439 lines
11 KiB
C++
439 lines
11 KiB
C++
//
|
|
// Copyright Aliaksei Levin (levlam@telegram.org), Arseny Smirnov (arseny30@gmail.com) 2014-2022
|
|
//
|
|
// Distributed under the Boost Software License, Version 1.0. (See accompanying
|
|
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
|
|
//
|
|
#include "td/utils/algorithm.h"
|
|
#include "td/utils/common.h"
|
|
#include "td/utils/FlatHashMap.h"
|
|
#include "td/utils/FlatHashMapChunks.h"
|
|
#include "td/utils/FlatHashSet.h"
|
|
#include "td/utils/logging.h"
|
|
#include "td/utils/Random.h"
|
|
#include "td/utils/Slice.h"
|
|
#include "td/utils/tests.h"
|
|
|
|
#include <algorithm>
|
|
#include <array>
|
|
#include <functional>
|
|
#include <random>
|
|
#include <unordered_map>
|
|
#include <unordered_set>
|
|
#include <utility>
|
|
|
|
template <class T>
|
|
static auto extract_kv(const T &reference) {
|
|
auto expected = td::transform(reference, [](auto &it) { return std::make_pair(it.first, it.second); });
|
|
std::sort(expected.begin(), expected.end());
|
|
return expected;
|
|
}
|
|
|
|
template <class T>
|
|
static auto extract_k(const T &reference) {
|
|
auto expected = td::transform(reference, [](auto &it) { return it; });
|
|
std::sort(expected.begin(), expected.end());
|
|
return expected;
|
|
}
|
|
|
|
TEST(FlatHashMapChunks, basic) {
|
|
td::FlatHashMapChunks<int, int> kv;
|
|
kv[5] = 3;
|
|
ASSERT_EQ(3, kv[5]);
|
|
kv[3] = 4;
|
|
ASSERT_EQ(4, kv[3]);
|
|
}
|
|
|
|
TEST(FlatHashMap, probing) {
|
|
auto test = [](int buckets, int elements) {
|
|
CHECK(buckets >= elements);
|
|
td::vector<bool> data(buckets, false);
|
|
std::random_device rnd;
|
|
std::mt19937 mt(rnd());
|
|
std::uniform_int_distribution<td::int32> d(0, buckets - 1);
|
|
for (int i = 0; i < elements; i++) {
|
|
int pos = d(mt);
|
|
while (data[pos]) {
|
|
pos++;
|
|
if (pos == buckets) {
|
|
pos = 0;
|
|
}
|
|
}
|
|
data[pos] = true;
|
|
}
|
|
int max_chain = 0;
|
|
int cur_chain = 0;
|
|
for (auto x : data) {
|
|
if (x) {
|
|
cur_chain++;
|
|
max_chain = td::max(max_chain, cur_chain);
|
|
} else {
|
|
cur_chain = 0;
|
|
}
|
|
}
|
|
LOG(INFO) << "Buckets=" << buckets << " elements=" << elements << " max_chain=" << max_chain;
|
|
};
|
|
test(8192, static_cast<int>(8192 * 0.8));
|
|
test(8192, static_cast<int>(8192 * 0.6));
|
|
test(8192, static_cast<int>(8192 * 0.3));
|
|
}
|
|
|
|
struct A {
|
|
int a;
|
|
};
|
|
|
|
struct AHash {
|
|
std::size_t operator()(A a) const {
|
|
return std::hash<int>()(a.a);
|
|
}
|
|
};
|
|
|
|
static bool operator==(const A &lhs, const A &rhs) {
|
|
return lhs.a == rhs.a;
|
|
}
|
|
|
|
TEST(FlatHashSet, foreach) {
|
|
td::FlatHashSet<A, AHash> s;
|
|
for (auto it : s) {
|
|
LOG(ERROR) << it.a;
|
|
}
|
|
s.insert({1});
|
|
LOG(INFO) << s.begin()->a;
|
|
}
|
|
|
|
TEST(FlatHashSet, TL) {
|
|
td::FlatHashSet<int> s;
|
|
int N = 100000;
|
|
for (int i = 0; i < 10000000; i++) {
|
|
s.insert((i + N / 2) % N + 1);
|
|
s.erase(i % N + 1);
|
|
}
|
|
}
|
|
|
|
