Spark SQL - Regexp_replace not giving desired result

Question

I am using Spark SQL to clean some data read from a CSV file.

The sample data looks like this:

1000616411022471|1000616415711839|0.10||ksetohouv|2