@Override
public Writable serialize(Object obj, ObjectInspector oi) throws SerDeException {
if (oi.getCategory() != Category.STRUCT) {
throw new VoltSerdeException(
getClass().toString()
+ " can only serialize struct types, but we got: "
+ oi.getTypeName());
}
VoltRecord vr = new VoltRecord(m_voltConf.getTableName());
StructObjectInspector soi = (StructObjectInspector) oi;
List<? extends StructField> structFields = soi.getAllStructFieldRefs();
List<Object> fieldValues = soi.getStructFieldsDataAsList(obj);
final int size = m_oig.getColumnTypes().size();
for (int i = 0; i < size; ++i) {
ObjectInspector fieldOI = structFields.get(i).getFieldObjectInspector();
PrimitiveObjectInspector poi = (PrimitiveObjectInspector) fieldOI;
Object fieldValue = poi.getPrimitiveJavaObject(fieldValues.get(i));
if (poi.getTypeInfo().equals(TypeInfoFactory.timestampTypeInfo)) {
fieldValue = fieldValue != null ? new Date(((Timestamp) fieldValue).getTime()) : null;
}
vr.add(fieldValue);
}
return vr;
}
public static PrimitiveObjectInspector asDoubleCompatibleOI(@Nonnull final ObjectInspector argOI)
throws UDFArgumentTypeException {
if (argOI.getCategory() != Category.PRIMITIVE) {
throw new UDFArgumentTypeException(
0,
"Only primitive type arguments are accepted but " + argOI.getTypeName() + " is passed.");
}
final PrimitiveObjectInspector oi = (PrimitiveObjectInspector) argOI;
switch (oi.getPrimitiveCategory()) {
case BYTE:
case SHORT:
case INT:
case LONG:
case FLOAT:
case DOUBLE:
case STRING:
case TIMESTAMP:
break;
default:
throw new UDFArgumentTypeException(
0,
"Only numeric or string type arguments are accepted but "
+ argOI.getTypeName()
+ " is passed.");
}
return oi;
}
public void printFileInfo() throws Exception {
System.out.println("Reader: " + m_reader);
System.out.println("# Rows: " + m_reader.getNumberOfRows());
System.out.println("# Types in the file: " + m_types.size());
for (int i = 0; i < m_types.size(); i++) {
System.out.println("Type " + i + ": " + m_types.get(i).getKind());
}
System.out.println("Compression: " + m_reader.getCompression());
if (m_reader.getCompression() != CompressionKind.NONE) {
System.out.println("Compression size: " + m_reader.getCompressionSize());
}
m_oi = (StructObjectInspector) m_reader.getObjectInspector();
System.out.println("object inspector type category: " + m_oi.getCategory());
System.out.println("object inspector type name : " + m_oi.getTypeName());
System.out.println("Number of columns in the table: " + m_fields.size());
// Print the type info:
for (int i = 0; i < m_fields.size(); i++) {
System.out.println("Column " + i + " name: " + m_fields.get(i).getFieldName());
ObjectInspector lv_foi = m_fields.get(i).getFieldObjectInspector();
System.out.println("Column " + i + " type category: " + lv_foi.getCategory());
System.out.println("Column " + i + " type name: " + lv_foi.getTypeName());
}
}
public Writable serialize(Object obj, ObjectInspector objInspector) throws SerDeException {
if (objInspector.getCategory() != Category.STRUCT) {
throw new SerDeException(
getClass().toString()
+ " can only serialize struct types, but we got: "
+ objInspector.getTypeName());
}
StructObjectInspector soi = (StructObjectInspector) objInspector;
List<? extends StructField> fields = soi.getAllStructFieldRefs();
StringBuilder sb = new StringBuilder();
for (int i = 0; i < fields.size(); i++) {
if (i > 0) sb.append(separator);
Object column = soi.getStructFieldData(obj, fields.get(i));
if (fields.get(i).getFieldObjectInspector().getCategory() == Category.PRIMITIVE) {
