/[pcre]/code/trunk/testdata/testinput15
ViewVC logotype

Contents of /code/trunk/testdata/testinput15

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1261 - (show annotations)
Wed Feb 27 16:27:01 2013 UTC (6 years, 5 months ago) by ph10
File size: 5889 byte(s)
Correct Unicode string checking in the light of corrigendum #9.
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 /./8
139 \x{fffe}
140 \x{ffff}
141 \x{1fffe}
142 \x{1ffff}
143 \x{2fffe}
144 \x{2ffff}
145 \x{3fffe}
146 \x{3ffff}
147 \x{4fffe}
148 \x{4ffff}
149 \x{5fffe}
150 \x{5ffff}
151 \x{6fffe}
152 \x{6ffff}
153 \x{7fffe}
154 \x{7ffff}
155 \x{8fffe}
156 \x{8ffff}
157 \x{9fffe}
158 \x{9ffff}
159 \x{afffe}
160 \x{affff}
161 \x{bfffe}
162 \x{bffff}
163 \x{cfffe}
164 \x{cffff}
165 \x{dfffe}
166 \x{dffff}
167 \x{efffe}
168 \x{effff}
169 \x{ffffe}
170 \x{fffff}
171 \x{10fffe}
172 \x{10ffff}
173 \x{fdd0}
174 \x{fdd1}
175 \x{fdd2}
176 \x{fdd3}
177 \x{fdd4}
178 \x{fdd5}
179 \x{fdd6}
180 \x{fdd7}
181 \x{fdd8}
182 \x{fdd9}
183 \x{fdda}
184 \x{fddb}
185 \x{fddc}
186 \x{fddd}
187 \x{fdde}
188 \x{fddf}
189 \x{fde0}
190 \x{fde1}
191 \x{fde2}
192 \x{fde3}
193 \x{fde4}
194 \x{fde5}
195 \x{fde6}
196 \x{fde7}
197 \x{fde8}
198 \x{fde9}
199 \x{fdea}
200 \x{fdeb}
201 \x{fdec}
202 \x{fded}
203 \x{fdee}
204 \x{fdef}
205
206 /\x{100}/8DZ
207
208 /\x{1000}/8DZ
209
210 /\x{10000}/8DZ
211
212 /\x{100000}/8DZ
213
214 /\x{10ffff}/8DZ
215
216 /[\x{ff}]/8DZ
217
218 /[\x{100}]/8DZ
219
220 /\x80/8DZ
221
222 /\xff/8DZ
223
224 /\x{D55c}\x{ad6d}\x{C5B4}/DZ8
225 \x{D55c}\x{ad6d}\x{C5B4}
226
227 /\x{65e5}\x{672c}\x{8a9e}/DZ8
228 \x{65e5}\x{672c}\x{8a9e}
229
230 /\x{80}/DZ8
231
232 /\x{084}/DZ8
233
234 /\x{104}/DZ8
235
236 /\x{861}/DZ8
237
238 /\x{212ab}/DZ8
239
240 /-- This one is here not because it's different to Perl, but because the way
241 the captured single-byte is displayed. (In Perl it becomes a character, and you
242 can't tell the difference.) --/
243
244 /X(\C)(.*)/8
245 X\x{1234}
246 X\nabc
247
248 /-- This one is here because Perl gives out a grumbly error message (quite
249 correctly, but that messes up comparisons). --/
250
251 /a\Cb/8
252 *** Failers
253 a\x{100}b
254
255 /[^ab\xC0-\xF0]/8SDZ
256 \x{f1}
257 \x{bf}
258 \x{100}
259 \x{1000}
260 *** Failers
261 \x{c0}
262 \x{f0}
263
264 /Ā{3,4}/8SDZ
