ex_twitter 0.1.1 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 5a6285b02440ee496486c9d354b6021886a2fd96
-  data.tar.gz: 0e1face89fa2dcdfe61f66f225ce53860e0bf527
+  metadata.gz: fae8a4c6b259764373fb52ca806eb66f8c38a472
+  data.tar.gz: b2c3d986eb8ed42504f00ae2841b8dab8e302436
 SHA512:
-  metadata.gz: 1eb8f2e80cbfd46564f03af107ca23b7c3eea9aba4fe4b673ab346f87bd323dc0459453437df59bd89f4f65307c36c53c5660c9195da46b0d062407460e446d9
-  data.tar.gz: 125f18acc4477cc0b8a1013c276004b1b87a75f88bd52fb6d5f28a1194fe0519a3c7c7bd9de01c193c8b1802bca5fc656e1e6fa0b222893032c9cc3a8c64142e
+  metadata.gz: b7c77240b80287e4d73935813b8185658eeac95b7ccaa93540a72ac01ec3b2cac751c12098cda953ff4b2bd0fc8cbcbb8d8f87f078be941845b6982d1e02fdaf
+  data.tar.gz: 28a1dc726814e7e503b6d179945ef214b3b025e6cfe0650c56e54855f0f46c057d11e68a10b4e30ef2edd74aeb07c5fb4d6c6f5b69af2e0ce3fff101d19578ce

data/README.md

@@ -50,11 +50,3 @@ end
 ```
 client.user_timeline
 ```
-
-## Memo
-
-```
-be rake build
-be rake release
-```
-

data/ex_twitter.gemspec

@@ -21,5 +21,5 @@ Gem::Specification.new do |spec|
   spec.require_paths = %w[lib]
   spec.summary = spec.description
   spec.test_files = Dir.glob('spec/**/*')
-  spec.version = '0.1.1'
+  spec.version = '0.2.0'
 end

