private long Merge(string srcIxFileName)
{
Log.InfoFormat("merging branch {0} with trunk {1}", _ixFilesToProcess[1], _ixFilesToProcess[0]);
var ix = IxInfo.Load(srcIxFileName);
var documentFileName = Path.Combine(_directory, ix.VersionId + ".rdoc");
long version;
using (var documentStream = new RDocStream(documentFileName, ix.PrimaryKeyFieldName))
{
using (var upsert = new UpsertTransaction(
_directory,
_analyzer,
ix.Compression,
documentStream))
{
version = upsert.Write();
upsert.Commit();
}
Log.InfoFormat("{0} merged with {1} creating a segmented index", srcIxFileName, _ixFilesToProcess[0]);
}
Util.RemoveAll(srcIxFileName);
return(version);
}
public Collector(string directory, IxInfo ix, ConcurrentDictionary <string, LazyTrie> trieFiles, ConcurrentDictionary <string, PostingsContainer> postingContainers)
{
_directory = directory;
_trieFiles = trieFiles;
_postingContainers = postingContainers;
_ix = ix;
}
private long Truncate(string srcIxFileName)
{
Log.InfoFormat("truncating {0}", srcIxFileName);
var srcIx = IxInfo.Load(srcIxFileName);
var documentFileName = Path.Combine(_directory, srcIx.VersionId + ".rdoc");
var docAddressFn = Path.Combine(_directory, srcIx.VersionId + ".da");
var docHashesFileName = Path.Combine(_directory, string.Format("{0}.{1}", srcIx.VersionId, "pk"));
long version;
using (var documentStream = new RDocStream(documentFileName, srcIx.PrimaryKeyFieldName))
{
Util.TryAquireWriteLock(_directory);
using (var upsert = new UpsertTransaction(
_directory,
_analyzer,
srcIx.Compression,
documentStream))
{
version = upsert.Write();
upsert.Commit();
}
Util.ReleaseFileLock(_directory);
Log.InfoFormat("ix {0} fully truncated", _ixFilesToProcess[0]);
}
Util.RemoveAll(srcIxFileName);
return(version);
}
private IEnumerable <ScoredDocument> GetDocs(IList <DocumentScore> scores, IxInfo ix)
{
var docAddressFileName = Path.Combine(_directory, ix.VersionId + ".da");
IList <BlockInfo> docAdrs;
using (var docAddressReader = new DocumentAddressReader(
new FileStream(docAddressFileName, FileMode.Open, FileAccess.Read, FileShare.Read, 4096 * 1, FileOptions.SequentialScan)))
{
var adrs = scores
.Select(s => new BlockInfo(s.DocumentId * _blockSize, _blockSize))
.OrderBy(b => b.Position)
.ToList();
docAdrs = docAddressReader.Get(adrs).ToList();
}
var docFileName = Path.Combine(_directory, ix.VersionId + ".rdoc");
using (var docReader = new DocumentReader(
new FileStream(docFileName, FileMode.Open, FileAccess.Read, FileShare.Read, 4096 * 4, FileOptions.SequentialScan),
(Compression)ix.Compression))
{
var dic = scores.ToDictionary(x => x.DocumentId, y => y.Score);
foreach (var doc in docReader.Get(docAdrs))
{
var score = dic[doc.Id];
yield return(new ScoredDocument {
Document = doc, Score = score
});
}
}
}
public DocumentScore(int documentId, UInt64 docHash, double score, IxInfo ix)
{
DocumentId = documentId;
Score = score;
Ix = ix;
DocHash = docHash;
}
public Collector(string directory, IxInfo ix, IScoringScheme scorerFactory = null, int documentCount = -1)
{
_directory = directory;
_ix = ix;
_scorerFactory = scorerFactory;
_documentCount = documentCount == -1 ? ix.DocumentCount : documentCount;
}
public Collector(string directory, IxInfo ix, IScoringScheme scorerFactory = null, IDistanceResolver distanceResolver = null, int documentCount = -1)
{
_directory = directory;
_ix = ix;
_scorerFactory = scorerFactory;
_distanceResolver = distanceResolver ?? new Levenshtein();
_documentCount = documentCount == -1 ? ix.DocumentCount : documentCount;
}
public Collector(string directory, IxInfo ix, IScoringScheme scorerFactory = null, IDistanceResolver distanceResolver = null, int documentCount = -1)
{
_directory = directory;
_ix = ix;
_scorerFactory = scorerFactory;
_distanceResolver = distanceResolver ?? new Levenshtein();
