[src/trunk]: src/usr.bin/make/unit-tests tests/make: test edge cases in patte...

[rillig <rillig%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/dd065bfc1b43
branches:  trunk
changeset: 366725:dd065bfc1b43
user:      rillig <rillig%NetBSD.org@localhost>
date:      Sat Jun 11 09:15:49 2022 +0000

description:
tests/make: test edge cases in pattern matching

diffstat:

 usr.bin/make/unit-tests/varmod-match.exp |    4 +-
 usr.bin/make/unit-tests/varmod-match.mk  |  107 ++++++++++++++++++++++++++----
 2 files changed, 93 insertions(+), 18 deletions(-)

diffs (147 lines):

diff -r ea1d9f04d602 -r dd065bfc1b43 usr.bin/make/unit-tests/varmod-match.exp
--- a/usr.bin/make/unit-tests/varmod-match.exp  Sat Jun 11 08:06:32 2022 +0000
+++ b/usr.bin/make/unit-tests/varmod-match.exp  Sat Jun 11 09:15:49 2022 +0000
@@ -10,8 +10,8 @@
 Comparing "$" != "$"
 CondParser_Eval: ${:Ua \$ sign any-asterisk:M*\$*} != "any-asterisk"
 Comparing "any-asterisk" != "any-asterisk"
-make: "varmod-match.mk" line 151: Unknown modifier "]"
-make: "varmod-match.mk" line 151: Malformed conditional (${ ${:U\:} ${:U\:\:} :L:M[:]} != ":")
+make: "varmod-match.mk" line 157: Unknown modifier "]"
+make: "varmod-match.mk" line 157: Malformed conditional (${ ${:U\:} ${:U\:\:} :L:M[:]} != ":")
 make: Fatal errors encountered -- cannot continue
 make: stopped in unit-tests
 exit status 1
diff -r ea1d9f04d602 -r dd065bfc1b43 usr.bin/make/unit-tests/varmod-match.mk
--- a/usr.bin/make/unit-tests/varmod-match.mk   Sat Jun 11 08:06:32 2022 +0000
+++ b/usr.bin/make/unit-tests/varmod-match.mk   Sat Jun 11 09:15:49 2022 +0000
@@ -1,4 +1,4 @@
-# $NetBSD: varmod-match.mk,v 1.10 2022/06/11 07:54:25 rillig Exp $
+# $NetBSD: varmod-match.mk,v 1.11 2022/06/11 09:15:49 rillig Exp $
 #
 # Tests for the :M variable modifier, which filters words that match the
 # given pattern.
@@ -130,6 +130,12 @@
 .  error
 .endif
 
+#      [\]     matches a single backslash
+WORDS=         a\b a[\]b ab
+.if ${WORDS:Ma[\]b} != "a\\b"
+.  error
+.endif
+
 #      :       terminates the pattern
 .if ${ A * :L:M:} != ""
 .  error
@@ -156,26 +162,95 @@
 
 #      [\]     matches exactly a backslash; no escaping takes place in
 #              character ranges
-# Without the 'a' in the below expressions, the backslash would end a word and
-# thus influence how the string is split into words.
-.if ${ ${:U\\a} ${:U\\\\a} :L:M[\]a} != "\\a"
+# Without the 'a' in the below words, the backslash would end a word and thus
+# influence how the string is split into words.
+WORDS=         1\a 2\\a
+.if ${WORDS:M?[\]a} != "1\\a"
+.  error
+.endif
+
+#      [[-]]   May look like it would match a single '[', '\' or ']', but
+#              the inner ']' has two roles: it is the upper bound of the
+#              character range as well as the closing character of the
+#              character list.  The outer ']' is just a regular character.
+WORDS=         [ ] [] \] ]]
+.if ${WORDS:M[[-]]} != "[] \\] ]]"
+.  error
+.endif
+
+#      [b[-]a]
+#              Same as for '[[-]]': the character list stops at the first
+#              ']', and the 'a]' is treated as a literal string.
+WORDS=         [a \a ]a []a \]a ]]a [a] \a] ]a] ba]
+.if ${WORDS:M[b[-]a]} != "[a] \\a] ]a] ba]"
+.  error
+.endif
+
+#      [-]     Matches a single '-' since the '-' only becomes part of a
+#              character range if it is preceded and followed by another
+#              character.
+WORDS=         - -]
+.if ${WORDS:M[-]} != "-"
+.  error
+.endif
+
+#      [       Incomplete empty character list, never matches.
+WORDS=         a a[
+.if ${WORDS:Ma[} != ""
+.  error
+.endif
+
+#      [^      Incomplete negated empty character list, matches any single
+#              character.
+WORDS=         a a[ aX
+.if ${WORDS:Ma[^} != "a[ aX"
 .  error
 .endif
 
-#.MAKEFLAGS: -dcv
+#      [-x1-3  Incomplete character list, matches those elements that can be
+#              parsed without lookahead.
+WORDS=         - + x xx 0 1 2 3 4 [x1-3
+.if ${WORDS:M[-x1-3} != "- x 1 2 3"
+.  error
+.endif
+
+#      [^-x1-3
+#              Incomplete negated character list, matches any character
+#              except those elements that can be parsed without lookahead.
+WORDS=         - + x xx 0 1 2 3 4 [x1-3
+.if ${WORDS:M[^-x1-3} != "+ 0 4"
+.  error
+.endif
+
+#      [\      Incomplete character list containing a single '\'.
 #
-# Incomplete patterns:
-#      [       matches TODO
-#      [x      matches TODO
-#      [^      matches TODO
-#      [-      matches TODO
-#      [xy     matches TODO
-#      [^x     matches TODO
-#      [\      matches TODO
+#              A word can only end with a backslash if the preceding
+#              character is a backslash as well; in all other cases the final
+#              backslash would escape the following space, making the space
+#              part of the word.  Only the very last word of a string can be
+#              '\', as there is no following space that could be escaped.
+WORDS=         \\ \a ${:Ux\\}
+.if ${WORDS:M?[\]} != "\\\\ x\\"
+.  error
+.endif
+
+#      [x-     Incomplete character list containing an incomplete character
+#              range, matches only the 'x'.
+WORDS=         [x- x x- y
+.if ${WORDS:M[x-} != "x"
+.  error
+.endif
+
+#      [^x-    Incomplete negated character list containing an incomplete
+#              character range; matches each word that does not have an 'x'
+#              at the position of the character list.
 #
-#      [x-     matches exactly 'x', doesn't match 'x-'
-#      [^x-    matches TODO
-#      \       matches never
+#              XXX: Even matches strings that are longer than a single
+#              character.
+WORDS=         [x- x x- y yyyyy
+.if ${WORDS:M[^x-} != "[x- y yyyyy"
+.  error
+.endif
 
 
 # The modifier ':tW' prevents splitting at whitespace.  Even leading and