Coverage Report

Created: 2026-05-22 14:16

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/exprs/function/function_string_digest.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 <cstddef>
19
#include <cstring>
20
#include <string_view>
21
#include <type_traits>
22
#include <vector>
23
24
#include "common/status.h"
25
#include "core/assert_cast.h"
26
#include "core/block/block.h"
27
#include "core/block/column_numbers.h"
28
#include "core/column/column_string.h"
29
#include "core/column/column_varbinary.h"
30
#include "core/column/column_vector.h"
31
#include "core/data_type/data_type_string.h"
32
#include "core/string_ref.h"
33
#include "exec/common/stringop_substring.h"
34
#include "exprs/function/function.h"
35
#include "exprs/function/simple_function_factory.h"
36
#include "exprs/function_context.h"
37
#include "util/md5.h"
38
#include "util/sha.h"
39
#include "util/sm3.h"
40
41
namespace doris {
42
#include "common/compile_check_avoid_begin.h"
43
44
struct SM3Sum {
45
    static constexpr auto name = "sm3sum";
46
    using ObjectData = SM3Digest;
47
};
48
49
struct MD5Sum {
50
    static constexpr auto name = "md5sum";
51
    using ObjectData = Md5Digest;
52
};
53
54
template <typename Impl>
55
class FunctionStringDigestMulti : public IFunction {
56
public:
57
    static constexpr auto name = Impl::name;
58
313
    static FunctionPtr create() { return std::make_shared<FunctionStringDigestMulti>(); }
_ZN5doris25FunctionStringDigestMultiINS_6SM3SumEE6createEv
Line
Count
Source
58
104
    static FunctionPtr create() { return std::make_shared<FunctionStringDigestMulti>(); }
_ZN5doris25FunctionStringDigestMultiINS_6MD5SumEE6createEv
Line
Count
Source
58
209
    static FunctionPtr create() { return std::make_shared<FunctionStringDigestMulti>(); }
59
0
    String get_name() const override { return name; }
Unexecuted instantiation: _ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE8get_nameB5cxx11Ev
Unexecuted instantiation: _ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE8get_nameB5cxx11Ev
60
0
    size_t get_number_of_arguments() const override { return 0; }
Unexecuted instantiation: _ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE23get_number_of_argumentsEv
Unexecuted instantiation: _ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE23get_number_of_argumentsEv
61
311
    bool is_variadic() const override { return true; }
_ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE11is_variadicEv
Line
Count
Source
61
103
    bool is_variadic() const override { return true; }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE11is_variadicEv
Line
Count
Source
61
208
    bool is_variadic() const override { return true; }
62
63
309
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
64
309
        return std::make_shared<DataTypeString>();
65
309
    }
_ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS7_EE
Line
Count
Source
63
102
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
64
102
        return std::make_shared<DataTypeString>();
65
102
    }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE20get_return_type_implERKSt6vectorISt10shared_ptrIKNS_9IDataTypeEESaIS7_EE
Line
Count
Source
63
207
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
64
207
        return std::make_shared<DataTypeString>();
65
207
    }
66
67
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
68
265
                        uint32_t result, size_t input_rows_count) const override {
69
265
        DCHECK_GE(arguments.size(), 1);
70
71
265
        auto res = ColumnString::create();
72
265
        auto& res_data = res->get_chars();
73
265
        auto& res_offset = res->get_offsets();
74
265
        res_offset.resize(input_rows_count);
75
76
265
        std::vector<ColumnPtr> argument_columns(arguments.size());
77
265
        std::vector<uint8_t> is_const(arguments.size(), 0);
78
809
        for (size_t i = 0; i < arguments.size(); ++i) {
79
544
            std::tie(argument_columns[i], is_const[i]) =
80
544
                    unpack_if_const(block.get_by_position(arguments[i]).column);
81
544
        }
82
83
265
        if (check_and_get_column<ColumnString>(argument_columns[0].get())) {
84
138
            vector_execute<ColumnString>(block, input_rows_count, argument_columns, is_const,
