path: root/tools/git-notify

#!/usr/bin/perl -w
#
# Tool to send git commit notifications
#
# Copyright 2005 Alexandre Julliard
# Copyright 2009 Nagios Plugins Development Team
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License as
# published by the Free Software Foundation; either version 2 of
# the License, or (at your option) any later version.
#
#
# This script is meant to be called from .git/hooks/post-receive.
#
# Usage: git-notify [options] [--] old-sha1 new-sha1 refname
#
#   -A        Omit the author name from the mail subject
#   -C        Show committer in the body if different from the author
#   -c name   Send CIA notifications under specified project name
#   -m addr   Send mail notifications to specified address
#   -n max    Set max number of individual mails to send
#   -r name   Set the git repository name
#   -s bytes  Set the maximum diff size in bytes (-1 for no limit)
#   -t file   Prevent duplicate notifications by saving state to this file
#   -U mask   Set the umask for creating the state file
#   -u url    Set the URL to the gitweb browser
#   -i branch If at least one -i is given, report only for specified branches
#   -x branch Exclude changes to the specified branch from reports
#   -X        Exclude merge commits
#   -z        Try to abbreviate the SHA1 name within gitweb URLs (unsafe)
#

use strict;
use Fcntl ':flock';
use Encode qw(encode decode);
use Cwd 'realpath';

sub git_config($);
sub get_repos_name();

# some parameters you may want to change

# set this to something that takes "-s"
my $mailer = "/usr/bin/mail";

# CIA notification address
my $cia_address = "cia\@cia.navi.cx";

# debug mode
my $debug = 0;

# configuration parameters

# omit the author from the mail subject (can be set with the -A option)
my $omit_author = git_config( "notify.omitauthor" );

# show the committer if different from the author (can be set with the -C option)
my $show_committer = git_config( "notify.showcommitter" );

# base URL of the gitweb repository browser (can be set with the -u option)
my $gitweb_url = git_config( "notify.baseurl" );

# abbreviate the SHA1 name within gitweb URLs (can be set with the -z option)
my $abbreviate_url = git_config( "notify.shorturls" );

# default repository name (can be changed with the -r option)
my $repos_name = git_config( "notify.repository" ) || get_repos_name();

# max size of diffs in bytes (can be changed with the -s option)
my $max_diff_size = git_config( "notify.maxdiff" ) || 10000;

# address for mail notices (can be set with -m option)
my $commitlist_address = git_config( "notify.mail" );

# project name for CIA notices (can be set with -c option)
my $cia_project_name = git_config( "notify.cia" );

# max number of individual notices before falling back to a single global notice (can be set with -n option)
my $max_individual_notices = git_config( "notify.maxnotices" ) || 100;

# branches to include
my @include_list = split /\s+/, git_config( "notify.include" ) || "";

# branches to exclude
my @exclude_list = split /\s+/, git_config( "notify.exclude" ) || "";

# the state file we use (can be set with the -t option)
my $state_file = git_config( "notify.statefile" );

# umask for creating the state file (can be set with -U option)
my $mode_mask = git_config( "notify.umask" ) || 002;

# Extra options to git rev-list
my @revlist_options;

sub usage()
{
    print "Usage: $0 [options] [--] old-sha1 new-sha1 refname\n";
    print "   -A        Omit the author name from the mail subject\n";
    print "   -C        Show committer in the body if different from the author\n";
    print "   -c name   Send CIA notifications under specified project name\n";
    print "   -m addr   Send mail notifications to specified address\n";
    print "   -n max    Set max number of individual mails to send\n";
    print "   -r name   Set the git repository name\n";
    print "   -s bytes  Set the maximum diff size in bytes (-1 for no limit)\n";
    print "   -t file   Prevent duplicate notifications by saving state to this file\n";
    print "   -U mask   Set the umask for creating the state file\n";
    print "   -u url    Set the URL to the gitweb browser\n";
    print "   -i branch If at least one -i is given, report only for specified branches\n";
    print "   -x branch Exclude changes to the specified branch from reports\n";
    print "   -X        Exclude merge commits\n";
    print "   -z        Try to abbreviate the SHA1 name within gitweb URLs (unsafe)\n";
    exit 1;
}

sub xml_escape($)
{
    my $str = shift;
    $str =~ s/&/&/g;
    $str =~ s/</&lt;/g;
    $str =~ s/>/&gt;/g;
    my @chars = unpack "U*", $str;
    $str = join "", map { ($_ > 127) ? sprintf "&#%u;", $_ : chr($_); } @chars;
    return $str;
}

