Capture tweets for a given lat/lon bounding box

stream.rb

#!/usr/bin/env ruby
$: << File.dirname(__FILE__)

require 'rubygems'
require 'eventmachine'
require 'em-http'
require 'yajl'
require 'yajl/json_gem'

class Streamer
  
  def initialize
    Signal.trap("INT")  { EM.stop }
    Signal.trap("TERM") { EM.stop }
    EventMachine.run {
      Hose.new
    }
    puts "shutting down!"
  end
  
end

class Hose
  
  PARSER = Yajl::Parser.new
  CREDS = ["TWITTER_USERNAME", "TWITTER_PASSWORD"]
  DATADIR = "/home/twittervision/baltimore"
  
  def initialize
    @http = EventMachine::HttpRequest.new('https://stream.twitter.com/1/statuses/filter.json',
        :connection_timeout => 0, :inactivity_timeout => 0).post(
      :head => {'authorization' => CREDS},
      :body => {:locations => '-76.75,39.18,-76.43,39.40'}
    )
    
    PARSER.on_parse_complete = proc { |h| process(h) }
    @http.stream { |json| PARSER << json }
    @http.errback { |e| restart }
    @http.callback { |c| restart }
  end

  def process(s)
    if s['geo']
      lat = s['geo']['coordinates'][0]
      lon = s['geo']['coordinates'][1]
    elsif s['coordinates']
      lat = s['coordinates']['coordinates'][1]
      lon = s['coordinates']['coordinates'][0]
    elsif s['place']
      c = s['place']['bounding_box']['coordinates'][0]
      lat = (c[0][1] + c[2][1])/2.0
      lon = (c[0][0] + c[1][0])/2.0
    end
    s['lat'] = lat
    s['lon'] = lon
    
    return unless lat && lon && (lat>=39.18 && lon>=-76.75) && (lat<=39.40 && lon<=76.43)
    
    ll = "#{lat},#{lon}"
    http2 = EventMachine::HttpRequest.new("http://maps.googleapis.com/maps/api/geocode/json?latlng=#{ll}\&sensor=false").get
    http2.callback { display(s, http2.response) }
  end
  
  def datestring
    Time.now.strftime("%Y%m%d")
  end

  def display(s,c)
    begin
      addr = JSON.parse(c)
      s['formatted_address'] = addr['results'].first['formatted_address']
    rescue
      s['formatted_address'] = "[#{s['lat']}, #{s['lon']}]"
    end
      
    ts = Time.now
    s['timestamp'] = ts.to_i
    File.open("#{DATADIR}/#{datestring}.txt", "a") do |f|
      f.write "#{s['user']['screen_name']}: #{s['text']}\n#{s['formatted_address']} (#{ts.strftime('%H:%M:%S')})\n\n"
    end
    
    File.open("#{DATADIR}/#{datestring}.json", "a") do |f|
      f.write s.to_json + "\n"
    end
  end
  
  def restart
    puts "restarting"
    Hose.new
  end
end

Streamer.new