85
138
                                         res_data, res_offset);
86
138
        } else if (check_and_get_column<ColumnVarbinary>(argument_columns[0].get())) {
87
127
            vector_execute<ColumnVarbinary>(block, input_rows_count, argument_columns, is_const,
88
127
                                            res_data, res_offset);
89
127
        } else {
90
0
            return Status::RuntimeError("Illegal column {} of argument of function {}",
91
0
                                        argument_columns[0]->get_name(), get_name());
92
0
        }
93
94
265
        block.replace_by_position(result, std::move(res));
95
265
        return Status::OK();
96
265
    }
_ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
68
80
                        uint32_t result, size_t input_rows_count) const override {
69
80
        DCHECK_GE(arguments.size(), 1);
70
71
80
        auto res = ColumnString::create();
72
80
        auto& res_data = res->get_chars();
73
80
        auto& res_offset = res->get_offsets();
74
80
        res_offset.resize(input_rows_count);
75
76
80
        std::vector<ColumnPtr> argument_columns(arguments.size());
77
80
        std::vector<uint8_t> is_const(arguments.size(), 0);
78
268
        for (size_t i = 0; i < arguments.size(); ++i) {
79
188
            std::tie(argument_columns[i], is_const[i]) =
80
188
                    unpack_if_const(block.get_by_position(arguments[i]).column);
81
188
        }
82
83
80
        if (check_and_get_column<ColumnString>(argument_columns[0].get())) {
84
40
            vector_execute<ColumnString>(block, input_rows_count, argument_columns, is_const,
85
40
                                         res_data, res_offset);
86
40
        } else if (check_and_get_column<ColumnVarbinary>(argument_columns[0].get())) {
87
40
            vector_execute<ColumnVarbinary>(block, input_rows_count, argument_columns, is_const,
88
40
                                            res_data, res_offset);
89
40
        } else {
90
0
            return Status::RuntimeError("Illegal column {} of argument of function {}",
91
0
                                        argument_columns[0]->get_name(), get_name());
92
0
        }
93
94
80
        block.replace_by_position(result, std::move(res));
95
80
        return Status::OK();
96
80
    }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE12execute_implEPNS_15FunctionContextERNS_5BlockERKSt6vectorIjSaIjEEjm
Line
Count
Source
68
185
                        uint32_t result, size_t input_rows_count) const override {
69
185
        DCHECK_GE(arguments.size(), 1);
70
71
185
        auto res = ColumnString::create();
72
185
        auto& res_data = res->get_chars();
73
185
        auto& res_offset = res->get_offsets();
74
185
        res_offset.resize(input_rows_count);
75
76
185
        std::vector<ColumnPtr> argument_columns(arguments.size());
77
185
        std::vector<uint8_t> is_const(arguments.size(), 0);
78
541
        for (size_t i = 0; i < arguments.size(); ++i) {
79
356
            std::tie(argument_columns[i], is_const[i]) =
80
356
                    unpack_if_const(block.get_by_position(arguments[i]).column);
81
356
        }
82
83
185
        if (check_and_get_column<ColumnString>(argument_columns[0].get())) {
84
98
            vector_execute<ColumnString>(block, input_rows_count, argument_columns, is_const,
85
98
                                         res_data, res_offset);
86
98
        } else if (check_and_get_column<ColumnVarbinary>(argument_columns[0].get())) {
87
87
            vector_execute<ColumnVarbinary>(block, input_rows_count, argument_columns, is_const,
88
87
                                            res_data, res_offset);
89
87
        } else {
90
0
            return Status::RuntimeError("Illegal column {} of argument of function {}",
91
0
                                        argument_columns[0]->get_name(), get_name());
92
0
        }
93
94
185
        block.replace_by_position(result, std::move(res));
95
185
        return Status::OK();
96
185
    }
97
98
private:
99
    template <typename ColumnType>
100
    void vector_execute(Block& block, size_t input_rows_count,
101
                        const std::vector<ColumnPtr>& argument_columns,
