internal static void Test(System.IO.TextReader reader, bool verbose, long bytes)
{
Analyzer analyzer = new SimpleAnalyzer();
TokenStream stream = analyzer.TokenStream(null, reader);
System.DateTime start = System.DateTime.Now;
int count = 0;
for (Token t = stream.Next(); t != null; t = stream.Next())
{
if (verbose)
{
System.Console.Out.WriteLine("Text=" + t.TermText() + " start=" + t.StartOffset() + " end=" + t.EndOffset());
}
count++;
}
System.DateTime end = System.DateTime.Now;
long time = end.Ticks - start.Ticks;
System.Console.Out.WriteLine(time + " milliseconds to extract " + count + " tokens");
System.Console.Out.WriteLine((time * 1000.0) / count + " microseconds/token");
System.Console.Out.WriteLine((bytes * 1000.0 * 60.0 * 60.0) / (time * 1000000.0) + " megabytes/hour");
}
private Directory MakeIndex()
{
Directory dir = new RAMDirectory();
try
{
System.Random r = NewRandom();
Analyzer analyzer = new SimpleAnalyzer();
IndexWriter writer = new IndexWriter(dir, analyzer, true, IndexWriter.MaxFieldLength.LIMITED);
writer.UseCompoundFile = false;
for (int d = 1; d <= NUM_DOCS; d++)
{
Document doc = new Document();
for (int f = 1; f <= NUM_FIELDS; f++)
{
doc.Add(new Field("f" + f, data[f % data.Length] + '#' + data[r.Next(data.Length)], Field.Store.YES, Field.Index.ANALYZED));
}
writer.AddDocument(doc);
}
writer.Close();
}
catch (System.Exception e)
{
throw new System.SystemException("", e);
}
return(dir);
}
public static Lucene.Net.Analysis.Analyzer GetAnalyzerByName(string analyzerName)
{
Lucene.Net.Analysis.Analyzer result;
Lucene.Net.Util.Version AppLuceneVersion = Lucene.Net.Util.Version.LUCENE_30;
switch (analyzerName)
{
case "SimpleAnalyzer":
result = new Lucene.Net.Analysis.SimpleAnalyzer();
break;
case "StandardAnalyzer":
result = new Lucene.Net.Analysis.Standard.StandardAnalyzer(AppLuceneVersion);
break;
case "KeywordAnalyzer":
result = new Lucene.Net.Analysis.KeywordAnalyzer();
break;
case "StopAnalyzer":
result = new Lucene.Net.Analysis.StopAnalyzer(AppLuceneVersion);
break;
case "WhitespaceAnalyzer":
result = new Lucene.Net.Analysis.WhitespaceAnalyzer();
break;
default:
result = new Lucene.Net.Analysis.SimpleAnalyzer();
break;
}
return(result);
}
private Directory MakeIndex()
{
Directory dir = new RAMDirectory();
try
{
System.Random r = NewRandom();
Analyzer analyzer = new SimpleAnalyzer();
IndexWriter writer = new IndexWriter(dir, analyzer, true, IndexWriter.MaxFieldLength.LIMITED);
writer.SetUseCompoundFile(false);
for (int d = 1; d <= NUM_DOCS; d++)
{
Document doc = new Document();
for (int f = 1; f <= NUM_FIELDS; f++)
{
doc.Add(new Field("f" + f, data[f % data.Length] + '#' + data[r.Next(data.Length)], Field.Store.YES, Field.Index.ANALYZED));
}
writer.AddDocument(doc);
}
writer.Close();
}
catch (System.Exception e)
{
throw new System.SystemException("", e);
}
return dir;
}
private static Directory MakeIndex()
{
Directory dir = new RAMDirectory();
try
{
System.Random r = new System.Random((System.Int32) (BASE_SEED + 42));
Analyzer analyzer = new SimpleAnalyzer();
IndexWriter writer = new IndexWriter(dir, analyzer, true);
writer.SetUseCompoundFile(false);
for (int d = 1; d <= NUM_DOCS; d++)
{
Lucene.Net.Documents.Document doc = new Lucene.Net.Documents.Document();
for (int f = 1; f <= NUM_FIELDS; f++)
{
doc.Add(new Field("f" + f, data[f % data.Length] + '#' + data[r.Next(data.Length)], Field.Store.YES, Field.Index.TOKENIZED));
}
writer.AddDocument(doc);
}
writer.Close();
}
catch (System.Exception e)
{
throw new System.SystemException("", e);
}
return dir;
}
public virtual QueryParsers.QueryParser GetParser(Analyzer a)
{
if (a == null)
{
a = new SimpleAnalyzer();
