Examples with UnionObjectInspector org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector used on opensource projects

Example 1 with UnionObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector in project hive by apache.

the class JsonSerDe method buildJSONString.

// TODO : code section copied over from SerDeUtils because of non-standard json production there
// should use quotes for all field names. We should fix this there, and then remove this copy.
// See http://jackson.codehaus.org/1.7.3/javadoc/org/codehaus/jackson/JsonParser.Feature.html#ALLOW_UNQUOTED_FIELD_NAMES
// for details - trying to enable Jackson to ignore that doesn't seem to work(compilation failure
// when attempting to use that feature, so having to change the production itself.
// Also, throws IOException when Binary is detected.
private static void buildJSONString(StringBuilder sb, Object o, ObjectInspector oi) throws IOException {
    switch(oi.getCategory()) {
        case PRIMITIVE:
            {
                PrimitiveObjectInspector poi = (PrimitiveObjectInspector) oi;
                if (o == null) {
                    sb.append("null");
                } else {
                    switch(poi.getPrimitiveCategory()) {
                        case BOOLEAN:
                            {
                                boolean b = ((BooleanObjectInspector) poi).get(o);
                                sb.append(b ? "true" : "false");
                                break;
                            }
                        case BYTE:
                            {
                                sb.append(((ByteObjectInspector) poi).get(o));
                                break;
                            }
                        case SHORT:
                            {
                                sb.append(((ShortObjectInspector) poi).get(o));
                                break;
                            }
                        case INT:
                            {
                                sb.append(((IntObjectInspector) poi).get(o));
                                break;
                            }
                        case LONG:
                            {
                                sb.append(((LongObjectInspector) poi).get(o));
                                break;
                            }
                        case FLOAT:
                            {
                                sb.append(((FloatObjectInspector) poi).get(o));
                                break;
                            }
                        case DOUBLE:
                            {
                                sb.append(((DoubleObjectInspector) poi).get(o));
                                break;
                            }
                        case STRING:
                            {
                                String s = SerDeUtils.escapeString(((StringObjectInspector) poi).getPrimitiveJavaObject(o));
                                appendWithQuotes(sb, s);
                                break;
                            }
                        case BINARY:
                            byte[] b = ((BinaryObjectInspector) oi).getPrimitiveJavaObject(o);
                            Text txt = new Text();
                            txt.set(b, 0, b.length);
                            appendWithQuotes(sb, SerDeUtils.escapeString(txt.toString()));
                            break;
                        case DATE:
                            Date d = ((DateObjectInspector) poi).getPrimitiveJavaObject(o);
                            appendWithQuotes(sb, d.toString());
                            break;
                        case TIMESTAMP:
                            {
                                Timestamp t = ((TimestampObjectInspector) poi).getPrimitiveJavaObject(o);
                                appendWithQuotes(sb, t.toString());
                                break;
                            }
                        case DECIMAL:
                            sb.append(((HiveDecimalObjectInspector) poi).getPrimitiveJavaObject(o));
                            break;
                        case VARCHAR:
                            {
                                String s = SerDeUtils.escapeString(((HiveVarcharObjectInspector) poi).getPrimitiveJavaObject(o).toString());
                                appendWithQuotes(sb, s);
                                break;
                            }
                        case CHAR:
                            {
                                // this should use HiveChar.getPaddedValue() but it's protected; currently (v0.13)
                                // HiveChar.toString() returns getPaddedValue()
                                String s = SerDeUtils.escapeString(((HiveCharObjectInspector) poi).getPrimitiveJavaObject(o).toString());
                                appendWithQuotes(sb, s);
                                break;
                            }
                        default:
                            throw new RuntimeException("Unknown primitive type: " + poi.getPrimitiveCategory());
                    }
                }
                break;
            }
        case LIST:
            {
                ListObjectInspector loi = (ListObjectInspector) oi;
                ObjectInspector listElementObjectInspector = loi.getListElementObjectInspector();
                List<?> olist = loi.getList(o);
                if (olist == null) {
                    sb.append("null");
                } else {
                    sb.append(SerDeUtils.LBRACKET);
                    for (int i = 0; i < olist.size(); i++) {
                        if (i > 0) {
                            sb.append(SerDeUtils.COMMA);
                        }
                        buildJSONString(sb, olist.get(i), listElementObjectInspector);
                    }
                    sb.append(SerDeUtils.RBRACKET);
                }
                break;
            }
        case MAP:
            {
                MapObjectInspector moi = (MapObjectInspector) oi;
                ObjectInspector mapKeyObjectInspector = moi.getMapKeyObjectInspector();
                ObjectInspector mapValueObjectInspector = moi.getMapValueObjectInspector();
                Map<?, ?> omap = moi.getMap(o);
                if (omap == null) {
                    sb.append("null");
                } else {
                    sb.append(SerDeUtils.LBRACE);
                    boolean first = true;
                    for (Object entry : omap.entrySet()) {
                        if (first) {
                            first = false;
                        } else {
                            sb.append(SerDeUtils.COMMA);
                        }
                        Map.Entry<?, ?> e = (Map.Entry<?, ?>) entry;
                        StringBuilder keyBuilder = new StringBuilder();
                        buildJSONString(keyBuilder, e.getKey(), mapKeyObjectInspector);
                        String keyString = keyBuilder.toString().trim();
                        if ((!keyString.isEmpty()) && (keyString.charAt(0) != SerDeUtils.QUOTE)) {
                            appendWithQuotes(sb, keyString);
                        } else {
                            sb.append(keyString);
                        }
                        sb.append(SerDeUtils.COLON);
                        buildJSONString(sb, e.getValue(), mapValueObjectInspector);
                    }
                    sb.append(SerDeUtils.RBRACE);
                }
                break;
            }
        case STRUCT:
            {
                StructObjectInspector soi = (StructObjectInspector) oi;
                List<? extends StructField> structFields = soi.getAllStructFieldRefs();
                if (o == null) {
                    sb.append("null");
                } else {
                    sb.append(SerDeUtils.LBRACE);
                    for (int i = 0; i < structFields.size(); i++) {
                        if (i > 0) {
                            sb.append(SerDeUtils.COMMA);
                        }
                        appendWithQuotes(sb, structFields.get(i).getFieldName());
                        sb.append(SerDeUtils.COLON);
                        buildJSONString(sb, soi.getStructFieldData(o, structFields.get(i)), structFields.get(i).getFieldObjectInspector());
                    }
                    sb.append(SerDeUtils.RBRACE);
                }
                break;
            }
        case UNION:
            {
                UnionObjectInspector uoi = (UnionObjectInspector) oi;
                if (o == null) {
                    sb.append("null");
                } else {
                    sb.append(SerDeUtils.LBRACE);
                    sb.append(uoi.getTag(o));
                    sb.append(SerDeUtils.COLON);
                    buildJSONString(sb, uoi.getField(o), uoi.getObjectInspectors().get(uoi.getTag(o)));
                    sb.append(SerDeUtils.RBRACE);
                }
                break;
            }
        default:
            throw new RuntimeException("Unknown type in ObjectInspector!");
    }
}

