convert.rb

require 'nokogiri'
require 'time'
require 'json'

module Parser
  extend self

  def message(el)
    raise "Requires Nokogiri::XML::Element" if el.class != Nokogiri::XML::Element

    names = el.css("> text()").text.split(", ")
    data = {}
    # data[:contact] = names[0].scan(/Scott Li/) ? names[1] : names[0]
    data[:contact] = names
    data[:messages] = []

    last = {}
    el.elements.each_with_index do |e, i|
      if i.even?
        last[:user] = e.css(".message_header .user").text
        last[:meta] = e.css(".message_header .meta").text
        last[:timestamp] = Time.parse(last[:meta]).to_i
      end

      if i.odd?
        e.text
        msg = {
          user: last[:user],
          meta: last[:meta],
          timestamp: last[:timestamp],
          content: e.text
        }

        data[:messages] << msg
      end
    end


    data[:messages].sort! {|before, after| before[:timestamp] <=> after[:timestamp]}
    return data
  end
end

file = File.open("html/messages.htm").read
noko = Nokogiri::HTML.parse(file)
threads = noko.css(".thread")
all_messages = []

job_n = 0
jobs = []

threads.each_with_index do |thread, index|
  jobs[index] = Thread.new do
    all_messages << Parser.message(thread)
    job_n += 1
    puts "#{job_n}\n"

  end
end

jobs.each(&:join)

puts "converting to json"
m_json = all_messages.to_json
puts "saving to file"
file = File.open("messages.json", "w+")
file.write(m_json)
file.close