refactor: 将布隆过滤器改为Cuckoo过滤器并优化代码结构
All checks were successful
构建上传工具 / build-tool (push) Successful in 59s
部署开发环境 / deploy-dev (push) Successful in 1m36s

This commit is contained in:
2025-09-12 22:26:45 +08:00
parent 5d304b6334
commit 1c56423ea4
4 changed files with 83 additions and 72 deletions

View File

@@ -31,6 +31,7 @@ func ReadDataHandler(c *gin.Context) {
} }
func WriteDataHandler(c *gin.Context) { func WriteDataHandler(c *gin.Context) {
//解析输入数据
input := struct { input := struct {
Token string `form:"token" binding:"required"` Token string `form:"token" binding:"required"`
Data string `form:"data" binding:"required"` Data string `form:"data" binding:"required"`
@@ -40,6 +41,7 @@ func WriteDataHandler(c *gin.Context) {
return return
} }
//数据获取
dedupObject, err := db.GetDedupObject(input.Token) dedupObject, err := db.GetDedupObject(input.Token)
if err != nil { if err != nil {
c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()}) c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
@@ -52,7 +54,7 @@ func WriteDataHandler(c *gin.Context) {
} }
dedupValue := strings.Split(input.Data, "----")[dataIndex[dedupObject]] dedupValue := strings.Split(input.Data, "----")[dataIndex[dedupObject]]
err = createBF(fmt.Sprintf("dedup:%s:%s", input.Token, dedupObject), 0.01, 100000000) err = createCF(fmt.Sprintf("dedup:%s:%s", input.Token, dedupObject), 100000000)
if err != nil && err.Error() != "ERR item exists" { if err != nil && err.Error() != "ERR item exists" {
c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()}) c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
return return
@@ -64,7 +66,7 @@ local dedupValue = ARGV[1] -- ARGV[1]: 去重值
local rawData = ARGV[2] -- ARGV[2]: 原始数据 local rawData = ARGV[2] -- ARGV[2]: 原始数据
-- 检查布隆过滤器中是否已存在该值 -- 检查布隆过滤器中是否已存在该值
local exists = redis.call('BF.EXISTS', dedupKey, dedupValue) local exists = redis.call('CF.EXISTS', dedupKey, dedupValue)
-- 如果已存在,返回已去重标记 -- 如果已存在,返回已去重标记
if exists == 1 then if exists == 1 then
@@ -72,7 +74,7 @@ if exists == 1 then
end end
-- 添加到布隆过滤器 -- 添加到布隆过滤器
redis.call('BF.ADD', dedupKey, dedupValue) redis.call('CF.ADD', dedupKey, dedupValue)
-- 添加到列表 -- 添加到列表
redis.call('LPUSH', listKey, rawData) redis.call('LPUSH', listKey, rawData)
@@ -102,8 +104,8 @@ return "ok"
c.JSON(http.StatusInternalServerError, gin.H{"error": "WriteDataHandler 错误"}) c.JSON(http.StatusInternalServerError, gin.H{"error": "WriteDataHandler 错误"})
} }
func createBF(bloomFilter string, errorRate float64, capacity int64) error { func createCF(bloomFilter string, capacity int64) error {
_, err := global.RDB.BFReserve(global.RCtx, bloomFilter, errorRate, capacity).Result() _, err := global.RDB.CFReserve(global.RCtx, bloomFilter, capacity).Result()
return err return err
} }

