1	# Test out Oil's regular expression syntax.
2
3	#### /^.$/
4	shopt -s oil:all
5	var pat = ''
6
7	setvar pat = /^.$/
8	echo pat=$pat
9
10	setvar pat = /%start dot %end/
11	echo pat=$pat
12
13	if ('' ~ pat) { # ERE syntax
14	echo yes
15	} else {
16	echo no
17	}
18	# $pat is same as pat
19	if ('f' ~ pat) { # ERE syntax
20	echo yes
21	} else {
22	echo no
23	}
24
25	## STDOUT:
26	pat=^.$
27	pat=^.$
28	no
29	yes
30	## END
31
32
33	#### /.+/
34	shopt -s oil:all
35
36	var pat = /.+/
37	echo $pat
38
39	var s = 'foo'
40	if (s ~ pat) { # ERE syntax
41	echo yes
42	}
43	var empty = ''
44	if (empty ~ pat) { echo yes } else { echo no }
45	## STDOUT:
46	.+
47	yes
48	no
49	## END
50
51	#### Positional captures with _match
52	shopt -s oil:all
53
54	var x = 'zz 2020-08-20'
55
56	if [[ $x =~ ([[:digit:]]+)-([[:digit:]]+) ]] {
57	argv.py "${BASH_REMATCH[@]}"
58	}
59
60	# THIS IS A NO-OP. The variable is SHADOWED by the special name.
61	# I think that's OK.
62	setvar BASH_REMATCH = %(reset)
63
64	if (x ~ /<d+> '-' <d+>/) {
65	argv.py "${BASH_REMATCH[@]}"
66	argv.py $_match(0) $_match(1) $_match(2)
67
68	argv.py $_match() # synonym for _match(0)
69
70	# TODO: Also test _start() and _end()
71	}
72	## STDOUT:
73	['2020-08', '2020', '08']
74	['2020-08', '2020', '08']
75	['2020-08', '2020', '08']
76	['2020-08']
77	## END
78
79	#### s ~ regex and s !~ regex
80	shopt -s oil:basic
81
82	var s = 'foo'
83	if (s ~ '.([[:alpha:]]+)') { # ERE syntax
84	echo matches
85	argv.py $_match(0) $_match(1)
86	}
87	if (s !~ '[[:digit:]]+') {
88	echo "does not match"
89	argv.py $_match(0) $_match(1)
90	}
91
92	if (s ~ '[[:digit:]]+') {
93	echo "matches"
94	}
95	# Should be cleared now
96	# should this be Undef rather than ''?
97	var x = _match(0)
98	var y = _match(1)
99	if (x === null and y === null) {
100	echo 'cleared'
101	}
102
103	## STDOUT:
104	matches
105	['foo', 'oo']
106	does not match
107	['foo', 'oo']
108	cleared
109	## END
110
111	#### _start() and _end()
112	shopt -s oil:basic
113
114	var s = 'foo123bar'
115	if (s ~ /digit+/) {
116	echo start=$_start() end=$_end()
117	}
118
119	if (s ~ / word+ <digit+> /) {
120	echo start=$_start(1) end=$_end(1)
121	}
122	## STDOUT:
123	start=3 end=6
124	start=3 end=6
125	## END
126
127	#### Repeat {1,3} etc.
