void test_ahtable_find_prev()
{
fprintf(stderr, "finding prev for %zu keys ... \n", k);
ahtable_build_index(T);
ahtable_iter_t i;
ahtable_iter_begin(T, &i, true);
value_t* u;
const char *key = NULL;
char *dkey = NULL;
size_t len = 0;
while (!ahtable_iter_finished(&i)) {
u = ahtable_iter_val(&i);
key = ahtable_iter_key(&i, &len);
/* increase key last byte by 1 and check result */
dkey = realloc(dkey, len); memcpy(dkey, key, len);
++dkey[len-1];
value_t *fp = NULL;
int r = ahtable_find_leq(T, dkey, len, &fp);
if (*fp != *u || r != -1) {
fprintf(stderr, "[error] ahtable_find_leq should find %lu, "
"but found prev=%lu and return -1, returned %d\n",
*u, *fp, r);
}
ahtable_iter_next(&i);
}
ahtable_iter_free(&i);
free(dkey);
fprintf(stderr, "done.\n");
}
static void hattrie_split_fill(node_ptr src, node_ptr left, node_ptr right, uint8_t split)
{
/* right should be most of the time hybrid */
/* keep or distribute keys to the new right node */
value_t* u;
const char* key;
size_t len;
ahtable_iter_t i;
ahtable_iter_begin(src.b, &i, false);
while (!ahtable_iter_finished(&i)) {
key = ahtable_iter_key(&i, &len);
u = ahtable_iter_val(&i);
assert(len > 0);
/* first char > split_point, move to the right */
if ((unsigned char) key[0] > split) {
if (src.b != right.b) {
/* insert to right (new bucket) */
if (*right.flag & NODE_TYPE_PURE_BUCKET) {
ahtable_insert(right.b, key + 1, len - 1, *u);
}
else {
ahtable_insert(right.b, key, len, *u);
}
/* transferred to right (from reused) */
if (src.b == left.b) {
ahtable_iter_del(&i);
continue;
}
} /* keep the node in right */
} else {
if (src.b != left.b) {
/* insert to left (new bucket) */
if (*left.flag & NODE_TYPE_PURE_BUCKET) {
ahtable_insert(left.b, key + 1, len - 1, *u);
}
else {
ahtable_insert(left.b, key, len, *u);
}
/* transferred to left (from reused) */
if (src.b == right.b) {
ahtable_iter_del(&i);
continue;
}
} /* keep the node in left */
}
ahtable_iter_next(&i);
}
ahtable_iter_free(&i);
}
void hattrie_walk (hattrie_t* T, const char* key, size_t len, void* user_data, hattrie_walk_cb cb) {
unsigned char* k = (unsigned char*)key;
node_ptr node = T->root;
size_t i, j;
ahtable_iter_t* it;
/* go down until a bucket is reached */
for (i = 0; i < len; i++, k++) {
if (!(*node.flag & NODE_TYPE_TRIE))
break;
node = node.t->xs[*k];
if (*node.flag & NODE_HAS_VAL) {
if (hattrie_walk_stop == cb(key, i, &node.t->val, user_data))
return;
}
}
if (i == len)
return;
assert(i);
if (*node.flag & NODE_TYPE_HYBRID_BUCKET) {
i--;
k--;
} else {
assert(*node.flag & NODE_TYPE_PURE_BUCKET);
}
/* dict order ensured short => long */
it = ahtable_iter_begin(node.b, true);
for(; !ahtable_iter_finished(it); ahtable_iter_next(it)) {
size_t stored_len;
unsigned char* stored_key = (unsigned char*)ahtable_iter_key(it, &stored_len);
int matched = 1;
if (stored_len + i > len) {
continue;
}
for (j = 0; j < stored_len; j++) {
if (stored_key[j] != k[j]) {
matched = 0;
break;
}
}
if (matched) {
value_t* val = ahtable_iter_val(it);
if (hattrie_walk_stop == cb(key, i + stored_len, val, user_data)) {
ahtable_iter_free(it);
return;
}
}
}
ahtable_iter_free(it);
}
void test_ahtable_sorted_iteration()
{
fprintf(stderr, "iterating in order through %zu keys ... \n", k);
ahtable_iter_t i;
ahtable_iter_begin(T, &i, true);
size_t count = 0;
value_t* u;
value_t v;
char* prev_key = malloc(m_high + 1);
size_t prev_len = 0;
const char *key = NULL;
size_t len = 0;
while (!ahtable_iter_finished(&i)) {
memcpy(prev_key, key, len);
prev_len = len;
++count;
key = ahtable_iter_key(&i, &len);
