Handle links that have a blank or missing href

lib/govspeak/link_extractor.rb

@@ -14,17 +14,22 @@ def call
     attr_reader :document, :website_root
 
     def extract_links
-      document_anchors.map do |link|
-        if website_root && link['href'].start_with?('/')
-          "#{website_root}#{link['href']}"
-        else
-          link['href']
-        end
+      document_anchors.
+        map { |link| extract_href_from_link(link) }.
+        reject(&:blank?)
+    end
+
+    def extract_href_from_link(link)
+      href = link['href'] || ''
+      if website_root && href.start_with?('/')
+        "#{website_root}#{href}"
+      else
+        href
       end
     end
 
     def document_anchors
-      processed_govspeak.css('a:not([href^="mailto"])').css('a:not([href^="#"])')
+      processed_govspeak.css('a[href]').css('a:not([href^="mailto"])').css('a:not([href^="#"])')
     end
 
     def processed_govspeak

test/govspeak_link_extractor_test.rb

@@ -14,6 +14,20 @@ def document_body
 [mailto:](mailto:[email protected])
 
 [absolute_path](/cais-trwydded-yrru-dros-dro)
+
+<a href="http://www.example.com/from/html">raw_html_link</a>
+
+[empty_markdown_link]()
+
+[a_different_empty_markdown_link]( )
+
+<a>empty_raw_html_link</a>
+
+<a href="">a_second_empty_raw_link</a>
+
+<a href=" ">a_third_empty_raw_link</a>
+
+<a href>a_fourth_empty_raw_link</a>
     }
   end
 
@@ -26,7 +40,7 @@ def links
   end
 
   test "Links are extracted from the body" do
-    expected_links = %w{http://www.example.com http://www.gov.com /cais-trwydded-yrru-dros-dro}
+    expected_links = %w{http://www.example.com http://www.gov.com /cais-trwydded-yrru-dros-dro http://www.example.com/from/html}
     assert_equal expected_links, links
   end
 
@@ -42,6 +56,11 @@ def links
     refute_includes ["mailto:[email protected]"], links
   end
 
+  test "Links are not extracted if they are blank" do
+    refute_includes [""], links
+    refute_includes [nil], links
+  end
+
   test "Absolute links are transformed to a url when website_root passed in" do
     urls = doc.extracted_links(website_root: "http://www.example.com")
     assert urls.include?("http://www.example.com/cais-trwydded-yrru-dros-dro")