128	var pat = null
129
130	setvar pat = /d{2}/
131	echo $pat
132	setvar pat = /d{1,3}/
133	echo $pat
134	setvar pat = /d{1,}/
135	echo $pat
136	setvar pat = /d{,3}/
137	echo $pat
138
139
140	## STDOUT:
141	[[:digit:]]{2}
142	[[:digit:]]{1,3}
143	[[:digit:]]{1,}
144	[[:digit:]]{,3}
145	## END
146
147
148	#### d+ digit+ !d+ !digit+
149	shopt -s oil:all
150
151	var pat = ''
152
153	setvar pat = /d+/
154	echo $pat
155	if ('42' ~ pat) { echo yes }
156
157	var empty = ''
158	if (empty ~ pat) { echo yes } else { echo no }
159
160	setvar pat = /digit+/
161	echo $pat
162	setvar pat = /!d+/
163	echo $pat
164	setvar pat = /!digit+/
165	echo $pat
166
167
168	## STDOUT:
169	[[:digit:]]+
170	yes
171	no
172	[[:digit:]]+
173	[^[:digit:]]+
174	[^[:digit:]]+
175	## END
176
177	#### Alternation and sequence
178	var pat = ''
179	setvar pat = /s d+ | w*/
180	echo $pat
181	setvar pat = /s d+ or w*/
182	echo $pat
183	## STDOUT:
184	[[:space:]][[:digit:]]+|[[:alpha:][:digit:]_]*
185	[[:space:]][[:digit:]]+|[[:alpha:][:digit:]_]*
186	## END
187
188	#### Char Class Ranges
189	shopt -s oil:all
190
191	var pat = ''
192	setvar pat = /[0-9 a-f]+/
193	echo $pat
194	# This is equivalent
195	setvar pat = /['0' - '9' 'a' - 'f']+/
196	echo $pat
197
198	if ('0123' ~ pat) { echo yes } else { echo no }
199	if ('zzz' ~ pat) { echo yes } else { echo no }
200	if ('' ~ pat) { echo yes } else { echo no }
201	## STDOUT:
202	[0-9a-f]+
203	[0-9a-f]+
204	yes
205	no
206	no
207	## END
208
209	#### Char Class Set
210	shopt -s oil:all
211	var pat = ''
212
213	# This is NOT allowed
214	# setvar pat = /[a b c]+/
215
216	setvar pat = /['abc']+/
217	echo $pat
218
219	if ('cbcb' ~ pat) { echo yes } else { echo no }
220	if ('0123' ~ pat) { echo yes } else { echo no }
221	if ('' ~ pat) { echo yes } else { echo no }
222	## STDOUT:
223	[abc]+
224	yes
225	no
226	no
227	## END
228
229	#### Range with escaped characters
230	shopt -s oil:all
231
232	var pat = null
233
234	setvar pat = / [ \x00 - \x0f ] /
235	echo $pat | od -A n -t x1
236
237	## STDOUT:
238	5b 00 2d 0f 5d 0a
239	## END
240
241
242	#### Group ()
243	shopt -s oil:all
244	var pat = ''
245
246	setvar pat = /(%start s or d d)/
247	echo $pat
248
249	if (' foo' ~ pat) { echo yes } else { echo no }
250	if ('-00-' ~ pat) { echo yes } else { echo no }
251	if ('foo' ~ pat) { echo yes } else { echo no }
252
253	## STDOUT:
254	(^[[:space:]]|[[:digit:]][[:digit:]])
255	yes
256	yes
257	no
258	## END
259
260	#### Capture is acceptable as a group
261	shopt -s oil:all
262	var pat = /<%start s | d d>/
263	echo $pat
264	## STDOUT:
265	(^[[:space:]]|[[:digit:]][[:digit:]])
266	## END
267
268	#### Named captures with _match
269	shopt -s oil:all
270
271	var x = 'zz 2020-08-20'
