oxedyne/fe2o3 — fe2o3_text/tests/regex

oxedyne/fe2o3/fe2o3_text/tests/regex_oracle/corpus.txt

6.8 KiB, 1 run

created by r1870400018:60336, which is this file's identity for as long as the history lasts, whatever it is later renamed to

download · who wrote it · its history

1	# The regular expression corpus that oracle.pl runs through Perl to write expected.txt.
2	#
3	# P a pattern, in the syntax of the Rust `regex` crate
4	# Q the same pattern in Perl's syntax, where the two differ (`\p{Greek}` is Script in Rust
5	# and Script_Extensions in Perl; `\u{..}`, `\<` and class set operations are Rust's only)
6	# R a replacement template, expanded as the `regex` crate expands one
7	# H a haystack; `\n`, `\t` and `\x{..}` are decoded, so a haystack cannot hold a backslash
8	#
9	# Every H is run against the P above it. A group inside a repetition keeps its capture from an
10	# earlier iteration in the `regex` crate (and in Python) but is reset in Perl, so no such case is
11	# here; tests/regex.rs checks those against Python's answers. Nor is a repetition whose later
12	# iteration can match nothing, `(a*)+`, where Perl and Python report the empty last iteration and
13	# the crate does not; rust_suite.txt carries the crate's own cases for that. Perl's `$` also matches before a final newline and
14	# Rust's does not, so no haystack under a `$` pattern ends in one; Perl's case-insensitivity
15	# folds `ß` to `ss` and Rust's does not, so no case-insensitive pattern meets an `ß`.
16
17	# Literals and dot
18	P abc
19	H xxabcxxabc
20	H ab
21	P a.c
22	H abc a\nc a\x{3bb}c
23	P (?s)a.c
24	H a\nc abc
25
26	# Classes
27	P [a-c]+
28	H zzabcbazz ab c
29	P [^a-c\s]+
30	H ab xyz\tcab
31	P [\d.]+x
32	H 3.14x and 1x
33	P [\]\-]+
34	H a]-]b
35	P [a-z&&[^aeiou]]+
36	Q (?:(?![aeiou])[a-z])+
37	H rhythm and blues
38	P [a-z--[aeiou]]+
39	Q (?:(?![aeiou])[a-z])+
40	H strengths
41	P [a-f~~c-h]+
42	Q [abgh]+
43	H abcdefghij
44	P [[:alpha:]]+
45	Q [A-Za-z]+
46	H abc\x{3bb}\x{3bc}DEF12
47	P [[:^digit:][:space:]]+
48	Q [^0-9]+
49	H 12ab 34
50	P [[:punct:]]+
51	Q [!-\/:-\@\[-`{-~]+
52	H a,b;c!?d
53
54	# Unicode shorthands
55	P \d+
56	H x\x{661}\x{662}\x{663}y 42 \x{ff11}\x{ff12} \x{966}\x{967}
57	P \w+
58	H caf\x{e9} na\x{ef}ve \x{3bb}\x{3cc}\x{3b3}\x{3bf}\x{3c2} e\x{301}t\x{e9} a_b \x{4e2d}\x{6587}
59	P \s+
60	H a\x{a0}b\x{3000}c\x{2028}d\x{85}e\x{1680}f
61	P \S+
62	H a\x{a0}b\x{3000}c
63	P \W+
64	H ab, \x{3bb}; \x{200d}!
65	P \D+
66	H \x{661}a\x{662}
67
68	# Boundaries
69	P \bcat\b
70	H cat concat cat. (cat)
71	P \Bcat\B
72	H concatenate cat
73	P \b\w+\b
74	H \x{3bb}\x{3cc}\x{3b3}\x{3bf}\x{3c2} \x{3ba}\x{3b1}\x{3b9}! caf\x{e9}
75	P \b
76	H ab cd
77	P \B
78	H ab c
79	P \<\w+
80	Q (?<!\w)(?=\w)\w+
81	H ab cd, ef
82	P \w+\>
83	Q \w+(?<=\w)(?!\w)
84	H ab cd, ef
85	P \b{start}.
86	Q (?<!\w)(?=\w).
87	H x y,z
88	P .\b{end}
89	Q .(?<=\w)(?!\w)
90	H x y,z
91	P \b{start-half}.
