Content-Transfer-Encoding: 7bit
Content-Type: multipart/mixed; boundary="_----------=_1660070143201900"
MIME-Version: 1.0
Date: Tue, 9 Aug 2022 18:35:43 +0000
From: "Roland Illig" <rillig@netbsd.org>
Subject: CVS commit: pkgsrc/pkgtools/lintpkgsrc/files
To: pkgsrc-changes@NetBSD.org
Reply-To: rillig@netbsd.org
Message-Id: <20220809183543.A045CFB1A@cvs.NetBSD.org>
Sender: pkgsrc-changes-owner@NetBSD.org
Precedence: bulk

This is a multi-part message in MIME format.

--_----------=_1660070143201900
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
Content-Type: text/plain; charset="US-ASCII"

Module Name:	pkgsrc
Committed By:	rillig
Date:		Tue Aug  9 18:35:43 UTC 2022

Modified Files:
	pkgsrc/pkgtools/lintpkgsrc/files: lintpkgsrc.pl
	pkgsrc/pkgtools/lintpkgsrc/files/t: glob.t

Log Message:
lintpksrc: fix parsing of the ':S' modifier in makefiles


To generate a diff of this commit:
cvs rdiff -u -r1.57 -r1.58 pkgsrc/pkgtools/lintpkgsrc/files/lintpkgsrc.pl
cvs rdiff -u -r1.5 -r1.6 pkgsrc/pkgtools/lintpkgsrc/files/t/glob.t

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.


--_----------=_1660070143201900
Content-Disposition: inline
Content-Length: 4545
Content-Transfer-Encoding: binary
Content-Type: text/x-diff; charset=us-ascii

Modified files:

Index: pkgsrc/pkgtools/lintpkgsrc/files/lintpkgsrc.pl
diff -u pkgsrc/pkgtools/lintpkgsrc/files/lintpkgsrc.pl:1.57 pkgsrc/pkgtools/lintpkgsrc/files/lintpkgsrc.pl:1.58
--- pkgsrc/pkgtools/lintpkgsrc/files/lintpkgsrc.pl:1.57	Tue Aug  9 18:14:22 2022
+++ pkgsrc/pkgtools/lintpkgsrc/files/lintpkgsrc.pl	Tue Aug  9 18:35:43 2022
@@ -1,6 +1,6 @@
 #!@PERL5@
 
-# $NetBSD: lintpkgsrc.pl,v 1.57 2022/08/09 18:14:22 rillig Exp $
+# $NetBSD: lintpkgsrc.pl,v 1.58 2022/08/09 18:35:43 rillig Exp $
 
 # Written by David Brownlee <abs@netbsd.org>.
 #
@@ -879,10 +879,10 @@ sub list_pkgsrc_pkgdirs($$) {
 # Return undef on error.
 sub glob2regex($) {
 	my ($glob) = @_;
-	my (@chars, $in_alt);
-	my ($regex);
 
-	@chars = split(//, $glob);
+	my @chars = split(//, $glob);
+	my $alternative_depth = 0;
+	my $regex = '';
 	while (defined($_ = shift @chars)) {
 		if ($_ eq '*') {
 			$regex .= '.*';
@@ -890,37 +890,43 @@ sub glob2regex($) {
 			$regex .= '.';
 		} elsif ($_ eq '+') {
 			$regex .= '\\+';
-		} elsif ($_ eq '\\') {
-			$regex .= $_ . shift @chars;
+		} elsif ($_ eq '\\' && @chars > 0) {
+			my $next = shift @chars;
+			$regex .= $next =~ /\w/ ? "$next" : "\\$next";
 		} elsif ($_ eq '.' || $_ eq '|') {
 			$regex .= quotemeta;
 		} elsif ($_ eq '{') {
 			$regex .= '(';
-			++$in_alt;
+			++$alternative_depth;
 		} elsif ($_ eq '}') {
-			if (!$in_alt) {
+			if ($alternative_depth == 0) {
 				# Error
 				return undef;
 			}
 			$regex .= ')';
-			--$in_alt;
-		} elsif ($_ eq ',' && $in_alt) {
+			--$alternative_depth;
+		} elsif ($_ eq ',' && $alternative_depth) {
 			$regex .= '|';
+		} elsif ($_ eq '[') {
+			$regex .= '[';
+			while (defined($_ = shift @chars)) {
+				$regex .= $_;
+				if ($_ eq ']') {
+					last;
+				} elsif ($_ eq '\\' && @chars > 0) {
+					$regex .= shift @chars;
+				}
+			}
+			return undef if $_ ne ']';
 		} else {
 			$regex .= $_;
 		}
 	}
 
