hive3从入门到精通

查询hive

架构

准备

HDFS配置

vim $HADOOP_HOME/etc/hadoop/core-site.xml 

  <!--配置所有节点的root用户都可作为代理用户-->
  <property>
    <name>hadoop.proxyuser.root.hosts</name>
    <value>*</value>
  </property>

  <!--配置root用户能够代理的用户组为任意组-->
  <property>
    <name>hadoop.proxyuser.root.groups</name>
    <value>*</value>
  </property>

  <!--配置root用户能够代理的用户为任意用户-->
  <property>
    <name>hadoop.proxyuser.root.users</name>
    <value>*</value>
  </property>

启动Hadoop

# 启动hadoop
start-all.sh
# 检查hadoop进程
jps
# 检查各端口
netstat -aplnt | grep java

检查MySQL是否启动成功

ps -aux | grep mysql
netstat -aplnt | grep 3306

安装hive

# 将软件上传到 /opt/soft 目录
# 解压hive
tar -zxvf apache-hive-3.1.3-bin.tar.gz 
# 目录改名
mv apache-hive-3.1.3-bin hive-3
# 进入配置文件目录
cd /opt/soft/hive-3/conf
# 复制配置文件
cp hive-env.sh.template  hive-env.sh
cp hive-default.xml.template  hive-site.xml
# 编辑环境配置文件
vim hive-env.sh
# 编辑配置文件
vim hive-site.xml

hive-env.sh

# hadoop 安装路径
export HADOOP_HOME=/opt/soft/hadoop-3/
# hive 配置文件路径
export HIVE_CONF_DIR=/opt/soft/hive-3/conf/

hive-site.xml

需要修改的位置提炼如下：

<configuration>
    <!-- 记录HIve中的元数据信息  记录在mysql中 -->
    <property>
        <name>javax.jdo.option.ConnectionURL</name>
        <value>jdbc:mysql://spark03:3306/hive?useUnicode=true&amp;createDatabaseIfNotExist=true&amp;characterEncoding=UTF8&amp;useSSL=false&amp;serverTimeZone=Asia/Shanghai</value>
    </property>

    <!-- jdbc mysql驱动 -->
    <property>
        <name>javax.jdo.option.ConnectionDriverName</name>
        <value>com.mysql.cj.jdbc.Driver</value>
    </property>

    <!-- mysql的用户名和密码 -->
    <property>
        <name>javax.jdo.option.ConnectionUserName</name>
        <value>root</value>
    </property>
    <property>
        <name>javax.jdo.option.ConnectionPassword</name>
        <value>Lihaozhe!!@@1122</value>
    </property>

    <property>
        <name>hive.metastore.warehouse.dir</name>
        <value>/user/hive/warehouse</value>
    </property>

    <property>
        <name>hive.exec.scratchdir</name>
        <value>/user/hive/tmp</value>
    </property>
    
    </property>

    <property>
        <name>hive.exec.local.scratchdir</name>
        <value>/user/hive/local</value>
        <description>Local scratch space for Hive jobs</description>
    </property>

    <property>
        <name>hive.downloaded.resources.dir</name>
        <value>/user/hive/resources</value>
        <description>Temporary local directory for added resources in the remote file system.</description>
    </property>

    <!-- 日志目录 -->
    <property>
        <name>hive.querylog.location</name>
        <value>/user/hive/log</value>
    </property>

    <!-- 设置metastore的节点信息 -->
    <property>
        <name>hive.metastore.uris</name>
        <value>thrift://spark01:9083</value>
    </property>

    <!-- 客户端远程连接的端口 -->
    <property> 
        <name>hive.server2.thrift.port</name> 
        <value>10000</value>
    </property>
    <property> 
        <name>hive.server2.thrift.bind.host</name> 
        <value>0.0.0.0</value>
    </property>
    <property>
        <name>hive.server2.webui.host</name>
        <value>0.0.0.0</value>
    </property>

    <!-- hive服务的页面的端口 -->
    <property>
        <name>hive.server2.webui.port</name>
        <value>10002</value>
    </property>

    <property> 
        <name>hive.server2.long.polling.timeout</name> 
        <value>5000ms</value>
    </property>

    <property>
        <name>hive.server2.enable.doAs</name>
        <value>true</value>
    </property>
    <!--
<property>
<name>datanucleus.autoCreateSchema</name>
<value>false</value>
</property>

<property>
<name>datanucleus.fixedDatastore</name>
<value>true</value>
</property>
-->
    <property>
        <name>hive.execution.engine</name>
        <value>mr</value>
    </property>
    <property>
        <name>hive.metastore.schema.verification</name>
        <value>false</value>
        <description>
          Enforce metastore schema version consistency.
          True: Verify that version information stored in is compatible with one from Hive jars.  Also disable automatic
                schema migration attempt. Users are required to manually migrate schema after Hive upgrade which ensures
                proper metastore schema migration. (Default)
          False: Warn if the version information stored in metastore doesn't match with one from in Hive jars.
        </description>
      </property>
</configuration>

注意：上面配置文件中的路径在 vi 编辑器下 全局替换

:%s@\${system:java.io.tmpdir}@/tmp/hive-log@g

不要使用图形化 不然每次保存后3215行都会有个 &#8 特殊字符 如果产生删除即可 具体报错信息 后面有单独的描述

上传 MySQL 连接驱动 jar 包到 hive 安装目录的lib目录下:

$HIVE_HOME/lib

删除原有的 protobuf-java-2.5.0.jar 文件

jar 包有两个 分别为：

• mysql-connector-j-8.0.33.jar
• protobuf-java-3.5.1.jar

guava版本冲突

删除 hive/lib目录中的 guava-19.0.jar

拷贝hadoop/share/hadoop/common/lib目录中的 guava-27.0-jre.jar 到 hive/lib 目录

rm -f $HIVE_HOME/lib/guava-19.0.jar
cp -v $HADOOP_HOME/share/hadoop/common/lib/guava-27.0-jre.jar $HIVE_HOME/lib

配置环境变量

vim /etc/profile.d/my_env.sh

export HIVE_HOME=/opt/soft/hive-3
export HCATALOG_HOME=/opt/soft/hive-3/hcatalog

export PATH=$PATH:$JAVA_HOME/bin:$ZOOKEEPER_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$HIVE_HOME/bin:$HCATALOG_HOME/bin:$HCATALOG_HOME/sbin

source /etc/profile

初始化hive的元数据库

注意：初始初始元数据库之前 保证 hadoop 和 mysql 正常启动

schematool -initSchema -dbType  mysql

报错解解决：

Exception in thread "main" java.lang.RuntimeException: com.ctc.wstx.exc.WstxParsingException: Illegal character entity: expansion character (code 0x8
at [row,col,system-id]: [3215,96,"file:/usr/local/hive/conf/hive-site.xml"]
at org.apache.hadoop.conf.Configuration.loadResource(Configuration.java:3051)
...
at org.apache.hadoop.util.RunJar.main(RunJar.java:236)
Caused by: com.ctc.wstx.exc.WstxParsingException: Illegal character entity: expansion character (code 0x8
at [row,col,system-id]: [3215,96,"file:/usr/local/hive/conf/hive-site.xml"]
at com.ctc.wstx.sr.StreamScanner.constructWfcException(StreamScanner.java:621)
...
at org.apache.hadoop.conf.Configuration.loadResource(Configuration.java:3034)
... 17 more

报错原因：

hive-site.xml配置文件中，3215行（见报错记录第二行）有特殊字符

解决办法：

进入hive-site.xml文件，跳转到对应行，删除里面的 &#8 特殊字符即可

Caused by: java.net.URISyntaxException: Relative path in absolute URI: ${system:java.io.tmpdir%7D/$%7Bsystem:user.name%7D
	at java.net.URI.checkPath(URI.java:1822)
	at java.net.URI.<init>(URI.java:745)
	at org.apache.hadoop.fs.Path.initialize(Path.java:260)

解决方案：将hive-site.xml配置文件的
hive.querylog.location
hive.exec.local.scratchdir
hive.downloaded.resources.dir
三个值（原始为$标识的相对路径）写成绝对值

# 全局替换
:%s@\${system:java.io.tmpdir}@/tmp/hive-log@g

远程模式

# 启动服务端
hive --service metastore
hive --service hiveserver2

# 启动服务端 后台运行
hive --service metastore &
hive --service hiveserver2 &

# 启动服务端 后台运行
nohup hive --service metastore > /dev/null 2>&1 &
nohup hive --service hiveserver2 > /dev/null 2>&1 &

hiveserver2 start
nohup hiveserver2 start >/dev/null 2>&1 &

bash 中 0、1、2 三个数字分别代表 STDIN_FILENO 、 STDOUT_FILENO 、STDERR_FILENO ，
即标准输入（一般是键盘），
标准输出（一般是显示屏，准确的说是用户终端控制台），
标准错误（输出错信息输出）。

