This is an automated email from the ASF dual-hosted git repository.
eldenmoon pushed a commit to branch branch-2.1
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/branch-2.1 by this push:
new 4cf769b39fb [Improve](table-function) explode json array with json
args (#39491)
4cf769b39fb is described below
commit 4cf769b39fb60e77630bd3a2d2c8b1309b000e2e
Author: amory <wangqian...@selectdb.com>
AuthorDate: Tue Aug 27 14:53:17 2024 +0800
[Improve](table-function) explode json array with json args (#39491)
---
.../exprs/table_function/vexplode_json_array.cpp | 24 +++-
.../vec/exprs/table_function/vexplode_json_array.h | 151 +++++++++++++++++++++
be/src/vec/functions/function_fake.cpp | 52 ++++---
be/src/vec/functions/function_fake.h | 5 +
.../generator/ExplodeJsonArrayDouble.java | 2 +
.../generator/ExplodeJsonArrayDoubleOuter.java | 2 +
.../functions/generator/ExplodeJsonArrayInt.java | 2 +
.../generator/ExplodeJsonArrayIntOuter.java | 2 +
.../functions/generator/ExplodeJsonArrayJson.java | 2 +
.../generator/ExplodeJsonArrayJsonOuter.java | 2 +
.../generator/ExplodeJsonArrayString.java | 2 +
.../generator/ExplodeJsonArrayStringOuter.java | 2 +
.../data/nereids_function_p0/gen_function/gen.out | 41 ++++++
.../table_function/explode_json_array.out | 16 ++-
.../nereids_function_p0/gen_function/gen.groovy | 3 +
.../table_function/explode_json_array.groovy | 6 +-
16 files changed, 291 insertions(+), 23 deletions(-)
diff --git a/be/src/vec/exprs/table_function/vexplode_json_array.cpp
b/be/src/vec/exprs/table_function/vexplode_json_array.cpp
index 00c4d92a359..3c22ef4e078 100644
--- a/be/src/vec/exprs/table_function/vexplode_json_array.cpp
+++ b/be/src/vec/exprs/table_function/vexplode_json_array.cpp
@@ -26,6 +26,8 @@
#include <limits>
#include "common/status.h"
+#include "util/jsonb_parser.h"
+#include "util/jsonb_utils.h"
#include "vec/columns/column.h"
#include "vec/columns/column_nullable.h"
#include "vec/columns/columns_number.h"
@@ -50,6 +52,7 @@ Status
VExplodeJsonArrayTableFunction<DataImpl>::process_init(Block* block, Runt
RETURN_IF_ERROR(_expr_context->root()->children()[0]->execute(_expr_context.get(),
block,
&text_column_idx));
_text_column = block->get_by_position(text_column_idx).column;
+ _text_datatype =
remove_nullable(block->get_by_position(text_column_idx).type);
return Status::OK();
}
@@ -59,10 +62,20 @@ void
VExplodeJsonArrayTableFunction<DataImpl>::process_row(size_t row_idx) {
StringRef text = _text_column->get_data_at(row_idx);
if (text.data != nullptr) {
- rapidjson::Document document;
- document.Parse(text.data, text.size);
- if (!document.HasParseError() && document.IsArray() &&
document.GetArray().Size()) {
- _cur_size = _parsed_data.set_output(document,
document.GetArray().Size());
+ if (WhichDataType(_text_datatype).is_json()) {
+ JsonbDocument* doc = JsonbDocument::createDocument(text.data,
text.size);
+ if (doc && doc->getValue() && doc->getValue()->isArray()) {
+ auto* a = (ArrayVal*)doc->getValue();
+ if (a->numElem() > 0) {
+ _cur_size = _parsed_data.set_output(*a, a->numElem());
+ }
+ }
+ } else {
+ rapidjson::Document document;
+ document.Parse(text.data, text.size);
+ if (!document.HasParseError() && document.IsArray() &&
document.GetArray().Size()) {
+ _cur_size = _parsed_data.set_output(document,
document.GetArray().Size());
+ }
}
}
}
@@ -70,6 +83,7 @@ void
VExplodeJsonArrayTableFunction<DataImpl>::process_row(size_t row_idx) {
template <typename DataImpl>
void VExplodeJsonArrayTableFunction<DataImpl>::process_close() {
_text_column = nullptr;
+ _text_datatype = nullptr;
