2022-untitled-game/code/src/engine/hash.c

#define HASH_CSTR(str) murmur_hash64(str, strlen(str), 0)
#define HASH_STR(str, len)  murmur_hash64(str, len, 0)

const uint64_t HASH_UNUSED = 0xFFFFFFFFFFFFFFFFU;

enum {
	NEAREST_PRIME_8 = 11,
	NEAREST_PRIME_16 = 17,
	NEAREST_PRIME_32 = 37,
	NEAREST_PRIME_64 = 67,
	NEAREST_PRIME_128 = 131,
	NEAREST_PRIME_256 = 257,
	NEAREST_PRIME_512 = 521,
	NEAREST_PRIME_1024 = 1031
};

typedef struct hash {
	int count;
	int capacity;
	int element_size;
	uint64_t* keys;
	void* data;
} hash_t;

hash_t
hash_create(size_t capacity, size_t element_size)
{
	hash_t result;

	result.count = 0;
	result.capacity = capacity;
	result.element_size = element_size;
	result.keys = engine_mem_alloc(capacity, sizeof(*result.keys));
	result.data = engine_mem_alloc(capacity, element_size);

	for (size_t i = 0; i < capacity; i++) {
		result.keys[i] = HASH_UNUSED;
	}

	return result;
}

void
hash_add(hash_t* hash, uint64_t key, void* element)
{
	int key_index = key % hash->capacity;

	LOG_ASSERT(hash->keys[key_index] == HASH_UNUSED, "Hash table collision");
	LOG_ASSERT(hash->count+1 < hash->capacity, "Hash table data array too small");

	uint64_t element_index = hash->count;
	hash->keys[key_index] = element_index;
	hash->count++;

	memcpy(&hash->data[element_index*hash->element_size], element, hash->element_size);
}

void*
hash_get(hash_t hash, uint64_t key)
{
	size_t key_index = key % hash.capacity;
	LOG_ASSERT(hash.keys[key_index] != HASH_UNUSED, "Key not found in hash table");

	uint64_t element_index = hash.keys[key_index];
	void* element = &hash.data[hash.element_size*element_index];

	return element;
}

uint32_t
murmur_hash32(const char* key, size_t len, uint32_t seed)
{
	const uint32_t C1 = 0xcc9e2d51;
	const uint32_t C2 = 0x1b873593;
	const uint32_t N  = 0xe6546b64;
	const int R1 = 15;
	const int R2 = 13;
	const int M  = 5;

	uint32_t hash = seed;

	const char* chunk = key;
	size_t chunk_size = sizeof(uint32_t);
	int chunk_count = len / chunk_size;
	for (int i = 0; i < chunk_count; i++)
	{
		uint32_t k = *(uint32_t*)chunk;
		k = k * C1;
		k = (k << R1) | (k >> (32-R1));
		k = k * C2;

		hash = hash ^ k;
		hash = (hash << R2) | (hash >> (32-R2));
		hash = (hash * M) + N;

		chunk += chunk_size;
	}

	int remainder_count = len % chunk_size;
	uint32_t k = 0;
	switch (remainder_count)
	{
		case 3: k |= chunk[2] << 16;
		case 2: k |= chunk[1] << 8;
		case 1:
		{
			k |= chunk[0];

			k = k * C1;
			k = (k << R1) | (k >> (32-R1));
			k = k * C2;

			hash = hash ^ k;
		}
		default: break;
	}

	hash = hash ^ len;
	hash = hash ^ (hash >> 16);
	hash = hash * 0x85ebca6b;
	hash = hash ^ (hash >> 13);
	hash = hash * 0xc2b2ae35;
	hash = hash ^ (hash >> 16);

	return hash;
}

uint64_t
murmur_hash64(const char* key, size_t len, uint64_t seed)
{
	const uint64_t M = 0xc6a4a7935bd1e995;
	const int R = 47;

	uint64_t hash = seed ^ (len * M);

	const char* chunk = key;
	size_t chunk_size = sizeof(uint64_t);
	int chunk_count = len / chunk_size;
	for (int i = 0; i < chunk_count; i++)
	{
		uint64_t k = *(uint64_t*)chunk;
		k = k * M;
		k = k ^ (k >> R);
		k = k * M;

		hash = hash ^ k;
		hash = hash * M;

		chunk += chunk_size;
	}

	int remainder_count = len % chunk_size;
	switch (remainder_count)
	{
		case 7: hash ^= (uint64_t)chunk[6] << 48;
		case 6: hash ^= (uint64_t)chunk[5] << 40;
		case 5: hash ^= (uint64_t)chunk[4] << 32;
		case 4: hash ^= (uint64_t)chunk[3] << 24;
		case 3: hash ^= (uint64_t)chunk[2] << 16;
		case 2: hash ^= (uint64_t)chunk[1] << 8;
		case 1:
		{
			hash ^= (uint64_t)chunk[0];
			hash = hash * M;
		}
		default: break;
	}

	hash = hash ^ (hash >> R);
	hash = hash * M;
	hash = hash ^ (hash >> R);

	return hash;
}