TagName サニタイズ（#281） (#289)

#281

#281

Merge remote-tracking branch 'origin/main' into feature/281

#281

#281 テストまだ通ってないので要確認

Co-authored-by: miteruzo <miteruzo@naver.com>
Reviewed-on: #289

backend/spec/factories/tags.rb

@@ -1,12 +1,22 @@
 FactoryBot.define do
   factory :tag do
+    transient do
+      name { nil }
+    end
+
     category { :general }
     post_count { 0 }
     association :tag_name
 
+    after(:build) do |tag, evaluator|
+      tag.name = evaluator.name if evaluator.name.present?
+    end
+
     trait :nico do
       category { :nico }
-      tag_name { association(:tag_name, name: "nico:#{ SecureRandom.hex(4) }") }
+      transient do
+        name { "nico:#{ SecureRandom.hex(4) }" }
+      end
     end
   end
 end

backend/spec/models/tag_name_sanitisation_rule_spec.rb

@@ -0,0 +1,101 @@
+require 'rails_helper'
+
+RSpec.describe TagNameSanitisationRule, type: :model do
+  describe '.sanitise' do
+    before do
+      described_class.create!(priority: 10, source_pattern: '_', replacement: '')
+      described_class.create!(priority: 20, source_pattern: 'ABC', replacement: 'abc')
+    end
+
+    it 'applies sanitisation rules sequentially in priority order' do
+      expect(described_class.sanitise('A_B_C')).to eq('abc')
+    end
+
+    it 'does not fail when a rule does not match' do
+      expect { described_class.sanitise('xyz') }.not_to raise_error
+      expect(described_class.sanitise('xyz')).to eq('xyz')
+    end
+  end
+
+  describe 'validations' do
+    it 'is invalid when source_pattern is not a valid regexp' do
+      rule = described_class.new(priority: 10, source_pattern: '[', replacement: '')
+      expect(rule).to be_invalid
+      expect(rule.errors[:source_pattern]).to be_present
+    end
+  end
+
+  describe '.apply!' do
+    before do
+      described_class.create!(priority: 10, source_pattern: '_', replacement: '')
+    end
+
+    context 'when no conflicting tag_name exists' do
+      let!(:tag_name) do
+        TagName.create!(name: 'tmp').tap do |tn|
+          tn.update_columns(name: 'foo_bar', updated_at: Time.current)
+        end
+      end
+
+      it 'renames the tag_name' do
+        described_class.apply!
+        expect(tag_name.reload.name).to eq('foobar')
+      end
+    end
+
+    context 'when a conflicting canonical tag_name exists' do
+      let!(:existing) { TagName.create!(name: 'foobar') }
+      let!(:source) do
+        TagName.create!(name: 'tmp').tap do |tn|
+          tn.update_columns(name: 'foo_bar', updated_at: Time.current)
+        end
+      end
+
+      it 'deletes the source tag_name' do
+        described_class.apply!
+        expect(TagName.exists?(source.id)).to be(false)
+        expect(existing.reload.name).to eq('foobar')
+      end
+    end
+
+    context 'when the source tag_name has a tag and the existing one has no tag' do
+      let!(:existing) { TagName.create!(name: 'foobar') }
+      let!(:source_tag) { create(:tag, name: 'tmp', category: :general) }
+      let!(:source_tag_name_id) { source_tag.tag_name_id }
+
+      before do
+        source_tag.tag_name.update_columns(name: 'foo_bar', updated_at: Time.current)
+      end
+
+      it 'moves the tag to the existing tag_name' do
+        described_class.apply!
+        expected_tag_name_id = existing.canonical_id || existing.id
+        expect(source_tag.reload.tag_name_id).to eq(expected_tag_name_id)
+        expect(TagName.exists?(source_tag_name_id)).to be(false)
+      end
+    end
+
+    context 'when both source and existing tag_names have tags' do
+      let!(:existing_tn) { TagName.create!(name: 'foobar') }
+      let!(:existing_tag) { Tag.create!(tag_name: existing_tn, category: :general) }
+
+      let!(:source_tn) { TagName.create!(name: 'tmp') }
+      let!(:source_tag) { Tag.create!(tag_name: source_tn, category: :general) }
+      let!(:source_tag_name_id) { source_tn.id }
+
+      before do
+        source_tn.update_columns(name: 'foo_bar', updated_at: Time.current)
+      end
+
+      it 'merges the source tag into the existing tag and deletes the source tag_name' do
+        expect(TagName.find_by(name: 'foobar')&.tag&.id).to eq(existing_tag.id)
+        expect(TagName.find_by(name: 'foo_bar')&.tag&.id).to eq(source_tag.id)
+
+        described_class.apply!
+
+        expect(Tag.exists?(source_tag.id)).to be(false)
+        expect(TagName.exists?(source_tag.tag_name_id)).to be(false)
+      end
+    end
+  end
+end

backend/spec/models/tag_spec.rb

@@ -2,16 +2,18 @@ require 'rails_helper'
 
