2026-10-18 02:06:11 +02:00

414 lines
9.0 KiB
C

#include "common.h"
#include <assert.h>
#include <ctype.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
static bool ch_is_punctuation(char ch)
{
return strchr(",.!?'", ch);
}
static bool ch_is_letters(char ch)
{
return isalpha(ch);
}
size_t next_word(char line[], size_t line_size, FILE *file, int *ch)
{
size_t len = 0;
if (!*ch)
*ch = fgetc(file);
if (*ch != EOF && !ch_is_punctuation(*ch) && !ch_is_letters(*ch)) {
assert(len < line_size - 1);
line[len] = ' ';
len += 1;
int newlines = 0;
while (*ch != EOF && !ch_is_punctuation(*ch) && !ch_is_letters(*ch)) {
newlines += *ch == '\n';
*ch = fgetc(file);
}
if (newlines >= 2) {
line[len - 1] = '\n';
return len;
}
}
bool matched = false;
while (*ch != EOF && ch_is_punctuation(*ch)) {
matched = true;
assert(len < line_size - 1);
line[len] = *ch;
len += 1;
*ch = fgetc(file);
}
if (matched)
return len;
while (*ch != EOF && ch_is_letters(*ch)) {
matched = true;
assert(len < line_size - 1);
line[len] = tolower(*ch);
len += 1;
*ch = fgetc(file);
}
return len;
}
void array_push(
void **data, size_t *capacity, size_t *size, void *elem, size_t elem_size)
{
if (!*capacity) {
*capacity = 4;
}
if (!*data) {
*data = malloc(*capacity * elem_size);
*size = 0;
}
if (*size >= *capacity) {
*capacity *= 2;
*data = realloc(*data, *capacity * elem_size);
}
memcpy((unsigned char *)*data + *size * elem_size, elem, elem_size);
*size += 1;
}
void array_push_n(void **data,
size_t *capacity,
size_t *size,
void *elems,
size_t elem_size,
size_t elem_count)
{
for (size_t i = 0; i < elem_count; ++i) {
array_push(data,
capacity,
size,
(unsigned char *)elems + elem_size * i,
elem_size);
}
}
typedef int (*CmpFn)(const void *, const void *);
static void *binary_search_internal(void *data,
size_t begin,
size_t end,
const void *entry,
size_t entry_size,
size_t *out_idx,
CmpFn cmp)
{
if (begin == end) {
if (out_idx)
*out_idx = begin;
return NULL;
}
size_t middle_idx = (end - begin) / 2 + begin;
void *middle_entry = (unsigned char *)data + entry_size * middle_idx;
int cmp_result = cmp(entry, middle_entry);
if (cmp_result == 0) {
if (out_idx)
*out_idx = middle_idx;
return middle_entry;
}
if (cmp_result < 0)
end = middle_idx;
else
begin = middle_idx + 1;
return binary_search_internal(
data, begin, end, entry, entry_size, out_idx, cmp);
}
void *binary_search(void *data,
size_t count,
const void *entry,
size_t entry_size,
size_t *out_idx,
CmpFn cmp)
{
return binary_search_internal(
data, 0, count, entry, entry_size, out_idx, cmp);
}
typedef struct RbTree RbTree;
RbTree *rbtree_new(size_t key_size, size_t val_size, CmpFn cmp);
void rbtree_free(RbTree *t);
void rbtree_insert(RbTree *t, const void *key, void *val);
void *rbtree_get(RbTree *t, const void *key);
typedef enum { Rb_Black, Rb_Red } RbColor;
typedef enum { Rb_Left, Rb_Right } RbDir;
typedef struct RbNode RbNode;
struct RbNode {
RbNode *parent;
union {
struct {
RbNode *left;
RbNode *right;
};
RbNode *child[2];
};
uint8_t color;
int8_t padding[7];
uint8_t key_and_val[];
};
struct RbTree {
size_t key_size;
size_t val_size;
CmpFn cmp;
size_t node_key_size;
size_t node_val_size;
size_t size_of_node;
RbNode *root;
};
RbTree *rbtree_new(size_t key_size, size_t val_size, CmpFn cmp)
{
size_t node_key_size = (key_size + 7ul) & ~7ul;
size_t node_val_size = (key_size + 7ul) & ~7ul;
size_t size_of_node = sizeof(RbNode) + node_key_size + node_val_size;
RbTree *t = malloc(sizeof(RbTree));
*t = (RbTree) {
.key_size = key_size,
.val_size = val_size,
.cmp = cmp,
.node_key_size = node_key_size,
.node_val_size = node_val_size,
.size_of_node = size_of_node,
.root = NULL,
};
return t;
}
static void t_free_node(RbNode *node)
{
if (!node)
return;
t_free_node(node->child[0]);
t_free_node(node->child[1]);
free(node);
}
