Package org.apache.iceberg

Interface DeletedDataFileScanTask

All Superinterfaces:
ChangelogScanTask, ContentScanTask<DataFile>, PartitionScanTask, ScanTask, Serializable
public interface DeletedDataFileScanTask extends ChangelogScanTask, ContentScanTask<DataFile>
A scan task for deletes generated by removing a data file from the table.

Note that all historical delete files added earlier must be applied while reading the data file. This is required to output only those data records that were live when the data file was removed.

Suppose snapshot S1 contains data files F1, F2, F3. Then snapshot S2 adds a position delete file, D1, that deletes records from F2 and snapshot S3 removes F2 entirely. A scan for changes generated by S3 should include the following task:

  • DeletedDataFileScanTask(file=F2, existing-deletes=[D1], snapshot=S3)

Readers consuming these tasks should produce deleted records with metadata like change ordinal and commit snapshot ID.

  • Method Details

    • existingDeletes

      List<DeleteFile> existingDeletes()
      A list of previously added delete files to apply when reading the data file in this task.
      Returns:
      a list of delete files to apply

    • operation

      default ChangelogOperation operation()
      Description copied from interface: ChangelogScanTask
      Returns the type of changes produced by this task (i.e. insert/delete).
      Specified by:
      operation in interface ChangelogScanTask

    • sizeBytes

      default long sizeBytes()
      Description copied from interface: ScanTask
      The number of bytes that should be read by this scan task.
      Specified by:
      sizeBytes in interface ContentScanTask<DataFile>
      Specified by:
      sizeBytes in interface ScanTask
      Returns:
      the total number of bytes to read

    • filesCount

      default int filesCount()
      Description copied from interface: ScanTask
      The number of files that will be opened by this scan task.
      Specified by:
      filesCount in interface ScanTask
      Returns:
      the number of files to open