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 5bc15c779d3 [refactor](predefine) get type info by visitor and fix
compaction (#49867)
5bc15c779d3 is described below
commit 5bc15c779d323d914ae5c24c924bc51dde4033ef
Author: lihangyu <lihan...@selectdb.com>
AuthorDate: Tue Apr 8 19:11:50 2025 +0800
[refactor](predefine) get type info by visitor and fix compaction (#49867)
---
be/src/olap/rowset/segment_v2/column_reader.cpp | 1 -
be/src/vec/columns/column_object.cpp | 79 ++++++++++++---------
be/src/vec/common/field_visitors.h | 7 ++
be/src/vec/common/schema_util.cpp | 3 +-
be/src/vec/data_types/convert_field_to_type.cpp | 10 +++
regression-test/data/variant_p0/desc.out | Bin 5397 -> 5529 bytes
regression-test/data/variant_p0/predefine/load.out | Bin 6803 -> 7262 bytes
.../schema_change/test_alter_add_drop_column.out | Bin 0 -> 187 bytes
regression-test/suites/variant_p0/desc.groovy | 10 +--
.../suites/variant_p0/predefine/load.groovy | 27 +++++++
.../test_alter_add_drop_column.groovy | 26 ++++---
11 files changed, 114 insertions(+), 49 deletions(-)
diff --git a/be/src/olap/rowset/segment_v2/column_reader.cpp
b/be/src/olap/rowset/segment_v2/column_reader.cpp
index a38871155ce..388200b88e2 100644
--- a/be/src/olap/rowset/segment_v2/column_reader.cpp
+++ b/be/src/olap/rowset/segment_v2/column_reader.cpp
@@ -1888,7 +1888,6 @@ Status DefaultValueColumnIterator::init(const
ColumnIteratorOptions& opts) {
// "NULL" is a special default value which means the default value is null.
if (_has_default_value) {
if (_default_value == "NULL") {
- DCHECK(_is_nullable);
_is_default_value_null = true;
} else {
_type_size = _type_info->size();
diff --git a/be/src/vec/columns/column_object.cpp
b/be/src/vec/columns/column_object.cpp
index 00830d13a94..18ff9033cf9 100644
--- a/be/src/vec/columns/column_object.cpp
+++ b/be/src/vec/columns/column_object.cpp
@@ -151,6 +151,7 @@ public:
}
return 1 + dimensions;
}
+ size_t operator()(const VariantField& x) { return apply_visitor(*this,
x.get_field()); }
template <typename T>
size_t operator()(const T&) const {
return 0;
@@ -203,17 +204,32 @@ public:
type = TypeIndex::VARIANT;
return 1;
}
+ size_t operator()(const VariantField& x) {
+ typed_field_info =
+ FieldInfo {x.get_type_id(), true, false, 0, x.get_scale(),
x.get_precision()};
+ return 1;
+ }
template <typename T>
size_t operator()(const T&) {
type = TypeId<NearestFieldType<T>>::value;
return 1;
}
- void get_scalar_type(TypeIndex* data_type) const { *data_type = type; }
+ void get_scalar_type(TypeIndex* data_type, int* precision, int* scale)
const {
+ if (typed_field_info.has_value()) {
+ *data_type = typed_field_info->scalar_type_id;
+ *precision = typed_field_info->precision;
+ *scale = typed_field_info->scale;
+ return;
+ }
+ *data_type = type;
+ }
bool contain_nulls() const { return have_nulls; }
bool need_convert_field() const { return false; }
private:
+ // initialized when operator()(const VariantField& x)
+ std::optional<FieldInfo> typed_field_info;
TypeIndex type = TypeIndex::Nothing;
bool have_nulls = false;
};
@@ -271,6 +287,15 @@ public:
type_indexes.insert(TypeIndex::VARIANT);
return 0;
}
+ size_t operator()(const VariantField& x) {
+ if (x.get_type_id() == TypeIndex::Array) {
+ apply_visitor(*this, x.get_field());
+ } else {
+ typed_field_info =
+ FieldInfo {x.get_type_id(), true, false, 0, x.get_scale(),
x.get_precision()};
+ }
+ return 0;
+ }
size_t operator()(const Null&) {
have_nulls = true;
return 0;
@@ -282,7 +307,14 @@ public:
type_indexes.insert(TypeId<NearestFieldType<T>>::value);
return 0;
}
- void get_scalar_type(TypeIndex* type) const {
+ void get_scalar_type(TypeIndex* type, int* precision, int* scale) const {
