YouTube の自動同期 (#314) (#340)

#314 #314 #314 #314 #314 Co-authored-by: miteruzo <miteruzo@naver.com> Reviewed-on: #340
2026-05-02 17:56:14 +09:00
parent fcd3b87b2a
commit 5002859fc8
10 changed files with 846 additions and 0 deletions
@@ -0,0 +1,168 @@
+require 'open-uri'
+require 'set'
+require 'time'
+
+
+module Youtube
+  class Sync
+    def initialize client: ApiClient.new
+      @client = client
+    end
+
+    def sync!
+      video_ids = discover_video_ids
+      return if video_ids.empty?
+
+      video_ids.each_slice(50) do |ids|
+        @client.videos(ids).fetch('items', []).each do |item|
+          sync_video!(VideoItem.new(item))
+        end
+      end
+    end
+
+    private
+
+    def discover_video_ids
+      ids = Set.new
+
+      query_terms.each do |q|
+        response = @client.search_videos(q:, published_after: sync_since)
+
+        response.fetch('items', []).each do |item|
+          video_id = item.dig('id', 'videoId')
+          ids << video_id if video_id.present?
+        end
+      end
+
+      playlist_ids.each do |playlist_id|
+        each_playlist_item(playlist_id) do |item|
+          video_id = item.dig('contentDetails', 'videoId')
+          video_id ||= item.dig('snippet', 'resourceId', 'videoId')
+
+          ids << video_id if video_id.present?
+        end
+      end
+
+      ids.to_a
+    end
+
+    def sync_video! video
+      post = Post.where('url REGEXP ?', youtube_url_regexp(video.id)).first
+
+      original_created_from = video.published_at.change(sec: 0)
+      original_created_before = original_created_from + 1.minute
+
+      post_created = false
+      post_changed = false
+
+      if post
+        post.assign_attributes(title: video.title,
+                               original_created_from:,
+                               original_created_before:,
+                               thumbnail_base: video.thumbnail_url)
+
+        post_changed = post.changed?
+        post.save! if post_changed
+
+        attach_thumbnail_if_needed!(post, video.thumbnail_url)
+      else
+        post_created = true
+        post = Post.create!(
+            title: video.title,
+            url: video.url,
+            thumbnail_base: video.thumbnail_url,
+            uploaded_user_id: nil,
+            original_created_from:,
+            original_created_before:)
+
+        attach_thumbnail_if_needed!(post, video.thumbnail_url)
+
+        sync_post_tags!(post, [Tag.tagme.id, Tag.bot.id, Tag.youtube.id, Tag.video.id])
+      end
+
+      kept_tag_ids = post.tags.pluck(:id).to_set
+      desired_tag_ids = kept_tag_ids.to_a
+
+      deerjikist = Deerjikist.find_by(platform: :youtube, code: video.channel_id)
+      if deerjikist
+        desired_tag_ids.delete(Tag.no_deerjikist.id)
+        desired_tag_ids << deerjikist.tag_id
+      elsif post.tags.where(category: :deerjikist).none?
+        desired_tag_ids << Tag.no_deerjikist.id
+      end
+
+      desired_tag_ids.uniq!
+
+      sync_post_tags!(post, desired_tag_ids, current_tag_ids: kept_tag_ids)
+
+      if post_created
+        PostVersionRecorder.record!(post:, event_type: :create, created_by_user: nil)
+      elsif post_changed || kept_tag_ids != desired_tag_ids.to_set
+        PostVersionRecorder.ensure_snapshot!(post, created_by_user: nil)
+        PostVersionRecorder.record!(post:, event_type: :update, created_by_user: nil)
+      end
+    end
+
+    def sync_post_tags! post, desired_tag_ids, current_tag_ids: nil
+      current_tag_ids ||= PostTag.kept.where(post_id: post.id).pluck(:tag_id).to_set
+      desired_tag_ids = desired_tag_ids.compact.to_set
+
+      to_add = desired_tag_ids - current_tag_ids
+      to_remove = current_tag_ids - desired_tag_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
+
+    def attach_thumbnail_if_needed! post, thumbnail_url
+      return if post.thumbnail.attached?
+      return if thumbnail_url.blank?
+
+      post.thumbnail.attach(
+          io:           URI.open(thumbnail_url),
+          filename:     File.basename(URI.parse(thumbnail_url).path),
+          content_type: 'image/jpeg')
+
+      post.resized_thumbnail!
+    end
+
+    def youtube_url_regexp id
+      escaped = Regexp.escape(id)
+      "(youtube\\.com/watch\\?v=#{ escaped }|youtu\\.be/#{ escaped })([^A-Za-z0-9_-]|$)"
+    end
+
+    def query_terms = ['ぼざろクリーチャーシリーズ', '伊地知ニジカ', '伊地知虹鹿']
+
+    def playlist_ids
+      ['PLrOch4zHkI5vu29b-f9umUQQ4tQkuWLPX',
+       'PLrOch4zHkI5vOK0RaytQq6PbucxQkkL0K',
+       'PLrOch4zHkI5tdwm9vSegiDQJOM-hgpcOC']
+    end
+
+    def sync_since = 14.days.ago
+
+    def each_playlist_item playlist_id
+      page_token = nil
+
+      loop do
+        response = @client.playlist_items(playlist_id:, page_token:)
+
+        response.fetch('items', []).each do |item|
+          yield item
+        end
+
+        page_token = response['nextPageToken']
+        break if page_token.blank?
+      end
+    end
+  end
+end