feedjira 2.0.0 → 2.2.0

data/lib/feedjira/parser/atom_feed_burner.rb

@@ -1,26 +1,43 @@
+# rubocop:disable Style/DocumentationMethod
 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_text_html, value: :href,
+                     with: { type: 'text/html' }
+      element :link, as: :url_notype, value: :href, with: { type: nil }
+      element :link, as: :feed_url_link, value: :href, with: { type: 'application/atom+xml' } # rubocop:disable Metrics/LineLength
+      element :"atom10:link", as: :feed_url_atom10_link, value: :href,
+                              with: { type: 'application/atom+xml' }
+      elements :"atom10:link", as: :hubs, value: :href, with: { rel: 'hub' }
+      elements :entry, as: :entries, class: AtomFeedBurnerEntry
+
+      attr_writer :url, :feed_url
+
+      def self.able_to_parse?(xml)
+        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false # rubocop:disable Metrics/LineLength
+      end
 
-      def self.able_to_parse?(xml) #:nodoc:
-        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false
+      # Feed url is <link> with type="text/html" if present,
+      # <link> with no type attribute otherwise
+      def url
+        @url || @url_text_html || @url_notype
+      end
+
+      # Feed feed_url is <link> with type="application/atom+xml" if present,
+      # <atom10:link> with type="application/atom+xml" otherwise
+      def feed_url
+        @feed_url || @feed_url_link || @feed_url_atom10_link
       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 +1,5 @@
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
-
   module Parser
     # Parser for dealing with Feedburner Atom feed entries.
     class AtomFeedBurnerEntry
@@ -7,29 +7,27 @@ 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 +1,21 @@
+# rubocop:disable Style/DocumentationMethod
+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 +1,30 @@
+# rubocop:disable Style/Documentation
+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

data/lib/feedjira/parser/google_docs_atom.rb

@@ -1,23 +1,24 @@
 require File.expand_path('./atom', File.dirname(__FILE__))
-
+# rubocop:disable Style/Documentation
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
   module Parser
     class GoogleDocsAtom
       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 :link, :as => :links, :value => :href
-      elements :entry, :as => :entries, :class => GoogleDocsAtomEntry
+      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 :link, as: :links, value: :href
+      elements :entry, as: :entries, class: GoogleDocsAtomEntry
 
       def url
         @url ||= links.first
       end
 
       def self.able_to_parse?(xml) #:nodoc:
-        %r{<id>https?://docs.google.com/.*\</id\>} =~ xml
+        %r{<id>https?://docs\.google\.com/.*\</id\>} =~ xml
       end
 
       def feed_url

data/lib/feedjira/parser/google_docs_atom_entry.rb

@@ -1,3 +1,5 @@
+# rubocop:disable Style/Documentation
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
   module Parser
     class GoogleDocsAtomEntry
@@ -5,21 +7,21 @@ 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 :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 :"docs:md5Checksum", :as => :checksum
-      element :"docs:filename", :as => :original_filename
-      element :"docs:suggestedFilename", :as => :suggested_filename
+      element :modified, as: :updated
+      elements :category, as: :categories, value: :term
+      elements :link, as: :links, value: :href
+      element :"docs:md5Checksum", as: :checksum
+      element :"docs:filename", as: :original_filename
+      element :"docs:suggestedFilename", as: :suggested_filename
 
       def url
         @url ||= links.first

data/lib/feedjira/parser/itunes_rss.rb

@@ -1,8 +1,11 @@
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
-
   module Parser
     # iTunes is RSS 2.0 + some apple extensions
-    # Source: http://www.apple.com/itunes/whatson/podcasts/specs.html
+    # Sources:
+    #   * https://cyber.harvard.edu/rss/rss.html
+    #   * http://lists.apple.com/archives/syndication-dev/2005/Nov/msg00002.html
+    #   * https://help.apple.com/itc/podcasts_connect/
     class ITunesRSS
       include SAXMachine
       include FeedUtilities
@@ -12,39 +15,55 @@ module Feedjira
       # RSS 2.0 elements that need including
       element :copyright
       element :description
+      element :image, class: RSSImage
       element :language
+      element :lastBuildDate, as: :last_built
+      element :link, as: :url
       element :managingEditor
+      element :rss, as: :version, value: :version
       element :title
-      element :link, :as => :url
+      element :ttl
 