102
                        const std::vector<uint8_t>& is_const, ColumnString::Chars& res_data,
103
265
                        ColumnString::Offsets& res_offset) const {
104
265
        if constexpr (std::is_same_v<Impl, MD5Sum>) {
105
185
            if (argument_columns.size() == 1) {
106
86
                const auto* col = assert_cast<const ColumnType*>(argument_columns[0].get());
107
86
                vector_execute_single_md5(col, input_rows_count, is_const[0], res_data, res_offset);
108
86
                return;
109
86
            }
110
185
        }
111
112
99
        using ObjectData = typename Impl::ObjectData;
113
487
        for (size_t i = 0; i < input_rows_count; ++i) {
114
222
            ObjectData digest;
115
768
            for (size_t j = 0; j < argument_columns.size(); ++j) {
116
546
                const auto* col = assert_cast<const ColumnType*>(argument_columns[j].get());
117
546
                StringRef data_ref = col->get_data_at(is_const[j] ? 0 : i);
118
546
                if (data_ref.size < 1) {
119
166
                    continue;
120
166
                }
121
380
                digest.update(data_ref.data, data_ref.size);
122
380
            }
123
222
            digest.digest();
124
222
            StringOP::push_value_string(std::string_view(digest.hex().c_str(), digest.hex().size()),
125
222
                                        i, res_data, res_offset);
126
222
        }
127
265
    }
_ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE14vector_executeINS_9ColumnStrIjEEEEvRNS_5BlockEmRKSt6vectorINS_3COWINS_7IColumnEE13immutable_ptrISA_EESaISD_EERKS8_IhSaIhEERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSM_IjLm4096ESP_Lm16ELm15EEE
Line
Count
Source
103
40
                        ColumnString::Offsets& res_offset) const {
104
        if constexpr (std::is_same_v<Impl, MD5Sum>) {
105
            if (argument_columns.size() == 1) {
106
                const auto* col = assert_cast<const ColumnType*>(argument_columns[0].get());
107
                vector_execute_single_md5(col, input_rows_count, is_const[0], res_data, res_offset);
108
                return;
109
            }
110
        }
111
112
40
        using ObjectData = typename Impl::ObjectData;
113
93
        for (size_t i = 0; i < input_rows_count; ++i) {
114
53
            ObjectData digest;
115
169
            for (size_t j = 0; j < argument_columns.size(); ++j) {
116
116
                const auto* col = assert_cast<const ColumnType*>(argument_columns[j].get());
117
116
                StringRef data_ref = col->get_data_at(is_const[j] ? 0 : i);
118
116
                if (data_ref.size < 1) {
119
30
                    continue;
120
30
                }
121
86
                digest.update(data_ref.data, data_ref.size);
122
86
            }
123
53
            digest.digest();
124
53
            StringOP::push_value_string(std::string_view(digest.hex().c_str(), digest.hex().size()),
125
53
                                        i, res_data, res_offset);
126
53
        }
127
40
    }
_ZNK5doris25FunctionStringDigestMultiINS_6SM3SumEE14vector_executeINS_15ColumnVarbinaryEEEvRNS_5BlockEmRKSt6vectorINS_3COWINS_7IColumnEE13immutable_ptrIS9_EESaISC_EERKS7_IhSaIhEERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSL_IjLm4096ESO_Lm16ELm15EEE
Line
Count
Source
103
40
                        ColumnString::Offsets& res_offset) const {
104
        if constexpr (std::is_same_v<Impl, MD5Sum>) {
105
            if (argument_columns.size() == 1) {
106
                const auto* col = assert_cast<const ColumnType*>(argument_columns[0].get());
107
                vector_execute_single_md5(col, input_rows_count, is_const[0], res_data, res_offset);
108
                return;
109
            }
110
        }
111
112
40
        using ObjectData = typename Impl::ObjectData;
113
93
        for (size_t i = 0; i < input_rows_count; ++i) {
114
53
            ObjectData digest;
115
169
            for (size_t j = 0; j < argument_columns.size(); ++j) {
116
116
                const auto* col = assert_cast<const ColumnType*>(argument_columns[j].get());
117
116
                StringRef data_ref = col->get_data_at(is_const[j] ? 0 : i);
118
116