_parsed_data.reset();
}
@@ -141,4 +155,4 @@ template class
VExplodeJsonArrayTableFunction<ParsedDataDouble>;
template class VExplodeJsonArrayTableFunction<ParsedDataString>;
template class VExplodeJsonArrayTableFunction<ParsedDataJSON>;
-} // namespace doris::vectorized
\ No newline at end of file
+} // namespace doris::vectorized
diff --git a/be/src/vec/exprs/table_function/vexplode_json_array.h
b/be/src/vec/exprs/table_function/vexplode_json_array.h
index 1d89d9fa57d..acf31ed40d9 100644
--- a/be/src/vec/exprs/table_function/vexplode_json_array.h
+++ b/be/src/vec/exprs/table_function/vexplode_json_array.h
@@ -32,6 +32,7 @@
#include "vec/core/types.h"
#include "vec/data_types/data_type.h"
#include "vec/exprs/table_function/table_function.h"
+#include "vec/functions/function_string.h"
namespace doris::vectorized {
@@ -44,6 +45,7 @@ struct ParsedData {
_values_null_flag.clear();
}
virtual int set_output(rapidjson::Document& document, int value_size) = 0;
+ virtual int set_output(ArrayVal& array_doc, int value_size) = 0;
virtual void insert_result_from_parsed_data(MutableColumnPtr& column,
int64_t cur_offset,
int max_step) = 0;
virtual void insert_many_same_value_from_parsed_data(MutableColumnPtr&
column,
@@ -90,6 +92,36 @@ struct ParsedDataInt : public ParsedData<int64_t> {
}
return value_size;
}
+ int set_output(ArrayVal& array_doc, int value_size) override {
+ _values_null_flag.resize(value_size, 0);
+ _backup_data.resize(value_size);
+ int i = 0;
+ for (auto& val : array_doc) {
+ if (val.isInt8()) {
+ _backup_data[i] = static_cast<const JsonbInt8Val&>(val).val();
+ } else if (val.isInt16()) {
+ _backup_data[i] = static_cast<const JsonbInt16Val&>(val).val();
+ } else if (val.isInt32()) {
+ _backup_data[i] = static_cast<const JsonbInt32Val&>(val).val();
+ } else if (val.isInt64()) {
+ _backup_data[i] = static_cast<const JsonbInt64Val&>(val).val();
+ } else if (val.isDouble()) {
+ auto value = static_cast<const JsonbDoubleVal&>(val).val();
+ if (value > MAX_VALUE) {
+ _backup_data[i] = MAX_VALUE;
+ } else if (value < MIN_VALUE) {
+ _backup_data[i] = MIN_VALUE;
+ } else {
+ _backup_data[i] = long(value);
+ }
+ } else {
+ _values_null_flag[i] = 1;
+ _backup_data[i] = 0;
+ }
+ ++i;
+ }
+ return value_size;
+ }
void insert_result_from_parsed_data(MutableColumnPtr& column, int64_t
cur_offset,
int max_step) override {
@@ -121,6 +153,22 @@ struct ParsedDataDouble : public ParsedData<double> {
return value_size;
}
+ int set_output(ArrayVal& array_doc, int value_size) override {
+ _values_null_flag.resize(value_size, 0);
+ _backup_data.resize(value_size);
+ int i = 0;
+ for (auto& val : array_doc) {
+ if (val.isDouble()) {
+ _backup_data[i] = static_cast<const
JsonbDoubleVal&>(val).val();
+ } else {
+ _backup_data[i] = 0;
+ _values_null_flag[i] = 1;
+ }
+ ++i;
+ }
+ return value_size;
+ }
+
void insert_result_from_parsed_data(MutableColumnPtr& column, int64_t
cur_offset,
int max_step) override {
assert_cast<ColumnFloat64*>(column.get())
@@ -220,6 +268,83 @@ struct ParsedDataString : public ParsedDataStringBase {
}
return value_size;
}
+
+ int set_output(ArrayVal& array_doc, int value_size) override {
+ _data_string_ref.clear();
+ _backup_data.clear();
+ _values_null_flag.clear();
+ int32_t wbytes = 0;
+ for (auto& val : array_doc) {
+ switch (val.type()) {
+ case JsonbType::T_String: {
+ _backup_data.emplace_back(static_cast<const
JsonbStringVal&>(val).getBlob(),
+ static_cast<const
JsonbStringVal&>(val).getBlobLen());
+ _values_null_flag.emplace_back(false);
+ break;
+ // do not set _data_string here.
