use of org.apache.hadoop.hive.common.type.HiveChar in project hive by apache.
the class TestDefaultHCatRecord method getHCat13TypesRecord.
private static HCatRecord getHCat13TypesRecord() {
List<Object> rec_hcat13types = new ArrayList<Object>(5);
// prec 5, scale 2
rec_hcat13types.add(HiveDecimal.create(new BigDecimal("123.45")));
rec_hcat13types.add(new HiveChar("hive_char", 10));
rec_hcat13types.add(new HiveVarchar("hive_varchar", 20));
rec_hcat13types.add(Date.valueOf("2014-01-06"));
rec_hcat13types.add(new Timestamp(System.currentTimeMillis()));
return new DefaultHCatRecord(rec_hcat13types);
}
use of org.apache.hadoop.hive.common.type.HiveChar in project hive by apache.
the class HCatBaseStorer method getJavaObj.
/**
* Convert from Pig value object to Hive value object
* This method assumes that {@link #validateSchema(org.apache.pig.impl.logicalLayer.schema.Schema.FieldSchema, org.apache.hive.hcatalog.data.schema.HCatFieldSchema, org.apache.pig.impl.logicalLayer.schema.Schema, org.apache.hive.hcatalog.data.schema.HCatSchema, int)}
* which checks the types in Pig schema are compatible with target Hive table, has been called.
*/
private Object getJavaObj(Object pigObj, HCatFieldSchema hcatFS) throws HCatException, BackendException {
try {
if (pigObj == null)
return null;
// The real work-horse. Spend time and energy in this method if there is
// need to keep HCatStorer lean and go fast.
Type type = hcatFS.getType();
switch(type) {
case BINARY:
return ((DataByteArray) pigObj).get();
case STRUCT:
HCatSchema structSubSchema = hcatFS.getStructSubSchema();
// Unwrap the tuple.
List<Object> all = ((Tuple) pigObj).getAll();
ArrayList<Object> converted = new ArrayList<Object>(all.size());
for (int i = 0; i < all.size(); i++) {
converted.add(getJavaObj(all.get(i), structSubSchema.get(i)));
}
return converted;
case ARRAY:
// Unwrap the bag.
DataBag pigBag = (DataBag) pigObj;
HCatFieldSchema tupFS = hcatFS.getArrayElementSchema().get(0);
boolean needTuple = tupFS.getType() == Type.STRUCT;
List<Object> bagContents = new ArrayList<Object>((int) pigBag.size());
Iterator<Tuple> bagItr = pigBag.iterator();
while (bagItr.hasNext()) {
// If there is only one element in tuple contained in bag, we throw away the tuple.
bagContents.add(getJavaObj(needTuple ? bagItr.next() : bagItr.next().get(0), tupFS));
}
return bagContents;
case MAP:
Map<?, ?> pigMap = (Map<?, ?>) pigObj;
Map<Object, Object> typeMap = new HashMap<Object, Object>();
for (Entry<?, ?> entry : pigMap.entrySet()) {
// the value has a schema and not a FieldSchema
typeMap.put(// Schema validation enforces that the Key is a String
(String) entry.getKey(), getJavaObj(entry.getValue(), hcatFS.getMapValueSchema().get(0)));
}
return typeMap;
case STRING:
case INT:
case BIGINT:
case FLOAT:
case DOUBLE:
return pigObj;
case SMALLINT:
if ((Integer) pigObj < Short.MIN_VALUE || (Integer) pigObj > Short.MAX_VALUE) {
handleOutOfRangeValue(pigObj, hcatFS);
return null;
}
return ((Integer) pigObj).shortValue();
case TINYINT:
if ((Integer) pigObj < Byte.MIN_VALUE || (Integer) pigObj > Byte.MAX_VALUE) {
handleOutOfRangeValue(pigObj, hcatFS);
return null;
}
return ((Integer) pigObj).byteValue();
case BOOLEAN:
if (pigObj instanceof String) {
