Add hash set implementation
Release / release (push) Successful in 4s

This commit is contained in:
2026-09-23 22:34:23 +01:00
parent 20e890964f
commit 68640dc5e2
12 changed files with 590 additions and 44 deletions
+51
View File
@@ -0,0 +1,51 @@
// vim:fileencoding=utf-8:foldmethod=marker
#include "set.h"
#include "../table/table.h"
WpHashSet hashSetAlloc(const WpAllocator *allocator, WpU8StreamEncoder encoder, WpKeyEqualTest key_eq,
u64 capacity, u64 key_size) {
return hashSetAllocWithCustomHasher(allocator, WP_HASH_TABLE_DEFAULT_HASHER, encoder, key_eq,
capacity, key_size);
}
WpHashSet hashSetAllocWithCustomHasher(const WpAllocator *allocator, WpHasher hasher,
WpU8StreamEncoder encoder, WpKeyEqualTest key_eq, u64 capacity,
u64 key_size) {
WpHashSet set = {0};
hashTableAlloc(allocator, &set, hasher, encoder, key_eq, capacity, key_size, 0, false);
return set;
}
WpHashLookupResult hashSetFind(WpHashSet *set, void *item, u64 item_size) {
u64 hash = hashTableCalcHash(set, item, item_size);
WpHashBucketRef ref = hashTableBucketRef(set, hash);
return hashTableLookup(set, item, ref, item_size);
}
void hashSetInsertCapped(WpHashSet *set, void *item, u64 item_size) {
WpHashLookupResult result = hashSetFind(set, item, item_size);
if (!result.found && set->load_factor < WP_HASH_MAX_LOAD_FACTOR) {
hashTableInsert(set, item, NULL, result.ref, item_size, 0);
}
}
void hashSetInsertAlloc(const WpAllocator *allocator, WpHashSet *set, void *item, u64 item_size) {
WpHashLookupResult result = hashSetFind(set, item, item_size);
if (!result.found) {
if (set->load_factor >= WP_HASH_MAX_LOAD_FACTOR) {
hashTableGrow(allocator, set, item_size, 0);
// Rerun item lookup after growing
result = hashSetFind(set, item, item_size);
}
hashTableInsert(set, item, NULL, result.ref, item_size, 0);
}
}
void hashSetRemove(WpHashSet *set, void *item, u64 item_size) {
WpHashLookupResult result = hashSetFind(set, item, item_size);
if (result.found) {
hashTableRemove(set, result.ref, item_size, 0);
}
}
+105
View File
@@ -0,0 +1,105 @@
// vim:fileencoding=utf-8:foldmethod=marker
#ifndef SET_H
#define SET_H
#include "../table/table.h"
#include "../../../common/misc/misc_utils.h"
#include "../../../common/platform/platform.h"
#ifdef WP_PLATFORM_CPP
BEGIN_C_LINKAGE
#endif // !WP_PLATFORM_CPP
typedef WpHashTable WpHashSet;
#ifdef WP_PLATFORM_CPP
#define wpHashSet(TYPE, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY) \
(WpHashSet{ \
WP_HASH_TABLE_DEFAULT_HASHER, \
U8_STREAM_ENCODER, \
ITEM_EQ_TEST, \
wpArrayWithCapacity(TYPE, wpMiscUtilsU64RoundUpPow2(CAPACITY), WP_ARRAY_INIT_NONE), \
NULL, \
wpArrayWithCapacity(WpHashBucket, wpMiscUtilsU64RoundUpPow2(CAPACITY), WP_ARRAY_INIT_FILLED), \
WP_HASH_TABLE_MAGIC, \
wpMiscUtilsU64RoundUpPow2(CAPACITY) - 1, \
0.0f, \
false, \
})
#define wpHashSetCustomHasher(TYPE, HASHER, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY) \
(WpHashSet{ \
HASHER, \
U8_STREAM_ENCODER, \
ITEM_EQ_TEST, \
wpArrayWithCapacity(TYPE, wpMiscUtilsU64RoundUpPow2(CAPACITY), WP_ARRAY_INIT_NONE), \
NULL, \
wpArrayWithCapacity(WpHashBucket, wpMiscUtilsU64RoundUpPow2(CAPACITY), WP_ARRAY_INIT_FILLED), \
