From: Don Armstrong <don@donarmstrong.com>
Date: Wed, 20 Nov 2013 23:26:17 +0000 (-0800)
Subject: handle more craziness at Science
X-Git-Url: https://git.donarmstrong.com/?a=commitdiff_plain;h=e5728e16aa74ce036fd8cf33b28a525f89d23735;p=bin.git

handle more craziness at Science
---

diff --git a/get_pdf b/get_pdf
index 935bfc8..1b43148 100755
--- a/get_pdf
+++ b/get_pdf
@@ -213,8 +213,15 @@ sub find_pdf_link {
         print STDERR $m->content() if $DEBUG > 1;
     }
     my @possible_links;
+    # this brings forward the actual link at Science
+    push @possible_links,
+        grep {my $temp = $_->attrs();
+              exists $temp->{rel} and $temp->{rel} =~ qr/view-/i and
+                  defined $_->text() and $_->text() =~ qr/Full\s*Text.*PDF/i
+              }
+        $m->find_all_links(text_regex => qr/PDF/i);
     # this is to prioritize the real link at science direct
-    push @possible_links, 
+    push @possible_links,
         grep {my $temp = $_->attrs();
               use Data::Dumper;
               print STDERR Dumper($temp);