feat: 類似度算出バッチ修正，ほか（#228） (#232)

#228 #228 #228 Co-authored-by: miteruzo <miteruzo@naver.com> Reviewed-on: #232
2026-01-22 23:30:08 +09:00
parent 86209dcc84
commit f6de272f55
18 changed files with 553 additions and 76 deletions
@@ -49,13 +49,19 @@ class TagsController < ApplicationController
    return head :unauthorized unless current_user
    return head :forbidden unless current_user.member?

+    name = params[:name].presence
+    category = params[:category].presence
+
    tag = Tag.find(params[:id])

-    attrs = { name: params[:name].presence,
-              category: params[:category].presence }.compact
+    if name.present?
+      tag.tag_name.update!(name:)
+    end

-    tag.update!(attrs) if attrs.present?
+    if category.present?
+      tag.update!(category:)
+    end

-    render json: tag
+    render json: tag.as_json(methods: [:name])
  end
 end
@@ -9,12 +9,7 @@ class Post < ApplicationRecord
  has_many :post_tags_with_discarded, -> { with_discarded }, class_name: 'PostTag'
  has_many :tags, through: :active_post_tags
  has_many :user_post_views, dependent: :destroy
-  has_many :post_similarities_as_post,
-           class_name: 'PostSimilarity',
-           foreign_key: :post_id
-  has_many :post_similarities_as_target_post,
-           class_name: 'PostSimilarity',
-           foreign_key: :target_post_id
+  has_many :post_similarities
  has_one_attached :thumbnail

  before_validation :normalise_url
@@ -34,18 +29,12 @@ class Post < ApplicationRecord
  end

  def related(limit: nil)
-    ids_with_cos =
-        post_similarities_as_post.select(:target_post_id, :cos)
-                                 .map { |ps| [ps.target_post_id, ps.cos] } +
-        post_similarities_as_target_post.select(:post_id, :cos)
-                                        .map { |ps| [ps.post_id, ps.cos] }
+    ids = post_similarities.select(:target_post_id).order(cos: :desc)
+    ids = ids.limit(limit) if limit
+    ids = ids.pluck(:target_post_id)
+    return [] if ids.empty?

-    sorted = ids_with_cos.sort_by { |_, cos| -cos }
-
-    ids = sorted.map(&:first)
-    ids = ids.first(limit) if limit
-
-    Post.where(id: ids).index_by(&:id).values_at(*ids)
+    Post.where(id: ids).order(Arel.sql("FIELD(id, #{ ids.join(',') })"))
  end

