git-commit-vandalism/git-shortlog

#!/usr/bin/perl -w

use strict;

#
# Even with git, we don't always have name translations.
# So have an email->real name table to translate the
# (hopefully few) missing names
#
my %mailmap = (
	'R.Marek@sh.cvut.cz' => 'Rudolf Marek',
	'Ralf.Wildenhues@gmx.de' => 'Ralf Wildenhues',
	'aherrman@de.ibm.com' => 'Andreas Herrmann',
	'akpm@osdl.org' => 'Andrew Morton',
	'andrew.vasquez@qlogic.com' => 'Andrew Vasquez',
	'aquynh@gmail.com' => 'Nguyen Anh Quynh',
	'axboe@suse.de' => 'Jens Axboe',
	'blaisorblade@yahoo.it' => 'Paolo \'Blaisorblade\' Giarrusso',
	'bunk@stusta.de' => 'Adrian Bunk',
	'domen@coderock.org' => 'Domen Puncer',
	'dougg@torque.net' => 'Douglas Gilbert',
	'dwmw2@shinybook.infradead.org' => 'David Woodhouse',
	'ecashin@coraid.com' => 'Ed L Cashin',
	'felix@derklecks.de' => 'Felix Moeller',
	'fzago@systemfabricworks.com' => 'Frank Zago',
	'gregkh@suse.de' => 'Greg Kroah-Hartman',
	'hch@lst.de' => 'Christoph Hellwig',
	'htejun@gmail.com' => 'Tejun Heo',
	'jejb@mulgrave.(none)' => 'James Bottomley',
	'jejb@titanic.il.steeleye.com' => 'James Bottomley',
	'jgarzik@pretzel.yyz.us' => 'Jeff Garzik',
	'johnpol@2ka.mipt.ru' => 'Evgeniy Polyakov',
	'kay.sievers@vrfy.org' => 'Kay Sievers',
	'minyard@acm.org' => 'Corey Minyard',
	'mshah@teja.com' => 'Mitesh shah',
	'pj@ludd.ltu.se' => 'Peter A Jonsson',
	'rmps@joel.ist.utl.pt' => 'Rui Saraiva',
	'santtu.hyrkko@gmail.com' => 'Santtu Hyrkkö',
	'simon@thekelleys.org.uk' => 'Simon Kelley',
	'ssant@in.ibm.com' => 'Sachin P Sant',
	'terra@gnome.org' => 'Morten Welinder',
	'tony.luck@intel.com' => 'Tony Luck',
	'welinder@anemone.rentec.com' => 'Morten Welinder',
	'welinder@darter.rentec.com' => 'Morten Welinder',
	'welinder@troll.com' => 'Morten Welinder',
);

my (%map);
my $pstate = 1;
my $n_records = 0;
my $n_output = 0;


sub shortlog_entry($$) {
	my ($name, $desc) = @_;
	my $key = $name;

	$desc =~ s#/pub/scm/linux/kernel/git/#/.../#g;
	$desc =~ s#\[PATCH\] ##g;

	# store description in array, in email->{desc list} map
	if (exists $map{$key}) {
		# grab ref
		my $obj = $map{$key};

		# add desc to array
		push(@$obj, $desc);
	} else {
		# create new array, containing 1 item
		my @arr = ($desc);

		# store ref to array
		$map{$key} = \@arr;
	}
}

# sort comparison function
sub by_name($$) {
	my ($a, $b) = @_;

	uc($a) cmp uc($b);
}

sub shortlog_output {
	my ($obj, $key, $desc);

	foreach $key (sort by_name keys %map) {
		# output author
		printf "%s:\n", $key;

		# output author's 1-line summaries
		$obj = $map{$key};
		foreach $desc (@$obj) {
			print "  $desc\n";
			$n_output++;
		}

		# blank line separating author from next author
		print "\n";
	}
}

sub changelog_input {
	my ($author, $desc);

	while (<>) {
		# get author and email
		if ($pstate == 1) {
			my ($email);

			next unless /^Author: (.*)<(.*)>.*$/;
	
			$n_records++;
	
			$author = $1;
			$email = $2;
			$desc = undef;

			# trim trailing whitespace.
			# why doesn't chomp work?
			while ($author && ($author =~ /\s$/)) {
				chop $author;
			}
	
			# cset author fixups
			if (exists $mailmap{$email}) {
				$author = $mailmap{$email};
			} elsif (exists $mailmap{$author}) {
				$author = $mailmap{$author};
			} elsif ((!$author) || ($author eq "")) {
				$author = $email;
			}
	
			$pstate++;
		}
	
		# skip to blank line
		elsif ($pstate == 2) {
			next unless /^\s*$/;
			$pstate++;
		}
	
		# skip to non-blank line
		elsif ($pstate == 3) {
			next unless /^\s*(\S.*)$/;

			# skip lines that are obviously not
			# a 1-line cset description
			next if /^\s*From: /;

			chomp;
			$desc = $1;
	
			&shortlog_entry($author, $desc);
	
			$pstate = 1;
		}
	
		else {
			die "invalid parse state $pstate";
		}
	}
}

sub finalize {
	#print "\n$n_records records parsed.\n";

	if ($n_records != $n_output) {
		die "parse error: input records != output records\n";
	}
}

&changelog_input;
&shortlog_output;
&finalize;
exit(0);