RandomizedCollection is a data structure that contains a collection of numbers, possibly duplicates (i.e., a multiset). It should support inserting and removing specific elements and also reporting a random element.

Implement the RandomizedCollection class:

  • RandomizedCollection() Initializes the empty RandomizedCollection object.
  • bool insert(int val) Inserts an item val into the multiset, even if the item is already present. Returns true if the item is not present, false otherwise.
  • bool remove(int val) Removes an item val from the multiset if present. Returns true if the item is present, false otherwise. Note that if val has multiple occurrences in the multiset, we only remove one of them.
  • int getRandom() Returns a random element from the current multiset of elements. The probability of each element being returned is linearly related to the number of the same values the multiset contains.

You must implement the functions of the class such that each function works on average O(1) time complexity.

Note: The test cases are generated such that getRandom will only be called if there is at least one item in the RandomizedCollection.

solution

Very similar to 380-insert-delete-getrandom-o(1), but we keep a set of all the indices where a value exists instead of just a single number.

Note that when we delete, we don’t care which one we delete, so an unordered set works perfectly fine here.

class RandomizedCollection:
	def __init__(self):
		self.a = []
		self.idx = defaultdict(set)
	  
	def insert(self, val: int) -> bool:
		self.idx[val].add(len(self.a))
		self.a.append(val)
		return len(self.idx[val]) == 1
	  
	def remove(self, val: int) -> bool:
		if not self.idx[val]:
			return False
		to_remove, last = self.idx[val].pop(), self.a[-1]
		# copy value at end into index that contains val
		self.a[to_remove] = last
		self.idx[last].add(to_remove)
		self.idx[last].remove(len(self.a)-1)
		  
		self.a.pop()
		return True
	  
	def getRandom(self) -> int:
		return random.choice(self.a)