From 39101ca422b8144c46e682118c741ab3a7af090d Mon Sep 17 00:00:00 2001
From: miteruzo <miteruzo@naver.com>
Date: Sat, 2 May 2026 02:48:40 +0900
Subject: [PATCH] #314

---
 backend/app/models/tag.rb                     |   1 +
 .../discovery/youtube_video_discoverer.rb     |  46 ++++++
 backend/app/services/post_importer.rb         |  51 ++++++
 .../services/video_sources/youtube/client.rb  |  87 ++++++++++
 backend/app/services/youtube/api_client.rb    |  68 ++++++++
 backend/app/services/youtube/search_client.rb |  34 ++++
 backend/app/services/youtube/sync.rb          | 152 ++++++++++++++++++
 backend/app/services/youtube/video_item.rb    |  29 ++++
 backend/lib/tasks/sync_posts.rake             |   6 +
 9 files changed, 474 insertions(+)
 create mode 100644 backend/app/services/discovery/youtube_video_discoverer.rb
 create mode 100644 backend/app/services/post_importer.rb
 create mode 100644 backend/app/services/video_sources/youtube/client.rb
 create mode 100644 backend/app/services/youtube/api_client.rb
 create mode 100644 backend/app/services/youtube/search_client.rb
 create mode 100644 backend/app/services/youtube/sync.rb
 create mode 100644 backend/app/services/youtube/video_item.rb
 create mode 100644 backend/lib/tasks/sync_posts.rake

diff --git a/backend/app/models/tag.rb b/backend/app/models/tag.rb
index 54c3d68..acdc5c7 100644
--- a/backend/app/models/tag.rb
+++ b/backend/app/models/tag.rb
@@ -84,6 +84,7 @@ class Tag < ApplicationRecord
   def self.no_deerjikist = find_or_create_by_tag_name!('ニジラー情報不詳', category: :meta)
   def self.video = find_or_create_by_tag_name!('動画', category: :meta)
   def self.niconico = find_or_create_by_tag_name!('ニコニコ', category: :meta)
+  def self.youtube = find_or_create_by_tag_name!('YouTube', category: :meta)
 
   def self.normalise_tags tag_names, with_tagme: true,
                                      with_no_deerjikist: true,
diff --git a/backend/app/services/discovery/youtube_video_discoverer.rb b/backend/app/services/discovery/youtube_video_discoverer.rb
new file mode 100644
index 0000000..70eac89
--- /dev/null
+++ b/backend/app/services/discovery/youtube_video_discoverer.rb
@@ -0,0 +1,46 @@
+class Discovery::YoutubeVideoDiscoverer
+  def initialize(client: Youtube::SearchClient.new)
+    @client = client
+  end
+
+  def call discovery_query:, published_after:, published_before:
+    body = @client.search_videos(
+        query: discovery_query.query,
+        published_after: published_after,
+        published_before: published_before)
+
+    body.fetch('items', []).each do |item|
+      next unless item.dig('id', 'kind') == 'youtube#video'
+
+      upsert_candidate!(discovery_query, item)
+    end
+  end
+
+  private
+
+  def upsert_candidate! discovery_query, item
+    snippet = item.fetch('snippet')
+    code = item.fetch('id').fetch('videoId')
+
+    candidate = VideoCandidate.find_or_initialize_by(provider: 'youtube', code:)
+
+    candidate.title = snippet['title'].to_s
+    candidate.description = snippet['description'].to_s
+    candidate.channel_code = snippet['channelId']
+    candidate.channel_title = snippet['channelTitle']
+    candidate.published_at = Time.zone.parse(snippet['publishedAt'])
+    candidate.thumbnail_url = snippet.dig('thumbnails', 'high', 'url')
+    candidate.raw_data = item
+    candidate.last_discovered_at = Time.current
+    candidate.save!
+
+    VideoCandidateHit.find_or_create_by!(
+      video_candidate: candidate,
+      discovery_query: discovery_query
+    ) do |hit|
+      hit.matched_field = 'youtube_search'
+      hit.score = 0
+      hit.searched_at = Time.current
+    end
+  end
+end
diff --git a/backend/app/services/post_importer.rb b/backend/app/services/post_importer.rb
new file mode 100644
index 0000000..6d86287
--- /dev/null
+++ b/backend/app/services/post_importer.rb
@@ -0,0 +1,51 @@
+class PostImporter
+  def initialize provider:
+    @provider = provider
+  end
+
+  def import_posts source_videos
+    alive_codes = []
+
+    ApplicationRecord.transaction do
+      source_videos.each do |source_video|
+        alive_codes << source_video.fetch(:code)
+
+        deerjikist_tag = deerjikist_tag_of(source_video)
+        video = upsert_post(source_video, deerjikist_tag)
+      end
+    end
+  end
+
+  private
+
+  attr_reader :provider
+
+  def deerjikist_tag_of source_video
+    user_code = source_video[:user_code]
+    return nil if user_code.blank?
