外部タグを分離 (#419) (#420)

**本番 DB のバックアップをかならずすること**

Reviewed-on: #420
Co-authored-by: miteruzo <miteruzo@naver.com>

backend/spec/tasks/nico_sync_spec.rb

@@ -15,6 +15,20 @@ RSpec.describe 'nico:sync' do
     NicoTagRelation.create!(nico_tag_id: nico_tag.id, tag_id: tag.id)
   end
 
+  def create_nico_sanitisation_rules!
+    TagNameSanitisationRule.create!(priority: 20,
+                                    source_pattern: '\\?',
+                                    replacement: '_')
+
+    TagNameSanitisationRule.create!(priority: 40,
+                                    source_pattern: '_$',
+                                    replacement: '')
+
+    TagNameSanitisationRule.create!(priority: 45,
+                                    source_pattern: '^([^:]+\\:)?_',
+                                    replacement: '\\1')
+  end
+
   it '既存 post を見つけて、nico tag と linked tag を追加し、差分が出たら bot を付ける' do
     # 既存 post（正規表現で拾われるURL）
     post = Post.create!(
@@ -29,7 +43,7 @@ RSpec.describe 'nico:sync' do
 
     # 追加される linked tag を準備（nico tag に紐付く一般タグ）
     linked = create_tag!('spec_linked', category: 'general')
-    nico = create_tag!('nico:AAA', category: 'nico')
+    nico = create_external_tag!('AAA')
     link_nico_to_tag!(nico, linked)
 
     # bot / tagme は task 内で使うので作っておく（Tag.bot/tagme がある前提）
@@ -53,7 +67,7 @@ RSpec.describe 'nico:sync' do
     active_tag_names = post.tags.joins(:tag_name).pluck('tag_names.name')
 
     expect(active_tag_names).to include('spec_kept')
-    expect(active_tag_names).to include('nico:AAA')
+    expect(post.external_tags).to contain_exactly(nico)
     expect(active_tag_names).to include('spec_linked')
 
     expect(post.original_created_from).to eq(Time.iso8601('2026-01-01T03:34:00Z'))
@@ -115,12 +129,12 @@ RSpec.describe 'nico:sync' do
     )
 
     # 旧nicoタグ（今回の同期結果に含まれない）
-    old_nico = create_tag!('nico:OLD', category: 'nico')
-    PostTag.create!(post:, tag: old_nico)
+    old_nico = create_external_tag!('OLD')
+    PostExternalTag.create!(post:, external_tag: old_nico)
     create_post_version_for!(post)
 
     # 今回は NEW のみ欲しい
-    new_nico = create_tag!('nico:NEW', category: 'nico')
+    new_nico = create_external_tag!('NEW')
 
     # bot/tagme 念のため
     Tag.bot
@@ -131,23 +145,21 @@ RSpec.describe 'nico:sync' do
 
     run_rake_task('nico:sync')
 
-    expect(PostTag.exists?(post:, tag: old_nico)).to be(false)
+    expect(PostExternalTag.exists?(post:, external_tag: old_nico)).to be(false)
     expect(old_nico.reload.post_count).to eq(0)
     expect(new_nico.reload.post_count).to eq(1)
 
     versions = post.post_versions.order(:version_no)
-    expect(versions.first.tags_json.map { |item| item.fetch('id') })
+    expect(versions.first.tags_json.filter_map { |item| item['external_tag_id'] })
         .to include(old_nico.id)
-    expect(versions.last.tags_json.map { |item| item.fetch('id') })
+    expect(versions.last.tags_json.filter_map { |item| item['external_tag_id'] })
         .to include(new_nico.id)
-    expect(versions.last.tags_json.map { |item| item.fetch('id') })
+    expect(versions.last.tags_json.filter_map { |item| item['external_tag_id'] })
         .not_to include(old_nico.id)
 
     # NEW は active にいる
     post.reload
-    active_names = post.tags.joins(:tag_name).pluck('tag_names.name')
-    expect(active_names).to include('nico:NEW')
-    expect(active_names).not_to include('nico:OLD')
+    expect(post.external_tags).to contain_exactly(new_nico)
   end
 
   def snapshot_tags(post)
@@ -213,7 +225,7 @@ RSpec.describe 'nico:sync' do
     create_post_version_for!(post)
 
     linked = create_tag!('spec_linked', category: 'general')
-    nico = create_tag!('nico:AAA', category: 'nico')
+    nico = create_external_tag!('AAA')
     link_nico_to_tag!(nico, linked)
 
