/// <summary>
/// 加载内存词库
/// </summary>
private void LoadDictionary()
{
if (DictionaryPath != string.Empty)
{
List <string> wordList = new List <string>();
Array.Clear(MEMORYLEXICON, 0, MEMORYLEXICON.Length);
string[] words = System.IO.File.ReadAllLines(DictionaryPath, System.Text.Encoding.UTF8);
foreach (string word in words)
{
string key = ToDBC(word);
wordList.Add(key);
wordList.Add(Strings.StrConv(key, VbStrConv.TraditionalChinese));
}
int Cmp(string key1, string key2)
{
return(String.Compare(key1, key2, StringComparison.Ordinal));
}
wordList.Sort(Cmp);
for (int i = wordList.Count - 1; i > 0; i--)
{
if (wordList[i] == wordList[i - 1])
{
wordList.RemoveAt(i);
}
}
foreach (var word in wordList)
{
WordGroup group = MEMORYLEXICON[word[0]];
if (group == null)
{
group = new WordGroup();
MEMORYLEXICON[word[0]] = group;
}
group.Add(word.Substring(1));
}
}
}
/// <summary>
/// 查找并替换
///使用方法
/// string returnstr = "";//返回已经替换的敏感词
///string str = System.AppDomain.CurrentDomain.SetupInformation.ApplicationBase;
///fw.DictionaryPath = str + "sensitive.txt";
///fw.SourctText = txttest.Value.Trim();
/// returnstr = fw.Filter('*');//*为要替换成的字符
/// </summary>
/// <param name="replaceChar">替换成的字符</param>
public string Filter(char replaceChar)
{
LoadDictionary();
if (SourctText != string.Empty)
{
char[] tempString = SourctText.ToCharArray();
for (int i = 0; i < SourctText.Length; i++)
{
//查询以该字为首字符的词组
WordGroup group = MEMORYLEXICON[ToDBC(SourctText)[i]];
if (group != null)
{
for (int z = 0; z < group.Count(); z++)
{
string word = group.GetWord(z);
if (word.Length == 0 || Check(word))
{
string blackword = string.Empty;
for (int pos = 0; pos < _wordlenght + 1; pos++)
{
blackword += tempString[pos + _cursor].ToString();
tempString[pos + _cursor] = replaceChar;
}
IllegalWords.Add(blackword);
_cursor = _cursor + _wordlenght;
i = i + _wordlenght;
}
}
}
_cursor++;
}
return(new string(tempString));
}
return(string.Empty);
}
