Hive自定义函数与transform的使用

hive是给了我们很多内置函数的，比如转大小写，截取字符串等，具体的都在官方文档里面。但是并不是所有的函数都能满足我们的需求，所以hive提供了给我们自定义函数的功能。
1、至于怎么测试hive为我们提供的函数
因为mysql或者oracle中都可以使用伪表，但是hive不行，所以可以使用以下方法
1）、创建表dual,create table dual(id string)
2）、在本地创建文件dual.data,内容为空格或者空一行
3）、将dual.data文件load到表dual
进行测试，比如：字符串截取

0: jdbc:hive2://localhost:10000> select substr('sichuan',1,3) from dual;
+------+--+
| _c0  |
+------+--+
| sic  |
+------+--+

当然也可以直接使用 select substr(‘sichuan’,1,3)，但是还是习惯用from dual;

2、自定义内置函数
添加maven依赖

  <dependency>
      <groupId>org.apache.hive</groupId>
      <artifactId>hive-exec</artifactId>
      <version>1.2.1</version>
    </dependency>
    <!-- https://mvnrepository.com/artifact/org.apache.hive/hive-metastore -->
    <dependency>
      <groupId>org.apache.hive</groupId>
      <artifactId>hive-metastore</artifactId>
      <version>1.2.1</version>
    </dependency>
    <!-- https://mvnrepository.com/artifact/org.apache.hive/hive-common -->
    <dependency>
      <groupId>org.apache.hive</groupId>
      <artifactId>hive-common</artifactId>
      <version>1.2.1</version>
    </dependency>
    <!-- https://mvnrepository.com/artifact/org.apache.hive/hive-service -->
    <dependency>
      <groupId>org.apache.hive</groupId>
      <artifactId>hive-service</artifactId>
      <version>1.2.1</version>
    </dependency>
    <!-- https://mvnrepository.com/artifact/org.apache.hive/hive-jdbc -->
    <dependency>
      <groupId>org.apache.hive</groupId>
      <artifactId>hive-jdbc</artifactId>
      <version>1.2.1</version>
    </dependency>

1）、大写转小写
可以先创建java类继承UDF,重载evaluate方法。

/**
 * 大写转小写
 * @author 12706
 */
public class UpperToLowerCase extends UDF {
    /*
     * 重载evaluate
     * 访问限制必须是public
     */
    public String evaluate(String word) {
        String lowerWord = word.toLowerCase();
        return lowerWord;
    }
}

打包上传到hadoop集群（打的jar包名字为hive.jar）。

0: jdbc:hive2://localhost:10000> select * from t5;
+--------+-----------+--+
| t5.id  |  t5.name  |
+--------+-----------+--+
| 13     | BABY      |
| 1      | zhangsan  |
| 2      | lisi      |
| 3      | wangwu    |
| 4      | furong    |
| 5      | fengjie   |
| 6      | aaa       |
| 7      | bbb       |
| 8      | ccc       |
| 9      | ddd       |
| 10     | eee       |
| 11     | fff       |
| 12     | ggg       |
+--------+-----------+--+
13 rows selected (0.221 seconds)

将jar包放到hive的classpath下

0: jdbc:hive2://localhost:10000> add jar /root/hive.jar;

创建临时函数，指定完整类名

0: jdbc:hive2://localhost:10000> create temporary function tolower as 'com.scu.hive.UpperToLowerCase';

到这就可以使用自定义临时函数tolower()了，测试t5表中的name输出小写

0: jdbc:hive2://localhost:10000> select id,tolower(name) from t5;
+-----+-----------+--+
| id  |    _c1    |
+-----+-----------+--+
| 13  | baby      |
| 1   | zhangsan  |
| 2   | lisi      |
| 3   | wangwu    |
| 4   | furong    |
| 5   | fengjie   |
| 6   | aaa       |
| 7   | bbb       |
| 8   | ccc       |
| 9   | ddd       |
| 10  | eee       |
| 11  | fff       |
| 12  | ggg       |
+-----+-----------+--+

根据电话号码显示归属地信息
jave类

/**
 * 根据电话号码前三位获取归属地
 * @author 12706
 *
 */
public class PhoneNumParse extends UDF{

    static HashMap<String, String> phoneMap = new HashMap<String, String>();

    static{
        phoneMap.put("136", "beijing");
        phoneMap.put("137", "shanghai");
        phoneMap.put("138", "shenzhen");
    }

    public static String evaluate(int phoneNum) {

        String num = String.valueOf(phoneNum);
        String province = phoneMap.get(num.substring(0, 3));
        return province==null?"foreign":province;
    }
    //测试
    public static void main(String[] args) {
        String string = evaluate(136666);
        System.out.println(string);
    }
}

