feedjira 2.2.0 → 3.1.2

data/lib/feedjira/date_time_utilities/date_time_pattern_parser.rb

@@ -1,32 +1,28 @@
-# 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
-      PATTERNS = ['%m/%d/%Y %T %p', '%d %m %Y %T %Z'].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
 
       def self.prepare(string)
         rgx = Regexp.new("^(#{JAPANESE_SYMBOLS.join('|')}),\s")
-        string.gsub(rgx, '')
+        string.gsub(rgx, "")
       end
       private_class_method :prepare
     end

data/lib/feedjira/feed.rb

@@ -1,60 +1,35 @@
-# rubocop:disable Style/Documentation
-# rubocop:disable Style/DocumentationMethod
+# frozen_string_literal: true
+
 module Feedjira
   class Feed
     class << self
-      def parse_with(parser, xml, &block)
-        parser.parse xml, &block
-      end
-
-      def parse(xml, &block)
-        parser = determine_feed_parser_for_xml(xml)
-        raise NoParserAvailable, 'No valid parser for XML.' unless parser
-        parse_with parser, xml, &block
-      end
-
-      def determine_feed_parser_for_xml(xml)
-        start_of_doc = xml.slice(0, 2000)
-        feed_classes.detect { |klass| klass.able_to_parse?(start_of_doc) }
-      end
-
-      def add_feed_class(klass)
-        feed_classes.unshift klass
-      end
-
-      def feed_classes
-        @feed_classes ||= Feedjira.parsers
-      end
-
-      def reset_parsers!
-        @feed_classes = nil
-      end
-
       def add_common_feed_element(element_tag, options = {})
-        feed_classes.each do |k|
-          k.element element_tag, options
+        Feedjira.parsers.each do |k|
+          k.element(element_tag, options)
         end
       end
 
       def add_common_feed_elements(element_tag, options = {})
-        feed_classes.each do |k|
-          k.elements element_tag, options
+        Feedjira.parsers.each do |k|
+          k.elements(element_tag, options)
         end
       end
 
       def add_common_feed_entry_element(element_tag, options = {})
-        call_on_each_feed_entry :element, element_tag, options
+        call_on_each_feed_entry(:element, element_tag, options)
       end
 
       def add_common_feed_entry_elements(element_tag, options = {})
-        call_on_each_feed_entry :elements, element_tag, options
+        call_on_each_feed_entry(:elements, element_tag, options)
       end
 
+      private
+
       def call_on_each_feed_entry(method, *parameters)
-        feed_classes.each do |klass|
+        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.class == Class
             end
 
             collection_configs.each do |config|
@@ -63,51 +38,6 @@ module Feedjira
           end
         end
       end
-
-      def fetch_and_parse(url)
-        response = connection(url).get
-        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.delete '"'
-
-        feed.last_modified = parse_last_modified(response)
-        feed
-      end
-
-      # rubocop:disable LineLength
-      def connection(url)
-        Faraday.new(url: url, headers: headers, request: request_options) do |conn|
-          conn.use FaradayMiddleware::FollowRedirects, limit: Feedjira.follow_redirect_limit
-          conn.adapter(*Faraday.default_adapter)
-        end
-      end
-      # rubocop:enable LineLength
-
-      private
-
-      def headers
-        {
-          user_agent: Feedjira.user_agent
-        }
-      end
-
-      def request_options
-        {
-          timeout: Feedjira.request_timeout
-        }
-      end
-
-      def parse_last_modified(response)
-        lm = response.headers['last-modified']
-        DateTime.parse(lm).to_time
-      rescue StandardError => e
-        Feedjira.logger.warn { "Failed to parse last modified '#{lm}'" }
-        Feedjira.logger.debug(e)
-        nil
-      end
     end
   end
 end

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
@@ -13,16 +13,18 @@ module Feedjira
       DateTime.parse(string).feed_utils_to_gm_time
     rescue StandardError => e
       Feedjira.logger.warn { "Failed to parse date #{string.inspect}" }
-      Feedjira.logger.debug(e)
+      Feedjira.logger.warn(e)
       nil
     end
 
     ##
     # 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.
@@ -39,9 +41,9 @@ module Feedjira
     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 +1,6 @@
-module Feedjira::Parser; end # rubocop:disable Style/Documentation
+# frozen_string_literal: true
+
+module Feedjira
+  module Parser
+  end
+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.
@@ -8,14 +9,14 @@ 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: { rel: 'self' }
+      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 :link, as: :hubs, value: :href, with: { rel: "hub" }
       elements :entry, as: :entries, class: AtomEntry
 
       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

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.
@@ -9,18 +10,18 @@ module Feedjira
       element :title
       element :subtitle, as: :description
       element :link, as: :url_text_html, value: :href,
-                     with: { type: 'text/html' }
+                     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' }
+                              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
+        ((/<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