topic Re: Is there a way to register a scala function that is available to other notebooks? in Data Engineering

Is there a way to register a scala function that is available to other notebooks?

qwerty1 — Tue, 25 Apr 2023 06:56:35 GMT

I am in a situation where I have a notebook that runs in a pipeline that creates a "live streaming table". So, I cannot use a language other than sql in the pipeline. I would like to format a certain column in the pipeline using a scala code (it's a complicated formatting and difficult to replicate in SQL).

Spark allows you to register scala methods as udf and access those registered methods in SQL.

But given my current situation (pipeline with DLT), I cannot include the scala method and the statement to register the method in spark context in the notebook.

Is there any work around here?

Re: Is there a way to register a scala function that is available to other notebooks?

-werners- — Tue, 25 Apr 2023 14:36:09 GMT

no, DLT does not work with Scala unfortunately.

Delta Live Tables are not vanilla spark.

Is python an option instead of scala?

Re: Is there a way to register a scala function that is available to other notebooks?

qwerty1 — Wed, 26 Apr 2023 06:57:42 GMT

yes, python is an option if I can use the library https://pypi.org/project/phonenumbers/

Re: Is there a way to register a scala function that is available to other notebooks?

-werners- — Wed, 26 Apr 2023 08:45:26 GMT

afaik you can create python udf's, but somehow I do not find the docs anymore.

https://docs.databricks.com/workflows/delta-live-tables/delta-live-tables-cookbook.html

and

https://docs.databricks.com/workflows/delta-live-tables/delta-live-tables-cookbook.html#import-python-modules-or-packages-from-a-databricks-repo

But they seem to be removed. If someone knows where to find these...