use of org.apache.hadoop.hive.serde2.proto.test.Complexpb.Complex in project hive by apache.
the class DeserializeRead method allocateCurrentWritable.
/*
* This class is used to read one field at a time. Simple fields like long, double, int are read
* into to primitive current* members; the non-simple field types like Date, Timestamp, etc, are
* read into a current object that this method will allocate.
*
* This method handles complex type fields by recursively calling this method.
*/
private void allocateCurrentWritable(TypeInfo typeInfo) {
switch(typeInfo.getCategory()) {
case PRIMITIVE:
switch(((PrimitiveTypeInfo) typeInfo).getPrimitiveCategory()) {
case DATE:
if (currentDateWritable == null) {
currentDateWritable = new DateWritable();
}
break;
case TIMESTAMP:
if (currentTimestampWritable == null) {
currentTimestampWritable = new TimestampWritable();
}
break;
case INTERVAL_YEAR_MONTH:
if (currentHiveIntervalYearMonthWritable == null) {
currentHiveIntervalYearMonthWritable = new HiveIntervalYearMonthWritable();
}
break;
case INTERVAL_DAY_TIME:
if (currentHiveIntervalDayTimeWritable == null) {
currentHiveIntervalDayTimeWritable = new HiveIntervalDayTimeWritable();
}
break;
case DECIMAL:
if (currentHiveDecimalWritable == null) {
currentHiveDecimalWritable = new HiveDecimalWritable();
}
break;
default:
}
break;
case LIST:
allocateCurrentWritable(((ListTypeInfo) typeInfo).getListElementTypeInfo());
break;
case MAP:
allocateCurrentWritable(((MapTypeInfo) typeInfo).getMapKeyTypeInfo());
allocateCurrentWritable(((MapTypeInfo) typeInfo).getMapValueTypeInfo());
break;
case STRUCT:
for (TypeInfo fieldTypeInfo : ((StructTypeInfo) typeInfo).getAllStructFieldTypeInfos()) {
allocateCurrentWritable(fieldTypeInfo);
}
break;
case UNION:
for (TypeInfo fieldTypeInfo : ((UnionTypeInfo) typeInfo).getAllUnionObjectTypeInfos()) {
allocateCurrentWritable(fieldTypeInfo);
}
break;
default:
throw new RuntimeException("Unexpected category " + typeInfo.getCategory());
}
}
use of org.apache.hadoop.hive.serde2.proto.test.Complexpb.Complex in project hive by apache.
the class ConstantPropagateProcFactory method evaluateFunction.
/**
* Evaluate UDF
*
* @param udf UDF object
* @param exprs
* @param oldExprs
* @return null if expression cannot be evaluated (not all parameters are constants). Or evaluated
* ExprNodeConstantDesc if possible.
* @throws HiveException
*/
private static ExprNodeDesc evaluateFunction(GenericUDF udf, List<ExprNodeDesc> exprs, List<ExprNodeDesc> oldExprs) {
DeferredJavaObject[] arguments = new DeferredJavaObject[exprs.size()];
ObjectInspector[] argois = new ObjectInspector[exprs.size()];
for (int i = 0; i < exprs.size(); i++) {
ExprNodeDesc desc = exprs.get(i);
if (desc instanceof ExprNodeConstantDesc) {
ExprNodeConstantDesc constant = (ExprNodeConstantDesc) exprs.get(i);
if (!constant.getTypeInfo().equals(oldExprs.get(i).getTypeInfo())) {
constant = typeCast(constant, oldExprs.get(i).getTypeInfo());
if (constant == null) {
return null;
}
}
if (constant.getTypeInfo().getCategory() != Category.PRIMITIVE) {
// nested complex types cannot be folded cleanly
return null;
}
Object value = constant.getValue();
PrimitiveTypeInfo pti = (PrimitiveTypeInfo) constant.getTypeInfo();
Object writableValue = null == value ? value : PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(pti).getPrimitiveWritableObject(value);
arguments[i] = new DeferredJavaObject(writableValue);
argois[i] = ObjectInspectorUtils.getConstantObjectInspector(constant.getWritableObjectInspector(), writableValue);
} else if (desc instanceof ExprNodeGenericFuncDesc) {
ExprNodeDesc evaluatedFn = foldExpr((ExprNodeGenericFuncDesc) desc);
if (null == evaluatedFn || !(evaluatedFn instanceof ExprNodeConstantDesc)) {
return null;
}
ExprNodeConstantDesc constant = (ExprNodeConstantDesc) evaluatedFn;
if (constant.getTypeInfo().getCategory() != Category.PRIMITIVE) {
// nested complex types cannot be folded cleanly
return null;
}
Object writableValue = PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector((PrimitiveTypeInfo) constant.getTypeInfo()).getPrimitiveWritableObject(constant.getValue());
arguments[i] = new DeferredJavaObject(writableValue);
argois[i] = ObjectInspectorUtils.getConstantObjectInspector(constant.getWritableObjectInspector(), writableValue);