}
QueryParsers.QueryParser qp = new QueryParsers.QueryParser("Field", a);
qp.SetOperator(QueryParsers.QueryParser.DEFAULT_OPERATOR_OR);
return(qp);
}
public virtual Query GetQueryDOA(System.String query, Analyzer a)
{
if (a == null)
{
a = new SimpleAnalyzer();
}
QueryParsers.QueryParser qp = new QueryParsers.QueryParser("Field", a);
qp.SetOperator(QueryParsers.QueryParser.DEFAULT_OPERATOR_AND);
return(qp.Parse(query));
}
public static void IndexDoc(System.String segment, System.String fileName)
{
Directory directory = FSDirectory.GetDirectory("test", false);
Analyzer analyzer = new SimpleAnalyzer();
DocumentWriter writer = new DocumentWriter(directory, analyzer, Similarity.GetDefault(), 1000);
System.IO.FileInfo file = new System.IO.FileInfo(fileName);
Document doc = Lucene.Net.Demo.FileDocument.Document(file);
writer.AddDocument(segment, doc);
directory.Close();
}
private void IndexDoc(System.String segment, System.String fileName)
{
Directory directory = FSDirectory.GetDirectory(indexDir, false);
Analyzer analyzer = new SimpleAnalyzer();
DocumentWriter writer = new DocumentWriter(directory, analyzer, Similarity.GetDefault(), 1000);
System.IO.FileInfo file = new System.IO.FileInfo(workDir.FullName + "\\" + fileName);
Document doc = FileDocument.Document(file);
writer.AddDocument(segment, doc);
directory.Close();
}
public static void IndexDoc(System.String segment, System.String fileName)
{
Directory directory = FSDirectory.GetDirectory("test", false);
Analyzer analyzer = new SimpleAnalyzer();
DocumentWriter writer = new DocumentWriter(directory, analyzer, Similarity.GetDefault(), 1000);
System.IO.FileInfo file = new System.IO.FileInfo(fileName);
Document doc = Lucene.Net.Demo.FileDocument.Document(file);
writer.AddDocument(segment, doc);
directory.Close();
}
internal static void Test(System.IO.TextReader reader, bool verbose, long bytes)
{
Analyzer analyzer = new SimpleAnalyzer();
TokenStream stream = analyzer.TokenStream(null, reader);
System.DateTime start = System.DateTime.Now;
int count = 0;
for (Token t = stream.Next(); t != null; t = stream.Next())
{
if (verbose)
{
System.Console.Out.WriteLine("Text=" + t.TermText() + " start=" + t.StartOffset() + " end=" + t.EndOffset());
}
count++;
}
System.DateTime end = System.DateTime.Now;
long time = end.Ticks - start.Ticks;
System.Console.Out.WriteLine(time + " milliseconds to extract " + count + " tokens");
System.Console.Out.WriteLine((time * 1000.0) / count + " microseconds/token");
System.Console.Out.WriteLine((bytes * 1000.0 * 60.0 * 60.0) / (time * 1000000.0) + " megabytes/hour");
}
private void IndexDoc(System.String segment, System.String fileName)
{
Directory directory = FSDirectory.GetDirectory(indexDir, false);
Analyzer analyzer = new SimpleAnalyzer();
DocumentWriter writer = new DocumentWriter(directory, analyzer, Similarity.GetDefault(), 1000);
System.IO.FileInfo file = new System.IO.FileInfo(workDir.FullName + "\\" + fileName);
Document doc = FileDocument.Document(file);
writer.AddDocument(segment, doc);
directory.Close();
}
public virtual QueryParsers.QueryParser GetParser(Analyzer a)
{
if (a == null)
a = new SimpleAnalyzer();
QueryParsers.QueryParser qp = new QueryParsers.QueryParser("Field", a);
qp.SetOperator(QueryParsers.QueryParser.DEFAULT_OPERATOR_OR);
return qp;
}
public virtual Query GetQueryDOA(System.String query, Analyzer a)
{
if (a == null)
a = new SimpleAnalyzer();
QueryParsers.QueryParser qp = new QueryParsers.QueryParser("Field", a);
qp.SetOperator(QueryParsers.QueryParser.DEFAULT_OPERATOR_AND);
return qp.Parse(query);
}
public virtual QueryParser GetParser(Analyzer a)
{
if (a == null)
a = new SimpleAnalyzer();