_documentCount = documentCount == -1 ? ix.DocumentCount : documentCount;
_scoreCache = new Dictionary <SubQuery, IList <DocumentScore> >();
}
public Searcher(string directory, QueryParser parser, IScoringScheme scorer)
{
_directory = directory;
_parser = parser;
_scorer = scorer;
_trieFiles = new ConcurrentDictionary <string, LazyTrie>();
_docContainers = new ConcurrentDictionary <string, DocContainer>();
_postingContainers = new ConcurrentDictionary <string, PostingsContainer>();
_ix = IxInfo.Load(Path.Combine(_directory, "0.ix"));
}
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);
}
public void Dispose()
{
foreach (var docId in _deletions)
{
DoRemove(docId);
}
Parallel.ForEach(_trieFiles, kvp =>
{
var field = kvp.Key;
var trie = kvp.Value;
using (var container = new TrieWriter(field.ToTrieContainerId()))
{
trie.Save(container, _directory);
}
});
_docWorker.Dispose();
_postingsWorker.Dispose();
Parallel.ForEach(_postingsContainers.Values, container =>
{
if (container.Count > 0)
{
container.Flush(_directory);
container.Dispose();
}
else
{
container.Dispose();
File.Delete(Path.Combine(_directory, container.Id + ".pc"));
}
});
Parallel.ForEach(_docContainers.Values, container => container.Dispose());
_ix.Save(Path.Combine(_directory, "1.ix"));
var ixInfo = new IxInfo();
foreach (var field in _ix.Fields)
{
ixInfo.DocCount[field.Key] = field.Value.Count;
}
ixInfo.Save(Path.Combine(_directory, "0.ix"));
}
public Collector(string directory, IxInfo ix, IScoringScheme scorer)
{
_directory = directory;
_ix = ix;
_scorer = scorer;
var initTimer = Time();
var dbOptions = new BPlusTree <Term, DocumentPosting[]> .OptionsV2(
new TermSerializer(),
new ArraySerializer <DocumentPosting>(new PostingSerializer()), new TermComparer());
dbOptions.FileName = Path.Combine(directory, string.Format("{0}-{1}.{2}", _ix.Name, "pos", "db"));
dbOptions.ReadOnly = true;
_postingDb = new BPlusTree <Term, DocumentPosting[]>(dbOptions);
Log.DebugFormat("init collector in {0}", initTimer.Elapsed);
}
private void GetDocs(IList <DocumentScore> scores, IxInfo ix, ConcurrentBag <ScoredDocument> result)
{
var documentIds = scores.Select(s => s.DocumentId).ToList();
var docAddressFileName = Path.Combine(_directory, ix.VersionId + ".da");
var docFileName = Path.Combine(_directory, ix.VersionId + ".rdoc");
using (var session = _sessionFactory.Create(docAddressFileName, docFileName, ix.Compression))
{
var dic = scores.ToDictionary(x => x.DocumentId, y => y.Score);
foreach (var doc in session.Read(documentIds))
{
var score = dic[doc.Id];
result.Add(new ScoredDocument(doc, score));
}
}
}
public long Commit()
{
if (_ixFilesToProcess.Length == 1)
{
// merge segments
var ix = IxInfo.Load(_ixFilesToProcess[0]);
if (Util.IsSegmented(_ixFilesToProcess[0]))
{
return(Truncate(_ixFilesToProcess[0]));
}
else
{
return(-1);
}
}
// merge branches
return(Merge(_ixFilesToProcess[1]));
}
public UpsertTransaction(
string directory,
IAnalyzer analyzer,
Compression compression,
DocumentStream documents,
IDocumentStoreWriter storeWriter = null)
{
_directory = directory;
_analyzer = analyzer;
_compression = compression;
_documents = documents;
var mainIndexVersion = Util.GetIndexFileNamesInChronologicalOrder(_directory)
.FirstOrDefault();
if (mainIndexVersion == null)
{
_indexVersionId = Util.GetNextChronologicalFileId();
}
else
{
if (Util.WriteLockExists(_directory) || !Util.TryAquireWriteLock(_directory))
{
_indexVersionId = Util.GetNextChronologicalFileId();
}
else
{
_indexVersionId = long.Parse(Path.GetFileNameWithoutExtension(mainIndexVersion));
var ix = IxInfo.Load(mainIndexVersion);
_count = ix.DocumentCount;
}
}
_storeWriter = storeWriter ??
new DocumentStoreWriter(directory, _indexVersionId, _compression);
}
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;
}
