Examples with UnionTypeInfo org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo used on opensource projects

Example 26 with UnionTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo in project hive by apache.

the class VectorSerializeRow method createField.

private Field createField(TypeInfo typeInfo) {
    final Field field = new Field();
    final Category category = typeInfo.getCategory();
    field.category = category;
    field.typeInfo = typeInfo;
    if (category == Category.PRIMITIVE) {
        field.isPrimitive = true;
        field.primitiveCategory = ((PrimitiveTypeInfo) typeInfo).getPrimitiveCategory();
        switch(field.primitiveCategory) {
            case BOOLEAN:
                field.writer = new VectorSerializeBooleanWriter();
                break;
            case BYTE:
                field.writer = new VectorSerializeByteWriter();
                break;
            case SHORT:
                field.writer = new VectorSerializeShortWriter();
                break;
            case INT:
                field.writer = new VectorSerializeIntWriter();
                break;
            case LONG:
                field.writer = new VectorSerializeLongWriter();
                break;
            case DATE:
                field.writer = new VectorSerializeDateWriter();
                break;
            case TIMESTAMP:
                field.writer = new VectorSerializeTimestampWriter();
                break;
            case FLOAT:
                field.writer = new VectorSerializeFloatWriter();
                break;
            case DOUBLE:
                field.writer = new VectorSerializeDoubleWriter();
                break;
            case STRING:
            case CHAR:
            case VARCHAR:
                field.writer = new VectorSerializeStringWriter();
                break;
            case BINARY:
                field.writer = new VectorSerializeBinaryWriter();
                break;
            case DECIMAL:
                field.writer = new VectorSerializeDecimalWriter();
                break;
            case INTERVAL_YEAR_MONTH:
                field.writer = new VectorSerializeHiveIntervalYearMonthWriter();
                break;
            case INTERVAL_DAY_TIME:
                field.writer = new VectorSerializeHiveIntervalDayTimeWriter();
                break;
            default:
                throw new RuntimeException("Unexpected primitive category " + field.primitiveCategory);
        }
    } else {
        field.isPrimitive = false;
        field.objectInspector = TypeInfoUtils.getStandardJavaObjectInspectorFromTypeInfo(typeInfo);
        switch(category) {
            case LIST:
                field.children = new Field[1];
                field.children[0] = createField(((ListTypeInfo) typeInfo).getListElementTypeInfo());
                field.writer = new VectorSerializeListWriter();
                break;
            case MAP:
                field.children = new Field[2];
                field.children[0] = createField(((MapTypeInfo) typeInfo).getMapKeyTypeInfo());
                field.children[1] = createField(((MapTypeInfo) typeInfo).getMapValueTypeInfo());
                field.writer = new VectorSerializeMapWriter();
                break;
            case STRUCT:
                StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
                List<TypeInfo> fieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
                field.children = createFields(fieldTypeInfos.toArray(new TypeInfo[fieldTypeInfos.size()]));
                field.writer = new VectorSerializeStructWriter();
                break;
            case UNION:
                UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
                List<TypeInfo> objectTypeInfos = unionTypeInfo.getAllUnionObjectTypeInfos();
                field.children = createFields(objectTypeInfos.toArray(new TypeInfo[objectTypeInfos.size()]));
                field.writer = new VectorSerializeUnionWriter();
                break;
            default:
                throw new RuntimeException();
        }
        field.count = field.children.length;
    }
    return field;
}

Example 27 with UnionTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo in project hive by apache.

the class VerifyLazy method lazyCompare.