void rbtree_free(RbTree *t)
{
t_free_node(t->root);
free(t);
}
static RbNode *t_find_node(RbTree *t,
RbNode *node,
const void *key,
RbNode **out_parent,
RbDir *out_dir)
{
int cmp_result = t->cmp(key, node->key_and_val);
if (cmp_result == 0)
return node;
RbDir dir = cmp_result < 0 ? Rb_Left : Rb_Right;
RbNode *next = node->child[dir];
if (next)
return t_find_node(t, next, key, out_parent, out_dir);
if (out_dir)
*out_dir = dir;
if (out_parent)
*out_parent = node;
return NULL;
}
static RbNode *t_node_new(
RbTree *t, RbNode *parent, RbColor color, const void *key, void *val)
{
RbNode *node = malloc(t->size_of_node);
*node = (RbNode) {
.parent = parent,
.child = { NULL, NULL },
.color = color,
};
memcpy(node->key_and_val, key, t->key_size);
memcpy(node->key_and_val + t->node_key_size, val, t->val_size);
return node;
}
static RbNode *t_rotate_subtree(RbTree *t, RbNode *sub, RbDir dir)
{
RbNode *sub_parent = sub->parent;
RbNode *new_root = sub->child[1 - dir];
RbNode *new_child = new_root->child[dir];
sub->child[1 - dir] = new_child;
if (new_child)
new_child->parent = sub;
new_root->child[dir] = sub;
new_root->parent = sub_parent;
sub->parent = new_root;
if (sub_parent)
sub_parent->child[sub == sub_parent->right] = new_root;
else
t->root = new_root;
return new_root;
}
void rbtree_insert(RbTree *t, const void *key, void *val)
{
// https://en.wikipedia.org/wiki/Red%E2%80%93black_tree#Insertion
RbNode *node, *parent = NULL;
RbDir dir = Rb_Left;
if (t->root && (node = t_find_node(t, t->root, key, &parent, &dir))) {
memcpy(node->key_and_val + t->node_key_size, val, t->val_size);
return;
}
node = t_node_new(t, parent, Rb_Red, key, val);
if (!t->root) {
t->root = node;
return;
}
parent->child[dir] = node;
do {
if (parent->color == Rb_Black)
// case 1
return;
RbNode *grandparent = parent->parent;
if (!grandparent) {
// case 4
parent->color = Rb_Black;
return;
}
dir = parent == grandparent->left ? Rb_Left : Rb_Right;
RbNode *uncle = grandparent->child[1 - dir];
if (!uncle || uncle->color == Rb_Black) {
if (node == parent->child[1 - dir]) {
// case 5
t_rotate_subtree(t, parent, dir);
node = parent;
parent = grandparent->child[dir];
}
// case 6
t_rotate_subtree(t, grandparent, 1 - dir);
parent->color = Rb_Black;
grandparent->color = Rb_Red;
return;
}
// case 2
parent->color = Rb_Black;
uncle->color = Rb_Black;
grandparent->color = Rb_Red;
node = grandparent;
} while ((parent = node->parent));
// case 3
}
void *rbtree_get(RbTree *t, const void *key)
{
RbNode *node;
if (!t->root || !(node = t_find_node(t, t->root, key, NULL, NULL)))
return NULL;
return node->key_and_val + t->node_key_size;
}
static bool t_red_node_children_valid(RbNode *node)
{
if (!node)
return true;
if (node->parent && node->color == Rb_Red)
return node->parent->color != Rb_Red;
return t_red_node_children_valid(node->left)
&& t_red_node_children_valid(node->right);
}
static int t_walk_count_black_nodes(RbNode *node)
{
if (!node)
return 1;
int left = t_walk_count_black_nodes(node->left);
int right = t_walk_count_black_nodes(node->right);
if (left != right)
return 0;
return left + (node->color == Rb_Black);
}
static bool rbtree_valid(RbTree *t)
{
return t_red_node_children_valid(t->root)
&& t_walk_count_black_nodes(t->root) != 0;
}
static int int_cmp(const int *a, const int *b)
{
return *a - *b;
}
void test_common(void)
{
RbTree *t = rbtree_new(sizeof(int), sizeof(int), (CmpFn)int_cmp);
int keys[] = { 3, 7, 5, 1, 4, 6 };
int vals[] = { 0, 2, 4, 6, 8, 10 };
int *result;
result = rbtree_get(t, &keys[0]);
assert(!result);
for (size_t i = 0; i < sizeof(keys) / sizeof(keys[0]); ++i) {
rbtree_insert(t, &keys[i], &vals[i]);
assert(rbtree_valid(t));
}
for (size_t i = 0; i < sizeof(keys) / sizeof(keys[0]); ++i) {
assert(*(int *)rbtree_get(t, &keys[i]) == vals[i]);
}
rbtree_free(t);
}