Support feeds with stories with scheme-less urls

Example is https://blog.golang.org/feed.atom.
Urls in this feed are referenced to as scheme-less urls
(//blog.golang.org/context).

This commit adds normalization of story urls using the feed url. If a
story url doesn't have a scheme it will use the scheme of feed's url.

Author: pascalw

app/repositories/story_repository.rb

@@ -3,6 +3,8 @@
 
 class StoryRepository
   def self.add(entry, feed)
+    entry.url = normalize_url(entry.url, feed.url)
+
     Story.create(feed: feed,
                 title: entry.title,
                 permalink: entry.url,
@@ -109,6 +111,17 @@ def self.expand_absolute_urls(content, base_url)
     doc.to_html
   end
 
+  def self.normalize_url(url, base_url)
+    uri = URI.parse(url)
+
+    unless uri.scheme
+      base_uri = URI.parse(base_url)
+      uri.scheme = base_uri.scheme || 'http'
+    end
+
+    uri.to_s
+  end
+
   def self.samples
     [
       SampleStory.new("Darin' Fireballs", "Why you should trade your firstborn for a Retina iPad"),

spec/repositories/story_repository_spec.rb

@@ -3,6 +3,20 @@
 app_require "repositories/story_repository"
 
 describe StoryRepository do
+  describe '.add' do
+    let(:feed) { double(url: 'http://blog.golang.org/feed.atom') }
+    before do
+      Story.stub(:create)
+    end
+
+    it 'normalizes story urls' do
+      entry = double(url: '//blog.golang.org/context', content: '').as_null_object
+      StoryRepository.should receive(:normalize_url).with(entry.url, feed.url)
+
+      StoryRepository.add(entry, feed)
+    end
+  end
+
   describe ".expand_absolute_urls" do
     it "preserves existing absolute urls" do
       content = '<a href="http://foo">bar</a>'
@@ -98,4 +112,26 @@
       end
     end
   end
+
+  describe ".normalize_url" do
+    it "resolves scheme-less urls" do
+      %w{http https}.each do |scheme|
+        feed_url = "#{scheme}://blog.golang.org/feed.atom"
+
+        url = StoryRepository.normalize_url("//blog.golang.org/context", feed_url)
+        url.should eq "#{scheme}://blog.golang.org/context"
+      end
+    end
+
+    it "leaves urls with a scheme intact" do
+      input = 'http://blog.golang.org/context'
+      normalized_url = StoryRepository.normalize_url(input, 'http://blog.golang.org/feed.atom')
+      normalized_url.should eq(input)
+    end
+
+    it "falls back to http if the base_url is also sheme less" do
+      url = StoryRepository.normalize_url("//blog.golang.org/context", "//blog.golang.org/feed.atom")
+      url.should eq 'http://blog.golang.org/context'
+    end
+  end
 end