AshinGau commented on code in PR #16024:
URL: https://github.com/apache/doris/pull/16024#discussion_r1080779915
##########
be/src/vec/exec/format/parquet/vparquet_group_reader.cpp:
##########
@@ -262,8 +258,9 @@ Status RowGroupReader::_do_lazy_read(Block* block, size_t
batch_size, size_t* re
// generated from next batch, so the filter column is removed
ahead.
DCHECK_EQ(block->rows(), 0);
} else {
- Block::filter_block(block, _lazy_read_ctx.all_predicate_col_ids,
filter_column_id,
- origin_column_num);
+ ColumnPtr filter_column =
block->get_by_position(filter_column_id).column;
Review Comment:
We should filter the predicate columns by `_filter_ptr` when there're delete
rows.
##########
be/src/vec/exec/format/parquet/vparquet_group_reader.cpp:
##########
@@ -434,6 +438,121 @@ Status RowGroupReader::_read_empty_batch(size_t
batch_size, size_t* read_rows, b
return Status::OK();
}
+Status RowGroupReader::_build_pos_delete_filter(size_t read_rows) {
+ _filter_ptr.reset(new IColumn::Filter(read_rows, 1));
+ _total_read_rows += read_rows;
+ if (!_position_delete_ctx.has_filter) {
+ return Status::OK();
+ }
+ size_t remaining_read_rows = _total_read_rows + read_rows;
+ while (_position_delete_ctx.index < _position_delete_ctx.end_index) {
+ const int64_t delete_row_index_in_row_group =
+ _position_delete_ctx.delete_rows[_position_delete_ctx.index] -
+ _position_delete_ctx.first_row_id;
+ int64_t read_range_rows = 0;
+ for (auto& range : _read_ranges) {
+ if (delete_row_index_in_row_group < range.first_row) {
+ break;
+ } else if (delete_row_index_in_row_group < range.last_row) {
Review Comment:
Maybe we can build an inline function to calculate the line number
##########
be/src/vec/exec/format/parquet/vparquet_group_reader.cpp:
##########
@@ -434,6 +438,121 @@ Status RowGroupReader::_read_empty_batch(size_t
batch_size, size_t* read_rows, b
return Status::OK();
}
+Status RowGroupReader::_build_pos_delete_filter(size_t read_rows) {
+ _filter_ptr.reset(new IColumn::Filter(read_rows, 1));
+ _total_read_rows += read_rows;
+ if (!_position_delete_ctx.has_filter) {
+ return Status::OK();
+ }
+ size_t remaining_read_rows = _total_read_rows + read_rows;
+ while (_position_delete_ctx.index < _position_delete_ctx.end_index) {
+ const int64_t delete_row_index_in_row_group =
+ _position_delete_ctx.delete_rows[_position_delete_ctx.index] -
+ _position_delete_ctx.first_row_id;
+ int64_t read_range_rows = 0;
+ for (auto& range : _read_ranges) {
Review Comment:
The two-level loop is time-consuming.
--
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: commits-unsubscr...@doris.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