if (((String) pigObj).trim().compareTo("0") == 0) {
return Boolean.FALSE;
}
if (((String) pigObj).trim().compareTo("1") == 0) {
return Boolean.TRUE;
}
throw new BackendException("Unexpected type " + type + " for value " + pigObj + " of class " + pigObj.getClass().getName(), PigHCatUtil.PIG_EXCEPTION_CODE);
}
return Boolean.parseBoolean(pigObj.toString());
case DECIMAL:
BigDecimal bd = (BigDecimal) pigObj;
DecimalTypeInfo dti = (DecimalTypeInfo) hcatFS.getTypeInfo();
if (bd.precision() > dti.precision() || bd.scale() > dti.scale()) {
handleOutOfRangeValue(pigObj, hcatFS);
return null;
}
return HiveDecimal.create(bd);
case CHAR:
String charVal = (String) pigObj;
CharTypeInfo cti = (CharTypeInfo) hcatFS.getTypeInfo();
if (charVal.length() > cti.getLength()) {
handleOutOfRangeValue(pigObj, hcatFS);
return null;
}
return new HiveChar(charVal, cti.getLength());
case VARCHAR:
String varcharVal = (String) pigObj;
VarcharTypeInfo vti = (VarcharTypeInfo) hcatFS.getTypeInfo();
if (varcharVal.length() > vti.getLength()) {
handleOutOfRangeValue(pigObj, hcatFS);
return null;
}
return new HiveVarchar(varcharVal, vti.getLength());
case TIMESTAMP:
DateTime dt = (DateTime) pigObj;
// getMillis() returns UTC time regardless of TZ
return new Timestamp(dt.getMillis());
case DATE:
/**
* We ignore any TZ setting on Pig value since java.sql.Date doesn't have it (in any
* meaningful way). So the assumption is that if Pig value has 0 time component (midnight)
* we assume it reasonably 'fits' into a Hive DATE. If time part is not 0, it's considered
* out of range for target type.
*/
DateTime dateTime = ((DateTime) pigObj);
if (dateTime.getMillisOfDay() != 0) {
handleOutOfRangeValue(pigObj, hcatFS, "Time component must be 0 (midnight) in local timezone; Local TZ val='" + pigObj + "'");
return null;
}
/*java.sql.Date is a poorly defined API. Some (all?) SerDes call toString() on it
[e.g. LazySimpleSerDe, uses LazyUtils.writePrimitiveUTF8()], which automatically adjusts
for local timezone. Date.valueOf() also uses local timezone (as does Date(int,int,int).
Also see PigHCatUtil#extractPigObject() for corresponding read op. This way a DATETIME from Pig,
when stored into Hive and read back comes back with the same value.*/
return new Date(dateTime.getYear() - 1900, dateTime.getMonthOfYear() - 1, dateTime.getDayOfMonth());
default:
throw new BackendException("Unexpected HCat type " + type + " for value " + pigObj + " of class " + pigObj.getClass().getName(), PigHCatUtil.PIG_EXCEPTION_CODE);
}
} catch (BackendException e) {
// provide the path to the field in the error message
throw new BackendException((hcatFS.getName() == null ? " " : hcatFS.getName() + ".") + e.getMessage(), e);
}
}
use of org.apache.hadoop.hive.common.type.HiveChar in project hive by apache.
the class VectorAssignRow method assignConvertRowColumn.
private void assignConvertRowColumn(ColumnVector columnVector, int batchIndex, TypeInfo targetTypeInfo, ObjectInspector sourceObjectInspector, Writable convertTargetWritable, Object object) {
final Category targetCategory = targetTypeInfo.getCategory();
if (targetCategory == null) {
/*
* This is a column that we don't want (i.e. not included) -- we are done.
