static TypeDescription convertTypeInfo(TypeInfo info) {
switch (info.getCategory()) {
case PRIMITIVE:
{
PrimitiveTypeInfo pinfo = (PrimitiveTypeInfo) info;
switch (pinfo.getPrimitiveCategory()) {
case BOOLEAN:
return TypeDescription.createBoolean();
case BYTE:
return TypeDescription.createByte();
case SHORT:
return TypeDescription.createShort();
case INT:
return TypeDescription.createInt();
case LONG:
return TypeDescription.createLong();
case FLOAT:
return TypeDescription.createFloat();
case DOUBLE:
return TypeDescription.createDouble();
case STRING:
return TypeDescription.createString();
case DATE:
return TypeDescription.createDate();
case TIMESTAMP:
return TypeDescription.createTimestamp();
case BINARY:
return TypeDescription.createBinary();
case DECIMAL:
{
DecimalTypeInfo dinfo = (DecimalTypeInfo) pinfo;
return TypeDescription.createDecimal()
.withScale(dinfo.getScale())
.withPrecision(dinfo.getPrecision());
}
case VARCHAR:
{
BaseCharTypeInfo cinfo = (BaseCharTypeInfo) pinfo;
return TypeDescription.createVarchar().withMaxLength(cinfo.getLength());
}
case CHAR:
{
BaseCharTypeInfo cinfo = (BaseCharTypeInfo) pinfo;
return TypeDescription.createChar().withMaxLength(cinfo.getLength());
}
default:
throw new IllegalArgumentException(
"ORC doesn't handle primitive" + " category " + pinfo.getPrimitiveCategory());
}
}
case LIST:
{
ListTypeInfo linfo = (ListTypeInfo) info;
return TypeDescription.createList(convertTypeInfo(linfo.getListElementTypeInfo()));
}
case MAP:
{
MapTypeInfo minfo = (MapTypeInfo) info;
return TypeDescription.createMap(
convertTypeInfo(minfo.getMapKeyTypeInfo()),
convertTypeInfo(minfo.getMapValueTypeInfo()));
}
case UNION:
{
UnionTypeInfo minfo = (UnionTypeInfo) info;
TypeDescription result = TypeDescription.createUnion();
for (TypeInfo child : minfo.getAllUnionObjectTypeInfos()) {
result.addUnionChild(convertTypeInfo(child));
}
return result;
}
case STRUCT:
{
StructTypeInfo sinfo = (StructTypeInfo) info;
TypeDescription result = TypeDescription.createStruct();
for (String fieldName : sinfo.getAllStructFieldNames()) {
result.addField(fieldName, convertTypeInfo(sinfo.getStructFieldTypeInfo(fieldName)));
}
return result;
}
default:
throw new IllegalArgumentException("ORC doesn't handle " + info.getCategory());
}
}
private Writable createPrimitive(final Object obj, final PrimitiveObjectInspector inspector)
throws SerDeException {
if (obj == null) {
return null;
}
switch (inspector.getPrimitiveCategory()) {
case VOID:
return null;
case BOOLEAN:
return new BooleanWritable(
((BooleanObjectInspector) inspector).get(obj) ? Boolean.TRUE : Boolean.FALSE);
case BYTE:
return new ByteWritable(((ByteObjectInspector) inspector).get(obj));
case DOUBLE:
return new DoubleWritable(((DoubleObjectInspector) inspector).get(obj));
case FLOAT:
return new FloatWritable(((FloatObjectInspector) inspector).get(obj));
case INT:
return new IntWritable(((IntObjectInspector) inspector).get(obj));
case LONG:
return new LongWritable(((LongObjectInspector) inspector).get(obj));
case SHORT:
return new ShortWritable(((ShortObjectInspector) inspector).get(obj));
case STRING:
String v = ((StringObjectInspector) inspector).getPrimitiveJavaObject(obj);
try {
return new BytesWritable(v.getBytes("UTF-8"));
} catch (UnsupportedEncodingException e) {
throw new SerDeException("Failed to encode string in UTF-8", e);
}
case DECIMAL:
HiveDecimal hd = (HiveDecimal) inspector.getPrimitiveJavaObject(obj);
DecimalTypeInfo decTypeInfo = (DecimalTypeInfo) inspector.getTypeInfo();
int prec = decTypeInfo.precision();
int scale = decTypeInfo.scale();
byte[] src = hd.setScale(scale).unscaledValue().toByteArray();
// Estimated number of bytes needed.
int bytes = PRECISION_TO_BYTE_COUNT[prec - 1];
if (bytes == src.length) {
// No padding needed.
return new BytesWritable(src);
}
byte[] tgt = new byte[bytes];
if (hd.signum() == -1) {
// For negative number, initializing bits to 1
for (int i = 0; i < bytes; i++) {
tgt[i] |= 0xFF;
}
}
System.arraycopy(
src, 0, tgt, bytes - src.length, src.length); // Padding leading zeroes/ones.
return new BytesWritable(tgt);
case TIMESTAMP:
return new TimestampWritable(
((TimestampObjectInspector) inspector).getPrimitiveJavaObject(obj));
case CHAR:
String strippedValue =
((HiveCharObjectInspector) inspector).getPrimitiveJavaObject(obj).getStrippedValue();
return new BytesWritable(Binary.fromString(strippedValue).getBytes());
case VARCHAR:
String value =
((HiveVarcharObjectInspector) inspector).getPrimitiveJavaObject(obj).getValue();
return new BytesWritable(Binary.fromString(value).getBytes());
case BINARY:
return new BytesWritable(((BinaryObjectInspector) inspector).getPrimitiveJavaObject(obj));
default:
throw new SerDeException("Unknown primitive : " + inspector.getPrimitiveCategory());
}
}