feature: add favicon scraper. Fixes #40

lib/link_thumbnailer/configuration.rb

@@ -41,7 +41,7 @@ def initialize
         %r{^http://pixel\.quantserve\.com/},
         %r{^http://s7\.addthis\.com/}
       ]
-      @attributes             = [:title, :images, :description, :videos]
+      @attributes             = [:title, :images, :description, :videos, :favicon]
       @graders                = [
         ->(description) { ::LinkThumbnailer::Graders::Length.new(description) },
         ->(description) { ::LinkThumbnailer::Graders::HtmlAttribute.new(description, :class) },

lib/link_thumbnailer/models/favicon.rb

@@ -0,0 +1,25 @@
+require 'link_thumbnailer/model'
+
+module LinkThumbnailer
+  module Models
+    class Favicon < ::LinkThumbnailer::Model
+
+      attr_reader :uri
+
+      def initialize(uri)
+        @uri = uri
+      end
+
+      def to_s
+        uri.to_s
+      end
+
+      def as_json(*)
+        {
+          src: to_s
+        }
+      end
+
+    end
+  end
+end

lib/link_thumbnailer/models/website.rb

@@ -4,7 +4,7 @@ module LinkThumbnailer
   module Models
     class Website < ::LinkThumbnailer::Model
 
-      attr_accessor :url, :title, :description, :images, :videos
+      attr_accessor :url, :title, :description, :images, :videos, :favicon
 
       def initialize
         @images = []
@@ -39,6 +39,7 @@ def images
       def as_json(*)
         {
           url:          url.to_s,
+          favicon:      favicon,
           title:        title,
           description:  description,
           images:       images,

lib/link_thumbnailer/scraper.rb

@@ -12,6 +12,8 @@
 require 'link_thumbnailer/scrapers/opengraph/images'
 require 'link_thumbnailer/scrapers/default/videos'
 require 'link_thumbnailer/scrapers/opengraph/videos'
+require 'link_thumbnailer/scrapers/default/favicon'
+require 'link_thumbnailer/scrapers/opengraph/favicon'
 
 module LinkThumbnailer
   class Scraper < ::SimpleDelegator

lib/link_thumbnailer/scrapers/default/favicon.rb

@@ -0,0 +1,36 @@
+require 'link_thumbnailer/scrapers/default/base'
+require 'link_thumbnailer/models/favicon'
+
+module LinkThumbnailer
+  module Scrapers
+    module Default
+      class Favicon < ::LinkThumbnailer::Scrapers::Default::Base
+
+        def value
+          modelize(to_uri(href)).to_s
+        end
+
+        private
+
+        def to_uri(href)
+          ::URI.parse(href)
+        rescue ::URI::InvalidURIError
+          nil
+        end
+
+        def href
+          node.attributes['href'].value.to_s if node
+        end
+
+        def node
+          document.xpath("//link[contains(@rel, 'icon')]").first
+        end
+
+        def modelize(uri)
+          model_class.new(uri)
+        end
+
+      end
+    end
+  end
+end

lib/link_thumbnailer/scrapers/opengraph/favicon.rb

@@ -0,0 +1,15 @@
+require 'link_thumbnailer/scrapers/opengraph/base'
+
+module LinkThumbnailer
+  module Scrapers
+    module Opengraph
+      class Favicon < ::LinkThumbnailer::Scrapers::Opengraph::Base
+
+        def value
+          nil
+        end
+
+      end
+    end
+  end
+end

spec/configuration_spec.rb

@@ -9,7 +9,7 @@
   it { expect(instance.verify_ssl).to             eq(true) }
   it { expect(instance.http_timeout).to           eq(5) }
   it { expect(instance.blacklist_urls).to_not     be_empty }
-  it { expect(instance.attributes).to             eq([:title, :images, :description, :videos]) }
+  it { expect(instance.attributes).to             eq([:title, :images, :description, :videos, :favicon]) }
   it { expect(instance.graders).to_not            be_empty }
   it { expect(instance.description_min_length).to eq(25) }
   it { expect(instance.positive_regex).to_not     be_nil }

spec/fixture_spec.rb

@@ -5,6 +5,7 @@
   let(:url)         { 'http://foo.com' }
   let(:png_url)     { 'http://foo.com/foo.png' }
   let(:video_url)   { 'http://foo.com/foo.swf' }
+  let(:favicon)     { 'http://foo.com/foo.ico' }
   let(:png)         { File.open(File.dirname(__FILE__) + '/fixtures/foo.png') }
   let(:action)      { LinkThumbnailer.generate(url) }
 
@@ -22,6 +23,7 @@
 
       let(:html) { File.open(File.dirname(__FILE__) + '/fixtures/og_valid_example.html').read() }
 
+      it { expect(action.favicon).to                eq(favicon) }
       it { expect(action.title).to                  eq(title) }
       it { expect(action.description).to            eq(description) }
       it { expect(action.images.count).to           eq(1) }
@@ -42,6 +44,7 @@
         stub_request(:get, png_url_2).to_return(status: 200, body: png_2, headers: {})
       end
 
+      it { expect(action.favicon).to                eq('') }
       it { expect(action.title).to                  eq(title) }
       it { expect(action.description).to            eq(description) }
       it { expect(action.images.count).to           eq(2) }
@@ -55,6 +58,7 @@
       let(:video_url_2) { 'http://foo.com/bar.swf' }
       let(:html)        { File.open(File.dirname(__FILE__) + '/fixtures/og_valid_multi_video_example.html').read() }
 
+      it { expect(action.favicon).to                eq('') }
       it { expect(action.title).to                  eq(title) }
       it { expect(action.description).to            eq(description) }
       it { expect(action.videos.count).to           eq(2) }
@@ -84,6 +88,7 @@
       let(:title)       { 'Title from meta' }
       let(:description) { 'Description from meta' }
 
+      it { expect(action.favicon).to      eq(favicon) }
       it { expect(action.title).to        eq(title) }
       it { expect(action.description).to  eq(description) }
 
@@ -94,6 +99,7 @@
       let(:html)        { File.open(File.dirname(__FILE__) + '/fixtures/default_from_body.html').read() }
       let(:description) { 'Description from body' }
 
+      it { expect(action.favicon).to                eq(favicon) }
       it { expect(action.description).to            eq(description) }
       it { expect(action.images.count).to           eq(1) }
       it { expect(action.images.first.src.to_s).to  eq(png_url) }

spec/fixtures/default_from_body.html

@@ -1,6 +1,7 @@
 <html>
 <head>
   <title>Title from meta</title>
+  <link rel="shortcut icon" href="http://foo.com/foo.ico">
 </head>
 <body>
 

spec/fixtures/default_from_meta.html

@@ -2,6 +2,7 @@
 <head>
   <title>Title from meta</title>
   <meta content="Description from meta" name="description">
+  <link rel="icon whatever" href="http://foo.com/foo.ico">
 </head>
 <body>