TEST(FlatHashMap, basic) {
|
|
{
|
|
td::FlatHashMap<int, int> map;
|
|
map[1] = 2;
|
|
ASSERT_EQ(2, map[1]);
|
|
ASSERT_EQ(1, map.find(1)->first);
|
|
ASSERT_EQ(2, map.find(1)->second);
|
|
// ASSERT_EQ(1, map.find(1)->key());
|
|
// ASSERT_EQ(2, map.find(1)->value());
|
|
for (auto &kv : map) {
|
|
ASSERT_EQ(1, kv.first);
|
|
ASSERT_EQ(2, kv.second);
|
|
}
|
|
map.erase(map.find(1));
|
|
}
|
|
|
|
td::FlatHashMap<int, std::array<td::unique_ptr<td::string>, 10>> x;
|
|
auto y = std::move(x);
|
|
x[12];
|
|
x.erase(x.find(12));
|
|
|
|
{
|
|
td::FlatHashMap<int, td::string> map = {{1, "hello"}, {2, "world"}};
|
|
ASSERT_EQ("hello", map[1]);
|
|
ASSERT_EQ("world", map[2]);
|
|
ASSERT_EQ(2u, map.size());
|
|
ASSERT_EQ("", map[3]);
|
|
ASSERT_EQ(3u, map.size());
|
|
}
|
|
|
|
{
|
|
td::FlatHashMap<int, td::string> map = {{1, "hello"}, {1, "world"}};
|
|
ASSERT_EQ("hello", map[1]);
|
|
ASSERT_EQ(1u, map.size());
|
|
}
|
|
|
|
using KV = td::FlatHashMap<td::string, td::string>;
|
|
using Data = td::vector<std::pair<td::string, td::string>>;
|
|
auto data = Data{{"a", "b"}, {"c", "d"}};
|
|
{ ASSERT_EQ(Data{}, extract_kv(KV())); }
|
|
|
|
{
|
|
KV kv;
|
|
for (auto &pair : data) {
|
|
kv.emplace(pair.first, pair.second);
|
|
}
|
|
ASSERT_EQ(data, extract_kv(kv));
|
|
|
|
KV moved_kv(std::move(kv));
|
|
ASSERT_EQ(data, extract_kv(moved_kv));
|
|
ASSERT_EQ(Data{}, extract_kv(kv));
|
|
ASSERT_TRUE(kv.empty());
|
|
kv = std::move(moved_kv);
|
|
ASSERT_EQ(data, extract_kv(kv));
|
|
|
|
KV assign_moved_kv;
|
|
assign_moved_kv = std::move(kv);
|
|
ASSERT_EQ(data, extract_kv(assign_moved_kv));
|
|
ASSERT_EQ(Data{}, extract_kv(kv));
|
|
ASSERT_TRUE(kv.empty());
|
|
kv = std::move(assign_moved_kv);
|
|
|
|
KV it_copy_kv;
|
|
for (auto &pair : kv) {
|
|
it_copy_kv.emplace(pair.first, pair.second);
|
|
}
|
|
ASSERT_EQ(data, extract_kv(it_copy_kv));
|
|
}
|
|
|
|
{
|
|
KV kv;
|
|
ASSERT_TRUE(kv.empty());
|
|
ASSERT_EQ(0u, kv.size());
|
|
for (auto &pair : data) {
|
|
kv.emplace(pair.first, pair.second);
|
|
}
|
|
ASSERT_TRUE(!kv.empty());
|
|
ASSERT_EQ(2u, kv.size());
|
|
|
|
ASSERT_EQ("a", kv.find("a")->first);
|
|
ASSERT_EQ("b", kv.find("a")->second);
|
|
kv.find("a")->second = "c";
|
|
ASSERT_EQ("c", kv.find("a")->second);
|
|
ASSERT_EQ("c", kv["a"]);
|
|
|
|
ASSERT_EQ(0u, kv.count("x"));
|
|
ASSERT_EQ(1u, kv.count("a"));
|
|
}
|
|
{
|
|
KV kv;
|
|
kv["d"];
|
|
ASSERT_EQ((Data{{"d", ""}}), extract_kv(kv));
|
|
kv.erase(kv.find("d"));
|
|
ASSERT_EQ(Data{}, extract_kv(kv));
|
|
}
|
|
}
|
|
|
|
TEST(FlatHashMap, remove_if_basic) {
|
|
td::Random::Xorshift128plus rnd(123);
|
|
|
|
constexpr int TESTS_N = 1000;
|
|
constexpr int MAX_TABLE_SIZE = 1000;
|
|
for (int test_i = 0; test_i < TESTS_N; test_i++) {
|
|
std::unordered_map<td::uint64, td::uint64> reference;
|
|
td::FlatHashMap<td::uint64, td::uint64> table;
|
|
int N = rnd.fast(1, MAX_TABLE_SIZE);
|
|
for (int i = 0; i < N; i++) {
|
|
auto key = rnd();
|
|
auto value = i;
|
|
reference[key] = value;
|
|
