apfl/src/hashmap.c

#include <assert.h>
#include <stdlib.h>
#include <string.h>

#include "hashmap.h"
#include "resizable.h"

struct bucket {
    void *keys;
    void *values;
    size_t len;
    size_t cap;
};

struct apfl_hashmap_struct {
    struct apfl_hashmap_callbacks callbacks;
    size_t keysize;
    size_t valsize;
    size_t nbuckets;
    struct bucket *buckets;
};

struct apfl_hashmap_cursor_struct {
    apfl_hashmap map;
    size_t bucket;
    size_t i;
};

#define FNV_PRIME 1099511628211U

apfl_hash
apfl_hash_fnv1a_add(const void *data, size_t len, apfl_hash hash)
{
    for (size_t i = 0; i < len; i++) {
        uint8_t byte = ((uint8_t *)data)[i];
        hash ^= byte;
        hash *= FNV_PRIME;
    }
    return hash;
}

apfl_hash
apfl_hash_fnv1a(const void *data, size_t len)
{
    return apfl_hash_fnv1a_add(data, len, APFL_HASH_FNV1A_INIT);
}

#define HAS_CALLBACK(map, cb) ((map)->callbacks.cb != NULL)
#define INVOKE_CALLBACK(map, cb, ...) (map)->callbacks.cb((map)->callbacks.opaque, __VA_ARGS__)

static bool
keys_eq(apfl_hashmap map, const void *a, const void *b)
{
    if (HAS_CALLBACK(map, keys_eq)) {
        return INVOKE_CALLBACK(map, keys_eq, a, b);
    } else {
        return memcmp(a, b, map->keysize) == 0;
    }
}

static apfl_hash
calc_hash(apfl_hashmap map, const void *key)
{
    if (HAS_CALLBACK(map, calc_hash)) {
        return INVOKE_CALLBACK(map, calc_hash, key);
    } else {
        return apfl_hash_fnv1a(key, map->keysize);
    }
}

static void
destroy_key(apfl_hashmap map, void *key)
{
    if (HAS_CALLBACK(map, destroy_key)) {
        INVOKE_CALLBACK(map, destroy_key, key);
    }
}

static void
destroy_value(apfl_hashmap map, void *value)
{
    if (HAS_CALLBACK(map, destroy_value)) {
        INVOKE_CALLBACK(map, destroy_value, value);
    }
}

static bool
copy_key(apfl_hashmap map, void *dest, const void *src)
{
    if (HAS_CALLBACK(map, copy_key)) {
        return INVOKE_CALLBACK(map, copy_key, dest, src);
    } else {
        memcpy(dest, src, map->keysize);
        return true;
    }
}

static bool
copy_value(apfl_hashmap map, void *dest, const void *src)
{
    if (HAS_CALLBACK(map, copy_value)) {
        return INVOKE_CALLBACK(map, copy_value, dest, src);
    } else {
        memcpy(dest, src, map->valsize);
        return true;
    }
}

#define CAP_GROW 5
static_assert(CAP_GROW >= 1, "CAP_GROW must be at least 1");

static size_t
calc_new_cap(size_t old_cap)
{
    return old_cap + CAP_GROW;
}

#define KVADDR(base, elemsize, off) (((char*)(base)) + ((elemsize)*(off)))

static bool
find_key_in_bucket(apfl_hashmap map, struct bucket *bucket, const void *key, size_t *off)
{
    size_t keysize = map->keysize;

    for (size_t i = 0; i < bucket->len; i++) {
        if (keys_eq(map, key, KVADDR(bucket->keys, keysize, i))) {
            *off = i;
            return true;
        }
    }

    return false;
}

static bool
set_in_bucket(apfl_hashmap map, struct bucket *bucket, const void *key, const void *value)
{
    size_t keysize = map->keysize;
    size_t valsize = map->valsize;

    size_t i;
    if (find_key_in_bucket(map, bucket, key, &i)) {
        void *dest = KVADDR(bucket->values, valsize, i);
        destroy_value(map, dest);

        return copy_value(map, dest, value);
    }

    if (bucket->len <= bucket->cap) {
        size_t new_cap = calc_new_cap(bucket->cap);

        void *newmem;

        newmem = realloc(bucket->keys, new_cap * keysize);
        if (newmem == NULL) {
            return false;
        }
        bucket->keys = newmem;

        newmem = realloc(bucket->values, new_cap * valsize);
        if (newmem == NULL) {
            return false;
        }
        bucket->values = newmem;

        bucket->cap = new_cap;
    }

    if (!copy_key(map, KVADDR(bucket->keys, keysize, bucket->len), key)) {
        return false;
    }

    if (!copy_value(map, KVADDR(bucket->values, valsize, bucket->len), value)) {
        destroy_key(map, KVADDR(bucket->keys, keysize, bucket->len));
        return false;
    }

    bucket->len++;

    return true;
}

static bool
get_in_bucket(apfl_hashmap map, struct bucket *bucket, const void *key, void *value)
{
    size_t i;
    if (!find_key_in_bucket(map, bucket, key, &i)) {
        return false;
    }

    if (value != NULL) {
        size_t valsize = map->valsize;
        if (!copy_value(map, value, KVADDR(bucket->values, valsize, i))) {
            return false; // TODO: This way, we cant distinguish an error in copy_value from a non-set key
        }
    }

    return true;
}

static struct bucket *
bucket_by_key(apfl_hashmap map, const void *key)
{
    apfl_hash hash = calc_hash(map, key);
    return &map->buckets[hash % map->nbuckets];
}

