备注:
Hive 版本 2.1.1

文章目录

  • 一.Hive 正则表达式概述
    • 1.1 字符集合:
    • 1.2 边界集合:
    • 1.3 重复次数集合:
    • 1.4 组合操作符:
    • 1.5 匹配操作符:
    • 1.6 转义操作符:
  • 二.Hive 正则表达式案例
    • 2.1 regexp
    • 2.2 regexp_replace
      • 2.2.1 截取字符串中汉字部分
      • 2.2.2 截取字符串中数字部分
      • 2.2.3 截取字符串中字母部分
      • 2.2.4 截取字符串中的大小写字母和数字
    • 2.3 regexp_extract
  • 参考：

一.Hive 正则表达式概述

Hive的正则表达式虽然没有关系型数据库的正则表达式那么强大，但是一样可以解决HQL开发过程中的诸多问题，数据工作者平时也离不开正则表达式。对此，特意做了个hive正则表达式的小结。所有代码都经过亲测，正常运行。

Hive支持如下三个正则表达式:

  1. regexp
  2. regexp_extract
  3. regexp_replace

1.1 字符集合:

字符 匹配的字符
\d 从０-９的任一数字
\D 任一非数字字符
\w 任一单词字符，包括A-Z,a-z,0-9和下划线
\W 任一非单词字符
\s 任一空白字符，包括制表符，换行符，回车符，换页符和垂直制表符
\S 任一非空白字符
. 任一字符

字符簇:

 [[:alpha:]] 任何字母。[[:digit:]] 任何数字。[[:alnum:]] 任何字母和数字。[[:space:]] 任何空白字符。[[:upper:]] 任何大写字母。[[:lower:]] 任何小写字母。[[:punct:]] 任何标点符号。[[:xdigit:]] 任何16进制的数字，相当于[0-9a-fA-F]。

1.2 边界集合:

字符 l 描述
^ 每一行的开头，单行模式下等价于字符串的开头
$ 每一行的结尾，单行模式下等价于字符串的结尾

1.3 重复次数集合:

贪婪模式会获取尽可能多的字符，而非贪婪模式会获取尽可能少的字符。

贪婪 非贪婪 描述
* *? 零次或多次
? ?? 零次或一次
+ +? 一次或多次
{m} {m}? 正好m次，贪婪与非贪婪一样的
{m,} {m,}? 至少m次
{m, n} {m, n}? 最少m最多n次

1.4 组合操作符:

优先级比较：圆括号>重复次数操作符>和>或

组合操作符 描述
[…] 方括号内任意字符或字符集合中的一个
[^…] 方括号内^为第一个字符时，表示与其后所有字符都不匹配的字符
(…) 圆括号，将复杂表达式当作单一表达式来处理
…|… 或
abc 和。直接将字符连在一起写

1.5 匹配操作符:

匹配操作符 描述
\n 即后向引用。n为1~9，标识由圆括号里取得的匹配字符串。方向是从左到右在regexp_replace函数中，允许在模式表达式和替换表达式中都使用\n

1.6 转义操作符:

转义操作符 描述
\ 将其后紧跟着的操作字符当作普通字符看待。例如 abcdef 可以匹配 abdef或abcccdef等，但无法匹配 abcdef，后者需要abc*def才能匹配

二.Hive 正则表达式案例

2.1 regexp

语法:

A REGEXP B
功能同RLIKE

如果字符串A或者字符串B为NULL，则返回NULL；如果字符串A符合表达式B 的正则语法，则为TRUE；否则为FALSE。B中字符”_”表示任意单个字符，而字符”%”表示任意数量的字符。

代码:

-- 匹配有8个连续数字的字符串
with tmp1 as
(
select '11145678abc' as rn
union all
select '111456789abc'
union all
select 'd111456789abc'
)
select rnfrom tmp1where rn regexp '\\d{8}';-- 匹配开头有8个及以上连续数字的字符
with tmp1 as
(
select '11145678abc' as rn
union all
select '111456789abc'
union all
select 'd111456789abc'
)
select rnfrom tmp1where rn regexp '^\\d{8}';-- 匹配开头只有8个连续数字的字符
with tmp1 as
(
select '11145678abc' as rn
union all
select '111456789abc'
union all
select 'd111456789abc'
)
select rnfrom tmp1where rn regexp '^\\d{8}\\D';

