btrc-hub/backend/lib/tasks/sync_nico.rake

namespace :nico do
  desc 'ニコニコ DB 同期'
  task sync: :environment do
    require 'open3'
    require 'open-uri'
    require 'nokogiri'

    fetch_thumbnail = -> url {
      html = URI.open(url, read_timeout: 60, 'User-Agent' => 'Mozilla/5.0').read
      doc = Nokogiri::HTML(html)

      doc.at('meta[name="thumbnail"]')&.[]('content').presence
    }

    mysql_user = ENV['MYSQL_USER']
    mysql_pass = ENV['MYSQL_PASS']
    nizika_nico_path = ENV['NIZIKA_NICO_PATH']
    stdout, stderr, status = Open3.capture3(
        { 'MYSQL_USER' => mysql_user, 'MYSQL_PASS' => mysql_pass },
        'python3', "#{ nizika_nico_path }/get_videos.py")

    if status.success?
      data = JSON.parse(stdout)
      data.each do |datum|
        post = Post.where('url LIKE ?', '%nicovideo.jp%').find { |post|
          post.url =~ %r{#{ Regexp.escape(datum['code']) }(?!\d)}
        }
        unless post
          title = datum['title']
          url = "https://www.nicovideo.jp/watch/#{ datum['code'] }"
          thumbnail_base = fetch_thumbnail.(url) || '' rescue ''
          post = Post.new(title:, url:, thumbnail_base:, uploaded_user: nil)
          if thumbnail_base.present?
            post.thumbnail.attach(
                io:           URI.open(thumbnail_base),
                filename:     File.basename(URI.parse(thumbnail_base).path),
                content_type: 'image/jpeg')
          end
          post.save!
          post.resized_thumbnail!
        end

        current_tags = post.tags.where(category: 'nico').pluck(:name).sort
        new_tags = datum['tags'].map { |tag| "nico:#{ tag }" }.sort
        if current_tags != new_tags
          post.tags.destroy(post.tags.where(name: current_tags))
          new_tags.each do |name|
            post.tags << Tag.find_or_initialize_by(name:) do |tag|
              tag.category = 'nico'
            end
          end
          post.tags << Tag.tagme if post.tags.size < 20 && post.tags.none?(Tag.tagme)
          post.tags << Tag.bot if post.tags.none?(Tag.bot)
        end
      end
    end
  end
end