private long getRawDataSize(TreeWriter child, TypeDescription schema)
{
long total = 0;
long numVals = child.fileStatistics.getNumberOfValues();
switch (schema.getCategory())
{
case Category.BOOLEAN:
case Category.BYTE:
case Category.SHORT:
case Category.INT:
case Category.FLOAT:
return numVals * JavaDataModel.Four;
case Category.LONG:
case Category.DOUBLE:
return numVals * JavaDataModel.Eight;
case Category.STRING:
case Category.VARCHAR:
case Category.CHAR:
// ORC strings are converted to java Strings. so use JavaDataModel to
// compute the overall size of strings
StringColumnStatistics scs = (StringColumnStatistics)child.fileStatistics;
numVals = numVals == 0 ? 1 : numVals;
int avgStringLen = (int)(scs.getSum() / numVals);
return numVals * JavaDataModel.lengthForStringOfLength(avgStringLen);
case Category.DECIMAL:
return numVals * JavaDataModel.lengthOfDecimal();
case Category.DATE:
return numVals * JavaDataModel.lengthOfDate();
case Category.BINARY:
// get total length of binary blob
BinaryColumnStatistics bcs = (BinaryColumnStatistics)child.fileStatistics;
return bcs.getSum();
case Category.TIMESTAMP:
return numVals * JavaDataModel.lengthOfTimestamp();
case Category.LIST:
case Category.MAP:
case Category.UNION:
case Category.STRUCT:
{
TreeWriter[] childWriters = child.getChildrenWriters();
IList<TypeDescription> childTypes = schema.getChildren();
for (int i = 0; i < childWriters.Length; ++i)
{
total += getRawDataSize(childWriters[i], childTypes[i]);
}
break;
}
default:
LOG.debug("Unknown object inspector category.");
break;
}
return total;
}
private void writeFileStatistics(OrcProto.Footer.Builder builder,
TreeWriter writer)
{
builder.AddStatistics(writer.fileStatistics.serialize());
foreach (TreeWriter child in writer.getChildrenWriters())
{
writeFileStatistics(builder, child);
}
}
private void writeStripeStatistics(OrcProto.StripeStatistics.Builder builder, TreeWriter treeWriter)
{
treeWriter.fileStatistics.merge(treeWriter.stripeColStatistics);
builder.AddColStats(treeWriter.stripeColStatistics.serialize().Build());
treeWriter.stripeColStatistics.reset();
foreach (TreeWriter child in treeWriter.getChildrenWriters())
{
writeStripeStatistics(builder, child);
}
}