/dev/null看作"黑洞". 它等价于一个只写文件.
所有写入它的内容都会永远丢失.
而尝试从它那儿读取内容则什么也读不到

# 客户端连接
hive
beeline -u jdbc:hive2://spark01:10000 -n root
beeline jdbc:hive2://spark01:10000> show databases;

体验

use default;
create table person (
	id int,
    phonenum bigint,
    salary dicimal,
    name string
);
show tables;
insert into person values (1001,13966668888,9999.99,"张三");

songsong,bingbing_lili,xiao song:18_xiaoxiao song:19
longlong,pingping_liuliu,xiao long:8_xiaoxiao long:9

drop table person;
create table person (
    name string,
    friends array<string>,
    childrens map<string,int>
)
 row format delimited fields terminated by ',' 
 collection items terminated by '_' 
 map keys terminated by ':' 
 lines terminated by '\n';

load data local inpath  '/root/person.txt' into table person; 

drop table data;
create table data (
    name string, 
   	amount int
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n';
 load data local inpath  '/root/data.txt' into table data; 

select count(*) from data;
select count(*) from data group by name;
select name,max(t) from data group by name;
select name,max(t) from data group by name order by max(t) ;

# 本地
set mapreduce.framework.name=local;
set hive.exec.mode.local.auto=true;
# yarn
set mapreduce.framework.name=yarn;
set hive.exec.mode.local.auto=false;

# 向量模式
set hive.vectorized.execution.enabled=true;
set hive.vectorized.execution.enabled=false;

set mapreduce.framework.name=local;
set hive.exec.mode.local.auto=true;
set hive.vectorized.execution.enabled=true; 

操作数据库

创建数据库

-- 创建数据库不指定路径
create database db_hive01;

-- 创建数据库指定 hdfs 路径
create database db_hive02 location '/db_hive02';

-- 创建数据库附加 dbproperties
create database  db_hive03  with dbproperties ('comment'='hive体验','create-date'='2023-11-13','create_author'='lihaozhe');

-- 创建数据表
create table product(
    `product_id`    int COMMENT '商品id',
    `product_name`  string COMMENT '商品名称',
    `category_id`   string COMMENT '所属分类id',
    `from_date`     string COMMENT '上架日期',
    `product_price` decimal(16, 2) COMMENT '商品单价'
) COMMENT '商品信息表';

describe product;
describe extended product;

查询数据库

-- 查看所有数据库
show databases;

-- 模糊查看所有数据库
-- * 代表所有
-- | 代表或
show databases like 'db_hive*';

-- 查看数据库信息
desc database db_hive03;

-- 查看数据库详尽信息
describe database db_hive03;

-- 查看数据库更详尽信息
describe database extended db_hive03;

-- 查看当前在使用哪个数据库
select current_database();

修改数据库

-- 修改 dbproperties
alter database db_hive03 SET dbproperties ('crate_data'='2023-04-18');

-- 修改location
alter database db_hive02 SET location '/db_hive002';

-- 修改 owner user
alter database database_name set owner user lhz;

使用数据库

use db_hive03;

删除数据库

-- 删除空数据库
drop database db_hive02 restrict;

-- 删除非空数据库
drop database db_hive03 cascade;

操作数据表

普通表

临时表 temporary

外部表 external

-- 利用 select 语句查询结果 创建一张表
create table as select

-- 复刻一张已经存在的表结构 但是 不包含数据
create table like

基本数据类型

内部表

create table product(
    `product_id`    int COMMENT '商品id',
    `product_name`  string COMMENT '商品名称',
    `category_id`   string COMMENT '所属分类id',
    `from_date`     string COMMENT '上架日期',
    `product_price` decimal(16, 2) COMMENT '商品单价'
) COMMENT '商品信息表';
show tables;
insert into product values (1,'huawei mate 60',1,'2023-10-25',6999.00);
insert into product values (2,'xiaomi 14',1,'2023-11-11',5999.00);
insert into product values (3,'xiaomi pad 3',2,'2023-11-11',5999.00);

先有鸡还是先有蛋

编写数据文件

vim atm.csv

李昊哲,10000
李胜龙,20000
李大宝,30000

先有表后有数据

创建数据表使用自定义分隔符

create table atm(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

上传数据文件

hdfs dfs -put atm.csv /user/hive/warehouse/db_hive.db/atm
hdfs dfs -cat /user/hive/warehouse/db_hive.db/atm/atm.csv

查询数据表数据

select * from atm;

先有数据后有表

创建数据目录

上传数据文件

hdfs dfs -mkdir -p /user/hive/warehouse/db_hive.db/atm_data
hdfs dfs -put atm.csv /user/hive/warehouse/db_hive.db/atm_data
hdfs dfs -cat /user/hive/warehouse/db_hive.db/atm_data/atm.csv

创建数据表

自定义分隔符

自定义存储路径

create table atm_data(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n'
    location '/user/hive/warehouse/db_hive.db/atm_data';

查询数据表数据

select * from atm_data;

数据加载

查询其它数据表加载

create table atm_01(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

insert into atm_01(name, amount) select name, amount from atm;
insert into atm_01 select name, amount from atm;

select * from atm_01;

通过数据文件加载

通过本地文件系统加载

create table atm_02(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

追加导入

load data local inpath  '/root/atm.csv' into table atm_02; 

select * from atm_02;

覆盖导入

load data local inpath  '/root/atm.csv' overwrite into table atm_02; 

通过HDFS加载

注意：使用该方式 原保存在HDFS上的数据会被移动到新数据表目录 相当于 mv 操作

create table atm_03(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

load data inpath  '/user/hive/warehouse/db_hive.db/atm/atm.csv' into table person; 

select * from atm_03;

使用hive脚本加载数据

create table atm_04(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

create table atm_05(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

create table atm_06(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

create table atm_07(
    name   string,
    amount int
)
    row format delimited fields terminated by ','
    lines terminated by '\n';

hive -e "insert into db_hive.atm_04 select name, amount from db_hive.atm_03;"

select * from db_hive.atm_04;

hive -e "load data local inpath '/root/atm.csv' into table db_hive.atm_05;"

select * from db_hive.atm_05;

编写 sql 文件 load.sql

insert into db_hive.atm_06 select name, amount from db_hive.atm_03;

hive 执行 load.sql 文件

hive -f /root/load.sql

select * from db_hive.atm_06;

编写 脚本文件 load.sh

load data local inpath '/root/atm.csv' into table db_hive.atm_07;

hive 执行 load.sh 文件

hive -f /root/load.sh

select * from db_hive.atm_07;

数据导出

hive -e "select name, amount from db_hive.atm_03;" > /root/atm_031.csv

cat /root/atm_031.csv

insert overwrite local directory '/root/result' select name, amount from db_hive.atm_03;

cat /root/result/000000_0

编写 export.sh 脚本

select name, amount from db_hive.atm_03;

执行 export.sh 脚本

hive -f /root/export.sql  > /root/atm_032.csv

cat /root/atm_032.csv

使用 HDFS 命令

hdfs dfs -ls /user/hive/warehouse/db_hive.db/product

hdfs dfs -cat /user/hive/warehouse/db_hive.db/product/* > /root/product.txt

cat /root/product.txt

复杂数据类型

array

准备数据 person01.txt

李昊哲,大漂亮_大美丽
李胜龙,丽丽_娜娜
李大宝,嘿嘿_哈哈

创建数据表

create table person01(
    name string,
    friends array<string>
)
    row format delimited fields terminated by ','
    collection items terminated by '_'
    lines terminated by '\n';

加载数据

load data local inpath '/root/person01.txt' into table db_hive.person01;

sql 查询

select name, friends,friends[0],friends[1] from person01;

map

准备数据 person02.txt

李昊哲,大漂亮:18
李胜龙,丽丽:16
李大宝,娜娜:17

创建数据表

create table person02(
    name string,
    friend map<string,int>
)
    row format delimited fields terminated by ','
    map keys terminated by ':'
    lines terminated by '\n';

加载数据

load data local inpath '/root/person02.txt' into table db_hive.person02;

sql 查询

select name,friend,friend['丽丽'] from person02;

struct

准备数据 person03.txt

1,李哲:18:0
2,李胜龙:16:1
3,李大宝:17:1

创建数据表

create table person03(
    id int,
    info struct<name:string,age:int,gender:int>
)
    row format delimited fields terminated by ','
    collection items terminated by ':'
    lines terminated by '\n';

加载数据

load data local inpath '/root/person03.txt' into table db_hive.person03;

sql 查询

select id,info,info.name,info.age,info.gender  from person03;

json数据类型

json函数

• ​ get_json_object

• ​ json_tuple

json serde 加载数据

{"name":"user01","amount":"100"}

{
    "name":"lihaozhe",
    "friends":["lizhe","lanlan"，"jinjin"],
    "student":[
        "xiaohuihui":15000,
        "huixiaoxiao":18000
    ],
    "address":{
        "province":"jilin",
        "city":"liaoyuan",
        "district":"dongliao"
    }
}