+
+    deerjikist&.tag
+  end
+
+  def upsert_post source_video, deerjikist_tag
+    url =
+        case provider
+        when 'youtube'
+          "https://youtube.com/watch/#{ source_video.fetch(:code) }"
+        end
+    raise 'ちんぽ!' if url.blank?
+
+    Post.find_or_initialize_by(url:).tap do |post|
+      post.title = source_viedo.fetch(:title)
+      post.uploaded_user_id = nil
+      post.original_created_from = source_video.fetch(:uploaded_at)
+      post.original_created_before = post.original_created_from + 1.min
+      post.tags << (deerjikist_tag || no_deerjikist)
+      post.tags << Tag.youtube
+      post.tags << Tag.video
+      post.tags << Tag.bot
+      post.tags << Tag.tagme
+      post.save!
+    end
+  end
+end
diff --git a/backend/app/services/video_sources/youtube/client.rb b/backend/app/services/video_sources/youtube/client.rb
new file mode 100644
index 0000000..ac797f2
--- /dev/null
+++ b/backend/app/services/video_sources/youtube/client.rb
@@ -0,0 +1,87 @@
+module VideoSources
+  module Youtube
+    class Client
+      API_BASE = 'https://www.googleapis.com/youtube/v3'
+
+      def initialize api_key: ENV.fetch('YOUTUBE_API_KEY')
+        @api_key = api_key
+      end
+
+      def videos video_ids
+        return [] if video_ids.empty?
+
+        response = connection.get('videos', part: 'snippet,statistics',
+                                            id: video_ids.join(','),
+                                            key: @api_key)
+
+        JSON.parse(response.body).fetch('items', []).map do |item|
+          build_video(item)
+        end
+      end
+
+      def comments video_id
+        comments = []
+        page_token = nil
+
+        loop do
+          response = connection.get('commentThreads', {
+              part: 'snippet',
+              videoId: video_id,
+              maxResults: 100,
+              textFormat: 'plainText',
+              pageToken: page_token,
+              key: @api_key }.compact)
+
+          body = JSON.parse(response.body)
+
+          comments.concat(body.fetch('items', []).map { |item| build_comment(item) })
+
+          page_token = body['nextPageToken']
+          break if page_token.blank?
+        end
+
+        comments
+      rescue Faraday::ForbiddenError
+        []
+      end
+
+      private
+
+      def connection
+        @connection ||= Faraday.new(url: API_BASE) do |faraday|
+          faraday.response :raise_error
+        end
+      end
+
+      def build_video item
+        snippet = item.fetch('snippet')
+        statistics = item.fetch('statistics', { })
+
+        { provider: 'youtube',
+          code: item.fetch('id'),
+          user_code: snippet['channelId'],
+          title: snippet['title'].to_s,
+          description: snippet['description'].to_s,
+          tag_names: snippet.fetch('tags', []),
+          views_count: statistics.fetch('viewCount', 0).to_i,
+          uploaded_at: Time.zone.parse(snippet.fetch('publishedAt')) }
+      end
+
+      def build_comment item
+        snippet =
+            item
+            .fetch('snippet')
+            .fetch('topLevelComment')
+            .fetch('snippet')
+
+        { provider_comment_id: item.fetch('id'),
+          user_code: snippet['authorChannelId']&.fetch('value', nil),
+          content: snippet['textDisplay'].to_s,
+          posted_at: Time.zone.parse(snippet.fetch('publishedAt')),
+          reaction_count: snippet.fetch('likeCount', 0).to_i,
+          comment_no: nil,
+          vpos_ms: nil }
+      end
+    end
+  end
+end
diff --git a/backend/app/services/youtube/api_client.rb b/backend/app/services/youtube/api_client.rb
new file mode 100644
index 0000000..8ab0b04
--- /dev/null
+++ b/backend/app/services/youtube/api_client.rb
@@ -0,0 +1,68 @@
+require 'json'
+require 'net/http'
+require 'uri'
+
+
+module Youtube
+  class ApiClient
+    ENDPOINT = 'https://www.googleapis.com/youtube/v3'
+
+    def initialize api_key: ENV.fetch('YOUTUBE_API_KEY')
+      @api_key = api_key
+    end
+
+    def search_videos q:, published_after:, page_token: nil
+      get_json('/search', {
+          part:           'snippet',
+          type:           'video',
+          q:              q,
+          order:          'date',
+          maxResults:     50,
+          publishedAfter: published_after.iso8601,
+          pageToken:      page_token }.compact)
+    end
+
+    def videos ids
+      return { 'items' => [] } if ids.empty?
