private Stream GetRequestedStream(Stream source, long offset, long lenght)
{
var putObjectRequestStream = new ObjectRequestStream(source, offset, lenght);
var requestStream = new MemoryStream();
if (putObjectRequestStream.Position != 0)
{
putObjectRequestStream.Seek(0, SeekOrigin.Begin);
}
putObjectRequestStream.CopyTo(requestStream, CopyBufferSize);
return requestStream;
}
private void PutObject(IDs3Client client, BulkObject obj, MasterObjectList bulkResult)
{
var fileToPut = File.OpenRead(testDirectoryBigFolderForMaxBlob + obj.Name);
var contentStream = new ObjectRequestStream(fileToPut, obj.Offset, obj.Length);
var putObjectRequest = new PutObjectRequest(
bulkResult.BucketName,
obj.Name,
contentStream
).WithJob(bulkResult.JobId)
.WithOffset(obj.Offset);
client.PutObject(putObjectRequest);
fileToPut.Close();
}
static void Main(string[] args)
{
// Create the IDs3Client instance
IDs3Client client = Ds3Builder.FromEnv().Build();
IDs3ClientHelpers helpers = new Ds3ClientHelpers(client);
string bucket = "bucket-name";
string directory = "TestData";
// Create a bucket if it does not already exist.
helpers.EnsureBucketExists(bucket);
// Generate the list of files to put for the job
var objectList = FileHelpers.ListObjectsForDirectory(directory);
// This is used later to create the stream for the object, this
// can be replaced with custom code to create the file stream
var streamBuilder = FileHelpers.BuildFilePutter(directory);
// Create the bulk put job
var putBulkResponse = client.PutBulkJobSpectraS3(
new PutBulkJobSpectraS3Request(bucket, objectList));
// Get the jobId from the response
Guid jobId = putBulkResponse.ResponsePayload.JobId;
// Create a set of chunk ids to know what has been processed and what as yet to be processed
var chunkIds = from chunk in putBulkResponse.ResponsePayload.Objects
select chunk.ChunkId;
var chunkSet = new HashSet<Guid>(chunkIds);
// Continue processing until all the chunks have been sent.
while (chunkSet.Count > 0)
{
// Get the set of chunks that are currently available for processing
var chunkResponse = client.GetJobChunksReadyForClientProcessingSpectraS3(
new GetJobChunksReadyForClientProcessingSpectraS3Request(jobId)
.WithPreferredNumberOfChunks(10)); // This can be changed to any number
// but 10 is a good default and you
// are not guaranteed to get this many
chunkResponse.Match((ts, response) =>
{
// If this matcher is called this means that we can safely process chunks without
// fear of the PutObject call failing due to cache unavailable conditions
// It is also safe to process all the chunks in parallel as well, or to process
// each chunk sequentially, while sending each object in parallel
foreach (var chunk in response.Objects)
{
chunkSet.Remove(chunk.ChunkId);
// this next step can be done in parallel
foreach (var obj in chunk.ObjectsList)
{
// Create the stream and seek to the correct position for that
// blob offset, and then wrap in a ObjectRequestStream to
// limit the amount of data transferred to the length of the
// blob being processed.
var stream = streamBuilder.Invoke(obj.Name);
stream.Seek(obj.Offset, System.IO.SeekOrigin.Begin);
var wrappedStream = new ObjectRequestStream(stream, obj.Length);
// Put the blob
client.PutObject(
new PutObjectRequest(bucket, obj.Name, wrappedStream)
.WithJob(jobId)
.WithOffset(obj.Offset));
}
}
},
Thread.Sleep // If this matcher is called this means that we need to wait before we can safely continue processing chunks
);
}
}
