| blob | 601cef8d72d18fc8f6050ddda8471ac579d6e795 |
1 # Future self: the delimiter is an ASCII vertical bar, which is also a
2 # REGEX special character, but hadn't already been used. Nearly every
3 # other plain ASCII character had been used by a test. Characters
4 # outside the plain ASCII range have a risk of being mangled by modern
5 # editors. So, avoid using | in a test, or if needed, select a new
6 # delimiter.
7 # 2.8.2 Regular Expression General Requirement
8 2|4|bb*|abbbc|
9 2|2|bb*|ababbbc|
10 7|9|A#*::|A:A#:qA::qA#::qA##::q|
11 1|5|A#*::|A##::A#::qA::qA#:q|
12 # 2.8.3.1.2 BRE Special Characters
13 # GA108
14 2|2|\.|a.c|
15 2|2|\[|a[c|
16 2|2|\\|a\c|
17 2|2|\*|a*c|
18 2|2|\^|a^c|
19 2|2|\$|a$c|
20 7|11|X\*Y\*8|Y*8X*8X*Y*8|
21 # GA109
22 2|2|[.]|a.c|
23 2|2|[[]|a[c|
24 -1|-1|[[]|ac|
25 2|2|[\]|a\c|
26 1|1|[\a]|abc|
27 2|2|[\.]|a\.c|
28 2|2|[\.]|a.\c|
29 2|2|[*]|a*c|
30 2|2|[$]|a$c|
31 2|2|[X*Y8]|7*8YX|
32 # GA110
33 2|2|*|a*c|
34 3|4|*a|*b*a*c|
35 1|5|**9=|***9=9|
36 # GA111
37 1|1|^*|*bc|
38 -1|-1|^*|a*c|
39 -1|-1|^*|^*ab|
40 1|5|^**9=|***9=|
41 -1|-1|^*5<*9|5<9*5<*9|
42 # GA112
43 2|3|\(*b\)|a*b|
44 -1|-1|\(*b\)|ac|
45 1|6|A\(**9\)=|A***9=79|
46 # GA113(1)
47 1|3|\(^*ab\)|*ab|
48 -1|-1|\(^*ab\)|^*ab|
49 -1|-1|\(^*b\)|a*b|
50 -1|-1|\(^*b\)|^*b|
51 ### GA113(2) GNU regex implements GA113(1)
52 ##-1|-1|\(^*ab\)|*ab|
53 ##-1|-1|\(^*ab\)|^*ab|
54 ##1|1|\(^*b\)|b|
55 ##1|3|\(^*b\)|^^b|
56 # GA114
57 1|3|a^b|a^b|
58 1|3|a\^b|a^b|
59 1|1|^^|^bc|
60 2|2|\^|a^c|
61 1|1|[c^b]|^abc|
62 1|1|[\^ab]|^ab|
63 2|2|[\^ab]|c\d|
64 -1|-1|[^^]|^|
65 1|3|\(a^b\)|a^b|
66 1|3|\(a\^b\)|a^b|
67 2|2|\(\^\)|a^b|
68 # GA115
69 3|3|$$|ab$|
70 -1|-1|$$|$ab|
71 2|3|$c|a$c|
72 2|2|[$]|a$c|
73 1|2|\$a|$a|
74 3|3|\$$|ab$|
75 2|6|A\([34]$[34]\)B|XA4$3BY|
76 # 2.8.3.1.3 Periods in BREs
77 # GA116
78 1|1|.|abc|
79 -1|-1|.ab|abc|
80 1|3|ab.|abc|
81 1|3|a.b|a,b|
82 -1|-1|.......|PqRs6|
83 1|7|.......|PqRs6T8|
84 # 2.8.3.2 RE Bracket Expression
85 # GA118
86 2|2|[abc]|xbyz|
87 -1|-1|[abc]|xyz|
88 2|2|[abc]|xbay|
89 # GA119
90 2|2|[^a]|abc|
91 4|4|[^]cd]|cd]ef|
92 2|2|[^abc]|axyz|
93 -1|-1|[^abc]|abc|
94 3|3|[^[.a.]b]|abc|
95 3|3|[^[=a=]b]|abc|
96 2|2|[^-ac]|abcde-|
97 2|2|[^ac-]|abcde-|
98 3|3|[^a-b]|abcde|
99 3|3|[^a-bd-e]|dec|
