use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project metacat by Netflix.
the class HiveTypeConverter method getCanonicalType.
/**
* Returns the canonical type.
*
* @param fieldInspector inspector
* @return type
*/
Type getCanonicalType(final ObjectInspector fieldInspector) {
switch(fieldInspector.getCategory()) {
case PRIMITIVE:
return getPrimitiveType(fieldInspector);
case MAP:
final MapObjectInspector mapObjectInspector = TypeUtils.checkType(fieldInspector, MapObjectInspector.class, "fieldInspector");
final Type keyType = getCanonicalType(mapObjectInspector.getMapKeyObjectInspector());
final Type valueType = getCanonicalType(mapObjectInspector.getMapValueObjectInspector());
if (keyType == null || valueType == null) {
return null;
}
return TypeRegistry.getTypeRegistry().getParameterizedType(TypeEnum.MAP, ImmutableList.of(keyType.getTypeSignature(), valueType.getTypeSignature()), ImmutableList.of());
case LIST:
final ListObjectInspector listObjectInspector = TypeUtils.checkType(fieldInspector, ListObjectInspector.class, "fieldInspector");
final Type elementType = getCanonicalType(listObjectInspector.getListElementObjectInspector());
if (elementType == null) {
return null;
}
return TypeRegistry.getTypeRegistry().getParameterizedType(TypeEnum.ARRAY, ImmutableList.of(elementType.getTypeSignature()), ImmutableList.of());
case STRUCT:
final StructObjectInspector structObjectInspector = TypeUtils.checkType(fieldInspector, StructObjectInspector.class, "fieldInspector");
final List<TypeSignature> fieldTypes = new ArrayList<>();
final List<Object> fieldNames = new ArrayList<>();
for (StructField field : structObjectInspector.getAllStructFieldRefs()) {
fieldNames.add(field.getFieldName());
final Type fieldType = getCanonicalType(field.getFieldObjectInspector());
if (fieldType == null) {
return null;
}
fieldTypes.add(fieldType.getTypeSignature());
}
return TypeRegistry.getTypeRegistry().getParameterizedType(TypeEnum.ROW, fieldTypes, fieldNames);
default:
log.info("Currently unsupported type {}, returning Unknown type", fieldInspector.getTypeName());
return BaseType.UNKNOWN;
}
}
use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project presto-hive-apache by prestodb.
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:
{
throw new IOException("JsonSerDe does not support BINARY type");
}
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!");
}
}
use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project presto by prestodb.
the class TestDataWritableWriter method writeValue.
/**
* It writes the field value to the Parquet RecordConsumer. It detects the field type, and calls
* the correct write function.
*
* @param value The writable object that contains the value.
* @param inspector The object inspector used to get the correct value type.
* @param type Type that contains information about the type schema.
*/
private void writeValue(final Object value, final ObjectInspector inspector, final Type type) {
if (type.isPrimitive()) {
checkInspectorCategory(inspector, ObjectInspector.Category.PRIMITIVE);
writePrimitive(value, (PrimitiveObjectInspector) inspector);
} else {
GroupType groupType = type.asGroupType();
OriginalType originalType = type.getOriginalType();
if (originalType != null && originalType.equals(OriginalType.LIST)) {
checkInspectorCategory(inspector, ObjectInspector.Category.LIST);
if (singleLevelArray) {
writeSingleLevelArray(value, (ListObjectInspector) inspector, groupType);
} else {
writeArray(value, (ListObjectInspector) inspector, groupType);
}
} else if (originalType != null && (originalType.equals(OriginalType.MAP) || originalType.equals(OriginalType.MAP_KEY_VALUE))) {
checkInspectorCategory(inspector, ObjectInspector.Category.MAP);
writeMap(value, (MapObjectInspector) inspector, groupType);
} else {
checkInspectorCategory(inspector, ObjectInspector.Category.STRUCT);
writeGroup(value, (StructObjectInspector) inspector, groupType);
}
}
}
use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project presto by prestodb.
the class TestDataWritableWriter method writeMap.
/**
* It writes a map type and its key-pair values to the Parquet RecordConsumer.
* This is called when the original type (MAP) is detected by writeValue().
* This function assumes the following schema:
* optional group mapCol (MAP) {
* repeated group map (MAP_KEY_VALUE) {
* required TYPE key;
* optional TYPE value;
* }
* }
*
* @param value The object that contains the map key-values.
* @param inspector The object inspector used to get the correct value type.
* @param type Type that contains information about the group (MAP) schema.
