Examples with PRIMITIVE org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category.PRIMITIVE used on opensource projects

Example 76 with PRIMITIVE

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category.PRIMITIVE 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;
                List<?> value = scloi.getWritableConstantValue();
                if (null == value) {
                    length = 0;
                } else {
                    length = value.size();
                }
                // check if list elements are primitive or Objects
                ObjectInspector leoi = scloi.getListElementObjectInspector();
                if (leoi.getCategory().equals(ObjectInspector.Category.PRIMITIVE)) {
                    int maxVarLen = HiveConf.getIntVar(conf, HiveConf.ConfVars.HIVE_STATS_MAX_VARIABLE_LENGTH);
                    result += getSizeOfPrimitiveTypeArraysFromType(leoi.getTypeName(), length, maxVarLen);
                } 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;
}

Example 77 with PRIMITIVE

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category.PRIMITIVE in project hive by apache.

the class GenericUDFTrunc method initializeDate.

private ObjectInspector initializeDate(ObjectInspector[] arguments) throws UDFArgumentLengthException, UDFArgumentTypeException {
    if (arguments.length != 2) {
        throw new UDFArgumentLengthException("trunc() requires 2 argument, got " + arguments.length);
    }
    if (arguments[0].getCategory() != ObjectInspector.Category.PRIMITIVE) {
        throw new UDFArgumentTypeException(0, "Only primitive type arguments are accepted but " + arguments[0].getTypeName() + " is passed. as first arguments");
    }
    if (arguments[1].getCategory() != ObjectInspector.Category.PRIMITIVE) {
        throw new UDFArgumentTypeException(1, "Only primitive type arguments are accepted but " + arguments[1].getTypeName() + " is passed. as second arguments");
    }
    ObjectInspector outputOI = PrimitiveObjectInspectorFactory.writableStringObjectInspector;
    inputType1 = ((PrimitiveObjectInspector) arguments[0]).getPrimitiveCategory();
    switch(inputType1) {
        case STRING:
        case VARCHAR:
        case CHAR:
        case VOID:
            inputType1 = PrimitiveCategory.STRING;
            textConverter1 = ObjectInspectorConverters.getConverter(arguments[0], PrimitiveObjectInspectorFactory.writableStringObjectInspector);
            break;
        case TIMESTAMP:
            timestampConverter = new TimestampConverter((PrimitiveObjectInspector) arguments[0], PrimitiveObjectInspectorFactory.writableTimestampObjectInspector);
            break;
        case DATE:
            dateWritableConverter = ObjectInspectorConverters.getConverter(arguments[0], PrimitiveObjectInspectorFactory.writableDateObjectInspector);
            break;
        default:
            throw new UDFArgumentTypeException(0, "TRUNC() only takes STRING/TIMESTAMP/DATEWRITABLE types as first argument, got " + inputType1);
    }
    inputType2 = ((PrimitiveObjectInspector) arguments[1]).getPrimitiveCategory();
    if (PrimitiveObjectInspectorUtils.getPrimitiveGrouping(inputType2) != PrimitiveGrouping.STRING_GROUP && PrimitiveObjectInspectorUtils.getPrimitiveGrouping(inputType2) != PrimitiveGrouping.VOID_GROUP) {
        throw new UDFArgumentTypeException(1, "trunc() only takes STRING/CHAR/VARCHAR types as second argument, got " + inputType2);
    }
    inputType2 = PrimitiveCategory.STRING;
    if (arguments[1] instanceof ConstantObjectInspector) {
        Object obj = ((ConstantObjectInspector) arguments[1]).getWritableConstantValue();
        fmtInput = obj != null ? obj.toString() : null;
    } else {
        textConverter2 = ObjectInspectorConverters.getConverter(arguments[1], PrimitiveObjectInspectorFactory.writableStringObjectInspector);
    }
    return outputOI;
}

Example 78 with PRIMITIVE

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category.PRIMITIVE in project hive by apache.

the class GenericUDFToVarchar method initialize.

