// Load a DryadLinqMetaData from an existing dsc stream.
internal static DryadLinqMetaData FromDscStream(HpcLinqContext context, string dscStreamName)
{
DryadLinqMetaData metaData;
try
{
DscFileSet fs = context.DscService.GetFileSet(dscStreamName);
metaData = new DryadLinqMetaData();
metaData.m_context = context;
metaData.m_dscStreamName = dscStreamName;
//metaData.m_fp = 0L;
//metaData.m_dataSetInfo = null;
byte[] metaDataBytes;
//record-type
metaDataBytes = fs.GetMetadata(DryadLinqMetaData.RECORD_TYPE_NAME);
if (metaDataBytes != null)
{
string recordTypeString = Encoding.UTF8.GetString(metaDataBytes);
metaData.m_elemType = Type.GetType(recordTypeString);
}
//Compression-scheme
metaData.m_compressionScheme = fs.CompressionScheme;
}
catch (Exception e)
{
throw new DryadLinqException(HpcLinqErrorCode.ErrorReadingMetadata,
String.Format(SR.ErrorReadingMetadata), e);
}
return(metaData);
}
/// <summary>
/// Open a DSC fileset as a LINQ-to-HPC IQueryable{T}.
/// </summary>
/// <typeparam name="T">The type of the records in the table.</typeparam>
/// <param name="dscFileSetName">The name of the DSC fileset. </param>
/// <returns>An IQueryable{T} representing the data and associated with the HPC LINQ query provider.</returns>
public IQueryable <T> FromDsc <T>(string fileSetName)
{
ThrowIfDisposed();
string fullPath = DataPath.MakeDscStreamUri(_dscService, fileSetName);
try {
DscFileSet fs = _dscService.GetFileSet(fileSetName);
if (!fs.IsSealed())
{
throw new DryadLinqException(HpcLinqErrorCode.FileSetMustBeSealed,
SR.FileSetMustBeSealed);
}
int fileCount = fs.GetFiles().Count();
if (fileCount < 1)
{
throw new DryadLinqException(HpcLinqErrorCode.FileSetMustHaveAtLeastOneFile,
SR.FileSetMustHaveAtLeastOneFile);
}
}
catch (DscException dscEx) {
throw new DryadLinqException(HpcLinqErrorCode.FileSetCouldNotBeOpened,
SR.FileSetCouldNotBeOpened, dscEx);
}
DryadLinqQuery <T> q = DataProvider.GetPartitionedTable <T>(this, fullPath);
q.CheckAndInitialize(); // force the data-info checks.
return(q);
}
// ingresses data, and also sets the temporary-length lease.
internal static DryadLinqQuery <T> IngressTemporaryDataDirectlyToDsc <T>(HpcLinqContext context, IEnumerable <T> source, string dscFileSetName, DryadLinqMetaData metaData, DscCompressionScheme outputScheme)
{
DryadLinqQuery <T> result = IngressDataDirectlyToDsc(context, source, dscFileSetName, metaData, outputScheme);
// try to set a temporary lease on the resulting fileset
try
{
DscFileSet fs = context.DscService.GetFileSet(dscFileSetName);
fs.SetLeaseEndTime(DateTime.Now.Add(StaticConfig.LeaseDurationForTempFiles));
}
catch (DscException)
{
// suppress
}
return(result);
}
public IEnumerator <T> GetEnumerator()
{
List <string[]> filePathList; // a list of dsc files, each of which is represented by an array holding the replica paths
DscCompressionScheme compressionScheme;
try
{
DscFileSet fileSet = m_context.DscService.GetFileSet(m_fileSetName);
filePathList = fileSet.GetFiles().Select(file => file.ReadPaths).ToList();
DryadLinqMetaData metaData = DryadLinqMetaData.FromDscStream(m_context, m_fileSetName);
compressionScheme = metaData.CompressionScheme;
}
catch (Exception e)
{
throw new DryadLinqException(HpcLinqErrorCode.FailedToGetReadPathsForStream,
String.Format(SR.FailedToGetReadPathsForStream, this.m_fileSetName), e);
}
return(new TableEnumerator(m_context, filePathList, m_fileSetName, compressionScheme));
}
public DscIOStream(string streamName, FileAccess access, FileMode createMode, DscCompressionScheme compressionScheme)
{
if (String.IsNullOrEmpty(streamName))
{
throw new ArgumentNullException("streamName");
}
Uri streamUri = new Uri(streamName);
this.m_dscClient = new DscService(streamUri.Host);
this.m_fileSetName = streamUri.LocalPath.TrimStart('/');
this.m_mode = access;
this.m_compressionScheme = compressionScheme;
bool streamExists = this.m_dscClient.FileSetExists(this.m_fileSetName);
if (access == FileAccess.Read)
{
switch (createMode)
{
case FileMode.Open:
case FileMode.OpenOrCreate:
if (!streamExists)
{
throw new FileNotFoundException(String.Format( SR.StreamDoesNotExist , streamName));
}
break;
case FileMode.Append:
case FileMode.Create:
case FileMode.CreateNew:
case FileMode.Truncate:
throw new NotSupportedException();
}
this.m_dscFileSet = this.m_dscClient.GetFileSet(streamName);
