342 lines
11 KiB
C++
342 lines
11 KiB
C++
/*
|
|
* SPDX-FileCopyrightText: Dora "cat" <cat@thenight.club>
|
|
* SPDX-License-Identifier: MPL-2.0
|
|
*
|
|
* This Source Code Form is subject to the terms of the Mozilla Public License,
|
|
* v. 2.0. If a copy of the MPL was not distributed with this file, You can
|
|
* obtain one at http://mozilla.org/MPL/2.0/.
|
|
*/
|
|
#ifndef GUARD_TOURMALINE_DUALKEYMAP_H
|
|
#define GUARD_TOURMALINE_DUALKEYMAP_H
|
|
#include "../Systems/Logging.hpp"
|
|
#include "Hashing.hpp"
|
|
|
|
#include <array>
|
|
#include <cmath>
|
|
#include <concepts>
|
|
#include <cstddef>
|
|
#include <cstdint>
|
|
#include <format>
|
|
#include <functional>
|
|
#include <optional>
|
|
#include <stack>
|
|
#include <tuple>
|
|
#include <type_traits>
|
|
#include <utility>
|
|
#include <variant>
|
|
#include <vector>
|
|
|
|
namespace Tourmaline::Containers {
|
|
template <Hashable AKey, Hashable BKey, typename Value,
|
|
uint64_t baseReservation = 2048>
|
|
class DualkeyMap {
|
|
public:
|
|
using QueryResult =
|
|
std::pair<std::variant<std::monostate, std::reference_wrapper<const AKey>,
|
|
std::reference_wrapper<const BKey>>,
|
|
Value &>;
|
|
using Entry = std::tuple<const AKey &, const BKey &, Value &>;
|
|
|
|
DualkeyMap() { hashList.reserve(baseReservation); }
|
|
~DualkeyMap() {
|
|
// I'm sure there is a better way to do this
|
|
for (DualkeyHash *hash : hashList) {
|
|
if (hash != nullptr) [[likely]] {
|
|
delete hash;
|
|
}
|
|
}
|
|
}
|
|
|
|
Entry Insert(AKey firstKey, BKey secondKey, Value value) {
|
|
DualkeyHash *hash = new DualkeyHash(std::move(firstKey),
|
|
std::move(secondKey), std::move(value));
|
|
|
|
if (graveyard.empty()) {
|
|
hashList.push_back(hash);
|
|
} else {
|
|
hashList[graveyard.top()] = hash;
|
|
graveyard.pop();
|
|
}
|
|
|
|
return {hash->firstKey, hash->secondKey, hash->value};
|
|
}
|
|
|
|
std::size_t Remove(std::optional<AKey> firstKey,
|
|
std::optional<BKey> secondKey) {
|
|
bool isFirstKeyGiven = firstKey.has_value();
|
|
bool isSecondKeyGiven = secondKey.has_value();
|
|
|
|
if (!isFirstKeyGiven && !isSecondKeyGiven) [[unlikely]] {
|
|
Systems::Logging::Log(
|
|
"Failed to Delete! DualkeyMap::Delete require at least 1 "
|
|
"key to be given! Terminating",
|
|
"Dualkey Map", Systems::Logging::LogLevel::Critical);
|
|
}
|
|
|
|
std::size_t firstKeyHash =
|
|
isFirstKeyGiven ? std::hash<AKey>{}(firstKey.value()) : 0;
|
|
std::size_t secondKeyHash =
|
|
isSecondKeyGiven ? std::hash<BKey>{}(secondKey.value()) : 0;
|
|
std::size_t index = 0, amountDeleted = 0;
|
|
uint8_t stateOfIndexing = isFirstKeyGiven + (isSecondKeyGiven << 1);
|
|
for (DualkeyHash *hash : hashList) {
|
|
// Tombstone
|
|
if (hash == nullptr) [[unlikely]] {
|
|
continue;
|
|
}
|
|
|
|
switch (stateOfIndexing) {
|
|
case 1: // Only first key is given
|
|
if (firstKeyHash == hash->firstKeyHash &&
|
|
firstKey.value() == hash->firstKey) {
|
|
delete hash;
|
|
hashList[index] = nullptr;
|
|
graveyard.push(index);
|
|
++amountDeleted;
|
|
}
|
|
break;
|
|
|
|
case 2: // Only second key is given
|
|
if (secondKeyHash == hash->secondKeyHash &&
|
|
secondKey.value() == hash->secondKey) {
|
|
delete hash;
|
|
hashList[index] = nullptr;
|
|
graveyard.push(index);
|
|
++amountDeleted;
|
|
}
|
|
break;
|
|
|
|
case 3:
|
|
if (firstKeyHash == hash->firstKeyHash &&
|
|
secondKeyHash == hash->secondKeyHash &&
|
|
firstKey.value() == hash->firstKey &&
|
|
secondKey.value() == hash->secondKey) {
|
|
delete hash;
|
|
hashList[index] = nullptr;
|
|
graveyard.push(index);
|
|
return 1;
|
|
}
|
|
break;
|
|
}
|
|
++index;
|
|
}
|
|
return amountDeleted;
|
|
}
|
|
|
|
[[nodiscard("Discarding an expensive query!")]]
