support <a> elements

[code/myrepos.git] / webcheckout

diff --git a/webcheckout b/webcheckout
index 6df790dffbd1f7be3eefb2e30a1cfc775b8af2d4..a6ae94fd09de897ac9498f99e28309997ebefb59 100755 (executable)
--- a/webcheckout
+++ b/webcheckout
@@ -182,15 +182,37 @@ sub parse {
 
        my @ret;
        my $parser=HTML::Parser->new(api_version => 3);
+       my $abody=undef;
+       my $aref=undef;
        $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 $attr->{type};
+
+               # need to collect the body of the <a> tag if there is no title
+               if ($tagname eq "a" && ! exists $attr->{title}) {
+                       $abody="";
+                       $aref=$attr;
+               }
+
                push @ret, $attr;
        }, "tagname, attr");
+       $parser->handler(text => sub {
+               if (defined $aref) {
+                       $abody.=join(" ", @_);
+               }
+       }, "text");
+       $parser->handler(end => sub {
+               my $tagname=shift;
+               if ($tagname eq "a" && defined $aref) {
+                       $aref->{title}=$abody;
+                       $aref=undef;
+                       $abody=undef;
+               }
+       }, "tagname");
+       $parser->report_tags(qw{link a});
        $parser->parse($page);
        $parser->eof;
 
@@ -209,6 +231,10 @@ if (! @repos) {
        die "no repositories found on $url\n";
 }
 
+#use Data::Dumper;
+#print Dumper(\@repos);
+#exit;
+
 if (defined $destdir && @repos > 1) {
        # create subdirs of $destdir for the multiple repos
        if (! $noact) {
@@ -234,6 +260,3 @@ foreach my $repo (@repos) {
        }
 }
 exit($errors > 0);
-
-#use Data::Dumper;
-#print Dumper(\@repos);