CODEWITHSUNDEEP
scalaapache-sparkapache-spark-sql
Zied Hermi
Zied Hermi

Reputation: 229

How to change order of string value in column

how to use a scala udf to change the order of a string 

   root
    |-- Loc: string (nullable = true)
    +----------------+
    |             Loc|
    +----------------+
    |8106f510000dc502|
    +----------------+
8106f510000dc502 to 08f150000dc50261

i want to convert it like this order [3,1,5,7,6,(8-16),4,2]

Upvotes: 1

Views: 180

Answers (3)

Ramesh Maharjan
Ramesh Maharjan

Reputation: 41957

If you are looking to turn from 100z200 to 200z100 on column Loc then defining udf function as below should be sufficient (assuming that you have z in middle of every string in the column)

def reverseReplace = udf((str: String) => {
  val index = str.indexOf("z")
  str.substring(index+1, str.length)+str.substring(index, index+1)+str.substring(0, index)
})

You can call the udf function as 

val m4=msc3.select("Loc").withColumn("Info", reverseReplace($"Loc"))
m4.show(false)

You will have following output

+-------+-------+
|Loc    |Info   |
+-------+-------+
|100z200|200z100|
|30z400 |400z30 |
|600z10 |10z600 |
+-------+-------+

Edited

According to what I understand from your updated question that you want your final result in [3,1,5,7,6,(8-16),4,2] order, following can be your udf function

def reverseReplace = udf((str: String) => {
  val len = str.length
  val index = 16 > len match {case true => len case false => 16}
  var finalStr = ""
  if(len > 2)
    finalStr += str.substring(3-1,3)
  if(len > 0)
    finalStr += str.substring(1-1,1)
  if(len > 4)
    finalStr += str.substring(5-1,5)
  if(len > 6)
    finalStr += str.substring(7-1,7)
  if(len > 5)
    finalStr += str.substring(6-1,6)
  if(len > 7)
    finalStr += str.substring(8-1, index)
  if(len > 3)
    finalStr += str.substring(4-1,4)
  if(len > 1)
    finalStr += str.substring(2-1,2)
  if(finalStr == "")
    finalStr = str
  finalStr
})

You can call this udf function as stated above

Upvotes: 0

Jacek Laskowski
Jacek Laskowski

Reputation: 74619

That appears like a Scala coding assignment and has almost nothing to do with Spark.

I'd do the following:

// the dataset
val loc = Seq("8106f510000dc502").toDF("Loc")

// the udf for decoding loc
def mydecode(codes: Seq[Int]) = udf { s: String =>
  codes.map(pos => s.charAt(pos)).mkString
}

val codes = Seq(3,1,5,7,6,4,2)
val decoded = loc.withColumn("decoded", mydecode(codes)($"loc"))
scala> decoded.show
+----------------+-------+
|             Loc|decoded|
+----------------+-------+
|8106f510000dc502|61501f0|
+----------------+-------+

I'm leaving the range in the codes array, i.e. (8-16) as your home exercise.

Upvotes: 2

Leo C
Leo C

Reputation: 22439

Another approach using regular expression and an UDF that can be assigned a separator (in this case "z"):

def flip(sep: String) = udf(
  (s: String) => {
    val pattern = s"""(.*?)${sep}(.*)""".r
    s match {
      case pattern(a, b) => b + sep + a
    }
  }
)

val df = Seq( ("100z200") ).toDF("Loc")

val dfFlipped = df.withColumn("Flipped", flip("z")($"Loc"))

dfFlipped.show
+-------+-------+
|    Loc|Flipped|
+-------+-------+
|100z200|200z100|
+-------+-------+

Upvotes: 0

Related Questions