prtwash.pl: allow user to toggle the parsing of pkgmk.conf

man1/pkgfoster.1

@@ -10,24 +10,25 @@
 .fam T
 .fi
 .SH DESCRIPTION
-\fBpkgfoster\fP is a simple interactive script you can use to clean up orphaned packages (i.e.,
-packages not listed in the "Depends on: " line of any other installed package). It uses
-prt-cache by default, so remember to build the cache with "prt-get cache". You can also
-use normal prt-get by modifying the PRT_GET variable. The bash array FOSTERED
-is read from and written to ~/.config/pkgfoster.conf, as a means of preserving your choices
-regarding which packages to keep. If you run \fBpkgfoster\fP as a non-root user, make sure the user
-is able to elevate permissions using sudo, doas, or 'su -c'.
+\fBpkgfoster\fP is a simple interactive script you can use to clean up orphaned
+packages (i.e., packages not listed in the "Depends on: " line of any other installed
+package). It uses prt-cache by default, so remember to build the cache with "prt-get
+cache". You can also use normal prt-get by modifying the PRT_GET variable. The file
+~/.config/pkgfoster.keep is used to preserve your decisions regarding which packages
+to keep. If you run \fBpkgfoster\fP as a non-root user, make sure the user is able to
+elevate permissions using sudo, doas, or 'su -c'.
 .PP
 Packages from the core collection are never considered for deletion, since some of them
 might provide runtime dependencies yet not be listed in the "Depends on: " line. See
 \fBPkgfile\fP(5) for an explanation of this practice.
 .PP
-Some packages flagged by \fBpkgfoster\fP might actually be optional (soft) dependencies of other
-packages on your system. Use \fBrevdep\fP(1) to identify any breakage that you introduce by
-removing these misidentified orphans, and rebuild the affected ports using
+Some packages flagged by \fBpkgfoster\fP might actually be optional (soft)
+dependencies of other packages on your system. Use \fBrevdep\fP(1) to identify any
+breakage that you introduce by removing these misidentified orphans, and rebuild the
+affected ports using
 .B prt\-get update \-fr
 (see \fBprt\-get\fP(8) for more detailed examples).
 .SH AUTHORS
 Jukka Heino <jukka@karsikkopuu.net>
 .SH SEE ALSO
-\fBprt\-cache\fP(8), \fBprt\-get\fP(8), \fBrevdep\fP(1)
+\fBprt\-cache\fP(8), \fBprt\-get\fP(8), \fBPkgfile\fP(5), \fBrevdep\fP(1)

man1/prtsweep.1

@@ -1,7 +1,7 @@
 .\" 
 .\" prtsweep manual page.
 .\" (C) 2e003 by Martin Opel
-.\" Revised 2021 by John McQuah
+.\" Revised 2022 by John McQuah
 .\"
 .TH prtsweep 1
 .SH NAME
@@ -9,7 +9,7 @@ prtsweep \- sweep old files from the ports directories
 .SH SYNOPSIS
 .PP
 .B prtsweep
-[\-a] [\-d] [\-n] [PORTDIR ...]
+[\-a] [\-d] [\-n] [\-q] [PORTDIR ...]
 .SH DESCRIPTION
 
 The \fIprtsweep\fP perl script sweeps port directories, deleting unneeded files.
@@ -17,7 +17,7 @@ The \fIprtsweep\fP perl script sweeps port directories, deleting unneeded files.
 built package
 .PP
 .nf
-	name#version-release.pkg.tar.?z*
+	name#version-release.pkg.tar.*
 .fi
 .PP
 All other files are removed. If a traversal of the ports collections in automatic mode

