private void processArrayField(CommonRecord record, GenericArray array, String fieldName) {
List<CommonValue> currentArray;
CommonValue arrayValue = record.getField(fieldName);
if (arrayValue != null && arrayValue.isArray()) {
currentArray = arrayValue.getArray().getList();
} else {
currentArray = new LinkedList<CommonValue>();
record.setField(
fieldName,
commonFactory.createCommonValue(
commonFactory.createCommonArray(array.getSchema(), currentArray)));
}
if (!array.isEmpty()) {
Object rawItem = array.get(0);
if (AvroGenericUtils.isRecord(rawItem)) {
GenericArray<GenericRecord> recordItems = (GenericArray<GenericRecord>) array;
// Adding new records
for (GenericRecord item : recordItems) {
CommonRecord newRecord = createCommonRecord(item);
updateRecord(newRecord, item);
currentArray.add(commonFactory.createCommonValue(newRecord));
}
} else if (AvroGenericUtils.isFixed(rawItem)) {
GenericArray<GenericFixed> fixedItems = (GenericArray<GenericFixed>) array;
if (AvroGenericUtils.isUuid(rawItem)) {
// Removing items with given uuids
for (GenericFixed item : fixedItems) {
UUID currentUuid = AvroGenericUtils.createUuidFromFixed(item);
Iterator<CommonValue> valueIt = currentArray.iterator();
while (valueIt.hasNext()) {
CommonRecord currentRecord = valueIt.next().getRecord();
if (currentRecord.getUuid().equals(currentUuid)) {
valueIt.remove();
records.remove(currentUuid);
break;
}
}
}
} else {
for (GenericFixed item : fixedItems) {
currentArray.add(
commonFactory.createCommonValue(
commonFactory.createCommonFixed(item.getSchema(), item.bytes())));
}
}
} else {
// Adding new primitive items
for (Object item : array) {
currentArray.add(commonFactory.createCommonValue(item));
}
}
}
}
/** Writes the given Avro datum into the given record, using the given Avro schema */
private void extractTree(Object datum, Schema schema, Record outputRecord, String prefix) {
// RECORD, ENUM, ARRAY, MAP, UNION, FIXED, STRING, BYTES, INT, LONG, FLOAT,
// DOUBLE, BOOLEAN, NULL
switch (schema.getType()) {
case RECORD:
{
IndexedRecord avroRecord = (IndexedRecord) datum;
String prefix2 = prefix + "/";
for (Field field : schema.getFields()) {
extractTree(
avroRecord.get(field.pos()),
field.schema(),
outputRecord,
prefix2 + field.name());
}
break;
}
case ENUM:
{
GenericEnumSymbol symbol = (GenericEnumSymbol) datum;
outputRecord.put(prefix, symbol.toString());
break;
}
case ARRAY:
{
Iterator iter = ((Collection) datum).iterator();
while (iter.hasNext()) {
extractTree(iter.next(), schema.getElementType(), outputRecord, prefix);
}
break;
}
case MAP:
{
Map<CharSequence, ?> map = (Map<CharSequence, ?>) datum;
for (Map.Entry<CharSequence, ?> entry : map.entrySet()) {
extractTree(
entry.getValue(),
schema.getValueType(),
outputRecord,
prefix + "/" + entry.getKey().toString());
}
break;
}
case UNION:
{
int index = GenericData.get().resolveUnion(schema, datum);
// String typeName = schema.getTypes().get(index).getName();
// String prefix2 = prefix + "/" + typeName;
String prefix2 = prefix;
extractTree(datum, schema.getTypes().get(index), outputRecord, prefix2);
break;
}
case FIXED:
{
GenericFixed fixed = (GenericFixed) datum;
outputRecord.put(prefix, fixed.bytes());
// outputRecord.put(prefix, utf8toString(fixed.bytes()));
break;
}
case BYTES:
{
ByteBuffer buf = (ByteBuffer) datum;
int pos = buf.position();
byte[] bytes = new byte[buf.remaining()];
buf.get(bytes);
buf.position(pos); // undo relative read
outputRecord.put(prefix, bytes);
// outputRecord.put(prefix, utf8toString(bytes));
break;
}
case STRING:
{
outputRecord.put(prefix, datum.toString());
break;
}
case INT:
{
outputRecord.put(prefix, datum);
break;
}
case LONG:
{
outputRecord.put(prefix, datum);
break;
}
case FLOAT:
{
outputRecord.put(prefix, datum);
break;
}
case DOUBLE:
{
outputRecord.put(prefix, datum);
break;
}
case BOOLEAN:
{
outputRecord.put(prefix, datum);
break;
}
case NULL:
{
break;
}
default:
throw new MorphlineRuntimeException("Unknown Avro schema type: " + schema.getType());
}
}
private Object translate(Object value, DataSchema dataSchema, Schema avroSchema) {
AvroOverride avroOverride = getAvroOverride(dataSchema);
if (avroOverride != null) {
return avroOverride
.getCustomDataTranslator()
.avroGenericToData(this, value, avroSchema, dataSchema);
}
DataSchema dereferencedDataSchema = dataSchema.getDereferencedDataSchema();
DataSchema.Type type = dereferencedDataSchema.getType();
Object result;
switch (type) {
case NULL:
if (value != null) {
appendMessage("value must be null for null schema");
result = BAD_RESULT;
break;
}
result = Data.NULL;
break;
case BOOLEAN:
result = ((Boolean) value).booleanValue();
break;
case INT:
