This commit is contained in:
Alessandro Mauri 2023-12-29 12:44:53 +01:00
parent 4aefe8b42d
commit 94837ed410
14 changed files with 1196 additions and 0 deletions

137
stuff/generic_hash.h Normal file
View file

@ -0,0 +1,137 @@
#ifndef _HASH_GENERIC
#define _HASH_GENERIC
#include <string.h>
#include <stdlib.h>
#include <stdint.h>
// FIXME: change the api to just one HASH_DECL to HASH_PROTO and HASH_DEFINE
#define HASH_MAXSIZE 4096
// for fibonacci hashing, 2^{32,64}/<golden ratio>
#define HASH_RATIO32 ((uint64_t)2654435769u)
#define HASH_RATIO64 ((uint64_t)11400714819322457583u)
// salt for string hashing
#define HASH_SALT ((uint64_t)0xbabb0cac)
/* Ready-made compares */
static inline int hash_cmp_u32(uint32_t a, uint32_t b) { return a == b; }
static inline int hash_cmp_u64(uint64_t a, uint64_t b) { return a == b; }
static inline int hash_cmp_str(const char *a, const char *b) { return strcmp(a, b) == 0; }
/* Ready-made hashes */
static inline uint32_t hash_u64(uint64_t c)
{
return (uint64_t)((((uint64_t)c+HASH_SALT)*HASH_RATIO64)>>32);
}
static inline uint32_t hash_u32(uint32_t c)
{
return (uint32_t)((((uint64_t)c<<31)*HASH_RATIO64)>>32);
}
static inline uint32_t hash_str(const char *s)
{
uint32_t h = HASH_SALT;
const uint8_t *v = (const uint8_t *)(s);
for (int x = *s; x; x--) {
h += v[x-1];
h += h << 10;
h ^= h >> 6;
}
h += h << 3;
h ^= h >> 11;
h += h << 15;
return h;
}
#define HASH_DECL(htname, codetype, datatype, hashfn, cmpfn) \
struct htname##_entry { \
codetype code; \
datatype data; \
}; \
\
struct htname##_ref { \
uint32_t items, size, exp; \
struct htname##_entry bucket[]; \
}; \
\
\
struct htname##_ref * htname##_create(uint32_t size) \
{ \
if (!size || size > HASH_MAXSIZE) \
return NULL; \
/* round to the greater power of two */ \
/* FIXME: check for intger overflow here */ \
uint32_t exp = 32-__builtin_clz(size-1); \
size = 1<<(exp); \
/* FIXME: check for intger overflow here */ \
struct htname##_ref *ht = malloc(sizeof(struct htname##_ref)+sizeof(struct htname##_entry)*size); \
if (ht) { \
ht->items = 0; \
ht->size = size; \
ht->exp = exp; \
memset(ht->bucket, 0, sizeof(struct htname##_entry)*size); \
} \
return ht; \
} \
\
\
void htname##_destroy(struct htname##_ref *ht) \
{ \
if (ht) free(ht); \
} \
\
\
static uint32_t htname##_lookup(struct htname##_ref *ht, uint32_t hash, uint32_t idx) \
{ \
if (!ht) return 0; \
uint32_t mask = ht->size-1; \
uint32_t step = (hash >> (32 - ht->exp)) | 1; \
return (idx + step) & mask; \
} \
\
\
/* Find and return the element by code */ \
struct htname##_entry * htname##_search(struct htname##_ref *ht, codetype code)\
{ \
if (!ht) return NULL; \
uint32_t h = hashfn(code); \
for (uint32_t i=h, x=0; ; x++) { \
i = htname##_lookup(ht, h, i); \
if (x > (ht->size<<1) || \
!ht->bucket[i].code || \
cmpfn(ht->bucket[i].code, code) \
) { \
return &(ht->bucket[i]); \
} \
} \
return NULL; \
} \
\
\
/* FIXME: this simply overrides the found item */ \
struct htname##_entry * htname##_insert(struct htname##_ref *ht, struct htname##_entry *entry) \
{ \
struct htname##_entry *r = htname##_search(ht, entry->code); \
if (r) { \
if (!r->code) \
ht->items++; \
*r = *entry; \
} \
return r; \
} \
\
\
struct htname##_entry * htname##_remove(struct htname##_ref *ht, codetype code)\
{ \
if (!ht) return NULL; \
