#!/usr/bin/env ruby
require "net/http"
require "uri"
require "cgi"
require "yaml"

start = 100
url = URI.parse("http://thepiratebay.org/")
cats = Hash.new

#starting from 100 look for empty pages.
while start<700
    Net::HTTP.start(url.host, url.port) {|http|
        res = http.get("/browse/#{start}")
        if res.body.size > 0
          cats[res.body.scan(/<title>(.*?)<\/title>/).to_s] = start
          start += 1
        else
          start = ((start/100) + 1) * 100
        end
     }
  pp start
end
# starting from 699 work backwards for empty pages
start = 699
while start>0
    Net::HTTP.start(url.host, url.port) {|http|
        res = http.get("/browse/#{start}")
        if res.body.size > 0
          cats[res.body.scan(/<title>(.*?)<\/title>/).to_s] = start
          start -= 1
        else
          start = ((start/100) - 1) * 100
          start -= 1
        end
     }
  pp start
end
#convert poorly named categories hash to something
# slightly useful
master_hash = Hash.new
master_hash["_RAW"] = cats
master_hash["_RAW"].keys.each{|raw_key|
  html = CGI::unescapeHTML(raw_key).split(">")
  key = html.first.strip
  master_hash[key] ||= Hash.new
  name = html.last.strip
  master_hash[key][name.sub(" - TPB","")] = master_hash["_RAW"][raw_key]
}

File.open(File.join(File.dirname(__FILE__),"config","pirate_categories.yml"),"w+"){|f|
  f.puts( YAML::dump(master_hash))
}