优化时间统计问题

app/services/subjects/shixun_used_info_service.rb

@@ -10,20 +10,23 @@ class Subjects::ShixunUsedInfoService < ApplicationService
     stages.each do |stage|
       position = stage.position
       shixuns = stage.shixuns.includes(myshixuns: :games, homework_commons: :course)
-      shixuns.each_with_index do |shixun, index|
-        stage = "#{position}-#{index+1}"
-        name = shixun.name
-        myshixuns = shixun.myshixuns
-        challenge_count = shixun.challenges_count
-        course_count = shixun.homework_commons.map{|hc| hc.course_id}.uniq.size
-        school_count = shixun.homework_commons.map{|hc| hc.course&.school_id}.uniq.size
-        used_count = shixun.myshixuns_count
-        passed_count = myshixuns.select{|m| m.status == 1}.size
-        evaluate_count = myshixuns.map{|m| m.output_times }.sum
-        passed_ave_time = passed_count > 0 ? myshixuns.map{|m| m.total_cost_time}.sum : 0
-        shixun_infos << {stage: stage, name: name, challenge_count: challenge_count, course_count: course_count,
-                         school_count: school_count, used_count: used_count, passed_count: passed_count,
-                         evaluate_count: evaluate_count, passed_ave_time: passed_ave_time, shixun_id: shixun.id}
+      shixuns.find_in_batches(batch_size: 1000) do |s|
+        Parallel.each_with_index(s, in_processes: 2) do |shixun, index|
+          stage = "#{position}-#{index+1}"
+          name = shixun.name
+          myshixuns = shixun.myshixuns
+          challenge_count = shixun.challenges_count
+          course_count = shixun.homework_commons.map{|hc| hc.course_id}.uniq.size
+          school_count = shixun.homework_commons.map{|hc| hc.course&.school_id}.uniq.size
+          used_count = shixun.myshixuns_count
+          passed_count = myshixuns.select{|m| m.status == 1}.size
+          evaluate_count = myshixuns.map{|m| m.output_times }.sum
+          passed_ave_time = passed_count > 0 ? myshixuns.map{|m| m.total_cost_time}.sum : 0
+          shixun_infos << {stage: stage, name: name, challenge_count: challenge_count, course_count: course_count,
+                           school_count: school_count, used_count: used_count, passed_count: passed_count,
+                           evaluate_count: evaluate_count, passed_ave_time: passed_ave_time, shixun_id: shixun.id}
+
+        end
       end
     end
     shixun_infos

