gistfile1.rb

require 'rubygems'
require 'nokogiri'

m = {}
f = {}
open("dist.male.first").readlines.each { |l| 
  d = l.split(/ +/) 
  m[d[0].downcase] = d[1].to_f
}
open("dist.female.first").readlines.each { |l| 
  d = l.split(/ +/) 
  f[d[0].downcase] = d[1].to_f
}
names = Dir.glob("page*").map { |f|
  Nokogiri::HTML.parse(open(f)).css("td.master").map { |td| td.children[2].text.split(/,/)[0].strip.downcase }
}.flatten.uniq
males = 0
females = 0
names.each { |n|
  n = n.split(/ /)[0]
  if m.has_key?(n) and f.has_key?(n)
    ratio = 1.0/(m[n] + f[n])
    puts "B #{n} #{ratio * m[n]} #{ratio * f[n]}"
    males += ratio * m[n]
    females += ratio * f[n]
  elsif m.has_key?(n)
    puts "M #{n}"
    males += 1
  elsif f.has_key?(n)
    puts "F #{n}"
    females += 1
  else
    puts "U #{n}"
  end
}
puts "male: #{males}"
puts "female: #{females}"