update external headers

This commit is contained in:
Nick Fisher
2022-02-06 13:28:28 +08:00
parent 6d5a63d398
commit bd3d0d080b
150 changed files with 27445 additions and 14805 deletions

View File

@@ -0,0 +1,307 @@
/**
* MIT License
*
* Copyright (c) 2017 Thibaut Goetghebuer-Planchon <tessil@gmx.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_ARRAY_GROWTH_POLICY_H
#define TSL_ARRAY_GROWTH_POLICY_H
#include <algorithm>
#include <array>
#include <climits>
#include <cmath>
#include <cstddef>
#include <iterator>
#include <limits>
#include <ratio>
#include <stdexcept>
#ifdef __EXCEPTIONS
# define THROW(_e, _m) throw _e(_m)
#else
# include <stdio.h>
# ifndef NDEBUG
# define THROW(_e, _m) do { fprintf(stderr, _m); std::terminate(); } while(0)
# else
# define THROW(_e, _m) std::terminate()
# endif
#endif
namespace tsl {
namespace ah {
/**
* Grow the hash table by a factor of GrowthFactor keeping the bucket count to a power of two. It allows
* the table to use a mask operation instead of a modulo operation to map a hash to a bucket.
*
* GrowthFactor must be a power of two >= 2.
*/
template<std::size_t GrowthFactor>
class power_of_two_growth_policy {
public:
/**
* Called on the hash table creation and on rehash. The number of buckets for the table is passed in parameter.
* This number is a minimum, the policy may update this value with a higher value if needed (but not lower).
*
* If 0 is given, min_bucket_count_in_out must still be 0 after the policy creation and
* bucket_for_hash must always return 0 in this case.
*/
explicit power_of_two_growth_policy(std::size_t& min_bucket_count_in_out) {
if(min_bucket_count_in_out > max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
if(min_bucket_count_in_out > 0) {
min_bucket_count_in_out = round_up_to_power_of_two(min_bucket_count_in_out);
m_mask = min_bucket_count_in_out - 1;
}
else {
m_mask = 0;
}
}
/**
* Return the bucket [0, bucket_count()) to which the hash belongs.
* If bucket_count() is 0, it must always return 0.
*/
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return hash & m_mask;
}
/**
* Return the number of buckets that should be used on next growth.
*/
std::size_t next_bucket_count() const {
if((m_mask + 1) > max_bucket_count() / GrowthFactor) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
return (m_mask + 1) * GrowthFactor;
}
/**
* Return the maximum number of buckets supported by the policy.
*/
std::size_t max_bucket_count() const {
// Largest power of two.
return (std::numeric_limits<std::size_t>::max() / 2) + 1;
}
/**
* Reset the growth policy as if it was created with a bucket count of 0.
* After a clear, the policy must always return 0 when bucket_for_hash is called.
*/
void clear() noexcept {
m_mask = 0;
}
private:
static std::size_t round_up_to_power_of_two(std::size_t value) {
if(is_power_of_two(value)) {
return value;
}
if(value == 0) {
return 1;
}
--value;
for(std::size_t i = 1; i < sizeof(std::size_t) * CHAR_BIT; i *= 2) {
value |= value >> i;
}
return value + 1;
}
static constexpr bool is_power_of_two(std::size_t value) {
return value != 0 && (value & (value - 1)) == 0;
}
protected:
static_assert(is_power_of_two(GrowthFactor) && GrowthFactor >= 2, "GrowthFactor must be a power of two >= 2.");
std::size_t m_mask;
};
/**
* Grow the hash table by GrowthFactor::num / GrowthFactor::den and use a modulo to map a hash
* to a bucket. Slower but it can be useful if you want a slower growth.
*/
template<class GrowthFactor = std::ratio<3, 2>>
class mod_growth_policy {
public:
explicit mod_growth_policy(std::size_t& min_bucket_count_in_out) {
if(min_bucket_count_in_out > max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
if(min_bucket_count_in_out > 0) {
m_mod = min_bucket_count_in_out;
}
else {
m_mod = 1;
}
}
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return hash % m_mod;
}
std::size_t next_bucket_count() const {
if(m_mod == max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
const double next_bucket_count = std::ceil(double(m_mod) * REHASH_SIZE_MULTIPLICATION_FACTOR);
if(!std::isnormal(next_bucket_count)) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
if(next_bucket_count > double(max_bucket_count())) {
return max_bucket_count();
}
else {
return std::size_t(next_bucket_count);
}
}
std::size_t max_bucket_count() const {
return MAX_BUCKET_COUNT;
}
void clear() noexcept {
m_mod = 1;
}
private:
static constexpr double REHASH_SIZE_MULTIPLICATION_FACTOR = 1.0 * GrowthFactor::num / GrowthFactor::den;
static const std::size_t MAX_BUCKET_COUNT =
std::size_t(double(
std::numeric_limits<std::size_t>::max() / REHASH_SIZE_MULTIPLICATION_FACTOR
));
static_assert(REHASH_SIZE_MULTIPLICATION_FACTOR >= 1.1, "Growth factor should be >= 1.1.");
std::size_t m_mod;
};
namespace detail {
static constexpr const std::array<std::size_t, 40> PRIMES = {{
1ul, 5ul, 17ul, 29ul, 37ul, 53ul, 67ul, 79ul, 97ul, 131ul, 193ul, 257ul, 389ul, 521ul, 769ul, 1031ul,
1543ul, 2053ul, 3079ul, 6151ul, 12289ul, 24593ul, 49157ul, 98317ul, 196613ul, 393241ul, 786433ul,
1572869ul, 3145739ul, 6291469ul, 12582917ul, 25165843ul, 50331653ul, 100663319ul, 201326611ul,
402653189ul, 805306457ul, 1610612741ul, 3221225473ul, 4294967291ul
}};
template<unsigned int IPrime>
static constexpr std::size_t mod(std::size_t hash) { return hash % PRIMES[IPrime]; }
// MOD_PRIME[iprime](hash) returns hash % PRIMES[iprime]. This table allows for faster modulo as the
// compiler can optimize the modulo code better with a constant known at the compilation.
static constexpr const std::array<std::size_t(*)(std::size_t), 40> MOD_PRIME = {{
&mod<0>, &mod<1>, &mod<2>, &mod<3>, &mod<4>, &mod<5>, &mod<6>, &mod<7>, &mod<8>, &mod<9>, &mod<10>,
&mod<11>, &mod<12>, &mod<13>, &mod<14>, &mod<15>, &mod<16>, &mod<17>, &mod<18>, &mod<19>, &mod<20>,
&mod<21>, &mod<22>, &mod<23>, &mod<24>, &mod<25>, &mod<26>, &mod<27>, &mod<28>, &mod<29>, &mod<30>,
&mod<31>, &mod<32>, &mod<33>, &mod<34>, &mod<35>, &mod<36>, &mod<37> , &mod<38>, &mod<39>
}};
}
/**
* Grow the hash table by using prime numbers as bucket count. Slower than tsl::ah::power_of_two_growth_policy in
* general but will probably distribute the values around better in the buckets with a poor hash function.
*
* To allow the compiler to optimize the modulo operation, a lookup table is used with constant primes numbers.
*
* With a switch the code would look like:
* \code
* switch(iprime) { // iprime is the current prime of the hash table
* case 0: hash % 5ul;
* break;
* case 1: hash % 17ul;
* break;
* case 2: hash % 29ul;
* break;
* ...
* }
* \endcode
*
* Due to the constant variable in the modulo the compiler is able to optimize the operation
* by a series of multiplications, substractions and shifts.
*
* The 'hash % 5' could become something like 'hash - (hash * 0xCCCCCCCD) >> 34) * 5' in a 64 bits environment.
*/
class prime_growth_policy {
public:
explicit prime_growth_policy(std::size_t& min_bucket_count_in_out) {
auto it_prime = std::lower_bound(detail::PRIMES.begin(),
detail::PRIMES.end(), min_bucket_count_in_out);
if(it_prime == detail::PRIMES.end()) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
m_iprime = static_cast<unsigned int>(std::distance(detail::PRIMES.begin(), it_prime));
if(min_bucket_count_in_out > 0) {
min_bucket_count_in_out = *it_prime;
}
else {
min_bucket_count_in_out = 0;
}
}
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return detail::MOD_PRIME[m_iprime](hash);
}
std::size_t next_bucket_count() const {
if(m_iprime + 1 >= detail::PRIMES.size()) {
THROW(std::length_error, "The hash table exceeds its maximum size.");
}
return detail::PRIMES[m_iprime + 1];
}
std::size_t max_bucket_count() const {
return detail::PRIMES.back();
}
void clear() noexcept {
m_iprime = 0;
}
private:
unsigned int m_iprime;
static_assert(std::numeric_limits<decltype(m_iprime)>::max() >= detail::PRIMES.size(),
"The type of m_iprime is not big enough.");
};
}
}
#endif

