一、表的操作
1、创建表、分区、分隔符
create table table_name(
name string,
age string)
partitioned by (
dt string)
row format delimited fields terminated by ‘\t’;
dt为分区，并且指定分隔符为\t hive默认分隔符\001
2、删除表数据、删除表和sql相同
truncate table ,drop table;
3、删除表的分区
alter table table_name drop if exists partition (dt = 20180808);
4、表添加字段
alter table table_name add columns(c1 int, c2 int);
5、分区表表结构的复制和包含数据的复制
二、数据的导入和导出
1、从本地文件加载
load data local inpath ‘/home/coco/demo/*’ into table test_table
2、hive查询数据写入到本地文件 并且指定分隔符
insert overwrite local directory ‘/home/coco/output’ row format delimited fields terminated by ‘\t’ select age,count(*) from test_name group by age order by age;
3、sqoop导入
sqoop import –connect jdbc:mysql://10.1.1.1:3306/user_center –username root –password 123456 –query “select * from user where age=18 and \$CONDITIONS ” –hive-import –target-dir /user/coco/user_all/city=beijing –hive-database user_center –hive-table test_table –split-by age –hive-partition-key city –hive-partition-value beijing –fields-terminated-by ‘\t’
\CONDITIONS为必填的
jdbc:mysql:// mysql地址/端口/库名 –username mysql用户名 –password mysql密码 –query 设为指定sql查询 –target-dir目标集群路径(可以查看表所在的集群路径,分区) –hive-database hive的库名 –hive-table hive的表名 –split-by 分割字段(尽量选取均匀分布的字段) –hive-partition-key 如果有是分区表分区的key –hive-partition-value 分区的值 –fields-terminated-by 分隔符
sqoop导入的防止只支持单分区表，如果需要导入多分区表，需要复杂些提供两种思路
直接把数据导入到表所在的路径，然后加载分区
或者建一个临时的不分区表，通过动态分区加载到分区表中
sqoop导入出错 报java.sql.SQLException: Streaming result set com.mysql.jdbc.RowDataDynamic@5f058f00 is still active. No statements may be…
–driver com.mysql.jdbc.Driver
接着出现The last packet sent successfully to the server was 0 milliseconds a
jdbc:mysql://10.1.1.1:3306/user_center?autoReconnect=true
4、sqoop导出
sqoop export –connect “jdbc:mysql://10.1.1.1:3306:3306/user_center?useUnicode=true&characterEncoding=utf-8” –username root –password 123456 –table user –export-dir /user/hive/warehouse/user_center.db/user_center –input-fields-terminated-by ‘\001’ –input-null-string ‘\\N’ –input-null-non-string ‘\\N’
三、数据处理
1、hive的去重
select * from( select *,row_number() over (partition by source_url order by phone asc) num from spider) t where t.num=1
partition by 后是需要去重的字段，可以添加多个，逗号分割。order by根据什么字段排序，决定着去重的时候保留去重哪些条数据，保留那一条数据
2、hive动态分区
hive -v -e “
use test_db;
set hive.exec.dynamic.partition=true;
set hive.exec.dynamic.partition.mode=nonstrict;
set hive.cli.print.header=true;
insert overwrite table user_table partition(city,dt) select a,b,c,city,dt from user_temp where age>18;
“
主要设置
set hive.exec.dynamic.partition=true;
set hive.exec.dynamic.partition.mode=nonstrict;
查询的最后字段就是分区的字段
3、hive引用python脚本处理数据
hive -v -e”
use test_db;
set hive.cli.print.header=false;
set hive.exec.dynamic.partition.mode=nonstrict;
set mapreduce.map.memory.mb=1025;
set mapreduce.reduce.memory.mb=1025;
add file /data/coco/house_.py;
drop table if exists test_table;
create table test_table as
SELECT TRANSFORM(a,b,c,d,e,f) USING ‘house_.py’
AS a,b,c FROM user_table;”
python脚本中
for line in sys.stdin:
(a,b,c) = line.split(‘\t’)
d=a+b
e=1
f=2
print ‘\t’.join([a,b,c,d,e,f])