struct htname##_entry *r = htname##_search(ht, code); \
if (r) r->code = 0; \
return r; \
} \
#endif

118
stuff/generic_stack.h Normal file
View file

@ -0,0 +1,118 @@
#ifndef _STACK_GENERIC_H
#define _STACK_GENERIC_H
#define STACK_STEP 8
#define STACK_SALT 0xbabb0cac
// FIXME: find a way to not re-hash the whole stack when removing one item
// incremental hash for every grow
#if STACK_DISABLE_HASH
#define STACK_HASH(p, s, h) {}
#else
#define STACK_HASH(p, s, h) \
{ \
unsigned char *v = (unsigned char *)(p); \
for (int x = (s); x; x--) { \
(h) += v[x-1]; \
(h) += (h) << 10; \
(h) ^= (h) >> 6; \
} \
(h) += (h) << 3; \
(h) ^= (h) >> 11; \
(h) += (h) << 15; \
}
#endif
// TODO: add a rolling hash
#define STACK_DECL(stackname, type) \
struct stackname { \
type *items; \
int size, idx, old_idx; \
unsigned int hash, old_hash; \
}; \
\
\
struct stackname stackname##_init(void) \
{ \
return (struct stackname){0, .hash = STACK_SALT}; \
} \
\
\
int stackname##_grow(struct stackname *stack, int step) \
{ \
if (!stack) \
return -1; \
stack->items = realloc(stack->items, (stack->size+step)*sizeof(type)); \
if(!stack->items) \
return -1; \
memset(&(stack->items[stack->size]), 0, step*sizeof(*(stack->items))); \
stack->size += step; \
return 0; \
} \
\
\
int stackname##_push(struct stackname *stack, type *e) \
{ \
if (!stack || !e) \
return -1; \
if (stack->idx >= stack->size) \
if (stackname##_grow(stack, STACK_STEP)) \
return -1; \
stack->items[stack->idx++] = *e; \
STACK_HASH(e, sizeof(type), stack->hash); \
return 0; \
} \
\
\
type stackname##_pop(struct stackname *stack) \
{ \
if (!stack || stack->idx == 0 || stack->size == 0) \
return (type){0}; \
stack->hash = STACK_SALT; \
STACK_HASH(stack->items, sizeof(type)*(stack->idx-1), stack->hash); \
return stack->items[stack->idx--]; \
} \
\
\
int stackname##_clear(struct stackname *stack) \
{ \
if (!stack) \
return -1; \
stack->old_idx = stack->idx; \
stack->old_hash = stack->hash; \
stack->hash = STACK_SALT; \
stack->idx = 0; \
return 0; \
} \
\
\
int stackname##_changed(struct stackname *stack) \
{ \
if (!stack) \
return -1; \
return stack->hash != stack->old_hash; \
} \
\
\
int stackname##_size_changed(struct stackname *stack) \
{ \
if (!stack) \
return -1; \
return stack->size != stack->old_idx; \
} \
\
\
int stackname##_free(struct stackname *stack) \
{ \
if (stack) { \
stackname##_clear(stack); \
if (stack->items) \
free(stack->items); \
} \
return 0; \
} \
#endif

214
stuff/main.c Executable file
View file

@ -0,0 +1,214 @@
//#!/usr/bin/tcc -run
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define HASH_SALT (0xbabb0cac)
#define HASH_RATIO64 ((unsigned long int)11400714819322457583u)
struct ntree_node
{
unsigned int id;
const char *name;
int children_no, children_size;
struct ntree_node *children, *parent;
};
static char tmp_name_buffer[16] = {"node:"};
unsigned int hash_str_to_uint(const char *s)
{
unsigned int h = HASH_SALT;
const unsigned char *v = (const unsigned char *)(s);
for (int x = *s; x; x--) {
h += v[x-1];
h += h << 10;
h ^= h >> 6;
}
h += h << 3;
h ^= h >> 11;
h += h << 15;
return h;
}
unsigned int hash_u32(unsigned int c)
{
return (unsigned int)((((unsigned long int)c<<31)*HASH_RATIO64)>>32);
}
const char * generate_new_name(void)
{
static int count = 0;
unsigned int h = hash_u32(count++);
