1	#!/usr/bin/env bash
2	#
3	# Only bash and zsh seem to implement [[ foo =~ '' ]]
4	#
5	# ^(a b)$ is a regex that should match 'a b' in a group.
6	#
7	# Not sure what bash is doing here... I think I have to just be empirical.
8	# Might need "compat" switch for parsing the regex. It should be an opaque
9	# string like zsh, not sure why it isn't.
10	#
11	# I think this is just papering over bugs...
12	# https://www.gnu.org/software/bash/manual/bash.html#Conditional-Constructs
13	#
14	# Storing the regular expression in a shell variable is often a useful way to
15	# avoid problems with quoting characters that are special to the shell. It is
16	# sometimes difficult to specify a regular expression literally without using
17	# quotes, or to keep track of the quoting used by regular expressions while
18	# paying attention to the shell’s quote removal. Using a shell variable to
19	# store the pattern decreases these problems. For example, the following is
20	# equivalent to the above:
21	#
22	# pattern='[[:space:]]*(a)?b'
23	# [[ $line =~ $pattern ]]
24	#
25	# If you want to match a character that’s special to the regular expression
26	# grammar, it has to be quoted to remove its special meaning. This means that in
27	# the pattern ‘xxx.txt’, the ‘.’ matches any character in the string (its usual
28	# regular expression meaning), but in the pattern ‘"xxx.txt"’ it can only match a
29	# literal ‘.’. Shell programmers should take special care with backslashes, since
30	# backslashes are used both by the shell and regular expressions to remove the
31	# special meaning from the following character. The following two sets of
32	# commands are not equivalent:
33	#
34	# From bash code: ( | ) are treated special. Normally they must be quoted, but
35	# they can be UNQUOTED in BASH_REGEX state. In fact they can't be quoted!
36
37	#### BASH_REMATCH
38	[[ foo123 =~ ([a-z]+)([0-9]+) ]]
39	argv.py "${BASH_REMATCH[@]}"
40	## STDOUT:
41	['foo123', 'foo', '123']
42	## END
43	## N-I zsh STDOUT:
44	['']
45	## END
46
47	#### Match is unanchored at both ends
48	[[ 'bar' =~ a ]] && echo true
49	## stdout: true
50
51	#### Failed match
52	[[ 'bar' =~ X ]] && echo true
53	## status: 1
54	## stdout-json: ""
55
56	#### Regex quoted with \ -- preferred in bash
57	[[ 'a b' =~ ^(a\ b)$ ]] && echo true
58	## stdout: true
59
60	#### Regex quoted with single quotes
61	# bash doesn't like the quotes
62	[[ 'a b' =~ '^(a b)$' ]] && echo true
63	## stdout-json: ""
64	## status: 1
65	## OK zsh stdout: true
66	## OK zsh status: 0
67
68	#### Regex quoted with double quotes
69	# bash doesn't like the quotes
70	[[ 'a b' =~ "^(a b)$" ]] && echo true
71	## stdout-json: ""
72	## status: 1
73	## OK zsh stdout: true
74	## OK zsh status: 0
75
76	#### Fix single quotes by storing in variable
77	pat='^(a b)$'
78	[[ 'a b' =~ $pat ]] && echo true
79	## stdout: true
80
81	#### Fix single quotes by storing in variable
82	pat="^(a b)$"
83	[[ 'a b' =~ $pat ]] && echo true
84	## stdout: true
85
86	#### Double quoting pat variable -- again bash doesn't like it.
87	pat="^(a b)$"
88	[[ 'a b' =~ "$pat" ]] && echo true
89	## stdout-json: ""
90	## status: 1
91	## OK zsh stdout: true
92	## OK zsh status: 0
93
94	#### Mixing quoted and unquoted parts
95	[[ 'a b' =~ 'a 'b ]] && echo true
96	[[ "a b" =~ "a "'b' ]] && echo true
97	## STDOUT:
98	true
99	true
100	## END
101
102	#### Regex with == and not =~ is parse error, different lexer mode required
103	# They both give a syntax error. This is lame.
104	[[ '^(a b)$' == ^(a\ b)$ ]] && echo true
105	## status: 2
106	## OK zsh status: 1
107
108	#### Omitting ( )
109	[[ '^a b$' == ^a\ b$ ]] && echo true
110	## stdout: true
111
112	#### Malformed regex
113	# Are they trying to PARSE the regex? Do they feed the buffer directly to
114	# regcomp()?
115	[[ 'a b' =~ ^)a\ b($ ]] && echo true
116	## status: 2
117	## OK zsh status: 1
118
119	#### Regex with char class containing space
120	# For some reason it doesn't work without parens?
121	[[ 'ba ba ' =~ ([a b]+) ]] && echo true
122	## stdout: true
123
124	#### Operators and space lose meaning inside ()
125	[[ '< >' =~ (< >) ]] && echo true
126	## stdout: true
127	## N-I zsh stdout-json: ""
128	## N-I zsh status: 1
129
130	#### Regex with |
131	[[ 'bar' =~ foo|bar ]] && echo true
132	## stdout: true
133	## N-I zsh stdout-json: ""
134	## N-I zsh status: 1
135
136	#### Regex to match literal brackets []
137
138	# bash-completion relies on this, so we're making it match bash.
