Add Number of days column to Date Column in same dataframe for Spark Scala App

梦想与她 提交于 2019-12-17 14:54:26

问题


I have a dataframe df of columns ("id", "current_date", "days") and I am trying to add the the "days" to "current_date" and create a new dataframe with new column called "new_date" using spark scala function date_add()

val newDF = df.withColumn("new_Date", date_add(df("current_date"), df("days").cast("Int")))

But looks like the function date_add only accepts Int values and not columns. How can get the desired output in such case? Are there any alternative functions i can use to get the desired output?

spark version: 1.6.0 scala version: 2.10.6


回答1:


A small custom udf can be used to make this date arithmetic possible.

import org.apache.spark.sql.functions.udf
import java.util.concurrent.TimeUnit
import java.util.Date
import java.text.SimpleDateFormat    

val date_add = udf((x: String, y: Int) => {
    val sdf = new SimpleDateFormat("yyyy-MM-dd")
    val result = new Date(sdf.parse(x).getTime() + TimeUnit.DAYS.toMillis(y))
  sdf.format(result)
} )

Usage:

scala> val df = Seq((1, "2017-01-01", 10), (2, "2017-01-01", 20)).toDF("id", "current_date", "days")
df: org.apache.spark.sql.DataFrame = [id: int, current_date: string, days: int]

scala> df.withColumn("new_Date", date_add($"current_date", $"days")).show()
+---+------------+----+----------+
| id|current_date|days|  new_Date|
+---+------------+----+----------+
|  1|  2017-01-01|  10|2017-01-11|
|  2|  2017-01-01|  20|2017-01-21|
+---+------------+----+----------+



回答2:


No need to use an UDF, you can do it using an SQL expression:

val newDF = df.withColumn("new_date", expr("date_add(current_date,days)"))


来源:https://stackoverflow.com/questions/44361332/add-number-of-days-column-to-date-column-in-same-dataframe-for-spark-scala-app

易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!