接入百度统计部分数据，按天存储

app/controllers/admins/dashboards_controller.rb

@@ -28,6 +28,41 @@ class Admins::DashboardsController < Admins::BaseController
     @day_new_project_count = Project.where(created_on: today).count
     @weekly_new_project_count = Project.where(created_on: current_week).count
     @month_new_project_count = Project.where(created_on: current_month).count
+
+    # 总的平台用户数
+    # 总的平台项目数
+    # 总的平台组织数
+    # 总的平台Issue数、评论数、PR数、Commit数
+    @user_count = User.count
+    @project_count = Project.count
+    @organization_count = Organization.count
+    @issue_count = Issue.count
+    @comment_count = Journal.count
+    @pr_count = PullRequest.count
+    @commit_count = CommitLog.count
+
+    @subject_name = ["用户数", "项目数", "组织数", "Issue数", "Issue评论数", "PR数", "Commit数"]
+    @subject_icon = ["fa-user","fa-git", "fa-sitemap", "fa-warning", "fa-comments", "fa-share-alt", "fa-upload"]
+    @subject_data = [@user_count, @project_count, @organization_count, @issue_count, @comment_count, @pr_count, @commit_count]
+
+
+    tongji_service = Baidu::TongjiService.new
+    @access_token = tongji_service.access_token
+    Rails.logger.info "baidu_tongji_auth access_token ===== #{@access_token}"
+    # @overview_data = tongji_service.api_overview
+    last_date = DailyPlatformStatistic.order(:date).last
+    start_date = last_date.date
+    end_date = Time.now
+    if @access_token.present?
+      @overview_data = tongji_service.overview_batch_add(start_date, end_date)
+      tongji_service.source_from_batch_add(start_date, end_date)
+    end
+
+    @current_week_statistic = DailyPlatformStatistic.where(date: current_week)
+    @pre_week_statistic = DailyPlatformStatistic.where(date: pre_week)
+
+
+
   end
 
   def month_active_user
@@ -42,6 +77,19 @@ class Admins::DashboardsController < Admins::BaseController
     render_ok(data: data)
   end
 
+  def baidu_tongji
+    tongji_service = Baidu::TongjiService.new
+    redirect_to tongji_service.code_url
+  end
+
+  def baidu_tongji_auth
+    if params[:code].present?
+      tongji_service = Baidu::TongjiService.new
+      tongji_service.get_access_token(params[:code])
+    end
+    redirect_to "/admins/"
+  end
+
   def evaluate
     names = []
     data = []
@@ -63,15 +111,20 @@ class Admins::DashboardsController < Admins::BaseController
     Time.now.beginning_of_day..Time.now.end_of_day
   end
 
-  def current_week
+  def pre_7_days
     7.days.ago.end_of_day..Time.now.end_of_day
     end
 
+  def current_week
+    Time.now.beginning_of_week..Time.now.end_of_day
+  end
+
   def current_month
     30.days.ago.end_of_day..Time.now.end_of_day
   end
 
   def pre_week
-    14.days.ago.end_of_day..7.days.ago.end_of_day
+    # 14.days.ago.end_of_day..7.days.ago.end_of_day
+    Time.now.prev_week..Time.now.prev_week.end_of_week
   end
 end

