Convert relative URLs in feed contents into absolute ones

app/repositories/story_repository.rb

@@ -3,10 +3,12 @@
 
 class StoryRepository
   def self.add(entry, feed)
+    url = entry.url
+    content = extract_content(entry)
     Story.create(feed: feed, 
                 title: entry.title, 
-                permalink: entry.url, 
-                body: StoryRepository.extract_content(entry),
+                permalink: url,
+                body: urls_to_absolute(content, url),
                 is_read: false,
                 published: entry.published || Time.now)
   end
@@ -45,11 +47,26 @@ def self.extract_content(entry)
     end
   end
 
+  def self.urls_to_absolute(content, base_url)
+    doc = Nokogiri::HTML.fragment(content)
+    abs_re = URI::DEFAULT_PARSER.regexp[:ABS_URI]
+    [["a", "href"], ["img", "src"], ["video", "src"]].each do |tag, attr|
+      doc.css(tag).each do |node|
+        url = node.get_attribute(attr)
+        unless url =~ abs_re
+          node.set_attribute(attr, URI.join(base_url, url).to_s)
+          URI.parse(url)
+        end
+      end
+    end
+    doc.to_html
+  end
+
   def self.samples
     [
       SampleStory.new("Darin' Fireballs", "Why you should trade your firstborn for a Retina iPad"),
       SampleStory.new("TechKrunch", "SugarGlidr raises $1.2M Series A for Social Network for Photo Filters"),
       SampleStory.new("Lambda Da Ultimate", "Flimsy types are the new hotness")
     ]
   end
-end
+end

spec/repositories/story_repository_spec.rb

@@ -0,0 +1,32 @@
+require "spec_helper"
+app_require "repositories/story_repository"
+
+describe StoryRepository do
+  klass = described_class
+
+  describe ".urls_to_absolute" do
+    it "preserves existing absolute urls" do
+      content = '<a href="http://foo">bar</a>'
+      expect(klass.urls_to_absolute(content, nil)).to eq(content)
+    end
+
+    it "replaces relative urls in a, img and video tags" do
+      content = <<-EOS
+<div>
+<img src="https://foo">
+<a href="/bar/baz">tee</a><img src="bar/bar">
+<video src="/tee"></video>
+</div>
+      EOS
+      expect(klass.urls_to_absolute(content, "http://oodl.io/d/").gsub(/\n/, ""))
+        .to eq((<<-EOS).gsub(/\n/, ""))
+<div>
+<img src="https://foo">
+<a href="http://oodl.io/bar/baz">tee</a>
+<img src="http://oodl.io/d/bar/bar">
+<video src="http://oodl.io/tee"></video>
+</div>
+      EOS
+    end
+  end
+end