFF }, |
| 667 |
{ 0x2FFFE, 0x2FFFF }, |
| 668 |
{ 0x3FFFE, 0x3FFFF }, |
| 669 |
{ 0x4FFFE, 0x4FFFF }, |
| 670 |
{ 0x5FFFE, 0x5FFFF }, |
| 671 |
{ 0x6FFFE, 0x6FFFF }, |
| 672 |
{ 0x7FFFE, 0x7FFFF }, |
| 673 |
{ 0x8FFFE, 0x8FFFF }, |
| 674 |
{ 0x9FFFE, 0x9FFFF }, |
| 675 |
{ 0xAFFFE, 0xAFFFF }, |
| 676 |
{ 0xBFFFE, 0xBFFFF }, |
| 677 |
{ 0xCFFFE, 0xCFFFF }, |
| 678 |
{ 0xDFFFE, 0xDFFFF }, |
| 679 |
{ 0xEFFFE, 0xEFFFF }, |
| 680 |
{ 0xFFFFE, 0xFFFFF }, |
| 681 |
{ 0x10FFFE, 0x10FFFF }, |
| 682 |
/* C.5 Surrogate codes */ |
| 683 |
{ 0xD800, 0xDFFF }, |
| 684 |
/* C.6 Inappropriate for plain text */ |
| 685 |
{ 0xFFF9, 0xFFF9 }, |
| 686 |
{ 0xFFFA, 0xFFFA }, |
| 687 |
{ 0xFFFB, 0xFFFB }, |
| 688 |
{ 0xFFFC, 0xFFFC }, |
| 689 |
{ 0xFFFD, 0xFFFD }, |
| 690 |
/* C.7 Inappropriate for canonical representation */ |
| 691 |
{ 0x2FF0, 0x2FFB }, |
| 692 |
/* C.8 Change display properties or are deprecated */ |
| 693 |
{ 0x0340, 0x0340 }, |
| 694 |
{ 0x0341, 0x0341 }, |
| 695 |
{ 0x200E, 0x200E }, |
| 696 |
{ 0x200F, 0x200F }, |
| 697 |
{ 0x202A, 0x202A }, |
| 698 |
{ 0x202B, 0x202B }, |
| 699 |
{ 0x202C, 0x202C }, |
| 700 |
{ 0x202D, 0x202D }, |
| 701 |
{ 0x202E, 0x202E }, |
| 702 |
{ 0x206A, 0x206A }, |
| 703 |
{ 0x206B, 0x206B }, |
| 704 |
{ 0x206C, 0x206C }, |
| 705 |
{ 0x206D, 0x206D }, |
| 706 |
{ 0x206E, 0x206E }, |
| 707 |
{ 0x206F, 0x206F }, |
| 708 |
/* C.9 Tagging characters */ |
| 709 |
{ 0xE0001, 0xE0001 }, |
| 710 |
{ 0xE0020, 0xE007F }, |
| 711 |
}; |
| 712 |
|
| 713 |
static int |
| 714 |
encode_utf8(u_int32_t c, char *s, size_t slen) |
| 715 |
{ |
| 716 |
DBG(("encode CP 0x%x", c)); |
| 717 |
if (c < 0x80) { |
| 718 |
if (slen >= 1) { |
| 719 |
s[0] = (char)c; |
| 720 |
} |
| 721 |
DBG(("CP 0x%x => %02x", c, (u_char)s[0])); |
| 722 |
return 1; |
| 723 |
} else if (c < 0x800) { |
| 724 |
if (slen >= 2) { |
| 725 |
s[0] = (char)(0xc0 | (c >> 6)); |
| 726 |
s[1] = (char)(0x80 | (c & 0x3f)); |
| 727 |
} |
| 728 |
DBG(("CP 0x%x => %02x %02x", c, (u_char)s[0], (u_char)s[1])); |
| 729 |
return 2; |
| 730 |
} else if (c < 0x10000) { |
| 731 |
if (slen >= 3) { |
| 732 |
s[0] = (char)(0xe0 | (c >> 12)); |
| 733 |
s[1] = (char)(0x80 | ((c >> 6) & 0x3f)); |
| 734 |
s[2] = (char)(0x80 | (c & 0x3f)); |
| 735 |
} |
| 736 |
DBG(("CP 0x%x => %02x %02x %02x", c, s[0], s[1], s[2])); |
| 737 |
return 3; |
| 738 |
} else if (c < 0x200000) { |
| 739 |
if (slen >= 4) { |
| 740 |
s[0] = (char)(0xf0 | (c >> 18)); |
| 741 |
s[1] = (char)(0x80 | ((c >> 12) & 0x3f)); |
| 742 |
s[2] = (char)(0x80 | ((c >> 6) & 0x3f)); |
| 743 |
s[3] = (char)(0x80 | (c & 0x3f)); |
| 744 |
} |
| 745 |
DBG(("CP 0x%x => %02x %02x %02x %02x", c, |
| 746 |
s[0], s[1], s[2], s[3])); |
| 747 |
return 4; |
| 748 |
#if BANED_RFC3629 |
| 749 |
} else if (c < 0x4000000) { |
| 750 |
if (slen >= 5) { |
| 751 |
s[0] = (char)(0xf8 | (c >> 24)); |
| 752 |
s[1] = (char)(0x80 | ((c >> 18) & 0x3f)); |
| 753 |
s[2] = (char)(0x80 | ((c >> 12) & 0x3f)); |
| 754 |
s[3] = (char)(0x80 | ((c >> 6) & 0x3f)); |
