Spark dataframe replaces empty characters (or other values) in each column with null

var offlineData: DataFrame = ......

val offlineDataFieldName: Array[String] = offlineData.schema.names
    (0 until offlineDataFieldName.length).foreach(idx => {
    
      offlineData = offlineData.withColumn(
        offlineDataFieldName.apply(idx)
        , when(col(offlineDataFieldName.apply(idx)) === "" or col(offlineDataFieldName.apply(idx)) === " Other values ", lit(null)).otherwise(col(offlineDataFieldName.apply(idx)))
      )
    })

Reference material ：
https://www.thinbug.com/q/44102133