/// <summary>
/// 在文本中查找第一个关键字
/// </summary>
/// <param name="text">文本</param>
/// <returns></returns>
public WordsSearchResult FindFirst(string text)
{
TrieNode2 ptr = null;
for (int i = 0; i < text.Length; i++)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[text[i]];
}
else
{
if (ptr.TryGetValue(text[i], out tn) == false)
{
tn = _first[text[i]];
}
}
if (tn != null)
{
if (tn.End)
{
var item = tn.Results[0];
return(new WordsSearchResult(_keywords[item], i + 1 - _keywords[item].Length, i, _indexs[item]));
}
}
ptr = tn;
}
return(WordsSearchResult.Empty);
}
/// <summary>
/// 在文本中查找所有的关键字
/// </summary>
/// <param name="text">文本</param>
/// <returns></returns>
public List <WordsSearchResult> FindAll(string text)
{
TrieNode2 ptr = null;
List <WordsSearchResult> list = new List <WordsSearchResult>();
for (int i = 0; i < text.Length; i++)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[text[i]];
}
else
{
if (ptr.TryGetValue(text[i], out tn) == false)
{
tn = _first[text[i]];
}
}
if (tn != null)
{
if (tn.End)
{
foreach (var item in tn.Results)
{
list.Add(new WordsSearchResult(_keywords[item], i + 1 - _keywords[item].Length, i, _indexs[item]));
}
}
}
ptr = tn;
}
return(list);
}
/// <summary>
/// 判断文本是否包含关键字
/// </summary>
/// <param name="text">文本</param>
/// <returns></returns>
public bool ContainsAny(string text)
{
TrieNode2 ptr = null;
foreach (char t in text)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[t];
}
else
{
if (ptr.TryGetValue(t, out tn) == false)
{
tn = _first[t];
}
}
if (tn != null)
{
if (tn.End)
{
return(true);
}
}
ptr = tn;
}
return(false);
}
/// <summary>
/// 在文本中查找所有的关键字
/// </summary>
/// <param name="text">文本</param>
/// <returns></returns>
public List <string> FindAll(string text)
{
TrieNode2 ptr = null;
List <string> list = new List <string>();
foreach (char t in text)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[t];
}
else
{
if (ptr.TryGetValue(t, out tn) == false)
{
tn = _first[t];
}
}
if (tn != null)
{
if (tn.End)
{
foreach (var item in tn.Results)
{
list.Add(_keywords[item]);
}
}
}
ptr = tn;
}
return(list);
}
/// <summary>
/// 在文本中查找第一个关键字
/// </summary>
/// <param name="text">文本</param>
/// <returns></returns>
public string FindFirst(string text)
{
TrieNode2 ptr = null;
foreach (char t in text)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[t];
}
else
{
if (ptr.TryGetValue(t, out tn) == false)
{
tn = _first[t];
}
}
if (tn != null)
{
if (tn.End)
{
return(_keywords[tn.Results[0]]);
}
}
ptr = tn;
}
return(null);
}
private void TryLinks(TrieNode2 node, TrieNode2 node2, Dictionary <TrieNode2, TrieNode2> links)
{
foreach (var item in node.m_values)
{
TrieNode2 tn;
if (node2 == null)
{
tn = _first[item.Key];
if (tn != null)
{
links[item.Value] = tn;
}
}
else
{
if (node2.TryGetValue(item.Key, out tn))
{
links[item.Value] = tn;
}
}
TryLinks(item.Value, tn, links);
}
}
/// <summary>
/// 在文本中替换所有的关键字
/// </summary>
/// <param name="text">文本</param>
/// <param name="replaceChar">替换符</param>
/// <returns></returns>
public string Replace(string text, char replaceChar = '*')
{
StringBuilder result = new StringBuilder(text);
TrieNode2 ptr = null;
for (int i = 0; i < text.Length; i++)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[text[i]];
}
else
{
if (ptr.TryGetValue(text[i], out tn) == false)
{
tn = _first[text[i]];
}
}
if (tn != null)
{
if (tn.End)
{
var maxLength = _keywords[tn.Results[0]].Length;
var start = i + 1 - maxLength;
for (int j = start; j <= i; j++)
{
result[j] = replaceChar;
}
}
}
ptr = tn;
}
return(result.ToString());
}
public bool Find(string text, string hz, string[] pinyins)
{
TrieNode2 ptr = null;
for (int i = 0; i < text.Length; i++)
{
TrieNode2 tn;
if (ptr == null)
{
tn = _first[text[i]];
}
else
{
if (ptr.TryGetValue(text[i], out tn) == false)
{
tn = _first[text[i]];
}
}
if (tn != null)
{
if (tn.End)
{
foreach (var result in tn.Results)
{
var keyword = _keywords[result];
var start = i + 1 - keyword.Length;
var end = i;
bool isok = true;
var keywordPinyins = _keywordPinyins[result];
for (int j = 0; j < keyword.Length; j++)
{
var idx = start + j;
var py = keywordPinyins[j];
if (py.Length == 1 && py[0] >= 0x3400 && py[0] <= 0x9fd5)
{
if (hz[idx] != py[0])
{
isok = false;
break;
}
}
else
{
if (pinyins[idx].StartsWith(py) == false)
{
isok = false;
break;
}
}
}
if (isok)
{
return(true);
}
}
}
}
ptr = tn;
}
return(false);
}