59、Flink CEP - Flink的复杂事件处理介绍及示例（4）- 延迟数据处理和三个实际应用示例

Flink 系列文章

一、Flink 专栏

Flink 专栏系统介绍某一知识点，并辅以具体的示例进行说明。

1、Flink 部署系列本部分介绍Flink的部署、配置相关基础内容。
2、Flink基础系列本部分介绍Flink 的基础部分，比如术语、架构、编程模型、编程指南、基本的datastream api用法、四大基石等内容。
3、Flik Table API和SQL基础系列本部分介绍Flink Table Api和SQL的基本用法，比如Table API和SQL创建库、表用法、查询、窗口函数、catalog等等内容。
4、Flik Table API和SQL提高与应用系列本部分是table api 和sql的应用部分，和实际的生产应用联系更为密切，以及有一定开发难度的内容。
5、Flink 监控系列本部分和实际的运维、监控工作相关。

二、Flink 示例专栏

Flink 示例专栏是 Flink 专栏的辅助说明，一般不会介绍知识点的信息，更多的是提供一个一个可以具体使用的示例。本专栏不再分目录，通过链接即可看出介绍的内容。

两专栏的所有文章入口点击：Flink 系列文章汇总索引

文章目录

本文介绍了Flink 的类库CEP的时间处理、可选的参数以及在实际工作中非常有用的三个示例。

如果需要了解更多内容，可以在本人Flink 专栏中了解更新系统的内容。

本文除了maven依赖外，没有其他依赖。

本专题分为以下几篇介绍：
59、Flink CEP - Flink的复杂事件处理介绍及示例（1）-入门
59、Flink CEP - Flink的复杂事件处理介绍及示例（2）- 模式API
59、Flink CEP - Flink的复杂事件处理介绍及示例（3）- 模式选取及超时处理
59、Flink CEP - Flink的复杂事件处理介绍及示例（4）- 延迟数据处理和三个实际应用示例
59、Flink CEP - Flink的复杂事件处理介绍及示例（完整版）

四、CEP库中的时间

1、按照事件时间处理迟到事件

在CEP中，事件的处理顺序很重要。在使用事件时间时，为了保证事件按照正确的顺序被处理，一个事件到来后会先被放到一个缓冲区中，在缓冲区里事件都按照时间戳从小到大排序，当水位线到达后，缓冲区中所有小于水位线的事件被处理。这意味着水位线之间的数据都按照时间戳被顺序处理。

这个库假定按照事件时间时水位线一定是正确的。
为了保证跨水位线的事件按照事件时间处理，Flink CEP库假定水位线一定是正确的，并且把时间戳小于最新水位线的事件看作是晚到的。晚到的事件不会被处理。你也可以指定一个侧输出标志来收集比最新水位线晚到的事件，你可以这样做：

PatternStream<Event> patternStream =CEP.pattern(input, pattern);OutputTag<String> lateDataOutputTag =newOutputTag<String>("late-data"){};SingleOutputStreamOperator<ComplexEvent> result = patternStream
    .sideOutputLateData(lateDataOutputTag).select(newPatternSelectFunction<Event,ComplexEvent>(){...});DataStream<String> lateData = result.getSideOutput(lateDataOutputTag);

2、时间上下文

在PatternProcessFunction中，用户可以和IterativeCondition中一样按照下面的方法使用实现了TimeContext的上下文：

/**
 * 支持获取事件属性比如当前处理事件或当前正处理的事件的时间。
 * 用在{@link PatternProcessFunction}和{@link org.apache.flink.cep.pattern.conditions.IterativeCondition}中
 */@PublicEvolvingpublicinterfaceTimeContext{/**
     * 当前正处理的事件的时间戳。
     *
     * <p>如果是{@link org.apache.flink.streaming.api.TimeCharacteristic#ProcessingTime}，这个值会被设置为事件进入CEP算子的时间。
     */longtimestamp();/** 返回当前的处理时间。 */longcurrentProcessingTime();}

这个上下文让用户可以获取处理的事件（在IterativeCondition时候是进来的记录，在PatternProcessFunction是匹配的结果）的时间属性。调用TimeContext#currentProcessingTime总是返回当前的处理时间，而且尽量去调用这个函数而不是调用其它的比如说System.currentTimeMillis()。

