* Copyright © 2008 Florent Bruneau
*/
-#include <sys/mman.h>
-
#include "array.h"
#include "str.h"
#include "trie.h"
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)
+#define str(trie, entry) array_ptr((trie)->c, (entry)->c_offset)
+#define key(trie, id) array_ptr((trie)->keys, array_elt((trie)->keys_offset, (id)))
+
struct trie_t {
A(trie_entry_t) entries;
A(char) c;
*/
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);
+ const char *c = str(trie, entry);
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_match(const trie_t *trie,
const trie_entry_t *entry, const char *key)
{
- return !!(strcmp(array_ptr(trie->c, entry->c_offset), key) == 0);
+ return !!(strcmp(str(trie, entry), key) == 0);
+}
+
+static inline bool trie_entry_prefix(const trie_t *trie,
+ const trie_entry_t *entry, const char *key)
+{
+ int len = entry->c_len;
+ if (len > 0 && str(trie, entry)[len -1] == '\0') {
+ --len;
+ }
+ return !!(strncmp(str(trie, entry), 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);
- const char c2 = array_elt(trie->c, child->c_offset);
+ uint32_t mid = (start + end) >> 1;
+ const trie_entry_t* child = array_ptr(trie->entries, mid);
+ const char c2 = str(trie, child)[0];
- 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;
+ 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;
- entry = array_ptr(trie->entries, trie_entry_new(trie));
+ int id = trie_entry_new(trie);
+ entry = array_ptr(trie->entries, id);
entry->c_offset = trie->c.len;
entry->c_len = len;
+#ifdef CHECK_INTEGRITY
+ for (int i = 0 ; i < len - 1 ; ++i) {
+ if (key[i] == '\0') {
+ printf("Found a '\\0' in the string of the leaf\n");
+ abort();
+ }
+ }
+ if (key[len - 1] != '\0') {
+ printf("Key does not end with a '\\0'");
+ abort();
+ }
+#endif
array_append(trie->c, key, len);
return trie->entries.len - 1;
}
-static inline void trie_entry_insert_child(trie_t *trie, trie_entry_t *entry,
- int pchild)
+static inline void trie_entry_insert_child(trie_t *trie, uint32_t id, uint32_t pchild)
{
+ trie_entry_t *entry = array_ptr(trie->entries, id);
if (entry->children_len == 0) {
entry->children_offset = pchild;
entry->children_len = 1;
}
}
-static inline void trie_entry_split(trie_t *trie, trie_entry_t *entry, int pos)
+static inline void trie_entry_split(trie_t *trie, uint32_t id, uint16_t pos)
{
trie_entry_t *child;
+ trie_entry_t *entry;
child = array_ptr(trie->entries, trie_entry_new(trie));
+ entry = array_ptr(trie->entries, id);
if (pos == 0) {
child->c_offset = entry->c_offset;
child->c_len = entry->c_len;
entry->c_offset = 0;
entry->c_len = 0;
} else {
+ assert(pos <= entry->c_len);
child->c_offset = entry->c_offset + pos;
child->c_len = entry->c_len - pos;
entry->c_len = pos;
}
-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';
+#ifdef CHECK_INTEGRITY
+ assert(strcmp(key(trie, first_key) + offset, str(trie, entry)) == 0);
+#endif
+
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) {
- const char *str = array_ptr(trie->keys, array_elt(trie->keys_offset, i));
- const char c = str[offset];
+ current = key(trie, first_key)[offset];
+ for (uint32_t i = first_key + 1 ; i < last_key ; ++i) {
+ const char *ckey = key(trie, i) + offset;
+ const char c = *ckey;
if (c != current) {
array_ensure_capacity_delta(trie->entries, 2);
if (fork_pos == 0) {
- trie_entry_split(trie, array_ptr(trie->entries, id), off_diff);
+ trie_entry_split(trie, id, off_diff);
}
- trie_entry_insert_child(trie, array_ptr(trie->entries, id),
- trie_add_leaf(trie, str + offset));
+ trie_entry_insert_child(trie, id, trie_add_leaf(trie, ckey));
forks[fork_pos++] = i;
current = c;
}
}
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);
# include "qsort.c"
}
- array_ensure_capacity(trie->entries, trie->keys.len);
- trie_compile_aux(trie, trie_add_leaf(trie, trie->keys.data),
+ array_ensure_capacity(trie->entries, trie->keys_offset.len);
+ trie_compile_aux(trie, trie_add_leaf(trie, key(trie, 0)),
0, trie->keys_offset.len, 0, 0);
trie_cleanup_build_data(trie);
array_adjust(trie->entries);
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;
+ }
+ } else {
+ return false;
+ }
+ }
+ }
+}
+
+bool trie_prefix(const trie_t *trie, const char *key)
+{
+ assert(trie->keys.len == 0L && "Can't lookup: trie not compiled");
+ if (trie->entries.len == 0) {
+ return false;
+ } else {
+ const trie_entry_t *current = array_ptr(trie->entries, 0);
+ while (true) {
+ if (trie_entry_is_leaf(current)) {
+ return trie_entry_prefix(trie, current, key);
+ } 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->locked) {
return;
}
- if (mlock(trie->entries.data, sizeof(trie_entry_t) * trie->entries.len) != 0) {
+ if (!array_lock(trie->entries)) {
UNIXERR("mlock");
- return;
}
- if (mlock(trie->c.data, trie->c.len) != 0) {
+ if (!array_lock(trie->c)) {
+ UNIXERR("mlock");
+ }
+ if (mlock(trie, sizeof(trie_t)) != 0) {
UNIXERR("mlock");
- munlock(trie->entries.data, sizeof(trie_entry_t) * trie->entries.len);
return;
}
trie->locked = true;
if (!trie->locked) {
return;
}
- munlock(trie->entries.data, sizeof(trie_entry_t) * trie->entries.len);
- munlock(trie->c.data, trie->c.len);
+ array_unlock(trie->entries);
+ array_unlock(trie->c);
+ munlock(trie, sizeof(trie_t));
trie->locked = 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(" ", stdout);
}
if (entry->c_len == 0) {
- fputs("(nil)", stdout);
+ fputs("(0)", stdout);
} else {
const char *c = array_ptr(trie->c, entry->c_offset);
printf("(%d) ", entry->c_len);
}
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));
}
}
projects / apps / pfixtools.git / blobdiff