public Collector(string directory, IxInfo ix, IScoringScheme scorerFactory = null, int documentCount = -1)
{
_directory = directory;
_ix = ix;
_scorerFactory = scorerFactory;
_documentCount = documentCount == -1 ? ix.DocumentCount : documentCount;
_scoreCache = new Dictionary <Query, IList <DocumentScore> >();
var docHashesFileName = Path.Combine(_directory, string.Format("{0}.{1}", _ix.VersionId, "pk"));
_posFileName = Path.Combine(directory, string.Format("{0}.{1}", ix.VersionId, "pos"));
_docHashReader = new DocHashReader(docHashesFileName);
}
private IEnumerable <DocumentScore> DoScore(IList <DocumentPosting> postings)
{
var docHashesFileName = Path.Combine(_directory, string.Format("{0}.{1}", _ix.VersionId, "pk"));
using (var docHashReader = new DocHashReader(docHashesFileName))
{
if (_scorerFactory == null)
{
foreach (var posting in postings)
{
var docHash = docHashReader.Read(posting.DocumentId);
if (!docHash.IsObsolete)
{
yield return(new DocumentScore(posting.DocumentId, docHash.Hash, 0, _ix));
}
}
}
else
{
if (postings.Any())
{
var docsWithTerm = postings.Count;
var scorer = _scorerFactory.CreateScorer(_documentCount, docsWithTerm);
foreach (var posting in postings)
{
var docHash = docHashReader.Read(posting.DocumentId);
if (!docHash.IsObsolete)
{
var score = scorer.Score(posting);
yield return(new DocumentScore(posting.DocumentId, docHash.Hash, score, _ix));
}
}
}
}
}
}
public RDocStream(string fileName, string primaryKeyFieldName = null, int skip = 0, int take = int.MaxValue)
: base(primaryKeyFieldName)
{
var versionId = Path.GetFileNameWithoutExtension(fileName);
var directory = Path.GetDirectoryName(fileName);
var docFileName = Path.Combine(directory, versionId + ".rdoc");
var docAddressFn = Path.Combine(directory, versionId + ".da");
var docHashesFileName = Path.Combine(directory, string.Format("{0}.{1}", versionId, "pk"));
var keyIndexFileName = Path.Combine(directory, versionId + ".kix");
var keyIndex = Util.GetKeyIndex(keyIndexFileName);
_ix = IxInfo.Load(Path.Combine(directory, versionId + ".ix"));
_hashReader = new DocHashReader(docHashesFileName);
_addressReader = new DocumentAddressReader(new FileStream(docAddressFn, FileMode.Open, FileAccess.Read));
_documentReader = new DocumentReader(
new FileStream(docFileName, FileMode.Open, FileAccess.Read),
_ix.Compression,
keyIndex);
_skip = skip;
_take = take;
_directory = directory;
}
public FullTextReadSession(SegmentInfo version, DocHashReader docHashReader, BlockInfoReader addressReader, Stream stream) : base(version, docHashReader, addressReader, stream)
{
}
public NetworkFullTextReadSession(
IPEndPoint postingsEndpoint, IPEndPoint documentsEndpoint, SegmentInfo version, DocHashReader docHashReader, BlockInfoReader addressReader, Stream stream)
: base(version, docHashReader, addressReader, stream)
{
_postingsReader = new NetworkPostingsReader(postingsEndpoint);
_documentsReader = new NetworkBlockReader(documentsEndpoint);
}
