广告数仓：数仓搭建(二)
2024-07-04 10:52:54
系列文章目录
广告数仓：采集通道创建
广告数仓：数仓搭建
广告数仓：数仓搭建(二)
文章目录
- 系列文章目录
- 前言
- DWD层创建
- 1.建表
- 广告事件事实表
- 2.数据装载
- 初步解析日志
- 解析IP和UA
- 标注无效流量
- 编写脚本
- 总结
前言
这次我们完成数仓剩下的内容
DWD层创建
1.建表
广告事件事实表
drop table if exists dwd_ad_event_inc;
create external table if not exists dwd_ad_event_inc
(event_time bigint comment '事件时间',event_type string comment '事件类型',ad_id string comment '广告id',ad_name string comment '广告名称',ad_product_id string comment '广告商品id',ad_product_name string comment '广告商品名称',ad_product_price decimal(16, 2) comment '广告商品价格',ad_material_id string comment '广告素材id',ad_material_url string comment '广告素材地址',ad_group_id string comment '广告组id',platform_id string comment '推广平台id',platform_name_en string comment '推广平台名称(英文)',platform_name_zh string comment '推广平台名称(中文)',client_country string comment '客户端所处国家',client_area string comment '客户端所处地区',client_province string comment '客户端所处省份',client_city string comment '客户端所处城市',client_ip string comment '客户端ip地址',client_device_id string comment '客户端设备id',client_os_type string comment '客户端操作系统类型',client_os_version string comment '客户端操作系统版本',client_browser_type string comment '客户端浏览器类型',client_browser_version string comment '客户端浏览器版本',client_user_agent string comment '客户端UA',is_invalid_traffic boolean comment '是否是异常流量'
) PARTITIONED BY (`dt` STRING)STORED AS ORCLOCATION '/warehouse/ad/dwd/dwd_ad_event_inc/'TBLPROPERTIES ('orc.compress' = 'snappy');
2.数据装载
初步解析日志
create temporary table coarse_parsed_log
as
selectparse_url('http://www.example.com' || request_uri, 'QUERY', 't') event_time,split(parse_url('http://www.example.com' || request_uri, 'PATH'), '/')[3] event_type,parse_url('http://www.example.com' || request_uri, 'QUERY', 'id') ad_id,split(parse_url('http://www.example.com' || request_uri, 'PATH'), '/')[2] platform,parse_url('http://www.example.com' || request_uri, 'QUERY', 'ip') client_ip,reflect('java.net.URLDecoder', 'decode', parse_url('http://www.example.com'||request_uri,'QUERY','ua'), 'utf-8') client_ua,parse_url('http://www.example.com'||request_uri,'QUERY','os_type') client_os_type,parse_url('http://www.example.com'||request_uri,'QUERY','device_id') client_device_id
from ods_ad_log_inc
where dt='2023-01-07';
解析IP和UA
这里我要用idea编写hive的udf自定义类
为pom.xml添加依赖
<dependencies><!-- hive-exec依赖无需打到jar包，故scope使用provided--><dependency><groupId>org.apache.hive</groupId><artifactId>hive-exec</artifactId><version>3.1.3</version><scope>provided</scope></dependency><!-- ip地址库--><dependency><groupId>org.lionsoul</groupId><artifactId>ip2region</artifactId><version>2.7.0</version></dependency><dependency><groupId>cn.hutool</groupId><artifactId>hutool-http</artifactId><version>5.8.18</version></dependency></dependencies><build><plugins><plugin><groupId>org.apache.maven.plugins</groupId><artifactId>maven-assembly-plugin</artifactId><version>3.0.0</version><configuration><!--将依赖编译到jar包中--><descriptorRefs><descriptorRef>jar-with-dependencies</descriptorRef></descriptorRefs></configuration><executions><!--配置执行器--><execution><id>make-assembly</id><!--绑定到package执行周期上--><phase>package</phase><goals><!--只运行一次--><goal>single</goal></goals></execution></executions></plugin></plugins></build>
com/atguigu/ad/hive/udf/ParseIP.java
package com.atguigu.ad.hive.udf;import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hive.ql.exec.UDFArgumentException;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDF;
import org.apache.hadoop.hive.serde2.objectinspector.ConstantObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory;
import org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory;
import org.apache.hadoop.io.IOUtils;
import org.lionsoul.ip2region.xdb.Searcher;import java.io.ByteArrayOutputStream;
