org.apache.hadoop.hive.ql.exec.vector.TimestampColumnVector Java Examples

private ColumnVector getPrimitiveColumnVector(PrimitiveObjectInspector poi) {
    switch (poi.getPrimitiveCategory()) {
    case BOOLEAN:
    case BYTE:
    case SHORT:
    case INT:
    case LONG:
    case DATE:
      return new LongColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case TIMESTAMP:
      return new TimestampColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case FLOAT:
    case DOUBLE:
      return new DoubleColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case BINARY:
    case STRING:
    case CHAR:
    case VARCHAR:
      return new BytesColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case DECIMAL:
      DecimalTypeInfo tInfo = (DecimalTypeInfo) poi.getTypeInfo();
      return new DecimalColumnVector(VectorizedRowBatch.DEFAULT_SIZE,
        tInfo.precision(), tInfo.scale()
      );
    default:
      throw UserException.unsupportedError()
        .message("Vectorized ORC reader is not supported for datatype: %s", poi.getPrimitiveCategory())
        .build(logger);
    }
}
 

private ColumnVector getPrimitiveColumnVector(PrimitiveObjectInspector poi) {
    switch (poi.getPrimitiveCategory()) {
    case BOOLEAN:
    case BYTE:
    case SHORT:
    case INT:
    case LONG:
    case DATE:
      return new LongColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case TIMESTAMP:
      return new TimestampColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case FLOAT:
    case DOUBLE:
      return new DoubleColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case BINARY:
    case STRING:
    case CHAR:
    case VARCHAR:
      return new BytesColumnVector(VectorizedRowBatch.DEFAULT_SIZE);
    case DECIMAL:
      DecimalTypeInfo tInfo = (DecimalTypeInfo) poi.getTypeInfo();
      return new DecimalColumnVector(VectorizedRowBatch.DEFAULT_SIZE,
        tInfo.precision(), tInfo.scale()
      );
    default:
      throw UserException.unsupportedError()
        .message("Vectorized ORC reader is not supported for datatype: %s", poi.getPrimitiveCategory())
        .build(logger);
    }
}
 

private static TimestampColumnVector createTimestampVector(int batchSize, Object value) {
	TimestampColumnVector lcv = new TimestampColumnVector(batchSize);
	if (value == null) {
		lcv.noNulls = false;
		lcv.isNull[0] = true;
		lcv.isRepeating = true;
	} else {
		Timestamp timestamp = value instanceof LocalDateTime ?
			Timestamp.valueOf((LocalDateTime) value) : (Timestamp) value;
		lcv.fill(timestamp);
		lcv.isNull[0] = false;
	}
	return lcv;
}
 

TimeStampMilliCopier(TimestampColumnVector inputVector, TimeStampMilliVector outputVector) {
  this.inputVector = inputVector;
  this.outputVector = outputVector;
}
 

TimestampToVarWidthCopier(TimestampColumnVector inputVector, BaseVariableWidthVector outputVector) {
  this.inputVector = inputVector;
  this.outputVector = outputVector;
}
 

TimeStampMilliCopier(TimestampColumnVector inputVector, TimeStampMilliVector outputVector) {
  this.inputVector = inputVector;
  this.outputVector = outputVector;
}
 

TimestampToVarWidthCopier(TimestampColumnVector inputVector, BaseVariableWidthVector outputVector) {
  this.inputVector = inputVector;
  this.outputVector = outputVector;
}
 

private void addCommonProperties(EntityContainer container) {
    LongColumnVector id = (LongColumnVector) batch.cols[0];
    BytesColumnVector type = (BytesColumnVector) batch.cols[1];
    MapColumnVector tags = (MapColumnVector) batch.cols[2];
    ListColumnVector nds = (ListColumnVector) batch.cols[5];
    ListColumnVector members = (ListColumnVector) batch.cols[6];
    LongColumnVector changeset = (LongColumnVector) batch.cols[7];
    TimestampColumnVector timestamp = (TimestampColumnVector) batch.cols[8];
    LongColumnVector uid = (LongColumnVector) batch.cols[9];
    BytesColumnVector user = (BytesColumnVector) batch.cols[10];
    LongColumnVector version = (LongColumnVector) batch.cols[11];
    LongColumnVector visible = (LongColumnVector) batch.cols[12];