app/jobs/daily_platform_statistics_job.rb

@@ -0,0 +1,44 @@
+# 按天获取百度统计数据，pv，访问，ip和来源分类占比
+# 其他统计：前一周用户留存率
+class DailyPlatformStatisticsJob < ApplicationJob
+  queue_as :default
+
+  def perform(*args)
+    Rails.logger.info("*********开始统计*********")
+
+    tongji_service = Baidu::TongjiService.new
+    access_token = tongji_service.access_token
+    Rails.logger.info "job baidu_tongji_auth access_token ===== #{access_token}"
+    ActiveJob::Base.logger.info "job baidu_tongji_auth access_token ===== #{access_token}"
+    # 从最后一个记录日期开始，如果遗漏日期数据可以补充数据
+    last_date = DailyPlatformStatistic.order(:date).last
+    start_date = last_date.date
+    end_date = Time.now
+    if access_token.present?
+      tongji_service.overview_batch_add(start_date, end_date)
+
+      # 本周访问来源占比，每天记录一次，如果遗漏日期数据可以补充数据
+      tongji_service.source_from_batch_add(start_date, end_date)
+    end
+    # 周用户留存率
+    pre_week_user_ids = User.where(created_on: pre_week).pluck(:id).uniq
+    weekly_keep_user_count = User.where(id: pre_week_user_ids).where(last_login_on: current_week).count
+    weekly_keep_rate = format("%.2f", pre_week_user_ids.size > 0 ? weekly_keep_user_count.to_f / pre_week_user_ids.size : 0)
+
+    job_date = 1.days.ago
+    daily_statistic = DailyPlatformStatistic.find_or_initialize_by(date: job_date)
+    daily_statistic.weekly_keep_rate = weekly_keep_rate
+    daily_statistic.save
+  end
+
+  private
+
+  def current_week
+    Time.now.beginning_of_week..Time.now.end_of_day
+  end
+
+  def pre_week
+    # 7.days.ago.beginning_of_week..7.days.ago.beginning_of_week.end_of_week
+    Time.now.prev_week..Time.now.prev_week.end_of_week
+  end
+end

app/models/daily_platform_statistic.rb

@@ -0,0 +1,24 @@
+# == Schema Information
+#
+# Table name: daily_platform_statistics
+#
+#  id               :integer          not null, primary key
+#  date             :date
+#  pv               :integer          default("0")
+#  visitor          :integer          default("0")
+#  ip               :integer          default("0")
+#  weekly_keep_rate :float(24)        default("0")
+#  source_through   :float(24)        default("0")
+#  source_link      :float(24)        default("0")
+#  source_search    :float(24)        default("0")
+#  source_custom    :float(24)        default("0")
+#  created_at       :datetime         not null
+#  updated_at       :datetime         not null
+#
+# Indexes
+#
+#  index_daily_platform_statistics_on_date  (date) UNIQUE
+#
+
+class DailyPlatformStatistic < ApplicationRecord
+end