     Tag.bot
@@ -241,7 +253,7 @@ RSpec.describe 'nico:sync' do
   end
 
   it '既存 post に差分が無いときは新しい version を作らない' do
-    nico = create_tag!('nico:AAA', category: 'nico')
+    nico = create_external_tag!('AAA')
     no_deerjikist = create_tag!('ニジラー情報不詳', category: 'meta')
 
     post = Post.create!(
@@ -252,7 +264,7 @@ RSpec.describe 'nico:sync' do
       original_created_before: Time.iso8601('2026-01-01T03:35:00Z')
     )
 
-    PostTag.create!(post: post, tag: nico)
+    PostExternalTag.create!(post:, external_tag: nico)
     PostTag.create!(post: post, tag: no_deerjikist)
     create_post_version_for!(post)
 
@@ -295,7 +307,7 @@ RSpec.describe 'nico:sync' do
       run_rake_task('nico:sync')
     }.to change(NicoTagVersion, :count).by(1)
 
-    nico_tag = Tag.joins(:tag_name).find_by!(tag_names: { name: 'nico:AAA' })
+    nico_tag = ExternalTag.find_by!(platform: :nico, name: 'AAA')
     version = nico_tag.nico_tag_versions.order(:version_no).last
 
     expect(version.version_no).to eq(1)
@@ -378,4 +390,203 @@ RSpec.describe 'nico:sync' do
     expect(versions.second.title).to eq('changed title')
     expect(versions.second.tags).to eq(snapshot_tags(post.reload))
   end
+
+  def create_external_tag!(name)
+    ExternalTag.create!(platform: :nico, name:)
+  end
+
+  def create_nico_sync_post!
+    post = Post.create!(
+      title: 't',
+      url: 'https://www.nicovideo.jp/watch/sm9',
+      uploaded_user: nil
+    )
+
+    PostTag.create!(post:, tag: Tag.no_deerjikist)
+
+    post
+  end
+
+  def run_nico_sync_with_tags! tags
+    stub_python([{
+      'code' => 'sm9',
+      'title' => 't',
+      'tags' => tags,
+      'user' => nil
+    }])
+
+    allow(URI).to receive(:open).and_return(StringIO.new('<html></html>'))
+
+    run_rake_task('nico:sync')
+  end
+
+  it '外部タグだけの変更では bot を付けず，投稿履歴を記録する' do
+    post = create_nico_sync_post!
+    PostVersionRecorder.record!(post:, event_type: :create, created_by_user: nil)
+
+    expect {
+      run_nico_sync_with_tags!(['raw tag[]', 'raw tag[]'])
+    }.to change(PostVersion, :count).by(1)
+        .and change(ExternalTag, :count).by(1)
+        .and change(PostExternalTag, :count).by(1)
+        .and change(NicoTagVersion, :count).by(1)
+        .and change(TagName, :count).by(0)
+
+    external = post.external_tags.sole
+    expect(external.name).to eq('raw tag[]')
+    expect(post.tags.map(&:name)).not_to include('bot操作')
+    expect(post.post_versions.order(:version_no).last.tags_json)
+        .to include('external_tag_id' => external.id)
+
+    expect {
+      run_nico_sync_with_tags!(['raw tag[]'])
+    }.to change(PostVersion, :count).by(0).and change(NicoTagVersion, :count).by(0)
+
+    expect {
+      run_nico_sync_with_tags!([])
+    }.to change(PostVersion, :count).by(1)
+    expect(post.reload.external_tags).to be_empty
+    expect(post.tags.map(&:name)).not_to include('bot操作')
+  end
+
+  it 'サニタイズ後の既存外部タグを再利用し、その連携タグを記載する' do
+    post = create_nico_sync_post!
+
+    create_nico_sanitisation_rules!
+
+    external_tag = create_external_tag!('AAA')
+    linked_tag = create_tag!('spec_linked', category: :general)
+    link_nico_to_tag!(external_tag, linked_tag)
+
+    expect {
+      run_nico_sync_with_tags!(['AAA?'])
+    }.not_to change(ExternalTag, :count)
+
+    post.reload
+
+    expect(post.external_tags).to contain_exactly(external_tag)
+    expect(post.tags).to include(linked_tag)
+    expect(ExternalTag.exists?(platform: :nico, name: 'AAA?')).to be(false)
+  end
+
+  it '外部タグに差分がない場合，内外マッピングが変はっても連携タグを再評価しない' do