WP_HASH_TABLE_MAGIC, \
wpMiscUtilsU64RoundUpPow2(CAPACITY) - 1, \
0.0f, \
false, \
})
#else
#define wpHashSet(TYPE, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY) \
((WpHashSet){ \
.hasher = WP_HASH_TABLE_DEFAULT_HASHER, \
.encoder = U8_STREAM_ENCODER, \
.key_eq = ITEM_EQ_TEST, \
.keys = wpArrayWithCapacity(TYPE, wpMiscUtilsU64RoundUpPow2(CAPACITY), \
WP_ARRAY_INIT_NONE), \
.buckets = wpArrayWithCapacity(WpHashBucket, wpMiscUtilsU64RoundUpPow2(CAPACITY), \
WP_ARRAY_INIT_FILLED), \
.magic = WP_HASH_TABLE_MAGIC, \
.bucket_mask = wpMiscUtilsU64RoundUpPow2(CAPACITY) - 1, \
})
#define wpHashSetCustomHasher(TYPE, HASHER, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY) \
((WpHashSet){ \
.hasher = HASHER, \
.encoder = U8_STREAM_ENCODER, \
.key_eq = ITEM_EQ_TEST, \
.keys = wpArrayWithCapacity(TYPE, wpMiscUtilsU64RoundUpPow2(CAPACITY), \
WP_ARRAY_INIT_NONE), \
.buckets = wpArrayWithCapacity(WpHashBucket, wpMiscUtilsU64RoundUpPow2(CAPACITY), \
WP_ARRAY_INIT_FILLED), \
.magic = WP_HASH_TABLE_MAGIC, \
.bucket_mask = wpMiscUtilsU64RoundUpPow2(CAPACITY) - 1, \
})
#endif
#define wpHashSetAlloc(TYPE, ALLOCATOR_PTR, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY) \
hashSetAlloc(ALLOCATOR_PTR, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY, sizeof(TYPE))
#define wpHashSetCustomHasherAlloc(TYPE, ALLOCATOR_PTR, HASHER, U8_STREAM_ENCODER, ITEM_EQ_TEST, \
CAPACITY) \
hashSetAllocWithCustomHasher(ALLOCATOR_PTR, HASHER, U8_STREAM_ENCODER, ITEM_EQ_TEST, CAPACITY, \
sizeof(TYPE))
#define wpHashSetFind(TYPE, SET_PTR, ITEM_PTR) hashSetFind(SET_PTR, ITEM_PTR, sizeof(TYPE))
#define wpHashSetInsertCapped(TYPE, SET_PTR, ITEM_PTR) \
hashSetInsertCapped(SET_PTR, ITEM_PTR, sizeof(TYPE))
#define wpHashSetInsertAlloc(TYPE, ALLOCATOR_PTR, SET_PTR, ITEM_PTR) \
hashSetInsertAlloc(ALLOCATOR_PTR, SET_PTR, ITEM_PTR, sizeof(TYPE))
#define wpHashSetRemove(TYPE, SET_PTR, ITEM_PTR) hashSetRemove(SET_PTR, ITEM_PTR, sizeof(TYPE))
#define wpHashSetItems(TYPE, SET_PTR) ((TYPE *)hashTableKeys(SET_PTR, sizeof(TYPE)))
#define wpHashSetCapacity(SET_PTR) hashTableCapacity(SET_PTR)
#define wpHashSetCount(SET_PTR) hashTableCount(SET_PTR)
#define wpHashSetItemSize(SET_PTR) hashTableKeySize(SET_PTR)
#define wpHashSetClear(TYPE, SET_PTR) hashTableClear(SET_PTR, sizeof(TYPE), 0)
#define wpHashSetDealloc(TYPE, ALLOCATOR_PTR, SET_PTR) \
hashTableDealloc(ALLOCATOR_PTR, SET_PTR, sizeof(TYPE), 0)
WpHashSet hashSetAlloc(const WpAllocator *allocator, WpU8StreamEncoder encoder,
WpKeyEqualTest key_eq, u64 capacity, u64 key_size);
WpHashSet hashSetAllocWithCustomHasher(const WpAllocator *allocator, WpHasher hasher,
WpU8StreamEncoder encoder, WpKeyEqualTest key_eq,
u64 capacity, u64 key_size);
WpHashLookupResult hashSetFind(WpHashSet *set, void *item, u64 item_size);
void hashSetInsertCapped(WpHashSet *set, void *item, u64 item_size);
void hashSetInsertAlloc(const WpAllocator *allocator, WpHashSet *set, void *item,
u64 item_size);
void hashSetRemove(WpHashSet *set, void *item, u64 item_size);
#ifdef WP_PLATFORM_CPP
END_C_LINKAGE
#endif // !WP_PLATFORM_CPP