100 2|2|[^---]|-ab|
101 16|16|[^a-zA-Z0-9]|pqrstVWXYZ23579#|
102 # GA120(1)
103 3|3|[]a]|cd]ef|
104 1|1|[]-a]|a_b|
105 3|3|[][.-.]-0]|ab0-]|
106 1|1|[]^a-z]|string|
107 # GA120(2)
108 4|4|[^]cd]|cd]ef|
109 0|0|[^]]*|]]]]]]]]X|
110 0|0|[^]]*|]]]]]]]]|
111 9|9|[^]]\{1,\}|]]]]]]]]X|
112 -1|-1|[^]]\{1,\}|]]]]]]]]|
113 # GA120(3)
114 3|3|[c[.].]d]|ab]cd|
115 2|8|[a-z]*[[.].]][A-Z]*|Abcd]DEFg|
116 # GA121
117 2|2|[[.a.]b]|Abc|
118 1|1|[[.a.]b]|aBc|
119 -1|-1|[[.a.]b]|ABc|
120 3|3|[^[.a.]b]|abc|
121 3|3|[][.-.]-0]|ab0-]|
122 3|3|[A-[.].]c]|ab]!|
123 # GA122
124 -2|-2|[[.ch.]]|abc|
125 -2|-2|[[.ab.][.CD.][.EF.]]|yZabCDEFQ9|
126 # GA125
127 2|2|[[=a=]b]|Abc|
128 1|1|[[=a=]b]|aBc|
129 -1|-1|[[=a=]b]|ABc|
130 3|3|[^[=a=]b]|abc|
131 # GA126
132 #W the expected result for [[:alnum:]]* is 2-7 which is wrong
133 0|0|[[:alnum:]]*| aB28gH|
134 2|7|[[:alnum:]][[:alnum:]]*| aB28gH|
135 #W the expected result for [^[:alnum:]]* is 2-5 which is wrong
136 0|0|[^[:alnum:]]*|2 ,\x7fa|
137 2|5|[^[:alnum:]][^[:alnum:]]*|2 ,\x7fa|
138 #W the expected result for [[:alpha:]]* is 2-5 which is wrong
139 0|0|[[:alpha:]]*| aBgH2|
140 2|5|[[:alpha:]][[:alpha:]]*| aBgH2|
141 1|6|[^[:alpha:]]*|2 8,\x7fa|
143 1|8|[^[:blank:]]*|aB28gH,\x7f |
144 1|2|[[:cntrl:]]*| \x7f |
145 1|8|[^[:cntrl:]]*|aB2 8gh,|
146 #W the expected result for [[:digit:]]* is 2-3 which is wrong
147 0|0|[[:digit:]]*|a28|
148 2|3|[[:digit:]][[:digit:]]*|a28|
149 1|8|[^[:digit:]]*|aB gH,\x7f|
150 1|7|[[:graph:]]*|aB28gH, |
151 1|3|[^[:graph:]]*| \x7f,|
152 1|2|[[:lower:]]*|agB|
153 1|8|[^[:lower:]]*|B2 8H,\x7fa|
154 1|8|[[:print:]]*|aB2 8gH, |
155 1|2|[^[:print:]]*| \x7f |
156 #W the expected result for [[:punct:]]* is 2-2 which is wrong
157 0|0|[[:punct:]]*|a,2|
158 2|3|[[:punct:]][[:punct:]]*|a,,2|
159 1|9|[^[:punct:]]*|aB2 8gH\x7f|
161 #W the expected result for [^[:space:]]* is 2-9 which is wrong
162 0|0|[^[:space:]]*| aB28gH,\x7f |
163 2|9|[^[:space:]][^[:space:]]*| aB28gH,\x7f |
164 #W the expected result for [[:upper:]]* is 2-3 which is wrong
165 0|0|[[:upper:]]*|aBH2|
166 2|3|[[:upper:]][[:upper:]]*|aBH2|
167 1|8|[^[:upper:]]*|a2 8g,\x7fB|
168 #W the expected result for [[:xdigit:]]* is 2-5 which is wrong
169 0|0|[[:xdigit:]]*|gaB28h|
