diff --git a/backend/app/models/tag.rb b/backend/app/models/tag.rb index eaf51da..1fd85e6 100644 --- a/backend/app/models/tag.rb +++ b/backend/app/models/tag.rb @@ -160,9 +160,15 @@ class Tag < ApplicationRecord end tag_name = st.tag_name + nico_flg = st.nico? st.destroy! tag_name.reload - tag_name.update!(canonical: target_tag.tag_name) + + if nico_flg + tag_name.destroy! + else + tag_name.update!(canonical: target_tag.tag_name) + end end # 投稿件数を再集計 diff --git a/backend/app/models/tag_name.rb b/backend/app/models/tag_name.rb index 225343d..5b39a06 100644 --- a/backend/app/models/tag_name.rb +++ b/backend/app/models/tag_name.rb @@ -1,4 +1,6 @@ class TagName < ApplicationRecord + before_validation :sanitise_name + has_one :tag has_one :wiki_page @@ -22,6 +24,10 @@ class TagName < ApplicationRecord private + def sanitise_name + self.name = TagNameSanitisationRule.sanitise(name) + end + def canonical_must_be_canonical if canonical&.canonical_id? errors.add :canonical, 'canonical は実体を示す必要があります.' diff --git a/backend/app/models/tag_name_sanitisation_rule.rb b/backend/app/models/tag_name_sanitisation_rule.rb new file mode 100644 index 0000000..d0c8ea1 --- /dev/null +++ b/backend/app/models/tag_name_sanitisation_rule.rb @@ -0,0 +1,55 @@ +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 + + if existing_tn.tag + Tag.merge_tags!(existing_tn.tag, tn.tag) if tn.tag + elsif tn.tag + tn.tag.update_columns(tag_name_id: existing_tn.id, updated_at: Time.current) + end + tn.destroy! + + 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 diff --git a/backend/db/migrate/20260309123200_create_tag_name_sanitisation_rules.rb b/backend/db/migrate/20260309123200_create_tag_name_sanitisation_rules.rb new file mode 100644 index 0000000..e187ce9 --- /dev/null +++ b/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 diff --git a/backend/db/schema.rb b/backend/db/schema.rb index 8077fac..58a7158 100644 --- a/backend/db/schema.rb +++ b/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_03_122700) do +ActiveRecord::Schema[8.0].define(version: 2026_03_09_123200) 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 @@ -127,6 +127,16 @@ ActiveRecord::Schema[8.0].define(version: 2026_03_03_122700) do t.index ["tag_id"], name: "index_tag_implications_on_tag_id" end + create_table "tag_name_sanitisation_rules", primary_key: "priority", id: :integer, charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t| + t.string "source_pattern", null: false + t.string "replacement", null: false + t.datetime "created_at", null: false + t.datetime "updated_at", null: false + t.datetime "discarded_at" + t.index ["discarded_at"], name: "index_tag_name_sanitisation_rules_on_discarded_at" + t.index ["source_pattern"], name: "index_tag_name_sanitisation_rules_on_source_pattern", unique: true + end + create_table "tag_names", charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t| t.string "name", null: false t.bigint "canonical_id" diff --git a/backend/lib/tasks/sync_nico.rake b/backend/lib/tasks/sync_nico.rake index 6ee9c26..09be474 100644 --- a/backend/lib/tasks/sync_nico.rake +++ b/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 diff --git a/backend/spec/models/tag_name_sanitisation_rule_spec.rb b/backend/spec/models/tag_name_sanitisation_rule_spec.rb new file mode 100644 index 0000000..9a95c3b --- /dev/null +++ b/backend/spec/models/tag_name_sanitisation_rule_spec.rb @@ -0,0 +1,63 @@ +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 rules 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) { TagName.create!(name: 'foo_bar') } + + 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) { TagName.create!(name: 'foo_bar') } + + 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: 'foo_bar', category: :general) } + + it 'moves the tag to the existing tag_name' do + described_class.apply! + expect(source_tag.reload.tag_name_id).to eq(existing.id) + end + end + end +end diff --git a/backend/spec/models/tag_name_spec.rb b/backend/spec/models/tag_name_spec.rb new file mode 100644 index 0000000..b89583c --- /dev/null +++ b/backend/spec/models/tag_name_spec.rb @@ -0,0 +1,28 @@ +require 'rails_helper' + +RSpec.describe TagName, type: :model do + describe 'before_validation :sanitise_name' do + before do + TagNameSanitisationRule.create!(priority: 10, source_pattern: '_', replacement: '') + end + + it 'sanitises name on create' do + tag_name = TagName.create!(name: 'foo_bar') + expect(tag_name.reload.name).to eq('foobar') + end + + it 'sanitises name on update' do + tag_name = TagName.create!(name: 'foobar') + tag_name.update!(name: 'foo_bar') + expect(tag_name.reload.name).to eq('foobar') + end + + it 'becomes invalid when sanitised name conflicts with another tag_name' do + TagName.create!(name: 'foobar') + tag_name = TagName.new(name: 'foo_bar') + + expect(tag_name).to be_invalid + expect(tag_name.errors[:name]).to be_present + end + end +end diff --git a/backend/spec/models/tag_spec.rb b/backend/spec/models/tag_spec.rb index 389f1a1..b63710a 100644 --- a/backend/spec/models/tag_spec.rb +++ b/backend/spec/models/tag_spec.rb @@ -2,10 +2,13 @@ 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) } @@ -15,7 +18,8 @@ RSpec.describe Tag, type: :model do expect(source_post_tag.reload.tag_id).to eq(target_tag.id) expect(Tag.exists?(source_tag.id)).to be(false) - expect(source_tag.tag_name.reload.canonical_id).to eq(target_tag.tag_name_id) + 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 @@ -31,8 +35,10 @@ RSpec.describe Tag, type: :model do expect(active.count).to eq(1) expect(discarded_source.discarded_at).to be_present + expect(discarded_source.tag_id).to eq(source_tag.id) expect(Tag.exists?(source_tag.id)).to be(false) - expect(source_tag.tag_name.reload.canonical_id).to eq(target_tag.tag_name_id) + 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 @@ -45,6 +51,7 @@ RSpec.describe Tag, type: :model do expect(Tag.exists?(target_tag.id)).to be(true) expect(Tag.exists?(source_tag.id)).to be(false) expect(source_post_tag.reload.tag_id).to eq(target_tag.id) + expect(target_tag.reload.post_count).to eq(1) end end @@ -52,9 +59,10 @@ RSpec.describe Tag, type: :model 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, - created_user: create_admin_user!, - updated_user: create_admin_user!) + tag_name: source_tag_name, + created_user: create_admin_user!, + updated_user: create_admin_user! + ) end it 'rolls back the transaction' do @@ -64,7 +72,20 @@ RSpec.describe Tag, type: :model do expect(Tag.exists?(source_tag.id)).to be(true) expect(source_post_tag.reload.tag_id).to eq(source_tag.id) - expect(source_tag.tag_name.reload.canonical_id).to be_nil + expect(source_tag_name.reload.canonical_id).to be_nil + 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 'deletes the source tag_name instead of aliasing it' do + described_class.merge_tags!(target_tag, [source_tag]) + + expect(Tag.exists?(source_tag.id)).to be(false) + expect(TagName.exists?(source_tag_name_id)).to be(false) end end end