|
|
std::vector<QueryResult> Query(std::optional<AKey> firstKey,
|
|
std::optional<BKey> secondKey) {
|
|
bool isFirstKeyGiven = firstKey.has_value();
|
|
bool isSecondKeyGiven = secondKey.has_value();
|
|
|
|
if (!isFirstKeyGiven && !isSecondKeyGiven) [[unlikely]] {
|
|
Systems::Logging::Log(
|
|
"Failed to Query! DualkeyMap::Query require at least 1 "
|
|
"key to be given! Terminating",
|
|
"Dualkey Map", Systems::Logging::LogLevel::Critical);
|
|
}
|
|
std::size_t firstKeyHash =
|
|
isFirstKeyGiven ? std::hash<AKey>{}(firstKey.value()) : 0;
|
|
std::size_t secondKeyHash =
|
|
isSecondKeyGiven ? std::hash<BKey>{}(secondKey.value()) : 0;
|
|
|
|
std::vector<QueryResult> finishedQuery{};
|
|
|
|
uint8_t stateOfIndexing = isFirstKeyGiven + (isSecondKeyGiven << 1);
|
|
// Putting hash checks first to benefit from short circuits
|
|
for (DualkeyHash *hash : hashList) {
|
|
// Tombstone
|
|
if (hash == nullptr) [[unlikely]] {
|
|
continue;
|
|
}
|
|
|
|
switch (stateOfIndexing) {
|
|
case 1: // Only first key is given
|
|
if (firstKeyHash == hash->firstKeyHash &&
|
|
firstKey.value() == hash->firstKey) {
|
|
finishedQuery.emplace_back(std::cref(hash->secondKey), hash->value);
|
|
}
|
|
continue;
|
|
case 2: // Only second key is given
|
|
if (secondKeyHash == hash->secondKeyHash &&
|
|
secondKey.value() == hash->secondKey) {
|
|
finishedQuery.emplace_back(std::cref(hash->firstKey), hash->value);
|
|
}
|
|
continue;
|
|
case 3: // Both are given
|
|
if (firstKeyHash == hash->firstKeyHash &&
|
|
secondKeyHash == hash->secondKeyHash &&
|
|
firstKey.value() == hash->firstKey &&
|
|
secondKey.value() == hash->secondKey) {
|
|
finishedQuery.emplace_back(std::monostate{}, hash->value);
|
|
break;
|
|
}
|
|
continue;
|
|
}
|
|
break;
|
|
}
|
|
|
|
return finishedQuery;
|
|
}
|
|
|
|
template <typename Key, std::size_t keyCount>
|
|
requires(std::same_as<Key, AKey> || std::same_as<Key, BKey>)
|
|
[[nodiscard("Discarding a very expensive query!")]]
|
|
int QueryWithAll(const Key (&keys)[keyCount]) {
|
|
auto queryResults = queryWithMany<Key>(keys);
|
|
|
|
for (const auto &queryRecord : queryResults) {
|
|
Systems::Logging::Log(
|
|
std::format("Opposite = {}, found = {}",
|
|
reinterpret_cast<uint64_t>(queryRecord.resultKey),
|
|
queryRecord.howManyFound),
|
|
"DKM", Systems::Logging::LogLevel::Info,
|
|
queryRecord.howManyFound == keyCount);
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
void Scan(std::function<bool(const std::size_t firstKeyHash,
|
|
const std::size_t secondKeyHash, Value &value)>
|
|
scanFunction) {
|
|
for (DualkeyHash *hash : hashList) {
|
|
if (hash == nullptr) {
|
|
continue;
|
|
}
|
|
if (scanFunction(hash->firstKeyHash, hash->secondKeyHash, hash->value)) {
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
void Scan(std::function<bool(const AKey &firstKey, const BKey &secondKey,
|
|
Value &value)>
|
|
scanFunction) {
|
|
for (DualkeyHash *hash : hashList) {
|
|
if (hash == nullptr) {
|
|
continue;
|
|
}
|
|
if (scanFunction(hash->firstKey, hash->secondKey, hash->value)) {
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
[[nodiscard]]
|
|
std::size_t Count() {
|
|
return hashList.size() - graveyard.size();
|
|
}
|
|
|
|
// No copying due to the container expected to be the sole
|
|
// owner of the data
|
|
DualkeyMap(const DualkeyMap &) = delete;
|
|
DualkeyMap &operator=(const DualkeyMap &) = delete;
|
|
|
|
private:
|
|
template <typename OppositeKey, std::size_t keyCount>
|
|
struct MultiQueryResult {