snprintf(tmp_name_buffer+sizeof("node:"), 16-sizeof("node:"), "%x", h);
return tmp_name_buffer;
}
int tree_prune(struct ntree_node *node)
{
if (node == NULL)
return 0;
if ((node->children_no == 0 && node->children != NULL) ||
(node->children_no != 0 && node->children == NULL)) {
printf("ERR: (%x %s) Inconsistent node children", node->id, node->name);
} else for (int i = 0; i < node->children_no; i++) {
tree_prune(&(node->children[i]));
}
if (node->children != NULL) {
free(node->children);
}
free((void *)node->name);
if (node->parent != NULL)
node->parent->children_no--;
*node = (struct ntree_node){0};
return 0;
}
struct ntree_node * tree_append(struct ntree_node *parent, const char *name)
{
if (name == NULL)
return NULL;
if (strlen(name) == 0)
name = generate_new_name();
// generate the children information
unsigned int id = hash_str_to_uint(name);
char *str = malloc(strlen(name)+1);
if (str == NULL)
return NULL;
strcpy(str, name);
// grow the parent buffer if necessary
if (parent->children_no >= parent->children_size) {
struct ntree_node *temp = NULL;
temp = realloc(parent->children, (parent->children_size+1)*sizeof(struct ntree_node));
if (temp == NULL) {
free(str);
return NULL;
}
parent->children = temp;
parent->children[parent->children_size] = (struct ntree_node){0};
parent->children_size++;
}
// find an open spot for the child
struct ntree_node *child = NULL;
for (int i = 0; i < parent->children_size; i++) {
if (parent->children[i].id == 0)
child = &(parent->children[i]);
}
if (child == NULL)
return NULL;
child->name = str;
child->id = id;
child->children = NULL;
child->children_no = 0;
child->parent = parent;
parent->children_no++;
//printf("append to %s, children: %d\n", parent->name, parent->children_no);
return child;
}
struct ntree_node * tree_find_id(struct ntree_node *root, unsigned int id)
{
if (id == 0 || root == NULL)
return NULL;
if (root->id == id)
return root;
else for (int i = 0; i < root->children_size; i++) {
if (root->children[i].id != 0 && tree_find_id(&(root->children[i]), id) != NULL)
return &(root->children[i]);
}
return NULL;
}
// TODO: add a foreach_child function
struct ntree_node * tree_find(struct ntree_node *root, const char *name)
{
if (name == NULL || strlen(name) == 0 || root == NULL)
return NULL;
unsigned int id = hash_str_to_uint(name);
return tree_find_id(root, id);
}
int tree_size(struct ntree_node *root)
{
if (root == NULL)
return 0;
int count = root->children_no;
if (count > 0) {
for (int i = 0; i < root->children_size; i++) {
if (root->children[i].id != 0)
count += tree_size(&(root->children[i]));
}
}
return count;
}
static int tree_print_ind(struct ntree_node *node, int dd)
{
for (int i = 0; i < dd; i++) printf(" ");
printf("[%s]\n", node->name);
for (int i = 0; i < node->children_size; i++) {
if (node->children[i].id == 0) continue;
tree_print_ind(&(node->children[i]), dd+1);
}
return node->children_no;
}
int tree_print(struct ntree_node *root)
{
if (root == NULL)
return 1;
tree_print_ind(root, 0);
return 0;
}
int main(void)
{
char *root_name = malloc(sizeof("root"));
strcpy(root_name, "root");
struct ntree_node root = {.name = root_name};
struct ntree_node *n;
n = tree_append(&root, "node 0:0");
tree_append(n, "node 0:0:0");
tree_append(&root, "node 0:1");
tree_append(&root, "node 0:2");
printf("Number of nodes %d\n", tree_size(&root));
