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: https://git.miteruzo.com/miteruzo/btrc-hub/pulls/289

backend/app/controllers/wiki_pages_controller.rb

@@ -90,7 +90,7 @@ class WikiPagesController < ApplicationController
 
     return head :unprocessable_entity if name.blank? || body.blank?
 
-    tag_name = TagName.find_or_create_by!(name:)
+    tag_name = TagName.find_undiscard_or_create_by!(name:)
     page = WikiPage.new(tag_name:, created_user: current_user, updated_user: current_user)
     if page.save
       message = params[:message].presence

backend/app/models/my_discard.rb

@@ -0,0 +1,20 @@
+module MyDiscard
+  extend ActiveSupport::Concern
+
+  included { include Discard::Model }
+
+  class_methods do
+    def find_undiscard_or_create_by! attrs, &block
+      record = with_discarded.find_by(attrs)
+
+      if record&.discarded?
+        record.undiscard!
+        record.update_columns(created_at: record.reload.updated_at)
+      end
+
+      record or create!(attrs, &block)
+    rescue ActiveRecord::RecordNotUnique
+      retry
+    end
+  end
+end

backend/app/models/tag.rb

@@ -1,5 +1,5 @@
 class Tag < ApplicationRecord
-  include Discard::Model
+  include MyDiscard
 
   class NicoTagNormalisationError < ArgumentError
     ;
@@ -134,10 +134,10 @@ class Tag < ApplicationRecord
   end
 
   def self.find_or_create_by_tag_name! name, category:
-    tn = TagName.find_or_create_by!(name: name.to_s.strip)
+    tn = TagName.find_undiscard_or_create_by!(name: name.to_s.strip)
     tn = tn.canonical if tn.canonical_id?
 
-    Tag.find_or_create_by!(tag_name_id: tn.id) do |t|
+    Tag.find_undiscard_or_create_by!(tag_name_id: tn.id) do |t|
       t.category = category
     end
   rescue ActiveRecord::RecordNotUnique
@@ -162,9 +162,19 @@ class Tag < ApplicationRecord
         end
 
         source_tag_name = source_tag.tag_name
+
+        if source_tag_name.wiki_page.present?
+          raise ActiveRecord::RecordInvalid.new(source_tag_name)
+        end
+
         source_tag.discard!
-        source_tag_name.reload
-        source_tag_name.update!(canonical: target_tag.tag_name)
+
+        if source_tag.nico?
+          source_tag_name.discard!
+        else
+          source_tag_name.update_columns(canonical_id: target_tag.tag_name_id,
+                                         updated_at: Time.current)
+        end
       end
 
       # 投稿件数を再集計

backend/app/models/tag_name.rb

@@ -1,4 +1,8 @@
 class TagName < ApplicationRecord
+  include MyDiscard
+
+  default_scope -> { kept }
+
   has_one :tag
   has_one :wiki_page
 
@@ -10,6 +14,7 @@ class TagName < ApplicationRecord
   validate :canonical_must_be_canonical
   validate :alias_name_must_not_have_prefix
   validate :canonical_must_not_be_present_with_tag_or_wiki_page
+  validate :name_must_be_sanitised
 
   def self.canonicalise names
     names = Array(names).map { |n| n.to_s.strip }.reject(&:blank?)
@@ -39,4 +44,10 @@ class TagName < ApplicationRecord
       errors.add :canonical, 'タグもしくは Wiki の参照がある名前はエーリアスになれません．'
     end
   end
+
+  def name_must_be_sanitised
+    if name? && name != TagNameSanitisationRule.sanitise(name)
+      errors.add :name, '名前に使用できない文字が含まれてゐます．'
+    end
+  end
 end

backend/app/models/tag_name_sanitisation_rule.rb

@@ -0,0 +1,58 @@
+class TagNameSanitisationRule < ApplicationRecord
+  include Discard::Model
+
+  self.primary_key = :priority
+
+  default_scope -> { kept }
+
+  validates :source_pattern, presence: true, uniqueness: true
+
+  validate :source_pattern_must_be_regexp
+
+  class << self
+    def sanitise(name) =
+        rules.reduce(name.dup) { |name, (pattern, replacement)| name.gsub(pattern, replacement) }
+
+    def apply!
+      TagName.find_each do |tn|
+        name = sanitise(tn.name)
+        next if name == tn.name
+
+        TagName.transaction do
+          existing_tn = TagName.find_by(name:)
+          if existing_tn
+            existing_tn = existing_tn.canonical || existing_tn
+            next if existing_tn.id == tn.id
+
+            existing_tag = Tag.find_by(tag_name_id: existing_tn.id)
+            source_tag = Tag.find_by(tag_name_id: tn.id)
+
+            if existing_tag
+              Tag.merge_tags!(existing_tag, source_tag) if tn.tag
+            elsif source_tag
+              source_tag.update_columns(tag_name_id: existing_tn.id, updated_at: Time.current)
+            end
+            tn.discard!
+
+            next
+          end
+
+          # TagName 側の自動サニタイズを回避
+          tn.update_columns(name:, updated_at: Time.current)
+        end
+      end
+    end
+
+    private
+
+    def rules = kept.order(:priority).map { |r| [Regexp.new(r.source_pattern), r.replacement] }
+  end
+
+  private
+
+  def source_pattern_must_be_regexp
+    Regexp.new(source_pattern)
+  rescue RegexpError
+    errors.add :source_pattern, '変な正規表現だね〜（笑）'
+  end
+end

