haohuaijin opened a new issue, #22939: URL: https://github.com/apache/datafusion/issues/22939
### Is your feature request related to a problem or challenge? DataFusion supports `ParquetAccessPlan` as a `PartitionedFile` extension, but that API is row-group based. Callers must know the row group layout and split any file-level row selection themselves. DataFusion already loads the row group metadata when opening the file, so it can split that selection internally. ### Describe the solution you'd like Add a `ParquetRowSelection` extension that wraps a file-level `RowSelection`. When opening the file, DataFusion should convert it to the existing `ParquetAccessPlan` form: - all rows skipped: `RowGroupAccess::Skip` - all rows selected: `RowGroupAccess::Scan` - mixed selected/skipped rows: `RowGroupAccess::Selection` The selection should be rejected if its total row count does not match the Parquet metadata. ### Describe alternatives you've considered _No response_ ### Additional context _No response_ -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
