private void btnExecuteSearch_Click(object sender, EventArgs e)
{
Directory indexDirectory = FSDirectory.Open(new System.IO.DirectoryInfo(tempPath));
IndexSearcher searcher = new IndexSearcher(indexDirectory, true); // read-only=true
QueryParser qp = new HebrewQueryParser(Lucene.Net.Util.Version.LUCENE_29, "content", analyzer);
qp.DefaultOperator = QueryParser.Operator.AND;
Query query = qp.Parse(txbSearchQuery.Text);
ScoreDoc[] hits = searcher.Search(query, null, 1000).ScoreDocs;
// Iterate through the results:
var l = new BindingList <SearchResult>();
for (int i = 0; i < hits.Length; i++)
{
Document hitDoc = searcher.Doc(hits[i].Doc);
var sr = new SearchResult(hitDoc.GetField("title").StringValue,
hitDoc.GetField("path").StringValue, hits[i].Score);
l.Add(sr);
}
searcher.Close();
indexDirectory.Close();
dgvResults.DataSource = l;
}
public void ParsesAcronymsCorrectly()
{
QueryParser qp = new HebrewQueryParser(LuceneUtil.Version.LUCENE_29, "f", new global::Lucene.Net.Analysis.Hebrew.SimpleAnalyzer());
qp.Parse(@"צה""ל");
qp.Parse(@"""צהל""");
qp.Parse(@"כל הכבוד לצה""ל");
qp.Parse(@"""כל הכבוד לצה""ל""");
qp.Parse(@"""כל הכבוד"" לצה""ל");
qp.Parse(@"מנכ""לית");
try
{
qp.Parse(@"צה""""ל");
qp.Parse(@"""צה""ל");
Assert.Fail("Expected exception was not thrown");
}
catch(ParseException) { }
}
public void ParsesAcronymsCorrectly()
{
QueryParser qp = new HebrewQueryParser(LuceneUtil.Version.LUCENE_30, "f", new global::Lucene.Net.Analysis.Hebrew.SimpleAnalyzer());
qp.Parse(@"צה""ל");
qp.Parse(@"""צהל""");
qp.Parse(@"כל הכבוד לצה""ל");
qp.Parse(@"""כל הכבוד לצה""ל""");
qp.Parse(@"""כל הכבוד"" לצה""ל");
qp.Parse(@"מנכ""לית");
try
{
qp.Parse(@"צה""""ל");
qp.Parse(@"""צה""ל");
Assert.True(false, "Expected exception was not thrown");
}
catch (ParseException) { }
}
private void RunQuery(string query, int expectedPosition)
{
var hqp = new HebrewQueryParser(global::Lucene.Net.Util.Version.LUCENE_29, "Text", analyzer);
Query q = hqp.Parse(query);
TopDocs td = searcher.Search(q, 10000);
int num = td.ScoreDocs[0].Doc;
var tf = searcher.IndexReader.GetTermFreqVectors(num)[0];
var tp = (TermPositionVector)tf;
var trms_list = new SortedSet <Term>();
q.ExtractTerms(trms_list);
foreach (var t in trms_list)
{
int[] pos = tp.GetTermPositions(tp.IndexOf(t.Text));
TermVectorOffsetInfo[] off = tp.GetOffsets(tp.IndexOf(t.Text));
AssertSinglePositionExists(pos, expectedPosition);
/*
* string sPos = "";
* string sOff = "";
* foreach (int p in pos)
* {
* sPos += " " + p;
* }
* foreach (TermVectorOffsetInfo o in off)
* {
* sOff += " (" + o.GetStartOffset() + "," + o.GetEndOffset() + ")";
* }
* Trace.WriteLine(string.Format("Term: {0} Pos:{1} Off:{2}", t.Text(), sPos, sOff));
*/
}
}
