/// <summary>
/// for given document and fields in that doc get fixed no of docs that are similar
/// assumes you have index that is up to date
/// </summary>
/// <returns>list of similar docs found</returns>
public IEnumerable<SearchResultItem> FindMoreLikeThis()
{
var results = new List<SearchResultItem>();
if (IsInit())
{
var moreLikeThis = new MoreLikeThis(reader);
moreLikeThis.SetFieldNames(fieldsToSearch.ToArray());
moreLikeThis.SetMinTermFreq(1);
moreLikeThis.SetMinDocFreq(1);
int currentLuceneDocId = GetLuceneDocNo(docId);
if (currentLuceneDocId != 0)
{
var query = moreLikeThis.Like(currentLuceneDocId);
var docs = searcher.Search(query, maxNo);
int count = docs.ScoreDocs.Length;
//start at 1 as first item will be current document itself which we dont want
for (int i = 1; i < count; i++)
{
var d = reader.Document(docs.ScoreDocs[i].doc);
var item = new SearchResultItem
{
PageName = d.GetField("nodeName").StringValue(),
NodeId = int.Parse(d.GetField("__NodeId").StringValue())
};
results.Add(item);
}
}
}
return results;
}
public static void Main(System.String[] a)
{
System.String indexName = "localhost_index";
System.String fn = "c:/Program Files/Apache Group/Apache/htdocs/manual/vhosts/index.html.en";
System.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 System.Uri(a[++i]);
}
}
System.IO.StreamWriter temp_writer;
temp_writer = new System.IO.StreamWriter(System.Console.OpenStandardOutput(), System.Console.Out.Encoding);
temp_writer.AutoFlush = true;
System.IO.StreamWriter o = temp_writer;
IndexReader r = IndexReader.Open(indexName);
o.WriteLine("Open index " + indexName + " which has " + r.NumDocs() + " docs");
MoreLikeThis mlt = new MoreLikeThis(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 System.IO.FileInfo(fn));
}
o.WriteLine("q: " + query);
o.WriteLine();
IndexSearcher searcher = new IndexSearcher(indexName);
Hits hits = searcher.Search(query);
int len = hits.Length();
o.WriteLine("found: " + len + " documents matching");
o.WriteLine();
for (int i = 0; i < System.Math.Min(25, len); i++)
{
Document d = hits.Doc(i);
System.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();
}
}
public static void Main(System.String[] a)
{
System.String indexName = "localhost_index";
System.String fn = "c:/Program Files/Apache Group/Apache/htdocs/manual/vhosts/index.html.en";
System.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 System.Uri(a[++i]);
}
}
System.IO.StreamWriter temp_writer;
temp_writer = new System.IO.StreamWriter(System.Console.OpenStandardOutput(), System.Console.Out.Encoding);
temp_writer.AutoFlush = true;
System.IO.StreamWriter o = temp_writer;
IndexReader r = IndexReader.Open(indexName);
o.WriteLine("Open index " + indexName + " which has " + r.NumDocs() + " docs");
MoreLikeThis mlt = new MoreLikeThis(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 System.IO.FileInfo(fn));
}
o.WriteLine("q: " + query);
o.WriteLine();
IndexSearcher searcher = new IndexSearcher(indexName);
Hits hits = searcher.Search(query);
int len = hits.Length();
o.WriteLine("found: " + len + " documents matching");
o.WriteLine();
for (int i = 0; i < System.Math.Min(25, len); i++)
{
Document d = hits.Doc(i);
System.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();
}
}