man1/prtwash.1

@@ -36,34 +36,28 @@ All items not included in the previous list WILL be deleted
 You can choose to delete some of the above by passing the proper
 option:
 .TP
-.B
--p
-also remove the built package (current version)
+.B \-p
+remove the built package (current version)
 .TP
-.B
--b
-also remove other built packages (older versions)
+.B \-b
+remove older builds (any package tarball that does NOT match
+the current version)
 .TP
-.B
--d
-also remove the add-on files provided by CRUX maintainers
+.B \-d
+remove add-on files provided by CRUX maintainers
 (patches and initscripts downloaded with \fBports -u\fP)
 .TP
-.B
--s
-also remove the upstream sources (those NOT obtained via \fBports -u\fP)
+.B \-s
+remove upstream sources (those NOT obtained via \fBports -u\fP)
 .TP
-.B
--t
+.B \-t
 (test mode) display the target files without deleting
 .TP
-.B
--q
+.B \-q
 (quiet mode) suppress messages about files that are kept;
 basically the same as \fBprtwash [OPTIONS] | grep -v "keeping"\fP
 .TP
-.B
--a
+.B \-a
 (automatic mode) take the port collections from the
 /etc/prt-get.conf file and do a recursive cleaning;
 if omitted, the command line must contain at least one <path> specifying
@@ -71,12 +65,14 @@ a valid port directory
 .PP
 Misc options
 .TP
-.B
--h
+.B \-\-parse\-pkgmk\-conf
+Revert to the v0.9 behaviour of parsing /etc/pkgmk.conf to determine the
+compression mode of built packages
+.TP
+.B \-h
 Display usage information
 .TP
-.B
--v
+.B \-v
 Display version
 .SH ENVIRONMENT
 In automatic mode, \fBprtwash\fP gets a list of repositories from
@@ -93,11 +89,24 @@ Sources under CRUX version control are only deleted if you pass the
 option \fB\-d\fP. Sources from upstream are only deleted if you pass the
 option \fB\-s\fP.
 .PP
-The current version of the package is only deleted if you pass the
-option \fB\-p\fP. Older versions of the package are only deleted if you
-pass the option \fB\-b\fP. Note that if the compression mode defined
-in /etc/pkgmk.conf has been changed since you last built the package,
-\fBprtwash\fP will not recognize the tarball in its regexp search.
+You can also delete the built package (current version) by passing
+the option \fB\-p\fP. To select only older versions for deletion,
+pass the option \fB\-b\fP instead. These two options operate independently,
+allowing you to target four distinct subsets (including the empty set)
+for deletion.
+.PP
+By passing the option \fB\-\-parse\-pkgmk\-conf\fP, you can modify the behaviour
+of the pattern matching performed to detect built packages. By default, the
+regexp searches are done with the patterns
+.B /${name}#${version}.*.pkg.tar.(gz|lz|xz|bz2)/
+and
+.B /${name}#.*pkg.tar.(gz|lz|xz|bz2)/
+using the variables defined in the \fBPkgfile\fP(5). Passing the option
+.B \-\-parse\-pkgmk\-conf
+will result in a more narrow regexp search, allowing the deletion of any package
+tarball that uses a different compression algorithm than the one currently defined
+in /etc/pkgmk.conf (which is how the v0.9 bash script worked).
+
 .SH COMPARISON WITH OTHER UTILITIES
 Because \fBprtwash\fP reads the location of port collections from \fB/etc/prt-get.conf\fP(5),
 you can easily control which collections are cleaned in automatic mode by commenting

scripts/pkgfoster

@@ -4,11 +4,11 @@
 # Jukka Heino <jukka@karsikkopuu.net>
 #
 # pkgfoster is a simple script you can use to clean up orphaned packages (i.e. 
-# packages which no other package depends on). It uses prt-cache by default, so 
-# remember to build the cache with "prt-get cache". You can also use normal 
-# prt-get by modifying the PRT_GET variable. An array of packages to keep 
-# is stored in the FOSTERED variable. Packages from core are never considered
-# for deletion.
+# packages which do not appear in the "Depends on: " line of any other installed
+# package). It uses prt-cache by default, so remember to build the cache with
+# "prt-get cache". You can also use normal prt-get by modifying the PRT_GET
+# variable. The file ~/.config/pkgfoster.keep contains a list of all packages
+# you've decided to keep. Packages from core are never considered for deletion.
 #
 
