Package org.apache.flink.connector.file.table

Class ColumnarRowIterator

java.lang.Object

org.apache.flink.connector.file.src.util.RecyclableIterator<org.apache.flink.table.data.RowData>

org.apache.flink.connector.file.table.ColumnarRowIterator

All Implemented Interfaces:

BulkFormat.RecordIterator<org.apache.flink.table.data.RowData>

@Internal
public class ColumnarRowIterator
extends RecyclableIterator<org.apache.flink.table.data.RowData>

A BulkFormat.RecordIterator that returns RowDatas. The next row is set by ColumnarRowData.setRowId(int).

Constructor Summary

Constructors

Constructor	Description
ColumnarRowIterator(org.apache.flink.table.data.columnar.ColumnarRowData rowData, Runnable recycler)

Method Summary

Modifier and Type	Method	Description
RecordAndPosition<org.apache.flink.table.data.RowData>	next()	Gets the next record from the file, together with its position.
void	set(int num, long recordSkipCount)
void	set(int num, long offset, long recordSkipCount)	Set number rows in this batch and updates the position.

Methods inherited from class org.apache.flink.connector.file.src.util.RecyclableIterator

releaseBatch

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

ColumnarRowIterator

public ColumnarRowIterator(org.apache.flink.table.data.columnar.ColumnarRowData rowData,
 @Nullable
 Runnable recycler)

Method Details

set

public void set(int num,
 long recordSkipCount)

Parameters:

num - number rows in this batch.

recordSkipCount - The number of rows that have been returned before this batch.

set

public void set(int num,
 long offset,
 long recordSkipCount)

Set number rows in this batch and updates the position.

next

@Nullable
public RecordAndPosition<org.apache.flink.table.data.RowData> next()

Description copied from interface: BulkFormat.RecordIterator

Gets the next record from the file, together with its position.

The position information returned with the record point to the record AFTER the returned record, because it defines the point where the reading should resume once the current record is emitted. The position information is put in the source's state when the record is emitted. If a checkpoint is taken directly after the record is emitted, the checkpoint must to describe where to resume the source reading from after that record.

Objects returned by this method may be reused by the iterator. By the time that this method is called again, no object returned from the previous call will be referenced any more. That makes it possible to have a single MutableRecordAndPosition object and return the same instance (with updated record and position) on every call.