Example 2 with UnionObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector in project hive by apache.

the class VectorRandomRowSource method getObjectInspector.

private ObjectInspector getObjectInspector(TypeInfo typeInfo) {
    final ObjectInspector objectInspector;
    switch(typeInfo.getCategory()) {
        case PRIMITIVE:
            {
                final PrimitiveTypeInfo primitiveType = (PrimitiveTypeInfo) typeInfo;
                objectInspector = PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(primitiveType);
            }
            break;
        case MAP:
            {
                final MapTypeInfo mapType = (MapTypeInfo) typeInfo;
                final MapObjectInspector mapInspector = ObjectInspectorFactory.getStandardMapObjectInspector(getObjectInspector(mapType.getMapKeyTypeInfo()), getObjectInspector(mapType.getMapValueTypeInfo()));
                objectInspector = mapInspector;
            }
            break;
        case LIST:
            {
                final ListTypeInfo listType = (ListTypeInfo) typeInfo;
                final ListObjectInspector listInspector = ObjectInspectorFactory.getStandardListObjectInspector(getObjectInspector(listType.getListElementTypeInfo()));
                objectInspector = listInspector;
            }
            break;
        case STRUCT:
            {
                final StructTypeInfo structType = (StructTypeInfo) typeInfo;
                final List<TypeInfo> fieldTypes = structType.getAllStructFieldTypeInfos();
                final List<ObjectInspector> fieldInspectors = new ArrayList<ObjectInspector>();
                for (TypeInfo fieldType : fieldTypes) {
                    fieldInspectors.add(getObjectInspector(fieldType));
                }
                final StructObjectInspector structInspector = ObjectInspectorFactory.getStandardStructObjectInspector(structType.getAllStructFieldNames(), fieldInspectors);
                objectInspector = structInspector;
            }
            break;
        case UNION:
            {
                final UnionTypeInfo unionType = (UnionTypeInfo) typeInfo;
                final List<TypeInfo> fieldTypes = unionType.getAllUnionObjectTypeInfos();
                final List<ObjectInspector> fieldInspectors = new ArrayList<ObjectInspector>();
                for (TypeInfo fieldType : fieldTypes) {
                    fieldInspectors.add(getObjectInspector(fieldType));
                }
                final UnionObjectInspector unionInspector = ObjectInspectorFactory.getStandardUnionObjectInspector(fieldInspectors);
                objectInspector = unionInspector;
            }
            break;
        default:
            throw new RuntimeException("Unexpected category " + typeInfo.getCategory());
    }
    Preconditions.checkState(objectInspector != null);
    return objectInspector;
}