| 755 |
s[4] = (char)(0x80 | (c & 0x3f)); |
| 756 |
} |
| 757 |
DBG(("CP 0x%x => %02x %02x %02x %02x %02x", c, |
| 758 |
s[0], s[1], s[2], s[3], s[4], s[5])); |
| 759 |
return 5; |
| 760 |
} else if (c < 0x80000000) { |
| 761 |
if (slen >= 6) { |
| 762 |
s[0] = (char)(0xfc | (c >> 30)); |
| 763 |
s[1] = (char)(0x80 | ((c >> 24) & 0x3f)); |
| 764 |
s[2] = (char)(0x80 | ((c >> 18) & 0x3f)); |
| 765 |
s[3] = (char)(0x80 | ((c >> 12) & 0x3f)); |
| 766 |
s[4] = (char)(0x80 | ((c >> 6) & 0x3f)); |
| 767 |
s[5] = (char)(0x80 | (c & 0x3f)); |
| 768 |
} |
| 769 |
DBG(("CP 0x%x => %02x %02x %02x %02x %02x %02x", c, |
| 770 |
s[0], s[1], s[2], s[3], s[4], s[5], s[6])); |
| 771 |
return 6; |
| 772 |
#endif |
| 773 |
} |
| 774 |
DBG(("INTERNAL ERROR: CP 0x%x not encodeable", c)); |
| 775 |
return -1; |
| 776 |
} |
| 777 |
|
| 778 |
static int |
| 779 |
code_in_table(u_int32_t c, const struct u32_range *table, size_t tlen) |
| 780 |
{ |
| 781 |
const struct u32_range *e, *end = (void *)(tlen + (char *)table); |
| 782 |
|
| 783 |
for (e = table; e < end; e++) { |
| 784 |
if (c >= e->lo && c <= e->hi) |
| 785 |
return 1; |
| 786 |
} |
| 787 |
return 0; |
| 788 |
} |
| 789 |
|
| 790 |
|
| 791 |
int |
| 792 |
ssh_utf8_stringprep(const char *in, char *_out, size_t olen) |
| 793 |
{ |
| 794 |
u_char *out = (u_char *)_out; |
| 795 |
int r, state = 0; |
| 796 |
size_t i, o, ilen = strlen(in); |
| 797 |
u_int32_t c, e; |
| 798 |
|
| 799 |
if (olen < 1) |
| 800 |
return -1; |
| 801 |
|
| 802 |
e = c = 0; |
| 803 |
for (i = o = 0; i < ilen; i++) { |
| 804 |
e = (u_char)in[i]; |
| 805 |
DBG(("top: i=%lu c=0x%02x e=0x%02x '%c' state=%d", |
| 806 |
(u_long)i, c, e, isprint(e) ? e : ' ', state)); |
| 807 |
/* Invalid code point state */ |
| 808 |
if (state == -1) { |
| 809 |
/* |
| 810 |
* Continue eating continuation characters until |
| 811 |
* a new start character comes along. |
| 812 |
*/ |
| 813 |
if ((e & 0xc0) == 0x80) |
| 814 |
continue; |
| 815 |
state = 0; |
| 816 |
} |
| 817 |
|
| 818 |
/* New code point state */ |
| 819 |
if (state == 0) { |
| 820 |
DBG(("new code point")); |
| 821 |
if ((e & 0x80) == 0) { |
| 822 |
/* 7 bit code point: skip to output */ |
| 823 |
c = e & 0x7f; |
| 824 |
DBG(("7 bit CP: c=0x%02x '%c'", |
| 825 |
c, isprint(c) ? c : '.')); |
| 826 |
goto have_code; |
| 827 |
} |
| 828 |
if ((e & 0xe0) == 0xc0) { |
| 829 |
/* 11 bit code point */ |
| 830 |
state = 1; |
| 831 |
c = (e & 0x1f) << 6; |
| 832 |
DBG(("start 11 bit CP: c=0x%02x", c)); |
| 833 |
if (c == 0) |
| 834 |
goto bad_encoding; |
| 835 |
continue; |
| 836 |
} |
| 837 |
if ((e & 0xf0) == 0xe0) { |
| 838 |
/* 16 bit code point */ |
| 839 |
state = 2; |
| 840 |
c = (e & 0xf) << 12; |
| 841 |
DBG(("start 16 bit CP: c=0x%02x", c)); |
| 842 |
if (c == 0) |
| 843 |
goto bad_encoding; |
| 844 |
continue; |
| 845 |
} |
| 846 |
if ((e & 0xf8) == 0xf0) { |
| 847 |
/* 21 bit code point */ |
| 848 |
state = 3; |
| 849 |
c = (e & 0x7) << 18; |
| 850 |
