Files
ec3/src/b-tree.c

437 lines
11 KiB
C
Raw Normal View History

#include <blue/core/queue.h>
#include <stdlib.h>
#include <assert.h>
#include <string.h>
#include "b-tree.h"
#define SWAP(type, x, y) \
do { \
type tmp = x; \
x = y; \
y = tmp; \
} while (0)
struct cache_entry {
int c_allocated;
b_queue_entry c_entry;
/* ... node data here ... */
};
#define GET_ENTRY(tree, entries, index) \
(b_tree_node_entry *)((unsigned char *)(entries) + ((index) * (tree)->tree_entry_size));
static b_tree_node *cache_alloc_node(struct b_tree *tree)
{
b_queue_iterator it = { 0 };
b_queue_foreach (&it, &tree->tree_cache) {
struct cache_entry *e = b_unbox(struct cache_entry, it.entry, c_entry);
if (!e->c_allocated) {
e->c_allocated = 1;
return (b_tree_node *)(e + 1);
}
}
struct cache_entry *e = malloc(sizeof *e + tree->tree_node_size);
if (!e) {
return NULL;
}
memset(e, 0x0, sizeof *e);
e->c_allocated = 1;
e->c_entry = B_QUEUE_ENTRY_INIT;
b_queue_push_back(&tree->tree_cache, &e->c_entry);
return (b_tree_node *)(e + 1);
}
static void cache_free_node(struct b_tree *tree, b_tree_node *node)
{
struct cache_entry *e = (struct cache_entry *)((unsigned char *)node - sizeof *e);
e->c_allocated = 0;
}
void b_tree_init(struct b_tree *tree, const struct b_tree_ops *ops, unsigned int node_size, unsigned int entry_size, unsigned int order)
{
memset(tree, 0x0, sizeof *tree);
assert((order % 2) == 0);
tree->tree_ops = ops;
tree->tree_node_size = node_size;
tree->tree_entry_size = entry_size;
tree->tree_order = order;
}
static long tree_alloc_id(struct b_tree *tree)
{
return tree->tree_ops->tree_alloc_node(tree);
}
static int node_read(struct b_tree *tree, long id, b_tree_node *out)
{
return tree->tree_ops->tree_get_node(tree, id, out);
}
static int node_write(struct b_tree *tree, long id, const b_tree_node *in)
{
return tree->tree_ops->tree_put_node(tree, id, in);
}
static unsigned long node_get_nr_entries(struct b_tree *tree, b_tree_node *node)
{
return tree->tree_ops->node_get_nr_entries(node);
}
static void node_set_nr_entries(struct b_tree *tree, b_tree_node *node, unsigned long val)
{
tree->tree_ops->node_set_nr_entries(node, val);
}
static long node_get_child(struct b_tree *tree, b_tree_node *node, unsigned long index)
{
uint16_t *children = tree->tree_ops->node_get_children(node);
if (children[index] == 0xFFFF) {
return B_TREE_INVALID_PTR;
}
return children[index];
}
static void node_set_child(struct b_tree *tree, b_tree_node *node, unsigned long index, long ptr)
{
uint16_t *children = tree->tree_ops->node_get_children(node);
if (ptr == B_TREE_INVALID_PTR) {
children[index] = 0xFFFF;
} else {
children[index] = ptr;
}
}
static b_tree_node_entry *node_get_entry(struct b_tree *tree, b_tree_node *node, unsigned long index)
{
b_tree_node_entry *entries = tree->tree_ops->node_get_entries(node);
return GET_ENTRY(tree, entries, index);
}
static void node_set_entry(struct b_tree *tree, b_tree_node *node, unsigned long index, const b_tree_node_entry *entry)
{
b_tree_node_entry *entries = tree->tree_ops->node_get_entries(node);
b_tree_node_entry *dest = GET_ENTRY(tree, entries, index);
memcpy(dest, entry, tree->tree_entry_size);
}
static void node_kill_entry(struct b_tree *tree, b_tree_node *node, unsigned long index)
{
b_tree_node_entry *entries = tree->tree_ops->node_get_entries(node);
b_tree_node_entry *dest = GET_ENTRY(tree, entries, index);
memset(dest, 0x0, tree->tree_entry_size);
}
static void node_shift_entries(struct b_tree *tree, b_tree_node *node, unsigned long shift_from, long shift_by)
{
unsigned long nr_entries = node_get_nr_entries(tree, node);
b_tree_node_entry *entries = tree->tree_ops->node_get_entries(node);