+ if (typed_field_info.has_value()) {
+ // fast path
+ *type = typed_field_info->scalar_type_id;
+ *precision = typed_field_info->precision;
+ *scale = typed_field_info->scale;
+ return;
+ }
DataTypePtr data_type;
get_least_supertype_jsonb(type_indexes, &data_type);
*type = data_type->get_type_id();
@@ -291,6 +323,8 @@ public:
bool need_convert_field() const { return field_types.size() > 1; }
private:
+ // initialized when operator()(const VariantField& x)
+ std::optional<FieldInfo> typed_field_info;
phmap::flat_hash_set<TypeIndex> type_indexes;
phmap::flat_hash_set<FieldType> field_types;
bool have_nulls = false;
@@ -334,49 +368,28 @@ void get_field_info_impl(const Field& field, FieldInfo*
info) {
Visitor to_scalar_type_visitor;
apply_visitor(to_scalar_type_visitor, field);
TypeIndex type_id;
- to_scalar_type_visitor.get_scalar_type(&type_id);
+ int precision = 0;
+ int scale = 0;
+ to_scalar_type_visitor.get_scalar_type(&type_id, &precision, &scale);
// array item's dimension may missmatch, eg. [1, 2, [1, 2, 3]]
*info = {
type_id,
to_scalar_type_visitor.contain_nulls(),
to_scalar_type_visitor.need_convert_field(),
apply_visitor(FieldVisitorToNumberOfDimensions(), field),
+ scale,
+ precision,
};
}
-void get_base_field_info(const Field& field, FieldInfo* info) {
- const auto& variant_field = field.get<const VariantField&>();
- const auto& wrapped_field = variant_field.get_field();
- if (variant_field.get_type_id() == TypeIndex::Array) {
- if (wrapped_field.safe_get<Array>().empty()) {
- info->scalar_type_id = TypeIndex::Nothing;
- ++info->num_dimensions;
- info->have_nulls = true;
- info->need_convert = false;
- } else {
- ++info->num_dimensions;
- get_base_field_info(wrapped_field.safe_get<Array>()[0], info);
- }
- return;
- }
-
- // handle scalar types
- info->scalar_type_id = variant_field.get_type_id();
- info->have_nulls = true;
- info->need_convert = false;
- info->scale = variant_field.get_scale();
- info->precision = variant_field.get_precision();
+bool is_complex_field(const Field& field) {
+ return field.is_complex_field() ||
+ (field.is_variant_field() &&
+ field.get<const VariantField&>().get_field().is_complex_field());
}
void get_field_info(const Field& field, FieldInfo* info) {
- if (field.is_variant_field()) {
- // Currently we support specify predefined schema for other types
include decimal, datetime ...etc
- // so we should set specified info to create correct types, and those
predefined types are static and
- // type no need to deduce
- get_base_field_info(field, info);
- return;
- }
- if (field.is_complex_field()) {
+ if (is_complex_field(field)) {
get_field_info_impl<FieldVisitorToScalarType>(field, info);
} else {
get_field_info_impl<SimpleFieldVisitorToScalarType>(field, info);
diff --git a/be/src/vec/common/field_visitors.h
b/be/src/vec/common/field_visitors.h
index eefa424c641..e5468867b07 100644
--- a/be/src/vec/common/field_visitors.h
+++ b/be/src/vec/common/field_visitors.h
@@ -25,6 +25,7 @@
#include "vec/common/demangle.h"
#include "vec/core/accurate_comparison.h"
#include "vec/core/field.h"
+#include "vec/core/types.h"
namespace doris::vectorized {
@@ -71,6 +72,12 @@ typename std::decay_t<Visitor>::ResultType
apply_visitor(Visitor&& visitor, F&&
return visitor(field.template get<DecimalField<Decimal256>>());
case Field::Types::JSONB:
return visitor(field.template get<JsonbField>());
+ case Field::Types::Variant:
+ return visitor(field.template get<VariantField>());
+ case Field::Types::IPv6:
+ return visitor(field.template get<IPv6>());
+ case Field::Types::Int256:
+ return visitor(field.template get<Int128>());
default:
throw doris::Exception(ErrorCode::INTERNAL_ERROR, "Bad type of Field