170 2|5|[[:xdigit:]][[:xdigit:]]*|gaB28h|
171 #W the expected result for [^[:xdigit:]]* is 2-7 which is wrong
172 2|7|[^[:xdigit:]][^[:xdigit:]]*|a gH,\x7f2|
173 # GA127
174 -2|-2|[b-a]|abc|
175 1|1|[a-c]|bbccde|
176 2|2|[a-b]|-bc|
177 3|3|[a-z0-9]|AB0|
178 3|3|[^a-b]|abcde|
179 3|3|[^a-bd-e]|dec|
180 1|1|[]-a]|a_b|
181 2|2|[+--]|a,b|
182 2|2|[--/]|a.b|
183 2|2|[^---]|-ab|
184 3|3|[][.-.]-0]|ab0-]|
185 3|3|[A-[.].]c]|ab]!|
186 2|6|bc[d-w]xy|abchxyz|
187 # GA129
188 1|1|[a-cd-f]|dbccde|
189 -1|-1|[a-ce-f]|dBCCdE|
190 2|4|b[n-zA-M]Y|absY9Z|
191 2|4|b[n-zA-M]Y|abGY9Z|
192 # GA130
193 3|3|[-xy]|ac-|
194 2|4|c[-xy]D|ac-D+|
195 2|2|[--/]|a.b|
196 2|4|c[--/]D|ac.D+b|
197 2|2|[^-ac]|abcde-|
198 1|3|a[^-ac]c|abcde-|
199 3|3|[xy-]|zc-|
200 2|4|c[xy-]7|zc-786|
201 2|2|[^ac-]|abcde-|
202 2|4|a[^ac-]c|5abcde-|
203 2|2|[+--]|a,b|
204 2|4|a[+--]B|Xa,By|
205 2|2|[^---]|-ab|
206 4|6|X[^---]Y|X-YXaYXbY|
207 # 2.8.3.3 BREs Matching Multiple Characters
208 # GA131
209 3|4|cd|abcdeabcde|
210 1|2|ag*b|abcde|
211 -1|-1|[a-c][e-f]|abcdef|
212 3|4|[a-c][e-f]|acbedf|
213 4|8|abc*XYZ|890abXYZ#*|
214 4|9|abc*XYZ|890abcXYZ#*|
215 4|15|abc*XYZ|890abcccccccXYZ#*|
216 -1|-1|abc*XYZ|890abc*XYZ#*|
217 # GA132
218 2|4|\(*bc\)|a*bc|
219 1|2|\(ab\)|abcde|
220 1|10|\(a\(b\(c\(d\(e\(f\(g\)h\(i\(j\)\)\)\)\)\)\)\)|abcdefghijk|
221 3|8|43\(2\(6\)*0\)AB|654320ABCD|
222 3|9|43\(2\(7\)*0\)AB|6543270ABCD|
223 3|12|43\(2\(7\)*0\)AB|6543277770ABCD|
224 # GA133
225 1|10|\(a\(b\(c\(d\(e\(f\(g\)h\(i\(j\)\)\)\)\)\)\)\)|abcdefghijk|
226 -1|-1|\(a\(b\(c\(d\(e\(f\(g\)h\(i\(k\)\)\)\)\)\)\)\)|abcdefghijk|
227 # GA134
228 2|4|\(bb*\)|abbbc|
229 2|2|\(bb*\)|ababbbc|
230 1|6|a\(.*b\)|ababbbc|
231 1|2|a\(b*\)|ababbbc|
232 1|20|a\(.*b\)c|axcaxbbbcsxbbbbbbbbc|
233 # GA135
234 1|7|\(a\(b\(c\(d\(e\)\)\)\)\)\4|abcdededede|
235 #W POSIX does not really specify whether a\(b\)*c\1 matches acb.
236 #W back references are supposed to expand to the last match, but what
237 #W if there never was a match as in this case?
238 -1|-1|a\(b\)*c\1|acb|
239 1|11|\(a\(b\(c\(d\(e\(f\(g\)h\(i\(j\)\)\)\)\)\)\)\)\9|abcdefghijjk|
240 # GA136
241 #W These two tests have the same problem as the test in GA135. No match
242 #W of a subexpression, why should the back reference be usable?
