sdf-jkl commented on code in PR #9118:
URL: https://github.com/apache/arrow-rs/pull/9118#discussion_r2722793094
##########
parquet/src/arrow/push_decoder/reader_builder/mod.rs:
##########
@@ -588,6 +582,23 @@ impl RowGroupReaderBuilder {
&mut self.buffers,
)?;
+ // before plan is build below
+ // check if plan is bitmask and if it is, put it in a variable
+ let page_offsets = if plan_builder.resolve_selection_strategy()
+ == RowSelectionStrategy::Mask
+ && plan_builder.selection().is_some_and(|selection| {
+ selection.requires_page_aware_mask(
+ &self.projection,
+ self.row_group_offset_index(row_group_idx),
+ )
+ }) {
+ self.row_group_offset_index(row_group_idx)
+ .and_then(|columns| columns.first())
Review Comment:
Another issue with the current implementation is that
`ParquetRecordBatchReader` is working page aware using pages offsets from a
single column.
However, read happens for all columns at once, using the same boolean
mask(which is col chunk specific).
It seems that supporting different page offsets for each col would require
us to push page awareness further down into the arrow readers.
--
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]