if (prev_key != NULL && cmpkey(prev_key, prev_len, key, len) > 0) {
fprintf(stderr, "[error] iteration is not correctly ordered.\n");
}
u = ahtable_iter_val(&i);
v = str_map_get(M, key, len);
if (*u != v) {
if (v == 0) {
fprintf(stderr, "[error] incorrect iteration (%lu, %lu)\n", *u, v);
}
else {
fprintf(stderr, "[error] incorrect iteration tally (%lu, %lu)\n", *u, v);
}
}
// this way we will see an error if the same key is iterated through
// twice
str_map_set(M, key, len, 0);
ahtable_iter_next(&i);
}
ahtable_iter_free(&i);
free(prev_key);
fprintf(stderr, "done.\n");
}
void test_ahtable_iteration()
{
fprintf(stderr, "iterating through %zu keys ... \n", k);
ahtable_iter_t i;
ahtable_iter_begin(T, &i, false);
size_t count = 0;
value_t* u;
value_t v;
size_t len;
const char* key;
while (!ahtable_iter_finished(&i)) {
++count;
key = ahtable_iter_key(&i, &len);
u = ahtable_iter_val(&i);
v = str_map_get(M, key, len);
if (*u != v) {
if (v == 0) {
fprintf(stderr, "[error] incorrect iteration (%lu, %lu)\n", *u, v);
}
else {
fprintf(stderr, "[error] incorrect iteration tally (%lu, %lu)\n", *u, v);
}
}
// this way we will see an error if the same key is iterated through
// twice
str_map_set(M, key, len, 0);
ahtable_iter_next(&i);
}
if (count != M->m) {
fprintf(stderr, "[error] iterated through %zu element, expected %zu\n",
count, M->m);
}
ahtable_iter_free(&i);
fprintf(stderr, "done.\n");
}
static bool hattrie_iter_prefix_not_match(hattrie_iter_t* i)
{
if (hattrie_iter_finished(i)) {
return false; // can not advance the iter
}
if (i->level >= i->prefix_len) {
return memcmp(i->key, i->prefix, i->prefix_len);
} else if (i->has_nil_key) {
return true; // subkey too short
}
size_t sublen;
const char* subkey;
subkey = ahtable_iter_key(i->i, &sublen);
if (i->level + sublen < i->prefix_len) {
return true; // subkey too short
}
return memcmp(i->key, i->prefix, i->level) ||
memcmp(subkey, i->prefix + i->level, (i->prefix_len - i->level));
}
int hattrie_split_mid(node_ptr node, unsigned *left_m, unsigned *right_m)
{
/* count the number of occourances of every leading character */
unsigned int cs[NODE_CHILDS]; // occurance count for leading chars
memset(cs, 0, NODE_CHILDS * sizeof(unsigned int));
size_t len;
const char* key;
/*! \todo expensive, maybe some heuristics or precalc would be better */
ahtable_iter_t i;
ahtable_iter_begin(node.b, &i, false);
while (!ahtable_iter_finished(&i)) {
key = ahtable_iter_key(&i, &len);
assert(len > 0);
cs[(unsigned char) key[0]] += 1;
ahtable_iter_next(&i);
}
ahtable_iter_free(&i);
/* choose a split point */
unsigned int all_m;
unsigned char j = node.b->c0;
all_m = ahtable_size(node.b);
*left_m = cs[j];
*right_m = all_m - *left_m;
int d;
while (j + 1 < node.b->c1) {
d = abs((int) (*left_m + cs[j + 1]) - (int) (*right_m - cs[j + 1]));
if (d <= abs(*left_m - *right_m) && *left_m + cs[j + 1] < all_m) {
j += 1;
*left_m += cs[j];
*right_m -= cs[j];
}
else break;
}
return j;
}
const char* hattrie_iter_key(hattrie_iter_t* i, size_t* len)
{
if (hattrie_iter_finished(i)) return NULL;
size_t sublen;
const char* subkey;
if (i->has_nil_key) {
subkey = NULL;
sublen = 0;
}
else subkey = ahtable_iter_key(i->i, &sublen);
if (i->keysize < i->level + sublen + 1) {
while (i->keysize < i->level + sublen + 1) i->keysize *= 2;
i->key = realloc_or_die(i->key, i->keysize * sizeof(char));
}
memcpy(i->key + i->level, subkey, sublen);
i->key[i->level + sublen] = '\0';
*len = i->level + sublen;
return i->key;
}
/* Perform one split operation on the given node with the given parent.
