feedjira 3.0.0.beta1 → 3.2.0

data/lib/feedjira/date_time_utilities/date_time_epoch_parser.rb

@@ -1,11 +1,12 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module DateTimeUtilities
     class DateTimeEpochParser
       def self.parse(string)
         epoch_time = string.to_i
         return Time.at(epoch_time).to_datetime if epoch_time.to_s == string
+
         raise "#{string} is not a valid epoch time"
       end
     end

data/lib/feedjira/date_time_utilities/date_time_language_parser.rb

@@ -1,12 +1,12 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module DateTimeUtilities
     class DateTimeLanguageParser
       MONTHS_ENGLISH =
-        %w(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec).freeze
+        %w[Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec].freeze
       MONTHS_SPANISH =
-        %w(Ene Feb Mar Abr May Jun Jul Ago Sep Oct Nov Dic).freeze
+        %w[Ene Feb Mar Abr May Jun Jul Ago Sep Oct Nov Dic].freeze
 
       def self.parse(string)
         DateTime.parse(translate(string))

data/lib/feedjira/date_time_utilities/date_time_pattern_parser.rb

@@ -1,25 +1,21 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module DateTimeUtilities
     class DateTimePatternParser
-      # rubocop:disable Style/AsciiComments
       # Japanese Symbols are required for strange Date Strings like
       # '水, 31 8 2016 07:37:00 PDT'
-      JAPANESE_SYMBOLS = %w(日 月 火 水 木 金 土).freeze
+      JAPANESE_SYMBOLS = %w[日 月 火 水 木 金 土].freeze
       PATTERNS = ["%m/%d/%Y %T %p", "%d %m %Y %T %Z"].freeze
 
-      # rubocop:disable Metrics/MethodLength
       def self.parse(string)
         PATTERNS.each do |p|
-          begin
-            datetime = DateTime.strptime(prepare(string), p)
-            return datetime
-          rescue StandardError => e
-            Feedjira.logger.debug("Failed to parse date #{string}")
-            Feedjira.logger.debug(e)
-            nil
-          end
+          datetime = DateTime.strptime(prepare(string), p)
+          return datetime
+        rescue StandardError => e
+          Feedjira.logger.debug("Failed to parse date #{string}")
+          Feedjira.logger.debug(e)
+          nil
         end
         raise "No pattern matched #{string}"
       end

data/lib/feedjira/feed.rb

@@ -1,5 +1,5 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   class Feed
     class << self
@@ -29,7 +29,7 @@ module Feedjira
         Feedjira.parsers.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
+              v.accessor == "entries" && v.data_class.is_a?(Class)
             end
 
             collection_configs.each do |config|

data/lib/feedjira/feed_entry_utilities.rb

@@ -1,5 +1,5 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module FeedEntryUtilities
     include Enumerable
@@ -20,28 +20,30 @@ module Feedjira
     ##
     # Returns the id of the entry or its url if not id is present, as some
     # formats don't support it
+    # rubocop:disable Naming/MemoizedInstanceVariableName
     def id
       @entry_id ||= @url
     end
+    # rubocop:enable Naming/MemoizedInstanceVariableName
 
     ##
     # Writer for published. By default, we keep the "oldest" publish time found.
     def published=(val)
       parsed = parse_datetime(val)
-      @published = parsed if !@published || parsed < @published
+      @published = parsed if parsed && (!@published || parsed < @published)
     end
 
     ##
     # Writer for updated. By default, we keep the most recent update time found.
     def updated=(val)
       parsed = parse_datetime(val)
-      @updated = parsed if !@updated || parsed > @updated
+      @updated = parsed if parsed && (!@updated || parsed > @updated)
     end
 
     def sanitize!
-      %w(title author summary content image).each do |name|
+      %w[title author summary content image].each do |name|
         if respond_to?(name) && send(name).respond_to?(:sanitize!)
-          send(name).send :sanitize!
+          send(name).send(:sanitize!)
         end
       end
     end
@@ -49,10 +51,15 @@ module Feedjira
     alias last_modified published
 
     def each
-      @rss_fields ||= instance_variables
+      @rss_fields ||= instance_variables.map do |ivar|
+        ivar.to_s.sub("@", "")
+      end.select do |field| # rubocop:disable Style/MultilineBlockChain
+        # select callable (public) methods only
+        respond_to?(field)
+      end
 
       @rss_fields.each do |field|
-        yield(field.to_s.sub("@", ""), instance_variable_get(field))
+        yield(field, instance_variable_get(:"@#{field}"))
       end
     end
 

data/lib/feedjira/feed_utilities.rb

@@ -1,8 +1,8 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module FeedUtilities
-    UPDATABLE_ATTRIBUTES = %w(title feed_url url last_modified etag).freeze
+    UPDATABLE_ATTRIBUTES = %w[title feed_url url last_modified etag].freeze
 
     attr_writer   :new_entries, :updated, :last_modified
     attr_accessor :etag
@@ -43,7 +43,7 @@ module Feedjira
     def last_modified
       @last_modified ||= begin
         published = entries.reject { |e| e.published.nil? }
-        entry = published.sort_by { |e| e.published if e.published }.last
+        entry = published.max_by(&:published)
         entry ? entry.published : nil
       end
     end
@@ -102,6 +102,7 @@ module Feedjira
 
       feed.entries.each do |entry|
         break unless new_entry?(entry, latest_entry)
+
         found_new_entries << entry
       end
 

data/lib/feedjira/parser.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
   end

data/lib/feedjira/parser/atom.rb

@@ -1,4 +1,5 @@
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
     # Parser for dealing with Atom feeds.
@@ -13,17 +14,14 @@ module Feedjira
       elements :link, as: :links, value: :href
       elements :link, as: :hubs, value: :href, with: { rel: "hub" }
       elements :entry, as: :entries, class: AtomEntry
+      element :icon
 