案例一

create table video01 (info string);
load data local inpath '/root/data.json' into table db_hive.video01;
select info from db_hive.video01;

select
    get_json_object(info,'$.rank') as rank,
    get_json_object(info,'$.title') as title,
    get_json_object(info,'$.dzl') as dzl,
    get_json_object(info,'$.bfl') as bfl,
    get_json_object(info,'$.zfl') as zfl,
    get_json_object(info,'$.type') as type,
    get_json_object(info,'$.time') as paly_time
from db_hive.video01;

select
   json_tuple(info,'rank','title','dzl','bfl','zfl','type','time') as (rank,title,dzl,bfl,zfl,type,pt)
from db_hive.video01;

案例二

create table video(
    rank int,
    title string,
    dzl decimal(16, 2),
    bfl decimal(16, 2),
    zfl decimal(16, 2),
    type string,
    `time` string
)
    row format serde 'org.apache.hive.hcatalog.data.JsonSerDe';

load data local inpath '/root/data.json' into table db_hive.video02;

select info from db_hive.video02;

查看表

-- 查看表
show tables;

-- 查看某数据库下的所有表
show tables in db_hive;

-- 查看表
show tables;

-- 模糊查看数据表
-- * 代表所有
-- | 代表或
show tables like 'atm';

-- 查看基本表信息
describe product;

-- 查看基本表详细信息
describe extended product;

-- 查看基本表详细信息并格式化展示
describe formatted product;

修改表

创建原始表

create table tb_user (
    user_id int,
    user_name string
);

show tables like 'tb_user';

修改表名称

格式 alter table 旧表名称 rename to 新表名称

alter table tb_user rename to  user;

show tables like 'user';

添加字段

格式 alter table 表名称 add columns (字段名称 数据类型)

describe `user`;

alter table `user` add columns (gender tinyint);

describe `user`;

修改字段名称

格式 alter table 表名称 change 旧字段名称 新字段名称 新数据类型 新约束

describe `user`;

alter table `user` change user_id id int;

describe `user`;

修改字段数据类型

格式 alter table 表名称 change 字段名称 字段名称 新的数据类型

describe `user`;

alter table `user` change gender gender smallint;

describe `user`;

修改所有字段

格式 alter table 表名称 replace columns (字段1 数据类型,字段2 数据类型,…)

describe `user`;

alter table `user` replace columns (user_id int, user_name string);

describe `user`;

删除表

drop table tb_user;

清空表

truncate table video;

外部表

create external table data (
    name string, 
   	amount int
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 location '/data';

describe formatted data;

drop table data;

create external table data (
    name string, 
   	amount int
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 location '/data';

Table Type:

• MANAGED_TABLE 内部表

• EXTERNAL_TABLE 外部表

describe formatted `product`;
-- 内部表转外部表
alter table product set tblproperties('EXTERNAL'='TRUE');
describe formatted `product`;
-- 外部表转内部表
alter table product set tblproperties('EXTERNAL'='FALSE');
describe formatted `product`;

hive查询优化

-- 本地
set mapreduce.framework.name=local;
set hive.exec.mode.local.auto=true;
set mapperd.job.tracker=local;
-- yarn
set mapreduce.framework.name=yarn;
set hive.exec.mode.local.auto=false;
set mapperd.job.tracker=yarn

-- 向量模式
set hive.vectorized.execution.enabled=true;
set hive.vectorized.execution.enabled=false;

set hive.exec.mode.local.auto=true;
set mapperd.job.tracker=local;
set mapreduce.framework.name=local;
set hive.vectorized.execution.enabled=true;

注意：在 datagrip 中执行优化后的 hive SQL 可能会有如下报错

Error while processing statement: FAILED: Execution Error, return code 2 from org.apache.hadoop.hive.q.exec.mr.MapRedTas

如果出现以上报错，可以通过 Hadoop 配置文件 yarn-site.xml 文件追加以下内容

  <!-- yarn 允许分配的最大最小内存 -->
  <property>
    <name>yarn.scheduler.minimum-allocation-mb</name>
    <value>1024</value>
  </property>
  <property>
    <name>yarn.scheduler.maximum-allocation-mb</name>
    <value>4096</value>
  </property>
  <property>
    <name>yarn.nodemanager.resource.memory-mb</name>
    <value>4096</value>
  </property>

追加后重启hadoop集群和hive

DQL

准备数据

部门表 dept.csv

10,行政部,1700
20,财务部,1800
30,教学部,1900
40,销售部,2000
50,新媒体,1600

hdfs dfs -mkdir -p /quiz01/dept
hdfs dfs -put /root/data/data02/dept.csv /quiz01/dept

create external table dept (
	dept_id int comment '部门ID',
	dept_name string comment '部门名称',
    location_code int comment '部门位置'
)
 row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/quiz01/dept';

load data local inpath '/root/data/data02/dept.csv' into table dept;

员工表 emp.csv

7369,张三,研发,800.00,30
7499,李四,财务,1600.00,20
7521,王五,行政,1250.00,10
7566,赵六,销售,2975.00,40
7654,侯七,研发,1250.00,30
7698,马八,研发,2850.00,30
7782,金九,行政,2450.0,30
7788,银十,行政,3000.00,10
7839,小芳,销售,5000.00,40
7844,小明,销售,1500.00,40
7876,小李,行政,1100.00,10
7900,小元,讲师,950.00,30
7902,小海,行政,3000.00,10
7934,小红明,讲师,1300.00,30
7934,小红,讲师,1300.00,

hdfs dfs -mkdir -p /quiz01/emp
hdfs dfs -put /root/data/data02/emp.csv /quiz01/emp

create external table emp (
	emp_id int comment '员工ID',
	emp_name string comment '员工姓名',
    emp_job string comment '员工岗位',
    emp_salary decimal(8,2) comment '员工薪资',
    dept_id int comment '员工隶属部门ID'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz01/emp';

load data local inpath '/root/data/data02/emp.csv' into table emp;

订单表 order.csv

30001,2008-09-01,10001
30002,2008-09-12,10003
30003,2008-09-30,10004
30004,2008-10-03,10005
30005,2008-10-08,10001

hdfs dfs -mkdir -p /quiz01/order
hdfs dfs -put /root/data/data01/order.csv /quiz01/order

create external table `order` (
  `o_num` int comment '销售记录编号',
  `o_date` string comment '销售时间',
  `c_id` int comment '消费者编号'
) row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/quiz01/order';

load data local inpath '/root/data/data01/order.csv' overwrite into table `order`;

订单详情表 order_item.csv

30001,1,a1,10,5.2
30001,2,b2,3,7.6
30001,3,bs1,5,11.2
30001,4,bs2,15,9.2
30002,1,b3,2,20
30003,1,c0,100,10
30004,1,o2,50,2.5
30005,1,c0,5,10
30005,2,b1,10,8.99
30005,3,a2,10,2.2
30005,4,m1,5,14.99

hdfs dfs -mkdir -p /quiz01/order_item
hdfs dfs -put /root/data/data01/order_item.csv /quiz01/order_item

create external table `order_item` (
  `o_num` int COMMENT '销售记录编号',
  `o_item` int COMMENT '销售记录流水',
  `f_id` string COMMENT '水果编号',
  `quantity` int COMMENT '数量',
  `item_price` decimal(8,2)COMMENT '单价'
) row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/quiz01/order_item';

load data local inpath '/root/data/data01/order_item.csv' overwrite into table `order_item`;

分类菜单 tb_category.csv

hdfs dfs -mkdir -p /quiz01/tb_category
hdfs dfs -put /root/data/data01/tb_category.csv /quiz01/tb_category

CREATE TABLE `tb_category` (
  `id` int COMMENT '分类ID',
  `name` string COMMENT '分类名称',
  `goods_num` int COMMENT '商品数量',
  `is_show` int COMMENT '是否显示',
  `is_menu` int COMMENT '是否导航',
  `seq` int COMMENT '排序',
  `parent_id` int COMMENT '上级ID',
  `template_id` int COMMENT '模板ID'
) row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/quiz01/tb_category';

load data local inpath '/root/data/data01/tb_category.csv' overwrite into table `tb_category`;

地区表 region.csv

hdfs dfs -mkdir -p /quiz02/region
hdfs dfs -put /root/data/data02/region.csv /quiz02/region

CREATE external TABLE `region` (
  `parent_code` int COMMENT '当前地区的上一级地区代码',  
  `region_code` int COMMENT '地区代码',
  `region_name` varchar(10) COMMENT '地区名称'
) 
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz02/region';

load data local inpath '/root/data/data02/region.csv' into table region;

居民表 person.csv

hdfs dfs -mkdir -p /quiz02/person
hdfs dfs -put /root/data/data02/person.csv /quiz02/person

CREATE external TABLE `person` (
  `id` int COMMENT '主键',
  `id_card` varchar(18) COMMENT '身份证号码',
  `mobile` varchar(11) COMMENT '中国手机号',
  `real_name` varchar(15) COMMENT '身份证姓名',
  `uuid` varchar(32) COMMENT '系统唯一身份标识符'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz02/person';

load data local inpath '/root/data/data02/person.csv' into table person;

分数表 score.csv

hdfs dfs -mkdir -p /quiz02/score
hdfs dfs -put /root/data/data02/score.csv /quiz02/score

