git-shortlog.perl

   1 #!/usr/bin/perl -w
   2
   3 use strict;
   4 use Getopt::Std;
   5 use File::Basename qw(basename dirname);
   6
   7 our ($opt_h, $opt_n, $opt_s);
   8 getopts('hns');
   9
  10 $opt_h && usage();
  11
  12 sub usage {
  13         print STDERR "Usage: ${\basename $0} [-h] [-n] [-s] < <log_data>\n";
  14         exit(1);
  15 }
  16
  17 my (%mailmap);
  18 my (%email);
  19 my (%map);
  20 my $pstate = 1;
  21 my $n_records = 0;
  22 my $n_output = 0;
  23
  24 sub shortlog_entry($$) {
  25         my ($name, $desc) = @_;
  26         my $key = $name;
  27
  28         $desc =~ s#/pub/scm/linux/kernel/git/#/.../#g;
  29         $desc =~ s#\[PATCH\] ##g;
  30
  31         # store description in array, in email->{desc list} map
  32         if (exists $map{$key}) {
  33                 # grab ref
  34                 my $obj = $map{$key};
  35
  36                 # add desc to array
  37                 push(@$obj, $desc);
  38         } else {
  39                 # create new array, containing 1 item
  40                 my @arr = ($desc);
  41
  42                 # store ref to array
  43                 $map{$key} = \@arr;
  44         }
  45 }
  46
  47 # sort comparison function
  48 sub by_name($$) {
  49         my ($a, $b) = @_;
  50
  51         uc($a) cmp uc($b);
  52 }
  53 sub by_nbentries($$) {
  54         my ($a, $b) = @_;
  55         my $a_entries = $map{$a};
  56         my $b_entries = $map{$b};
  57
  58         @$b_entries - @$a_entries || by_name $a, $b;
  59 }
  60
  61 my $sort_method = $opt_n ? \&by_nbentries : \&by_name;
  62
  63 sub summary_output {
  64         my ($obj, $num, $key);
  65
  66         foreach $key (sort $sort_method keys %map) {
  67                 $obj = $map{$key};
  68                 $num = @$obj;
  69                 printf "%s: %u\n", $key, $num;
  70                 $n_output += $num;
  71         }
  72 }
  73
  74 sub shortlog_output {
  75         my ($obj, $num, $key, $desc);
  76
  77         foreach $key (sort $sort_method keys %map) {
  78                 $obj = $map{$key};
  79                 $num = @$obj;
  80
  81                 # output author
  82                 printf "%s (%u):\n", $key, $num;
  83
  84                 # output author's 1-line summaries
  85                 foreach $desc (reverse @$obj) {
  86                         print "  $desc\n";
  87                         $n_output++;
  88                 }
  89
  90                 # blank line separating author from next author
  91                 print "\n";
  92         }
  93 }
  94
  95 sub changelog_input {
  96         my ($author, $desc);
  97
  98         while (<>) {
  99                 # get author and email
 100                 if ($pstate == 1) {
 101                         my ($email);
 102
 103                         next unless /^[Aa]uthor:?\s*(.*?)\s*<(.*)>/;
 104
 105                         $n_records++;
 106
 107                         $author = $1;
 108                         $email = $2;
 109                         $desc = undef;
 110
 111                         # cset author fixups
 112                         if (exists $mailmap{$email}) {
 113                                 $author = $mailmap{$email};
 114                         } elsif (exists $mailmap{$author}) {
 115                                 $author = $mailmap{$author};
 116                         } elsif (!$author) {
 117                                 $author = $email;
 118                         }
 119                         $email{$author}{$email}++;
 120                         $pstate++;
 121                 }
 122
 123                 # skip to blank line
 124                 elsif ($pstate == 2) {