+    post = create_nico_sync_post!
+
+    external_tag = create_external_tag!('AAA')
+    old_linked_tag = create_tag!('spec_old_linked', category: :general)
+    new_linked_tag = create_tag!('spec_new_linked', category: :general)
+
+    relation = link_nico_to_tag!(external_tag, old_linked_tag)
+
+    run_nico_sync_with_tags!(['AAA'])
+
+    expect(post.reload.tags).to include(old_linked_tag)
+
+    # 人手で連携タグを消除する．
+    PostTag.find_by!(post:, tag: old_linked_tag).destroy!
+
+    # 内外マッピングを変更する．
+    relation.destroy!
+    link_nico_to_tag!(external_tag, new_linked_tag)
+
+    # 外部タグ自体には差分が無い．
+    run_nico_sync_with_tags!(['AAA'])
+
+    post.reload
+
+    expect(post.external_tags).to include(external_tag)
+    expect(post.tags).not_to include(old_linked_tag)
+    expect(post.tags).not_to include(new_linked_tag)
+  end
+
+  it '外部タグの消除では連携タグを消除せず，再記載時にその時点の内外マッピングを適用する' do
+    post = create_nico_sync_post!
+
+    external_tag = create_external_tag!('AAA')
+    old_linked_tag = create_tag!('spec_old_linked', category: :general)
+    new_linked_tag = create_tag!('spec_new_linked', category: :general)
+
+    relation = link_nico_to_tag!(external_tag, old_linked_tag)
+
+    # 外部タグを新規記載する．
+    run_nico_sync_with_tags!(['AAA'])
+
+    post.reload
+    expect(post.external_tags).to include(external_tag)
+    expect(post.tags).to include(old_linked_tag)
+
+    # 外部タグを消除する．
+    run_nico_sync_with_tags!([])
+
+    post.reload
+    expect(post.external_tags).not_to include(external_tag)
+
+    # 外部タグの消除によって連携タグまでは消除されない．
+    expect(post.tags).to include(old_linked_tag)
+
+    # 外部タグが記載されてゐない間に内外マッピングを変更する．
+    relation.destroy!
+    link_nico_to_tag!(external_tag, new_linked_tag)
+
+    # 同じ外部タグを再記載する．
+    run_nico_sync_with_tags!(['AAA'])
+
+    post.reload
+
+    expect(post.external_tags).to include(external_tag)
+
+    # 旧連携タグは自動的には消除されない．
+    expect(post.tags).to include(old_linked_tag)
+
+    # 再記載時点の内外マッピングが新たに適用される．
+    expect(post.tags).to include(new_linked_tag)
+  end
+
+  it '外部タグだけの変更では bot を付けず、投稿履歴を記録する' do
+    post = create_nico_sync_post!
+    PostVersionRecorder.record!(post:, event_type: :create, created_by_user: nil)
+
+    create_nico_sanitisation_rules!
+
+    expect {
+      run_nico_sync_with_tags!(['AAA?', 'AAA?'])
+    }.to change(PostVersion, :count).by(1)
+        .and change(ExternalTag, :count).by(1)
+        .and change(PostExternalTag, :count).by(1)
+        .and change(NicoTagVersion, :count).by(1)
+        .and change(TagName, :count).by(0)
+
+    external = post.reload.external_tags.sole
+
+    expect(external).to have_attributes(
+        platform: 'nico',
+        name: 'AAA')
+    expect(ExternalTag.exists?(platform: :nico, name: 'AAA?')).to be(false)
+
+    expect(post.tags.map(&:name)).not_to include('bot操作')
+    expect(post.post_versions.order(:version_no).last.tags_json)
+        .to include('external_tag_id' => external.id)
+
+    expect {
+      run_nico_sync_with_tags!(['AAA?'])
+    }.to change(PostVersion, :count).by(0)
+        .and change(NicoTagVersion, :count).by(0)
+
+    expect {
+      run_nico_sync_with_tags!([])
+    }.to change(PostVersion, :count).by(1)
+
+    expect(post.reload.external_tags).to be_empty
+    expect(post.tags.map(&:name)).not_to include('bot操作')
+  end
+
+  it 'nico: prefix を含む従来の規則で外部タグ名をサニタイズする' do
+    post = create_nico_sync_post!
+    create_nico_sanitisation_rules!
+
+    run_nico_sync_with_tags!(['foo:_bar'])
+
+    expect(post.reload.external_tags.sole)
+        .to have_attributes(platform: 'nico', name: 'foo:_bar')
+  end
 end