       # If author is not present use managingEditor on the channel
-      element :"itunes:author", :as => :itunes_author
-      element :"itunes:block", :as => :itunes_block
-      element :"itunes:image", :value => :href, :as => :itunes_image
-      element :"itunes:explicit", :as => :itunes_explicit
-      element :"itunes:keywords", :as => :itunes_keywords
+      element :"itunes:author", as: :itunes_author
+      element :"itunes:block", as: :itunes_block
+      element :"itunes:image", value: :href, as: :itunes_image
+      element :"itunes:explicit", as: :itunes_explicit
+      element :"itunes:complete", as: :itunes_complete
+      element :"itunes:keywords", as: :itunes_keywords
+      element :"itunes:type", as: :itunes_type
+
       # New URL for the podcast feed
-      element :"itunes:new_feed_url", :as => :itunes_new_feed_url
-      element :"itunes:subtitle", :as => :itunes_subtitle
+      element :"itunes:new_feed_url", as: :itunes_new_feed_url
+      element :"itunes:subtitle", as: :itunes_subtitle
+
       # If summary is not present, use the description tag
-      element :"itunes:summary", :as => :itunes_summary
+      element :"itunes:summary", as: :itunes_summary
 
-      # iTunes RSS feeds can have multiple main categories...
-      # ...and multiple sub-categories per category
-      # TODO subcategories not supported correctly - they are at the same level
-      #   as the main categories
-      elements :"itunes:category", :as => :itunes_categories, :value => :text
+      # iTunes RSS feeds can have multiple main categories and multiple
+      # sub-categories per category.
+      elements :"itunes:category", as: :_itunes_categories,
+                                   class: ITunesRSSCategory
+      private :_itunes_categories
 
-      elements :"itunes:owner", :as => :itunes_owners, :class => ITunesRSSOwner
+      def itunes_categories
+        _itunes_categories.flat_map do |itunes_category|
+          itunes_category.enum_for(:each_subcategory).to_a
+        end
+      end
 
-      elements :item, :as => :entries, :class => ITunesRSSItem
+      def itunes_category_paths
+        _itunes_categories.flat_map do |itunes_category|
+          itunes_category.enum_for(:each_path).to_a
+        end
+      end
+
+      elements :"itunes:owner", as: :itunes_owners, class: ITunesRSSOwner
+      elements :item, as: :entries, class: ITunesRSSItem
 
       def self.able_to_parse?(xml)
-        /xmlns:itunes\s?=\s?\"http:\/\/www.itunes.com\/dtds\/podcast-1.0.dtd\"/i =~ xml
+        %r{xmlns:itunes\s?=\s?[\"\']http://www\.itunes\.com/dtds/podcast-1\.0\.dtd[\"\']}i =~ xml # rubocop:disable Metrics/LineLength
       end
-
     end
-
   end
-
 end

data/lib/feedjira/parser/itunes_rss_category.rb

@@ -0,0 +1,39 @@
+# rubocop:disable Style/DocumentationMethod
+module Feedjira
+  module Parser
+    # iTunes extensions to the standard RSS2.0 item
+    # Source: https://help.apple.com/itc/podcasts_connect/#/itcb54353390
+    class ITunesRSSCategory
+      include SAXMachine
+
+      attribute :text
+
+      elements :"itunes:category", as: :itunes_categories,
+                                   class: ITunesRSSCategory
+
+      def each_subcategory
+        return to_enum(__method__) unless block_given?
+
+        yield text
+
+        itunes_categories.each do |itunes_category|
+          itunes_category.each_subcategory(&proc)
+        end
+      end
+
+      def each_path(ancestors = [])
+        return to_enum(__method__, ancestors) unless block_given?
+
+        category_hierarchy = ancestors + [text]
+
+        if itunes_categories.empty?
+          yield category_hierarchy
+        else
+          itunes_categories.each do |itunes_category|
+            itunes_category.each_path(category_hierarchy, &proc)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/feedjira/parser/itunes_rss_item.rb

@@ -1,5 +1,4 @@
 module Feedjira
-
   module Parser
     # iTunes extensions to the standard RSS2.0 item
     # Source: http://www.apple.com/itunes/whatson/podcasts/specs.html
@@ -8,29 +7,42 @@ module Feedjira
       include FeedEntryUtilities
 
       element :author
-      element :guid, :as => :entry_id
+      element :guid, as: :entry_id
       element :title
-      element :link, :as => :url
-      element :description, :as => :summary
-      element :"content:encoded", :as => :content
-      element :pubDate, :as => :published
+      element :link, as: :url
+      element :description, as: :summary
+      element :"content:encoded", as: :content
+      element :pubDate, as: :published
 