+ // Because the address of the string stored in `_backup_data`
may
+ // change each time `emplace_back()` is called.
+ }
+ case JsonbType::T_Int8: {
+ wbytes = snprintf(tmp_buf, sizeof(tmp_buf), "%d",
+ static_cast<const JsonbInt8Val&>(val).val());
+ _backup_data.emplace_back(tmp_buf, wbytes);
+ _values_null_flag.emplace_back(false);
+ break;
+ }
+ case JsonbType::T_Int16: {
+ wbytes = snprintf(tmp_buf, sizeof(tmp_buf), "%d",
+ static_cast<const
JsonbInt16Val&>(val).val());
+ _backup_data.emplace_back(tmp_buf, wbytes);
+ _values_null_flag.emplace_back(false);
+ break;
+ }
+ case JsonbType::T_Int64: {
+ wbytes = snprintf(tmp_buf, sizeof(tmp_buf), "%" PRId64,
+ static_cast<const
JsonbInt64Val&>(val).val());
+ _backup_data.emplace_back(tmp_buf, wbytes);
+ _values_null_flag.emplace_back(false);
+ break;
+ }
+ case JsonbType::T_Double: {
+ wbytes = snprintf(tmp_buf, sizeof(tmp_buf), "%f",
+ static_cast<const
JsonbDoubleVal&>(val).val());
+ _backup_data.emplace_back(tmp_buf, wbytes);
+ _values_null_flag.emplace_back(false);
+ break;
+ }
+ case JsonbType::T_Int32: {
+ wbytes = snprintf(tmp_buf, sizeof(tmp_buf), "%d",
+ static_cast<const
JsonbInt32Val&>(val).val());
+ _backup_data.emplace_back(tmp_buf, wbytes);
+ _values_null_flag.emplace_back(false);
+ break;
+ }
+ case JsonbType::T_True:
+ _backup_data.emplace_back(TRUE_VALUE);
+ _values_null_flag.emplace_back(false);
+ break;
+ case JsonbType::T_False:
+ _backup_data.emplace_back(FALSE_VALUE);
+ _values_null_flag.emplace_back(false);
+ break;
+ case JsonbType::T_Null:
+ _backup_data.emplace_back();
+ _values_null_flag.emplace_back(true);
+ break;
+ default:
+ _backup_data.emplace_back();
+ _values_null_flag.emplace_back(true);
+ break;
+ }
+ }
+ // Must set _data_string at the end, so that we can
+ // save the real addr of string in `_backup_data` to `_data_string`.
+ for (auto& str : _backup_data) {
+ _data_string_ref.emplace_back(str.data(), str.length());
+ }
+ return value_size;
+ }
};
struct ParsedDataJSON : public ParsedDataStringBase {
@@ -246,6 +371,31 @@ struct ParsedDataJSON : public ParsedDataStringBase {
}
return value_size;
}
+
+ int set_output(ArrayVal& array_doc, int value_size) override {
+ _data_string_ref.clear();
+ _backup_data.clear();
+ _values_null_flag.clear();
+ auto writer = std::make_unique<JsonbWriter>();
+ for (auto& v : array_doc) {
+ if (v.isObject()) {
+ writer->reset();
+ writer->writeValue(&v);
+ _backup_data.emplace_back(writer->getOutput()->getBuffer(),
+ writer->getOutput()->getSize());
+ _values_null_flag.emplace_back(false);
+ } else {
+ _backup_data.emplace_back();
+ _values_null_flag.emplace_back(true);
+ }
+ }
+ // Must set _data_string at the end, so that we can
+ // save the real addr of string in `_backup_data` to `_data_string`.
