/[pcre]/code/trunk/testdata/testinput17
ViewVC logotype

Contents of /code/trunk/testdata/testinput17

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1056 - (show annotations)
Tue Oct 16 15:53:41 2012 UTC (7 years, 1 month ago) by chpe
File size: 12679 byte(s)
pcre32: compile: Fix \H and \V character ranges for pcre32

Go up to 0xffffffff, and move the tests to the split 16- and 32-bit
tests because the output differs.

TODO: these character ranges look rather odd for non-UTF mode... bug?
1 /-- This set of tests is for the 16- and 32-bit library's basic (non-UTF-16 or -32) features
2 that are not compatible with the 8-bit library, or which give different
3 output in 16- or 32-bit mode. --/
4
5 /a\Cb/
6 aXb
7 a\nb
8
9 /[^\x{c4}]/DZ
10
11
12 /\x{100}/I
13
14 / (?: [\040\t] | \(
15 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
16 \) )* # optional leading comment
17 (?: (?:
18 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
19 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
20 |
21 " (?: # opening quote...
22 [^\\\x80-\xff\n\015"] # Anything except backslash and quote
23 | # or
24 \\ [^\x80-\xff] # Escaped something (something != CR)
25 )* " # closing quote
26 ) # initial word
27 (?: (?: [\040\t] | \(
28 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
29 \) )* \. (?: [\040\t] | \(
30 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
31 \) )* (?:
32 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
33 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
34 |
35 " (?: # opening quote...
36 [^\\\x80-\xff\n\015"] # Anything except backslash and quote
37 | # or
38 \\ [^\x80-\xff] # Escaped something (something != CR)
39 )* " # closing quote
40 ) )* # further okay, if led by a period
41 (?: [\040\t] | \(
42 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
43 \) )* @ (?: [\040\t] | \(
44 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
45 \) )* (?:
46 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
47 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
48 | \[ # [
49 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
50 \] # ]
51 ) # initial subdomain
52 (?: #
53 (?: [\040\t] | \(
54 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
55 \) )* \. # if led by a period...
56 (?: [\040\t] | \(
57 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
58 \) )* (?:
59 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
60 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
61 | \[ # [
62 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
63 \] # ]
64 ) # ...further okay
65 )*
66 # address
67 | # or
68 (?:
69 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
70 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
71 |
72 " (?: # opening quote...
73 [^\\\x80-\xff\n\015"] # Anything except backslash and quote
74 | # or
75 \\ [^\x80-\xff] # Escaped something (something != CR)
76 )* " # closing quote
77 ) # one word, optionally followed by....
78 (?:
79 [^()<>@,;:".\\\[\]\x80-\xff\000-\010\012-\037] | # atom and space parts, or...
80 \(
81 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
82 \) | # comments, or...
83
84 " (?: # opening quote...
85 [^\\\x80-\xff\n\015"] # Anything except backslash and quote
86 | # or
87 \\ [^\x80-\xff] # Escaped something (something != CR)
88 )* " # closing quote
89 # quoted strings
90 )*
91 < (?: [\040\t] | \(
92 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
93 \) )* # leading <
94 (?: @ (?: [\040\t] | \(
95 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
96 \) )* (?:
97 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
98 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
99 | \[ # [
100 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
101 \] # ]
102 ) # initial subdomain
103 (?: #
104 (?: [\040\t] | \(
105 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
106 \) )* \. # if led by a period...
107 (?: [\040\t] | \(
108 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
109 \) )* (?:
110 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
111 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
112 | \[ # [
113 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
114 \] # ]
115 ) # ...further okay
116 )*
117
118 (?: (?: [\040\t] | \(
119 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
120 \) )* , (?: [\040\t] | \(
121 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
122 \) )* @ (?: [\040\t] | \(
123 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
124 \) )* (?:
125 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
126 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
127 | \[ # [
128 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
129 \] # ]
130 ) # initial subdomain
