gistfile1.rb

# Some more pondering for my ScrapeKit normalizer builder. I keep throwing
# different use-cases at it which keeps causing me to change my design. I think
# I am starting to get to a happy place.

normalizer do |doc|
  doc.collect :inventories, 'table[width="66%"] tr', :exclude => [0, -1, 'UNKNOWN QUANTITY'], :missing => :halt do |tr|
    tr.all 'td[width="25%"]', :include => [1..3] do |(name, _, quantity)|
      name.select(:name, :format => [:titlecase])
      quantity.select(:quantity, :format => [:to_i])
    end
  end
  doc.locate '.user-details', :missing => :halt, :include_if => :user_is_active do |div|
    div.select '.name' do |content, node, doc|
      first, last = *content.split
      { :first_name => first, :last_name => last }
    end
    div.select :age, '.age', :format => [/[0-9]+/, :to_i]
    div.select :address, '.address', :format => [:strip]
    div.select :country, '.country-code', :format => [:country_code_to_name]
  end
  doc.collect :attendees, 'ul.attendees li' do |li|
    li.select do |content|
      name, email = *content.split('|').map { |n| n.strip }
      { :name => name, :email => email }
    end
  end
  doc.collect :job_histories, '.user-job-history ul li' do |li|
    li.select :company, '.company-name'
    li.select :title, '.job-title'
    li.select '.job-timeframe', :helper => :job_dates
  end
end