profile-list

#!/usr/bin/env ruby
# encoding: UTF-8
$LOAD_PATH.unshift File.expand_path(File.dirname(__FILE__) + '/lib')

require 'pp'
require 'domain-profiler'
require 'rubygems'
require 'gchart'
require 'erb'

def fullname(name)
  Name.new.full(name)
end

class Array
  def lookup(host)
    self.map{|name| Name.new.simplify(name,host) }.uniq
  end
end


if ARGV.length == 0
   print "Usage: ./profile-list [file containing one domain per line] 'header to put in HTML output'\n"
   exit
end

filename = ARGV[0]
file = File.new(filename)

hosts = {}
list_name = ARGV[1] ? ARGV[1] : 'REPLACE ME'

file.map {|host|
  if match_data = host.match(/^#List:(.*)/)
    list_name = match_data[1]
  end
  next if host.match(/^#/)
  host.chomp!
  profile = DomainProfiler.new(host)

  begin
    out = {}
    out[:web_host]   = profile.dns.a.map{|record| orgname(record.answer) }.lookup(host)
    out[:dns_host]   = profile.dns.ns.map{|record| Name.new.shorten(record.answer) }.lookup(host)
    out[:mail_host]  = profile.dns.mx.map{|record| Name.new.shorten(record.host) }.lookup(host)
    out[:registrar]  = profile.whois.registrar.lookup(host)
    out[:ssl_issuer] = profile.ssl.ca.lookup(host)
    out[:ssl_type]   = profile.ssl.cn
    hosts[host] = out
  rescue
  end

}

# Modify this to make stats on hosting provider, dns, mail, whois, ssl, ssn type and count HTML

count = {}
types =  [:web_host,:mail_host,:dns_host,:registrar,:ssl_issuer,:ssl_type]

types.each { |kind| count[kind] = [] }

# Turn the list of host data into a hash of type data
hosts.each do |hostname,data|
  data.each do |kind,value|
    if kind === :ssl_type and not value[0].is_a? Symbol
      if value[0].match(/^\*/)
        ssl_type = :star
      else
        ssl_type = :normal
      end

      value = ssl_type
    end

    count[kind].push(value) 
  end
end

output = {}
count.each do |kind,values|
  summary = {}
  values.flatten.each do |value|
    value = :other unless value.is_a? Symbol
      
    if summary[value].is_a? Integer
      summary[value] += 1
    else
      summary[value] = 1
    end
  end
  output[kind] = summary
end

# Collapse any item less than smallest_percent into the :other category
smallest_percent = 0.02  
output.each do |kind,values|
  total = 1
  values.each {|k,v| total += v}
  smallest_value = total * smallest_percent
  values.each {|k,v| 
    if v < smallest_value
      values[:other] = 1 unless values[:other].is_a? Integer
      values[:other] += v
      values.delete(k)
    end
  }
end

  full_name = {
    :web_host => 'Web Host',
    :mail_host => 'Email Host',
    :dns_host => 'DNS Host',
    :registrar => 'Registrar',
    :ssl_issuer => 'SSL Issuer',
    :ssl_type => 'Certificate Type',
  }


output.each do |kind, summary_data|
  keys = []
  values = []
  summary_data.sort { |a,b| a[1] <=> b[1] }.each do |k,v|
    keys.push(Name.new.full(k))
    values.push(v)
  end

  output[kind] = Gchart.pie(:size => '400x200', 
#                            :background => 'F8ECDC',
                            :bar_color => ['CFF09E','A8DBA8','79BD9A','3B8686','0B486B'],
                            :color => '', 
                            :title => full_name[kind], 
                            :labels => keys, 
                            :data => values
                            ) 
end

charts = []
types.each do |kind|
  charts.push( {:kind => kind, :url => output[kind] } )
end
puts ERB.new(File.read("view/html")).result