鲁春利的工作笔记，谁说程序员不能有文艺范？
在hive中创建mywork数据库，以后的测试在该数据库中进行，避免每次都使用default数据库。
hive> create database mywork; OK Time taken: 0.487 seconds hive> show databases; OK default mywork Time taken: 0.614 seconds, Fetched: 2 row(s) hive> hive> use mywork; OK Time taken: 0.064 seconds hive> create table student(id int, name string); OK Time taken: 0.519 seconds hive>
查看Hive在HDFS上的存储
[hadoop@dnode1 ~]$ hdfs dfs -ls -R /user/hive drwxrw-rw- - hadoop hadoop 0 2015-12-08 21:37 /user/hive/warehouse drwxrw-rw- - hadoop hadoop 0 2015-12-08 21:36 /user/hive/warehouse/mywork.db drwxrw-rw- - hadoop hadoop 0 2015-12-08 21:36 /user/hive/warehouse/mywork.db/student [hadoop@dnode1 ~]$
Hive支持的数据类型如下:
原生类型:
TINYINT 1字节 SMALLINT 2字节 INT 4字节 BIGINT 8字节 BOOLEAN true/false FLOAT 4字节 DOUBLE 8字节 STRING 字符串 BINARY (Hive 0.8.0以上才可用) TIMESTAMP (Hive 0.8.0以上才可用)
复合类型:
arrays: ARRAY<data_type> 有序字段，类型必须相同 maps: MAP<primitive_type, data_type> 无序的键/值对 structs: STRUCT<col_name : data_type [COMMENT col_comment], ...> 一组命名的字段 union: UNIONTYPE<data_type, data_type, ...>
说明：ARRAY数据类型通过下标来获取值，如arrays[0]，MAP通过[“指定域名称”]访问， STRUCT类型通过点方式访问（如structs.col_name)。
建表示例：
hive> create table employee ( > eno int comment 'the no of employee', > ename string comment 'name of employee', > salary float comment 'salary of employee', > subordinates array<string> comment 'employees managed by current employee', > deductions map<string, float> comment 'deductions', > address struct<province : string, city : string, street : string, zip : int> comment 'address' > ) comment 'This is table of employee info'; OK Time taken: 0.33 seconds hive>
在Hive中各列之间，以及复合类型内部使用了不同的分隔符来指定，每行数据对应一条记录。
在${HIVE_HOME}/data目录下创建文件data_default.txt文件，采用默认分隔符，内容为：
Hive默认的字段分隔符为ascii码的控制符/001，建表的时候用fields terminated by ‘/001’。造数据在vi 打开文件里面，用ctrl+v然后再ctrl+a可以输入这个控制符/001（即^A）。按顺序，/002的输入方式为ctrl+v,ctrl+b。以此类推。
说明：
1000 员工编号 zhangsan 员工姓名 5000.0 员工工资 lisi^Bwangwu 下属员工 ptax^C200^Bpension^C200 工资扣除金额（如税收等） shandong^Bheze^Bdingtao^B274106 家庭住址（struct结构只需指定值即可）
加载数据
hive> load data local inpath 'data/data_default.txt' into table employee;
Loading data to table mywork.employee
Table mywork.employee stats: [numFiles=1, numRows=0, totalSize=83, rawDataSize=0]
OK
Time taken: 0.426 seconds
hive> select * from employee;
OK
1000 zhangsan 5000.0 ["lisi","wangwu"] {"ptax":200.0,"pension":200.0} {"province":"shandong","city":"heze","street":"dingtao","zip":274106}
Time taken: 0.114 seconds, Fetched: 1 row(s)
hive>
# 对于复合类型数据查询方式如下
hive> select eno, ename, salary, subordinates[0], deductions['ptax'], address.province from employee;
OK
1000 zhangsan 5000.0 lisi 200.0 shandong
Time taken: 0.129 seconds, Fetched: 1 row(s)
hive>
查看HDFS数据结构
[hadoop@dnode1 ~]$ hdfs dfs -ls -R /user/hive/warehouse/ drwxrw-rw- - hadoop hadoop 0 2015-12-09 00:00 /user/hive/warehouse/mywork.db drwxrw-rw- - hadoop hadoop 0 2015-12-09 00:00 /user/hive/warehouse/mywork.db/employee -rwxrw-rw- 2 hadoop hadoop 83 2015-12-09 00:00 /user/hive/warehouse/mywork.db/employee/data_default.txt drwxrw-rw- - hadoop hadoop 0 2015-12-08 23:03 /user/hive/warehouse/mywork.db/student [hadoop@dnode1 ~]$ hdfs dfs -text /user/hive/warehouse/mywork.db/employee/data_default.txt 1000zhangsan5000.0lisiwangwuptax200pension200shandonghezedingtao274106 [hadoop@dnode1 ~]$
自定义分隔符：
hive> create table employee_02 (
> eno int comment 'the no of employee',
> ename string comment 'name of employee',
> salary float comment 'salary of employee',
> subordinates array<string> comment 'employees managed by current employee',
> deductions map<string, float> comment 'deductions',
> address struct<province : string, city : string, street : string, zip : int> comment 'address'
