public MetaInfoObjExtractor(CompressionKind compressionKind, int bufferSize, int metadataSize,
ByteBuffer footerBuffer)
{
this.compressionKind = compressionKind;
this.bufferSize = bufferSize;
this.codec = WriterImpl.createCodec(compressionKind);
this.metadataSize = metadataSize;
int position = footerBuffer.position();
int footerBufferSize = footerBuffer.limit() - footerBuffer.position() - metadataSize;
this.metadata = extractMetadata(footerBuffer, position, metadataSize, codec, bufferSize);
this.footer = extractFooter(
footerBuffer, position + metadataSize, footerBufferSize, codec, bufferSize);
footerBuffer.position(position);
this.inspector = OrcStruct.createObjectInspector(0, footer.TypesList);
}
/** Extracts the necessary metadata from an externally store buffer (fullFooterBuffer). */
public static FooterInfo extractMetaInfoFromFooter(ByteBuffer bb, string srcPath)
{
// Read the PostScript. Be very careful as some parts of this historically use bb position
// and some use absolute offsets that have to take position into account.
int baseOffset = bb.position();
int lastByteAbsPos = baseOffset + bb.remaining() - 1;
int psLen = bb.get(lastByteAbsPos) & 0xff;
int psAbsPos = lastByteAbsPos - psLen;
OrcProto.PostScript ps = extractPostScript(bb, srcPath, psLen, psAbsPos);
Debug.Assert(baseOffset == bb.position());
// Extract PS information.
int footerSize = (int)ps.FooterLength, metadataSize = (int)ps.MetadataLength,
footerAbsPos = psAbsPos - footerSize, metadataAbsPos = footerAbsPos - metadataSize;
CompressionKind compressionKind = (CompressionKind)Enum.Parse(typeof(CompressionKind), ps.Compression.ToString(), true);
CompressionCodec codec = WriterImpl.createCodec(compressionKind);
int bufferSize = (int)ps.CompressionBlockSize;
bb.position(metadataAbsPos);
bb.mark();
// Extract metadata and footer.
OrcProto.Metadata metadata = extractMetadata(
bb, metadataAbsPos, metadataSize, codec, bufferSize);
List <StripeStatistics> stats = new List <StripeStatistics>(metadata.StripeStatsCount);
foreach (OrcProto.StripeStatistics ss in metadata.StripeStatsList)
{
stats.Add(new StripeStatistics(ss.ColStatsList));
}
OrcProto.Footer footer = extractFooter(bb, footerAbsPos, footerSize, codec, bufferSize);
bb.position(metadataAbsPos);
bb.limit(psAbsPos);
// TODO: do we need footer buffer here? FileInfo/FileMetaInfo is a mess...
FileMetaInfo fmi = new FileMetaInfo(
compressionKind, bufferSize, metadataSize, bb, extractWriterVersion(ps));
return(new FooterInfo(stats, footer, fmi));
}
/**
* Constructor that let's the user specify additional options.
* @param path pathname for file
* @param options options for reading
* @
*/
public ReaderImpl(Func <Stream> streamCreator, string path, OrcFile.ReaderOptions options)
{
this.streamCreator = streamCreator;
this.path = path;
this.conf = options.getConfiguration();
FileMetadata fileMetadata = options.getFileMetadata();
if (fileMetadata != null)
{
this.compressionKind = fileMetadata.getCompressionKind();
this.bufferSize = fileMetadata.getCompressionBufferSize();
this.codec = WriterImpl.createCodec(compressionKind);
this.metadataSize = fileMetadata.getMetadataSize();
this.stripeStats = fileMetadata.getStripeStats();
this.versionList = fileMetadata.getVersionList();
this.writerVersion = OrcFile.WriterVersionHelpers.from(fileMetadata.getWriterVersionNum());
this.types = fileMetadata.getTypes();
this.rowIndexStride = fileMetadata.getRowIndexStride();
this.contentLength = fileMetadata.getContentLength();
this.numberOfRows = fileMetadata.getNumberOfRows();
this.fileStats = fileMetadata.getFileStats();
this.stripes = fileMetadata.getStripes();
this.inspector = OrcStruct.createObjectInspector(0, fileMetadata.getTypes());
this.footerByteBuffer = null; // not cached and not needed here
this.userMetadata = null; // not cached and not needed here
this.footerMetaAndPsBuffer = null;
}
else
{
FileMetaInfo footerMetaData;
if (options.getFileMetaInfo() != null)
{
footerMetaData = options.getFileMetaInfo();
this.footerMetaAndPsBuffer = null;
}
else
{
using (Stream file = streamCreator())
{
footerMetaData = extractMetaInfoFromFooter(file, path, options.getMaxLength());
this.footerMetaAndPsBuffer = footerMetaData.footerMetaAndPsBuffer;
}
}
MetaInfoObjExtractor rInfo =
new MetaInfoObjExtractor(footerMetaData.compressionKind,
footerMetaData.bufferSize,
footerMetaData.metadataSize,
footerMetaData.footerBuffer
);
this.footerByteBuffer = footerMetaData.footerBuffer;
this.compressionKind = rInfo.compressionKind;
this.codec = rInfo.codec;
this.bufferSize = rInfo.bufferSize;
this.metadataSize = rInfo.metadataSize;
this.stripeStats = rInfo.metadata.StripeStatsList;
this.types = rInfo.footer.TypesList;
this.rowIndexStride = (int)rInfo.footer.RowIndexStride;
this.contentLength = (int)rInfo.footer.ContentLength;
this.numberOfRows = (int)rInfo.footer.NumberOfRows;
this.userMetadata = rInfo.footer.MetadataList;
this.fileStats = rInfo.footer.StatisticsList;
this.inspector = rInfo.inspector;
this.versionList = footerMetaData.versionList.Select(v => (int)v).ToList();
this.writerVersion = footerMetaData.writerVersion;
this.stripes = convertProtoStripesToStripes(rInfo.footer.StripesList);
}
}