#!/usr/bin/env ruby

### Don't need to edit below here
%w{ rubygems hpricot open-uri }.each {|gem| require gem}

class Twitterchive
  def initialize(tag)
    @tag = tag
  end
  def parse
    fields = ["id","title", "published","author/name","author/uri"]
    @entries = [fields.join(",") + ",link"]
    next_url = "http://search.twitter.com/search.atom?tag=#{@tag}&lang=all&rpp=50"
    while (next_url)
      begin
        puts next_url
        doc = Hpricot.parse(open(next_url))

        @entries << (doc/:entry).collect do |entry|
          fields.collect {|f| (entry/f).inner_text }.join(",") + ",#{(entry/"link").first["href"]}"
        end
        next_url = (doc/"link[@rel=next]").first["href"]
        sleep(0.2)
      rescue
        next_url = nil
      end
    end
  end

  def save(filename = nil)
    File.open(filename || "#{@tag}.csv","w") do |file|
      file << @entries.join("\n")
    end
  end
end

if __FILE__ == $0
  %w{ ostruct optparse }.each {|gem| require gem}
  # Parse the command line options
  options = OpenStruct.new
  opts = OptionParser.new
  opts.banner = "Usage: twitterchive.rb [options] archive_filename"
  opts.on("-t [tag]", "--tag [tag]", "Twitter hashtag to archive") {|tag| options.tag = tag }
  
  opts.on_tail("-h", "--help", "Show this message") do
    puts opts
    exit
  end
  file = opts.parse(ARGV)
  
  unless options.tag.nil? || options.tag == ""
    tag_archive = Twitterchive.new(options.tag)
    tag_archive.parse
    tag_archive.save(file.first)
  end
end