spider.rb

require 'rubygems'
require 'tumblr_client'
require 'active_record'


GIF_MAX_WIDTH  = 1024
GIF_MAX_HEIGHT = 768

Tumblr.configure do |config|
  config.consumer_key = "AcHYlG5GaS9xf842H1RC8WF81cvAH9zqf7fvxGSrfHQSy7ksqO"
  config.consumer_secret = "SYfZQUh6kGgEG3zAtQwO9jJwGEDcVXBMRg1QUTRIMZqBqDHMUR"
end

$tumblr = Tumblr.new

class IndexedGif
  def initializer()
  end
  attr_accessor :gif_url, :source_url, :source_name, :source_id, :tags, :caption, :individual_caption

  def store
    @indexed_time = Time.now.getutc
    ## Make this serialize and store to database
    puts self.inspect
  end
end

class String
  def ends_with?(str)
    str = str.to_str
    tail = self[-str.length, str.length]
    tail == str
  end
end

def scour_account(baseurl)

end


def process_post(post)
  if(post['source_url'])
    # This isn't the original source of this post! Go get it from there.
    #TODO: finish this
  else
    # This is the original source of this post

    post['photos'].each do |photo|
      image = photo['original_size']

      unless image['url'].ends_with?(".gif")
        # Not a gif, discard
        next
      end

      if image['width'] > GIF_MAX_WIDTH or image['height'] > GIF_MAX_HEIGHT
        # Image too large, discard
        next
      end

      gif = IndexedGif.new
      gif.gif_url             = image['url']
      gif.source_url          = post['post_url']
      gif.source_name         = post['blog_name']
      gif.source_id           = post['id']
      gif.tags                = post['tags']
      gif.caption             = post['caption']
      gif.individual_caption  = photo['caption']


      gif.store
    end
  end
end
#scour_account "gaysexistheanswer.tumblr.com"