static void
delete_in_bucket(apfl_hashmap map, struct bucket *bucket, const void *key)
{
    size_t i;
    if (!find_key_in_bucket(map, bucket, key, &i)) {
        return;
    }

    size_t keysize = map->keysize;
    size_t valsize = map->valsize;

    destroy_key(map, KVADDR(bucket->keys, keysize, i));
    destroy_value(map, KVADDR(bucket->values, valsize, i));

    assert(bucket->len >= (i+1));
    memmove(
        KVADDR(bucket->keys, keysize, i),
        KVADDR(bucket->keys, keysize, i+1),
        (bucket->len - (i+1)) * keysize
    );
    memmove(
        KVADDR(bucket->values, valsize, i),
        KVADDR(bucket->values, valsize, i+1),
        (bucket->len - (i+1)) * valsize
    );

    assert(bucket->len > 0); // if len == 0, we would not have found an entry
    bucket->len--;
}

#define INITIAL_NBUCKETS 16 // Must be a power of 2

apfl_hashmap
apfl_hashmap_new(struct apfl_hashmap_callbacks callbacks, size_t keysize, size_t valsize)
{
    apfl_hashmap map = malloc(sizeof(struct apfl_hashmap_struct));
    if (map == NULL) {
        goto fail;
    }

    map->callbacks = callbacks;
    map->keysize = keysize;
    map->valsize = valsize;
    map->nbuckets = INITIAL_NBUCKETS;
    map->buckets = malloc(sizeof(struct bucket) * INITIAL_NBUCKETS);
    if (map->buckets == NULL) {
        goto fail;
    }

    for (size_t i = 0; i < INITIAL_NBUCKETS; i++) {
        map->buckets[i] = (struct bucket) {
            .keys   = NULL,
            .values = NULL,
            .len    = 0,
            .cap    = 0,
        };
    }

    return map;

fail:
    free(map);
    return NULL;
}

void
apfl_hashmap_delete(apfl_hashmap map, const void *key)
{
    delete_in_bucket(map, bucket_by_key(map, key), key);
}

bool
apfl_hashmap_get(apfl_hashmap map, const void *key, void *value)
{
    return get_in_bucket(map, bucket_by_key(map, key), key, value);
}

bool
apfl_hashmap_set(apfl_hashmap map, const void *key, const void *value)
{
    return set_in_bucket(map, bucket_by_key(map, key), key, value);
}

static void
destroy_bucket(apfl_hashmap map, struct bucket *bucket)
{
    for (size_t i = 0; i < bucket->len; i++) {
        destroy_key(map, KVADDR(bucket->keys, map->keysize, i));
        destroy_value(map, KVADDR(bucket->values, map->valsize, i));
    }
    free(bucket->keys);
    free(bucket->values);
    bucket->len = 0;
    bucket->cap = 0;
}

void
apfl_hashmap_destroy(apfl_hashmap map)
{
    if (map == NULL) {
        return;
    }

    if (map->buckets != NULL) {
        for (size_t i = 0; i < map->nbuckets; i++) {
            destroy_bucket(map, &map->buckets[i]);
        }
        free(map->buckets);
    }

    free(map);
}

static void
cursor_skip_empty_buckets(apfl_hashmap_cursor cur)
{
    apfl_hashmap map = cur->map;
    while (cur->bucket < map->nbuckets && map->buckets[cur->bucket].len == 0) {
        cur->bucket++;
    }
}

apfl_hashmap_cursor
apfl_hashmap_get_cursor(apfl_hashmap map)
{
    apfl_hashmap_cursor cursor = malloc(sizeof(struct apfl_hashmap_cursor_struct));
    if (cursor != NULL) {
        cursor->map = map;
        cursor->i = 0;
        cursor->bucket = 0;

        cursor_skip_empty_buckets(cursor);
    }
    return cursor;
}

bool
apfl_hashmap_cursor_is_end(apfl_hashmap_cursor cursor)
{
    return cursor->bucket >= cursor->map->nbuckets;
}

static struct bucket *
cursor_get_bucket(apfl_hashmap_cursor cursor)
{
    return apfl_hashmap_cursor_is_end(cursor)
        ? NULL
        : &cursor->map->buckets[cursor->bucket];
}

void
apfl_hashmap_cursor_next(apfl_hashmap_cursor cursor)
{
    struct bucket *bucket = cursor_get_bucket(cursor);

    if (bucket == NULL) {
        return; // End already reached
    }

    cursor->i++;
    if (cursor->i < bucket->len) {
        return;
    }

    cursor->bucket++;
    cursor->i = 0;
    cursor_skip_empty_buckets(cursor);
}

#define CURSOR_GET(cursor, out, bucketmemb, sizememb, copy) \
    struct bucket *bucket = cursor_get_bucket(cursor);      \
                                                            \
    if (bucket == NULL) {                                   \
        return false; /* End already reached */             \
    }                                                       \
                                                            \
    if (cursor->i >= bucket->len) {                         \
        return false;                                       \
    }                                                       \
                                                            \
    size_t size = cursor->map->sizememb;                    \
                                                            \
    return copy(                                            \
        cursor->map,                                        \
        out,                                                \
        KVADDR(bucket->bucketmemb, size, bucket->len)       \
    );                                                      \

bool
apfl_hashmap_cursor_get_key(apfl_hashmap_cursor cursor, void *key)
{
    CURSOR_GET(cursor, key, keys, keysize, copy_key)
}

bool
apfl_hashmap_cursor_get_value(apfl_hashmap_cursor cursor, void *value)
{
    CURSOR_GET(cursor, value, values, valsize, copy_value)
}