  def resized_thumbnail!
@@ -1,4 +1,6 @@
 class PostSimilarity < ApplicationRecord
-  belongs_to :post, class_name: 'Post', foreign_key: 'post_id'
-  belongs_to :target_post, class_name: 'Post', foreign_key: 'target_post_id'
+  self.primary_key = :post_id, :target_post_id
+
+  belongs_to :post
+  belongs_to :target_post, class_name: 'Post'
 end
@@ -1,4 +1,6 @@
 class TagSimilarity < ApplicationRecord
-  belongs_to :tag, class_name: 'Tag', foreign_key: 'tag_id'
-  belongs_to :target_tag, class_name: 'Tag', foreign_key: 'target_tag_id'
+  self.primary_key = :tag_id, :target_tag_id
+
+  belongs_to :tag
+  belongs_to :target_tag, class_name: 'Tag'
 end
@@ -0,0 +1,106 @@
+module Similarity
+  class Calc
+    def self.call model, tgt
+      similarity_model = "#{ model.name }Similarity".constantize
+
+      # 最大保存件数
+      n = 20
+
+      similarity_model.delete_all
+
+      posts = model.includes(tgt).select(:id).to_a
+
+      tag_ids = { }
+      tag_cnts = { }
+
+      posts.each do |p|
+        arr = p.public_send(tgt).map(&:id).sort
+        tag_ids[p.id] = arr
+        tag_cnts[p.id] = arr.size
+      end
+
+      intersection_size = -> a, b do
+        i = 0
+        j = 0
+        cnt = 0
+        while i < a.size && j < b.size
+          a_i = a[i]
+          b_j = b[j]
+          if a_i == b_j
+            cnt += 1
+            i += 1
+            j += 1
+          elsif a_i < b_j
+            i += 1
+          else
+            j += 1
+          end
+        end
+        cnt
+      end
+
+      push_topk = -> list, cos, target_id do
+        return if list.size >= n && cos <= list[-1][0]
+
+        idx = nil
+        list.each_with_index do |(c, tid), i|
+          if tid == target_id
+            idx = i
+            break
+          end
+        end
+        if idx
+          return if cos <= list[idx][0]
+          list.delete_at(idx)
+        end
+
+        insert_at = list.size
+        list.each_with_index do |(c, _), i|
+          if cos > c
+            insert_at = i
+            break
+          end
+        end
+        list.insert(insert_at, [cos, target_id])
+        list.pop if list.size > n
+      end
+
+      top = Hash.new { |h, key| h[key] = [] }
+
+      ids = posts.map(&:id)
+      ids.each_with_index do |post_id, i|
+        a = tag_ids[post_id]
+        a_cnt = tag_cnts[post_id]
+
+        ((i + 1)...ids.size).each do |j|
+          target_id = ids[j]
+          b = tag_ids[target_id]
+          b_cnt = tag_cnts[target_id]
+
+          norm = Math.sqrt(a_cnt * b_cnt)
+          cos = norm.zero? ? 0.0 : intersection_size.(a, b).fdiv(norm)
+
+          push_topk.(top[post_id], cos, target_id)
+          push_topk.(top[target_id], cos, post_id)
+        end
+      end
+
+      buf = []
+      flush = -> do
+        return if buf.empty?
+        similarity_model.insert_all!(buf)
+        buf.clear
+      end
+
+      top.each do |post_id, list|
+        list.each do |cos, target_post_id|
+          buf << { "#{ model.name.underscore }_id".to_sym => post_id,
+                   "target_#{ model.name.underscore }_id".to_sym => target_post_id,
+                   cos: }
+          flush.call if buf.size >= 1_000
+        end
+      end
+      flush.call
+    end
+  end
+end
@@ -0,0 +1,43 @@
+class ChangePostSimilaritiesToCompositePk < ActiveRecord::Migration[8.0]
+  def up
+    execute <<~SQL
+      ALTER TABLE
+        post_similarities
+          MODIFY COLUMN id BIGINT NOT NULL
+      ;
+    SQL
+
+    execute <<~SQL
+      ALTER TABLE
+        post_similarities
+          DROP PRIMARY KEY
+      ;
+    SQL
+
+    remove_column :post_similarities, :id
+
+    execute <<~SQL
+      ALTER TABLE
+        post_similarities
+          ADD PRIMARY KEY (post_id, target_post_id)
+      ;
+    SQL
+  end
+
+  def down
+    execute <<~SQL
+      ALTER TABLE
+        post_similarities
+          DROP PRIMARY KEY
+      ;
+    SQL
+
+    execute <<~SQL
+      ALTER TABLE
+        post_similarities
+          ADD COLUMN id BIGINT NOT NULL AUTO_INCREMENT FIRST
+        , ADD PRIMARY KEY (id)
+      ;
+    SQL
+  end
+end
@@ -0,0 +1,43 @@
+class ChangeTagSimilaritiesToCompositePk < ActiveRecord::Migration[8.0]
+  def up
+    execute <<~SQL
+      ALTER TABLE
+        tag_similarities
+          MODIFY COLUMN id BIGINT NOT NULL
+      ;
+    SQL
+
+    execute <<~SQL
+      ALTER TABLE
+        tag_similarities
+          DROP PRIMARY KEY
+      ;
+    SQL
+
+    remove_column :tag_similarities, :id
+
+    execute <<~SQL
+      ALTER TABLE
+        tag_similarities
+          ADD PRIMARY KEY (tag_id, target_tag_id)
+      ;
+    SQL
+  end
+
+  def down
+    execute <<~SQL
+      ALTER TABLE
+        tag_similarities
+          DROP PRIMARY KEY
+      ;
+    SQL
+
+    execute <<~SQL
+      ALTER TABLE
+        tag_similarities
+          ADD COLUMN id BIGINT NOT NULL AUTO_INCREMENT FIRST
+        , ADD PRIMARY KEY (id)
+      ;
+    SQL
+  end
+end
@@ -0,0 +1,9 @@
+class AddIndexToPostSimilarities < ActiveRecord::Migration[8.0]
+  def change
+    remove_index :post_similarities, name: 'index_post_similarities_on_post_id'
+
+    add_index :post_similarities, [:post_id, :cos],
+              order: { cos: :desc },
+              name: 'index_post_similarities_on_post_id_and_cos'
+  end
+end
@@ -0,0 +1,9 @@
+class AddIndexToTagSimilarities < ActiveRecord::Migration[8.0]
+  def change
+    remove_index :tag_similarities, name: 'index_tag_similarities_on_tag_id'
+
+    add_index :tag_similarities, [:tag_id, :cos],
+              order: { cos: :desc },
+              name: 'index_tag_similarities_on_tag_id_and_cos'
+  end
+end
@@ -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_01_18_144400) do
+ActiveRecord::Schema[8.0].define(version: 2026_01_21_225600) 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
@@ -55,11 +55,11 @@ ActiveRecord::Schema[8.0].define(version: 2026_01_18_144400) do
    t.index ["tag_id"], name: "index_nico_tag_relations_on_tag_id"
  end

-  create_table "post_similarities", charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t|
+  create_table "post_similarities", primary_key: ["post_id", "target_post_id"], charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t|
    t.bigint "post_id", null: false
    t.bigint "target_post_id", null: false
    t.float "cos", null: false
-    t.index ["post_id"], name: "index_post_similarities_on_post_id"
+    t.index ["post_id", "cos"], name: "index_post_similarities_on_post_id_and_cos", order: { cos: :desc }
    t.index ["target_post_id"], name: "index_post_similarities_on_target_post_id"
  end

@@ -126,11 +126,11 @@ ActiveRecord::Schema[8.0].define(version: 2026_01_18_144400) do
    t.index ["name"], name: "index_tag_names_on_name", unique: true
  end

-  create_table "tag_similarities", charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t|
+  create_table "tag_similarities", primary_key: ["tag_id", "target_tag_id"], charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t|
    t.bigint "tag_id", null: false
    t.bigint "target_tag_id", null: false
    t.float "cos", null: false
-    t.index ["tag_id"], name: "index_tag_similarities_on_tag_id"
+    t.index ["tag_id", "cos"], name: "index_tag_similarities_on_tag_id_and_cos", order: { cos: :desc }
    t.index ["target_tag_id"], name: "index_tag_similarities_on_target_tag_id"
  end