#endif // !SET_H
+77 -33
View File
@@ -8,26 +8,62 @@
#include "../../../common/assert/assert.h"
#include "../../../common/misc/misc_utils.h"
wp_persist inline void hashTableValidate(const WpHashTable *table, u64 key_size, u64 value_size);
wp_persist inline void *hashTableGetKeyFromBucket(WpHashTable *table, WpHashBucket bucket, u64 key_size);
wp_persist inline void *hashTableGetValueFromBucket(WpHashTable *table, WpHashBucket bucket, u64 value_size);
wp_persist inline u64 hashTableGetDist(u64 dist_and_tag);
wp_persist inline u64 hashTableNextBucket(WpHashTable *table, u64 bucket_index);
wp_persist inline u64 hashTableDistInc(u64 dist_and_tag);
wp_persist inline u64 hashTableDistDec(u64 dist_and_tag);
wp_persist inline void hashTableValidate(const WpHashTable *table, u64 key_size, u64 value_size);
void *hashTableKeys(WpHashTable *table) {
void hashTableAlloc(const WpAllocator *allocator, WpHashTable *table, WpHasher hasher,
WpU8StreamEncoder encoder, WpKeyEqualTest key_eq, u64 capacity, u64 key_size,
u64 value_size, b8 has_values) {
wpDebugAssert(allocator != NULL && table != NULL && encoder != NULL && key_eq != NULL,
"`allocator`, `table`, `encoder` and `key_eq` should not be NULL");
u64 table_capacity = wpMiscUtilsU64RoundUpPow2(capacity);
table->hasher = hasher;
table->encoder = encoder;
table->key_eq = key_eq;
table->keys = arrayAllocCapacity(allocator, table_capacity, WP_ARRAY_INIT_NONE, key_size);;
table->buckets = wpArrayAllocCapacity(WpHashBucket, allocator, table_capacity, WP_ARRAY_INIT_FILLED);
table->magic = WP_HASH_TABLE_MAGIC;
table->bucket_mask = table_capacity - 1;
table->load_factor = WP_HASH_MAX_LOAD_FACTOR;
table->has_values = has_values;
wpRuntimeAssert(table->keys != NULL && table->buckets != NULL, "Failed to allocate keys, buckets or both");
if (has_values) {
table->values = arrayAllocCapacity(allocator, table_capacity, WP_ARRAY_INIT_NONE, value_size);;
wpRuntimeAssert(table->values != NULL, "Failed to allocate values");
}
}
u64 hashTableCalcHash(WpHashTable *table, void *key, u64 key_size) {
hashTableValidate(table, key_size, 0);
wpDebugAssert(key != NULL, "`key` should not be NULL");
WpU8Stream stream = table->encoder(key);
return wpHasherGetHash64(&table->hasher, &stream);
}
WpHashBucketRef hashTableBucketRef(WpHashTable *table, u64 hash) {
hashTableValidate(table, 0, 0);
return (WpHashBucketRef){
.dist_and_tag = WP_HASH_DIST_INC | (hash & WP_HASH_TAG_MASK),
.bucket_index = hash & table->bucket_mask,
};
}
void *hashTableKeys(WpHashTable *table, u64 key_size) {
hashTableValidate(table, key_size, 0);
return table->keys;
}
void *hashTableValues(WpHashTable *table) {
hashTableValidate(table, 0, 0);
return table->values;
}
void *hashTableGetKeyFromBucket(WpHashTable *table, WpHashBucket bucket, u64 key_size) {
hashTableValidate(table, key_size, 0);
return arrayGet(table->keys, bucket.item_index, key_size);
}
void *hashTableGetValueFromBucket(WpHashTable *table, WpHashBucket bucket, u64 value_size) {
void *hashTableValues(WpHashTable *table, u64 value_size) {
hashTableValidate(table, 0, value_size);