@Override
public ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException {
    if (arguments.length != 1) {
        throw new UDFArgumentException("VARCHAR cast requires a value argument");
    }
    try {
        argumentOI = (PrimitiveObjectInspector) arguments[0];
    } catch (ClassCastException e) {
        throw new UDFArgumentException("The function VARCHAR takes only primitive types");
    }
    // Check if this UDF has been provided with type params for the output varchar type
    SettableHiveVarcharObjectInspector outputOI;
    outputOI = (SettableHiveVarcharObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(typeInfo);
    converter = new HiveVarcharConverter(argumentOI, outputOI);
    return outputOI;
}

Example 79 with PRIMITIVE

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category.PRIMITIVE in project hive by apache.

the class GenericUDFToChar method initialize.

@Override
public ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException {
    if (arguments.length != 1) {
        throw new UDFArgumentException("CHAR cast requires a value argument");
    }
    try {
        argumentOI = (PrimitiveObjectInspector) arguments[0];
    } catch (ClassCastException e) {
        throw new UDFArgumentException("The function CHAR takes only primitive types");
    }
    // Check if this UDF has been provided with type params for the output char type
    SettableHiveCharObjectInspector outputOI;
    outputOI = (SettableHiveCharObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(typeInfo);
    converter = new HiveCharConverter(argumentOI, outputOI);
    return outputOI;
}

Example 80 with PRIMITIVE

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category.PRIMITIVE in project hive by apache.

the class VerifyFastRow method serializeWrite.