-	if ($in_alt) {
-		# Error
-		return undef;
-	}
-	if ($regex eq $glob) {
-		return ('');
-	}
-	if ($opt{D}) {
-		print "glob2regex: $glob -> $regex\n";
-	}
+	return undef if $alternative_depth > 0;
+	return '' if $regex eq $glob; # XXX: why?
+
+	$opt{D} and print "glob2regex: $glob -> $regex\n";
 	'^' . $regex . '$';
 }
 

Index: pkgsrc/pkgtools/lintpkgsrc/files/t/glob.t
diff -u pkgsrc/pkgtools/lintpkgsrc/files/t/glob.t:1.5 pkgsrc/pkgtools/lintpkgsrc/files/t/glob.t:1.6
--- pkgsrc/pkgtools/lintpkgsrc/files/t/glob.t:1.5	Tue Aug  9 18:14:22 2022
+++ pkgsrc/pkgtools/lintpkgsrc/files/t/glob.t	Tue Aug  9 18:35:43 2022
@@ -1,4 +1,4 @@
-# $NetBSD: glob.t,v 1.5 2022/08/09 18:14:22 rillig Exp $
+# $NetBSD: glob.t,v 1.6 2022/08/09 18:35:43 rillig Exp $
 
 use strict;
 use warnings;
@@ -11,46 +11,49 @@ require('../lintpkgsrc.pl');
 sub test_glob2regex() {
 
 	ok(glob2regex('*'), '^.*$');
+	ok(glob2regex('\*'), '');
 
 	ok(glob2regex('?'), '^.$');
+	ok(glob2regex('\?'), '');
 
-	ok(glob2regex('[a-z]'), '');
+	# Ordinary characters in glob patterns.
+	ok(glob2regex('+'), '^\+$');
+	ok(glob2regex('\+'), '');
+	ok(glob2regex('|'), '^\|$');
+	ok(glob2regex('\|'), '');
+
+	ok(glob2regex('\.'), '');
+	ok(glob2regex('\n'), '^n$');
+	ok(glob2regex('\\\\'), '');
+	ok(glob2regex('\['), '');
+	ok(glob2regex('\{'), '');
+	ok(glob2regex('\-'), '');
 
+	ok(glob2regex('[a-z]'), '');
 	ok(glob2regex('[a-z0-9]'), '');
-
 	ok(glob2regex('[a-z0-9_]'), '');
-
-	# Outside of braces, the ',' is a regular character.
-	ok(glob2regex('a,b'), '');
-
-	# FIXME: Inside brackets, the '*' is a literal '*'.
-	ok(glob2regex('[*]'), '^[.*]$');
-
-	ok(glob2regex('\*'), '');
+	ok(glob2regex('[*]'), '');
 
 	ok(glob2regex('*.[ch]'), '^.*\.[ch]$');
 
+	# Outside of braces, the ',' is a regular character.
+	ok(glob2regex('a,b'), '');
 	ok(glob2regex('{one,two}'), '^(one|two)$');
-
 	ok(glob2regex('{{thi,fou}r,fif}teen'), '^((thi|fou)r|fif)teen$');
 
 	# There is an unbalanced '}' at the very end.
-	ok(glob2regex('{{thi,fou}r,fif}teen}'), undef);
-
-	ok(glob2regex('a+b|c'), '^a\+b\|c$');
+	ok(glob2regex('{four,fif}teen}'), undef);
 
+	# An escaped '[' does not start a character class.
 	ok(glob2regex('a\[b*'), '^a\[b.*$');
 
-	ok(glob2regex('a\+b'), '');
-
-	ok(glob2regex('a\?b'), '');
-
-	# XXX: Depending on the exact implementation, the '\n' may be
-	# interpreted as a newline, a literal 'n' or a literal '\' 'n'.
-	ok(glob2regex('a\n*'), '^a\n.*$');
+	ok(glob2regex('a\n*'), '^an.*$');
 
 	# https://gnats.netbsd.org/12996
 	ok(glob2regex('libsigc++'), '^libsigc\+\+$');
+
+	my $re = 'a\nb';
+	ok("a\nb" =~ $re, 1);
 }
 
 test_glob2regex();


--_----------=_1660070143201900--