huffman_f 在huffman_e的基础上，将排序改为利用堆排序原理选择最小的两个权值。也即，将所有元素的权值组织成堆后，每次堆内的根结点就是最小值了。每取出一个根结点后，就把堆尾元素调到根结点重建堆。取出两个最小值合并成一个子树后，再把子树作为叶子结点放到堆中，并让其上升到合适的位置，保持堆性质不变。因为每次不必完成整个排序过程，而只是组织成堆，因此，这种方法要比使用快速排序更快。上述算法参考了mg-1.2.1中Huffman编码的实现，见http://www.cs.mu.oz.au/mg-huffman_f in huffman_e, on the basis of order to use Heap Sort tenets of the two smallest choose the right values. That is, all the elements of the weights organizations piles, each pile of Root is a minimum. Root out every one after, they pile tail elements transferred Root reconstruction reactor. Two out minimum son into a tree, then the tree-leaf nodes as into rubble, and allowed to rise to a suitable location, to maintain the nature of sculpture. Because each time without complete the entire sequencing process, and the only organization piles, therefore, this method than using the rapid sequencing faster. Reference to the above algorithm mg-1.2.1 Huffman coding to achieve, see http://www.cs.mu.oz.au/mg