 PRT_GET=prt-cache
@@ -26,17 +26,20 @@ else
 fi
 
 BASE=$(awk '/^[[:space:]]*prtdir.*\/core/ {print $2}' /etc/prt-get.conf)
-CONF=${XDG_CONFIG_HOME:="$HOME/.config"}/pkgfoster.conf
-[ -f $CONF ] && . $CONF || mkdir -p ${XDG_CONFIG_HOME}
+CONF="${XDG_CONFIG_HOME:="$HOME/.config"}/pkgfoster.keep"
+if [ ! -f "$CONF" ]; then
+	mkdir -p "${XDG_CONFIG_HOME}"
+	touch "$CONF"
+fi
 
 echo "Checking packages for orphans..."
 
 while true ; do
 	RECHECK=0
 
-	orphans=$(comm -13 <(cat <(find "$BASE" -maxdepth 1 -type d -printf "%f\n") \
-		<(for (( k=0; k<${#FOSTERED[@]}; k++ )); do echo "${FOSTERED[$k]}"; done) \
-		| sort -u) <($PRT_GET listorphans | sort) | tr '\n' ' ')
+	orphans=$(comm -23 <($PRT_GET listorphans | sort) \
+		<(cat <(find "$BASE" -maxdepth 1 -type d -printf "%f\n") "$CONF" \
+		| sort -u) | tr '\n' ' ')
 
 	for PACKAGE in ${orphans[@]}; do
 		echo
@@ -50,16 +53,11 @@ while true ; do
 			$PKGRM $PACKAGE
 			RECHECK=1
 		else
-			FOSTERED=( $PACKAGE ${FOSTERED[@]} )
+			echo $PACKAGE >> "$CONF"
 		fi
 	done
 
 	if [ "$RECHECK" = "0" ] ; then
-		echo -n "FOSTERED=(" > $CONF
-		for (( k=0; k<${#FOSTERED[@]}; k++ )); do
-			echo -n "${FOSTERED[$k]} " >> $CONF
-		done
-		echo ")" >> $CONF
 		exit 0
 	fi
 

scripts/prtsweep.pl

@@ -19,15 +19,34 @@ our %options = ( auto => 0, dryrun => 0, rmdir => 0, pkgtoo => 0, quiet => 0 );
 our @portdirs;
 our $argports;
 
+######################### main routine ################################
+parse_args();
+print_usage() if ((2*$argports-1)*(1-2*$options{auto}) < 0);
+
+if ($options{auto} == 1) {
+	my @basedirs = getportdirs();
+	foreach my $collection (@basedirs) {
+		print "====> Sweeping port collection $collection\n";
+		foreach my $port (list_subdirs($collection)) {
+			do_sweep($port);
+		}
+	}
+} else {
+	foreach my $port (@portdirs) {
+		do_sweep($port);
+	}
+}
+
+######################### subroutines #################################
 sub print_usage {
 	print <<EOT;
 Usage: prtsweep [OPTION]... [PORTDIRS]... 
 
   -a     automatic mode, only valid when [PORTDIRS] are omitted
-  -d     remove directories of dropped ports (no Pkgfile present)
+  -d     remove directories of dropped ports (signature file not found)
   -n     dry-run, don't actually delete anything
   -p     delete any built packages too
-  -q     quiet mode
+  -q     quiet mode, only print messages when files are removed
 