public static void serializeWrite(SerializeWrite serializeWrite, TypeInfo typeInfo, Object object) throws IOException {
    if (object == null) {
        serializeWrite.writeNull();
        return;
    }
    switch(typeInfo.getCategory()) {
        case PRIMITIVE:
            {
                PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
                switch(primitiveTypeInfo.getPrimitiveCategory()) {
                    case BOOLEAN:
                        {
                            boolean value = ((BooleanWritable) object).get();
                            serializeWrite.writeBoolean(value);
                        }
                        break;
                    case BYTE:
                        {
                            byte value = ((ByteWritable) object).get();
                            serializeWrite.writeByte(value);
                        }
                        break;
                    case SHORT:
                        {
                            short value = ((ShortWritable) object).get();
                            serializeWrite.writeShort(value);
                        }
                        break;
                    case INT:
                        {
                            int value = ((IntWritable) object).get();
                            serializeWrite.writeInt(value);
                        }
                        break;
                    case LONG:
                        {
                            long value = ((LongWritable) object).get();
                            serializeWrite.writeLong(value);
                        }
                        break;
                    case FLOAT:
                        {
                            float value = ((FloatWritable) object).get();
                            serializeWrite.writeFloat(value);
                        }
                        break;
                    case DOUBLE:
                        {
                            double value = ((DoubleWritable) object).get();
                            serializeWrite.writeDouble(value);
                        }
                        break;
                    case STRING:
                        {
                            Text value = (Text) object;
                            byte[] stringBytes = value.getBytes();
                            int stringLength = stringBytes.length;
                            serializeWrite.writeString(stringBytes, 0, stringLength);
                        }
                        break;
                    case CHAR:
                        {
                            HiveChar value = ((HiveCharWritable) object).getHiveChar();
                            serializeWrite.writeHiveChar(value);
                        }
                        break;
                    case VARCHAR:
                        {
                            HiveVarchar value = ((HiveVarcharWritable) object).getHiveVarchar();
                            serializeWrite.writeHiveVarchar(value);
                        }
                        break;
                    case DECIMAL:
                        {
                            HiveDecimal value = ((HiveDecimalWritable) object).getHiveDecimal();
                            DecimalTypeInfo decTypeInfo = (DecimalTypeInfo) primitiveTypeInfo;
                            serializeWrite.writeHiveDecimal(value, decTypeInfo.scale());
                        }
                        break;
                    case DATE:
                        {
                            Date value = ((DateWritableV2) object).get();
                            serializeWrite.writeDate(value);
                        }
                        break;
                    case TIMESTAMP:
                        {
                            Timestamp value = ((TimestampWritableV2) object).getTimestamp();
                            serializeWrite.writeTimestamp(value);
                        }
                        break;
                    case INTERVAL_YEAR_MONTH:
                        {
                            HiveIntervalYearMonth value = ((HiveIntervalYearMonthWritable) object).getHiveIntervalYearMonth();
                            serializeWrite.writeHiveIntervalYearMonth(value);
                        }
                        break;
                    case INTERVAL_DAY_TIME:
                        {
                            HiveIntervalDayTime value = ((HiveIntervalDayTimeWritable) object).getHiveIntervalDayTime();
                            serializeWrite.writeHiveIntervalDayTime(value);
                        }
                        break;
                    case BINARY:
                        {
                            BytesWritable byteWritable = (BytesWritable) object;
                            byte[] binaryBytes = byteWritable.getBytes();
                            int length = byteWritable.getLength();
                            serializeWrite.writeBinary(binaryBytes, 0, length);
                        }
                        break;
                    default:
                        throw new Error("Unknown primitive category " + primitiveTypeInfo.getPrimitiveCategory().name());
                }
            }
            break;
        case LIST:
            {
                ListTypeInfo listTypeInfo = (ListTypeInfo) typeInfo;
                TypeInfo elementTypeInfo = listTypeInfo.getListElementTypeInfo();
                ArrayList<Object> elements = (ArrayList<Object>) object;
                serializeWrite.beginList(elements);
                boolean isFirst = true;
                for (Object elementObject : elements) {
                    if (isFirst) {
                        isFirst = false;
                    } else {
                        serializeWrite.separateList();
                    }
                    if (elementObject == null) {
                        serializeWrite.writeNull();
                    } else {
                        serializeWrite(serializeWrite, elementTypeInfo, elementObject);
                    }
                }
                serializeWrite.finishList();
            }
            break;
        case MAP:
            {
                MapTypeInfo mapTypeInfo = (MapTypeInfo) typeInfo;
                TypeInfo keyTypeInfo = mapTypeInfo.getMapKeyTypeInfo();
                TypeInfo valueTypeInfo = mapTypeInfo.getMapValueTypeInfo();
                Map<Object, Object> hashMap = (Map<Object, Object>) object;
                serializeWrite.beginMap(hashMap);
                boolean isFirst = true;
                for (Map.Entry<Object, Object> entry : hashMap.entrySet()) {
                    if (isFirst) {
                        isFirst = false;
                    } else {
                        serializeWrite.separateKeyValuePair();
                    }
                    if (entry.getKey() == null) {
                        serializeWrite.writeNull();
                    } else {
                        serializeWrite(serializeWrite, keyTypeInfo, entry.getKey());
                    }
                    serializeWrite.separateKey();
                    if (entry.getValue() == null) {
                        serializeWrite.writeNull();
                    } else {
                        serializeWrite(serializeWrite, valueTypeInfo, entry.getValue());
                    }
                }
                serializeWrite.finishMap();
            }
            break;
        case STRUCT:
            {
                StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
                List<TypeInfo> fieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
                List<Object> fieldValues = (List<Object>) object;
                final int size = fieldValues.size();
                serializeWrite.beginStruct(fieldValues);
                boolean isFirst = true;
                for (int i = 0; i < size; i++) {
                    if (isFirst) {
                        isFirst = false;
                    } else {
                        serializeWrite.separateStruct();
                    }
                    serializeWrite(serializeWrite, fieldTypeInfos.get(i), fieldValues.get(i));
                }
                serializeWrite.finishStruct();
            }
            break;
        case UNION:
            {
                UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
                List<TypeInfo> fieldTypeInfos = unionTypeInfo.getAllUnionObjectTypeInfos();
                final int size = fieldTypeInfos.size();
                StandardUnionObjectInspector.StandardUnion standardUnion = (StandardUnionObjectInspector.StandardUnion) object;
                byte tag = standardUnion.getTag();
                serializeWrite.beginUnion(tag);
                serializeWrite(serializeWrite, fieldTypeInfos.get(tag), standardUnion.getObject());
                serializeWrite.finishUnion();
            }
            break;
        default:
            throw new Error("Unknown category " + typeInfo.getCategory().name());
    }
}