将工程打包上传到linux，注意：如果名字还是跟上面一样，那么需要重新连接hive服务端了，否则jar包是不会覆盖的，建议打的jar包名字别一样

编辑文件vi prov.data
创建表flow(phonenum int,flow int)
将文件load到flow表

[root@mini1 ~]# vi prov.data;
1367788,1
1367788,10
1377788,80
1377788,97
1387788,98
1387788,99
1387788,100
1555118,99

0: jdbc:hive2://localhost:10000> create table flow(phonenum int,flow int)
0: jdbc:hive2://localhost:10000> row format delimited fields terminated by ',';
No rows affected (0.143 seconds)
0: jdbc:hive2://localhost:10000> load data local inpath '/root/prov.data' into table flow;
INFO  : Loading data to table myhive3.flow from file:/root/prov.data
INFO  : Table myhive3.flow stats: [numFiles=1, totalSize=88]
No rows affected (0.316 seconds)
0: jdbc:hive2://localhost:10000> select * from flow;
+----------------+------------+--+
| flow.phonenum  | flow.flow  |
+----------------+------------+--+
| 1367788        | 1          |
| 1367788        | 10         |
| 1377788        | 80         |
| 1377788        | 97         |
| 1387788        | 98         |
| 1387788        | 99         |
| 1387788        | 100        |
| 1555118        | 99         |
+----------------+------------+--+

classpath下加入jar包，创建临时函数，测试

0: jdbc:hive2://localhost:10000> add jar /root/hive.jar;
INFO  : Added [/root/hive.jar] to class path
INFO  : Added resources: [/root/hive.jar]
No rows affected (0.236 seconds)
0: jdbc:hive2://localhost:10000> create temporary function getprovince as 'com.scu.hive.PhoneNumParse';
No rows affected (0.038 seconds)
0: jdbc:hive2://localhost:10000> select phonenum,getprovince(phonenum),flow from flow;
+-----------+-----------+-------+--+
| phonenum  |    _c1    | flow  |
+-----------+-----------+-------+--+
| 1367788   | beijing   | 1     |
| 1367788   | beijing   | 10    |
| 1377788   | shanghai  | 80    |
| 1377788   | shanghai  | 97    |
| 1387788   | shenzhen  | 98    |
| 1387788   | shenzhen  | 99    |
| 1387788   | shenzhen  | 100   |
| 1555118   | foreign   | 99    |
+-----------+-----------+-------+--+

Json数据解析UDF开发
有文件，内容一部分如下，里面都是json串，现在需要将它展示输出到表中，并解析对应为4个字段。

{"movie":"1193","rate":"5","timeStamp":"978300760","uid":"1"}
{"movie":"661","rate":"3","timeStamp":"978302109","uid":"1"}
{"movie":"914","rate":"3","timeStamp":"978301968","uid":"1"}
{"movie":"3408","rate":"4","timeStamp":"978300275","uid":"1"}
{"movie":"2355","rate":"5","timeStamp":"978824291","uid":"1"}
{"movie":"1197","rate":"3","timeStamp":"978302268","uid":"1"}

java类

public class JsonParse extends UDF{
    //{"movie":"1193","rate":"5","timeStamp":"978300760","uid":"1"}
    //输出字符串 1193 5 978300760 1
    public static String evaluate(String line){
        ObjectMapper objectMapper = new ObjectMapper();
        //json串转java对象
        String json = "";
        try {
            MovieRateBean bean = objectMapper.readValue(line,MovieRateBean.class);
            json = bean.toString();
        } catch (Exception e) {
            e.printStackTrace();
        }
        return json;
    }

}

public class MovieRateBean {
    private String movie;
    private String rate;//评分
    private String timeStamp;
    private String uid;
    @Override
    public String toString() {
        return  this.movie+"\t"+this.rate+"\t"+this.timeStamp+"\t"+this.uid;
    }
    get、set方法
}

工程打包上传到linux下。
创建表json

create table json(line string);

将文件导入到json表

load data local inpath ‘/root/json.data’ into table json;

0: jdbc:hive2://localhost:10000> select * from json limit 10;
+----------------------------------------------------------------+--+
|                           json.line                            |
+----------------------------------------------------------------+--+
| {"movie":"1193","rate":"5","timeStamp":"978300760","uid":"1"}  |
| {"movie":"661","rate":"3","timeStamp":"978302109","uid":"1"}   |
| {"movie":"914","rate":"3","timeStamp":"978301968","uid":"1"}   |
| {"movie":"3408","rate":"4","timeStamp":"978300275","uid":"1"}  |
| {"movie":"2355","rate":"5","timeStamp":"978824291","uid":"1"}  |
| {"movie":"1197","rate":"3","timeStamp":"978302268","uid":"1"}  |
| {"movie":"1287","rate":"5","timeStamp":"978302039","uid":"1"}  |
| {"movie":"2804","rate":"5","timeStamp":"978300719","uid":"1"}  |
| {"movie":"594","rate":"4","timeStamp":"978302268","uid":"1"}   |
| {"movie":"919","rate":"4","timeStamp":"978301368","uid":"1"}   |
+----------------------------------------------------------------+--+

