Reputation: 53
I have a sequence of string
val listOfString : Seq[String] = Seq("a","b","c")
How can I make a transform like
def addColumn(example: Seq[String]): DataFrame => DataFrame {
some code which returns a transform which add these String as column to dataframe
}
input
+-------
| id
+-------
| 1
+-------
output
+-------+-------+----+-------
| id | a | b | c
+-------+-------+----+-------
| 1 | 0 | 0 | 0
+-------+-------+----+-------
I am only interested in making it as transform
Upvotes: 1
Views: 1577
Reputation: 7336
You can use the transform
method of the datasets together with a single select
statement:
import org.apache.spark.sql.DataFrame
import org.apache.spark.sql.functions.lit
def addColumns(extraCols: Seq[String])(df: DataFrame): DataFrame = {
val selectCols = df.columns.map{col(_)} ++ extraCols.map{c => lit(0).as(c)}
df.select(selectCols :_*)
}
// usage example
val yourExtraColumns : Seq[String] = Seq("a","b","c")
df.transform(addColumns(yourExtraColumns))
Resources
https://towardsdatascience.com/dataframe-transform-spark-function-composition-eb8ec296c108
https://mungingdata.com/apache-spark/chaining-custom-dataframe-transformations/
Upvotes: 1
Reputation: 31540
Use .toDF()
and pass your listOfString.
Example:
//sample dataframe
df.show()
//+---+---+---+
//| _1| _2| _3|
//+---+---+---+
//| 0| 0| 0|
//+---+---+---+
df.toDF(listOfString:_*).show()
//+---+---+---+
//| a| b| c|
//+---+---+---+
//| 0| 0| 0|
//+---+---+---+
UPDATE:
Use foldLeft
to add the columns to the existing dataframe with values.
val df=Seq(("1")).toDF("id")
val listOfString : Seq[String] = Seq("a","b","c")
val new_df=listOfString.foldLeft(df){(df,colName) => df.withColumn(colName,lit("0"))}
//+---+---+---+---+
//| id| a| b| c|
//+---+---+---+---+
//| 1| 0| 0| 0|
//+---+---+---+---+
//or creating a function
import org.apache.spark.sql.DataFrame
def addColumns(extraCols: Seq[String],df: DataFrame): DataFrame = {
val new_df=extraCols.foldLeft(df){(df,colName) => df.withColumn(colName,lit("0"))}
return new_df
}
addColumns(listOfString,df).show()
//+---+---+---+---+
//| id| a| b| c|
//+---+---+---+---+
//| 1| 0| 0| 0|
//+---+---+---+---+
Upvotes: 1