RSS Feed processor class

gistfile1.rb

require 'rss/1.0'
require 'rss/2.0'
require 'open-uri'
require 'json'

# Class handles reading, processing RSS feeds
#
# Examples
#
#   # GOOD
#   Feed.run("http://celebuzz.com/feed", 1)
#   # => [ [ "[item 1]", "[title 1]", "[image link 1]" ] ]
#
#   Feed.run("http://celebuzz.com/feed", 4, [ :link, :title, :description ])
#   # => [ [ "[item 1]", "[title 1]", "[description 1]", "[image link 1]" ],
#          [ "[item 2]", "[title 2]", "[description 2]", "[image link 2]" ],
#          [ "[item 3]", "[title 3]", "[description 3]", "[image link 3]" ],
#          [ "[item 4]", "[title 4]", "[description 4]", "[image link 4]" ] ]
#
#   # BAD
#   Feed.run("http://celebuzz.com/fe", 1)
#   # => [ [ :exception, ""http://celebuzz.com/fe", "404..." ], ]
class Feed

  # Public: Initialize a feed
  #
  # options - Optional values to instantiate attributes to
  #         :source - RSS url
  #
  # Examples
  #
  #   Feed.new :source => "http://foo.com"
  #   # => #<Feed..
  #
  # Returns a Feed instance
  def initialize(options={})
    @source = options[:source] if options.key?(:source)
  end

  # Public: Sets attributes hash with values from rss feed
  #
  # attrs - Array of atttribute keys as symbols
  #
  # Examples
  #
  #   #info([ :link ])
  #   # => [ :link ]
  #
  # Returns an array
  def info(attrs = [:link,:title,:description])
    attrs.each { |x| self.attributes[x] = channel.send x }
  end

  # Public: Return an array of values from an rss feed
  #
  # n     - number of items to return
  # attrs - array of attributes to return for each rss item
  # NOTE: the first image in the description is added to the return array
  #
  # Example
  #   to_ary(1, attrs[:link])
  #   # => [ [ "http://example.net", "http://example.net/image.png" ] ]
  #
  # Returns a collection of arrays
  def to_ary(n = 5, attrs=[:link,:title])
    @ary ||= channel.items.first(n).map do |item|
        tmp = attrs.map { |attr| item.send attr }

        # Find the first image
        if item.description
          item.description =~ /<img [^>]*src=["|\']([^"|\']+)/i
          tmp << $1
        end
        tmp
    end
  end

  # Public: Processes feeds and returns array of data
  #
  # feeds - One or more feed urls
  #
  # Examples
  #
  #   Feed.run("http://example.net")
  #   # => [ [ "http://example.net/item1.html", "[image]" [, ... ]
  #
  #
  #
  # NOTE: Array may contain :exception as first element if error
  # Returns an array
  def self.run(feeds, n = 5, attrs = [ :link, :title, :description ])
    Array(feeds).map do |source|
      begin
        new(source: source).to_ary(n, attrs)
      rescue Exception => blasto
        [ :exception, source, blasto.to_s ]
      end
    end
  end

  private

  attr_accessor :content
  attr_accessor :source

  def channel
    @channel ||= parse.channel
  end

  def parse
    open(source) { |s| self.content = s.read }
    rss = RSS::Parser.parse(content, false)
  end

  def attributes
    @attributes ||= {}
  end
end

if $0 == __FILE__
  feeds = [
    "http://celebuzz.com/feed", # good
    "http://celebuzz.com/fee",  # bad
  ]
  p Feed.run(feeds, 1)
end