*/
static void hattrie_split(hattrie_t* T, node_ptr parent, node_ptr node)
{
/* only buckets may be split */
assert(*node.flag & NODE_TYPE_PURE_BUCKET ||
*node.flag & NODE_TYPE_HYBRID_BUCKET);
assert(*parent.flag & NODE_TYPE_TRIE);
if (*node.flag & NODE_TYPE_PURE_BUCKET) {
/* turn the pure bucket into a hybrid bucket */
parent.t->xs[node.b->c0].t = alloc_trie_node(T, node);
/* if the bucket had an empty key, move it to the new trie node */
value_t* val = ahtable_tryget(node.b, NULL, 0);
if (val) {
parent.t->xs[node.b->c0].t->val = *val;
parent.t->xs[node.b->c0].t->flag |= NODE_HAS_VAL;
*val = 0;
ahtable_del(node.b, NULL, 0);
}
node.b->c0 = 0x00;
node.b->c1 = NODE_MAXCHAR;
node.b->flag = NODE_TYPE_HYBRID_BUCKET;
return;
}
/* This is a hybrid bucket. Perform a proper split. */
/* count the number of occourances of every leading character */
unsigned int cs[NODE_CHILDS]; // occurance count for leading chars
memset(cs, 0, NODE_CHILDS * sizeof(unsigned int));
size_t len;
const char* key;
ahtable_iter_t* i = ahtable_iter_begin(node.b, false);
while (!ahtable_iter_finished(i)) {
key = ahtable_iter_key(i, &len);
assert(len > 0);
cs[(unsigned char) key[0]] += 1;
ahtable_iter_next(i);
}
ahtable_iter_free(i);
/* choose a split point */
unsigned int left_m, right_m, all_m;
unsigned char j = node.b->c0;
all_m = ahtable_size(node.b);
left_m = cs[j];
right_m = all_m - left_m;
int d;
while (j + 1 < node.b->c1) {
d = abs((int) (left_m + cs[j + 1]) - (int) (right_m - cs[j + 1]));
if (d <= abs(left_m - right_m) && left_m + cs[j + 1] < all_m) {
j += 1;
left_m += cs[j];
right_m -= cs[j];
}
else break;
}
/* now split into two node cooresponding to ranges [0, j] and
* [j + 1, NODE_MAXCHAR], respectively. */
/* create new left and right nodes */
/* TODO: Add a special case if either node is a hybrid bucket containing all
* the keys. In such a case, do not build a new table, just use the old one.
* */
size_t num_slots;
for (num_slots = ahtable_initial_size;
(double) left_m > ahtable_max_load_factor * (double) num_slots;
num_slots *= 2);
node_ptr left, right;
left.b = ahtable_create_n(num_slots);
left.b->c0 = node.b->c0;
left.b->c1 = j;
left.b->flag = left.b->c0 == left.b->c1 ?
NODE_TYPE_PURE_BUCKET : NODE_TYPE_HYBRID_BUCKET;
for (num_slots = ahtable_initial_size;
(double) right_m > ahtable_max_load_factor * (double) num_slots;
num_slots *= 2);
right.b = ahtable_create_n(num_slots);
right.b->c0 = j + 1;
right.b->c1 = node.b->c1;
right.b->flag = right.b->c0 == right.b->c1 ?
NODE_TYPE_PURE_BUCKET : NODE_TYPE_HYBRID_BUCKET;
/* update the parent's pointer */
unsigned int c;
for (c = node.b->c0; c <= j; ++c) parent.t->xs[c] = left;
for (; c <= node.b->c1; ++c) parent.t->xs[c] = right;
/* distribute keys to the new left or right node */
value_t* u;
value_t* v;
i = ahtable_iter_begin(node.b, false);
while (!ahtable_iter_finished(i)) {
key = ahtable_iter_key(i, &len);
u = ahtable_iter_val(i);
assert(len > 0);
/* left */
if ((unsigned char) key[0] <= j) {
if (*left.flag & NODE_TYPE_PURE_BUCKET) {
v = ahtable_get(left.b, key + 1, len - 1);
}
else {
v = ahtable_get(left.b, key, len);
}
*v = *u;
}
/* right */
else {
if (*right.flag & NODE_TYPE_PURE_BUCKET) {
v = ahtable_get(right.b, key + 1, len - 1);
}
else {
v = ahtable_get(right.b, key, len);
}
*v = *u;
}
ahtable_iter_next(i);
}
ahtable_iter_free(i);
ahtable_free(node.b);
}