使用EventTime时，TimeContext#timestamp()返回的值等于分配的时间戳。使用ProcessingTime时，这个值等于事件进入CEP算子的时间点（在PatternProcessFunction中是匹配产生的时间）。这意味着多次调用这个方法得到的值是一致的。

五、可选的参数设置

用于配置 Flink CEP 的 SharedBuffer 缓存容量的选项。它可以加快 CEP 算子的处理速度，并限制内存中缓存的元素数量。

仅当 state.backend.type 设置为 rocksdb 时限制内存使用才有效，这会将超过缓存数量的元素传输到 rocksdb 状态存储而不是内存状态存储。当 state.backend.type 设置为 rocksdb 时，这些配置项有助于限制内存。相比之下，当 state.backend 设置为非 rocksdb 时，缓存会导致性能下降。与使用 Map 实现的旧缓存相比，状态部分将包含更多从 guava-cache 换出的元素，这将使得 copy on write 时的状态处理增加一些开销。

在这里插入图片描述

六、示例

本部分通过几个示例展示CEP的使用方式，有些是工作中实际的例子简化版，有些是为了演示CEP功能构造的例子，其运行结果均在代码的注释中。

1、maven依赖

本文示例均使用此处的依赖。

<properties><encoding>UTF-8</encoding><project.build.sourceEncoding>UTF-8</project.build.sourceEncoding><maven.compiler.source>1.8</maven.compiler.source><maven.compiler.target>1.8</maven.compiler.target><java.version>1.8</java.version><scala.version>2.12</scala.version><flink.version>1.17.0</flink.version></properties><dependencies><dependency><groupId>org.apache.flink</groupId><artifactId>flink-clients</artifactId><version>${flink.version}</version><scope>provided</scope></dependency><dependency><groupId>org.apache.flink</groupId><artifactId>flink-java</artifactId><version>${flink.version}</version><scope>provided</scope></dependency><dependency><groupId>org.apache.flink</groupId><artifactId>flink-streaming-java</artifactId><version>${flink.version}</version><!-- <scope>provided</scope> --></dependency><dependency><groupId>org.apache.flink</groupId><artifactId>flink-csv</artifactId><version>${flink.version}</version><scope>provided</scope></dependency><dependency><groupId>org.apache.flink</groupId><artifactId>flink-json</artifactId><version>${flink.version}</version><scope>provided</scope></dependency><dependency><groupId>org.projectlombok</groupId><artifactId>lombok</artifactId><version>1.18.2</version><!-- <scope>provided</scope> --></dependency><dependency><groupId>org.apache.flink</groupId><artifactId>flink-cep</artifactId><version>1.17.2</version></dependency></dependencies>

