true_url 0.0.3 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: bba8a863389ecda1e8c6de716c3243d68e22af78
-  data.tar.gz: 620de279d14cefa673b61b9024b903c1131452d4
+  metadata.gz: fe4a8259bda1a0227d7c85cebd88661f208bd4ce
+  data.tar.gz: 14b26af3018b5d04472c06a0cbb6a925cf06332a
 SHA512:
-  metadata.gz: db04004b8be8e1071fb05cd5c6f39406f22023d1f46bb2ece6f6f9d5f01a33e056f5182e68d76148a9f6ac077c85f68ab494f1982744459cb9de57113ff85910
-  data.tar.gz: 54dafb4f8b4656cde6318bbff7f3f656ce869df3efe510b38b59ac96c59c7788b3bc65ed5a4293e6749c6fabaea199faa1a8ef5e60918f2e638a4453b336793b
+  metadata.gz: b6a37a1a3087e12d242949196983b3d65da44c76f7c82c17f7e1e3058c28cc398b0ed87dbd80a8d2ada9895c22490e04a8bef1d3aeae3be0031a950bf064cfe3
+  data.tar.gz: 6770fdfb57aff9f0a471986bc61e980c6a15400f0ec15b841c325f0def08202c24f6b00da25d9ceb7c21082a21587a11d2115fe9629bb65c96e6f40eb0f04653

data/lib/true_url/fetch.rb

@@ -0,0 +1,38 @@
+require 'http'
+require 'nokogiri'
+
+class TrueURL
+  module Fetch
+    extend self
+    
+    def execute(context)
+      starting_url = context.working_url
+
+      response = HTTP.follow
+                     .get(starting_url)
+
+      canonical_url = find_canonical_header(response.headers) || find_canonical_url(response.to_s) || response.uri
+      context.set_working_url(canonical_url, starting_url)
+    end
+
+    def find_canonical_header(headers)
+      return if headers['Link'].nil?
+
+      links = headers['Link'].is_a?(String) ? [headers['Link']] : headers['Link']
+      links.each { |link| return link.split(/[<>;]/)[1] if link.end_with?('rel="canonical"') }
+      nil
+    end
+
+    def find_canonical_url(html)
+      doc = Nokogiri::HTML(html)
+
+      elem = doc.at('link[rel="canonical"]')
+      canonical_url = elem['href'] unless elem.nil?
+
+      elem = doc.at('meta[property="og:url"]')
+      og_url = elem['content'] unless elem.nil?
+
+      canonical_url || og_url
+    end
+  end
+end

data/lib/true_url/version.rb

@@ -1,3 +1,3 @@
 class TrueURL
-  VERSION = '0.0.3'.freeze
+  VERSION = '0.0.4'.freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: true_url
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
 platform: ruby
 authors:
 - Jonathan Wong
@@ -116,6 +116,7 @@ files:
 - Rakefile
 - lib/true_url.rb
 - lib/true_url/context.rb
+- lib/true_url/fetch.rb
 - lib/true_url/strategy.rb
 - lib/true_url/strategy/dailymotion.rb
 - lib/true_url/strategy/nicovideo.rb