Creating custom User-Defined Functions (UDF)

Built-in functions in Hive sometimes do not fit the requirements of a business use case or when data analytics required some custom manipulation of data based on certain conditions. For such cases, the user needs to define custom logic as a UDF and run it over the data.

How to do it…

For writing a custom function in Hive, you will have to extend a Hive class: org.apache.hadoop.hive.ql.exec.UDF.

Let's understand the concept of creating a custom UDF with the example of creating a function to reverse a string.

The following are the steps to create a custom UDF:

1. Create a new Java project using any IDE, such as Eclipse. Give this any name; let's say "HiveUDF".

2. When writing a custom UDF, there should be two libraries in the classpath of the project. To do so:

   1. Create a folder "lib" under the project.

   2. Add the following two JAR files to the lib folder:

      • hadoop-common-2.6.0

      • hive-exec-1.2.1

   3. Add these JAR files to the classpath of the project. Right-click on Project...