How to operate a row depending on condition of previous row in Spark?

Question

We can easily operate a row in Spark, e.g. transform to capital letter, drop the last character, etc.

Question: how can we operate a row depending on the condition fr