305 lines
7.8 KiB
Go
305 lines
7.8 KiB
Go
// 演示数据分析和报告功能
|
||
package main
|
||
|
||
import (
|
||
"encoding/json"
|
||
"fmt"
|
||
"log"
|
||
"math/rand"
|
||
"os"
|
||
"sort"
|
||
"time"
|
||
|
||
"code.tczkiot.com/wlw/pipelinedb"
|
||
"code.tczkiot.com/wlw/pipelinedb/examples/common"
|
||
)
|
||
|
||
// UserEvent 用户事件结构
|
||
type UserEvent struct {
|
||
UserID string `json:"user_id"`
|
||
Action string `json:"action"`
|
||
Page string `json:"page"`
|
||
Timestamp time.Time `json:"timestamp"`
|
||
Value float64 `json:"value,omitempty"`
|
||
}
|
||
|
||
// AnalyticsReport 分析报告
|
||
type AnalyticsReport struct {
|
||
TotalEvents int `json:"total_events"`
|
||
UniqueUsers int `json:"unique_users"`
|
||
TopActions []ActionCount `json:"top_actions"`
|
||
TopPages []PageCount `json:"top_pages"`
|
||
UserActivity map[string]int `json:"user_activity"`
|
||
HourlyActivity map[int]int `json:"hourly_activity"`
|
||
TotalValue float64 `json:"total_value"`
|
||
}
|
||
|
||
type ActionCount struct {
|
||
Action string `json:"action"`
|
||
Count int `json:"count"`
|
||
}
|
||
|
||
type PageCount struct {
|
||
Page string `json:"page"`
|
||
Count int `json:"count"`
|
||
}
|
||
|
||
func main() {
|
||
// 创建临时数据库文件
|
||
dbFile := "analytics_example.db"
|
||
defer os.Remove(dbFile)
|
||
|
||
// 确保文件可以创建
|
||
if _, err := os.Create(dbFile); err != nil {
|
||
log.Fatalf("创建数据库文件失败: %v", err)
|
||
}
|
||
|
||
fmt.Println("🚀 数据分析示例")
|
||
fmt.Println("================")
|
||
|
||
// 配置数据库
|
||
fmt.Println("\n📂 步骤1: 配置数据库")
|
||
config := &pipelinedb.Config{
|
||
CacheSize: 100,
|
||
}
|
||
|
||
// 创建处理器
|
||
handler := common.NewExampleHandler("数据分析")
|
||
|
||
pdb, err := pipelinedb.Open(pipelinedb.Options{
|
||
Filename: dbFile,
|
||
Config: config,
|
||
Handler: handler,
|
||
})
|
||
if err != nil {
|
||
log.Fatalf("打开数据库失败: %v", err)
|
||
}
|
||
defer pdb.Stop()
|
||
|
||
fmt.Println("✅ 数据库已配置")
|
||
|
||
// 生成模拟用户行为数据
|
||
fmt.Println("\n📊 步骤2: 生成模拟用户行为数据")
|
||
|
||
users := []string{"user001", "user002", "user003", "user004", "user005"}
|
||
actions := []string{"页面访问", "按钮点击", "表单提交", "文件下载", "搜索"}
|
||
pages := []string{"/home", "/products", "/about", "/contact", "/login", "/checkout"}
|
||
|
||
// 生成1000个随机事件
|
||
events := make([]UserEvent, 1000)
|
||
baseTime := time.Now().Add(-24 * time.Hour) // 从24小时前开始
|
||
|
||
for i := 0; i < 1000; i++ {
|
||
event := UserEvent{
|
||
UserID: users[rand.Intn(len(users))],
|
||
Action: actions[rand.Intn(len(actions))],
|
||
Page: pages[rand.Intn(len(pages))],
|
||
Timestamp: baseTime.Add(time.Duration(i) * time.Minute),
|
||
}
|
||
|
||
// 为某些动作添加价值
|
||