CREATE external TABLE `score` (
  `stu_id` string COMMENT '学生ID',
  `course_id` string COMMENT '课程ID',
  `score` string COMMENT '分数'
)
 row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/quiz02/score';
load data local inpath '/root/data/data02/score.csv' overwrite into table score;

load data local inpath '/root/data/data02/score.csv' overwrite into table score;

单表查询

-- 查询所有
select * from dept;

-- 按照指定字段查询
select dept_name from dept;

-- 列别名
select dept_name as name from dept;

-- limit 分页查询
select * from emp limit 5,5;

-- where 按条件查询
select * from emp where dept_id = 10;

-- 关系运算符
-- = != > >= < <=
select * from emp where emp_salary < 2000;

-- in
select * from emp where dept_id in (10,30);

-- not in
select * from emp where dept_id not in (10,30);

-- like
select * from emp where emp_name like '小%';

-- not like
select * from emp where emp_name not like '小%';

-- 逻辑运算符
-- and
select  * from emp where  dept_id = 30 and emp_salary > 1000;

-- between and
select  * from emp where  emp_salary >= 1300 and emp_salary <= 1600;
select  * from emp where  emp_salary between 1300 and 1600;

-- or
select * from emp where dept_id = 10 or dept_id = 30;
select * from emp where emp.dept_id in (10,30);

-- not !
select * from emp where dept_id != 10;
select * from emp where not dept_id = 10;

-- 聚合函数
-- count(*) count(1) count(column_name)
select count(*) from emp;
select count(*) as total from emp;

-- max
select max(emp_salary) from emp;

-- min
select min(emp_salary) from emp;

-- sum
select sum(emp_salary) from emp;

-- avg
select avg(emp_salary) from emp;

-- 聚合函数
-- count(*) count(1) count(column_name)
select count(*) from emp;
select count(*) as total from emp;

-- max
select max(emp_salary) from emp;

-- min
select min(emp_salary) from emp;

-- sum
select sum(emp_salary) from emp;

-- avg
select avg(emp_salary) from emp;
select round(avg(emp_salary),2) from emp;

-- group by 分组查询
select dept_id , round(avg(emp_salary),2) from emp group by dept_id;
-- having
select dept_id, avg(emp_salary) as avg_salary from emp group by dept_id having avg_salary > 2000;

-- where having
select dept_id, round(avg(emp_salary),2) as avg_salary from emp where dept_id != 40 group by dept_id having avg(emp_salary) > 2000;

-- order by 全局排序
select dept_id , round(avg(emp_salary),2) as avg_salary from emp group by dept_id order by avg_salary desc;
-- sort by (每个map或reduce)内部排序
select * from emp sort by dept_id desc;
-- 设置 reduce 数量 仅在当前连接有效 连接断开失效
set mapreduce.job.reduce=3;
select * from emp sort by dept_id desc;
-- 将查询结果写入到文件
insert overwrite local directory '/root/sort-result' select * from emp sort by dept_id desc;

-- distribute by 分区 类似与 mapreduce 中的 partation 自定义分区
set mapreduce.job.reduces=4;
insert overwrite local directory '/root/distribute-result' select * from emp distribute by dept_id sort by emp_salary desc;
-- distribute by 分区规则 根据字段的hash值 与 reduce 的数量 进行相除 余数 相同的在到一个分区
-- hvie 要求 distribute by 语句执行 在 sort by 语句之前
-- 执行结束之后 将 mapreduce.job.reduces 设置为 -1 不然 会影响 分区 分桶 load

-- cluster by 只能升序 不能降序 cluster by = sort by + distribute by
select * from emp cluster by dept_id;

多表查询

-- 多表查询
-- 交叉查询 避免笛卡尔积
select * from dept, emp where dept.dept_id = emp.dept_id;

-- 等值join
select * from dept join emp where dept.dept_id = emp.dept_id;

-- inner join 内连接
select * from dept inner join emp where dept.dept_id = emp.dept_id;

-- left join 左外连接
select * from customer left join `order` o on customer.c_id = o.c_id;

-- right join 右外连接
select * from customer right join `order` o on customer.c_id = o.c_id;

-- full join 全外连接
select * from customer full join `order` o on customer.c_id = o.c_id;

-- union 上下拼接 去重
select * from emp where dept_id = 10 or dept_id = 30 or dept_id = 40;
select * from emp where dept_id in (10,30,40);

select * from emp where dept_id in (10,30) union select * from emp where dept_id in (30,40);

-- union all 上下拼接 不去重
select * from emp where dept_id in (10,30) union all select * from emp where dept_id in (30,40);

-- 自关联
select * from region where region_code = '220422';

子查询

-- 子查询作为查询条件
select * from tb_category;
select count(*) from tb_category;
-- 一级导航
select * from tb_category where parent_id = 0;
-- 二级导航
select * from tb_category where parent_id in (select id from tb_category where parent_id = 0);

-- 三级导航
select * from tb_category where parent_id in (
    select id from tb_category where parent_id in (
        select id from tb_category where parent_id = 0
    )
);

-- 子查询作为数据源 也就是 当中一张数据表看待
-- 一级导航
select * from tb_category where parent_id = 0;

-- 二级导航
select t1.id, t1.name, t1.goods_num, t1.is_show, t1.is_menu, t1.seq, t1.parent_id, t1.template_id from tb_category t1
    inner join
    (select id, name, goods_num, is_show, is_menu, seq, parent_id, template_id from tb_category where parent_id = 0) t2
    on t1.parent_id = t2.id;
    
-- 三级导航
select t0.id, t0.name, t0.goods_num, t0.is_show, t0.is_menu, t0.seq, t0.parent_id, t0.template_id from tb_category t0
    inner join
    (select t1.id, t1.name, t1.goods_num, t1.is_show, t1.is_menu, t1.seq, t1.parent_id, t1.template_id from tb_category t1
        inner join
        (select id, name, goods_num, is_show, is_menu, seq, parent_id, template_id from tb_category where parent_id = 0) t2
        on t1.parent_id = t2.id) t3
    on t0.parent_id = t3.id;
    
-- 将一级导航和该导航下的二级导航数据封装一次性送到前端
select id, name, goods_num, is_show, is_menu, seq, t4.parent_id, template_id, second_ids, second_names from tb_category t4
    inner join
    (select parent_id,collect_list(t2.id) second_ids,collect_list(t2.name) second_names from tb_category t2
        inner join
       (select id from tb_category where parent_id = 0) t1
        on t2.parent_id = t1.id group by parent_id) t3
    on t3.parent_id = t4.id;

函数

-- 算数运算函数
-- + - * / % & | ^ ~

-- 数值函数
-- round 四舍五入
select rount(3.3) as num;
select rount(3.5) as num;

-- ceil 向上取整
select ceil(3.3) as num;

-- floor 向下取整
select floor(3.9) as num;

-- mod
select mod(9,2);

-- 字符串函数
-- substr 字符串截取 substr(column_name,start_index,length)
select substr(id_card,3,3),id_card from person;

-- substring 字符串截取 substring(column_name,start_index,length)
select substring(id_card,3,3),id_card from person;

-- split 字符串分割
select split('2023-04-19','-')

-- nvl 判空 替换 null 值
select nvl("李昊哲",1)
select nvl(null,1)

-- concat 字符串拼接
select  concat('slogan',' - ','桃李不言下自成蹊')

-- concat 字符串拼接
select  concat_ws('-',array('2022','04','19'))

-- get_json_object 解析 json 字符串
select get_json_object('{"name":"李昊哲","age":41}','$.name') as name;

select get_json_object('[
{"name":"李昊哲","age":41},
{"name":"李哲","age":16}
]','$.[0].name') as name;

-- json_tuple
select json_tuple('{"name":"李昊哲","age":41}','name','age') as (name,age);

-- 日期函数
-- unix_timestamp
select unix_timestamp();
select unix_timestamp('1983-11-22 20:30:00','yyyy-MM-dd HH:mm:ss');

-- from_unixtime
select from_unixtime(438381000);

-- current_date
select current_date();

-- current_timestamp
select current_timestamp();

-- year
select year('1983-11-22 20:30:00');

-- month
select month('1983-11-22 20:30:00');

-- day
select day('1983-11-22 20:30:00');

-- hours
select year('1983-11-22 20:30:00');

-- minute
select minute('1983-11-22 20:30:00');

-- second
select second('1983-11-22 20:30:00');

-- datediff 两个日期相差天数 (结束日期减去开始日期)
select datediff('1983-11-22','1983-10-18');

-- date_add 日期增加几天
select date_add('1983-11-22',3);

-- date_sub 日期减少几天
select date_sub('1983-11-22',3);

-- date_format 日期格式化
select date_format('1983-11-22 20:30:00','yyyy年MM月dd日 HH时mm分ss秒');

-- 读取身份证获取出生日期 输出格式为unix_stamp
-- 1、字符串截取 2、日期转换
select substr(id_card,7,8) from person limit 3;
select unix_timestamp(substr(id_card,7,8),'yyyyMMdd') from person limit 3;