public static boolean lazyCompare(TypeInfo typeInfo, Object lazyObject, Object expectedObject) {
    if (expectedObject == null) {
        if (lazyObject != null) {
            throw new RuntimeException("Expected object is null but object is not null " + lazyObject.toString() + " typeInfo " + typeInfo.toString());
        }
        return true;
    } else if (lazyObject == null) {
        throw new RuntimeException("Expected object is not null \"" + expectedObject.toString() + "\" typeInfo " + typeInfo.toString() + " but object is null");
    }
    if (lazyObject instanceof Writable) {
        if (!lazyObject.equals(expectedObject)) {
            throw new RuntimeException("Expected object " + expectedObject.toString() + " and actual object " + lazyObject.toString() + " is not equal typeInfo " + typeInfo.toString());
        }
        return true;
    }
    if (lazyObject instanceof LazyPrimitive) {
        Object primitiveObject = ((LazyPrimitive) lazyObject).getObject();
        PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
        switch(primitiveTypeInfo.getPrimitiveCategory()) {
            case BOOLEAN:
                {
                    if (!(primitiveObject instanceof LazyBoolean)) {
                        throw new RuntimeException("Expected LazyBoolean");
                    }
                    boolean value = ((LazyBoolean) primitiveObject).getWritableObject().get();
                    boolean expected = ((BooleanWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Boolean field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case BYTE:
                {
                    if (!(primitiveObject instanceof LazyByte)) {
                        throw new RuntimeException("Expected LazyByte");
                    }
                    byte value = ((LazyByte) primitiveObject).getWritableObject().get();
                    byte expected = ((ByteWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Byte field mismatch (expected " + (int) expected + " found " + (int) value + ")");
                    }
                }
                break;
            case SHORT:
                {
                    if (!(primitiveObject instanceof LazyShort)) {
                        throw new RuntimeException("Expected LazyShort");
                    }
                    short value = ((LazyShort) primitiveObject).getWritableObject().get();
                    short expected = ((ShortWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Short field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case INT:
                {
                    if (!(primitiveObject instanceof LazyInteger)) {
                        throw new RuntimeException("Expected LazyInteger");
                    }
                    int value = ((LazyInteger) primitiveObject).getWritableObject().get();
                    int expected = ((IntWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Int field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case LONG:
                {
                    if (!(primitiveObject instanceof LazyLong)) {
                        throw new RuntimeException("Expected LazyLong");
                    }
                    long value = ((LazyLong) primitiveObject).getWritableObject().get();
                    long expected = ((LongWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Long field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case FLOAT:
                {
                    if (!(primitiveObject instanceof LazyFloat)) {
                        throw new RuntimeException("Expected LazyFloat");
                    }
                    float value = ((LazyFloat) primitiveObject).getWritableObject().get();
                    float expected = ((FloatWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Float field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case DOUBLE:
                {
                    if (!(primitiveObject instanceof LazyDouble)) {
                        throw new RuntimeException("Expected LazyDouble");
                    }
                    double value = ((LazyDouble) primitiveObject).getWritableObject().get();
                    double expected = ((DoubleWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Double field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case STRING:
                {
                    if (!(primitiveObject instanceof LazyString)) {
                        throw new RuntimeException("Text expected writable not Text");
                    }
                    Text value = ((LazyString) primitiveObject).getWritableObject();
                    Text expected = ((Text) expectedObject);
                    if (!value.equals(expected)) {
                        throw new RuntimeException("String field mismatch (expected '" + expected + "' found '" + value + "')");
                    }
                }
                break;