table[key] = value;
|
|
}
|
|
ASSERT_EQ(extract_kv(reference), extract_kv(table));
|
|
|
|
td::vector<std::pair<td::uint64, td::uint64>> kv;
|
|
td::table_remove_if(table, [&](auto &it) {
|
|
kv.emplace_back(it.first, it.second);
|
|
return it.second % 2 == 0;
|
|
});
|
|
std::sort(kv.begin(), kv.end());
|
|
ASSERT_EQ(extract_kv(reference), kv);
|
|
|
|
td::table_remove_if(reference, [](auto &it) { return it.second % 2 == 0; });
|
|
ASSERT_EQ(extract_kv(reference), extract_kv(table));
|
|
}
|
|
}
|
|
|
|
static constexpr size_t MAX_TABLE_SIZE = 1000;
|
|
TEST(FlatHashMap, stress_test) {
|
|
td::Random::Xorshift128plus rnd(123);
|
|
size_t max_table_size = MAX_TABLE_SIZE; // dynamic value
|
|
std::unordered_map<td::uint64, td::uint64> ref;
|
|
td::FlatHashMap<td::uint64, td::uint64> tbl;
|
|
|
|
auto validate = [&] {
|
|
ASSERT_EQ(ref.empty(), tbl.empty());
|
|
ASSERT_EQ(ref.size(), tbl.size());
|
|
ASSERT_EQ(extract_kv(ref), extract_kv(tbl));
|
|
for (auto &kv : ref) {
|
|
auto tbl_it = tbl.find(kv.first);
|
|
ASSERT_TRUE(tbl_it != tbl.end());
|
|
ASSERT_EQ(kv.second, tbl_it->second);
|
|
}
|
|
};
|
|
|
|
td::vector<td::RandomSteps::Step> steps;
|
|
auto add_step = [&](td::Slice step_name, td::uint32 weight, auto f) {
|
|
auto g = [&, f = std::move(f)] {
|
|
//ASSERT_EQ(ref.size(), tbl.size());
|
|
f();
|
|
ASSERT_EQ(ref.size(), tbl.size());
|
|
//validate();
|
|
};
|
|
steps.emplace_back(td::RandomSteps::Step{std::move(g), weight});
|
|
};
|
|
|
|
auto gen_key = [&] {
|
|
auto key = rnd() % 4000 + 1;
|
|
return key;
|
|
};
|
|
|
|
add_step("Reset hash table", 1, [&] {
|
|
validate();
|
|
td::reset_to_empty(ref);
|
|
td::reset_to_empty(tbl);
|
|
max_table_size = rnd.fast(1, MAX_TABLE_SIZE);
|
|
});
|
|
add_step("Clear hash table", 1, [&] {
|
|
validate();
|
|
ref.clear();
|
|
tbl.clear();
|
|
max_table_size = rnd.fast(1, MAX_TABLE_SIZE);
|
|
});
|
|
|
|
add_step("Insert random value", 1000, [&] {
|
|
if (tbl.size() > max_table_size) {
|
|
return;
|
|
}
|
|
auto key = gen_key();
|
|
auto value = rnd();
|
|
ref[key] = value;
|
|
tbl[key] = value;
|
|
ASSERT_EQ(ref[key], tbl[key]);
|
|
});
|
|
|
|
add_step("Emplace random value", 1000, [&] {
|
|
if (tbl.size() > max_table_size) {
|
|
return;
|
|
}
|
|
auto key = gen_key();
|
|
auto value = rnd();
|
|
auto ref_it = ref.emplace(key, value);
|
|
auto tbl_it = tbl.emplace(key, value);
|
|
ASSERT_EQ(ref_it.second, tbl_it.second);
|
|
ASSERT_EQ(key, tbl_it.first->first);
|
|
});
|
|
|
|
add_step("empty operator[]", 1000, [&] {
|
|
if (tbl.size() > max_table_size) {
|
|
return;
|
|
}
|
|
auto key = gen_key();
|
|
ASSERT_EQ(ref[key], tbl[key]);
|
|
});
|
|
|
|
add_step("reserve", 10, [&] { tbl.reserve(static_cast<size_t>(rnd() % max_table_size)); });
|
|
|
|
add_step("find", 1000, [&] {
|
|
auto key = gen_key();
|
|
auto ref_it = ref.find(key);
|
|
auto tbl_it = tbl.find(key);
|
|
ASSERT_EQ(ref_it == ref.end(), tbl_it == tbl.end());
|
|
if (ref_it != ref.end()) {
|
|
ASSERT_EQ(ref_it->first, tbl_it->first);
