|
| 1 | +// |
| 2 | +// hashtable.h |
| 3 | +// vm_persistence |
| 4 | +// |
| 5 | +// Created by Jinglei Ren on 12/24/14. |
| 6 | +// Copyright (c) 2014 Jinglei Ren <jinglei@ren.systems>. |
| 7 | +// |
| 8 | + |
| 9 | +#ifndef VM_PERSISTENCE_SLIB_HASHTABLE_H_ |
| 10 | +#define VM_PERSISTENCE_SLIB_HASHTABLE_H_ |
| 11 | + |
| 12 | +#include <cstdint> |
| 13 | +#include <cassert> |
| 14 | +#include "lib/list.h" |
| 15 | +#include "lib/mem_alloc.h" |
| 16 | + |
| 17 | +namespace slib { |
| 18 | + |
| 19 | +template<typename T> |
| 20 | +struct HashEqual { |
| 21 | + virtual uint64_t hash(const T &key) const { |
| 22 | + std::hash<T> hasher; |
| 23 | + return hasher(key); |
| 24 | + } |
| 25 | + |
| 26 | + virtual bool equal(const T &x, const T &y) const { |
| 27 | + std::equal_to<T> equal; |
| 28 | + return equal(x, y); |
| 29 | + } |
| 30 | + |
| 31 | + virtual ~HashEqual() { } |
| 32 | +}; |
| 33 | + |
| 34 | +struct hlist_bucket { |
| 35 | + std::size_t size; |
| 36 | + struct hlist_head head; |
| 37 | +}; |
| 38 | + |
| 39 | +template <typename K, typename V> |
| 40 | +struct hlist_pair { |
| 41 | + K key; |
| 42 | + V value; |
| 43 | + hlist_node node; |
| 44 | + |
| 45 | + hlist_pair(const K &k, const V &v) : key(k), value(v) { |
| 46 | + INIT_HLIST_NODE(&node); |
| 47 | + } |
| 48 | +}; |
| 49 | + |
| 50 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 51 | +class hashtable { |
| 52 | + public: |
| 53 | + hashtable(std::size_t num_buckets = 11, std::size_t local_load_factor = 4); |
| 54 | + size_t local_load_factor() const { return local_load_factor_; } |
| 55 | + void set_local_load_factor(std::size_t f) { local_load_factor_ = f; } |
| 56 | + std::size_t bucket_count() const { return bucket_count_; } |
| 57 | + |
| 58 | + bool find(const K &key, V &value) const; |
| 59 | + bool update(const K &key, V &value); |
| 60 | + bool insert(const K &key, const V &value); |
| 61 | + bool erase(const K &key, std::pair<K, V> &erased); |
| 62 | + std::vector<std::pair<K, V>> entries(const K *key = NULL, |
| 63 | + std::size_t n = -1) const; |
| 64 | + std::size_t clear(); |
| 65 | + void rehash(std::size_t buckets); |
| 66 | + |
| 67 | + std::size_t size() const; |
| 68 | + std::size_t bucket_size(size_t i) const { return buckets_[i].size; } |
| 69 | + float load_factor() const; |
| 70 | + |
| 71 | + ~hashtable(); |
| 72 | + |
| 73 | + private: |
| 74 | + hlist_bucket *get_bucket(const K &key) const { |
| 75 | + return buckets_ + hash_equal_.hash(key) % bucket_count_; |
| 76 | + } |
| 77 | + |
| 78 | + std::size_t local_load_factor_; |
| 79 | + hlist_bucket *buckets_; |
| 80 | + std::size_t bucket_count_; |
| 81 | + HashEqual hash_equal_; |
| 82 | +}; |
| 83 | + |
| 84 | +// Accessory functions |
| 85 | + |
| 86 | +template <class Alloc> |
| 87 | +inline hlist_bucket *new_buckets(std::size_t n) { |
| 88 | + hlist_bucket *bkts = (hlist_bucket *)Alloc::Malloc(n * sizeof(hlist_bucket)); |
| 89 | + for (std::size_t i = 0; i < n; ++i) { |
| 90 | + bkts[i].size = 0; |
| 91 | + INIT_HLIST_HEAD(&bkts[i].head); |
| 92 | + } |
| 93 | + return bkts; |
| 94 | +} |
| 95 | + |
| 96 | +inline std::size_t size_sum(const hlist_bucket *bkts, std::size_t n) { |
| 97 | + std::size_t count = 0; |
| 98 | + for (std::size_t i = 0; i < n; ++i) { |
| 99 | + count += bkts[i].size; |
| 100 | + } |
| 101 | + return count; |
| 102 | +} |
| 103 | + |