-- 读取身份证获取出生日期 输出格式为 yyyy-MM-dd
-- 1、字符串截取 2、日期格式化
select substr(id_card,7,8) from person limit 3;
select unix_timestamp(substr(id_card,7,8),'yyyyMMdd') from person limit 3;
select from_unixtime(unix_timestamp(substr(id_card,7,8),'yyyyMMdd')) from person limit 3;

select substr(from_unixtime(unix_timestamp(substr(id_card,7,8),'yyyyMMdd')),1,10) from person limit 3;
select date_format(from_unixtime( unix_timestamp(substr(id_card,7,8),'yyyyMMdd')),'yyyy-MM-dd') from person limit 3;

-- 流程控制函数
-- case when
-- 90以上A 80~90B 70~80C 60~70D 60以下E
select
stu_id,course_id,
case
when score >= 90 then 'A'
when score >= 80 then 'B'
when score >= 70 then 'C'
when score >= 60 then 'D'
else 'E'
end as grade
from score;

-- if 三目运算 if(条件表达式,条件为真的返回结果,条件为假的返回结果)
select if(1=2,'托尼','玛丽') as `发型师`

-- 结合字符串函数 时间函数 流程控制函数 计算身份证信息
-- 根据身份证号 判断性别 身份证号 第十七位 奇数为男性 偶数为女性
select id_card, if(mod(substr(id_card,17,1),2) = 1,'精神小伙儿','扒蒜老妹儿') gender from person;

-- 根据身份证号 找出所有男性信息
select *, mod(substr(id_card,17,1),2) gender from person where mod(substr(id_card,17,1),2) = 1;

-- 根据身份证号 计算男性人数和女性人数
select 
if(mod(substr(id_card,17,1),2) = 1,'精神小伙儿','扒蒜老妹儿') gender ,
count(*) gender_count
from person group by mod(substr(id_card,17,1),2) limit 10;

-- 根据身份证号 计算生日排序
select
date_format(from_unixtime( unix_timestamp(substr(id_card,7,8),'yyyyMMdd')),'yyyy-MM-dd') as `birthday`
from person
order by unix_timestamp(`birthday`,'yyyy-MM-dd') desc

-- 根据身份证号 计算年龄
-- 1、当前月份-出生月份 > 0 说明 已经过完生日 及 使用 当前年份 - 出生年份 = 年龄
-- 2、当前月份-出生月份 < 0 说明 未过生日 及 使用 当前年份 - 出生年份 -1 = 年龄
-- 3、当前月份-出生月份 = 0 
-- 3.1、当前日-出生日 > 0 说明 已经过完生日 及 使用 当前年份 - 出生年份 = 年龄
-- 3.2、当前日-出生日 < 0 说明 未过生日 及 使用 当前年份 - 出生年份 -1 = 年龄
-- 3.3、当前日-出生日 = 0 说明 生日视作过完了 及 使用 当前年份 - 出生年份  = 年龄
select if(month(`current_date`()) - month(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))) > 0,
      year(`current_date`()) - year(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))),
      if(month(`current_date`()) - month(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))) < 0,
         year(`current_date`()) - year(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))) - 1,
         if(day(`current_date`()) - day(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))) > 0,
            year(`current_date`()) - year(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))),
            if(day(`current_date`()) - day(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))) < 0,
               year(`current_date`()) - year(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd'))) - 1,
               year(`current_date`()) - year(from_unixtime(unix_timestamp(substr(id_card, 7, 8), 'yyyyMMdd')))
                )
             )
          )
       ) as `age`
from person;

-- 集合函数
-- size
-- array 声明一个集合
select array(0,1,2,3,4) as nums;

-- array_contains 判断 array 中是否包含某元素
select array_contains(array(0,1,2,3,4),3) as num;

-- sort_array array中元素排序 目前只有升序
select sort_array(array(4,1,3,2,0)) as nums;

-- size 集合中元素的数量
select size(array(0,1,2,3,4)) as nums;

-- map
select map('校长',10000,'讲师',8000);
-- {"校长":10000,"讲师":8000} 

-- map_keys 返回 map 中 所有的 key
select map_keys(`map`('校长',10000,'讲师',8000)) as keys;

-- map_values 返回 map 中 所有的 value
select map_values(`map`('校长',10000,'讲师',8000)) as v

-- struct 声明结构体属性名称
select struct('real_name','李昊哲','age',41);
-- {"col1":"real_name","col2":"李昊哲","col3":"age","col4":41} 

-- named_struct 声明结构体属性和值
select named_struct('real_name','李昊哲','age',41);
-- {"real_name":"李昊哲","age":41}

综合练习

练习数据

学生表

讲师表

课程表

分数表

学生表 student.csv

hdfs dfs -mkdir -p /quiz03/student
hdfs dfs -put /root/data/data02/student.csv /quiz03/student

create external table student (
	stu_id string comment '学生ID',
	stu_name string comment '学生姓名',
    birthday string comment '出生日期',
    gender string comment '学生性别'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz03/student';

load data local inpath '/root/data/data02/student.csv' overwrite into table student;

讲师表 teacher.csv

hdfs dfs -mkdir -p /quiz03/teacher
hdfs dfs -put /root/data/data02/teacher.csv /quiz03/teacher

create external table teacher (
	tea_id string comment '课程ID',
	tea_name string comment '课程名称'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz03/teacher';

load data local inpath '/root/data/data02/teacher.csv' overwrite into table teacher;

课程表 course.csv

hdfs dfs -mkdir -p /quiz03/course
hdfs dfs -put /root/data/data02/course.csv /quiz03/course

create external table course (
	course_id string comment '课程ID',
	course_name string comment '课程名称',
    tea_id string comment '讲师ID'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz03/course';

load data local inpath '/root/data/data02/course.csv' overwrite into table course;

分数表 score.csv

hdfs dfs -mkdir -p /quiz03/score
hdfs dfs -put /root/data/data02/score.csv /quiz03/score

create external table score (
	stu_id string comment '学生ID',
	course_id string comment '课程ID',
    score int comment '成绩'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz03/score';

load data local inpath '/root/data/data02/score.csv' overwrite into table score;

练习题目

-- 查询所有学生信息
select * from student;

-- 查询周姓学生信息
select * from student where stu_name like '周%';

-- 查询周姓学生数量
select count(*) from student where stu_name like '周%';

-- 查询 学生ID 004 的分数 超过 85 的成绩
select stu.stu_id, stu_name, birthday, gender, course_id, score from student stu inner join score s
   on stu.stu_id = 004 and stu.stu_id = s.stu_id and score > 85;

-- 查询 学生程ID 004 的成绩降序
select stu.stu_id, stu_name, birthday, gender, course_id, score from student stu inner join score s
   on stu.stu_id = 004  and  stu.stu_id = s.stu_id order by score desc;

-- 查询 数学成绩不及格学生及其对应的成绩 学生成绩
select stu.stu_id, stu_name, birthday, gender, course_name, score from student stu
    inner join score s inner join course c on s.course_id = c.course_id and  stu.stu_id = s.stu_id
    and c.course_name = '数学' and score < 60 order by score;

-- 查询男女生人数
select gender,count(*) from student group by gender;

-- 查询编号为 02 的课程平均成绩
select round(avg(score),2) from score where course_id = 02;

-- 查询每科课程平均成绩
select course_id,round(avg(score),2) from score group by course_id;

-- 查询参加考试学生人数
select count(distinct stu_id) from score where score is not null and score >= 0;
select count(stu_id) from (select stu_id from score where score is not null and score >= 0 group by stu_id) t;

-- 查询每科有多少学生参加考试
select course_id,count(stu_id) from score where score is not null and score >= 0 group by course_id;

-- 查询未参加考试的学生信息
select * from student where stu_id in (
    select stu.stu_id from student stu
        left join score s on stu.stu_id = s.stu_id
        group by stu.stu_id having count(*) != (select count(*) from course)
);

-- 查询选修至少 4 门 以上课程学生的学号
select stu_id,count(course_id) course_count from score
 group by stu_id
 having course_count >= 4;

-- 查询姓氏相同学生名单 并且同姓人数大于 2 的姓氏
select first_name ,count(*) first_name_count from (
 select stu_id,stu_name,substr(stu_name,1,1) as first_name
 from student
) ts
 group by ts.first_name
 having first_name_count > 1;

-- 查询每门功课的学生的平均成绩 按照平均成绩升序 平均成绩相同按照课程编号降序
select course_id, avg(score) avg_score
 from score
 group by course_id
 order by avg_score,course_id desc;

-- 统计参加考试人数大于等于 15 的学科
select course_id,count(*) as stu_count from score group by course_id having stu_count > 15;

-- 查询学生总成绩并按照总成绩降序排序
select stu_id, sum(score) sum_score
 from score
 group by stu_id
 order by sum_score desc;

-- 按照指定格式显示 stu_id 语文 数学 英语 选课数 平均成绩
select
 s.stu_id,
 sum(`if`(c.course_name='语文',score,0)) as `语文`,
 sum(`if`(c.course_name='数学',score,0)) as `数学`,
 sum(`if`(c.course_name='英语',score,0)) as `英语`,
 count(s.course_id) as `选课数`,
 avg(s.score) as `平均成绩`
 from course c left join score s
 on c.course_id = s.course_id
 group by s.stu_id
 order by `平均成绩` desc;

