chiark / gitweb /
pcre3 (2:8.35-7.1) unstable; urgency=medium
[pcre3.git] / testdata / testinput18
1 /-- This set of tests is for UTF-16 and UTF-32 support, and is relevant only to
2     the 16- and 32-bit libraries. --/
3     
4 < forbid W 
5
6 /ÃÃÃxxx/8?DZSS
7
8 /abc/8
9     Ã]
10
11 /X(\C{3})/8
12     X\x{11234}Y
13     X\x{11234}YZ
14
15 /X(\C{4})/8
16     X\x{11234}YZ
17     X\x{11234}YZW
18
19 /X\C*/8
20     XYZabcdce
21
22 /X\C*?/8
23     XYZabcde
24
25 /X\C{3,5}/8
26     Xabcdefg
27     X\x{11234}Y
28     X\x{11234}YZ
29     X\x{11234}\x{512}
30     X\x{11234}\x{512}YZ
31     X\x{11234}\x{512}\x{11234}Z
32
33 /X\C{3,5}?/8
34     Xabcdefg
35     X\x{11234}Y
36     X\x{11234}YZ
37     X\x{11234}\x{512}YZ
38     *** Failers
39     X\x{11234}
40
41 /a\Cb/8
42     aXb
43     a\nb
44
45 /a\C\Cb/8
46     a\x{12257}b
47     a\x{12257}\x{11234}b
48     ** Failers
49     a\x{100}b
50
51 /ab\Cde/8
52     abXde
53
54 /-- Check maximum character size --/
55
56 /\x{ffff}/8DZ
57
58 /\x{10000}/8DZ
59
60 /\x{100}/8DZ
61
62 /\x{1000}/8DZ
63
64 /\x{10000}/8DZ
65
66 /\x{100000}/8DZ
67
68 /\x{10ffff}/8DZ
69
70 /[\x{ff}]/8DZ
71
72 /[\x{100}]/8DZ
73
74 /\x80/8DZ
75
76 /\xff/8DZ
77
78 /\x{D55c}\x{ad6d}\x{C5B4}/DZ8
79     \x{D55c}\x{ad6d}\x{C5B4}
80
81 /\x{65e5}\x{672c}\x{8a9e}/DZ8
82     \x{65e5}\x{672c}\x{8a9e}
83
84 /\x{80}/DZ8
85
86 /\x{084}/DZ8
87
88 /\x{104}/DZ8
89
90 /\x{861}/DZ8
91
92 /\x{212ab}/DZ8
93
94 /-- This one is here not because it's different to Perl, but because the way
95 the captured single-byte is displayed. (In Perl it becomes a character, and you
96 can't tell the difference.) --/
97
98 /X(\C)(.*)/8
99     X\x{1234}
100     X\nabc
101
102 /-- This one is here because Perl gives out a grumbly error message (quite
103 correctly, but that messes up comparisons). --/
104
105 /a\Cb/8
106     *** Failers
107     a\x{100}b
108
109 /[^ab\xC0-\xF0]/8SDZ
110     \x{f1}
111     \x{bf}
112     \x{100}
113     \x{1000}
114     *** Failers
115     \x{c0}
116     \x{f0}
117
118 /Ā{3,4}/8SDZ
119   \x{100}\x{100}\x{100}\x{100\x{100}
120
121 /(\x{100}+|x)/8SDZ
122
123 /(\x{100}*a|x)/8SDZ
124
125 /(\x{100}{0,2}a|x)/8SDZ
126
127 /(\x{100}{1,2}a|x)/8SDZ
128
129 /\x{100}/8DZ
130
131 /a\x{100}\x{101}*/8DZ
132
133 /a\x{100}\x{101}+/8DZ
134
135 /[^\x{c4}]/DZ
136
137 /[\x{100}]/8DZ
138     \x{100}
139     Z\x{100}
140     \x{100}Z
141     *** Failers
142
143 /[\xff]/DZ8
144     >\x{ff}<
145
146 /[^\xff]/8DZ
147
148 /\x{100}abc(xyz(?1))/8DZ
149
150 /\777/8I
151   \x{1ff}
152   \777
153
154 /\x{100}+\x{200}/8DZ
155
156 /\x{100}+X/8DZ
157
158 /^[\QĀ\E-\QŐ\E/BZ8
159
160 /X/8
161     \x{d800}
162     \x{d800}\?