{}",
static_cast<int>(field.get_type()));
diff --git a/be/src/vec/common/schema_util.cpp
b/be/src/vec/common/schema_util.cpp
index feadabc46da..17430380310 100644
--- a/be/src/vec/common/schema_util.cpp
+++ b/be/src/vec/common/schema_util.cpp
@@ -643,7 +643,8 @@ TabletColumn create_sparse_column(const TabletColumn&
variant) {
res.set_aggregation_method(variant.aggregation());
res.set_path_info(PathInData {variant.name_lower_case() + "." +
SPARSE_COLUMN_PATH});
res.set_parent_unique_id(variant.unique_id());
-
+ // set default value to "NULL" DefaultColumnIterator will call
insert_many_defaults
+ res.set_default_value("NULL");
TabletColumn child_tcolumn;
child_tcolumn.set_type(FieldType::OLAP_FIELD_TYPE_STRING);
res.add_sub_column(child_tcolumn);
diff --git a/be/src/vec/data_types/convert_field_to_type.cpp
b/be/src/vec/data_types/convert_field_to_type.cpp
index 55613e21921..2646d0e831d 100644
--- a/be/src/vec/data_types/convert_field_to_type.cpp
+++ b/be/src/vec/data_types/convert_field_to_type.cpp
@@ -44,6 +44,8 @@
#include "vec/data_types/data_type.h"
#include "vec/data_types/data_type_array.h"
#include "vec/data_types/data_type_nullable.h"
+#include "vec/io/io_helper.h"
+#include "vec/runtime/ipv6_value.h"
namespace doris::vectorized {
#include "common/compile_check_begin.h"
@@ -61,7 +63,12 @@ class FieldVisitorToStringSimple : public
StaticVisitor<String> {
public:
String operator()(const Null& x) const { return "NULL"; }
String operator()(const UInt64& x) const { return std::to_string(x); }
+ String operator()(const IPv6& x) const {
+ auto value = IPv6Value(x);
+ return value.to_string();
+ }
String operator()(const Int64& x) const { return std::to_string(x); }
+ String operator()(const Int128& x) const { return int128_to_string(x); }
String operator()(const Float64& x) const { return std::to_string(x); }
String operator()(const String& x) const { return x; }
[[noreturn]] String operator()(const UInt128& x) const {
@@ -73,6 +80,9 @@ public:
[[noreturn]] String operator()(const Tuple& x) const {
throw doris::Exception(ErrorCode::NOT_IMPLEMENTED_ERROR, "Not
implemeted");
}
+ [[noreturn]] String operator()(const VariantField& x) const {
+ throw doris::Exception(ErrorCode::NOT_IMPLEMENTED_ERROR, "Not
implemeted");
+ }
[[noreturn]] String operator()(const DecimalField<Decimal32>& x) const {
throw doris::Exception(ErrorCode::NOT_IMPLEMENTED_ERROR, "Not
implemeted");
}
diff --git a/regression-test/data/variant_p0/desc.out
b/regression-test/data/variant_p0/desc.out
index ffd87b9d05b..b6afa3f2101 100644
Binary files a/regression-test/data/variant_p0/desc.out and
b/regression-test/data/variant_p0/desc.out differ
diff --git a/regression-test/data/variant_p0/predefine/load.out
b/regression-test/data/variant_p0/predefine/load.out
index 06e06e16c58..ca4161c3a01 100644
Binary files a/regression-test/data/variant_p0/predefine/load.out and
b/regression-test/data/variant_p0/predefine/load.out differ
diff --git
a/regression-test/data/variant_p0/schema_change/test_alter_add_drop_column.out
b/regression-test/data/variant_p0/schema_change/test_alter_add_drop_column.out
new file mode 100644
index 00000000000..33fb39ee1d0
Binary files /dev/null and
b/regression-test/data/variant_p0/schema_change/test_alter_add_drop_column.out
differ
diff --git a/regression-test/suites/variant_p0/desc.groovy
b/regression-test/suites/variant_p0/desc.groovy
index 90ca4595fec..57ec6f360f3 100644
--- a/regression-test/suites/variant_p0/desc.groovy
+++ b/regression-test/suites/variant_p0/desc.groovy
@@ -57,7 +57,7 @@ suite("regression_test_variant_desc", "nonConcurrent"){
)
DUPLICATE KEY(`k`)
DISTRIBUTED BY HASH(k) BUCKETS ${buckets}
