Screenscraping London Open House venues into GeoJSON

gistfile1.rb

# curl -O 'http://events.londonopenhouse.org/Venues?q=&Page=[1-80]'
# grep 'href.*/building/' * | cut -d\" -f 2 | sort -u | sed -e 's/.*/http:\/\/events.londonopenhouse.org\/&/' > buildings.txt
# gem install nokogiri
# ruby this_script.rb > loh.json
# ogr2ogr -f KML loh.kml loh.json

require 'nokogiri'
require 'json'

DETAILED = false # Google Maps complains if the KML gets too big

features = []
Dir.glob("[0-9-]*").each { |f|
  html = open(f).read
  doc = Nokogiri::HTML(html)
  if f.match(/^-/)
    f = f.slice(1,1000)
  end
  url = "http://events.londonopenhouse.org/building/#{f}"
  x, lat, lng = doc.to_s.match(/new Microsoft.Maps.Location\((.*),(.*)\)/).to_a
  lat = lat.to_f
  lng = lng.to_f
  if DETAILED
    description = doc.css("#body").children.to_s.strip
  else
    description = doc.css("address").first.children.to_s.strip + "<br><a href='#{url}'>#{url}</a>"
  end
  properties = {
    'name' => doc.title,
    'description' => description,
    'href' => url
  }
  if DETAILED
    doc.css("fieldset.venueView > table > tr").each { |kv|
      label = kv.css("td label").first.children.to_s.strip
      value = kv.css("td")[1].children.to_s.strip
      if label == "Event Types"
        value = kv.css("td")[1].children.css("a").children.to_s.strip
      end
      properties[label] = value
    }
  end
  feature = {
    'type' => "Feature",
    "id" => url,
    "properties" => properties,
    "geometry" => {
      "type" => "Point",
      "coordinates" => [lng, lat]
    }
  }
  if lat != 0 and lng != 0
    features << feature
  end
}

data = {
  'type' => "FeatureCollection",
  'features' => features
}
puts JSON.generate(data)