this.m_dscFileEnumerator = this.m_dscFileSet.GetFiles().GetEnumerator();
}
else if (access == FileAccess.Write)
{
switch (createMode)
{
case FileMode.Append:
if (!streamExists)
{
this.m_dscFileSet = this.m_dscClient.CreateFileSet(this.m_fileSetName, this.m_compressionScheme);
}
break;
case FileMode.Create:
if (streamExists)
{
this.m_dscClient.DeleteFileSet(this.m_fileSetName);
}
this.m_dscFileSet = this.m_dscClient.CreateFileSet(this.m_fileSetName, this.m_compressionScheme);
break;
case FileMode.CreateNew:
if (streamExists)
{
throw new IOException(String.Format(SR.StreamAlreadyExists, streamName));
}
break;
case FileMode.Truncate:
if (streamExists)
{
this.m_dscClient.DeleteFileSet(this.m_fileSetName);
}
this.m_dscFileSet = this.m_dscClient.CreateFileSet(this.m_fileSetName, this.m_compressionScheme);
break;
case FileMode.Open:
case FileMode.OpenOrCreate: // TODO: this should be dealt with correctly,
// although it's not obvious what open should do
throw new NotSupportedException();
}
}
else
{
throw new ArgumentException(SR.ReadWriteNotSupported, "access");
}
this.m_fstream = null;
this.m_atEOF = false;
}
public DscIOStream(string streamName, FileAccess access, DscCompressionScheme compressionScheme)
{
if (String.IsNullOrEmpty(streamName))
{
throw new ArgumentNullException("streamName");
}
Uri streamUri = new Uri(streamName);
this.m_dscClient = new DscService(streamUri.Host);
this.m_fileSetName = streamUri.LocalPath;
this.m_mode = access;
this.m_fstream = null;
this.m_atEOF = false;
this.m_compressionScheme = compressionScheme;
if (access == FileAccess.Read)
{
this.m_dscFileSet = this.m_dscClient.GetFileSet(streamName);
this.m_dscFileEnumerator = this.m_dscFileSet.GetFiles().GetEnumerator();
}
else if (access == FileAccess.Write)
{
this.m_dscFileSet = this.m_dscClient.CreateFileSet(streamName, compressionScheme);
}
else
{
throw new ArgumentException(SR.ReadWriteNotSupported, "access");
}
}
//* streams plain enumerable data directly to DSC
internal static DryadLinqQuery <T> IngressDataDirectlyToDsc <T>(HpcLinqContext context,
IEnumerable <T> source,
string dscFileSetName,
DryadLinqMetaData metaData,
DscCompressionScheme outputScheme)
{
try
{
string dscFileSetUri = DataPath.MakeDscStreamUri(context.DscService.HostName, dscFileSetName);
if (source.Take(1).Count() == 0)
{
//there is no data.. we must create a FileSet with an empty file
//(the factory/stream approach opens files lazily and thus never opens a file if there is no data)
if (context.DscService.FileSetExists(dscFileSetName))
{
context.DscService.DeleteFileSet(dscFileSetName);
}
DscFileSet fileSet = context.DscService.CreateFileSet(dscFileSetName, outputScheme);
DscFile file = fileSet.AddNewFile(0);
string writePath = file.WritePath;
if (outputScheme == DscCompressionScheme.Gzip)
{
//even zero-byte file must go through the gzip-compressor (for headers etc).
using (Stream s = new FileStream(writePath, FileMode.Create))
{
var gzipStream = new GZipStream(s, CompressionMode.Compress);
gzipStream.Close();
}
}
else
{
StreamWriter sw = new StreamWriter(writePath, false);
sw.Close();
}
fileSet.Seal();
}
else
{
HpcLinqFactory <T> factory = (HpcLinqFactory <T>)HpcLinqCodeGen.GetFactory(context, typeof(T));
// new DscBlockStream(uri,Create,Write,compress) provides a DSC stream with one partition.
NativeBlockStream nativeStream = new DscBlockStream(dscFileSetUri, FileMode.Create, FileAccess.Write, outputScheme);
HpcRecordWriter <T> writer = factory.MakeWriter(nativeStream);
try
{
if (context.Configuration.AllowConcurrentUserDelegatesInSingleProcess)
{
foreach (T item in source)
{
writer.WriteRecordAsync(item);
}
}
else
{
foreach (T item in source)
{
writer.WriteRecordSync(item);
}
}
}
finally
{
writer.Close(); // closes the NativeBlockStream, which seals the dsc stream.
}
}
if (metaData != null)
{
DscFileSet fileSet = context.DscService.GetFileSet(dscFileSetName);
fileSet.SetMetadata(DryadLinqMetaData.RECORD_TYPE_NAME, Encoding.UTF8.GetBytes(metaData.ElemType.AssemblyQualifiedName));
}
return(DataProvider.GetPartitionedTable <T>(context, dscFileSetUri));
}
catch
{
// if we had a problem creating the empty fileset, try to delete it to avoid cruft being left in DSC.
try
{
context.DscService.DeleteFileSet(dscFileSetName);
}
catch
{
// suppress error during delete
}
throw; // rethrow the original exception.
}
}