# execute git-rev-list(1) with the given parameters and return the output
sub git_rev_list(@)
{
    my @args = @_;
    my $revlist = [];
    my $pid = open REVLIST, "-|";

    die "Cannot open pipe: $!" if not defined $pid;
    if (!$pid)
    {
        exec "git", "rev-list", @revlist_options, @args or die "Cannot execute rev-list: $!";
    }
    while (<REVLIST>)
    {
        chomp;
        die "Invalid commit: $_" if not /^[0-9a-f]{40}$/;
        push @$revlist, $_;
    }
    close REVLIST or die $! ? "Cannot execute rev-list: $!" : "rev-list exited with status: $?";
    return $revlist;
}

# append the given commit hashes to the state file
sub save_commits($)
{
    my $commits = shift;

    open STATE, ">>", $state_file or die "Cannot open $state_file: $!";
    flock STATE, LOCK_EX or die "Cannot lock $state_file";
    print STATE "$_\n" for @$commits;
    flock STATE, LOCK_UN or die "Cannot unlock $state_file";
    close STATE or die "Cannot close $state_file: $!";
}

# for the given range, return the new hashes (and append them to the state file)
sub get_new_commits($$)
{
    my ($old_sha1, $new_sha1) = @_;
    my ($seen, @args);
    my $newrevs = [];

    @args = ( "^$old_sha1" ) unless $old_sha1 eq '0' x 40;
    push @args, $new_sha1, @exclude_list;

    my $revlist = git_rev_list(@args);

    if (not defined $state_file or not -e $state_file)
    {
        save_commits(git_rev_list("--all", "--full-history")) if defined $state_file;
        return $revlist;
    }

    open STATE, $state_file or die "Cannot open $state_file: $!";
    flock STATE, LOCK_SH or die "Cannot lock $state_file";
    while (<STATE>)
    {
        chomp;
        die "Invalid commit: $_" if not /^[0-9a-f]{40}$/;
        $seen->{$_} = 1;
    }
    flock STATE, LOCK_UN or die "Cannot unlock $state_file";
    close STATE or die "Cannot close $state_file: $!";

    # FIXME: if another git-notify process reads the $state_file at *this*
    # point, that process might generate duplicates of our notifications.

    save_commits($revlist);

    foreach my $commit (@$revlist)
    {
        push @$newrevs, $commit unless $seen->{$commit};
    }
    return $newrevs;
}

# truncate the given string if it exceeds the specified number of characters
sub truncate_str($$)
{
    my ($str, $max) = @_;

    if (length($str) > $max)
    {
        $str = substr($str, 0, $max);
        $str =~ s/\s+\S+$//;
        $str .= " ...";
    }
    return $str;
}

# right-justify the left column of "left: right" elements, omit undefined elements
sub format_table(@)
{
    my @lines = @_;
    my @table;
    my $max = 0;

    foreach my $line (@lines)
    {
       next if not defined $line;
       my $pos = index($line, ":");

       $max = $pos if $pos > $max;
    }

    foreach my $line (@lines)
    {
       next if not defined $line;
       my ($left, $right) = split(/: */, $line, 2);

       push @table, (defined $left and defined $right)
           ? sprintf("%*s: %s", $max + 1, $left, $right)
           : $line;
    }
    return @table;
}

# format an integer date + timezone as string
# algorithm taken from git's date.c
sub format_date($$)
{
    my ($time,$tz) = @_;

    if ($tz < 0)
    {
        my $minutes = (-$tz / 100) * 60 + (-$tz % 100);
        $time -= $minutes * 60;
    }
    else
    {
        my $minutes = ($tz / 100) * 60 + ($tz % 100);
        $time += $minutes * 60;
    }
    return gmtime($time) . sprintf " %+05d", $tz;
}

# fetch a parameter from the git config file
sub git_config($)
{
    my ($param) = @_;

    open CONFIG, "-|" or exec "git", "config", $param;
    my $ret = <CONFIG>;
    chomp $ret if $ret;
    close CONFIG or $ret = undef;
    return $ret;
}