2、示例：输出每个用户连续三次登录失败的信息，允许数据延迟10s

importjava.time.Duration;importjava.util.Arrays;importjava.util.List;importjava.util.Map;importorg.apache.flink.api.common.eventtime.WatermarkStrategy;importorg.apache.flink.cep.CEP;importorg.apache.flink.cep.PatternSelectFunction;importorg.apache.flink.cep.PatternStream;importorg.apache.flink.cep.pattern.Pattern;importorg.apache.flink.cep.pattern.conditions.SimpleCondition;importorg.apache.flink.streaming.api.datastream.DataStream;importorg.apache.flink.streaming.api.environment.StreamExecutionEnvironment;importlombok.AllArgsConstructor;importlombok.Data;importlombok.NoArgsConstructor;/*
 * @Author: alanchan
 * @LastEditors: alanchan
 * @Description: 输出每个用户连续三次登录失败的信息，允许数据延迟10s
 */publicclassTestLoginFailDemo{@Data@NoArgsConstructor@AllArgsConstructorstaticclassLoginEvent{privateInteger userId;privateString ip;privateString status;privateLong timestamp;@Overridepublicbooleanequals(Object obj){if(obj instanceofLoginEvent){LoginEvent loginEvent =(LoginEvent) obj;returnthis.userId == loginEvent.getUserId()&&this.ip.equals(loginEvent.ip)&&this.status.equals(loginEvent.getStatus())&&this.timestamp == loginEvent.getTimestamp();}else{returnfalse;}}@OverridepublicinthashCode(){returnsuper.hashCode()+Long.hashCode(timestamp);}}finalstaticList<LoginEvent> loginEventList =Arrays.asList(newLoginEvent(1001,"192.168.10.1","F",2L),newLoginEvent(1001,"192.168.10.2","F",3L),newLoginEvent(1002,"192.168.10.8","F",4L),newLoginEvent(1001,"192.168.10.6","F",5L),newLoginEvent(1002,"192.168.10.8","F",7L),newLoginEvent(1002,"192.168.10.8","F",8L),newLoginEvent(1002,"192.168.10.8","S",6L));publicstaticvoidmain(String[] args)throwsException{StreamExecutionEnvironment env =StreamExecutionEnvironment.getExecutionEnvironment();// 按用户id分组DataStream<LoginEvent> loginEventDS = env.fromCollection(loginEventList).assignTimestampsAndWatermarks(WatermarkStrategy.<LoginEvent>forBoundedOutOfOrderness(Duration.ofSeconds(10)).withTimestampAssigner((loginEvent, rs)-> loginEvent.getTimestamp())).keyBy(loginEvent -> loginEvent.getUserId());// 定义模式,连续的三个登录失败事件Pattern<LoginEvent,?> loginEventPattern =Pattern.<LoginEvent>begin("first").where(newSimpleCondition<LoginEvent>(){@Overridepublicbooleanfilter(LoginEvent value)throwsException{return value.getStatus().equals("F");}}).next("second").where(newSimpleCondition<LoginEvent>(){@Overridepublicbooleanfilter(LoginEvent value)throwsException{return value.getStatus().equals("F");}}).next("third").where(newSimpleCondition<LoginEvent>(){@Overridepublicbooleanfilter(LoginEvent value)throwsException{return value.getStatus().equals("F");}});// 将Pattern应用到流上，检测匹配的复杂事件，得到一个PatternStreamPatternStream<LoginEvent> patternStream =CEP.pattern(loginEventDS, loginEventPattern);// 将匹配到的流选择出来输出
        patternStream
                .select(newPatternSelectFunction<LoginEvent,String>(){@OverridepublicStringselect(Map<String,List<LoginEvent>> map)throwsException{// 个体模式是单例的,List 中只有一个元素LoginEvent first = map.get("first").get(0);LoginEvent second = map.get("second").get(0);LoginEvent third = map.get("third").get(0);// return first.toString() + "\n" + second.toString() + "\n" + third.toString();return map.get("first").toString()+" \n"+ map.get("second").toString()+" \n"+ map.get("third").toString();}}).print("连续三次登录失败用户信息：\n");// 连续三次登录失败用户信息：// :9> TestLoginFailDemo.LoginEvent(userId=1001, ip=192.168.10.1, status=F,timestamp=2)// TestLoginFailDemo.LoginEvent(userId=1001, ip=192.168.10.2, status=F, timestamp=3)// TestLoginFailDemo.LoginEvent(userId=1001, ip=192.168.10.6, status=F, timestamp=5)

