private void AddPdfToIndexe(string fileName)
{
Task t = new Task(() =>
{
var Pdfdirectory = "C:\\SearchMe";
var pathPdf = Pdfdirectory + "\\" + Guid.NewGuid().ToString() + ".txt";
if (!Directory.Exists(Pdfdirectory))
{
Directory.CreateDirectory(Pdfdirectory);
}
//should change to application startup path
if (!File.Exists(pathPdf))
{
File.Create(pathPdf).Dispose();
}
Utility.WritePdfContentToTxtFile(Pdfdirectory, fileName, pathPdf);
DataForIndex dfi = Utility.CreatePdfIndexedDocument(fileName, pathPdf);
LuceneBussines lucene = new LuceneBussines();
lucene.CreateIndex(dfi);
IndexDocCount++;
LableTextChange(DocumentCountLable, IndexDocCount.ToString());
});
t.Start();
}
public static DataForIndex CreateSoundIndexDoucment(string fn)
{
DataForIndex dfi = new DataForIndex();
try
{
var file = TagLib.File.Create(fn);
dfi.ID = new Random().Next(int.MaxValue);
dfi.Label = "Music";
dfi.FileExtension = Path.GetExtension(fn);
dfi.AudioDuration = file.Properties.Duration.ToString();
dfi.AudioBitrate = file.Properties.AudioBitrate.ToString();
foreach (var item in file.Tag.Genres)
{
dfi.AudioGenre += item + " , ";
}
dfi.AudioAlbum = file.Tag.Album;
dfi.FileName = fn;
dfi.Body = fn + " , " + dfi.AudioGenre + " , " + dfi.AudioDuration + " , " + dfi.AudioAlbum;
}
catch (Exception ex)
{
FileInfo fi = new FileInfo(fn);
fi.GetType();
}
return(dfi);
}
public void indexsite(DataForIndex LuceneForSite, string url, string sitetxt)
{
dfi2.ID = new Random().Next(int.MaxValue);
dfi2.FileName = url;
dfi2.Body = Utility.RemoveHtmlTags(sitetxt);
Thread t = new Thread(GoForIndexSite);
t.Start();
}
public static DataForIndex CreateDocumentIndex(FilterReader reader, string fileName)
{
var dfi = new DataForIndex();
dfi.ID = new Random().Next(int.MaxValue);
dfi.Label = "Docs";
dfi.FileExtension = Path.GetExtension(fileName);
dfi.FileName = fileName;
dfi.Body = reader.ReadToEnd();
return(dfi);
}
private void AddSoundToIndexed(string fn)
{
Task t = new Task(() =>
{
DataForIndex dfi = Utility.CreateSoundIndexDoucment(fn);
LuceneBussines lucene = new LuceneBussines();
lucene.CreateIndex(dfi);
IndexMp3Count++;
LableTextChange(CountFileIndexedLabel, IndexMp3Count.ToString());
});
t.Start();
}
private void AddVideoToIndexed(string fn)
{
Task t = new Task(() =>
{
DataForIndex dfi = Utility.CreateVideoIndexDoucment(fn);
LuceneBussines lucene = new LuceneBussines();
lucene.CreateIndex(dfi);
IndexMovieCount++;
LableTextChange(Videolblcount, IndexMovieCount.ToString());
});
t.Start();
}
private void AddDocumentToIndexed(string fileName, FilterReaderTimeout timeoutOption)
{
Task t = new Task(() =>
{
try
{
var reader = new FilterReader(fileName,
string.Empty,
disableEmbeddedContent: false,
includeProperties: false,
readIntoMemory: false,
filterReaderTimeout: timeoutOption,
timeout: -1);
DataForIndex dfi = Utility.CreateDocumentIndex(reader, fileName);
LuceneBussines lucene = new LuceneBussines();
lucene.CreateIndex(dfi);
IndexDocCount++;
LableTextChange(DocumentCountLable, IndexDocCount.ToString());
}
catch (Exception ex)
{
try
{
var reader = new FilterReader(fileName,
string.Empty,
disableEmbeddedContent: false,
includeProperties: false,
readIntoMemory: true,
filterReaderTimeout: FilterReaderTimeout.TimeoutWithException,