| 104 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 105 | +hlist_pair<K, V> *find_in(const hlist_bucket *bkt, const K &key, |
| 106 | + const HashEqual &hash_equal = HashEqual()) { |
| 107 | + hlist_node *node; |
| 108 | + hlist_for_each(node, &bkt->head) { |
| 109 | + hlist_pair<K, V> *pair = container_of(node, &hlist_pair<K, V>::node); |
| 110 | + if (hash_equal.equal(pair->key, key)) { |
| 111 | + return pair; |
| 112 | + } |
| 113 | + } |
| 114 | + return NULL; |
| 115 | +} |
| 116 | + |
| 117 | +template <typename K, typename V> |
| 118 | +void insert_to(hlist_bucket *bkt, hlist_pair<K, V> *pair) { |
| 119 | + hlist_add_head(&pair->node, &bkt->head); |
| 120 | + ++bkt->size; |
| 121 | +} |
| 122 | + |
| 123 | +template <typename K, typename V, class Alloc> |
| 124 | +void erase_from(hlist_bucket *bkt, hlist_pair<K, V> *pair) { |
| 125 | + hlist_del(&pair->node); |
| 126 | + Alloc::Delete(pair); |
| 127 | + --bkt->size; |
| 128 | + assert((bool)bkt->size != hlist_empty(&bkt->head)); |
| 129 | +} |
| 130 | + |
| 131 | +template <typename K, typename V, class Alloc> |
| 132 | +std::size_t clear_all(hlist_bucket *bkts, std::size_t n) { |
| 133 | + std::size_t num = 0; |
| 134 | + for (std::size_t i = 0; i < n; ++i) { |
| 135 | + hlist_node *pos, *next; |
| 136 | + hlist_for_each_safe(pos, next, &bkts[i].head) { |
| 137 | + hlist_pair<K, V> *pair = container_of(pos, &hlist_pair<K, V>::node); |
| 138 | + erase_from<K, V, Alloc>(bkts + i, pair); |
| 139 | + ++num; |
| 140 | + } |
| 141 | + assert(!bkts[i].size && hlist_empty(&bkts[i].head)); |
| 142 | + } |
| 143 | + return num; |
| 144 | +} |
| 145 | + |
| 146 | +// Implementation of hashtable |
| 147 | + |
| 148 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 149 | +hashtable<K, V, HashEqual, Alloc>::hashtable(std::size_t n, std::size_t f) { |
| 150 | + bucket_count_ = n; |
| 151 | + buckets_ = new_buckets<Alloc>(bucket_count_); |
| 152 | + local_load_factor_ = f; |
| 153 | +} |
| 154 | + |
| 155 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 156 | +bool hashtable<K, V, HashEqual, Alloc>::find(const K &key, V &value) const { |
| 157 | + hlist_bucket *bkt = get_bucket(key); |
| 158 | + hlist_pair<K, V> *pair = find_in<K, V, HashEqual, Alloc>(bkt, key); |
| 159 | + if (!pair) return false; |
| 160 | + value = pair->value; |
| 161 | + return true; |
| 162 | +} |
| 163 | + |
| 164 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 165 | +bool hashtable<K, V, HashEqual, Alloc>::update(const K &key, V &value) { |
| 166 | + hlist_bucket *bkt = get_bucket(key); |
| 167 | + hlist_pair<K, V> *pair = find_in<K, V, HashEqual, Alloc>(bkt, key); |
| 168 | + if (!pair) return false; |
| 169 | + V old = pair->value; |
| 170 | + pair->value = value; |
| 171 | + value = old; |
| 172 | + return true; |
| 173 | +} |
| 174 | + |
| 175 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 176 | +bool hashtable<K, V, HashEqual, Alloc>::insert(const K &key, const V &value) { |
| 177 | + hlist_bucket *bkt = get_bucket(key); |
| 178 | + hlist_pair<K, V> *pair = find_in<K, V, HashEqual, Alloc>(bkt, key); |
| 179 | + if (pair) return false; |
| 180 | + pair = Alloc::template New<hlist_pair<K, V>>(key, value); |
| 181 | + insert_to(bkt, pair); |
| 182 | + |
| 183 | + if (bkt->size >= local_load_factor_) { |
| 184 | + rehash((bucket_count_ << 1) + 1); |
| 185 | + } |
| 186 | + return true; |