        env.execute();}}

3、示例：查找同一个ip在10秒内访问同一个链接超过3次的ip，可以不连续

可以监控接口是否被攻击，应用应该比较广泛。

importjava.time.Duration;importjava.util.Arrays;importjava.util.HashMap;importjava.util.List;importjava.util.Map;importorg.apache.flink.api.common.eventtime.WatermarkStrategy;importorg.apache.flink.cep.CEP;importorg.apache.flink.cep.PatternStream;importorg.apache.flink.cep.functions.PatternProcessFunction;importorg.apache.flink.cep.nfa.aftermatch.AfterMatchSkipStrategy;importorg.apache.flink.cep.pattern.Pattern;importorg.apache.flink.streaming.api.datastream.DataStream;importorg.apache.flink.streaming.api.environment.StreamExecutionEnvironment;importorg.apache.flink.streaming.api.windowing.time.Time;importorg.apache.flink.util.Collector;importlombok.AllArgsConstructor;importlombok.Data;importlombok.NoArgsConstructor;/*
 * @Author: alanchan
 * @LastEditors: alanchan
 * @Description: 查找 同一个ip在10秒内访问同一个链接超过3次的ip，可以不连续
 */publicclassTestRepeatAccessDemo{@Data@NoArgsConstructor@AllArgsConstructorstaticclassLogMessage{privateString ip;privateString url;privateLong timestamp;@Overridepublicbooleanequals(Object obj){if(obj instanceofLogMessage){LogMessage logMessage =(LogMessage) obj;returnthis.ip.equals(logMessage.getIp())&&this.url.equals(logMessage.getUrl())&&this.timestamp == logMessage.getTimestamp();}else{returnfalse;}}@OverridepublicinthashCode(){returnsuper.hashCode()+Long.hashCode(timestamp);}}finalstaticList<LogMessage> logMessageList =Arrays.asList(newLogMessage("192.168.10.1","URL1",2L),newLogMessage("192.168.10.1","URL1",3L),newLogMessage("192.168.10.1","URL2",4L),newLogMessage("192.168.10.1","URL2",5L),newLogMessage("192.168.10.8","URL1",6L),newLogMessage("192.168.10.1","URL1",7L));@Data@NoArgsConstructor@AllArgsConstructorstaticclassRiskLogListextendsLogMessage{privateint count;}staticvoidtest1()throwsException{StreamExecutionEnvironment env =StreamExecutionEnvironment.getExecutionEnvironment();// 按用户id分组DataStream<LogMessage> logMessageDS = env.fromCollection(logMessageList).assignTimestampsAndWatermarks(WatermarkStrategy.<LogMessage>forBoundedOutOfOrderness(Duration.ofSeconds(10)).withTimestampAssigner((logMessage, rs)-> logMessage.getTimestamp())).keyBy(logMessage -> logMessage.getIp()+ logMessage.getUrl());// 根据ip和url分组// 定义模式Pattern<LogMessage,?> logMessagePattern =Pattern.<LogMessage>begin("first").followedBy("second").times(2).within(Time.seconds(10));// 将Pattern应用到流上，检测匹配的复杂事件，得到一个PatternStreamPatternStream<LogMessage> patternStream =CEP.pattern(logMessageDS, logMessagePattern);// 将匹配到的流选择出来输出
        patternStream.process(newPatternProcessFunction<LogMessage,String>(){@OverridepublicvoidprocessMatch(Map<String,List<LogMessage>> match,Context ctx,Collector<String> out)throwsException{LogMessage logMessage1 = match.get("first").get(0);LogMessage logMessage2 = match.get("second").get(0);LogMessage logMessage3 = match.get("second").get(1);boolean flag = logMessage1.getUrl().equals(logMessage2.getUrl())&& logMessage1.getUrl().equals(logMessage3.getUrl());if(flag){
                    out.collect(logMessage1.getIp()+"  url:"+ logMessage1.getUrl()+"   timestamp:"+ logMessage1.getTimestamp()+"  timestamp2:"+ logMessage2.getTimestamp()+"  timestamp3:"+ logMessage3.getTimestamp());}}}).print("输出信息：\n");// 控制台输出：// 输出信息：:1> 192.168.10.1 url:URL1 timestamp:2 timestamp2:3 timestamp3:7
        env.execute();}publicstaticvoidmain(String[] args)throwsException{test1();}}