            case CHAR:
                {
                    if (!(primitiveObject instanceof LazyHiveChar)) {
                        throw new RuntimeException("Expected LazyHiveChar");
                    }
                    HiveChar value = ((LazyHiveChar) primitiveObject).getWritableObject().getHiveChar();
                    HiveChar expected = ((HiveCharWritable) expectedObject).getHiveChar();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveChar field mismatch (expected '" + expected + "' found '" + value + "')");
                    }
                }
                break;
            case VARCHAR:
                {
                    if (!(primitiveObject instanceof LazyHiveVarchar)) {
                        throw new RuntimeException("Expected LazyHiveVarchar");
                    }
                    HiveVarchar value = ((LazyHiveVarchar) primitiveObject).getWritableObject().getHiveVarchar();
                    HiveVarchar expected = ((HiveVarcharWritable) expectedObject).getHiveVarchar();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveVarchar field mismatch (expected '" + expected + "' found '" + value + "')");
                    }
                }
                break;
            case DECIMAL:
                {
                    if (!(primitiveObject instanceof LazyHiveDecimal)) {
                        throw new RuntimeException("Expected LazyDecimal");
                    }
                    HiveDecimal value = ((LazyHiveDecimal) primitiveObject).getWritableObject().getHiveDecimal();
                    HiveDecimal expected = ((HiveDecimalWritable) expectedObject).getHiveDecimal();
                    if (!value.equals(expected)) {
                        DecimalTypeInfo decimalTypeInfo = (DecimalTypeInfo) primitiveTypeInfo;
                        int precision = decimalTypeInfo.getPrecision();
                        int scale = decimalTypeInfo.getScale();
                        throw new RuntimeException("Decimal field mismatch (expected " + expected.toString() + " found " + value.toString() + ") precision " + precision + ", scale " + scale);
                    }
                }
                break;
            case DATE:
                {
                    if (!(primitiveObject instanceof LazyDate)) {
                        throw new RuntimeException("Expected LazyDate");
                    }
                    Date value = ((LazyDate) primitiveObject).getWritableObject().get();
                    Date expected = ((DateWritableV2) expectedObject).get();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("Date field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case TIMESTAMP:
                {
                    if (!(primitiveObject instanceof LazyTimestamp)) {
                        throw new RuntimeException("TimestampWritableV2 expected writable not TimestampWritableV2");
                    }
                    Timestamp value = ((LazyTimestamp) primitiveObject).getWritableObject().getTimestamp();
                    Timestamp expected = ((TimestampWritableV2) expectedObject).getTimestamp();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("Timestamp field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case INTERVAL_YEAR_MONTH:
                {
                    if (!(primitiveObject instanceof LazyHiveIntervalYearMonth)) {
                        throw new RuntimeException("Expected LazyHiveIntervalYearMonth");
                    }
                    HiveIntervalYearMonth value = ((LazyHiveIntervalYearMonth) primitiveObject).getWritableObject().getHiveIntervalYearMonth();
                    HiveIntervalYearMonth expected = ((HiveIntervalYearMonthWritable) expectedObject).getHiveIntervalYearMonth();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveIntervalYearMonth field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case INTERVAL_DAY_TIME:
                {
                    if (!(primitiveObject instanceof LazyHiveIntervalDayTime)) {
                        throw new RuntimeException("Expected writable LazyHiveIntervalDayTime");
                    }
                    HiveIntervalDayTime value = ((LazyHiveIntervalDayTime) primitiveObject).getWritableObject().getHiveIntervalDayTime();
                    HiveIntervalDayTime expected = ((HiveIntervalDayTimeWritable) expectedObject).getHiveIntervalDayTime();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveIntervalDayTime field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case BINARY:
                {
                    if (!(primitiveObject instanceof LazyBinary)) {
                        throw new RuntimeException("Expected LazyBinary");
                    }
                    BytesWritable bytesWritable = ((LazyBinary) primitiveObject).getWritableObject();
                    byte[] value = Arrays.copyOfRange(bytesWritable.getBytes(), 0, bytesWritable.getLength());