243 #W 1 2 a\(b\)*c\1 acb
244 #W 4 7 a\(b\(c\(d\(f\)*\)\)\)\4|xYzabcdePQRST
245 -1|-1|a\(b\)*c\1|acb|
246 -1|-1|a\(b\(c\(d\(f\)*\)\)\)\4|xYzabcdePQRST|
247 # GA137
248 -2|-2|\(a\(b\)\)\3|foo|
249 -2|-2|\(a\(b\)\)\(a\(b\)\)\5|foo|
250 # GA138
251 1|2|ag*b|abcde|
252 1|10|a.*b|abababvbabc|
253 2|5|b*c|abbbcdeabbbbbbcde|
254 2|5|bbb*c|abbbcdeabbbbbbcde|
255 1|5|a\(b\)*c\1|abbcbbb|
256 -1|-1|a\(b\)*c\1|abbdbd|
257 0|0|\([a-c]*\)\1|abcacdef|
258 1|6|\([a-c]*\)\1|abcabcabcd|
259 1|2|a^*b|ab|
260 1|5|a^*b|a^^^b|
261 # GA139
262 1|2|a\{2\}|aaaa|
263 1|7|\([a-c]*\)\{0,\}|aabcaab|
264 1|2|\(a\)\1\{1,2\}|aabc|
265 1|3|\(a\)\1\{1,2\}|aaaabc|
266 #W the expression \(\(a\)\1\)\{1,2\} is ill-formed, using \2
267 1|4|\(\(a\)\2\)\{1,2\}|aaaabc|
268 # GA140
269 1|2|a\{2\}|aaaa|
270 -1|-1|a\{2\}|abcd|
271 0|0|a\{0\}|aaaa|
272 1|64|a\{64\}|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa|
273 # GA141
274 1|7|\([a-c]*\)\{0,\}|aabcaab|
275 #W the expected result for \([a-c]*\)\{2,\} is failure which isn't correct
276 1|3|\([a-c]*\)\{2,\}|abcdefg|
277 1|3|\([a-c]*\)\{1,\}|abcdefg|
278 -1|-1|a\{64,\}|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa|
279 # GA142
280 1|3|a\{2,3\}|aaaa|
281 -1|-1|a\{2,3\}|abcd|
282 0|0|\([a-c]*\)\{0,0\}|foo|
283 1|63|a\{1,63\}|aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa|
284 # 2.8.3.4 BRE Precedence
285 # GA143
286 #W There are numerous bugs in the original version.
287 2|19|\^\[[[.].]]\\(\\1\\)\*\\{1,2\\}\$|a^[]\(\1\)*\{1,2\}$b|
288 1|6|[[=*=]][[=\=]][[=]=]][[===]][[...]][[:punct:]]|*\]=.;|
289 1|6|[$\(*\)^]*|$\()*^|
290 1|1|[\1]|1|
291 1|1|[\{1,2\}]|{|
292 #W the expected result for \(*\)*\1* is 2-2 which isn't correct
293 0|0|\(*\)*\1*|a*b*11|
294 2|3|\(*\)*\1*b|a*b*11|
295 #W the expected result for \(a\(b\{1,2\}\)\{1,2\}\) is 1-5 which isn't correct
296 1|3|\(a\(b\{1,2\}\)\{1,2\}\)|abbab|
297 1|5|\(a\(b\{1,2\}\)\)\{1,2\}|abbab|
298 1|1|^\(^\(^a$\)$\)$|a|
299 1|2|\(a\)\1$|aa|
300 1|3|ab*|abb|
301 1|4|ab\{2,4\}|abbbc|
302 # 2.8.3.5 BRE Expression Anchoring
303 # GA144
304 1|1|^a|abc|
305 -1|-1|^b|abc|
306 -1|-1|^[a-zA-Z]|99Nine|
307 1|4|^[a-zA-Z]*|Nine99|
308 # GA145(1)
309 1|2|\(^a\)\1|aabc|
310 -1|-1|\(^a\)\1|^a^abc|
311 1|2|\(^^a\)|^a|
312 1|1|\(^^\)|^^|
313 1|3|\(^abc\)|abcdef|
314 -1|-1|\(^def\)|abcdef|
315 ### GA145(2) GNU regex implements GA145(1)
316 ##-1|-1|\(^a\)\1|aabc|
317 ##1|4|\(^a\)\1|^a^abc|
318 ##-1|-1|\(^^a\)|^a|
319 ##1|2|\(^^\)|^^|
320 # GA146
321 3|3|a$|cba|
322 -1|-1|a$|abc|
323 5|7|[a-z]*$|99ZZxyz|
324 #W the expected result for [a-z]*$ is failure which isn't correct
325 10|9|[a-z]*$|99ZZxyz99|
326 3|3|$$|ab$|
327 -1|-1|$$|$ab|
328 3|3|\$$|ab$|
329 # GA147(1)
330 -1|-1|\(a$\)\1|bcaa|
331 -1|-1|\(a$\)\1|ba$|
332 -1|-1|\(ab$\)|ab$|
333 1|2|\(ab$\)|ab|
334 4|6|\(def$\)|abcdef|
335 -1|-1|\(abc$\)|abcdef|
336 ### GA147(2) GNU regex implements GA147(1)
337 ##-1|-1|\(a$\)\1|bcaa|
338 ##2|5|\(a$\)\1|ba$a$|
339 ##-1|-1|\(ab$\)|ab|
340 ##1|3|\(ab$\)|ab$|
341 # GA148
342 0|0|^$||
343 1|3|^abc$|abc|
344 -1|-1|^xyz$|^xyz^|
345 -1|-1|^234$|^234$|
346 1|9|^[a-zA-Z0-9]*$|2aA3bB9zZ|
347 -1|-1|^[a-z0-9]*$|2aA3b#B9zZ|