if event.Action == "表单提交" || event.Action == "文件下载" {
|
||
event.Value = rand.Float64() * 100
|
||
}
|
||
|
||
events[i] = event
|
||
}
|
||
|
||
fmt.Printf("✅ 生成了 %d 个用户事件\n", len(events))
|
||
|
||
// 将事件存储到数据库
|
||
fmt.Println("\n💾 步骤3: 存储事件数据")
|
||
|
||
for i, event := range events {
|
||
// 序列化事件数据
|
||
eventData, err := json.Marshal(event)
|
||
if err != nil {
|
||
log.Fatalf("序列化事件失败: %v", err)
|
||
}
|
||
|
||
// 创建元数据
|
||
metadata := fmt.Sprintf(`{"user_id": "%s", "action": "%s", "page": "%s"}`,
|
||
event.UserID, event.Action, event.Page)
|
||
|
||
// 根据动作类型分组
|
||
group := "用户行为"
|
||
if event.Action == "搜索" {
|
||
group = "搜索行为"
|
||
} else if event.Value > 0 {
|
||
group = "有价值行为"
|
||
}
|
||
|
||
recordID, err := pdb.AcceptData(group, eventData, metadata)
|
||
if err != nil {
|
||
log.Fatalf("存储事件失败: %v", err)
|
||
}
|
||
|
||
if (i+1)%200 == 0 {
|
||
fmt.Printf(" 📝 已存储 %d/%d 个事件，最新ID: %d\n", i+1, len(events), recordID)
|
||
}
|
||
}
|
||
|
||
// 数据分析
|
||
fmt.Println("\n🔍 步骤4: 执行数据分析")
|
||
|
||
report := &AnalyticsReport{
|
||
UserActivity: make(map[string]int),
|
||
HourlyActivity: make(map[int]int),
|
||
}
|
||
|
||
// 获取所有组的数据
|
||
groups := []string{"用户行为", "搜索行为", "有价值行为"}
|
||
allEvents := []UserEvent{}
|
||
|
||
for _, group := range groups {
|
||
fmt.Printf("📋 分析组: %s\n", group)
|
||
|
||
pageReq := &pipelinedb.PageRequest{
|
||
Page: 1,
|
||
PageSize: 1000, // 获取所有数据
|
||
}
|
||
|
||
response, err := pdb.GetRecordsByGroup(group, pageReq)
|
||
if err != nil {
|
||
fmt.Printf(" ❌ 查询组失败: %v\n", err)
|
||
continue
|
||
}
|
||
|
||
fmt.Printf(" 📊 找到 %d 条记录\n", len(response.Records))
|
||
|
||
// 解析事件数据
|
||
for _, record := range response.Records {
|
||
var event UserEvent
|
||
if err := json.Unmarshal(record.Data, &event); err != nil {
|
||
continue
|
||
}
|
||
allEvents = append(allEvents, event)
|
||
}
|
||
}
|
||
|
||
// 分析数据
|
||
fmt.Println("\n📈 步骤5: 生成分析报告")
|
||
|
||
report.TotalEvents = len(allEvents)
|
||
|
||
// 统计唯一用户
|
||
uniqueUsers := make(map[string]bool)
|
||
actionCounts := make(map[string]int)
|
||
pageCounts := make(map[string]int)
|
||
|
||
for _, event := range allEvents {
|
||
// 唯一用户
|
||
uniqueUsers[event.UserID] = true
|
||
|
||
// 用户活跃度
|
||
report.UserActivity[event.UserID]++
|
||
|
||
// 动作统计
|
||
actionCounts[event.Action]++
|
||
|
||
// 页面统计
|
||
pageCounts[event.Page]++
|
||
|
||
// 小时活跃度
|
||
hour := event.Timestamp.Hour()
|
||
report.HourlyActivity[hour]++
|
||
|
||
// 总价值
|
||
report.TotalValue += event.Value
|
||
}
|
||
|
||
report.UniqueUsers = len(uniqueUsers)
|
||
|
||
// 排序Top动作
|
||
for action, count := range actionCounts {
|
||
report.TopActions = append(report.TopActions, ActionCount{