// For primitive object, serialize to plain string
sb.append(column == null ? nullString : column.toString());
} else {
// For complex object, serialize to JSON format
sb.append(SerDeUtils.getJSONString(column, fields.get(i).getFieldObjectInspector()));
}
}
serializeCache.set(sb.toString());
return serializeCache;
}
@Override
protected String extractField(Object target) {
if (target instanceof HiveType) {
HiveType type = (HiveType) target;
ObjectInspector inspector = type.getObjectInspector();
if (inspector instanceof StructObjectInspector) {
StructObjectInspector soi = (StructObjectInspector) inspector;
StructField field = soi.getStructFieldRef(fieldName);
ObjectInspector foi = field.getFieldObjectInspector();
Assert.isTrue(
foi.getCategory() == ObjectInspector.Category.PRIMITIVE,
String.format(
"Field [%s] needs to be a primitive; found [%s]", fieldName, foi.getTypeName()));
// expecting a writeable - simply do a toString
Object data = soi.getStructFieldData(type.getObject(), field);
if (data == null || data instanceof NullWritable) {
return StringUtils.EMPTY;
}
return data.toString();
}
}
return null;
}
public static IntObjectInspector asIntOI(@Nonnull final ObjectInspector argOI)
throws UDFArgumentException {
if (!INT_TYPE_NAME.equals(argOI.getTypeName())) {
throw new UDFArgumentException("Argument type must be INT: " + argOI.getTypeName());
}
return (IntObjectInspector) argOI;
}
@Override
public Writable serialize(final Object obj, final ObjectInspector objInspector)
throws SerDeException {
if (!objInspector.getCategory().equals(Category.STRUCT)) {
throw new SerDeException(
"Cannot serialize " + objInspector.getCategory() + ". Can only serialize a struct");
}
final ArrayWritable serializeData = createStruct(obj, (StructObjectInspector) objInspector);
serializedSize = serializeData.get().length;
status = LAST_OPERATION.SERIALIZE;
return serializeData;
}
public static StructObjectInspector getTableObjectInspector(
@SuppressWarnings("deprecation") Deserializer deserializer) {
try {
ObjectInspector inspector = deserializer.getObjectInspector();
checkArgument(
inspector.getCategory() == Category.STRUCT,
"expected STRUCT: %s",
inspector.getCategory());
return (StructObjectInspector) inspector;
} catch (SerDeException e) {
throw Throwables.propagate(e);
}
}
/*
* validate and setup resultExprStr
*/
private void validateAndSetupResultExprStr(
MatchPath evaluator, List<PTFExpressionDef> args, int argsNum) throws SemanticException {
PTFExpressionDef resultExprArg = args.get(argsNum - 1);
ObjectInspector resultExprArgOI = resultExprArg.getOI();
if (!ObjectInspectorUtils.isConstantObjectInspector(resultExprArgOI)
|| (resultExprArgOI.getCategory() != ObjectInspector.Category.PRIMITIVE)
|| ((PrimitiveObjectInspector) resultExprArgOI).getPrimitiveCategory()
!= PrimitiveObjectInspector.PrimitiveCategory.STRING) {
throwErrorWithSignature("Currently the result Expr parameter must be a Constant String.");
}
evaluator.resultExprStr =
((ConstantObjectInspector) resultExprArgOI).getWritableConstantValue().toString();
}
public static Block serializeObject(
Type type, BlockBuilder builder, Object object, ObjectInspector inspector) {
switch (inspector.getCategory()) {
case PRIMITIVE:
serializePrimitive(type, builder, object, (PrimitiveObjectInspector) inspector);
return null;
case LIST:
return serializeList(type, builder, object, (ListObjectInspector) inspector);
case MAP:
return serializeMap(type, builder, object, (MapObjectInspector) inspector);