data/lib/ex_twitter/client.rb

@@ -0,0 +1,194 @@
+require 'active_support'
+require 'active_support/cache'
+require 'active_support/core_ext/string'
+
+require 'ex_twitter/log_subscriber'
+require 'ex_twitter/utils'
+require 'ex_twitter/existing_api'
+require 'ex_twitter/new_api'
+
+require 'twitter'
+require 'hashie'
+require 'parallel'
+
+module ExTwitter
+  class Client < Twitter::REST::Client
+    def initialize(options = {})
+      @cache = ActiveSupport::Cache::FileStore.new(File.join('tmp', 'api_cache'))
+      @call_count = 0
+
+      ExTwitter::LogSubscriber.attach_to :ex_twitter
+      ExTwitter::LogSubscriber.attach_to :active_support
+      @@logger = @logger =
+        if options[:logger]
+          options.delete(:logger)
+        else
+          Dir.mkdir('log') unless File.exists?('log')
+          Logger.new('log/ex_twitter.log')
+        end
+
+      super
+    end
+
+    def self.logger
+      @@logger
+    end
+
+    attr_accessor :call_count
+    attr_reader :cache, :authenticated_user, :logger
+
+    INDENT = 4
+
+    include ExTwitter::Utils
+
+    alias :old_friendship? :friendship?
+    alias :old_user? :user?
+    alias :old_user :user
+    alias :old_friend_ids :friend_ids
+    alias :old_follower_ids :follower_ids
+    alias :old_friends :friends
+    alias :old_followers :followers
+    alias :old_users :users
+    alias :old_home_timeline :home_timeline
+    alias :old_user_timeline :user_timeline
+    alias :old_mentions_timeline :mentions_timeline
+    alias :old_favorites :favorites
+    alias :old_search :search
+
+    include ExTwitter::ExistingApi
+    include ExTwitter::NewApi
+
+    def usage_stats_wday_series_data(times)
+      wday_count = times.each_with_object((0..6).map { |n| [n, 0] }.to_h) do |time, memo|
+        memo[time.wday] += 1
+      end
+      wday_count.map { |k, v| [I18n.t('date.abbr_day_names')[k], v] }.map do |key, value|
+        {name: key, y: value, drilldown: key}
+      end
+    end
+
+    def usage_stats_wday_drilldown_series(times)
+      hour_count =
+        (0..6).each_with_object((0..6).map { |n| [n, nil] }.to_h) do |wday, wday_memo|
+          wday_memo[wday] =
+            times.select { |t| t.wday == wday }.map { |t| t.hour }.each_with_object((0..23).map { |n| [n, 0] }.to_h) do |hour, hour_memo|
+              hour_memo[hour] += 1
+            end
+        end
+      hour_count.map { |k, v| [I18n.t('date.abbr_day_names')[k], v] }.map do |key, value|
+        {name: key, id: key, data: value.to_a.map{|a| [a[0].to_s, a[1]] }}
+      end
+    end
+
+    def usage_stats_hour_series_data(times)
+      hour_count = times.each_with_object((0..23).map { |n| [n, 0] }.to_h) do |time, memo|
+        memo[time.hour] += 1
+      end
+      hour_count.map do |key, value|
+        {name: key.to_s, y: value, drilldown: key.to_s}
+      end
+    end
+
+    def usage_stats_hour_drilldown_series(times)
+      wday_count =
+        (0..23).each_with_object((0..23).map { |n| [n, nil] }.to_h) do |hour, hour_memo|
+          hour_memo[hour] =
+            times.select { |t| t.hour == hour }.map { |t| t.wday }.each_with_object((0..6).map { |n| [n, 0] }.to_h) do |wday, wday_memo|
+              wday_memo[wday] += 1
+            end
+        end
+      wday_count.map do |key, value|
+        {name: key.to_s, id: key.to_s, data: value.to_a.map{|a| [I18n.t('date.abbr_day_names')[a[0]], a[1]] }}
+      end
+    end
+
+    def twitter_addiction_series(times)
+      five_mins = 5.minutes
+      wday_expended_seconds =
+        (0..6).each_with_object((0..6).map { |n| [n, nil] }.to_h) do |wday, wday_memo|
+          target_times = times.select { |t| t.wday == wday }
+          wday_memo[wday] = target_times.empty? ? nil : target_times.each_cons(2).map {|a, b| (a - b) < five_mins ? a - b : five_mins }.sum
+        end
+      days = times.map{|t| t.to_date.to_s(:long) }.uniq.size
+      weeks = (days > 7) ? days / 7.0 : 1.0
+      wday_expended_seconds.map { |k, v| [I18n.t('date.abbr_day_names')[k], (v.nil? ? nil : v / weeks / 60)] }.map do |key, value|
+        {name: key, y: value}
+      end
+    end
+
+    def usage_stats(user, options = {})
+      n_days_ago = options.has_key?(:days) ? options[:days].days.ago : 100.years.ago
+      tweets = options.has_key?(:tweets) ? options.delete(:tweets) : user_timeline(user)
+      times =
+        # TODO Use user specific time zone
+        tweets.map { |t| ActiveSupport::TimeZone['Tokyo'].parse(t.created_at.to_s) }.
+          select { |t| t > n_days_ago }
+      [
+        usage_stats_wday_series_data(times),
+        usage_stats_wday_drilldown_series(times),
+        usage_stats_hour_series_data(times),
+        usage_stats_hour_drilldown_series(times),
+        twitter_addiction_series(times)
+      ]
+    end
+
+
+    def calc_scores_from_users(users, options)
+      min = options.has_key?(:min) ? options[:min] : 0
+      max = options.has_key?(:max) ? options[:max] : 1000
+      users.each_with_object(Hash.new(0)) { |u, memo| memo[u.id] += 1 }.
+        select { |_k, v| min <= v && v <= max }.
+        sort_by { |_, v| -v }.to_h
+    end
+
+    def calc_scores_from_tweets(tweets, options = {})
+      calc_scores_from_users(tweets.map { |t| t.user }, options)
+    end
+
+    def select_favoriting_from_favs(favs, options = {})
+      return [] if favs.empty?
+      uids = calc_scores_from_tweets(favs)
+      result = uids.map { |uid, score| f = favs.
+        find { |f| f.user.id.to_i == uid.to_i }; Array.new(score, f) }.flatten.map { |f| f.user }
+      (options.has_key?(:uniq) && !options[:uniq]) ? result : result.uniq { |u| u.id }
+    end
+
+    def favoriting(user, options= {})
+      favs = options.has_key?(:favorites) ? options.delete(:favorites) : favorites(user, options)
+      select_favoriting_from_favs(favs, options)
+    rescue => e
+      logger.warn "#{__method__} #{user.inspect} #{e.class} #{e.message}"
+      raise e
+    end
+
+    def favorited_by(user)
+    end
+
+    def close_friends(_uid, options = {})
+      min = options.has_key?(:min) ? options[:min] : 0
+      max = options.has_key?(:max) ? options[:max] : 1000
+      uid_i = _uid.to_i
+      _replying = options.has_key?(:replying) ? options.delete(:replying) : replying(uid_i, options)
+      _replied = options.has_key?(:replied) ? options.delete(:replied) : replied(uid_i, options)
+      _favoriting = options.has_key?(:favoriting) ? options.delete(:favoriting) : favoriting(uid_i, options)
+
+      min_max = {min: min, max: max}
+      _users = _replying + _replied + _favoriting
+      return [] if _users.empty?
+
+      scores = calc_scores_from_users(_users, min_max)
+      replying_scores = calc_scores_from_users(_replying, min_max)
+      replied_scores = calc_scores_from_users(_replied, min_max)
+      favoriting_scores = calc_scores_from_users(_favoriting, min_max)
+
+      scores.keys.map { |uid| _users.find { |u| u.id.to_i == uid.to_i } }.
+        map do |u|
+        u[:score] = scores[u.id]
+        u[:replying_score] = replying_scores[u.id]
+        u[:replied_score] = replied_scores[u.id]
+        u[:favoriting_score] = favoriting_scores[u.id]
+        u
+      end
+    end
+  end
+end

