Java Code Examples for org.apache.parquet.schema.Type#asPrimitiveType()

/**
 * Changes the list inner '$data$' vector name to 'element' in the schema
 */
private Type renameChildTypeToElement(Type childType) {
  if (childType.isPrimitive()) {
    PrimitiveType childPrimitiveType = childType.asPrimitiveType();
    return new PrimitiveType(childType.getRepetition(),
      childPrimitiveType.getPrimitiveTypeName(),
      childPrimitiveType.getTypeLength(),
      "element",
      childPrimitiveType.getOriginalType(),
      childPrimitiveType.getDecimalMetadata(),
      childPrimitiveType.getId());
  } else {
    GroupType childGroupType = childType.asGroupType();
    Type.ID id = childGroupType.getId();
    GroupType groupType = new GroupType(childType.getRepetition(),
      "element",
      childType.getOriginalType(),
      childGroupType.getFields());
    if (id != null) {
      groupType = groupType.withId(id.hashCode());
    }
    return groupType;
  }
}
 

/**
 * Creates an empty {@code Statistics} instance for the specified type to be
 * used for reading/writing the new min/max statistics used in the V2 format.
 *
 * @param type
 *          type of the column
 * @return instance of a typed statistics class
 */
public static Statistics<?> createStats(Type type) {
  PrimitiveType primitive = type.asPrimitiveType();
  switch (primitive.getPrimitiveTypeName()) {
    case INT32:
      return new IntStatistics(primitive);
    case INT64:
      return new LongStatistics(primitive);
    case FLOAT:
      return new FloatStatistics(primitive);
    case DOUBLE:
      return new DoubleStatistics(primitive);
    case BOOLEAN:
      return new BooleanStatistics(primitive);
    case BINARY:
    case INT96:
    case FIXED_LEN_BYTE_ARRAY:
      return new BinaryStatistics(primitive);
    default:
      throw new UnknownColumnTypeException(primitive.getPrimitiveTypeName());
  }
}
 

public static PrimitiveType primitive(MessageType schema, String[] path) {
  Type current = schema;
  for (String part : path) {
    current = current.asGroupType().getType(part);
    if (current.isPrimitive()) {
      return current.asPrimitiveType();
    }
  }
  return null;
}
 

@Override
final public void start() {
  currentTuple = TF.newTuple(schemaSize);
  if (elephantBirdCompatible) {
    try {
      int i = 0;
      for (Type field : parquetSchema.getFields()) {
        if (field.isPrimitive() && field.isRepetition(Repetition.OPTIONAL)) {
          PrimitiveType primitiveType = field.asPrimitiveType();
          switch (primitiveType.getPrimitiveTypeName()) {
          case INT32:
            currentTuple.set(i, I32_ZERO);
            break;
          case INT64:
            currentTuple.set(i, I64_ZERO);
            break;
          case FLOAT:
            currentTuple.set(i, FLOAT_ZERO);
            break;
          case DOUBLE:
            currentTuple.set(i, DOUBLE_ZERO);
            break;
          case BOOLEAN:
            currentTuple.set(i, I32_ZERO);
            break;
          }
        }
        ++ i;
      }
    } catch (ExecException e) {
      throw new RuntimeException(e);
    }
  }
}
 

private static Converter newConverter(Schema schema, Type type,
    GenericData model, Class<?> knownClass, ParentValueContainer setter) {
  LogicalType logicalType = schema.getLogicalType();
  Conversion<?> conversion;

  if (knownClass != null) {
    conversion = model.getConversionByClass(knownClass, logicalType);
  } else {
    conversion = model.getConversionFor(logicalType);
  }

  ParentValueContainer parent = ParentValueContainer
      .getConversionContainer(setter, conversion, schema);

  switch (schema.getType()) {
  case BOOLEAN:
    return new AvroConverters.FieldBooleanConverter(parent);
  case INT:
    Class<?> intDatumClass = getDatumClass(conversion, knownClass, schema, model);
    if (intDatumClass == null) {
      return new AvroConverters.FieldIntegerConverter(parent);
    }
    if (intDatumClass == byte.class || intDatumClass == Byte.class) {
      return new AvroConverters.FieldByteConverter(parent);
    }
    if (intDatumClass == char.class || intDatumClass == Character.class) {
      return new AvroConverters.FieldCharConverter(parent);
    }
    if (intDatumClass == short.class || intDatumClass == Short.class) {
      return new AvroConverters.FieldShortConverter(parent);
    }
    return new AvroConverters.FieldIntegerConverter(parent);
  case LONG:
    return new AvroConverters.FieldLongConverter(parent);
  case FLOAT:
    return new AvroConverters.FieldFloatConverter(parent);
  case DOUBLE:
    return new AvroConverters.FieldDoubleConverter(parent);
  case BYTES:
    Class<?> byteDatumClass = getDatumClass(conversion, knownClass, schema, model);
    if (byteDatumClass == null) {
      return new AvroConverters.FieldByteBufferConverter(parent);
    }
    if (byteDatumClass.isArray() && byteDatumClass.getComponentType() == byte.class) {
      return new AvroConverters.FieldByteArrayConverter(parent);
    }
    return new AvroConverters.FieldByteBufferConverter(parent);
  case STRING:
    if (logicalType != null && logicalType.getName().equals(LogicalTypes.uuid().getName())) {
      return new AvroConverters.FieldUUIDConverter(parent, type.asPrimitiveType());
    }
    return newStringConverter(schema, model, parent);
  case RECORD:
    return new AvroRecordConverter(parent, type.asGroupType(), schema, model);
  case ENUM:
    return new AvroConverters.FieldEnumConverter(parent, schema, model);
  case ARRAY:
    Class<?> arrayDatumClass = getDatumClass(conversion, knownClass, schema, model);
    if (arrayDatumClass != null && arrayDatumClass.isArray()) {
      return new AvroArrayConverter(parent, type.asGroupType(), schema, model,
          arrayDatumClass);
    }
    return new AvroCollectionConverter(parent, type.asGroupType(), schema,
        model, arrayDatumClass);
  case MAP:
    return new MapConverter(parent, type.asGroupType(), schema, model);
  case UNION:
    return new AvroUnionConverter(parent, type, schema, model);
  case FIXED:
    return new AvroConverters.FieldFixedConverter(parent, schema, model);
  default:
    throw new UnsupportedOperationException(String.format(
        "Cannot convert Avro type: %s to Parquet type: %s", schema, type));
  }
}