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,863 @@
/**
* MIT License
*
* Copyright (c) 2017 Thibaut Goetghebuer-Planchon <tessil@gmx.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_ARRAY_MAP_H
#define TSL_ARRAY_MAP_H
#include <cstddef>
#include <cstdint>
#include <initializer_list>
#include <iterator>
#include <string>
#include <type_traits>
#include <utility>
#include "array_hash.h"
namespace tsl {
/**
* Implementation of a cache-conscious string hash map.
*
* The map stores the strings as `const CharT*`. If `StoreNullTerminator` is true,
* the strings are stored with the a null-terminator (the `key()` method of the iterators
* will return a pointer to this null-terminated string). Otherwise the null character
* is not stored (which allow an economy of 1 byte per string).
*
* The value `T` must be either nothrow move-constructible, copy-constructible or both.
*
* The size of a key string is limited to `std::numeric_limits<KeySizeT>::max() - 1`.
* That is 65 535 characters by default, but can be raised with the `KeySizeT` template parameter.
* See `max_key_size()` for an easy access to this limit.
*
* The number of elements in the map is limited to `std::numeric_limits<IndexSizeT>::max()`.
* That is 4 294 967 296 elements, but can be raised with the `IndexSizeT` template parameter.
* See `max_size()` for an easy access to this limit.
*
* Iterators invalidation:
* - clear, operator=: always invalidate the iterators.
* - insert, emplace, operator[]: always invalidate the iterators.
* - erase: always invalidate the iterators.
* - shrink_to_fit: always invalidate the iterators.
*/
template<class CharT,
class T,
class Hash = tsl::ah::str_hash<CharT>,
class KeyEqual = tsl::ah::str_equal<CharT>,
bool StoreNullTerminator = true,
class KeySizeT = std::uint16_t,
class IndexSizeT = std::uint32_t,
class GrowthPolicy = tsl::ah::power_of_two_growth_policy<2>>
class array_map {
private:
template<typename U>
using is_iterator = tsl::detail_array_hash::is_iterator<U>;
using ht = tsl::detail_array_hash::array_hash<CharT, T, Hash, KeyEqual, StoreNullTerminator,
KeySizeT, IndexSizeT, GrowthPolicy>;
public:
using char_type = typename ht::char_type;
using mapped_type = T;
using key_size_type = typename ht::key_size_type;
using index_size_type = typename ht::index_size_type;
using size_type = typename ht::size_type;
using hasher = typename ht::hasher;
using key_equal = typename ht::key_equal;
using iterator = typename ht::iterator;
using const_iterator = typename ht::const_iterator;
public:
array_map(): array_map(ht::DEFAULT_INIT_BUCKET_COUNT) {
}
explicit array_map(size_type bucket_count,
const Hash& hash = Hash()): m_ht(bucket_count, hash, ht::DEFAULT_MAX_LOAD_FACTOR)
{
}
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
array_map(InputIt first, InputIt last,
size_type bucket_count = ht::DEFAULT_INIT_BUCKET_COUNT,
const Hash& hash = Hash()): array_map(bucket_count, hash)
{
insert(first, last);
}
#ifdef TSL_AH_HAS_STRING_VIEW
array_map(std::initializer_list<std::pair<std::basic_string_view<CharT>, T>> init,
size_type bucket_count = ht::DEFAULT_INIT_BUCKET_COUNT,
const Hash& hash = Hash()): array_map(bucket_count, hash)
{
insert(init);
}
#else
array_map(std::initializer_list<std::pair<const CharT*, T>> init,
size_type bucket_count = ht::DEFAULT_INIT_BUCKET_COUNT,
const Hash& hash = Hash()): array_map(bucket_count, hash)
{
insert(init);
}
#endif
#ifdef TSL_AH_HAS_STRING_VIEW
array_map& operator=(std::initializer_list<std::pair<std::basic_string_view<CharT>, T>> ilist) {
clear();
reserve(ilist.size());
insert(ilist);
return *this;
}
#else
array_map& operator=(std::initializer_list<std::pair<const CharT*, T>> ilist) {
clear();
reserve(ilist.size());
insert(ilist);
return *this;
}
#endif
/*
* Iterators
*/
iterator begin() noexcept { return m_ht.begin(); }
const_iterator begin() const noexcept { return m_ht.begin(); }
const_iterator cbegin() const noexcept { return m_ht.cbegin(); }
iterator end() noexcept { return m_ht.end(); }
const_iterator end() const noexcept { return m_ht.end(); }
const_iterator cend() const noexcept { return m_ht.cend(); }
/*
* Capacity
*/
bool empty() const noexcept { return m_ht.empty(); }
size_type size() const noexcept { return m_ht.size(); }
size_type max_size() const noexcept { return m_ht.max_size(); }
size_type max_key_size() const noexcept { return m_ht.max_key_size(); }
void shrink_to_fit() { m_ht.shrink_to_fit(); }
/*
* Modifiers
*/
void clear() noexcept { m_ht.clear(); }
#ifdef TSL_AH_HAS_STRING_VIEW
std::pair<iterator, bool> insert(const std::basic_string_view<CharT>& key, const T& value) {
return m_ht.emplace(key.data(), key.size(), value);
}
#else
std::pair<iterator, bool> insert(const CharT* key, const T& value) {
return m_ht.emplace(key, std::char_traits<CharT>::length(key), value);
}
std::pair<iterator, bool> insert(const std::basic_string<CharT>& key, const T& value) {
return m_ht.emplace(key.data(), key.size(), value);
}
#endif
std::pair<iterator, bool> insert_ks(const CharT* key, size_type key_size, const T& value) {
return m_ht.emplace(key, key_size, value);
}
#ifdef TSL_AH_HAS_STRING_VIEW
std::pair<iterator, bool> insert(const std::basic_string_view<CharT>& key, T&& value) {
return m_ht.emplace(key.data(), key.size(), std::move(value));
}
#else
std::pair<iterator, bool> insert(const CharT* key, T&& value) {
return m_ht.emplace(key, std::char_traits<CharT>::length(key), std::move(value));
}
std::pair<iterator, bool> insert(const std::basic_string<CharT>& key, T&& value) {
return m_ht.emplace(key.data(), key.size(), std::move(value));
}
#endif
std::pair<iterator, bool> insert_ks(const CharT* key, size_type key_size, T&& value) {
return m_ht.emplace(key, key_size, std::move(value));
}
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
void insert(InputIt first, InputIt last) {
if(std::is_base_of<std::forward_iterator_tag,
typename std::iterator_traits<InputIt>::iterator_category>::value)
{
const auto nb_elements_insert = std::distance(first, last);
const std::size_t nb_free_buckets = std::size_t(float(bucket_count())*max_load_factor()) - size();
if(nb_elements_insert > 0 && nb_free_buckets < std::size_t(nb_elements_insert)) {
reserve(size() + std::size_t(nb_elements_insert));
}
}
for(auto it = first; it != last; ++it) {
insert_pair(*it);
}
}
#ifdef TSL_AH_HAS_STRING_VIEW
void insert(std::initializer_list<std::pair<std::basic_string_view<CharT>, T>> ilist) {
insert(ilist.begin(), ilist.end());
}
#else
void insert(std::initializer_list<std::pair<const CharT*, T>> ilist) {
insert(ilist.begin(), ilist.end());
}
#endif
#ifdef TSL_AH_HAS_STRING_VIEW
template<class M>
std::pair<iterator, bool> insert_or_assign(const std::basic_string_view<CharT>& key, M&& obj) {
return m_ht.insert_or_assign(key.data(), key.size(), std::forward<M>(obj));
}
#else
template<class M>
std::pair<iterator, bool> insert_or_assign(const CharT* key, M&& obj) {
return m_ht.insert_or_assign(key, std::char_traits<CharT>::length(key), std::forward<M>(obj));
}
template<class M>
std::pair<iterator, bool> insert_or_assign(const std::basic_string<CharT>& key, M&& obj) {
return m_ht.insert_or_assign(key.data(), key.size(), std::forward<M>(obj));
}
#endif
template<class M>
std::pair<iterator, bool> insert_or_assign_ks(const CharT* key, size_type key_size, M&& obj) {
return m_ht.insert_or_assign(key, key_size, std::forward<M>(obj));
}
#ifdef TSL_AH_HAS_STRING_VIEW
template<class... Args>
std::pair<iterator, bool> emplace(const std::basic_string_view<CharT>& key, Args&&... args) {
return m_ht.emplace(key.data(), key.size(), std::forward<Args>(args)...);
}
#else
template<class... Args>
std::pair<iterator, bool> emplace(const CharT* key, Args&&... args) {
return m_ht.emplace(key, std::char_traits<CharT>::length(key), std::forward<Args>(args)...);
}
template<class... Args>
std::pair<iterator, bool> emplace(const std::basic_string<CharT>& key, Args&&... args) {
return m_ht.emplace(key.data(), key.size(), std::forward<Args>(args)...);
}
#endif
template<class... Args>
std::pair<iterator, bool> emplace_ks(const CharT* key, size_type key_size, Args&&... args) {
return m_ht.emplace(key, key_size, std::forward<Args>(args)...);
}
/**
* Erase has an amortized O(1) runtime complexity, but even if it removes the key immediately,
* it doesn't do the same for the associated value T.
*
* T will only be removed when the ratio between the size of the map and
* the size of the map + the number of deleted values still stored is low enough.
*
* To force the deletion you can call shrink_to_fit.
*/
iterator erase(const_iterator pos) { return m_ht.erase(pos); }
/**
* @copydoc erase(const_iterator pos)
*/
iterator erase(const_iterator first, const_iterator last) { return m_ht.erase(first, last); }
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc erase(const_iterator pos)
*/
size_type erase(const std::basic_string_view<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#else
/**
* @copydoc erase(const_iterator pos)
*/
size_type erase(const CharT* key) {
return m_ht.erase(key, std::char_traits<CharT>::length(key));
}
/**
* @copydoc erase(const_iterator pos)
*/
size_type erase(const std::basic_string<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#endif
/**
* @copydoc erase(const_iterator pos)
*/
size_type erase_ks(const CharT* key, size_type key_size) {
return m_ht.erase(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
size_type erase(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.erase(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
size_type erase(const CharT* key, std::size_t precalculated_hash) {
return m_ht.erase(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
size_type erase(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.erase(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* @copydoc erase(const_iterator pos)
*
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
size_type erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.erase(key, key_size, precalculated_hash);
}
void swap(array_map& other) { other.m_ht.swap(m_ht); }