data/lib/ex_twitter/existing_api.rb

@@ -0,0 +1,122 @@
+module ExTwitter
+  module ExistingApi
+    def friendship?(*args)
+      options = args.extract_options!
+      fetch_cache_or_call_api(__method__, args) {
+        call_old_method("old_#{__method__}", *args, options)
+      }
+    end
+
+    def user?(*args)
+      options = args.extract_options!
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        call_old_method("old_#{__method__}", args[0], options)
+      }
+    end
+
+    def user(*args)
+      options = args.extract_options!
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        call_old_method("old_#{__method__}", args[0], options)
+      }
+    end
+
+    def friend_ids(*args)
+      options = {count: 5000, cursor: -1}.merge(args.extract_options!)
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_cursor("old_#{__method__}", *args, options)
+      }
+    end
+
+    def follower_ids(*args)
+      options = {count: 5000, cursor: -1}.merge(args.extract_options!)
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_cursor("old_#{__method__}", *args, options)
+      }
+    end
+
+    # specify reduce: false to use tweet for inactive_*
+    def friends(*args)
+      options = {count: 200, include_user_entities: true, cursor: -1}.merge(args.extract_options!)
+      options[:reduce] = false unless options.has_key?(:reduce)
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_cursor("old_#{__method__}", *args, options)
+      }
+    end
+
+    # specify reduce: false to use tweet for inactive_*
+    def followers(*args)
+      options = {count: 200, include_user_entities: true, cursor: -1}.merge(args.extract_options!)
+      options[:reduce] = false unless options.has_key?(:reduce)
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_cursor("old_#{__method__}", *args, options)
+      }
+    end
+
+    # use compact, not use sort and uniq
+    # specify reduce: false to use tweet for inactive_*
+    def users(*args)
+      options = args.extract_options!
+      options[:reduce] = false
+      users_per_workers = args.first.compact.each_slice(100).to_a
+      processed_users = []
+
+      Parallel.each_with_index(users_per_workers, in_threads: [users_per_workers.size, 10].min) do |users_per_worker, i|
+        _users = fetch_cache_or_call_api(__method__, users_per_worker, options) {
+          call_old_method("old_#{__method__}", users_per_worker, options)
+        }
+
+        processed_users << {i: i, users: _users}
+      end
+
+      processed_users.sort_by{|p| p[:i] }.map{|p| p[:users] }.flatten.compact
+    rescue => e # debug
+      logger.warn "#{__method__}: #{args.inspect} #{e.class} #{e.message}"
+      raise e
+    end
+
+    def home_timeline(*args)
+      options = {count: 200, include_rts: true, call_limit: 3}.merge(args.extract_options!)
+      fetch_cache_or_call_api(__method__, user.screen_name, options) {
+        collect_with_max_id("old_#{__method__}", options)
+      }
+    end
+
+    def user_timeline(*args)
+      options = {count: 200, include_rts: true, call_limit: 3}.merge(args.extract_options!)
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_max_id("old_#{__method__}", *args, options)
+      }
+    end
+
+    def mentions_timeline(*args)
+      options = {count: 200, include_rts: true, call_limit: 1}.merge(args.extract_options!)
+      fetch_cache_or_call_api(__method__, user.screen_name, options) {
+        collect_with_max_id("old_#{__method__}", options)
+      }
+    end
+
+    def favorites(*args)
+      options = {count: 100, call_count: 1}.merge(args.extract_options!)
+      args[0] = verify_credentials(skip_status: true).id if args.empty?
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_max_id("old_#{__method__}", *args, options)
+      }
+    end
+
+    def search(*args)
+      options = {count: 100, result_type: :recent, call_limit: 1}.merge(args.extract_options!)
+      options[:reduce] = false
+      fetch_cache_or_call_api(__method__, args[0], options) {
+        collect_with_max_id("old_#{__method__}", *args, options) { |response| response.attrs[:statuses] }
+      }
+    end
+  end
+end