View File

@@ -3,9 +3,9 @@ package controller
import ( import (
"dypid/db" "dypid/db"
"dypid/global" "dypid/global"
"fmt"
"net/http" "net/http"
"strconv" "strconv"
"strings"
"github.com/gin-gonic/gin" "github.com/gin-gonic/gin"
) )
@@ -116,9 +116,10 @@ func GetTokenInfoHandler(c *gin.Context) {
func DeleteTokenInfoHandler(c *gin.Context) { func DeleteTokenInfoHandler(c *gin.Context) {
//解析输入数据 //解析输入数据
input := struct { input := struct {
Token string `form:"token" binding:"required"` Token string `form:"token" binding:"required"`
DedupBF string `form:"dedup_bf"` DedupBF string `form:"dedup_bf"`
CacheList string `form:"cache_list"` CacheList string `form:"cache_list"`
BothNumber string `form:"both_number"`
}{} }{}
if err := c.ShouldBindQuery(&input); err != nil { if err := c.ShouldBindQuery(&input); err != nil {
c.JSON(http.StatusBadRequest, gin.H{"error": "Token不能为空"}) c.JSON(http.StatusBadRequest, gin.H{"error": "Token不能为空"})
@@ -132,6 +133,12 @@ func DeleteTokenInfoHandler(c *gin.Context) {
return return
} }
dedupObject, err := db.GetDedupObject(input.Token)
if err != nil {
c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
return
}
//删除去重对象 //删除去重对象
switch input.DedupBF { switch input.DedupBF {
case "": case "":
@@ -139,6 +146,7 @@ func DeleteTokenInfoHandler(c *gin.Context) {
keys := global.RDB.Keys(global.RCtx, "dedup:"+input.Token+":*").Val() keys := global.RDB.Keys(global.RCtx, "dedup:"+input.Token+":*").Val()
global.RDB.Del(global.RCtx, keys...) global.RDB.Del(global.RCtx, keys...)
default: default:
//TODO 不考虑单独删除指定数量去重参考值
_, err := strconv.Atoi(input.DedupBF) _, err := strconv.Atoi(input.DedupBF)
if err != nil { if err != nil {
c.JSON(http.StatusBadRequest, gin.H{"error": "dedup_bf设置错误 " + err.Error()}) c.JSON(http.StatusBadRequest, gin.H{"error": "dedup_bf设置错误 " + err.Error()})
@@ -155,9 +163,25 @@ func DeleteTokenInfoHandler(c *gin.Context) {
c.JSON(http.StatusBadRequest, gin.H{"error": "cache_list设置错误 " + err.Error()}) c.JSON(http.StatusBadRequest, gin.H{"error": "cache_list设置错误 " + err.Error()})
return return
} }
fmt.Println(-i)
global.RDB.LTrim(global.RCtx, "list:"+input.Token, 1, int64(-i)) global.RDB.LTrim(global.RCtx, "list:"+input.Token, 1, int64(-i))
} }
//TODO
switch input.BothNumber {
case "":
default:
i, err := strconv.Atoi(input.BothNumber)
if err != nil {
c.JSON(http.StatusBadRequest, gin.H{"error": "both_number设置错误 " + err.Error()})
return
}
result := global.RDB.LRange(global.RCtx, "list:"+input.Token, 1, int64(i)).Val()
global.RDB.LTrim(global.RCtx, "list:"+input.Token, 1, int64(-i))
dataIndex, err := getDataIndex(input.Token)
for _, s := range result {
s2 := strings.Split(s, "----")[dataIndex[dedupObject]]
global.RDB.CFDel(global.RCtx, "dedup:"+input.Token+":"+dedupObject, s2)
}
}
//输出信息 //输出信息
c.JSON(http.StatusOK, gin.H{"result": "ok"}) c.JSON(http.StatusOK, gin.H{"result": "ok"})

View File

@@ -26,6 +26,41 @@ var httpClient = &http.Client{
} }
func main() { func main() {
initConfig()
//检测./upload
fmt.Println("程序启动成功正在检测txt文件")
for {
files, err := getTxtFiles("./")
if err != nil {
fmt.Println(err)
return
}
if files != nil {
start := time.Now()
wg := sync.WaitGroup{}
for _, filePath := range files {
fmt.Println("正在上传文件:", filePath)
wg.Add(1)
go func() {
processFile(filePath)
err := os.Truncate(filePath, 0)
if err != nil {
fmt.Println("清空文件失败:", err)
}
wg.Done()
}()
}
wg.Wait()
fmt.Println("上传完成,耗时:", time.Since(start))
}
time.Sleep(time.Minute)
}
}
func initConfig() {
//程序配置 //程序配置
viper.SetDefault("url", "http://localhost:8080") viper.SetDefault("url", "http://localhost:8080")
viper.SetDefault("token", "") viper.SetDefault("token", "")
@@ -45,33 +80,6 @@ func main() {
fmt.Errorf("无法读取配置文件: %w", err) fmt.Errorf("无法读取配置文件: %w", err)
} }
}) })
//检测./upload
fmt.Println("程序启动成功正在检测txt文件")
for {
files, err := getTxtFiles("./")
if err != nil {
fmt.Println(err)
return
}
if files != nil {
wg := sync.WaitGroup{}
for _, filePath := range files {
fmt.Println("正在上传文件:", filePath)
wg.Add(1)
go func() {
processFile(filePath)
err := os.Truncate(filePath, 0)
if err != nil {
fmt.Println("清空文件失败:", err)
}
wg.Done()
}()
}
wg.Wait()
}
time.Sleep(time.Minute)
}
} }
func uploadDataToServer(data string) error { func uploadDataToServer(data string) error {