# parse command line options
sub parse_options()
{
    while (@ARGV && $ARGV[0] =~ /^-/)
    {
        my $arg = shift @ARGV;

        if ($arg eq '--') { last; }
        elsif ($arg eq '-A') { $omit_author = 1; }
        elsif ($arg eq '-C') { $show_committer = 1; }
        elsif ($arg eq '-c') { $cia_project_name = shift @ARGV; }
        elsif ($arg eq '-m') { $commitlist_address = shift @ARGV; }
        elsif ($arg eq '-n') { $max_individual_notices = shift @ARGV; }
        elsif ($arg eq '-r') { $repos_name = shift @ARGV; }
        elsif ($arg eq '-s') { $max_diff_size = shift @ARGV; }
        elsif ($arg eq '-t') { $state_file = shift @ARGV; }
        elsif ($arg eq '-U') { $mode_mask = shift @ARGV; }
        elsif ($arg eq '-u') { $gitweb_url = shift @ARGV; }
        elsif ($arg eq '-i') { push @include_list, shift @ARGV; }
        elsif ($arg eq '-x') { push @exclude_list, shift @ARGV; }
        elsif ($arg eq '-X') { push @revlist_options, "--no-merges"; }
        elsif ($arg eq '-z') { $abbreviate_url = 1; }
        elsif ($arg eq '-d') { $debug++; }
        else { usage(); }
    }
    if (@ARGV && $#ARGV != 2) { usage(); }
    @exclude_list = map { "^$_"; } @exclude_list;
}

# send an email notification
sub mail_notification($$$@)
{
    my ($name, $subject, $content_type, @text) = @_;
    $subject = encode("MIME-Q",$subject);
    if ($debug)
    {
        binmode STDOUT, ":utf8";
        print "---------------------\n";
        print "To: $name\n";
        print "Subject: $subject\n";
        print "Content-Type: $content_type\n";
        print "\n", join("\n", @text), "\n";
    }
    else
    {
        my $pid = open MAIL, "|-";
        return unless defined $pid;
        if (!$pid)
        {
            exec $mailer, "-s", $subject, "-a", "Content-Type: $content_type", $name or die "Cannot exec $mailer";
        }
        binmode MAIL, ":utf8";
        print MAIL join("\n", @text), "\n";
        close MAIL or warn $! ? "Cannot execute $mailer: $!" : "$mailer exited with status: $?";
    }
}

# get the default repository name
sub get_repos_name()
{
    my $dir = `git rev-parse --git-dir`;
    chomp $dir;
    my $repos = realpath($dir);
    $repos =~ s/(.*?)((\.git\/)?\.git)$/$1/;
    $repos =~ s/(.*)\/([^\/]+)\/?$/$2/;
    return $repos;
}

# extract the information from a commit or tag object and return a hash containing the various fields
sub get_object_info($)
{
    my $obj = shift;
    my %info = ();
    my @log = ();
    my $do_log = 0;

    $info{"encoding"} = "utf-8";

    open TYPE, "-|" or exec "git", "cat-file", "-t", $obj or die "cannot run git-cat-file";
    my $type = <TYPE>;
    chomp $type;
    close TYPE or die $! ? "Cannot execute cat-file: $!" : "cat-file exited with status: $?";

    open OBJ, "-|" or exec "git", "cat-file", $type, $obj or die "cannot run git-cat-file";
    while (<OBJ>)
    {
        chomp;
        if ($do_log)
        {
            last if /^-----BEGIN PGP SIGNATURE-----/;
            push @log, $_;
        }
        elsif (/^(author|committer|tagger) ((.*) (<.*>)) (\d+) ([+-]\d+)$/)
        {
            $info{$1} = $2;
            $info{$1 . "_name"} = $3;
            $info{$1 . "_email"} = $4;
            $info{$1 . "_date"} = $5;
            $info{$1 . "_tz"} = $6;
        }
        elsif (/^tag (.+)/)
        {
            $info{"tag"} = $1;
        }
        elsif (/^encoding (.+)/)
        {
            $info{"encoding"} = $1;
        }
        elsif (/^$/) { $do_log = 1; }
    }
    close OBJ or die $! ? "Cannot execute cat-file: $!" : "cat-file exited with status: $?";

    $info{"type"} = $type;
    $info{"log"} = \@log;
    return %info;
}

# send a ref change notice to a mailing list
sub send_ref_notice($$@)
{
    my ($ref, $action, @notice) = @_;
    my ($reftype, $refname) = ($ref =~ /^refs\/(head|tag)s\/(.+)/);

    $reftype =~ s/^head$/branch/;

