As far as I know, in general, there isn't a way to distinguish explicit null values from missing ones. (Someone please correct me if I'm wrong, since I would love to be able to do this for my own reasons). If you really must do it, and don't care about performance at all (since it will be horrible), read each object as a separate batch, while inferring the schema. If the schema contains the column, but the value is null, you will know it was explicitly set that way. If the schema doesn't contain the column, you'll know it was missing.
On Tue, Jun 23, 2020 at 7:34 AM Harmanat Singh <wish.man...@gmail.com> wrote: > Hi > > Please look at my issue from the link below. > > https://stackoverflow.com/questions/62526118/how-to-differentiate-between-null-and-missing-mongogdb-values-in-a-spark-datafra > > > Kindly Help > > > Best > Mannat >