Python and Java support for serverless tasks¶

Serverless tasks can invoke the following object types and functions: user-defined functions (UDFs) and stored procedures written in Python, Java, and Scala.

You can use Python or Java in your tasks in a few different ways. To understand the difference between these options, see Choosing whether to write a stored procedure or a user-defined function.

User-defined functions¶

You can create UDFs to call in your task’s AS clause. You can use UDFs to perform operations not available in SQL. For more information about UDFs, see User-defined functions overview.

The following examples in Python and Java create a function that adds one to the input value.

CREATE OR REPLACE FUNCTION addone(i int)
  RETURNS int
  LANGUAGE python
  RUNTIME_VERSION = '3.8'
  HANDLER = 'addone_py'
  AS
    $$
    def addone_py(i):
      return i+1
    $$;

Copy

CREATE OR REPLACE FUNCTION add_one(i int)
  RETURNS int
  LANGUAGE java
  CALLED ON NULL INPUT
  HANDLER = 'TestFunc.addOne'
  TARGET_PATH = '@~/testfunc.jar'
  AS
    'class TestFunc {
      public static int addOne(int i) {
        return i+1;
      }
    }';

Copy

The following examples create my_task2 that adds one to the return value of my_task1.

CREATE OR REPLACE TASK IF NOT EXISTS my_task2
  AFTER my_task1
  AS
    SELECT addone(SYSTEM$GET_PREDECESSOR_RETURN_VALUE());

Copy

CREATE OR REPLACE TASK IF NOT EXISTS my_task2
  AFTER my_task1
  AS
    SELECT add_one(SYSTEM$GET_PREDECESSOR_RETURN_VALUE());

Copy

Stored procedures¶

You can create stored procedures to call in your task’s AS clause. Stored procedures generally perform administrative operations by executing SQL statements. For more information about stored procedures, see Stored procedures overview.

The following examples in Python and Java accept a table name and role name to return a filtered table with rows that match the specified role.

CREATE OR REPLACE PROCEDURE filterByRole(tableName VARCHAR, role VARCHAR)
  RETURNS TABLE(id NUMBER, name VARCHAR, role VARCHAR)
  LANGUAGE PYTHON
  RUNTIME_VERSION = '3.8'
  PACKAGES = ('snowflake-snowpark-python')
  HANDLER = 'filter_by_role'
  AS
    $$
    from snowflake.snowpark.functions import col

    def filter_by_role(session, table_name, role):
      df = session.table(table_name)
      return df.filter(col("role") == role)
    $$;

Copy

CREATE OR REPLACE PROCEDURE filter_by_role(table_name VARCHAR, role VARCHAR)
  RETURNS TABLE()
  LANGUAGE JAVA
  RUNTIME_VERSION = '11'
  PACKAGES = ('com.snowflake:snowpark:latest')
  HANDLER = 'FilterClass.filterByRole'
  AS
    $$
    import com.snowflake.snowpark_java.*;

    public class FilterClass {
      public DataFrame filterByRole(Session session, String tableName, String role) {
        DataFrame table = session.table(tableName);
        DataFrame filteredRows = table.filter(Functions.col("role").equal_to(Functions.lit(role)));
        return filteredRows;
      }
    }
    $$;

Copy

The following examples create task2 that calls the stored procedure with the table returned from task1 and the role of dev.

CREATE OR REPLACE TASK IF NOT EXISTS my_task2
  AFTER my_task1
  AS
    CALL filterByRole(SYSTEM$GET_PREDECESSOR_RETURN_VALUE(), 'dev');

Copy

CREATE OR REPLACE TASK IF NOT EXISTS my_task2
  AFTER my_task1
  AS
    CALL filter_by_role(SYSTEM$GET_PREDECESSOR_RETURN_VALUE(), 'dev');

Copy

SQL AS clause¶

You can also define Python or Java code directly in the AS clause of your task definition.

The following example uses Python to set the return value of task2 to a string.

CREATE OR REPLACE TASK IF NOT EXISTS task2
  SCHEDULE = '1 minute'
  AS
    $$
    print(Task completed successfully.)
    $$
  ;

Copy