app/services/baidu/tongji_service.rb

@@ -0,0 +1,221 @@
+module Baidu
+  class TongjiService < ApplicationService
+    attr_reader :client_id, :client_secret, :site_id
+    # login、code、password、password_confirmation
+    def initialize
+      @client_id      = "6dMO2kqKUaMZkBrMaUMxQSNAT49v0Mjq"
+      @client_secret  = "qvWqF33AOmGs1tPCgsROvis9EQCuNmd3"
+      @site_id = 18657013
+    end
+
+    def call
+
+    end
+
+
+    def init_overview_data_by(start_date = nil, end_date = nil)
+      start_date = Time.now.prev_year.beginning_of_year if start_date.nil?
+      end_date = Time.now
+      Rails.logger.info("*********开始百度统计-概览:#{start_date}-#{end_date}*********")
+      sql_connection = ActiveRecord::Base.connection
+      sql_connection.begin_db_transaction
+
+      # 如果存在数据 先清空
+      # sql_connection.execute("delete from daily_platform_statistics where date between '#{start_date}' and '#{end_date}'")
+      multiple_days_data = overview_multiple_days_data(start_date, end_date)
+      if multiple_days_data.present?
+        sql = "replace into daily_platform_statistics (date,pv,visitor,ip,created_at,updated_at) values #{multiple_days_data.join(",")}"
+        sql_connection.execute(sql)
+      end
+      sql_connection.commit_db_transaction
+      Rails.logger.info("*********结束百度统计-概览:#{start_date}-#{end_date}*********")
+    end
+
+    def init_source_from_data_by(start_date = nil, end_date = nil)
+      start_date = Time.now.prev_year.beginning_of_year if start_date.nil?
+      end_date = Time.now
+      Rails.logger.info("*********开始百度统计-来源:#{start_date}-#{end_date}*********")
+      source_from_batch_add(start_date, end_date)
+      Rails.logger.info("*********结束百度统计-来源:#{start_date}-#{end_date}*********")
+    end
+
+    # 按日期获取来源数据
+    def source_from_batch_add(start_date,end_date)
+      # 补充更新开始时间的当天数据
+      source_from_by_date(start_date)
+      diff_days(start_date, end_date).times.each do |t|
+        new_start_date = start_date + (t + 1).days
+        source_from_by_date(new_start_date)
+      end
+      # 补充更新最后时间一天数据
+      source_from_by_date(end_date)
+    end
+
+    # 按天获取来源数据
+    def source_from_by_date(start_date)
+      return [] unless access_token.present? && start_date.present?
+      source_from_data = api("source/all/a", start_date, start_date, "pv_count,visitor_count,ip_count")
+      source_from = []
+      source_from_data['items'][1].each_with_index do |source, index|
+        source_from.push(((source[0].to_f / source_from_data['sum'][0][0].to_f) * 100).round(2))
+      end
+      daily_statistic = DailyPlatformStatistic.find_or_initialize_by(date: start_date)
+      daily_statistic.source_through = source_from[0]
+      daily_statistic.source_link = source_from[1]
+      daily_statistic.source_search = source_from[2]
+      daily_statistic.source_custom = source_from[3]
+      daily_statistic.save
+    end
+
+    def diff_days(start_date, end_date)
+      (end_date.beginning_of_day.to_i - start_date.beginning_of_day.to_i) / (24 * 3600)
+    end
+
+    def overview_batch_add(start_date, end_date)
+      return [] unless access_token.present? && start_date.present? && end_date.present?
+      start_date = Time.now - 1.days if start_date.strftime("%Y%m%d") == end_date.strftime("%Y%m%d")
+      overview_data = api("overview/getTimeTrendRpt", start_date, end_date, "pv_count,visitor_count,ip_count")
+      overview_data['items'][0].each_with_index do |date, index|
+        pv = overview_data['items'][1][index][0]
+        visitor = overview_data['items'][1][index][1]
+        ip = overview_data['items'][1][index][2]
+        job_date = date[0].to_s.gsub("/", "-")
+        daily_statistic = DailyPlatformStatistic.find_or_initialize_by(date: job_date)
+        daily_statistic.date = job_date
+        daily_statistic.pv = pv
+        daily_statistic.visitor = visitor
+        daily_statistic.ip = ip
+        daily_statistic.save
+      end
+      overview_data
+    end
+
+    def overview_multiple_days_data(start_date, end_date)
+      return [] unless access_token.present? && start_date.present? && end_date.present?
+      overview_data = api("overview/getTimeTrendRpt", start_date, end_date, "pv_count,visitor_count,ip_count")
+      data = []
+      created_at = Time.now.strftime("%Y-%m-%d 00:00:00")
+      overview_data['items'][0].each_with_index do |date, index|
+        pv = overview_data['items'][1][index][0]
+        visitor = overview_data['items'][1][index][1]
+        ip = overview_data['items'][1][index][2]
+        data.push("('#{date[0].to_s.gsub("/", "-")}', #{pv.to_s.gsub("--","0")}, #{visitor.to_s.gsub("--","0")}, #{ip.to_s.gsub("--","0")},\"#{created_at}\",\"#{created_at}\")")
