mr: Make check() even more pedantic

[code/myrepos.git] / webcheckout

diff --git a/webcheckout b/webcheckout
index a6ae94fd09de897ac9498f99e28309997ebefb59..e48d9de34dd9ed6ecd2fe6d8fb824711edc31e54 100755 (executable)
--- a/webcheckout
+++ b/webcheckout
@@ -16,7 +16,7 @@ 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
+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
@@ -35,25 +35,33 @@ 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
+=item --no-act, -n
 
 Do not actually check anything out, just print out the commands that would
 be run to check out the repositories.
 
-=item -q
+=item --quiet, -q
 
 Quiet mode. Do not print out the commands being run. (The VCS commands
 may still be noisy however.)
 
 =back
 
+=head1 PREREQUISITES
+
+To use this program you will need lots of VCS programs installed,
+obviously. It also depends on the perl LWP and HTML::Parser modules.
+
+If the perl URI module is installed, webcheckout can heuristically guess
+what you mean by partial URLs, such as "kitenet.net/~joey"'
+
 =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/>
+This program is included in myrepos <http://myrepos.branchable.com/>
 
 =cut
 
@@ -75,7 +83,7 @@ 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
+# Controls where to check out to. If not set, the VCS is allowed to
 # decide.
 my $destdir;
 
@@ -111,6 +119,11 @@ sub getopts {
        $url=shift @ARGV;
        $destdir=shift @ARGV;
 
+       eval q{use URI::Heuristic};
+       if (! $@) {
+               $url=URI::Heuristic::uf_uristr($url);
+       }
+
        if ($noact) {
                $quiet=0;
        }
@@ -187,9 +200,10 @@ sub parse {
        $parser->handler(start => sub {
                my $tagname=shift;
                my $attr=shift;
-               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};
+               return if ! exists $attr->{rel} || $attr->{rel} !~ /^vcs-(.+)/i;
+               $attr->{type}=lc($1);
 
                # need to collect the body of the <a> tag if there is no title
                if ($tagname eq "a" && ! exists $attr->{title}) {