wedata_from_google_cache.rb

#!/usr/bin/env ruby -Ku
require 'rubygems'
require 'json'
require 'mechanize'

agent = WWW::Mechanize.new
#sample iview
url = "http://209.85.173.132/search?hl=ja&q=cache%3Awedata.net%2Fdatabases%2Fiview%2Fitems"
page = agent.get(url);
resource_url = 'http://wedata.net' + page.root.search('#content div ul.links li a')[0]['href']
ary = page.root.search('li.hentry').to_ary.map {|entry|
  info = {
    "name" => entry.search('a.entry-title')[0].text,
    "update_at" => entry.search('span.updated')[0].text,
    "created_by" => entry.search('div.vcard a')[0].text,
    "resource_url" => 'http://wedata.net' + entry.search('a.entry-title')[0]["href"],
    "database_resource_url" => resource_url,
    "data" => "",
    "created_at" => entry.search('span.published')[0].text
  }
  data = Hash.new
  entry.search('tr').to_ary.each { |tr|
    data[tr.children[0].text] = tr.children[1].text
  }
  info["data"] = data
  info
}
File.open('iview.json', 'w'){|file|
  file.write(JSON.pretty_generate(ary, nil))
}