+      end
+      data
+    end
+
+    def code_url
+      "http://openapi.baidu.com/oauth/2.0/authorize?response_type=code&client_id=#{client_id}&redirect_uri=oob&scope=basic&display=popup"
+    end
+
+    def oauth_url(code)
+      "http://openapi.baidu.com/oauth/2.0/token?grant_type=authorization_code&code=#{code}&client_id=#{client_id}&client_secret=#{client_secret}&redirect_uri=oob"
+    end
+
+    def get_access_token(code)
+      uri = URI.parse(oauth_url(code))
+      response = Net::HTTP.get_response(uri)
+      Rails.logger.info "baidu_tongji_auth response.body ===== #{response.body}"
+      if response.code.to_i == 200
+        data = JSON.parse(response.body)
+        access_token = data['access_token']
+        refresh_token = data['refresh_token']
+        expires_in = data['expires_in']
+        if access_token.present?
+          Rails.cache.write("baidu_tongji_auth/access_token", access_token, expires_in: expires_in)
+          Rails.cache.write("baidu_tongji_auth/refresh_token", refresh_token, expires_in: 1.year)
+        end
+      end
+    end
+
+    def refresh_access_token
+      url = "http://openapi.baidu.com/oauth/2.0/token?grant_type=refresh_token&refresh_token=#{refresh_token}&client_id=#{client_id}&client_secret=#{client_secret}"
+      uri = URI.parse(url)
+      response = Net::HTTP.get_response(uri)
+      Rails.logger.info "baidu_tongji_auth response.body ===== #{response.body}"
+      if response.code.to_i == 200
+        data = JSON.parse(response.body)
+        access_token = data['access_token']
+        refresh_token = data['refresh_token']
+        expires_in = data['expires_in']
+        if access_token.present?
+          Rails.cache.write("baidu_tongji_auth/access_token", access_token, expires_in: expires_in)
+          Rails.cache.write("baidu_tongji_auth/refresh_token", refresh_token, expires_in: 1.year)
+        end
+      end
+    end
+
+    def access_token
+      access_token = Rails.cache.read("baidu_tongji_auth/access_token")
+      if access_token.blank?
+        refresh_access_token
+        access_token = Rails.cache.read("baidu_tongji_auth/access_token")
+      end
+      access_token
+    end
+
+    def refresh_token
+      refresh_token = Rails.cache.read("baidu_tongji_auth/refresh_token")
+      # 如果刷新token失效，access_token也重置
+      if refresh_token.blank?
+        Rails.cache.delete("baidu_tongji_auth/access_token")
+      end
+      refresh_token
+    end
+
+    # 网站概况(趋势数据)
+    def api_overview
+      start_date = Time.now.beginning_of_week
+      end_date = Time.now
+      start_date = Time.now - 1.days if start_date.strftime("%Y%m%d") == end_date.strftime("%Y%m%d")
+      api("overview/getTimeTrendRpt", start_date, end_date, "pv_count,visitor_count,ip_count")
+    end
+
+    # 网站概况(来源网站、搜索词、入口页面、受访页面)
+    def api_overview_getCommonTrackRpt
+      start_date = Time.now.beginning_of_week
+      end_date = Time.now
+      api("overview/getCommonTrackRpt", start_date, end_date, "pv_count")
+    end
+
+    # 全部来源
+    def source_from
+      start_date = Time.now.beginning_of_week
+      end_date = Time.now
+      api("source/all/a", start_date, end_date, "pv_count,visitor_count,ip_count")
+    end
+
+    def api(api_method, start_date, end_date, metrics = nil)
+      start_date_fmt = start_date.strftime("%Y%m%d")
+      end_date_fmt = end_date.strftime("%Y%m%d")
+      api_url = "https://openapi.baidu.com/rest/2.0/tongji/report/getData?access_token=#{access_token}&site_id=#{site_id}&method=#{api_method}&start_date=#{start_date_fmt}&end_date=#{end_date_fmt}&metrics=#{metrics}"
+      data = url_http_post(api_url, {})
+      data['result']
+    end
+
+    def url_http_post(api_url, params)
+      Rails.logger.info "api_url==#{api_url}"
+      uri = URI.parse(api_url)
+      http = Net::HTTP.new uri.host, uri.port
+      http.open_timeout = 60
+      http.read_timeout = 60
+      if uri.scheme == 'https'
+        http.verify_mode = OpenSSL::SSL::VERIFY_NONE
+        http.use_ssl = true
+      end
+      begin
+        request = Net::HTTP::Post.new(uri)
+        request.set_form_data(params) if params.present?
+        request['Content-Type'] = 'application/json;charset=utf-8'
+        # request['Content-Type'] = 'application/x-www-form-urlencoded;charset=utf-8'
+        response = http.start { |http| http.request(request) }
+        Rails.logger.info "api response.body==#{response.body}"
+        JSON.parse response.body
+      rescue => err
+        Rails.logger.error("#############api_url:#{api_url},error:#{err.message.size}")
+        # Rails.logger.error("#############api_url:#{api_url},error:#{err.message}")
+        return {}
+      end
+    end
+  end
+end