} else {
return null;
}
}
try {
ObjectInspector oi = udf.initialize(argois);
Object o = udf.evaluate(arguments);
if (LOG.isDebugEnabled()) {
LOG.debug(udf.getClass().getName() + "(" + exprs + ")=" + o);
}
if (o == null) {
return new ExprNodeConstantDesc(TypeInfoUtils.getTypeInfoFromObjectInspector(oi), o);
}
Class<?> clz = o.getClass();
if (PrimitiveObjectInspectorUtils.isPrimitiveWritableClass(clz)) {
PrimitiveObjectInspector poi = (PrimitiveObjectInspector) oi;
TypeInfo typeInfo = poi.getTypeInfo();
o = poi.getPrimitiveJavaObject(o);
if (typeInfo.getTypeName().contains(serdeConstants.DECIMAL_TYPE_NAME) || typeInfo.getTypeName().contains(serdeConstants.VARCHAR_TYPE_NAME) || typeInfo.getTypeName().contains(serdeConstants.CHAR_TYPE_NAME) || typeInfo.getTypeName().contains(serdeConstants.TIMESTAMPLOCALTZ_TYPE_NAME)) {
return new ExprNodeConstantDesc(typeInfo, o);
}
} else if (udf instanceof GenericUDFStruct && oi instanceof StandardConstantStructObjectInspector) {
// do not fold named_struct, only struct()
ConstantObjectInspector coi = (ConstantObjectInspector) oi;
TypeInfo structType = TypeInfoUtils.getTypeInfoFromObjectInspector(coi);
return new ExprNodeConstantDesc(structType, ObjectInspectorUtils.copyToStandardJavaObject(o, coi));
} else if (!PrimitiveObjectInspectorUtils.isPrimitiveJavaClass(clz)) {
if (LOG.isErrorEnabled()) {
LOG.error("Unable to evaluate " + udf + ". Return value unrecoginizable.");
}
return null;
} else {
// fall through
}
String constStr = null;
if (arguments.length == 1 && FunctionRegistry.isOpCast(udf)) {
// remember original string representation of constant.
constStr = arguments[0].get().toString();
}
return new ExprNodeConstantDesc(o).setFoldedFromVal(constStr);
} catch (HiveException e) {
LOG.error("Evaluation function " + udf.getClass() + " failed in Constant Propagation Optimizer.");
throw new RuntimeException(e);
}
}
use of org.apache.hadoop.hive.serde2.proto.test.Complexpb.Complex in project hive by apache.
the class StatsUtils method getSizeOfComplexTypes.
/**
* Get the size of complex data types
* @return raw data size
*/
public static long getSizeOfComplexTypes(HiveConf conf, ObjectInspector oi) {
long result = 0;
int length = 0;
int listEntries = HiveConf.getIntVar(conf, HiveConf.ConfVars.HIVE_STATS_LIST_NUM_ENTRIES);
int mapEntries = HiveConf.getIntVar(conf, HiveConf.ConfVars.HIVE_STATS_MAP_NUM_ENTRIES);
switch(oi.getCategory()) {
case PRIMITIVE:
String colTypeLowerCase = oi.getTypeName().toLowerCase();
if (colTypeLowerCase.equals(serdeConstants.STRING_TYPE_NAME) || colTypeLowerCase.startsWith(serdeConstants.VARCHAR_TYPE_NAME) || colTypeLowerCase.startsWith(serdeConstants.CHAR_TYPE_NAME)) {
int avgColLen = (int) getAvgColLenOf(conf, oi, colTypeLowerCase);
result += JavaDataModel.get().lengthForStringOfLength(avgColLen);
} else if (colTypeLowerCase.equals(serdeConstants.BINARY_TYPE_NAME)) {
int avgColLen = (int) getAvgColLenOf(conf, oi, colTypeLowerCase);
result += JavaDataModel.get().lengthForByteArrayOfSize(avgColLen);
} else {
result += getAvgColLenOfFixedLengthTypes(colTypeLowerCase);
}
break;
case LIST:
if (oi instanceof StandardConstantListObjectInspector) {
// constant list projection of known length
StandardConstantListObjectInspector scloi = (StandardConstantListObjectInspector) oi;
length = scloi.getWritableConstantValue().size();
// check if list elements are primitive or Objects
ObjectInspector leoi = scloi.getListElementObjectInspector();
if (leoi.getCategory().equals(ObjectInspector.Category.PRIMITIVE)) {
result += getSizeOfPrimitiveTypeArraysFromType(leoi.getTypeName(), length, conf);
} else {
result += JavaDataModel.get().lengthForObjectArrayOfSize(length);
}
} else {
StandardListObjectInspector sloi = (StandardListObjectInspector) oi;
// list overhead + (configured number of element in list * size of element)
long elemSize = getSizeOfComplexTypes(conf, sloi.getListElementObjectInspector());
result += JavaDataModel.get().arrayList() + (listEntries * elemSize);
}
break;
case MAP:
if (oi instanceof StandardConstantMapObjectInspector) {
// constant map projection of known length
StandardConstantMapObjectInspector scmoi = (StandardConstantMapObjectInspector) oi;