       # If author is not present use author tag on the item
-      element :"itunes:author", :as => :itunes_author
-      element :"itunes:block", :as => :itunes_block
-      element :"itunes:duration", :as => :itunes_duration
-      element :"itunes:explicit", :as => :itunes_explicit
-      element :"itunes:keywords", :as => :itunes_keywords
-      element :"itunes:subtitle", :as => :itunes_subtitle
-      element :"itunes:image", :value => :href, :as => :itunes_image
-      element :"itunes:isClosedCaptioned", :as => :itunes_closed_captioned
-      element :"itunes:order", :as => :itunes_order
+      element :"itunes:author", as: :itunes_author
+      element :"itunes:block", as: :itunes_block
+      element :"itunes:duration", as: :itunes_duration
+      element :"itunes:explicit", as: :itunes_explicit
+      element :"itunes:keywords", as: :itunes_keywords
+      element :"itunes:subtitle", as: :itunes_subtitle
+      element :"itunes:image", value: :href, as: :itunes_image
+      element :"itunes:isClosedCaptioned", as: :itunes_closed_captioned
+      element :"itunes:order", as: :itunes_order
+      element :"itunes:season", as: :itunes_season
+      element :"itunes:episode", as: :itunes_episode
+      element :"itunes:title", as: :itunes_title
+      element :"itunes:episodeType", as: :itunes_episode_type
+
       # If summary is not present, use the description tag
-      element :"itunes:summary", :as => :itunes_summary
-      element :enclosure, :value => :length, :as => :enclosure_length
-      element :enclosure, :value => :type, :as => :enclosure_type
-      element :enclosure, :value => :url, :as => :enclosure_url
+      element :"itunes:summary", as: :itunes_summary
+      element :enclosure, value: :length, as: :enclosure_length
+      element :enclosure, value: :type, as: :enclosure_type
+      element :enclosure, value: :url, as: :enclosure_url
+      elements 'psc:chapter', as: :raw_chapters, class: Feedjira::Parser::PodloveChapter # rubocop:disable Metrics/LineLength
+
+      # Podlove requires clients to re-order by start time in the
+      # event the publisher doesn't provide them in that
+      # order. SAXMachine doesn't have any sort capability afaik, so
+      # we have to sort chapters manually.
+      def chapters
+        raw_chapters.sort_by(&:start)
+      end
     end
   end
-
 end

data/lib/feedjira/parser/itunes_rss_owner.rb

@@ -1,12 +1,12 @@
+# rubocop:disable Style/Documentation
 module Feedjira
-
   module Parser
     class ITunesRSSOwner
       include SAXMachine
       include FeedUtilities
-      element :"itunes:name", :as => :name
-      element :"itunes:email", :as => :email
+
+      element :"itunes:name", as: :name
+      element :"itunes:email", as: :email
     end
   end
-
 end

data/lib/feedjira/parser/podlove_chapter.rb

@@ -0,0 +1,22 @@
+# rubocop:disable Style/Documentation
+# rubocop:disable Style/DocumentationMethod
+module Feedjira
+  module Parser
+    class PodloveChapter
+      include SAXMachine
+      include FeedEntryUtilities
+      attribute :start, as: :start_ntp
+      attribute :title
+      attribute :href, as: :url
+      attribute :image
+
+      def start
+        return unless start_ntp
+        parts = start_ntp.split(':')
+        parts.reverse.to_enum.with_index.map do |part, index|
+          part.to_f * (60**index)
+        end.reduce(:+)
+      end
+    end
+  end
+end

data/lib/feedjira/parser/rss.rb

@@ -1,24 +1,27 @@
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
-
   module Parser
     # Parser for dealing with RSS feeds.
+    # Source: https://cyber.harvard.edu/rss/rss.html
     class RSS
       include SAXMachine
       include FeedUtilities
+      element :description
+      element :image, class: RSSImage
+      element :language
+      element :lastBuildDate, as: :last_built
+      element :link, as: :url
       element :rss, as: :version, value: :version
       element :title
-      element :description
-      element :link, :as => :url
-      elements :item, :as => :entries, :class => RSSEntry
-      elements :"atom:link", :as => :hubs, :value => :href, :with => {:rel => "hub"}
+      element :ttl
+      elements :"atom:link", as: :hubs, value: :href, with: { rel: 'hub' }
+      elements :item, as: :entries, class: RSSEntry
 
       attr_accessor :feed_url
 
-      def self.able_to_parse?(xml) #:nodoc:
+      def self.able_to_parse?(xml)
         (/\<rss|\<rdf/ =~ xml) && !(/feedburner/ =~ xml)
       end
     end
-
   end
-
 end

data/lib/feedjira/parser/rss_entry.rb

@@ -1,5 +1,4 @@
 module Feedjira
-
   module Parser
     # Parser for dealing with RDF feed entries.
     class RSSEntry
@@ -7,31 +6,28 @@ module Feedjira
       include FeedEntryUtilities
 
       element :title
-      element :link, :as => :url
-
-      element :"dc:creator", :as => :author
-      element :author, :as => :author
-      element :"content:encoded", :as => :content
-      element :description, :as => :summary
+      element :link, as: :url
 