-- 查询一共参加了三门功课且其中一门为语文的学生id 和 姓名
select s.stu_id,stu_name from
(select t1.stu_id ,count(t1.course_id) course_count  from
	(select stu_id,course_id from score
		where stu_id in ( select stu_id from score where course_id = "01")
	) t1 group by  t1.stu_id having course_count >=3
) t2 join student s on t2.stu_id = s.stu_id;
-- 分解
-- 查询该学生的姓名
select s.stu_id,stu_name from
-- 成绩表中学习科目数量 >=3 科的学生
(select t1.stu_id ,count(t1.course_id) course_count  from
 	--  报名了语文的学生还报名了那些学科
	(select stu_id,course_id from score
		where stu_id in (
            -- 查询报名了语文的学生ID
            select stu_id from score where course_id = "01"
        )
	) t1 group by  t1.stu_id having course_count >=3
) t2 join student s on t2.stu_id = s.stu_id;

-- 查询两门以上的课程不及格学生的学号及其平均成绩
-- 1、先按照学生分组 过滤出成绩低于60的数量 大于1
-- 2、计算所有学生的平均成绩
-- 3、两个子查询相互join
select  t1.stu_id,t2.avg_score from
(select stu_id, sum(if(score < 60, 1, 0)) as result from score group by stu_id having result > 1) t1
 left join
(select stu_id,avg(score) as avg_score from score group by stu_id) t2 on t1.stu_id =t2.stu_id;

-- 查询所有学生的学号、姓名、选课数、总成绩
select
    stu.stu_id,stu.stu_name,count(s.course_id) count_course ,nvl(sum(s.score),0) total_score
from student stu left join score s on stu.stu_id = s.stu_id
group by stu.stu_id, stu.stu_name order by stu.stu_id;

-- 平均成绩大于 85 的所有学生的学号、姓名、平均成绩
select
    stu.stu_id,stu.stu_name ,nvl(avg(s.score),0) as `avg_score`
from student stu left join score s on stu.stu_id = s.stu_id
group by stu.stu_id, stu.stu_name having nvl(avg(s.score),0) > 85 order by stu.stu_id

-- 查询学生的选课情况：学号，姓名，课程号，课程名称
select student.stu_id,student.stu_name,c.course_id,c.course_name from student
right join score s on student.stu_id = s.stu_id
left join course c on s.course_id = c.course_id

-- 查询学生的没有选课情况：学号，姓名
select stu_id,stu_name from
(
select student.stu_id,student.stu_name, s.course_id from student
left join score s on student.stu_id = s.stu_id
left join course c on s.course_id = c.course_id
) t where course_id is null

-- 查询出每门课程的及格人数和不及格人数
select c.course_id,course_name,pass,fail
from course c join
(
select
 course_id,sum(if(score >= 60,1,0)) as `pass`, sum(if(score < 60,1,0)) as `fail`
 from score group by course_id
) t on c.course_id = t.course_id

-- 查询课程编号为03且课程成绩在80分以上的学生的学号和姓名及课程信息
select t1.stu_id,s.stu_name,t1.course_id,c.course_name,t1.score from
(select * from score where course_id = '03' and score > 80) t1
left join student s on s.stu_id = t1.stu_id
left join course c on t1.course_id = c.course_id

-- 查询语文成绩低于平均分数的学生是谁，教师是谁
select t3.stu_id,t3.stu_name,t3.`avg_score`,t.tea_name from
(select t2.stu_id,t2.`avg_score`,s.stu_name,t2.course_id,c.tea_id from
    (select t1.stu_id,t1.course_id,t1.`avg_score` from
        (select stu_id,s.course_id, avg(score) as `avg_score` from score s right join
         (select course_id from course where course_name = '语文') t1 on t1.course_id = s.course_id
         group by stu_id,s.course_id) t1
        where t1.`avg_score` < (select avg(score) as `avg_score` from score s right join (select course_id from course where course_name = '语文') t1 on t1.course_id = s.course_id)
    ) t2 left join student s on t2.stu_id = s.stu_id
    left join course c on t2.course_id = c.course_id
)t3 left join teacher t on t3.tea_id = t.tea_id;

-- 查询所有学生总成绩和平均成绩，
-- 且他们的总成绩低于平均成绩的有多少个人，
-- 高于平均成绩的有多少人，
-- 低于平均成绩的男生和女生分别有多少人，
-- 且他们的任课老师是谁。

-- 统计各科成绩各分数段人数：课程编号,课程名称,[100-85],[85-75],[70-60],[0-60]及所占百分比
select c.course_id, course_name, a, b, c, d from course c left join (
    select course_id,
       concat(round((sum(`if`(score >= 85,1,0)) / count(*)) * 100,2), '%') as a,
       concat(round((sum(`if`(score between 75 and 84,1,0)) / count(*)) * 100,2), '%') as b,
       concat(round((sum(`if`(score between 60 and 74,1,0)) / count(*)) * 100,2), '%') as c,
       concat(round((sum(`if`(score < 60,1,0)) / count(*)) * 100,2), '%') as d
    from score group by course_id
) t on t.course_id = c.course_id;

-- 查询各科成绩最高分、最低分和平均分，以如下形式显示：
-- 课程ID，课程name，最高分，最低分，平均分，中下率，中等率，优良率，优秀率
-- 及格为>=60，中等为：70-80，优良为：80-90，优秀为：>=90
select c.course_id, course_name, `最高分`,`最低分`,`平均分`,`优秀率`, `优良率`, `中等率`, `中下率`,`芸芸众生` from course c left join (
    select course_id,
           max(score) `最高分`,
           min(score) `最低分`,
           round(avg(score),2) `平均分`,
       concat(round((sum(`if`(score >= 90,1,0)) / count(*)) * 100,2), '%') as `优秀率`,
       concat(round((sum(`if`(score between 80 and 89,1,0)) / count(*)) * 100,2), '%') as `优良率`,
       concat(round((sum(`if`(score between 70 and 79,1,0)) / count(*)) * 100,2), '%') as `中等率`,
       concat(round((sum(`if`(score between 60 and 69,1,0)) / count(*)) * 100,2), '%') as `中下率`,
       concat(round((sum(`if`(score < 60,1,0)) / count(*)) * 100,2), '%') as `芸芸众生`
    from score group by course_id
) t on t.course_id = c.course_id;

-- 查询每门课程的男生和女生的比例是多少
select t1.course_id,t1.gender, concat(round((count_gender / count_course_student) * 100,2), '%') as proportion from
    (select course_id,gender,count(*) count_gender from score s1 inner join student s2 on s1.stu_id = s2.stu_id group by course_id,gender) t1
    inner join
    (select course_id,count(*) count_course_student from score s1 inner join student s2 on s1.stu_id = s2.stu_id group by course_id) t2
    on t2.course_id = t1.course_id;

-- 每门学科的成绩是男生比较优一些还是女生比较优一些，并且每门课程的最高分是谁。
select battle_t.course_id,male_avg_score, female_avg_score, battle, max_stu_id,min_stu_id, max_score, min_score from
    (select male_t.course_id,round(male_avg_score,2) male_avg_score,round(female_avg_score,2) female_avg_score,
           case
               when male_avg_score > female_avg_score then '男性优秀'
               when male_avg_score < female_avg_score then '女性优秀'
               else '势均力敌'
           end battle
        from
        (select course_id,avg(score) male_avg_score from score s1 inner join student s2 on s1.stu_id = s2.stu_id and gender = '男' group by course_id) male_t
        inner join
        (select course_id,avg(score) female_avg_score from score s1 inner join student s2 on s1.stu_id = s2.stu_id and gender = '女' group by course_id) female_t
        on male_t.course_id = female_t.course_id) battle_t
    inner join
    (select max_t.course_id,max_t.stu_id max_stu_id,max_score,min_t.stu_id min_stu_id,min_score from
        (select stu_id, s.course_id, max_score from score s
                inner join
                (select course_id, max(score) max_score from score group by course_id) t
                on s.course_id = t.course_id and max_score = score) max_t
        full join
        (select stu_id, s.course_id, min_score from score s
                inner join
                (select course_id,min(score) min_score from score group by course_id) t
                on s.course_id = t.course_id and min_score = score) min_t
        on max_t.course_id = min_t.course_id) info
    on battle_t.course_id = info.course_id;

-- 课程编号为"01"且课程分数小于60，按分数降序排列的学生信息
select s.stu_id, stu.stu_name, stu.birthday, stu.gender,s.score
    from score s join student stu on s.stu_id = stu.stu_id
    where s.score < 60  order by s.score desc;

-- 查询所有课程成绩在70分以上的学生的姓名、课程名称和分数，按分数升序
select stu.stu_name, c.course_name, s2.score
    from student stu join
    (select s.stu_id, sum(`if`(s.score >= 70, 0, 1)) as `is_ok` from score s group by s.stu_id having is_ok = 0) t1
    on stu.stu_id = t1.stu_id left join score s2 on stu.stu_id = s2.stu_id left join course c on s2.course_id = c.course_id
    order by s2.score;