4、示例：监测服务器的温度并告警

监测机器温度一小时内三次大于设定温度进行风险记录，将风险记录中的数据上一次大于平均值进行报警。

importjava.time.Duration;importjava.util.ArrayList;importjava.util.Arrays;importjava.util.HashMap;importjava.util.Iterator;importjava.util.List;importjava.util.Map;importorg.apache.flink.api.common.eventtime.WatermarkStrategy;importorg.apache.flink.cep.CEP;importorg.apache.flink.cep.PatternStream;importorg.apache.flink.cep.functions.PatternProcessFunction;importorg.apache.flink.cep.pattern.Pattern;importorg.apache.flink.cep.pattern.conditions.SimpleCondition;importorg.apache.flink.streaming.api.datastream.DataStream;importorg.apache.flink.streaming.api.environment.StreamExecutionEnvironment;importorg.apache.flink.streaming.api.windowing.time.Time;importorg.apache.flink.util.Collector;importlombok.AllArgsConstructor;importlombok.Data;importlombok.NoArgsConstructor;/*
 * @Author: alanchan
 * @LastEditors: alanchan
 * @Description: 监测 机器温度一小时内三次大于设定温度进行风险记录，将风险记录中的数据上一次大于平均值进行报警
 */publicclassTestMachineMonitoring{// 机器的基本信息@Data@NoArgsConstructor@AllArgsConstructorstaticclassMechineInfo{privateint mechineId;privateString mechineName;privateint temperature;privateLong timestamp;@Overridepublicbooleanequals(Object obj){if(obj instanceofMechineInfo){MechineInfo mechineInfo =(MechineInfo) obj;returnthis.mechineId == mechineInfo.getMechineId()&&this.mechineName.equals(mechineInfo.getMechineName())&&this.timestamp == mechineInfo.getTimestamp()&&this.temperature == mechineInfo.getTemperature();}else{returnfalse;}}@OverridepublicinthashCode(){returnsuper.hashCode()+Long.hashCode(timestamp);}}// 机器的三次平均温度@Data@NoArgsConstructor@AllArgsConstructorstaticclassMechineRiskInfo{privateint mechineId;privatedouble avgTemperature;privateLong timestamp;@Overridepublicbooleanequals(Object obj){if(obj instanceofMechineRiskInfo){MechineRiskInfo mechineRiskInfo =(MechineRiskInfo) obj;returnthis.mechineId == mechineRiskInfo.getMechineId()&&this.avgTemperature == mechineRiskInfo.getAvgTemperature()&&this.timestamp == mechineRiskInfo.getTimestamp();}else{returnfalse;}}@OverridepublicinthashCode(){returnsuper.hashCode()+Long.hashCode(timestamp);}}// 预警通知信息@Data@NoArgsConstructor@AllArgsConstructorstaticclassMechineAlertInfo{privateint mechineId;privateString email;privatedouble avgTemperature;privateLong timestamp;@Overridepublicbooleanequals(Object obj){if(obj instanceofMechineAlertInfo){MechineAlertInfo mechineAlertInfo =(MechineAlertInfo) obj;returnthis.mechineId == mechineAlertInfo.getMechineId()&&this.email == mechineAlertInfo.getEmail()&&this.avgTemperature == mechineAlertInfo.getAvgTemperature()&&this.timestamp == mechineAlertInfo.getTimestamp();}else{returnfalse;}}@OverridepublicinthashCode(){returnsuper.hashCode()+Long.hashCode(timestamp);}}// 初始化流数据staticList<MechineInfo> mechineInfoList =Arrays.asList(newMechineInfo(1,"m1",331,2L),newMechineInfo(1,"m1",321,4L),newMechineInfo(1,"m1",311,5L),newMechineInfo(1,"m1",361,7L),newMechineInfo(1,"m1",351,9L),newMechineInfo(1,"m1",341,11L),newMechineInfo(2,"m11",121,3L),newMechineInfo(3,"m21",101,4L),newMechineInfo(4,"m31",98,5L),newMechineInfo(5,"m41",123,6L));// 风险数据集合// static List<MechineRiskInfo> mechineRiskInfoList = new ArrayList();// 预警数据集合// static Map<String, MechineAlertInfo> mechineAlertInfoMap = new HashMap<String, MechineAlertInfo>();// 预警温度privatestaticfinaldoubleTEMPERATURE_SETTING=100;// 超时数据staticvoidtest1()throwsException{StreamExecutionEnvironment env =StreamExecutionEnvironment.getExecutionEnvironment();// 按用户id分组DataStream<MechineInfo> mechineInfoStream = env.fromCollection(mechineInfoList).assignTimestampsAndWatermarks(WatermarkStrategy.<MechineInfo>forBoundedOutOfOrderness(Duration.ofSeconds(10)).withTimestampAssigner((logMessage, rs)-> logMessage.getTimestamp())).keyBy(mechineInfo -> mechineInfo.getMechineId());// 根据ip和url分组// 定义模式1，过滤温度大于设置温度Pattern<MechineInfo,?> mechineInfoPattern =Pattern.<MechineInfo>begin("first").where(newSimpleCondition<MechineInfo>(){@Overridepublicbooleanfilter(MechineInfo value)throwsException{return value.getTemperature()>=TEMPERATURE_SETTING;}}).followedBy("second").where(newSimpleCondition<MechineInfo>(){@Overridepublicbooleanfilter(MechineInfo value)throwsException{return value.getTemperature()>=TEMPERATURE_SETTING;}}).times(2).within(Time.minutes(10));PatternStream<MechineInfo> patternStream =CEP.pattern(mechineInfoStream, mechineInfoPattern);// 筛选，并计算 三次温度平均值DataStream<MechineRiskInfo> mechineRiskInfoStream = patternStream
                .process(newPatternProcessFunction<TestMachineMonitoring.MechineInfo,MechineRiskInfo>(){@OverridepublicvoidprocessMatch(Map<String,List<MechineInfo>> match,Context ctx,Collector<MechineRiskInfo> out)throwsException{MechineInfo firstMechineInfo = match.get("first").get(0);MechineInfo secondMechineInfo1 = match.get("second").get(0);MechineInfo secondMechineInfo2 = match.get("second").get(1);// System.out.printf("mechineInfo:id=%s,name=%s,t=%s,ts=%s",//         firstMechineInfo.getMechineId(),//         firstMechineInfo.getMechineName(), firstMechineInfo.getTemperature(),//         firstMechineInfo.getTimestamp() + "\n");// System.out.printf("secondMechineInfo1:id=%s,name=%s,t=%s,ts=%s",//         secondMechineInfo1.getMechineId(),//         secondMechineInfo1.getMechineName(), secondMechineInfo1.getTemperature(),//         secondMechineInfo1.getTimestamp() + "\n");// System.out.printf("secondMechineInfo2:id=%s,name=%s,t=%s,ts=%s",//         secondMechineInfo2.getMechineId(),//         secondMechineInfo2.getMechineName(), secondMechineInfo2.getTemperature(),//         secondMechineInfo2.getTimestamp() + "\n");

