When getting the different counts (single_field, multi_field, etc) the sum() is returning an int64 (since the df.sum() function returns that type). This isn't serializable and causes clients to have to cast the counts to int if they want to stream the results. Cast to int here so downstream clients don't have to worry about that.
When getting the different counts (single_field, multi_field, etc) the sum() is returning an int64 (since the df.sum() function returns that type). This isn't serializable and causes clients to have to cast the counts to int if they want to stream the results. Cast to int here so downstream clients don't have to worry about that.