feedjira 2.0.0 → 2.1.0

data/lib/feedjira/feed.rb

@@ -5,16 +5,14 @@ module Feedjira
     end
 
     def self.parse(xml, &block)
-      if parser = determine_feed_parser_for_xml(xml)
-        parse_with parser, xml, &block
-      else
-        raise NoParserAvailable.new("No valid parser for XML.")
-      end
+      parser = determine_feed_parser_for_xml(xml)
+      raise NoParserAvailable, 'No valid parser for XML.' unless parser
+      parse_with parser, xml, &block
     end
 
     def self.determine_feed_parser_for_xml(xml)
       start_of_doc = xml.slice(0, 2000)
-      feed_classes.detect {|klass| klass.able_to_parse?(start_of_doc)}
+      feed_classes.detect { |klass| klass.able_to_parse?(start_of_doc) }
     end
 
     def self.add_feed_class(klass)
@@ -25,6 +23,7 @@ module Feedjira
       @feed_classes ||= [
         Feedjira::Parser::RSSFeedBurner,
         Feedjira::Parser::GoogleDocsAtom,
+        Feedjira::Parser::AtomYoutube,
         Feedjira::Parser::AtomFeedBurner,
         Feedjira::Parser::Atom,
         Feedjira::Parser::ITunesRSS,
@@ -53,10 +52,14 @@ module Feedjira
     end
 
     def self.call_on_each_feed_entry(method, *parameters)
-      feed_classes.each do |k|
-        k.sax_config.collection_elements.each_value do |vl|
-          vl.find_all{|v| (v.accessor == 'entries') && (v.data_class.class == Class)}.each do |v|
-              v.data_class.send(method, *parameters)
+      feed_classes.each do |klass|
+        klass.sax_config.collection_elements.each_value do |value|
+          collection_configs = value.select do |v|
+            v.accessor == 'entries' && v.data_class.class == Class
+          end
+
+          collection_configs.each do |config|
+            config.data_class.send(method, *parameters)
           end
         end
       end
@@ -64,15 +67,14 @@ module Feedjira
 
     def self.fetch_and_parse(url)
       response = connection(url).get
-      raise FetchFailure.new("Fetch failed - #{response.status}") unless response.success?
-      xml = response.body
-      parser_klass = determine_feed_parser_for_xml xml
-      raise NoParserAvailable.new("No valid parser for XML.") unless parser_klass
-
-      feed = parse_with parser_klass, xml
+      unless response.success?
+        raise FetchFailure, "Fetch failed - #{response.status}"
+      end
+      feed = parse response.body
       feed.feed_url = url
-      feed.etag = response.headers['etag'].to_s.gsub(/"/, '')
-      feed.last_modified = response.headers['last-modified']
+      feed.etag = response.headers['etag'].to_s.delete '"'
+
+      feed.last_modified = parse_last_modified(response)
       feed
     end
 
@@ -82,5 +84,12 @@ module Feedjira
         conn.adapter :net_http
       end
     end
+
+    def self.parse_last_modified(response)
+      DateTime.parse(response.headers['last-modified']).to_time
+    rescue
+      nil
+    end
+    private_class_method :parse_last_modified
   end
 end

data/lib/feedjira/feed_entry_utilities.rb

@@ -1,23 +1,22 @@
 module Feedjira
   module FeedEntryUtilities
-
     include Enumerable
+    include DateTimeUtilities
 
     def published
       @published ||= @updated
     end
 
     def parse_datetime(string)
-      begin
-        DateTime.parse(string).feed_utils_to_gm_time
-      rescue
-        warn "Failed to parse date #{string.inspect}"
-        nil
-      end
+      DateTime.parse(string).feed_utils_to_gm_time
+    rescue
+      warn "Failed to parse date #{string.inspect}"
+      nil
     end
 