timeout: 5000);
if (reader == null)
{
return;
}
;
DataForIndex dfi = Utility.CreateDocumentIndex(reader, fileName);
LuceneBussines lucene = new LuceneBussines();
lucene.CreateIndex(dfi);
IndexDocCount++;
LableTextChange(DocumentCountLable, IndexDocCount.ToString());
}
catch (Exception)
{
}
}
});
t.Start();
}
public static DataForIndex CreatePdfIndexedDocument(string fileName, string pathPdf)
{
var dfi = new DataForIndex();
dfi.ID = new Random().Next(int.MaxValue);
dfi.Label = "Docs";
dfi.FileName = fileName;
dfi.FileExtension = Path.GetExtension(fileName);
StreamReader sr = new StreamReader(pathPdf);
dfi.Body = sr.ReadToEnd();
sr.Close();
sr.Dispose();
File.Delete(pathPdf);
return(dfi);
}
public static DataForIndex CreateVideoIndexDoucment(string fn)
{
DataForIndex dfi = new DataForIndex();
try
{
var file = new FileInfo(fn);
dfi.ID = new Random().Next(int.MaxValue);
dfi.Label = "Music";
dfi.FileExtension = Path.GetExtension(fn);
dfi.FileName = fn;
dfi.Body = fn + " , " + dfi.FileExtension + " , " + dfi.Label;
}
catch
{
}
return(dfi);
}
public void CreateIndexSite(DataForIndex dfi)
{
sem.WaitOne();
var directory = FSDirectory.Open(new DirectoryInfo(Environment.CurrentDirectory + "\\LuceneIndexSite"));
using (
var writer = new IndexWriter(directory, Analyzer, create: false,
mfl: IndexWriter.MaxFieldLength.UNLIMITED))
{
lock (writer)
{
Document doc = new Document();
doc.Add(new Field("ID", dfi.ID.ToString(), Field.Store.YES, Field.Index.NO, Field.TermVector.NO));
doc.Add(new Field("FileName", dfi.FileName ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.NO));
doc.Add(new Field("Body", dfi.Body ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.WITH_POSITIONS_OFFSETS));
doc.Add(new Field("FileExtension", dfi.FileExtension ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.YES));
doc.Add(new Field("AudioGenre", dfi.AudioGenre ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.YES));
doc.Add(new Field("Label", dfi.Label ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.YES));
doc.Add(new Field("AudioAlbum", dfi.AudioAlbum ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.YES));
doc.Add(new Field("AudioBitrate", dfi.AudioBitrate ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.YES));
doc.Add(new Field("AudioDuration", dfi.AudioDuration ?? "-", Field.Store.YES, Field.Index.ANALYZED,
Field.TermVector.YES));
//sem.WaitOne();
writer.AddDocument(doc);
CountIndex++;
//sem.Release();
//writer.Optimize();
writer.Commit();
writer.Dispose();
}
}
sem.Release();
}
public static void CreateIndexForStatistics(DataForIndex dfi)
{
FileInfo file = new FileInfo("stopwords.txt");
var directory = FSDirectory.Open(new DirectoryInfo(Environment.CurrentDirectory + "\\StatisticsIndex"));
var analyzer = new StandardAnalyzer(_version, file);
using (var writer = new IndexWriter(directory, analyzer, create: false, mfl: IndexWriter.MaxFieldLength.UNLIMITED))
{
var section = string.Empty;
foreach (var line in File.ReadAllLines(dfi.FileName))
{
Document postDocument = new Document();
postDocument.Add(new Field("Id", new Random().Next().ToString(), Field.Store.YES, Field.Index.NOT_ANALYZED));
postDocument.Add(new Field("Body", line, Field.Store.YES, Field.Index.ANALYZED, Field.TermVector.WITH_POSITIONS_OFFSETS));
