Insert Delete GetRandom O(1) - Duplicates Allowed

Process commands on a multiset supporting average O(1) insertion, removal, and uniform random sampling over stored occurrences.
insert x adds one occurrence and returns true exactly when x was previously absent.
remove x removes one occurrence and returns whether removal occurred.
getRandom returns a uniformly random stored occurrence.
Return one string result per command. For deterministic judging, every getRandom command occurs when all stored occurrences have the same value.

Function
processRandomizedCollection(operations: String[]) → String[]

Examples
Example 1
operations = ["insert 1","insert 1","insert 2","remove 1","remove 2","getRandom"]
return = ["true","false","true","true","true","1"]
One occurrence of 1 remains before getRandom.

Constraints
1 <= operations.length <= 10^4.
Every getRandom is issued on a nonempty collection whose distinct-value count is one.

The trick is two structures working together. Keep a dynamic array of every stored occurrence, and a hash map from value to a set of indices where that value sits in the array. Insert appends to the array, adds the index to the set, and returns true only if the set was empty before. Remove grabs any index of x, swaps the last array element into that slot, updates both sets, then pops the tail. The pitfall is the order of updates when the removed element is itself the last one. Do the set add and remove carefully or you'll corrupt the index. Here getRandom only fires when all stored values are identical, so you can return any element. Still, implement it properly. If the swap logic goes blank on you live, StealthCoder is the hedge.

Drill it cold or hedge it with StealthCoder. Either way, don't walk into the OA hoping you remember the trick.