feat: allow skips up to 8 characters long
This commit is contained in:
parent
4bcdd5c4d9
commit
88ea0db2ee
4 changed files with 224 additions and 196 deletions
147
src/trie_node.c
147
src/trie_node.c
|
|
@ -8,10 +8,10 @@
|
|||
* Represents a node of the binary tree contained within each non-leaf
|
||||
* TrieNode.
|
||||
*/
|
||||
typedef struct ttinode {
|
||||
struct ttinode *left;
|
||||
struct ttinode *right;
|
||||
struct ttnode *next;
|
||||
typedef struct tinode {
|
||||
struct tinode *left;
|
||||
struct tinode *right;
|
||||
struct tnode *next;
|
||||
char key;
|
||||
} TrieInnerNode;
|
||||
|
||||
|
|
@ -26,25 +26,21 @@ typedef struct ttinode {
|
|||
* to be stored as a single node. Its size will be zero, represents its true,
|
||||
* and its string pointer is initialized.
|
||||
*/
|
||||
typedef struct ttnode {
|
||||
union {
|
||||
TrieInnerNode *root;
|
||||
char *string;
|
||||
} ptr;
|
||||
typedef struct tnode {
|
||||
Entry *entry;
|
||||
// What type of node this is
|
||||
// 0: regular non-representing node
|
||||
// 1: regular representing node
|
||||
// 2: full leaf
|
||||
uint8_t type;
|
||||
// Dependent on type
|
||||
// 0, 1: size of underlying binary tree
|
||||
// 2: length of string
|
||||
uint8_t size;
|
||||
|
||||
TrieInnerNode* tree;
|
||||
uint8_t tree_size;
|
||||
|
||||
// Skips are at most 8 characters, and are stored in the nodes
|
||||
char string[TRIE_MAX_SKIP_SIZE];
|
||||
uint8_t string_len: 4;
|
||||
|
||||
bool represents : 1;
|
||||
} TrieNode;
|
||||
|
||||
// Required for recursively freeing tree structure
|
||||
void ttnode_free(TrieNode *node);
|
||||
void tnode_free(TrieNode *node);
|
||||
|
||||
/**
|
||||
* Allocate and initialize a new TrieInnerNode representing a given
|
||||
|
|
@ -53,7 +49,7 @@ void ttnode_free(TrieNode *node);
|
|||
* @param c character to represent
|
||||
* @return pointer to newly allocated struct
|
||||
*/
|
||||
TrieInnerNode *ttinode_init(char c) {
|
||||
TrieInnerNode *tinode_init(char c) {
|
||||
TrieInnerNode *node = calloc(1, sizeof(TrieInnerNode));
|
||||
node->key = c;
|
||||
|
||||
|
|
@ -65,7 +61,15 @@ TrieInnerNode *ttinode_init(char c) {
|
|||
*
|
||||
* @return pointer to newly allocated struct
|
||||
*/
|
||||
TrieNode *ttnode_init() { return calloc(1, sizeof(TrieNode)); }
|
||||
TrieNode *tnode_init() {
|
||||
TrieNode *node = malloc(sizeof(TrieNode));
|
||||
|
||||
node->tree_size = 0;
|
||||
node->string_len = 0;
|
||||
node->represents = false;
|
||||
|
||||
return node;
|
||||
}
|
||||
|
||||
/**
|
||||
* Free a TrieInnerNode and its underlying tree structure. This should
|
||||
|
|
@ -74,17 +78,17 @@ TrieNode *ttnode_init() { return calloc(1, sizeof(TrieNode)); }
|
|||
*
|
||||
* @param node node whose tree to free
|
||||
*/
|
||||
void ttinode_free_cascade(TrieInnerNode *node) {
|
||||
void tinode_free_cascade(TrieInnerNode *node) {
|
||||
if (node->left != NULL) {
|
||||
ttinode_free_cascade(node->left);
|
||||
tinode_free_cascade(node->left);
|
||||
}
|
||||
|
||||
if (node->right != NULL) {
|
||||
ttinode_free_cascade(node->right);
|
||||
tinode_free_cascade(node->right);
|
||||
}
|
||||
|
||||
if (node->next != NULL) {
|
||||
ttnode_free(node->next);
|
||||
tnode_free(node->next);
|
||||
}
|
||||
|
||||
free(node);
|
||||
|
|
@ -95,11 +99,9 @@ void ttinode_free_cascade(TrieInnerNode *node) {
|
|||
*
|
||||
* @param node node to free
|
||||
*/
|
||||
void ttnode_free(TrieNode *node) {
|
||||
if (node->type == 2) {
|
||||
free(node->ptr.string);
|
||||
} else if (node->size != 0) {
|
||||
ttinode_free_cascade(node->ptr.root);
|
||||
void tnode_free(TrieNode *node) {
|
||||
if (node->tree_size > 0) {
|
||||
tinode_free_cascade(node->tree);
|
||||
}
|
||||
|
||||
// TODO properly free entry
|
||||
|
|
@ -110,18 +112,6 @@ void ttnode_free(TrieNode *node) {
|
|||
free(node);
|
||||
}
|
||||
|
||||
/**
|
||||
* Add the string to the given node & set its type accordingely.
|
||||
*
|
||||
* @param node node to add string to
|
||||
* @param string string to add
|
||||
*/
|
||||
void ttnode_set_string(TrieNode *node, const char *string) {
|
||||
node->type = 2;
|
||||
node->size = strlen(string);
|
||||
node->ptr.string = strdup(string);
|
||||
}
|
||||
|
||||
/**
|
||||
* This function performs a lookup in the underlying binary tree of the given
|
||||
* TrieNode. If found, the return value is a pointer to the memory
|
||||
|
|
@ -140,26 +130,21 @@ void ttnode_set_string(TrieNode *node, const char *string) {
|
|||
* node represents a leaf with a string, because the struct and therefore the
|
||||
* address is created if it doesn't exist yet.
