| 1 | # This file is derived from |
|---|
| 2 | # |
|---|
| 3 | # http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt |
|---|
| 4 | # |
|---|
| 5 | # Which was created by Markus Kuhn <mkuhn@acm.org> - 2000-09-02 |
|---|
| 6 | # |
|---|
| 7 | # lines begining with # and blank lines are ignored |
|---|
| 8 | # |
|---|
| 9 | # Beyond that, this file consists of a series of test cases. Each test case consists of |
|---|
| 10 | # 2 or 3 lines: |
|---|
| 11 | # |
|---|
| 12 | # 1. A UTF-8 string |
|---|
| 13 | # 2. A status |
|---|
| 14 | # VALID : The string is a valid UTF-8 representation of valid Unicode |
|---|
| 15 | # INCOMPLETE : The string has a partial character at the end |
|---|
| 16 | # NOTUNICODE : The string is valid UTF-8, but the characters represented |
|---|
| 17 | # are not valid unicode ( |
|---|
| 18 | # OVERLONG : The string includes overlong sequences |
|---|
| 19 | # MALFORMED : The string is not valid UTF-8 |
|---|
| 20 | # 3. If the status is VALID or NOTUNICODE, the UCS-4 representation of the string, |
|---|
| 21 | # as a series of hex numbers. |
|---|
| 22 | |
|---|
| 23 | # 1 Some correct UTF-8 text |
|---|
| 24 | κόÏμε |
|---|
| 25 | VALID |
|---|
| 26 | 03ba 1f79 03c3 03bc 03b5 |
|---|
| 27 | |
|---|
| 28 | # 2.1 First possible sequence of a certain length |
|---|
| 29 | # |
|---|
| 30 | # FIXME - handle NULLS? |
|---|
| 31 | # |
|---|
| 32 | # [ NULL BYTE ] |
|---|
| 33 | #VALID |
|---|
| 34 | #0000 |
|---|
| 35 | |
|---|
| 36 | Â |
|---|
| 37 | VALID |
|---|
| 38 | 0080 |
|---|
| 39 | |
|---|
| 40 | à |
|---|
| 41 | VALID |
|---|
| 42 | 0800 |
|---|
| 43 | |
|---|
| 44 | ð |
|---|
| 45 | VALID |
|---|
| 46 | 00010000 |
|---|
| 47 | |
|---|
| 48 | ø |
|---|
| 49 | NOTUNICODE |
|---|
| 50 | 00200000 |
|---|
| 51 | |
|---|
| 52 | ü |
|---|
| 53 | NOTUNICODE |
|---|
| 54 | 04000000 |
|---|
| 55 | |
|---|
| 56 | |
|---|
| 57 | VALID |
|---|
| 58 | 0000007f |
|---|
| 59 | |
|---|
| 60 | ß¿ |
|---|
| 61 | VALID |
|---|
| 62 | 000007ff |
|---|
| 63 | |
|---|
| 64 | ï¿¿ |
|---|
| 65 | NOTUNICODE |
|---|
| 66 | 0000ffff |
|---|
| 67 | |
|---|
| 68 | ÷¿¿¿ |
|---|
| 69 | NOTUNICODE |
|---|
| 70 | 001fffff |
|---|
| 71 | |
|---|
| 72 | û¿¿¿¿ |
|---|
| 73 | NOTUNICODE |
|---|
| 74 | 03ffffff |
|---|
| 75 | |
|---|
| 76 | ý¿¿¿¿¿ |
|---|
| 77 | NOTUNICODE |
|---|
| 78 | 7fffffff |
|---|
| 79 | |
|---|
| 80 | # 2.3 Other boundary conditions |
|---|
| 81 | |
|---|
| 82 | í¿ |
|---|
| 83 | VALID |
|---|
| 84 | d7ff |
|---|
| 85 | |
|---|
| 86 | î |
|---|
| 87 | VALID |
|---|