+ for (auto& str : _backup_data) {
+ _data_string_ref.emplace_back(str);
+ }
+ return value_size;
+ }
};
template <typename DataImpl>
@@ -267,6 +417,7 @@ private:
void _insert_values_into_column(MutableColumnPtr& column, int max_step);
DataImpl _parsed_data;
ColumnPtr _text_column;
+ DataTypePtr _text_datatype;
};
} // namespace doris::vectorized
diff --git a/be/src/vec/functions/function_fake.cpp
b/be/src/vec/functions/function_fake.cpp
index 0353b3a2a7c..c7edcf4df8f 100644
--- a/be/src/vec/functions/function_fake.cpp
+++ b/be/src/vec/functions/function_fake.cpp
@@ -38,7 +38,7 @@
namespace doris::vectorized {
-template <typename ReturnType, bool AlwaysNullable = false>
+template <typename ReturnType, bool AlwaysNullable = false, bool VARIADIC =
false>
struct FunctionFakeBaseImpl {
static DataTypePtr get_return_type_impl(const DataTypes& arguments) {
if constexpr (AlwaysNullable) {
@@ -46,6 +46,16 @@ struct FunctionFakeBaseImpl {
}
return std::make_shared<ReturnType>();
}
+ static DataTypes get_variadic_argument_types() {
+ if constexpr (VARIADIC) {
+ if constexpr (AlwaysNullable) {
+ return {make_nullable(std::make_shared<ReturnType>())};
+ }
+ return {std::make_shared<ReturnType>()};
+ } else {
+ return {};
+ }
+ }
static std::string get_error_msg() { return "Fake function do not support
execute"; }
};
@@ -55,6 +65,7 @@ struct FunctionExplode {
return make_nullable(
check_and_get_data_type<DataTypeArray>(arguments[0].get())->get_nested_type());
}
+ static DataTypes get_variadic_argument_types() { return {}; }
static std::string get_error_msg() { return "Fake function do not support
execute"; }
};
@@ -67,6 +78,7 @@ struct FunctionExplodeMap {
fieldTypes[1] =
check_and_get_data_type<DataTypeMap>(arguments[0].get())->get_value_type();
return
make_nullable(std::make_shared<vectorized::DataTypeStruct>(fieldTypes));
}
+ static DataTypes get_variadic_argument_types() { return {}; }
static std::string get_error_msg() { return "Fake function do not support
execute"; }
};
@@ -80,6 +92,7 @@ struct FunctionExplodeJsonObject {
fieldTypes[1] = make_nullable(std::make_shared<DataTypeJsonb>());
return
make_nullable(std::make_shared<vectorized::DataTypeStruct>(fieldTypes));
}
+ static DataTypes get_variadic_argument_types() { return {}; }
static std::string get_error_msg() { return "Fake function do not support
execute"; }
};
@@ -87,6 +100,7 @@ struct FunctionEsquery {
static DataTypePtr get_return_type_impl(const DataTypes& arguments) {
return
FunctionFakeBaseImpl<DataTypeUInt8>::get_return_type_impl(arguments);
}
+ static DataTypes get_variadic_argument_types() { return {}; }
static std::string get_error_msg() { return "esquery only supported on es
table"; }
};
@@ -102,11 +116,13 @@ void
register_table_function_expand(SimpleFunctionFactory& factory, const std::s
factory.register_function<FunctionFake<FunctionImpl>>(name + suffix);
};
-template <typename ReturnType>
+template <typename ReturnType, bool VARIADIC>
void register_table_function_expand_default(SimpleFunctionFactory& factory,
const std::string& name,
const std::string& suffix) {
-
factory.register_function<FunctionFake<FunctionFakeBaseImpl<ReturnType>>>(name);
