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 |
/badutf/8
|
53 |
\xdf
|
54 |
\xef
|
55 |
\xef\x80
|
56 |
\xf7
|
57 |
\xf7\x80
|
58 |
\xf7\x80\x80
|
59 |
\xfb
|
60 |
\xfb\x80
|
61 |
\xfb\x80\x80
|
62 |
\xfb\x80\x80\x80
|
63 |
\xfd
|
64 |
\xfd\x80
|
65 |
\xfd\x80\x80
|
66 |
\xfd\x80\x80\x80
|
67 |
\xfd\x80\x80\x80\x80
|
68 |
\xdf\x7f
|
69 |
\xef\x7f\x80
|
70 |
\xef\x80\x7f
|
71 |
\xf7\x7f\x80\x80
|
72 |
\xf7\x80\x7f\x80
|
73 |
\xf7\x80\x80\x7f
|
74 |
\xfb\x7f\x80\x80\x80
|
75 |
\xfb\x80\x7f\x80\x80
|
76 |
\xfb\x80\x80\x7f\x80
|
77 |
\xfb\x80\x80\x80\x7f
|
78 |
\xfd\x7f\x80\x80\x80\x80
|
79 |
\xfd\x80\x7f\x80\x80\x80
|
80 |
\xfd\x80\x80\x7f\x80\x80
|
81 |
\xfd\x80\x80\x80\x7f\x80
|
82 |
\xfd\x80\x80\x80\x80\x7f
|
83 |
\xed\xa0\x80
|
84 |
\xc0\x8f
|
85 |
\xe0\x80\x8f
|
86 |
\xf0\x80\x80\x8f
|
87 |
\xf8\x80\x80\x80\x8f
|
88 |
\xfc\x80\x80\x80\x80\x8f
|
89 |
\x80
|
90 |
\xfe
|
91 |
\xff
|
92 |
\xef\xb7\x90
|
93 |
|
94 |
/badutf/8
|
95 |
\xfb\x80\x80\x80\x80
|
96 |
\xfd\x80\x80\x80\x80\x80
|
97 |
\xf7\xbf\xbf\xbf
|
98 |
|
99 |
/shortutf/8
|
100 |
\P\P\xdf
|
101 |
\P\P\xef
|
102 |
\P\P\xef\x80
|
103 |
\P\P\xf7
|
104 |
\P\P\xf7\x80
|
105 |
\P\P\xf7\x80\x80
|
106 |
\P\P\xfb
|
107 |
\P\P\xfb\x80
|
108 |
\P\P\xfb\x80\x80
|
109 |
\P\P\xfb\x80\x80\x80
|
110 |
\P\P\xfd
|
111 |
\P\P\xfd\x80
|
112 |
\P\P\xfd\x80\x80
|
113 |
\P\P\xfd\x80\x80\x80
|
114 |
\P\P\xfd\x80\x80\x80\x80
|
115 |
|
116 |
/anything/8
|
117 |
\xc0\x80
|
118 |
\xc1\x8f
|
119 |
\xe0\x9f\x80
|
120 |
\xf0\x8f\x80\x80
|
121 |
\xf8\x87\x80\x80\x80
|
122 |
\xfc\x83\x80\x80\x80\x80
|
123 |
\xfe\x80\x80\x80\x80\x80
|
124 |
\xff\x80\x80\x80\x80\x80
|
125 |
\xc3\x8f
|
126 |
\xe0\xaf\x80
|
127 |
\xe1\x80\x80
|
128 |
\xf0\x9f\x80\x80
|
129 |
\xf1\x8f\x80\x80
|
130 |
\xf8\x88\x80\x80\x80
|
131 |
\xf9\x87\x80\x80\x80
|
132 |
\xfc\x84\x80\x80\x80\x80
|
133 |
\xfd\x83\x80\x80\x80\x80
|
134 |
\?\xf8\x88\x80\x80\x80
|
135 |
\?\xf9\x87\x80\x80\x80
|
136 |
\?\xfc\x84\x80\x80\x80\x80
|
137 |
\?\xfd\x83\x80\x80\x80\x80
|
138 |
|
139 |
/noncharacter/8
|
140 |
\x{fffe}
|
141 |
\x{ffff}
|
142 |
\x{1fffe}
|
143 |
\x{1ffff}
|
144 |
\x{2fffe}
|
145 |
\x{2ffff}
|
146 |
\x{3fffe}
|
147 |
\x{3ffff}
|
148 |
\x{4fffe}
|
149 |
\x{4ffff}
|
150 |
\x{5fffe}
|
151 |
\x{5ffff}
|
152 |
\x{6fffe}
|
153 |
\x{6ffff}
|
154 |
\x{7fffe}
|
155 |
\x{7ffff}
|
156 |
\x{8fffe}
|
157 |
\x{8ffff}
|
158 |
\x{9fffe}
|
159 |
\x{9ffff}
|
160 |
\x{afffe}
|
161 |
\x{affff}
|
162 |
\x{bfffe}
|
163 |
\x{bffff}
|
164 |
\x{cfffe}
|
165 |
\x{cffff}
|
166 |
\x{dfffe}
|
167 |
\x{dffff}
|
168 |
\x{efffe}
|
169 |
\x{effff}
|
170 |
\x{ffffe}
|
171 |
\x{fffff}
|
172 |
\x{10fffe}
|
173 |
\x{10ffff}
|
174 |
\x{fdd0}
|
175 |
\x{fdd1}
|
176 |
