| 1 |
/-- This set of tests is for UTF-8 support, and is relevant only to the 8-bit
|
| 2 |
library. --/
|
| 3 |
|
| 4 |
/X(\C{3})/8
|
| 5 |
X\x{1234}
|
| 6 |
|
| 7 |
/X(\C{4})/8
|
| 8 |
X\x{1234}YZ
|
| 9 |
|
| 10 |
/X\C*/8
|
| 11 |
XYZabcdce
|
| 12 |
|
| 13 |
/X\C*?/8
|
| 14 |
XYZabcde
|
| 15 |
|
| 16 |
/X\C{3,5}/8
|
| 17 |
Xabcdefg
|
| 18 |
X\x{1234}
|
| 19 |
X\x{1234}YZ
|
| 20 |
X\x{1234}\x{512}
|
| 21 |
X\x{1234}\x{512}YZ
|
| 22 |
|
| 23 |
/X\C{3,5}?/8
|
| 24 |
Xabcdefg
|
| 25 |
X\x{1234}
|
| 26 |
X\x{1234}YZ
|
| 27 |
X\x{1234}\x{512}
|
| 28 |
|
| 29 |
/a\Cb/8
|
| 30 |
aXb
|
| 31 |
a\nb
|
| 32 |
|
| 33 |
/a\C\Cb/8
|
| 34 |
a\x{100}b
|
| 35 |
|
| 36 |
/ab\Cde/8
|
| 37 |
abXde
|
| 38 |
|
| 39 |
/a\C\Cb/8
|
| 40 |
a\x{100}b
|
| 41 |
** Failers
|
| 42 |
a\x{12257}b
|
| 43 |
|
| 44 |
/[]/8
|
| 45 |
|
| 46 |
//8
|
| 47 |
|
| 48 |
/xxx/8
|
| 49 |
|
| 50 |
/xxx/8?DZSS
|
| 51 |
|
| 52 |
/abc/8
|
| 53 |
]
|
| 54 |
|
| 55 |
|
| 56 |
\?
|
| 57 |
\xe1\x88
|
| 58 |
\P\xe1\x88
|
| 59 |
\P\P\xe1\x88
|
| 60 |
XX\xea
|
| 61 |
\O0XX\xea
|
| 62 |
\O1XX\xea
|
| 63 |
\O2XX\xea
|
| 64 |
XX\xf1
|
| 65 |
XX\xf8
|
| 66 |
XX\xfc
|
| 67 |
ZZ\xea\xaf\x20YY
|
| 68 |
ZZ\xfd\xbf\xbf\x2f\xbf\xbfYY
|
| 69 |
ZZ\xfd\xbf\xbf\xbf\x2f\xbfYY
|
| 70 |
ZZ\xfd\xbf\xbf\xbf\xbf\x2fYY
|
| 71 |
ZZ\xffYY
|
| 72 |
ZZ\xfeYY
|
| 73 |
|
| 74 |
/anything/8
|
| 75 |
\xc0\x80
|
| 76 |
\xc1\x8f
|
| 77 |
\xe0\x9f\x80
|
| 78 |
\xf0\x8f\x80\x80
|
| 79 |
\xf8\x87\x80\x80\x80
|
| 80 |
\xfc\x83\x80\x80\x80\x80
|
| 81 |
\xfe\x80\x80\x80\x80\x80
|
| 82 |
\xff\x80\x80\x80\x80\x80
|
| 83 |
\xc3\x8f
|
| 84 |
\xe0\xaf\x80
|
| 85 |
\xe1\x80\x80
|
| 86 |
\xf0\x9f\x80\x80
|
| 87 |
\xf1\x8f\x80\x80
|
| 88 |
\xf8\x88\x80\x80\x80
|
| 89 |
\xf9\x87\x80\x80\x80
|
| 90 |
\xfc\x84\x80\x80\x80\x80
|
| 91 |
\xfd\x83\x80\x80\x80\x80
|
| 92 |
\?\xf8\x88\x80\x80\x80
|
| 93 |
\?\xf9\x87\x80\x80\x80
|
| 94 |
\?\xfc\x84\x80\x80\x80\x80
|
| 95 |
\?\xfd\x83\x80\x80\x80\x80
|
| 96 |
|
| 97 |
/\x{100}/8DZ
|
| 98 |
|
| 99 |
/\x{1000}/8DZ
|
| 100 |
|
| 101 |
/\x{10000}/8DZ
|
| 102 |
|
| 103 |
/\x{100000}/8DZ
|
| 104 |
|
| 105 |
/\x{10ffff}/8DZ
|
| 106 |
|
| 107 |
/[\x{ff}]/8DZ
|
| 108 |
|
| 109 |
/[\x{100}]/8DZ