- factory.register_function<FunctionFake<FunctionFakeBaseImpl<ReturnType,
true>>>(name + suffix);
+ factory.register_function<FunctionFake<FunctionFakeBaseImpl<ReturnType,
false, VARIADIC>>>(
+ name);
+ factory.register_function<FunctionFake<FunctionFakeBaseImpl<ReturnType,
true, VARIADIC>>>(
+ name + suffix);
};
template <typename FunctionImpl>
@@ -114,10 +130,11 @@ void
register_table_function_expand_outer(SimpleFunctionFactory& factory, const
register_table_function_expand<FunctionImpl>(factory, name,
COMBINATOR_SUFFIX_OUTER);
};
-template <typename ReturnType>
+template <typename ReturnType, bool VARIADIC>
void register_table_function_expand_outer_default(SimpleFunctionFactory&
factory,
const std::string& name) {
- register_table_function_expand_default<ReturnType>(factory, name,
COMBINATOR_SUFFIX_OUTER);
+ register_table_function_expand_default<ReturnType, VARIADIC>(factory, name,
+
COMBINATOR_SUFFIX_OUTER);
};
void register_function_fake(SimpleFunctionFactory& factory) {
@@ -127,16 +144,19 @@ void register_function_fake(SimpleFunctionFactory&
factory) {
register_table_function_expand_outer<FunctionExplodeMap>(factory,
"explode_map");
register_table_function_expand_outer<FunctionExplodeJsonObject>(factory,
"explode_json_object");
- register_table_function_expand_outer_default<DataTypeString>(factory,
"explode_split");
- register_table_function_expand_outer_default<DataTypeInt32>(factory,
"explode_numbers");
- register_table_function_expand_outer_default<DataTypeInt64>(factory,
"explode_json_array_int");
- register_table_function_expand_outer_default<DataTypeString>(factory,
-
"explode_json_array_string");
- register_table_function_expand_outer_default<DataTypeString>(factory,
-
"explode_json_array_json");
- register_table_function_expand_outer_default<DataTypeFloat64>(factory,
-
"explode_json_array_double");
- register_table_function_expand_outer_default<DataTypeInt64>(factory,
"explode_bitmap");
+ register_table_function_expand_outer_default<DataTypeString,
false>(factory, "explode_split");
+ register_table_function_expand_outer_default<DataTypeInt32,
false>(factory, "explode_numbers");
+ register_table_function_expand_outer_default<DataTypeInt64, false>(factory,
+
"explode_json_array_int");
+ register_table_function_expand_outer_default<DataTypeString, false>(
+ factory, "explode_json_array_string");
+ register_table_function_expand_outer_default<DataTypeJsonb, true>(factory,
+
"explode_json_array_json");
+ register_table_function_expand_outer_default<DataTypeString, true>(factory,
+
"explode_json_array_json");
+ register_table_function_expand_outer_default<DataTypeFloat64, false>(
+ factory, "explode_json_array_double");
+ register_table_function_expand_outer_default<DataTypeInt64,
false>(factory, "explode_bitmap");
}
} // namespace doris::vectorized
diff --git a/be/src/vec/functions/function_fake.h
b/be/src/vec/functions/function_fake.h
index 0dabdfb3c83..dfbfce2127d 100644
--- a/be/src/vec/functions/function_fake.h
+++ b/be/src/vec/functions/function_fake.h
@@ -35,6 +35,7 @@ class Block;
} // namespace doris
namespace doris::vectorized {
+
// FunctionFake is use for some function call expr only work at prepare/open
phase, do not support execute().
