hive 常用操作

参考:https://www.cnblogs.com/jonban/p/10779938.html

Hive 启动:hive
退出:hive>quit;
show databases;
show tables;
desc tab_name; --查看表的结构及表的路径
show partitions fact_measured_cft_hive ;展示表分区 fact_measured_cft_hive

 

1、 创建表 create-table.sql

复制代码
create table if not exists db_hive.tb_user
(
    id int,
    username string comment '用户名',
    age int comment '年龄',
    address string comment '地址'
)
comment '用户表'
row format delimited fields terminated by ',' 
stored as textfile
location '/user/hive/warehouse/db_hive.db/db_user'
复制代码

 

 

2、执行创建表

hive -f 'create-table.sql'

 

3、加载数据到 tb_user 表中

数据文件  /root/files/tb_user.txt

1001,Logan,16,shenzhen
1002,Herry potter,12,Magic school
1003,孙悟空,500,花果山

 

Hive交互式命令行执行命令  load data local inpath '/root/files/tb_user.txt' into table db_hive.tb_user;

如下所示:

hive (db_hive)> load data local inpath '/root/files/tb_user.txt' into table db_hive.tb_user;

 

如果要覆盖旧数据,可以加 overwrite,如下所示

hive (db_hive)> load data local inpath '/root/files/tb_user.txt' overwrite into table db_hive.tb_user;

 

 

4、查询数据

 

hive -e "select id,username from db_hive.tb_user"

 

 

 

 

5、根据已有表创建只有部分字段的子表

create table if not exists db_hive.tb_user_sub 
as 
select id,username from db_hive.tb_user;

 

 

6、 like 创建表

create table if not exists db_hive.tb_user_like like db_hive.tb_user;

 

插入数据

insert into table db_hive.tb_user_like select * from db_hive.tb_user;

 

 

7、重命名表

alter table tb_user_like rename to tb_user_rename ;

 

 

8、 创建外部表,删除时只删除元数据,不会删除表数据

create external table if not exists db_hive.tb_ext(id string);

 

 

9、创建分区表

复制代码
create table if not exists db_hive.tb_logs(
    ip string,
    text string,
    log_time string
)
partitioned by (month string)
row format delimited fields terminated by "\t";
复制代码

 

 数据文件 /root/files/tb_logs.txt

192.168.32.100  login   20190429072650
192.168.32.100  order   20190429072730
192.168.32.101  browse  20190429072812

 

载入数据

load data local inpath '/root/files/tb_logs.txt' into table db_hive.tb_logs partition (month = '201904')

 

查询数据

select ip,text,log_time from tb_logs where month = '201904';

 

 

10、手工创建分区数据及修复分区表

创建分区目录

hdfs dfs -mkdir -p /user/hive/warehouse/db_hive.db/tb_logs/month=201905

 

上传数据文件到分区目录下

hdfs dfs -put /root/files/tb_logs.txt /user/hive/warehouse/db_hive.db/tb_logs/month=201905

 

此时执行查询

select count(distinct ip) from db_hive.tb_logs where month = '201905';

查询结果为0。

【原因】:数据并未添加到分区中,查看配置的MySQL元数据信息

mysql> use hive_metastore;
mysql> select * from PARTITIONS;

 

示例配置的Hive元数据存放为MySQL数据库中的 hive_metastore 数据库

查询分区表 PARTITIONS 中的数据,发现只有一条记录,如下所示:

+---------+-------------+------------------+--------------+-------+--------+
| PART_ID | CREATE_TIME | LAST_ACCESS_TIME | PART_NAME    | SD_ID | TBL_ID |
+---------+-------------+------------------+--------------+-------+--------+
|       1 |  1556494255 |                0 | month=201904 |    29 |     28 |
+---------+-------------+------------------+--------------+-------+--------+

 

【修复方法一】直接执行修复命令

msck repair table tb_logs

此时分区表中的数据如下:

复制代码
+---------+-------------+------------------+--------------+-------+--------+
| PART_ID | CREATE_TIME | LAST_ACCESS_TIME | PART_NAME    | SD_ID | TBL_ID |
+---------+-------------+------------------+--------------+-------+--------+
|       1 |  1556494255 |                0 | month=201904 |    29 |     28 |
|       2 |  1556495227 |                0 | month=201905 |    30 |     28 |
+---------+-------------+------------------+--------------+-------+--------+
复制代码

 

执行查询命令

select count(distinct ip) from db_hive.tb_logs where month = '201905';

返回结果为2,数据已正常加入分区。

 

 

【修复方法二】 使用增加分区命令

操作步骤:创建新分区目录并上传数据文件,命令如下:

hive (db_hive)> dfs -mkdir -p /user/hive/warehouse/db_hive.db/tb_logs/month=201906;
hive (db_hive)> dfs -put /root/files/tb_logs.txt /user/hive/warehouse/db_hive.db/tb_logs/month=201906;

 

执行增加分区命令

alter table tb_logs add partition(month = '201906');

 

查询数据,测试结果正常。

此时元数据分区表中数据如下:

复制代码
+---------+-------------+------------------+--------------+-------+--------+
| PART_ID | CREATE_TIME | LAST_ACCESS_TIME | PART_NAME    | SD_ID | TBL_ID |
+---------+-------------+------------------+--------------+-------+--------+
|       1 |  1556494255 |                0 | month=201904 |    29 |     28 |
|       2 |  1556495227 |                0 | month=201905 |    30 |     28 |
|       3 |  1556495635 |                0 | month=201906 |    31 |     28 |
+---------+-------------+------------------+--------------+-------+--------+
复制代码

 

查看表分区命令

show partitions db_hive.tb_logs;

 

 

11、 导出表数据

export table db_hive.tb_logs to '/user/hive/warehouse/export/db_hive/tb_logs';

 

 

12、 导入表数据

创建表

create table tb_logs_like like tb_logs;

 

导入数据

import table tb_logs_like from '/user/hive/warehouse/export/db_hive/tb_logs';

 

 

13、导出数据到本地文件

insert overwrite local directory '/root/files/hive_out' 
row format delimited fields terminated by '\t' collection items terminated by '\n' 
select * from db_hive.tb_logs;

 

 

Hive 常用命令和语句

#!/usr/bin/python
# -*- coding: utf-8 -*-
from sqlalchemy.engine import create_engine
from sqlalchemy import text
import pandas as pd
import datetime

starttime = datetime.datetime.now()

sql = """
    select *
    from fact_five_data_cft
    where source = 'classics' and  wfid='320924' and yyyy='2019'
    limit 10
    """
engine = create_engine('presto://node1:8085/hive/cnyb')
df = pd.read_sql(text(sql),engine)
print(df)
endtime = datetime.datetime.now()
runtime = endtime - starttime
print "presto run time -->"+str(runtime)

 

你可能感兴趣的