public Searcher(IReadSessionFactory sessionFactory)
{
_parser = new QueryParser();
_scorerFactory = new TfIdfFactory();
_versions = Util.GetIndexVersionListInChronologicalOrder(sessionFactory.DirectoryName);
_blockSize = BlockSerializer.SizeOfBlock();
_sessionFactory = sessionFactory;
}
public Searcher(string directory, IQueryParser parser = null, IScoringSchemeFactory scorerFactory = null, IReadSessionFactory sessionFactory = null)
{
_parser = parser ?? new QueryParser();
_scorerFactory = scorerFactory ?? new TfIdfFactory();
_versions = Util.GetIndexVersionListInChronologicalOrder(directory);
_blockSize = BlockSerializer.SizeOfBlock();
_sessionFactory = sessionFactory ?? new FullTextReadSessionFactory(directory);
}
public Searcher(string directory, long version, QueryParser parser, IScoringSchemeFactory scorerFactory, IReadSessionFactory sessionFactory = null)
{
_directory = directory;
_parser = parser;
_scorerFactory = scorerFactory;
_versions = new[] { BatchInfo.Load(Path.Combine(directory, version + ".ix")) };
_blockSize = BlockSerializer.SizeOfBlock();
_sessionFactory = sessionFactory ?? new ReadSessionFactory(directory);
}
public MappedTrieReader(string fileName)
{
var dir = Path.GetDirectoryName(fileName);
var version = Path.GetFileNameWithoutExtension(fileName);
_stream = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.Read, 4096 * 1, FileOptions.SequentialScan);
_blockSize = Serializer.SizeOfNode() + BlockSerializer.SizeOfBlock();
}
public Searcher(string directory, QueryParser parser, IScoringSchemeFactory scorerFactory, IReadSessionFactory sessionFactory = null)
{
_directory = directory;
_parser = parser;
_scorerFactory = scorerFactory;
_versions = Util.GetIndexFileNamesInChronologicalOrder(directory)
.Select(f => BatchInfo.Load(f)).ToArray();
_blockSize = BlockSerializer.SizeOfBlock();
_sessionFactory = sessionFactory ?? new ReadSessionFactory(directory);
}
public ReadSession(
SegmentInfo version,
DocHashReader docHashReader,
BlockInfoReader addressReader,
Stream stream)
{
Version = version;
Stream = stream;
DocHashReader = docHashReader;
AddressReader = addressReader;
BlockSize = BlockSerializer.SizeOfBlock();
Stream.Seek(Version.KeyIndexOffset, SeekOrigin.Begin);
KeyIndex = DocumentSerializer.ReadKeyIndex(Stream, Version.KeyIndexSize);
}
public ReadSession(
BatchInfo version,
PostingsReader postingsReader,
DocHashReader docHashReader,
DocumentAddressReader addressReader,
Stream stream)
{
Version = version;
_stream = stream;
_docHashReader = docHashReader;
_postingsReader = postingsReader;
_addressReader = addressReader;
_blockSize = BlockSerializer.SizeOfBlock();
}
private IEnumerable <DocumentTableRow> StreamDocuments()
{
var skipped = 0;
var took = 0;
for (int docId = 0; docId < _ix.DocumentCount; docId++)
{
var hash = _hashReader.Read(docId);
var address = _addressReader.Read(new[]
{
new BlockInfo(docId * BlockSerializer.SizeOfBlock(), BlockSerializer.SizeOfBlock())
}).First();
var document = _documentReader.Read(new List <BlockInfo> {
address
}).First();
if (!hash.IsObsolete)
{
if (skipped == _skip && took < _take)
{
yield return(document);
took++;
}
else if (skipped < _skip)
{
skipped++;
}
else if (took == _take)
{
break;
}
}
}
}
public MappedTrieReader(Stream stream)
{
_stream = stream;
_blockSize = Serializer.SizeOfNode() + BlockSerializer.SizeOfBlock();
_leaveOpen = true;
}