template <typename Impl>
class FunctionFake : public IFunction {
@@ -55,6 +56,10 @@ public:
bool use_default_implementation_for_nulls() const override { return true; }
+ DataTypes get_variadic_argument_types_impl() const override {
+ return Impl::get_variadic_argument_types();
+ }
+
bool use_default_implementation_for_constants() const override { return
false; }
Status execute_impl(FunctionContext* context, Block& block, const
ColumnNumbers& arguments,
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDouble.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDouble.java
index e6072931325..c24477c20d1 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDouble.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDouble.java
@@ -23,6 +23,7 @@ import
org.apache.doris.nereids.trees.expressions.functions.PropagateNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
import org.apache.doris.nereids.types.DoubleType;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -36,6 +37,7 @@ import java.util.List;
public class ExplodeJsonArrayDouble extends TableGeneratingFunction implements
UnaryExpression, PropagateNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+ FunctionSignature.ret(DoubleType.INSTANCE).args(JsonType.INSTANCE),
FunctionSignature.ret(DoubleType.INSTANCE).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDoubleOuter.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDoubleOuter.java
index 1bb8c0383f4..a2e3609c48c 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDoubleOuter.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayDoubleOuter.java
@@ -23,6 +23,7 @@ import
org.apache.doris.nereids.trees.expressions.functions.AlwaysNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
import org.apache.doris.nereids.types.DoubleType;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -36,6 +37,7 @@ import java.util.List;
public class ExplodeJsonArrayDoubleOuter extends TableGeneratingFunction
implements UnaryExpression, AlwaysNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+ FunctionSignature.ret(DoubleType.INSTANCE).args(JsonType.INSTANCE),
FunctionSignature.ret(DoubleType.INSTANCE).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayInt.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayInt.java
index 17277e1c967..86db7573375 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayInt.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayInt.java
@@ -23,6 +23,7 @@ import
org.apache.doris.nereids.trees.expressions.functions.PropagateNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
import org.apache.doris.nereids.types.BigIntType;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -36,6 +37,7 @@ import java.util.List;
public class ExplodeJsonArrayInt extends TableGeneratingFunction implements
UnaryExpression, PropagateNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+ FunctionSignature.ret(BigIntType.INSTANCE).args(JsonType.INSTANCE),
FunctionSignature.ret(BigIntType.INSTANCE).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayIntOuter.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayIntOuter.java
index 2a8820c717c..fb7360959a4 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayIntOuter.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayIntOuter.java
@@ -23,6 +23,7 @@ import
org.apache.doris.nereids.trees.expressions.functions.AlwaysNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
import org.apache.doris.nereids.types.BigIntType;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -36,6 +37,7 @@ import java.util.List;
public class ExplodeJsonArrayIntOuter extends TableGeneratingFunction
implements UnaryExpression, AlwaysNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+ FunctionSignature.ret(BigIntType.INSTANCE).args(JsonType.INSTANCE),
FunctionSignature.ret(BigIntType.INSTANCE).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJson.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJson.java
index a07e0e5d8fa..6209f23a7dd 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJson.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJson.java
@@ -22,6 +22,7 @@ import org.apache.doris.nereids.trees.expressions.Expression;
