read

While preparing for a technical coding interview, and practising LeetCode questions, I created this cheatsheet.

As you know, coding interview requires interviewee to manipulate collection types (array, dictionary, set) expertly to solve mathematic puzzles.

Loop array

for value in array
for index in array.indices
for (index, value) in array.enumerated()

Loop array with range

for i in 0..<array.count // Traditional loop with indices

for v in array[2...] // Skip the first 2 elements
for v in array[..<(array.count-2)] // Skip the last 2 elements
for v in array.prefix(3) // The first 3
for v in array.suffix(3) // The last 3

Loop dictionary

for (key, value) in dictionary

Check if dictionary has key

dictionary.keys.contains("k")

// Alternatively, if the value type is not an `Optional`
if dictionary["k"] == nil // "k" is not in keys

Dictionary with Optional Value

Beware if you have a dictionary such as [String: Int?]. If you set a key to nil using dictionary["k"] = nil, what it does is to remove the key “k”.

The designed way to set is dictionary.updateValue(nil, forKey: "k") 😱 The key “k” will exist.

Special sequences

let tenOnes = repeatElement(1, count: 10) // Array(tenOnes)

stride(from: 0, to: 60, by: 5) // Every 5 min

Array CRUD

// Find
firstIndex(of: x) // O(n)

// Insert
append(x) // O(1)
append(contentsOf: anArray)
insert(x, at: i) // O(n)

It is important to note that append O(1) is much faster than insert.

// Remove
remove(at: i) // O(n)
removeAll(where: {..})
dropFirst(x)
let last = popLast() // O(1)

// Move
move(fromOffsets:toOffset:)

Mutate

sorted()
reversed()
shuffled()
swapAt()

There are more ops using your own predicate: max(by:), sorted(by:), filter(), map()

Partition

partition(by:) is a powerful API, and quite advanced. It reorders the sequence by a predicate, such that the right half satisfy the predicate (while left half does not).

It returns the index of the first index in the right half.

let p = array.partition { .. }
let h1 = list[..<p] // Left half
let h2 = list[p...] // Right half

Note that it is an unstable partition, which means the order in both half are not preserved. There is an internal halfStablePartition which will have the 1st half retain the original order, and also a stablePartition which will have both half retain the original order.

If don’t know Crusty, read this up.

Comparable, Equatable

To compare and sort, the type must implement the Comparable protocol.

extension MyClass: Comparable {
    static func < (lhs: MyClass, MyClass: Date) -> Bool {
        return lhs.foo < rhs.foo
    }
}

Comparable inherits Equatable protocol, so you might need to implement it too. In cases like struct where Equatable is provided by default, you need not implement it.

extension MyClass: Equatable {
    public static func == (lhs: MyClass, rhs: MyClass) -> Bool {
        return lhs.foo == rhs.foo
    }
}

Identity Comparison ===

There is an op === for identity comparison eg. are they the same instances?

// For example, in the above Equatable implementation, we could also compare their identities
return lhs === rhs

Dictionary O(1) access

An optimizing trick is to make use of the quick O(1) operation when accessing a dictionary, since they are hashed.

The trade off is that the setup takes O(n), and a space of O(n). And you need the type to be hashable.

Hashable

extension MyClass: Hashable {
    public func hash(into hasher: inout Hasher) {
        hasher.combine(ObjectIdentifier(self).hashValue)
    }
}