Presto UDFs开发示例
GitHub presto-udfs开发支持了一部分HIVE UDFs，多谢分享，在此基础上进行我们的UDF开发学习。
- git代码presto-udfs，新建一个测试udf，
UdfTest输入一个字符串str，返回str_hjw
package com.qubole.presto.udfs.udfLearn;
import com.facebook.presto.spi.function.Description;
import com.facebook.presto.spi.function.ScalarFunction;
import com.facebook.presto.spi.function.SqlType;
import com.facebook.presto.spi.type.StandardTypes;
import io.airlift.slice.Slice;
import static io.airlift.slice.Slices.utf8Slice;
public class UdfTest
{
private UdfTest() {}
@Description("This is a udf test, return a result equals concat(input,'_hjw')")
@ScalarFunction("udf_test")
@SqlType(StandardTypes.VARCHAR)
public static Slice locateString(@SqlType(StandardTypes.VARCHAR) Slice input)
{
StringBuilder result = new StringBuilder();
result.append(input).append("_hjw");
return utf8Slice(result.toString());
}
}打包jar
mvn package在Preto根目录plugin目录下新建udfs文件夹
/presto-server-0.191/plugin$ ls
accumulo geospatial memory presto-thrift sqlserver
atop hive-hadoop2 ml raptor teradata-functions
blackhole jmx mongodb redis tpcds
cassandra kafka mysql redshift tpch
example-http localfile postgresql resource-group-managers udfs- 将打包好的udfs-2.0.2-SNAPSHOT.jar复制过来
/plugin/udfs$ ls
udfs-2.0.2-SNAPSHOT.jar- 重启服务(这里是单机，只启动本机服务即可，集群见
参考目录)
/bin$ ./launcher start
Started as 790- 启动CLI
Presto/CLI/presto$ ./presto --server localhost:8080 --catalog hive --schema default
presto:default>
presto:default> select udf_test('str')
-> ;
_col0
---------------------------------------------------
Slice{base=[B@6d9ac4da, address=16, length=3}_hjw
(1 row)
Query 20180212_052513_00002_ntdq9, FINISHED, 1 node
Splits: 17 total, 17 done (100.00%)
0:00 [0 rows, 0B] [0 rows/s, 0B/s]返回值的类型有点问题，暂时先放放…有空再检查下类型问题
Slice{base=[B@6d9ac4da, address=16, length=3}_hjw