+
+      get_json('/videos', part: 'snippet,status,contentDetails', id: ids.join (','))
+    end
+
+    def playlist_items playlist_id:, page_token: nil
+      get_json('/playlistItems', {
+          part:       'snippet,contentDetails,status',
+          playlistId: playlist_id,
+          maxResults: 50,
+          pageToken:  page_token }.compact)
+    end
+
+    def channel id: nil, handle: nil
+      params = { part: 'snippet,contentDetails' }
+      params[:id] = id if id
+      params[:forHandle] = handle if handle
+
+      get_json('/channels', params)
+    end
+
+    private
+
+    def get_json path, params
+      uri = URI(ENDPOINT + path)
+      uri.query = URI.encode_www_form(params.merge(key: @api_key))
+
+      response = Net::HTTP.start(uri.host,
+                                 uri.port,
+                                 use_ssl: true,
+                                 open_timeout: 10,
+                                 read_timeout: 30) do |http|
+        http.get(uri)
+      end
+
+      unless response.is_a?(Net::HTTPSuccess)
+        raise "YouTube API error: #{ response.code } #{ response.body }"
+      end
+
+      JSON.parse(response.body)
+    end
+  end
+end
diff --git a/backend/app/services/youtube/search_client.rb b/backend/app/services/youtube/search_client.rb
new file mode 100644
index 0000000..66bed09
--- /dev/null
+++ b/backend/app/services/youtube/search_client.rb
@@ -0,0 +1,34 @@
+module Youtube
+  class SearchClient
+    API_BASE = 'https://www.googleapis.com/youtube/v3'
+
+    def initialize api_key: ENV.fetch('YOUTUBE_API_KEY')
+      @api_key = api_key
+    end
+
+    def search_videos query:, published_after: nil, published_before: nil, page_token: nil
+      response = connection.get('search', {
+          part: 'snippet',
+          q: query,
+          type: 'video',
+          order: 'date',
+          maxResults: 50,
+          regionCode: 'JP',
+          relevanceLanguage: 'ja',
+          publishedAfter: published_after&.iso8601,
+          publishedBefore: published_before&.iso8601,
+          pageToken: page_token,
+          key: @api_key }.compact)
+
+      JSON.parse(response.body)
+    end
+
+    private
+
+    def connection
+      @connection ||= Faraday.new(url: API_BASE) do |faraday|
+        faraday.response :raise_error
+      end
+    end
+  end
+end
diff --git a/backend/app/services/youtube/sync.rb b/backend/app/services/youtube/sync.rb
new file mode 100644
index 0000000..146441b
--- /dev/null
+++ b/backend/app/services/youtube/sync.rb
@@ -0,0 +1,152 @@
+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|
+        response = @client.playlist_items(playlist_id:)
+        response.fetch('items', []).each do |item|
+          video_id = item.dig('contentDetails', '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:           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 << 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
+  end
+end
diff --git a/backend/app/services/youtube/video_item.rb b/backend/app/services/youtube/video_item.rb
new file mode 100644
index 0000000..b8bad47
--- /dev/null
+++ b/backend/app/services/youtube/video_item.rb
@@ -0,0 +1,29 @@
+module Youtube
+  class VideoItem
+    attr_reader :id, :title, :channel_id, :published_at, :thumbnail_url, :raw_tags
+
+    def initialize item
+      snippet = item.fetch('snippet')
+
+      @id = item.fetch('id')
+      @title = snippet['title']
+      @channel_id = snippet['channelId']
+      @published_at = Time.iso8601(snippet['publishedAt'])
+      @thumbnail_url = pick_thumbnail(snippet['thumbnails'] || { })
+      @raw_tags = snippet['tags'] || []
+    end
+
+    def url = "https://www.youtube.com/watch?v=#{ @id }"
+
+    private
+
+    def pick_thumbnail thumbnails
+      ['maxres', 'standard', 'high', 'medium', 'default'].each do |key|
+        url = thumbnails.dig(key, 'url')
+        return url if url.present?
+      end
+
+      nil
+    end
+  end
+end
diff --git a/backend/lib/tasks/sync_posts.rake b/backend/lib/tasks/sync_posts.rake
new file mode 100644
index 0000000..267d474
--- /dev/null
+++ b/backend/lib/tasks/sync_posts.rake
@@ -0,0 +1,6 @@
+namespace :post do
+  desc '投稿同期（ニコニコ以外）'
+  task sync: :environment do
+    Youtube::Sync.new.sync!
+  end
+end