public void CreateIndexFromDirectory()
{
string DirecotryPath = outputDirectoryName == null
? Path.GetDirectoryName(Assembly.GetEntryAssembly().Location)
: Path.GetDirectoryName(outputDirectoryName);
var directory = FSDirectory.Open(DirecotryPath);
Analyzer analyzer = new TermAnalyzer();
IndexWriter writer =
new IndexWriter(directory, analyzer, true,
IndexWriter.MaxFieldLength.UNLIMITED);
var files = FileHelper.GetFilesFromRelativePath(inputDirectoryName);
writer.DeleteAll();
foreach (var fileInfo in files)
{
//var detailedPath = DirecotryPath + "\\" + fileInfo.Name;
//var detailedDirectory = FSDirectory.Open(detailedPath);
//IndexWriter detailedlWriter = new IndexWriter(detailedDirectory, analyzer, true,
// IndexWriter.MaxFieldLength.UNLIMITED);
Document doc = new Document();
Document detailedDocument = new Document();
var path = fileInfo.Directory + "\\" + fileInfo.Name;
var reader = new StreamReader(path);
doc.Add(new Field("content", reader, Field.TermVector.WITH_POSITIONS));
writer.AddDocument(doc);
reader.Close();
reader = new StreamReader(path);
detailedDocument.Add(new Field(fileInfo.Name, reader, Field.TermVector.WITH_POSITIONS));
//detailedlWriter.AddDocument(detailedDocument);
//detailedlWriter.Optimize();
//detailedlWriter.Commit();
}
writer.Optimize();
writer.Commit();
}
public List<ArticleInfo> SearchForString(string stringQuery)
{
var files = FileHelper.GetFilesFromRelativePath(inputDirectoryName);
string DirecotryPath = outputDirectoryName == null
? Path.GetDirectoryName(Assembly.GetEntryAssembly().Location)
: Path.GetDirectoryName(outputDirectoryName);
var dir = FSDirectory.Open(DirecotryPath);
IndexReader reader = DirectoryReader.Open(dir, true);
IndexSearcher indexSearcher = new IndexSearcher(reader);
Analyzer analyzer = new TermAnalyzer();
QueryParser queryParser = new QueryParser(Version.LUCENE_30, "content", analyzer);
Query query = queryParser.Parse(stringQuery);
Stopwatch searchStopwatch = new Stopwatch();
searchStopwatch.Start();
var hits = indexSearcher.Search(query, null, 100000);
searchStopwatch.Stop();
this.LuceneTime = searchStopwatch.Elapsed;
List<ArticleInfo> articles = new List<ArticleInfo>();
hits.ScoreDocs.ToList().ForEach(s =>
{
var y = reader.GetTermFreqVector(s.Doc, "content");
ArticleInfo info = new ArticleInfo()
{
FileInfo = files[s.Doc],
Rank = s.Score,
//Terms = y.GetTerms().ToList()
};
articles.Add(info);
});
return articles;
}