Transforming a git repository into a co-modification graph

gitlog2graph.pl

#!/usr/bin/perl

# Converts a gitlog file to a list of edges in an undirected weighted graph
# Author: Aron Lurie
# Date: 2018-08-15
# Usage: git log --pretty=oneline --name-only | perl git.pl > mylogfile.txt

my @files = ();
my %adjacent = ();
while (<STDIN>) {
	chomp;
	if ($_ =~ /[a-f0-9]{40} /) {
		if (@files) {
			for my $filename (@files) {
				for my $other_filename (@files) {
					next if ($filename > $other_filename);
					$adjacent{$filename}{$other_filename}++;
				}
			}
		}
		@files = ();
	} else {
		push @files,  $_;
	}
}

for my $filename (keys %adjacent) {
	for my $other_filename (keys %{ $adjacent{$filename} }) {
		my $conn = $adjacent{$filename}{$other_filename};
		print join(",", ($filename, $other_filename, 1000/$conn)), "\n";
	}
}