result += getSizeOfMap(scmoi);
} else {
StandardMapObjectInspector smoi = (StandardMapObjectInspector) oi;
result += getSizeOfComplexTypes(conf, smoi.getMapKeyObjectInspector());
result += getSizeOfComplexTypes(conf, smoi.getMapValueObjectInspector());
// hash map overhead
result += JavaDataModel.get().hashMap(mapEntries);
}
break;
case STRUCT:
if (oi instanceof StandardConstantStructObjectInspector) {
// constant map projection of known length
StandardConstantStructObjectInspector scsoi = (StandardConstantStructObjectInspector) oi;
result += getSizeOfStruct(scsoi);
} else {
StructObjectInspector soi = (StructObjectInspector) oi;
// add constant object overhead for struct
result += JavaDataModel.get().object();
// add constant struct field names references overhead
result += soi.getAllStructFieldRefs().size() * JavaDataModel.get().ref();
for (StructField field : soi.getAllStructFieldRefs()) {
result += getSizeOfComplexTypes(conf, field.getFieldObjectInspector());
}
}
break;
case UNION:
UnionObjectInspector uoi = (UnionObjectInspector) oi;
// add constant object overhead for union
result += JavaDataModel.get().object();
// add constant size for unions tags
result += uoi.getObjectInspectors().size() * JavaDataModel.get().primitive1();
for (ObjectInspector foi : uoi.getObjectInspectors()) {
result += getSizeOfComplexTypes(conf, foi);
}
break;
default:
break;
}
return result;
}
use of org.apache.hadoop.hive.serde2.proto.test.Complexpb.Complex in project hive by apache.
the class VerifyFastRow method doVerifyDeserializeRead.
public static void doVerifyDeserializeRead(DeserializeRead deserializeRead, TypeInfo typeInfo, Object object, boolean isNull) throws IOException {
if (isNull) {
if (object != null) {
TestCase.fail("Field reports null but object is not null (class " + object.getClass().getName() + ", " + object.toString() + ")");
}
return;
} else if (object == null) {
TestCase.fail("Field report not null but object is null");
}
switch(typeInfo.getCategory()) {
case PRIMITIVE:
{
PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
switch(primitiveTypeInfo.getPrimitiveCategory()) {
case BOOLEAN:
{
boolean value = deserializeRead.currentBoolean;
if (!(object instanceof BooleanWritable)) {
TestCase.fail("Boolean expected writable not Boolean");
}
boolean expected = ((BooleanWritable) object).get();
if (value != expected) {
TestCase.fail("Boolean field mismatch (expected " + expected + " found " + value + ")");
}
}
break;
case BYTE:
{
byte value = deserializeRead.currentByte;
if (!(object instanceof ByteWritable)) {
TestCase.fail("Byte expected writable not Byte");
}
byte expected = ((ByteWritable) object).get();
if (value != expected) {
TestCase.fail("Byte field mismatch (expected " + (int) expected + " found " + (int) value + ")");
}
}
break;
case SHORT:
{
short value = deserializeRead.currentShort;
if (!(object instanceof ShortWritable)) {
TestCase.fail("Short expected writable not Short");
}
short expected = ((ShortWritable) object).get();
if (value != expected) {
TestCase.fail("Short field mismatch (expected " + expected + " found " + value + ")");
}
}
break;
case INT:
{
int value = deserializeRead.currentInt;
if (!(object instanceof IntWritable)) {
TestCase.fail("Integer expected writable not Integer");
}
int expected = ((IntWritable) object).get();
if (value != expected) {
TestCase.fail("Int field mismatch (expected " + expected + " found " + value + ")");
}
}
break;
case LONG:
{
long value = deserializeRead.currentLong;
if (!(object instanceof LongWritable)) {
TestCase.fail("Long expected writable not Long");
}
Long expected = ((LongWritable) object).get();
if (value != expected) {
TestCase.fail("Long field mismatch (expected " + expected + " found " + value + ")");
}
}
break;
case FLOAT:
{
float value = deserializeRead.currentFloat;
if (!(object instanceof FloatWritable)) {
TestCase.fail("Float expected writable not Float");
}
float expected = ((FloatWritable) object).get();
if (value != expected) {
TestCase.fail("Float field mismatch (expected " + expected + " found " + value + ")");
}
}
break;
case DOUBLE:
{
double value = deserializeRead.currentDouble;
if (!(object instanceof DoubleWritable)) {
TestCase.fail("Double expected writable not Double");
}