import java.util.ArrayList;public class ParseIP extends GenericUDF {Searcher searcher = null;/*** 判断函数传入的参数个数以及类型 同时确定返回值类型**/@Overridepublic ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException {//传入参数的个数if (arguments.length != 2) {throw new UDFArgumentException("parseIP必须填写2个参数");}// 校验参数的类型ObjectInspector hdfsPathOI = arguments[0];if (hdfsPathOI.getCategory() != ObjectInspector.Category.PRIMITIVE) {throw new UDFArgumentException("parseIP第一个参数必须是基本数据类型");}PrimitiveObjectInspector hdfsPathOI1 = (PrimitiveObjectInspector) hdfsPathOI;if (hdfsPathOI1.getPrimitiveCategory() != PrimitiveObjectInspector.PrimitiveCategory.STRING) {throw new UDFArgumentException("parseIP第一个参数必须STRING类型");}ObjectInspector ipOI = arguments[1];if (ipOI.getCategory() != ObjectInspector.Category.PRIMITIVE) {throw new UDFArgumentException("parseIP第一个参数必须是基本数据类型");}PrimitiveObjectInspector ipOI1 = (PrimitiveObjectInspector) ipOI;if (ipOI1.getPrimitiveCategory() != PrimitiveObjectInspector.PrimitiveCategory.STRING) {throw new UDFArgumentException("parseIP第二个参数必须STRING类型");}// 读取ip静态库进入内存中//获取hdfsPath地址if (hdfsPathOI instanceof ConstantObjectInspector) {String hafsPath = ((ConstantObjectInspector) hdfsPathOI).getWritableConstantValue().toString();// 从hdfs读取静态库Path path = new Path(hafsPath);try {FileSystem fileSystem = FileSystem.get(new Configuration());FSDataInputStream inputStream = fileSystem.open(path);ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream();IOUtils.copyBytes(inputStream, byteArrayOutputStream, 1024);byte[] bytes = byteArrayOutputStream.toByteArray();//创建静态库，解析IPsearcher = Searcher.newWithBuffer(bytes);} catch (Exception e) {e.printStackTrace();}}// 确定函数返回值的类型ArrayList<String> structFieldNames = new ArrayList<>();structFieldNames.add("country");structFieldNames.add("area");structFieldNames.add("province");structFieldNames.add("city");structFieldNames.add("isp");ArrayList<ObjectInspector> structFieldObjectInspectors = new ArrayList<>();structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);return ObjectInspectorFactory.getStandardStructObjectInspector(structFieldNames, structFieldObjectInspectors);}/*** 处理数据**/@Overridepublic Object evaluate(DeferredObject[] deferredObjects) throws HiveException {String ip = deferredObjects[1].get().toString();ArrayList<Object> result = new ArrayList<>();try {String search = searcher.search(ip);String[] split = search.split("\\|");result.add(split[0]);result.add(split[1]);result.add(split[2]);result.add(split[3]);result.add(split[4]);} catch (Exception e) {e.printStackTrace();}return result;}/*** 描述函数*/@Overridepublic String getDisplayString(String[] children) {return getStandardDisplayString("parseIP", children);}
}
com/atguigu/ad/hive/udf/ParseUA.java
package com.atguigu.ad.hive.udf;import cn.hutool.http.useragent.UserAgent;
import cn.hutool.http.useragent.UserAgentUtil;
import org.apache.hadoop.hive.ql.exec.UDFArgumentException;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDF;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory;
import org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory;import java.util.ArrayList;public class ParseUA extends GenericUDF {@Overridepublic ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException {//传入参数的个数if (arguments.length != 1) {throw new UDFArgumentException("parseIP必须填写1个参数");}// 校验参数的类型ObjectInspector uaOI = arguments[0];if (uaOI.getCategory() != ObjectInspector.Category.PRIMITIVE) {throw new UDFArgumentException("parseUA第一个参数必须是基本数据类型");}PrimitiveObjectInspector uaOI1 = (PrimitiveObjectInspector) uaOI;if (uaOI1.getPrimitiveCategory() != PrimitiveObjectInspector.PrimitiveCategory.STRING) {throw new UDFArgumentException("parseUA第一个参数必须STRING类型");}// 确定函数返回值的类型ArrayList<String> structFieldNames = new ArrayList<>();structFieldNames.add("browser");structFieldNames.add("browserVersion");structFieldNames.add("engine");structFieldNames.add("engineVersion");structFieldNames.add("os");structFieldNames.add("osVersion");structFieldNames.add("platform");structFieldNames.add("isMobile");ArrayList<ObjectInspector> structFieldObjectInspectors = new ArrayList<>();structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);structFieldObjectInspectors.