*/
return;
}
if (object == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
try {
switch(targetCategory) {
case PRIMITIVE:
final PrimitiveObjectInspector sourcePrimitiveOI = (PrimitiveObjectInspector) sourceObjectInspector;
final PrimitiveCategory targetPrimitiveCategory = ((PrimitiveTypeInfo) targetTypeInfo).getPrimitiveCategory();
switch(targetPrimitiveCategory) {
case VOID:
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
case BOOLEAN:
((LongColumnVector) columnVector).vector[batchIndex] = (PrimitiveObjectInspectorUtils.getBoolean(object, sourcePrimitiveOI) ? 1 : 0);
break;
case BYTE:
((LongColumnVector) columnVector).vector[batchIndex] = PrimitiveObjectInspectorUtils.getByte(object, sourcePrimitiveOI);
break;
case SHORT:
((LongColumnVector) columnVector).vector[batchIndex] = PrimitiveObjectInspectorUtils.getShort(object, sourcePrimitiveOI);
break;
case INT:
((LongColumnVector) columnVector).vector[batchIndex] = PrimitiveObjectInspectorUtils.getInt(object, sourcePrimitiveOI);
break;
case LONG:
((LongColumnVector) columnVector).vector[batchIndex] = PrimitiveObjectInspectorUtils.getLong(object, sourcePrimitiveOI);
break;
case TIMESTAMP:
{
final Timestamp timestamp = PrimitiveObjectInspectorUtils.getTimestamp(object, sourcePrimitiveOI);
if (timestamp == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
((TimestampColumnVector) columnVector).set(batchIndex, timestamp);
}
break;
case DATE:
{
final Date date = PrimitiveObjectInspectorUtils.getDate(object, sourcePrimitiveOI);
if (date == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
DateWritable dateWritable = (DateWritable) convertTargetWritable;
if (dateWritable == null) {
dateWritable = new DateWritable();
}
dateWritable.set(date);
((LongColumnVector) columnVector).vector[batchIndex] = dateWritable.getDays();
}
break;
case FLOAT:
((DoubleColumnVector) columnVector).vector[batchIndex] = PrimitiveObjectInspectorUtils.getFloat(object, sourcePrimitiveOI);
break;
case DOUBLE:
((DoubleColumnVector) columnVector).vector[batchIndex] = PrimitiveObjectInspectorUtils.getDouble(object, sourcePrimitiveOI);
break;
case BINARY:
{
final BytesWritable bytesWritable = PrimitiveObjectInspectorUtils.getBinary(object, sourcePrimitiveOI);
if (bytesWritable == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
((BytesColumnVector) columnVector).setVal(batchIndex, bytesWritable.getBytes(), 0, bytesWritable.getLength());
}
break;
case STRING:
{
final String string = PrimitiveObjectInspectorUtils.getString(object, sourcePrimitiveOI);
if (string == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
Text text = (Text) convertTargetWritable;
if (text == null) {
text = new Text();
}
text.set(string);
((BytesColumnVector) columnVector).setVal(batchIndex, text.getBytes(), 0, text.getLength());
}
break;
case VARCHAR:
{
// UNDONE: Performance problem with conversion to String, then bytes...
final HiveVarchar hiveVarchar = PrimitiveObjectInspectorUtils.getHiveVarchar(object, sourcePrimitiveOI);
if (hiveVarchar == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
// TODO: Do we need maxLength checking?
byte[] bytes = hiveVarchar.getValue().getBytes();
((BytesColumnVector) columnVector).setVal(batchIndex, bytes, 0, bytes.length);
}
break;
case CHAR:
{
// UNDONE: Performance problem with conversion to String, then bytes...
final HiveChar hiveChar = PrimitiveObjectInspectorUtils.getHiveChar(object, sourcePrimitiveOI);
if (hiveChar == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
// We store CHAR in vector row batch with padding stripped.
// TODO: Do we need maxLength checking?
