* 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 {
- int c_offset;
- int c_len;
+ uint32_t c_offset;
+ uint32_t children_offset;
- int children_offset;
- int children_len;
+ uint16_t c_len;
+ uint16_t children_len;
};
+#define TRIE_ENTRY_INIT { 0, 0, 0, 0 }
+ARRAY(trie_entry_t)
struct trie_t {
- trie_entry_t *entries;
- int entries_len;
- int entries_size;
-
- char *c;
- int c_len;
- int c_size;
-
- char **keys;
- int keys_len;
- int keys_size;
+ A(trie_entry_t) entries;
+ A(char) c;
+ A(char) keys;
+ A(int) keys_offset;
bool locked;
};
static inline void trie_cleanup_build_data(trie_t *trie)
{
- for (int i = 0 ; i < trie->keys_len ; ++i) {
- p_delete(&trie->keys[i]);
- }
- p_delete(&trie->keys);
- trie->keys_len = trie->keys_size = 0;
+ array_wipe(trie->keys);
+ array_wipe(trie->keys_offset);
}
void trie_delete(trie_t **trie)
{
if (*trie) {
trie_cleanup_build_data(*trie);
- p_delete(&(*trie)->entries);
- p_delete(&(*trie)->c);
+ trie_unlock(*trie);
+ array_wipe((*trie)->entries);
+ array_wipe((*trie)->c);
p_delete(trie);
}
}
*/
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 = trie->c + entry->c_offset;
+ 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_match(const trie_t *trie,
const trie_entry_t *entry, const char *key)
{
- return !!(strcmp(trie->c + entry->c_offset, key) == 0);
+ return !!(strcmp(array_ptr(trie->c, entry->c_offset), 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 && 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 = &trie->entries[mid];
- const char c2 = trie->c[child->c_offset];
+ 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;
}
-#define GROW(Buffer, Added, Len, Size) \
- do { \
- const int required_size = (Len) + (Added); \
- int next_size = (Size); \
- if (next_size >= required_size) { \
- break; \
- } \
- do { \
- next_size = p_alloc_nr(next_size); \
- } while (next_size < required_size); \
- p_allocgrow(&(Buffer), next_size, &(Size)); \
- } while(0)
-
-static inline void trie_grow(trie_t *trie, int delta)
+static inline uint32_t trie_entry_new(trie_t *trie)
{
- GROW(trie->entries, delta, trie->entries_len, trie->entries_size);
+ const trie_entry_t e = TRIE_ENTRY_INIT;
+ array_add(trie->entries, e);
+ return trie->entries.len - 1;
}
-static inline int trie_entry_new(trie_t *trie)
-{
- memset(trie->entries + trie->entries_len, 0, sizeof(trie_entry_t));
- return trie->entries_len++;
-}
-
-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 = &trie->entries[trie_entry_new(trie)];
- GROW(trie->c, len, trie->c_len, trie->c_size);
- memcpy(trie->c + trie->c_len, key, len);
- entry->c_offset = trie->c_len;
+ entry = array_ptr(trie->entries, trie_entry_new(trie));
+ entry->c_offset = trie->c.len;
entry->c_len = len;
- trie->c_len += len;
- return trie->entries_len - 1;
+ 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)
+ uint32_t pchild)
{
if (entry->children_len == 0) {
entry->children_offset = pchild;
entry->children_len = 1;
} else {
if (entry->children_offset + entry->children_len != pchild) {
- trie_inspect(trie);
printf("Inserting child %d while offset is %d[%d]\n",
pchild, entry->children_offset, entry->children_len);
abort();
}
}
-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 = &trie->entries[trie_entry_new(trie)];
+ child = array_ptr(trie->entries, trie_entry_new(trie));
if (pos == 0) {
child->c_offset = entry->c_offset;
child->c_len = entry->c_len;
}
child->children_offset = entry->children_offset;
child->children_len = entry->children_len;
- entry->children_offset = trie->entries_len - 1;
+ entry->children_offset = trie->entries.len - 1;
entry->children_len = 1;
}
void trie_insert(trie_t *trie, const char* key)
{
- GROW(trie->keys, 1, trie->keys_len, trie->keys_size);
- trie->keys[trie->keys_len++] = strdup(key);
+ assert(trie->entries.len == 0 && "Trie already compiled");
+
+ int len = m_strlen(key) + 1;
+ array_add(trie->keys_offset, trie->keys.len);
+ array_append(trie->keys, key, len);
}
-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 = trie->keys[first_key][offset];
- for (int i = first_key + 1 ; i < last_key ; ++i) {
- if (trie->keys[i][offset] != current) {
- trie_grow(trie, 2);
+ current = array_elt(trie->keys, array_elt(trie->keys_offset, first_key) + offset);
+ 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) {
+ array_ensure_capacity_delta(trie->entries, 2);
if (fork_pos == 0) {
- trie_entry_split(trie, &trie->entries[id], off_diff);
+ trie_entry_split(trie, array_ptr(trie->entries, id), off_diff);
}
- trie_entry_insert_child(trie, &trie->entries[id],
- trie_add_leaf(trie, trie->keys[i] +
- offset));
+ trie_entry_insert_child(trie, array_ptr(trie->entries, id),
+ trie_add_leaf(trie, str + offset));
forks[fork_pos++] = i;
- current = trie->keys[i][offset];
+ current = c;
}
}
