All Products
Document Center

DataWorks:Create and use a MaxCompute UDF

Last Updated:Feb 22, 2024

If the built-in functions provided by MaxCompute cannot meet your business requirements, you can write code logic to create user-defined functions (UDFs) to meet your diverse business requirements.

Background information

UDFs allow you to write specific code based on your business requirements to extend the data processing capabilities. This topic describes how to create a MaxCompute UDF in a visualized manner in DataWorks. You can also use MaxCompute Studio or the CLI in MaxCompute to create a UDF. For more information, see Use MaxCompute Studio to create a UDF and Use the CLI in MaxCompute to create a UDF.


Before you can register a MaxCompute UDF, you must upload an existing resource or create a resource and add the resource to DataWorks in a visualized manner.


DataWorks allows you to view and manage UDFs that are uploaded in a visualized manner only in the DataWorks console. If you add UDFs to a MaxCompute compute engine by using other tools such as MaxCompute Studio, you must use the MaxCompute function feature in DataWorks DataStudio to manually load the UDFs to DataWorks. You can view and manage the UDFs in DataWorks after the loading is complete. For more information, see MaxCompute functions.

Register a function

  1. Go to the DataStudio page.

    Log on to the DataWorks console. In the left-side navigation pane, choose Data Modeling and Development > DataStudio. On the page that appears, select the desired workspace from the drop-down list and click Go to DataStudio.

  2. Create a workflow. For more information, see Create an auto triggered workflow.

  3. Create a function.

    1. Expand the desired workflow, right-click MaxCompute, and then select Create Function.

    2. In the Create Function dialog box, enter a function name and a path.

    3. Click Create.

    4. In the Register Function section of the configuration tab that appears, configure the parameters that are described in the following table.




      Function Type

      The type of the function. Valid values: Mathematical Operation Functions, Aggregate Functions, String Processing Functions, Date Functions, Window Functions, and Other Functions. For more information, see Use built-in functions.

      Engine Instance MaxCompute

      The MaxCompute compute engine instance. The value of this parameter cannot be changed.

      Function Name

      The name of the UDF. You can use this name to reference the UDF in SQL statements. The UDF name must be globally unique and cannot be changed after the UDF is registered.


      The owner of the function. The default owner is the account that is used to log on to the DataWorks console. You can change the value of this parameter.

      Class Name

      The name of the class that implements the UDF. Configure this parameter in the Resource name.Class name format. The resource name can be the name of a Java or Python package.

      When you register UDFs in the DataWorks console, you can reference MaxCompute resources including JAR packages and Python resources. The value format of this parameter varies based on the resource type:

      • If the resource type is JAR, configure the Class Name parameter in the JAR package name.Actual class name format. You can query the class name by executing the copy reference statement in IntelliJ IDEA.

        For example, if com.aliyun.odps.examples.udf is the Java package name and UDAFExample is the class name, the value of the Class Name parameter is com.aliyun.odps.examples.udf.UDAFExample.

      • If the resource type is Python, configure the Class Name parameter in the Python resource name.Actual class name format.

        For example, if LcLognormDist_sh is the Python resource name and LcLognormDist_sh is the class name, the value of the Class Name parameter is LcLognormDist_sh.LcLognormDist_sh.

        • You do not need to include the .jar or .py suffix in the resource name.

        • You can use a resource after the resource is committed and deployed. For information about how to create a MaxCompute resource, see Create and use MaxCompute resources.


      The resource that has been uploaded or added to DataWorks. You can select a resource by using the codeless UI or code editor.

      • You do not need to specify the path of the added resources.

      • If multiple resources are referenced in the UDF, separate the resource names with commas (,).


      The description of the UDF.

      Expression Syntax

      The syntax of the UDF. Example: test.

      Parameter Description

      The description of the input and output parameters that are supported.

      Return Value

      Optional. The return value. Example: 1.


      Optional. The example of the UDF.

  4. Click the 保存 icon in the top toolbar to save the UDF.

  5. Commit the UDF.

    1. Click the 提交 icon in the top toolbar.

    2. In the Submit dialog box, enter your comments in the Change description field.

    3. Click Confirm.

For information about how to view the functions in a MaxCompute compute engine and the change history of the functions, and how to perform other operations, see MaxCompute functions.

View the function version and roll back the function

For example, in the MaxCompute folder in the Business Flow section in the Scheduled Workflow pane, you can right-click a MaxCompute function and select View Earlier Versions to view the earlier versions of the function or roll back the function.


Use UDFs in nodes

If you want to use a UDF in a node, you can directly reference the name of the UDF in the code of the node. In detail, you can find the UDF in the Scheduled Workflow pane, right-click the UDF name, and then select Insert Function. This way, the UDF is displayed on the configuration tab of the node.节点中使用函数

Appendix 1: View UDFs

  • You can run the SHOW FUNCTIONS command to view all registered UDFs in a MaxCompute project that is associated with a specific DataWorks workspace as a compute engine instance.

  • MaxCompute provides various types of built-in functions. For more information, see Overview.

// View functions in the current project. 

Appendix 2: View the details of a UDF

  • You can run the DESCRIBE or abbreviated DESC command followed by a UDF name to view the details of a UDF.

    // Use the abbreviated DESC command to view the details of a UDF.
    DESC FUNCTION <function_name>;
  • In DataWorks, if the processing logic that is required in your workflow cannot be implemented by using existing functions, you can write a MaxCompute UDF and upload and associate a corresponding resource such as a JAR package or a Python file to manage and expand your data processing capabilities. For more information, see Manage MaxCompute resources.