final byte[] bytes = hiveChar.getStrippedValue().getBytes();
((BytesColumnVector) columnVector).setVal(batchIndex, bytes, 0, bytes.length);
}
break;
case DECIMAL:
{
final HiveDecimal hiveDecimal = PrimitiveObjectInspectorUtils.getHiveDecimal(object, sourcePrimitiveOI);
if (hiveDecimal == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
if (columnVector instanceof Decimal64ColumnVector) {
Decimal64ColumnVector dec64ColVector = (Decimal64ColumnVector) columnVector;
dec64ColVector.set(batchIndex, hiveDecimal);
if (dec64ColVector.isNull[batchIndex]) {
return;
}
} else {
((DecimalColumnVector) columnVector).set(batchIndex, hiveDecimal);
}
}
break;
case INTERVAL_YEAR_MONTH:
{
final HiveIntervalYearMonth intervalYearMonth = PrimitiveObjectInspectorUtils.getHiveIntervalYearMonth(object, sourcePrimitiveOI);
if (intervalYearMonth == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
((LongColumnVector) columnVector).vector[batchIndex] = intervalYearMonth.getTotalMonths();
}
break;
case INTERVAL_DAY_TIME:
{
final HiveIntervalDayTime intervalDayTime = PrimitiveObjectInspectorUtils.getHiveIntervalDayTime(object, sourcePrimitiveOI);
if (intervalDayTime == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
((IntervalDayTimeColumnVector) columnVector).set(batchIndex, intervalDayTime);
}
break;
default:
throw new RuntimeException("Primitive category " + targetPrimitiveCategory.name() + " not supported");
}
break;
case LIST:
{
final ListColumnVector listColumnVector = (ListColumnVector) columnVector;
final ListObjectInspector sourceListOI = (ListObjectInspector) sourceObjectInspector;
final ObjectInspector sourceElementOI = sourceListOI.getListElementObjectInspector();
final int size = sourceListOI.getListLength(object);
final TypeInfo targetElementTypeInfo = ((ListTypeInfo) targetTypeInfo).getListElementTypeInfo();
listColumnVector.offsets[batchIndex] = listColumnVector.childCount;
listColumnVector.childCount += size;
listColumnVector.ensureSize(listColumnVector.childCount, true);
listColumnVector.lengths[batchIndex] = size;
for (int i = 0; i < size; i++) {
final Object element = sourceListOI.getListElement(object, i);
final int offset = (int) (listColumnVector.offsets[batchIndex] + i);
assignConvertRowColumn(listColumnVector.child, offset, targetElementTypeInfo, sourceElementOI, null, element);
}
}
break;
case MAP:
{
final MapColumnVector mapColumnVector = (MapColumnVector) columnVector;
final MapObjectInspector mapObjectInspector = (MapObjectInspector) sourceObjectInspector;
final MapTypeInfo mapTypeInfo = (MapTypeInfo) targetTypeInfo;
final Map<?, ?> map = mapObjectInspector.getMap(object);
for (Map.Entry<?, ?> entry : map.entrySet()) {
assignConvertRowColumn(mapColumnVector.keys, batchIndex, mapTypeInfo.getMapKeyTypeInfo(), mapObjectInspector.getMapKeyObjectInspector(), null, entry.getKey());
assignConvertRowColumn(mapColumnVector.values, batchIndex, mapTypeInfo.getMapValueTypeInfo(), mapObjectInspector.getMapValueObjectInspector(), null, entry.getValue());
}
}
break;
case STRUCT:
{
final StructColumnVector structColumnVector = (StructColumnVector) columnVector;
final StructObjectInspector sourceStructOI = (StructObjectInspector) sourceObjectInspector;
final List<? extends StructField> sourceFields = sourceStructOI.getAllStructFieldRefs();
final StructTypeInfo targetStructTypeInfo = (StructTypeInfo) targetTypeInfo;
final List<TypeInfo> targetTypeInfos = targetStructTypeInfo.getAllStructFieldTypeInfos();
final int size = targetTypeInfos.size();
for (int i = 0; i < size; i++) {
if (i < sourceFields.size()) {
final StructField sourceStructField = sourceFields.get(i);
final ObjectInspector sourceFieldOI = sourceStructField.getFieldObjectInspector();
final Object sourceData = sourceStructOI.getStructFieldData(object, sourceStructField);
assignConvertRowColumn(structColumnVector.fields[i], batchIndex, targetTypeInfos.get(i), sourceFieldOI, null, sourceData);
} else {
final ColumnVector fieldColumnVector = structColumnVector.fields[i];
VectorizedBatchUtil.setNullColIsNullValue(fieldColumnVector, batchIndex);
}
}
}
break;
case UNION:
{
final UnionColumnVector unionColumnVector = (UnionColumnVector) columnVector;
final UnionObjectInspector unionObjectInspector = (UnionObjectInspector) sourceObjectInspector;
final UnionTypeInfo unionTypeInfo = (UnionTypeInfo) targetTypeInfo;
final int tag = unionObjectInspector.getTag(object);
assignConvertRowColumn(unionColumnVector.fields[tag], batchIndex, unionTypeInfo.getAllUnionObjectTypeInfos().get(tag), unionObjectInspector.getObjectInspectors().get(tag), null, unionObjectInspector.getField(tag));
}
break;
default:
throw new RuntimeException("Category " + targetCategory.name() + " not supported");
}
} catch (NumberFormatException e) {
// Some of the conversion methods throw this exception on numeric parsing errors.