case STRUCT:
return serializeStruct(type, builder, object, (StructObjectInspector) inspector);
}
throw new RuntimeException("Unknown object inspector category: " + inspector.getCategory());
}
/*
* validate and setup patternStr
*/
private void validateAndSetupPatternStr(MatchPath evaluator, List<PTFExpressionDef> args)
throws SemanticException {
PTFExpressionDef symboPatternArg = args.get(0);
ObjectInspector symbolPatternArgOI = symboPatternArg.getOI();
if (!ObjectInspectorUtils.isConstantObjectInspector(symbolPatternArgOI)
|| (symbolPatternArgOI.getCategory() != ObjectInspector.Category.PRIMITIVE)
|| ((PrimitiveObjectInspector) symbolPatternArgOI).getPrimitiveCategory()
!= PrimitiveObjectInspector.PrimitiveCategory.STRING) {
throwErrorWithSignature("Currently the symbol Pattern must be a Constant String.");
}
evaluator.patternStr =
((ConstantObjectInspector) symbolPatternArgOI).getWritableConstantValue().toString();
}
private ObjectInspector solveOi(ObjectInspector arg) {
switch (arg.getCategory()) {
case PRIMITIVE:
// VOID, BOOLEAN, BYTE, SHORT, INT, LONG, FLOAT, DOUBLE, STRING, TIMESTAMP, BINARY, DECIMAL,
// UNKNOWN
PrimitiveObjectInspector poi = (PrimitiveObjectInspector) arg;
return PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(
poi.getPrimitiveCategory());
case LIST:
return ObjectInspectorFactory.getStandardListObjectInspector(
solveOi(((ListObjectInspector) arg).getListElementObjectInspector()));
case MAP:
return ObjectInspectorFactory.getStandardMapObjectInspector(
solveOi(((MapObjectInspector) arg).getMapKeyObjectInspector()),
solveOi(((MapObjectInspector) arg).getMapValueObjectInspector()));
case STRUCT:
StructObjectInspector soi = (StructObjectInspector) arg;
int size = soi.getAllStructFieldRefs().size();
ArrayList<String> fnl = new ArrayList<String>(size);
ArrayList<ObjectInspector> foil = new ArrayList<ObjectInspector>(size);
for (StructField sf : ((StructObjectInspector) arg).getAllStructFieldRefs()) {
fnl.add(sf.getFieldName());
foil.add(solveOi(sf.getFieldObjectInspector()));
}
return JsonStructObjectInspector.getJsonStructObjectInspector(fnl, foil);
default:
return arg;
}
}
// array<类型对象>
@Override
public String getTypeName() {
return org.apache.hadoop.hive.serde.serdeConstants.LIST_TYPE_NAME
+ "<"
+ listElementObjectInspector.getTypeName()
+ ">";
}
public static double getAsConstDouble(@Nonnull final ObjectInspector numberOI)
throws UDFArgumentException {
final String typeName = numberOI.getTypeName();
if (DOUBLE_TYPE_NAME.equals(typeName)) {
DoubleWritable v = getConstValue(numberOI);
return v.get();
} else if (FLOAT_TYPE_NAME.equals(typeName)) {
FloatWritable v = getConstValue(numberOI);
return v.get();
} else if (INT_TYPE_NAME.equals(typeName)) {
IntWritable v = getConstValue(numberOI);
return v.get();
} else if (BIGINT_TYPE_NAME.equals(typeName)) {
LongWritable v = getConstValue(numberOI);
return v.get();
} else if (SMALLINT_TYPE_NAME.equals(typeName)) {
ShortWritable v = getConstValue(numberOI);
return v.get();
} else if (TINYINT_TYPE_NAME.equals(typeName)) {
ByteWritable v = getConstValue(numberOI);
return v.get();
}
throw new UDFArgumentException(
"Unexpected argument type to cast as double: "
+ TypeInfoUtils.getTypeInfoFromObjectInspector(numberOI));
}
/**
* Convert a LazyObject to a standard Java object in compliance with JDBC 3.0 (see JDBC 3.0
* Specification, Table B-3: Mapping from JDBC Types to Java Object Types).
*
* <p>This method is kept consistent with {@link HiveResultSetMetaData#hiveTypeToSqlType}.