 RSpec.describe Tag, type: :model do
   describe '.merge_tags!' do
-    let!(:target_tag) { create(:tag) }
-    let!(:source_tag) { create(:tag) }
+    let!(:target_tag) { create(:tag, category: :general) }
+    let!(:source_tag) { create(:tag, category: :general) }
+    let!(:source_tag_name) { source_tag.tag_name }
 
-    let!(:post_record) { Post.create!(url: 'https://example.com/posts/1', title: 'test post') }
+    let!(:post_record) do
+      Post.create!(url: 'https://example.com/posts/1', title: 'test post')
+    end
 
     context 'when merging a simple source tag' do
       let!(:source_post_tag) { PostTag.create!(post: post_record, tag: source_tag) }
 
-      it 'discards the source post_tag, creates an active target post_tag, ' +
-         'discards the source tag, and aliases the source tag_name' do
+      it 'discards the source post_tag, creates an active target post_tag, discards the source tag, and aliases the source tag_name' do
         described_class.merge_tags!(target_tag, [source_tag])
 
         source_pt = PostTag.with_discarded.find(source_post_tag.id)
@@ -22,7 +24,9 @@ RSpec.describe Tag, type: :model do
         expect(active_target).to be_present
 
         expect(Tag.with_discarded.find(source_tag.id)).to be_discarded
-        expect(source_tag.tag_name.reload.canonical_id).to eq(target_tag.tag_name_id)
+        expect(TagName.with_discarded.find(source_tag_name.id)).not_to be_discarded
+        expect(source_tag_name.reload.canonical_id).to eq(target_tag.tag_name_id)
+        expect(target_tag.reload.post_count).to eq(1)
       end
     end
 
@@ -30,8 +34,7 @@ RSpec.describe Tag, type: :model do
       let!(:target_post_tag) { PostTag.create!(post: post_record, tag: target_tag) }
       let!(:source_post_tag) { PostTag.create!(post: post_record, tag: source_tag) }
 
-      it 'discards the source post_tag, keeps one active target post_tag, ' +
-         'and aliases the source tag_name' do
+      it 'discards the source post_tag, keeps one active target post_tag, discards the source tag, and aliases the source tag_name' do
         described_class.merge_tags!(target_tag, [source_tag])
 
         source_pt = PostTag.with_discarded.find(source_post_tag.id)
@@ -43,7 +46,9 @@ RSpec.describe Tag, type: :model do
         expect(active.first.id).to eq(target_post_tag.id)
 
         expect(Tag.with_discarded.find(source_tag.id)).to be_discarded
-        expect(source_tag.tag_name.reload.canonical_id).to eq(target_tag.tag_name_id)
+        expect(TagName.with_discarded.find(source_tag_name.id)).not_to be_discarded
+        expect(source_tag_name.reload.canonical_id).to eq(target_tag.tag_name_id)
+        expect(target_tag.reload.post_count).to eq(1)
       end
     end
 
@@ -61,14 +66,49 @@ RSpec.describe Tag, type: :model do
         expect(source_pt.discarded_at).to be_present
         expect(source_pt.tag_id).to eq(source_tag.id)
         expect(active_target).to be_present
+        expect(source_tag_name.reload.canonical_id).to eq(target_tag.tag_name_id)
+        expect(target_tag.reload.post_count).to eq(1)
       end
     end
 
