This is an automated email from the ASF dual-hosted git repository.
eldenmoon pushed a commit to branch variant-sparse
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/variant-sparse by this push:
new f64ddfebad1 [Improvement](Variant) optimize performance while loading
variant data (#49316)
f64ddfebad1 is described below
commit f64ddfebad1daffa8cad9add91596c8c1f3961b5
Author: lihangyu <lihan...@selectdb.com>
AuthorDate: Thu Mar 20 17:23:39 2025 +0800
[Improvement](Variant) optimize performance while loading variant data
(#49316)
---
.../rowset/segment_v2/vertical_segment_writer.cpp | 2 +-
be/src/vec/columns/column_object.cpp | 32 ++++++++++++++++------
be/src/vec/columns/column_object.h | 15 ++++++++++
be/src/vec/json/parse2column.cpp | 12 ++++++--
be/src/vec/olap/olap_data_convertor.cpp | 4 +++
be/src/vec/olap/olap_data_convertor.h | 1 +
6 files changed, 53 insertions(+), 13 deletions(-)
diff --git a/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
b/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
index 8e8c1a5d209..1f559ae6217 100644
--- a/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
+++ b/be/src/olap/rowset/segment_v2/vertical_segment_writer.cpp
@@ -1097,7 +1097,7 @@ Status
VerticalSegmentWriter::_append_block_with_variant_subcolumns(RowsInBlock&
RETURN_IF_ERROR(_column_writers[current_column_id]->append(
column->get_nullmap(), column->get_data(), data.num_rows));
_flush_schema->append_column(tablet_column);
- _olap_data_convertor->clear_source_content();
+ _olap_data_convertor->clear_source_content(current_column_id);
}
}
diff --git a/be/src/vec/columns/column_object.cpp
b/be/src/vec/columns/column_object.cpp
index 90720c57235..15b24589d46 100644
--- a/be/src/vec/columns/column_object.cpp
+++ b/be/src/vec/columns/column_object.cpp
@@ -368,7 +368,10 @@ DataTypeSerDeSPtr
ColumnObject::Subcolumn::generate_data_serdes(DataTypePtr type
ColumnObject::Subcolumn::Subcolumn(MutableColumnPtr&& data_, DataTypePtr type,
bool is_nullable_,
bool is_root_)
- : least_common_type(type), is_nullable(is_nullable_),
is_root(is_root_) {
+ : least_common_type(type),
+ is_nullable(is_nullable_),
+ is_root(is_root_),
+ num_rows(data_->size()) {
data.push_back(std::move(data_));
data_types.push_back(type);
data_serdes.push_back(generate_data_serdes(type, is_root));
@@ -380,14 +383,11 @@ ColumnObject::Subcolumn::Subcolumn(size_t size_, bool
is_nullable_, bool is_root
: least_common_type(std::make_shared<DataTypeNothing>()),
is_nullable(is_nullable_),
num_of_defaults_in_prefix(size_),
- is_root(is_root_) {}
+ is_root(is_root_),
+ num_rows(size_) {}
size_t ColumnObject::Subcolumn::Subcolumn::size() const {
- size_t res = num_of_defaults_in_prefix;
- for (const auto& part : data) {
- res += part->size();
- }
- return res;
+ return num_rows + current_num_of_defaults;
}
size_t ColumnObject::Subcolumn::Subcolumn::byteSize() const {
@@ -423,6 +423,7 @@ void
ColumnObject::Subcolumn::add_new_column_part(DataTypePtr type) {
void ColumnObject::Subcolumn::insert(Field field, FieldInfo info) {
auto base_type = WhichDataType(info.scalar_type_id);
+ ++num_rows;
if (base_type.is_nothing() && info.num_dimensions == 0) {
insert_default();
return;
@@ -554,6 +555,7 @@ void ColumnObject::Subcolumn::insert_range_from(const
