写这个的出发点是吧之前做的trie能写到文件而且能很容易mmap到内存,　可以用于输入法　搜索引擎分词　词表的功能,　共享前缀存储,　支持最大２０亿节点.

简单一点可以当一个hash表使用,　查找的时间复杂度为　N* Log(256) N为字符长度,　key可以为二进制的数据,　value　必须为int类型,　不能为０,　因为mmtrie_get()/find() 没有结果的情况下为０,

如果需要set　value为０的话自己+１,　出来的时候-１就好了.-Write the starting point is to do the trie before it can write files and can very easily mmap to memory, can be used to enter France search engine sub-word vocabulary, shared storage prefix supports up to two billion nodes. A simple thing can be when a Use hash table to find the time complexity is N* Log (256) N for the characters in length, key data for binary, value must be int type, can not be 0, because mmtrie_get ()/find () does not result in the situation is 0, if set value to 0 if its+1, out of time-1 enough.