/// <summary>
/// 从文本文件读取字典
/// </summary>
/// <param name="fileName"></param>
static public T_DictFile LoadFromTextDict(String fileName)
{
T_DictFile dictFile = new T_DictFile();
String dictStr = CFile.ReadFileToString(fileName, "utf-8");
String[] words = CRegex.Split(dictStr, "\r\n");
foreach (String word in words)
{
String[] wp = CRegex.Split(word, @"\|");
if (wp == null)
{
continue;
}
if (wp.Length != 2)
{
continue;
}
int pos = 0;
try
{
pos = int.Parse(wp[1]);
}
catch
{
continue;
}
T_DictStruct dict = new T_DictStruct();
dict.Word = wp[0];
dict.Pos = pos;
if (dict.Word.Contains("一") || dict.Word.Contains("二") ||
dict.Word.Contains("三") || dict.Word.Contains("四") ||
dict.Word.Contains("五") || dict.Word.Contains("六") ||
dict.Word.Contains("七") || dict.Word.Contains("八") ||
dict.Word.Contains("九") || dict.Word.Contains("十"))
{
dict.Pos |= (int)T_POS.POS_A_M;
}
if (dict.Word == "字典")
{
dict.Pos = (int)T_POS.POS_D_N;
}
dictFile.Dicts.Add(dict);
}
return(dictFile);
}
/// <summary>
/// 分词,不屏蔽停用词
/// </summary>
/// <param name="str"></param>
/// <returns></returns>
private List <String> SegmentNoStopWord(String str)
{
List <String> preWords = PreSegment(str);
List <String> retWords = new List <String>();
int index = 0;
while (index < preWords.Count)
{
int next = -1;
foreach (IRule rule in _Rules)
{
if (!_MatchName && rule is MatchName)
{
continue;
}
next = rule.ProcRule(preWords, index, retWords);
if (next > 0)
{
index = next;
break;
}
}
if (next > 0)
{
continue;
}
retWords.Add(preWords[index]);
index++;
}
//return retWords;
List <String> retStrings = RecoverUnknowWord(retWords);
if (AutoStudy)
{
foreach (String word in retStrings)
{
T_DictStruct dict = (T_DictStruct)_ExtractWords.GetTag(word);
if (dict != null)
{
dict.Frequency++;
}
}
}
return(retStrings);
}
protected virtual double GetFreqWeight(List <T_WordInfo> words, List <int> list)
{
double weight = 0;
for (int i = 0; i < list.Count; i++)
{
T_WordInfo w = (T_WordInfo)words[(int)list[i]];
T_DictStruct dict = (T_DictStruct)w.Tag;
weight += dict.Frequency;
}
return(weight);
}
public virtual void DeleteWord(String word)
{
word = word.Trim();
T_DictStruct w = GetWord(word);
if (w == null)
{
return;
}
_DictTbl.Remove(w.Word);
_Dict.Dicts.Remove(w);
}
public virtual void UpdateWord(String word, double frequency, int pos)
{
word = word.Trim();
T_DictStruct w = GetWord(word);
if (w == null)
{
return;
}
w.Frequency = frequency;
w.Pos = pos;
}
public virtual void SaveDict()
{
_MatchNameRule.SaveNameTraffic(_DictPath + "Name.dct");
foreach (T_DictStruct word in _Dict.Dicts)
{
T_DictStruct dict = (T_DictStruct)_ExtractWords.GetTag(word.Word);
if (dict != null)
{
word.Frequency = dict.Frequency;
}
}
Dict.SaveToBinFileEx(_DictPath + "Dict.dct", _Dict);
Dict.SaveToBinFileEx(_DictPath + "UnknownWords.dct", _UnknownWordsDict);
}
private void TrafficUnknownWord(String word, T_POS Pos)
{
if (word.Length <= 1 || word.Length > 3)
{
return;
}
T_DictStruct unknownWord = _UnknownWordsDictMgr.GetWord(word);
if (unknownWord == null)
{
_UnknownWordsDictMgr.InsertWord(word, 1, (int)Pos);
return;
}
//如果是屏蔽的未登录词，则不加入
//屏蔽的未登录词用词性等于0来表示
if (unknownWord.Pos == 0)
{
return;
}
unknownWord.Pos |= (int)Pos;
unknownWord.Frequency++;
if (unknownWord.Frequency > UnknownWordsThreshold && AutoInsertUnknownWords)
{
T_DictStruct w = _DictMgr.GetWord(word);
if (w == null)
{
_DictMgr.InsertWord(word, unknownWord.Frequency, unknownWord.Pos);
_ExtractWords.InsertWordToDfa(word, unknownWord);
_POS.AddWordPos(word, unknownWord.Pos);
}
else
{
w.Pos |= unknownWord.Pos;
w.Frequency += unknownWord.Frequency;
}
unknownWord.Frequency = 0;
}
}
static public T_DictFile LoadFromBinFileEx(string fileName)
{
T_DictFile dictFile = new T_DictFile();
dictFile.Dicts = new List <T_DictStruct>();
File.SetAttributes(fileName, FileAttributes.Normal);
FileStream fs = new FileStream(fileName, FileMode.Open);
