Dynamically specify pivot column in SQL

[erigaud]

Hello everyone !

I am looking for a way to dynamically specify pivot columns in a SQL query, so it can be used in a view. However we don't want to hard code the values that need to become columns, and would rather extract it from another table.

I've seen other posts on internet asking the same question but without any luck 😕

Here's what we have currently :

SELECT * FROM table PIVOT (sum(my_value)
FOR mapping in ('col1', 'col2', ...)

 Instead we want to use a table table_mapping containing all the values that should go in the list, is there any way to do that in sql ? 

Thank you !

[Kaniz · Accepted solution]

Hi @erigaud, In Databricks SQL, you can’t use a dynamic list of columns directly in the PIVOT clause. 

However, there is a workaround using DataFrames in PySpark. 

This approach allows you to pivot on the mapping column dynamically. The distinct values in the mapping column will become the columns in the pivoted DataFrame.

If you strictly want to use SQL, you might need to generate your SQL query string with the pivot col....