Getting a reference to an immutable Map - scala

Question

I'm parallelising over a collection to count the number same item values in a List. The list in this case is uniqueSetOfLinks :
for (iListVal <- uniqueSetOfLinks.par) {
try {
val num : Int = listOfLinks.count(_.equalsIgnoreCase(iListVal))
linkTotals + iListVal -> num
}
catch {
case e : Exception => {
e.printStackTrace()
}
}
}
linkTotals is an immutable Map. To gain a reference to the total number of links do I need to update linkTotals so that it is immutable ?
I can then do something like :
linkTotals.put(iListVal, num)

Answer 1

You can't update immutable collection, all you can do is to combine immutable collection with addition element to get new immutable collection, like this:
val newLinkTotals = linkTotals + (iListVal -> num)
In case of collection you could create new collection of pairs and than add all pairs to the map:
val optPairs =
for (iListVal <- uniqueSetOfLinks.par)
yield
try {
val num : Int = listOfLinks.count(_.equalsIgnoreCase(iListVal))
Some(iListVal -> num)
}
catch {
case e : Exception => e.printStackTrace()
None
}
val newLinkTotals = linkTotals ++ optPairs.flatten // for non-empty initial map
val map = optPairs.flatten.toMap // in case there is no initial map
Note that you are using parallel collections (.par), so you should not use mutable state, like linkTotals += iListVal -> num.

Answer 2

Possible variation of #senia's answer (got rid of explicit flatten):
val optPairs =
(for {
iListVal <- uniqueSetOfLinks.par
count <- {
try
Some(listOfLinks.count(_.equalsIgnoreCase(iListVal)))
catch {
case e: Exception =>
e.printStackTrace()
None
}
}
} yield iListVal -> count) toMap

Answer 3

I think that you need some form of MapReduce in order to have parallel number of items estimation.
In your problem you already have all unique links. The partial intermediate result of map is simply a pair. And "reduce" is just toMap. So you can simply par-map the link to pair (link-> count) and then finally construct a map:
def count(iListVal:String) = listOfLinks.count(_.equalsIgnoreCase(iListVal))
val listOfPairs = uniqueSetOfLinks.par.map(iListVal => Try( (iListVal, count(iListVal)) ))
("map" operation is par-map)
Then remove exceptions:
val clearListOfPairs = listOfPairs.flatMap(_.toOption)
And then simply convert it to a map ("reduce"):
val linkTotals = clearListOfPairs.toMap
(if you need to check for exceptions, use Try.failure)