 125                         next unless /^\s*$/;
 126                         $pstate++;
 127                 }
 128
 129                 # skip to non-blank line
 130                 elsif ($pstate == 3) {
 131                         next unless /^\s*?(.*)/;
 132
 133                         # skip lines that are obviously not
 134                         # a 1-line cset description
 135                         next if /^\s*From: /;
 136
 137                         chomp;
 138                         $desc = $1;
 139
 140                         &shortlog_entry($author, $desc);
 141
 142                         $pstate = 1;
 143                 }
 144
 145                 else {
 146                         die "invalid parse state $pstate";
 147                 }
 148         }
 149 }
 150
 151 sub read_mailmap {
 152         my ($fh, $mailmap) = @_;
 153         while (<$fh>) {
 154                 chomp;
 155                 if (/^([^#].*?)\s*<(.*)>/) {
 156                         $mailmap->{$2} = $1;
 157                 }
 158         }
 159 }
 160
 161 sub setup_mailmap {
 162         read_mailmap(\*DATA, \%mailmap);
 163         if (-f '.mailmap') {
 164                 my $fh = undef;
 165                 open $fh, '<', '.mailmap';
 166                 read_mailmap($fh, \%mailmap);
 167                 close $fh;
 168         }
 169 }
 170
 171 sub finalize {
 172         #print "\n$n_records records parsed.\n";
 173
 174         if ($n_records != $n_output) {
 175                 die "parse error: input records != output records\n";
 176         }
 177         if (0) {
 178                 for my $author (sort keys %email) {
 179                         my $e = $email{$author};
 180                         for my $email (sort keys %$e) {
 181                                 print STDERR "$author <$email>\n";
 182                         }
 183                 }
 184         }
 185 }
 186
 187 &setup_mailmap;
 188 &changelog_input;
 189 $opt_s ? &summary_output : &shortlog_output;
 190 &finalize;
 191 exit(0);
 192
 193
 194 __DATA__
 195 #
 196 # Even with git, we don't always have name translations.
 197 # So have an email->real name table to translate the
 198 # (hopefully few) missing names
 199 #
 200 Adrian Bunk <bunk@stusta.de>
 201 Andreas Herrmann <aherrman@de.ibm.com>
 202 Andrew Morton <akpm@osdl.org>
 203 Andrew Vasquez <andrew.vasquez@qlogic.com>
 204 Christoph Hellwig <hch@lst.de>
 205 Corey Minyard <minyard@acm.org>
 206 David Woodhouse <dwmw2@shinybook.infradead.org>
 207 Domen Puncer <domen@coderock.org>
 208 Douglas Gilbert <dougg@torque.net>
 209 Ed L Cashin <ecashin@coraid.com>
 210 Evgeniy Polyakov <johnpol@2ka.mipt.ru>
 211 Felix Moeller <felix@derklecks.de>
 212 Frank Zago <fzago@systemfabricworks.com>
 213 Greg Kroah-Hartman <gregkh@suse.de>
 214 James Bottomley <jejb@mulgrave.(none)>
 215 James Bottomley <jejb@titanic.il.steeleye.com>
 216 Jeff Garzik <jgarzik@pretzel.yyz.us>
 217 Jens Axboe <axboe@suse.de>
 218 Kay Sievers <kay.sievers@vrfy.org>
 219 Mitesh shah <mshah@teja.com>
 220 Morten Welinder <terra@gnome.org>
 221 Morten Welinder <welinder@anemone.rentec.com>
 222 Morten Welinder <welinder@darter.rentec.com>
 223 Morten Welinder <welinder@troll.com>
 224 Nguyen Anh Quynh <aquynh@gmail.com>
 225 Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
 226 Peter A Jonsson <pj@ludd.ltu.se>
 227 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
 228 Rudolf Marek <R.Marek@sh.cvut.cz>
 229 Rui Saraiva <rmps@joel.ist.utl.pt>
 230 Sachin P Sant <ssant@in.ibm.com>
 231 Santtu Hyrkk\e,Av\e(B <santtu.hyrkko@gmail.com>
 232 Simon Kelley <simon@thekelleys.org.uk>
 233 Tejun Heo <htejun@gmail.com>
 234 Tony Luck <tony.luck@intel.com>