131 (?: #
132 (?: [\040\t] | \(
133 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
134 \) )* \. # if led by a period...
135 (?: [\040\t] | \(
136 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
137 \) )* (?:
138 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
139 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
140 | \[ # [
141 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
142 \] # ]
143 ) # ...further okay
144 )*
145 )* # further okay, if led by comma
146 : # closing colon
147 (?: [\040\t] | \(
148 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
149 \) )* )? # optional route
150 (?:
151 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
152 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
153 |
154 " (?: # opening quote...
155 [^\\\x80-\xff\n\015"] # Anything except backslash and quote
156 | # or
157 \\ [^\x80-\xff] # Escaped something (something != CR)
158 )* " # closing quote
159 ) # initial word
160 (?: (?: [\040\t] | \(
161 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
162 \) )* \. (?: [\040\t] | \(
163 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
164 \) )* (?:
165 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
166 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
167 |
168 " (?: # opening quote...
169 [^\\\x80-\xff\n\015"] # Anything except backslash and quote
170 | # or
171 \\ [^\x80-\xff] # Escaped something (something != CR)
172 )* " # closing quote
173 ) )* # further okay, if led by a period
174 (?: [\040\t] | \(
175 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
176 \) )* @ (?: [\040\t] | \(
177 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
178 \) )* (?:
179 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
180 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
181 | \[ # [
182 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
183 \] # ]
184 ) # initial subdomain
185 (?: #
186 (?: [\040\t] | \(
187 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
188 \) )* \. # if led by a period...
189 (?: [\040\t] | \(
190 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
191 \) )* (?:
192 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+ # some number of atom characters...
193 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
194 | \[ # [
195 (?: [^\\\x80-\xff\n\015\[\]] | \\ [^\x80-\xff] )* # stuff
196 \] # ]
197 ) # ...further okay
198 )*
199 # address spec
200 (?: [\040\t] | \(
201 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
202 \) )* > # trailing >
203 # name and address
204 ) (?: [\040\t] | \(
205 (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] | \( (?: [^\\\x80-\xff\n\015()] | \\ [^\x80-\xff] )* \) )*
206 \) )* # optional trailing comment
207 /xSI
208
209 /[\h]/BZ
210 >\x09<
211
212 /[\h]+/BZ
213 >\x09\x20\xa0<
214
215 /[\v]/BZ
216
217 /[^\h]/BZ
218
219 /\h+/SI
220 \x{1681}\x{200b}\x{1680}\x{2000}\x{202f}\x{3000}
221 \x{3001}\x{2fff}\x{200a}\xa0\x{2000}
222
223 /[\h\x{dc00}]+/BZSI
224 \x{1681}\x{200b}\x{1680}\x{2000}\x{202f}\x{3000}
225 \x{3001}\x{2fff}\x{200a}\xa0\x{2000}
226
227 /\H+/SI
228 \x{1680}\x{180e}\x{167f}\x{1681}\x{180d}\x{180f}
229 \x{2000}\x{200a}\x{1fff}\x{200b}
230 \x{202f}\x{205f}\x{202e}\x{2030}\x{205e}\x{2060}
231 \xa0\x{3000}\x9f\xa1\x{2fff}\x{3001}
232
233 /[\H\x{d800}]+/
234 \x{1680}\x{180e}\x{167f}\x{1681}\x{180d}\x{180f}
235 \x{2000}\x{200a}\x{1fff}\x{200b}
236 \x{202f}\x{205f}\x{202e}\x{2030}\x{205e}\x{2060}
237 \xa0\x{3000}\x9f\xa1\x{2fff}\x{3001}
238
239 /\v+/SI
240 \x{2027}\x{2030}\x{2028}\x{2029}
241 \x09\x0e\x84\x86\x85\x0a\x0b\x0c\x0d
242
243 /[\v\x{dc00}]+/BZSI
244 \x{2027}\x{2030}\x{2028}\x{2029}
245 \x09\x0e\x84\x86\x85\x0a\x0b\x0c\x0d
246
247 /\V+/SI
248 \x{2028}\x{2029}\x{2027}\x{2030}
249 \x85\x0a\x0b\x0c\x0d\x09\x0e\x84\x86
250
251 /[\V\x{d800}]+/
252 \x{2028}\x{2029}\x{2027}\x{2030}
253 \x85\x0a\x0b\x0c\x0d\x09\x0e\x84\x86
254
255 /\R+/SI<bsr_unicode>
256 \x{2027}\x{2030}\x{2028}\x{2029}
257 \x09\x0e\x84\x86\x85\x0a\x0b\x0c\x0d
258
259 /\x{d800}\x{d7ff}\x{dc00}\x{dc00}\x{dcff}\x{dd00}/I
260 \x{d800}\x{d7ff}\x{dc00}\x{dc00}\x{dcff}\x{dd00}
261
262 /[^\x{80}][^\x{ff}][^\x{100}][^\x{1000}][^\x{ffff}]/BZ
263
264 /[^\x{80}][^\x{ff}][^\x{100}][^\x{1000}][^\x{ffff}]/BZi
265
266 /[^\x{100}]*[^\x{1000}]+[^\x{ffff}]??[^\x{8000}]{4,}[^\x{7fff}]{2,9}?[^\x{100}]{5,6}+/BZ
267
268 /[^\x{100}]*[^\x{1000}]+[^\x{ffff}]??[^\x{8000}]{4,}[^\x{7fff}]{2,9}?[^\x{100}]{5,6}+/BZi
269
270 /(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF)XX/K
271 XX
272
273 /(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDE)XX/K
274 XX
275
276 /\u0100/<JS>BZ
277
278 /[\u0100-\u0200]/<JS>BZ
279
280 /\ud800/<JS>BZ
281
282 /-- End of testinput17 --/

  ViewVC Help
Powered by ViewVC 1.1.5