org.apache.spark.sql.vectorized.ColumnVector Java Examples

The following examples show how to use org.apache.spark.sql.vectorized.ColumnVector. You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example #1
Source File: ArrowBinaryIterator.java    From spark-bigquery-connector with Apache License 2.0 6 votes vote down vote up
private Iterator<InternalRow> toArrowRows(VectorSchemaRoot root, List<String> namesInOrder) {
    ColumnVector[] columns = namesInOrder.stream()
            .map(name -> root.getVector(name))
            .map(vector -> new ArrowSchemaConverter(vector))
            .collect(Collectors.toList()).toArray(new ColumnVector[0]);

    ColumnarBatch batch = new ColumnarBatch(columns);
    batch.setNumRows(root.getRowCount());
    return batch.rowIterator();
}
 
Example #2
Source File: ColumnarBatchReader.java    From iceberg with Apache License 2.0 6 votes vote down vote up
@Override
public final ColumnarBatch read(ColumnarBatch reuse, int numRowsToRead) {
  Preconditions.checkArgument(numRowsToRead > 0, "Invalid number of rows to read: %s", numRowsToRead);
  ColumnVector[] arrowColumnVectors = new ColumnVector[readers.length];

  if (reuse == null) {
    closeVectors();
  }

  for (int i = 0; i < readers.length; i += 1) {
    vectorHolders[i] = readers[i].read(vectorHolders[i], numRowsToRead);
    int numRowsInVector = vectorHolders[i].numValues();
    Preconditions.checkState(
        numRowsInVector == numRowsToRead,
        "Number of rows in the vector %s didn't match expected %s ", numRowsInVector,
        numRowsToRead);
    arrowColumnVectors[i] =
        IcebergArrowColumnVector.forHolder(vectorHolders[i], numRowsInVector);
  }
  ColumnarBatch batch = new ColumnarBatch(arrowColumnVectors);
  batch.setNumRows(numRowsToRead);
  return batch;
}
 
Example #3
Source File: TestHelpers.java    From iceberg with Apache License 2.0 6 votes vote down vote up
public static void assertEqualsBatch(Types.StructType struct, Iterator<Record> expected, ColumnarBatch batch,
                                     boolean checkArrowValidityVector) {
  for (int rowId = 0; rowId < batch.numRows(); rowId++) {
    List<Types.NestedField> fields = struct.fields();
    InternalRow row = batch.getRow(rowId);
    Record rec = expected.next();
    for (int i = 0; i < fields.size(); i += 1) {
      Type fieldType = fields.get(i).type();
      Object expectedValue = rec.get(i);
      Object actualValue = row.isNullAt(i) ? null : row.get(i, convert(fieldType));
      assertEqualsUnsafe(fieldType, expectedValue, actualValue);

      if (checkArrowValidityVector) {
        ColumnVector columnVector = batch.column(i);
        ValueVector arrowVector = ((IcebergArrowColumnVector) columnVector).vectorAccessor().getVector();
        Assert.assertEquals("Nullability doesn't match", expectedValue == null, arrowVector.isNull(rowId));
      }
    }
  }
}
 
Example #4
Source File: HiveWarehouseDataReader.java    From spark-llap with Apache License 2.0 6 votes vote down vote up
@Override public ColumnarBatch get() {
  //Spark asks you to convert one column at a time so that different
  //column types can be handled differently.
  //NumOfCols << NumOfRows so this is negligible
  List<FieldVector> fieldVectors = wrapperWritable.getVectorSchemaRoot().getFieldVectors();
  if(columnVectors == null) {
    //Lazy create ColumnarBatch/ColumnVector[] instance
    columnVectors = new ColumnVector[fieldVectors.size()];
    columnarBatch = new ColumnarBatch(columnVectors);
  }
  Iterator<FieldVector> iterator = fieldVectors.iterator();
  int rowCount = -1;
  for (int i = 0; i < columnVectors.length; i++) {
    FieldVector fieldVector = iterator.next();
    columnVectors[i] = new ArrowColumnVector(fieldVector);
    if (rowCount == -1) {
      //All column vectors have same length so we can get rowCount from any column
      rowCount = fieldVector.getValueCount();
    }
  }
  columnarBatch.setNumRows(rowCount);
  return columnarBatch;
}
 
Example #5
Source File: FlightDataReader.java    From flight-spark-source with Apache License 2.0 5 votes vote down vote up
@Override
public ColumnarBatch get() {
  start();
  ColumnarBatch batch = new ColumnarBatch(
    stream.getRoot().getFieldVectors()
      .stream()
      .map(FlightArrowColumnVector::new)
      .toArray(ColumnVector[]::new)
  );
  batch.setNumRows(stream.getRoot().getRowCount());
  return batch;
}
 
Example #6
Source File: CountDataReader.java    From spark-llap with Apache License 2.0 5 votes vote down vote up
@Override public ColumnarBatch get() {
  int size = (numRows >= 1000) ? 1000 : (int) numRows;
  OnHeapColumnVector vector = new OnHeapColumnVector(size, DataTypes.LongType);
  for(int i = 0; i < size; i++) {
    vector.putLong(0, numRows);
  }
  numRows -= size;
  ColumnarBatch batch = new ColumnarBatch(new ColumnVector[] {vector});
  batch.setNumRows(size);
  return batch;
}
 
Example #7
Source File: IcebergArrowColumnVector.java    From iceberg with Apache License 2.0 4 votes vote down vote up
static ColumnVector forHolder(VectorHolder holder, int numRows) {
  return holder.isDummy() ? new NullValuesColumnVector(numRows) :
      new IcebergArrowColumnVector(holder);
}
 
Example #8
Source File: NullValuesColumnVector.java    From iceberg with Apache License 2.0 4 votes vote down vote up
@Override
protected ColumnVector getChild(int ordinal) {
  throw new UnsupportedOperationException();
}