     ##
-    # Returns the id of the entry or its url if not id is present, as some formats don't support it
+    # Returns the id of the entry or its url if not id is present, as some
+    # formats don't support it
     def id
       @entry_id ||= @url
     end
@@ -37,30 +36,29 @@ module Feedjira
     end
 
     def sanitize!
-      %w[title author summary content image].each do |name|
-        if self.respond_to?(name) && self.send(name).respond_to?(:sanitize!)
-          self.send(name).send :sanitize!
+      %w(title author summary content image).each do |name|
+        if respond_to?(name) && send(name).respond_to?(:sanitize!)
+          send(name).send :sanitize!
         end
       end
     end
 
-    alias_method :last_modified, :published
+    alias last_modified published
 
     def each
-      @rss_fields ||= self.instance_variables
+      @rss_fields ||= instance_variables
 
       @rss_fields.each do |field|
-        yield(field.to_s.sub('@', ''), self.instance_variable_get(field))
+        yield(field.to_s.sub('@', ''), instance_variable_get(field))
       end
     end
 
     def [](field)
-      self.instance_variable_get("@#{field.to_s}")
+      instance_variable_get("@#{field}")
     end
 
     def []=(field, value)
-      self.instance_variable_set("@#{field.to_s}", value)
+      instance_variable_set("@#{field}", value)
     end
-
   end
 end

data/lib/feedjira/feed_utilities.rb

@@ -1,6 +1,6 @@
 module Feedjira
   module FeedUtilities
-    UPDATABLE_ATTRIBUTES = %w(title feed_url url last_modified etag)
+    UPDATABLE_ATTRIBUTES = %w(title feed_url url last_modified etag).freeze
 
     attr_writer   :new_entries, :updated, :last_modified
     attr_accessor :etag
@@ -32,7 +32,8 @@ module Feedjira
 
     def last_modified
       @last_modified ||= begin
-        entry = entries.reject {|e| e.published.nil? }.sort_by { |entry| entry.published if entry.published }.last
+        published = entries.reject { |e| e.published.nil? }
+        entry = published.sort_by { |e| e.published if e.published }.last
         entry ? entry.published : nil
       end
     end
@@ -45,13 +46,13 @@ module Feedjira
       @new_entries ||= []
     end
 
-    def has_new_entries?
-      new_entries.size > 0
+    def new_entries?
+      !new_entries.empty?
     end
 
     def update_from_feed(feed)
       self.new_entries += find_new_entries_for(feed)
-      self.entries.unshift(*self.new_entries)
+      entries.unshift(*self.new_entries)
 
       @updated = false
 
@@ -61,7 +62,8 @@ module Feedjira
     end
 
     def update_attribute(feed, name)
-      old_value, new_value = send(name), feed.send(name)
+      old_value = send(name)
+      new_value = feed.send(name)
 
       if old_value != new_value
         send("#{name}=", new_value)
@@ -72,33 +74,36 @@ module Feedjira
     end
 
     def sanitize_entries!
-      entries.each {|entry| entry.sanitize!}
+      entries.each(&:sanitize!)
     end
 
     private
 
+    # This implementation is a hack, which is why it's so ugly. It's to get
+    # around the fact that not all feeds have a published date. However,
+    # they're always ordered with the newest one first. So we go through the
+    # entries just parsed and insert each one as a new entry until we get to
+    # one that has the same id as the the newest for the feed.
     def find_new_entries_for(feed)
-      # this implementation is a hack, which is why it's so ugly.
-      # it's to get around the fact that not all feeds have a published date.
-      # however, they're always ordered with the newest one first.
-      # So we go through the entries just parsed and insert each one as a new entry
-      # until we get to one that has the same id as the the newest for the feed
-      return feed.entries if self.entries.length == 0
-      latest_entry = self.entries.first
+      return feed.entries if entries.length.zero?
+
+      latest_entry = entries.first
       found_new_entries = []
+
       feed.entries.each do |entry|
-        if entry.entry_id.nil? && latest_entry.entry_id.nil?
-          break if entry.url == latest_entry.url
-        else
-          break if entry.entry_id == latest_entry.entry_id || entry.url == latest_entry.url
-        end
+        break unless new_entry?(entry, latest_entry)
         found_new_entries << entry
       end
+
       found_new_entries
     end
 
