hashmap

Hashmap

A Red-Black Tree Hash Map implementation in Golang that uses user-supplied hashing algorithms.

Usage

The hashmap supports the classic Get, Insert, Remove operations you'd expect.

Inserting

//hash function must take interface{} and return int64
hashFunc := func(i interface{}) int64 {
    return int64(i.(int))
}

m := hashmap.New(hashFunc)
//insertion of key, value. keep in mind the key will be used as input to your hashFunc
m.Insert(4, 0)
//you can store different types
m.Insert(19, "hello")
//panics, because the hashFunc doesn't support string keys
m.Insert("fail", "oh no")

Selecting

hashFunc := func(i interface{}) int64 {
    return int64(i.(int))
}

m := hashmap.New(hashFunc)
m.Insert(4, 0)
m.Insert(19, 10)

//returns value as interface{} and found flag (true if the key was found)
value, found := m.Get(19)
// found will be false
value, found = m.Get(123)

Removing

hashFunc := func(i interface{}) int64 {
    return int64(i.(int))
}

m := hashmap.New(hashFunc)
m.Insert(4, 0)
m.Insert(19, 10)

//returns found flag (true if the key was found)
found := m.Remove(19)
// found will be false
found = m.Remove(123)

Type safety concerns

As this hash map supports keys and values of any type (by type hinting interface{}), there could be concerns of type safety and runtime problems. The suggested way to work around this is to wrap the hash map into type-specific proxy with methods such as Get(key KeyType) (value ValueType, found bool) and do the type assertions there.

Direct support for code generation by this package is still considered but not yet implemented.

TODO

• implement as thread safe
• threadsafety: don't lock the whole tree but separate nodes?
• CI
• Performance optimizations
• Performance tests and docs