/*
* Lookup
*/
#ifdef TSL_AH_HAS_STRING_VIEW
T& at(const std::basic_string_view<CharT>& key) {
return m_ht.at(key.data(), key.size());
}
const T& at(const std::basic_string_view<CharT>& key) const {
return m_ht.at(key.data(), key.size());
}
#else
T& at(const CharT* key) {
return m_ht.at(key, std::char_traits<CharT>::length(key));
}
const T& at(const CharT* key) const {
return m_ht.at(key, std::char_traits<CharT>::length(key));
}
T& at(const std::basic_string<CharT>& key) {
return m_ht.at(key.data(), key.size());
}
const T& at(const std::basic_string<CharT>& key) const {
return m_ht.at(key.data(), key.size());
}
#endif
T& at_ks(const CharT* key, size_type key_size) {
return m_ht.at(key, key_size);
}
const T& at_ks(const CharT* key, size_type key_size) const {
return m_ht.at(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
T& at(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.at(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const T& at(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.at(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
T& at(const CharT* key, std::size_t precalculated_hash) {
return m_ht.at(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const T& at(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.at(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
T& at(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.at(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const T& at(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.at(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
T& at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.at(key, key_size, precalculated_hash);
}
/**
* @copydoc at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const T& at_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.at(key, key_size, precalculated_hash);
}
#ifdef TSL_AH_HAS_STRING_VIEW
T& operator[](const std::basic_string_view<CharT>& key) { return m_ht.access_operator(key.data(), key.size()); }
#else
T& operator[](const CharT* key) { return m_ht.access_operator(key, std::char_traits<CharT>::length(key)); }
T& operator[](const std::basic_string<CharT>& key) { return m_ht.access_operator(key.data(), key.size()); }
#endif
#ifdef TSL_AH_HAS_STRING_VIEW
size_type count(const std::basic_string_view<CharT>& key) const {
return m_ht.count(key.data(), key.size());
}
#else
size_type count(const CharT* key) const {
return m_ht.count(key, std::char_traits<CharT>::length(key));
}
size_type count(const std::basic_string<CharT>& key) const {
return m_ht.count(key.data(), key.size());
}
#endif
size_type count_ks(const CharT* key, size_type key_size) const {
return m_ht.count(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const
*/
size_type count(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.count(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const
*/
size_type count(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.count(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const
*/
size_type count(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.count(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
size_type count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.count(key, key_size, precalculated_hash);
}
#ifdef TSL_AH_HAS_STRING_VIEW
iterator find(const std::basic_string_view<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string_view<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#else
iterator find(const CharT* key) {
return m_ht.find(key, std::char_traits<CharT>::length(key));
}
const_iterator find(const CharT* key) const {
return m_ht.find(key, std::char_traits<CharT>::length(key));
}
iterator find(const std::basic_string<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#endif
iterator find_ks(const CharT* key, size_type key_size) {
return m_ht.find(key, key_size);
}
const_iterator find_ks(const CharT* key, size_type key_size) const {
return m_ht.find(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
iterator find(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
iterator find(const CharT* key, std::size_t precalculated_hash) {
return m_ht.find(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.find(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
iterator find(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
iterator find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.find(key, key_size, precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.find(key, key_size, precalculated_hash);
}
#ifdef TSL_AH_HAS_STRING_VIEW
std::pair<iterator, iterator> equal_range(const std::basic_string_view<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string_view<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#else
std::pair<iterator, iterator> equal_range(const CharT* key) {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key));
}
std::pair<const_iterator, const_iterator> equal_range(const CharT* key) const {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key));
}
std::pair<iterator, iterator> equal_range(const std::basic_string<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#endif
std::pair<iterator, iterator> equal_range_ks(const CharT* key, size_type key_size) {
return m_ht.equal_range(key, key_size);
}
std::pair<const_iterator, const_iterator> equal_range_ks(const CharT* key, size_type key_size) const {
return m_ht.equal_range(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<iterator, iterator> equal_range(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<iterator, iterator> equal_range(const CharT* key, std::size_t precalculated_hash) {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<iterator, iterator> equal_range(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
std::pair<iterator, iterator> equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.equal_range(key, key_size, precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.equal_range(key, key_size, precalculated_hash);
}
/*
* Bucket interface
*/
size_type bucket_count() const { return m_ht.bucket_count(); }
size_type max_bucket_count() const { return m_ht.max_bucket_count(); }
/*
* Hash policy
*/
float load_factor() const { return m_ht.load_factor(); }
float max_load_factor() const { return m_ht.max_load_factor(); }
void max_load_factor(float ml) { m_ht.max_load_factor(ml); }
void rehash(size_type count) { m_ht.rehash(count); }
void reserve(size_type count) { m_ht.reserve(count); }
/*
* Observers
*/
hasher hash_function() const { return m_ht.hash_function(); }
key_equal key_eq() const { return m_ht.key_eq(); }
/*
* Other
*/
/**
* Return the `const_iterator it` as an `iterator`.
*/
iterator mutable_iterator(const_iterator it) noexcept { return m_ht.mutable_iterator(it); }
/**
* Serialize the map through the `serializer` parameter.
*
* The `serializer` parameter must be a function object that supports the following calls:
* - `template<typename U> void operator()(const U& value);` where the types `std::uint64_t`, `float` and `T` must be supported for U.
* - `void operator()(const CharT* value, std::size_t value_size);`
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, ...) of the types it serializes
* in the hands of the `Serializer` function object if compatibility is required.
*/
template<class Serializer>
void serialize(Serializer& serializer) const {
m_ht.serialize(serializer);
}
/**
* Deserialize a previously serialized map through the `deserializer` parameter.
*
* The `deserializer` parameter must be a function object that supports the following calls:
* - `template<typename U> U operator()();` where the types `std::uint64_t`, `float` and `T` must be supported for U.
* - `void operator()(CharT* value_out, std::size_t value_size);`
*
* If the deserialized hash map type is hash compatible with the serialized map, the deserialization process can be
* sped up by setting `hash_compatible` to true. To be hash compatible, the Hash (take care of the 32-bits vs 64 bits),
* KeyEqual, GrowthPolicy, StoreNullTerminator, KeySizeT and IndexSizeT must behave the same than the ones used on the
* serialized map. Otherwise the behaviour is undefined with `hash_compatible` sets to true.
*
* The behaviour is undefined if the type `CharT` and `T` of the `array_map` are not the same as the
* types used during serialization.
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, size of int, ...) of the types it
* deserializes in the hands of the `Deserializer` function object if compatibility is required.
*/
template<class Deserializer>
static array_map deserialize(Deserializer& deserializer, bool hash_compatible = false) {
array_map map(0);
map.m_ht.deserialize(deserializer, hash_compatible);
return map;
}
friend bool operator==(const array_map& lhs, const array_map& rhs) {
if(lhs.size() != rhs.size()) {
return false;
}
for(auto it = lhs.cbegin(); it != lhs.cend(); ++it) {
const auto it_element_rhs = rhs.find_ks(it.key(), it.key_size());
if(it_element_rhs == rhs.cend() || it.value() != it_element_rhs.value()) {
return false;
}
}
return true;
}
friend bool operator!=(const array_map& lhs, const array_map& rhs) {
return !operator==(lhs, rhs);
}
friend void swap(array_map& lhs, array_map& rhs) {
lhs.swap(rhs);
}
private:
template<class U, class V>
void insert_pair(const std::pair<U, V>& value) {
insert(value.first, value.second);
}
template<class U, class V>
void insert_pair(std::pair<U, V>&& value) {
insert(value.first, std::move(value.second));
}
public:
static const size_type MAX_KEY_SIZE = ht::MAX_KEY_SIZE;
private:
ht m_ht;
};
/**
* Same as
* `tsl::array_map<CharT, T, Hash, KeyEqual, StoreNullTerminator, KeySizeT, IndexSizeT, tsl::ah::prime_growth_policy>`.
*/
template<class CharT,
class T,
class Hash = tsl::ah::str_hash<CharT>,
class KeyEqual = tsl::ah::str_equal<CharT>,
bool StoreNullTerminator = true,
class KeySizeT = std::uint16_t,
class IndexSizeT = std::uint32_t>
using array_pg_map = array_map<CharT, T, Hash, KeyEqual, StoreNullTerminator,
KeySizeT, IndexSizeT, tsl::ah::prime_growth_policy>;
} //end namespace tsl
#endif