double expected = ((DoubleWritable) object).get();
if (value != expected) {
TestCase.fail("Double field mismatch (expected " + expected + " found " + value + ")");
}
}
break;
case STRING:
{
byte[] stringBytes = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
Text text = new Text(stringBytes);
String string = text.toString();
String expected = ((Text) object).toString();
if (!string.equals(expected)) {
TestCase.fail("String field mismatch (expected '" + expected + "' found '" + string + "')");
}
}
break;
case CHAR:
{
byte[] stringBytes = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
Text text = new Text(stringBytes);
String string = text.toString();
HiveChar hiveChar = new HiveChar(string, ((CharTypeInfo) primitiveTypeInfo).getLength());
HiveChar expected = ((HiveCharWritable) object).getHiveChar();
if (!hiveChar.equals(expected)) {
TestCase.fail("Char field mismatch (expected '" + expected + "' found '" + hiveChar + "')");
}
}
break;
case VARCHAR:
{
byte[] stringBytes = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
Text text = new Text(stringBytes);
String string = text.toString();
HiveVarchar hiveVarchar = new HiveVarchar(string, ((VarcharTypeInfo) primitiveTypeInfo).getLength());
HiveVarchar expected = ((HiveVarcharWritable) object).getHiveVarchar();
if (!hiveVarchar.equals(expected)) {
TestCase.fail("Varchar field mismatch (expected '" + expected + "' found '" + hiveVarchar + "')");
}
}
break;
case DECIMAL:
{
HiveDecimal value = deserializeRead.currentHiveDecimalWritable.getHiveDecimal();
if (value == null) {
TestCase.fail("Decimal field evaluated to NULL");
}
HiveDecimal expected = ((HiveDecimalWritable) object).getHiveDecimal();
if (!value.equals(expected)) {
DecimalTypeInfo decimalTypeInfo = (DecimalTypeInfo) primitiveTypeInfo;
int precision = decimalTypeInfo.getPrecision();
int scale = decimalTypeInfo.getScale();
TestCase.fail("Decimal field mismatch (expected " + expected.toString() + " found " + value.toString() + ") precision " + precision + ", scale " + scale);
}
}
break;
case DATE:
{
Date value = deserializeRead.currentDateWritable.get();
Date expected = ((DateWritable) object).get();
if (!value.equals(expected)) {
TestCase.fail("Date field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
}
}
break;
case TIMESTAMP:
{
Timestamp value = deserializeRead.currentTimestampWritable.getTimestamp();
Timestamp expected = ((TimestampWritable) object).getTimestamp();
if (!value.equals(expected)) {
TestCase.fail("Timestamp field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
}
}
break;
case INTERVAL_YEAR_MONTH:
{
HiveIntervalYearMonth value = deserializeRead.currentHiveIntervalYearMonthWritable.getHiveIntervalYearMonth();
HiveIntervalYearMonth expected = ((HiveIntervalYearMonthWritable) object).getHiveIntervalYearMonth();
if (!value.equals(expected)) {
TestCase.fail("HiveIntervalYearMonth field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
}
}
break;
case INTERVAL_DAY_TIME:
{
HiveIntervalDayTime value = deserializeRead.currentHiveIntervalDayTimeWritable.getHiveIntervalDayTime();
HiveIntervalDayTime expected = ((HiveIntervalDayTimeWritable) object).getHiveIntervalDayTime();
if (!value.equals(expected)) {
TestCase.fail("HiveIntervalDayTime field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
}
}
break;
case BINARY:
{
byte[] byteArray = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
BytesWritable bytesWritable = (BytesWritable) object;
byte[] expected = Arrays.copyOfRange(bytesWritable.getBytes(), 0, bytesWritable.getLength());
if (byteArray.length != expected.length) {
TestCase.fail("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(byteArray) + ")");
}
for (int b = 0; b < byteArray.length; b++) {
if (byteArray[b] != expected[b]) {
TestCase.fail("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(byteArray) + ")");
}
}
}
break;
default:
throw new Error("Unknown primitive category " + primitiveTypeInfo.getPrimitiveCategory());
}
}
break;
case LIST:
case MAP:
case STRUCT:
case UNION:
throw new Error("Complex types need to be handled separately");
default:
throw new Error("Unknown category " + typeInfo.getCategory());
}
}
use of org.apache.hadoop.hive.serde2.proto.test.Complexpb.Complex in project hive by apache.
