database/doris
2
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

[bugfix](topn) fix topn optimzation wrong result for NULL values (#18121)

Browse Source 
1. add PassNullPredicate to fix topn wrong result for NULL values
2. refactor RuntimePredicate to avoid using TCondition
3. refactor using ordering_exprs in fe and vsort_node
This commit is contained in:
Kang
2023-03-31 10:01:34 +08:00
committed by GitHub
parent 8be43857ef
commit 4e1e0ce06d
369 changed files with 3193 additions and 161 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
be/src/runtime/runtime_predicate.cpp
View File

@ -17,19 +17,22 @@
#include "runtime/runtime_predicate.h"
#include "olap/accept_null_predicate.h"
#include "olap/predicate_creator.h"
namespace doris {
namespace vectorized {
Status RuntimePredicate::init(const PrimitiveType type) {
Status RuntimePredicate::init(const PrimitiveType type, const bool nulls_first) {
std::unique_lock<std::shared_mutex> wlock(_rwlock);
if (_inited) {
return Status::OK();
}
_nulls_first = nulls_first;
_predicate_arena.reset(new Arena());
// set get value function
@ -111,6 +114,11 @@ Status RuntimePredicate::init(const PrimitiveType type) {
}
Status RuntimePredicate::update(const Field& value, const String& col_name, bool is_reverse) {
// skip null value
if (value.is_null()) {
return Status::OK();
}
std::unique_lock<std::shared_mutex> wlock(_rwlock);
// TODO why null
@ -139,18 +147,31 @@ Status RuntimePredicate::update(const Field& value, const String& col_name, bool
return Status::OK();
}
TCondition condition;
condition.__set_column_name(col_name);
condition.__set_column_unique_id(_tablet_schema->column(col_name).unique_id());
condition.__set_condition_op(is_reverse ? ">=" : "<=");
// get value string from _orderby_extrem and push back to condition_values
condition.condition_values.push_back(_get_value_fn(_orderby_extrem));
VLOG_DEBUG << "update runtime predicate condition " << condition;
// update _predictate
_predictate.reset(parse_to_predicate(_tablet_schema, condition, _predicate_arena.get(), false));
int32_t col_unique_id = _tablet_schema->column(col_name).unique_id();
const TabletColumn& column = _tablet_schema->column_by_uid(col_unique_id);
uint32_t index = _tablet_schema->field_index(col_unique_id);
auto val = _get_value_fn(_orderby_extrem);
ColumnPredicate* pred = nullptr;
if (is_reverse) {
// For DESC sort, create runtime predicate col_name >= min_top_value
// since values that < min_top_value are less than any value in current topn values
pred = create_comparison_predicate<PredicateType::GE>(column, index, val, false,
_predicate_arena.get());
} else {
// For ASC sort, create runtime predicate col_name <= max_top_value
// since values that > min_top_value are large than any value in current topn values
pred = create_comparison_predicate<PredicateType::LE>(column, index, val, false,
_predicate_arena.get());
}
// For NULLS FIRST, wrap a AcceptNullPredicate to return true for NULL
// since ORDER BY ASC/DESC should get NULL first but pred returns NULL
// and NULL in where predicate will be treated as FALSE
if (_nulls_first) {
pred = new AcceptNullPredicate(pred);
}
_predictate.reset(pred);
return Status::OK();
}