272
273	if (x ~ /<d+ : year> '-' <d+ : month>/) {
274	argv.py $_match('year') $_match('month')
275	}
276	## STDOUT:
277	['2020', '08']
278	## END
279
280	#### Named Capture Decays Without Name
281	shopt -s oil:all
282	var pat = /<d+ : month>/
283	echo $pat
284
285	if ('123' ~ pat) {
286	echo yes
287	}
288
289	## STDOUT:
290	([[:digit:]]+)
291	yes
292	## END
293
294	#### Named Capture With ~ Assigns Variable
295	shopt -s oil:all
296	var pat = /<d+ : month>/
297	echo $pat
298
299	if ('123' ~ pat) {
300	echo yes
301	= month
302	}
303	## STDOUT:
304	([[:digit:]]+)
305	yes
306	TODO MONTH
307	## END
308
309	#### literal ''
310	shopt -s oil:all
311	var pat = ''
312
313	setvar pat = /'abc' 'def'/
314	echo $pat
315
316	#setvar pat = /'abc' '^ + * ?'/
317	#echo $pat
318
319	if ('abcde' ~ pat) { echo yes } else { echo no }
320	if ('abcdef' ~ pat) { echo yes } else { echo no }
321
322	## STDOUT:
323	abcdef
324	no
325	yes
326	## END
327
328	#### double quoted, $x, and ${x}
329	shopt -s oil:all
330	var pat = ''
331
332	var x = 'x'
333	var y = 'y'
334	setvar pat = / $x ${x} "abc" "$x${y}"/
335	echo $pat
336
337	if ('xxabcx' ~ pat) { echo yes } else { echo no }
338	if ('xxabcxyf' ~ pat) { echo yes } else { echo no }
339
340	## STDOUT:
341	xxabcxy
342	no
343	yes
344	## END
345
346	#### @splice
347	shopt -s oil:all
348	var d = /d+/;
349	var ip = / @d '.' @d '.' @d '.' @d /
350	echo $ip
351	if ('0.0.0.0' ~ ip) { echo yes } else { echo no }
352	if ('0.0.0' ~ ip) { echo yes } else { echo no }
353	## STDOUT:
354	[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+
355	yes
356	no
357	## END
358
359	#### splice with capital letters
360	shopt -s oil:all
361	var D = /d+/;
362	var ip = / D '.' D '.' D '.' D /
363	echo $ip
364	if ('0.0.0.0' ~ ip) { echo yes } else { echo no }
365	if ('0.0.0' ~ ip) { echo yes } else { echo no }
366	## STDOUT:
367	[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+
368	yes
369	no
370	## END
371
372	#### Matching escaped tab character
373	shopt -s oil:all
374
375	var lines=%($'aa\tbb' $'cc\tdd')
376
377	var pat = / ('a' [\t] 'b') /
378	write pat=$pat
379	write @lines | egrep $pat
380
381	## stdout-json: "pat=(a[\t]b)\naa\tbb\n"
382
383	#### Match unicode char
384	shopt -s oil:all
385	var pat = / 'a' dot 'b' /
386
387	if ('axb' ~ pat ) { echo yes } else { echo no }
388
389	# mu character
390	if ($'a\xce\xbcb' ~ pat ) { echo yes } else { echo no }
391
392	if ('aZZb' ~ pat ) { echo yes } else { echo no }
393	## STDOUT:
394	yes
395	yes
396	no
397	## END
398
399	#### Match non-ASCII byte denoted using $'\xff' (TODO: LANG=C)
400
401	# NOTE: This pattern doesn't work with en_US.UTF-8. I think the user should
402	# set LANG=C or shopt --unset libc_utf8.