*/
private static Object convertLazyToJava(Object o, ObjectInspector oi) {
Object obj = ObjectInspectorUtils.copyToStandardObject(o, oi, ObjectInspectorCopyOption.JAVA);
if (obj == null) {
return null;
}
if (oi.getTypeName().equals(serdeConstants.BINARY_TYPE_NAME)) {
return new String((byte[]) obj);
}
// for now, expose non-primitive as a string
// TODO: expose non-primitive as a structured object while maintaining JDBC compliance
if (oi.getCategory() != ObjectInspector.Category.PRIMITIVE) {
return SerDeUtils.getJSONString(o, oi);
}
return obj;
}
// Serializes a row of data into a query string
@Override
public Writable serialize(Object obj, ObjectInspector objInspector) throws SerDeException {
LOG.info(obj.toString());
LOG.info(objInspector.toString());
return null;
}
/**
* Hive will call this to serialize an object. Returns a writable object of the same class
* returned by <a href="#getSerializedClass">getSerializedClass</a>
*
* @param obj The object to serialize
* @param objInspector The ObjectInspector that knows about the object's structure
* @return a serialized object in form of a Writable. Must be the same type returned by <a
* href="#getSerializedClass">getSerializedClass</a>
* @throws SerDeException
*/
@Override
public Writable serialize(Object obj, ObjectInspector objInspector) throws SerDeException {
// make sure it is a struct record
if (objInspector.getCategory() != Category.STRUCT) {
throw new SerDeException(
getClass().toString()
+ " can only serialize struct types, but we got: "
+ objInspector.getTypeName());
}
JSONObject serializer = serializeStruct(obj, (StructObjectInspector) objInspector, columnNames);
Text t = new Text(serializer.toString());
serializedDataSize = t.getBytes().length;
return t;
}
public static PrimitiveObjectInspector asPrimitiveObjectInspector(
@Nonnull final ObjectInspector oi) throws UDFArgumentException {
if (oi.getCategory() != Category.PRIMITIVE) {
throw new UDFArgumentException(
"Is not PrimitiveObjectInspector: " + TypeInfoUtils.getTypeInfoFromObjectInspector(oi));
}
return (PrimitiveObjectInspector) oi;
}
@Nonnull
public static ListObjectInspector asListOI(@Nonnull final ObjectInspector oi)
throws UDFArgumentException {
Category category = oi.getCategory();
if (category != Category.LIST) {
throw new UDFArgumentException("Expected List OI but was: " + oi);
}
return (ListObjectInspector) oi;
}
private Writable createObject(final Object obj, final ObjectInspector inspector)
throws SerDeException {
if (obj == null) {
return null;
}
switch (inspector.getCategory()) {
case STRUCT:
return createStruct(obj, (StructObjectInspector) inspector);
case LIST:
return createArray(obj, (ListObjectInspector) inspector);
case MAP:
return createMap(obj, (MapObjectInspector) inspector);
case PRIMITIVE:
return createPrimitive(obj, (PrimitiveObjectInspector) inspector);
default:
throw new SerDeException("Unknown data type" + inspector.getCategory());
}
}
/**
* Serializes a field. Since we have nested structures, it may be called recursively for instance
* when defining a list<struct<>>
*
* @param obj Object holding the fields' content
* @param oi The field's objec inspector
* @return the serialized object
*/
Object serializeField(Object obj, ObjectInspector oi) {
if (obj == null) {
return null;
}
Object result = null;
switch (oi.getCategory()) {
case PRIMITIVE:
PrimitiveObjectInspector poi = (PrimitiveObjectInspector) oi;
switch (poi.getPrimitiveCategory()) {
case VOID:
result = null;
break;
case BOOLEAN:
result = (((BooleanObjectInspector) poi).get(obj) ? Boolean.TRUE : Boolean.FALSE);
break;
case BYTE:
result = (((ShortObjectInspector) poi).get(obj));
break;
case DOUBLE:
result = (((DoubleObjectInspector) poi).get(obj));
break;
case FLOAT:
result = (((FloatObjectInspector) poi).get(obj));
break;
case INT:
result = (((IntObjectInspector) poi).get(obj));
break;
case LONG:
result = (((LongObjectInspector) poi).get(obj));
break;
case SHORT:
result = (((ShortObjectInspector) poi).get(obj));
break;
case STRING:
