codders-feedzirra 0.2.0.rc2

data/lib/feedzirra/feed_entry_utilities.rb

@@ -0,0 +1,65 @@
+module Feedzirra
+  module FeedEntryUtilities
+
+    include Enumerable
+
+    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
+    end
+
+    ##
+    # 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
+
+    ##
+    # 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
+    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
+    end
+
+    def sanitize!
+      self.title.sanitize! if self.title
+      self.author.sanitize! if self.author
+      self.summary.sanitize! if self.summary
+      self.content.sanitize! if self.content
+    end
+
+    alias_method :last_modified, :published
+
+    def each
+      @rss_fields ||= self.instance_variables
+
+      @rss_fields.each do |field|
+        yield(field.to_s.sub('@', ''), self.instance_variable_get(field))
+      end
+    end
+
+    def [](field)
+      self.instance_variable_get("@#{field.to_s}")
+    end
+
+    def []=(field, value)
+      self.instance_variable_set("@#{field.to_s}", value)
+    end
+
+  end
+end

data/lib/feedzirra/feed_utilities.rb

@@ -0,0 +1,72 @@
+module Feedzirra
+  module FeedUtilities
+    UPDATABLE_ATTRIBUTES = %w(title feed_url url last_modified etag)
+
+    attr_writer   :new_entries, :updated, :last_modified
+    attr_accessor :etag
+
+    def last_modified
+      @last_modified ||= begin
+        entry = entries.reject {|e| e.published.nil? }.sort_by { |entry| entry.published if entry.published }.last
+        entry ? entry.published : nil
+      end
+    end
+
+    def updated?
+      @updated
+    end
+
+    def new_entries
+      @new_entries ||= []
+    end
+
+    def has_new_entries?
+      new_entries.size > 0
+    end
+
+    def update_from_feed(feed)
+      self.new_entries += find_new_entries_for(feed)
+      self.entries.unshift(*self.new_entries)
+
+      @updated = false
+      UPDATABLE_ATTRIBUTES.each do |name|
+        updated = update_attribute(feed, name)
+        @updated ||= updated
+      end
+    end
+
+    def update_attribute(feed, name)
+      old_value, new_value = send(name), feed.send(name)
+
+      if old_value != new_value
+        send("#{name}=", new_value)
+      end
+    end
+
+    def sanitize_entries!
+      entries.each {|entry| entry.sanitize!}
+    end
+
+    private
+
+    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 url as the the newest for the feed
+      return feed.entries if self.entries.length == 0
+      latest_entry = self.entries.first
+      found_new_entries = []
+      feed.entries.each do |entry|
+        break if entry.url == latest_entry.url
+        found_new_entries << entry
+      end
+      found_new_entries
+    end
+
+    def existing_entry?(test_entry)
+      entries.any? { |entry| entry.url == test_entry.url }
+    end
+  end
+end

data/lib/feedzirra/parser.rb

@@ -0,0 +1,20 @@
+module Feedzirra
+  module Parser
+    autoload :RSS,                'feedzirra/parser/rss'
+    autoload :RSSEntry,           'feedzirra/parser/rss_entry'
+    autoload :RSSFeedBurner,      'feedzirra/parser/rss_feed_burner'
+    autoload :RSSFeedBurnerEntry, 'feedzirra/parser/rss_feed_burner_entry'
+    
+    autoload :ITunesRSS,      'feedzirra/parser/itunes_rss'
+    autoload :ITunesRSSItem,  'feedzirra/parser/itunes_rss_item'
+    autoload :ITunesRSSOwner, 'feedzirra/parser/itunes_rss_owner'
+
+    autoload :GoogleDocsAtom,      'feedzirra/parser/google_docs_atom'
+    autoload :GoogleDocsAtomEntry, 'feedzirra/parser/google_docs_atom_entry'
+
+    autoload :Atom,                'feedzirra/parser/atom'
+    autoload :AtomEntry,           'feedzirra/parser/atom_entry'
+    autoload :AtomFeedBurner,      'feedzirra/parser/atom_feed_burner'
+    autoload :AtomFeedBurnerEntry, 'feedzirra/parser/atom_feed_burner_entry'
+  end
+end

data/lib/feedzirra/parser/atom.rb

@@ -0,0 +1,29 @@
+module Feedzirra
+
+  module Parser
+    # Parser for dealing with Atom feeds.
+    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 => {:type => "application/atom+xml"}
+      elements :link, :as => :links, :value => :href
+      elements :entry, :as => :entries, :class => AtomEntry
+
+      def self.able_to_parse?(xml) #:nodoc:
+        /\<feed[^\>]+xmlns=[\"|\'](http:\/\/www\.w3\.org\/2005\/Atom|http:\/\/purl\.org\/atom\/ns\#)[\"|\'][^\>]*\>/ =~ xml
+      end
+
+      def url
+        @url || links.last
+      end
+
+      def feed_url
+        @feed_url ||= links.first
+      end
+    end
+  end
+
+end

data/lib/feedzirra/parser/atom_entry.rb

@@ -0,0 +1,30 @@
+module Feedzirra
+
+  module Parser
+    # Parser for dealing with Atom feed entries.
+    class AtomEntry
+      include SAXMachine
+      include FeedEntryUtilities
+      
+      element :title
+      element :link, :as => :url, :value => :href, :with => {:type => "text/html", :rel => "alternate"}
+      element :name, :as => :author
+      element :content
+      element :summary
+      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/feedzirra/parser/atom_feed_burner.rb