result = ((Number) value).intValue();
break;
case LONG:
result = ((Number) value).longValue();
break;
case FLOAT:
result = ((Number) value).floatValue();
break;
case DOUBLE:
result = ((Number) value).doubleValue();
break;
case STRING:
result = value.toString();
break;
case BYTES:
ByteBuffer byteBuffer = (ByteBuffer) value;
ByteString byteString = ByteString.copy(byteBuffer);
byteBuffer.rewind();
result = byteString;
break;
case ENUM:
String enumValue = value.toString();
EnumDataSchema enumDataSchema = (EnumDataSchema) dereferencedDataSchema;
if (enumDataSchema.getSymbols().contains(enumValue) == false) {
appendMessage(
"enum value %1$s not one of %2$s", enumValue, enumDataSchema.getSymbols());
result = BAD_RESULT;
break;
}
result = enumValue;
break;
case FIXED:
GenericFixed fixed = (GenericFixed) value;
byte[] fixedBytes = fixed.bytes();
FixedDataSchema fixedDataSchema = (FixedDataSchema) dereferencedDataSchema;
if (fixedDataSchema.getSize() != fixedBytes.length) {
appendMessage(
"GenericFixed size %1$d != FixedDataSchema size %2$d",
fixedBytes.length, fixedDataSchema.getSize());
result = BAD_RESULT;
break;
}
byteString = ByteString.copy(fixedBytes);
result = byteString;
break;
case MAP:
@SuppressWarnings("unchecked")
Map<?, Object> map = (Map<?, Object>) value;
DataSchema valueDataSchema = ((MapDataSchema) dereferencedDataSchema).getValues();
Schema valueAvroSchema = avroSchema.getValueType();
DataMap dataMap = new DataMap(map.size());
for (Map.Entry<?, Object> entry : map.entrySet()) {
String key = entry.getKey().toString();
_path.addLast(key);
Object entryValue = translate(entry.getValue(), valueDataSchema, valueAvroSchema);
_path.removeLast();
dataMap.put(key, entryValue);
}
result = dataMap;
break;
case ARRAY:
GenericArray<?> list = (GenericArray<?>) value;
DataSchema elementDataSchema = ((ArrayDataSchema) dereferencedDataSchema).getItems();
Schema elementAvroSchema = avroSchema.getElementType();
DataList dataList = new DataList(list.size());
for (int i = 0; i < list.size(); i++) {
_path.addLast(i);
Object entryValue = translate(list.get(i), elementDataSchema, elementAvroSchema);
_path.removeLast();
dataList.add(entryValue);
}
result = dataList;
break;
case RECORD:
GenericRecord record = (GenericRecord) value;
RecordDataSchema recordDataSchema = (RecordDataSchema) dereferencedDataSchema;
dataMap = new DataMap(avroSchema.getFields().size());
for (RecordDataSchema.Field field : recordDataSchema.getFields()) {
String fieldName = field.getName();
Object fieldValue = record.get(fieldName);
// fieldValue could be null if the Avro schema does not contain the named field or
// the field is present with a null value. In either case we do not add a value
// to the translated DataMap. We do not consider optional/required/default here
// either (i.e. it is not an error if a required field is missing); the user can
// later call ValidateDataAgainstSchema with various
// settings for RequiredMode to obtain the desired behaviour.
if (fieldValue == null) {
continue;
}
boolean isOptional = field.getOptional();
DataSchema fieldDataSchema = field.getType();
Schema fieldAvroSchema = avroSchema.getField(fieldName).schema();
if (isOptional && (fieldDataSchema.getDereferencedType() != DataSchema.Type.UNION)) {
// Avro schema should be union with 2 types: null and the field's type.
Map.Entry<String, Schema> fieldAvroEntry =
findUnionMember(fieldDataSchema, fieldAvroSchema);
if (fieldAvroEntry == null) {
continue;
}
fieldAvroSchema = fieldAvroEntry.getValue();
}
_path.addLast(fieldName);
dataMap.put(fieldName, translate(fieldValue, fieldDataSchema, fieldAvroSchema));
_path.removeLast();
}
result = dataMap;
break;
case UNION:
UnionDataSchema unionDataSchema = (UnionDataSchema) dereferencedDataSchema;
Map.Entry<DataSchema, Schema> memberSchemas =
findUnionMemberSchema(value, unionDataSchema, avroSchema);
if (memberSchemas == null) {
result = BAD_RESULT;
break;
}
if (value == null) {
// schema must be "null" schema
result = Data.NULL;
} else {
DataSchema memberDataSchema = memberSchemas.getKey();
Schema memberAvroSchema = memberSchemas.getValue();
String key = memberDataSchema.getUnionMemberKey();
dataMap = new DataMap(1);
_path.addLast(key);
dataMap.put(key, translate(value, memberDataSchema, memberAvroSchema));
_path.removeLast();
result = dataMap;
}
break;
default:
appendMessage("schema type unknown %1$s", dereferencedDataSchema.getType());
result = BAD_RESULT;
break;
}
return result;
}
private void processFixedField(CommonRecord record, GenericFixed fixed, String fieldName) {
record.setField(
fieldName,
commonFactory.createCommonValue(
commonFactory.createCommonFixed(fixed.getSchema(), fixed.bytes())));
}