163     \x{da00}
164     \x{da00}\?
165     \x{dc00}
166     \x{dc00}\?
167     \x{de00}
168     \x{de00}\?
169     \x{dfff}
170     \x{dfff}\?
171     \x{110000}
172     \x{d800}\x{1234}
173
174 /(*UTF16)\x{11234}/
175   abcd\x{11234}pqr
176
177 /(*UTF)\x{11234}/I
178   abcd\x{11234}pqr
179
180 /(*UTF-32)\x{11234}/
181   abcd\x{11234}pqr
182
183 /(*CRLF)(*UTF16)(*BSR_UNICODE)a\Rb/I
184
185 /(*CRLF)(*UTF32)(*BSR_UNICODE)a\Rb/I
186
187 /\h/SI8
188     ABC\x{09}
189     ABC\x{20}
190     ABC\x{a0}
191     ABC\x{1680}
192     ABC\x{180e}
193     ABC\x{2000}
194     ABC\x{202f}
195     ABC\x{205f}
196     ABC\x{3000}
197
198 /\v/SI8
199     ABC\x{0a}
200     ABC\x{0b}
201     ABC\x{0c}
202     ABC\x{0d}
203     ABC\x{85}
204     ABC\x{2028}
205
206 /\h*A/SI8
207     CDBABC
208     \x{2000}ABC 
209
210 /\R*A/SI8<bsr_unicode>
211     CDBABC
212     \x{2028}A  
213
214 /\v+A/SI8
215
216 /\s?xxx\s/8SI
217
218 /\sxxx\s/I8ST1
219     AB\x{85}xxx\x{a0}XYZ
220     AB\x{a0}xxx\x{85}XYZ
221
222 /\S \S/I8ST1
223     \x{a2} \x{84}
224     A Z
225
226 /a+/8
227     a\x{123}aa\>1
228     a\x{123}aa\>2
229     a\x{123}aa\>3
230     a\x{123}aa\>4
231     a\x{123}aa\>5
232     a\x{123}aa\>6
233
234 /\x{1234}+/iS8I
235
236 /\x{1234}+?/iS8I
237
238 /\x{1234}++/iS8I
239
240 /\x{1234}{2}/iS8I
241
242 /[^\x{c4}]/8DZ
243
244 /X+\x{200}/8DZ
245
246 /\R/SI8
247
248 /-- Check bad offset --/
249
250 /a/8
251     \x{10000}\>1
252     \x{10000}ab\>1
253     \x{10000}ab\>2
254     \x{10000}ab\>3
255     \x{10000}ab\>4
256     \x{10000}ab\>5
257
258 //8
259
260 /\w+\x{C4}/8BZ
261     a\x{C4}\x{C4}
262
263 /\w+\x{C4}/8BZT1
264     a\x{C4}\x{C4}
265     
266 /\W+\x{C4}/8BZ
267     !\x{C4}
268  
269 /\W+\x{C4}/8BZT1
270     !\x{C4}
271
272 /\W+\x{A1}/8BZ
273     !\x{A1}
274  
275 /\W+\x{A1}/8BZT1
276     !\x{A1}
277
278 /X\s+\x{A0}/8BZ
279     X\x20\x{A0}\x{A0}
280
281 /X\s+\x{A0}/8BZT1
282     X\x20\x{A0}\x{A0}
283
284 /\S+\x{A0}/8BZ
285     X\x{A0}\x{A0}
286
287 /\S+\x{A0}/8BZT1
288     X\x{A0}\x{A0}
289
290 /\x{a0}+\s!/8BZ
291     \x{a0}\x20!
292
293 /\x{a0}+\s!/8BZT1
294     \x{a0}\x20!
295
296 /(*UTF)abc/9
297
298 /abc/89
299
300 /-- End of testinput18 --/