@@ -0,0 +1,21 @@
+module Feedzirra
+
+  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 :entry, :as => :entries, :class => AtomFeedBurnerEntry
+
+      def self.able_to_parse?(xml) #:nodoc:
+        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false
+      end
+    end
+
+  end
+
+end

data/lib/feedzirra/parser/atom_feed_burner_entry.rb

@@ -0,0 +1,31 @@
+module Feedzirra
+
+  module Parser
+    # Parser for dealing with Feedburner Atom feed entries.
+    class AtomFeedBurnerEntry
+      include SAXMachine
+      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 :summary
+      element :content
+      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
+      end
+
+    end
+  end
+
+end

data/lib/feedzirra/parser/google_docs_atom.rb

@@ -0,0 +1,28 @@
+require File.expand_path('./atom', File.dirname(__FILE__))
+
+module Feedzirra
+  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
+
+      def url
+        @url ||= links.first
+      end
+
+      def self.able_to_parse?(xml) #:nodoc:
+        %r{<id>https?://docs.google.com/.*\</id\>} =~ xml
+      end
+
+      def feed_url
+        @feed_url ||= links.first
+      end
+    end
+  end
+end

data/lib/feedzirra/parser/google_docs_atom_entry.rb

@@ -0,0 +1,29 @@
+module Feedzirra
+  module Parser
+    class GoogleDocsAtomEntry
+      include SAXMachine
+      include FeedEntryUtilities
+      
+      element :title
+      element :link, :as => :url, :value => :href, :with => {:type => "text/html", :rel => "alternate"}
+      element :name, :as => :author
+      element :content
+      element :summary
+      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
+      element :"docs:md5Checksum", :as => :checksum
+      element :"docs:filename", :as => :original_filename
+      element :"docs:suggestedFilename", :as => :suggested_filename
+
+      def url
+        @url ||= links.first
+      end
+    end
+  end
+end

data/lib/feedzirra/parser/itunes_rss.rb

@@ -0,0 +1,50 @@
+module Feedzirra
+
+  module Parser
+    # iTunes is RSS 2.0 + some apple extensions
+    # Source: http://www.apple.com/itunes/whatson/podcasts/specs.html
+    class ITunesRSS
+      include SAXMachine
+      include FeedUtilities
+
+      attr_accessor :feed_url
+
+      # RSS 2.0 elements that need including
+      element :copyright
+      element :description
+      element :language
+      element :managingEditor
+      element :title
+      element :link, :as => :url
+
+      # 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
+      # New URL for the podcast feed
+      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
+
+      # 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
+
+      elements :"itunes:owner", :as => :itunes_owners, :class => ITunesRSSOwner
+
+      elements :item, :as => :entries, :class => ITunesRSSItem
+
+      def self.able_to_parse?(xml)
+        /xmlns:itunes=\"http:\/\/www.itunes.com\/dtds\/podcast-1.0.dtd\"/i =~ xml
+      end
+
+    end
+
+  end
+
+end

data/lib/feedzirra/parser/itunes_rss_item.rb

@@ -0,0 +1,32 @@
+module Feedzirra
+  
+  module Parser
+    # iTunes extensions to the standard RSS2.0 item
+    # Source: http://www.apple.com/itunes/whatson/podcasts/specs.html
+    class ITunesRSSItem
+      include SAXMachine
+      include FeedEntryUtilities
+
+      element :author
+      element :guid
+      element :title
+      element :link, :as => :url
+      element :description, :as => :summary
+      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
+      # 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
+    end
+  end
+  
+end

data/lib/feedzirra/parser/itunes_rss_owner.rb

@@ -0,0 +1,12 @@
+module Feedzirra
+  
+  module Parser
+    class ITunesRSSOwner
+      include SAXMachine
+      include FeedUtilities
+      element :"itunes:name", :as => :name
+      element :"itunes:email", :as => :email
+    end
+  end
+  
+end

data/lib/feedzirra/parser/rss.rb

@@ -0,0 +1,22 @@
+module Feedzirra
+
+  module Parser
+    # Parser for dealing with RSS feeds.
+    class RSS
+      include SAXMachine
+      include FeedUtilities
+      element :title
+      element :description
+      element :link, :as => :url
+      elements :item, :as => :entries, :class => RSSEntry
+
+      attr_accessor :feed_url
+
+      def self.able_to_parse?(xml) #:nodoc:
+        (/\<rss|\<rdf/ =~ xml) && !(/feedburner/ =~ xml)
+      end
+    end
+
+  end
+
+end