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
// We always set the null flag to false when there is a value.
columnVector.isNull[batchIndex] = false;
}
use of org.apache.hadoop.hive.common.type.HiveChar in project hive by apache.
the class VectorizationContext method getConstantVectorExpression.
private VectorExpression getConstantVectorExpression(Object constantValue, TypeInfo typeInfo, VectorExpressionDescriptor.Mode mode) throws HiveException {
String typeName = typeInfo.getTypeName();
VectorExpressionDescriptor.ArgumentType vectorArgType = VectorExpressionDescriptor.ArgumentType.fromHiveTypeName(typeName);
if (vectorArgType == VectorExpressionDescriptor.ArgumentType.NONE) {
throw new HiveException("No vector argument type for type name " + typeName);
}
int outCol = -1;
if (mode == VectorExpressionDescriptor.Mode.PROJECTION) {
outCol = ocm.allocateOutputColumn(typeInfo);
}
if (constantValue == null) {
return new ConstantVectorExpression(outCol, typeInfo, true);
}
// Boolean is special case.
if (typeName.equalsIgnoreCase("boolean")) {
if (mode == VectorExpressionDescriptor.Mode.FILTER) {
if (((Boolean) constantValue).booleanValue()) {
return new FilterConstantBooleanVectorExpression(1);
} else {
return new FilterConstantBooleanVectorExpression(0);
}
} else {
if (((Boolean) constantValue).booleanValue()) {
return new ConstantVectorExpression(outCol, 1, typeInfo);
} else {
return new ConstantVectorExpression(outCol, 0, typeInfo);
}
}
}
switch(vectorArgType) {
case INT_FAMILY:
return new ConstantVectorExpression(outCol, ((Number) constantValue).longValue(), typeInfo);
case DATE:
return new ConstantVectorExpression(outCol, DateWritable.dateToDays((Date) constantValue), typeInfo);
case TIMESTAMP:
return new ConstantVectorExpression(outCol, (Timestamp) constantValue, typeInfo);
case INTERVAL_YEAR_MONTH:
return new ConstantVectorExpression(outCol, ((HiveIntervalYearMonth) constantValue).getTotalMonths(), typeInfo);
case INTERVAL_DAY_TIME:
return new ConstantVectorExpression(outCol, (HiveIntervalDayTime) constantValue, typeInfo);
case FLOAT_FAMILY:
return new ConstantVectorExpression(outCol, ((Number) constantValue).doubleValue(), typeInfo);
case DECIMAL:
return new ConstantVectorExpression(outCol, (HiveDecimal) constantValue, typeInfo);
case STRING:
return new ConstantVectorExpression(outCol, ((String) constantValue).getBytes(), typeInfo);
case CHAR:
return new ConstantVectorExpression(outCol, ((HiveChar) constantValue), typeInfo);
case VARCHAR:
return new ConstantVectorExpression(outCol, ((HiveVarchar) constantValue), typeInfo);
default:
throw new HiveException("Unsupported constant type: " + typeName + ", object class " + constantValue.getClass().getSimpleName());
}
}
use of org.apache.hadoop.hive.common.type.HiveChar in project hive by apache.
the class VectorAssignRow method assignRowColumn.