@@ -1,28 +1,6 @@
 namespace :post_similarity do
  desc '関聯投稿テーブル作成'
  task calc: :environment do
-    dot = -> a, b { (a.keys & b.keys).sum { |k| a[k] * b[k] } }
-    norm = -> v { Math.sqrt(v.values.sum { |e| e * e }) }
-    cos = -> a, b do
-      na = norm.(a)
-      nb = norm.(b)
-      if na.zero? || nb.zero?
-        0.0
-      else
-        dot.(a, b) / na / nb
-      end
-    end
-
-    posts = Post.includes(:tags).to_a
-    posts.each_with_index do |post, i|
-      existence_of_tags = post.tags.index_with(1)
-      ((i + 1)...posts.size).each do |j|
-        target_post = posts[j]
-        existence_of_target_tags = target_post.tags.index_with(1)
-        PostSimilarity.find_or_initialize_by(post:, target_post:).tap { |ps|
-          ps.cos = cos.(existence_of_tags, existence_of_target_tags)
-        }.save!
-      end
-    end
+    Similarity::Calc.call(Post, :tags)
  end
 end
@@ -1,28 +1,6 @@
 namespace :tag_similarity do
  desc '関聯タグ・テーブル作成'
  task calc: :environment do
-    dot = -> a, b { (a.keys & b.keys).sum { |k| a[k] * b[k] } }
-    norm = -> v { Math.sqrt(v.values.sum { |e| e * e }) }
-    cos = -> a, b do
-      na = norm.(a)
-      nb = norm.(b)
-      if na.zero? || nb.zero?
-        0.0
-      else
-        dot.(a, b) / na / nb
-      end
-    end
-
-    tags = Tag.includes(:posts).to_a
-    tags.each_with_index do |tag, i|
-      existence_of_posts = tag.posts.index_with(1)
-      ((i + 1)...tags.size).each do |j|
-        target_tag = tags[j]
-        existence_of_target_posts = target_tag.posts.index_with(1)
-        TagSimilarity.find_or_initialize_by(tag:, target_tag:).tap { |ts|
-          ts.cos = cos.(existence_of_posts, existence_of_target_posts)
-        }.save!
-      end
-    end
+    Similarity::Calc.call(Tag, :posts)
  end
 end
@@ -7,5 +7,9 @@ FactoryBot.define do
    trait :member do
      role { "member" }
    end
+
+    trait :admin do
+      role { 'admin' }
+    end
  end
 end
@@ -0,0 +1,134 @@
+# spec/requests/tag_children_spec.rb
+require "rails_helper"
+
+RSpec.describe "TagChildren", type: :request do
+  let!(:parent) { create(:tag) }
+  let!(:child)  { create(:tag) }
+
+  # ここは君のUser factoryに合わせて調整
+  let(:user) { create_member_user! }
+  let(:admin) { create_admin_user! }
+
+  # current_user を ApplicationController でスタブ
+  def stub_current_user(user_or_nil)
+    allow_any_instance_of(ApplicationController)
+      .to receive(:current_user)
+      .and_return(user_or_nil)
+  end
+
+  describe "POST /tag_children" do
+    subject(:do_request) do
+      post "/tags/#{ parent_id }/children/#{ child_id }"
+    end
+
+    context "when not logged in" do
+      let(:parent_id) { parent.id }
+      let(:child_id) { child.id }
+
+      it "returns 401" do
+        stub_current_user(nil)
+        do_request
+        expect(response).to have_http_status(:unauthorized)
+      end
+    end
+
+    context "when logged in but not admin" do
+      let(:parent_id) { parent.id }
+      let(:child_id) { child.id }
+
+      it "returns 403" do
+        stub_current_user(user)
+        do_request
+        expect(response).to have_http_status(:forbidden)
+      end
+    end
+
+    context "when admin and params are present" do
+      before { stub_current_user(admin) }
+      let(:parent_id) { parent.id }
+      let(:child_id) { child.id }
+
+      it "returns 204 and adds child to parent.children" do
+        expect(parent.children).not_to include(child)
+
+        expect { do_request }
+          .to change { parent.reload.children.ids.include?(child.id) }
+          .from(false).to(true)
+
+        expect(response).to have_http_status(:no_content)
+      end
+    end
+
+    context "when Tag.find raises (invalid ids) it still returns 204" do
+      before { stub_current_user(admin) }
+
+      let(:parent_id) { -1 }
+      let(:child_id) { -1 }
+
+      it "returns 204 (rescue nil)" do
+        do_request
+        expect(response).to have_http_status(:no_content)
+      end
+    end
+  end
+
+  describe "DELETE /tag_children" do
+    subject(:do_request) do
+      delete "/tags/#{ parent_id }/children/#{ child_id }"
+    end
+
+    context "when not logged in" do
+      let(:parent_id) { parent.id }
+      let(:child_id) { child.id }
+
+      it "returns 401" do
+        stub_current_user(nil)
+        do_request
+        expect(response).to have_http_status(:unauthorized)
+      end
+    end
+
+    context "when logged in but not admin" do
+      let(:parent_id) { parent.id }
+      let(:child_id) { child.id }
+
+      it "returns 403" do
+        stub_current_user(user)
+        do_request
+        expect(response).to have_http_status(:forbidden)
+      end
+    end
+
+    context "when admin and params are present" do
+      before do
+        stub_current_user(admin)
+        parent.children << child
+      end
+
+      let(:parent_id) { parent.id }
+      let(:child_id) { child.id }
+
+      it "returns 204 and removes child from parent.children" do
+        expect(parent.reload.children).to include(child)
+
+        expect { do_request }
+          .to change { parent.reload.children.ids.include?(child.id) }
+          .from(true).to(false)
+
+        expect(response).to have_http_status(:no_content)
+      end
+    end
+
+    context "when Tag.find raises (invalid ids) it still returns 204" do
+      before { stub_current_user(admin) }
+
+      let(:parent_id) { -1 }
+      let(:child_id) { -1 }
+
+      it "returns 204 (rescue nil)" do
+        do_request
+        expect(response).to have_http_status(:no_content)
+      end
+    end
+  end
+end
@@ -76,4 +76,103 @@ RSpec.describe 'Tags API', type: :request do
      expect(response).to have_http_status(:not_found)
    end
  end
+
+  # member? を持つ user を想定（Factory 側で trait 作ってもOK）
+  let(:member_user) { create(:user) }
+  let(:non_member_user) { create(:user) }
+
+  def stub_current_user(user)
+    allow_any_instance_of(ApplicationController).to receive(:current_user).and_return(user)
+  end
+
+  before do
+    allow(member_user).to receive(:member?).and_return(true)
+    allow(non_member_user).to receive(:member?).and_return(false)
+  end
+
+  describe "PATCH /tags/:id" do
+    context "未ログイン" do
+      before { stub_current_user(nil) }
+
+      it "401 を返す" do
+        patch "/tags/#{tag.id}", params: { name: "new" }
+        expect(response).to have_http_status(:unauthorized)
+      end
+    end
+
+    context "ログインしてゐるが member でない" do
+      before { stub_current_user(non_member_user) }
+
+      it "403 を返す" do
+        patch "/tags/#{tag.id}", params: { name: "new" }
+        expect(response).to have_http_status(:forbidden)
+      end
+    end
+
+    context "member" do
+      before { stub_current_user(member_user) }
+
+      it "name だけ更新できる" do
+        patch "/tags/#{tag.id}", params: { name: "new" }
+
+        expect(response).to have_http_status(:ok)
+
+        tag.reload
+        expect(tag.name).to eq("new")
+        expect(tag.category).to eq("general")
+
+        json = JSON.parse(response.body)
+        expect(json["id"]).to eq(tag.id)
+        expect(json["name"]).to eq("new")
+        expect(json["category"]).to eq("general")
+      end
+
+      it "category だけ更新できる" do
+        patch "/tags/#{tag.id}", params: { category: "meme" }
+
+        expect(response).to have_http_status(:ok)
+
+        tag.reload
+        expect(tag.name).to eq("spec_tag")
+        expect(tag.category).to eq("meme")
+      end
+
+      it "空文字は presence により無視され、更新は走らない（値が変わらない）" do
+        patch "/tags/#{tag.id}", params: { name: "", category: "   " }
+
+        expect(response).to have_http_status(:ok)
+
+        tag.reload
+        expect(tag.name).to eq("spec_tag")
+        expect(tag.category).to eq("general")
+      end
+
+      it "両方更新できる" do
+        patch "/tags/#{tag.id}", params: { name: "n", category: "meta" }
+
+        expect(response).to have_http_status(:ok)
+
+        tag.reload
+        expect(tag.name).to eq("n")
+        expect(tag.category).to eq("meta")
+      end
+
+      it "存在しない id だと RecordNotFound になる（通常は 404）" do
+        # Rails 設定次第で例外がそのまま上がる/404になる
+        # APIなら rescue_from で 404 にしてることが多いので、その場合は 404 を期待。
+        patch "/tags/999999999", params: { name: "x" }
+
+        expect(response.status).to be_in([404, 500])
+      end
+
+      it "バリデーションで update! が失敗したら（通常は 422 か 500）" do
+        patch "/tags/#{tag.id}", params: { name: 'new', category: 'nico' }
+
+        # rescue_from の実装次第で変はる:
+        # - RecordInvalid を 422 にしてるなら 422
+        # - 未処理なら 500
+        expect(response.status).to be_in([422, 500])
+      end
+    end
+  end
 end
@@ -5,4 +5,11 @@ module TestRecords
                 role: 'member',
                 banned: false)
  end
+
+  def create_admin_user!
+    User.create!(name: 'spec admin',
+                 inheritance_code: SecureRandom.hex(16),
+                 role: 'admin',
+                 banned: false)
+  end
 end
@@ -0,0 +1,34 @@
+require 'rails_helper'
+
+
+RSpec.describe 'post_similarity:calc' do
+  include RakeTaskHelper
+
+  it 'calls Similarity::Calc with Post and :tags' do
+    # 必要最低限のデータ
+    t1 = Tag.create!(name: "t1")
+    t2 = Tag.create!(name: "t2")
+    t3 = Tag.create!(name: "t3")
+
+    p1 = Post.create!(url: "https://example.com/1")
+    p2 = Post.create!(url: "https://example.com/2")
+    p3 = Post.create!(url: "https://example.com/3")
+
+    # kept スコープが絡むなら、PostTag がデフォで kept になる前提
+    PostTag.create!(post: p1, tag: t1)
+    PostTag.create!(post: p1, tag: t2)
+
+    PostTag.create!(post: p2, tag: t1)
+    PostTag.create!(post: p2, tag: t3)
+
+    PostTag.create!(post: p3, tag: t3)
+
+    expect { run_rake_task("post_similarity:calc") }
+      .to change { PostSimilarity.count }.from(0)
+
+    ps = PostSimilarity.find_by!(post_id: p1.id, target_post_id: p2.id)
+    ps_rev = PostSimilarity.find_by!(post_id: p2.id, target_post_id: p1.id)
+    expect(ps_rev.cos).to eq(ps.cos)
+  end
+end
+
@@ -0,0 +1,34 @@
+require 'rails_helper'
+
+
+RSpec.describe 'tag_similarity:calc' do
+  include RakeTaskHelper
+
+  it 'calls Similarity::Calc with Tag and :posts' do
+    # 必要最低限のデータ
+    t1 = Tag.create!(name: "t1")
+    t2 = Tag.create!(name: "t2")
+    t3 = Tag.create!(name: "t3")
+
+    p1 = Post.create!(url: "https://example.com/1")
+    p2 = Post.create!(url: "https://example.com/2")
+    p3 = Post.create!(url: "https://example.com/3")
+
+    # kept スコープが絡むなら、PostTag がデフォで kept になる前提
+    PostTag.create!(post: p1, tag: t1)
+    PostTag.create!(post: p1, tag: t2)
+
+    PostTag.create!(post: p2, tag: t1)
+    PostTag.create!(post: p2, tag: t3)
+
+    PostTag.create!(post: p3, tag: t3)
+
+    expect { run_rake_task("tag_similarity:calc") }
+      .to change { TagSimilarity.count }.from(0)
+
+    ps = TagSimilarity.find_by!(tag_id: t1.id, target_tag_id: t2.id)
+    ps_rev = TagSimilarity.find_by!(tag_id: t2.id, target_tag_id: t1.id)
+    expect(ps_rev.cos).to eq(ps.cos)
+  end
+end
+