byte[] version = new byte[32];
fs.Read(version, 0, version.Length);
String ver = Encoding.UTF8.GetString(version, 0, version.Length);
String verNumStr = CRegex.GetMatch(ver, "KTDictSeg Dict V(.+)", true);
if (verNumStr == null || verNumStr == "")
{
//1.3以前版本
fs.Close();
return(LoadFromBinFile(fileName));
}
while (fs.Position < fs.Length)
{
byte[] buf = new byte[sizeof(int)];
fs.Read(buf, 0, buf.Length);
int length = BitConverter.ToInt32(buf, 0);
buf = new byte[length];
T_DictStruct dict = new T_DictStruct();
fs.Read(buf, 0, buf.Length);
dict.Word = Encoding.UTF8.GetString(buf, 0, length - sizeof(int) - sizeof(double));
dict.Pos = BitConverter.ToInt32(buf, length - sizeof(int) - sizeof(double));
dict.Frequency = BitConverter.ToDouble(buf, length - sizeof(double));
dictFile.Dicts.Add(dict);
}
fs.Close();
return(dictFile);
}
public virtual void InsertWord(String word, double frequency, int pos)
{
word = word.Trim();
if (GetWord(word) != null)
{
return;
}
T_DictStruct w = new T_DictStruct();
w.Word = word;
w.Frequency = frequency;
w.Pos = pos;
_Dict.Dicts.Add(w);
_DictTbl[word] = w;
}
/// <summary>
/// 预分词
/// </summary>
/// <param name="str">要分词的句子</param>
/// <returns>预分词后的字符串输出</returns>
private List <String> PreSegment(String str)
{
ArrayList initSeg = new ArrayList();
if (!CRegex.GetSingleMatchStrings(str, PATTERNS, true, ref initSeg))
{
return(new List <String>());
}
List <String> retWords = new List <String>();
int i = 0;
_ExtractWords.MatchDirection = MatchDirection;
while (i < initSeg.Count)
{
String word = (String)initSeg[i];
if (word == "")
{
word = " ";
}
if (i < initSeg.Count - 1)
{
bool mergeOk = false;
if (((word[0] >= '0' && word[0] <= '9') || (word[0] >= '０' && word[0] <= '９')) &&
((word[word.Length - 1] >= '0' && word[word.Length - 1] <= '9') ||
(word[word.Length - 1] >= '０' && word[word.Length - 1] <= '９'))
)
{
//合并浮点数
word = MergeFloat(initSeg, i, ref i);
mergeOk = true;
}
else if ((word[0] >= 'a' && word[0] <= 'z') ||
(word[0] >= 'A' && word[0] <= 'Z')
)
{
//合并成英文专业名词
String specialEnglish = MergeEnglishSpecialWord(_ExtractWords, initSeg, i, ref i);
if (specialEnglish != null)
{
InsertWordToArray(specialEnglish, retWords);
continue;
}
//合并邮件地址
if ((String)initSeg[i + 1] != "")
{
if (((String)initSeg[i + 1])[0] == '@')
{
word = MergeEmail(initSeg, i, ref i);
mergeOk = true;
}
}
}
if (mergeOk)
{
InsertWordToArray(word, retWords);
continue;
}
}
if (word[0] < 0x4e00 || word[0] > 0x9fa5)
{
//英文或符号，直接加入
InsertWordToArray(word, retWords);
}
else
{
List <T_WordInfo> words = _ExtractWords.ExtractFullTextMaxMatch(word);
int lastPos = 0;
bool lstIsName = false; //前一个词是人名
foreach (T_WordInfo wordInfo in words)
{
if (lastPos < wordInfo.Position)
{
/*
* String unMatchWord = word.Substring(lastPos, wordInfo.Position - lastPos);
*
* InsertWordToArray(unMatchWord, retWords);
*/
//中间有未匹配词，将单个字逐个加入
for (int j = lastPos; j < wordInfo.Position; j++)
{
InsertWordToArray(word[j].ToString(), retWords);
}
}
lastPos = wordInfo.Position + wordInfo.Word.Length;
//统计中文姓名的后缀
if (AutoStudy && lstIsName)
{
T_DictStruct wordDict = (T_DictStruct)wordInfo.Tag;
if ((wordDict.Pos & (int)T_POS.POS_A_NR) == 0)
{
_MatchNameRule.AddBefore(wordInfo.Word);
}
lstIsName = false;
}
//统计中文姓名的前缀
//如总统，主席等
if ((((T_DictStruct)wordInfo.Tag).Pos & (int)T_POS.POS_A_NR) != 0)
{
if (wordInfo.Word.Length > 1 && wordInfo.Word.Length <= 4 && retWords.Count > 0 && AutoStudy && !lstIsName)
{
T_DictStruct wordDict = (T_DictStruct)wordInfo.Tag;
_MatchNameRule.AddBefore(retWords[retWords.Count - 1]);
}
lstIsName = true;
}
InsertWordToArray(wordInfo.Word, retWords);
}
if (lastPos < word.Length)
{
//尾部有未匹配词，将单个字逐个加入
for (int j = lastPos; j < word.Length; j++)
{
InsertWordToArray(word[j].ToString(), retWords);
}
//InsertWordToArray(word.Substring(lastPos, word.Length - lastPos), retWords);
}
}
i++;
}
return(retWords);
}