- properties("replication_num" = "1", "disable_auto_compaction" =
"false");
+ properties("replication_num" = "1", "disable_auto_compaction" =
"false", "variant_max_subcolumns_count" = "0");
"""
}
@@ -76,7 +76,7 @@ suite("regression_test_variant_desc", "nonConcurrent"){
PARTITION p3 VALUES LESS THAN (100000)
)
DISTRIBUTED BY HASH(k) BUCKETS ${buckets}
- properties("replication_num" = "1", "disable_auto_compaction" =
"false");
+ properties("replication_num" = "1", "disable_auto_compaction" =
"false", "variant_max_subcolumns_count" = "0");
"""
}
@@ -185,7 +185,7 @@ suite("regression_test_variant_desc", "nonConcurrent"){
)
DUPLICATE KEY(`k`)
DISTRIBUTED BY HASH(k) BUCKETS 5
- properties("replication_num" = "1", "disable_auto_compaction" =
"false");
+ properties("replication_num" = "1", "disable_auto_compaction" =
"false", "variant_max_subcolumns_count" = "0");
"""
sql """ insert into ${table_name} values (0, '{"a": 1123, "b" : [123,
{"xx" : 1}], "c" : {"c" : 456, "d" : null, "e" : 7.111}, "zzz" : null, "oooo" :
{"akakaka" : null, "xxxx" : {"xxx" : 123}}}')"""
sql "select * from ${table_name} limit 1"
@@ -228,7 +228,7 @@ suite("regression_test_variant_desc", "nonConcurrent"){
)
DUPLICATE KEY(`k`)
DISTRIBUTED BY HASH(k) BUCKETS 5
- properties("replication_num" = "1", "disable_auto_compaction" =
"false");
+ properties("replication_num" = "1", "disable_auto_compaction" =
"false", "variant_max_subcolumns_count" = "0");
"""
sql """ insert into ${table_name} values (0, '{"名字" : "jack",
"!@#^&*()": "11111", "金额" : 200, "画像" : {"地址" : "北京", "\\\u4E2C\\\u6587":
"unicode"}}')"""
sql """set describe_extend_variant_column = true"""
@@ -244,7 +244,7 @@ suite("regression_test_variant_desc", "nonConcurrent"){
)
DUPLICATE KEY(`k`)
DISTRIBUTED BY HASH(k) BUCKETS 5
- properties("replication_num" = "1", "disable_auto_compaction" =
"false");
+ properties("replication_num" = "1", "disable_auto_compaction" =
"false", "variant_max_subcolumns_count" = "0");
"""
sql """ insert into ${table_name} values (0, '{}')"""
sql """ insert into ${table_name} values (0, '100')"""
diff --git a/regression-test/suites/variant_p0/predefine/load.groovy
b/regression-test/suites/variant_p0/predefine/load.groovy
index 1093ee3a25f..7ccb8068024 100644
--- a/regression-test/suites/variant_p0/predefine/load.groovy
+++ b/regression-test/suites/variant_p0/predefine/load.groovy
@@ -292,4 +292,31 @@ suite("regression_test_variant_predefine_schema", "p0"){
"variant_max_subcolumns_count" = "0"
);
"""
+
+ // array with nulls
+
+ sql "DROP TABLE IF EXISTS test_array_with_nulls"
+ // test bf with bool
+ sql """
+ CREATE TABLE `test_array_with_nulls` (
+ `k` bigint NULL,
+ `var` variant<array_decimal:array<decimalv3(27,9)>>
+ ) ENGINE=OLAP
+ DUPLICATE KEY(`k`)
+ DISTRIBUTED BY HASH(`k`) BUCKETS 1
+ PROPERTIES (
+ "replication_allocation" = "tag.location.default: 1",
+ "min_load_replica_num" = "-1",
+ "variant_max_subcolumns_count" = "0"
+ );
+ """
+ sql """insert into test_array_with_nulls values(3, '{"array_decimal" :
[null, 2.2, 3.3, 4.4]}')"""
+ qt_sql_arr_null_1 "select * from test_array_with_nulls order by k"
+ sql """insert into test_array_with_nulls values(1, '{"array_decimal" :
[1.1, 2.2, 3.3, null]}')"""
+ sql """insert into test_array_with_nulls values(2, '{"array_decimal" :
[1.1, 2.2, null, 4.4]}')"""
+ sql """insert into test_array_with_nulls values(4, '{"array_decimal" :
[1.1, null, 3.3, 4.4]}')"""
+ sql """insert into test_array_with_nulls values(5, '{"array_decimal" :
[1.1, 2.2, 3.3, 4.4]}')"""
+ sql """insert into test_array_with_nulls values(6, '{"array_decimal" :
[]}')"""
+ sql """insert into test_array_with_nulls values(7, '{"array_decimal" :
[null, null]}')"""