writer.AddDocument(postDocument);
section = string.Empty;
}
writer.Optimize();
writer.Commit();
writer.Dispose();
}
}
private bool ShowResultinRichtxtsites(DataForIndex finddfi, string item, string searchvalue)
{
int index = item.IndexOf(searchvalue, StringComparison.Ordinal);
if (index != -1)
{
if (finddfi.IsFarsiArabic == true)
{
string term = searchvalue;
if (!finddfi.HarfArabic.Contains("#"))
{
term = term.Replace(finddfi.HarfArabic, finddfi.HarfFarsi);
index = item.IndexOf(term);
}
}
if (index >= 0)
{
RichtxtResultSite.AppendText(item.Substring(0, index));
RichtxtResultSite.SelectionBackColor = Color.Yellow;
Font fonttext = RichtxtResult.SelectionFont;
RichtxtResultSite.SelectionFont = new Font("Tahoma", 12, FontStyle.Bold);
RichtxtResultSite.AppendText(item.Substring(index, searchvalue.Length));
RichtxtResultSite.SelectionBackColor = Color.White;
RichtxtResultSite.SelectionFont = new Font(fonttext, FontStyle.Regular);
RichtxtResultSite.AppendText(
item.Substring(index + searchvalue.Length, item.Length - (index + searchvalue.Length)) +
Environment.NewLine + "//////////////////" + Environment.NewLine);
return(true);
}
}
else
{
return(false);
}
return(false);
}
private static List <DataForIndex> getResult(List <DataForIndex> dfi, ScoreDoc[] hits, string term, IndexSearcher searcher, Query query, MultiFieldQueryParser parser)
{
if (hits.Length == 0)
{
term = searchByPartialWords(term);
query = parseQuery(term, parser);
hits = searcher.Search(query, 100).ScoreDocs;
}
foreach (var scoreDoc in hits)
{
var doc = searcher.Doc(scoreDoc.Doc);
var score = scoreDoc.Score;
DataForIndex listdata = new DataForIndex();
listdata.ID = int.Parse(doc.Get("ID"));
listdata.FileName = doc.Get("FileName");
listdata.SearchWord = term;
listdata.FileExtension = doc.Get("FileExtension");
listdata.AudioGenre = doc.Get("AudioGenre");
listdata.AudioAlbum = doc.Get("AudioAlbum");
listdata.AudioBitrate = doc.Get("AudioBitrate");
listdata.AudioDuration = doc.Get("AudioDuration");
listdata.Label = doc.Get("Label");
if (isFarsiArabic)
{
listdata.IsFarsiArabic = true;
listdata.HarfArabic = HarfArabic;
listdata.HarfFarsi = HarfFarsi;
}
listdata.Score = score;
TermPositionVector obj_vector =
(TermPositionVector)searcher.IndexReader.GetTermFreqVector(scoreDoc.Doc, "Body");
int int_phraseIndex = obj_vector.IndexOf(term.Split(' ').FirstOrDefault());
TermVectorOffsetInfo[] obj_offsetInfo = obj_vector.GetOffsets(int_phraseIndex);
StringBuilder text = new StringBuilder();
for (int i = 0; i < obj_offsetInfo.Length; i++)
{
string body = doc.Get("Body");
int start = obj_offsetInfo[i].StartOffset;
int end = body.Length;
int count = 100;
if (start + count <= end)
{
end = start + count;
}
if (start > count)
{
start = start - count;
}
else
{
start = 0;
}
text.Append(body.Substring(start, end - start) + " # ");
}
listdata.ResultText = text.ToString();
if (dfi.FirstOrDefault(x => x.FileName == listdata.FileName) == null)
{
dfi.Add(listdata);
}
else
{
var del = dfi.FirstOrDefault(x => x.FileName == listdata.FileName);
dfi.Remove(del);
del.SearchWord = del.SearchWord + " + " + listdata.SearchWord;
del.ResultText = del.ResultText + " " + listdata.ResultText;
dfi.Add(del);
}
}
return(dfi);
}