0: jdbc:hive2://localhost:10000> add jar /root/hive3.jar;
INFO  : Added [/root/hive3.jar] to class path
INFO  : Added resources: [/root/hive3.jar]
No rows affected (0.023 seconds)
0: jdbc:hive2://localhost:10000> create temporary function parsejson as 'com.scu.hive.JsonParse';
No rows affected (0.07 seconds)
0: jdbc:hive2://localhost:10000> select parsejson(line) from json limit 10;
+---------------------+--+
|         _c0         |
+---------------------+--+
| 1193  5       978300760       1  |
| 661   3       978302109       1   |
| 914   3       978301968       1   |
| 3408  4       978300275       1  |
| 2355  5       978824291       1  |
| 1197  3       978302268       1  |
| 1287  5       978302039       1  |
| 2804  5       978300719       1  |
| 594   4       978302268       1   |
| 919   4       978301368       1   |
+---------------------+--+

到这里发现还有不足的地方，就是没显示字段。可以使用函数来实现重写建表来命名。

0: jdbc:hive2://localhost:10000> create table t_rating as
0: jdbc:hive2://localhost:10000> select split(parsejson(line),'\t')[0]as movieid,
0: jdbc:hive2://localhost:10000> split(parsejson(line),'\t')[1] as rate,
0: jdbc:hive2://localhost:10000> split(parsejson(line),'\t')[2] as timestring,
0: jdbc:hive2://localhost:10000> split(parsejson(line),'\t')[3] as uid 
0: jdbc:hive2://localhost:10000> from json limit 10;

0: jdbc:hive2://localhost:10000> select * from t_rating;
+-------------------+----------------+----------------------+---------------+--+
| t_rating.movieid  | t_rating.rate  | t_rating.timestring  | t_rating.uid  |
+-------------------+----------------+----------------------+---------------+--+
| 919               | 4              | 978301368            | 1             |
| 594               | 4              | 978302268            | 1             |
| 2804              | 5              | 978300719            | 1             |
| 1287              | 5              | 978302039            | 1             |
| 1197              | 3              | 978302268            | 1             |
| 2355              | 5              | 978824291            | 1             |
| 3408              | 4              | 978300275            | 1             |
| 914               | 3              | 978301968            | 1             |
| 661               | 3              | 978302109            | 1             |
| 1193              | 5              | 978300760            | 1             |
+-------------------+----------------+----------------------+---------------+--+

transform关键字使用
需求，创建新表，内容与t_rating表一致，但是第三个字段时间戳要改为输出周几。
Hive的 TRANSFORM 关键字提供了在SQL中调用自写脚本的功能
适合实现Hive中没有的功能又不想写UDF的情况。

1、编写python脚本(先看看机器有没有python)，用来将表时间戳转为周几
2、加入编写的py文件
3、创建新表，字段值为t_rating表传入py函数后输出的字段值

[root@mini1 ~]# python
Python 2.6.6 (r266:84292, Feb 21 2013, 23:54:59) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-3)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>>  print 'hello';
hello
>>> quit()
[root@mini1 ~]# vi weekday_mapper.py;
#import sys
import datetime

for line in sys.stdin:
  line = line.strip()
  movieid, rating, unixtime,userid = line.split('\t')
  weekday = datetime.datetime.fromtimestamp(float(unixtime)).isoweekday()
  print '\t'.join([movieid, rating, str(weekday),userid])

切换到hive客户端

0: jdbc:hive2://localhost:10000> add FILE /root/weekday_mapper.py;
1
0: jdbc:hive2://localhost:10000> create TABLE u_data_new as
0: jdbc:hive2://localhost:10000> SELECT
0: jdbc:hive2://localhost:10000>   TRANSFORM (movieid, rate, timestring,uid)
0: jdbc:hive2://localhost:10000>   USING 'python weekday_mapper.py'
0: jdbc:hive2://localhost:10000>   AS (movieid, rate, weekday,uid)
0: jdbc:hive2://localhost:10000> FROM t_rating;
...
0: jdbc:hive2://localhost:10000> select * from u_data_new;
+---------------------+------------------+---------------------+-----------------+--+
| u_data_new.movieid  | u_data_new.rate  | u_data_new.weekday  | u_data_new.uid  |
+---------------------+------------------+---------------------+-----------------+--+
| 919                 | 4                | 1                   | 1               |
| 594                 | 4                | 1                   | 1               |
| 2804                | 5                | 1                   | 1               |
| 1287                | 5                | 1                   | 1               |
| 1197                | 3                | 1                   | 1               |
| 2355                | 5                | 7                   | 1               |
| 3408                | 4                | 1                   | 1               |
| 914                 | 3                | 1                   | 1               |
| 661                 | 3                | 1                   | 1               |
| 1193                | 5                | 1                   | 1               |
+---------------------+------------------+---------------------+-----------------+--+