                    BytesWritable bytesWritableExpected = (BytesWritable) expectedObject;
                    byte[] expected = Arrays.copyOfRange(bytesWritableExpected.getBytes(), 0, bytesWritableExpected.getLength());
                    if (value.length != expected.length) {
                        throw new RuntimeException("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(value) + ")");
                    }
                    for (int b = 0; b < value.length; b++) {
                        if (value[b] != expected[b]) {
                            throw new RuntimeException("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(value) + ")");
                        }
                    }
                }
                break;
            default:
                throw new Error("Unknown primitive category " + primitiveTypeInfo.getPrimitiveCategory());
        }
    } else if (lazyObject instanceof LazyArray) {
        LazyArray lazyArray = (LazyArray) lazyObject;
        List<Object> list = lazyArray.getList();
        List<Object> expectedList = (List<Object>) expectedObject;
        ListTypeInfo listTypeInfo = (ListTypeInfo) typeInfo;
        if (list.size() != expectedList.size()) {
            throw new RuntimeException("SerDe deserialized list length does not match (list " + list.toString() + " list.size() " + list.size() + " expectedList " + expectedList.toString() + " expectedList.size() " + expectedList.size() + ")" + " elementTypeInfo " + listTypeInfo.getListElementTypeInfo().toString());
        }
        return lazyCompareList((ListTypeInfo) typeInfo, list, expectedList);
    } else if (typeInfo instanceof ListTypeInfo) {
        List<Object> list;
        if (lazyObject instanceof LazyBinaryArray) {
            list = ((LazyBinaryArray) lazyObject).getList();
        } else {
            list = (List<Object>) lazyObject;
        }
        List<Object> expectedList = (List<Object>) expectedObject;
        if (list.size() != expectedList.size()) {
            throw new RuntimeException("SerDe deserialized list length does not match (list " + list.toString() + " list.size() " + list.size() + " expectedList " + expectedList.toString() + " expectedList.size() " + expectedList.size() + ")");
        }
        return lazyCompareList((ListTypeInfo) typeInfo, list, expectedList);
    } else if (lazyObject instanceof LazyMap) {
        LazyMap lazyMap = (LazyMap) lazyObject;
        Map<Object, Object> map = lazyMap.getMap();
        Map<Object, Object> expectedMap = (Map<Object, Object>) expectedObject;
        return lazyCompareMap((MapTypeInfo) typeInfo, map, expectedMap);
    } else if (typeInfo instanceof MapTypeInfo) {
        Map<Object, Object> map;
        Map<Object, Object> expectedMap = (Map<Object, Object>) expectedObject;
        if (lazyObject instanceof LazyBinaryMap) {
            map = ((LazyBinaryMap) lazyObject).getMap();
        } else {
            map = (Map<Object, Object>) lazyObject;
        }
        return lazyCompareMap((MapTypeInfo) typeInfo, map, expectedMap);
    } else if (lazyObject instanceof LazyStruct) {
        LazyStruct lazyStruct = (LazyStruct) lazyObject;
        List<Object> fields = lazyStruct.getFieldsAsList();
        List<Object> expectedFields = (List<Object>) expectedObject;
        StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
        return lazyCompareStruct(structTypeInfo, fields, expectedFields);
    } else if (typeInfo instanceof StructTypeInfo) {
        ArrayList<Object> fields;
        if (lazyObject instanceof LazyBinaryStruct) {
            fields = ((LazyBinaryStruct) lazyObject).getFieldsAsList();
        } else {
            fields = (ArrayList<Object>) lazyObject;
        }
        List<Object> expectedFields = (List<Object>) expectedObject;
        StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
        return lazyCompareStruct(structTypeInfo, fields, expectedFields);
    } else if (lazyObject instanceof LazyUnion) {
        LazyUnion union = (LazyUnion) lazyObject;
        StandardUnionObjectInspector.StandardUnion expectedUnion = (StandardUnionObjectInspector.StandardUnion) expectedObject;
        UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
        return lazyCompareUnion(unionTypeInfo, union, expectedUnion);
    } else if (typeInfo instanceof UnionTypeInfo) {
        StandardUnionObjectInspector.StandardUnion expectedUnion = (StandardUnionObjectInspector.StandardUnion) expectedObject;
        UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
        if (lazyObject instanceof LazyBinaryUnion) {
            return lazyCompareUnion(unionTypeInfo, (LazyBinaryUnion) lazyObject, expectedUnion);
        } else {
            return lazyCompareUnion(unionTypeInfo, (UnionObject) lazyObject, expectedUnion);
        }
    } else {
        System.err.println("Not implemented " + typeInfo.getClass().getName());
    }
    return true;
}