 Report bugs on libera.chat #crux-devel
 EOT
@@ -36,22 +55,24 @@ EOT
 
 sub parse_args {
 	foreach my $arg (@ARGV) {
-		next if (-f $arg); # no filenames, only options or directories
-		if ($arg =~ /^-a$/) {
+		if ($arg eq "-a") {
 			$options{auto} = 1;
-		} elsif ($arg =~ /^-d$/) {
+		} elsif ($arg eq "-d") {
 			$options{rmdir} = 1;
-		} elsif ($arg =~ /^-n$/) {
+		} elsif ($arg eq "-n") {
 			$options{dryrun} = 1;
-		} elsif ($arg =~ /^-p$/) {
+		} elsif ($arg eq "-p") {
 			$options{pkgtoo} = 1;
-		} elsif ($arg =~ /^-q$/) {
+		} elsif ($arg eq "-q") {
 			$options{quiet} = 1;
-		} elsif ($arg =~ /^--version$/) {
+		} elsif ($arg eq "--version") {
 			print $version."\n";
 			exit 0;
-		} elsif (-d "$arg") {
+		} elsif ((-d "$arg") ||        # false for symlink to a portdir,
+		       (-f "$arg/.signature")) {  # so a second test is needed
 			push (@portdirs, $arg);
+		} elsif (-f "$arg") {
+			print "WARN: $arg is not a port directory or recognized option, ignoring.\n";
 		} else {
 			print_usage();
 		}
@@ -68,7 +89,8 @@ sub list_subdirs {
 	opendir(DIR, $path) or return;
 	foreach my $entry(sort(readdir(DIR))) {
 		next if ( substr($entry,0,1) eq '.' );
-		push (@list, "$path/$entry") if -d "$path/$entry";
+		push (@list, "$path/$entry") if ((-d "$path/$entry")
+			or (-f "$path/$entry/.signature"));
 	}
 	closedir(DIR);
 	return @list;
@@ -97,15 +119,15 @@ sub sweep {
 
 	print "=======> $port\n" unless $options{quiet}==1;
 	my @wanted = parse_signature ("$port/.signature");
-	my $builtpkg=($options{pkgtoo} != 1) ? $path[-1].'#.*pkg\.tar.*' :
-		'ReallyLongStringThatWouldNeverMatchAnyBuiltPackage';
-	$builtpkg =~ s/\+/\\\+/;
+	my $builtpkg=$path[-1].'#.*pkg\.tar\.(bz2|gz|lz|xz)$';
+	$builtpkg =~ s/\+/\\\+/; # plus sign in filenames interferes with regex search
 
 	opendir (DIR, $port) or return;
 	foreach my $f (sort(readdir(DIR))) {
 		next if ( $f eq '.' or $f eq '..' );
-		$f =~ s/\+/\\\+/; # plus sign in filenames interferes with regex search
-		if ((grep /$f/, @wanted) >= 1 or ($f =~ /$builtpkg/)) {
+		$f =~ s/\+/\\\+/;
+		if ((grep /$f/, @wanted) >= 1 or
+			($f =~ /$builtpkg/)*($options{pkgtoo}==0)) {
 			print "... keeping file $port/$f.\n" unless $options{quiet} == 1;
 		} else {
 			remove ("$port/$f");
@@ -127,15 +149,17 @@ sub remove {
 }
 