测试记录:

hive> > -- 匹配有8个连续数字的字符串> with tmp1 as> (> select '11145678abc' as rn> union all> select '111456789abc'> union all> select 'd111456789abc'> )> select rn>   from tmp1>  where rn regexp '\\d{8}';
Query ID = root_20201217151846_3f0eebcd-6f5a-455d-b8c9-afd5cddbc358
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0295, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0295/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0295
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2020-12-17 15:18:52,063 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:18:58,242 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.73 sec
MapReduce Total cumulative CPU time: 1 seconds 730 msec
Ended Job = job_1606698967173_0295
MapReduce Jobs Launched:
Stage-Stage-1: Map: 1   Cumulative CPU: 1.73 sec   HDFS Read: 5150 HDFS Write: 162 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 1 seconds 730 msec
OK
rn
11145678abc
111456789abc
d111456789abc
Time taken: 13.291 seconds, Fetched: 3 row(s)
hive> -- 匹配开头有8个及以上连续数字的字符
hive> with tmp1 as> (> select '11145678abc' as rn> union all> select '111456789abc'> union all> select 'd111456789abc'> )> select rn>   from tmp1>  where rn regexp '^\\d{8}';
Query ID = root_20201217151946_c5102f51-5e70-4a80-afa6-3678f92091f0
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0296, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0296/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0296
Hadoop job information for Stage-1: number of mappers: 2; number of reducers: 0
2020-12-17 15:19:54,003 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:20:00,174 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 3.85 sec
MapReduce Total cumulative CPU time: 3 seconds 850 msec
Ended Job = job_1606698967173_0296
MapReduce Jobs Launched:
Stage-Stage-1: Map: 2   Cumulative CPU: 3.85 sec   HDFS Read: 10856 HDFS Write: 223 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 3 seconds 850 msec
OK
rn
11145678abc
111456789abc
Time taken: 15.275 seconds, Fetched: 2 row(s)
hive> -- 匹配开头只有8个连续数字的字符
hive> with tmp1 as> (> select '11145678abc' as rn> union all> select '111456789abc'> union all> select 'd111456789abc'> )> select rn>   from tmp1>  where rn regexp '^\\d{8}\\D';
Query ID = root_20201217152016_c920f47c-663c-4d4f-a9df-e9c77d20a126
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0297, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0297/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0297
Hadoop job information for Stage-1: number of mappers: 2; number of reducers: 0
2020-12-17 15:20:22,628 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:20:28,804 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 3.61 sec
MapReduce Total cumulative CPU time: 3 seconds 610 msec
Ended Job = job_1606698967173_0297
MapReduce Jobs Launched:
Stage-Stage-1: Map: 2   Cumulative CPU: 3.61 sec   HDFS Read: 10990 HDFS Write: 198 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 3 seconds 610 msec
OK
rn
11145678abc
Time taken: 13.633 seconds, Fetched: 1 row(s)
hive>

2.2 regexp_replace

语法:

regexp_replace(string INITIAL_STRING, string PATTERN, string REPLACEMENT)

返回用替换实例替换INITIAL_STRING中与模式中定义的java正则表达式语法匹配的所有子字符串所产生的字符串。例如，regexp_replace(“foobar”， “oo|ar”， “”)返回’fb。‘注意，在使用预定义字符类时需要注意:使用’\s’作为第二个参数将匹配字母s;’\s’必须匹配空格，等等。

测试数据

create table test_reg(id int,str string);insert into test_reg values (1,'我在学习Hive，大数据。');
insert into test_reg values (2,'Hive，我来了，Coming666。');
insert into test_reg values (3,'666,Hive居然拥有关系型数据库的诸多特性。');
insert into test_reg values (4,'wuwuwu，Hive学习起来还是存在一定难度。');
insert into test_reg values (5,'Hive数据仓库，6666。');