View File

@@ -16,12 +16,8 @@ value.value = useCounterStore().token
const options = ref([] as string[]) const options = ref([] as string[])
// 控制删除指定Redis键的确认对话框的显示状态 // 控制删除指定Redis键的确认对话框的显示状态
const deleteSpecifyRedisVisible = ref(false) const deleteSpecifyDataVisible = ref(false)
const inputSpecifyRedis = ref('') const inputSpecifyData = ref('')
// 控制删除指定去重项的确认对话框的显示状态
const deleteSpecifyDedupVisible = ref(false)
const inputSpecifyDedup = ref('')
const getInfo = () => { const getInfo = () => {
@@ -76,27 +72,15 @@ axios.get('/api/token').then(res => {
} }
}) })
const deleteSpecifyRedis = () => { const deleteSpecifyData = () => {
axios.delete('/api/token/info', { axios.delete('/api/token/info', {
params: { params: {
token: value.value, token: value.value,
cache_list: inputSpecifyRedis.value, cache_list: inputSpecifyData.value,
} }
}).then(res => { }).then(res => {
getInfo() getInfo()
deleteSpecifyRedisVisible.value = false deleteSpecifyDataVisible.value = false
})
}
const deleteSpecifyDedup = () => {
axios.delete('/api/token/info', {
params: {
token: value.value,
dedup_bf: inputSpecifyDedup.value,
}
}).then(res => {
getInfo()
deleteSpecifyDedupVisible.value = false
}) })
} }
</script> </script>
@@ -107,6 +91,7 @@ const deleteSpecifyDedup = () => {
<el-alert title="您没有权限访问此页面" type="error" center show-icon/> <el-alert title="您没有权限访问此页面" type="error" center show-icon/>
</div> </div>
<div v-if="useCounterStore().isAdmin"> <div v-if="useCounterStore().isAdmin">
<b>当前Token</b> <b>当前Token</b>
<el-select v-model="value" placeholder="选择Token" style="width: 240px"> <el-select v-model="value" placeholder="选择Token" style="width: 240px">
@@ -144,25 +129,17 @@ const deleteSpecifyDedup = () => {
<el-button type="danger" @click="deleteDedup">删除全部去重参考值</el-button> <el-button type="danger" @click="deleteDedup">删除全部去重参考值</el-button>
<el-button type="danger" @click="deleteRedis">删除全部原始数据</el-button> <el-button type="danger" @click="deleteRedis">删除全部原始数据</el-button>
<div style="margin-top: 10px"> <div style="margin-top: 10px">
<el-button type="danger" @click="deleteSpecifyDedupVisible=true">删除指定数量去重参考值</el-button> <el-button type="danger" @click="deleteSpecifyDataVisible=true">
<el-button type="danger" @click="deleteSpecifyRedisVisible=true">删除指定数量原始数据</el-button> 删除指定数量数据去重参考值+原始数据
</el-button>
</div> </div>
<!--弹窗输入--> <!--弹窗输入-->
<el-dialog v-model="deleteSpecifyDedupVisible" title="删除指定数量去重参考值" width="400"> <el-dialog v-model="deleteSpecifyDataVisible" title="删除指定数量的数据" width="400">
<el-input v-model="inputSpecifyDedup" style="width: 200px" placeholder="请输入删除数量"/> <el-input v-model="inputSpecifyData" style="width: 200px" placeholder="请输入删除数量"/>
<template #footer> <template #footer>
<el-button type="primary" @click="deleteSpecifyDedup"> <el-button type="primary" @click="deleteSpecifyData">
确定
</el-button>
</template>
</el-dialog>
<el-dialog v-model="deleteSpecifyRedisVisible" title="删除指定数量原始数据" width="400">
<el-input v-model="inputSpecifyRedis" style="width: 200px" placeholder="请输入删除数量"/>
<template #footer>
<el-button type="primary" @click="deleteSpecifyRedis">
确定 确定
</el-button> </el-button>
</template> </template>