apparatus twitter usernames

scraper.rb

require 'nokogiri'
require 'open-uri'
require 'json'
require 'csv'

URL = 'http://apparatus.si/oddaja/pogovor/page/%d/'

def last_page
  doc = Nokogiri::HTML(open(URL%1))
  pagination = doc.css('.archive-pagination a')
  link = pagination[-2].attr('href').gsub(/\D/, '').to_i
end

def shows
  1.upto(last_page).map { |page|
    Nokogiri::HTML(open(URL%page)).css('h1.entry-title a')
  }.flatten.map { |show|
    regex = show.text.match(/^(.*): (.*)$/)
    {
      no: regex[1].to_i,
      person: regex[2],
      link: show.attr('href')
    }
  }.reverse
end

def get_user(link)
  url = URI(link.attr('href'))
  user = url.path.sub('/', '')
end

def get_twitter_links(html)
  html.css('a').select{ |link|
    link.attr('href') =~ /twitter.com/
  }.reject{ |link|
    %w(anzet apparatus_si).include?(get_user(link).downcase) || link.attr('href') =~ /share/
  }
end

twitter = shows.map { |show|
  html = Nokogiri::HTML(open(show[:link]))
  links = get_twitter_links(html)
  if links.first
    show.merge(twitter: get_user(links.first))
  else
    show
  end
}

CSV.open('twitter.csv', 'w') do |csv|
  csv << twitter.last.keys
  twitter.each do |hash|
    csv << hash.values
  end
end

Get all twitter usernames of people who were on Apparatus podcast/Storming Mortal for fun and profit.
Take the first twitter link from show page that's not anzet|apparatus|share link. As with every 80/20 concept it works for majority, but fails miserably for minority which also includes me - I get parishilton 😆