    @notice = (format_table(
        "Module: $repos_name",
        ($reftype eq "tag" ? "Tag:" : "Branch:") . $refname,
        @notice,
        ($action ne "removed" and $gitweb_url)
            ? "URL: $gitweb_url/?a=shortlog;h=$ref" : undef),
        "",
        "The $refname $reftype has been $action.");

    mail_notification($commitlist_address, "$refname $reftype $action",
        "text/plain; charset=us-ascii", @notice);
}

# send a commit notice to a mailing list
sub send_commit_notice($$)
{
    my ($ref,$obj) = @_;
    my %info = get_object_info($obj);
    my @notice = ();
    my ($url,$subject,$obj_string);

    if ($gitweb_url)
    {
        if ($abbreviate_url)
        {
            open REVPARSE, "-|" or exec "git", "rev-parse", "--short", $obj or die "cannot exec git-rev-parse";
            $obj_string = <REVPARSE>;
            chomp $obj_string if defined $obj_string;
            close REVPARSE or die $! ? "Cannot execute rev-parse: $!" : "rev-parse exited with status: $?";
        }
        $obj_string = $obj if not defined $obj_string;
        $url = "$gitweb_url/?a=$info{type};h=$obj_string";
    }

    if ($info{"type"} eq "tag")
    {
        push @notice, format_table(
          "Module: $repos_name",
          "Branch: $ref",
          "Tag: $obj",
          "Tagger:" . $info{"tagger"},
          "Date:" . format_date($info{"tagger_date"},$info{"tagger_tz"}),
          $url ? "URL: $url" : undef),
          "",
          join "\n", @{$info{"log"}};

        $subject = "Tag " . $info{"tag"} . ": ";
        $subject .= $info{"tagger_name"} . ": " unless $omit_author;
    }
    else
    {
        push @notice, format_table(
          "Module: $repos_name",
          "Branch: $ref",
          "Commit: $obj",
          "Author:" . $info{"author"},
          $show_committer && $info{"committer"} ne $info{"author"} ? "Committer:" . $info{"committer"} : undef,
          "Date:" . format_date($info{"author_date"},$info{"author_tz"}),
          $url ? "URL: $url" : undef),
          "",
          @{$info{"log"}},
          "",
          "---",
          "";

        open STAT, "-|" or exec "git", "diff-tree", "--stat", "-M", "--no-commit-id", $obj or die "cannot exec git-diff-tree";
        push @notice, join("", <STAT>);
        close STAT or die $! ? "Cannot execute diff-tree: $!" : "diff-tree exited with status: $?";

        open DIFF, "-|" or exec "git", "diff-tree", "-p", "-M", "--no-commit-id", $obj or die "cannot exec git-diff-tree";
        my $diff = join("", <DIFF>);
        close DIFF or die $! ? "Cannot execute diff-tree: $!" : "diff-tree exited with status: $?";

        if (($max_diff_size == -1) || (length($diff) < $max_diff_size))
        {
            push @notice, $diff;
        }
        else
        {
            push @notice, "Diff: $gitweb_url/?a=commitdiff;h=$obj_string" if $gitweb_url;
        }
        $subject = $info{"author_name"} . ": " unless $omit_author;
    }

    $subject .= truncate_str(${$info{"log"}}[0],50);
    $_ = decode($info{"encoding"}, $_) for @notice;
    mail_notification($commitlist_address, $subject, "text/plain; charset=UTF-8", @notice);
}

# send a commit notice to the CIA server
sub send_cia_notice($$)
{
    my ($ref,$commit) = @_;
    my %info = get_object_info($commit);
    my @cia_text = ();

    return if $info{"type"} ne "commit";

    push @cia_text,
        "<message>",
        "  <generator>",
        "    <name>git-notify script for CIA</name>",
        "  </generator>",
        "  <source>",
        "    <project>" . xml_escape($cia_project_name) . "</project>",
        "    <module>" . xml_escape($repos_name) . "</module>",
        "    <branch>" . xml_escape($ref). "</branch>",
        "  </source>",
        "  <body>",
        "    <commit>",
        "      <revision>" . substr($commit,0,10) . "</revision>",
        "      <author>" . xml_escape($info{"author"}) . "</author>",
        "      <log>" . xml_escape(join "\n", @{$info{"log"}}) . "</log>",
        "      <files>";