+ if (fork_pos == 0 && current == '\0') {
+ return;
+ }
}
forks[fork_pos] = last_key;
- const int children_len = trie->entries[id].children_len;
- for (int i = 0 ; i < children_len ; ++i) {
- int child = trie->entries[id].children_offset + 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);
}
}
}
-
-static inline void trie_shrink(trie_t *trie)
-{
- p_shrink(&trie->entries, trie->entries_len, &trie->entries_size);
- p_shrink(&trie->c, trie->c_len, &trie->c_size);
-}
-
-static inline void trie_lock(trie_t *trie)
-{
- if (mlock(trie->entries, sizeof(trie_entry_t) * trie->entries_len) != 0) {
- UNIXERR("mlock");
- return;
- }
- if (mlock(trie->c, trie->c_len) != 0) {
- UNIXERR("mlock");
- munlock(trie->entries, sizeof(trie_entry_t) * trie->entries_len);
- return;
- }
- trie->locked = true;
-}
-
-typedef char *str_t;
-
void trie_compile(trie_t *trie, bool memlock)
{
+ assert(trie->entries.len == 0 && "Trie already compiled");
+ assert(trie->keys.len != 0 && "Trying to compile an empty trie");
{
-# define QSORT_TYPE str_t
-# define QSORT_BASE trie->keys
-# define QSORT_NELT trie->keys_len
-# define QSORT_LT(a,b) strcmp(*a, *b) < 0
+# define QSORT_TYPE int
+# define QSORT_BASE trie->keys_offset.data
+# define QSORT_NELT trie->keys_offset.len
+# define QSORT_LT(a,b) strcmp(trie->keys.data + *a, trie->keys.data + *b) < 0
# include "qsort.c"
}
- trie_grow(trie, trie->keys_len);
- trie_compile_aux(trie, trie_add_leaf(trie, trie->keys[0]),
- 0, trie->keys_len, 0, 0);
+ array_ensure_capacity(trie->entries, trie->keys.len);
+ trie_compile_aux(trie, trie_add_leaf(trie, trie->keys.data),
+ 0, trie->keys_offset.len, 0, 0);
trie_cleanup_build_data(trie);
- trie_shrink(trie);
+ array_adjust(trie->entries);
+ array_adjust(trie->c);
if (memlock) {
trie_lock(trie);
}
bool trie_lookup(const trie_t *trie, const char *key)
{
- if (trie->entries_len == 0) {
+ assert(trie->keys.len == 0L && "Can't lookup: trie not compiled");
+ if (trie->entries.len == 0) {
return false;
} else {
- trie_entry_t *current = trie->entries;
+ 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;
}
}
+void trie_lock(trie_t *trie)
+{
+ if (trie->locked) {
+ return;
+ }
+ if (!array_lock(trie->entries)) {
+ UNIXERR("mlock");
+ }
+ if (!array_lock(trie->c)) {
+ UNIXERR("mlock");
+ }
+ if (mlock(trie, sizeof(trie_t)) != 0) {
+ UNIXERR("mlock");
+ return;
+ }
+ trie->locked = true;
+}
+
+void trie_unlock(trie_t *trie)
+{
+ if (!trie->locked) {
+ return;
+ }
+ array_unlock(trie->entries);
+ array_unlock(trie->c);
+ munlock(trie, sizeof(trie_t));
+ trie->locked = false;
+}
/* Debug {{{1
*/
-static inline void trie_entry_inspect(const trie_t *trie,
+static inline void trie_entry_inspect(const trie_t *trie, bool show_content,
const trie_entry_t *entry, int level)
{
static int max_depth = 0;
- static int leafs = 0;
+ 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;
}
- ++leafs;
+ ++leaves;
depth_sum += level;
}
- for (int i = 0 ; i < level ; ++i) {
- fputs(" ", stdout);
- }
- if (entry->c_len == 0) {
- fputs("(nil)", stdout);
- } else {
- const char *c = trie->c + entry->c_offset;
- printf("(%d) ", entry->c_len);
- for (int i = 0 ; i < entry->c_len ; ++i) {
- if (c[i]) {
- printf("%c ", c[i]);
- } else {
- fputs("\\0 ", stdout);
+ if (show_content) {
+ for (int i = 0 ; i < level ; ++i) {
+ fputs(" ", stdout);
+ }
+ if (entry->c_len == 0) {
+ fputs("(nil)", stdout);
+ } else {
+ const char *c = array_ptr(trie->c, entry->c_offset);
+ printf("(%d) ", entry->c_len);
+ for (int i = 0 ; i < entry->c_len ; ++i) {
+ if (c[i]) {
+ printf("%c ", c[i]);
+ } else {
+ fputs("\\0 ", stdout);
+ }
}
}
+ fputs("\n", stdout);
}
- 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, &trie->entries[i], level + 1);
+ trie_entry_inspect(trie, show_content, array_ptr(trie->entries, i), level + 1);
}
if (level == 0) {
- printf("Average char per node: %d\n", trie->c_len / trie->entries_len);
- printf("Number of nodes: %d\n", trie->entries_len);
- printf("Number of leafs: %d\n", leafs);
+ printf("Average char per node: %d\n", trie->c.len / trie->entries.len);
+ printf("Number of nodes: %d\n", trie->entries.len);
+ printf("Number of leaves: %d\n", leaves);
printf("Max depth: %d\n", max_depth);
- printf("Average leaf depth: %d\n", depth_sum / leafs);
- printf("Memory used: %d\n", (trie->entries_size * sizeof(trie_entry_t))
- + (trie->c_size) + sizeof(trie_t));
+ printf("Average leaf depth: %d\n", depth_sum / leaves);
+ printf("Memory used: %zd\n", (trie->entries.size * sizeof(trie_entry_t))
+ + (trie->c.size) + sizeof(trie_t));
}
}
-void trie_inspect(const trie_t *trie)
+void trie_inspect(const trie_t *trie, bool show_content)
{
- trie_entry_inspect(trie, trie->entries, 0);
+ trie_entry_inspect(trie, show_content, trie->entries.data, 0);
}
projects / apps / pfixtools.git / blobdiff