| 88 | e000 |
|---|
| 89 | |
|---|
| 90 | � |
|---|
| 91 | VALID |
|---|
| 92 | fffd |
|---|
| 93 | |
|---|
| 94 | ô¿¿ |
|---|
| 95 | VALID |
|---|
| 96 | 0010ffff |
|---|
| 97 | |
|---|
| 98 | ô |
|---|
| 99 | NOTUNICODE |
|---|
| 100 | 00110000 |
|---|
| 101 | |
|---|
| 102 | # 3.1 Unexpected continuation bytes |
|---|
| 103 | |
|---|
| 104 | |
|---|
| 105 | MALFORMED |
|---|
| 106 | ¿ |
|---|
| 107 | MALFORMED |
|---|
| 108 | ¿ |
|---|
| 109 | MALFORMED |
|---|
| 110 | ¿ |
|---|
| 111 | MALFORMED |
|---|
| 112 | ¿¿ |
|---|
| 113 | MALFORMED |
|---|
| 114 | ¿¿ |
|---|
| 115 | MALFORMED |
|---|
| 116 | ¿¿¿ |
|---|
| 117 | MALFORMED |
|---|
| 118 | ¿¿¿ |
|---|
| 119 | MALFORMED |
|---|
| 120 |
¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ |
|---|
| 121 | MALFORMED |
|---|
| 122 | |
|---|
| 123 | # 3.2 Lonely start characters |
|---|
| 124 | |
|---|
| 125 | À Á Â Ã Ä Å Æ Ç È É Ê Ë Ì Í Î Ï Ð Ñ Ò Ó Ô Õ Ö × Ø Ù Ú Û Ü Ý Þ ß |
|---|
| 126 | MALFORMED |
|---|
| 127 | à á â ã ä å æ ç è é ê ë ì í î ï |
|---|
| 128 | MALFORMED |
|---|
| 129 | ð ñ ò ó ô õ ö ÷ |
|---|
| 130 | MALFORMED |
|---|
| 131 | ø ù ú û |
|---|
| 132 | MALFORMED |
|---|
| 133 | ü ý |
|---|
| 134 | MALFORMED |
|---|
| 135 | |
|---|
| 136 | # 3.3 Sequences with last continuation byte missing |
|---|
| 137 | |
|---|
| 138 | À |
|---|
| 139 | INCOMPLETE |
|---|
| 140 | à |
|---|
| 141 | INCOMPLETE |
|---|
| 142 | ð |
|---|
| 143 | INCOMPLETE |
|---|
| 144 | ø |
|---|
| 145 | INCOMPLETE |
|---|
| 146 | ü |
|---|
| 147 | INCOMPLETE |
|---|
| 148 | ß |
|---|
| 149 | INCOMPLETE |
|---|
| 150 | ï¿ |
|---|
| 151 | INCOMPLETE |
|---|
| 152 | ÷¿¿ |
|---|
| 153 | INCOMPLETE |
|---|
| 154 | û¿¿¿ |
|---|
| 155 | INCOMPLETE |
|---|
| 156 | ý¿¿¿¿ |
|---|
| 157 | INCOMPLETE |
|---|
| 158 | |
|---|
| 159 | # 3.4 Concatenation of incomplete sequences |
|---|
| 160 | |
|---|
| 161 | Ààðøüßï¿÷¿¿û¿¿¿ý¿¿¿¿ |
|---|
| 162 | MALFORMED |
|---|
| 163 | |
|---|
| 164 | # 3.5 Impossible bytes |
|---|
| 165 | |
|---|
| 166 | þ |
|---|
| 167 | MALFORMED |
|---|
| 168 | ÿ |
|---|
| 169 | MALFORMED |
|---|
| 170 | þþÿÿ |
|---|
| 171 | MALFORMED |
|---|
| 172 | |
|---|
| 173 | # Examples of an overlong ASCII character |
|---|
| 174 | |
|---|
| 175 | À¯ |
|---|
| 176 | OVERLONG |
|---|
| 177 | ௠|
|---|
| 178 | OVERLONG |
|---|
| 179 | ð¯ |
|---|
| 180 | OVERLONG |
|---|
| 181 | ø¯ |
|---|
| 182 | OVERLONG |
|---|
| 183 | ü¯ |
|---|
| 184 | OVERLONG |
|---|
| 185 | |
|---|
| 186 | # Maximum overlong sequences |
|---|
| 187 | |
|---|
| 188 | Á¿ |
|---|
