scc/runtime/scc_utils/src/hashtable.c

#include <scc_hashtable.h>

#ifndef SCC_INIT_HASHMAP_SIZE
#define SCC_INIT_HASHMAP_SIZE (32)
#endif

void scc_hashtable_init(scc_hashtable_t *ht) {
    scc_vec_init(ht->entries);
    ht->count = 0;
    ht->tombstone_count = 0;
    Assert(ht->key_cmp != NULL && ht->hash_func != NULL);
}

static int next_power_of_two(int n) {
    n--;
    n |= n >> 1;
    n |= n >> 2;
    n |= n >> 4;
    n |= n >> 8;
    n |= n >> 16;
    return n + 1;
}

static scc_hashtable_entry_t *find_entry(scc_hashtable_t *ht, const void *key,
                                         u32 hash) {
    if (ht->entries.cap == 0)
        return NULL;

    u32 index = hash & (ht->entries.cap - 1); // 容量是2的幂
    u32 probe = 0;

    scc_hashtable_entry_t *tombstone = NULL;

    while (1) {
        scc_hashtable_entry_t *entry = &scc_vec_at(ht->entries, index);
        if (entry->state == ENTRY_EMPTY) {
            return tombstone ? tombstone : entry;
        }

        if (entry->state == ENTRY_TOMBSTONE) {
            if (!tombstone)
                tombstone = entry;
        } else if (entry->hash == hash && ht->key_cmp(entry->key, key) == 0) {
            return entry;
        }

        // Liner finding
        index = (index + 1) & (ht->entries.cap - 1);
        probe++;
        if (probe >= ht->entries.cap)
            break;
    }
    LOG_ERROR("hashset_find: hash table is full");
    return NULL;
}

static void adjust_capacity(scc_hashtable_t *ht, usize new_cap) {
    new_cap = next_power_of_two(new_cap);
    Assert(new_cap >= ht->entries.cap);

    SCC_VEC(scc_hashtable_entry_t) old_entries;
    old_entries.data = ht->entries.data;
    old_entries.cap = ht->entries.cap;

    // Not used size but for gdb python extention debug
    ht->entries.size = new_cap;
    ht->entries.cap = new_cap;
    ht->entries.data =
        scc_realloc(NULL, new_cap * sizeof(scc_hashtable_entry_t));
    scc_memset(ht->entries.data, 0, new_cap * sizeof(scc_hashtable_entry_t));

    // rehash the all of the old data
    for (usize i = 0; i < old_entries.cap; i++) {
        scc_hashtable_entry_t *entry = &scc_vec_at(old_entries, i);
        if (entry->state == ENTRY_ACTIVE) {
            scc_hashtable_entry_t *dest =
                find_entry(ht, entry->key, entry->hash);
            *dest = *entry;
        }
    }

    scc_vec_free(old_entries);
    ht->tombstone_count = 0;
}

void *scc_hashtable_set(scc_hashtable_t *ht, const void *key, void *value) {
    if (ht->count + ht->tombstone_count >= ht->entries.cap * 0.75) {
        int new_cap = ht->entries.cap < SCC_INIT_HASHMAP_SIZE
                          ? SCC_INIT_HASHMAP_SIZE
                          : ht->entries.cap * 2;
        adjust_capacity(ht, new_cap);
    }

    u32 hash = ht->hash_func(key);
    scc_hashtable_entry_t *entry = find_entry(ht, key, hash);

    void *old_value = NULL;
    if (entry->state == ENTRY_ACTIVE) {
        old_value = entry->value;
    } else {
        if (entry->state == ENTRY_TOMBSTONE)
            ht->tombstone_count--;
        ht->count++;
    }

    entry->key = key;
    entry->value = value;
    entry->hash = hash;
    entry->state = ENTRY_ACTIVE;
    return old_value;
}

void *scc_hashtable_get(scc_hashtable_t *ht, const void *key) {
    if (ht->entries.cap == 0)
        return NULL;

    u32 hash = ht->hash_func(key);
    scc_hashtable_entry_t *entry = find_entry(ht, key, hash);
    return (entry && entry->state == ENTRY_ACTIVE) ? entry->value : NULL;
}

void *scc_hashtable_del(scc_hashtable_t *ht, const void *key) {
    if (ht->entries.cap == 0)
        return NULL;

    u32 hash = ht->hash_func(key);
    scc_hashtable_entry_t *entry = find_entry(ht, key, hash);

    if (entry == NULL || entry->state != ENTRY_ACTIVE)
        return NULL;

    void *value = entry->value;
    entry->state = ENTRY_TOMBSTONE;
    ht->count--;
    ht->tombstone_count++;
    return value;
}

void scc_hashtable_drop(scc_hashtable_t *ht) {
    scc_vec_free(ht->entries);
    ht->count = 0;
    ht->tombstone_count = 0;
}

void scc_hashtable_foreach(scc_hashtable_t *ht, scc_hashtable_iter_fn iter_func,
                           void *context) {
    for (usize i = 0; i < ht->entries.cap; i++) {
        scc_hashtable_entry_t *entry = &scc_vec_at(ht->entries, i);
        if (entry->state == ENTRY_ACTIVE) {
            if (iter_func(entry->key, entry->value, context)) {
                break; // enable callback function terminal the iter
            }
        }
    }
}