-      element :"media:content", :as => :image, :value => :url
-      element :enclosure, :as => :image, :value => :url
+      element :"dc:creator", as: :author
+      element :author, as: :author
+      element :"content:encoded", as: :content
+      element :description, as: :summary
 
-      element :pubDate, :as => :published
-      element :pubdate, :as => :published
-      element :"dc:date", :as => :published
-      element :"dc:Date", :as => :published
-      element :"dcterms:created", :as => :published
+      element :"media:content", as: :image, value: :url
+      element :enclosure, as: :image, value: :url
 
+      element :pubDate, as: :published
+      element :pubdate, as: :published
+      element :"dc:date", as: :published
+      element :"dc:Date", as: :published
+      element :"dcterms:created", as: :published
 
-      element :"dcterms:modified", :as => :updated
-      element :issued, :as => :published
-      elements :category, :as => :categories
+      element :"dcterms:modified", as: :updated
+      element :issued, as: :published
+      elements :category, as: :categories
 
-      element :guid, :as => :entry_id
-      element :"dc:identifier", :as => :entry_id
+      element :guid, as: :entry_id
+      element :"dc:identifier", as: :entry_id
     end
-
   end
-
 end

data/lib/feedjira/parser/rss_feed_burner.rb

@@ -1,5 +1,5 @@
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
-
   module Parser
     # Parser for dealing with RSS feeds.
     class RSSFeedBurner
@@ -7,9 +7,10 @@ module Feedjira
       include FeedUtilities
       element :title
       element :description
-      element :link, :as => :url
-      elements :"atom10:link", :as => :hubs, :value => :href, :with => {:rel => "hub"}
-      elements :item, :as => :entries, :class => RSSFeedBurnerEntry
+      element :link, as: :url
+      element :lastBuildDate, as: :last_built
+      elements :"atom10:link", as: :hubs, value: :href, with: { rel: 'hub' }
+      elements :item, as: :entries, class: RSSFeedBurnerEntry
 
       attr_accessor :feed_url
 
@@ -17,7 +18,5 @@ module Feedjira
         (/\<rss|\<rdf/ =~ xml) && (/feedburner/ =~ xml)
       end
     end
-
   end
-
 end

data/lib/feedjira/parser/rss_feed_burner_entry.rb

@@ -1,43 +1,39 @@
+# rubocop:disable Style/DocumentationMethod
 module Feedjira
-
   module Parser
     # Parser for dealing with RDF feed entries.
     class RSSFeedBurnerEntry
-        include SAXMachine
-        include FeedEntryUtilities
-
-        element :title
-
-        element :"feedburner:origLink", :as => :url
-        element :link, :as => :url
+      include SAXMachine
+      include FeedEntryUtilities
 
-        element :"dc:creator", :as => :author
-        element :author, :as => :author
-        element :"content:encoded", :as => :content
-        element :description, :as => :summary
+      element :title
 
-        element :"media:content", :as => :image, :value => :url
-        element :enclosure, :as => :image, :value => :url
+      element :"feedburner:origLink", as: :url
+      element :link, as: :url
 
-        element :pubDate, :as => :published
-        element :pubdate, :as => :published
-        element :"dc:date", :as => :published
-        element :"dc:Date", :as => :published
-        element :"dcterms:created", :as => :published
+      element :"dc:creator", as: :author
+      element :author, as: :author
+      element :"content:encoded", as: :content
+      element :description, as: :summary
 
+      element :"media:content", as: :image, value: :url
+      element :enclosure, as: :image, value: :url
 
-        element :"dcterms:modified", :as => :updated
-        element :issued, :as => :published
-        elements :category, :as => :categories
+      element :pubDate, as: :published
+      element :pubdate, as: :published
+      element :"dc:date", as: :published
+      element :"dc:Date", as: :published
+      element :"dcterms:created", as: :published
 
-        element :guid, :as => :entry_id
+      element :"dcterms:modified", as: :updated
+      element :issued, as: :published
+      elements :category, as: :categories
 
-        def url
-          @url || @link
-        end
+      element :guid, as: :entry_id
 
+      def url
+        @url || @link
+      end
     end
-
   end
-
 end