lib/tasks/statistic_subject_info.rake

@@ -9,25 +9,27 @@ namespace :subjects do
     buffer_size = 0
     column_value = "subject_id, study_count, course_study_count, initiative_study, passed_count, course_used_count, " +
         "school_used_count, created_at, updated_at"
-    subjects.find_each do |subject|
-      puts("---------------------data_statistic: #{subject.id}")
-      Rails.logger.info("---------------------data_statistic: #{subject.id}")
-      data = Subjects::DataStatisticService.new(subject)
-      study_count = data.study_count
-      next if study_count == 0
-      course_study_count = data.course_study_count
-      initiative_study = study_count - course_study_count
-      str += ", " unless str.empty?
-      str += ("(#{subject.id}, #{study_count}, #{course_study_count}, #{initiative_study}, " +
-          "#{data.passed_count}, #{data.course_used_count}, #{data.school_used_count}, " +
-          "'#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}', '#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}')")
-      buffer_size += 1
-      if buffer_size == 1000
-        sql = "REPLACE INTO subject_records(#{column_value}) VALUES #{str}"
-        puts sql
-        ActiveRecord::Base.connection.execute sql
-        str = ""
-        buffer_size = 0
+    subjects.find_in_batches(batch_size: 50) do |s|
+      Parallel.each(s, in_processes: 4) do |subject|
+        puts("---------------------data_statistic: #{subject.id}")
+        Rails.logger.info("---------------------data_statistic: #{subject.id}")
+        data = Subjects::DataStatisticService.new(subject)
+        study_count = data.study_count
+        next if study_count == 0
+        course_study_count = data.course_study_count
+        initiative_study = study_count - course_study_count
+        str += ", " unless str.empty?
+        str += ("(#{subject.id}, #{study_count}, #{course_study_count}, #{initiative_study}, " +
+            "#{data.passed_count}, #{data.course_used_count}, #{data.school_used_count}, " +
+            "'#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}', '#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}')")
+        buffer_size += 1
+        if buffer_size == 1000
+          sql = "REPLACE INTO subject_records(#{column_value}) VALUES #{str}"
+          puts sql
+          ActiveRecord::Base.connection.execute sql
+          str = ""
+          buffer_size = 0
+        end
       end
     end
     if buffer_size > 0
@@ -47,23 +49,26 @@ namespace :subjects do
     buffer_size = 0
     column_value = "subject_id, school_id, school_name, course_count, student_count, choice_shixun_num, " +
         "choice_shixun_frequency, created_at, updated_at"
-    subjects.find_each do |subject|
-      puts("---------------------course_info_statistic: #{subject.id}")
-      Rails.logger.info("---------------------course_info_statistic: #{subject.id}")
-      data = Subjects::CourseUsedInfoService.call(subject)
-      Parallel.map(data) do |key|
-        next if key[:school_id].nil?
-        str += ", " unless str.empty?
-        str += ("(#{subject.id}, #{key[:school_id]}, '#{key[:school_name]}', #{key[:course_count]}, " +
-            "#{key[:student_count]}, #{key[:choice_shixun_num]}, #{key[:choice_shixun_frequency]}, " +
-            "'#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}', '#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}')")
-        buffer_size += 1
-        if buffer_size == 1000
-          sql = "REPLACE INTO subject_course_records(#{column_value}) VALUES #{str}"
-          puts sql
-          ActiveRecord::Base.connection.execute sql
-          str = ""
-          buffer_size = 0
+
+    subjects.find_in_batches(batch_size: 50) do |s|
+      Parallel.each(s, in_processes: 4) do |subject|
+        puts("---------------------course_info_statistic: #{subject.id}")
+        Rails.logger.info("---------------------course_info_statistic: #{subject.id}")
+        data = Subjects::CourseUsedInfoService.call(subject)
+        Parallel.map(data) do |key|
+          next if key[:school_id].nil?
+          str += ", " unless str.empty?
+          str += ("(#{subject.id}, #{key[:school_id]}, '#{key[:school_name]}', #{key[:course_count]}, " +
+              "#{key[:student_count]}, #{key[:choice_shixun_num]}, #{key[:choice_shixun_frequency]}, " +
+              "'#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}', '#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}')")
+          buffer_size += 1
+          if buffer_size == 1000
+            sql = "REPLACE INTO subject_course_records(#{column_value}) VALUES #{str}"
+            puts sql
+            ActiveRecord::Base.connection.execute sql
+            str = ""
+            buffer_size = 0
+          end
         end
       end
     end
@@ -84,24 +89,26 @@ namespace :subjects do
     buffer_size = 0
     column_value = "subject_id, shixun_id, stage, shixun_name, challenge_count, course_count, " +
         "school_count, used_count, passed_count, evaluate_count, passed_ave_time, created_at, updated_at"
-    subjects.find_each(batch_size: 100) do |subject|
-      puts("---------------------shixun_info_statistic: #{subject.id}")
-      Rails.logger.info("---------------------shixun_info_statistic: #{subject.id}")
-      data = Subjects::ShixunUsedInfoService.call(subject)
-      data.each do |key|
-        next if key[:shixun_id].nil?
-        str += ", " unless str.empty?
-        str += ("(#{subject.id}, #{key[:shixun_id]}, '#{key[:stage]}', '#{key[:name]}', #{key[:challenge_count]}, " +
-            "#{key[:course_count]}, #{key[:school_count]}, #{key[:used_count]}, #{key[:passed_count]}, " +
-            "#{key[:evaluate_count]}, #{key[:passed_ave_time]}, " +
-            "'#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}', '#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}')")
-        buffer_size += 1
-        if buffer_size == 1000
-          sql = "REPLACE INTO subject_shixun_infos(#{column_value}) VALUES #{str}"
-          puts sql
-          ActiveRecord::Base.connection.execute sql
-          str = ""
-          buffer_size = 0
+    subjects.find_each(batch_size: 50) do |s|
+      Parallel.each(s, in_processes: 4) do |subject|
+        puts("---------------------shixun_info_statistic: #{subject.id}")
+        Rails.logger.info("---------------------shixun_info_statistic: #{subject.id}")
+        data = Subjects::ShixunUsedInfoService.call(subject)
+        data.each do |key|
+          next if key[:shixun_id].nil?
+          str += ", " unless str.empty?
+          str += ("(#{subject.id}, #{key[:shixun_id]}, '#{key[:stage]}', '#{key[:name]}', #{key[:challenge_count]}, " +
+              "#{key[:course_count]}, #{key[:school_count]}, #{key[:used_count]}, #{key[:passed_count]}, " +
+              "#{key[:evaluate_count]}, #{key[:passed_ave_time]}, " +
+              "'#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}', '#{Time.now.strftime('%Y-%m-%d %H:%M:%S')}')")
+          buffer_size += 1
+          if buffer_size == 1000
+            sql = "REPLACE INTO subject_shixun_infos(#{column_value}) VALUES #{str}"
+            puts sql
+            ActiveRecord::Base.connection.execute sql
+            str = ""
+            buffer_size = 0
+          end
         end
       end
     end