app/views/admins/dashboards/_baidu_tongji.html.erb

@@ -0,0 +1,50 @@
+<div>
+  数据来源百度统计，本周 [<%= @current_week_statistic.first&.date %> / <%= @current_week_statistic.last&.date %>]
+</div>
+<table class="table table-hover text-center subject-list-table">
+  <thead class="thead-light">
+  <tr>
+    <th width="20%">日期</th>
+    <th width="15%">访问量</th>
+    <th width="15%">访客数</th>
+    <th width="15%">IP数</th>
+    <th width="15%">直接访问占比</th>
+    <th width="15%">外部链接占比</th>
+    <th width="15%">搜索引擎占比</th>
+    <th width="20%">自定义</th>
+  </tr>
+  </thead>
+  <tbody>
+  <% @current_week_statistic.each_with_index do |week, index| %>
+    <tr class="">
+      <td><%= week.date %> </td>
+      <td><%= week.pv %></td>
+      <td><%= week.visitor %></td>
+      <td><%= week.ip %></td>
+      <td><%= week.source_through %>%</td>
+      <td><%= week.source_link %>%</td>
+      <td><%= week.source_search %>%</td>
+      <td><%= week.source_custom.to_f %>%</td>
+      </td>
+    </tr>
+  <% end %>
+<!--  <tr class="">-->
+<!--    <td><span style="color: #ff8200">合计</span></td>-->
+<!--    <td><span style="color: #ff8200"><%#= @current_week_statistic.sum(:pv) %></span></td>-->
+<!--    <td><span style="color: #ff8200"><%#= @current_week_statistic.sum(:visitor) %></span></td>-->
+<!--    <td><span style="color: #ff8200"><%#= @current_week_statistic.sum(:ip) %></span></td>-->
+<!--    </td>-->
+<!--  </tr>-->
+  </tbody>
+</table>
+
+
+<% unless @access_token.present? && @overview_data.present? %>
+  <div>
+    <a href="/admins/baidu_tongji" target="_blank">1.百度统计需人工授权,点击去授权</a>
+  </div>
+  <div>
+    <a href="/admins/baidu_tongji_auth" onclick="this.href = '/admins/baidu_tongji_auth?code=' + document.getElementById('auth_code').value">2.获取百度统计授权码</a>
+    <input name="code" id="auth_code" style="width: 380px" placeholder="请输入1.百度统计返回的code后，点击2.获取百度统计Token"/>
+  </div>
+<% end %>

app/views/admins/dashboards/_baidu_tongji_api.html.erb

@@ -0,0 +1,67 @@
+<% if @access_token.present? && @overview_data.present? %>
+  <div>
+    数据来源百度统计，本周 <%= @overview_data['timeSpan'] %>
+  </div>
+  <table class="table table-hover text-center subject-list-table">
+    <thead class="thead-light">
+    <tr>
+      <th width="20%">日期</th>
+      <th width="15%">访问量</th>
+      <th width="15%">访客数</th>
+      <th width="15%">IP数</th>
+    </tr>
+    </thead>
+    <tbody>
+    <% pv_count = [] %>
+    <% visitor_count = [] %>
+    <% ip_count = [] %>
+    <% @overview_data['items'][0].each_with_index do |date, index| %>
+      <% pv = @overview_data['items'][1][index][0] %>
+      <% visitor = @overview_data['items'][1][index][1] %>
+      <% ip = @overview_data['items'][1][index][2] %>
+      <tr class="">
+        <td><%= date[0] %> </td>
+        <td><%= pv %></td>
+        <td><%= visitor %></td>
+        <td><%= ip %></td>
+        </td>
+      </tr>
+      <% pv_count.push(pv) %>
+      <% visitor_count.push(visitor) %>
+      <% ip_count.push(ip) %>
+    <% end %>
+    <tr class="">
+      <td><span style="color: #ff8200">合计</span></td>
+      <td><span style="color: #ff8200"><%= pv_count %></span></td>
+      <td><span style="color: #ff8200"><%= visitor_count %></span></td>
+      <td><span style="color: #ff8200"><%= ip_count %></span></td>
+      </td>
+    </tr>
+    </tbody>
+  </table>
+  <table class="table table-hover text-center subject-list-table">
+    <thead class="thead-light">
+    <tr>
+      <th width="15%">直接访问占比</th>
+      <th width="15%">外部链接占比</th>
+      <th width="15%">搜索引擎占比</th>
+      <th width="20%">自定义</th>
+    </tr>
+    </thead>
+    <tbody>
+    <tr class="">
+      <% @source_from_data['items'][1].each_with_index do |source, index| %>
+        <td><%= ((source[0].to_f / @source_from_data['sum'][0][0].to_f) * 100).round(2).to_s %>%</td>
+      <% end %>
+    </tr>
+    </tbody>
+  </table>
+<% else %>
+  <div>
+    <a href="/admins/baidu_tongji" target="_blank">1.百度统计需人工授权,点击去授权</a>
+  </div>
+  <div>
+    <a href="/admins/baidu_tongji_auth" onclick="this.href = '/admins/baidu_tongji_auth?code=' + document.getElementById('auth_code').value">2.获取百度统计授权码</a>
+    <input name="code" id="auth_code" style="width: 380px" placeholder="请输入1.百度统计返回的code后，点击2.获取百度统计Token"/>
+  </div>
+<% end %>