403
404	shopt -s oil:all
405	var pat = /[ $'\xff' ]/;
406
407	echo $pat | od -A n -t x1
408	if ($'\xff' ~ pat) { echo yes } else { echo no }
409	if ($'\xfe' ~ pat) { echo yes } else { echo no }
410
411	## STDOUT:
412	5b ff 5d 0a
413	yes
414	no
415	## END
416
417	#### Match non-ASCII byte denoted using \xff (TODO: LANG=C)
418	shopt -s oil:all
419	var pat = /[ \xff ]/;
420
421	echo $pat | od -A n -t x1
422	if ($'\xff' ~ pat) { echo yes } else { echo no }
423	if ($'\xfe' ~ pat) { echo yes } else { echo no }
424
425	## STDOUT:
426	5b ff 5d 0a
427	yes
428	no
429	## END
430
431	#### ERE can express Unicode escapes that are in the ASCII range
432	shopt -s oil:all
433	var pat = /[ \u{7f} ]/;
434
435	echo $pat | od -A n -t x1
436	if ($'\x7f' ~ pat) { echo yes } else { echo no }
437	if ($'\x7e' ~ pat) { echo yes } else { echo no }
438
439	var pat2 = /[ \u{7f} ]/;
440	var pat3 = /[ \u{0007f} ]/;
441	test "$pat2" = "$pat3" && echo 'equal'
442
443	## STDOUT:
444	5b 7f 5d 0a
445	yes
446	no
447	equal
448	## END
449
450	#### ERE can't express higher Unicode escapes
451	shopt -s oil:all
452	var pat = /[ \u{ff} ]/;
453
454	echo $pat | od -A n -t x1
455	if ($'\x7f' ~ pat) { echo yes } else { echo no }
456	if ($'\x7e' ~ pat) { echo yes } else { echo no }
457
458	## status: 1
459	## stdout-json: ""
460
461	#### non-ASCII bytes must be singleton terms, e.g. '\x7f\xff' is disallowed
462	var bytes = $'\x7f\xff'
463	var pat = / [ $bytes ] /
464	echo $pat
465	## status: 1
466	## stdout-json: ""
467
468	#### Matching escaped tab character
469	shopt -s oil:all
470
471	# BUG: need C strings in array literal
472	var lines=%($'aa\tbb' $'cc\tdd')
473
474	var pat = / ('a' [\t] 'b') /
475	write pat=$pat
476	write @lines | egrep $pat
477
478	## stdout-json: "pat=(a[\t]b)\naa\tbb\n"
479
480	#### Matching ] and \ and ' and " in character classes
481	shopt -s oil:all
482
483	# BUG: need C strings in array literal
484	var lines=%(
485	'backslash \'
486	'rbracket ]'
487	'lbracket ['
488	"sq '"
489	'dq "'
490	)
491
492	# Weird GNU quirk: ] has to come first!
493	# []abc] works. But [abc\]] does NOT work. Stupid rule!
494
495	var pat = / [ ']' \\ \' \" ] /
496	write pat=$pat
497	write @lines | egrep $pat
498
499	## STDOUT:
500	pat=[]\\'"]
501	backslash \
502	rbracket ]
503	sq '
504	dq "
505	## END
506
507	#### Matching literal hyphen in character classes
508	shopt -s oil:all
509
510	var literal = '-'
511	var pat = / [ 'a' $literal 'b' ${literal} "-" ] /
512	write pat=$pat
513	write 'c-d' 'ab' 'cd' | grep $pat
514	## STDOUT:
515	pat=[a\-b\-\-]
516	c-d
517	ab
518	## END
519
520	#### Repeated String Literal With Single Char
521	shopt -s oil:all
522
523	var literal = 'f'
524	var pat = null
525
526	setvar pat = / %start $literal+ %end /
527	echo $pat
528	setvar pat = / %start ($literal)+ %end /
529	echo $pat
530
531	if ('fff' ~ pat) { echo yes }
532	if ('foo' !~ pat) { echo no }
533
534	## STDOUT:
535	^f+$
536	^(f)+$
537	yes
538	no
539	## END
540
541	#### Error when unparenthesized string of more than one character is repeated
542	shopt -s oil:all
543
544	var literal = 'foo'
545	var pat = null
546
547	setvar pat = / %start $literal+ %end /
548	echo $pat
549	setvar pat = / %start ($literal)+ %end /
550	echo $pat
551
552	if ('foofoo' ~ pat) { echo yes }
553	if ('foof' !~ pat) { echo no }
554
555	## status: 1
556	## stdout-json: ""
557
558	#### Instead of $'foo\\bar' use 'foo' \\ 'bar'
559	shopt -s oil:all
560	var pat = /'foo' \\ 'bar'/
561	echo $pat
562
563	if (r'foo\bar' ~ pat) { echo yes }
564	if (r'foo.bar' !~ pat) { echo no }
565	## STDOUT:
566	foo\\bar
567	yes
568	no
569	## END
570
571	#### Negation of Character Class ![a-z]
572	shopt -s oil:all
573
574	var pat = / ![ a-z ] /
575	echo $pat
576
577	if ('0' ~ pat) { echo yes }
578	if ('a' !~ pat) { echo no }
579
580	## STDOUT:
581	[^a-z]
582	yes
583	no
584	## END
585
586	#### Posix and Perl class in class literals
587	shopt -s oil:all
588
589	var pat = null
590
591	setvar pat = / [ space 'z' ] /
592	echo $pat
593	#setvar pat = / [ ~space 'z' ] /
594	#echo $pat
595
596	# PROBLEM: can't negate individual POSIX classes. They would have to be a Perl
597	# class to be \D or \S.
598	# [[:space:]z] negates the whole thing!
599	# [^[:space:]]
600
601	setvar pat = / [ digit 'z' ] /
602	echo $pat
603	#setvar pat = / [ ~digit 'z' ] /
604	#echo $pat
605
606	## STDOUT:
607	[[:space:]z]
608	[[:digit:]z]
609	## END
610
611	#### [!d] can't be negated because it's a literal character
612	setvar pat = / [ !d 'z' ] /
613	echo $pat
614	## status: 2
615	## stdout-json: ""
616
617	#### [!digit] can't be negated in POSIX ERE (but yes in Perl)
618	var pat = null
619	setvar pat = / [ !digit 'z' ] /
620	echo $pat
621	## status: 1
622	## stdout-json: ""
623
624	#### Long Python Example
625
626	# https://docs.python.org/3/reference/lexical_analysis.html#integer-literals
627
628	# integer ::= decinteger | bininteger | octinteger | hexinteger
629	# decinteger ::= nonzerodigit (["_"] digit)* | "0"+ (["_"] "0")*
630	# bininteger ::= "0" ("b" | "B") (["_"] bindigit)+
631	# octinteger ::= "0" ("o" | "O") (["_"] octdigit)+
632	# hexinteger ::= "0" ("x" | "X") (["_"] hexdigit)+
633	# nonzerodigit ::= "1"..."9"
634	# digit ::= "0"..."9"
635	# bindigit ::= "0" | "1"
636	# octdigit ::= "0"..."7"
637	# hexdigit ::= digit | "a"..."f" | "A"..."F"
638
639	shopt -s oil:all
640
641	DecDigit = / [0-9] /
642	BinDigit = / [0-1] /
643	OctDigit = / [0-7] /
644	HexDigit = / [0-9 a-f A-F] / # note: not splicing Digit into character class
645
646	DecInt = / [1-9] ('_'? DecDigit)* | '0'+ ('_'? '0')* /
647	BinInt = / '0' [b B] ('_'? BinDigit)+ /
648	OctInt = / '0' [o O] ('_'? OctDigit)+ /
649	HexInt = / '0' [x X] ('_'? HexDigit)+ /
650
651	Integer = / %start (DecInt | BinInt | OctInt | HexInt) %end /
652
653	#echo $Integer
654
655	if ( '123' ~ Integer) { echo 'Y' }
656	if ( 'zzz' !~ Integer) { echo 'N' }
657
658	if ('123_000' ~ Integer) { echo 'Y decimal' }
659	if ('000_123' !~ Integer) { echo 'N decimal' }
660
661	if ( '0b100' ~ Integer) { echo 'Y binary' }
662	if ( '0b102' !~ Integer) { echo 'N binary' }
663
664	if ( '0o755' ~ Integer) { echo 'Y octal' }
665	if ( '0o778' !~ Integer) { echo 'N octal' }
666
667	if ( '0xFF' ~ Integer) { echo 'Y hex' }
668	if ( '0xFG' !~ Integer) { echo 'N hex' }
669
670	## STDOUT:
671	Y
672	N
673	Y decimal
674	N decimal
675	Y binary
676	N binary
677	Y octal
678	N octal
679	Y hex
680	N hex
681	## END
682
683	#### Invalid sh operation on eggex
684	var pat = / d+ /
685	#pat[invalid]=1
686	pat[invalid]+=1
687	## status: 1
688	## stdout-json: ""
689