\x{fdd2}
|
177 |
\x{fdd3}
|
178 |
\x{fdd4}
|
179 |
\x{fdd5}
|
180 |
\x{fdd6}
|
181 |
\x{fdd7}
|
182 |
\x{fdd8}
|
183 |
\x{fdd9}
|
184 |
\x{fdda}
|
185 |
\x{fddb}
|
186 |
\x{fddc}
|
187 |
\x{fddd}
|
188 |
\x{fdde}
|
189 |
\x{fddf}
|
190 |
\x{fde0}
|
191 |
\x{fde1}
|
192 |
\x{fde2}
|
193 |
\x{fde3}
|
194 |
\x{fde4}
|
195 |
\x{fde5}
|
196 |
\x{fde6}
|
197 |
\x{fde7}
|
198 |
\x{fde8}
|
199 |
\x{fde9}
|
200 |
\x{fdea}
|
201 |
\x{fdeb}
|
202 |
\x{fdec}
|
203 |
\x{fded}
|
204 |
\x{fdee}
|
205 |
\x{fdef}
|
206 |
|
207 |
/\x{100}/8DZ
|
208 |
|
209 |
/\x{1000}/8DZ
|
210 |
|
211 |
/\x{10000}/8DZ
|
212 |
|
213 |
/\x{100000}/8DZ
|
214 |
|
215 |
/\x{10ffff}/8DZ
|
216 |
|
217 |
/[\x{ff}]/8DZ
|
218 |
|
219 |
/[\x{100}]/8DZ
|
220 |
|
221 |
/\x80/8DZ
|
222 |
|
223 |
/\xff/8DZ
|
224 |
|
225 |
/\x{D55c}\x{ad6d}\x{C5B4}/DZ8
|
226 |
\x{D55c}\x{ad6d}\x{C5B4}
|
227 |
|
228 |
/\x{65e5}\x{672c}\x{8a9e}/DZ8
|
229 |
\x{65e5}\x{672c}\x{8a9e}
|
230 |
|
231 |
/\x{80}/DZ8
|
232 |
|
233 |
/\x{084}/DZ8
|
234 |
|
235 |
/\x{104}/DZ8
|
236 |
|
237 |
/\x{861}/DZ8
|
238 |
|
239 |
/\x{212ab}/DZ8
|
240 |
|
241 |
/-- This one is here not because it's different to Perl, but because the way
|
242 |
the captured single-byte is displayed. (In Perl it becomes a character, and you
|
243 |
can't tell the difference.) --/
|
244 |
|
245 |
/X(\C)(.*)/8
|
246 |
X\x{1234}
|
247 |
X\nabc
|
248 |
|
249 |
/-- This one is here because Perl gives out a grumbly error message (quite
|
250 |
correctly, but that messes up comparisons). --/
|
251 |
|
252 |
/a\Cb/8
|
253 |
*** Failers
|
254 |
a\x{100}b
|
255 |
|
256 |
/[^ab\xC0-\xF0]/8SDZ
|
257 |
\x{f1}
|
258 |
\x{bf}
|
259 |
\x{100}
|
260 |
\x{1000}
|
261 |
*** Failers
|
262 |
\x{c0}
|
263 |
\x{f0}
|
264 |
|
265 |
/Ā{3,4}/8SDZ
|
266 |
\x{100}\x{100}\x{100}\x{100\x{100}
|
267 |
|
268 |
/(\x{100}+|x)/8SDZ
|
269 |
|
270 |
/(\x{100}*a|x)/8SDZ
|
271 |
|
272 |
/(\x{100}{0,2}a|x)/8SDZ
|
273 |
|
274 |
/(\x{100}{1,2}a|x)/8SDZ
|
275 |
|
276 |
/\x{100}/8DZ
|
277 |
|
278 |
/a\x{100}\x{101}*/8DZ
|
279 |
|
280 |
/a\x{100}\x{101}+/8DZ
|
281 |
|
282 |
/[^\x{c4}]/DZ
|
283 |
|
284 |
/[\x{100}]/8DZ
|
285 |
\x{100}
|
286 |
Z\x{100}
|
287 |
\x{100}Z
|
288 |
*** Failers
|
289 |
|
290 |
/[\xff]/DZ8
|
291 |
>\x{ff}<
|
292 |
|
293 |
/[^\xff]/8DZ
|
294 |
|
295 |
/\x{100}abc(xyz(?1))/8DZ
|
296 |
|
297 |
/a\x{1234}b/P8
|
298 |
a\x{1234}b
|
299 |
|
300 |
/\777/8I
|
301 |
\x{1ff}
|
302 |
\777
|
303 |
|
304 |
/\x{100}+\x{200}/8DZ
|
305 |
|
306 |
/\x{100}+X/8DZ
|
307 |
|
308 |
/^[\QĀ\E-\QŐ\E/BZ8
|
309 |
|
310 |
/-- This tests the stricter UTF-8 check according to RFC 3629. --/
|
311 |
|
312 |
/X/8
|
313 |
\x{0}\x{d7ff}\x{e000}\x{10ffff}
|
314 |
\x{d800}
|
315 |
\x{d800}\?