-- 查询某学生不同课程的成绩相同的学生编号、课程编号、学生成绩
select s1.stu_id,collect_list(s1.course_id) as course_id,collect_set(s1.score) as score
    from score s1 join score s2 on s1.stu_id = s2.stu_id
    and s1.course_id != s2.course_id
    and s1.score == s2.score
    group by s1.stu_id;

分组排序取TopN

-- 查询各科成绩前五名的学生
select a.course_id,a.stu_id,a.score from score a
    left join score b
    on a.course_id = b.course_id and a.score <= b.score
    group by a.stu_id,a.course_id,a.score
    having count(a.stu_id) <=5
    order by a.course_id,a.score desc;

select S1.course_id,s1.stu_id,s1.score from score s1 where
    (select count(*) from score s2
        where s2.course_id=s1.course_id AND s2.score > s1.score
        ) <= 5 order by s1.course_id,s1.score desc;

-- row_number() over () 连续序号
-- over()里头的分组以及排序的执行晚于 where 、group by、order by 的执行。
select * from
    (select course_id, stu_id,  score,
        row_number() over (partition by course_id order by score desc ) as mum
        from score) t where mum <= 5;

-- rank() over () 排名 跳跃排序 序号不是连续的
select * from
    (select course_id, stu_id,  score,
        rank() over (partition by course_id order by score desc ) as mum
        from score) t where mum <= 5;

-- dense_rank() over () 排名 连续排序
select * from
    (select course_id, stu_id,  score,
        dense_rank() over (partition by course_id order by score desc ) as mum
        from score) t where mum <= 5;

行列转换

列转行

create table test (
    stu_name string,
    course_name string,
    score int
);
insert into test values ('张三','语文','80'),('张三','数学','90'), ('李四','语文','85'),('李四','数学','95');
select * from test;

select stu_name,
       case when course_name = '语文' then score end as yuwen,
       case when course_name = '数学' then score end as shuxue
       from test;

-- null 在 比较的时候 是最小值
select stu_name,
       max(case when course_name = '语文' then score end) as yuwen,
       max(case when course_name = '数学' then score end) as shuxue
       from test group by stu_name;

-- 员工表 emp.csv
create external table emp (
	emp_id int comment '员工ID',
	emp_name string comment '员工姓名',
    emp_job string comment '员工岗位',
    emp_salary decimal(8,2) comment '员工薪资',
    dept_id int comment '员工隶属部门ID'
)
 row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/quiz01/emp';
load data local inpath '/root/data/data02/emp.csv' overwrite into table emp;
select * from emp;

-- 多个值转为集合 collect_list 不会去重
select collect_list(emp_job) job_list from emp;
-- 多个值转为集合 collect_set 会去重
select collect_set(emp_job) job_set from emp;
-- size 获取结合中元素的数量
select size(collect_set(emp_job)) job_count from emp;
-- concat_ws 将多个数据 以分隔符形式 拼接 concat_ws(分隔符,数据1,数据2,...)
select concat_ws('-',collect_set(emp_job)) job_string from emp;
-- split 字符串切分 以分隔符切分字符串 为集合
select split(concat_ws('-',collect_set(emp_job)),'-') job_item from emp;

行转列

create table sales (
    emp_name string,
    january int,
    february int,
    march int
);
insert into sales values ('张三',1000,2000,3000),('李四',1500,2500,3500);
select * from sales;

select emp_name,'january' yue, january sale from sales
union all
select emp_name,'february' yue,february sale from sales
union all
select emp_name,'march' yue,march sale from sales;

select t1.emp_name,
    yue_list[0] january,
    yue_list[1] february,
    yue_list[2] march
    from(
        select t.emp_name,collect_list(yue) yue_list from(
            select emp_name,january yue from sales
            union all
            select emp_name,february yue from sales
            union all
            select emp_name,march yue from sales) t
            group by t.emp_name) t1;

-- collect_list
select collect_list(emp_job) as `job` from emp;

-- collect_set
select collect_set(emp_job) as `job` from emp;

-- concat_ws
select concat_ws(',',collect_set(emp_job)) as `job` from emp;

-- split
select split(concat_ws(',',collect_set(emp_job)),',') as `job_list` from emp;

用户函数

• UDF，即用户定义函数(user-defined function)，作用于单行数据，并且产生一个数据行作为输出。

Hive中大多数函数都属于这一类，比如数学函数和字符串函数。UDF函数的输入与输出值是1:1关系。

• UDTF，即用户定义表生成函数（user-defined table-generating function），

作用于单行数据，并且产生多个数据行。UDTF函数的输入与输出值是1:n的关系。

• UDAF，用户定义聚集函数（user-defined aggregate function），作用于多行数据，产生一个输出数据行。
  Hive中像COUNT、MAX、MIN和SUM这样的函数就是聚集函数。UDAF函数的输入与输出值是n:1的关系。

雇员表 employee.csv

hdfs dfs -mkdir -p /quiz04/employee
hdfs dfs -put /root/data/data02/employee.csv /quiz04/employee

create external table employee(
    name string comment '姓名',
    sex  string comment '性别',
    birthday string comment '出生年月',
    hiredate string comment '入职日期',
    job string comment '岗位',
    salary int comment '薪资',
    bonus int comment '奖金',
    friends array<string> comment '朋友',
    children map<string,int> comment '孩子'
)
 row format delimited fields terminated by ',' 
 collection items terminated by '_' 
 map keys terminated by ':' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz04/employee';

load data local inpath '/root/data/data02/employee.csv' into table employee;

UDTF

-- explode
select explode(array('java','python','scala','go')) as course;
select explode(map('name','李昊哲','gender','1')) as (key,value);

-- posexplode
select posexplode(array('java','python','scala','go')) as (pos,course);

-- inline
select inline(array(named_struct('id',1,'name','李昊哲','gender','1'),
                   named_struct('id',2,'name','李哲','gender','0'),
                   named_struct('id',3,'name','李大宝','gender','1'))) as (id,name,gender);

-- lateral view   
select * from employee lateral view explode(friends) t as friend;

select e.name,e.friends,t1.friend from employee e lateral view explode(friends) t1 as `friend`;

select * from employee e lateral view explode(children) t1 as `children_name`,`children_friend_count`;

select e.name,e.children,t1.children_name,t1.nvl(t2.children_friend_count,0) from employee e
lateral view explode(children) t1 as `children_name`,`children_friend_count`;

select e.name,e.friends,e.children,t1.friend,t2.children_name,nvl(t2.children_friend_count,0) from employee e
lateral view explode(friends) t1 as `friend`
lateral view explode(children) t2 as `children_name`,`children_friend_count`;

-- lateral view outer

电影表 movie.txt

hdfs dfs -mkdir -p /quiz04/movie
hdfs dfs -put /root/data/data02/movie.txt /quiz04/movie

create external table movie(
    name string comment '电影名称',
    category string comment '电影分类'
) 
 row format delimited fields terminated by '-' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz04/movie';

load data local inpath '/root/data/data02/movie.txt' into table movie;

UDTF 案例

-- 根据上述电影信息表，统计各分类的电影数量
select cate,count(name) as `quantity` from movie
 lateral view explode(split(category,',')) tmp as cate
 group by cate;

分组和去重

-- 统计岗位数量
select count(distinct emp_job) from emp;
select count(*) from (select emp_job from emp group by emp_job) t;

窗口函数(开窗函数)

能为每行数据划分一个窗口，然后对窗口范围内的数据进行计算，最后将计算结果返回给该行

Function(arg1,..., argn) OVER ([PARTITION BY <...>] [ORDER BY <....>] [<window_expression>])
-- 其中Function(arg1,..., argn) 可以是下面分类中的任意一个
    -- 聚合函数：比如sum max min avg count等
    -- 分析函数：比如lead lag first_value last_value等
    -- 排序函数：比如row_number rank dense_rank等    
-- OVER [PARTITION BY <...>] 类似于group by 用于指定分组  每个分组你可以把它叫做窗口
-- 如果没有PARTITION BY 那么整张表的所有行就是一组
-- [ORDER BY <....>]  用于指定每个分组内的数据排序规则 支持ASC、DESC
-- [<window_expression>] 用于指定每个窗口中 操作的数据范围 默认是窗口中所有行

hdfs dfs -mkdir /quiz04/order
hdfs dfs -put /root/data/data02/order.csv /quiz04/order

create external table `order`
(
    order_id     string comment '订单id',
    user_id      string comment '用户id',
    user_name    string comment '用户姓名',
    order_date   string comment '下单日期',
    order_amount int comment '订单金额'
)
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/quiz04/order';

load data local inpath '/root/data/data02/order.csv' into table `order`;

聚合函数

rows 基于行

range 基于值

函数() over(rows between and 3)

• unbounded preceding 表示从前面的起点
• number preceding 往前
• current row 当前行
• number following 往后
• unbounded following 表示到后面的终点

-- 统计每个用户截至每次下单的累计下单总额

select *,
       sum(order_amount) over (
           partition by user_id ,substr(order_date,1,7)
           order by order_date
           rows between unbounded preceding and current row
           ) sum_order_amount
       from `order`;

select *,
       sum(order_amount) over (
           partition by user_id ,substr(order_date,1,7)
           order by order_date
           rows unbounded preceding
           ) sum_order_amount
       from `order`;

