cgmailstats.rb

require 'nokogumbo'
require 'date'
require 'json'

start = Date.today - 105

results = {}
index = 'http://www.w3.org/community/groups/'
groups = Nokogiri::HTML5.get(index)
groups = groups.search('a.h3').map {|a| a['href'].sub(/^#/, '')}
groups.each do |group|
  list = "http://lists.w3.org/Archives/Public/public-#{group}"
  doc = Nokogiri::HTML5.get(list)
  count = 0
  doc.search('tbody tr').each do |tr|
    next if tr.at('td').text == 'n/a'
    date=Date.parse(tr.at('td').text)
    count += tr.search('td').last.text.to_i if date > start
  end
  results[group] = count
end

puts JSON.pretty_generate(results)