|
316 |
\x{da00}
|
317 |
\x{da00}\?
|
318 |
\x{dfff}
|
319 |
\x{dfff}\?
|
320 |
\x{110000}
|
321 |
\x{110000}\?
|
322 |
\x{2000000}
|
323 |
\x{2000000}\?
|
324 |
\x{7fffffff}
|
325 |
\x{7fffffff}\?
|
326 |
|
327 |
/(*UTF8)\x{1234}/
|
328 |
abcd\x{1234}pqr
|
329 |
|
330 |
/(*CRLF)(*UTF8)(*BSR_UNICODE)a\Rb/I
|
331 |
|
332 |
/\h/SI8
|
333 |
ABC\x{09}
|
334 |
ABC\x{20}
|
335 |
ABC\x{a0}
|
336 |
ABC\x{1680}
|
337 |
ABC\x{180e}
|
338 |
ABC\x{2000}
|
339 |
ABC\x{202f}
|
340 |
ABC\x{205f}
|
341 |
ABC\x{3000}
|
342 |
|
343 |
/\v/SI8
|
344 |
ABC\x{0a}
|
345 |
ABC\x{0b}
|
346 |
ABC\x{0c}
|
347 |
ABC\x{0d}
|
348 |
ABC\x{85}
|
349 |
ABC\x{2028}
|
350 |
|
351 |
/\h*A/SI8
|
352 |
CDBABC
|
353 |
|
354 |
/\v+A/SI8
|
355 |
|
356 |
/\s?xxx\s/8SI
|
357 |
|
358 |
/\sxxx\s/I8ST1
|
359 |
AB\x{85}xxx\x{a0}XYZ
|
360 |
AB\x{a0}xxx\x{85}XYZ
|
361 |
|
362 |
/\S \S/I8ST1
|
363 |
\x{a2} \x{84}
|
364 |
A Z
|
365 |
|
366 |
/a+/8
|
367 |
a\x{123}aa\>1
|
368 |
a\x{123}aa\>2
|
369 |
a\x{123}aa\>3
|
370 |
a\x{123}aa\>4
|
371 |
a\x{123}aa\>5
|
372 |
a\x{123}aa\>6
|
373 |
|
374 |
/\x{1234}+/iS8I
|
375 |
|
376 |
/\x{1234}+?/iS8I
|
377 |
|
378 |
/\x{1234}++/iS8I
|
379 |
|
380 |
/\x{1234}{2}/iS8I
|
381 |
|
382 |
/[^\x{c4}]/8DZ
|
383 |
|
384 |
/X+\x{200}/8DZ
|
385 |
|
386 |
/\R/SI8
|
387 |
|
388 |
/\777/8DZ
|
389 |
|
390 |
/\w+\x{C4}/8BZ
|
391 |
a\x{C4}\x{C4}
|
392 |
|
393 |
/\w+\x{C4}/8BZT1
|
394 |
a\x{C4}\x{C4}
|
395 |
|
396 |
/\W+\x{C4}/8BZ
|
397 |
!\x{C4}
|
398 |
|
399 |
/\W+\x{C4}/8BZT1
|
400 |
!\x{C4}
|
401 |
|
402 |
/\W+\x{A1}/8BZ
|
403 |
!\x{A1}
|
404 |
|
405 |
/\W+\x{A1}/8BZT1
|
406 |
!\x{A1}
|
407 |
|
408 |
/X\s+\x{A0}/8BZ
|
409 |
X\x20\x{A0}\x{A0}
|
410 |
|
411 |
/X\s+\x{A0}/8BZT1
|
412 |
X\x20\x{A0}\x{A0}
|
413 |
|
414 |
/\S+\x{A0}/8BZ
|
415 |
X\x{A0}\x{A0}
|
416 |
|
417 |
/\S+\x{A0}/8BZT1
|
418 |
X\x{A0}\x{A0}
|
419 |
|
420 |
/\x{a0}+\s!/8BZ
|
421 |
\x{a0}\x20!
|
422 |
|
423 |
/\x{a0}+\s!/8BZT1
|
424 |
\x{a0}\x20!
|
425 |
|
426 |
/A/8
|
427 |
\x{ff000041}
|
428 |
\x{7f000041}
|
429 |
|
430 |
/-- End of testinput15 --/
|