1 /-- This set of tests is for UTF-8 support but not Unicode property support, 2 and is relevant only to the 8-bit library. --/ 3 4 < forbid W 5 6 /X(\C{3})/8 7 X\x{1234} 8 9 /X(\C{4})/8 10 X\x{1234}YZ 11 12 /X\C*/8 13 XYZabcdce 14 15 /X\C*?/8 16 XYZabcde 17 18 /X\C{3,5}/8 19 Xabcdefg 20 X\x{1234} 21 X\x{1234}YZ 22 X\x{1234}\x{512} 23 X\x{1234}\x{512}YZ 24 25 /X\C{3,5}?/8 26 Xabcdefg 27 X\x{1234} 28 X\x{1234}YZ 29 X\x{1234}\x{512} 30 31 /a\Cb/8 32 aXb 33 a\nb 34 35 /a\C\Cb/8 36 a\x{100}b 37 38 /ab\Cde/8 39 abXde 40 41 /a\C\Cb/8 42 a\x{100}b 43 ** Failers 44 a\x{12257}b 45 46 /[]/8 47 48 //8 49 50 /xxx/8 51 52 /xxx/8?DZSSO 53 54 /badutf/8 55 \xdf 56 \xef 57 \xef\x80 58 \xf7 59 \xf7\x80 60 \xf7\x80\x80 61 \xfb 62 \xfb\x80 63 \xfb\x80\x80 64 \xfb\x80\x80\x80 65 \xfd 66 \xfd\x80 67 \xfd\x80\x80 68 \xfd\x80\x80\x80 69 \xfd\x80\x80\x80\x80 70 \xdf\x7f 71 \xef\x7f\x80 72 \xef\x80\x7f 73 \xf7\x7f\x80\x80 74 \xf7\x80\x7f\x80 75 \xf7\x80\x80\x7f 76 \xfb\x7f\x80\x80\x80 77 \xfb\x80\x7f\x80\x80 78 \xfb\x80\x80\x7f\x80 79 \xfb\x80\x80\x80\x7f 80 \xfd\x7f\x80\x80\x80\x80 81 \xfd\x80\x7f\x80\x80\x80 82 \xfd\x80\x80\x7f\x80\x80 83 \xfd\x80\x80\x80\x7f\x80 84 \xfd\x80\x80\x80\x80\x7f 85 \xed\xa0\x80 86 \xc0\x8f 87 \xe0\x80\x8f 88 \xf0\x80\x80\x8f 89 \xf8\x80\x80\x80\x8f 90 \xfc\x80\x80\x80\x80\x8f 91 \x80 92 \xfe 93 \xff 94 95 /badutf/8 96 \xfb\x80\x80\x80\x80 97 \xfd\x80\x80\x80\x80\x80 98 \xf7\xbf\xbf\xbf 99 100 /shortutf/8 101 \P\P\xdf 102 \P\P\xef 103 \P\P\xef\x80 104 \P\P\xf7 105 \P\P\xf7\x80 106 \P\P\xf7\x80\x80 107 \P\P\xfb 108 \P\P\xfb\x80 109 \P\P\xfb\x80\x80 110 \P\P\xfb\x80\x80\x80 111 \P\P\xfd 112 \P\P\xfd\x80 113 \P\P\xfd\x80\x80 114 \P\P\xfd\x80\x80\x80 115 \P\P\xfd\x80\x80\x80\x80 116 117 /anything/8 118 \xc0\x80 119 \xc1\x8f 120 \xe0\x9f\x80 121 \xf0\x8f\x80\x80 122 \xf8\x87\x80\x80\x80 123 \xfc\x83\x80\x80\x80\x80 124 \xfe\x80\x80\x80\x80\x80 125 \xff\x80\x80\x80\x80\x80 126 \xc3\x8f 127 \xe0\xaf\x80 128 \xe1\x80\x80 129 \xf0\x9f\x80\x80 130 \xf1\x8f\x80\x80 131 \xf8\x88\x80\x80\x80 132 \xf9\x87\x80\x80\x80 133 \xfc\x84\x80\x80\x80\x80 134 \xfd\x83\x80\x80\x80\x80 135 \?\xf8\x88\x80\x80\x80 136 \?\xf9\x87\x80\x80\x80 137 \?\xfc\x84\x80\x80\x80\x80 138 \?\xfd\x83\x80\x80\x80\x80 139 140 /\x{100}/8DZ 141 142 /\x{1000}/8DZ 143 144 /\x{10000}/8DZ 145 146 /\x{100000}/8DZ 147 148 /\x{10ffff}/8DZ 149 150 /[\x{ff}]/8DZ 151 152 /[\x{100}]/8DZ 153 154 /\x80/8DZ 155 156 /\xff/8DZ 157 158 /\x{D55c}\x{ad6d}\x{C5B4}/DZ8 159 \x{D55c}\x{ad6d}\x{C5B4} 160 161 /\x{65e5}\x{672c}\x{8a9e}/DZ8 162 \x{65e5}\x{672c}\x{8a9e} 163 164 /\x{80}/DZ8 165 166 /\x{084}/DZ8 167 168 /\x{104}/DZ8 169 170 /\x{861}/DZ8 171 172 /\x{212ab}/DZ8 173 174 /-- This one is here not because it's different to Perl, but because the way 175 the captured single-byte is displayed. (In Perl it becomes a character, and you 176 can't tell the difference.) --/ 177 178 /X(\C)(.