Subcolumn& src, size_t sta
length, src.size());
}
size_t end = start + length;
+ num_rows += length;
// num_rows += length;
if (data.empty()) {
add_new_column_part(src.get_least_common_type());
@@ -634,7 +636,7 @@ void ColumnObject::Subcolumn::insert_range_from(const
Subcolumn& src, size_t sta
}
bool ColumnObject::Subcolumn::is_finalized() const {
- return num_of_defaults_in_prefix == 0 && (data.empty() || (data.size() ==
1));
+ return current_num_of_defaults == 0 && num_of_defaults_in_prefix == 0 &&
(data.empty() || (data.size() == 1));
}
template <typename Func>
@@ -681,6 +683,10 @@ void ColumnObject::resize(size_t n) {
}
void ColumnObject::Subcolumn::finalize(FinalizeMode mode) {
+ if (current_num_of_defaults) {
+ insert_many_defaults(current_num_of_defaults);
+ current_num_of_defaults = 0;
+ }
if (!is_root && data.size() == 1 && num_of_defaults_in_prefix == 0) {
data[0] = data[0]->convert_to_full_column_if_const();
return;
@@ -722,11 +728,12 @@ void ColumnObject::Subcolumn::finalize(FinalizeMode mode)
{
}
void ColumnObject::Subcolumn::insert_default() {
- if (data.empty()) {
+ if (UNLIKELY(data.empty())) {
++num_of_defaults_in_prefix;
} else {
data.back()->insert_default();
}
+ ++num_rows;
}
void ColumnObject::Subcolumn::insert_many_defaults(size_t length) {
@@ -735,6 +742,7 @@ void ColumnObject::Subcolumn::insert_many_defaults(size_t
length) {
} else {
data.back()->insert_many_defaults(length);
}
+ num_rows += length;
}
void ColumnObject::Subcolumn::pop_back(size_t n) {
@@ -742,6 +750,7 @@ void ColumnObject::Subcolumn::pop_back(size_t n) {
throw doris::Exception(ErrorCode::OUT_OF_BOUND,
"Invalid number of elements to pop: {}, size:
{}", n, size());
}
+ num_rows -= n;
size_t num_removed = 0;
for (auto it = data.rbegin(); it != data.rend(); ++it) {
if (n == 0) {
@@ -2127,6 +2136,7 @@ void ColumnObject::ensure_root_node_type(const
DataTypePtr& expected_root_type)
root.data[0] = casted_column;
root.data_types[0] = expected_root_type;
root.least_common_type = Subcolumn::LeastCommonType
{expected_root_type, true};
+ root.num_rows = casted_column->size();
}
}
@@ -2181,6 +2191,9 @@ size_t ColumnObject::filter(const Filter& filter) {
if (count == 0) {
clear();
} else {
+ for (auto& subcolumn : subcolumns) {
+ subcolumn->data.num_rows = count;
+ }
for_each_subcolumn([&](auto& part) {
if (part->size() != count) {
if (part->is_exclusive()) {
@@ -2210,6 +2223,7 @@ void ColumnObject::clear_column_data() {
(*std::move(part)).clear();
}
entry->data.num_of_defaults_in_prefix = 0;
+ entry->data.num_rows = 0;
}
serialized_sparse_column->clear();
num_rows = 0;
diff --git a/be/src/vec/columns/column_object.h
b/be/src/vec/columns/column_object.h
index c367da89da4..422bb38a704 100644
--- a/be/src/vec/columns/column_object.h
+++ b/be/src/vec/columns/column_object.h
@@ -154,6 +154,18 @@ public:
void insert_default();
+ void increment_default_counter() {
+ ++current_num_of_defaults;
+ }
+
+ void reset_current_num_of_defaults() {
+ current_num_of_defaults = 0;
+ }
+
+ size_t cur_num_of_defaults() {
+ return current_num_of_defaults;
+ }
+
void insert_many_defaults(size_t length);
void insert_range_from(const Subcolumn& src, size_t start, size_t
length);
@@ -252,6 +264,9 @@ public:
// If it is the root subcolumn of SubcolumnsTree,
