public List <string> GetDocumentsSimilarToDocument(string document_filename)
{
List <string> fingerprints = new List <string>();
IndexReader index_reader = IndexReader.Open(LIBRARY_INDEX_BASE_PATH, true);
Searcher index_searcher = new IndexSearcher(index_reader);
LuceneMoreLikeThis mlt = new LuceneMoreLikeThis(index_reader);
mlt.SetFieldNames(new string[] { "content" });
mlt.SetMinTermFreq(0);
Query query = mlt.Like(new StreamReader(document_filename));
Hits hits = index_searcher.Search(query);
var i = hits.Iterator();
while (i.MoveNext())
{
Hit hit = (Hit)i.Current;
string fingerprint = hit.Get("fingerprint");
fingerprints.Add(fingerprint);
}
// Close the index
index_searcher.Close();
index_reader.Close();
return(fingerprints);
}
public List <string> GetDocumentsSimilarToDocument(string document_filename)
{
List <string> fingerprints = new List <string>();
try
{
using (IndexReader index_reader = IndexReader.Open(LIBRARY_INDEX_BASE_PATH, true))
{
using (IndexSearcher index_searcher = new IndexSearcher(index_reader))
{
LuceneMoreLikeThis mlt = new LuceneMoreLikeThis(index_reader);
mlt.SetFieldNames(new string[] { "content" });
mlt.SetMinTermFreq(0);
Query query = mlt.Like(new StreamReader(document_filename));
Hits hits = index_searcher.Search(query);
var i = hits.Iterator();
while (i.MoveNext())
{
Hit hit = (Hit)i.Current;
string fingerprint = hit.Get("fingerprint");
fingerprints.Add(fingerprint);
}
// Close the index
index_searcher.Close();
}
index_reader.Close();
}
}
catch (Exception ex)
{
Logging.Warn(ex, "GetDocumentsSimilarToDocument: There was a problem opening the index file for searching.");
}
return(fingerprints);
}
public static void Main(String[] a)
{
String indexName = "localhost_index";
String fn = "c:/Program Files/Apache Group/Apache/htdocs/manual/vhosts/index.html.en";
Uri url = null;
for (int i = 0; i < a.Length; i++)
{
if (a[i].Equals("-i"))
{
indexName = a[++i];
}
else if (a[i].Equals("-f"))
{
fn = a[++i];
}
else if (a[i].Equals("-url"))
{
url = new Uri(a[++i]);
}
}
StreamWriter temp_writer;
temp_writer = new StreamWriter(Console.OpenStandardOutput(), Console.Out.Encoding);
temp_writer.AutoFlush = true;
StreamWriter o = temp_writer;
IndexReader r = IndexReader.Open(indexName);
o.WriteLine("Open index " + indexName + " which has " + r.NumDocs() + " docs");
LuceneMoreLikeThis mlt = new LuceneMoreLikeThis(r);
o.WriteLine("Query generation parameters:");
o.WriteLine(mlt.DescribeParams());
o.WriteLine();
Query query = null;
if (url != null)
{
o.WriteLine("Parsing URL: " + url);
query = mlt.Like(url);
}
else if (fn != null)
{
o.WriteLine("Parsing file: " + fn);
query = mlt.Like(new FileInfo(fn));
}
o.WriteLine("q: " + query);
o.WriteLine();
Lucene.Net.Search.IndexSearcher searcher = new Lucene.Net.Search.IndexSearcher(indexName);
Lucene.Net.Search.Hits hits = searcher.Search(query);
int len = hits.Length();
o.WriteLine("found: " + len + " documents matching");
o.WriteLine();
for (int i = 0; i < Math.Min(25, len); i++)
{
Lucene.Net.Documents.Document d = hits.Doc(i);
String summary = d.Get("summary");
o.WriteLine("score : " + hits.Score(i));
o.WriteLine("url : " + d.Get("url"));
o.WriteLine("\ttitle : " + d.Get("title"));
if (summary != null)
{
o.WriteLine("\tsummary: " + d.Get("summary"));
}
o.WriteLine();
}
}