                        out.collect(newMechineRiskInfo(
                                firstMechineInfo.getMechineId(),(firstMechineInfo.getTemperature()+ secondMechineInfo1.getTemperature()+ secondMechineInfo2.getTemperature())/3,
                                ctx.timestamp()));}}).keyBy(mechineRiskInfo -> mechineRiskInfo.getMechineId());

        mechineRiskInfoStream.print("mechineRiskInfoStream:");// 定义模式2，比较风险数据的前后两条，如果是上升的趋势，则报警，并设置报警联系人Pattern<MechineRiskInfo,?> mechineRiskInfoPattern =Pattern.<MechineRiskInfo>begin("step1").next("step2").within(Time.hours(1));PatternStream<MechineRiskInfo> patternStream2 =CEP.pattern(mechineRiskInfoStream, mechineRiskInfoPattern);// 筛选 警告信息，并设置发送邮箱DataStream<MechineAlertInfo> mechineAlertInfoList = patternStream2
                .process(newPatternProcessFunction<TestMachineMonitoring.MechineRiskInfo,MechineAlertInfo>(){@OverridepublicvoidprocessMatch(Map<String,List<MechineRiskInfo>> match,Context ctx,Collector<MechineAlertInfo> out)throwsException{MechineRiskInfo mechineRiskInfo1 = match.get("step1").get(0);MechineRiskInfo mechineRiskInfo2 = match.get("step2").get(0);MechineAlertInfoMechineAlertInfo=null;if(mechineRiskInfo1.getAvgTemperature()<= mechineRiskInfo2.getAvgTemperature()){MechineAlertInfo=newMechineAlertInfo(mechineRiskInfo1.getMechineId(),"[email protected]",
                                    mechineRiskInfo2.getAvgTemperature(), ctx.currentProcessingTime());

                            out.collect(MechineAlertInfo);}}});
        mechineAlertInfoList.print("mechineAlertInfoList:");// mechineAlertInfoList::11> TestMachineMonitoring.MechineAlertInfo(mechineId=1, [email protected], avgTemperature=331.0, timestamp=1705366481553)// mechineAlertInfoList::11> TestMachineMonitoring.MechineAlertInfo(mechineId=1, [email protected], avgTemperature=341.0, timestamp=1705366481566)// mechineAlertInfoList::11> TestMachineMonitoring.MechineAlertInfo(mechineId=1, [email protected], avgTemperature=351.0, timestamp=1705366481567)
        env.execute();}publicstaticvoidmain(String[] args)throwsException{test1();}}

以上，本文介绍了Flink 的类库CEP的时间处理、可选的参数以及在实际工作中非常有用的三个示例。

标签： flink 大数据 kafka

本文转载自: https://blog.csdn.net/chenwewi520feng/article/details/135615696
版权归原作者 一瓢一瓢的饮 alanchanchn 所有，如有侵权，请联系我们删除。