public List <IndexPageResult> GetDocumentPagesWithQuery(string query)
{
List <IndexPageResult> results = new List <IndexPageResult>();
Dictionary <string, IndexPageResult> fingerprints_already_seen = new Dictionary <string, IndexPageResult>();
try
{
using (IndexReader index_reader = IndexReader.Open(LIBRARY_INDEX_BASE_PATH, true))
{
using (IndexSearcher index_searcher = new IndexSearcher(index_reader))
{
QueryParser query_parser = new QueryParser(Version.LUCENE_29, "content", analyzer);
Query query_object = query_parser.Parse(query);
Lucene.Net.Search.Hits hits = index_searcher.Search(query_object);
var i = hits.Iterator();
while (i.MoveNext())
{
Hit hit = (Hit)i.Current;
string fingerprint = hit.Get("fingerprint");
int page = Convert.ToInt32(hit.Get("page"));
double score = hit.GetScore();
// If this is the first time we have seen this fingerprint, make the top-level record
if (!fingerprints_already_seen.ContainsKey(fingerprint))
{
IndexPageResult result = new IndexPageResult();
result.fingerprint = fingerprint;
result.score = score;
// Add to our structures
results.Add(result);
fingerprints_already_seen[fingerprint] = result;
}
// And add the page record
{
IndexPageResult result = fingerprints_already_seen[fingerprint];
result.page_results.Add(new PageResult {
page = page, score = score
});
}
}
// Close the index
index_searcher.Close();
}
index_reader.Close();
}
}
catch (Exception ex)
{
Logging.Warn(ex, "GetDocumentPagesWithQuery: There was a problem opening the index file for searching.");
}
return(results);
}
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);
}
/***
* Understands the lucene query syntax
*/
public List <IndexResult> GetDocumentsWithQuery(string query)
{
List <IndexResult> fingerprints = new List <IndexResult>();
HashSet <string> fingerprints_already_seen = new HashSet <string>();
try
{
IndexReader index_reader = IndexReader.Open(LIBRARY_INDEX_BASE_PATH, true);
Searcher index_searcher = new IndexSearcher(index_reader);
QueryParser query_parser = new QueryParser(Version.LUCENE_29, "content", analyzer);
Query query_object = query_parser.Parse(query);
Hits hits = index_searcher.Search(query_object);
var i = hits.Iterator();
while (i.MoveNext())
{
Hit hit = (Hit)i.Current;
string fingerprint = hit.Get("fingerprint");
string page = hit.Get("page");
if (!fingerprints_already_seen.Contains(fingerprint))
{
fingerprints_already_seen.Add(fingerprint);
IndexResult index_result = new IndexResult {
fingerprint = fingerprint, score = hit.GetScore()
};
fingerprints.Add(index_result);
}
}
// Close the index
index_searcher.Close();
index_reader.Close();
}
catch (Exception ex)
{
Logging.Warn(ex, "There was a problem opening the index file for searching.");
}
return(fingerprints);
}
public HashSet <string> GetDocumentsWithWord(string keyword)
{
HashSet <string> fingerprints = new HashSet <string>();
try
{
keyword = ReasonableWord.MakeReasonableWord(keyword);
if (null != keyword)
{
////Do a quick check for whether there are actually any segments files, otherwise we throw many exceptions in the IndexReader.Open in a very tight loop.
////Added by Nik to cope with some exception...will uncomment this when i know what the problem is...
//var segments_files = Directory.GetFiles(LIBRARY_INDEX_BASE_PATH, "segments*", SearchOption.AllDirectories);
//if (segments_files.Length <= 0)
//{
// Logging.Debug("No index segments files found");
// return fingerprints;
//}
using (IndexReader index_reader = IndexReader.Open(LIBRARY_INDEX_BASE_PATH, true))
{
using (IndexSearcher index_searcher = new IndexSearcher(index_reader))
{
TermQuery term_query = new TermQuery(new Term("content", keyword));
Hits hits = index_searcher.Search(term_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, "GetDocumentsWithWord: There was a problem opening the index file for searching.");
}
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);
}