result = (((StringObjectInspector) poi).getPrimitiveJavaObject(obj));
break;
case UNKNOWN:
throw new RuntimeException("Unknown primitive");
}
break;
case MAP:
result = serializeMap(obj, (MapObjectInspector) oi);
break;
case LIST:
result = serializeList(obj, (ListObjectInspector) oi);
break;
case STRUCT:
result = serializeStruct(obj, (StructObjectInspector) oi, null);
break;
}
return result;
}
private ObjectInspectorConverters.Converter getConverter(ObjectInspector arg) {
switch (arg.getCategory()) {
case PRIMITIVE:
return ObjectInspectorConverters.getConverter(arg, arg);
case LIST:
case MAP:
case STRUCT:
return ObjectInspectorConverters.getConverter(arg, solveOi(arg));
default:
return null;
}
}
/*
* validate and setup SymbolInfo
*/
private void validateAndSetupSymbolInfo(
MatchPath evaluator, List<PTFExpressionDef> args, int argsNum) throws SemanticException {
int symbolArgsSz = argsNum - 2;
if (symbolArgsSz % 2 != 0) {
throwErrorWithSignature(
"Symbol Name, Expression need to be specified in pairs: "
+ "there are odd number of symbol args");
}
evaluator.symInfo = new SymbolsInfo(symbolArgsSz / 2);
for (int i = 1; i <= symbolArgsSz; i += 2) {
PTFExpressionDef symbolNameArg = args.get(i);
ObjectInspector symbolNameArgOI = symbolNameArg.getOI();
if (!ObjectInspectorUtils.isConstantObjectInspector(symbolNameArgOI)
|| (symbolNameArgOI.getCategory() != ObjectInspector.Category.PRIMITIVE)
|| ((PrimitiveObjectInspector) symbolNameArgOI).getPrimitiveCategory()
!= PrimitiveObjectInspector.PrimitiveCategory.STRING) {
throwErrorWithSignature(
String.format(
"Currently a Symbol Name(%s) must be a Constant String",
symbolNameArg.getExpressionTreeString()));
}
String symbolName =
((ConstantObjectInspector) symbolNameArgOI).getWritableConstantValue().toString();
PTFExpressionDef symolExprArg = args.get(i + 1);
ObjectInspector symolExprArgOI = symolExprArg.getOI();
if ((symolExprArgOI.getCategory() != ObjectInspector.Category.PRIMITIVE)
|| ((PrimitiveObjectInspector) symolExprArgOI).getPrimitiveCategory()
!= PrimitiveObjectInspector.PrimitiveCategory.BOOLEAN) {
throwErrorWithSignature(
String.format(
"Currently a Symbol Expression(%s) " + "must be a boolean expression",
symolExprArg.getExpressionTreeString()));
}
evaluator.symInfo.add(symbolName, symolExprArg);
}
}
/*
* For primitive types, use LazyBinary's object.
* For complex types, make a standard (Java) object from LazyBinary's object.
*/
public static List<Object> getComplexFieldsAsList(
LazyBinaryStruct lazyBinaryStruct,
ArrayList<Object> objectArrayBuffer,
LazyBinaryStructObjectInspector lazyBinaryStructObjectInspector) {
List<? extends StructField> fields = lazyBinaryStructObjectInspector.getAllStructFieldRefs();
for (int i = 0; i < fields.size(); i++) {
StructField field = fields.get(i);
ObjectInspector objectInspector = field.getFieldObjectInspector();
Category category = objectInspector.getCategory();
Object object = lazyBinaryStruct.getField(i);
if (category == Category.PRIMITIVE) {
objectArrayBuffer.set(i, object);
} else {
objectArrayBuffer.set(
i,
ObjectInspectorUtils.copyToStandardObject(
object, objectInspector, ObjectInspectorCopyOption.WRITABLE));
}
}
return objectArrayBuffer;
}
/** Create a hierarchical LazyBinaryObject based on the given typeInfo. */
public static LazyBinaryObject createLazyBinaryObject(ObjectInspector oi) {
ObjectInspector.Category c = oi.getCategory();
switch (c) {
case PRIMITIVE:
return createLazyBinaryPrimitiveClass((PrimitiveObjectInspector) oi);
case MAP:
return new LazyBinaryMap((LazyBinaryMapObjectInspector) oi);
case LIST:
return new LazyBinaryArray((LazyBinaryListObjectInspector) oi);
case STRUCT:
return new LazyBinaryStruct((LazyBinaryStructObjectInspector) oi);
}
throw new RuntimeException("Hive LazyBinarySerDe Internal error.");
}
/**
* Convert a Object to a standard Java object in compliance with JDBC 3.0 (see JDBC 3.0
* Specification, Table B-3: Mapping from JDBC Types to Java Object Types).
*
* <p>This method is kept consistent with {@link HiveResultSetMetaData#hiveTypeToSqlType}.