    Entity entity = container.getEntity();

    id.vector[row] = entity.getId();
    changeset.vector[row] = entity.getChangesetId();
    type.setVal(row, entity.getType().toString().toLowerCase().getBytes());

    tags.offsets[row] = tags.childCount;
    tags.lengths[row] = entity.getTags().size(); // number of key/value pairings
    tags.childCount += tags.lengths[row];
    tags.keys.ensureSize(tags.childCount, tags.offsets[row] != 0);
    tags.values.ensureSize(tags.childCount, tags.offsets[row] != 0);

    int i = 0;
    for (Tag tag : entity.getTags()) {
        ((BytesColumnVector) tags.keys).setVal((int) tags.offsets[row] + i, tag.getKey().getBytes());
        ((BytesColumnVector) tags.values).setVal((int) tags.offsets[row] + i, tag.getValue().getBytes());

        i++;
    }

    timestamp.time[row] = entity.getTimestamp().getTime();
    timestamp.nanos[row] = 0;

    uid.vector[row] = entity.getUser().getId();

    user.setVal(row, entity.getUser().getName().getBytes());

    version.vector[row] = entity.getVersion();

    visible.vector[row] = 1;
    if (entity.getMetaTags().get("visible") == Boolean.FALSE) {
        visible.vector[row] = 0;
    }

    nds.offsets[row] = nds.childCount;
    nds.lengths[row] = 0;

    members.offsets[row] = members.childCount;
    members.lengths[row] = 0;
}
 

public OrcTimestampColumnVector(TimestampColumnVector vector) {
	super(vector);
	this.vector = vector;
}
 

protected void prepareReadFileWithTypes(String file, int rowSize) throws IOException {
	// NOTE: orc has field name information, so name should be same as orc
	TypeDescription schema =
			TypeDescription.fromString(
					"struct<" +
							"f0:float," +
							"f1:double," +
							"f2:timestamp," +
							"f3:tinyint," +
							"f4:smallint" +
							">");

	org.apache.hadoop.fs.Path filePath = new org.apache.hadoop.fs.Path(file);
	Configuration conf = new Configuration();

	Writer writer =
			OrcFile.createWriter(filePath,
					OrcFile.writerOptions(conf).setSchema(schema));

	VectorizedRowBatch batch = schema.createRowBatch(rowSize);
	DoubleColumnVector col0 = (DoubleColumnVector) batch.cols[0];
	DoubleColumnVector col1 = (DoubleColumnVector) batch.cols[1];
	TimestampColumnVector col2 = (TimestampColumnVector) batch.cols[2];
	LongColumnVector col3 = (LongColumnVector) batch.cols[3];
	LongColumnVector col4 = (LongColumnVector) batch.cols[4];

	col0.noNulls = false;
	col1.noNulls = false;
	col2.noNulls = false;
	col3.noNulls = false;
	col4.noNulls = false;
	for (int i = 0; i < rowSize - 1; i++) {
		col0.vector[i] = i;
		col1.vector[i] = i;

		Timestamp timestamp = toTimestamp(i);
		col2.time[i] = timestamp.getTime();
		col2.nanos[i] = timestamp.getNanos();

		col3.vector[i] = i;
		col4.vector[i] = i;
	}

	col0.isNull[rowSize - 1] = true;
	col1.isNull[rowSize - 1] = true;
	col2.isNull[rowSize - 1] = true;
	col3.isNull[rowSize - 1] = true;
	col4.isNull[rowSize - 1] = true;

	batch.size = rowSize;
	writer.addRowBatch(batch);
	batch.reset();
	writer.close();
}
 

