Script to generate an Atom feed from the VirtualBox News page

virtualbox_feed.rb

#!/usr/bin/ruby

require 'rubygems'
require 'nokogiri'
require 'open-uri'
require 'uuidtools'
require 'date'
require 'activesupport' # For DateTime.to_time
require 'uri'

# Create an Atom feed from the VirtualBox news page
class FeedBuilder

    def initialize(url)
        @url = url
        @doc = Nokogiri::HTML(open(url))
        build_feed
    end

    def parse_date(node)
        month, day, year = node.content.split(' ')
        year = year[0..3] # First four chars (get rid of the '.')
        months = %w(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec)
        DateTime.civil(year.to_i, months.index(month) + 1, day.to_i)
    end

    def extract_title(content)
        if content =~ /^(.*?)[,!\.] /
            title = $1.strip.split(' ')
            if title.size > 10
                title = title[0..9] # First 10 words
                title[-1] += '...'
            end
            return title.join(' ')
        else
            raise "Unable to extract title from: #{content}"
        end
    end

    # Convert relative links to absolute
    def resolve_links
        @doc.css('a').each do |a|
            url = URI.parse(a['href'])
            if url.host.nil?
                url = URI::HTTP.build({
                    :host => 'www.virtualbox.org',
                    :path => url.path
                })
                a['href'] = url.to_s
            end
        end
    end

    def build_feed
        resolve_links
        @builder = Nokogiri::XML::Builder.new do |feed|
            feed.feed(:xmlns => "http://www.w3.org/2005/Atom") do
                last_updated = parse_date(@doc.css('div#searchable ul li strong').first)
                feed.title "VirtualBox News"
                feed.link(:href => @url)
                feed.link(
                    :rel => "self",
                    :type => "application/atom+xml",
                    :href => 'http://home.wezm.net/files/virtualbox.atom'
                )
                feed.updated last_updated.to_s
                feed.author {
                    feed.name "Sun VirtualBox"
                }
                feed.id_ 'urn:uuid:' + UUID.timestamp_create(last_updated.to_time).to_s

                # Add the entries, limit to 10 items
                @doc.css('div#searchable ul li').each_with_index do |item, index|
                    break if index >= 10
                    feed.entry {
                        updated = parse_date(item.css('strong').first.remove)
                        content = item.inner_html.lstrip
                        feed.title extract_title(item.content)
                        feed.link(:rel => "alternate", :type => "text/html", :href => @url)
                        feed.id_ 'urn:uuid:' + UUID.timestamp_create(updated.to_time).to_s
                        feed.updated updated.to_s
                        feed.content(content, :type => "html")
                    }
                end
            end
        end
    end

    def to_s
        @builder.to_xml
    end
end

if ARGV.size < 1
    puts "Usage: virtualbox_feed output.atom"
    exit 2
end

feed = FeedBuilder.new('http://www.virtualbox.org/wiki/News')
File.open(ARGV[0], 'w') do |atom|
    atom << feed.to_s
end