tree_print(&root);
tree_prune(tree_find(&root, "node 0:0"));
printf("Number of nodes %d\n", tree_size(&root));
tree_prune(&root);
printf("Number of nodes %d\n", tree_size(&root));
return 0;
}

348
stuff/vectree.h Normal file
View file

@ -0,0 +1,348 @@
#ifndef _VECTREE_H
#define _VECTREE_H
#ifdef VTREE_DTYPE
typedef struct {
int size, elements;
VTREE_DTYPE *vector;
int *refs;
} Vtree;
int vtree_init(Vtree *tree, unsigned int size);
int vtree_pack(Vtree *tree);
int vtree_resize(Vtree *tree, unsigned int newsize);
int vtree_add(Vtree *tree, VTREE_DTYPE elem, int parent);
int vtree_prune(Vtree *tree, int ref);
int vtree_subtree_size(Vtree *tree, int ref);
int vtree_children_it(Vtree *tree, int parent, int *cursor);
int vtree_level_order_it(Vtree *tree, int ref, int **queue_p, int *cursor);
int vtree_destroy(Vtree *tree);
#ifdef VTREE_IMPL
#define IS_VALID_REF(t, r) ((r) >= 0 && (r) < (t)->size)
#define REF_IS_PRESENT(t, r) ((t)->refs[r] >= 0)
int vtree_init(Vtree *tree, unsigned int size)
{
if (tree == NULL) {
return -1;
}
tree->vector = malloc(sizeof(VTREE_DTYPE) * size);
if (tree->vector == NULL) {
return -1;
}
tree->refs = malloc(sizeof(int) * size);
if (tree->refs == NULL) {
free(tree->vector);
return -1;
}
// set all refs to -1, meaning invalid (free) element
for (unsigned int i = 0; i < size; i++) {
tree->refs[i] = -1;
}
// fill vector with zeroes
memset(tree->vector, 0, size * sizeof(VTREE_DTYPE));
tree->size = size;
tree->elements = 0;
return 0;
}
int vtree_destroy(Vtree *tree)
{
if (tree == NULL) {
return -1;
}
free(tree->vector);
free(tree->refs);
return 0;
}
int vtree_pack(Vtree *tree)
{
if (tree == NULL) {
return -1;
}
// TODO: add a PACKED flag to skip this
int free_spot = -1;
for (int i = 0; i < tree->size; i++) {
if (tree->refs[i] == -1) {
free_spot = i;
continue;
}
// find a item that can be packed
if (free_spot >= 0 && tree->refs[i] >= 0) {
int old_ref = i;
// move the item
tree->vector[free_spot] = tree->vector[i];
tree->refs[free_spot] = tree->refs[i];
tree->vector[i] = (VTREE_DTYPE){0};
tree->refs[i] = -1;
// and move all references
for (int x = 0; x < tree->size; x++) {
if (tree->refs[x] == old_ref) {
tree->refs[x] = free_spot;
}
}
// mark the free spot as used
free_spot = -1;
}
}
return 0;
}
int vtree_resize(Vtree *tree, unsigned int newsize)
{
if (tree == NULL) {
return -1;
}
// return error when shrinking with too many elements
if ((int)newsize < tree->elements) {
return -1;
}
// pack the vector when shrinking to avoid data loss
if ((int)newsize < tree->size) {
//if (vtree_pack(tree) < 0) {
// return -1;
//}
// TODO: allow shrinking, since packing destroys all references
return -1;
}
VTREE_DTYPE *newvec = realloc(tree->vector, newsize * sizeof(VTREE_DTYPE));
if (newvec == NULL) {
return -1;
}
int *newrefs = realloc(tree->refs, newsize * sizeof(int));
if (newrefs == NULL) {
return -1;
}
tree->vector = newvec;
tree->refs = newrefs;
if ((int)newsize > tree->size) {
for (int i = tree->size; i < (int)newsize; i++) {
tree->vector[i] = (VTREE_DTYPE){0};
tree->refs[i] = -1;
}
}
tree->size = newsize;
return 0;
}
// add an element to the tree, return it's ref
int vtree_add(Vtree *tree, VTREE_DTYPE elem, int parent)
{
if (tree == NULL) {
return -1;
}
// invalid parent
if (!IS_VALID_REF(tree, parent)) {
return -1;
}
// no space left