    open COMMIT, "-|" or exec "git", "diff-tree", "--name-status", "-r", "-M", $commit or die "cannot run git-diff-tree";
    while (<COMMIT>)
    {
        chomp;
        if (/^([AMD])\t(.*)$/)
        {
            my ($action, $file) = ($1, $2);
            my %actions = ( "A" => "add", "M" => "modify", "D" => "remove" );
            next unless defined $actions{$action};
            push @cia_text, "        <file action=\"$actions{$action}\">" . xml_escape($file) . "</file>";
        }
        elsif (/^R\d+\t(.*)\t(.*)$/)
        {
            my ($old, $new) = ($1, $2);
            push @cia_text, "        <file action=\"rename\" to=\"" . xml_escape($new) . "\">" . xml_escape($old) . "</file>";
        }
    }
    close COMMIT or die $! ? "Cannot execute diff-tree: $!" : "diff-tree exited with status: $?";

    push @cia_text,
        "      </files>",
        $gitweb_url ? "      <url>" . xml_escape("$gitweb_url/?a=commit;h=$commit") . "</url>" : "",
        "    </commit>",
        "  </body>",
        "  <timestamp>" . $info{"author_date"} . "</timestamp>",
        "</message>";

    mail_notification($cia_address, "DeliverXML", "text/xml", @cia_text);
}

# send a global commit notice when there are too many commits for individual mails
sub send_global_notice($$$)
{
    my ($ref, $old_sha1, $new_sha1) = @_;
    my $notice = git_rev_list("--pretty", "^$old_sha1", "$new_sha1", @exclude_list);

    foreach my $rev (@$notice)
    {
        $rev =~ s/^commit /URL:    $gitweb_url\/?a=commit;h=/ if $gitweb_url;
    }

    mail_notification($commitlist_address, "New commits on branch $ref", "text/plain; charset=UTF-8", @$notice);
}

# send all the notices
sub send_all_notices($$$)
{
    my ($old_sha1, $new_sha1, $ref) = @_;
    my ($reftype, $refname, $action, @notice);

    return if ($ref =~ /^refs\/remotes\//
        or (@include_list && !grep {$_ eq $ref} @include_list));
    die "The name \"$ref\" doesn't sound like a local branch or tag"
        if not (($reftype, $refname) = ($ref =~ /^refs\/(head|tag)s\/(.+)/));

    if ($new_sha1 eq '0' x 40)
    {
        $action = "removed";
        @notice = ( "Old SHA1: $old_sha1" );
    }
    elsif ($old_sha1 eq '0' x 40)
    {
        $action = "created";
        @notice = ( "SHA1: $new_sha1" );
    }
    elsif ($reftype eq "tag")
    {
        $action = "updated";
        @notice = ( "Old SHA1: $old_sha1", "New SHA1: $new_sha1" );
    }
    elsif (not grep( $_ eq $old_sha1, @{ git_rev_list( $new_sha1, "--full-history" ) } ))
    {
        $action = "rewritten";
        @notice = ( "Old SHA1: $old_sha1", "New SHA1: $new_sha1" );
    }

    send_ref_notice( $ref, $action, @notice ) if ($commitlist_address and $action);

    unless ($reftype eq "tag" or $new_sha1 eq '0' x 40)
    {
        my $commits = get_new_commits ( $old_sha1, $new_sha1 );

        if (@$commits > $max_individual_notices)
        {
            send_global_notice( $refname, $old_sha1, $new_sha1 ) if $commitlist_address;
        }
        elsif (@$commits > 0)
        {
            foreach my $commit (@$commits)
            {
                send_commit_notice( $refname, $commit ) if $commitlist_address;
                send_cia_notice( $refname, $commit ) if $cia_project_name;
            }
        }
        elsif ($commitlist_address)
        {
            @notice = ( "Old SHA1: $old_sha1", "New SHA1: $new_sha1" );
            send_ref_notice( $ref, "modified", @notice );
        }
    }
}

parse_options();

umask( $mode_mask );

# append repository path to URL
$gitweb_url .= "/$repos_name.git" if $gitweb_url;

if (@ARGV)
{
    send_all_notices( $ARGV[0], $ARGV[1], $ARGV[2] );
}
else  # read them from stdin
{
    while (<>)
    {
        chomp;
        if (/^([0-9a-f]{40}) ([0-9a-f]{40}) (.*)$/) { send_all_notices( $1, $2, $3 ); }
    }
}

exit 0;