if (shift_from >= nr_entries) {
return;
}
b_tree_node_entry *src = GET_ENTRY(tree, entries, shift_from);
b_tree_node_entry *dest = GET_ENTRY(tree, entries, shift_from + shift_by);
unsigned int count = nr_entries - shift_from;
memmove(dest, src, tree->tree_entry_size * count);
if (shift_by < 0) {
dest = GET_ENTRY(tree, entries, nr_entries + shift_by);
count = abs(shift_by);
}
else {
dest = src;
count = shift_by;
}
memset(dest, 0x0, count * tree->tree_entry_size);
return;
}
static void node_shift_children(struct b_tree *tree, b_tree_node *node, unsigned long shift_from, long shift_by)
{
unsigned long nr_entries = node_get_nr_entries(tree, node);
uint16_t *children = tree->tree_ops->node_get_children(node);
if (shift_from >= nr_entries + 1) {
return;
}
if (shift_from == nr_entries && shift_by > 0) {
return;
}
uint16_t *src = &children[shift_from];
uint16_t *dest = &children[shift_from + shift_by];
unsigned int count = nr_entries + 1 - shift_from;
if (shift_by > 0) {
count -= shift_by;
}
memmove(dest, src, sizeof(uint16_t) * count);
if (shift_by < 0) {
dest = &children[nr_entries + 1 + shift_from];
count = abs(shift_by);
}
else {
dest = src;
count = shift_by;
}
memset(dest, 0xFF, count * sizeof(uint16_t));
}
static void node_move_entries(struct b_tree *tree, b_tree_node *from, b_tree_node *to, unsigned long src_index, unsigned long dest_index, unsigned long count)
{
unsigned long from_nr_entries = node_get_nr_entries(tree, from);
unsigned long to_nr_entries = node_get_nr_entries(tree, to);
b_tree_node_entry *from_entries = tree->tree_ops->node_get_entries(from);
b_tree_node_entry *to_entries = tree->tree_ops->node_get_entries(to);
b_tree_node_entry *src = GET_ENTRY(tree, from_entries, src_index);
b_tree_node_entry *dest = GET_ENTRY(tree, to_entries, dest_index);
memmove(dest, src, count * tree->tree_entry_size);
if (src_index + count >= from_nr_entries) {
memset(src, 0x0, count * tree->tree_entry_size);
} else {
node_shift_entries(tree, from, src_index + count, (int)count * -1);
}
from_nr_entries -= count;
to_nr_entries += count;
node_set_nr_entries(tree, from, from_nr_entries);
node_set_nr_entries(tree, to, to_nr_entries);
}
static void node_move_children(struct b_tree *tree, b_tree_node *from, b_tree_node *to, unsigned long src_index, unsigned long dest_index, unsigned long count)
{
unsigned long from_nr_entries = node_get_nr_entries(tree, from);
uint16_t *from_children = tree->tree_ops->node_get_children(from);
uint16_t *to_children = tree->tree_ops->node_get_children(to);
uint16_t *src = &from_children[src_index];
uint16_t *dest = &to_children[dest_index];
memmove(dest, src, count * sizeof(uint16_t));
if (src_index + count >= from_nr_entries + 1) {
memset(src, 0xFF, count * sizeof(uint16_t));
}
else {
node_shift_children(tree, from, src_index + count, (int)count * -1);
}
}
static int node_put(struct b_tree *tree, b_tree_node *n, const b_tree_node_entry *e)
{
unsigned long nr_entries = node_get_nr_entries(tree, n);
if (nr_entries == 0) {
node_set_entry(tree, n, 0, e);
node_set_nr_entries(tree, n, 1);
return 0;
}
int insert_at = -1;
for (unsigned long i = 0; i < nr_entries; i++) {
const b_tree_node_entry *cur = node_get_entry(tree, n, i);
int cmp = entry_compare(tree, cur, e);
if (cmp == 1) {
insert_at = i;
break;
}
}
if (insert_at != -1) {
node_shift_entries(tree, n, insert_at, 1);
}
else {
insert_at = nr_entries;
}
node_set_entry(tree, n, insert_at, e);
nr_entries++;
node_set_nr_entries(tree, n, nr_entries);
return insert_at;
}
static int node_is_leaf(struct b_tree *tree, b_tree_node *node)
{
long first_child = node_get_child(tree, node, 0);