265 \x{100}\x{100}\x{100}\x{100\x{100}
266
267 /(\x{100}+|x)/8SDZ
268
269 /(\x{100}*a|x)/8SDZ
270
271 /(\x{100}{0,2}a|x)/8SDZ
272
273 /(\x{100}{1,2}a|x)/8SDZ
274
275 /\x{100}/8DZ
276
277 /a\x{100}\x{101}*/8DZ
278
279 /a\x{100}\x{101}+/8DZ
280
281 /[^\x{c4}]/DZ
282
283 /[\x{100}]/8DZ
284 \x{100}
285 Z\x{100}
286 \x{100}Z
287 *** Failers
288
289 /[\xff]/DZ8
290 >\x{ff}<
291
292 /[^\xff]/8DZ
293
294 /\x{100}abc(xyz(?1))/8DZ
295
296 /a\x{1234}b/P8
297 a\x{1234}b
298
299 /\777/8I
300 \x{1ff}
301 \777
302
303 /\x{100}+\x{200}/8DZ
304
305 /\x{100}+X/8DZ
306
307 /^[\QĀ\E-\QŐ\E/BZ8
308
309 /-- This tests the stricter UTF-8 check according to RFC 3629. --/
310
311 /X/8
312 \x{d800}
313 \x{d800}\?
314 \x{da00}
315 \x{da00}\?
316 \x{dfff}
317 \x{dfff}\?
318 \x{110000}
319 \x{110000}\?
320 \x{2000000}
321 \x{2000000}\?
322 \x{7fffffff}
323 \x{7fffffff}\?
324
325 /(*UTF8)\x{1234}/
326 abcd\x{1234}pqr
327
328 /(*CRLF)(*UTF)(*BSR_UNICODE)a\Rb/I
329
330 /\h/SI8
331 ABC\x{09}
332 ABC\x{20}
333 ABC\x{a0}
334 ABC\x{1680}
335 ABC\x{180e}
336 ABC\x{2000}
337 ABC\x{202f}
338 ABC\x{205f}
339 ABC\x{3000}
340
341 /\v/SI8
342 ABC\x{0a}
343 ABC\x{0b}
344 ABC\x{0c}
345 ABC\x{0d}
346 ABC\x{85}
347 ABC\x{2028}
348
349 /\h*A/SI8
350 CDBABC
351
352 /\v+A/SI8
353
354 /\s?xxx\s/8SI
355
356 /\sxxx\s/I8ST1
357 AB\x{85}xxx\x{a0}XYZ
358 AB\x{a0}xxx\x{85}XYZ
359
360 /\S \S/I8ST1
361 \x{a2} \x{84}
362 A Z
363
364 /a+/8
365 a\x{123}aa\>1
366 a\x{123}aa\>2
367 a\x{123}aa\>3
368 a\x{123}aa\>4
369 a\x{123}aa\>5
370 a\x{123}aa\>6
371
372 /\x{1234}+/iS8I
373
374 /\x{1234}+?/iS8I
375
376 /\x{1234}++/iS8I
377
378 /\x{1234}{2}/iS8I
379
380 /[^\x{c4}]/8DZ
381
382 /X+\x{200}/8DZ
383
384 /\R/SI8
385
386 /\777/8DZ
387
388 /\w+\x{C4}/8BZ
389 a\x{C4}\x{C4}
390
391 /\w+\x{C4}/8BZT1
392 a\x{C4}\x{C4}
393
394 /\W+\x{C4}/8BZ
395 !\x{C4}
396
397 /\W+\x{C4}/8BZT1
398 !\x{C4}
399
400 /\W+\x{A1}/8BZ
401 !\x{A1}
402
403 /\W+\x{A1}/8BZT1
404 !\x{A1}
405
406 /X\s+\x{A0}/8BZ
407 X\x20\x{A0}\x{A0}
408
409 /X\s+\x{A0}/8BZT1
410 X\x20\x{A0}\x{A0}
411
412 /\S+\x{A0}/8BZ
413 X\x{A0}\x{A0}
414
415 /\S+\x{A0}/8BZT1
416 X\x{A0}\x{A0}
417
418 /\x{a0}+\s!/8BZ
419 \x{a0}\x20!
420
421 /\x{a0}+\s!/8BZT1
422 \x{a0}\x20!
423
424 /A/8
425 \x{ff000041}
426 \x{7f000041}
427
428 /-- End of testinput15 --/

  ViewVC Help
Powered by ViewVC 1.1.5