139	# zsh understandably differs.
140	[[ '[]' =~ \[\] ]] && echo true
141
142	# Another way to write this.
143	pat='\[\]'
144	[[ '[]' =~ $pat ]] && echo true
145	## STDOUT:
146	true
147	true
148	## END
149	## OK zsh STDOUT:
150	true
151	## END
152
153	#### Regex to match literals . ^ $ etc.
154	[[ 'x' =~ \. ]] || echo false
155	[[ '.' =~ \. ]] && echo true
156
157	[[ 'xx' =~ \^\$ ]] || echo false
158	[[ '^$' =~ \^\$ ]] && echo true
159
160	[[ 'xxx' =~ \+\*\? ]] || echo false
161	[[ '*+?' =~ \*\+\? ]] && echo true
162
163	[[ 'xx' =~ \{\} ]] || echo false
164	[[ '{}' =~ \{\} ]] && echo true
165	## STDOUT:
166	false
167	true
168	false
169	true
170	false
171	true
172	false
173	true
174	## END
175	## BUG zsh STDOUT:
176	true
177	false
178	false
179	false
180	## END
181	## BUG zsh status: 1
182
183	#### Unquoted { is a regex parse error
184	[[ { =~ { ]] && echo true
185	echo status=$?
186	## stdout-json: ""
187	## status: 2
188	## BUG bash stdout-json: "status=2\n"
189	## BUG bash status: 0
190	## BUG zsh stdout-json: "status=1\n"
191	## BUG zsh status: 0
192
193	#### Fatal error inside [[ =~ ]]
194
195	# zsh and osh are stricter than bash. bash treats [[ like a command.
196
197	[[ a =~ $(( 1 / 0 )) ]]
198	echo status=$?
199	## stdout-json: ""
200	## status: 1
201	## BUG bash stdout: status=1
202	## BUG bash status: 0
203
204	#### Quoted { and +
205	[[ { =~ "{" ]] && echo 'yes {'
206	[[ + =~ "+" ]] && echo 'yes +'
207	[[ * =~ "*" ]] && echo 'yes *'
208	[[ ? =~ "?" ]] && echo 'yes ?'
209	[[ ^ =~ "^" ]] && echo 'yes ^'
210	[[ $ =~ "$" ]] && echo 'yes $'
211	[[ '(' =~ '(' ]] && echo 'yes ('
212	[[ ')' =~ ')' ]] && echo 'yes )'
213	[[ '|' =~ '|' ]] && echo 'yes |'
214	[[ '\' =~ '\' ]] && echo 'yes \'
215	echo ---
216
217	[[ . =~ "." ]] && echo 'yes .'
218	[[ z =~ "." ]] || echo 'no .'
219	echo ---
220
221	# This rule is weird but all shells agree. I would expect that the - gets
222	# escaped? It's an operator? but it behaves like a-z.
223	[[ a =~ ["a-z"] ]]; echo "a $?"
224	[[ - =~ ["a-z"] ]]; echo "- $?"
225	[[ b =~ ['a-z'] ]]; echo "b $?"
226	[[ z =~ ['a-z'] ]]; echo "z $?"
227
228	echo status=$?
229	## STDOUT:
230	yes {
231	yes +
232	yes *
233	yes ?
234	yes ^
235	yes $
236	yes (
237	yes )
238	yes |
239	yes \
240	---
241	yes .
242	no .
243	---
244	a 0
245	- 1
246	b 0
247	z 0
248	status=0
249	## END
250	## N-I zsh STDOUT:
251	yes ^
252	yes $
253	yes )
254	yes |
255	---
256	yes .
257	---
258	a 0
259	- 1
260	b 0
261	z 0
262	status=0
263	## END
264
265	#### Escaped {
266	# from bash-completion
267	[[ '$PA' =~ ^(\$\{?)([A-Za-z0-9_]*)$ ]] && argv.py "${BASH_REMATCH[@]}"
268	## STDOUT:
269	['$PA', '$', 'PA']
270	## END
271	## BUG zsh stdout-json: ""
272	## BUG zsh status: 1
273
274	#### Escaped { stored in variable first
275	# from bash-completion
276	pat='^(\$\{?)([A-Za-z0-9_]*)$'
277	[[ '$PA' =~ $pat ]] && argv.py "${BASH_REMATCH[@]}"
278	## STDOUT:
279	['$PA', '$', 'PA']
280	## END
281	## BUG zsh STDOUT:
282	['']
283	## END
284
285	#### regex with ?
286	[[ 'c' =~ c? ]] && echo true
287	[[ '' =~ c? ]] && echo true
288	## STDOUT:
289	true
290	true
291	## END
292
293	#### regex with unprintable characters
294	# can't have nul byte
295
296	# This pattern has literal characters
297	pat=$'^[\x01\x02]+$'
298
299	[[ $'\x01\x02\x01' =~ $pat ]]; echo status=$?
300	[[ $'a\x01' =~ $pat ]]; echo status=$?
301
302	# NOTE: There doesn't appear to be any way to escape these!
303	pat2='^[\x01\x02]+$'
304
305	## STDOUT:
306	status=0
307	status=1
308	## END