// the root Node should be JSONB type when finalize
bool is_root = false;
+ size_t num_rows = 0;
+ // distinguish from num_of_defaults_in_prefix when data is not empty
+ size_t current_num_of_defaults = 0;
};
using Subcolumns = SubcolumnsTree<Subcolumn>;
diff --git a/be/src/vec/json/parse2column.cpp b/be/src/vec/json/parse2column.cpp
index 0e8472928a5..4ef416dc865 100644
--- a/be/src/vec/json/parse2column.cpp
+++ b/be/src/vec/json/parse2column.cpp
@@ -154,7 +154,7 @@ void parse_json_to_variant(IColumn& column, const char*
src, size_t length,
}
auto& [paths, values] = *result;
assert(paths.size() == values.size());
- size_t old_num_rows = column_object.size();
+ size_t old_num_rows = column_object.rows();
for (size_t i = 0; i < paths.size(); ++i) {
FieldInfo field_info;
get_field_info(values[i], &field_info);
@@ -173,6 +173,10 @@ void parse_json_to_variant(IColumn& column, const char*
src, size_t length,
throw doris::Exception(ErrorCode::INVALID_ARGUMENT, "Failed to
find sub column {}",
paths[i].get_path());
}
+ if (subcolumn->cur_num_of_defaults() > 0) {
+ subcolumn->insert_many_defaults(subcolumn->cur_num_of_defaults());
+ subcolumn->reset_current_num_of_defaults();
+ }
if (subcolumn->size() != old_num_rows) {
throw doris::Exception(ErrorCode::INVALID_ARGUMENT,
"subcolumn {} size missmatched, may
contains duplicated entry",
@@ -184,9 +188,10 @@ void parse_json_to_variant(IColumn& column, const char*
src, size_t length,
const auto& subcolumns = column_object.get_subcolumns();
for (const auto& entry : subcolumns) {
if (entry->data.size() == old_num_rows) {
- bool inserted =
column_object.try_insert_default_from_nested(entry);
+ bool inserted = UNLIKELY(entry->path.has_nested_part() &&
+
column_object.try_insert_default_from_nested(entry));
if (!inserted) {
- entry->data.insert_default();
+ entry->data.increment_default_counter();
}
}
}
@@ -213,6 +218,7 @@ void parse_json_to_variant(IColumn& column, const
ColumnString& raw_json_column,
StringRef raw_json = raw_json_column.get_data_at(i);
parse_json_to_variant(column, raw_json.data, raw_json.size,
parser.get(), config);
}
+ column.finalize();
}
} // namespace doris::vectorized
diff --git a/be/src/vec/olap/olap_data_convertor.cpp
b/be/src/vec/olap/olap_data_convertor.cpp
index 47e85633488..c2887c41934 100644
--- a/be/src/vec/olap/olap_data_convertor.cpp
+++ b/be/src/vec/olap/olap_data_convertor.cpp
@@ -285,6 +285,10 @@ void OlapBlockDataConvertor::clear_source_content() {
}
}
+void OlapBlockDataConvertor::clear_source_content(size_t cid) {
+ _convertors[cid]->clear_source_column();
+}
+
std::pair<Status, IOlapColumnDataAccessor*>
OlapBlockDataConvertor::convert_column_data(
size_t cid) {
assert(cid < _convertors.size());
diff --git a/be/src/vec/olap/olap_data_convertor.h
b/be/src/vec/olap/olap_data_convertor.h
index 936179356bb..c2a8e6ace47 100644
--- a/be/src/vec/olap/olap_data_convertor.h
+++ b/be/src/vec/olap/olap_data_convertor.h
@@ -84,6 +84,7 @@ public:
size_t row_pos, size_t
num_rows, uint32_t cid);
void clear_source_content();
+ void clear_source_content(size_t cid);
std::pair<Status, IOlapColumnDataAccessor*> convert_column_data(size_t
cid);
void add_column_data_convertor(const TabletColumn& column);
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