> ) comment 'This is table of employee info'
> row format delimited fields terminated by '/t'
> collection items terminated by ','
> map keys terminated by ':'
> lines terminated by '/n';
OK
Time taken: 0.228 seconds
hive> load data local inpath 'data/data_employee02.txt' into table employee_02;
Loading data to table mywork.employee_02
Table mywork.employee_02 stats: [numFiles=1, totalSize=99]
OK
Time taken: 0.371 seconds
hive> select * from employee_02;
OK
1000 'zhangsan' 5000.0 ["'lisi'","'wangwu'"] {"'ptax'":200.0,"'pension'":200.0} {"province":"'shandong'","city":"'heze'","street":"'dingtao'","zip":274106}
Time taken: 0.101 seconds, Fetched: 1 row(s)
hive>
data/employee02.txt文件内容为
[hadoop@nnode data]$ pwd /usr/local/hive1.2.0/data [hadoop@nnode data]$ cat data_employee02.txt 1000 'zhangsan' 5000.0 'lisi','wangwu' 'ptax':200,'pension':200 'shandong','heze','dingtao',274106 [hadoop@nnode data]$
说明：由于在文本文件中包含有单引号，在load到hive的表之后表示方式为属性加双引号，这里的单引号被认为了是属性或值的一部分了，需要注意。
查看详细表定义
# 建表时为默认设置 hive> describe formatted employee; OK # col_name data_type comment eno int the no of employee ename string name of employee salary float salary of employee subordinates array<string> employees managed by current employee deductions map<string,float> deductions address struct<province:string,city:string,street:string,zip:int> address # Detailed Table Information Database: mywork Owner: hadoop CreateTime: Tue Dec 08 23:10:07 CST 2015 LastAccessTime: UNKNOWN Protect Mode: None Retention: 0 Location: hdfs://cluster/user/hive/warehouse/mywork.db/employee Table Type: MANAGED_TABLE Table Parameters: COLUMN_STATS_ACCURATE true comment This is table of employee info numFiles 1 numRows 0 rawDataSize 0 totalSize 83 transient_lastDdlTime 1449590423 # Storage Information SerDe Library: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe InputFormat: org.apache.hadoop.mapred.TextInputFormat OutputFormat: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat Compressed: No Num Buckets: -1 Bucket Columns: [] Sort Columns: [] Storage Desc Params: serialization.format 1 Time taken: 0.098 seconds, Fetched: 37 row(s) # 建表时自定义了分隔符 hive> describe formatted employee_02; OK # col_name data_type comment eno int the no of employee ename string name of employee salary float salary of employee subordinates array<string> employees managed by current employee deductions map<string,float> deductions address struct<province:string,city:string,street:string,zip:int> address # Detailed Table Information Database: mywork Owner: hadoop CreateTime: Wed Dec 09 00:12:53 CST 2015 LastAccessTime: UNKNOWN Protect Mode: None Retention: 0 Location: hdfs://cluster/user/hive/warehouse/mywork.db/employee_02 Table Type: MANAGED_TABLE Table Parameters: COLUMN_STATS_ACCURATE true comment This is table of employee info numFiles 1 totalSize 99 transient_lastDdlTime 1449591260 # Storage Information SerDe Library: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe InputFormat: org.apache.hadoop.mapred.TextInputFormat OutputFormat: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat Compressed: No Num Buckets: -1 Bucket Columns: [] Sort Columns: [] Storage Desc Params: colelction.delim , field.delim /t line.delim /n mapkey.delim : serialization.format /t Time taken: 0.116 seconds, Fetched: 39 row(s) hive>
遗留问题：
hive> delete from student; FAILED: SemanticException [Error 10294]: Attempt to do update or delete using transaction manager that does not support these operations. hive>
注意事项：如果sql语句中含有tab格式的内容，则会出现如下问题
原创文章，作者：kepupublish，如若转载，请注明出处：https://blog.ytso.com/193735.html