import org.apache.doris.nereids.trees.expressions.functions.PropagateNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -35,6 +36,7 @@ import java.util.List;
*/
public class ExplodeJsonArrayJson extends TableGeneratingFunction implements
UnaryExpression, PropagateNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+ FunctionSignature.ret(JsonType.INSTANCE).args(JsonType.INSTANCE),
FunctionSignature.ret(VarcharType.SYSTEM_DEFAULT).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJsonOuter.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJsonOuter.java
index bb4a34905a4..ab358855196 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJsonOuter.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayJsonOuter.java
@@ -22,6 +22,7 @@ import org.apache.doris.nereids.trees.expressions.Expression;
import org.apache.doris.nereids.trees.expressions.functions.PropagateNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -35,6 +36,7 @@ import java.util.List;
*/
public class ExplodeJsonArrayJsonOuter extends TableGeneratingFunction
implements UnaryExpression, PropagateNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+ FunctionSignature.ret(JsonType.INSTANCE).args(JsonType.INSTANCE),
FunctionSignature.ret(VarcharType.SYSTEM_DEFAULT).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayString.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayString.java
index 653cb36ca21..04717cd5c09 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayString.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayString.java
@@ -22,6 +22,7 @@ import org.apache.doris.nereids.trees.expressions.Expression;
import org.apache.doris.nereids.trees.expressions.functions.PropagateNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -35,6 +36,7 @@ import java.util.List;
public class ExplodeJsonArrayString extends TableGeneratingFunction implements
UnaryExpression, PropagateNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+
FunctionSignature.ret(VarcharType.SYSTEM_DEFAULT).args(JsonType.INSTANCE),
FunctionSignature.ret(VarcharType.SYSTEM_DEFAULT).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayStringOuter.java
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayStringOuter.java
index dfb21ab826f..03507aa9799 100644
---
a/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayStringOuter.java
+++
b/fe/fe-core/src/main/java/org/apache/doris/nereids/trees/expressions/functions/generator/ExplodeJsonArrayStringOuter.java
@@ -22,6 +22,7 @@ import org.apache.doris.nereids.trees.expressions.Expression;
import org.apache.doris.nereids.trees.expressions.functions.AlwaysNullable;
import org.apache.doris.nereids.trees.expressions.shape.UnaryExpression;
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor;
+import org.apache.doris.nereids.types.JsonType;
import org.apache.doris.nereids.types.VarcharType;
import com.google.common.base.Preconditions;
@@ -35,6 +36,7 @@ import java.util.List;
public class ExplodeJsonArrayStringOuter extends TableGeneratingFunction
implements UnaryExpression, AlwaysNullable {
public static final List<FunctionSignature> SIGNATURES = ImmutableList.of(
+
FunctionSignature.ret(VarcharType.SYSTEM_DEFAULT).args(JsonType.INSTANCE),
FunctionSignature.ret(VarcharType.SYSTEM_DEFAULT).args(VarcharType.SYSTEM_DEFAULT)
);
diff --git a/regression-test/data/nereids_function_p0/gen_function/gen.out
b/regression-test/data/nereids_function_p0/gen_function/gen.out
index 17f86d875e0..286a05ee85b 100644
--- a/regression-test/data/nereids_function_p0/gen_function/gen.out
+++ b/regression-test/data/nereids_function_p0/gen_function/gen.out
@@ -757,6 +757,47 @@
11 {"id":2,"name":"Mary"}
11 {"id":3,"name":"Bob"}
+-- !sql_explode_json_array_json_Json --
+\N {"id":1,"name":"John"}
+\N {"id":2,"name":"Mary"}
+\N {"id":3,"name":"Bob"}
+0 {"id":1,"name":"John"}
+0 {"id":2,"name":"Mary"}
+0 {"id":3,"name":"Bob"}
+1 {"id":1,"name":"John"}
+1 {"id":2,"name":"Mary"}
+1 {"id":3,"name":"Bob"}
+2 {"id":1,"name":"John"}
+2 {"id":2,"name":"Mary"}
+2 {"id":3,"name":"Bob"}
+3 {"id":1,"name":"John"}
+3 {"id":2,"name":"Mary"}
+3 {"id":3,"name":"Bob"}
+4 {"id":1,"name":"John"}
+4 {"id":2,"name":"Mary"}
+4 {"id":3,"name":"Bob"}
+5 {"id":1,"name":"John"}
+5 {"id":2,"name":"Mary"}
+5 {"id":3,"name":"Bob"}
+6 {"id":1,"name":"John"}
+6 {"id":2,"name":"Mary"}