return first_child == B_TREE_INVALID_PTR;
}
static void node_init(struct b_tree *tree, b_tree_node *out)
{
memset(out, 0x0, tree->tree_node_size);
for (unsigned int i = 0; i < tree->tree_order; i++) {
node_set_child(tree, out, i, B_TREE_INVALID_PTR);
}
}
static int entry_compare(struct b_tree *tree, const b_tree_node_entry *a, const b_tree_node_entry *b)
{
return tree->tree_ops->entry_compare(a, b);
}
int b_tree_put(struct b_tree *tree, const b_tree_node_entry *to_put)
{
int depth = 0;
long current_id = tree->tree_root;
long next_id = -1;
b_tree_node *current = cache_alloc_node(tree);
b_tree_node *next = cache_alloc_node(tree);
int err = node_read(tree, current_id, current);
if (err != 0) {
return err;
}
unsigned long nr_entries = node_get_nr_entries(tree, current);
if (nr_entries == tree->tree_order - 1) {
/* root node is full. split it pre-emptively. */
b_tree_node *a = next;
b_tree_node *b = cache_alloc_node(tree);
node_init(tree, a);
node_init(tree, b);
int median = nr_entries / 2;
node_move_entries(tree, current, a, 0, 0, median);
node_move_entries(tree, current, b, 1, 0, median);
node_move_children(tree, current, a, 0, 0, median + 1);
node_move_children(tree, current, b, median + 1, 0, median + 1);
int a_index = tree_alloc_id(tree);
node_write(tree, a_index, a);
int b_index = tree_alloc_id(tree);
node_write(tree, b_index, b);
node_set_child(tree, current, 0, a_index);
node_set_child(tree, current, 1, b_index);
node_write(tree, current_id, current);
cache_free_node(tree, b);
}
bool leaf = node_is_leaf(tree, current);
while (!leaf) {
unsigned int i;
bool found_bigger_key = false;
nr_entries = node_get_nr_entries(tree, current);
next_id = node_get_child(tree, current, nr_entries);
for (i = 0; i < nr_entries; i++) {
b_tree_node_entry *entry = node_get_entry(tree, current, i);
int cmp = entry_compare(tree, entry, to_put);
if (cmp == 1) {
next_id = node_get_child(tree, current, i);
break;
}
}
err = node_read(tree, next_id, next);
if (err != 0) {
return err;
}
nr_entries = node_get_nr_entries(tree, next);
if (nr_entries < tree->tree_order - 1) {
/* swap current and next pointers. */
SWAP(b_tree_node *, current, next);
current_id = next_id;
leaf = node_is_leaf(tree, current);
depth++;
continue;
}
/* this node is full, split it. */
/* half the nodes from `next` will be moved into `n3`. the other
half will stay put. the median entry will be moved to
`current`.
`n3` will become the right sibling of `next` */
b_tree_node *n3 = cache_alloc_node(tree);
node_init(tree, n3);
bool n3_is_next = false;
nr_entries = node_get_nr_entries(tree, next);
int median = nr_entries / 2;
b_tree_node_entry *median_node = node_get_entry(tree, next, median);
int cmp = entry_compare(tree, median_node, to_put);
if (cmp == -1) {
// median++;
n3_is_next = true;
}
node_move_entries(
tree,
next,
n3,
median + 1,
0,
nr_entries - median - 1);
node_move_children(
tree,
next,
n3,
median + 1,
0,
nr_entries - median);
int index = node_put(tree, current, median_node);
node_kill_entry(tree, next, median);
nr_entries = node_get_nr_entries(tree, next);
nr_entries--;
node_set_nr_entries(tree, next, nr_entries);
node_shift_children(tree, current, index + 1, 1);
int n3_id = tree_alloc_id(tree);
node_set_child(tree, current, index + 1, n3_id);
node_write(tree, current_id, current);
node_write(tree, next_id, next);
node_write(tree, n3_id, n3);
if (n3_is_next) {
cache_free_node(tree, current);
current = n3;
current_id = n3_id;
}
else {
SWAP(b_tree_node *, current, next);
current_id = next_id;
}
depth++;
leaf = node_is_leaf(tree, current);
}
node_put(tree, current, to_put);
node_write(tree, current_id, current);
return 0;
}