View File

@@ -0,0 +1,664 @@
/**
* MIT License
*
* Copyright (c) 2017 Thibaut Goetghebuer-Planchon <tessil@gmx.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_ARRAY_SET_H
#define TSL_ARRAY_SET_H
#include <cstddef>
#include <cstdint>
#include <initializer_list>
#include <iterator>
#include <string>
#include <type_traits>
#include <utility>
#include "array_hash.h"
namespace tsl {
/**
* Implementation of a cache-conscious string hash set.
*
* The set stores the strings as `const CharT*`. If `StoreNullTerminator` is true,
* the strings are stored with the a null-terminator (the `key()` method of the iterators
* will return a pointer to this null-terminated string). Otherwise the null character
* is not stored (which allow an economy of 1 byte per string).
*
* The size of a key string is limited to `std::numeric_limits<KeySizeT>::max() - 1`.
* That is 65 535 characters by default, but can be raised with the `KeySizeT` template parameter.
* See `max_key_size()` for an easy access to this limit.
*
* The number of elements in the set is limited to `std::numeric_limits<IndexSizeT>::max()`.
* That is 4 294 967 296 elements, but can be raised with the `IndexSizeT` template parameter.
* See `max_size()` for an easy access to this limit.
*
* Iterators invalidation:
* - clear, operator=: always invalidate the iterators.
* - insert, emplace, operator[]: always invalidate the iterators.
* - erase: always invalidate the iterators.
* - shrink_to_fit: always invalidate the iterators.
*/
template<class CharT,
class Hash = tsl::ah::str_hash<CharT>,
class KeyEqual = tsl::ah::str_equal<CharT>,
bool StoreNullTerminator = true,
class KeySizeT = std::uint16_t,
class IndexSizeT = std::uint32_t,
class GrowthPolicy = tsl::ah::power_of_two_growth_policy<2>>
class array_set {
private:
template<typename U>
using is_iterator = tsl::detail_array_hash::is_iterator<U>;
using ht = tsl::detail_array_hash::array_hash<CharT, void, Hash, KeyEqual, StoreNullTerminator,
KeySizeT, IndexSizeT, GrowthPolicy>;
public:
using char_type = typename ht::char_type;
using key_size_type = typename ht::key_size_type;
using index_size_type = typename ht::index_size_type;
using size_type = typename ht::size_type;
using hasher = typename ht::hasher;
using key_equal = typename ht::key_equal;
using iterator = typename ht::iterator;
using const_iterator = typename ht::const_iterator;
array_set(): array_set(ht::DEFAULT_INIT_BUCKET_COUNT) {
}
explicit array_set(size_type bucket_count,
const Hash& hash = Hash()): m_ht(bucket_count, hash, ht::DEFAULT_MAX_LOAD_FACTOR)
{
}
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
array_set(InputIt first, InputIt last,
size_type bucket_count = ht::DEFAULT_INIT_BUCKET_COUNT,
const Hash& hash = Hash()): array_set(bucket_count, hash)
{
insert(first, last);
}
#ifdef TSL_AH_HAS_STRING_VIEW
array_set(std::initializer_list<std::basic_string_view<CharT>> init,
size_type bucket_count = ht::DEFAULT_INIT_BUCKET_COUNT,
const Hash& hash = Hash()): array_set(bucket_count, hash)
{
insert(init);
}
#else
array_set(std::initializer_list<const CharT*> init,
size_type bucket_count = ht::DEFAULT_INIT_BUCKET_COUNT,
const Hash& hash = Hash()): array_set(bucket_count, hash)
{
insert(init);
}
#endif
#ifdef TSL_AH_HAS_STRING_VIEW
array_set& operator=(std::initializer_list<std::basic_string_view<CharT>> ilist) {
clear();
reserve(ilist.size());
insert(ilist);
return *this;
}
#else
array_set& operator=(std::initializer_list<const CharT*> ilist) {
clear();
reserve(ilist.size());
insert(ilist);
return *this;
}
#endif
/*
* Iterators
*/
iterator begin() noexcept { return m_ht.begin(); }
const_iterator begin() const noexcept { return m_ht.begin(); }
const_iterator cbegin() const noexcept { return m_ht.cbegin(); }
iterator end() noexcept { return m_ht.end(); }
const_iterator end() const noexcept { return m_ht.end(); }
const_iterator cend() const noexcept { return m_ht.cend(); }
/*
* Capacity
*/
bool empty() const noexcept { return m_ht.empty(); }
size_type size() const noexcept { return m_ht.size(); }
size_type max_size() const noexcept { return m_ht.max_size(); }
size_type max_key_size() const noexcept { return m_ht.max_key_size(); }
void shrink_to_fit() { m_ht.shrink_to_fit(); }
/*
* Modifiers
*/
void clear() noexcept { m_ht.clear(); }
#ifdef TSL_AH_HAS_STRING_VIEW
std::pair<iterator, bool> insert(const std::basic_string_view<CharT>& key) {
return m_ht.emplace(key.data(), key.size());
}
#else
std::pair<iterator, bool> insert(const CharT* key) {
return m_ht.emplace(key, std::char_traits<CharT>::length(key));
}
std::pair<iterator, bool> insert(const std::basic_string<CharT>& key) {
return m_ht.emplace(key.data(), key.size());
}
#endif
std::pair<iterator, bool> insert_ks(const CharT* key, size_type key_size) {
return m_ht.emplace(key, key_size);
}
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
void insert(InputIt first, InputIt last) {
if(std::is_base_of<std::forward_iterator_tag,
typename std::iterator_traits<InputIt>::iterator_category>::value)
{
const auto nb_elements_insert = std::distance(first, last);
const std::size_t nb_free_buckets = std::size_t(float(bucket_count())*max_load_factor()) - size();
if(nb_elements_insert > 0 && nb_free_buckets < std::size_t(nb_elements_insert)) {
reserve(size() + std::size_t(nb_elements_insert));
}
}
for(auto it = first; it != last; ++it) {
insert(*it);
}
}
#ifdef TSL_AH_HAS_STRING_VIEW
void insert(std::initializer_list<std::basic_string_view<CharT>> ilist) {
insert(ilist.begin(), ilist.end());
}
#else
void insert(std::initializer_list<const CharT*> ilist) {
insert(ilist.begin(), ilist.end());
}
#endif
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc emplace_ks(const CharT* key, size_type key_size)
*/
std::pair<iterator, bool> emplace(const std::basic_string_view<CharT>& key) {
return m_ht.emplace(key.data(), key.size());
}
#else
/**
* @copydoc emplace_ks(const CharT* key, size_type key_size)
*/
std::pair<iterator, bool> emplace(const CharT* key) {
return m_ht.emplace(key, std::char_traits<CharT>::length(key));
}
/**
* @copydoc emplace_ks(const CharT* key, size_type key_size)
*/
std::pair<iterator, bool> emplace(const std::basic_string<CharT>& key) {
return m_ht.emplace(key.data(), key.size());
}
#endif
/**
* No difference compared to the insert method. Mainly here for coherence with array_map.
*/
std::pair<iterator, bool> emplace_ks(const CharT* key, size_type key_size) {
return m_ht.emplace(key, key_size);
}
iterator erase(const_iterator pos) { return m_ht.erase(pos); }
iterator erase(const_iterator first, const_iterator last) { return m_ht.erase(first, last); }
#ifdef TSL_AH_HAS_STRING_VIEW
size_type erase(const std::basic_string_view<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#else
size_type erase(const CharT* key) {
return m_ht.erase(key, std::char_traits<CharT>::length(key));
}
size_type erase(const std::basic_string<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#endif
size_type erase_ks(const CharT* key, size_type key_size) {
return m_ht.erase(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
size_type erase(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.erase(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
size_type erase(const CharT* key, std::size_t precalculated_hash) {
return m_ht.erase(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
size_type erase(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.erase(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
size_type erase_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.erase(key, key_size, precalculated_hash);
}
void swap(array_set& other) { other.m_ht.swap(m_ht); }
/*
* Lookup
*/
#ifdef TSL_AH_HAS_STRING_VIEW
size_type count(const std::basic_string_view<CharT>& key) const { return m_ht.count(key.data(), key.size()); }
#else
size_type count(const CharT* key) const { return m_ht.count(key, std::char_traits<CharT>::length(key)); }
size_type count(const std::basic_string<CharT>& key) const { return m_ht.count(key.data(), key.size()); }
#endif
size_type count_ks(const CharT* key, size_type key_size) const { return m_ht.count(key, key_size); }
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const
*/
size_type count(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.count(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const
*/
size_type count(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.count(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const
*/
size_type count(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.count(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
size_type count_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.count(key, key_size, precalculated_hash);
}
#ifdef TSL_AH_HAS_STRING_VIEW
iterator find(const std::basic_string_view<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string_view<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#else
iterator find(const CharT* key) {
return m_ht.find(key, std::char_traits<CharT>::length(key));
}
const_iterator find(const CharT* key) const {
return m_ht.find(key, std::char_traits<CharT>::length(key));
}
iterator find(const std::basic_string<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#endif
iterator find_ks(const CharT* key, size_type key_size) {
return m_ht.find(key, key_size);
}
const_iterator find_ks(const CharT* key, size_type key_size) const {
return m_ht.find(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
iterator find(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
iterator find(const CharT* key, std::size_t precalculated_hash) {
return m_ht.find(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.find(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
iterator find(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.find(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
iterator find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.find(key, key_size, precalculated_hash);
}
/**
* @copydoc find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
const_iterator find_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.find(key, key_size, precalculated_hash);
}
#ifdef TSL_AH_HAS_STRING_VIEW
std::pair<iterator, iterator> equal_range(const std::basic_string_view<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string_view<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#else
std::pair<iterator, iterator> equal_range(const CharT* key) {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key));
}
std::pair<const_iterator, const_iterator> equal_range(const CharT* key) const {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key));
}
std::pair<iterator, iterator> equal_range(const std::basic_string<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#endif
std::pair<iterator, iterator> equal_range_ks(const CharT* key, size_type key_size) {
return m_ht.equal_range(key, key_size);
}
std::pair<const_iterator, const_iterator> equal_range_ks(const CharT* key, size_type key_size) const {
return m_ht.equal_range(key, key_size);
}
#ifdef TSL_AH_HAS_STRING_VIEW
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<iterator, iterator> equal_range(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string_view<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
#else
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<iterator, iterator> equal_range(const CharT* key, std::size_t precalculated_hash) {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range(const CharT* key, std::size_t precalculated_hash) const {
return m_ht.equal_range(key, std::char_traits<CharT>::length(key), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<iterator, iterator> equal_range(const std::basic_string<CharT>& key, std::size_t precalculated_hash) {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string<CharT>& key, std::size_t precalculated_hash) const {
return m_ht.equal_range(key.data(), key.size(), precalculated_hash);
}
#endif
/**
* Use the hash value 'precalculated_hash' instead of hashing the key. The hash value should be the same
* as hash_function()(key). Useful to speed-up the lookup to the value if you already have the hash.
*/
std::pair<iterator, iterator> equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) {
return m_ht.equal_range(key, key_size, precalculated_hash);
}
/**
* @copydoc equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash)
*/
std::pair<const_iterator, const_iterator> equal_range_ks(const CharT* key, size_type key_size, std::size_t precalculated_hash) const {
return m_ht.equal_range(key, key_size, precalculated_hash);
}
/*
* Bucket interface
*/
size_type bucket_count() const { return m_ht.bucket_count(); }
size_type max_bucket_count() const { return m_ht.max_bucket_count(); }
/*
* Hash policy
*/
float load_factor() const { return m_ht.load_factor(); }
float max_load_factor() const { return m_ht.max_load_factor(); }
void max_load_factor(float ml) { m_ht.max_load_factor(ml); }
void rehash(size_type count) { m_ht.rehash(count); }
void reserve(size_type count) { m_ht.reserve(count); }
/*
* Observers
*/
hasher hash_function() const { return m_ht.hash_function(); }
key_equal key_eq() const { return m_ht.key_eq(); }
/*
* Other
*/
/**
* Return the `const_iterator it` as an `iterator`.
*/
iterator mutable_iterator(const_iterator it) noexcept { return m_ht.mutable_iterator(it); }
/**
* Serialize the set through the `serializer` parameter.
*
* The `serializer` parameter must be a function object that supports the following calls:
* - `template<typename U> void operator()(const U& value);` where the types `std::uint64_t` and `float` must be supported for U.
* - `void operator()(const CharT* value, std::size_t value_size);`
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, ...) of the types it serializes
* in the hands of the `Serializer` function object if compatibility is required.
*/
template<class Serializer>
void serialize(Serializer& serializer) const {
m_ht.serialize(serializer);
}
/**
* Deserialize a previously serialized set through the `deserializer` parameter.
*
* The `deserializer` parameter must be a function object that supports the following calls:
* - `template<typename U> U operator()();` where the types `std::uint64_t` and `float` must be supported for U.
* - `void operator()(CharT* value_out, std::size_t value_size);`
*
* If the deserialized hash set type is hash compatible with the serialized set, the deserialization process can be
* sped up by setting `hash_compatible` to true. To be hash compatible, the Hash (take care of the 32-bits vs 64 bits),
* KeyEqual, GrowthPolicy, StoreNullTerminator, KeySizeT and IndexSizeT must behave the same than the ones used on the
* serialized set. Otherwise the behaviour is undefined with `hash_compatible` sets to true.
*
* The behaviour is undefined if the type `CharT` of the `array_set` is not the same as the
* type used during serialization.
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, size of int, ...) of the types it
* deserializes in the hands of the `Deserializer` function object if compatibility is required.
*/
template<class Deserializer>
static array_set deserialize(Deserializer& deserializer, bool hash_compatible = false) {
array_set set(0);
set.m_ht.deserialize(deserializer, hash_compatible);
return set;
}
friend bool operator==(const array_set& lhs, const array_set& rhs) {
if(lhs.size() != rhs.size()) {
return false;
}
for(auto it = lhs.cbegin(); it != lhs.cend(); ++it) {
const auto it_element_rhs = rhs.find_ks(it.key(), it.key_size());
if(it_element_rhs == rhs.cend()) {
return false;
}
}
return true;
}
friend bool operator!=(const array_set& lhs, const array_set& rhs) {
return !operator==(lhs, rhs);
}
friend void swap(array_set& lhs, array_set& rhs) {
lhs.swap(rhs);
}
public:
static const size_type MAX_KEY_SIZE = ht::MAX_KEY_SIZE;
private:
ht m_ht;
};
/**
* Same as
* `tsl::array_set<CharT, Hash, KeyEqual, StoreNullTerminator, KeySizeT, IndexSizeT, tsl::ah::prime_growth_policy>`.
*/
template<class CharT,
class Hash = tsl::ah::str_hash<CharT>,
class KeyEqual = tsl::ah::str_equal<CharT>,
bool StoreNullTerminator = true,
class KeySizeT = std::uint16_t,
class IndexSizeT = std::uint32_t>
using array_pg_set = array_set<CharT, Hash, KeyEqual, StoreNullTerminator,
KeySizeT, IndexSizeT, tsl::ah::prime_growth_policy>;
} //end namespace tsl
#endif

2090
ios/include/tsl/htrie_hash.h Normal file

File diff suppressed because it is too large Load Diff

647
ios/include/tsl/htrie_map.h Normal file
View File

@@ -0,0 +1,647 @@
/**
* MIT License
*
* Copyright (c) 2017 Thibaut Goetghebuer-Planchon <tessil@gmx.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_HTRIE_MAP_H
#define TSL_HTRIE_MAP_H
#include <cstddef>
#include <cstring>
#include <initializer_list>
#include <string>
#include <utility>
#include "htrie_hash.h"
namespace tsl {
/**
* Implementation of a hat-trie map.
*
* The value T must be either nothrow move-constructible/assignable, copy-constructible or both.
*
* The size of a key string is limited to std::numeric_limits<KeySizeT>::max() - 1.
* That is 65 535 characters by default, but can be raised with the KeySizeT template parameter.
* See max_key_size() for an easy access to this limit.
*
* Iterators invalidation:
* - clear, operator=: always invalidate the iterators.
* - insert, emplace, operator[]: always invalidate the iterators.
* - erase: always invalidate the iterators.
*/
template<class CharT,
class T,
class Hash = tsl::ah::str_hash<CharT>,
class KeySizeT = std::uint16_t>
class htrie_map {
private:
template<typename U>
using is_iterator = tsl::detail_array_hash::is_iterator<U>;
using ht = tsl::detail_htrie_hash::htrie_hash<CharT, T, Hash, KeySizeT>;
public:
using char_type = typename ht::char_type;
using mapped_type = T;
using key_size_type = typename ht::key_size_type;
using size_type = typename ht::size_type;
using hasher = typename ht::hasher;
using iterator = typename ht::iterator;
using const_iterator = typename ht::const_iterator;
using prefix_iterator = typename ht::prefix_iterator;
using const_prefix_iterator = typename ht::const_prefix_iterator;
public:
explicit htrie_map(const Hash& hash = Hash()): m_ht(hash, ht::HASH_NODE_DEFAULT_MAX_LOAD_FACTOR,
ht::DEFAULT_BURST_THRESHOLD)
{
}
explicit htrie_map(size_type burst_threshold,
const Hash& hash = Hash()): m_ht(hash, ht::HASH_NODE_DEFAULT_MAX_LOAD_FACTOR,
burst_threshold)
{
}
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
htrie_map(InputIt first, InputIt last,
const Hash& hash = Hash()): htrie_map(hash)
{
insert(first, last);
}
#ifdef TSL_HT_HAS_STRING_VIEW
htrie_map(std::initializer_list<std::pair<std::basic_string_view<CharT>, T>> init,
const Hash& hash = Hash()): htrie_map(hash)
{
insert(init);
}
#else
htrie_map(std::initializer_list<std::pair<const CharT*, T>> init,
const Hash& hash = Hash()): htrie_map(hash)
{
insert(init);
}
#endif
#ifdef TSL_HT_HAS_STRING_VIEW
htrie_map& operator=(std::initializer_list<std::pair<std::basic_string_view<CharT>, T>> ilist) {
clear();
insert(ilist);
return *this;
}
#else
htrie_map& operator=(std::initializer_list<std::pair<const CharT*, T>> ilist) {
clear();
insert(ilist);
return *this;
}
#endif
/*
* Iterators
*/
iterator begin() noexcept { return m_ht.begin(); }
const_iterator begin() const noexcept { return m_ht.begin(); }
const_iterator cbegin() const noexcept { return m_ht.cbegin(); }
iterator end() noexcept { return m_ht.end(); }
const_iterator end() const noexcept { return m_ht.end(); }
const_iterator cend() const noexcept { return m_ht.cend(); }
/*
* Capacity
*/
bool empty() const noexcept { return m_ht.empty(); }
size_type size() const noexcept { return m_ht.size(); }
size_type max_size() const noexcept { return m_ht.max_size(); }
size_type max_key_size() const noexcept { return m_ht.max_key_size(); }
/**
* Call shrink_to_fit() on each hash node of the hat-trie to reduce its size.
*/
void shrink_to_fit() { m_ht.shrink_to_fit(); }
/*
* Modifiers
*/
void clear() noexcept { m_ht.clear(); }
std::pair<iterator, bool> insert_ks(const CharT* key, size_type key_size, const T& value) {
return m_ht.insert(key, key_size, value);
}
#ifdef TSL_HT_HAS_STRING_VIEW
std::pair<iterator, bool> insert(const std::basic_string_view<CharT>& key, const T& value) {
return m_ht.insert(key.data(), key.size(), value);
}
#else
std::pair<iterator, bool> insert(const CharT* key, const T& value) {
return m_ht.insert(key, std::strlen(key), value);
}
std::pair<iterator, bool> insert(const std::basic_string<CharT>& key, const T& value) {
return m_ht.insert(key.data(), key.size(), value);
}
#endif
std::pair<iterator, bool> insert_ks(const CharT* key, size_type key_size, T&& value) {
return m_ht.insert(key, key_size, std::move(value));
}
#ifdef TSL_HT_HAS_STRING_VIEW
std::pair<iterator, bool> insert(const std::basic_string_view<CharT>& key, T&& value) {
return m_ht.insert(key.data(), key.size(), std::move(value));
}
#else
std::pair<iterator, bool> insert(const CharT* key, T&& value) {
return m_ht.insert(key, std::strlen(key), std::move(value));
}
std::pair<iterator, bool> insert(const std::basic_string<CharT>& key, T&& value) {
return m_ht.insert(key.data(), key.size(), std::move(value));
}
#endif
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
void insert(InputIt first, InputIt last) {
for(auto it = first; it != last; ++it) {
insert_pair(*it);
}
}
#ifdef TSL_HT_HAS_STRING_VIEW
void insert(std::initializer_list<std::pair<std::basic_string_view<CharT>, T>> ilist) {
insert(ilist.begin(), ilist.end());
}
#else
void insert(std::initializer_list<std::pair<const CharT*, T>> ilist) {
insert(ilist.begin(), ilist.end());
}
#endif
template<class... Args>
std::pair<iterator, bool> emplace_ks(const CharT* key, size_type key_size, Args&&... args) {
return m_ht.insert(key, key_size, std::forward<Args>(args)...);
}
#ifdef TSL_HT_HAS_STRING_VIEW
template<class... Args>
std::pair<iterator, bool> emplace(const std::basic_string_view<CharT>& key, Args&&... args) {
return m_ht.insert(key.data(), key.size(), std::forward<Args>(args)...);
}
#else
template<class... Args>
std::pair<iterator, bool> emplace(const CharT* key, Args&&... args) {
return m_ht.insert(key, std::strlen(key), std::forward<Args>(args)...);
}
template<class... Args>
std::pair<iterator, bool> emplace(const std::basic_string<CharT>& key, Args&&... args) {
return m_ht.insert(key.data(), key.size(), std::forward<Args>(args)...);
}
#endif
iterator erase(const_iterator pos) { return m_ht.erase(pos); }
iterator erase(const_iterator first, const_iterator last) { return m_ht.erase(first, last); }
size_type erase_ks(const CharT* key, size_type key_size) {
return m_ht.erase(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
size_type erase(const std::basic_string_view<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#else
size_type erase(const CharT* key) {
return m_ht.erase(key, std::strlen(key));
}
size_type erase(const std::basic_string<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#endif
/**
* Erase all the elements which have 'prefix' as prefix. Return the number of erase elements.
*/
size_type erase_prefix_ks(const CharT* prefix, size_type prefix_size) {
return m_ht.erase_prefix(prefix, prefix_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
/**
* @copydoc erase_prefix_ks(const CharT* prefix, size_type prefix_size)
*/
size_type erase_prefix(const std::basic_string_view<CharT>& prefix) {
return m_ht.erase_prefix(prefix.data(), prefix.size());
}
#else
/**
* @copydoc erase_prefix_ks(const CharT* prefix, size_type prefix_size)
*/
size_type erase_prefix(const CharT* prefix) {
return m_ht.erase_prefix(prefix, std::strlen(prefix));
}
/**
* @copydoc erase_prefix_ks(const CharT* prefix, size_type prefix_size)
*/
size_type erase_prefix(const std::basic_string<CharT>& prefix) {
return m_ht.erase_prefix(prefix.data(), prefix.size());
}
#endif
void swap(htrie_map& other) { other.m_ht.swap(m_ht); }
/*
* Lookup
*/
T& at_ks(const CharT* key, size_type key_size) { return m_ht.at(key, key_size); }
const T& at_ks(const CharT* key, size_type key_size) const { return m_ht.at(key, key_size); }
#ifdef TSL_HT_HAS_STRING_VIEW
T& at(const std::basic_string_view<CharT>& key) { return m_ht.at(key.data(), key.size()); }
const T& at(const std::basic_string_view<CharT>& key) const { return m_ht.at(key.data(), key.size()); }
#else
T& at(const CharT* key) { return m_ht.at(key, std::strlen(key)); }
const T& at(const CharT* key) const { return m_ht.at(key, std::strlen(key)); }
T& at(const std::basic_string<CharT>& key) { return m_ht.at(key.data(), key.size()); }
const T& at(const std::basic_string<CharT>& key) const { return m_ht.at(key.data(), key.size()); }
#endif
#ifdef TSL_HT_HAS_STRING_VIEW
T& operator[](const std::basic_string_view<CharT>& key) { return m_ht.access_operator(key.data(), key.size()); }
#else
T& operator[](const CharT* key) { return m_ht.access_operator(key, std::strlen(key)); }
T& operator[](const std::basic_string<CharT>& key) { return m_ht.access_operator(key.data(), key.size()); }
#endif
size_type count_ks(const CharT* key, size_type key_size) const { return m_ht.count(key, key_size); }
#ifdef TSL_HT_HAS_STRING_VIEW
size_type count(const std::basic_string_view<CharT>& key) const { return m_ht.count(key.data(), key.size()); }
#else
size_type count(const CharT* key) const { return m_ht.count(key, std::strlen(key)); }
size_type count(const std::basic_string<CharT>& key) const { return m_ht.count(key.data(), key.size()); }
#endif
iterator find_ks(const CharT* key, size_type key_size) {
return m_ht.find(key, key_size);
}
const_iterator find_ks(const CharT* key, size_type key_size) const {
return m_ht.find(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
iterator find(const std::basic_string_view<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string_view<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#else
iterator find(const CharT* key) {
return m_ht.find(key, std::strlen(key));
}
const_iterator find(const CharT* key) const {
return m_ht.find(key, std::strlen(key));
}
iterator find(const std::basic_string<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#endif
std::pair<iterator, iterator> equal_range_ks(const CharT* key, size_type key_size) {
return m_ht.equal_range(key, key_size);
}
std::pair<const_iterator, const_iterator> equal_range_ks(const CharT* key, size_type key_size) const {
return m_ht.equal_range(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
std::pair<iterator, iterator> equal_range(const std::basic_string_view<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string_view<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#else
std::pair<iterator, iterator> equal_range(const CharT* key) {
return m_ht.equal_range(key, std::strlen(key));
}
std::pair<const_iterator, const_iterator> equal_range(const CharT* key) const {
return m_ht.equal_range(key, std::strlen(key));
}
std::pair<iterator, iterator> equal_range(const std::basic_string<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#endif
/**
* Return a range containing all the elements which have 'prefix' as prefix. The range is defined by a pair
* of iterator, the first being the begin iterator and the second being the end iterator.
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range_ks(const CharT* prefix, size_type prefix_size) {
return m_ht.equal_prefix_range(prefix, prefix_size);
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range_ks(const CharT* prefix, size_type prefix_size) const {
return m_ht.equal_prefix_range(prefix, prefix_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range(const std::basic_string_view<CharT>& prefix) {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range(const std::basic_string_view<CharT>& prefix) const {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
#else
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range(const CharT* prefix) {
return m_ht.equal_prefix_range(prefix, std::strlen(prefix));
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range(const CharT* prefix) const {
return m_ht.equal_prefix_range(prefix, std::strlen(prefix));
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range(const std::basic_string<CharT>& prefix) {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range(const std::basic_string<CharT>& prefix) const {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
#endif
/**
* Return the element in the trie which is the longest prefix of `key`. If no
* element in the trie is a prefix of `key`, the end iterator is returned.
*
* Example:
*
* tsl::htrie_map<char, int> map = {{"/foo", 1}, {"/foo/bar", 1}};
*
* map.longest_prefix("/foo"); // returns {"/foo", 1}
* map.longest_prefix("/foo/baz"); // returns {"/foo", 1}
* map.longest_prefix("/foo/bar/baz"); // returns {"/foo/bar", 1}
* map.longest_prefix("/foo/bar/"); // returns {"/foo/bar", 1}
* map.longest_prefix("/bar"); // returns end()
* map.longest_prefix(""); // returns end()
*/
iterator longest_prefix_ks(const CharT* key, size_type key_size) {
return m_ht.longest_prefix(key, key_size);
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix_ks(const CharT* key, size_type key_size) const {
return m_ht.longest_prefix(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
iterator longest_prefix(const std::basic_string_view<CharT>& key) {
return m_ht.longest_prefix(key.data(), key.size());
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix(const std::basic_string_view<CharT>& key) const {
return m_ht.longest_prefix(key.data(), key.size());
}
#else
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
iterator longest_prefix(const CharT* key) {
return m_ht.longest_prefix(key, std::strlen(key));
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix(const CharT* key) const {
return m_ht.longest_prefix(key, std::strlen(key));
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
iterator longest_prefix(const std::basic_string<CharT>& key) {
return m_ht.longest_prefix(key.data(), key.size());
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix(const std::basic_string<CharT>& key) const {
return m_ht.longest_prefix(key.data(), key.size());
}
#endif
/*
* Hash policy
*/
float max_load_factor() const { return m_ht.max_load_factor(); }
void max_load_factor(float ml) { m_ht.max_load_factor(ml); }
/*
* Burst policy
*/
size_type burst_threshold() const { return m_ht.burst_threshold(); }
void burst_threshold(size_type threshold) { m_ht.burst_threshold(threshold); }
/*
* Observers
*/
hasher hash_function() const { return m_ht.hash_function(); }
/*
* Other
*/
/**
* Serialize the map through the `serializer` parameter.
*
* The `serializer` parameter must be a function object that supports the following calls:
* - `void operator()(const U& value);` where the types `std::uint64_t`, `float` and `T` must be supported for U.
* - `void operator()(const CharT* value, std::size_t value_size);`
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, ...) of the types it serializes
* in the hands of the `Serializer` function object if compatibility is required.
*/
template<class Serializer>
void serialize(Serializer& serializer) const {
m_ht.serialize(serializer);
}
/**
* Deserialize a previously serialized map through the `deserializer` parameter.
*
* The `deserializer` parameter must be a function object that supports the following calls:
* - `template<typename U> U operator()();` where the types `std::uint64_t`, `float` and `T` must be supported for U.
* - `void operator()(CharT* value_out, std::size_t value_size);`
*
* If the deserialized hash map part of the hat-trie is hash compatible with the serialized map, the deserialization process
* can be sped up by setting `hash_compatible` to true. To be hash compatible, the Hash (take care of the 32-bits vs 64 bits),
* and KeySizeT must behave the same than the ones used in the serialized map. Otherwise the behaviour is undefined
* with `hash_compatible` sets to true.
*
* The behaviour is undefined if the type `CharT` and `T` of the `htrie_map` are not the same as the
* types used during serialization.
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, size of int, ...) of the types it
* deserializes in the hands of the `Deserializer` function object if compatibility is required.
*/
template<class Deserializer>
static htrie_map deserialize(Deserializer& deserializer, bool hash_compatible = false) {
htrie_map map;
map.m_ht.deserialize(deserializer, hash_compatible);
return map;
}
friend bool operator==(const htrie_map& lhs, const htrie_map& rhs) {
if(lhs.size() != rhs.size()) {
return false;
}
std::string key_buffer;
for(auto it = lhs.cbegin(); it != lhs.cend(); ++it) {
it.key(key_buffer);
const auto it_element_rhs = rhs.find(key_buffer);
if(it_element_rhs == rhs.cend() || it.value() != it_element_rhs.value()) {
return false;
}
}
return true;
}
friend bool operator!=(const htrie_map& lhs, const htrie_map& rhs) {
return !operator==(lhs, rhs);
}
friend void swap(htrie_map& lhs, htrie_map& rhs) {
lhs.swap(rhs);
}
private:
template<class U, class V>
void insert_pair(const std::pair<U, V>& value) {
insert(value.first, value.second);
}
template<class U, class V>
void insert_pair(std::pair<U, V>&& value) {
insert(value.first, std::move(value.second));
}
private:
ht m_ht;
};
} // end namespace tsl
#endif