return table->has_values ? arrayGet(table->values, bucket.item_index, value_size) : NULL;
return table->values;
}
void *hashTableGetKeyFromBucketRef(WpHashTable *table, WpHashBucketRef ref, u64 key_size) {
@@ -58,25 +94,21 @@ u64 hashTableValueSize(WpHashTable *table) {
return table->has_values ? wpArrayItemSize(table->values) : 0;
}
u64 hashTableCalcHash(WpHashTable *table, void *key, u64 key_size) {
hashTableValidate(table, key_size, 0);
wpDebugAssert(key != NULL, "`key` should not be NULL");
WpU8Stream stream = table->encoder(key);
return wpHasherGetHash64(&table->hasher, &stream);
}
void hashTableGrow(WpHashTable *table, const WpAllocator *allocator, u64 key_size, u64 value_size) {
void hashTableGrow(const WpAllocator *allocator, WpHashTable *table, u64 key_size, u64 value_size) {
hashTableValidate(table, key_size, value_size);
wpDebugAssert(allocator != NULL, "`allocator` should not be NULL");
u64 capacity = hashTableCapacity(table) * 2;
u64 capacity = wpMiscUtilsU64RoundUpPow2(hashTableCapacity(table) * 2);
u64 mask = capacity - wpU64Const(1);
WpArray keys = arrayAllocCapacity(allocator, capacity, WP_ARRAY_INIT_NONE, key_size);
WpArray values = NULL;
WpHashBucketArray buckets = wpArrayAllocCapacity(WpHashBucket, allocator, capacity, WP_ARRAY_INIT_FILLED);
wpRuntimeAssert(keys != NULL && buckets != NULL, "Failed to allocate keys, buckets or both");
if (table->has_values) {
values = arrayAllocCapacity(allocator, capacity, WP_ARRAY_INIT_NONE, value_size);
wpRuntimeAssert(values != NULL, "Failed to allocate values");
}
WpArray old_keys = table->keys;
@@ -212,27 +244,39 @@ void hashTableClear(WpHashTable *table, u64 key_size, u64 value_size) {
wpArrayZero(WpHashBucket, table->buckets);
}
WpHashBucketRef hashTableBucketRef(WpHashTable *table, u64 hash) {
hashTableValidate(table, 0, 0);
return (WpHashBucketRef){
.dist_and_tag = WP_HASH_DIST_INC | (hash & WP_HASH_TAG_MASK),
.bucket_index = hash & table->bucket_mask,
};
void hashTableDealloc(const WpAllocator *allocator, WpHashTable *table, u64 key_size, u64 value_size) {
wpDebugAssert(allocator != NULL, "`allocator` should not be NULL");
hashTableValidate(table, key_size, value_size);
arrayDealloc(allocator, table->keys, key_size);
wpArrayDealloc(WpHashBucket, allocator, table->buckets);
if (table->has_values) {
arrayDealloc(allocator, table->values, value_size);
}
}
u64 hashTableGetDist(u64 dist_and_tag) {
wp_persist inline void *hashTableGetKeyFromBucket(WpHashTable *table, WpHashBucket bucket, u64 key_size) {
hashTableValidate(table, key_size, 0);
return arrayGet(table->keys, bucket.item_index, key_size);
}
wp_persist inline void *hashTableGetValueFromBucket(WpHashTable *table, WpHashBucket bucket, u64 value_size) {
hashTableValidate(table, 0, value_size);
return table->has_values ? arrayGet(table->values, bucket.item_index, value_size) : NULL;
}
wp_persist inline u64 hashTableGetDist(u64 dist_and_tag) {
return dist_and_tag & (~WP_HASH_TAG_MASK);
}
u64 hashTableNextBucket(WpHashTable *table, u64 bucket_index) {
wp_persist inline u64 hashTableNextBucket(WpHashTable *table, u64 bucket_index) {