*/
public static Object toThriftPayload(Object val, ObjectInspector valOI, int version) {
if (valOI.getCategory() == ObjectInspector.Category.PRIMITIVE) {
if (val == null) {
return null;
}
Object obj =
ObjectInspectorUtils.copyToStandardObject(
val, valOI, ObjectInspectorUtils.ObjectInspectorCopyOption.JAVA);
// uses string type for binary before HIVE_CLI_SERVICE_PROTOCOL_V6
if (version < 5
&& ((PrimitiveObjectInspector) valOI).getPrimitiveCategory()
== PrimitiveObjectInspector.PrimitiveCategory.BINARY) {
// todo HIVE-5269
return new String((byte[]) obj);
}
return obj;
}
// for now, expose non-primitive as a string
// TODO: expose non-primitive as a structured object while maintaining JDBC compliance
return SerDeUtils.getJSONString(val, valOI);
}
@Override
@SuppressWarnings("unchecked")
public void process(Object row, int tag) throws HiveException {
try {
ObjectInspector rowInspector = inputObjInspectors[tag];
if (firstRow) {
firstRow = false;
// TODO: this is fishy - we init object inspectors based on first tag. We
// should either init for each tag, or if rowInspector doesn't really
// matter, then we can create this in ctor and get rid of firstRow.
if (conf.getWriteType() == AcidUtils.Operation.UPDATE
|| conf.getWriteType() == AcidUtils.Operation.DELETE) {
assert rowInspector instanceof StructObjectInspector
: "Exptected rowInspector to be instance of StructObjectInspector but it is a "
+ rowInspector.getClass().getName();
acidRowInspector = (StructObjectInspector) rowInspector;
// The record identifier is always in the first column
recIdField = acidRowInspector.getAllStructFieldRefs().get(0);
recIdInspector = (StructObjectInspector) recIdField.getFieldObjectInspector();
// The bucket field is in the second position
bucketField = recIdInspector.getAllStructFieldRefs().get(1);
bucketInspector = (IntObjectInspector) bucketField.getFieldObjectInspector();
}
if (isLogInfoEnabled) {
LOG.info(
"keys are "
+ conf.getOutputKeyColumnNames()
+ " num distributions: "
+ conf.getNumDistributionKeys());
}
keyObjectInspector =
initEvaluatorsAndReturnStruct(
keyEval,
distinctColIndices,
conf.getOutputKeyColumnNames(),
numDistributionKeys,
rowInspector);
valueObjectInspector =
initEvaluatorsAndReturnStruct(
valueEval, conf.getOutputValueColumnNames(), rowInspector);
partitionObjectInspectors = initEvaluators(partitionEval, rowInspector);
if (bucketEval != null) {
bucketObjectInspectors = initEvaluators(bucketEval, rowInspector);
}
int numKeys = numDistinctExprs > 0 ? numDistinctExprs : 1;
int keyLen = numDistinctExprs > 0 ? numDistributionKeys + 1 : numDistributionKeys;
cachedKeys = new Object[numKeys][keyLen];
cachedValues = new Object[valueEval.length];
}
// Determine distKeyLength (w/o distincts), and then add the first if present.
populateCachedDistributionKeys(row, 0);
// replace bucketing columns with hashcode % numBuckets
int bucketNumber = -1;
if (bucketEval != null) {
bucketNumber = computeBucketNumber(row, conf.getNumBuckets());
cachedKeys[0][buckColIdxInKey] = new Text(String.valueOf(bucketNumber));
} else if (conf.getWriteType() == AcidUtils.Operation.UPDATE
|| conf.getWriteType() == AcidUtils.Operation.DELETE) {
// In the non-partitioned case we still want to compute the bucket number for updates and
// deletes.
bucketNumber = computeBucketNumber(row, conf.getNumBuckets());
}
HiveKey firstKey = toHiveKey(cachedKeys[0], tag, null);
int distKeyLength = firstKey.getDistKeyLength();
if (numDistinctExprs > 0) {
populateCachedDistinctKeys(row, 0);
firstKey = toHiveKey(cachedKeys[0], tag, distKeyLength);
}
final int hashCode;
// distKeyLength doesn't include tag, but includes buckNum in cachedKeys[0]
if (useUniformHash && partitionEval.length > 0) {
hashCode = computeMurmurHash(firstKey);
} else {
hashCode = computeHashCode(row, bucketNumber);
}
firstKey.setHashCode(hashCode);
/*
* in case of TopN for windowing, we need to distinguish between rows with
* null partition keys and rows with value 0 for partition keys.