static void setValue(JSONWriter writer, ColumnVector vector,
        TypeDescription schema, int row) throws JSONException {
    if (vector.isRepeating) {
        row = 0;
    }
    if (vector.noNulls || !vector.isNull[row]) {
        switch (schema.getCategory()) {
        case BOOLEAN:
            writer.value(((LongColumnVector) vector).vector[row] != 0);
            break;
        case BYTE:
        case SHORT:
        case INT:
        case LONG:
            writer.value(((LongColumnVector) vector).vector[row]);
            break;
        case FLOAT:
        case DOUBLE:
            writer.value(((DoubleColumnVector) vector).vector[row]);
            break;
        case STRING:
        case CHAR:
        case VARCHAR:
            writer.value(((BytesColumnVector) vector).toString(row));
            break;
        case DECIMAL:
            writer.value(((DecimalColumnVector) vector).vector[row]
                    .toString());
            break;
        case DATE:
            writer.value(new DateWritable(
                    (int) ((LongColumnVector) vector).vector[row])
                    .toString());
            break;
        case TIMESTAMP:
            writer.value(((TimestampColumnVector) vector)
                    .asScratchTimestamp(row).toString());
            break;
        case LIST:
            setList(writer, (ListColumnVector) vector, schema, row);
            break;
        case STRUCT:
            setStruct(writer, (StructColumnVector) vector, schema, row);
            break;
        case UNION:
            setUnion(writer, (UnionColumnVector) vector, schema, row);
            break;
        case BINARY:
            // To prevent similar mistakes like the one described in https://github.com/pinterest/secor/pull/1018,
            // it would be better to explicitly throw an exception here rather than ignore the incoming values,
            // which causes silent failures in a later stage.
            throw new UnsupportedOperationException();
        case MAP:
            setMap(writer, (MapColumnVector) vector, schema, row);
            break;
        default:
            throw new IllegalArgumentException("Unknown type "
                    + schema.toString());
        }
    } else {
        writer.value(null);
    }
}
 

protected static Object convertFromSourceToTargetDataType( ColumnVector columnVector, int currentBatchRow,
                                                           int orcValueMetaInterface ) {

  if ( columnVector.isNull[ currentBatchRow ] ) {
    return null;
  }
  switch ( orcValueMetaInterface ) {
    case ValueMetaInterface.TYPE_INET:
      try {
        return InetAddress.getByName( new String( ( (BytesColumnVector) columnVector ).vector[ currentBatchRow ],
          ( (BytesColumnVector) columnVector ).start[ currentBatchRow ],
          ( (BytesColumnVector) columnVector ).length[ currentBatchRow ] ) );
      } catch ( UnknownHostException e ) {
        e.printStackTrace();
      }

    case ValueMetaInterface.TYPE_STRING:
      return new String( ( (BytesColumnVector) columnVector ).vector[ currentBatchRow ],
        ( (BytesColumnVector) columnVector ).start[ currentBatchRow ],
        ( (BytesColumnVector) columnVector ).length[ currentBatchRow ] );

    case ValueMetaInterface.TYPE_INTEGER:
      return (long) ( (LongColumnVector) columnVector ).vector[ currentBatchRow ];

    case ValueMetaInterface.TYPE_NUMBER:
      return ( (DoubleColumnVector) columnVector ).vector[ currentBatchRow ];

    case ValueMetaInterface.TYPE_BIGNUMBER:
      HiveDecimalWritable obj = ( (DecimalColumnVector) columnVector ).vector[ currentBatchRow ];
      return obj.getHiveDecimal().bigDecimalValue();

    case ValueMetaInterface.TYPE_TIMESTAMP:
      Timestamp timestamp = new Timestamp( ( (TimestampColumnVector) columnVector ).time[ currentBatchRow ] );
      timestamp.setNanos( ( (TimestampColumnVector) columnVector ).nanos[ currentBatchRow ] );
      return timestamp;

    case ValueMetaInterface.TYPE_DATE:
      LocalDate localDate =
        LocalDate.ofEpochDay( 0 ).plusDays( ( (LongColumnVector) columnVector ).vector[ currentBatchRow ] );
      Date dateValue = Date.from( localDate.atStartOfDay( ZoneId.systemDefault() ).toInstant() );
      return dateValue;

    case ValueMetaInterface.TYPE_BOOLEAN:
      return ( (LongColumnVector) columnVector ).vector[ currentBatchRow ] == 0 ? false : true;

    case ValueMetaInterface.TYPE_BINARY:
      byte[] origBytes = ( (BytesColumnVector) columnVector ).vector[ currentBatchRow ];
      int startPos = ( (BytesColumnVector) columnVector ).start[ currentBatchRow ];
      byte[] newBytes = Arrays.copyOfRange( origBytes, startPos,
        startPos + ( (BytesColumnVector) columnVector ).length[ currentBatchRow ] );
      return newBytes;
  }

  //if none of the cases match return a null
  return null;
}