vlv-search/lib/parser.rb

require 'nokogiri'
require 'time'

class Parser
  def threads(page)
    threads = Array.new

    page.css('.even, .odd').each do |row|
      thread = Hash.new
      thread_link = row.at_css('.subject > a')
      next if thread_link.nil?
      thread[:remote_id] = thread_link['href'].split('/')[3]
      thread[:title] = thread_link.text.strip
      creator = row.at_css('.memberlink').text.strip
      creator = creator[0..-2] if creator.match(/\+$/)
      thread[:creator] = creator
      thread[:is_sticky] = !!thread[:title].match(/^Sticky:/)
      threads << thread
    end

    threads
  end

  def posts(thread, page)
    posts = Array.new

    page.css('.post').each do |_post|
      post = Hash.new
      post[:remote_id] = _post.at_css('ul.view')[:id].split('_')[1].to_i
      post[:creator] = _post.at_css('.memberlink').text.strip
      date, time = _post.at_css('.postinfo').text.split('posted this')[1].split('@')
      post[:created_at] = Time.parse("#{date} #{time}")
      post[:body] = _post.at_css('.postbody').children.map(&:to_html).join.strip
      post[:thread_id] = thread[:id]
      posts << post
    end

    posts
  end

  def thread_created_at(first_post)
    post_info = first_post.text.split('posted this')
    date, time = post_info[1].split('@')
    Time.parse("#{date} #{time}")
  end
end