HappenLee commented on code in PR #17594:
URL: https://github.com/apache/doris/pull/17594#discussion_r1133428175
##########
be/src/vec/exec/format/parquet/vparquet_group_reader.cpp:
##########
@@ -71,11 +97,136 @@ Status RowGroupReader::init(const FieldDescriptor& schema,
std::vector<RowRange>
}
_column_readers[read_col._file_slot_name] = std::move(reader);
}
+ // Check if single slot can be filtered by dict.
+ if (!_slot_id_to_filter_conjuncts) {
+ return Status::OK();
+ }
+ for (auto& predicate_col_name : _lazy_read_ctx.predicate_columns) {
+ auto field =
const_cast<FieldSchema*>(schema.get_column(predicate_col_name));
+ if (_can_filter_by_dict(predicate_col_name,
+
_row_group_meta.columns[field->physical_column_index].meta_data)) {
+ _dict_filter_col_names.emplace_back(predicate_col_name);
+ } else {
+ int slot_id = _col_name_to_slot_id->at(predicate_col_name);
+ if (_slot_id_to_filter_conjuncts->find(slot_id) !=
+ _slot_id_to_filter_conjuncts->end()) {
+ for (VExprContext* ctx :
_slot_id_to_filter_conjuncts->at(slot_id)) {
+ _filter_conjuncts.push_back(ctx);
+ }
+ }
+ }
+ }
+ RETURN_IF_ERROR(_rewrite_dict_predicates());
return Status::OK();
}
+bool RowGroupReader::_can_filter_by_dict(const string& predicate_col_name,
+ const tparquet::ColumnMetaData&
column_metadata) {
+ SlotDescriptor* slot = nullptr;
+ const std::vector<SlotDescriptor*>& slots = _tuple_descriptor->slots();
+ int slot_id = _col_name_to_slot_id->at(predicate_col_name);
+ for (auto each : slots) {
+ if (each->id() == slot_id) {
+ slot = each;
+ break;
+ }
+ }
+ if (!slot->type().is_string_type()) {
+ return false;
+ }
+
+ if (_slot_id_to_filter_conjuncts->find(slot_id) ==
_slot_id_to_filter_conjuncts->end()) {
+ return false;
+ }
+
+ if (!is_dictionary_encoded(column_metadata)) {
+ return false;
+ }
+
+ for (VExprContext* ctx : _slot_id_to_filter_conjuncts->at(slot_id)) {
+ const VExpr* root_expr = ctx->root();
+ if (root_expr->node_type() == TExprNodeType::FUNCTION_CALL) {
Review Comment:
if the conjunt is `a > 10 is null ` ? the logic seems do the consider the
case
--
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]