586
ios/include/tsl/htrie_set.h Normal file
View File

@@ -0,0 +1,586 @@
/**
* MIT License
*
* Copyright (c) 2017 Thibaut Goetghebuer-Planchon <tessil@gmx.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_HTRIE_SET_H
#define TSL_HTRIE_SET_H
#include <cstddef>
#include <cstring>
#include <initializer_list>
#include <string>
#include <utility>
#include "htrie_hash.h"
namespace tsl {
/**
* Implementation of a hat-trie set.
*
* The size of a key string is limited to std::numeric_limits<KeySizeT>::max() - 1.
* That is 65 535 characters by default, but can be raised with the KeySizeT template parameter.
* See max_key_size() for an easy access to this limit.
*
* Iterators invalidation:
* - clear, operator=: always invalidate the iterators.
* - insert: always invalidate the iterators.
* - erase: always invalidate the iterators.
*/
template<class CharT,
class Hash = tsl::ah::str_hash<CharT>,
class KeySizeT = std::uint16_t>
class htrie_set {
private:
template<typename U>
using is_iterator = tsl::detail_array_hash::is_iterator<U>;
using ht = tsl::detail_htrie_hash::htrie_hash<CharT, void, Hash, KeySizeT>;
public:
using char_type = typename ht::char_type;
using key_size_type = typename ht::key_size_type;
using size_type = typename ht::size_type;
using hasher = typename ht::hasher;
using iterator = typename ht::iterator;
using const_iterator = typename ht::const_iterator;
using prefix_iterator = typename ht::prefix_iterator;
using const_prefix_iterator = typename ht::const_prefix_iterator;
public:
explicit htrie_set(const Hash& hash = Hash()): m_ht(hash, ht::HASH_NODE_DEFAULT_MAX_LOAD_FACTOR,
ht::DEFAULT_BURST_THRESHOLD)
{
}
explicit htrie_set(size_type burst_threshold,
const Hash& hash = Hash()): m_ht(hash, ht::HASH_NODE_DEFAULT_MAX_LOAD_FACTOR,
burst_threshold)
{
}
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
htrie_set(InputIt first, InputIt last,
const Hash& hash = Hash()): htrie_set(hash)
{
insert(first, last);
}
#ifdef TSL_HT_HAS_STRING_VIEW
htrie_set(std::initializer_list<std::basic_string_view<CharT>> init,
const Hash& hash = Hash()): htrie_set(hash)
{
insert(init);
}
#else
htrie_set(std::initializer_list<const CharT*> init,
const Hash& hash = Hash()): htrie_set(hash)
{
insert(init);
}
#endif
#ifdef TSL_HT_HAS_STRING_VIEW
htrie_set& operator=(std::initializer_list<std::basic_string_view<CharT>> ilist) {
clear();
insert(ilist);
return *this;
}
#else
htrie_set& operator=(std::initializer_list<const CharT*> ilist) {
clear();
insert(ilist);
return *this;
}
#endif
/*
* Iterators
*/
iterator begin() noexcept { return m_ht.begin(); }
const_iterator begin() const noexcept { return m_ht.begin(); }
const_iterator cbegin() const noexcept { return m_ht.cbegin(); }
iterator end() noexcept { return m_ht.end(); }
const_iterator end() const noexcept { return m_ht.end(); }
const_iterator cend() const noexcept { return m_ht.cend(); }
/*
* Capacity
*/
bool empty() const noexcept { return m_ht.empty(); }
size_type size() const noexcept { return m_ht.size(); }
size_type max_size() const noexcept { return m_ht.max_size(); }
size_type max_key_size() const noexcept { return m_ht.max_key_size(); }
/**
* Call shrink_to_fit() on each hash node of the hat-trie to reduce its size.
*/
void shrink_to_fit() { m_ht.shrink_to_fit(); }
/*
* Modifiers
*/
void clear() noexcept { m_ht.clear(); }
std::pair<iterator, bool> insert_ks(const CharT* key, size_type key_size) {
return m_ht.insert(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
std::pair<iterator, bool> insert(const std::basic_string_view<CharT>& key) {
return m_ht.insert(key.data(), key.size());
}
#else
std::pair<iterator, bool> insert(const CharT* key) {
return m_ht.insert(key, std::strlen(key));
}
std::pair<iterator, bool> insert(const std::basic_string<CharT>& key) {
return m_ht.insert(key.data(), key.size());
}
#endif
template<class InputIt, typename std::enable_if<is_iterator<InputIt>::value>::type* = nullptr>
void insert(InputIt first, InputIt last) {
for(auto it = first; it != last; ++it) {
insert(*it);
}
}
#ifdef TSL_HT_HAS_STRING_VIEW
void insert(std::initializer_list<std::basic_string_view<CharT>> ilist) {
insert(ilist.begin(), ilist.end());
}
#else
void insert(std::initializer_list<const CharT*> ilist) {
insert(ilist.begin(), ilist.end());
}
#endif
std::pair<iterator, bool> emplace_ks(const CharT* key, size_type key_size) {
return m_ht.insert(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
std::pair<iterator, bool> emplace(const std::basic_string_view<CharT>& key) {
return m_ht.insert(key.data(), key.size());
}
#else
std::pair<iterator, bool> emplace(const CharT* key) {
return m_ht.insert(key, std::strlen(key));
}
std::pair<iterator, bool> emplace(const std::basic_string<CharT>& key) {
return m_ht.insert(key.data(), key.size());
}
#endif
iterator erase(const_iterator pos) { return m_ht.erase(pos); }
iterator erase(const_iterator first, const_iterator last) { return m_ht.erase(first, last); }
size_type erase_ks(const CharT* key, size_type key_size) {
return m_ht.erase(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
size_type erase(const std::basic_string_view<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#else
size_type erase(const CharT* key) {
return m_ht.erase(key, std::strlen(key));
}
size_type erase(const std::basic_string<CharT>& key) {
return m_ht.erase(key.data(), key.size());
}
#endif
/**
* Erase all the elements which have 'prefix' as prefix. Return the number of erase elements.
*/
size_type erase_prefix_ks(const CharT* prefix, size_type prefix_size) {
return m_ht.erase_prefix(prefix, prefix_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
/**
* @copydoc erase_prefix_ks(const CharT* prefix, size_type prefix_size)
*/
size_type erase_prefix(const std::basic_string_view<CharT>& prefix) {
return m_ht.erase_prefix(prefix.data(), prefix.size());
}
#else
/**
* @copydoc erase_prefix_ks(const CharT* prefix, size_type prefix_size)
*/
size_type erase_prefix(const CharT* prefix) {
return m_ht.erase_prefix(prefix, std::strlen(prefix));
}
/**
* @copydoc erase_prefix_ks(const CharT* prefix, size_type prefix_size)
*/
size_type erase_prefix(const std::basic_string<CharT>& prefix) {
return m_ht.erase_prefix(prefix.data(), prefix.size());
}
#endif
void swap(htrie_set& other) { other.m_ht.swap(m_ht); }
/*
* Lookup
*/
size_type count_ks(const CharT* key, size_type key_size) const { return m_ht.count(key, key_size); }
#ifdef TSL_HT_HAS_STRING_VIEW
size_type count(const std::basic_string_view<CharT>& key) const { return m_ht.count(key.data(), key.size()); }
#else
size_type count(const CharT* key) const { return m_ht.count(key, std::strlen(key)); }
size_type count(const std::basic_string<CharT>& key) const { return m_ht.count(key.data(), key.size()); }
#endif
iterator find_ks(const CharT* key, size_type key_size) {
return m_ht.find(key, key_size);
}
const_iterator find_ks(const CharT* key, size_type key_size) const {
return m_ht.find(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
iterator find(const std::basic_string_view<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string_view<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#else
iterator find(const CharT* key) {
return m_ht.find(key, std::strlen(key));
}
const_iterator find(const CharT* key) const {
return m_ht.find(key, std::strlen(key));
}
iterator find(const std::basic_string<CharT>& key) {
return m_ht.find(key.data(), key.size());
}
const_iterator find(const std::basic_string<CharT>& key) const {
return m_ht.find(key.data(), key.size());
}
#endif
std::pair<iterator, iterator> equal_range_ks(const CharT* key, size_type key_size) {
return m_ht.equal_range(key, key_size);
}
std::pair<const_iterator, const_iterator> equal_range_ks(const CharT* key, size_type key_size) const {
return m_ht.equal_range(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
std::pair<iterator, iterator> equal_range(const std::basic_string_view<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string_view<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#else
std::pair<iterator, iterator> equal_range(const CharT* key) {
return m_ht.equal_range(key, std::strlen(key));
}
std::pair<const_iterator, const_iterator> equal_range(const CharT* key) const {
return m_ht.equal_range(key, std::strlen(key));
}
std::pair<iterator, iterator> equal_range(const std::basic_string<CharT>& key) {
return m_ht.equal_range(key.data(), key.size());
}
std::pair<const_iterator, const_iterator> equal_range(const std::basic_string<CharT>& key) const {
return m_ht.equal_range(key.data(), key.size());
}
#endif
/**
* Return a range containing all the elements which have 'prefix' as prefix. The range is defined by a pair
* of iterator, the first being the begin iterator and the second being the end iterator.
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range_ks(const CharT* prefix, size_type prefix_size) {
return m_ht.equal_prefix_range(prefix, prefix_size);
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range_ks(const CharT* prefix, size_type prefix_size) const {
return m_ht.equal_prefix_range(prefix, prefix_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range(const std::basic_string_view<CharT>& prefix) {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range(const std::basic_string_view<CharT>& prefix) const {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
#else
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range(const CharT* prefix) {
return m_ht.equal_prefix_range(prefix, std::strlen(prefix));
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range(const CharT* prefix) const {
return m_ht.equal_prefix_range(prefix, std::strlen(prefix));
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<prefix_iterator, prefix_iterator> equal_prefix_range(const std::basic_string<CharT>& prefix) {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
/**
* @copydoc equal_prefix_range_ks(const CharT* prefix, size_type prefix_size)
*/
std::pair<const_prefix_iterator, const_prefix_iterator> equal_prefix_range(const std::basic_string<CharT>& prefix) const {
return m_ht.equal_prefix_range(prefix.data(), prefix.size());
}
#endif
/**
* Return the element in the trie which is the longest prefix of `key`. If no
* element in the trie is a prefix of `key`, the end iterator is returned.
*
* Example:
*
* tsl::htrie_set<char> set = {"/foo", "/foo/bar"};
*
* set.longest_prefix("/foo"); // returns "/foo"
* set.longest_prefix("/foo/baz"); // returns "/foo"
* set.longest_prefix("/foo/bar/baz"); // returns "/foo/bar"
* set.longest_prefix("/foo/bar/"); // returns "/foo/bar"
* set.longest_prefix("/bar"); // returns end()
* set.longest_prefix(""); // returns end()
*/
iterator longest_prefix_ks(const CharT* key, size_type key_size) {
return m_ht.longest_prefix(key, key_size);
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix_ks(const CharT* key, size_type key_size) const {
return m_ht.longest_prefix(key, key_size);
}
#ifdef TSL_HT_HAS_STRING_VIEW
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
iterator longest_prefix(const std::basic_string_view<CharT>& key) {
return m_ht.longest_prefix(key.data(), key.size());
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix(const std::basic_string_view<CharT>& key) const {
return m_ht.longest_prefix(key.data(), key.size());
}
#else
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
iterator longest_prefix(const CharT* key) {
return m_ht.longest_prefix(key, std::strlen(key));
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix(const CharT* key) const {
return m_ht.longest_prefix(key, std::strlen(key));
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
iterator longest_prefix(const std::basic_string<CharT>& key) {
return m_ht.longest_prefix(key.data(), key.size());
}
/**
* @copydoc longest_prefix_ks(const CharT* key, size_type key_size)
*/
const_iterator longest_prefix(const std::basic_string<CharT>& key) const {
return m_ht.longest_prefix(key.data(), key.size());
}
#endif
/*
* Hash policy
*/
float max_load_factor() const { return m_ht.max_load_factor(); }
void max_load_factor(float ml) { m_ht.max_load_factor(ml); }
/*
* Burst policy
*/
size_type burst_threshold() const { return m_ht.burst_threshold(); }
void burst_threshold(size_type threshold) { m_ht.burst_threshold(threshold); }
/*
* Observers
*/
hasher hash_function() const { return m_ht.hash_function(); }
/*
* Other
*/
/**
* Serialize the set through the `serializer` parameter.
*
* The `serializer` parameter must be a function object that supports the following calls:
* - `void operator()(const U& value);` where the types `std::uint64_t` and `float` must be supported for U.
* - `void operator()(const CharT* value, std::size_t value_size);`
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, ...) of the types it serializes
* in the hands of the `Serializer` function object if compatibility is required.
*/
template<class Serializer>
void serialize(Serializer& serializer) const {
m_ht.serialize(serializer);
}
/**
* Deserialize a previously serialized set through the `deserializer` parameter.
*
* The `deserializer` parameter must be a function object that supports the following calls:
* - `template<typename U> U operator()();` where the types `std::uint64_t` and `float` must be supported for U.
* - `void operator()(CharT* value_out, std::size_t value_size);`
*
* If the deserialized hash set part of the hat-trie is hash compatible with the serialized set, the deserialization process
* can be sped up by setting `hash_compatible` to true. To be hash compatible, the Hash (take care of the 32-bits vs 64 bits),
* and KeySizeT must behave the same than the ones used in the serialized set. Otherwise the behaviour is undefined
* with `hash_compatible` sets to true.
*
* The behaviour is undefined if the type `CharT` of the `htrie_set` is not the same as the
* type used during serialization.
*
* The implementation leaves binary compatibility (endianness, IEEE 754 for floats, size of int, ...) of the types it
* deserializes in the hands of the `Deserializer` function object if compatibility is required.
*/
template<class Deserializer>
static htrie_set deserialize(Deserializer& deserializer, bool hash_compatible = false) {
htrie_set set;
set.m_ht.deserialize(deserializer, hash_compatible);
return set;
}
friend bool operator==(const htrie_set& lhs, const htrie_set& rhs) {
if(lhs.size() != rhs.size()) {
return false;
}
std::string key_buffer;
for(auto it = lhs.cbegin(); it != lhs.cend(); ++it) {
it.key(key_buffer);
const auto it_element_rhs = rhs.find(key_buffer);
if(it_element_rhs == rhs.cend()) {
return false;
}
}
return true;
}
friend bool operator!=(const htrie_set& lhs, const htrie_set& rhs) {
return !operator==(lhs, rhs);
}
friend void swap(htrie_set& lhs, htrie_set& rhs) {
lhs.swap(rhs);
}
private:
ht m_ht;
};
} // end namespace tsl
#endif