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);return ObjectInspectorFactory.getStandardStructObjectInspector(structFieldNames, structFieldObjectInspectors);}@Overridepublic Object evaluate(DeferredObject[] deferredObjects) throws HiveException {String ua = deferredObjects[0].get().toString();UserAgent parse = UserAgentUtil.parse(ua);ArrayList<Object> result = new ArrayList<>();result.add(parse.getBrowser().getName());result.add(parse.getVersion());result.add(parse.getEngine().getName());result.add(parse.getEngineVersion());result.add(parse.getOs().getName());result.add(parse.getOsVersion());result.add(parse.getPlatform().getName());result.add(parse.isMobile());return result;}@Overridepublic String getDisplayString(String[] strings) {return getStandardDisplayString("parseUA", strings);}
}
打包上传到hadoop集群
上传到/user/hive/jars目录，没有就创建一个
ip2region.xdb到HDFS/ip2region/
这个文件可以自己生成 也可以用提供的
在hive中注册自定义函数
create function parse_ipas 'com.atguigu.ad.hive.udf.ParseIP'using jar 'hdfs://hadoop102:8020//user/hive/jars/ad_hive_udf-1.0-SNAPSHOT-jar-with-dependencies.jar';create function parse_uaas 'com.atguigu.ad.hive.udf.ParseUA'using jar 'hdfs://hadoop102:8020//user/hive/jars/ad_hive_udf-1.0-SNAPSHOT-jar-with-dependencies.jar';
测试一下
select parse_ip("hdfs://hadoop102:8020/ip2region/ip2region.xdb","120.245.112.30")select parse_ua("Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/109.0.0.0 Safari/537.36");
创建临时表
set hive.vectorized.execution.enabled=false;
create temporary table fine_parsed_log
as
selectevent_time,event_type,ad_id,platform,client_ip,client_ua,client_os_type,client_device_id,parse_ip('hdfs://hadoop102:8020/ip2region/ip2region.xdb',client_ip) region_struct,if(client_ua != '',parse_ua(client_ua),null) ua_struct
from coarse_parsed_log;
标注无效流量
1.根据已知爬虫列表进行判断
建表
drop table if exists dim_crawler_user_agent;
create external table if not exists dim_crawler_user_agent
(pattern STRING comment '正则表达式',addition_date STRING comment '收录日期',url STRING comment '爬虫官方url',instances ARRAY<STRING> comment 'UA实例'
)STORED AS ORCLOCATION '/warehouse/ad/dim/dim_crawler_user_agent'TBLPROPERTIES ('orc.compress' = 'snappy');
创建过度表
create temporary table if not exists tmp_crawler_user_agent
(pattern STRING comment '正则表达式',addition_date STRING comment '收录日期',url STRING comment '爬虫官方url',instances ARRAY<STRING> comment 'UA实例'
)ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.JsonSerDe'STORED AS TEXTFILELOCATION '/warehouse/ad/tmp/tmp_crawler_user_agent';
上传数据
导入数据
insert overwrite table dim_crawler_user_agent select * from tmp_crawler_user_agent;
2.同一ip访问过快
5分钟内超过100次，SQL实现逻辑如下
create temporary table high_speed_ip
as
selectdistinct client_ip
from
(selectevent_time,client_ip,ad_id,count(1) over(partition by client_ip,ad_id order by cast(event_time as bigint) range between 300000 preceding and current row) event_count_last_5minfrom coarse_parsed_log
)t1
where event_count_last_5min>100;
3.同一ip固定周期访问
固定周期访问超过5次，SQL实现逻辑如下
create temporary table cycle_ip
as
selectdistinct client_ip
from
(selectclient_ip,ad_id,sfrom(selectevent_time,client_ip,ad_id,sum(num) over(partition by client_ip,ad_id order by event_time) sfrom(selectevent_time,client_ip,ad_id,time_diff,if(lag(time_diff,1,0) over(partition by client_ip,ad_id order by event_time)!=time_diff,1,0) numfrom(selectevent_time,client_ip,ad_id,lead(event_time,1,0) over(partition by client_ip,ad_id order by event_time)-event_time time_difffrom coarse_parsed_log)t1)t2)t3group by client_ip,ad_id,shaving count(*)>=5