the class TestNewInputOutputFormat method testNewOutputFormatComplex.
@SuppressWarnings("unchecked")
@Test
public // Test outputformat with complex data type, and with reduce
void testNewOutputFormatComplex() throws Exception {
Path inputPath = new Path(workDir, "TestOrcFile." + testCaseName.getMethodName() + ".txt");
Path outputPath = new Path(workDir, "TestOrcFile." + testCaseName.getMethodName() + ".orc");
localFs.delete(outputPath, true);
PrintWriter pw = new PrintWriter(new OutputStreamWriter(localFs.create(inputPath)));
pw.println("I have eaten");
pw.println("the plums");
pw.println("that were in");
pw.println("the icebox");
pw.println("and which");
pw.println("you were probably");
pw.println("saving");
pw.println("for breakfast");
pw.println("Forgive me");
pw.println("they were delicious");
pw.println("so sweet");
pw.println("and so cold");
pw.close();
Job job = new Job(conf, "orc test");
job.setOutputFormatClass(OrcNewOutputFormat.class);
job.setJarByClass(TestNewInputOutputFormat.class);
job.setMapperClass(OrcTestMapper3.class);
job.setReducerClass(OrcTestReducer3.class);
job.setMapOutputKeyClass(IntWritable.class);
job.setMapOutputValueClass(Text.class);
job.setOutputKeyClass(NullWritable.class);
job.setOutputValueClass(OrcSerdeRow.class);
FileInputFormat.addInputPath(job, inputPath);
FileOutputFormat.setOutputPath(job, outputPath);
boolean result = job.waitForCompletion(true);
assertTrue(result);
Path outputFilePath = new Path(outputPath, "part-r-00000");
Reader reader = OrcFile.createReader(outputFilePath, OrcFile.readerOptions(conf).filesystem(localFs));
RecordReader rows = reader.rows();
ObjectInspector orcOi = reader.getObjectInspector();
ObjectInspector stoi = TypeInfoUtils.getStandardJavaObjectInspectorFromTypeInfo(OrcTestReducer3.typeInfo);
ObjectInspectorConverters.Converter converter = ObjectInspectorConverters.getConverter(orcOi, stoi);
Object row = rows.next(null);
List<Object> converted = (List<Object>) converter.convert(row);
assertEquals(1, converted.get(0));
assertEquals(1, converted.get(1));
List<Object> list = (List<Object>) converted.get(2);
assertEquals(list.size(), 1);
assertEquals("saving", ((List<Object>) list.get(0)).get(0));
assertEquals(6, ((List<Object>) list.get(0)).get(1));
Map<String, Integer> map = (Map<String, Integer>) converted.get(3);
assertEquals(map.size(), 1);
assertEquals(map.get("saving"), new Integer(1));
row = rows.next(null);
converted = (List<Object>) converter.convert(row);
assertEquals(2, converted.get(0));
assertEquals(6, converted.get(1));
list = (List<Object>) converted.get(2);
assertEquals(list.size(), 6);
assertEquals("breakfast", ((List<Object>) list.get(0)).get(0));
assertEquals(9, ((List<Object>) list.get(0)).get(1));
map = (Map<String, Integer>) converted.get(3);
assertEquals(map.size(), 11);
assertEquals(map.get("the"), new Integer(2));
row = rows.next(null);
converted = (List<Object>) converter.convert(row);
assertEquals(3, converted.get(0));
assertEquals(5, converted.get(1));
list = (List<Object>) converted.get(2);
assertEquals(list.size(), 5);
assertEquals("cold", ((List<Object>) list.get(0)).get(0));
assertEquals(4, ((List<Object>) list.get(0)).get(1));
map = (Map<String, Integer>) converted.get(3);
assertEquals(map.size(), 13);
assertEquals(map.get("were"), new Integer(3));
assertFalse(rows.hasNext());
localFs.delete(outputPath, true);
}
Aggregations