|
|
ASSERT_EQ(ref_it->second, tbl_it->second);
|
|
}
|
|
});
|
|
|
|
add_step("find_and_erase", 100, [&] {
|
|
auto key = gen_key();
|
|
auto ref_it = ref.find(key);
|
|
auto tbl_it = tbl.find(key);
|
|
ASSERT_EQ(ref_it == ref.end(), tbl_it == tbl.end());
|
|
if (ref_it != ref.end()) {
|
|
ref.erase(ref_it);
|
|
tbl.erase(tbl_it);
|
|
}
|
|
});
|
|
|
|
add_step("remove_if", 5, [&] {
|
|
auto mul = rnd();
|
|
auto bit = rnd() % 64;
|
|
auto condition = [&](auto &it) {
|
|
return (((it.second * mul) >> bit) & 1) == 0;
|
|
};
|
|
td::table_remove_if(tbl, condition);
|
|
td::table_remove_if(ref, condition);
|
|
});
|
|
|
|
td::RandomSteps runner(std::move(steps));
|
|
for (size_t i = 0; i < 1000000; i++) {
|
|
runner.step(rnd);
|
|
}
|
|
}
|
|
|
|
TEST(FlatHashSet, stress_test) {
|
|
td::vector<td::RandomSteps::Step> steps;
|
|
auto add_step = [&steps](td::Slice, td::uint32 weight, auto f) {
|
|
steps.emplace_back(td::RandomSteps::Step{std::move(f), weight});
|
|
};
|
|
|
|
td::Random::Xorshift128plus rnd(123);
|
|
size_t max_table_size = MAX_TABLE_SIZE; // dynamic value
|
|
std::unordered_set<td::uint64> ref;
|
|
td::FlatHashSet<td::uint64> tbl;
|
|
|
|
auto validate = [&] {
|
|
ASSERT_EQ(ref.empty(), tbl.empty());
|
|
ASSERT_EQ(ref.size(), tbl.size());
|
|
ASSERT_EQ(extract_k(ref), extract_k(tbl));
|
|
};
|
|
auto gen_key = [&] {
|
|
auto key = rnd() % 4000 + 1;
|
|
return key;
|
|
};
|
|
|
|
add_step("Reset hash table", 1, [&] {
|
|
validate();
|
|
td::reset_to_empty(ref);
|
|
td::reset_to_empty(tbl);
|
|
max_table_size = rnd.fast(1, MAX_TABLE_SIZE);
|
|
});
|
|
add_step("Clear hash table", 1, [&] {
|
|
validate();
|
|
ref.clear();
|
|
tbl.clear();
|
|
max_table_size = rnd.fast(1, MAX_TABLE_SIZE);
|
|
});
|
|
|
|
add_step("Insert random value", 1000, [&] {
|
|
if (tbl.size() > max_table_size) {
|
|
return;
|
|
}
|
|
auto key = gen_key();
|
|
ref.insert(key);
|
|
tbl.insert(key);
|
|
});
|
|
|
|
add_step("reserve", 10, [&] { tbl.reserve(static_cast<size_t>(rnd() % max_table_size)); });
|
|
|
|
add_step("find", 1000, [&] {
|
|
auto key = gen_key();
|
|
auto ref_it = ref.find(key);
|
|
auto tbl_it = tbl.find(key);
|
|
ASSERT_EQ(ref_it == ref.end(), tbl_it == tbl.end());
|
|
if (ref_it != ref.end()) {
|
|
ASSERT_EQ(*ref_it, *tbl_it);
|
|
}
|
|
});
|
|
|
|
add_step("find_and_erase", 100, [&] {
|
|
auto key = gen_key();
|
|
auto ref_it = ref.find(key);
|
|
auto tbl_it = tbl.find(key);
|
|
ASSERT_EQ(ref_it == ref.end(), tbl_it == tbl.end());
|
|
if (ref_it != ref.end()) {
|
|
ref.erase(ref_it);
|
|
tbl.erase(tbl_it);
|
|
}
|
|
});
|
|
|
|
add_step("remove_if", 5, [&] {
|
|
auto mul = rnd();
|
|
auto bit = rnd() % 64;
|
|
auto condition = [&](auto &it) {
|
|
return (((it * mul) >> bit) & 1) == 0;
|
|
};
|
|
td::table_remove_if(tbl, condition);
|
|
td::table_remove_if(ref, condition);
|
|
});
|
|
|
|
td::RandomSteps runner(std::move(steps));
|
|
for (size_t i = 0; i < 10000000; i++) {
|
|
runner.step(rnd);
|
|
}
|
|
}
|