平衡二叉树(Balanced Binary Tree)又被称为AVL树
且具有以下性质:
它是一 棵空树或它的左右两个子树的高度差的绝对值不超过1,并且左右两个子树都是一棵平衡二叉树。
构造与调整方法 平衡二叉树的常用算法有红黑树、AVL、Treap等。
最小二叉平衡树的节点的公式如下 F(n)=F(n-1)+F(n-2)+1 这个类似于一个递归的数列,可以参考Fibonacci数列,1是根节点,F(n-1)是左子树的节点数量,F(n-2)是右子树的节点数量。
AVL树是最先发明的自平衡二叉查找算法,是平衡二叉树的一种。在AVL中任何节点的两个儿子子树的高度最大差别为1,所以它又被成为高度平衡树。查找、插入和删除在平均和最坏情况下都是O(log n)。增加和删除可能需要通过一次或多次树旋转来平衡这棵树。
总的来说平衡二叉树是一种存储结构,这种存储结构会提高遍历和查找的效率,所以今天讲的数据结构是用来存储数据,并且目的是提高查找效率的,作用有点像哈希表。
先来说明文档网址:http://web.mit.edu/barnowl/share/gtk-doc/html/glib/glib-Balanced-Binary-Trees.html
这里需要注意的是我在编译的程序的时候使用的是2.0.6版本,而这个帮助文档的版本是2.16.3,其中g_tree_remove函数的返回值类型不同。
这个数据结构体:
typedef struct _GTree GTree;
功能函数:
#include <glib.h> GTree; GTree* g_tree_new (GCompareFunc key_compare_func); GTree* g_tree_new_with_data (GCompareDataFunc key_compare_func, gpointer key_compare_data); GTree* g_tree_new_full (GCompareDataFunc key_compare_func, gpointer key_compare_data, GDestroyNotify key_destroy_func, GDestroyNotify value_destroy_func); void g_tree_insert (GTree *tree, gpointer key, gpointer value); void g_tree_replace (GTree *tree, gpointer key, gpointer value); gint g_tree_nnodes (GTree *tree); gint g_tree_height (GTree *tree); gpointer g_tree_lookup (GTree *tree, gconstpointer key); gboolean g_tree_lookup_extended (GTree *tree, gconstpointer lookup_key, gpointer *orig_key, gpointer *value); void g_tree_foreach (GTree *tree, GTraverseFunc func, gpointer user_data); void g_tree_traverse (GTree *tree, GTraverseFunc traverse_func, GTraverseType traverse_type, gpointer user_data); gboolean (*GTraverseFunc) (gpointer key, gpointer value, gpointer data); enum GTraverseType; gpointer g_tree_search (GTree *tree, GCompareFunc search_func, gconstpointer user_data); gboolean g_tree_remove (GTree *tree, gconstpointer key); gboolean g_tree_steal (GTree *tree, gconstpointer key); void g_tree_destroy (GTree *tree);
下面先看一小段代码:
#include <glib.h> struct map { int key; char *value; } m[10] = { {0,"zero"}, {1,"one"}, {2,"two"}, {3,"three"}, {4,"four"}, {5,"five"}, {6,"six"}, {7,"seven"}, {8,"eight"}, {9,"nine"}, }; typedef struct map map; static gint myCompare(gconstpointer p1, gconstpointer p2) { const char *a = p1; const char *b = p2; return *a - *b; } static gint mySearch(gconstpointer p1, gconstpointer p2) { return myCompare(p1, p2); } static gint myTraverse(gpointer key, gpointer value, gpointer fmt) { g_printf(fmt, *(gint*)key, (gchar*)value); return FALSE; } static void test_avl_tree(void) { GTree *tree; gint i; // GTree* g_tree_new(GCompareFunc key_compare_func); 用来创建一个平衡二叉树,key_compare_func函数用来排序节点规则 tree = g_tree_new(myCompare); // void g_tree_insert(GTree *tree, gpointer key, gpointer value); 插入一个键值到树中 // 需要注意的是如果key值重复的话会将以前的值进行去除然后将新的键值加入树中,去除的包括key值 // 如果使用的是new_full函数并且实现了销毁key和value的函数,在插入重复key时会新的key被销毁 for (i = 0; i < sizeof(m)/sizeof(m[0]); i++) g_tree_insert(tree, &m[i].key, m[i].value); // void g_tree_foreach(GTree *tree, GTraverseFunc func, gpointer user_data); 遍历树,将键值依次传入func函数中 g_printf("Now the tree:\n"); g_tree_foreach(tree, myTraverse, "Key:\t%d\t\tVaule:\t%s\n"); // gint g_tree_nnodes(GTree *tree); 取得树中的节点数 g_printf("The tree should have '%d' items now.