data/lib/ex_twitter/log_subscriber.rb

@@ -0,0 +1,79 @@
+require 'active_support'
+require 'active_support/core_ext'
+
+
+module ExTwitter
+  class LogSubscriber < ActiveSupport::LogSubscriber
+
+    def initialize
+      super
+      @odd = false
+    end
+
+    def cache_any(event)
+      return unless logger.debug?
+
+      payload = event.payload
+      name  = "#{payload.delete(:name)} (#{event.duration.round(1)}ms)"
+      name = colorize_payload_name(name, payload[:name], AS: true)
+      debug { "#{name}: #{(payload.inspect)}" }
+    end
+
+    %w(read write fetch_hit generate delete exist?).each do |operation|
+      class_eval <<-METHOD, __FILE__, __LINE__ + 1
+          def cache_#{operation}(event)
+            event.payload[:name] = '#{operation}'
+            cache_any(event)
+          end
+      METHOD
+    end
+
+    def call(event)
+      return unless logger.debug?
+
+      payload = event.payload
+      name = "#{payload.delete(:operation)} (#{event.duration.round(1)}ms)"
+
+      name = colorize_payload_name(name, payload[:name])
+      # sql  = color(sql, sql_color(sql), true)
+
+      key = payload.delete(:key)
+      debug { "#{name}: #{key} #{(payload.inspect)}" }
+    end
+
+    private
+
+    def colorize_payload_name(name, payload_name, options = {})
+      if options[:AS]
+        color(name, MAGENTA, true)
+      else
+        color(name, CYAN, true)
+      end
+    end
+
+    def sql_color(sql)
+      case sql
+        when /\A\s*rollback/mi
+          RED
+        when /select .*for update/mi, /\A\s*lock/mi
+          WHITE
+        when /\A\s*select/i
+          BLUE
+        when /\A\s*insert/i
+          GREEN
+        when /\A\s*update/i
+          YELLOW
+        when /\A\s*delete/i
+          RED
+        when /transaction\s*\Z/i
+          CYAN
+        else
+          MAGENTA
+      end
+    end
+
+    def logger
+      ExTwitter::Client.logger
+    end
+  end
+end