92	Q (?<!\w).
93	H ab c
94	P .\b{end-half}
95	Q .(?!\w)
96	H ab c
97
98	# Anchors and multi-line
99	P ^\w+
100	H ab cd
101	H \x{3bb}x
102	P \w+$
103	H ab cd
104	P (?m)^\w+
105	H ab\ncd\nef
106	P (?m)\w+$
107	H ab\ncd\nef
108	P \A\w
109	H ab
110	P \w\z
111	H ab
112	P (?m)^$
113	H a\n\nb
114
115	# Quantifiers
116	P a{2,3}
117	H aaaaaaa
118	P a{2,}?
119	H aaaaa
120	P a{3}
121	H aaaaaaa
122	P a+?b
123	H aaab
124	P <.*?>
125	H <a><bc>
126	P <.*>
127	H <a><bc>
128	P x*
129	H axxb
130	P (ab)*c
131	H abababc c abc
132	P a?
133	H baa
134
135	# Alternation
136	P a\|ab
137	H ab
138	P ab\|a
139	H ab
140	P cat\|dog\|bird
141	H a dog and a bird and a cat
142	P (?:a\|b)+c
143	H abbac
144
145	# Capture groups
146	P (\w+)@(\w+)\.com
147	H mail joe@site.com and ann@web.com
148	P (a)\|(b)
149	H ab
150	P (a\|(b))+
151	H ba
152	P (?P<year>\d{4})-(?P<month>\d{2})
153	R $month/$year
154	H 2026-09 and 1999-12
155	P (?<first>\w)(?<rest>\w*)
156	R ${rest}${first}ay
157	H pig latin
158	P (\d)(\d)?
159	R [$2$1]
160	H 1 23 456
161	P (a)
162	H b
163	P ()
164	H ab
165	P (x)?y
166	H y xy
167	P (\p{Lu})(\p{Ll}+)
168	R $2$1
169	H Alpha Beta gamma \x{394}\x{3ad}\x{3bb}\x{3c4}\x{3b1}
170
171	# Unicode properties
172	P \p{L}+
173	H ab12\x{3bb}\x{3bc}-\x{4e2d}\x{6587} \x{627}\x{644}
174	P \pL+
175	H ab12\x{3bb}
176	P \PL+
177	H ab12\x{3bb}--c
178	P \p{Lu}
179	H aBc\x{394}\x{3b4}
180	P \p{Ll}+
181	H aBc\x{394}\x{3b4}
182	P \p{Uppercase_Letter}+
183	H aBC\x{394}\x{3b4}
184	P \p{gc=Nd}+
185	H ab\x{661}\x{662}12
186	P \p{General_Category=Decimal_Number}+
187	H ab\x{661}\x{662}12
188	P \p{Letter}+
189	H x1y
190	P \p{N}+
191	H x\x{bd}\x{2163}9y
192	P \p{Nl}
193	H x\x{2163}9y
194	P \p{No}
195	H x\x{bd}9y
196	P \p{P}+
197	H a,.!?b\x{201c}c\x{201d}
198	P \p{Ps}\p{Pe}
199	H ()[]{}
200	P \p{S}+
201	H a+\x{20ac}$\x{a9}b
202	P \p{Sc}
203	H a$\x{20ac}\x{a3}b
204	P \p{Z}+
205	H a \x{a0}\x{2028}b
206	P \p{Zs}+
207	H a \x{a0}\x{2028}b
208	P \p{Cc}+
209	H a\tb\x{7f}c
210	P \p{Cf}
211	H a\x{200b}\x{ad}b
212	P \p{M}+
213	H e\x{301}\x{302}a\x{93f}
214	P \p{Mn}+
215	H e\x{301}\x{302}a\x{93f}
216	P \p{Mc}
217	H e\x{301}a\x{93f}
218	P \p{LC}+
219	H a\x{2c6}B\x{1c5}
220	P \p{gc!=L}+
221	Q \P{L}+
222	H ab12cd
223	P \p{Greek}+
224	Q \p{sc=Greek}+
225	H abc\x{3bb}\x{3cc}\x{3b3}\x{3bf}\x{3c2}\x{b7}x
226	P \p{sc=Grek}+
227	H \x{3bb}\x{3cc}\x{342}\x{3b3}
228	P \p{scx=Grek}+
229	H \x{3bb}\x{3cc}\x{342}\x{3b3}
230	P \p{Script_Extensions=Greek}+
231	Q \p{scx=Grek}+
232	H \x{3bb}\x{3cc}\x{342}\x{3b3}
233	P \p{Latin}+
234	Q \p{sc=Latin}+
235	H caf\x{e9} \x{3bb}
236	P \p{Han}+
237	Q \p{sc=Han}+
238	H \x{4e2d}\x{6587}\x{3001}x
239	P \p{scx=Hani}+
240	H \x{4e2d}\x{6587}\x{3001}x
241	P \p{Cyrillic}+
242	Q \p{sc=Cyrillic}+
243	H \x{41f}\x{440}\x{438}\x{432}\x{435}\x{442} mir
244	P \p{Arabic}+
245	Q \p{sc=Arabic}+
246	H \x{627}\x{644}\x{633}\x{644}\x{627}\x{645} \x{660}
247	P \p{scx=Arab}+
248	H \x{627}\x{644}\x{60c}\x{660}