return (++bucket_index) & table->bucket_mask;
}
u64 hashTableDistInc(u64 dist_and_tag) {
wp_persist inline u64 hashTableDistInc(u64 dist_and_tag) {
return dist_and_tag + WP_HASH_DIST_INC;
}
u64 hashTableDistDec(u64 dist_and_tag) {
wp_persist inline u64 hashTableDistDec(u64 dist_and_tag) {
return dist_and_tag - WP_HASH_DIST_INC;
}
+11 -11
View File
@@ -69,18 +69,21 @@ typedef struct WpHashLookupResult {
b8 found;
} WpHashLookupResult;
void *hashTableKeys(WpHashTable *table);
void *hashTableValues(WpHashTable *table);
void *hashTableGetKeyFromBucket(WpHashTable *table, WpHashBucket bucket, u64 key_size);
void *hashTableGetValueFromBucket(WpHashTable *table, WpHashBucket bucket, u64 value_size);
void hashTableAlloc(const WpAllocator *allocator, WpHashTable *table,
WpHasher hasher, WpU8StreamEncoder encoder,
WpKeyEqualTest key_eq, u64 capacity, u64 key_size,
u64 value_size, b8 has_values);
u64 hashTableCalcHash(WpHashTable *table, void *key, u64 key_size);
WpHashBucketRef hashTableBucketRef(WpHashTable *table, u64 hash);
void *hashTableKeys(WpHashTable *table, u64 key_size);
void *hashTableValues(WpHashTable *table, u64 value_size);
void *hashTableGetKeyFromBucketRef(WpHashTable *table, WpHashBucketRef ref, u64 key_size);
void *hashTableGetValueFromBucketRef(WpHashTable *table, WpHashBucketRef ref, u64 value_size);
u64 hashTableCapacity(WpHashTable *table);
u64 hashTableCount(WpHashTable *table);
u64 hashTableKeySize(WpHashTable *table);
u64 hashTableValueSize(WpHashTable *table);
u64 hashTableCalcHash(WpHashTable *table, void *key, u64 key_size);
void hashTableGrow(WpHashTable *table, const WpAllocator *allocator, u64 key_size,
void hashTableGrow(const WpAllocator *allocator, WpHashTable *table, u64 key_size,
u64 value_size);
WpHashLookupResult hashTableLookup(WpHashTable *table, void *key, WpHashBucketRef ref, u64 key_size);
void hashTableUpdate(WpHashTable *table, void *value, WpHashBucketRef ref, u64 value_size);
@@ -88,11 +91,8 @@ void hashTableInsert(WpHashTable *table, void *key, void *value, W
u64 key_size, u64 value_size);
void hashTableRemove(WpHashTable *table, WpHashBucketRef ref, u64 key_size, u64 value_size);
void hashTableClear(WpHashTable *table, u64 key_size, u64 value_size);
WpHashBucketRef hashTableBucketRef(WpHashTable *table, u64 hash);
u64 hashTableGetDist(u64 dist_and_tag);
u64 hashTableNextBucket(WpHashTable *table, u64 bucket_index);
u64 hashTableDistInc(u64 dist_and_tag);
u64 hashTableDistDec(u64 dist_and_tag);
void hashTableDealloc(const WpAllocator *allocator, WpHashTable *table, u64 key_size,
u64 value_size);
#ifdef WP_PLATFORM_CPP
END_C_LINKAGE
+1
View File
@@ -10,6 +10,7 @@
#include "hash/hasher/murmur3.c"
#include "hash/hasher/siphash.c"
#include "hash/table/table.c"
#include "hash/set/set.c"
#include "queue/queue.c"
#include "stream/stream.c"
#include "mem/allocator/mem_allocator.c"
+1
View File
@@ -9,6 +9,7 @@
#include "hash/hasher/murmur3.h"
#include "hash/hasher/siphash.h"
#include "hash/table/table.h"
#include "hash/set/set.h"
#include "queue/queue.h"
#include "stream/stream.h"
#include "mem/allocator/mem_allocator.h"