|
|
OppositeKey *resultKey = nullptr;
|
|
std::size_t howManyFound = 1;
|
|
std::array<Value *, keyCount> valueQueryResults;
|
|
};
|
|
|
|
struct DualkeyHash {
|
|
DualkeyHash(AKey &&firstKey, BKey &&secondKey, Value &&value)
|
|
: firstKey(std::move(firstKey)), secondKey(std::move(secondKey)),
|
|
firstKeyHash(std::hash<AKey>{}(this->firstKey)),
|
|
secondKeyHash(std::hash<BKey>{}(this->secondKey)),
|
|
value(std::move(value)) {}
|
|
|
|
const AKey firstKey;
|
|
const BKey secondKey;
|
|
const std::size_t firstKeyHash;
|
|
const std::size_t secondKeyHash;
|
|
mutable Value value;
|
|
};
|
|
|
|
template <typename Key, std::size_t keyCount>
|
|
inline std::vector<MultiQueryResult<
|
|
std::conditional_t<std::is_same_v<Key, AKey>, BKey, AKey>, keyCount>>
|
|
queryWithMany(const Key (&keys)[keyCount]) {
|
|
constexpr bool searchingInFirstKey = std::is_same_v<Key, AKey>;
|
|
|
|
// I really can't wait for C++26 contracts
|
|
if constexpr (keyCount == 0) {
|
|
Systems::Logging::Log("Failed to Query! QueryWithAll require at least 2 "
|
|
"key to be given, zero was given! Terminating",
|
|
"Dualkey Map",
|
|
Systems::Logging::LogLevel::Critical);
|
|
}
|
|
|
|
// Hoping this never ever gets triggered :sigh:
|
|
if constexpr (keyCount == 1) {
|
|
Systems::Logging::Log("QueryWithAll should not be used for single key "
|
|
"entry! Please use Query for this instead.",
|
|
"Dualkey Map", Systems::Logging::LogLevel::Error);
|
|
}
|
|
|
|
// While we don't necessary need the hashes,
|
|
// it just helps us tremendously benefit from short circuit checks
|
|
std::array<std::size_t, keyCount> keyHashes;
|
|
for (uint64_t index = 0; index < keyCount; index++) {
|
|
keyHashes[index] = std::hash<Key>{}(keys[index]);
|
|
}
|
|
|
|
uint64_t hashToCompare;
|
|
Key *keyToCompare;
|
|
std::conditional_t<searchingInFirstKey, BKey *, AKey *> resultKey;
|
|
std::vector<MultiQueryResult<
|
|
std::conditional_t<searchingInFirstKey, BKey, AKey>, keyCount>>
|
|
queryResults;
|
|
for (DualkeyHash *hash : hashList) {
|
|
// The hell of doing 2 conditions with similar logics in
|
|
// the same logical block
|
|
if constexpr (searchingInFirstKey) {
|
|
hashToCompare = hash->firstKeyHash;
|
|
keyToCompare = const_cast<AKey *>(&hash->firstKey);
|
|
resultKey = const_cast<BKey *>(&hash->secondKey);
|
|
} else {
|
|
hashToCompare = hash->secondKeyHash;
|
|
keyToCompare = const_cast<BKey *>(&hash->secondKey);
|
|
resultKey = const_cast<AKey *>(&hash->firstKey);
|
|
}
|
|
|
|
// The code above was done to make this code more uniform
|
|
for (uint64_t index = 0; index < keyCount; index++) {
|
|
if (keyHashes[index] == hashToCompare && keys[index] == *keyToCompare) {
|
|
|
|
bool doesExist = false;
|
|
for (auto &queryRecord : queryResults) {
|
|
if (*queryRecord.resultKey == *resultKey) {
|
|
queryRecord.valueQueryResults[index] = &hash->value;
|
|
++queryRecord.howManyFound;
|
|
doesExist = true;
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (doesExist) {
|
|
break;
|
|
}
|
|
|
|
// Since the result record is not present
|
|
// we have to make it
|
|
queryResults.emplace_back();
|
|
auto &newRecord = queryResults.back();
|
|
newRecord.resultKey = resultKey;
|
|
newRecord.valueQueryResults[index] = &hash->value;
|
|
}
|
|
}
|
|
}
|
|
|
|
return queryResults;
|
|
}
|
|
|
|
// It makes more sense to store the individual hash
|
|
std::vector<DualkeyHash *> hashList;
|
|
std::stack<std::size_t> graveyard;
|
|
};
|
|
} // namespace Tourmaline::Containers
|
|
#endif
|