-    context 'when aliasing the source tag_name is invalid' do
+    context 'when the source tag_name is invalid under sanitisation rules' do
+      let!(:source_post_tag) { PostTag.create!(post: post_record, tag: source_tag) }
+      let!(:sanitisation_rule) do
+        TagNameSanitisationRule.create!(
+          priority: 99_999,
+          source_pattern: 'INVALIDTOKEN',
+          replacement: ''
+        )
+      end
+
+      before do
+        source_tag_name.update_columns(
+          name: "#{ source_tag_name.name }INVALIDTOKEN",
+          updated_at: Time.current
+        )
+      end
+
+      it 'still merges, but discards the source tag_name instead of aliasing it' do
+        described_class.merge_tags!(target_tag, [source_tag])
+
+        source_pt = PostTag.with_discarded.find(source_post_tag.id)
+        active_target = PostTag.kept.find_by(post_id: post_record.id, tag_id: target_tag.id)
+        discarded_source_tag_name = TagName.with_discarded.find(source_tag_name.id)
+
+        expect(source_pt.discarded_at).to be_present
+        expect(source_pt.tag_id).to eq(source_tag.id)
+        expect(active_target).to be_present
+
+        expect(Tag.with_discarded.find(source_tag.id)).to be_discarded
+        expect(target_tag.reload.post_count).to eq(1)
+      end
+    end
+
+    context 'when the source tag_name has a wiki_page' do
       let!(:source_post_tag) { PostTag.create!(post: post_record, tag: source_tag) }
       let!(:wiki_page) do
         WikiPage.create!(
-          tag_name: source_tag.tag_name,
+          tag_name: source_tag_name,
           created_user: create_admin_user!,
           updated_user: create_admin_user!
         )
@@ -80,8 +120,29 @@ RSpec.describe Tag, type: :model do
         }.to raise_error(ActiveRecord::RecordInvalid)
 
         expect(Tag.with_discarded.find(source_tag.id)).not_to be_discarded
+        expect(TagName.with_discarded.find(source_tag_name.id)).not_to be_discarded
         expect(PostTag.kept.find(source_post_tag.id).tag_id).to eq(source_tag.id)
-        expect(source_tag.tag_name.reload.canonical_id).to be_nil
+        expect(PostTag.kept.find_by(post_id: post_record.id, tag_id: target_tag.id)).to be_nil
+        expect(source_tag_name.reload.canonical_id).to be_nil
+        expect(target_tag.reload.post_count).to eq(0)
+      end
+    end
+
+    context 'when merging a nico source tag' do
+      let!(:target_tag) { create(:tag, category: :nico, name: 'nico:foo') }
+      let!(:source_tag) { create(:tag, category: :nico, name: 'nico:bar') }
+      let!(:source_tag_name_id) { source_tag.tag_name_id }
+
+      it 'discards the source tag_name instead of aliasing it' do
+        described_class.merge_tags!(target_tag, [source_tag])
+
+        discarded_source_tag = Tag.with_discarded.find(source_tag.id)
+        discarded_source_tag_name = TagName.with_discarded.find(source_tag_name_id)
+
+        expect(discarded_source_tag).to be_discarded
+        expect(discarded_source_tag_name).to be_discarded
+        expect(discarded_source_tag_name.canonical_id).to be_nil
+        expect(target_tag.reload.post_count).to eq(0)
       end
     end
   end

backend/spec/requests/posts_spec.rb

@@ -525,8 +525,9 @@ RSpec.describe 'Posts API', type: :request do
 
     context "when nico tag already exists in tags" do
       before do
-        Tag.find_or_create_by!(tag_name: TagName.find_or_create_by!(name: 'nico:nico_tag'),
-                               category: :nico)
+        Tag.find_undiscard_or_create_by!(
+            tag_name: TagName.find_undiscard_or_create_by!(name: 'nico:nico_tag'),
+            category: :nico)
       end
 
       it 'return 400' do
@@ -610,8 +611,9 @@ RSpec.describe 'Posts API', type: :request do
 
     context "when nico tag already exists in tags" do
       before do
-        Tag.find_or_create_by!(tag_name: TagName.find_or_create_by!(name: 'nico:nico_tag'),
-                               category: :nico)
+        Tag.find_undiscard_or_create_by!(
+            tag_name: TagName.find_undiscard_or_create_by!(name: 'nico:nico_tag'),
+            category: :nico)
       end
 
       it 'return 400' do

backend/spec/tasks/nico_sync_spec.rb

@@ -8,8 +8,8 @@ RSpec.describe "nico:sync" do
   end
 
   def create_tag!(name, category:)
-    tn = TagName.find_or_create_by!(name: name.to_s.strip)
-    Tag.find_or_create_by!(tag_name_id: tn.id) { |t| t.category = category }
+    tn = TagName.find_undiscard_or_create_by!(name: name.to_s.strip)
+    Tag.find_undiscard_or_create_by!(tag_name_id: tn.id) { |t| t.category = category }
   end
 
   def link_nico_to_tag!(nico_tag, tag)