View File

@@ -1,290 +1,290 @@
/**
* MIT License
*
* Copyright (c) 2017 Tessil
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_ROBIN_GROWTH_POLICY_H
#define TSL_ROBIN_GROWTH_POLICY_H
#include <algorithm>
#include <array>
#include <climits>
#include <cmath>
#include <cstddef>
#include <iterator>
#include <limits>
#include <ratio>
#include <stdexcept>
#ifdef __EXCEPTIONS
# define THROW(_e, _m) throw _e(_m)
#else
# include <stdio.h>
# ifndef NDEBUG
# define THROW(_e, _m) do { fprintf(stderr, _m); std::terminate(); } while(0)
# else
# define THROW(_e, _m) std::terminate()
# endif
#endif
#ifndef __has_builtin
#define __has_builtin(x) 0
#endif
#if __has_builtin(__builtin_expect)
# define TSL_LIKELY( exp ) (__builtin_expect( !!(exp), true ))
#else
# define TSL_LIKELY( exp ) (exp)
#endif
namespace tsl {
namespace rh {
/**
* Grow the hash table by a factor of GrowthFactor keeping the bucket count to a power of two. It allows
* the table to use a mask operation instead of a modulo operation to map a hash to a bucket.
*
* GrowthFactor must be a power of two >= 2.
*/
template<std::size_t GrowthFactor>
class power_of_two_growth_policy {
public:
/**
* Called on the hash table creation and on rehash. The number of buckets for the table is passed in parameter.
* This number is a minimum, the policy may update this value with a higher value if needed (but not lower).
*/
power_of_two_growth_policy(std::size_t& min_bucket_count_in_out) {
if(min_bucket_count_in_out > max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
static_assert(MIN_BUCKETS_SIZE > 0, "MIN_BUCKETS_SIZE must be > 0.");
const std::size_t min_bucket_count = MIN_BUCKETS_SIZE;
min_bucket_count_in_out = std::max(min_bucket_count, min_bucket_count_in_out);
min_bucket_count_in_out = round_up_to_power_of_two(min_bucket_count_in_out);
m_mask = min_bucket_count_in_out - 1;
}
/**
* Return the bucket [0, bucket_count()) to which the hash belongs.
*/
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return hash & m_mask;
}
/**
* Return the bucket count to use when the bucket array grows on rehash.
*/
std::size_t next_bucket_count() const {
if((m_mask + 1) > max_bucket_count() / GrowthFactor) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
return (m_mask + 1) * GrowthFactor;
}
/**
* Return the maximum number of buckets supported by the policy.
*/
std::size_t max_bucket_count() const {
// Largest power of two.
return (std::numeric_limits<std::size_t>::max() / 2) + 1;
}
private:
static std::size_t round_up_to_power_of_two(std::size_t value) {
if(is_power_of_two(value)) {
return value;
}
if(value == 0) {
return 1;
}
--value;
for(std::size_t i = 1; i < sizeof(std::size_t) * CHAR_BIT; i *= 2) {
value |= value >> i;
}
return value + 1;
}
static constexpr bool is_power_of_two(std::size_t value) {
return value != 0 && (value & (value - 1)) == 0;
}
protected:
static const std::size_t MIN_BUCKETS_SIZE = 2;
static_assert(is_power_of_two(GrowthFactor) && GrowthFactor >= 2, "GrowthFactor must be a power of two >= 2.");
std::size_t m_mask;
};
/**
* Grow the hash table by GrowthFactor::num / GrowthFactor::den and use a modulo to map a hash
* to a bucket. Slower but it can be usefull if you want a slower growth.
*/
template<class GrowthFactor = std::ratio<3, 2>>
class mod_growth_policy {
public:
mod_growth_policy(std::size_t& min_bucket_count_in_out) {
if(min_bucket_count_in_out > max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
static_assert(MIN_BUCKETS_SIZE > 0, "MIN_BUCKETS_SIZE must be > 0.");
const std::size_t min_bucket_count = MIN_BUCKETS_SIZE;
min_bucket_count_in_out = std::max(min_bucket_count, min_bucket_count_in_out);
m_bucket_count = min_bucket_count_in_out;
}
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return hash % m_bucket_count;
}
std::size_t next_bucket_count() const {
if(m_bucket_count == max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
const double next_bucket_count = std::ceil(double(m_bucket_count) * REHASH_SIZE_MULTIPLICATION_FACTOR);
if(!std::isnormal(next_bucket_count)) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
if(next_bucket_count > double(max_bucket_count())) {
return max_bucket_count();
}
else {
return std::size_t(next_bucket_count);
}
}
std::size_t max_bucket_count() const {
return MAX_BUCKET_COUNT;
}
private:
static const std::size_t MIN_BUCKETS_SIZE = 2;
static constexpr double REHASH_SIZE_MULTIPLICATION_FACTOR = 1.0 * GrowthFactor::num / GrowthFactor::den;
static const std::size_t MAX_BUCKET_COUNT =
std::size_t(double(
std::numeric_limits<std::size_t>::max() / REHASH_SIZE_MULTIPLICATION_FACTOR
));
static_assert(REHASH_SIZE_MULTIPLICATION_FACTOR >= 1.1, "Growth factor should be >= 1.1.");
std::size_t m_bucket_count;
};
namespace detail {
static constexpr const std::array<std::size_t, 39> PRIMES = {{
5ul, 17ul, 29ul, 37ul, 53ul, 67ul, 79ul, 97ul, 131ul, 193ul, 257ul, 389ul, 521ul, 769ul, 1031ul, 1543ul, 2053ul,
3079ul, 6151ul, 12289ul, 24593ul, 49157ul, 98317ul, 196613ul, 393241ul, 786433ul, 1572869ul, 3145739ul,
6291469ul, 12582917ul, 25165843ul, 50331653ul, 100663319ul, 201326611ul, 402653189ul, 805306457ul,
1610612741ul, 3221225473ul, 4294967291ul
}};
template<unsigned int IPrime>
static constexpr std::size_t mod(std::size_t hash) { return hash % PRIMES[IPrime]; }
// MOD_PRIME[iprime](hash) returns hash % PRIMES[iprime]. This table allows for faster modulo as the
// compiler can optimize the modulo code better with a constant known at the compilation.
static constexpr const std::array<std::size_t(*)(std::size_t), 39> MOD_PRIME = {{
&mod<0>, &mod<1>, &mod<2>, &mod<3>, &mod<4>, &mod<5>, &mod<6>, &mod<7>, &mod<8>, &mod<9>, &mod<10>,
&mod<11>, &mod<12>, &mod<13>, &mod<14>, &mod<15>, &mod<16>, &mod<17>, &mod<18>, &mod<19>, &mod<20>,
&mod<21>, &mod<22>, &mod<23>, &mod<24>, &mod<25>, &mod<26>, &mod<27>, &mod<28>, &mod<29>, &mod<30>,
&mod<31>, &mod<32>, &mod<33>, &mod<34>, &mod<35>, &mod<36>, &mod<37> , &mod<38>
}};
}
/**
* Grow the hash table by using prime numbers as bucket count. Slower than tsl::rh::power_of_two_growth_policy in
* general but will probably distribute the values around better in the buckets with a poor hash function.
*
* To allow the compiler to optimize the modulo operation, a lookup table is used with constant primes numbers.
*
* With a switch the code would look like:
* \code
* switch(iprime) { // iprime is the current prime of the hash table
* case 0: hash % 5ul;
* break;
* case 1: hash % 17ul;
* break;
* case 2: hash % 29ul;
* break;
* ...
* }
* \endcode
*
* Due to the constant variable in the modulo the compiler is able to optimize the operation
* by a series of multiplications, substractions and shifts.
*
* The 'hash % 5' could become something like 'hash - (hash * 0xCCCCCCCD) >> 34) * 5' in a 64 bits environement.
*/
class prime_growth_policy {
public:
prime_growth_policy(std::size_t& min_bucket_count_in_out) {
auto it_prime = std::lower_bound(detail::PRIMES.begin(),
detail::PRIMES.end(), min_bucket_count_in_out);
if(it_prime == detail::PRIMES.end()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
m_iprime = static_cast<unsigned int>(std::distance(detail::PRIMES.begin(), it_prime));
min_bucket_count_in_out = *it_prime;
}
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return detail::MOD_PRIME[m_iprime](hash);
}
std::size_t next_bucket_count() const {
if(m_iprime + 1 >= detail::PRIMES.size()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
return detail::PRIMES[m_iprime + 1];
}
std::size_t max_bucket_count() const {
return detail::PRIMES.back();
}
private:
unsigned int m_iprime;
static_assert(std::numeric_limits<decltype(m_iprime)>::max() >= detail::PRIMES.size(),
"The type of m_iprime is not big enough.");
};
}
}
#endif
/**
* MIT License
*
* Copyright (c) 2017 Tessil
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef TSL_ROBIN_GROWTH_POLICY_H
