This is an automated email from the ASF dual-hosted git repository.
panxiaolei pushed a commit to branch new_join2
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/new_join2 by this push:
new 05d723dab44 fix
05d723dab44 is described below
commit 05d723dab449f68ece332c7880ec76549bbe23a8
Author: BiteTheDDDDt <pxl...@qq.com>
AuthorDate: Fri Nov 24 19:06:08 2023 +0800
fix
---
be/src/vec/common/hash_table/hash_map.h | 12 +++++++++--
be/src/vec/exec/join/process_hash_table_probe.h | 2 +-
.../vec/exec/join/process_hash_table_probe_impl.h | 23 ++++++++++------------
be/src/vec/exec/join/vhash_join_node.h | 3 ++-
4 files changed, 23 insertions(+), 17 deletions(-)
diff --git a/be/src/vec/common/hash_table/hash_map.h
b/be/src/vec/common/hash_table/hash_map.h
index 08ca7628a03..39872e205b3 100644
--- a/be/src/vec/common/hash_table/hash_map.h
+++ b/be/src/vec/common/hash_table/hash_map.h
@@ -224,7 +224,8 @@ public:
}
template <int JoinOpType>
- void prepare_build(size_t num_elem, int batch_size) {
+ void prepare_build(size_t num_elem, int batch_size, bool has_null_key) {
+ _has_null_key = has_null_key;
max_batch_size = batch_size;
bucket_size = calc_bucket_size(num_elem + 1);
first.resize(bucket_size + 1);
@@ -314,6 +315,8 @@ public:
return iter_idx >= elem_num;
}
+ bool has_null_key() { return _has_null_key; }
+
private:
// only LEFT_ANTI_JOIN/LEFT_SEMI_JOIN/NULL_AWARE_LEFT_ANTI_JOIN/CROSS_JOIN
support mark join
template <int JoinOpType, bool with_other_conjuncts>
@@ -338,7 +341,11 @@ private:
} else {
bool matched = JoinOpType ==
doris::TJoinOp::LEFT_SEMI_JOIN ? build_idx != 0
: build_idx == 0;
- mark_column->insert_value(matched);
+ if (!matched && _has_null_key) {
+ mark_column->insert_null();
+ } else {
+ mark_column->insert_value(matched);
+ }
}
}
@@ -515,6 +522,7 @@ private:
mutable uint32_t iter_idx = 1;
Cell cell;
doris::vectorized::Arena* pool;
+ bool _has_null_key = false;
};
template <typename Key, typename Mapped, typename Hash = DefaultHash<Key>,
diff --git a/be/src/vec/exec/join/process_hash_table_probe.h
b/be/src/vec/exec/join/process_hash_table_probe.h
index e1e893ce190..995c3992245 100644
--- a/be/src/vec/exec/join/process_hash_table_probe.h
+++ b/be/src/vec/exec/join/process_hash_table_probe.h
@@ -68,7 +68,7 @@ struct ProcessHashTableProbe {
// and output block may be different
// The output result is determined by the other join conjunct result and
same_to_prev struct
Status do_other_join_conjuncts(Block* output_block, bool is_mark_join,
- bool is_the_last_sub_block,
std::vector<uint8_t>& visited);
+ std::vector<uint8_t>& visited, bool
has_null_in_build_side);
template <typename HashTableType>
typename HashTableType::State _init_probe_side(HashTableType&
hash_table_ctx, size_t probe_rows,
diff --git a/be/src/vec/exec/join/process_hash_table_probe_impl.h
b/be/src/vec/exec/join/process_hash_table_probe_impl.h
index a93736b4371..d38189087c7 100644
--- a/be/src/vec/exec/join/process_hash_table_probe_impl.h
+++ b/be/src/vec/exec/join/process_hash_table_probe_impl.h
@@ -164,9 +164,6 @@ Status ProcessHashTableProbe<JoinOpType,
Parent>::do_process(HashTableType& hash
bool all_match_one = false;
size_t probe_size = 0;
- // Is the last sub block of splitted block
- bool is_the_last_sub_block = false;
-
std::unique_ptr<ColumnFilterHelper> mark_column;
if (is_mark_join) {
mark_column = std::make_unique<ColumnFilterHelper>(*mcol[mcol.size() -
1]);
@@ -199,8 +196,9 @@ Status ProcessHashTableProbe<JoinOpType,
Parent>::do_process(HashTableType& hash
output_block->swap(mutable_block.to_block());
if constexpr (with_other_conjuncts) {
- return do_other_join_conjuncts(output_block, is_mark_join,
is_the_last_sub_block,
-
hash_table_ctx.hash_table->get_visited());
+ return do_other_join_conjuncts(output_block, is_mark_join,
+
hash_table_ctx.hash_table->get_visited(),
+
hash_table_ctx.hash_table->has_null_key());
}
return Status::OK();
@@ -208,8 +206,8 @@ Status ProcessHashTableProbe<JoinOpType,
Parent>::do_process(HashTableType& hash
template <int JoinOpType, typename Parent>
Status ProcessHashTableProbe<JoinOpType, Parent>::do_other_join_conjuncts(
- Block* output_block, bool is_mark_join, bool is_the_last_sub_block,
- std::vector<uint8_t>& visited) {
+ Block* output_block, bool is_mark_join, std::vector<uint8_t>& visited,
+ bool has_null_in_build_side) {
// dispose the other join conjunct exec
auto row_count = output_block->rows();
if (!row_count) {
@@ -280,13 +278,12 @@ Status ProcessHashTableProbe<JoinOpType,
Parent>::do_other_join_conjuncts(
for (size_t i = 0; i < row_count; ++i) {
filter_map[i] = true;
- if constexpr (JoinOpType != TJoinOp::LEFT_SEMI_JOIN) {
- if (!_build_indexs[i]) {
- helper.insert_null();
- continue;
- }
+ if (has_null_in_build_side &&
+ (_build_indexs[i] != 0) ^ (JoinOpType ==
TJoinOp::LEFT_SEMI_JOIN)) {
+ helper.insert_null();
+ } else {
+ helper.insert_value(filter_column_ptr[i]);
}
- helper.insert_value(filter_column_ptr[i]);
}
} else {
if constexpr (JoinOpType == TJoinOp::LEFT_SEMI_JOIN) {
diff --git a/be/src/vec/exec/join/vhash_join_node.h
b/be/src/vec/exec/join/vhash_join_node.h
index aecd49c8dbc..7f66fa85905 100644
--- a/be/src/vec/exec/join/vhash_join_node.h
+++ b/be/src/vec/exec/join/vhash_join_node.h
@@ -133,7 +133,8 @@ struct ProcessHashTableBuild {
}
SCOPED_TIMER(_parent->_build_table_insert_timer);
- hash_table_ctx.hash_table->template prepare_build<JoinOpType>(_rows,
_batch_size);
+ hash_table_ctx.hash_table->template prepare_build<JoinOpType>(_rows,
_batch_size,
+
*has_null_key);
hash_table_ctx.init_serialized_keys(_build_raw_ptrs, _rows,
null_map ? null_map->data() :
nullptr, true, true,
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