                if (data_ref.size < 1) {
119
30
                    continue;
120
30
                }
121
86
                digest.update(data_ref.data, data_ref.size);
122
86
            }
123
53
            digest.digest();
124
53
            StringOP::push_value_string(std::string_view(digest.hex().c_str(), digest.hex().size()),
125
53
                                        i, res_data, res_offset);
126
53
        }
127
40
    }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE14vector_executeINS_9ColumnStrIjEEEEvRNS_5BlockEmRKSt6vectorINS_3COWINS_7IColumnEE13immutable_ptrISA_EESaISD_EERKS8_IhSaIhEERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSM_IjLm4096ESP_Lm16ELm15EEE
Line
Count
Source
103
98
                        ColumnString::Offsets& res_offset) const {
104
98
        if constexpr (std::is_same_v<Impl, MD5Sum>) {
105
98
            if (argument_columns.size() == 1) {
106
52
                const auto* col = assert_cast<const ColumnType*>(argument_columns[0].get());
107
52
                vector_execute_single_md5(col, input_rows_count, is_const[0], res_data, res_offset);
108
52
                return;
109
52
            }
110
98
        }
111
112
46
        using ObjectData = typename Impl::ObjectData;
113
152
        for (size_t i = 0; i < input_rows_count; ++i) {
114
54
            ObjectData digest;
115
203
            for (size_t j = 0; j < argument_columns.size(); ++j) {
116
149
                const auto* col = assert_cast<const ColumnType*>(argument_columns[j].get());
117
149
                StringRef data_ref = col->get_data_at(is_const[j] ? 0 : i);
118
149
                if (data_ref.size < 1) {
119
51
                    continue;
120
51
                }
121
98
                digest.update(data_ref.data, data_ref.size);
122
98
            }
123
54
            digest.digest();
124
54
            StringOP::push_value_string(std::string_view(digest.hex().c_str(), digest.hex().size()),
125
54
                                        i, res_data, res_offset);
126
54
        }
127
98
    }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE14vector_executeINS_15ColumnVarbinaryEEEvRNS_5BlockEmRKSt6vectorINS_3COWINS_7IColumnEE13immutable_ptrIS9_EESaISC_EERKS7_IhSaIhEERNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNSL_IjLm4096ESO_Lm16ELm15EEE
Line
Count
Source
103
87
                        ColumnString::Offsets& res_offset) const {
104
87
        if constexpr (std::is_same_v<Impl, MD5Sum>) {
105
87
            if (argument_columns.size() == 1) {
106
34
                const auto* col = assert_cast<const ColumnType*>(argument_columns[0].get());
107
34
                vector_execute_single_md5(col, input_rows_count, is_const[0], res_data, res_offset);
108
34
                return;
109
34
            }
110
87
        }
111
112
53
        using ObjectData = typename Impl::ObjectData;
113
149
        for (size_t i = 0; i < input_rows_count; ++i) {
114
62
            ObjectData digest;
115
227
            for (size_t j = 0; j < argument_columns.size(); ++j) {
116
165
                const auto* col = assert_cast<const ColumnType*>(argument_columns[j].get());
117
165
                StringRef data_ref = col->get_data_at(is_const[j] ? 0 : i);
118
165
                if (data_ref.size < 1) {
119
55
                    continue;
120
55
                }
121
110
                digest.update(data_ref.data, data_ref.size);
122
110
            }
123
62
            digest.digest();
124
62
            StringOP::push_value_string(std::string_view(digest.hex().c_str(), digest.hex().size()),
125
62
                                        i, res_data, res_offset);
126
62
        }
127
87
    }
128
129
    template <typename ColumnType>
130
    void vector_execute_single_md5(const ColumnType* col, size_t input_rows_count, bool is_const,
131
                                   ColumnString::Chars& res_data,
132
86
                                   ColumnString::Offsets& res_offset) const {
133
86
        ColumnString::check_chars_length(input_rows_count * MD5_HEX_LENGTH, input_rows_count);
134
86
        res_data.resize(input_rows_count * MD5_HEX_LENGTH);
135
248
        for (size_t i = 0; i < input_rows_count; ++i) {
136
162
            res_offset[i] = (i + 1) * MD5_HEX_LENGTH;