#define TSL_ROBIN_GROWTH_POLICY_H
#include <algorithm>
#include <array>
#include <climits>
#include <cmath>
#include <cstddef>
#include <iterator>
#include <limits>
#include <ratio>
#include <stdexcept>
#ifdef __EXCEPTIONS
# define THROW(_e, _m) throw _e(_m)
#else
# include <stdio.h>
# ifndef NDEBUG
# define THROW(_e, _m) do { fprintf(stderr, _m); std::terminate(); } while(0)
# else
# define THROW(_e, _m) std::terminate()
# endif
#endif
#ifndef __has_builtin
#define __has_builtin(x) 0
#endif
#if __has_builtin(__builtin_expect)
# define TSL_LIKELY( exp ) (__builtin_expect( !!(exp), true ))
#else
# define TSL_LIKELY( exp ) (exp)
#endif
namespace tsl {
namespace rh {
/**
* Grow the hash table by a factor of GrowthFactor keeping the bucket count to a power of two. It allows
* the table to use a mask operation instead of a modulo operation to map a hash to a bucket.
*
* GrowthFactor must be a power of two >= 2.
*/
template<std::size_t GrowthFactor>
class power_of_two_growth_policy {
public:
/**
* Called on the hash table creation and on rehash. The number of buckets for the table is passed in parameter.
* This number is a minimum, the policy may update this value with a higher value if needed (but not lower).
*/
power_of_two_growth_policy(std::size_t& min_bucket_count_in_out) {
if(min_bucket_count_in_out > max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
static_assert(MIN_BUCKETS_SIZE > 0, "MIN_BUCKETS_SIZE must be > 0.");
const std::size_t min_bucket_count = MIN_BUCKETS_SIZE;
min_bucket_count_in_out = std::max(min_bucket_count, min_bucket_count_in_out);
min_bucket_count_in_out = round_up_to_power_of_two(min_bucket_count_in_out);
m_mask = min_bucket_count_in_out - 1;
}
/**
* Return the bucket [0, bucket_count()) to which the hash belongs.
*/
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return hash & m_mask;
}
/**
* Return the bucket count to use when the bucket array grows on rehash.
*/
std::size_t next_bucket_count() const {
if((m_mask + 1) > max_bucket_count() / GrowthFactor) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
return (m_mask + 1) * GrowthFactor;
}
/**
* Return the maximum number of buckets supported by the policy.
*/
std::size_t max_bucket_count() const {
// Largest power of two.
return (std::numeric_limits<std::size_t>::max() / 2) + 1;
}
private:
static std::size_t round_up_to_power_of_two(std::size_t value) {
if(is_power_of_two(value)) {
return value;
}
if(value == 0) {
return 1;
}
--value;
for(std::size_t i = 1; i < sizeof(std::size_t) * CHAR_BIT; i *= 2) {
value |= value >> i;
}
return value + 1;
}
static constexpr bool is_power_of_two(std::size_t value) {
return value != 0 && (value & (value - 1)) == 0;
}
protected:
static const std::size_t MIN_BUCKETS_SIZE = 2;
static_assert(is_power_of_two(GrowthFactor) && GrowthFactor >= 2, "GrowthFactor must be a power of two >= 2.");
std::size_t m_mask;
};
/**
* Grow the hash table by GrowthFactor::num / GrowthFactor::den and use a modulo to map a hash
* to a bucket. Slower but it can be usefull if you want a slower growth.
*/
template<class GrowthFactor = std::ratio<3, 2>>
class mod_growth_policy {
public:
mod_growth_policy(std::size_t& min_bucket_count_in_out) {
if(min_bucket_count_in_out > max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
static_assert(MIN_BUCKETS_SIZE > 0, "MIN_BUCKETS_SIZE must be > 0.");
const std::size_t min_bucket_count = MIN_BUCKETS_SIZE;
min_bucket_count_in_out = std::max(min_bucket_count, min_bucket_count_in_out);
m_bucket_count = min_bucket_count_in_out;
}
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return hash % m_bucket_count;
}
std::size_t next_bucket_count() const {
if(m_bucket_count == max_bucket_count()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
const double next_bucket_count = std::ceil(double(m_bucket_count) * REHASH_SIZE_MULTIPLICATION_FACTOR);
if(!std::isnormal(next_bucket_count)) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
if(next_bucket_count > double(max_bucket_count())) {
return max_bucket_count();
}
else {
return std::size_t(next_bucket_count);
}
}
std::size_t max_bucket_count() const {
return MAX_BUCKET_COUNT;
}
private:
static const std::size_t MIN_BUCKETS_SIZE = 2;
static constexpr double REHASH_SIZE_MULTIPLICATION_FACTOR = 1.0 * GrowthFactor::num / GrowthFactor::den;
static const std::size_t MAX_BUCKET_COUNT =
std::size_t(double(
std::numeric_limits<std::size_t>::max() / REHASH_SIZE_MULTIPLICATION_FACTOR
));
static_assert(REHASH_SIZE_MULTIPLICATION_FACTOR >= 1.1, "Growth factor should be >= 1.1.");
std::size_t m_bucket_count;
};
namespace detail {
static constexpr const std::array<std::size_t, 39> PRIMES = {{
5ul, 17ul, 29ul, 37ul, 53ul, 67ul, 79ul, 97ul, 131ul, 193ul, 257ul, 389ul, 521ul, 769ul, 1031ul, 1543ul, 2053ul,
3079ul, 6151ul, 12289ul, 24593ul, 49157ul, 98317ul, 196613ul, 393241ul, 786433ul, 1572869ul, 3145739ul,
6291469ul, 12582917ul, 25165843ul, 50331653ul, 100663319ul, 201326611ul, 402653189ul, 805306457ul,
1610612741ul, 3221225473ul, 4294967291ul
}};
template<unsigned int IPrime>
static constexpr std::size_t mod(std::size_t hash) { return hash % PRIMES[IPrime]; }
// MOD_PRIME[iprime](hash) returns hash % PRIMES[iprime]. This table allows for faster modulo as the
// compiler can optimize the modulo code better with a constant known at the compilation.
static constexpr const std::array<std::size_t(*)(std::size_t), 39> MOD_PRIME = {{
&mod<0>, &mod<1>, &mod<2>, &mod<3>, &mod<4>, &mod<5>, &mod<6>, &mod<7>, &mod<8>, &mod<9>, &mod<10>,
&mod<11>, &mod<12>, &mod<13>, &mod<14>, &mod<15>, &mod<16>, &mod<17>, &mod<18>, &mod<19>, &mod<20>,
&mod<21>, &mod<22>, &mod<23>, &mod<24>, &mod<25>, &mod<26>, &mod<27>, &mod<28>, &mod<29>, &mod<30>,
&mod<31>, &mod<32>, &mod<33>, &mod<34>, &mod<35>, &mod<36>, &mod<37> , &mod<38>
}};
}
/**
* Grow the hash table by using prime numbers as bucket count. Slower than tsl::rh::power_of_two_growth_policy in
* general but will probably distribute the values around better in the buckets with a poor hash function.
*
* To allow the compiler to optimize the modulo operation, a lookup table is used with constant primes numbers.
*
* With a switch the code would look like:
* \code
* switch(iprime) { // iprime is the current prime of the hash table
* case 0: hash % 5ul;
* break;
* case 1: hash % 17ul;
* break;
* case 2: hash % 29ul;
* break;
* ...
* }
* \endcode
*
* Due to the constant variable in the modulo the compiler is able to optimize the operation
* by a series of multiplications, substractions and shifts.
*
* The 'hash % 5' could become something like 'hash - (hash * 0xCCCCCCCD) >> 34) * 5' in a 64 bits environement.
*/
class prime_growth_policy {
public:
prime_growth_policy(std::size_t& min_bucket_count_in_out) {
auto it_prime = std::lower_bound(detail::PRIMES.begin(),
detail::PRIMES.end(), min_bucket_count_in_out);
if(it_prime == detail::PRIMES.end()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
m_iprime = static_cast<unsigned int>(std::distance(detail::PRIMES.begin(), it_prime));
min_bucket_count_in_out = *it_prime;
}
std::size_t bucket_for_hash(std::size_t hash) const noexcept {
return detail::MOD_PRIME[m_iprime](hash);
}
std::size_t next_bucket_count() const {
if(m_iprime + 1 >= detail::PRIMES.size()) {
THROW(std::length_error, "The hash table exceeds its maxmimum size.");
}
return detail::PRIMES[m_iprime + 1];
}
std::size_t max_bucket_count() const {
return detail::PRIMES.back();
}
private:
unsigned int m_iprime;
static_assert(std::numeric_limits<decltype(m_iprime)>::max() >= detail::PRIMES.size(),
"The type of m_iprime is not big enough.");
};
}
}
#endif

File diff suppressed because it is too large Load Diff

File diff suppressed because it is too large Load Diff

File diff suppressed because it is too large Load Diff