)t4;
4.同一设备访问过快
5分钟内超过100次，SQL实现逻辑如下
create temporary table high_speed_device
as
selectdistinct client_device_id
from
(selectevent_time,client_device_id,ad_id,count(1) over(partition by client_device_id,ad_id order by cast(event_time as bigint) range between 300000 preceding and current row) event_count_last_5minfrom coarse_parsed_logwhere client_device_id != ''
)t1
where event_count_last_5min>100;
5.同一设备固定周期访问
固定周期访问超过5次。
create temporary table cycle_device
as
selectdistinct client_device_id
from
(selectclient_device_id,ad_id,sfrom(selectevent_time,client_device_id,ad_id,sum(num) over(partition by client_device_id,ad_id order by event_time) sfrom(selectevent_time,client_device_id,ad_id,time_diff,if(lag(time_diff,1,0) over(partition by client_device_id,ad_id order by event_time)!=time_diff,1,0) numfrom(selectevent_time,client_device_id,ad_id,lead(event_time,1,0) over(partition by client_device_id,ad_id order by event_time)-event_time time_difffrom coarse_parsed_logwhere client_device_id != '')t1)t2)t3group by client_device_id,ad_id,shaving count(*)>=5
)t4;
6.标识异常流量并做维度退化
insert overwrite table dwd_ad_event_inc partition (dt='2023-01-07')
selectevent_time,event_type,event.ad_id,ad_name,product_id,product_name,product_price,material_id,material_url,group_id,plt.id,platform_name_en,platform_name_zh,region_struct.country,region_struct.area,region_struct.province,region_struct.city,event.client_ip,event.client_device_id,if(event.client_os_type!='',event.client_os_type,ua_struct.os),nvl(ua_struct.osVersion,''),nvl(ua_struct.browser,''),nvl(ua_struct.browserVersion,''),event.client_ua,if(coalesce(pattern,hsi.client_ip,ci.client_ip,hsd.client_device_id,cd.client_device_id) is not null,true,false)
from fine_parsed_log event
left join dim_crawler_user_agent crawler on event.client_ua regexp crawler.pattern
left join high_speed_ip hsi on event.client_ip = hsi.client_ip
left join cycle_ip ci on event.client_ip = ci.client_ip
left join high_speed_device hsd on event.client_device_id = hsd.client_device_id
left join cycle_device cd on event.client_device_id = cd.client_device_id
left join
(selectad_id,ad_name,product_id,product_name,product_price,material_id,material_url,group_idfrom dim_ads_info_fullwhere dt='2023-01-07'
)ad
on event.ad_id=ad.ad_id
left join
(selectid,platform_name_en,platform_name_zhfrom dim_platform_info_fullwhere dt='2023-01-07'
)plt
on event.platform=plt.platform_name_en;
编写脚本
#!/bin/bashAPP=ad# 如果是输入的日期按照取输入日期；如果没输入日期取当前时间的前一天
if [ -n "$2" ] ;thendo_date=$2
else do_date=`date -d "-1 day" +%F`
fidwd_ad_event_inc="
set hive.vectorized.execution.enabled=false;
--初步解析
create temporary table coarse_parsed_log
as
selectparse_url('http://www.example.com' || request_uri, 'QUERY', 't') event_time,split(parse_url('http://www.example.com' || request_uri, 'PATH'), '/')[3] event_type,parse_url('http://www.example.com' || request_uri, 'QUERY', 'id') ad_id,split(parse_url('http://www.example.com' || request_uri, 'PATH'), '/')[2] platform,parse_url('http://www.example.com' || request_uri, 'QUERY', 'ip') client_ip,reflect('java.net.URLDecoder', 'decode', parse_url('http://www.example.com'||request_uri,'QUERY','ua'), 'utf-8') client_ua,parse_url('http://www.example.com'||request_uri,'QUERY','os_type') client_os_type,parse_url('http://www.example.com'||request_uri,'QUERY','device_id') client_device_id
from ${APP}.ods_ad_log_inc
where dt='$do_date';
--进一步解析ip和ua
create temporary table fine_parsed_log
as
selectevent_time,event_type,ad_id,platform,client_ip,client_ua,client_os_type,client_device_id,${APP}.parse_ip('hdfs://hadoop102:8020/ip2region/ip2region.xdb',client_ip) region_struct,if(client_ua != '',${APP}.parse_ua(client_ua),null) ua_struct