Example 3 with UnionObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector in project hive by apache.

the class VectorExtractRow method extractRowColumn.

public Object extractRowColumn(ColumnVector colVector, TypeInfo typeInfo, ObjectInspector objectInspector, int batchIndex) {
    if (colVector == null) {
        // may ask for them..
        return null;
    }
    final int adjustedIndex = (colVector.isRepeating ? 0 : batchIndex);
    if (!colVector.noNulls && colVector.isNull[adjustedIndex]) {
        return null;
    }
    final Category category = typeInfo.getCategory();
    switch(category) {
        case PRIMITIVE:
            {
                final PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
                final PrimitiveCategory primitiveCategory = primitiveTypeInfo.getPrimitiveCategory();
                final Writable primitiveWritable = VectorizedBatchUtil.getPrimitiveWritable(primitiveCategory);
                switch(primitiveCategory) {
                    case VOID:
                        return null;
                    case BOOLEAN:
                        ((BooleanWritable) primitiveWritable).set(((LongColumnVector) colVector).vector[adjustedIndex] == 0 ? false : true);
                        return primitiveWritable;
                    case BYTE:
                        ((ByteWritable) primitiveWritable).set((byte) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case SHORT:
                        ((ShortWritable) primitiveWritable).set((short) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case INT:
                        ((IntWritable) primitiveWritable).set((int) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case LONG:
                        ((LongWritable) primitiveWritable).set(((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case TIMESTAMP:
                        // From java.sql.Timestamp used by vectorization to serializable org.apache.hadoop.hive.common.type.Timestamp
                        java.sql.Timestamp ts = ((TimestampColumnVector) colVector).asScratchTimestamp(adjustedIndex);
                        Timestamp serializableTS = Timestamp.ofEpochMilli(ts.getTime(), ts.getNanos());
                        ((TimestampWritableV2) primitiveWritable).set(serializableTS);
                        return primitiveWritable;
                    case DATE:
                        ((DateWritableV2) primitiveWritable).set((int) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case FLOAT:
                        ((FloatWritable) primitiveWritable).set((float) ((DoubleColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case DOUBLE:
                        ((DoubleWritable) primitiveWritable).set(((DoubleColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case BINARY:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            BytesWritable bytesWritable = (BytesWritable) primitiveWritable;
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                bytesWritable.set(EMPTY_BYTES, 0, 0);
                            } else {
                                bytesWritable.set(bytes, start, length);
                            }
                            return primitiveWritable;
                        }
                    case STRING:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                ((Text) primitiveWritable).set(EMPTY_BYTES, 0, 0);
                            } else {
                                // Use org.apache.hadoop.io.Text as our helper to go from byte[] to String.
                                ((Text) primitiveWritable).set(bytes, start, length);
                            }
                            return primitiveWritable;
                        }
                    case VARCHAR:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            final HiveVarcharWritable hiveVarcharWritable = (HiveVarcharWritable) primitiveWritable;
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                hiveVarcharWritable.set(EMPTY_STRING, -1);
                            } else {
                                final int adjustedLength = StringExpr.truncate(bytes, start, length, ((VarcharTypeInfo) primitiveTypeInfo).getLength());
                                if (adjustedLength == 0) {
                                    hiveVarcharWritable.set(EMPTY_STRING, -1);
                                } else {
                                    hiveVarcharWritable.set(new String(bytes, start, adjustedLength, Charsets.UTF_8), -1);
                                }
                            }
                            return primitiveWritable;
                        }
                    case CHAR:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            final HiveCharWritable hiveCharWritable = (HiveCharWritable) primitiveWritable;
                            final int maxLength = ((CharTypeInfo) primitiveTypeInfo).getLength();
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                hiveCharWritable.set(EMPTY_STRING, maxLength);
                            } else {
                                final int adjustedLength = StringExpr.rightTrimAndTruncate(bytes, start, length, ((CharTypeInfo) primitiveTypeInfo).getLength());
                                if (adjustedLength == 0) {
                                    hiveCharWritable.set(EMPTY_STRING, maxLength);
                                } else {
                                    hiveCharWritable.set(new String(bytes, start, adjustedLength, Charsets.UTF_8), maxLength);
