备注:
Hive 版本 2.1.1
Table of Contents
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:
- regexp
- regexp_extract
- 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 rn
from tmp1
where 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 rn
from tmp1
where 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 rn
from tmp1
where 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_str
from 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_str
from 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_str
from 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_str
from 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 filterid
from 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