backend/app/models/wiki_page.rb

@@ -2,6 +2,10 @@ require 'set'
 
 
 class WikiPage < ApplicationRecord
+  include MyDiscard
+
+  default_scope -> { kept }
+
   has_many :wiki_revisions, dependent: :destroy
   belongs_to :created_user, class_name: 'User'
   belongs_to :updated_user, class_name: 'User'

backend/db/migrate/20260309123200_create_tag_name_sanitisation_rules.rb

@@ -0,0 +1,31 @@
+class CreateTagNameSanitisationRules < ActiveRecord::Migration[8.0]
+  def up
+    create_table :tag_name_sanitisation_rules, id: :integer, primary_key: :priority do |t|
+      t.string :source_pattern, null: false
+      t.string :replacement, null: false
+      t.timestamps
+      t.datetime :discarded_at
+      t.index :source_pattern, unique: true
+      t.index :discarded_at
+    end
+
+    now = ActiveRecord::Base.connection.quote(Time.current)
+    execute <<~SQL
+      INSERT INTO
+        tag_name_sanitisation_rules(priority, source_pattern, replacement, created_at, updated_at)
+      VALUES
+        (10, '\\\\*',           '_',         #{ now }, #{ now })
+      , (20, '\\\\?',           '_',         #{ now }, #{ now })
+      , (25, '\\\\/',           '_',         #{ now }, #{ now })
+      , (30, '_+',              '_',         #{ now }, #{ now })
+      , (40, '_$',              '',          #{ now }, #{ now })
+      , (45, '^([^:]+\\\\:)?_', '\\\\1',     #{ now }, #{ now })
+      , (50, '^([^:]+\\\\:)?$', '\\\\1null', #{ now }, #{ now })
+      ;
+    SQL
+  end
+
+  def down
+    drop_table :tag_name_sanitisation_rules
+  end
+end

backend/db/migrate/20260311232100_add_discarded_at_to_tag_names.rb

@@ -0,0 +1,6 @@
+class AddDiscardedAtToTagNames < ActiveRecord::Migration[8.0]
+  def change
+    add_column :tag_names, :discarded_at, :datetime
+    add_index :tag_names, :discarded_at
+  end
+end

backend/db/migrate/20260311232300_add_discarded_at_to_wiki_pages.rb

@@ -0,0 +1,6 @@
+class AddDiscardedAtToWikiPages < ActiveRecord::Migration[8.0]
+  def change
+    add_column :wiki_pages, :discarded_at, :datetime
+    add_index :wiki_pages, :discarded_at
+  end
+end

backend/db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema[8.0].define(version: 2026_03_11_123100) do
+ActiveRecord::Schema[8.0].define(version: 2026_03_11_232300) do
   create_table "active_storage_attachments", charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t|
     t.string "name", null: false
     t.string "record_type", null: false
@@ -142,7 +142,9 @@ ActiveRecord::Schema[8.0].define(version: 2026_03_11_123100) do
     t.bigint "canonical_id"
     t.datetime "created_at", null: false
     t.datetime "updated_at", null: false
+    t.datetime "discarded_at"
     t.index ["canonical_id"], name: "index_tag_names_on_canonical_id"
+    t.index ["discarded_at"], name: "index_tag_names_on_discarded_at"
     t.index ["name"], name: "index_tag_names_on_name", unique: true
   end
 
@@ -204,7 +206,9 @@ ActiveRecord::Schema[8.0].define(version: 2026_03_11_123100) do
     t.bigint "updated_user_id", null: false
     t.datetime "created_at", null: false
     t.datetime "updated_at", null: false
+    t.datetime "discarded_at"
     t.index ["created_user_id"], name: "index_wiki_pages_on_created_user_id"
+    t.index ["discarded_at"], name: "index_wiki_pages_on_discarded_at"
     t.index ["tag_name_id"], name: "index_wiki_pages_on_tag_name_id", unique: true
     t.index ["updated_user_id"], name: "index_wiki_pages_on_updated_user_id"
   end

backend/lib/tasks/sync_nico.rake

@@ -108,7 +108,7 @@ namespace :nico do
       desired_non_nico_tag_ids = []
 
       datum['tags'].each do |raw|
-        name = "nico:#{ raw }"
+        name = TagNameSanitisationRule.sanitise("nico:#{ raw }")
         tag = Tag.find_or_create_by_tag_name!(name, category: :nico)
         desired_nico_tag_based_ids << tag.id
 

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 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 aliasing the source tag_name is invalid' do
+    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)