DBG(("start 21 bit CP: c=0x%02x", c)); |
| 851 |
if (c == 0) |
| 852 |
goto bad_encoding; |
| 853 |
continue; |
| 854 |
} |
| 855 |
#if BANNED_RFC3629 |
| 856 |
if ((e & 0xfc) == 0xf8) { |
| 857 |
/* 26 bit code point */ |
| 858 |
state = 4; |
| 859 |
c = (e & 0x3) << 24; |
| 860 |
DBG(("start 26 bit CP: c=0x%02x", c)); |
| 861 |
if (c == 0) |
| 862 |
goto bad_encoding; |
| 863 |
continue; |
| 864 |
} |
| 865 |
if ((e & 0xfe) == 0xfc) { |
| 866 |
/* 31 bit code point */ |
| 867 |
state = 5; |
| 868 |
c = (e & 0x1) << 30; |
| 869 |
DBG(("start 31 bit CP: c=0x%02x", c)); |
| 870 |
if (c == 0) |
| 871 |
goto bad_encoding; |
| 872 |
continue; |
| 873 |
} |
| 874 |
#endif |
| 875 |
bad_encoding: |
| 876 |
DBG(("bad encoding")); |
| 877 |
c = 0; |
| 878 |
state = -1; |
| 879 |
continue; |
| 880 |
} |
| 881 |
|
| 882 |
/* In multibyte code point state */ |
| 883 |
if (state < 1 || state > 5) { |
| 884 |
DBG(("INTERNAL ERROR: invalid state %d", state)); |
| 885 |
return -1; |
| 886 |
} |
| 887 |
DBG(("CP continuation; before c=0x%02x state=%d", c, state)); |
| 888 |
state--; |
| 889 |
c |= (e & 0x3f) << (state * 6); |
| 890 |
DBG(("CP continuation; after c=0x%02x state=%d", c, state)); |
| 891 |
if (state > 0) |
| 892 |
continue; |
| 893 |
|
| 894 |
have_code: |
| 895 |
DBG(("CP done; final c=0x%02x", c)); |
| 896 |
/* Character finished. prepare, encode and output */ |
| 897 |
if (c == 0) |
| 898 |
break; /* shouldn't happen */ |
| 899 |
|
| 900 |
/* RFC3629 bans codepoints > U+10FFFF */ |
| 901 |
if (c > 0x10FFFF) { |
| 902 |
DBG(("CP 0x%x > 0x10FFFF", c)); |
| 903 |
goto bad_encoding; |
| 904 |
} |
| 905 |
/* Mapping */ |
| 906 |
if (code_in_table(c, map_to_nothing, sizeof(map_to_nothing))) { |
| 907 |
DBG(("CP 0x%x mapped to nothing", c)); |
| 908 |
continue; |
| 909 |
} |
| 910 |
/* Prohibitied output */ |
| 911 |
if (code_in_table(c, prohibited, sizeof(prohibited)) && |
| 912 |
!code_in_table(c, whitelist, sizeof(whitelist))) { |
| 913 |
DBG(("CP 0x%x is in prohibited list", c)); |
| 914 |
return -1; |
| 915 |
} |
| 916 |
/* Map unassigned code points to U+FFFD */ |
| 917 |
if (code_in_table(c, unassigned, sizeof(unassigned))) { |
| 918 |
DBG(("CP 0x%x is unassigned", c)); |
| 919 |
c = 0xFFFD; |
| 920 |
} |
| 921 |
|
| 922 |
/* Encode the character */ |
| 923 |
r = encode_utf8(c, out + o, olen - o - 1); |
| 924 |
DBG(("CP 0x%x encode returned %d o=%lu of %lu", |
| 925 |
c, r, (u_long)o, (u_long)olen)); |
| 926 |
if (r < 0) |
| 927 |
return -1; |
| 928 |
o += r; |
| 929 |
} |
| 930 |
out[o] = '\0'; |
| 931 |
return 0; |
| 932 |
} |
| 933 |
|
| 934 |
#if 0 |
| 935 |
#include <err.h> |
| 936 |
|
| 937 |
int |
| 938 |
main(int argc, char **argv) |
| 939 |
{ |
| 940 |
char in[8192], out[8192]; |
| 941 |
|
| 942 |
while (fgets(in, sizeof(in), stdin) != NULL) { |
| 943 |
if (ssh_utf8_stringprep(in, out, sizeof(out)) != 0) |
| 944 |
errx(1, "bad input"); |
| 945 |
fputs(out, stdout); |
| 946 |
} |
| 947 |
return 0; |
| 948 |
} |
| 949 |
#endif |