| 189 | OVERLONG |
|---|
| 190 | à¿ |
|---|
| 191 | OVERLONG |
|---|
| 192 | ð¿¿ |
|---|
| 193 | OVERLONG |
|---|
| 194 | ø¿¿¿ |
|---|
| 195 | OVERLONG |
|---|
| 196 | ü¿¿¿¿ |
|---|
| 197 | OVERLONG |
|---|
| 198 | |
|---|
| 199 | # Overlong representation of the NUL character |
|---|
| 200 | |
|---|
| 201 | À |
|---|
| 202 | OVERLONG |
|---|
| 203 | à |
|---|
| 204 | OVERLONG |
|---|
| 205 | ð |
|---|
| 206 | OVERLONG |
|---|
| 207 | ø |
|---|
| 208 | OVERLONG |
|---|
| 209 | ü |
|---|
| 210 | OVERLONG |
|---|
| 211 | |
|---|
| 212 | # Illegal code positions |
|---|
| 213 | |
|---|
| 214 | # Single UTF-16 surrogates |
|---|
| 215 | |
|---|
| 216 | í |
|---|
| 217 | NOTUNICODE |
|---|
| 218 | d800 |
|---|
| 219 | |
|---|
| 220 | í¿ |
|---|
| 221 | NOTUNICODE |
|---|
| 222 | db7f |
|---|
| 223 | |
|---|
| 224 | í® |
|---|
| 225 | NOTUNICODE |
|---|
| 226 | db80 |
|---|
| 227 | |
|---|
| 228 | í¯¿ |
|---|
| 229 | NOTUNICODE |
|---|
| 230 | dbff |
|---|
| 231 | |
|---|
| 232 | í° |
|---|
| 233 | NOTUNICODE |
|---|
| 234 | dc00 |
|---|
| 235 | |
|---|
| 236 | í¾ |
|---|
| 237 | NOTUNICODE |
|---|
| 238 | df80 |
|---|
| 239 | |
|---|
| 240 | í¿¿ |
|---|
| 241 | NOTUNICODE |
|---|
| 242 | dfff |
|---|
| 243 | |
|---|
| 244 | # Paired UTF-16 surrogates |
|---|
| 245 | |
|---|
| 246 | í í° |
|---|
| 247 | NOTUNICODE |
|---|
| 248 | d800 dc00 |
|---|
| 249 | |
|---|
| 250 | í í¿¿ |
|---|
| 251 | NOTUNICODE |
|---|
| 252 | d800 dfff |
|---|
| 253 | |
|---|
| 254 | í¿í° |
|---|
| 255 | NOTUNICODE |
|---|
| 256 | db7f dc00 |
|---|
| 257 | |
|---|
| 258 | í¿í¿¿ |
|---|
| 259 | NOTUNICODE |
|---|
| 260 | db7f dfff |
|---|
| 261 | |
|---|
| 262 | í®í° |
|---|
| 263 | NOTUNICODE |
|---|
| 264 | db80 dc00 |
|---|
| 265 | |
|---|
| 266 | í®í¿¿ |
|---|
| 267 | NOTUNICODE |
|---|
| 268 | db80 dfff |
|---|
| 269 | |
|---|
| 270 | í¯¿í° |
|---|
| 271 | NOTUNICODE |
|---|
| 272 | dbff dc00 |
|---|
| 273 | |
|---|
| 274 | 􏿿 |
|---|
| 275 | NOTUNICODE |
|---|
| 276 | dbff dfff |
|---|
| 277 | |
|---|
| 278 | # Other illegal code positions |
|---|
| 279 | |
|---|
| 280 | ￾ |
|---|
| 281 | NOTUNICODE |
|---|
| 282 | fffe |
|---|
| 283 | |
|---|
| 284 | ï¿¿ |
|---|
| 285 | NOTUNICODE |
|---|
| 286 | ffff |
|---|
| 287 | |
|---|
| 288 | ################ |
|---|
| 289 | # |
|---|
| 290 | # Some more tests, not from Markus Kuhn's file |
|---|
| 291 | # |
|---|
| 292 | |
|---|
| 293 | # Mixed plane 0 and higher planes |
|---|
| 294 | |
|---|
| 295 | AðBô¿¿C |
|---|
| 296 | VALID |
|---|
| 297 | 41 00010000 42 10ffff 43 |
|---|
