be/src/exprs/function/function_variant_parse.cpp
Line | Count | Source |
1 | | // Licensed to the Apache Software Foundation (ASF) under one |
2 | | // or more contributor license agreements. See the NOTICE file |
3 | | // distributed with this work for additional information |
4 | | // regarding copyright ownership. The ASF licenses this file |
5 | | // to you under the Apache License, Version 2.0 (the |
6 | | // "License"); you may not use this file except in compliance |
7 | | // with the License. You may obtain a copy of the License at |
8 | | // |
9 | | // http://www.apache.org/licenses/LICENSE-2.0 |
10 | | // |
11 | | // Unless required by applicable law or agreed to in writing, |
12 | | // software distributed under the License is distributed on an |
13 | | // "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
14 | | // KIND, either express or implied. See the License for the |
15 | | // specific language governing permissions and limitations |
16 | | // under the License. |
17 | | |
18 | | #include <cstdint> |
19 | | |
20 | | #include "common/config.h" |
21 | | #include "common/exception.h" |
22 | | #include "core/assert_cast.h" |
23 | | #include "core/column/column_nullable.h" |
24 | | #include "core/column/column_string.h" |
25 | | #include "core/column/column_variant.h" |
26 | | #include "core/column/variant_v2/column_variant_v2.h" |
27 | | #include "core/data_type/data_type_nullable.h" |
28 | | #include "core/data_type/data_type_string.h" |
29 | | #include "core/data_type/data_type_variant.h" |
30 | | #include "core/data_type/data_type_variant_v2.h" |
31 | | #include "exec/common/variant_util.h" |
32 | | #include "exprs/function/parse/variant_string_parse.h" |
33 | | #include "exprs/function/simple_function_factory.h" |
34 | | #include "util/json/json_parser.h" |
35 | | |
36 | | namespace doris { |
37 | | namespace { |
38 | | |
39 | | template <bool ERROR_TO_NULL> |
40 | | class FunctionVariantParse final : public IFunction { |
41 | | public: |
42 | | static constexpr auto name = ERROR_TO_NULL ? "try_parse_to_variant" : "parse_to_variant"; |
43 | | |
44 | 43 | static FunctionPtr create() { return std::make_shared<FunctionVariantParse>(); }function_variant_parse.cpp:_ZN5doris12_GLOBAL__N_120FunctionVariantParseILb0EE6createEv Line | Count | Source | 44 | 25 | static FunctionPtr create() { return std::make_shared<FunctionVariantParse>(); } |
function_variant_parse.cpp:_ZN5doris12_GLOBAL__N_120FunctionVariantParseILb1EE6createEv Line | Count | Source | 44 | 18 | static FunctionPtr create() { return std::make_shared<FunctionVariantParse>(); } |
|
45 | | |
46 | 2 | String get_name() const override { return name; }function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb0EE8get_nameB5cxx11Ev Line | Count | Source | 46 | 1 | String get_name() const override { return name; } |
function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb1EE8get_nameB5cxx11Ev Line | Count | Source | 46 | 1 | String get_name() const override { return name; } |
|
47 | 0 | size_t get_number_of_arguments() const override { return 1; }Unexecuted instantiation: function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb0EE23get_number_of_argumentsEv Unexecuted instantiation: function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb1EE23get_number_of_argumentsEv |
48 | 25 | bool use_default_implementation_for_nulls() const override { return false; }function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb0EE36use_default_implementation_for_nullsEv Line | Count | Source | 48 | 16 | bool use_default_implementation_for_nulls() const override { return false; } |
function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb1EE36use_default_implementation_for_nullsEv Line | Count | Source | 48 | 9 | bool use_default_implementation_for_nulls() const override { return false; } |
|
