typedef struct trie_entry_t trie_entry_t;
struct trie_entry_t {
- int32_t c_offset;
- int32_t c_len;
+ uint32_t c_offset;
+ uint32_t children_offset;
- int32_t children_offset;
- int32_t children_len;
+ uint16_t c_len;
+ uint16_t children_len;
};
#define TRIE_ENTRY_INIT { 0, 0, 0, 0 }
ARRAY(trie_entry_t)
*/
static inline bool trie_entry_c_match(const trie_t *trie,
const trie_entry_t *entry,
- const char *key, int *pos)
+ const char *key)
{
const char *c = array_ptr(trie->c, entry->c_offset);
int i = 0;
for (i = 0 ; i < entry->c_len ; ++i) {
if (key[i] != c[i]) {
- if (pos) {
- *pos = i;
- }
return false;
}
}
- if (pos) {
- *pos = i;
- }
return true;
}
static inline bool trie_entry_prefix(const trie_t *trie,
const trie_entry_t *entry, const char *key)
{
- return !!(strncmp(array_ptr(trie->c, entry->c_offset), key, entry->c_len) == 0);
+ int len = entry->c_len;
+ if (len > 0 && array_elt(trie->c, entry->c_offset + len - 1) == '\0') {
+ --len;
+ }
+ return !!(strncmp(array_ptr(trie->c, entry->c_offset), key, len) == 0);
}
static inline bool trie_entry_is_leaf(const trie_entry_t *entry)
* Only the first character of the children is taken into account in the
* lookup. The current entry is assumed to match the key.
*/
-static inline trie_entry_t* trie_entry_child(const trie_t *trie,
- const trie_entry_t* entry,
- const char *key)
+static inline const trie_entry_t* trie_entry_child(const trie_t *trie,
+ const trie_entry_t* entry,
+ const char *key)
{
- int start = entry->children_offset;
- int end = start + entry->children_len;
+ uint32_t start = entry->children_offset;
+ uint32_t end = start + entry->children_len;
const char c = *key;
while (start < end) {
- int mid = (start + end) / 2;
- trie_entry_t* child = array_ptr(trie->entries, mid);
+ uint32_t mid = (start + end) >> 1;
+ const trie_entry_t* child = array_ptr(trie->entries, mid);
const char c2 = array_elt(trie->c, child->c_offset);
- if (child->c_len) {
- if (c2 == c) {
- return child;
- }
- if (c < c2) {
- end = mid;
- } else {
- start = mid + 1;
- }
+ if (c2 == c) {
+ return child;
+ }
+ if (c < c2) {
+ end = mid;
} else {
- abort();
+ start = mid + 1;
}
}
return NULL;
}
-static inline int trie_entry_new(trie_t *trie)
+static inline uint32_t trie_entry_new(trie_t *trie)
{
const trie_entry_t e = TRIE_ENTRY_INIT;
array_add(trie->entries, e);
return trie->entries.len - 1;
}
-static inline int trie_add_leaf(trie_t *trie, const char *key)
+static inline uint32_t trie_add_leaf(trie_t *trie, const char *key)
{
trie_entry_t *entry;
int len = m_strlen(key) + 1;
}
static inline void trie_entry_insert_child(trie_t *trie, trie_entry_t *entry,
- int pchild)
+ uint32_t pchild)
{
if (entry->children_len == 0) {
entry->children_offset = pchild;
}
}
-static inline void trie_entry_split(trie_t *trie, trie_entry_t *entry, int pos)
+static inline void trie_entry_split(trie_t *trie, trie_entry_t *entry, uint16_t pos)
{
trie_entry_t *child;
child = array_ptr(trie->entries, trie_entry_new(trie));
}
-static inline void trie_compile_aux(trie_t *trie, int id,
- int first_key, int last_key, int offset,
- int initial_diff)
+static inline void trie_compile_aux(trie_t *trie, uint32_t id,
+ uint32_t first_key, uint32_t last_key,
+ int offset, int initial_diff)
{
- int forks[256];
- int fork_pos = 0;
+ uint32_t forks[256];
+ uint32_t fork_pos = 0;
char current = '\0';
for (int off_diff = initial_diff ; fork_pos == 0 ; ++off_diff, ++offset) {
current = array_elt(trie->keys, array_elt(trie->keys_offset, first_key) + offset);
- for (int i = first_key + 1 ; i < last_key ; ++i) {
+ for (uint32_t i = first_key + 1 ; i < last_key ; ++i) {
const char *str = array_ptr(trie->keys, array_elt(trie->keys_offset, i));
const char c = str[offset];
if (c != current) {
}
forks[fork_pos] = last_key;
- const int children_len = array_elt(trie->entries, id).children_len;
- for (int i = 0 ; i < children_len ; ++i) {
+ const uint8_t children_len = array_elt(trie->entries, id).children_len;
+ for (uint16_t i = 0 ; i < children_len ; ++i) {
int child = array_elt(trie->entries, id).children_offset + i;
if (forks[i] - 1 > first_key) {
trie_compile_aux(trie, child, first_key, forks[i], offset, 1);
if (trie->entries.len == 0) {
return false;
} else {
- trie_entry_t *current = array_ptr(trie->entries, 0);
+ const trie_entry_t *current = array_ptr(trie->entries, 0);
while (true) {
- int pos = 0;
if (trie_entry_is_leaf(current)) {
return trie_entry_match(trie, current, key);
- } else if (trie_entry_c_match(trie, current, key, &pos)) {
- key += pos;
+ } else if (trie_entry_c_match(trie, current, key)) {
+ key += current->c_len;
current = trie_entry_child(trie, current, key);
if (current == NULL) {
return false;
if (trie->entries.len == 0) {
return false;
} else {
- trie_entry_t *current = array_ptr(trie->entries, 0);
+ const trie_entry_t *current = array_ptr(trie->entries, 0);
while (true) {
- int pos = 0;
if (trie_entry_is_leaf(current)) {
return trie_entry_prefix(trie, current, key);
- } else if (trie_entry_c_match(trie, current, key, &pos)) {
- key += pos;
+ } else if (trie_entry_c_match(trie, current, key)) {
+ key += current->c_len;
current = trie_entry_child(trie, current, key);
if (current == NULL) {
return false;
static int leaves = 0;
static int depth_sum = 0;
+ if (entry == array_ptr(trie->entries, 0)) {
+ max_depth = 0;
+ leaves = 0;
+ depth_sum = 0;
+ }
if (trie_entry_is_leaf(entry)) {
if (level > max_depth) {
max_depth = level;
}
fputs("\n", stdout);
}
- for (int i = entry->children_offset ;
+ for (uint32_t i = entry->children_offset ;
i < entry->children_offset + entry->children_len ; ++i) {
trie_entry_inspect(trie, show_content, array_ptr(trie->entries, i), level + 1);
}
printf("Number of leaves: %d\n", leaves);
printf("Max depth: %d\n", max_depth);
printf("Average leaf depth: %d\n", depth_sum / leaves);
- printf("Memory used: %d\n", (trie->entries.size * sizeof(trie_entry_t))
+ printf("Memory used: %zd\n", (trie->entries.size * sizeof(trie_entry_t))
+ (trie->c.size) + sizeof(trie_t));
}
}