2.2.1 截取字符串中汉字部分

代码:

select id,regexp_replace(str,'([^\\u4E00-\\u9FA5]+)','') new_strfrom test_reg;

测试记录:

hive> > select id,regexp_replace(str,'([^\\u4E00-\\u9FA5]+)','') new_str>   from test_reg;
Query ID = root_20201217153549_3a7163ba-365b-4f65-adad-930d42ec385d
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0303, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0303/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0303
Hadoop job information for Stage-1: number of mappers: 2; number of reducers: 0
2020-12-17 15:35:56,518 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:36:02,702 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 6.32 sec
MapReduce Total cumulative CPU time: 6 seconds 320 msec
Ended Job = job_1606698967173_0303
MapReduce Jobs Launched:
Stage-Stage-1: Map: 2   Cumulative CPU: 6.32 sec   HDFS Read: 8951 HDFS Write: 495 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 6 seconds 320 msec
OK
id      new_str
3       居然拥有关系型数据库的诸多特性
4       学习起来还是存在一定难度
5       数据仓库
1       我在学习大数据
2       我来了
Time taken: 14.196 seconds, Fetched: 5 row(s)
hive>

2.2.2 截取字符串中数字部分

代码:

select id,regexp_replace(str,'([^0-9]+)','') new_strfrom test_reg;

测试记录:

hive> > select id,regexp_replace(str,'([^0-9]+)','') new_str>   from test_reg;
Query ID = root_20201217153822_cf2389b9-8533-4b82-b472-57df3a1da418
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0304, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0304/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0304
Hadoop job information for Stage-1: number of mappers: 2; number of reducers: 0
2020-12-17 15:38:30,220 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:38:37,427 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 6.19 sec
MapReduce Total cumulative CPU time: 6 seconds 190 msec
Ended Job = job_1606698967173_0304
MapReduce Jobs Launched:
Stage-Stage-1: Map: 2   Cumulative CPU: 6.19 sec   HDFS Read: 9139 HDFS Write: 259 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 6 seconds 190 msec
OK
id      new_str
3       666
4
5       6666
1
2       666
Time taken: 16.038 seconds, Fetched: 5 row(s)
hive>

2.2.3 截取字符串中字母部分

代码:

select id,regexp_replace(str,'([^a-zA-Z]+)','') new_strfrom test_reg;

测试记录:

hive> > select id,regexp_replace(str,'([^a-zA-Z]+)','') new_str>   from test_reg;
Query ID = root_20201217154102_65e60007-0a8a-45d2-94de-bb0bfe11f9e8
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0305, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0305/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0305
Hadoop job information for Stage-1: number of mappers: 2; number of reducers: 0
2020-12-17 15:41:08,696 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:41:14,864 Stage-1 map = 50%,  reduce = 0%, Cumulative CPU 3.02 sec
2020-12-17 15:41:15,891 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 6.13 sec
MapReduce Total cumulative CPU time: 6 seconds 130 msec
Ended Job = job_1606698967173_0305
MapReduce Jobs Launched:
Stage-Stage-1: Map: 2   Cumulative CPU: 6.13 sec   HDFS Read: 9145 HDFS Write: 281 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 6 seconds 130 msec
OK
id      new_str
3       Hive
4       wuwuwuHive
5       Hive
1       Hive
2       HiveComing
Time taken: 14.205 seconds, Fetched: 5 row(s)
hive>

2.2.4 截取字符串中的大小写字母和数字

代码:

select id,regexp_replace(str,'([^a-zA-Z0-9]+)','') new_strfrom test_reg;

测试记录:

hive> select id,regexp_replace(str,'([^a-zA-Z0-9]+)','') new_str>   from test_reg;
Query ID = root_20201217154722_738edcd7-4c2a-4aa1-a66a-acac268f199b
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0310, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0310/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0310
Hadoop job information for Stage-1: number of mappers: 2; number of reducers: 0
2020-12-17 15:47:29,743 Stage-1 map = 0%,  reduce = 0%
2020-12-17 15:47:36,950 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 6.13 sec
MapReduce Total cumulative CPU time: 6 seconds 130 msec
Ended Job = job_1606698967173_0310
MapReduce Jobs Launched:
Stage-Stage-1: Map: 2   Cumulative CPU: 6.13 sec   HDFS Read: 9123 HDFS Write: 291 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 6 seconds 130 msec
OK
id      new_str
3       666Hive
4       wuwuwuHive
5       Hive6666
1       Hive
2       HiveComing666
Time taken: 15.686 seconds, Fetched: 5 row(s)