| 187 | +} |
| 188 | + |
| 189 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 190 | +bool hashtable<K, V, HashEqual, Alloc>::erase(const K &key, std::pair<K, V> &erased) { |
| 191 | + hlist_bucket *bkt = get_bucket(key); |
| 192 | + hlist_pair<K, V> *pair = find_in<K, V, HashEqual, Alloc>(bkt, key); |
| 193 | + if (!pair) return false; |
| 194 | + erased.first = pair->key; |
| 195 | + erased.second = pair->value; |
| 196 | + erase_from<K, V, Alloc>(bkt, pair); |
| 197 | + return true; |
| 198 | +} |
| 199 | + |
| 200 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 201 | +std::vector<std::pair<K, V>> hashtable<K, V, HashEqual, Alloc>::entries( |
| 202 | + const K *key, std::size_t num) const { |
| 203 | + std::vector<std::pair<K, V>> pairs; |
| 204 | + hlist_bucket *bkt = key ? get_bucket(*key) : buckets_; |
| 205 | + hlist_pair<K, V> *pos = key ? find_in<K, V, HashEqual, Alloc>(bkt, *key) : |
| 206 | + container_of(bkt->head.first, &hlist_pair<K, V>::node); |
| 207 | + if (!pos) return pairs; |
| 208 | + |
| 209 | + hlist_bucket *bkt_end = buckets_ + bucket_count_; |
| 210 | + hlist_node *node = &pos->node; |
| 211 | + for (; bkt < bkt_end; ++bkt) { |
| 212 | + if (!node) node = bkt->head.first; |
| 213 | + for (; node; node = node->next) { |
| 214 | + hlist_pair<K, V> *pair = container_of(node, &hlist_pair<K, V>::node); |
| 215 | + pairs.push_back(std::make_pair(pair->key, pair->value)); |
| 216 | + if (pairs.size() == num) return pairs; |
| 217 | + } |
| 218 | + } |
| 219 | + return pairs; |
| 220 | +} |
| 221 | + |
| 222 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 223 | +std::size_t hashtable<K, V, HashEqual, Alloc>::clear() { |
| 224 | + return clear_all<K, V, Alloc>(buckets_, bucket_count_); |
| 225 | +} |
| 226 | + |
| 227 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 228 | +hashtable<K, V, HashEqual, Alloc>::~hashtable() { |
| 229 | + clear_all<K, V, Alloc>(buckets_, bucket_count_); |
| 230 | + Alloc::Free(buckets_, sizeof(hlist_bucket) * bucket_count_); |
| 231 | +} |
| 232 | + |
| 233 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 234 | +void hashtable<K, V, HashEqual, Alloc>::rehash(std::size_t n) { |
| 235 | + hlist_bucket *bkts = new_buckets<Alloc>(n); |
| 236 | + std::size_t num = 0; |
| 237 | + for (std::size_t i = 0; i < bucket_count_; ++i) { |
| 238 | + hlist_node *pos, *next; |
| 239 | + hlist_for_each_safe(pos, next, &buckets_[i].head) { |
| 240 | + hlist_pair<K, V> *pair = container_of(pos, &hlist_pair<K, V>::node); |
| 241 | + std::size_t j = hash_equal_.hash(pair->key) % n; |
| 242 | + insert_to(bkts + j, pair); |
| 243 | + ++num; |
| 244 | + } |
| 245 | + } |
| 246 | + Alloc::Free(buckets_, sizeof(hlist_bucket) * bucket_count_); |
| 247 | + buckets_ = bkts; |
| 248 | + bucket_count_ = n; |
| 249 | +} |
| 250 | + |
| 251 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 252 | +std::size_t hashtable<K, V, HashEqual, Alloc>::size() const { |
| 253 | + return size_sum(buckets_, bucket_count_); |
| 254 | +} |
| 255 | + |
| 256 | +template <typename K, typename V, class HashEqual, class Alloc> |
| 257 | +float hashtable<K, V, HashEqual, Alloc>::load_factor() const { |
| 258 | + return size() / (float)bucket_count_; |
| 259 | +} |
| 260 | + |
| 261 | +} // namespace slib |
| 262 | + |
| 263 | +#endif // VM_PERSISTENCE_SLIB_HASHTABLE_H_ |
| 264 | + |
0 commit comments