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

backend/app/controllers/tags_controller.rb

@@ -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

backend/app/models/post.rb

@@ -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!

backend/app/models/post_similarity.rb

@@ -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

backend/app/models/tag_similarity.rb

@@ -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

backend/app/services/similarity/calc.rb

@@ -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