+6 {"id":3,"name":"Bob"}
+7 {"id":1,"name":"John"}
+7 {"id":2,"name":"Mary"}
+7 {"id":3,"name":"Bob"}
+8 {"id":1,"name":"John"}
+8 {"id":2,"name":"Mary"}
+8 {"id":3,"name":"Bob"}
+9 {"id":1,"name":"John"}
+9 {"id":2,"name":"Mary"}
+9 {"id":3,"name":"Bob"}
+10 {"id":1,"name":"John"}
+10 {"id":2,"name":"Mary"}
+10 {"id":3,"name":"Bob"}
+11 {"id":1,"name":"John"}
+11 {"id":2,"name":"Mary"}
+11 {"id":3,"name":"Bob"}
+
-- !sql_explode_Double --
0 0.1
1 0.2
diff --git
a/regression-test/data/query_p0/sql_functions/table_function/explode_json_array.out
b/regression-test/data/query_p0/sql_functions/table_function/explode_json_array.out
index ccc012e1121..f75b56b3305 100644
---
a/regression-test/data/query_p0/sql_functions/table_function/explode_json_array.out
+++
b/regression-test/data/query_p0/sql_functions/table_function/explode_json_array.out
@@ -79,7 +79,21 @@
\N 80 3
\N 80 b
--- !outer_join_explode_json_array11 --
+-- !outer_join_explode_json_array111 --
+\N \N {"id":1,"name":"John"}
+\N \N {"id":2,"name":"Mary"}
+\N \N {"id":3,"name":"Bob"}
+\N 30 {"id":1,"name":"John"}
+\N 30 {"id":2,"name":"Mary"}
+\N 30 {"id":3,"name":"Bob"}
+\N 50 {"id":1,"name":"John"}
+\N 50 {"id":2,"name":"Mary"}
+\N 50 {"id":3,"name":"Bob"}
+\N 80 {"id":1,"name":"John"}
+\N 80 {"id":2,"name":"Mary"}
+\N 80 {"id":3,"name":"Bob"}
+
+-- !outer_join_explode_json_array112 --
\N \N {"id":1,"name":"John"}
\N \N {"id":2,"name":"Mary"}
\N \N {"id":3,"name":"Bob"}
diff --git a/regression-test/suites/nereids_function_p0/gen_function/gen.groovy
b/regression-test/suites/nereids_function_p0/gen_function/gen.groovy
index 547ee40a220..7f30c9a2b6a 100644
--- a/regression-test/suites/nereids_function_p0/gen_function/gen.groovy
+++ b/regression-test/suites/nereids_function_p0/gen_function/gen.groovy
@@ -62,6 +62,9 @@ suite("nereids_gen_fn") {
qt_sql_explode_json_array_json_Varchar '''
select id, e from fn_test lateral view
explode_json_array_json('[{"id":1,"name":"John"},{"id":2,"name":"Mary"},{"id":3,"name":"Bob"}]')
lv as e order by id, e'''
+ qt_sql_explode_json_array_json_Json '''
+ select id, e from fn_test lateral view
explode_json_array_json(cast('[{"id":1,"name":"John"},{"id":2,"name":"Mary"},{"id":3,"name":"Bob"}]'
as json)) lv as e order by id, cast(e as string); '''
+
// explode
order_qt_sql_explode_Double "select id, e from fn_test lateral view
explode(kadbl) lv as e order by id, e"
order_qt_sql_explode_Double_notnull "select id, e from
fn_test_not_nullable lateral view explode(kadbl) lv as e order by id, e"
diff --git
a/regression-test/suites/query_p0/sql_functions/table_function/explode_json_array.groovy
b/regression-test/suites/query_p0/sql_functions/table_function/explode_json_array.groovy
index e6ed1b62a24..edc1bc7fa1a 100644
---
a/regression-test/suites/query_p0/sql_functions/table_function/explode_json_array.groovy
+++
b/regression-test/suites/query_p0/sql_functions/table_function/explode_json_array.groovy
@@ -60,10 +60,14 @@ suite("explode_json_array") {
qt_outer_join_explode_json_array11 """SELECT id, age, e1 FROM (SELECT id,
age, e1 FROM (SELECT b.id, a.age FROM
person a LEFT JOIN person b ON
a.id=b.age)T LATERAL VIEW EXPLODE_JSON_ARRAY_STRING('[1, "b", 3]')
TMP AS e1) AS T ORDER BY age, e1"""
- qt_outer_join_explode_json_array11 """SELECT id, age, e1 FROM (SELECT id,
age, e1 FROM (SELECT b.id, a.age FROM
+ qt_outer_join_explode_json_array111 """SELECT id, age, e1 FROM (SELECT id,
age, e1 FROM (SELECT b.id, a.age FROM
person a LEFT JOIN person b ON
a.id=b.age)T LATERAL VIEW
EXPLODE_JSON_ARRAY_JSON('[{"id":1,"name":"John"},{"id":2,"name":"Mary"},{"id":3,"name":"Bob"}]')
TMP AS e1) AS T ORDER BY age, e1"""
+ qt_outer_join_explode_json_array112 """SELECT id, age, e1 FROM (SELECT id,
age, e1 FROM (SELECT b.id, a.age FROM
+ person a LEFT JOIN person b ON
a.id=b.age)T LATERAL VIEW
EXPLODE_JSON_ARRAY_JSON(cast('[{"id":1,"name":"John"},{"id":2,"name":"Mary"},{"id":3,"name":"Bob"}]'
as Json))
+ TMP AS e1) AS T ORDER BY age, cast(e1
as string)"""
+
qt_explode_json_array12 """ SELECT c_age, COUNT(1) FROM person
LATERAL VIEW
EXPLODE_JSON_ARRAY_INT('[9223372036854775807,9223372036854775808]') t1 as c_age
GROUP BY c_age ORDER BY c_age """
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