app/views/admins/dashboards/index.html.erb

@@ -1,6 +1,35 @@
 <% define_admin_breadcrumbs do %>
   <% add_admin_breadcrumb('概览', admins_path) %>
 <% end %>
+<div class="header bg-gradient-primary pb-8 pt-md-8">
+  <div class="container-fluid">
+    <div class="header-body">
+      <!-- Card stats -->
+      <div class="row">
+        <%@subject_name.each_with_index do |subject, index| %>
+          <div class="col-xl-3 col-lg-6" style="padding-bottom: 15px;">
+            <div class="card card-stats mb-4 mb-xl-0">
+              <div class="card-body">
+                <div class="row">
+                  <div class="col">
+                    <h5 class="card-title text-uppercase11 text-muted mb-0"><%=subject %></h5>
+                    <span class="h2 font-weight-bold mb-0"><%= @subject_data[index] %></span>
+                  </div>
+                  <div class="col-auto">
+                    <div class="icon icon-shape rounded-circle shadow">
+                      <i class="fa <%=@subject_icon[index] %>"></i>
+                    </div>
+                  </div>
+                </div>
+              </div>
+            </div>
+          </div>
+        <% end %>
+
+      </div>
+    </div>
+  </div>
+</div>
 
 <div class="box admin-list-container project-language-list-container">
   <table class="table table-hover text-center subject-list-table">
@@ -53,6 +82,7 @@
     </tr>
     </tbody>
   </table>
+  <%= render partial: 'admins/dashboards/baidu_tongji' %>
 </div>
 <div id="project-language-modals">
 </div>

config/routes.rb

@@ -795,6 +795,8 @@ Rails.application.routes.draw do
   namespace :admins do
     mount Sidekiq::Web => '/sidekiq'
     get '/', to: 'dashboards#index'
+    get '/baidu_tongji', to: 'dashboards#baidu_tongji'
+    get '/baidu_tongji_auth', to: 'dashboards#baidu_tongji_auth'
     namespace :topic do
       resources :activity_forums
       resources :banners

config/sidekiq_cron.yml

@@ -12,3 +12,8 @@ create_daily_project_statistics:
   cron: "0 1 * * *"
   class: "DailyProjectStatisticsJob"
   queue: cache
+
+daily_platform_statistics:
+  cron: "0 1 * * *"
+  class: "DailyPlatformStatisticsJob"
+  queue: default

db/migrate/202401321314370_create_daily_platform_statistics.rb

@@ -0,0 +1,18 @@
+class CreateDailyPlatformStatistics < ActiveRecord::Migration[5.2]
+  def change
+    create_table :daily_platform_statistics do |t|
+      t.date :date
+      t.index :date, unique: true
+      t.bigint :pv, default: 0
+      t.bigint :visitor, default: 0
+      t.bigint :ip, default: 0
+      t.float :weekly_keep_rate, default: 0
+      t.float :source_through, default: 0
+      t.float :source_link, default: 0
+      t.float :source_search, default: 0
+      t.float :source_custom, default: 0
+
+      t.timestamps
+    end
+  end
+end

spec/models/daily_platform_statistic_spec.rb

@@ -0,0 +1,5 @@
+require 'rails_helper'
+
+RSpec.describe DailyPlatformStatistic, type: :model do
+  pending "add some examples to (or delete) #{__FILE__}"
+end