敏感词过滤 DFA 状态机 cocos2d-js 实现版

// 定义命名空间
var GameClient = GameClient || {};

// DFA 实现敏感词过滤 // improve from http://blog.sina.com.cn/s/blog_53f6148101016wq1.html
GameClient.FilterWord = cc.Class.extend({
    m_wordList: null,// @private
    m_map: null,// @private

    ctor: function (wordList) {
        if (wordList == null) {
            wordList = [];
        }
//        cc.warn("wordList: %o",wordList);
        this.m_wordList = wordList;
        this.m_map = this.BuildMap(wordList);
    },// 替换 str 中的敏感词为 replaceStr,step 为筛选强度， 最小为 0，最大为 5
    RunFilterWord: function (str,replaceStr,step) {
        if (str == undefined) {
            return "";
        }

        if (step > 5) {
            step = 5;
        }
        if (step < 0) {
            step = 0;
        }
        for (var j = 0; j < step; j++) {
            var result = this.Check(this.m_map,str);
            // 排序结果，长度长的在前面
            result.sort(function(a,b){
                return b.length - a.length});
            console.log("result: " + result);
            for (var i = 0; i < result.length; i++) {
                str = str.replace(result[i],replaceStr);
            }
        }

        return str;
    },// @private
    BuildMap: function (wordList) {
        var result = {};
        var count = wordList.length;
        // 遍历单词
        for (var i = 0; i < count; ++i) {
            var map = result;
            var word = wordList[i];
            var wordLength = word.length;
            // 遍历单词的每个字母
            for (var j = 0; j < wordLength; ++j) {
                var ch = word.charat(j);
                var stateInfo = map[ch];

                if (stateInfo == null) {
                    stateInfo = {};
                    map[ch] = stateInfo;
                }
                // 如果是最后一个字母，设置一个完结标识
                if (j == wordLength - 1) {
                    stateInfo["isFinish"] = true;
                }
                else {
                    map = stateInfo;
                }
            }
        }

        return result;
    },// @private // 找出 content 中的敏感词，会返回一个包含敏感词的列表
    Check: function (map,content) {
        var result = [];
        var count = content.length;
        var stack = [];
        var point = map;
        // 用于标记找到关键词的标记
        var isFound = false;
        var foundStack = null;

        for (var i = 0; i < count; ++i) {
            var ch = content.charat(i);
            var item = point[ch];
            // 如果没找到则复位，让主循环
            if (item == null) {
                if (isFound) {
                    isFound = false;
                    i = i - (stack.length - foundStack.length + 1); // 计算回退距离
                    result.push(foundStack.join("")); // 把单个的字母数组连成一串字符串
                }
                else {
                    i = i - stack.length; // 计算回退距离
                }
                stack = []; // 清空字符堆栈
                point = map;
            }
            else if (item["isFinish"]) {
                stack.push(ch);
                point = item;
                // 标记找到了目标词
                isFound = true;
                foundStack = stack.concat(); // 复制数组
            }
            else {
                stack.push(ch);
                point = item;
            }
        }
        // 这里还要补充检查
        if (isFound) {
            result.push(foundStack.join("")); // 把单个的字母数组连成一串字符串
        }
        return result;
    },Destroy: function () {
        this.m_wordList = null;
        this.m_map = null;
    }
});

var filterWord = new GameClient.FilterWord(["ab","bad","abc","ddd","aaa","ccc","bbb","adsf","edsdf","ss","fff","aaa"]);
console.log("1: " + filterWord.RunFilterWord("asdfsdsddaadaaabbbffdassdf sdf bbb aaa a","#",1));

直接上代码