spark RDD sort by two values

Question

I have a RDDof (name:String, popularity:Int, rank:Int). I want to sort this by rank and if rank matches then by popularity. I am doing so by two transformations.

var result = myRDD
        .sortBy(_._2, ascending = false)
        .sortBy(_._3, ascending = false)
        .take(10)

Can I do the it in one transformation?

Solution

You can try make an RDD of key value where key will be Tuple composed from rank and popularity and value will be name and sort by the key.

For example:

// _._1 - name

// _._2 - popularity

// _._3 - rank

var tupledRDD = myRDD.map(line => ((line._3, line._2), line._1))
.sortBy(_._1, ascending=false)
.take(10)