137
162
        }
138
86
        if (input_rows_count == 0) {
139
0
            return;
140
0
        }
141
142
86
        if (is_const) {
143
0
            StringRef data_ref = col->get_data_at(0);
144
0
            const unsigned char* input = reinterpret_cast<const unsigned char*>(data_ref.data);
145
0
            size_t length = data_ref.size;
146
0
            char digest[MD5_HEX_LENGTH];
147
0
            md5_hex_batch(&input, &length, digest, 1);
148
0
            for (size_t i = 0; i < input_rows_count; ++i) {
149
0
                std::memcpy(res_data.data() + i * MD5_HEX_LENGTH, digest, MD5_HEX_LENGTH);
150
0
            }
151
0
            return;
152
0
        }
153
154
86
        std::vector<const unsigned char*> inputs(input_rows_count);
155
86
        std::vector<size_t> lengths(input_rows_count);
156
248
        for (size_t i = 0; i < input_rows_count; ++i) {
157
162
            StringRef data_ref = col->get_data_at(i);
158
162
            inputs[i] = reinterpret_cast<const unsigned char*>(data_ref.data);
159
162
            lengths[i] = data_ref.size;
160
162
        }
161
86
        md5_hex_batch(inputs.data(), lengths.data(), reinterpret_cast<char*>(res_data.data()),
162
86
                      input_rows_count);
163
86
    }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE25vector_execute_single_md5INS_9ColumnStrIjEEEEvPKT_mbRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS9_IjLm4096ESC_Lm16ELm15EEE
Line
Count
Source
132
52
                                   ColumnString::Offsets& res_offset) const {
133
52
        ColumnString::check_chars_length(input_rows_count * MD5_HEX_LENGTH, input_rows_count);
134
52
        res_data.resize(input_rows_count * MD5_HEX_LENGTH);
135
151
        for (size_t i = 0; i < input_rows_count; ++i) {
136
99
            res_offset[i] = (i + 1) * MD5_HEX_LENGTH;
137
99
        }
138
52
        if (input_rows_count == 0) {
139
0
            return;
140
0
        }
141
142
52
        if (is_const) {
143
0
            StringRef data_ref = col->get_data_at(0);
144
0
            const unsigned char* input = reinterpret_cast<const unsigned char*>(data_ref.data);
145
0
            size_t length = data_ref.size;
146
0
            char digest[MD5_HEX_LENGTH];
147
0
            md5_hex_batch(&input, &length, digest, 1);
148
0
            for (size_t i = 0; i < input_rows_count; ++i) {
149
0
                std::memcpy(res_data.data() + i * MD5_HEX_LENGTH, digest, MD5_HEX_LENGTH);
150
0
            }
151
0
            return;
152
0
        }
153
154
52
        std::vector<const unsigned char*> inputs(input_rows_count);
155
52
        std::vector<size_t> lengths(input_rows_count);
156
151
        for (size_t i = 0; i < input_rows_count; ++i) {
157
99
            StringRef data_ref = col->get_data_at(i);
158
99
            inputs[i] = reinterpret_cast<const unsigned char*>(data_ref.data);
159
99
            lengths[i] = data_ref.size;
160
99
        }
161
52
        md5_hex_batch(inputs.data(), lengths.data(), reinterpret_cast<char*>(res_data.data()),
162
52
                      input_rows_count);
163
52
    }
_ZNK5doris25FunctionStringDigestMultiINS_6MD5SumEE25vector_execute_single_md5INS_15ColumnVarbinaryEEEvPKT_mbRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Line
Count
Source
132
34
                                   ColumnString::Offsets& res_offset) const {
133
34
        ColumnString::check_chars_length(input_rows_count * MD5_HEX_LENGTH, input_rows_count);
134
34
        res_data.resize(input_rows_count * MD5_HEX_LENGTH);
135
97
        for (size_t i = 0; i < input_rows_count; ++i) {
136
63
            res_offset[i] = (i + 1) * MD5_HEX_LENGTH;
137
63
        }
138
34
        if (input_rows_count == 0) {
139
0
            return;
140
0
        }
141
142
34
        if (is_const) {
143
0
            StringRef data_ref = col->get_data_at(0);
144
0
            const unsigned char* input = reinterpret_cast<const unsigned char*>(data_ref.data);
145
0
            size_t length = data_ref.size;
146
0
            char digest[MD5_HEX_LENGTH];
147
0
            md5_hex_batch(&input, &length, digest, 1);
148
0
            for (size_t i = 0; i < input_rows_count; ++i) {
149
0