249	P \p{Common}+
250	Q \p{sc=Common}+
251	H ab 12,\x{3000}
252	P \p{Inherited}+
253	Q \p{sc=Inherited}+
254	H e\x{301}\x{302}x
255	P \p{Devanagari}+
256	Q \p{sc=Devanagari}+
257	H \x{928}\x{92e}\x{938}\x{94d}\x{924}\x{947} \x{966}
258	P \p{Hiragana}+
259	Q \p{sc=Hiragana}+
260	H \x{3072}\x{3089}\x{304c}\x{306a}\x{30ab}\x{30bf}
261	P \P{Greek}+
262	Q \P{sc=Greek}+
263	H ab\x{3bb}cd
264	P \p{sc!=Latin}+
265	Q \P{sc=Latin}+
266	H ab\x{3bb}\x{3bc}cd
267	P \p{Alphabetic}+
268	H x\x{2163}y1
269	P \p{Alpha}+
270	Q \p{Alphabetic}+
271	H x\x{2163}y1
272	P \p{White_Space}+
273	H a \x{85}\x{2029}b
274	P \p{WSpace}+
275	Q \p{White_Space}+
276	H a \x{85}b
277	P \p{Uppercase}+
278	H a\x{2160}B\x{24b6}c
279	P \p{Lowercase}+
280	H A\x{aa}b\x{2170}C
281	P \p{Emoji}+
282	H a\x{1f600}\x{2764}#1
283	P \p{Extended_Pictographic}+
284	H a\x{1f600}\x{2764}\x{a9}#1
285	P \p{Emoji_Presentation}+
286	H a\x{1f600}\x{2764}#
287	P \p{Ideographic}+
288	H \x{4e2d}\x{3007}a
289	P \p{Math}+
290	H a+\x{2211}=b
291	P \p{Dash}+
292	H a-\x{2014}b
293	P \p{Hex_Digit}+
294	H xyzAF09\x{ff21}g
295	P \p{Any}+
296	H a\x{3bb}\n\x{1f600}
297	P \p{ASCII}+
298	H ab\x{3bb}cd
299	P \p{Assigned}+
300	H ab\x{378}cd
301	P \p{Cn}+
302	H ab\x{378}\x{379}cd
303	P \p{is Greek}+
304	Q \p{sc=Greek}+
305	H ab\x{3bb}
306	P \p{ letter }+
307	Q \p{L}+
308	H 1ab2
309	P \p{lowercase-letter}+
310	Q \p{Ll}+
311	H ABcdEF
312	P \p{gc=lowercaseletter}+
313	Q \p{Ll}+
314	H ABcdEF
315	P [\p{Greek}\d]+
316	Q [\p{sc=Greek}\d]+
317	H ab\x{3bb}12\x{3bc}c
318	P [^\p{L}\s]+
319	H ab, 12 \x{3bb};
320	P [\p{L}&&\p{Greek}]+
321	Q (?:(?=\p{sc=Greek})\p{L})+
322	H a\x{3bb}\x{375}\x{3bc}b
323	P [\p{L}--\p{Latin}]+
324	Q (?:(?!\p{sc=Latin})\p{L})+
325	H ab\x{3bb}\x{3bc}\x{4e2d}c
326	P [\P{L}a]+
327	H xa1b2
328	P [\w--\d]+
329	Q (?:(?!\d)\w)+
330	H ab12c_d3
331
332	# Flags
333	P (?i)hello
334	H say HeLLo there
335	P (?i)[a-c]+
336	H xABCy
337	P (?i)[^a]
338	H aAb
339	P (?i)\x{3a3}\x{39f}\x{3a6}
340	H \x{3c3}\x{3bf}\x{3c6} \x{3c2}\x{3bf}\x{3c6}
341	P (?i)k
342	H K\x{212a}k
343	P (?i)s
344	H \x{17f}S
345	P (?i)\p{Lu}+
346	H aBc
347	P (?i)\P{Lu}+
348	Q (?i)[^\p{Lu}\p{Ll}\p{Lt}]+
349	H aB1c;
350	P a(?i:b)c
351	H abc aBc abC
352	P (?i)a(?-i)b
353	H Ab AB ab
354	P (?i:a\|b)c
355	H Ac Bc aC
356	P (?x) a \s+ b # comment
357	Q (?x) a \s+ b
358	H a b ab
359	P (?x) \d{2} - \d{2}
360	H 12-34
361	P (?s:.)+
362	H a\nb
363	P (?m:^b)
364	H a\nb
365
366	# Escapes
367	P \x41\x{3bb}\u{3bc}
368	Q \x41\x{3bb}\x{3bc}
369	H xA\x{3bb}\x{3bc}y
370	P A\U0001F600
371	Q \x{41}\x{1f600}
372	H A\x{1f600}
373	P \t\x{0A}
374	H a\t\nb
375	P \.\+\*\?\\|\[\]\{\}\^\$\#\&\-\~
376	Q \.\+\*\?\\|\[\]\{\}\^\$\#\&\-\~
377	H .+*?()\|[]{}^$#&-~
378
379	# Multibyte haystacks
380	P \x{1f600}+
381	H a\x{1f600}\x{1f600}b\x{1f600}
382	P \w+
383	H \x{928}\x{92e}\x{938}\x{94d}\x{924}\x{947} \x{1f600}x
384	P .
385	H \x{1f600}\x{e9}a
386	P .\x{301}
387	H e\x{301}