                                }
                            }
                            return primitiveWritable;
                        }
                    case DECIMAL:
                        if (colVector instanceof Decimal64ColumnVector) {
                            Decimal64ColumnVector dec32ColVector = (Decimal64ColumnVector) colVector;
                            ((HiveDecimalWritable) primitiveWritable).deserialize64(dec32ColVector.vector[adjustedIndex], dec32ColVector.scale);
                        } else {
                            // The HiveDecimalWritable set method will quickly copy the deserialized decimal writable fields.
                            ((HiveDecimalWritable) primitiveWritable).set(((DecimalColumnVector) colVector).vector[adjustedIndex]);
                        }
                        return primitiveWritable;
                    case INTERVAL_YEAR_MONTH:
                        ((HiveIntervalYearMonthWritable) primitiveWritable).set((int) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case INTERVAL_DAY_TIME:
                        ((HiveIntervalDayTimeWritable) primitiveWritable).set(((IntervalDayTimeColumnVector) colVector).asScratchIntervalDayTime(adjustedIndex));
                        return primitiveWritable;
                    default:
                        throw new RuntimeException("Primitive category " + primitiveCategory.name() + " not supported");
                }
            }
        case LIST:
            {
                final ListColumnVector listColumnVector = (ListColumnVector) colVector;
                final ListTypeInfo listTypeInfo = (ListTypeInfo) typeInfo;
                final ListObjectInspector listObjectInspector = (ListObjectInspector) objectInspector;
                final int offset = (int) listColumnVector.offsets[adjustedIndex];
                final int size = (int) listColumnVector.lengths[adjustedIndex];
                final List list = new ArrayList();
                for (int i = 0; i < size; i++) {
                    list.add(extractRowColumn(listColumnVector.child, listTypeInfo.getListElementTypeInfo(), listObjectInspector.getListElementObjectInspector(), offset + i));
                }
                return list;
            }
        case MAP:
            {
                final MapColumnVector mapColumnVector = (MapColumnVector) colVector;
                final MapTypeInfo mapTypeInfo = (MapTypeInfo) typeInfo;
                final MapObjectInspector mapObjectInspector = (MapObjectInspector) objectInspector;
                final int offset = (int) mapColumnVector.offsets[adjustedIndex];
                final int size = (int) mapColumnVector.lengths[adjustedIndex];
                final Map<Object, Object> map = new LinkedHashMap<Object, Object>();
                for (int i = 0; i < size; i++) {
                    final Object key = extractRowColumn(mapColumnVector.keys, mapTypeInfo.getMapKeyTypeInfo(), mapObjectInspector.getMapKeyObjectInspector(), offset + i);
                    final Object value = extractRowColumn(mapColumnVector.values, mapTypeInfo.getMapValueTypeInfo(), mapObjectInspector.getMapValueObjectInspector(), offset + i);
                    map.put(key, value);
                }
                return map;
            }
        case STRUCT:
            {
                final StructColumnVector structColumnVector = (StructColumnVector) colVector;
                final StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
                final StandardStructObjectInspector structInspector = (StandardStructObjectInspector) objectInspector;
                final List<TypeInfo> fieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
                final int size = fieldTypeInfos.size();
                final List<? extends StructField> structFields = structInspector.getAllStructFieldRefs();
                final Object struct = structInspector.create();
                for (int i = 0; i < size; i++) {
                    final StructField structField = structFields.get(i);
                    final TypeInfo fieldTypeInfo = fieldTypeInfos.get(i);
                    final Object value = extractRowColumn(structColumnVector.fields[i], fieldTypeInfo, structField.getFieldObjectInspector(), adjustedIndex);
                    structInspector.setStructFieldData(struct, structField, value);
                }
                return struct;
            }
        case UNION:
            {
                final UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
                final List<TypeInfo> objectTypeInfos = unionTypeInfo.getAllUnionObjectTypeInfos();
                final UnionObjectInspector unionInspector = (UnionObjectInspector) objectInspector;
                final List<ObjectInspector> unionInspectors = unionInspector.getObjectInspectors();
                final UnionColumnVector unionColumnVector = (UnionColumnVector) colVector;
                final byte tag = (byte) unionColumnVector.tags[adjustedIndex];
                final Object object = extractRowColumn(unionColumnVector.fields[tag], objectTypeInfos.get(tag), unionInspectors.get(tag), adjustedIndex);
                final StandardUnion standardUnion = new StandardUnion();
                standardUnion.setTag(tag);
                standardUnion.setObject(object);
                return standardUnion;
            }
        default:
            throw new RuntimeException("Category " + category.name() + " not supported");
    }
}