*/
boolean partKeyNull = conf.isPTFReduceSink() && partitionKeysAreNull(row);
// Try to store the first key. If it's not excluded, we will proceed.
int firstIndex = reducerHash.tryStoreKey(firstKey, partKeyNull);
if (firstIndex == TopNHash.EXCLUDE) return; // Nothing to do.
// Compute value and hashcode - we'd either store or forward them.
BytesWritable value = makeValueWritable(row);
if (firstIndex == TopNHash.FORWARD) {
collect(firstKey, value);
} else {
assert firstIndex >= 0;
reducerHash.storeValue(firstIndex, firstKey.hashCode(), value, false);
}
// All other distinct keys will just be forwarded. This could be optimized...
for (int i = 1; i < numDistinctExprs; i++) {
System.arraycopy(cachedKeys[0], 0, cachedKeys[i], 0, numDistributionKeys);
populateCachedDistinctKeys(row, i);
HiveKey hiveKey = toHiveKey(cachedKeys[i], tag, distKeyLength);
hiveKey.setHashCode(hashCode);
collect(hiveKey, value);
}
} catch (HiveException e) {
throw e;
} catch (Exception e) {
throw new HiveException(e);
}
}
public String getType() {
return objectInspector.getTypeName();
}
// This method is just for experimentation.
public void testRead() throws Exception {
m_reader = OrcFile.createReader(m_file_path, OrcFile.readerOptions(m_conf));
System.out.println("Reader: " + m_reader);
System.out.println("# Rows: " + m_reader.getNumberOfRows());
m_types = m_reader.getTypes();
System.out.println("# Types in the file: " + m_types.size());
for (int i = 0; i < m_types.size(); i++) {
System.out.println("Type " + i + ": " + m_types.get(i).getKind());
}
System.out.println("Compression: " + m_reader.getCompression());
if (m_reader.getCompression() != CompressionKind.NONE) {
System.out.println("Compression size: " + m_reader.getCompressionSize());
}
StructObjectInspector m_oi = (StructObjectInspector) m_reader.getObjectInspector();
System.out.println("object inspector type category: " + m_oi.getCategory());
System.out.println("object inspector type name : " + m_oi.getTypeName());
m_fields = m_oi.getAllStructFieldRefs();
System.out.println("Number of columns in the table: " + m_fields.size());
RecordReader m_rr = m_reader.rows();
// Print the type info:
for (int i = 0; i < m_fields.size(); i++) {
System.out.println("Column " + i + " name: " + m_fields.get(i).getFieldName());
ObjectInspector lv_foi = m_fields.get(i).getFieldObjectInspector();
System.out.println("Column " + i + " type category: " + lv_foi.getCategory());
System.out.println("Column " + i + " type name: " + lv_foi.getTypeName());
// Object lv_column_val = m_oi.getStructFieldData(lv_row, m_fields.get(i));
// System.out.print("Column " + i + " value: " + lv_row.getFieldValue(i));
}
OrcStruct lv_row = null;
Object lv_field_val = null;
StringBuilder lv_row_string = new StringBuilder(1024);
while (m_rr.hasNext()) {
lv_row = (OrcStruct) m_rr.next(lv_row);
lv_row_string.setLength(0);
for (int i = 0; i < m_fields.size(); i++) {
lv_field_val = lv_row.getFieldValue(i);
if (lv_field_val != null) {
lv_row_string.append(lv_field_val);
}
lv_row_string.append('|');
}
System.out.println(lv_row_string);
}
/**
* Typecasting to appropriate type based on the 'kind' if (OrcProto.Type.Kind.INT ==
* m_types.get(1).getKind()) { IntWritable lvf_1_val = (IntWritable) lv_row.getFieldValue(0);
* System.out.println("Column 1 value: " + lvf_1_val); }
*/
}
public static boolean isBooleanOI(@Nonnull final ObjectInspector oi) {
String typeName = oi.getTypeName();
return BOOLEAN_TYPE_NAME.equals(typeName);
}