 sub do_sweep {
+	# argument either a real directory (not symlink) or has a signature;
+	# this subroutine determines which condition was satisfied.
 	my $port = shift; my $nf = 0;
 	if (! -f "$port/.signature") {
-		opendir (PORTDIR,$port) or die "cannot open $port for reading\n";
+		opendir (PORTDIR,$port) or return;
 		foreach my $f (readdir PORTDIR) {
 			next if ($f eq '.' or $f eq '..');
 			$nf += 1;
 		}
 		closedir (PORTDIR);
-		print "no signature file, $port does not define a valid port.";
+		print "WARN: $port/.signature not found, invalid port directory.";
 		rm_emptydir($port,$nf);
 	} else {
 		sweep($port);
@@ -145,10 +169,11 @@ sub do_sweep {
 sub rm_emptydir {
 	my $port = shift; my $nf = shift;
 	my $msg = ($options{rmdir}==1) ? "\n":
-		" use -d to remove empty directories.\n";
+		"\n  Use -d to remove empty directories.\n";
 	my $modal = ($options{dryrun}==0) ? "" : "would be";
-	$msg = ($nf == 0) ? "$msg empty directory $port $modal deleted.\n" :
-		"$msg cannot remove $port: directory not empty\n";
+	my $post = ($nf == 0) ? "  Empty directory $port $modal deleted.\n" :
+		"  Cannot remove $port: directory not empty\n";
+	$msg = ($options{rmdir}==1) ? "$msg $post" : $msg ;
 	print $msg;
 	rmdir ($port) if (($nf == 0) and ($options{dryrun} == 0));
 }
@@ -179,21 +204,3 @@ sub getportdirs {
 	closedir PORTS_DEFS;
 	return @basedirs ;
 }
-
-# main
-parse_args();
-print_usage() if ((2*$argports-1)*(1-2*$options{auto}) < 0);
-
-if ($options{auto} == 1) {
-	my @basedirs = getportdirs();
-	foreach my $collection (@basedirs) {
-		print "====> Sweeping port collection $collection\n";
-		foreach my $port (list_subdirs($collection)) {
-			do_sweep($port);
-		}
-	}
-} else {
-	foreach my $port (@portdirs) {
-		do_sweep($port);
-	}
-}

scripts/prtwash.pl

@@ -18,7 +18,7 @@ our @portdirs;
 our $argports;
 my $version = 1.2.1;
 our %options = ( oldver=>0, pkgtoo=>0, srctoo=>0, addons=>0,
-  dryrun=>0, auto=>0, quiet=>0 );
+  dryrun=>0, auto=>0, quiet=>0, parse_pkgmk=>0 );
 
 ####################   main routine   ################################
 parse_args();
@@ -46,9 +46,8 @@ if ($options{auto} == 1) {
 
 ####################   subroutines   #################################
 sub parse_args {
-  while ( my $arg = pop(@ARGV) ) {
-  next if (-f $arg); # no filenames, only directories and options
- if ($arg eq "-b") {
+  foreach my $arg (@ARGV) {
+  if ($arg eq "-b") {
     $options{oldver} = 1;
    } elsif ($arg eq "-p") {
     $options{pkgtoo} = 1;
@@ -62,11 +61,16 @@ sub parse_args {
     $options{auto} = 1;
    } elsif ($arg eq "-q") {
     $options{quiet} = 1;
+   } elsif ($arg eq "--parse-pkgmk-conf") {
+    $options{parse_pkgmk} = 1;
    } elsif ($arg eq "-v") {
     print "$version\n";
     exit(0);
-   } elsif (-d $arg) {
+   } elsif ((-d $arg) ||          # false for symlink to a portdir, so a
+           (-f "$arg/Pkgfile")) { # followup test is performed to catch those
     push (@portdirs, $arg);
+   } elsif (-f $arg) {
+    print "WARN: ignoring invalid port directory $arg\n";
    } else {
     print_usage();
    }
@@ -124,7 +128,7 @@ sub parse_pkgfile {
   return \$name, \$version, \$release, \@source, \@renames;
 }
 
-sub keeplist { # remember to pop off the last element for regex purposes
+sub keeplist { # remember to pop off the last two elements for regex purposes
   my $port = shift;
   my @keepers = ("Pkgfile",".footprint",".signature");
     push (@keepers,"pre-install","post-install","README","README.md",
@@ -140,7 +144,7 @@ sub keeplist { # remember to pop off the last element for regex purposes
   my $i; my $si; my $ki;
   ENTRY: for ($i=0; $i<=$#source; $i++) {
     $si = $source[$i];
-    $ki = ($renames[$i] eq "SKIP" or $renames[$i] eq "") ? basename($si) :
+    $ki = ("$renames[$i]" eq "SKIP" or "$renames[$i]" eq "") ? basename($si) :
      $renames[$i];
     if ($si =~ /^remote:/) {
       next ENTRY if ($options{srctoo}==1);
@@ -150,21 +154,23 @@ sub keeplist { # remember to pop off the last element for regex purposes
       push(@keepers, $ki);
     }
   }
-  push (@keepers, $name.'#'.$version.'-'.$release.".pkg.tar.$compression_mode") if $options{pkgtoo}==0;
-  push (@keepers, $name."#.*pkg.tar.$compression_mode");
+  push (@keepers, "$name#$version.*\.pkg\.tar\.$compression_mode");
+  push (@keepers, "$name#.*\.pkg\.tar\.$compression_mode");
   return @keepers;
 }
 
