* add more debugging information

 * handle relative urls properly
 * look in meta tags too

diff --git a/get_pdf b/get_pdf
index 23daa99caf8a3a14132df6a732add40992738b3d..7daca792fc7f6332a7e11121b83bb3ad4b211b0b 100755 (executable)
--- a/get_pdf
+++ b/get_pdf
@@ -160,22 +160,28 @@ sub find_pdf_link {
        $m->get($inputs[0]->value);
        print STDERR $m->content() if $DEBUG;
     }
-    my @possible_links = $m->find_all_links(text_regex => qr/pdf/i);
+    my @possible_links;
+    push @possible_links, $m->find_all_links(tag_regex => qr/meta/,
+                                            url_regex  => qr/(reprint|\.pdf)/i,
+                                           );
+    push @possible_links, $m->find_all_links(text_regex => qr/pdf/i);
     push @possible_links,$m->find_all_links(text_regex => qr/manual\s*download/i);
-    print STDERR map{$_->url,qq(\n)} @possible_links if $DEBUG;
+    print STDERR $m->uri() if $DEBUG;
+    print STDERR $m->content() if $DEBUG;
+    print STDERR map{$_->url_abs(),qq(\n)} @possible_links if $DEBUG;
     if (not @possible_links and $DEBUG) {
        print STDERR $m->content();
     }
     my $best_guess = $possible_links[0] if @possible_links;
     for my $link (@possible_links) {
-       my $r = $m->get($link->url());
+       my $r = $m->get($link->url_abs());
        if ($r->header('Content-Type') =~ /pdf/) {
            return $m;
        }
     }
     my @sub_frames = $m->find_all_links(tag_regex=>qr/^i?frame$/);
     for my $frame (@sub_frames) {
-       $m->get($frame->url());
+       $m->get($frame->url_abs());
        my $pdf_m = find_pdf_link($m,
                                  0,
                                  $call+1,
@@ -185,7 +191,7 @@ sub find_pdf_link {
        }
     }
     if ($guess and defined $best_guess) {
-       $m->get($best_guess->url());
+       $m->get($best_guess->url_abs());
        return $m;
     }
     return undef;