support <a> elements

[code/myrepos.git] / webcheckout

diff --git a/webcheckout b/webcheckout
index 4841feedb7511d4866a725b4e439d6fc8ec3327c..a6ae94fd09de897ac9498f99e28309997ebefb59 100755 (executable)
--- a/webcheckout
+++ b/webcheckout
@@ -127,23 +127,26 @@ sub doit {
 sub better {
        my ($a, $b)=@_;
 
-       my $firstanon=$b;
+       my @anon;
        foreach my $r (@anon_urls) {
                if ($a->{href} =~ /$r/) {
-                       $firstanon=$a;
-                       last;
+                       push @anon, $a;
                }
                elsif ($b->{href} =~ /$r/) {
-                       $firstanon=$b;
-                       last;
+                       push @anon, $b;
                }
        }
 
        if ($want_auth) {
-               return $firstanon != $a;
+               # Whichever is authed is better.
+               return 1 if ! @anon || ! grep { $_ eq $a } @anon;
+               return 0 if ! grep { $_ eq $b } @anon;
+               # Neither is authed, so the better anon method wins.
+               return $anon[0] == $a;
        }
        else {
-               return $firstanon == $a;
+               # Better anon method wins.
+               return @anon && $anon[0] == $a;
        }
 }
 
@@ -179,15 +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;
 
@@ -206,6 +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) {
@@ -231,6 +260,3 @@ foreach my $repo (@repos) {
        }
 }
 exit($errors > 0);
-
-#use Data::Dumper;
-#print Dumper(\@repos);