49 | 27 | bool skip_return_type_check() const override { return true; }function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb0EE22skip_return_type_checkEv Line | Count | Source | 49 | 17 | bool skip_return_type_check() const override { return true; } |
function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb1EE22skip_return_type_checkEv Line | Count | Source | 49 | 10 | bool skip_return_type_check() const override { return true; } |
|
50 | | |
51 | 0 | DataTypePtr get_return_type_impl(const DataTypes& arguments) const override { |
52 | 0 | auto result = std::make_shared<DataTypeVariant>(); |
53 | 0 | return ERROR_TO_NULL || arguments[0]->is_nullable() ? make_nullable(std::move(result)) |
54 | 0 | : std::move(result); |
55 | 0 | } Unexecuted instantiation: function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb0EE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS7_EE Unexecuted instantiation: function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb1EE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS7_EE |
56 | | |
57 | | // Keep strict/error-to-null, SQL-null, and legacy/V2 parse transitions in one auditable state machine. |
58 | | // NOLINTNEXTLINE(readability-function-size,readability-function-cognitive-complexity) |
59 | | Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments, |
60 | 25 | uint32_t result, size_t input_rows_count) const override { |
61 | 25 | const ColumnPtr source = |
62 | 25 | block.get_by_position(arguments[0]).column->convert_to_full_column_if_const(); |
63 | 25 | const ColumnString* strings = nullptr; |
64 | 25 | const NullMap* input_nulls = nullptr; |
65 | 25 | if (const auto* nullable = check_and_get_column<ColumnNullable>(source.get())) { |
66 | 3 | strings = &assert_cast<const ColumnString&>(nullable->get_nested_column()); |
67 | 3 | input_nulls = &nullable->get_null_map_data(); |
68 | 22 | } else { |
69 | 22 | strings = &assert_cast<const ColumnString&>(*source); |
70 | 22 | } |
71 | 25 | DORIS_CHECK_EQ(strings->size(), input_rows_count); |
72 | | |
73 | 25 | const bool result_is_nullable = block.get_by_position(result).type->is_nullable(); |
74 | 25 | DORIS_CHECK_EQ(result_is_nullable, ERROR_TO_NULL || input_nulls != nullptr); |
75 | 25 | auto result_nulls = ColumnUInt8::create(input_rows_count, uint8_t {0}); |
76 | | |
77 | 25 | const IDataType* result_type = remove_nullable(block.get_by_position(result).type).get(); |
78 | 25 | if (const auto* variant_type = dynamic_cast<const DataTypeVariant*>(result_type)) { |
79 | 6 | const int32_t max_subcolumns_count = variant_type->variant_max_subcolumns_count(); |
80 | 6 | const bool enable_doc_mode = variant_type->enable_doc_mode(); |
81 | 6 | auto values = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); |
82 | 6 | ParseConfig parse_config; |
83 | 6 | parse_config.check_duplicate_json_path = |
84 | 6 | config::variant_enable_duplicate_json_path_check; |
85 | 6 | parse_config.parse_to = enable_doc_mode ? ParseConfig::ParseTo::OnlyDocValueColumn |
86 | 6 | : ParseConfig::ParseTo::OnlySubcolumns; |
87 | 6 | JsonParser parser; |
88 | 6 | const StringRef empty_json("", 0); |
89 | 6 | if constexpr (!ERROR_TO_NULL) { |
90 | 11 | for (size_t row = 0; row < input_rows_count; ++row) { |
91 | 7 | const bool input_is_null = input_nulls != nullptr && (*input_nulls)[row] != 0; |
92 | 7 | try { |
93 | 7 | variant_util::parse_json_to_variant( |
94 | 7 | *values, input_is_null ? empty_json : strings->get_data_at(row), |
95 | 7 | &parser, parse_config); |
96 | 7 | } catch (const Exception& exception) { |
97 | 1 | const Status status = exception.to_status(); |
98 | 1 | if (!status.is<ErrorCode::INVALID_ARGUMENT>()) { |
99 | 0 | return status; |
100 | 0 | } |
101 | 1 | return Status::InvalidArgument( |
102 | 1 | "Parse json document failed at row {}, error: {}", row, |
103 | 1 | status.to_string()); |
104 | 1 | } |
105 | 6 | if (input_is_null) { |
