public DocumentDescriptorId FindDuplicateDocumentId(
DocumentDescriptorId sourceDocumentId,
FileHash sourceHash,
BlobId sourceBlobId
)
{
if (!_config.IsDeduplicationActive)
return null;
var original = _blobStore.GetDescriptor(sourceBlobId);
var matches = _hashReader.FindDocumentByHash(sourceHash);
Logger.DebugFormat("Deduplicating document {0}", sourceDocumentId);
foreach (var match in matches)
{
if (match.DocumentDescriptorId == sourceDocumentId)
continue;
Logger.DebugFormat("Checking document {0}", match.DocumentDescriptorId);
var candidate = this._blobStore.GetDescriptor(match.BlobId);
// only within same content type!
if (candidate.ContentType != original.ContentType)
{
Logger.DebugFormat("document {0} has different ContentType ({1}), skipping",
match.DocumentDescriptorId, candidate.ContentType
);
continue;
}
// and same length
if (candidate.Length != original.Length)
{
Logger.DebugFormat("document {0} has different length ({1}), skipping",
match.DocumentDescriptorId, candidate.Length
);
continue;
}
// binary check
using (var candidateStream = candidate.OpenRead())
using (var originalStream = original.OpenRead())
{
if (StreamHelper.StreamsContentsAreEqual(candidateStream, originalStream))
{
Logger.DebugFormat("{0} has same content of {1}: match found!",
match.DocumentDescriptorId, sourceDocumentId
);
return match.DocumentDescriptorId;
}
else
{
Logger.DebugFormat("{0} has different content of {1}, skipping",
match.DocumentDescriptorId, sourceDocumentId
);
}
}
}
return null;
}
public IEnumerable<Match> FindDocumentByHash(FileHash hash)
{
return _reader.Collection
.Find(Builders<DocumentDescriptorReadModel>.Filter.Eq(x => x.Hash, hash))
.Sort(Builders<DocumentDescriptorReadModel>.Sort.Ascending(x=>x.SequenceNumber))
.ToList()
.Select(x => new Match(x.Id, x.GetOriginalBlobId()));
}
public DocumentDescriptorInitialized(
BlobId blobId,
DocumentHandleInfo handleInfo,
FileHash hash)
{
Hash = hash;
HandleInfo = handleInfo;
BlobId = blobId;
}
public InitializeDocumentDescriptor(
DocumentDescriptorId aggregateId,
BlobId blobId,
DocumentHandleInfo handleInfo,
FileHash hash,
FileNameWithExtension fileName
) : base(aggregateId)
{
FileName = fileName;
Hash = hash;
BlobId = blobId;
HandleInfo = handleInfo;
}
public InitializeDocumentDescriptorAsAttach(
DocumentDescriptorId aggregateId,
BlobId blobId,
DocumentHandleInfo handleInfo,
DocumentHandle fatherHandle,
DocumentDescriptorId fatherDocumentDescriptorId,
FileHash hash,
FileNameWithExtension fileName)
: base(aggregateId, blobId, handleInfo, hash, fileName)
{
FatherHandle = fatherHandle;
FatherDocumentDescriptorId = fatherDocumentDescriptorId;
}