                std::memcpy(res_data.data() + i * MD5_HEX_LENGTH, digest, MD5_HEX_LENGTH);
150
0
            }
151
0
            return;
152
0
        }
153
154
34
        std::vector<const unsigned char*> inputs(input_rows_count);
155
34
        std::vector<size_t> lengths(input_rows_count);
156
97
        for (size_t i = 0; i < input_rows_count; ++i) {
157
63
            StringRef data_ref = col->get_data_at(i);
158
63
            inputs[i] = reinterpret_cast<const unsigned char*>(data_ref.data);
159
63
            lengths[i] = data_ref.size;
160
63
        }
161
34
        md5_hex_batch(inputs.data(), lengths.data(), reinterpret_cast<char*>(res_data.data()),
162
34
                      input_rows_count);
163
34
    }
164
};
165
166
class FunctionStringDigestSHA1 : public IFunction {
167
public:
168
    static constexpr auto name = "sha1";
169
14
    static FunctionPtr create() { return std::make_shared<FunctionStringDigestSHA1>(); }
170
0
    String get_name() const override { return name; }
171
0
    size_t get_number_of_arguments() const override { return 1; }
172
13
    bool is_variadic() const override { return true; }
173
174
12
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
175
12
        return std::make_shared<DataTypeString>();
176
12
    }
177
178
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
179
10
                        uint32_t result, size_t input_rows_count) const override {
180
10
        DCHECK_EQ(arguments.size(), 1);
181
10
        ColumnPtr data_col = block.get_by_position(arguments[0]).column;
182
183
10
        auto res_col = ColumnString::create();
184
10
        auto& res_data = res_col->get_chars();
185
10
        auto& res_offset = res_col->get_offsets();
186
10
        res_offset.resize(input_rows_count);
187
10
        if (const auto* str_col = check_and_get_column<ColumnString>(data_col.get())) {
188
5
            vector_execute(str_col, input_rows_count, res_data, res_offset);
189
5
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(data_col.get())) {
190
5
            vector_execute(vb_col, input_rows_count, res_data, res_offset);
191
5
        } else {
192
0
            return Status::RuntimeError("Illegal column {} of argument of function {}",
193
0
                                        data_col->get_name(), get_name());
194
0
        }
195
196
10
        block.replace_by_position(result, std::move(res_col));
197
10
        return Status::OK();
198
10
    }
199
200
private:
201
    template <typename ColumnType>
202
    void vector_execute(const ColumnType* col, size_t input_rows_count,
203
10
                        ColumnString::Chars& res_data, ColumnString::Offsets& res_offset) const {
204
10
        SHA1Digest digest;
205
28
        for (size_t i = 0; i < input_rows_count; ++i) {
206
18
            StringRef data_ref = col->get_data_at(i);
207
18
            digest.reset(data_ref.data, data_ref.size);
208
18
            std::string_view ans = digest.digest();
209
210
18
            StringOP::push_value_string(ans, i, res_data, res_offset);
211
18
        }
212
10
    }
_ZNK5doris24FunctionStringDigestSHA114vector_executeINS_9ColumnStrIjEEEEvPKT_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS7_IjLm4096ESA_Lm16ELm15EEE
Line
Count
Source
203
5
                        ColumnString::Chars& res_data, ColumnString::Offsets& res_offset) const {
204
5
        SHA1Digest digest;
205
14
        for (size_t i = 0; i < input_rows_count; ++i) {
206
9
            StringRef data_ref = col->get_data_at(i);
207
9
            digest.reset(data_ref.data, data_ref.size);
208
9
            std::string_view ans = digest.digest();
209
210
9
            StringOP::push_value_string(ans, i, res_data, res_offset);
211
9
        }
212
5
    }
_ZNK5doris24FunctionStringDigestSHA114vector_executeINS_15ColumnVarbinaryEEEvPKT_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS6_IjLm4096ES9_Lm16ELm15EEE
Line
Count
Source
203
5
                        ColumnString::Chars& res_data, ColumnString::Offsets& res_offset) const {
204
5
        SHA1Digest digest;
205
14
        for (size_t i = 0; i < input_rows_count; ++i) {
206
9
            StringRef data_ref = col->get_data_at(i);
207
9
            digest.reset(data_ref.data, data_ref.size);
208
9
