Fast INDEX generation. See bsd.port.mk rev. 1.300 for details.

Submitted by:	steve

1 files changed, 117 insertions, 0 deletions

diff --git a/Tools/make_index b/Tools/make_index
new file mode 100644
index 000000000000..aa465b820e8a
--- /dev/null
+++ b/Tools/make_index
@@ -0,0 +1,117 @@
+#!/usr/bin/perl
+#
+# $Id$
+#
+# Written to speed-up INDEX file generation.  The new scheme
+# basically visits each port once and writes out each port's
+# build-depends and run-depends as a list of directories.  This
+# script goes back in and maps the directories back to pkgnames,
+# fixes up the build-depends and run-depends list, and writes
+# out the new INDEX file.
+
+# Helper function to map a directory to a pkgname.
+sub by_path {
+    my $name = shift;
+
+  # If a direct mapping exists, then use it.
+    return $by_path{$name} if (defined $by_path{$name});
+
+  # Make sure we have /usr/ports at the beginning.
+    $name =~ s!^$pwd!/usr/ports!o;
+    return $by_path{$name} if (defined $by_path{$name});
+
+  # Collapse all the '..' sequences.
+    my @f = split('/', $name), @p = ();
+    foreach (@f) { (/\.\./) ? pop(@p) : push(@p, $_); }
+    return $by_path{join('/', @p)};
+}
+
+# This routine replaces what used to be the time-consuming
+# recursive 'depends-list' and 'package-depends' targets.
+sub recurse {
+    my $pkg = shift(@_);
+    return if $pkg->{checked};
+
+  # build-depends = build-depends + recursive list of run-depends
+  #     for each build-depends
+    my @deps = ();
+    foreach $name (@{$pkg->{bdep}}) {
+        recurse($index{$name});
+	push(@deps, @{$index{$name}->{rdep}});
+    }    
+    $pkg->{bdep} = uniqify(@{$pkg->{bdep}}, @deps);
+
+  # same as above except for run-depends this time
+    @deps = ();
+    foreach $name (@{$pkg->{rdep}}) {
+	recurse($index{$name});
+	push(@deps, @{$index{$name}->{rdep}});
+    }
+    $pkg->{rdep} = uniqify(@{$pkg->{rdep}}, @deps);
+    $pkg->{checked} = 1;
+}
+
+# Given one or more lists as arguments return the set
+# of unique elements among them.
+sub uniqify {
+    my %seen = ();
+    my @unique = grep {! $seen{$_}++} (@_);
+    return \@unique;
+}
+
+# Save where we are so that we can map all directories formed
+# from ${PORTSDIR} to their canonical location '/usr/ports/...'.
+chomp($pwd = `pwd`);
+
+# Read each line of output generated by the 'index' target.
+while (<>) {
+    chomp;
+    my @f = split(/\|/);
+
+  # Force to canonical form.
+    $f[1] =~ s!^$pwd!/usr/ports!o;
+    $f[4] =~ s!^$pwd!/usr/ports!o;
+
+  # Save directory -> pkgname relationship.
+  # Note: $f[0] gets clobbered by the splice below so we'll save
+  # it to a new $name first.
+    $by_path{$f[1]} = $name = $f[0];
+
+  # Create a hash table of the infomation we need about this port.
+    my $pkg = {
+	'bdep'		=> [split(/ /, $f[7])],
+	'rdep'		=> [split(/ /, $f[8])],
+	'text'		=> join('|', splice(@f, 0, 7))
+    };
+    $index{$name} = $pkg;
+
+  # This is a cheap way of preserving the order of the entries.
+    push(@names, $name);
+}
+
+# For each port perform the mapping between directory and pkgnames.
+foreach $name (keys %index) {
+    my $pkg = $index{$name};
+  # first the build dependencies
+    if (@{$pkg->{bdep}}) {
+	my @bdep = map { by_path($_) } @{$pkg->{bdep}};
+	$pkg->{bdep} = \@bdep;
+    }
+  # and now the run dependencies
+    if (@{$pkg->{rdep}}) {
+	my @rdep = map { by_path($_) } @{$pkg->{rdep}};
+	$pkg->{rdep} = \@rdep;
+    }
+}
+
+# With all that done we're finally ready to write out the new
+# INDEX file one port at a time.
+foreach $name (@names) {
+    my $pkg = $index{$name};
+    recurse($pkg);
+    print "$pkg->{text}|";
+    print join(' ', sort(@{$pkg->{bdep}})) if @{$pkg->{bdep}};
+    print "|";
+    print join(' ', sort(@{$pkg->{rdep}})) if @{$pkg->{rdep}};
+    print "\n";
+}