RubyTrie 2.1 → 2.2

data/ChangeLog

@@ -1,4 +1,5 @@
 1.0 added children and each methods
 1.1 added levenshtein algorithm implementation - inspired from: http://stevehanov.ca/blog/index.php?id=114
 2.0 completely rewritten the trie implementation to use more optimized data structure. For /usr/share/dict/web2 it takes the memory to about 6Mb from 40Mb. The levenshtein now takes a block and passes in the value but it can be easy extened to return an array.  It also has a glitch where some values are for some reason just false... :|
-2.1 bug fixes, fixing levenshtein search
+2.1 bug fixes, fixing levenshtein search
+2.2 fixes child search

data/ext/trie/otrie2.c

@@ -13,6 +13,8 @@
 
 #include "otrie2.h"
 
+Node* node_find_impl(Node* this, const char* string, bool include_partial);
+
 Node* new_node() {
 	Node *node = malloc(sizeof(Node));
 	memset(node, 0, sizeof(Node));
@@ -61,6 +63,14 @@ void node_insert(Node* node, const char* string, const VALUE value) {
 }
 
 Node* node_find(Node* this, const char* string) {
+	return node_find_impl(this, string, false);
+}
+
+Node* partial_node_find(Node* this, const char* string) {
+	return node_find_impl(this, string, true);
+}
+
+Node* node_find_impl(Node* this, const char* string, bool include_partial) {
 	int len = strlen(string);
 	Pos *cur = new_pos(this, 0);
 	int i=0;
@@ -68,7 +78,7 @@ Node* node_find(Node* this, const char* string) {
 		pos_next(cur, string + i, false);
 		if (cur->node == NULL) { return NULL; }
 	}
-	if (strlen(cur->node->data) == cur->offset + 1)
+	if (strlen(cur->node->data) == cur->offset + 1 || include_partial)
 		return cur->node;
 	return NULL;
 }
@@ -129,11 +139,10 @@ void pos_next(Pos *this, const char* string, bool insert) {
 				Node *splitChild = new_node_string(this->node->data + this->offset + 1);
 				splitChild -> value = this -> node -> value;
 				Node *newChild = new_node_string(string); 
-        newChild -> value = Qnil;
+				newChild -> value = Qnil;
 				node_update_data(this->node, this->node->data, this->offset + 1);
 				splitChild -> next_sibling = newChild;
 				this->node -> first_child = splitChild;
-        this->node -> value = Qnil;
 				
 				this->node = newChild;
 				this->offset = 0;

data/ext/trie/otrie2.h

@@ -33,6 +33,7 @@ Node* new_node();
 void free_node(Node*);
 void node_insert(Node* node, const char* string, const VALUE value);
 Node* node_find(Node* this, const char* string);
+Node* partial_node_find(Node* this, const char* string);
 
 Pos* new_pos(Node *node, int offset);
 Node* pos_find_or_create_child(Pos* this, const char* string, bool down, bool insert);

data/ext/trie/ruby_trie.c

@@ -117,7 +117,7 @@ static VALUE rb_trie_find_children(VALUE self, VALUE key) {
 	key_cstring = StringValuePtr(key);
 	Data_Get_Struct(self, Node, root);
 	
-	node = node_find(root, key_cstring);
+	node = partial_node_find(root, key_cstring);
 	
 	if (node != NULL && node->value != Qnil) {
 		rb_ary_push(rary, node->value);
@@ -144,7 +144,7 @@ static VALUE rb_trie_find_children_with_block(VALUE self, VALUE key) {
 	key_cstring = StringValuePtr(key);
 	Data_Get_Struct(self, Node, root);
 	
-	node = node_find(root, key_cstring);
+	node = partial_node_find(root, key_cstring);
 	
 	if (node != NULL && node->value != Qnil) {
 		rb_yield(node->value);
@@ -168,12 +168,10 @@ static VALUE rb_trie_set_key_to_value(VALUE self, VALUE key, VALUE value) {
 	
 	node = node_find(root, key_cstring);
 	if (node == NULL || node -> value == Qnil) {
-    // printf("New node for %s -> %d\n", key_cstring, value);
 		VALUE arr = rb_ary_new();
 		rb_ary_push(arr, value);
 		node_insert(root, key_cstring, arr);
 	} else {
-    // printf("Append value %s to %s -> %d\n", node->data, key_cstring, node->value);
 		rb_ary_push(node->value, value);
 	}
 	

metadata

@@ -1,12 +1,12 @@
 --- !ruby/object:Gem::Specification 
 name: RubyTrie
 version: !ruby/object:Gem::Version 
-  hash: 1
+  hash: 7
   prerelease: false
   segments: 
   - 2
-  - 1
-  version: "2.1"
+  - 2
+  version: "2.2"
 platform: ruby
 authors: 
 - Petrica Ghiurca
@@ -15,7 +15,7 @@ autorequire: trie
 bindir: bin
 cert_chain: []
 
-date: 2011-03-21 00:00:00 +02:00
+date: 2011-03-26 23:00:00 +02:00
 default_executable: 
 dependencies: []