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