from coarse_parsed_log;
--高速访问ip
create temporary table high_speed_ip
as
selectdistinct client_ip
from
(selectevent_time,client_ip,ad_id,count(1) over(partition by client_ip,ad_id order by cast(event_time as bigint) range between 300000 preceding and current row) event_count_last_5minfrom coarse_parsed_log
)t1
where event_count_last_5min>100;
--周期访问ip
create temporary table cycle_ip
as
selectdistinct client_ip
from
(selectclient_ip,ad_id,sfrom(selectevent_time,client_ip,ad_id,sum(num) over(partition by client_ip,ad_id order by event_time) sfrom(selectevent_time,client_ip,ad_id,time_diff,if(lag(time_diff,1,0) over(partition by client_ip,ad_id order by event_time)!=time_diff,1,0) numfrom(selectevent_time,client_ip,ad_id,lead(event_time,1,0) over(partition by client_ip,ad_id order by event_time)-event_time time_difffrom coarse_parsed_log)t1)t2)t3group by client_ip,ad_id,shaving count(*)>=5
)t4;
--高速访问设备
create temporary table high_speed_device
as
selectdistinct client_device_id
from
(selectevent_time,client_device_id,ad_id,count(1) over(partition by client_device_id,ad_id order by cast(event_time as bigint) range between 300000 preceding and current row) event_count_last_5minfrom coarse_parsed_logwhere client_device_id != ''
)t1
where event_count_last_5min>100;
--周期访问设备
create temporary table cycle_device
as
selectdistinct client_device_id
from
(selectclient_device_id,ad_id,sfrom(selectevent_time,client_device_id,ad_id,sum(num) over(partition by client_device_id,ad_id order by event_time) sfrom(selectevent_time,client_device_id,ad_id,time_diff,if(lag(time_diff,1,0) over(partition by client_device_id,ad_id order by event_time)!=time_diff,1,0) numfrom(selectevent_time,client_device_id,ad_id,lead(event_time,1,0) over(partition by client_device_id,ad_id order by event_time)-event_time time_difffrom coarse_parsed_logwhere client_device_id != '')t1)t2)t3group by client_device_id,ad_id,shaving count(*)>=5
)t4;
--维度退化
insert overwrite table ${APP}.dwd_ad_event_inc partition (dt='$do_date')
selectevent_time,event_type,event.ad_id,ad_name,product_id,product_name,product_price,material_id,material_url,group_id,plt.id,platform_name_en,platform_name_zh,region_struct.country,region_struct.area,region_struct.province,region_struct.city,event.client_ip,event.client_device_id,if(event.client_os_type!='',event.client_os_type,ua_struct.os),nvl(ua_struct.osVersion,''),nvl(ua_struct.browser,''),nvl(ua_struct.browserVersion,''),event.client_ua,if(coalesce(pattern,hsi.client_ip,ci.client_ip,hsd.client_device_id,cd.client_device_id) is not null,true,false)
from fine_parsed_log event
left join ${APP}.dim_crawler_user_agent crawler on event.client_ua regexp crawler.pattern
left join high_speed_ip hsi on event.client_ip = hsi.client_ip
left join cycle_ip ci on event.client_ip = ci.client_ip
left join high_speed_device hsd on event.client_device_id = hsd.client_device_id
left join cycle_device cd on event.client_device_id = cd.client_device_id
left join
(selectad_id,ad_name,product_id,product_name,product_price,material_id,material_url,group_idfrom ${APP}.dim_ads_info_fullwhere dt='$do_date'
)ad
on event.ad_id=ad.ad_id
left join
(selectid,platform_name_en,platform_name_zh`在这里插入代码片`from ${APP}.dim_platform_info_fullwhere dt='$do_date'
)plt
on event.platform=plt.platform_name_en;
"case $1 in
"dwd_ad_event_inc")hive -e "$dwd_ad_event_inc"
;;
"all")hive -e "$dwd_ad_event_inc"
;;
esac
添加权限测试一下
测试之前可以先关掉DataGrip节省一点内存,然后重启一下hiveserver2服务，清空之前的内存。
chmod +x ~/bin/ad_ods_to_dwd.sh
ad_ods_to_dwd.sh all 2023-01-07
由于每次调用需要创建多个临时表，所以时间会稍微长一点，大概几分钟。
总结
至此输仓搭建全部完成。
广告数仓：数仓搭建(二)相关推荐
- （130）System Verilog忽略仓数与违反仓数用法
(130)System Verilog忽略仓数与违反仓数用法 1.1 目录 1)目录 2)FPGA简介 3)System Verilog简介 4)System Verilog忽略仓数与违反仓数用法 5 ...
