remove_budstikka_missing_items.rb

# Delete missing (deleted/draft) news_items from Budstikka internal articles
# Using 404 amedia response code to identify such articles

newspaper = Newspaper.find(4);

min_news_age = 5.days.ago

news_items_scope = NewsItem.internal.where(newspaper_id: newspaper.id).where("created_at > ?", min_news_age);

duplicated_titles = news_items_scope.group('title').count.select{|k,v| v > 1};
news_items = news_items_scope.where(title: duplicated_titles.keys).to_a;

destroyed_items_count = 0

news_items.each do |item|
  url = "#{Content::Amedia::Client::CONTENT_URL}#{item.external_id}"
  response = Excon.head(url)

  if response.status == 404
    puts "Destroyed NewsItem #{item.id} (#{item.url}"

    item.destroy
    destroyed_items_count = destroyed_items_count + 1
  else
    #puts "Item OK #{item.id} #{item.url}"
  end
end;

puts "Items destroyed: #{destroyed_items_count}"
puts "Done"