DFA算法 使用C# 实现 敏感词过滤

DFA，全称 Deterministic Finite Automaton 即确定有穷自动机：从一个状态通过一系列的事件转换到另一个状态，即 state -> event -> state。状态以及引起状态转换的事件都是可确定的，状态以及事件的数量都是可穷举的。

下面是使用C#代码进行实现的逻辑，下面贴出来整个类的实现。相互学习，勿喷。

using System.Collections;
using System.Text;

namespace WeChatCapture
{
    public class WordSensitiveDFA
    {
        /// <summary>
        /// 原始过滤词数据集
        /// </summary>
        public HashSet<string> sensitiveWordSet;

        /// <summary>
        /// 过滤词库
        /// </summary>
        public Hashtable sensitiveWordTable = new Hashtable();

        /// <summary>
        /// 构造函数
        /// </summary>
        /// <param name="SensitiveWordsArray">过滤词 词库 array</param>
        public WordSensitiveDFA(string[] SensitiveWordsArray)
        {
            sensitiveWordSet = new HashSet<string>(SensitiveWordsArray);
            InitSensitiveWordMap();
        }

        /// <summary>
        /// 初始化 过滤词 词库
        /// </summary>
        public void InitSensitiveWordMap()
        {
            sensitiveWordTable = new Hashtable(sensitiveWordSet.Count);

            foreach (string word in sensitiveWordSet)
            {
                Hashtable newTable = sensitiveWordTable;
                for (int i = 0; i < word.Length; i++)
                {
                    char w = word[i];
                    if (newTable.ContainsKey(w))
                    {
                        newTable = (Hashtable)newTable[w];
                    }
                    else
                    {
                        var newMap = new Hashtable();
                        newMap.Add("IsEnd", 0);
                        newTable.Add(w, newMap);
                        newTable = newMap;
                    }

                    if (i == word.Length - 1)
                    {
                        if (newTable.ContainsKey("IsEnd"))
                        {
                            newTable["IsEnd"] = 1;
                        }
                        else
                        {
                            newTable.Add("IsEnd", 1);
                        }
                    }
                }
            }
        }

        /// <summary>
        /// 查找所有敏感词，找到则返回敏感词长度
        /// </summary>
        /// <param name="content">需要过滤的字符串</param>
        /// <param name="startIndex">查找的起始位置</param>
        /// <returns></returns>
        public int SearchSensitiveWord(string content, int startIndex)
        {
            Hashtable newMap = sensitiveWordTable;
            bool flag = false;
            int len = 0;

            for (int i = startIndex; i < content.Length; i++)
            {
                char word = content[i];
                Hashtable temp = (Hashtable)newMap[word];
                if (temp != null)
                {
                    if ((int)temp["IsEnd"] == 1) flag = true;
                    else newMap = temp;
                    len++;
                }
                else break;
            }
            if (!flag) len = 0;
            return len;
        }

        /// <summary>
        /// 找到内容字符串内所有敏感词
        /// </summary>
        /// <param name="txt">需要过滤的文本内容</param>
        /// <returns></returns>
        public List<string> GetAllSensitiveWords(string txt)
        {
            List<string> result = new List<string>();
            for (int i = 0; i < txt.Length; i++)
            {
                int length = SearchSensitiveWord(txt, i);
                if (length > 0)
                {
                    result.Add(txt.Substring(i, length));
                    i = i + length - 1;
                }
            }
            return result;
        }

        /// <summary>
        /// 替换 需要剔除的 敏感字 \0 char 空
        /// </summary>
        /// <param name="txt">需要过滤的文本内容</param>
        /// <returns></returns>
        public string ReplaceSensitiveWords(string txt)
        {
            int i = 0;
            StringBuilder sbstr = new StringBuilder(txt);
            while (i < txt.Length)
            {
                int len = SearchSensitiveWord(txt, i);
                if (len > 0)
                {
                    for (int j = 0; j < len; j++)
                    {
                        sbstr[i + j] = '\0';
                    }
                    i += len;
                }
                else
                    ++i;
            }
            return sbstr.ToString();
        }

    }
}