|
||||
*/
|
||||
TrieNode **ttnode_search(TrieNode *node, const char c,
|
||||
TrieNode **tnode_search(TrieNode *node, const char c,
|
||||
bool create) {
|
||||
// Full leafs will always return NULL
|
||||
if (node->type == 2) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
// It can happen that the node has no initialized root yet
|
||||
if (node->size == 0) {
|
||||
if (node->tree_size == 0) {
|
||||
if (create) {
|
||||
node->size++;
|
||||
node->ptr.root = ttinode_init(c);
|
||||
node->tree_size++;
|
||||
node->tree = tinode_init(c);
|
||||
|
||||
return &node->ptr.root->next;
|
||||
return &node->tree->next;
|
||||
}
|
||||
|
||||
return NULL;
|
||||
}
|
||||
|
||||
TrieInnerNode *parent = node->ptr.root;
|
||||
TrieInnerNode *parent = node->tree;
|
||||
TrieInnerNode *child;
|
||||
|
||||
// Iterate through the tree until we either find the character or realize it's
|
||||
|
|
@ -186,7 +171,7 @@ TrieNode **ttnode_search(TrieNode *node, const char c,
|
|||
// If create is true, we create the new node so that we can still return a
|
||||
// non-NULL pointer.
|
||||
if (create) {
|
||||
TrieInnerNode *new_node = ttinode_init(c);
|
||||
TrieInnerNode *new_node = tinode_init(c);
|
||||
|
||||
if (c < parent->key) {
|
||||
parent->left = new_node;
|
||||
|
|
@ -194,7 +179,7 @@ TrieNode **ttnode_search(TrieNode *node, const char c,
|
|||
parent->right = new_node;
|
||||
}
|
||||
|
||||
node->size++;
|
||||
node->tree_size++;
|
||||
|
||||
return &new_node->next;
|
||||
}
|
||||
|
|
@ -208,37 +193,37 @@ TrieNode **ttnode_search(TrieNode *node, const char c,
|
|||
*
|
||||
* @param node node to split
|
||||
*/
|
||||
void ttnode_split(TrieNode *node) {
|
||||
TrieNode *new_node = ttnode_init();
|
||||
char key = node->ptr.string[0];
|
||||
/* void tnode_split(TrieNode *node) { */
|
||||
/* TrieNode *new_node = tnode_init(); */
|
||||
/* char key = node->ptr.string[0]; */
|
||||
|
||||
// There's a chance the remaining string was only 1 character, meaning the new
|
||||
// node doesn't have to store a string
|
||||
if (node->ptr.string[1] != DELIMITER) {
|
||||
ttnode_set_string(new_node, node->ptr.string + 1);
|
||||
} else {
|
||||
new_node->type = 1;
|
||||
}
|
||||
/* // There's a chance the remaining string was only 1 character, meaning the new */
|
||||
/* // node doesn't have to store a string */
|
||||
/* if (node->ptr.string[1] != DELIMITER) { */
|
||||
/* tnode_set_string(new_node, node->ptr.string + 1); */
|
||||
/* } else { */
|
||||
/* new_node->type = 1; */
|
||||
/* } */
|
||||
|
||||
new_node->entry = node->entry;
|
||||
/* new_node->entry = node->entry; */
|
||||
|
||||
node->type = 0;
|
||||
node->size = 0;
|
||||
node->entry = NULL;
|
||||
/* node->type = 0; */
|
||||
/* node->size = 0; */
|
||||
/* node->entry = NULL; */
|
||||
|
||||
free(node->ptr.string);
|
||||
node->ptr.string = NULL;
|
||||
/* free(node->ptr.string); */
|
||||
/* node->ptr.string = NULL; */
|
||||
|
||||
// Initialize node's binary tree with the correct character
|
||||
TrieNode **node_ptr = ttnode_search(node, key, true);
|
||||
*node_ptr = new_node;
|
||||
}
|
||||
/* // Initialize node's binary tree with the correct character */
|
||||
/* TrieNode **node_ptr = tnode_search(node, key, true); */
|
||||
/* *node_ptr = new_node; */
|
||||
/* } */
|
||||
|
||||
/*
|
||||
* Remove the given character from a TrieInnerNode's subtree. The
|
||||
* function assumes the character is indeed in the subtree.
|
||||
*/
|
||||
void ttinode_remove(TrieInnerNode *node, const char c) {
|
||||
void tinode_remove(TrieInnerNode *node, const char c) {
|
||||
TrieInnerNode **to_remove_ptr = &node;
|
||||
|
||||
// We use pointers to pointers here so we can later free the removed node
|
||||
|
|
@ -311,12 +296,8 @@ void ttinode_remove(TrieInnerNode *node, const char c) {
|
|||
* @param node node to remove character from
|
||||
* @param c character to remove
|
||||
*/
|
||||
void ttnode_remove(TrieNode *node, const char c) {
|
||||
ttinode_remove(node->ptr.root, c);
|
||||
void tnode_remove(TrieNode *node, const char c) {
|
||||
tinode_remove(node->tree, c);
|
||||
|
||||
node->size--;
|
||||
|
||||
if (node->size == 0) {
|
||||
node->ptr.root = NULL;
|
||||
}
|
||||
node->tree_size--;
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue