xref: /freebsd/contrib/bmake/unit-tests/varmod-subst-regex.mk (revision 2c3632d14fe37fa35c262ee9fb66835be0a52621)
1# $NetBSD: varmod-subst-regex.mk,v 1.3 2020/08/28 17:15:04 rillig Exp $
2#
3# Tests for the :C,from,to, variable modifier.
4
5all: mod-regex-compile-error
6all: mod-regex-limits
7all: mod-regex-errors
8
9# The variable expression expands to 4 words.  Of these words, none matches
10# the regular expression "a b" since these words don't contain any
11# whitespace.
12.if ${:Ua b b c:C,a b,,} != "a b b c"
13.error
14.endif
15
16# Using the '1' modifier does not change anything.  The '1' modifier just
17# means to apply at most 1 replacement in the whole variable expression.
18.if ${:Ua b b c:C,a b,,1} != "a b b c"
19.error
20.endif
21
22# The 'W' modifier treats the whole variable value as a single big word,
23# containing whitespace.  This big word matches the regular expression,
24# therefore it gets replaced.  Whitespace is preserved after replacing.
25.if ${:Ua b b c:C,a b,,W} != " b c"
26.error
27.endif
28
29# The 'g' modifier does not have any effect here since each of the words
30# contains the character 'b' a single time.
31.if ${:Ua b b c:C,b,,g} != "a c"
32.error
33.endif
34
35# The first :C modifier has the 'W' modifier, which makes the whole
36# expression a single word.  The 'g' modifier then replaces all occurrences
37# of "1 2" with "___".  The 'W' modifier only applies to this single :C
38# modifier.  This is demonstrated by the :C modifier that follows.  If the
39# 'W' modifier would be preserved, only a single underscore would have been
40# replaced with an 'x'.
41.if ${:U1 2 3 1 2 3:C,1 2,___,Wg:C,_,x,} != "x__ 3 x__ 3"
42.error
43.endif
44
45# The regular expression does not match in the first word.
46# It matches once in the second word, and the \0\0 doubles that word.
47# In the third word, the regular expression matches as early as possible,
48# and since the matches must not overlap, the next possible match would
49# start at the 6, but at that point, there is only one character left,
50# and that cannot match the regular expression "..".  Therefore only the
51# "45" is doubled in the result.
52.if ${:U1 23 456:C,..,\0\0,} != "1 2323 45456"
53.error
54.endif
55
56# The modifier '1' applies the replacement at most once, across the whole
57# variable value, no matter whether it is a single big word or many small
58# words.
59#
60# Up to 2020-08-28, the manual page said that the modifiers '1' and 'g'
61# were orthogonal, which was wrong.
62.if ${:U12345 12345:C,.,\0\0,1} != "112345 12345"
63.error
64.endif
65
66# Multiple asterisks form an invalid regular expression.  This produces an
67# error message and (as of 2020-08-28) stops parsing in the middle of the
68# variable expression.  The unparsed part of the expression is then copied
69# verbatim to the output, which is unexpected and can lead to strange shell
70# commands being run.
71mod-regex-compile-error:
72	@echo $@: ${:Uword1 word2:C,****,____,g:C,word,____,:Q}.
73
74# These tests generate error messages but as of 2020-08-28 just continue
75# parsing and execution as if nothing bad had happened.
76mod-regex-limits:
77	@echo $@:11-missing:${:U1 23 456:C,..,\1\1,:Q}
78	@echo $@:11-ok:${:U1 23 456:C,(.).,\1\1,:Q}
79	@echo $@:22-missing:${:U1 23 456:C,..,\2\2,:Q}
80	@echo $@:22-missing:${:U1 23 456:C,(.).,\2\2,:Q}
81	@echo $@:22-ok:${:U1 23 456:C,(.)(.),\2\2,:Q}
82	# The :C modifier only handles single-digit capturing groups,
83	# which is more than enough for daily use.
84	@echo $@:capture:${:UabcdefghijABCDEFGHIJrest:C,(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.),\9\8\7\6\5\4\3\2\1\0\10\11\12,}
85
86mod-regex-errors:
87	@echo $@: ${UNDEF:Uvalue:C,[,,}
88