public override TokenStream TokenStream(string fieldName, System.IO.TextReader reader) { TokenStream result = new PanGuTokenizer(reader, _OriginalResult); result = new LowerCaseFilter(result); return(result); }
public void ContainAnalyzerTest() { LuceneEngine engine = new LuceneEngine(); engine.Init(); //string text = "三国演义"; //string text = "a|f|g|i|m"; string text = "NBA常规赛-快船vs凯尔特人"; TextReader tr = new StringReader(text); PanGuTokenizer ct = new PanGuTokenizer(tr); int end = 0; Lucene.Net.Analysis.Token t; string ss = string.Empty; while (end < text.Length) { t = ct.Next(); end = t.EndOffset(); ss = ss + t.TermText() + "/ "; } string xxx = ss; }
public void SpellTest() { LuceneEngine engine = new LuceneEngine(); engine.Init(); var firsts = SpellUtils.GetCnSegment("NBA常规赛-快船vs凯尔特人"); var spells = SpellUtils.GetSpellSegment("战重警和"); var segment = new Segment(); var collection = segment.DoSegment("国际足球100509K联赛釜山-大田"); var list = new List<string>(); foreach (WordInfo word in collection) { if (word == null) continue; list.Add(word.Word); } //var list = SpellUtils.GetSpellSegment("战警"); string sss = "san国yan义"; string ssss = Synacast.LuceneNetSearcher.Searcher.Searcher.SegmentKeyWord(sss); string text = "重庆"; TextReader tr = new StringReader(text); PanGuTokenizer ct = new PanGuTokenizer(tr); int end = 0; Lucene.Net.Analysis.Token t; string ss = string.Empty; while (end < text.Length) { t = ct.Next(); end = t.EndOffset(); ss = ss + t.TermText() + "/ "; } }
public override TokenStream TokenStream(string fieldName, System.IO.TextReader reader) { TokenStream result = new PanGuTokenizer(reader, _OriginalResult); result = new LowerCaseFilter(result); return result; }