-- 统计每个用户截至每次下单的当月累积下单总额
select *,
       sum(order_amount) over (
           partition by user_id ,substr(order_date,1,7)
           order by order_date
           rows between unbounded preceding and unbounded following
           ) sum_order_amount
       from `order`;

-- 最近三笔订单总金额
-- 当前订单金额与前两笔订单金额的总和
select *,
       sum(order_amount) over (
           partition by user_id
           order by order_date
           rows 2 preceding
           ) sum_order_amount
       from `order`;

-- 当前订单金额与后两笔订单金额的总和
select *,
       sum(order_amount) over (
           partition by user_id
           order by order_date
           rows between current row and 2 following
           ) sum_order_amount
       from `order`;

-- 当前订单金额与前一笔订单和后一笔订单金额的总和
select *,
       sum(order_amount) over (
           partition by user_id
           order by order_date
           rows between 1 preceding and 1 following
           ) sum_order_amount
       from `order`;

分析函数

lag lead

lag() over() 与 lead() over() 函数是跟偏移量相关的两个分析函数，
通过这两个函数可以在一次查询中取出同一字段的前 N 行的数据 (lag) 和后 N 行的数据 (lead) 作为独立的列,
从而更方便地进行进行数据过滤。这种操作可以代替表的自联接，并且 LAG 和 LEAD 有更高的效率。
over() 表示 lag() 与 lead() 操作的数据都在 over() 的范围内，可以使用 partition by 语句（用于分组） order by 语句（用于排序）。
partition by a order by b 表示以 a 字段进行分组，再 以 b 字段进行排序，对数据进行查询。

例如：lag(field, num, defaultvalue) field 需要查找的字段，num 往前查找的 num 行的数据，defaultvalue 没有符合条件的默认值

例如：lead(field, num, defaultvalue) field 需要查找的字段，num 往后查找的 num 行的数据，defaultvalue 没有符合条件的默认值

first_value last_value

irst_value 取每个分区内某列的第一个值
语法：first_value(col,true/false) over (partition by col1 order by col2)
第二个参数为true，跳过空值（默认为false）

ast_value 取每个分区内某列的最后一个值
语法：last_value(col,true/false) over (partition by col1 order by col2)
第二个参数为true，跳过空值（默认为false）

lead lag

-- 统计每个用户每次下单距离上次下单相隔的天数（首次下单按0天算）
select order_id, user_id, user_name, order_date, order_amount, datediff(order_date,pre_order_date) diff_data from (
    select order_id, user_id, user_name, order_date, order_amount,
       lag(order_date,1,order_date) over (partition by user_id order by order_date) pre_order_date
       from `order`) t;

-- 每个用户首笔订单时间
select order_id, user_id, user_name, order_date, order_amount from (
    select order_id, user_id, user_name, order_date, order_amount,
       lag(order_date,1,order_date) over (partition by user_id order by order_date) pre_order_date
       from `order`) t where datediff(order_date,pre_order_date) = 0;

-- 每个用户每个月首笔订单时间
select order_id, user_id, user_name, order_date, order_amount from (
    select order_id, user_id, user_name, order_date, order_amount,
       lag(order_date,1,order_date) over (partition by user_id,substr(order_date,1,7) order by order_date) pre_order_date
       from `order`) t where datediff(order_date,pre_order_date) = 0;
       
-- 每个用户每个月最后笔订单时间
select order_id, user_id, user_name, order_date, order_amount from (
    select order_id, user_id, user_name, order_date, order_amount,
       lead(order_date,1,order_date) over (partition by user_id,substr(order_date,1,7) order by order_date) next_order_date
       from `order`) t where datediff(order_date,next_order_date) = 0;
       
-- 每个岗位先先入职的远哥和后入在的员工工资差
select name, sex, birthday, hiredate, job, salary, bonus, friends, children, new_salary,(salary - new_salary) salary_diff from (
    select name, sex, birthday, hiredate, job, salary, bonus, friends, children,
       lead(salary,1,salary) over (partition by job order by hiredate) new_salary
       from employee) t;

first_value last_value

-- 查询所有下单记录以及每个用户的每个下单记录所在月份的首/末次下单日期
select order_id, user_id, user_name, order_date, order_amount, first_order_date, last_order_date from (
    select order_id, user_id, user_name, order_date, order_amount,
        first_value(order_date) over(partition by user_id,substr(order_date,1,7) order by order_date) `first_order_date`,
        last_value(order_date) over (partition by user_id,substr(order_date,1,7) order by order_date
            rows between current row and unbounded following) `last_order_date`
    from `order`)t
    where t.order_date = first_order_date or t.order_date = last_order_date;

排序函数

-- 查询各科成绩前五名的学生
select a.course_id,a.stu_id,a.score from score a
    left join score b
    on a.course_id = b.course_id and a.score <= b.score
    group by a.stu_id,a.course_id,a.score
    having count(a.stu_id) <=5
    order by a.course_id,a.score desc;

select S1.course_id,s1.stu_id,s1.score from score s1 where
    (select count(*) from score s2
        where s2.course_id=s1.course_id AND s2.score > s1.score
        ) <= 5 order by s1.course_id,s1.score desc;

-- row_number() over () 连续序号
-- over()里头的分组以及排序的执行晚于 where 、group by、order by 的执行。
select * from
    (select course_id, stu_id,  score,
        row_number() over (partition by course_id order by score desc ) as mum
        from score) t where mum <= 5;

-- rank() over () 排名 跳跃排序 序号不是连续的
select * from
    (select course_id, stu_id,  score,
        rank() over (partition by course_id order by score desc ) as mum
        from score) t where mum <= 5;

-- dense_rank() over () 排名 连续排序
select * from
    (select course_id, stu_id,  score,
        dense_rank() over (partition by course_id order by score desc ) as mum
        from score) t where mum <= 5;

-- 每个月每个消费总金额前三名的用户
select order_id, user_id, user_name, order_date, order_amount, total_order_amount, rank_total_order_amount from
(select order_id, user_id, user_name, order_date, order_amount, total_order_amount,
    dense_rank() over (partition by substr(order_date,1,7) order by total_order_amount desc) rank_total_order_amount
    from (
        select order_id, user_id, user_name, order_date, order_amount,
           sum(order_amount) over(partition by substr(order_date,1,7),user_id order by order_date
               rows between unbounded preceding and unbounded following) total_order_amount
           from `order`) t) t1 where rank_total_order_amount <= 3;

案例分析

准备数据

用户信息表 user.csv

hdfs dfs -mkdir -p /tmall/user
hdfs dfs -put /root/user.csv /tmall/user

create external table `user` (
	`user_id`  string COMMENT '用户id',
    `gender`   string COMMENT '性别',
    `birthday` string COMMENT '生日'
) COMMENT '用户信息表'
 row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/tmall/user';

load data local inpath '/root/user.csv' into table user;

商品信息表 sku.csv

hdfs dfs -mkdir -p /tmall/sku
hdfs dfs -put /root/sku.csv /tmall/sku

create external table sku (
	`sku_id`      string COMMENT '商品id',
    `name`        string COMMENT '商品名称',
    `category_id` string COMMENT '所属分类id',
    `from_date`   string COMMENT '上架日期',
    `price`       decimal(16, 2) COMMENT '商品单价'
) COMMENT '商品信息表'
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/tmall/sku';

load data local inpath '/root/sku.csv' into table sku;

商品分类信息表 category.csv

hdfs dfs -mkdir -p /tmall/category
hdfs dfs -put /root/category.csv /tmall/category

create external table category (
	`category_id`   string COMMENT '商品分类ID',
    `category_name` string COMMENT '商品分类名称'
) COMMENT '商品分类信息表'
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/tmall/category';

load data local inpath '/root/category.csv' into table category;

订单信息表 order.csv

hdfs dfs -mkdir -p /tmall/order
hdfs dfs -put /root/order.csv /tmall/order

create external table `order` (
	`order_id`     string COMMENT '订单id',
    `user_id`      string COMMENT '用户id',
    `create_date`  string COMMENT '下单日期',
    `total_amount` decimal(16, 2) COMMENT '订单总金额'
) COMMENT '订单信息表'
 row format delimited fields terminated by ','
 lines terminated by '\n'
 stored as textfile
 location '/tmall/order';

load data local inpath '/root/order.csv' into table order;

订单明细表 order_detail.csv

hdfs dfs -mkdir -p /tmall/order_detail
hdfs dfs -put /root/order_detail.csv /tmall/order_detail

create external table order_detail (
	`order_detail_id` string COMMENT '订单明细id',
    `order_id`        string COMMENT '订单id',
    `sku_id`          string COMMENT '商品id',
    `create_date`     string COMMENT '下单日期',
    `price`           decimal(16, 2) COMMENT '下单时的商品单价',
    `sku_num`         int COMMENT '下单商品件数'
) COMMENT '订单明细表'
 row format delimited fields terminated by ',' 
 lines terminated by '\n'
 stored as textfile
 location '/tmall/order_detail';