andrearota

// Problem: creating a Spark UDF that take extra parameter at invocation time.
// Solution: using currying
// http://stackoverflow.com/questions/35546576/how-can-i-pass-extra-parameters-to-udfs-in-sparksql

// We want to create hideTabooValues, a Spark UDF that set to -1 fields that contains any of given taboo values.
// E.g. forbiddenValues = [1, 2, 3]
//      dataframe = [1, 2, 3, 4, 5, 6]
//      dataframe.select(hideTabooValues(forbiddenValues)) :> [-1, -1, -1, 4, 5, 6]
//
// Implementing this in Spark, we find two major issues:

raelg

package services

import java.lang.reflect.Type

import com.google.gson._
import org.joda.time.format.ISODateTimeFormat
import org.joda.time.{DateTime, DateTimeZone}

object Serializers {