Example 4 with UnionObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector 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.toSqlTimestamp());
                        }
                        break;
                    case DATE:
                        {
                            final Date date = PrimitiveObjectInspectorUtils.getDate(object, sourcePrimitiveOI);
                            if (date == null) {
                                VectorizedBatchUtil.setNullColIsNullValue(columnVector, batchIndex);
                                return;
                            }
                            DateWritableV2 dateWritable = (DateWritableV2) convertTargetWritable;
                            if (dateWritable == null) {
                                dateWritable = new DateWritableV2();
                            }
                            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;
}

Example 5 with UnionObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector in project hive by apache.

the class VectorRandomRowSource method getObjectInspector.

public static ObjectInspector getObjectInspector(TypeInfo typeInfo, DataTypePhysicalVariation dataTypePhysicalVariation) {
    final ObjectInspector objectInspector;
    switch(typeInfo.getCategory()) {
        case PRIMITIVE:
            {
                final PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
                if (primitiveTypeInfo instanceof DecimalTypeInfo && dataTypePhysicalVariation == DataTypePhysicalVariation.DECIMAL_64) {
                    objectInspector = PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(TypeInfoFactory.longTypeInfo);
                } else {
                    objectInspector = PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(primitiveTypeInfo);
                }
            }
            break;
        case MAP:
            {
                final MapTypeInfo mapType = (MapTypeInfo) typeInfo;
                final MapObjectInspector mapInspector = ObjectInspectorFactory.getStandardMapObjectInspector(getObjectInspector(mapType.getMapKeyTypeInfo()), getObjectInspector(mapType.getMapValueTypeInfo()));
                objectInspector = mapInspector;
            }
            break;
        case LIST:
            {
                final ListTypeInfo listType = (ListTypeInfo) typeInfo;
                final ListObjectInspector listInspector = ObjectInspectorFactory.getStandardListObjectInspector(getObjectInspector(listType.getListElementTypeInfo()));
                objectInspector = listInspector;
            }
            break;
        case STRUCT:
            {
                final StructTypeInfo structType = (StructTypeInfo) typeInfo;
                final List<TypeInfo> fieldTypes = structType.getAllStructFieldTypeInfos();
                final List<ObjectInspector> fieldInspectors = new ArrayList<ObjectInspector>();
                for (TypeInfo fieldType : fieldTypes) {
                    fieldInspectors.add(getObjectInspector(fieldType));
                }
                final StructObjectInspector structInspector = ObjectInspectorFactory.getStandardStructObjectInspector(structType.getAllStructFieldNames(), fieldInspectors);
                objectInspector = structInspector;
            }
            break;
        case UNION:
            {
                final UnionTypeInfo unionType = (UnionTypeInfo) typeInfo;
                final List<TypeInfo> fieldTypes = unionType.getAllUnionObjectTypeInfos();
                final List<ObjectInspector> fieldInspectors = new ArrayList<ObjectInspector>();
                for (TypeInfo fieldType : fieldTypes) {
                    fieldInspectors.add(getObjectInspector(fieldType));
                }
                final UnionObjectInspector unionInspector = ObjectInspectorFactory.getStandardUnionObjectInspector(fieldInspectors);
                objectInspector = unionInspector;
            }
            break;
        default:
            throw new RuntimeException("Unexpected category " + typeInfo.getCategory());
    }
    Preconditions.checkState(objectInspector != null);
    return objectInspector;
}