This is an automated email from the ASF dual-hosted git repository.
panxiaolei pushed a commit to branch refactor_rf
in repository https://gitbox.apache.org/repos/asf/doris.git
commit 6e4173933d9927a5f97c8d81f596ac4a617d5a0e
Author: Gabriel <liwenqi...@selectdb.com>
AuthorDate: Tue Mar 4 19:05:10 2025 +0800
test (#48630)
---
be/src/runtime_filter/runtime_filter_wrapper.cpp | 5 +-
.../runtime_filter/runtime_filter_wrapper_test.cpp | 64 ++++++++++++++++++++++
be/test/runtime_filter/utils_test.cpp | 2 +-
3 files changed, 68 insertions(+), 3 deletions(-)
diff --git a/be/src/runtime_filter/runtime_filter_wrapper.cpp
b/be/src/runtime_filter/runtime_filter_wrapper.cpp
index df850312150..e248082667d 100644
--- a/be/src/runtime_filter/runtime_filter_wrapper.cpp
+++ b/be/src/runtime_filter/runtime_filter_wrapper.cpp
@@ -164,7 +164,7 @@ Status RuntimeFilterWrapper::merge(const
RuntimeFilterWrapper* other) {
DCHECK(_state != State::IGNORED);
DCHECK(other->_state == State::READY);
- DCHECK(_filter_type == other->_filter_type) << debug_string();
+ DCHECK(_filter_type == other->_filter_type);
switch (_filter_type) {
case RuntimeFilterType::IN_FILTER: {
@@ -580,7 +580,8 @@ std::string RuntimeFilterWrapper::debug_string() const {
filter_type_to_string(get_real_type()))
: filter_type_to_string(_filter_type);
auto result = fmt::format("[id: {}, state: {}, type: {}, column_type: {}",
_filter_id,
- to_string(_state), type_string,
type_to_string(_column_return_type));
+
states_to_string<RuntimeFilterWrapper>({_state}), type_string,
+ type_to_string(_column_return_type));
if (_state == State::READY) {
if (get_real_type() == RuntimeFilterType::BLOOM_FILTER) {
diff --git a/be/test/runtime_filter/runtime_filter_wrapper_test.cpp
b/be/test/runtime_filter/runtime_filter_wrapper_test.cpp
index 5ca0443bfb9..af3f154b13a 100644
--- a/be/test/runtime_filter/runtime_filter_wrapper_test.cpp
+++ b/be/test/runtime_filter/runtime_filter_wrapper_test.cpp
@@ -1063,6 +1063,70 @@ TEST_F(RuntimeFilterWrapperTest, TestInOrBloom) {
EXPECT_EQ(wrapper->hybrid_set()->size(), col->size() * 2);
EXPECT_EQ(wrapper->get_real_type(), RuntimeFilterType::IN_FILTER);
}
+ {
+ // In + In -> Bloom
+ max_in_num = 16;
+ runtime_size = 10;
+ RuntimeFilterParams params {
+ .filter_id = filter_id,
+ .filter_type = filter_type,
+ .column_return_type = column_return_type,
+ .null_aware = null_aware,
+ .max_in_num = max_in_num,
+ .runtime_bloom_filter_min_size = runtime_bloom_filter_min_size,
+ .runtime_bloom_filter_max_size = runtime_bloom_filter_max_size,
+ .bloom_filter_size = bloom_filter_size,
+ .build_bf_by_runtime_size = build_bf_by_runtime_size,
+ .bloom_filter_size_calculated_by_ndv =
bloom_filter_size_calculated_by_ndv,
+ .enable_fixed_len_to_uint32_v2 = enable_fixed_len_to_uint32_v2,
+ .bitmap_filter_not_in = bitmap_filter_not_in};
+ wrapper = std::make_shared<RuntimeFilterWrapper>(¶ms);
+ EXPECT_EQ(wrapper->get_real_type(), RuntimeFilterType::IN_FILTER);