-    def existing_entry?(test_entry)
-      entries.any? { |entry| entry.id == test_entry.id }
+    def new_entry?(entry, latest)
+      nil_ids = entry.entry_id.nil? && latest.entry_id.nil?
+      new_id = entry.entry_id != latest.entry_id
+      new_url = entry.url != latest.url
+
+      (nil_ids || new_id) && new_url
     end
   end
 end

data/lib/feedjira/parser/atom.rb

@@ -4,16 +4,17 @@ module Feedjira
     class Atom
       include SAXMachine
       include FeedUtilities
+
       element :title
-      element :subtitle, :as => :description
-      element :link, :as => :url, :value => :href, :with => {:type => "text/html"}
-      element :link, :as => :feed_url, :value => :href, :with => {:rel => "self"}
-      elements :link, :as => :links, :value => :href
-      elements :link, :as => :hubs, :value => :href, :with => {:rel => "hub"}
-      elements :entry, :as => :entries, :class => AtomEntry
+      element :subtitle, as: :description
+      element :link, as: :url, value: :href, with: { type: 'text/html' }
+      element :link, as: :feed_url, value: :href, with: { rel: 'self' }
+      elements :link, as: :links, value: :href
+      elements :link, as: :hubs, value: :href, with: { rel: 'hub' }
+      elements :entry, as: :entries, class: AtomEntry
 
-      def self.able_to_parse?(xml) #:nodoc:
-        /\<feed[^\>]+xmlns\s?=\s?[\"|\'](http:\/\/www\.w3\.org\/2005\/Atom|http:\/\/purl\.org\/atom\/ns\#)[\"|\'][^\>]*\>/ =~ xml
+      def self.able_to_parse?(xml)
+        %r{\<feed[^\>]+xmlns\s?=\s?[\"\'](http://www\.w3\.org/2005/Atom|http://purl\.org/atom/ns\#)[\"\'][^\>]*\>} =~ xml # rubocop:disable Metrics/LineLength
       end
 
       def url

data/lib/feedjira/parser/atom_entry.rb

@@ -1,5 +1,4 @@
 module Feedjira
-
   module Parser
     # Parser for dealing with Atom feed entries.
     class AtomEntry
@@ -7,28 +6,26 @@ module Feedjira
       include FeedEntryUtilities
 
       element :title
-      element :link, :as => :url, :value => :href, :with => {:type => "text/html", :rel => "alternate"}
-      element :name, :as => :author
+      element :link, as: :url, value: :href, with: { type: 'text/html', rel: 'alternate' } # rubocop:disable Metrics/LineLength
+      element :name, as: :author
       element :content
       element :summary
 
-      element :"media:content", :as => :image, :value => :url
-      element :enclosure, :as => :image, :value => :href
+      element :"media:content", as: :image, value: :url
+      element :enclosure, as: :image, value: :href
 
       element :published
-      element :id, :as => :entry_id
-      element :created, :as => :published
-      element :issued, :as => :published
+      element :id, as: :entry_id
+      element :created, as: :published
+      element :issued, as: :published
       element :updated
-      element :modified, :as => :updated
-      elements :category, :as => :categories, :value => :term
-      elements :link, :as => :links, :value => :href
+      element :modified, as: :updated
+      elements :category, as: :categories, value: :term
+      elements :link, as: :links, value: :href
 
       def url
         @url ||= links.first
       end
     end
-
   end
-
 end

data/lib/feedjira/parser/atom_feed_burner.rb

@@ -1,26 +1,24 @@
 module Feedjira
-
   module Parser
     # Parser for dealing with Feedburner Atom feeds.
     class AtomFeedBurner
       include SAXMachine
       include FeedUtilities
+
       element :title
-      element :subtitle, :as => :description
-      element :link, :as => :url, :value => :href, :with => {:type => "text/html"}
-      element :link, :as => :feed_url, :value => :href, :with => {:type => "application/atom+xml"}
-      elements :"atom10:link", :as => :hubs, :value => :href, :with => {:rel => "hub"}
-      elements :entry, :as => :entries, :class => AtomFeedBurnerEntry
+      element :subtitle, as: :description
+      element :link, as: :url, value: :href, with: { type: 'text/html' }
+      element :link, as: :feed_url, value: :href, with: { type: 'application/atom+xml' } # rubocop:disable Metrics/LineLength
+      elements :"atom10:link", as: :hubs, value: :href, with: { rel: 'hub' }
+      elements :entry, as: :entries, class: AtomFeedBurnerEntry
 
