github-actions[bot] commented on code in PR #63389:
URL: https://github.com/apache/doris/pull/63389#discussion_r3272205384
##########
be/src/format/parquet/vparquet_reader.cpp:
##########
@@ -1443,6 +1577,76 @@ Status ParquetReader::_process_column_stat_filter(
return Status::OK();
}
+Status ParquetReader::_process_expr_zonemap_filter(const tparquet::RowGroup&
row_group,
+ bool* filter_group) {
+ DORIS_CHECK(filter_group != nullptr);
+ if (!config::enable_expr_zonemap_filter || _zonemap_filter_conjuncts ==
nullptr ||
+ _zonemap_filter_conjuncts->empty() || !_enable_filter_by_min_max) {
+ return Status::OK();
+ }
+
+ std::set<int> column_ids;
+ for (const auto& conjunct : *_zonemap_filter_conjuncts) {
+ if (conjunct->root() != nullptr) {
+ conjunct->root()->collect_slot_column_ids(column_ids);
+ }
+ }
+ if (column_ids.empty()) {
+ return Status::OK();
+ }
+
+ ZoneMapEvalContext ctx;
+ for (const int cid : column_ids) {
+ if (cid < 0 || cid >= _tuple_descriptor->slots().size()) {
+ continue;
+ }
+ auto* slot = _tuple_descriptor->slots()[cid];
+ ZoneMapEvalContext::SlotZoneMap slot_zone_map;
+ slot_zone_map.data_type = slot->type();
+ if (!_table_info_node_ptr->children_column_exists(slot->col_name())) {
+ ctx.slots.emplace(cid, std::move(slot_zone_map));
+ continue;
+ }
+ const auto& file_col_name =
+
_table_info_node_ptr->children_file_column_name(slot->col_name());
+ const FieldSchema* col_schema =
_file_metadata->schema().get_column(file_col_name);
+ int parquet_col_id = col_schema->physical_column_index;
+ const auto& meta_data = row_group.columns[parquet_col_id].meta_data;
+
Review Comment:
This row-group expr-zonemap path needs the same complex-column guard used by
the page-index path above. For an expression such as `complex_col IS NULL`,
`collect_slot_column_ids()` can include a complex slot,
`children_column_exists()` can be true, but
`FieldSchema::physical_column_index` is `-1` for complex parent fields. This
then indexes `row_group.columns[-1]` and can crash instead of conservatively
skipping zonemap pruning. Please check `parquet_col_id < 0` (and fall back by
emplacing the slot without a zonemap or continuing) before accessing
`row_group.columns[parquet_col_id]`.
--
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]