+ qt_sql_arr_null_2 "select * from test_array_with_nulls order by k"
}
\ No newline at end of file
diff --git
a/regression-test/suites/variant_p0/schema_change/test_alter_add_drop_column.groovy
b/regression-test/suites/variant_p0/schema_change/test_alter_add_drop_column.groovy
index ce55a62af65..df2b3c46a60 100644
---
a/regression-test/suites/variant_p0/schema_change/test_alter_add_drop_column.groovy
+++
b/regression-test/suites/variant_p0/schema_change/test_alter_add_drop_column.groovy
@@ -25,7 +25,7 @@ suite("regression_test_variant_add_drop_column",
"variant_type"){
)
DUPLICATE KEY(`k`)
DISTRIBUTED BY HASH(k) BUCKETS 1
- properties("replication_num" = "1");
+ properties("replication_num" = "1", "disable_auto_compaction" =
"true");
"""
sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}')"""
@@ -34,20 +34,28 @@ suite("regression_test_variant_add_drop_column",
"variant_type"){
sql "alter table variant_add_drop_column add column t2 datetime default
null"
sql """insert into variant_add_drop_column values (1, '{"a" :
12345234567,"b" : 2}', '{"xxx" : 1}', "2021-01-01 01:01:01", "2021-01-01
01:01:01")"""
sql "alter table variant_add_drop_column add column i1 int default null"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}', "2021-01-01 01:01:01", "2021-01-01 01:01:01", 12345)"""
+ sql """insert into variant_add_drop_column values (2, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}', "2021-01-01 01:01:01", "2021-01-01 01:01:01", 12345)"""
sql "alter table variant_add_drop_column drop column t1"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}', "2021-01-01 01:01:01", 12345)"""
+ sql """insert into variant_add_drop_column values (3, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}', "2021-01-01 01:01:01", 12345)"""
sql "alter table variant_add_drop_column drop column t2"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}', 12345)"""
+ sql """insert into variant_add_drop_column values (4, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}', 12345)"""
sql "alter table variant_add_drop_column drop column i1"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}')"""
+ sql """insert into variant_add_drop_column values (5, '{"a" : 12345,"b" :
2}', '{"xxx" : 1}')"""
sql "alter table variant_add_drop_column drop column v"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}')"""
+ sql """insert into variant_add_drop_column values (6, '{"a" : 12345,"b" :
2}')"""
sql "alter table variant_add_drop_column add column v variant default null"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}', '{"a" : 12345,"b" : 2}')"""
+ sql """insert into variant_add_drop_column values (7, '{"a" : 12345,"b" :
2}', '{"a" : 12345,"b" : 2}')"""
sql "alter table variant_add_drop_column add column v3 variant default
null"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}', '{"a" : 12345,"b" : 2}', '{"a" : 12345,"b" : 2}')"""
+ sql """insert into variant_add_drop_column values (8, '{"a" : 12345,"b" :
2}', '{"a" : 12345,"b" : 2}', '{"a" : 12345,"b" : 2}')"""
sql "alter table variant_add_drop_column drop column v"
sql "alter table variant_add_drop_column drop column v2"
- sql """insert into variant_add_drop_column values (1, '{"a" : 12345,"b" :
2}')"""
+ sql """insert into variant_add_drop_column values (9, '{"a" : 12345,"b" :
2}')"""
+
+ // trigger compactions for all tablets in ${tableName}
+ def tablets = sql_return_maparray """ show tablets from ${table_name}; """
+
+ // trigger compactions for all tablets in ${tableName}
+ trigger_and_wait_compaction(table_name, "cumulative")
+
+ qt_sql "select * from variant_add_drop_column order by k limit 10"
}
\ No newline at end of file
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