- 【离线数仓-3-数仓建模方法理论汇总】
离线数仓-3-数仓建模方法理论汇总 离线数仓-3-数仓建模方法理论汇总 1.数仓概述 2.数据仓库核心架构(Hive) 3.数据仓库建模概述 4.数据仓库建模方法论 1.ER(Entity Relat ...
- 数字经济时代，达尔文平台助力广告投放走向数智化
早晨起床,先打开手机查看一下智能手环的睡眠记录,看下自己的睡眠质量,洗漱之后踩上智能体重秤,记录下体重.体脂数据,然后根据APP推荐菜谱,准备早餐.午餐-身体的多项行为.数据开始被记录分析,人们也开始 ...
- 数商云：搭建B2B撮合交易平台，以“三新”战略驱动化工企业利润高增长
数商云--驱动全链数字化运营 自2014年伊始,一批着眼于化工类的垂直B2B电商开始兴起,目的是用互联网方式降低行业信息不对称,提高流通分销效率,降低产业链运营成本.然而,在经过几年的"野蛮 ...
- 【Kafka】Kafka-分区数-备份数-如何设置-怎么确定-怎么修改
Kafka-分区数-备份数-如何设置-怎么确定-怎么修改 kafka partition 数量 更新_百度搜索kafka重新分配partition - - CSDN博客如何为Kafka集群选择合适的P ...
- js一个大盒子中有三个小盒子_寒假预习：一万有多大，数一数估一估，亲身感觉一万有多大...
空中课堂的教学进行到二年级数学万以内数的认识了.怎样让学生感受到一万有多大呢?我给学生布置了一个数数的作业.布置作业前一天,我自己先数了数. 我数的是大米,先一粒一粒的数出一百粒大米放在一个小盒子里, ...
- 数一数你连听都没听过的古典小说有多少？
作者:www.gudianxiaoshuo.com 数一数你连名字都没听过的古典小说有多少? 先秦 先秦汉魏晋南北朝诗 屈原全集 汉代诗歌 诗经 词 [全唐五代词] 宋词三百首 晏几道词全集 欧阳修词 ...
- 【算法讲11：卡特兰数】默慈金数 | 那罗延数 | 施罗德数
[算法讲11:卡特兰数]默慈金数 | 那罗延数 | 施罗德数 ⌈\lceil⌈卡特兰数⌋\rfloor⌋Catalan Number 引入 思考 ⌈\lceil⌈卡特兰数⌋\rfloor⌋的性质 ⌈\ ...
- 数钱数到手抽筋html5,数钱数到手抽筋的经典句子
1.他两手抓住竹竿,像敏捷的猴子那样,双脚一蹬,就嘈嘈嘈地爬了上去. 2.父母会在历史博物馆里徜徉度过愉快的时光. 3.只要我的生命没有停止,对他的记忆就不会熄灭. 4.对于像数学物理这样的学科,我只 ...
最新文章
- java 匿名对象有引用_java 匿名内部类对象持有外部类对象的引用
- 上交大博士生：博三经历5轮审稿被拒后，如今发10篇论文成为审稿人
- 设计模式之策略模式(strategy)--游戏角色使用武器
- [Noip模拟赛] Polygon
- 钉钉小程序数据传递——子传父，父传子
- 集群没有leader_ZooKeeper 集群中 Leader 与 Follower 的4种数据同步策略
- 异常处理记录: Servlet class X is not a javax.servlet.Servlet
- Flutter基础—定位对齐之大小比例
- 五十二 常用第三方模块 图形界面
- Centos7 Kubernetes(k8s) 开发服务器(单服务器)部署 路由 IngressRoute【traefik2.X】
- 在Windows环境下搭建Nginx文件服务器(简单实用版)
- 达人管理系统(CRM)原型图
- html求视频的原尺寸,PR怎样导出原尺寸视频？
- 学术论文检索--搜索引擎篇
- 马士兵_JAVA自学之路
- 【安全】Shellshock漏洞
- 洛谷P2404 自然数的拆分问题(回溯)
- 解决方法 curl: (35) OpenSSL SSL_connect: SSL_ERROR_SYSCALL in connection to bit.ly:443
- 用python写一个圣诞互换礼物的抽奖小程序
- 【新词发现】基于SNS的文本数据挖掘、短语挖掘