人面猴
序言：七十年代末，一起剥皮案震惊了整个滨河市，随后出现的几起案子，更是在滨河造成了极大的恐慌，老刑警刘岩，带你破解...
沈念sama阅读 156,757评论 4赞 359
死咒
序言：滨河连续发生了三起死亡事件，死亡现场离奇诡异，居然都是意外死亡，警方通过查阅死者的电脑和手机，发现死者居然都...
沈念sama阅读 66,478评论 1赞 289
救了他两次的神仙让他今天三更去死
文/潘晓璐我一进店门，熙熙楼的掌柜王于贵愁眉苦脸地迎上来，“玉大人，你说我怎么就摊上这事。” “怎么了？”我有些...
开封第一讲书人阅读 106,540评论 0赞 237
道士缉凶录：失踪的卖姜人
文/不坏的土叔我叫张陵，是天一观的道长。经常有香客问我，道长，这世上最难降的妖魔是什么？我笑而不...
开封第一讲书人阅读 43,593评论 0赞 203
港岛之恋（遗憾婚礼）
正文为了忘掉前任，我火速办了婚礼，结果婚礼上，老公的妹妹穿的比我还像新娘。我一直安慰自己，他们只是感情好，可当我...
茶点故事阅读 51,903评论 3赞 285
恶毒庶女顶嫁案：这布局不是一般人想出来的
文/花漫我一把揭开白布。她就那样静静地躺着，像睡着了一般。火红的嫁衣衬着肌肤如雪。梳的纹丝不乱的头发上，一...
开封第一讲书人阅读 40,329评论 1赞 210
城市分裂传说
那天，我揣着相机与录音，去河边找鬼。笑死，一个胖子当着我的面吹牛，可吹牛的内容都是我干的。我是一名探鬼主播，决...
沈念sama阅读 31,659评论 2赞 309
双鸳鸯连环套：你想象不到人心有多黑
文/苍兰香墨我猛地睁开眼，长吁一口气：“原来是场噩梦啊……” “哼！你这毒妇竟也来了？” 一声冷哼从身侧响起，我...
开封第一讲书人阅读 30,383评论 0赞 195
万荣杀人案实录
序言：老挝万荣一对情侣失踪，失踪者是张志新（化名）和其女友刘颖，没想到半个月后，有当地人在树林里发现了一具尸体，经...
沈念sama阅读 34,055评论 1赞 238
护林员之死
正文独居荒郊野岭守林人离奇死亡，尸身上长有42处带血的脓包…… 初始之章·张勋以下内容为张勋视角年9月15日...
茶点故事阅读 30,337评论 2赞 241
白月光启示录
正文我和宋清朗相恋三年，在试婚纱的时候发现自己被绿了。大学时的朋友给我发了我未婚夫和他白月光在一起吃饭的照片。...
茶点故事阅读 31,864评论 1赞 256
活死人
序言：一个原本活蹦乱跳的男人离奇死亡，死状恐怖，灵堂内的尸体忽然破棺而出，到底是诈尸还是另有隐情，我是刑警宁泽，带...
沈念sama阅读 28,227评论 2赞 251
日本核电站爆炸内幕
正文年R本政府宣布，位于F岛的核电站，受9级特大地震影响，放射性物质发生泄漏。R本人自食恶果不足惜，却给世界环境...
茶点故事阅读 32,820评论 3赞 231
男人毒药：我在死后第九天来索命
文/蒙蒙一、第九天我趴在偏房一处隐蔽的房顶上张望。院中可真热闹，春花似锦、人声如沸。这庄子的主人今日做“春日...
开封第一讲书人阅读 25,999评论 0赞 8
一桩弑父案，背后竟有这般阴谋
文/苍兰香墨我抬头看了看天上的太阳。三九已至，却和暖如春，着一层夹袄步出监牢的瞬间，已是汗流浃背。一阵脚步声响...
开封第一讲书人阅读 26,750评论 0赞 192
情欲美人皮
我被黑心中介骗来泰国打工，没想到刚下飞机就差点儿被人妖公主榨干…… 1. 我叫王不留，地道东北人。一个月前我还...
沈念sama阅读 35,365评论 2赞 269
代替公主和亲
正文我出身青楼，却偏偏与公主长得像，于是被迫代替她去往敌国和亲。传闻我的和亲对象是个残疾皇子，可洞房花烛夜当晚...
茶点故事阅读 35,260评论 2赞 258

Hive自定义函数与transform的使用

推荐阅读更多精彩内容