|
| 110 |
|
| 111 |
/\x80/8DZ
|
| 112 |
|
| 113 |
/\xff/8DZ
|
| 114 |
|
| 115 |
/\x{D55c}\x{ad6d}\x{C5B4}/DZ8
|
| 116 |
\x{D55c}\x{ad6d}\x{C5B4}
|
| 117 |
|
| 118 |
/\x{65e5}\x{672c}\x{8a9e}/DZ8
|
| 119 |
\x{65e5}\x{672c}\x{8a9e}
|
| 120 |
|
| 121 |
/\x{80}/DZ8
|
| 122 |
|
| 123 |
/\x{084}/DZ8
|
| 124 |
|
| 125 |
/\x{104}/DZ8
|
| 126 |
|
| 127 |
/\x{861}/DZ8
|
| 128 |
|
| 129 |
/\x{212ab}/DZ8
|
| 130 |
|
| 131 |
/-- This one is here not because it's different to Perl, but because the way
|
| 132 |
the captured single-byte is displayed. (In Perl it becomes a character, and you
|
| 133 |
can't tell the difference.) --/
|
| 134 |
|
| 135 |
/X(\C)(.*)/8
|
| 136 |
X\x{1234}
|
| 137 |
X\nabc
|
| 138 |
|
| 139 |
/-- This one is here because Perl gives out a grumbly error message (quite
|
| 140 |
correctly, but that messes up comparisons). --/
|
| 141 |
|
| 142 |
/a\Cb/8
|
| 143 |
*** Failers
|
| 144 |
a\x{100}b
|
| 145 |
|
| 146 |
/[^ab\xC0-\xF0]/8SDZ
|
| 147 |
\x{f1}
|
| 148 |
\x{bf}
|
| 149 |
\x{100}
|
| 150 |
\x{1000}
|
| 151 |
*** Failers
|
| 152 |
\x{c0}
|
| 153 |
\x{f0}
|
| 154 |
|
| 155 |
/Ā{3,4}/8SDZ
|
| 156 |
\x{100}\x{100}\x{100}\x{100\x{100}
|
| 157 |
|
| 158 |
/(\x{100}+|x)/8SDZ
|
| 159 |
|
| 160 |
/(\x{100}*a|x)/8SDZ
|
| 161 |
|
| 162 |
/(\x{100}{0,2}a|x)/8SDZ
|
| 163 |
|
| 164 |
/(\x{100}{1,2}a|x)/8SDZ
|
| 165 |
|
| 166 |
/\x{100}/8DZ
|
| 167 |
|
| 168 |
/a\x{100}\x{101}*/8DZ
|
| 169 |
|
| 170 |
/a\x{100}\x{101}+/8DZ
|
| 171 |
|
| 172 |
/[^\x{c4}]/DZ
|
| 173 |
|
| 174 |
/[\x{100}]/8DZ
|
| 175 |
\x{100}
|
| 176 |
Z\x{100}
|
| 177 |
\x{100}Z
|
| 178 |
*** Failers
|
| 179 |
|
| 180 |
/[\xff]/DZ8
|
| 181 |
>\x{ff}<
|
| 182 |
|
| 183 |
/[^\xff]/8DZ
|
| 184 |
|
| 185 |
/\x{100}abc(xyz(?1))/8DZ
|
| 186 |
|
| 187 |
/a\x{1234}b/P8
|
| 188 |
a\x{1234}b
|
| 189 |
|
| 190 |
/\777/8I
|
| 191 |
\x{1ff}
|
| 192 |
\777
|
| 193 |
|
| 194 |
/\x{100}+\x{200}/8DZ
|
| 195 |
|
| 196 |
/\x{100}+X/8DZ
|
| 197 |
|
| 198 |
/^[\QĀ\E-\QŐ\E/BZ8
|
| 199 |
|
| 200 |
/-- This tests the stricter UTF-8 check according to RFC 3629. --/
|
| 201 |
|
| 202 |
/X/8
|
| 203 |
\x{0}\x{d7ff}\x{e000}\x{10ffff}
|
| 204 |
\x{d800}
|
| 205 |
\x{d800}\?
