Set User-Agent to "mr" when downloading the mrconfig file with mr bootstrap. Closes...

[code/myrepos.git] / webcheckout

diff --git a/webcheckout b/webcheckout
index a6ae94fd09de897ac9498f99e28309997ebefb59..0c93eefcd8d10e41d9ba649438a353debe91f6b8 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
 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
 <http://kitenet.net/~joey/rfc/rel-vcs/>.
 
 If the optional destdir parameter is specified, VCS programs will be asked


@@ -35,18 +35,26 @@ anonymous repositories when possible. If you have an account that
 allows you to use authenticated repositories, you might want to use this
 option.
 
 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.
 
 
 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
 
 
 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>
 =head1 AUTHOR
 
 Copyright 2009 Joey Hess <joey@kitenet.net>


@@ -75,7 +83,7 @@ my $noact=0;
 # Controls whether to perfer repos that use authentication.
 my $want_auth=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;
 
 # decide.
 my $destdir;
 


@@ -111,6 +119,11 @@ sub getopts {
        $url=shift @ARGV;
        $destdir=shift @ARGV;
 
        $url=shift @ARGV;
        $destdir=shift @ARGV;
 

+       eval q{use URI::Heuristic};
+       if (! $@) {
+               $url=URI::Heuristic::uf_uristr($url);
+       }
+

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


@@ -187,9 +200,10 @@ sub parse {
        $parser->handler(start => sub {
                my $tagname=shift;
                my $attr=shift;
        $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->{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}) {
 
                # need to collect the body of the <a> tag if there is no title
                if ($tagname eq "a" && ! exists $attr->{title}) {