*/
private void writeMap(final Object value, final MapObjectInspector inspector, final GroupType type) {
// Get the internal map structure (MAP_KEY_VALUE)
GroupType repeatedType = type.getType(0).asGroupType();
recordConsumer.startGroup();
Map<?, ?> mapValues = inspector.getMap(value);
if (mapValues != null && mapValues.size() > 0) {
recordConsumer.startField(repeatedType.getName(), 0);
Type keyType = repeatedType.getType(0);
String keyName = keyType.getName();
ObjectInspector keyInspector = inspector.getMapKeyObjectInspector();
Type valuetype = repeatedType.getType(1);
String valueName = valuetype.getName();
ObjectInspector valueInspector = inspector.getMapValueObjectInspector();
for (Map.Entry<?, ?> keyValue : mapValues.entrySet()) {
recordConsumer.startGroup();
if (keyValue != null) {
// write key element
Object keyElement = keyValue.getKey();
recordConsumer.startField(keyName, 0);
writeValue(keyElement, keyInspector, keyType);
recordConsumer.endField(keyName, 0);
// write value element
Object valueElement = keyValue.getValue();
if (valueElement != null) {
recordConsumer.startField(valueName, 1);
writeValue(valueElement, valueInspector, valuetype);
recordConsumer.endField(valueName, 1);
}
}
recordConsumer.endGroup();
}
recordConsumer.endField(repeatedType.getName(), 0);
}
recordConsumer.endGroup();
}
use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project flink by apache.
the class HiveInspectors method getConversion.
/**
* Get conversion for converting Flink object to Hive object from an ObjectInspector and the
* corresponding Flink DataType.
*/
public static HiveObjectConversion getConversion(ObjectInspector inspector, LogicalType dataType, HiveShim hiveShim) {
if (inspector instanceof PrimitiveObjectInspector) {
HiveObjectConversion conversion;
if (inspector instanceof BooleanObjectInspector || inspector instanceof StringObjectInspector || inspector instanceof ByteObjectInspector || inspector instanceof ShortObjectInspector || inspector instanceof IntObjectInspector || inspector instanceof LongObjectInspector || inspector instanceof FloatObjectInspector || inspector instanceof DoubleObjectInspector || inspector instanceof BinaryObjectInspector || inspector instanceof VoidObjectInspector) {
conversion = IdentityConversion.INSTANCE;
} else if (inspector instanceof DateObjectInspector) {
conversion = hiveShim::toHiveDate;
} else if (inspector instanceof TimestampObjectInspector) {
conversion = hiveShim::toHiveTimestamp;
} else if (inspector instanceof HiveCharObjectInspector) {
conversion = o -> o == null ? null : new HiveChar((String) o, ((CharType) dataType).getLength());
} else if (inspector instanceof HiveVarcharObjectInspector) {
conversion = o -> o == null ? null : new HiveVarchar((String) o, ((VarCharType) dataType).getLength());
} else if (inspector instanceof HiveDecimalObjectInspector) {
conversion = o -> o == null ? null : HiveDecimal.create((BigDecimal) o);
} else {
throw new FlinkHiveUDFException("Unsupported primitive object inspector " + inspector.getClass().getName());
}
// currently this happens for constant arguments for UDFs
if (((PrimitiveObjectInspector) inspector).preferWritable()) {
conversion = new WritableHiveObjectConversion(conversion, hiveShim);
}
return conversion;
}
if (inspector instanceof ListObjectInspector) {
HiveObjectConversion eleConvert = getConversion(((ListObjectInspector) inspector).getListElementObjectInspector(), ((ArrayType) dataType).getElementType(), hiveShim);
return o -> {
if (o == null) {
return null;
}
Object[] array = (Object[]) o;
List<Object> result = new ArrayList<>();
for (Object ele : array) {
result.add(eleConvert.toHiveObject(ele));
}
return result;
};
}
if (inspector instanceof MapObjectInspector) {
MapObjectInspector mapInspector = (MapObjectInspector) inspector;
MapType kvType = (MapType) dataType;
HiveObjectConversion keyConversion = getConversion(mapInspector.getMapKeyObjectInspector(), kvType.getKeyType(), hiveShim);
HiveObjectConversion valueConversion = getConversion(mapInspector.getMapValueObjectInspector(), kvType.getValueType(), hiveShim);
return o -> {
if (o == null) {
return null;
}
Map<Object, Object> map = (Map) o;
Map<Object, Object> result = new HashMap<>(map.size());
for (Map.Entry<Object, Object> entry : map.entrySet()) {
result.put(keyConversion.toHiveObject(entry.getKey()), valueConversion.toHiveObject(entry.getValue()));
}
return result;
};
}
if (inspector instanceof StructObjectInspector) {
StructObjectInspector structInspector = (StructObjectInspector) inspector;
List<? extends StructField> structFields = structInspector.getAllStructFieldRefs();
List<RowType.RowField> rowFields = ((RowType) dataType).getFields();
HiveObjectConversion[] conversions = new HiveObjectConversion[structFields.size()];
for (int i = 0; i < structFields.size(); i++) {
conversions[i] = getConversion(structFields.get(i).getFieldObjectInspector(), rowFields.get(i).getType(), hiveShim);
}
return o -> {
if (o == null) {
return null;
}
Row row = (Row) o;
List<Object> result = new ArrayList<>(row.getArity());
for (int i = 0; i < row.getArity(); i++) {
result.add(conversions[i].toHiveObject(row.getField(i)));
}
return result;
};
}
throw new FlinkHiveUDFException(String.format("Flink doesn't support convert object conversion for %s yet", inspector));
}
Aggregations