2.3 regexp_extract

语法:

regexp_extract(string subject, string pattern, int index)

返回使用模式提取的字符串。例如，regexp_extract(‘foothebar’， ‘foo(.*?)(bar)’， 2)返回’bar ‘。‘注意，在使用预定义字符类时需要注意:使用’\s’作为第二个参数将匹配字母s;’\s’必须匹配空格，等等。

测试json数据

create table test_reg2(id int,str string);
insert into test_reg2 values (1,'{"filtertype":"29","filtername":"成人Node","filtertitle":"成人Group","filtersubtype":"","filterid":"29|1","filterValue":"1|4"}');
insert into test_reg2 values (2,'{"filtertitle":"钻级","filtertype":"16","filtersubtype":"","filtername":"四钻/高档","filterid":"16|4",}');

代码:

select id,regexp_extract(str,'(filtertype"\\:")(\\d+)(",)',2)               as filtertype,regexp_extract(str,'(filtername"\\:")((\\W*\\w*)|(\\W*))(",)',2)  as filtername,regexp_extract(str,'(filtertitle"\\:")((\\W*\\w*)|(\\W*))(",)',2) as filtertitle,regexp_extract(str,'(filterid"\\:")(\\d+\\|\\d+)(",)',2)          as filteridfrom test_reg2;

测试记录:

hive> > select id>       ,regexp_extract(str,'(filtertype"\\:")(\\d+)(",)',2)               as filtertype>       ,regexp_extract(str,'(filtername"\\:")((\\W*\\w*)|(\\W*))(",)',2)  as filtername>       ,regexp_extract(str,'(filtertitle"\\:")((\\W*\\w*)|(\\W*))(",)',2) as filtertitle>       ,regexp_extract(str,'(filterid"\\:")(\\d+\\|\\d+)(",)',2)          as filterid>  from test_reg2;
Query ID = root_20201217164358_dbda76fc-2bc1-41d6-b140-72f590b05502
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1606698967173_0314, Tracking URL = http://hp1:8088/proxy/application_1606698967173_0314/
Kill Command = /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/hadoop/bin/hadoop job  -kill job_1606698967173_0314
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2020-12-17 16:44:05,767 Stage-1 map = 0%,  reduce = 0%
2020-12-17 16:44:12,987 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 2.93 sec
MapReduce Total cumulative CPU time: 2 seconds 930 msec
Ended Job = job_1606698967173_0314
MapReduce Jobs Launched:
Stage-Stage-1: Map: 1   Cumulative CPU: 2.93 sec   HDFS Read: 5038 HDFS Write: 205 HDFS EC Read: 0 SUCCESS
Total MapReduce CPU Time Spent: 2 seconds 930 msec
OK
id      filtertype      filtername      filtertitle     filterid
1       29      成人Node        成人Group       29|1
2       16      四钻/高档       钻级    16|4
Time taken: 15.265 seconds, Fetched: 2 row(s)

参考：

1.https://cwiki.apache.org/confluence/display/Hive/LanguageManual+UDF#LanguageManualUDF-MathematicalFunctions
2.https://blog.csdn.net/weixin_37536446/article/details/81053172
3.https://www.cnblogs.com/db-record/p/11454325.html

大数据开发之Hive篇12-Hive正则表达式相关推荐

  1. 大数据开发之Hive优化篇8-Hive Job优化

    备注: Hive 版本 2.1.1 文章目录 Hive job优化概述 一.并行执行 二.本地执行 三.合并输入小文件 四.合并输出小文件 五.控制Map/Reduce数 5.1 控制Hive job ...

  2. 大数据开发之Hive篇3-Hive数据定义语言

    备注: Hive 版本 2.1.1 文章目录 一.Hive关系模型概述 1.1.Database 1.2 Table 1.2.1 管理表和外部表 1.2.2 永久表和临时表 1.3 Partition ...