QueryParser qp = new QueryParser(Version.LUCENE_CURRENT, "field", a);
qp.DefaultOperator = QueryParser.OR_OPERATOR;
return qp;
}
public virtual Query GetQueryDOA(System.String query, Analyzer a)
{
if (a == null)
a = new SimpleAnalyzer();
QueryParser qp = new QueryParser("field", a);
qp.SetDefaultOperator(QueryParser.AND_OPERATOR);
return qp.Parse(query);
}
public virtual Query GetQueryDOA(System.String query, Analyzer a)
{
if (a == null)
a = new SimpleAnalyzer();
QueryParser qp = new QueryParser(Version.LUCENE_CURRENT, "field", a);
qp.DefaultOperator = QueryParser.AND_OPERATOR;
return qp.Parse(query);
}
public virtual QueryParser GetParser(Analyzer a)
{
if (a == null)
a = new SimpleAnalyzer();
QueryParser qp = new QueryParser("field", a);
qp.SetDefaultOperator(QueryParser.OR_OPERATOR);
return qp;
}
private void lblSearch_Click(object sender, EventArgs e)
{
//Check the saving path
if (String.IsNullOrEmpty(txt_Saving.Text))
{
MessageBox.Show("Please choose a path to save results.");
return;
}
if (String.IsNullOrEmpty(txt_FileName.Text))
{
MessageBox.Show("Please input a file name.");
return;
}
// Check is existing or new file.
String filePath = String.Format("{0}\\{1}.txt", txt_Saving.Text, txt_FileName.Text);
if (File.Exists(filePath))
{
if (MessageBox.Show("The file is existing, do you want to append text into the file? (Yes: append it, No: Set a new file name.)",
"Confirm", MessageBoxButtons.YesNo) == DialogResult.Yes)
{
// User wanna append it, so get the Topic ID from document...
var lastLine = File.ReadLines(filePath).Count() != 0 ? File.ReadLines(filePath).Last() : "000";
if (!String.IsNullOrEmpty(lastLine))
{
var firstword = lastLine.ToString().Split(' ')[0];
try
{
TopicID = Convert.ToInt32(firstword) + 1;
}
catch (Exception ex) {
MessageBox.Show("Fail appending! The format in the file is wrong. Please set a new file name.");
return;
}
}
}
else
{
return;
}
}
btn_SaveBrowse.Visible = false;
txt_Saving.ReadOnly = true;
txt_FileName.ReadOnly = true;
// Calculate the Process Time
Lucene.Net.Analysis.Analyzer analyzer;
if (ckb_Analzer.Checked)
{
analyzer = new Lucene.Net.Analysis.SimpleAnalyzer();
}
else
{
analyzer = generator.analyzer;
}
Stopwatch stopWatch = new Stopwatch();
stopWatch.Start();
Searcher searcher = new Searcher(txt_IndexPath.Text, analyzer, generator.writer);
searcher.CreateSearcher();
searcher.CreateParser(ddl_Type.SelectedItem.ToString(), ddl_Fields.SelectedItem.ToString());
// Searching and generate result
Dictionary <string, float> resultList = searcher.DisplayResults(searcher.SearchIndex(txt_InformationNeeds.Text), generator.collectionList);
lbl_QueryText.Text = searcher.query.ToString();
ResultCollectionList =
(from c in generator.collectionList
join r in resultList on c.DocID equals r.Key
select new Collection
{
DocID = c.DocID,
Title = c.Title,
Author = c.Author,
Bibliographic = c.Bibliographic,
Words = c.Words,
Score = r.Value
}).OrderByDescending(n => n.Score).ToList();
stopWatch.Stop();
long ts = stopWatch.ElapsedMilliseconds;
lbl_SearchingTime.Text = ts.ToString() + " ms";
// Set columns to listview
GenerateListView();
// Set the total number of documents
total = resultList.Count();
lblTotalDocs.Text = total + " docs";
// Saving
using (System.IO.StreamWriter file =
new System.IO.StreamWriter(filePath, true))
{
int rank = 1;
foreach (Collection c in ResultCollectionList)
{
string TopicIDString = string.Format("{0:000}", TopicID);
file.NewLine = "\n";
file.WriteLine(String.Format("{0} {1} {2} {3} {4} {5}", TopicIDString, "Q0", c.DocID, rank.ToString(), c.Score, "9913661_9913351_10032711_RetrievalHero"));
rank = rank + 1;
}
TopicID++;
}
}