if (tree->elements >= tree->size) {
return -1;
}
// check if the parent exists
// if there are no elements in the tree the first add will set the root
if (!REF_IS_PRESENT(tree, parent) && tree->elements != 0) {
return -1;
}
// get the first free spot
int free_spot = -1;
for (int i = 0; i < tree->size; i++) {
if (tree->refs[i] == -1) {
free_spot = i;
break;
}
}
if (free_spot < 0) {
return -1;
}
// finally add the element
tree->vector[free_spot] = elem;
tree->refs[free_spot] = parent;
tree->elements++;
return free_spot;
}
// prune the tree starting from the ref
// returns the number of pruned elements
int vtree_prune(Vtree *tree, int ref)
{
if (tree == NULL) {
return -1;
}
if (!IS_VALID_REF(tree, ref)) {
return -1;
}
if (!REF_IS_PRESENT(tree, ref)) {
return 0;
}
tree->vector[ref] = (VTREE_DTYPE){0};
tree->refs[ref] = -1;
int count = 1;
for (int i = 0; i < tree->size; i++) {
if (tree->refs[i] == ref) {
count += vtree_prune(tree, i);
}
}
return count;
}
// find the size of the subtree starting from ref
int vtree_subtree_size(Vtree *tree, int ref)
{
if (tree == NULL) {
return -1;
}
if (!IS_VALID_REF(tree, ref)) {
return -1;
}
if (!REF_IS_PRESENT(tree, ref)) {
return 0;
}
int count = 1;
for (int i = 0; i < tree->size; i++) {
// only root has the reference to itself
if (tree->refs[i] == ref && ref != i) {
count += vtree_subtree_size(tree, i);
}
}
return count;
}
// iterate through the first level children, use a cursor like strtok_r
int vtree_children_it(Vtree *tree, int parent, int *cursor)
{
if (tree == NULL || cursor == NULL) {
return -1;
}
// if the cursor is out of bounds then we are done for sure
if (!IS_VALID_REF(tree, *cursor)) {
return -1;
}
// same for the parent, if it's invalid it can't have children
if (!IS_VALID_REF(tree, parent) || !REF_IS_PRESENT(tree, parent)) {
return -1;
}
// find the first child, update the cursor and return the ref
for (int i = *cursor; i < tree->size; i++) {
if (tree->refs[i] == parent) {
*cursor = i + 1;
return i;
}
}
// if no children are found return -1
*cursor = -1;
return -1;
}
/* iterates trough every leaf of the subtree in the following manner
* node [x], x: visit order
* [0]
* / | \
* / [2] [3]
* [1] |
* / \ [6]
* [4] [5]
*/
int vtree_level_order_it(Vtree *tree, int ref, int **queue_p, int *cursor)
{
if (tree == NULL || queue_p == NULL || cursor == NULL) {
return -1;
}
int *queue = *queue_p;
// TODO: this could also be done when adding or removing elements
// first call, create a ref array ordered like we desire
if (queue == NULL) {
*cursor = 0;
// create a queue of invalid refs, size is the worst case
queue = malloc(sizeof(int) * tree->size);
if (queue == NULL) {
return -1;
}
for (int i = 0; i < tree->size; i++) {
queue[i] = -1;
}
*queue_p = queue;
// iterate through the queue appending found children
int pos = 0, off = 0;
do {
//printf ("ref=%d\n", ref);
for (int i = 0; i < tree->size; i++) {
if (tree->refs[i] == ref) {
queue[pos++] = i;
}
}
for (;ref == queue[off] && off < tree->size; off++);
ref = queue[off];
} while (IS_VALID_REF(tree, ref));
}
//PRINT_ARR(queue, tree->size);
//return -1;
// on successive calls just iterate through the queue until we find an
// invalid ref
int ret = queue[(*cursor)++];
if (!IS_VALID_REF(tree, ret)) {
free(queue);
}
return ret;
}
#endif // VTREE_IMPL
#endif // VTREE_DTYPE
#endif