|
||
Action: action,
|
||
Count: count,
|
||
})
|
||
}
|
||
sort.Slice(report.TopActions, func(i, j int) bool {
|
||
return report.TopActions[i].Count > report.TopActions[j].Count
|
||
})
|
||
|
||
// 排序Top页面
|
||
for page, count := range pageCounts {
|
||
report.TopPages = append(report.TopPages, PageCount{
|
||
Page: page,
|
||
Count: count,
|
||
})
|
||
}
|
||
sort.Slice(report.TopPages, func(i, j int) bool {
|
||
return report.TopPages[i].Count > report.TopPages[j].Count
|
||
})
|
||
|
||
// 显示报告
|
||
fmt.Println("\n📋 数据分析报告")
|
||
fmt.Println("================")
|
||
|
||
fmt.Printf("📊 总体统计:\n")
|
||
fmt.Printf(" 总事件数: %d\n", report.TotalEvents)
|
||
fmt.Printf(" 唯一用户数: %d\n", report.UniqueUsers)
|
||
fmt.Printf(" 总价值: %.2f\n", report.TotalValue)
|
||
fmt.Printf(" 平均每用户事件: %.1f\n", float64(report.TotalEvents)/float64(report.UniqueUsers))
|
||
|
||
fmt.Printf("\n🔥 热门动作 (Top 5):\n")
|
||
for i, action := range report.TopActions {
|
||
if i >= 5 {
|
||
break
|
||
}
|
||
percentage := float64(action.Count) / float64(report.TotalEvents) * 100
|
||
fmt.Printf(" %d. %s: %d 次 (%.1f%%)\n",
|
||
i+1, action.Action, action.Count, percentage)
|
||
}
|
||
|
||
fmt.Printf("\n📄 热门页面 (Top 5):\n")
|
||
for i, page := range report.TopPages {
|
||
if i >= 5 {
|
||
break
|
||
}
|
||
percentage := float64(page.Count) / float64(report.TotalEvents) * 100
|
||
fmt.Printf(" %d. %s: %d 次 (%.1f%%)\n",
|
||
i+1, page.Page, page.Count, percentage)
|
||
}
|
||
|
||
fmt.Printf("\n👥 用户活跃度:\n")
|
||
for userID, count := range report.UserActivity {
|
||
fmt.Printf(" %s: %d 次事件\n", userID, count)
|
||
}
|
||
|
||
// 数据库统计
|
||
fmt.Println("\n💾 数据库统计")
|
||
stats, err := pdb.GetStats()
|
||
if err != nil {
|
||
log.Fatalf("获取数据库统计失败: %v", err)
|
||
}
|
||
|
||
fmt.Printf("📈 存储统计:\n")
|
||
fmt.Printf(" 数据库总记录: %d\n", stats.TotalRecords)
|
||
fmt.Printf(" 总组数: %d\n", len(stats.GroupStats))
|
||
|
||
for group, groupStats := range stats.GroupStats {
|
||
fmt.Printf(" [%s]: 热:%d 温:%d 冷:%d\n",
|
||
group, groupStats.HotRecords, groupStats.WarmRecords, groupStats.ColdRecords)
|
||
}
|
||
|
||
// 导出报告
|
||
fmt.Println("\n📤 步骤6: 导出分析报告")
|
||
reportJSON, err := json.MarshalIndent(report, "", " ")
|
||
if err != nil {
|
||
log.Fatalf("序列化报告失败: %v", err)
|
||
}
|
||
|
||
reportFile := "analytics_report.json"
|
||
err = os.WriteFile(reportFile, reportJSON, 0644)
|
||
if err != nil {
|
||
log.Fatalf("写入报告文件失败: %v", err)
|
||
}
|
||
|
||
fmt.Printf("✅ 分析报告已导出到: %s\n", reportFile)
|
||
defer os.Remove(reportFile) // 清理示例文件
|
||
|
||
fmt.Println("\n🎉 数据分析示例完成!")
|
||
fmt.Println("💡 提示: 这个示例展示了如何使用Pipeline Database进行复杂的数据分析")
|
||
}
|