  3. 大数据开发之Hive篇17-Hive锁机制

    备注: Hive 版本 2.1.1 文章目录 一.Hive锁概述 二.Hive 锁相关操作 2.1 Hive的并发性 2.2 查看表的锁 2.3 解锁 三.Hive 事务表锁机制 四.Hive 锁测试 ...

  4. 大数据开发之Sqoop详细介绍

    备注: 测试环境 CDH 6.3.1 Sqoop 1.4.7 文章目录 一.Sqoop概述 二.Sqoop 工具概述 三.Sqoon工具详解 3.1 codegen 3.2 create-hive-t ...

  5. 高效大数据开发之 bitmap 思想的应用

    作者:xmxiong,PCG 运营开发工程师 数据仓库的数据统计,可以归纳为三类:增量类.累计类.留存类.而累计类又分为历史至今的累计与最近一段时间内的累计(比如滚动月活跃天,滚动周活跃天,最近 N ...

  6. 萌新Java开发实战记录：大数据开发之”IP热力图、地点热门TopN（文章底部附源码）

    提示:文章写完后,目录可以自动生成,如何生成可参考右边的帮助文档 目录 一. 课程设计背景概述 1. <IP经纬热力图>概述 2. <电商分析系统>概述 二.需求分析 1.&l ...

  7. 大数据开发之Hive篇14-Hive归档(Archiving)

    备注: Hive 版本 2.1.1 文章目录 一.Hive归档简介 二.Hive 归档操作 参考 一.Hive归档简介 由于HDFS的设计,文件系统中的文件数量直接影响namenode中的内存消耗.虽 ...

  8. 大数据开发之Hive篇18-Hive的回收站

    备注: Hive 版本 2.1.1 一.模拟误删表 误删除了这张表 hive> > drop table ods_fact_sale_orc; OK 二.从回收站恢复表 查看回收表 [ro ...

  9. 大数据开发之Hive篇8-Hive视图

    备注: Hive 版本 2.1.1 文章目录 一.Hive视图介绍 二.视图案例 2.1 视图能够简化用户内的操作 2.2 视图使用户能以多种角度看待同一数据 2.3 视图对重构数据库提供了一定程度的 ...

最新文章

  1. python_day2基本数据类型
  2. opencv 通道分离合并
  3. 【解放日报】除了CEO首席执行官，你了解CIO吗？
  4. php新闻列表排序,javascript 新闻列表排序简单封装
  5. linux pid t 头文件_pid和tid及线程调度
  6. 接口规范 14.转码接口
  7. 2019 CCF CSP-J2题解
  8. RAR Extractor Max for Mac(解压缩软件)
  9. 回顾 2018: 革新的一年
  10. ubuntu安装vasp_Ubuntu下p4vasp的安装
  11. Unicode字符集和多字节字符集
  12. SAP中的输入历史记录设定
  13. 哈哈哈……~好敷衍的第一篇博客标题~
  14. 【QNX Hypervisor 2.2 用户手册】4 构建QNX Hypervisor系统
  15. Qt读写Excel--QXlsx合并单元格、文本对齐7
  16. k-means均值聚类
  17. OAK 3D人工智能相机和RealSense系列相机的对比
  18. python的占位符——%
  19. 人工智能换脸技术python_人工智能几行代码实现换脸，python+dlib实现图文教程
  20. centos 安装 Go环境

热门文章

  1. OkHttpUtils | okhttp-OkGo的使用，完美支持RxJava
  2. 学习3ds max—做自行车车轮
  3. Target “pango_windowing“ links to target “Eigen3::Eigen“ but the target was not found. Perhaps a
  4. 证书文件编码格式介绍
  5. 高效程序猿的狂暴之路
  6. 平克四部曲之《白板》
  7. 1389：买零食【C++】
  8. Lam Research和VELO3D达成战略协议，使用金属增材制造应用生产半导体资本设备
  9. HTML简易时钟教程,html5 svg简单的模拟时钟特效-HTML5动画
  10. python算式运算题目_python的四则运算练习