data/lib/ex_twitter/new_api.rb

@@ -0,0 +1,238 @@
+module ExTwitter
+  module NewApi
+    def friends_parallelly(*args)
+      options = {super_operation: __method__}.merge(args.extract_options!)
+      _friend_ids = friend_ids(*(args + [options]))
+      users(_friend_ids.map { |id| id.to_i }, options)
+    end
+
+    def followers_parallelly(*args)
+      options = {super_operation: __method__}.merge(args.extract_options!)
+      _follower_ids = follower_ids(*(args + [options]))
+      users(_follower_ids.map { |id| id.to_i }, options)
+    end
+
+    def _fetch_parallelly(signatures) # [{method: :friends, args: ['ts_3156', ...], {...}]
+      result = Array.new(signatures.size)
+
+      Parallel.each_with_index(signatures, in_threads: result.size) do |signature, i|
+        result[i] = send(signature[:method], *signature[:args])
+      end
+
+      result
+    end
+
+    def friends_and_followers(*args)
+      _fetch_parallelly(
+        [
+          {method: :friends_parallelly, args: args},
+          {method: :followers_parallelly, args: args}])
+    end
+
+    def friends_followers_and_statuses(*args)
+      _fetch_parallelly(
+        [
+          {method: :friends_parallelly, args: args},
+          {method: :followers_parallelly, args: args},
+          {method: :user_timeline, args: args}])
+    end
+
+    def one_sided_following(me)
+      if uid_or_screen_name?(me)
+        # TODO use friends_and_followers
+        friends_parallelly(me).to_a - followers_parallelly(me).to_a
+      elsif me.respond_to?(:friends) && me.respond_to?(:followers)
+        me.friends.to_a - me.followers.to_a
+      else
+        raise
+      end
+    end
+
+    def one_sided_followers(me)
+      if uid_or_screen_name?(me)
+        # TODO use friends_and_followers
+        followers_parallelly(me).to_a - friends_parallelly(me).to_a
+      elsif me.respond_to?(:friends) && me.respond_to?(:followers)
+        me.followers.to_a - me.friends.to_a
+      else
+        raise
+      end
+    end
+
+    def mutual_friends(me)
+      if uid_or_screen_name?(me)
+        # TODO use friends_and_followers
+        friends_parallelly(me).to_a & followers_parallelly(me).to_a
+      elsif me.respond_to?(:friends) && me.respond_to?(:followers)
+        me.friends.to_a & me.followers.to_a
+      else
+        raise
+      end
+    end
+
+    def common_friends(me, you)
+      if uid_or_screen_name?(me) && uid_or_screen_name?(you)
+        friends_parallelly(me).to_a & friends_parallelly(you).to_a
+      elsif me.respond_to?(:friends) && you.respond_to?(:friends)
+        me.friends.to_a & you.friends.to_a
+      else
+        raise
+      end
+    end
+
+    def common_followers(me, you)
+      if uid_or_screen_name?(me) && uid_or_screen_name?(you)
+        followers_parallelly(me).to_a & followers_parallelly(you).to_a
+      elsif me.respond_to?(:followers) && you.respond_to?(:followers)
+        me.followers.to_a & you.followers.to_a
+      else
+        raise
+      end
+    end
+
+    def removing(pre_me, cur_me)
+      if uid_or_screen_name?(pre_me) && uid_or_screen_name?(cur_me)
+        friends_parallelly(pre_me).to_a - friends_parallelly(cur_me).to_a
+      elsif pre_me.respond_to?(:friends) && cur_me.respond_to?(:friends)
+        pre_me.friends.to_a - cur_me.friends.to_a
+      else
+        raise
+      end
+    end
+
+    def removed(pre_me, cur_me)
+      if uid_or_screen_name?(pre_me) && uid_or_screen_name?(cur_me)
+        followers_parallelly(pre_me).to_a - followers_parallelly(cur_me).to_a
+      elsif pre_me.respond_to?(:followers) && cur_me.respond_to?(:followers)
+        pre_me.followers.to_a - cur_me.followers.to_a
+      else
+        raise
+      end
+    end
+
+    def _extract_screen_names(tweets, options = {})
+      result = tweets.map do |t|
+        $1 if t.text =~ /^(?:\.)?@(\w+)( |\W)/ # include statuses starts with .
+      end.compact
+      (options.has_key?(:uniq) && !options[:uniq]) ? result : result.uniq
+    end
+
+    # users which specified user is replying
+    # in_reply_to_user_id and in_reply_to_status_id is not used because of distinguishing mentions from replies
+    def replying(*args)
+      options = args.extract_options!
+      tweets =
