fix handling of repos w/o titles

[code/myrepos.git] / webcheckout

#!/usr/bin/perl

=head1 NAME

webcheckout - check out repositories referenced on a web page

=head1 SYNOPSIS

B<webcheckout> [options] url [destdir]

=head1 DESCRIPTION

B<webcheckout> downloads an url and parses it, looking for version control 
repositories referenced by the page. It checks out each repository into
a subdirectory of the current directory, using whatever VCS program is
appropriate for that repository (git, svn, etc).

The information about the repositories is embedded in the web page using
the rel=vcs microformat, which is documented at
<http://kitenet.net/~joey/rfc/rel-vcs/>.

If the optional destdir parameter is specified, VCS programs will be asked
to check out repositories into that directory. If there are multiple
repositories to check out, each will be checked out into a separate
subdirectory of the destdir.

=head1 OPTIONS

=over 4

=item -a, --auth

Prefer authenticated repositories. By default, webcheckout will use
anonymous repositories when possible. If you have an account that
allows you to use authenticated repositories, you might want to use this
option.

=item -n

Do not actually check anything out, just print out the commands that would
be run to check out the repositories.

=item -q

Quiet mode. Do not print out the commands being run. (The VCS commands
may still be noisy however.)

=back

=head1 AUTHOR

Copyright 2009 Joey Hess <joey@kitenet.net>

Licensed under the GNU GPL version 2 or higher.

This program is included in mr <http://kitenet.net/~joey/code/mr/>

=cut

use LWP::Simple;
use HTML::Parser;
use Getopt::Long;
use warnings;
use strict;

# What to download.
my $url;

# Controls whether to print what is being done.
my $quiet=0;

# Controls whether to actually check anything out.
my $noact=0;

# Controls whether to perfer repos that use authentication.
my $want_auth=0;

# Controls where to check out to. If not set, the vcs is allowed to
# decide.
my $destdir;

# how to perform checkouts
my %handlers=(
        git => sub { doit("git", "clone", shift, $destdir) },
        svn => sub { doit("svn", "checkout", shift, $destdir) },
        bzr => sub { doit("bzr", "branch", shift, $destdir) },
);

# Regexps matching urls that are used for anonymous
# repository checkouts. The order is significant:
# urls matching earlier in the list are preferred over
# those matching later.
my @anon_urls=(
        qr/^git:\/\//i,
        qr/^bzr:\/\//i,
        qr/^svn:\/\//i,
        qr/^http:\/\//i, # generally the worst transport
);

sub getopts {
        Getopt::Long::Configure("bundling", "no_permute");
        my $result=GetOptions(
                "q|quiet" => \$quiet,
                "n|noact" => \$noact,
                "a|auth", => \$want_auth,
        );
        if (! $result || @ARGV < 1) {
                die "usage: webcheckout [options] url [destdir]\n";
        }

        $url=shift @ARGV;
        $destdir=shift @ARGV;

        if ($noact) {
                $quiet=0;
        }
}

sub doit {
        my @args=grep { defined } @_;
        print join(" ", @args)."\n" unless $quiet;
        return 0 if $noact;
        return system(@args);
}

# Is repo a better than repo b?
sub better {
        my ($a, $b)=@_;

        my $firstanon=$b;
        foreach my $r (@anon_urls) {
                if ($a->{href} =~ /$r/) {
                        $firstanon=$a;
                        last;
                }
                elsif ($b->{href} =~ /$r/) {
                        $firstanon=$b;
                        last;
                }
        }

        if ($want_auth) {
                return $firstanon != $a;
        }
        else {
                return $firstanon == $a;
        }
}

# Eliminate duplicate repositories from list.
# Duplicate repositories have the same title, or the same href.
sub dedup {
        my %seenhref;
        my %bytitle;
        my @others;
        foreach my $repo (@_) {
                if (exists $repo->{title} &&
                    length $repo->{title}) {
                        if (exists $bytitle{$repo->{title}}) {
                                my $other=$bytitle{$repo->{title}};
                                next unless better($repo, $other);
                                delete $bytitle{$other->{title}}
                        }

                        if (! $seenhref{$repo->{href}}++) {
                                $bytitle{$repo->{title}}=$repo;
                        }
                }
                else {
                        push @others, $repo;
                }
        }

        return values %bytitle, @others;
}

sub parse {
        my $page=shift;

        my @ret;
        my $parser=HTML::Parser->new(api_version => 3);
        $parser->handler(start => sub {
                my $tagname=shift;
                my $attr=shift;
                return if lc $tagname ne 'link';
                return if ! exists $attr->{rel} || lc $attr->{rel} ne 'vcs';
                return if ! exists $attr->{href} || ! length $attr->{href};
                return if ! exists $attr->{type} || ! length $attr->{type};
                push @ret, $attr;
        }, "tagname, attr");
        $parser->parse($page);
        $parser->eof;

        return @ret;
}

getopts();

my $page=get($url);
if (! defined $page) {
        die "failed to download $url\n";
}

my @repos=dedup(parse($page));
if (! @repos) {
        die "no repositories found on $url\n";
}

if (defined $destdir && @repos > 1) {
        # create subdirs of $destdir for the multiple repos
        mkdir($destdir);
        chdir($destdir) || die "failed to chdir to $destdir: $!";
        $destdir=undef;
}

my $errors=0;
foreach my $repo (@repos) {
        my $handler=$handlers{$repo->{type}};
        if ($handler) {
                if ($handler->($repo->{href}) != 0) {
                        print STDERR "failed to checkout ".$repo->{href}."\n";
                        $errors++;
                }
        }
        else {
                print STDERR "unknown repository type ".$repo->{type}.
                        " for ".$repo->{href}."\n";
                $errors++;
        }
}
exit($errors > 0);

#use Data::Dumper;
#print Dumper(\@repos);