private void assignRowColumn(ColumnVector columnVector, int batchIndex, TypeInfo targetTypeInfo, Object object) {
if (object == null) {
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
}
switch(targetTypeInfo.getCategory()) {
case PRIMITIVE:
{
final PrimitiveCategory targetPrimitiveCategory = ((PrimitiveTypeInfo) targetTypeInfo).getPrimitiveCategory();
switch(targetPrimitiveCategory) {
case VOID:
VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
return;
case BOOLEAN:
if (object instanceof Boolean) {
((LongColumnVector) columnVector).vector[batchIndex] = (((Boolean) object) ? 1 : 0);
} else {
((LongColumnVector) columnVector).vector[batchIndex] = (((BooleanWritable) object).get() ? 1 : 0);
}
break;
case BYTE:
if (object instanceof Byte) {
((LongColumnVector) columnVector).vector[batchIndex] = ((Byte) object);
} else {
((LongColumnVector) columnVector).vector[batchIndex] = ((ByteWritable) object).get();
}
break;
case SHORT:
if (object instanceof Short) {
((LongColumnVector) columnVector).vector[batchIndex] = ((Short) object);
} else {
((LongColumnVector) columnVector).vector[batchIndex] = ((ShortWritable) object).get();
}
break;
case INT:
if (object instanceof Integer) {
((LongColumnVector) columnVector).vector[batchIndex] = ((Integer) object);
} else {
((LongColumnVector) columnVector).vector[batchIndex] = ((IntWritable) object).get();
}
break;
case LONG:
if (object instanceof Long) {
((LongColumnVector) columnVector).vector[batchIndex] = ((Long) object);
} else {
((LongColumnVector) columnVector).vector[batchIndex] = ((LongWritable) object).get();
}
break;
case TIMESTAMP:
if (object instanceof Timestamp) {
((TimestampColumnVector) columnVector).set(batchIndex, ((Timestamp) object));
} else {
((TimestampColumnVector) columnVector).set(batchIndex, ((TimestampWritable) object).getTimestamp());
}
break;
case DATE:
if (object instanceof Date) {
((LongColumnVector) columnVector).vector[batchIndex] = DateWritable.dateToDays((Date) object);
} else {
((LongColumnVector) columnVector).vector[batchIndex] = ((DateWritable) object).getDays();
}
break;
case FLOAT:
if (object instanceof Float) {
((DoubleColumnVector) columnVector).vector[batchIndex] = ((Float) object);
} else {
((DoubleColumnVector) columnVector).vector[batchIndex] = ((FloatWritable) object).get();
}
break;
case DOUBLE:
if (object instanceof Double) {
((DoubleColumnVector) columnVector).vector[batchIndex] = ((Double) object);
} else {
((DoubleColumnVector) columnVector).vector[batchIndex] = ((DoubleWritable) object).get();
}
break;
case BINARY:
{
if (object instanceof byte[]) {
byte[] bytes = (byte[]) object;
((BytesColumnVector) columnVector).setVal(batchIndex, bytes, 0, bytes.length);
} else {
BytesWritable bw = (BytesWritable) object;
((BytesColumnVector) columnVector).setVal(batchIndex, bw.getBytes(), 0, bw.getLength());
}
}
break;
case STRING:
{
if (object instanceof String) {
String string = (String) object;
byte[] bytes = string.getBytes();
((BytesColumnVector) columnVector).setVal(batchIndex, bytes, 0, bytes.length);
} else {
Text tw = (Text) object;
((BytesColumnVector) columnVector).setVal(batchIndex, tw.getBytes(), 0, tw.getLength());
}
}
break;
case VARCHAR:
{
// UNDONE: Performance problem with conversion to String, then bytes...
// We store VARCHAR type stripped of pads.
HiveVarchar hiveVarchar;
if (object instanceof HiveVarchar) {
hiveVarchar = (HiveVarchar) object;
} else {
hiveVarchar = ((HiveVarcharWritable) object).getHiveVarchar();
}
// TODO: HIVE-13624 Do we need maxLength checking?
byte[] bytes = hiveVarchar.getValue().getBytes();
((BytesColumnVector) columnVector).setVal(batchIndex, bytes, 0, bytes.length);
}
break;
case CHAR:
{
// UNDONE: Performance problem with conversion to String, then bytes...
// We store CHAR type stripped of pads.
HiveChar hiveChar;
if (object instanceof HiveChar) {
hiveChar = (HiveChar) object;
} else {
hiveChar = ((HiveCharWritable) object).getHiveChar();
}
// TODO: HIVE-13624 Do we need maxLength checking?