+ EXPECT_TRUE(wrapper->init(runtime_size).ok());
+ EXPECT_EQ(wrapper->get_real_type(), RuntimeFilterType::IN_FILTER);
+ RuntimeFilterParams new_params {
+ .filter_id = filter_id,
+ .filter_type = filter_type,
+ .column_return_type = column_return_type,
+ .null_aware = null_aware,
+ .max_in_num = max_in_num,
+ .runtime_bloom_filter_min_size = runtime_bloom_filter_min_size,
+ .runtime_bloom_filter_max_size = runtime_bloom_filter_max_size,
+ .bloom_filter_size = bloom_filter_size,
+ .build_bf_by_runtime_size = build_bf_by_runtime_size,
+ .bloom_filter_size_calculated_by_ndv =
bloom_filter_size_calculated_by_ndv,
+ .enable_fixed_len_to_uint32_v2 = enable_fixed_len_to_uint32_v2,
+ .bitmap_filter_not_in = bitmap_filter_not_in};
+ auto new_wrapper = std::make_shared<RuntimeFilterWrapper>(&new_params);
+ EXPECT_EQ(new_wrapper->get_real_type(), RuntimeFilterType::IN_FILTER);
+ EXPECT_TRUE(new_wrapper->init(runtime_size).ok());
+ EXPECT_EQ(new_wrapper->get_real_type(), RuntimeFilterType::IN_FILTER);
+ // Insert
+ auto col =
vectorized::ColumnHelper::create_column<DataType>(data_vector);
+ EXPECT_TRUE(wrapper->insert(col, 0).ok());
+ EXPECT_EQ(wrapper->get_state(), RuntimeFilterWrapper::State::UNINITED);
+ EXPECT_EQ(wrapper->hybrid_set()->size(), col->size());
+
+ std::vector<int> new_data_vector(10);
+ std::iota(new_data_vector.begin(), new_data_vector.end(), 10);
+ col =
vectorized::ColumnHelper::create_column<DataType>(new_data_vector);
+ EXPECT_TRUE(new_wrapper->insert(col, 0).ok());
+ EXPECT_EQ(new_wrapper->get_state(),
RuntimeFilterWrapper::State::UNINITED);
+ EXPECT_EQ(new_wrapper->hybrid_set()->size(), col->size());
+ new_wrapper->_state = RuntimeFilterWrapper::State::READY;
+ // Merge
+ EXPECT_TRUE(wrapper->merge(new_wrapper.get()).ok());
+
+
+ std::vector<int> final_data_vector(20);
+ std::iota(final_data_vector.begin(), final_data_vector.end(), 0);
+ col =
vectorized::ColumnHelper::create_column<DataType>(final_data_vector);
+ std::vector<uint8_t> final_res(20);
+ wrapper->bloom_filter_func()->find_fixed_len(col, final_res.data());
+ EXPECT_TRUE(std::all_of(final_res.begin(), final_res.end(),
+ [](uint8_t i) -> bool { return i; }));
+ EXPECT_EQ(wrapper->get_real_type(), RuntimeFilterType::BLOOM_FILTER);
+ }
{
// In + Bloom -> Bloom
max_in_num = 18;
diff --git a/be/test/runtime_filter/utils_test.cpp
b/be/test/runtime_filter/utils_test.cpp
index ca7a91a1fc7..6f41bf2c2f3 100644
--- a/be/test/runtime_filter/utils_test.cpp
+++ b/be/test/runtime_filter/utils_test.cpp
@@ -105,7 +105,7 @@ TEST_F(RuntimeFilterUtilsTest, TestCreateBinaryPredicate) {
TExprNode pred_node;
TypeDescriptor type(PrimitiveType::TYPE_INT);
auto op = TExprOpcode::GE;
- EXPECT_TRUE(create_vbin_predicate(type, op, expr, &pred_node,
false).ok());
+ EXPECT_TRUE(create_vbin_predicate(type, op, expr, &pred_node,
true).ok());
}
{
vectorized::VExprSPtr expr;
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
