scratch/proglangs/nsforth/dict.c

#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include "dict.h"
#include <stdio.h>

#define FREE_DICT_NAMES(env) 											\
	while ((env)->capacity > 0) {										\
		char *word_name = (env)->entries[--(env)->capacity].word_name;	\
		if (word_name != NULL && word_name != DELETED_ENTRY_MARK)		\
			free(word_name);											\
	}

const size_t DEFAULT_HASHMAP_SIZE = 255;
const size_t HASHMAP_REHASH_MARK = 70;
struct dict_ent *dict_last;

static size_t dict_table_fnv1a(const char *key, const size_t key_len)
{
	const size_t fnv_offset = 0xcbf29ce484222325;
	const size_t fnv_prime = 0x100000001b3;
	size_t hash = fnv_offset;

	for (size_t ki = 0; ki < key_len; ++ki) {
		hash ^= (uint8_t)key[ki];
		hash *= fnv_prime;
	}
	return hash;
}

static void dict_destroy_enviroment(struct dict_table *env)
{
	FREE_DICT_NAMES(env);
	free(env->entries);
	free(env);
}

struct dict_table *dict_new_env(struct dict_table *prev_env)
{
	struct dict_table *new_env = xcalloc(1, sizeof(*new_env));
	new_env->entries = xcalloc(DEFAULT_HASHMAP_SIZE, sizeof(struct dict_ent));
	new_env->prev = prev_env;
	new_env->capacity = DEFAULT_HASHMAP_SIZE;

	if (prev_env != NULL)
		new_env->level = prev_env->level + 1;
	return new_env;
}

void dict_destroy_env(struct dict_table *env)
{
	if (env->prev == NULL) {
		dict_destroy_enviroment(env);
		return;
	}

	while (env != NULL) {
		struct dict_table *tmp = env;
		env = env->prev;
		dict_destroy_enviroment(tmp);
	}
}

struct dict_ent *_dict_insert(
	struct dict_table *env, const char *word_name, uint8_t flags, dict_fn nat_fn,
	struct comp_word *fn_def
);
// Resizes the bucket size if needed (shrink or expand), also removes deleted entries
void dict_table_rehash(struct dict_table *env)
{
	const size_t new_capacity = env->capacity * 2;

	struct dict_table new_dict_table = {0};
	new_dict_table.entries = xcalloc(new_capacity, sizeof(struct dict_ent));
	new_dict_table.capacity = new_capacity;

	// copy elements to the new dict_table, removing deleted entries in the process
	for (size_t i = 0; i < env->capacity; ++i) {
		struct dict_ent *oldent = &env->entries[i];
		//if (oldent->word_name != NULL && oldent->word_name != DELETED_ENTRY_MARK) {
		if (IS_ENTRY(oldent)) {
			dict_insert(&new_dict_table, oldent->word_name, oldent->flags, oldent->nat_fn, oldent->fn_def);
		}
	}

	FREE_DICT_NAMES(env);
	free(env->entries); // free old buckets
	env->capacity = new_capacity;
	env->entries = new_dict_table.entries;
}

struct dict_ent *_dict_insert(
	struct dict_table *env, const char *word_name, uint8_t flags, dict_fn nat_fn,
	struct comp_word *fn_def
)
{
	if ((env->used * 100) / env->capacity >= HASHMAP_REHASH_MARK)
		dict_table_rehash(env);

	size_t hash_index = dict_table_fnv1a(word_name, strlen(word_name)) % env->capacity;

	while (env->entries[hash_index].word_name != NULL) {
		++hash_index;
		if (hash_index >= env->capacity)
			hash_index = 0;
	}

	if (env->entries[hash_index].word_name != DELETED_ENTRY_MARK)
		++env->used;

	struct dict_ent *dslot = &env->entries[hash_index];
	dslot->word_name = (char *)word_name;
	dslot->flags = flags;
	dslot->nat_fn = nat_fn;
	dslot->fn_def = fn_def;

	dict_last = dslot;
	return dslot;
}

struct dict_ent *dict_insert(
	struct dict_table *env, const char *word_name, uint8_t flags, dict_fn nat_fn,
	struct comp_word *fn_def
)
{
	return _dict_insert(env, strdup(word_name), flags, nat_fn, fn_def);
}

// like dict_lookup it searches a symbol, but limited to the current environment (scope)
struct dict_ent *dict_lookup(struct dict_table *env, const char *word_name)
{
	if (env->entries == NULL)
		return NULL;
	size_t hash_index = dict_table_fnv1a(word_name, strlen(word_name)) % env->capacity;

	while (env->entries[hash_index].word_name != NULL) {
		struct dict_ent *ent = &env->entries[hash_index];
		if (ent->word_name != DELETED_ENTRY_MARK && !strcmp(ent->word_name, word_name))
			return ent;

		++hash_index;
		if (hash_index >= env->capacity)
			hash_index = 0;
	}

	// entry not found
	return NULL;
}

// searches for a symbol in the current table and in all linked environments (scopes)
//struct dict_ent *dict_lookup(struct dict_table *env, char *word_name)
//{
//	if (env->entries == NULL)
//		return NULL;
//	size_t hash_index = dict_table_fnv1a(word_name, strlen(word_name)) % env->capacity;
//
//	for (struct dict_table *s = env; s != NULL; s = s->prev) {
//		while (s->entries[hash_index].word_name != NULL) {
//			if (s->entries[hash_index].word_name != DELETED_ENTRY_MARK && !strcmp(s->entries[hash_index].word_name, word_name))
//				return &s->entries[hash_index];
//
//			++hash_index;
//			if (hash_index >= s->capacity)
//				hash_index = 0;
//		}
//	}
//
//	// entry not found
//	return NULL;
//}