 sub get_compress {
-  my $conf = "/etc/pkgmk.conf";
-  $compression_mode = "gz";
-  open(CONFIG,$conf) or return $compression_mode;
-  while(<CONFIG>) {
-    $compression_mode = $1 if m/^[\t ]*PKGMK_COMPRESSION_MODE=(.*)\n/;
+  my $suffix = "(gz|lz|xz|bz2)";
+  if ($options{parse_pkgmk} == 1) {
+    my $conf = "/etc/pkgmk.conf";
+    open(CONFIG,$conf) or return $suffix;
+      while(<CONFIG>) {
+        $suffix = $1 if m/^[\t ]*PKGMK_COMPRESSION_MODE=(.*)\n/;
+      }
+    close(CONFIG);
+    $suffix =~ s/"//g;          # ensure no double-quotes remain (thanks jaeger)
   }
-  close(CONFIG);
-  $compression_mode =~ s/"//g;  # ensure no double-quotes remain (thanks jaeger)
-  return $compression_mode;
+  return $suffix;
 }
 
 sub do_wash {
@@ -174,16 +180,18 @@ sub do_wash {
     print "WARN: no Pkgfile found in $port. Skipping.\n";
     return;
   } else {
-    my @keepers = keeplist($port);
-    my $nevermatch = "ReallyLongStringThatWouldNeverMatchAnyBuiltPackage";
-    my $pkgrex = pop(@keepers);
-    $pkgrex = ($options{oldver} == 1 or $options{pkgtoo} == 1) ? $nevermatch : $pkgrex;
+    my @keepers = keeplist($port); my $iswanted;
+    my $allbuilds = pop(@keepers);
+    my $currbuild = pop(@keepers);
 
     opendir (DIR,$port) or return;
+    print "=====> washing $port\n" unless $options{quiet} == 1;
     foreach my $f (sort(readdir(DIR))) {
       next if ($f eq '.' or $f eq '..');
       $f =~ s/\+/\\\+/g; # plus sign in filenames interferes with regex search
-      if ((grep /$f/, @keepers) or ($f =~ /$pkgrex/)) {
+      $iswanted = ( grep (/$f/, @keepers ) >= 1 );
+      if ($iswanted or ($options{pkgtoo}==0)*($f =~ /$currbuild/)
+     or ($options{oldver}==0)*($f =~ /$allbuilds/)*($f !~ /$currbuild/)) {
         print "... keeping file $port/$f.\n" unless $options{quiet} == 1;
       } else {
         remove ("$port/$f");
@@ -194,8 +202,7 @@ sub do_wash {
 }
 
 sub getportdirs { # returns scalar references to two arrays
-  my @basedirs;
-  my @localports;
+  my @basedirs; my @localports;
   my $conf = "/etc/prt-get.conf";
 
   open(PORTS, $conf) or die "could not open $conf";
@@ -211,7 +218,7 @@ sub getportdirs { # returns scalar references to two arrays
         push @basedirs, $line if (-d $line);
       } else {
         my @a = split(/:/, $line);
-        my @b = split(/,/, @a[1]);
+        my @b = split(/,/, $a[1]);
         while ( my $c = pop @b ) {
         my $port = $a[0] . "/" . $c;
           push @localports, $port if (-d $port);