|
| 206 |
\x{da00}
|
| 207 |
\x{da00}\?
|
| 208 |
\x{dfff}
|
| 209 |
\x{dfff}\?
|
| 210 |
\x{110000}
|
| 211 |
\x{110000}\?
|
| 212 |
\x{2000000}
|
| 213 |
\x{2000000}\?
|
| 214 |
\x{7fffffff}
|
| 215 |
\x{7fffffff}\?
|
| 216 |
|
| 217 |
/(*UTF8)\x{1234}/
|
| 218 |
abcd\x{1234}pqr
|
| 219 |
|
| 220 |
/(*CRLF)(*UTF8)(*BSR_UNICODE)a\Rb/I
|
| 221 |
|
| 222 |
/\h/SI8
|
| 223 |
ABC\x{09}
|
| 224 |
ABC\x{20}
|
| 225 |
ABC\x{a0}
|
| 226 |
ABC\x{1680}
|
| 227 |
ABC\x{180e}
|
| 228 |
ABC\x{2000}
|
| 229 |
ABC\x{202f}
|
| 230 |
ABC\x{205f}
|
| 231 |
ABC\x{3000}
|
| 232 |
|
| 233 |
/\v/SI8
|
| 234 |
ABC\x{0a}
|
| 235 |
ABC\x{0b}
|
| 236 |
ABC\x{0c}
|
| 237 |
ABC\x{0d}
|
| 238 |
ABC\x{85}
|
| 239 |
ABC\x{2028}
|
| 240 |
|
| 241 |
/\h*A/SI8
|
| 242 |
CDBABC
|
| 243 |
|
| 244 |
/\v+A/SI8
|
| 245 |
|
| 246 |
/\s?xxx\s/8SI
|
| 247 |
|
| 248 |
/\sxxx\s/I8ST1
|
| 249 |
AB\x{85}xxx\x{a0}XYZ
|
| 250 |
AB\x{a0}xxx\x{85}XYZ
|
| 251 |
|
| 252 |
/\S \S/I8ST1
|
| 253 |
\x{a2} \x{84}
|
| 254 |
A Z
|
| 255 |
|
| 256 |
/a+/8
|
| 257 |
a\x{123}aa\>1
|
| 258 |
a\x{123}aa\>2
|
| 259 |
a\x{123}aa\>3
|
| 260 |
a\x{123}aa\>4
|
| 261 |
a\x{123}aa\>5
|
| 262 |
a\x{123}aa\>6
|
| 263 |
|
| 264 |
/\x{1234}+/iS8I
|
| 265 |
|
| 266 |
/\x{1234}+?/iS8I
|
| 267 |
|
| 268 |
/\x{1234}++/iS8I
|
| 269 |
|
| 270 |
/\x{1234}{2}/iS8I
|
| 271 |
|
| 272 |
/[^\x{c4}]/8DZ
|
| 273 |
|
| 274 |
/X+\x{200}/8DZ
|
| 275 |
|
| 276 |
/\R/SI8
|
| 277 |
|
| 278 |
/-- End of testinput15 --/
|