rexylib/include/rexy/cx/hashmap.hpp

/**
	This file is a part of rexy's general purpose library
	Copyright (C) 2020 rexy712

	This program is free software: you can redistribute it and/or modify
	it under the terms of the GNU Affero General Public License as published by
	the Free Software Foundation, either version 3 of the License, or
	(at your option) any later version.

	This program is distributed in the hope that it will be useful,
	but WITHOUT ANY WARRANTY; without even the implied warranty of
	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
	GNU Affero General Public License for more details.

	You should have received a copy of the GNU Affero General Public License
	along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/

#ifndef REXY_CX_HASHMAP_HPP
#define REXY_CX_HASHMAP_HPP

#include "vector.hpp"
#include "array.hpp"
#include "algorithm.hpp"
#include "hash.hpp"

#include <climits> //CHAR_BIT
#include <initializer_list>

namespace rexy::cx{

	template<class Key, class Value>
	struct element
	{
		Key key;
		Value value;
	};

	template<class Key, class Value, size_t N, class Hash = hash<Key>>
	class hashmap
	{
	public:
		static constexpr size_t single_bucket_bit = size_t{1} << ((sizeof(size_t)*CHAR_BIT) - 1);
		static constexpr size_t max_size = N;

		static_assert((max_size & single_bucket_bit) == 0);

		using key_type = Key;
		using value_type = Value;
		using hash_type = Hash;

	private:
		array<Value,N>  m_values; //perfect hash table
		array<size_t,N> m_g;      //'salt' values for indexing into the perfect hash table

	public:
		constexpr hashmap(const element<Key,Value>(&elements)[N]){
			array<vector<element<Key,Value>,N>,N> buckets;
			array<bool,N> slots_used;
			size_t current_bucket = 0;

			//place all keys into buckets
			for(auto& element : elements){
				buckets[Hash{}(element.key) % max_size].push_back(element);
			}

			//sort the buckets based on size, largest first
			quicksort(buckets.begin(), buckets.end(), [](auto&& left, auto&& right) -> bool{
				return left.size() > right.size();
			});

			//for each bucket, try different values of 'd' to try to find a hash that doesn't collide
			for(current_bucket = 0;current_bucket < buckets.size();++current_bucket){
				auto& bucket = buckets[current_bucket];
				//only handle buckets containing collisions
				if(bucket.size() <= 1)
					break;

				array<bool,N> pass_slots_used;
				vector<size_t,N> pass_slots;
				size_t d = 1;

				for(size_t i = 0;i < bucket.size();){
					size_t slot = Hash{}(bucket[i].key, d) % max_size;
					if(pass_slots_used[slot] || slots_used[slot]){
						//slot already in use, try another value for 'd'
						++d;
						i = 0;
						pass_slots_used.fill(false);
						pass_slots.clear();
					}else{
						//slot is good to go
						pass_slots_used[slot] = true;
						pass_slots.push_back(slot);
						++i;
					}
				}
				//store the successful value of 'd' at index of the first hash for this bucket
				m_g[Hash{}(bucket[0].key) % max_size] = d;

				//take the value from the temporary bucket into the permanent slot
				for(size_t i = 0;i < bucket.size();++i){
					m_values[pass_slots[i]] = std::move(bucket[i].value);
					slots_used[pass_slots[i]] = true;
				}
			}

			//Handle remaining single value buckets
			size_t next_free_slot = 0;

			for(;current_bucket < buckets.size();++current_bucket){
				auto& bucket = buckets[current_bucket];
				if(bucket.size() == 0)
					break;
				for(;slots_used[next_free_slot];++next_free_slot);

				m_g[Hash{}(bucket[0].key) % max_size] = (next_free_slot | single_bucket_bit);
				m_values[next_free_slot] = std::move(bucket[0].value);
				slots_used[next_free_slot] = true;
			}
		}

		//no key checks. give a correct key or get a random answer :)
		constexpr value_type& operator[](const Key& key){
			size_t d = m_g[Hash{}(key) % max_size];
			if(d & single_bucket_bit)
				return m_values[d & ~single_bucket_bit];
			return m_values[Hash{}(key, d) % max_size];
		}
		constexpr const value_type& operator[](const Key& key)const{
			size_t d = m_g[Hash{}(key) % max_size];
			if(d & single_bucket_bit)
				return m_values[d & ~single_bucket_bit];
			return m_values[Hash{}(key, d) % max_size];
		}
	};

	template<class Key, class Value, size_t N, class Hash = hash<Key>>
	constexpr auto make_hashmap(const element<Key,Value>(&list)[N]){
		return hashmap<Key,Value,N,Hash>(list);
	}
}

#ifdef REXY_STRING_BASE_HPP
#include "string_hash.hpp"
#endif

#endif