Hi ,

I am getting an syntax error when I run @@ROW_COUNT after the insert statement.

I am running the code in Databricks

Hi @Kaniz Fatma​ ,

I have tried the way you have mentioned but it still throws an error.

Hi ,

My requirement here is I will be creating a function using Python code to perform insert operation to a Delta table , that is why I am running it in an Python cell.

I will be passing a table name to that function and I need to get the number of records inserted into the table once the function is executed.

So any solution to achieve this?

Hi @Hubert Dudek​ 

Your approach is working for me.

Thank you.

Great! Please when you can select as best answer.

@Hubert Dudek​, when I execute a similar piece of code in VSCode executed through databricks-connect, the dataframe contains 1 row with no columns, which is a problem. Executing the same code in a notebook on the same cluster works as you stated. Is this possibly a bug in databricks-connect?

Hi Hubert, will this work with all DML operations and on all Spark versions? And can i use NUM_AFFECTED_ROWS for all DML operations?

Almost same advice than Hubert, I use the history of the delta table :

df_history.select(F.col('operationMetrics')).collect()[0].operationMetrics['numOutputRows']

You can find also other 'operationMetrics' values, like 'numTargetRowsDeleted'.

in case someone is looking for purely SQL based solution: (add LIMIT 1 to the query if you are looking for last op only)

select t.timestamp, t.operation, t.operationMetrics.numOutputRows as numOutputRows
from (
DESCRIBE HISTORY <catalog>.<schema>.<table>
) t
where t.operation like "%INSERT%"
order by t.timestamp desc