106 | 1 | result_nulls->get_data()[row] = 1; |
107 | 1 | } |
108 | 6 | } |
109 | 5 | } else { |
110 | 4 | for (size_t row = 0; row < input_rows_count; ++row) { |
111 | 3 | const bool input_is_null = input_nulls != nullptr && (*input_nulls)[row] != 0; |
112 | 3 | auto one_row = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); |
113 | 3 | try { |
114 | 3 | variant_util::parse_json_to_variant( |
115 | 3 | *one_row, input_is_null ? empty_json : strings->get_data_at(row), |
116 | 3 | &parser, parse_config); |
117 | 3 | one_row->finalize(); |
118 | 3 | } catch (const Exception& exception) { |
119 | 1 | const Status status = exception.to_status(); |
120 | 1 | if (!status.is<ErrorCode::INVALID_ARGUMENT>()) { |
121 | 0 | return status; |
122 | 0 | } |
123 | 1 | one_row = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); |
124 | 1 | variant_util::parse_json_to_variant(*one_row, empty_json, &parser, |
125 | 1 | parse_config); |
126 | 1 | one_row->finalize(); |
127 | 1 | result_nulls->get_data()[row] = 1; |
128 | 1 | } |
129 | 3 | if (input_is_null) { |
130 | 0 | result_nulls->get_data()[row] = 1; |
131 | 0 | } |
132 | 3 | values->insert_range_from(*one_row, 0, 1); |
133 | 3 | } |
134 | 1 | } |
135 | 5 | values->finalize(); |
136 | | |
137 | 6 | ColumnPtr output; |
138 | 6 | if (result_is_nullable) { |
139 | 2 | output = ColumnNullable::create(std::move(values), std::move(result_nulls)); |
140 | 4 | } else { |
141 | 4 | output = std::move(values); |
142 | 4 | } |
143 | 6 | DORIS_CHECK_EQ(output->size(), input_rows_count); |
144 | 6 | block.replace_by_position(result, std::move(output)); |
145 | 6 | return Status::OK(); |
146 | 6 | } |
147 | 19 | DORIS_CHECK(dynamic_cast<const DataTypeVariantV2*>(result_type) != nullptr); |
148 | | |
149 | 19 | JsonStringToVariantEncoder encoder(JsonToVariantOptions::current_config()); |
150 | 19 | const StringRef null_json("null", 4); |
151 | 47 | for (size_t row = 0; row < input_rows_count; ++row) { |
152 | 28 | if (input_nulls != nullptr && (*input_nulls)[row] != 0) { |
153 | 2 | encoder.add_json(null_json); |
154 | 2 | result_nulls->get_data()[row] = 1; |
155 | 2 | continue; |
156 | 2 | } |
157 | | |
158 | 26 | const Status status = encoder.try_add_json(strings->get_data_at(row)); |
159 | 26 | if (status.ok()) { |
160 | 16 | continue; |
161 | 16 | } |
162 | 10 | if constexpr (!ERROR_TO_NULL) { |
163 | 5 | return Status::InvalidArgument("Parse json document failed at row {}, error: {}", |
164 | 5 | row, status.to_string()); |
165 | 5 | } |
166 | 0 | encoder.add_json(null_json); |
167 | 10 | result_nulls->get_data()[row] = 1; |
168 | 10 | } |
169 | | |
170 | 19 | VariantBatchBuilder encoded = encoder.finish_batch(); |
171 | 19 | auto values = ColumnVariantV2::create(); |
172 | 19 | values->insert_encoded_batch(encoded); |
173 | | |
174 | 19 | ColumnPtr output; |
175 | 19 | if (result_is_nullable) { |
176 | 10 | output = ColumnNullable::create(std::move(values), std::move(result_nulls)); |
177 | 10 | } else { |
178 | 9 | output = std::move(values); |
179 | 9 | } |
180 | 19 | DORIS_CHECK_EQ(output->size(), input_rows_count); |
181 | 19 | block.replace_by_position(result, std::move(output)); |
182 | 19 | return Status::OK(); |
183 | 25 | } function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb0EE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm Line | Count | Source | 60 | 16 | uint32_t result, size_t input_rows_count) const override { | 61 | 16 | const ColumnPtr source = | 62 | 16 | block.get_by_position(arguments[0]).column->convert_to_full_column_if_const(); | 63 | 16 | const ColumnString* strings = nullptr; | 64 | 16 | const NullMap* input_nulls = nullptr; | 65 | 16 | if (const auto* nullable = check_and_get_column<ColumnNullable>(source.get())) { | 66 | 3 | strings = &assert_cast<const ColumnString&>(nullable->get_nested_column()); | 67 | 3 | input_nulls = &nullable->get_null_map_data(); | 68 | 13 | } else { | 69 | 13 | strings = &assert_cast<const ColumnString&>(*source); | 70 | 13 | } | 71 | 16 | DORIS_CHECK_EQ(strings->size(), input_rows_count); | 72 | | | 73 | 16 | const bool result_is_nullable = block.get_by_position(result).type->is_nullable(); | 74 | 16 | DORIS_CHECK_EQ(result_is_nullable, ERROR_TO_NULL || input_nulls != nullptr); | 75 | 16 | auto result_nulls = ColumnUInt8::create(input_rows_count, uint8_t {0}); | 76 | | | 77 | 16 | const IDataType* result_type = remove_nullable(block.get_by_position(result).type).get(); | 78 | 16 | if (const auto* variant_type = dynamic_cast<const DataTypeVariant*>(result_type)) { | 79 | 5 | const int32_t max_subcolumns_count = variant_type->variant_max_subcolumns_count(); | 80 | 5 | const bool enable_doc_mode = variant_type->enable_doc_mode(); | 81 | 5 | auto values = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); | 82 | 5 | ParseConfig parse_config; | 83 | 5 | parse_config.check_duplicate_json_path = | 84 | 5 | config::variant_enable_duplicate_json_path_check; | 85 | 5 | parse_config.parse_to = enable_doc_mode ? ParseConfig::ParseTo::OnlyDocValueColumn | 86 | 5 | : ParseConfig::ParseTo::OnlySubcolumns; | 87 | 5 | JsonParser parser; | 88 | 5 | const StringRef empty_json("", 0); | 89 | 5 | if constexpr (!ERROR_TO_NULL) { | 90 | 11 | for (size_t row = 0; row < input_rows_count; ++row) { | 91 | 7 | const bool input_is_null = input_nulls != nullptr && (*input_nulls)[row] != 0; | 92 | 7 | try { | 93 | 7 | variant_util::parse_json_to_variant( | 94 | 7 | *values, input_is_null ? empty_json : strings->get_data_at(row), | 95 | 7 | &parser, parse_config); | 96 | 7 | } catch (const Exception& exception) { | 97 | 1 | const Status status = exception.to_status(); | 98 | 1 | if (!status.is<ErrorCode::INVALID_ARGUMENT>()) { | 99 | 0 | return status; | 100 | 0 | } | 101 | 1 | return Status::InvalidArgument( | 102 | 1 | "Parse json document failed at row {}, error: {}", row, | 103 | 1 | status.to_string()); | 104 | 1 | } | 105 | 6 | if (input_is_null) { | 106 | 1 | result_nulls->get_data()[row] = 1; | 107 | 1 | } | 108 | 6 | } | 109 | | } else { | 110 | | for (size_t row = 0; row < input_rows_count; ++row) { | 111 | | const bool input_is_null = input_nulls != nullptr && (*input_nulls)[row] != 0; | 112 | | auto one_row = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); | 113 | | try { | 114 | | variant_util::parse_json_to_variant( | 115 | | *one_row, input_is_null ? empty_json : strings->get_data_at(row), | 116 | | &parser, parse_config); | 117 | | one_row->finalize(); | 118 | | } catch (const Exception& exception) { | 119 | | const Status status = exception.to_status(); | 120 | | if (!status.is<ErrorCode::INVALID_ARGUMENT>()) { | 121 | | return status; | 122 | | } | 123 | | one_row = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); | 124 | | variant_util::parse_json_to_variant(*one_row, empty_json, &parser, | 125 | | parse_config); | 126 | | one_row->finalize(); | 127 | | result_nulls->get_data()[row] = 1; | 128 | | } | 129 | | if (input_is_null) { | 130 | | result_nulls->get_data()[row] = 1; | 131 | | } | 132 | | values->insert_range_from(*one_row, 0, 1); | 133 | | } | 134 | | } | 135 | 4 | values->finalize(); | 136 | | | 137 | 5 | ColumnPtr output; | 138 | 5 | if (result_is_nullable) { | 139 | 1 | output = ColumnNullable::create(std::move(values), std::move(result_nulls)); | 140 | 4 | } else { | 141 | 4 | output = std::move(values); | 142 | 4 | } | 143 | 5 | DORIS_CHECK_EQ(output->size(), input_rows_count); | 144 | 5 | block.replace_by_position(result, std::move(output)); | 145 | 5 | return Status::OK(); | 146 | 5 | } | 147 | 11 | DORIS_CHECK(dynamic_cast<const DataTypeVariantV2*>(result_type) != nullptr); | 148 | | | 149 | 11 | JsonStringToVariantEncoder encoder(JsonToVariantOptions::current_config()); | 