// We store CHAR in vector row batch with padding stripped.
byte[] bytes = hiveChar.getStrippedValue().getBytes();
((BytesColumnVector) columnVector).setVal(batchIndex, bytes, 0, bytes.length);
}
break;
case DECIMAL:
if (object instanceof HiveDecimal) {
((DecimalColumnVector) columnVector).set(batchIndex, (HiveDecimal) object);
} else {
((DecimalColumnVector) columnVector).set(batchIndex, (HiveDecimalWritable) object);
}
break;
case INTERVAL_YEAR_MONTH:
if (object instanceof HiveIntervalYearMonth) {
((LongColumnVector) columnVector).vector[batchIndex] = ((HiveIntervalYearMonth) object).getTotalMonths();
} else {
((LongColumnVector) columnVector).vector[batchIndex] = ((HiveIntervalYearMonthWritable) object).getHiveIntervalYearMonth().getTotalMonths();
}
break;
case INTERVAL_DAY_TIME:
if (object instanceof HiveIntervalDayTime) {
((IntervalDayTimeColumnVector) columnVector).set(batchIndex, (HiveIntervalDayTime) object);
} else {
((IntervalDayTimeColumnVector) columnVector).set(batchIndex, ((HiveIntervalDayTimeWritable) object).getHiveIntervalDayTime());
}
break;
default:
throw new RuntimeException("Primitive category " + targetPrimitiveCategory.name() + " not supported");
}
}
break;
case LIST:
{
final ListColumnVector listColumnVector = (ListColumnVector) columnVector;
final ListTypeInfo listTypeInfo = (ListTypeInfo) targetTypeInfo;
final TypeInfo elementTypeInfo = listTypeInfo.getListElementTypeInfo();
final List list = (List) object;
final int size = list.size();
final int childCount = listColumnVector.childCount;
listColumnVector.offsets[batchIndex] = childCount;
listColumnVector.lengths[batchIndex] = size;
listColumnVector.childCount = childCount + size;
listColumnVector.child.ensureSize(childCount + size, true);
for (int i = 0; i < size; i++) {
assignRowColumn(listColumnVector.child, childCount + i, elementTypeInfo, list.get(i));
}
}
break;
case MAP:
{
final MapColumnVector mapColumnVector = (MapColumnVector) columnVector;
final MapTypeInfo mapTypeInfo = (MapTypeInfo) targetTypeInfo;
final Map<Object, Object> map = (Map<Object, Object>) object;
final int size = map.size();
int childCount = mapColumnVector.childCount;
mapColumnVector.offsets[batchIndex] = childCount;
mapColumnVector.lengths[batchIndex] = size;
mapColumnVector.keys.ensureSize(childCount + size, true);
mapColumnVector.values.ensureSize(childCount + size, true);
for (Map.Entry<Object, Object> entry : map.entrySet()) {
assignRowColumn(mapColumnVector.keys, childCount, mapTypeInfo.getMapKeyTypeInfo(), entry.getKey());
assignRowColumn(mapColumnVector.values, childCount, mapTypeInfo.getMapValueTypeInfo(), entry.getValue());
childCount++;
}
mapColumnVector.childCount = childCount;
}
break;
case STRUCT:
{
final StructColumnVector structColumnVector = (StructColumnVector) columnVector;
final StructTypeInfo targetStructTypeInfo = (StructTypeInfo) targetTypeInfo;
final List<TypeInfo> targetFieldTypeInfos = targetStructTypeInfo.getAllStructFieldTypeInfos();
final int size = targetFieldTypeInfos.size();
if (object instanceof List) {
final List struct = (List) object;
for (int i = 0; i < size; i++) {
assignRowColumn(structColumnVector.fields[i], batchIndex, targetFieldTypeInfos.get(i), struct.get(i));
}
} else {
final Object[] array = (Object[]) object;
for (int i = 0; i < size; i++) {
assignRowColumn(structColumnVector.fields[i], batchIndex, targetFieldTypeInfos.get(i), array[i]);
}
}
}
break;
case UNION:
{
final StandardUnion union = (StandardUnion) object;
final UnionColumnVector unionColumnVector = (UnionColumnVector) columnVector;
final UnionTypeInfo unionTypeInfo = (UnionTypeInfo) targetTypeInfo;
final List<TypeInfo> objectTypeInfos = unionTypeInfo.getAllUnionObjectTypeInfos();
final byte tag = union.getTag();
unionColumnVector.tags[batchIndex] = tag;
assignRowColumn(unionColumnVector.fields[tag], batchIndex, objectTypeInfos.get(tag), union.getObject());
}
break;
default:
throw new RuntimeException("Category " + targetTypeInfo.getCategory().name() + " not supported");
}
/*
* We always set the null flag to false when there is a value.
*/
columnVector.isNull[batchIndex] = false;
}
Aggregations