generate_data.rb

#!/usr/bin/env ruby

$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), 'vendor', 'rpodcast', 'lib'))

options = {}
ARGV.each do |arg|
  key, value = arg.split('=')
  options[key.intern] = value.to_i
end
options[:limit] ||= 1000
options[:offset] ||= 0

require 'rubygems'
require 'rpodcast'
require 'open-uri'
require 'json'

def leading_zero(num)
  num < 10 ? "0#{num}" : num.to_s
end

class String
  def truncate(size)
    self.size > size ? self[0..size] : self
  end
end

# Audios

content = open("http://www-tracey.archive.org/services/collection-rss.php?collection=78rpm").read

feed = RPodcast::Feed.new(content)
feed.parse

entries = feed.episodes.slice(options[:offset], options[:limit]).map{|e|
  artist, title = e.title.split('-')
  duration = "#{rand(8)}:#{leading_zero(rand(60))}"
  album = e.title.split(' ').reverse.join(' ').truncate(48)
  { 'Artist' => artist.truncate(39), 'Link' => e.enclosure.url, 'Name' => title.truncate(39), 'Album' => album.truncate(39), 'Length' => duration, 'Type' => 'Audio' } if title
} rescue nil

# Videos 

content = open("http://www-tracey.archive.org/services/collection-rss.php?mediatype=movies").read

feed = RPodcast::Feed.new(content)
feed.parse

entries << feed.episodes.slice(options[:offset], options[:limit]).map{|e|
  title = e.title.truncate(90)
  duration = "#{rand(15)}:#{leading_zero(rand(60))}"
  { 'Name' => e.title, 'Link' => e.enclosure.url, 'Length' => duration, 'Type' => 'Video' } if title
} rescue nil

# Images

content = open("http://api.flickr.com/services/feeds/photos_public.gne?id=51035599008@N01&lang=en-us&format=rss_200").read

feed = RPodcast::Feed.new(content)
feed.parse

entries << feed.episodes.slice(options[:offset], options[:limit]).map{|e|
  title = e.title.truncate(80)
  duration = "#{rand(15)}:#{leading_zero(rand(60))}"
  { 'Name' => e.title, 'Link' => e.enclosure.url, 'Type' => 'Image' } if title
} rescue nil

# Documents

content = open("http://www.scribd.com/feeds/group_rss/1257").read

doc = Hpricot.XML(content)

entries << feed.episodes.slice(options[:offset], options[:limit]).map{|e|
  title = e.at('title').inner_html.truncate(80)
  { 'Name' => title, 'Size' => e.at('description').to_s.size * rand(25), 'Type' => 'Document' }
} rescue nil

schema = {
  'Audio' => ['Sharing', 'Name', 'Artist', 'Album', 'Length'],
  'Video' => ['Sharing', 'Name', 'Length'],
  'Images' => ['Sharing', 'Name'],
  'Documents' => ['Sharing', 'Name', 'Type']
}

response = {'schema' => schema, 'files' => entries.flatten.compact}

puts JSON.pretty_generate(response)