\t\tResult: %d.\n", 10, g_tree_nnodes(tree)); // gint g_tree_height(GTree *tree); 取得树的高度,如果是空树将返回0,如果只有根节点返回1,以此类推 g_printf("The height of tree is '%d' now.\n", g_tree_height(tree)); // void g_tree_replace(GTree *tree, gpointer key, gpointer value); 将key的值用value替换,功能和g_tree_insert很相似,只是在销毁的时候会去销毁旧的key,而不是新的 g_tree_replace(tree, &m[3].key, "3333"); g_printf("Now the vaule of '%d' should be '3333' now.\n", m[3].key); g_tree_foreach(tree, myTraverse, "Key:\t%d\t\tVaule:\t%s\n"); gchar *tmp = NULL; // gpointer g_tree_lookup(GTree *tree, gconstpointer key); 查找key的值 g_printf("Now the vaule of '%d' should be '%s' now[lookup].\n", m[3].key, (tmp = (gchar *)g_tree_lookup(tree, &m[3].key)) != NULL ? tmp : NULL); // gboolean g_tree_remove(GTree *tree, gconstpointer key); 移除key对应的键值,需要注意的是在2.0版本中这个函数返回类型是void,在之后的版本中返回是bool,代表是否成功 // gboolean b = g_tree_remove(tree, &m[3].key); g_tree_remove(tree, &m[3].key); g_printf("The key '%d' has been found and removed now.\n", m[3].key); // gpointer g_tree_search(GTree *tree, GCompareFunc search_func, gconstpointer user_data); 通过search_func形式去查找树,可以使用这种方式暂时改变查找方式,比如正序或倒序 g_printf("Now the vaule which should be removed of '%d' should be '%s' now[search].\n", m[3].key, (tmp = (gchar *)g_tree_search(tree, mySearch, &m[3].key)) != NULL ? tmp : NULL); g_printf("Now the tree look like:\n"); g_tree_foreach(tree, myTraverse, "Key:\t%d\t\tVaule:\t%s\n"); // void g_tree_destroy(GTree *tree);释放树,如果使用g_tree_new_full()函数创建的树会使用销毁函数同时销毁键值 g_tree_destroy(tree); } int main(void) { g_printf("BEGIN:\n************************************************************\n"); test_avl_tree(); g_printf("\n************************************************************\nDONE\n"); return 0; }
看一下运行结果:
linux@ubuntu:~/16021/glibDemo$ gcc GTree.c -o GTree -lglib-2.0 linux@ubuntu:~/16021/glibDemo$ ./GTree BEGIN: ************************************************************ Now the tree: Key: 0 Vaule: zero Key: 1 Vaule: one Key: 2 Vaule: two Key: 3 Vaule: three Key: 4 Vaule: four Key: 5 Vaule: five Key: 6 Vaule: six Key: 7 Vaule: seven Key: 8 Vaule: eight Key: 9 Vaule: nine The tree should have '10' items now. Result: 10. The height of tree is '4' now. Now the vaule of '3' should be '3333' now. Key: 0 Vaule: zero Key: 1 Vaule: one Key: 2 Vaule: two Key: 3 Vaule: 3333 Key: 4 Vaule: four Key: 5 Vaule: five Key: 6 Vaule: six Key: 7 Vaule: seven Key: 8 Vaule: eight Key: 9 Vaule: nine Now the vaule of '3' should be '3333' now[lookup]. The key '3' has been found and removed now. Now the vaule which should be removed of '3' should be '(null)' now[search]. Now the tree look like: Key: 0 Vaule: zero Key: 1 Vaule: one Key: 2 Vaule: two Key: 4 Vaule: four Key: 5 Vaule: five Key: 6 Vaule: six Key: 7 Vaule: seven Key: 8 Vaule: eight Key: 9 Vaule: nine ************************************************************ DONE linux@ubuntu:~/16021/glibDemo$
如果使用先序遍历存储模型就是下图:
从上面这个图中也能明白为什么g_tree_height函数返回值为4了。
在glib中还有一种数据结构叫做n叉树 N-ary Trees
这是树的最一般形式,但是应用却是最少的,现实生活中的N叉树是最多的,但是对于它的操作、算法却是非常的麻烦,在实际使用时往往通过人为方法将n叉树转换为二叉树去进行处理,所以对于n叉树的数据结构就不去了解了。