+        if args.empty?
+          user_timeline(user.screen_name, options)
+        elsif uid_or_screen_name?(args[0])
+          user_timeline(args[0], options)
+        else
+          raise
+        end
+      screen_names = _extract_screen_names(tweets, options)
+      users(screen_names, {super_operation: __method__}.merge(options))
+    rescue Twitter::Error::NotFound => e
+      e.message == 'No user matches for specified terms.' ? [] : (raise e)
+    rescue => e
+      logger.warn "#{__method__} #{args.inspect} #{e.class} #{e.message}"
+      raise e
+    end
+
+    def _extract_uids(tweets, options)
+      result = tweets.map do |t|
+        t.user.id.to_i if t.text =~ /^(?:\.)?@(\w+)( |\W)/ # include statuses starts with .
+      end.compact
+      (options.has_key?(:uniq) && !options[:uniq]) ? result : result.uniq
+    end
+
+    def _extract_users(tweets, uids, options = {})
+      uids.map { |u| tweets.find { |t| t.user.id.to_i == u.to_i } }.map { |t| t.user }
+    end
+
+    # users which specified user is replied
+    # when user is login you had better to call mentions_timeline
+    def replied(*args)
+      options = args.extract_options!
+
+      if args.empty? || (uid_or_screen_name?(args[0]) && authenticating_user?(args[0]))
+        mentions_timeline.uniq { |m| m.user.id }.map { |m| m.user }
+      else
+        searched_result = search('@' + user(args[0]).screen_name, options)
+        uids = _extract_uids(searched_result, options)
+        _extract_users(searched_result, uids, options)
+      end
+    end
+
+    def _extract_inactive_users(users, options = {})
+      authorized = options.delete(:authorized)
+      two_weeks_ago = 2.weeks.ago.to_i
+      users.select do |u|
+        if authorized
+          (Time.parse(u.status.created_at).to_i < two_weeks_ago) rescue false
+        else
+          false
+        end
+      end
+    end
+
+    def inactive_friends(user = nil)
+      if user.blank?
+        _extract_inactive_users(friends_parallelly, authorized: true)
+      elsif uid_or_screen_name?(user)
+        authorized = authenticating_user?(user) || authorized_user?(user)
+        _extract_inactive_users(friends_parallelly(user), authorized: authorized)
+      elsif user.respond_to?(:friends)
+        authorized = authenticating_user?(user.uid.to_i) || authorized_user?(user.uid.to_i)
+        _extract_inactive_users(user.friends, authorized: authorized)
+      else
+        raise
+      end
+    end
+
+    def inactive_followers(user = nil)
+      if user.blank?
+        _extract_inactive_users(followers_parallelly, authorized: true)
+      elsif uid_or_screen_name?(user)
+        authorized = authenticating_user?(user) || authorized_user?(user)
+        _extract_inactive_users(followers_parallelly(user), authorized: authorized)
+      elsif user.respond_to?(:followers)
+        authorized = authenticating_user?(user.uid.to_i) || authorized_user?(user.uid.to_i)
+        _extract_inactive_users(user.followers, authorized: authorized)
+      else
+        raise
+      end
+    end
+
+    def clusters_belong_to(text)
+      return [] if text.blank?
+
+      exclude_words = JSON.parse(File.read(Rails.configuration.x.constants['cluster_bad_words_path']))
+      special_words = JSON.parse(File.read(Rails.configuration.x.constants['cluster_good_words_path']))
+
+      # クラスタ用の単語の出現回数を記録
+      cluster_word_counter =
+        special_words.map { |sw| [sw, text.scan(sw)] }
+          .delete_if { |item| item[1].empty? }
+          .each_with_object(Hash.new(1)) { |item, memo| memo[item[0]] = item[1].size }
+
+      # 同一文字種の繰り返しを見付ける。漢字の繰り返し、ひらがなの繰り返し、カタカナの繰り返し、など
+      text.scan(/[一-龠〆ヵヶ々]+|[ぁ-んー～]+|[ァ-ヴー～]+|[ａ-ｚＡ-Ｚ０-９]+|[、。！!？?]+/).
+
+        # 複数回繰り返される文字を除去
+        map { |w| w.remove /[？！?!。、ｗ]|(ー{2,})/ }.
+
+        # 文字数の少なすぎる単語、ひらがなだけの単語、除外単語を除去する
+        delete_if { |w| w.length <= 1 || (w.length <= 2 && w =~ /^[ぁ-んー～]+$/) || exclude_words.include?(w) }.
+
+        # 出現回数を記録
+        each { |w| cluster_word_counter[w] += 1 }
+
+      # 複数個以上見付かった単語のみを残し、出現頻度順にソート
+      cluster_word_counter.select { |_, v| v > 3 }.sort_by { |_, v| -v }.to_h
+    end
+
+    def clusters_assigned_to
+      raise NotImplementedError.new
+    end
+  end
+end