internal WordAnalyResult SentenceAnaly()
{
if (ConfigBase.Default.IsTraceStack)
{
LogService.AnyLog("Stack", new StackTrace().GetFrame(0).GetMethod().ToString());
}
WordAnalyResult result = new WordAnalyResult();
stops.Clear();
stops.Add(0);//开始值
//计算停止处
for (var fi = 0; fi < Raw.Length; fi++)
{
//句断符号停止
if (SentenceStopChars.Contains(Raw[fi]))
{
if (Raw[fi] != '.' || Raw[fi] == '.' && (fi == Raw.Length - 1 || Raw[fi++] == ' '))
{
stops.Add(fi);
}
}
}
stops.Add(Raw.Length);//结束值
//截句
for (var fi = 1; fi < stops.Count; fi++)
{
var spi = stops[fi - 1];
if (spi > 0)
{
var splitword = Raw[spi];
//words.Add(splitword.ToString());
spi++;
}
var epi = stops[fi];
var sch = '\x0';
if (epi < Raw.Length)
{
sch = Raw[epi];
//epi++;//包含符号//注释后不包含符号
}
var slen = epi - spi;
if (slen < 1)
{
break;
}
var span = Raw.Slice(spi, slen);
//$"P>{span.ToString()}".WriteInfoLine();
var words = new LinyeeNLWordAnalyzer(span, Enc, WordStopChars).WordAnaly();
var res = new Sentence(words, sch);
sentences.Add(res);;
}
//
return(result.SetData(sentences).SetOk());
}
internal WordAnalyResult ParagraphAnaly()
{
if (ConfigBase.Default.IsTraceStack)
{
LogService.AnyLog("Stack", new StackTrace().GetFrame(0).GetMethod().ToString());
}
WordAnalyResult result = new WordAnalyResult();
stops.Clear();
stops.Add(0);//开始值
//计算停止处
for (var fi = 0; fi < Raw.Length; fi++)
{
//段落符号停止
if (ParagraphStopChars.Contains(Raw[fi]))
{
stops.Add(fi);
}
}
stops.Add(Raw.Length);//结束值
//截段
for (var fi = 1; fi < stops.Count; fi++)
{
var spi = stops[fi - 1];
if (spi > 0)
{
var splitword = Raw[spi];
//words.Add(splitword.ToString());
spi++;
}
var epi = stops[fi];
//if (epi >= Raw.Length && paragraphs.Count > 0)
//{
// break;//退出
//}
//else if(epi< Raw.Length)
//{
// epi++;
//}
var slen = epi - spi;
if (slen < 1)
{
break;
}
var span = Raw.Slice(spi, slen);
paragraphs.Add(new Paragraph(new LinyeeNLSentenceAnalyzer(span, Enc, SentenceStopChars, WordStopChars).SentenceAnaly()));
}
//
return(result.SetData(paragraphs).SetOk());
}
internal WordAnalyResult <List <WordInfoOnce> > WordAnaly()
{
if (ConfigBase.Default.IsTraceStack)
{
LogService.AnyLog("Stack", new StackTrace().GetFrame(0).GetMethod().ToString());
}
WordAnalyResult <List <WordInfoOnce> > result = new WordAnalyResult <List <WordInfoOnce> >();
var winfos = WordDict.WordAnaly(Raw, 0);
words.AddRange(winfos);
//
return(result.SetData(words).SetOk());
}
public WordAnalyResult AddNew(string key, WordInfoOnce value)
{
WordAnalyResult result = new WordAnalyResult();
if (ContainsKey(key))
{
return(result.SetFail("已经存在"));//已有
}
else
{
base.Add(key, value);
savetimes++;
return(result.SetOk());
}
}
internal WordAnalyResult <List <string> > WordAnaly()
{
if (ConfigBase.Default.IsTraceStack)
{
LogService.AnyLog("Stack", new StackTrace().GetFrame(0).GetMethod().ToString());
}
WordAnalyResult <List <string> > result = new WordAnalyResult <List <string> >();
#region 标点分词
stops.Clear();
stops.Add(0);//开始值
//计算停止处
for (var fi = 0; fi < Raw.Length; fi++)
{
//标点符号停止
if (StopChars.Contains(Raw[fi]))
{
stops.Add(fi);
}
}
stops.Add(Raw.Length);//结束值
//截词
for (var fi = 1; fi < stops.Count; fi++)
{
var spi = stops[fi - 1];
if (spi > 0)
{
var splitword = Raw[spi];
words.Add(splitword.ToString());
spi++;
}
var epi = stops[fi];
//if (epi >= Raw.Length && words.Count>0)
//{
// break;//退出
//}
var wlen = epi - spi;
if (wlen < 1)
{
break;
}
var word = Raw.Slice(spi, wlen);
var wordstr = word.ToString();
//词中是否含有中文
if (wordstr.ContainChinese())
{
try
{
var winfos = new LinyeeNLZHWordAnalyzer(word, Enc, LinyeeWrodDict.Default).WordAnaly();
if (winfos.IsOK)
{
words.AddRange(winfos.Data.Select(p => p.w));
}
}
catch (Exception ex)
{
ex.ToString().WriteErrorLine();
}
}
else
{
words.Add(wordstr);
var info = new WordInfoOnce(wordstr);
}
}
#endregion
//
return(result.SetData(words).SetOk());
}