-      def self.able_to_parse?(xml) #:nodoc:
-        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false
+      def self.able_to_parse?(xml)
+        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false # rubocop:disable Metrics/LineLength
       end
 
       def self.preprocess(xml)
         Preprocessor.new(xml).to_xml
       end
     end
-
   end
-
 end

data/lib/feedjira/parser/atom_feed_burner_entry.rb

@@ -1,5 +1,4 @@
 module Feedjira
-
   module Parser
     # Parser for dealing with Feedburner Atom feed entries.
     class AtomFeedBurnerEntry
@@ -7,29 +6,27 @@ module Feedjira
       include FeedEntryUtilities
 
       element :title
-      element :name, :as => :author
-      element :link, :as => :url, :value => :href, :with => {:type => "text/html", :rel => "alternate"}
-      element :"feedburner:origLink", :as => :url
+      element :name, as: :author
+      element :link, as: :url, value: :href, with: { type: 'text/html', rel: 'alternate' } # rubocop:disable Metrics/LineLength
+      element :"feedburner:origLink", as: :url
       element :summary
       element :content
 
-      element :"media:content", :as => :image, :value => :url
-      element :enclosure, :as => :image, :value => :href
+      element :"media:content", as: :image, value: :url
+      element :enclosure, as: :image, value: :href
 
       element :published
-      element :id, :as => :entry_id
-      element :issued, :as => :published
-      element :created, :as => :published
+      element :id, as: :entry_id
+      element :issued, as: :published
+      element :created, as: :published
       element :updated
-      element :modified, :as => :updated
-      elements :category, :as => :categories, :value => :term
-      elements :link, :as => :links, :value => :href
+      element :modified, as: :updated
+      elements :category, as: :categories, value: :term
+      elements :link, as: :links, value: :href
 
       def url
         @url ||= links.first
       end
-
     end
   end
-
 end

data/lib/feedjira/parser/atom_youtube.rb

@@ -0,0 +1,20 @@
+module Feedjira
+  module Parser
+    # Parser for dealing with RSS feeds.
+    class AtomYoutube
+      include SAXMachine
+      include FeedUtilities
+      element :title
+      element :link, as: :url, value: :href, with: { rel: 'alternate' }
+      element :link, as: :feed_url, value: :href, with: { rel: 'self' }
+      element :name, as: :author
+      element :"yt:channelId", as: :youtube_channel_id
+
+      elements :entry, as: :entries, class: AtomYoutubeEntry
+
+      def self.able_to_parse?(xml) #:nodoc:
+        %r{xmlns:yt="http://www.youtube.com/xml/schemas/2015"} =~ xml
+      end
+    end
+  end
+end

data/lib/feedjira/parser/atom_youtube_entry.rb

@@ -0,0 +1,29 @@
+module Feedjira
+  module Parser
+    class AtomYoutubeEntry
+      include SAXMachine
+      include FeedEntryUtilities
+
+      element :title
+      element :link, as: :url, value: :href, with: { rel: 'alternate' }
+      element :name, as: :author
+      element :"media:description", as: :content
+      element :summary
+      element :published
+      element :id, as: :entry_id
+      element :updated
+      element :"yt:videoId", as: :youtube_video_id
+      element :"media:title", as: :media_title
+      element :"media:content", as: :media_url, value: :url
+      element :"media:content", as: :media_type, value: :type
+      element :"media:content", as: :media_width, value: :width
+      element :"media:content", as: :media_height, value: :height
+      element :"media:thumbnail", as: :media_thumbnail_url, value: :url
+      element :"media:thumbnail", as: :media_thumbnail_width, value: :width
+      element :"media:thumbnail", as: :media_thumbnail_height, value: :height
+      element :"media:starRating", as: :media_star_count, value: :count
+      element :"media:starRating", as: :media_star_average, value: :average
+      element :"media:statistics", as: :media_views, value: :views
+    end
+  end
+end