spix_parser 1.6.4 → 1.6.5

data/lib/spix_parser/tools/feed_discovery/document.rb

@@ -1,3 +1,4 @@
+require 'ruby-debug'
 module Spix
   module FeedDiscovery
     class Document
@@ -51,7 +52,7 @@ module Spix
 
       def feed_unsing_address uri, &block
         begin
-          Feed.new(uri, favicon).tap do |feed|
+          Feed.new(uri).tap do |feed|
             block.call feed if block_given?
           end
         rescue => error
@@ -101,71 +102,6 @@ module Spix
       def feed?
         %w[rss feed].include? @document.root.name
       end
-
-      def favicon
-        shortcut_in_document or shortcut_from_original_page or shortcut_from(base_uri) or default_favico_if_exist
-      end
-
-      def shortcut_in_document
-        shortcuts = find_shortcut_in @document
-        shortcuts.any? ? base_uri.merge(shortcuts.first.to_s).to_s : nil 
-      end
-      private :shortcut_in_document
-
-      def shortcut_from_original_page
-        if feed?
-          if node = @document.search('link').first
-            path = URI.parse node.content.strip
-            shortcut_from URI.parse path.select(:scheme, :host).join("://") rescue nil
-          end
-        end
-      end
-      private :shortcut_from_original_page
-
-      def shortcut_from base_uri
-        doc = get base_uri
-        shortcuts = find_shortcut_in doc
-        shortcuts.any? ? base_uri.merge(shortcuts.first.to_s).to_s : nil 
-
-      rescue Net::HTTPError, Net::HTTPFatalError
-        logger.warn "error opening favicon: #{$!}"
-        nil
-      end
-      private :shortcut_from
-
-      def find_shortcut_in doc
-        doc.xpath(
-          '//link[contains(translate(@rel, "ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz"), "shortcut")]', 
-          '//link[contains(translate(@rel, "ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz"), "ico")]'
-        ).map { |node| node.get_attribute "href" }
-      end
-
-      def default_favico_if_exist
-        http = Net::HTTP.new base_uri.host, base_uri.port
-        resp = http.request_head 'favicon.ico'  
-        base_uri.merge('favicon.ico').to_s unless resp.kind_of? Net::HTTPError
-      rescue
-        nil
-      end
-      private :default_favico_if_exist
-
-      def get uri
-        resp = Net::HTTP.get_response uri
-        doc = Nokogiri::HTML(resp.body)
-        if resp.kind_of?(Net::HTTPRedirection) or (refresh_metatags = Nokogiri::HTML(resp.body).search('meta[@http-equiv=REFRESH]')).any?
-          path = resp['location'] || refresh_metatags.first.get_attribute('content')[/http:\/\/.*/]
-          get URI.parse path
-        else
-          doc 
-        end
-      rescue
-        Nokogiri::HTML('')
-      end
-
-      def base_uri
-        @base_uri ||= URI.parse @uri.select(:scheme, :host).join("://")
-      end
-      private :base_uri
       
     end
   end

data/lib/spix_parser/tools/feed_discovery/feed.rb

@@ -2,9 +2,9 @@ module Spix
   module FeedDiscovery
     class Feed < Hash 
       
-      def initialize url, favicon
+      def initialize url
         self.url = url.to_s
-        self.favicon = favicon
+        self.favicon = get_favicon
         self.title = get_title
       end
 
@@ -24,13 +24,61 @@ module Spix
       end
       private :get_title
 
+      def get_favicon
+        if node = content.search('link').first
+          path = URI.parse node.content.strip
+          shortcut_from URI.parse path.select(:scheme, :host).join("://") rescue nil
+        end
+      end
+
+      def shortcut_from base_uri
+        doc = get base_uri
+        shortcuts = find_shortcut_in doc
+        shortcuts.any? ? base_uri.merge(shortcuts.first.to_s).to_s : nil 
+
+      rescue Net::HTTPError, Net::HTTPFatalError
+        logger.warn "error opening favicon: #{$!}"
+        nil
+      end
+      private :shortcut_from
+
+      def find_shortcut_in doc
+        doc.xpath(
+          '//link[contains(translate(@rel, "ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz"), "shortcut")]', 
+          '//link[contains(translate(@rel, "ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz"), "ico")]'
+        ).map { |node| node.get_attribute "href" }
+      end
+
+      def get uri
+        resp = Net::HTTP.get_response uri
+        doc = Nokogiri::HTML(resp.body)
+        if resp.kind_of?(Net::HTTPRedirection) or (refresh_metatags = Nokogiri::HTML(resp.body).search('meta[@http-equiv=REFRESH]')).any?
+          path = resp['location'] || refresh_metatags.first.get_attribute('content')[/http:\/\/.*/]
+          get URI.parse path
+        else
+          doc 
+        end
+      rescue
+        Nokogiri::HTML('')
+      end
+
+      def base_uri
+        @base_uri ||= URI.parse uri.select(:scheme, :host).join("://")
+      end
+      private :base_uri
+
       def content
+        @content ||= load_content
+      end
+      private :content
+
+      def load_content
         req = Net::HTTP.new uri.host, uri.port
         path = uri - uri.select(:scheme, :host).join("://")
         resp = req.request_get path.to_s
         Nokogiri::XML(resp.body)
       end
-      private :content
+      private :load_content
 
       def uri
         @uri ||= URI.parse url

data/lib/spix_parser/version.rb

@@ -4,7 +4,7 @@ module Spix
     module Version
       MAJOR = 1
       MINOR = 6
-      TINY  = 4
+      TINY  = 5
 
       def self.current_version
         "#{MAJOR}.#{MINOR}.#{TINY}"

metadata

@@ -2,7 +2,7 @@
 name: spix_parser
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 1.6.4
+  version: 1.6.5
 platform: ruby
 authors: 
 - Marcio Lopes de Faria