*)/8 179 X\x{1234} 180 X\nabc 181 182 /-- This one is here because Perl gives out a grumbly error message (quite 183 correctly, but that messes up comparisons). --/ 184 185 /a\Cb/8 186 *** Failers 187 a\x{100}b 188 189 /[^ab\xC0-\xF0]/8SDZ 190 \x{f1} 191 \x{bf} 192 \x{100} 193 \x{1000} 194 *** Failers 195 \x{c0} 196 \x{f0} 197 198 /{3,4}/8SDZ 199 \x{100}\x{100}\x{100}\x{100\x{100} 200 201 /(\x{100}+|x)/8SDZ 202 203 /(\x{100}*a|x)/8SDZ 204 205 /(\x{100}{0,2}a|x)/8SDZ 206 207 /(\x{100}{1,2}a|x)/8SDZ 208 209 /\x{100}/8DZ 210 211 /a\x{100}\x{101}*/8DZ 212 213 /a\x{100}\x{101}+/8DZ 214 215 /[^\x{c4}]/DZ 216 217 /[\x{100}]/8DZ 218 \x{100} 219 Z\x{100} 220 \x{100}Z 221 *** Failers 222 223 /[\xff]/DZ8 224 >\x{ff}< 225 226 /[^\xff]/8DZ 227 228 /\x{100}abc(xyz(?1))/8DZ 229 230 /a\x{1234}b/P8 231 a\x{1234}b 232 233 /\777/8I 234 \x{1ff} 235 \777 236 237 /\x{100}+\x{200}/8DZ 238 239 /\x{100}+X/8DZ 240 241 /^[\Q\E-\Q\E/BZ8 242 243 /-- This tests the stricter UTF-8 check according to RFC 3629. --/ 244 245 /X/8 246 \x{d800} 247 \x{d800}\? 248 \x{da00} 249 \x{da00}\? 250 \x{dfff} 251 \x{dfff}\? 252 \x{110000} 253 \x{110000}\? 254 \x{2000000} 255 \x{2000000}\? 256 \x{7fffffff} 257 \x{7fffffff}\? 258 259 /(*UTF8)\x{1234}/ 260 abcd\x{1234}pqr 261 262 /(*CRLF)(*UTF)(*BSR_UNICODE)a\Rb/I 263 264 /\h/SI8 265 ABC\x{09} 266 ABC\x{20} 267 ABC\x{a0} 268 ABC\x{1680} 269 ABC\x{180e} 270 ABC\x{2000} 271 ABC\x{202f} 272 ABC\x{205f} 273 ABC\x{3000} 274 275 /\v/SI8 276 ABC\x{0a} 277 ABC\x{0b} 278 ABC\x{0c} 279 ABC\x{0d} 280 ABC\x{85} 281 ABC\x{2028} 282 283 /\h*A/SI8 284 CDBABC 285 286 /\v+A/SI8 287 288 /\s?xxx\s/8SI 289 290 /\sxxx\s/I8ST1 291 AB\x{85}xxx\x{a0}XYZ 292 AB\x{a0}xxx\x{85}XYZ 293 294 /\S \S/I8ST1 295 \x{a2} \x{84} 296 A Z 297 298 /a+/8 299 a\x{123}aa\>1 300 a\x{123}aa\>2 301 a\x{123}aa\>3 302 a\x{123}aa\>4 303 a\x{123}aa\>5 304 a\x{123}aa\>6 305 306 /\x{1234}+/iS8I 307 308 /\x{1234}+?/iS8I 309 310 /\x{1234}++/iS8I 311 312 /\x{1234}{2}/iS8I 313 314 /[^\x{c4}]/8DZ 315 316 /X+\x{200}/8DZ 317 318 /\R/SI8 319 320 /\777/8DZ 321 322 /\w+\x{C4}/8BZ 323 a\x{C4}\x{C4} 324 325 /\w+\x{C4}/8BZT1 326 a\x{C4}\x{C4} 327 328 /\W+\x{C4}/8BZ 329 !\x{C4} 330 331 /\W+\x{C4}/8BZT1 332 !\x{C4} 333 334 /\W+\x{A1}/8BZ 335 !\x{A1} 336 337 /\W+\x{A1}/8BZT1 338 !\x{A1} 339 340 /X\s+\x{A0}/8BZ 341 X\x20\x{A0}\x{A0} 342 343 /X\s+\x{A0}/8BZT1 344 X\x20\x{A0}\x{A0} 345 346 /\S+\x{A0}/8BZ 347 X\x{A0}\x{A0} 348 349 /\S+\x{A0}/8BZT1 350 X\x{A0}\x{A0} 351 352 /\x{a0}+\s!/8BZ 353 \x{a0}\x20! 354 355 /\x{a0}+\s!/8BZT1 356 \x{a0}\x20! 357 358 /A/8 359 \x{ff000041} 360 \x{7f000041} 361 362 /(*UTF8)abc/9 363 364 /abc/89 365 366 /-- End of testinput15 --/ 367