       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
+        %r{<feed[^>]+xmlns\s?=\s?["'](http://www\.w3\.org/2005/Atom|http://purl\.org/atom/ns\#)["'][^>]*>} =~ xml
       end
 
       def url
-        @url || (links - [feed_url]).last || links.last
-      end
-
-      def feed_url
-        @feed_url ||= links.first
+        @url || (links - [feed_url]).last
       end
 
       def self.preprocess(xml)

data/lib/feedjira/parser/atom_entry.rb

@@ -1,32 +1,15 @@
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
     # Parser for dealing with Atom feed entries.
     class AtomEntry
       include SAXMachine
       include FeedEntryUtilities
+      include AtomEntryUtilities
 
-      element :title
-      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:thumbnail", as: :image, value: :url
       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 :updated
-      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

@@ -1,4 +1,5 @@
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
     # Parser for dealing with Feedburner Atom feeds.
@@ -11,7 +12,7 @@ module Feedjira
       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 :link, as: :feed_url_link, value: :href, with: { type: "application/atom+xml" }
       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" }
@@ -20,7 +21,7 @@ module Feedjira
       attr_writer :url, :feed_url
 
       def self.able_to_parse?(xml)
-        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false # rubocop:disable Metrics/LineLength
+        ((/<feed/ =~ xml) && (/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/<rss|<rdf/ =~ xml)) || false
       end
 
       # Feed url is <link> with type="text/html" if present,

data/lib/feedjira/parser/atom_feed_burner_entry.rb

@@ -1,32 +1,21 @@
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
     # Parser for dealing with Feedburner Atom feed entries.
     class AtomFeedBurnerEntry
       include SAXMachine
       include FeedEntryUtilities
+      include AtomEntryUtilities
 
-      element :title
-      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 :"feedburner:origLink", as: :orig_link
+      private :orig_link
 
+      element :"media:thumbnail", as: :image, value: :url
       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 :updated
-      element :modified, as: :updated
-      elements :category, as: :categories, value: :term
-      elements :link, as: :links, value: :href
 
       def url
-        @url ||= links.first
+        orig_link || super
       end
     end
   end

data/lib/feedjira/parser/atom_google_alerts.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Feedjira
+  module Parser
+    # Parser for dealing with Feedburner Atom feeds.
+    class AtomGoogleAlerts
+      include SAXMachine
+      include FeedUtilities
+
+      element :title
+      element :subtitle, as: :description
+      element :link, as: :feed_url, value: :href, with: { rel: "self" }
+      element :link, as: :url, value: :href, with: { rel: "self" }
+      elements :link, as: :links, value: :href
+      elements :entry, as: :entries, class: AtomGoogleAlertsEntry
+
+      def self.able_to_parse?(xml)
+        Atom.able_to_parse?(xml) && (%r{<id>tag:google\.com,2005:[^<]+/com\.google/alerts/} === xml) # rubocop:disable Style/CaseEquality
+      end
+
+      def self.preprocess(xml)
+        Preprocessor.new(xml).to_xml
+      end
+    end
+  end
+end

data/lib/feedjira/parser/atom_google_alerts_entry.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module Feedjira
+  module Parser
+    # Parser for dealing with Feedburner Atom feed entries.
+    class AtomGoogleAlertsEntry
+      include SAXMachine
+      include FeedEntryUtilities
+      include AtomEntryUtilities
+
+      def url
+        url = super
+        return unless url&.start_with?("https://www.google.com/url?")
+
+        uri = URI(url)
+        cons = URI.decode_www_form(uri.query).assoc("url")
+        cons && cons[1]
+      end
+    end
+  end
+end

data/lib/feedjira/parser/atom_youtube.rb

@@ -1,4 +1,5 @@
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
     # Parser for dealing with RSS feeds.

data/lib/feedjira/parser/atom_youtube_entry.rb

@@ -1,19 +1,20 @@
-# rubocop:disable Style/Documentation
+# frozen_string_literal: true
+
 module Feedjira
   module Parser
     class AtomYoutubeEntry
       include SAXMachine
       include FeedEntryUtilities
+      include AtomEntryUtilities
+
+      sax_config.top_level_elements["link"].clear
+      sax_config.collection_elements["link"].clear
 
-      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 :"yt:channelId", as: :youtube_channel_id
       element :"media:title", as: :media_title
       element :"media:content", as: :media_url, value: :url
       element :"media:content", as: :media_type, value: :type

data/lib/feedjira/parser/globally_unique_identifier.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module Feedjira
+  module Parser
+    class GloballyUniqueIdentifier
+      include SAXMachine
+
+      attribute :isPermaLink, as: :is_perma_link
+
+      value :guid
+
+      def perma_link?
+        is_perma_link != "false"
+      end
+
+      def url
+        perma_link? ? guid : nil
+      end
+    end
+  end
+end

data/lib/feedjira/parser/google_docs_atom.rb

@@ -1,6 +1,6 @@
+# frozen_string_literal: true
+
 require File.expand_path("./atom", File.dirname(__FILE__))
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
 module Feedjira
   module Parser
     class GoogleDocsAtom
@@ -9,7 +9,7 @@ module Feedjira
       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" } # rubocop:disable Metrics/LineLength
+      element :link, as: :feed_url, value: :href, with: { type: "application/atom+xml" }
       elements :link, as: :links, value: :href
       elements :entry, as: :entries, class: GoogleDocsAtomEntry
 
@@ -18,7 +18,7 @@ module Feedjira
       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