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'
    require 'set'

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

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

    def sync_post_tags! post, desired_tag_ids, current_ids: nil
      current_ids ||= PostTag.kept.where(post_id: post.id).pluck(:tag_id).to_set
      desired_ids = desired_tag_ids.compact.to_set

      to_add = desired_ids - current_ids
      to_remove = current_ids - desired_ids

      Tag.where(id: to_add.to_a).find_each do |tag|
        begin
          PostTag.create!(post:, tag:)
        rescue ActiveRecord::RecordNotUnique
          ;
        end
      end

      PostTag.where(post_id: post.id, tag_id: to_remove.to_a).kept.find_each do |pt|
        pt.discard_by!(nil)
      end
    end

    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")

    abort unless status.success?

    data = JSON.parse(stdout)
    data.each do |datum|
      code = datum['code']
      post = Post.where('url REGEXP ?', "nicovideo\\.jp/watch/#{ Regexp.escape(code) }([^0-9]|$)")
                 .first
      title = datum['title']
      original_created_at = datum['uploaded_at'] &&
          Time.strptime(datum['uploaded_at'], '%Y-%m-%d %H:%M:%S')
      original_created_from = original_created_at&.change(sec: 0)
      original_created_before = original_created_from&.+(1.minute)
      if post
        attrs = { title:, original_created_from:, original_created_before: }
        post.update!(attrs) if attrs.any? { |k, v| post.public_send(k) != v }
      else
        url = "https://www.nicovideo.jp/watch/#{ code }"
        thumbnail_base = fetch_thumbnail.(url) rescue nil
        post = Post.new(title:, url:, thumbnail_base:, uploaded_user: nil,
                        original_created_from:, original_created_before:)
        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!
        sync_post_tags!(post, [Tag.tagme.id])
      end

      kept_ids = PostTag.kept.where(post_id: post.id).pluck(:tag_id).to_set
      kept_non_nico_ids = post.tags.where.not(category: 'nico').pluck(:id).to_set

      desired_nico_ids = []
      desired_non_nico_ids = []
      datum['tags'].each do |raw|
        name = "nico:#{ raw }"
        tag = Tag.find_or_create_by_tag_name!(name, category: 'nico')
        desired_nico_ids << tag.id
        unless tag.id.in?(kept_ids)
          linked_ids = tag.linked_tags.pluck(:id)
          desired_non_nico_ids.concat(linked_ids)
          desired_nico_ids.concat(linked_ids)
        end
      end
      desired_nico_ids.uniq!

      desired_all_ids = kept_non_nico_ids.to_a + desired_nico_ids
      desired_non_nico_ids.concat(kept_non_nico_ids.to_a)
      desired_non_nico_ids.uniq!
      if kept_non_nico_ids.to_set != desired_non_nico_ids.to_set
        desired_all_ids << Tag.bot.id
      end
      desired_all_ids.uniq!

      sync_post_tags!(post, desired_all_ids, current_ids: kept_ids)
    end
  end
end