150 | 11 | const StringRef null_json("null", 4); | 151 | 28 | for (size_t row = 0; row < input_rows_count; ++row) { | 152 | 17 | if (input_nulls != nullptr && (*input_nulls)[row] != 0) { | 153 | 2 | encoder.add_json(null_json); | 154 | 2 | result_nulls->get_data()[row] = 1; | 155 | 2 | continue; | 156 | 2 | } | 157 | | | 158 | 15 | const Status status = encoder.try_add_json(strings->get_data_at(row)); | 159 | 15 | if (status.ok()) { | 160 | 10 | continue; | 161 | 10 | } | 162 | 5 | if constexpr (!ERROR_TO_NULL) { | 163 | 5 | return Status::InvalidArgument("Parse json document failed at row {}, error: {}", | 164 | 5 | row, status.to_string()); | 165 | 5 | } | 166 | 0 | encoder.add_json(null_json); | 167 | 5 | result_nulls->get_data()[row] = 1; | 168 | 5 | } | 169 | | | 170 | 11 | VariantBatchBuilder encoded = encoder.finish_batch(); | 171 | 11 | auto values = ColumnVariantV2::create(); | 172 | 11 | values->insert_encoded_batch(encoded); | 173 | | | 174 | 11 | ColumnPtr output; | 175 | 11 | if (result_is_nullable) { | 176 | 2 | output = ColumnNullable::create(std::move(values), std::move(result_nulls)); | 177 | 9 | } else { | 178 | 9 | output = std::move(values); | 179 | 9 | } | 180 | | DORIS_CHECK_EQ(output->size(), input_rows_count); | 181 | 11 | block.replace_by_position(result, std::move(output)); | 182 | 11 | return Status::OK(); | 183 | 16 | } |
function_variant_parse.cpp:_ZNK5doris12_GLOBAL__N_120FunctionVariantParseILb1EE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm Line | Count | Source | 60 | 9 | uint32_t result, size_t input_rows_count) const override { | 61 | 9 | const ColumnPtr source = | 62 | 9 | block.get_by_position(arguments[0]).column->convert_to_full_column_if_const(); | 63 | 9 | const ColumnString* strings = nullptr; | 64 | 9 | const NullMap* input_nulls = nullptr; | 65 | 9 | if (const auto* nullable = check_and_get_column<ColumnNullable>(source.get())) { | 66 | 0 | strings = &assert_cast<const ColumnString&>(nullable->get_nested_column()); | 67 | 0 | input_nulls = &nullable->get_null_map_data(); | 68 | 9 | } else { | 69 | 9 | strings = &assert_cast<const ColumnString&>(*source); | 70 | 9 | } | 71 | 9 | DORIS_CHECK_EQ(strings->size(), input_rows_count); | 72 | | | 73 | 9 | const bool result_is_nullable = block.get_by_position(result).type->is_nullable(); | 74 | 9 | DORIS_CHECK_EQ(result_is_nullable, ERROR_TO_NULL || input_nulls != nullptr); | 75 | 9 | auto result_nulls = ColumnUInt8::create(input_rows_count, uint8_t {0}); | 76 | | | 77 | 9 | const IDataType* result_type = remove_nullable(block.get_by_position(result).type).get(); | 78 | 9 | if (const auto* variant_type = dynamic_cast<const DataTypeVariant*>(result_type)) { | 79 | 1 | const int32_t max_subcolumns_count = variant_type->variant_max_subcolumns_count(); | 80 | 1 | const bool enable_doc_mode = variant_type->enable_doc_mode(); | 81 | 1 | auto values = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); | 82 | 1 | ParseConfig parse_config; | 83 | 1 | parse_config.check_duplicate_json_path = | 84 | 1 | config::variant_enable_duplicate_json_path_check; | 85 | 1 | parse_config.parse_to = enable_doc_mode ? ParseConfig::ParseTo::OnlyDocValueColumn | 86 | 1 | : ParseConfig::ParseTo::OnlySubcolumns; | 87 | 1 | JsonParser parser; | 88 | 1 | const StringRef empty_json("", 0); | 89 | | if constexpr (!ERROR_TO_NULL) { | 90 | | for (size_t row = 0; row < input_rows_count; ++row) { | 91 | | const bool input_is_null = input_nulls != nullptr && (*input_nulls)[row] != 0; | 92 | | try { | 93 | | variant_util::parse_json_to_variant( | 94 | | *values, input_is_null ? empty_json : strings->get_data_at(row), | 95 | | &parser, parse_config); | 96 | | } catch (const Exception& exception) { | 97 | | const Status status = exception.to_status(); | 98 | | if (!status.is<ErrorCode::INVALID_ARGUMENT>()) { | 99 | | return status; | 100 | | } | 101 | | return Status::InvalidArgument( | 102 | | "Parse json document failed at row {}, error: {}", row, | 103 | | status.to_string()); | 104 | | } | 105 | | if (input_is_null) { | 106 | | result_nulls->get_data()[row] = 1; | 107 | | } | 108 | | } | 109 | 1 | } else { | 110 | 4 | for (size_t row = 0; row < input_rows_count; ++row) { | 111 | 3 | const bool input_is_null = input_nulls != nullptr && (*input_nulls)[row] != 0; | 112 | 3 | auto one_row = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); | 113 | 3 | try { | 114 | 3 | variant_util::parse_json_to_variant( | 115 | 3 | *one_row, input_is_null ? empty_json : strings->get_data_at(row), | 116 | 3 | &parser, parse_config); | 117 | 3 | one_row->finalize(); | 118 | 3 | } catch (const Exception& exception) { | 119 | 1 | const Status status = exception.to_status(); | 120 | 1 | if (!status.is<ErrorCode::INVALID_ARGUMENT>()) { | 121 | 0 | return status; | 122 | 0 | } | 123 | 1 | one_row = ColumnVariant::create(max_subcolumns_count, enable_doc_mode); | 124 | 1 | variant_util::parse_json_to_variant(*one_row, empty_json, &parser, | 125 | 1 | parse_config); | 126 | 1 | one_row->finalize(); | 127 | 1 | result_nulls->get_data()[row] = 1; | 128 | 1 | } | 129 | 3 | if (input_is_null) { | 130 | 0 | result_nulls->get_data()[row] = 1; | 131 | 0 | } | 132 | 3 | values->insert_range_from(*one_row, 0, 1); | 133 | 3 | } | 134 | 1 | } | 135 | 1 | values->finalize(); | 136 | | | 137 | 1 | ColumnPtr output; | 138 | 1 | if (result_is_nullable) { | 139 | 1 | output = ColumnNullable::create(std::move(values), std::move(result_nulls)); | 140 | 1 | } else { | 141 | 0 | output = std::move(values); | 142 | 0 | } | 143 | 1 | DORIS_CHECK_EQ(output->size(), input_rows_count); | 144 | 1 | block.replace_by_position(result, std::move(output)); | 145 | 1 | return Status::OK(); | 146 | 1 | } | 147 | 8 | DORIS_CHECK(dynamic_cast<const DataTypeVariantV2*>(result_type) != nullptr); | 148 | | | 149 | 8 | JsonStringToVariantEncoder encoder(JsonToVariantOptions::current_config()); | 150 | 8 | const StringRef null_json("null", 4); | 151 | 19 | for (size_t row = 0; row < input_rows_count; ++row) { | 152 | 11 | if (input_nulls != nullptr && (*input_nulls)[row] != 0) { | 153 | 0 | encoder.add_json(null_json); | 154 | 0 | result_nulls->get_data()[row] = 1; | 155 | 0 | continue; | 156 | 0 | } | 157 | | | 158 | 11 | const Status status = encoder.try_add_json(strings->get_data_at(row)); | 159 | 11 | if (status.ok()) { | 160 | 6 | continue; | 161 | 6 | } | 162 | | if constexpr (!ERROR_TO_NULL) { | 163 | | return Status::InvalidArgument("Parse json document failed at row {}, error: {}", | 164 | | row, status.to_string()); | 165 | | } | 166 | 5 | encoder.add_json(null_json); | 167 | 5 | result_nulls->get_data()[row] = 1; | 168 | 5 | } | 169 | | | 170 | 8 | VariantBatchBuilder encoded = encoder.finish_batch(); | 171 | 8 | auto values = ColumnVariantV2::create(); | 172 | 8 | values->insert_encoded_batch(encoded); | 173 | | | 174 | 8 | ColumnPtr output; | 175 | 8 | if (result_is_nullable) { | 176 | 8 | output = ColumnNullable::create(std::move(values), std::move(result_nulls)); | 177 | 8 | } else { | 178 | 0 | output = std::move(values); | 179 | 0 | } | 180 | | DORIS_CHECK_EQ(output->size(), input_rows_count); | 181 | 8 | block.replace_by_position(result, std::move(output)); | 182 | 8 | return Status::OK(); | 183 | 9 | } |
|
184 | | }; |
185 | | |
186 | | using FunctionParseToVariant = FunctionVariantParse<false>; |
187 | | using FunctionTryParseToVariant = FunctionVariantParse<true>; |
188 | | |
189 | | } // namespace |
190 | | |
191 | 7 | void register_function_variant_parse(SimpleFunctionFactory& factory) { |
192 | 7 | factory.register_function<FunctionParseToVariant>(); |
193 | 7 | factory.register_function<FunctionTryParseToVariant>(); |
194 | 7 | } |
195 | | |
196 | | } // namespace doris |