Add a tool that I've written to help with finding what other

files/revisions were changed with a particular commit.

3 files changed, 165 insertions, 0 deletions

diff --git a/tools/tools/README b/tools/tools/README
index 0d47beb92cf0..6bdc8c4f0897 100644
--- a/tools/tools/README
+++ b/tools/tools/README
@@ -7,6 +7,8 @@ other categories.
 
 Please make a subdir per program, and add a brief description to this file.
 
+commitsdb	A tool for reconstructing commit history using md5
+		checksums of the commit logs.
 editing		Editor modes and the like to help editing FreeBSD code.
 epfe 		Extract printing filter examples from printing.sgml.
 html-mv         Rename HTML generated filenames to human readable filenames.
diff --git a/tools/tools/commitsdb/make_commit_db b/tools/tools/commitsdb/make_commit_db
new file mode 100644
index 000000000000..7b0fee780c78
--- /dev/null
+++ b/tools/tools/commitsdb/make_commit_db
@@ -0,0 +1,102 @@
+#!/usr/bin/perl -w
+
+# $FreeBSD$
+
+# This script walks the tree from the current directory
+# and spits out a database generated by md5'ing the cvs log
+# messages of each revision of every file in the tree.
+
+use strict;
+use Digest::MD5 qw(md5_hex);
+
+my $dbname = "commitsdb";
+open DB, "> $dbname" or die "$!\n";
+
+# Extract all the logs for the current directory.
+my @dirs = ".";
+while (@dirs) {
+	my $dir = shift @dirs;
+	my %logs;
+
+	opendir DIR, $dir or die $!;
+	foreach (grep { /[^\.]/ } readdir DIR) {
+		my $filename = "$dir/$_";
+		if (-f $filename) {
+			my %loghash = parse_log_message($filename);
+			next unless %loghash;
+
+			$logs{$filename} = {%loghash};
+		} elsif (-d $_) {
+			next if /^CVS$/;
+			push @dirs, $_;
+		}
+	}
+	close DIR;
+
+	# Product a database of the commits
+	foreach my $f (keys %logs) {
+		my $file = $logs{$f};
+		foreach my $rev (keys %$file) {
+			my $hash = $$file{$rev};
+
+			print DB "$f $rev $hash\n";
+		}
+	}
+
+	print "\r" . " " x 30 . "\r$dir";
+}
+print "\n";
+
+close DB;
+
+
+
+##################################################
+# Run a cvs log on a file and return a parse entry.
+##################################################
+sub parse_log_message {
+	my $file = shift;
+
+	# Get a log of the file.
+	open LOG, "cvs -R log $file |" or die $!;
+	my @log = <LOG>;
+	my $log = join "", @log;
+	close LOG;
+
+	# Split the log into revisions.
+	my @entries = split /----------------------------\n/, $log;
+
+	# Throw away the first entry.
+	shift @entries;
+
+	# Record the hash of the message against the revision.
+	my %loghash = ();
+	foreach my $e (@entries) {
+		# Get the revision number
+		$e =~ s/^revision\s*(\S*)\n//s;
+		my $rev = $1;
+
+		# Strip off any other headers.
+		while ($e =~ s/^(date|branches):[^\n]*\n//sg) {
+		};
+
+		my $hash = string_to_hash($e);
+		$loghash{$rev} = $hash;
+	}
+
+	return %loghash;
+}
+
+
+##################################################
+# Convert a log message into an md5 checksum.
+##################################################
+sub string_to_hash {
+	my $logmsg = shift;
+
+	return md5_hex($logmsg);
+}
+
+
+
+#end
diff --git a/tools/tools/commitsdb/query_commit_db b/tools/tools/commitsdb/query_commit_db
new file mode 100644
index 000000000000..f753f327b898
--- /dev/null
+++ b/tools/tools/commitsdb/query_commit_db
@@ -0,0 +1,61 @@
+#!/usr/bin/perl -w
+
+# $FreeBSD$
+
+# This script takes a filename and revision number as arguments
+# and spits out a list of other files and their revisions that share
+# the same log message.  This is done by referring to the database
+# previously written by running make_commit_db.
+
+use strict;
+use Digest::MD5 qw(md5_hex);
+
+my $dbname = "commitsdb";
+
+# Take the filename and revision number from the command line.
+my ($file, $revision) = (shift, shift);
+
+# Find the checksum of the named revision.
+my %possible_files;
+open DB, "< $dbname" or die "$!\n";
+my $cksum;
+while (<DB>) {
+	chomp;
+	my ($name, $rev, $hash) = split;
+	$name =~ s/^\.\///g;
+
+	$possible_files{$name} = 1 if $file !~ /\// && $name =~ /^.*\/$file/;
+
+	next unless $name eq $file and $rev eq $revision;
+	$cksum = $hash;
+}
+close DB;
+
+# Handle the fall-out if the file/revision wasn't matched.
+unless ($cksum) {
+	if (%possible_files) {
+		print "Couldn't find the file. Maybe you meant:\n";
+		foreach (sort keys %possible_files) {
+			print "\t$_\n";
+		}
+	}
+	die "Can't find $file rev $revision in database\n";
+}
+
+
+# Look for similar revisions.
+my @results;
+open DB, "< $dbname" or die "$!\n";
+while (<DB>) {
+	chomp;
+	my ($name, $rev, $hash) = split;
+
+	next unless $hash eq $cksum;
+
+	push @results, "$name $rev";
+}
+close DB;
+
+print map { "$_\n" } sort @results;
+
+#end