Example 28 with UnionTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo in project hive by apache.

the class LazyFactory method createLazyObjectInspector.

/**
 * Create a hierarchical ObjectInspector for LazyObject with the given typeInfo.
 *
 * @param typeInfo The type information for the LazyObject
 * @param separatorIndex The current level (for separators). List(array), struct uses 1 level of
 *          separator, and map uses 2 levels: the first one for delimiting entries, the second one
 *          for delimiting key and values.
 * @param lazyParams Params for lazy types
 * @param option the {@link ObjectInspectorOptions}
 * @return The ObjectInspector
 * @throws SerDeException
 */
public static ObjectInspector createLazyObjectInspector(TypeInfo typeInfo, int separatorIndex, LazyObjectInspectorParameters lazyParams, ObjectInspectorOptions option) throws SerDeException {
    ObjectInspector.Category c = typeInfo.getCategory();
    switch(c) {
        case PRIMITIVE:
            return LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector((PrimitiveTypeInfo) typeInfo, lazyParams);
        case MAP:
            return LazyObjectInspectorFactory.getLazySimpleMapObjectInspector(createLazyObjectInspector(((MapTypeInfo) typeInfo).getMapKeyTypeInfo(), separatorIndex + 2, lazyParams, option), createLazyObjectInspector(((MapTypeInfo) typeInfo).getMapValueTypeInfo(), separatorIndex + 2, lazyParams, option), LazyUtils.getSeparator(lazyParams.getSeparators(), separatorIndex), LazyUtils.getSeparator(lazyParams.getSeparators(), separatorIndex + 1), lazyParams);
        case LIST:
            return LazyObjectInspectorFactory.getLazySimpleListObjectInspector(createLazyObjectInspector(((ListTypeInfo) typeInfo).getListElementTypeInfo(), separatorIndex + 1, lazyParams, option), LazyUtils.getSeparator(lazyParams.getSeparators(), separatorIndex), lazyParams);
        case STRUCT:
            StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
            List<String> fieldNames = structTypeInfo.getAllStructFieldNames();
            List<TypeInfo> fieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
            List<ObjectInspector> fieldObjectInspectors = new ArrayList<ObjectInspector>(fieldTypeInfos.size());
            for (int i = 0; i < fieldTypeInfos.size(); i++) {
                fieldObjectInspectors.add(createLazyObjectInspector(fieldTypeInfos.get(i), separatorIndex + 1, lazyParams, option));
            }
            return LazyObjectInspectorFactory.getLazySimpleStructObjectInspector(fieldNames, fieldObjectInspectors, null, LazyUtils.getSeparator(lazyParams.getSeparators(), separatorIndex), lazyParams, option);
        case UNION:
            UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
            List<ObjectInspector> lazyOIs = new ArrayList<ObjectInspector>();
            for (TypeInfo uti : unionTypeInfo.getAllUnionObjectTypeInfos()) {
                lazyOIs.add(createLazyObjectInspector(uti, separatorIndex + 1, lazyParams, option));
            }
            return LazyObjectInspectorFactory.getLazyUnionObjectInspector(lazyOIs, LazyUtils.getSeparator(lazyParams.getSeparators(), separatorIndex), lazyParams);
    }
    throw new RuntimeException("Hive LazySerDe Internal error.");
}

Example 29 with UnionTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo in project hive by apache.

the class TestTypeInfoToSchema method createAvroUnionSchema.

@Test
public void createAvroUnionSchema() {
    UnionTypeInfo unionTypeInfo = new UnionTypeInfo();
    unionTypeInfo.setAllUnionObjectTypeInfos(Arrays.asList(INT, FLOAT, STRING));
    final String specificSchema = Schema.createUnion(Arrays.asList(Schema.create(Schema.Type.NULL), Schema.create(Schema.Type.INT), Schema.create(Schema.Type.FLOAT), Schema.create(Schema.Type.STRING))).toString();
    String expectedSchema = genSchemaWithoutNull(specificSchema);
    Assert.assertEquals("Test for union's avro schema failed", expectedSchema, getAvroSchemaString(unionTypeInfo));
}

Example 30 with UnionTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo in project hive by apache.

the class TestTypeInfoToSchema method createAvroUnionSchemaOfOne.

@Test
public void createAvroUnionSchemaOfOne() {
    UnionTypeInfo unionTypeInfo = new UnionTypeInfo();
    unionTypeInfo.setAllUnionObjectTypeInfos(Arrays.asList(STRING));
    final String specificSchema = Schema.createUnion(Arrays.asList(Schema.create(Schema.Type.NULL), Schema.create(Schema.Type.STRING))).toString();
    String expectedSchema = genSchemaWithoutNull(specificSchema);
    Assert.assertEquals("Test for union's avro schema failed", expectedSchema, getAvroSchemaString(unionTypeInfo));
}