            std::string_view ans = digest.digest();
209
210
9
            StringOP::push_value_string(ans, i, res_data, res_offset);
211
9
        }
212
5
    }
213
};
214
215
class FunctionStringDigestSHA2 : public IFunction {
216
public:
217
    static constexpr auto name = "sha2";
218
2
    static FunctionPtr create() { return std::make_shared<FunctionStringDigestSHA2>(); }
219
0
    String get_name() const override { return name; }
220
0
    size_t get_number_of_arguments() const override { return 2; }
221
1
    bool is_variadic() const override { return true; }
222
223
0
    DataTypePtr get_return_type_impl(const DataTypes& arguments) const override {
224
0
        return std::make_shared<DataTypeString>();
225
0
    }
226
227
    Status execute_impl(FunctionContext* context, Block& block, const ColumnNumbers& arguments,
228
0
                        uint32_t result, size_t input_rows_count) const override {
229
0
        DCHECK(!is_column_const(*block.get_by_position(arguments[0]).column));
230
231
0
        ColumnPtr data_col = block.get_by_position(arguments[0]).column;
232
233
0
        [[maybe_unused]] const auto& [right_column, right_const] =
234
0
                unpack_if_const(block.get_by_position(arguments[1]).column);
235
0
        auto digest_length = assert_cast<const ColumnInt32*>(right_column.get())->get_data()[0];
236
237
0
        auto res_col = ColumnString::create();
238
0
        auto& res_data = res_col->get_chars();
239
0
        auto& res_offset = res_col->get_offsets();
240
0
        res_offset.resize(input_rows_count);
241
242
0
        if (digest_length == 224) {
243
0
            execute_base<SHA224Digest>(data_col, input_rows_count, res_data, res_offset);
244
0
        } else if (digest_length == 256) {
245
0
            execute_base<SHA256Digest>(data_col, input_rows_count, res_data, res_offset);
246
0
        } else if (digest_length == 384) {
247
0
            execute_base<SHA384Digest>(data_col, input_rows_count, res_data, res_offset);
248
0
        } else if (digest_length == 512) {
249
0
            execute_base<SHA512Digest>(data_col, input_rows_count, res_data, res_offset);
250
0
        } else {
251
0
            return Status::InvalidArgument(
252
0
                    "sha2's digest length only support 224/256/384/512 but meet {}", digest_length);
253
0
        }
254
255
0
        block.replace_by_position(result, std::move(res_col));
256
0
        return Status::OK();
257
0
    }
258
259
private:
260
    template <typename T>
261
    void execute_base(ColumnPtr data_col, int input_rows_count, ColumnString::Chars& res_data,
262
0
                      ColumnString::Offsets& res_offset) const {
263
0
        if (const auto* str_col = check_and_get_column<ColumnString>(data_col.get())) {
264
0
            vector_execute<T>(str_col, input_rows_count, res_data, res_offset);
265
0
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(data_col.get())) {
266
0
            vector_execute<T>(vb_col, input_rows_count, res_data, res_offset);
267
0
        } else {
268
0
            throw Exception(ErrorCode::RUNTIME_ERROR,
269
0
                            "Illegal column {} of argument of function {}", data_col->get_name(),
270
0
                            get_name());
271
0
        }
272
0
    }
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA212execute_baseINS_12SHA224DigestEEEvNS_3COWINS_7IColumnEE13immutable_ptrIS4_EEiRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA212execute_baseINS_12SHA256DigestEEEvNS_3COWINS_7IColumnEE13immutable_ptrIS4_EEiRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA212execute_baseINS_12SHA384DigestEEEvNS_3COWINS_7IColumnEE13immutable_ptrIS4_EEiRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA212execute_baseINS_12SHA512DigestEEEvNS_3COWINS_7IColumnEE13immutable_ptrIS4_EEiRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
273
274
    template <typename DigestType, typename ColumnType>
275
    void vector_execute(const ColumnType* col, size_t input_rows_count,
276
0
                        ColumnString::Chars& res_data, ColumnString::Offsets& res_offset) const {
277
0
        DigestType digest;
278
0
        for (size_t i = 0; i < input_rows_count; ++i) {
279
0
            StringRef data_ref = col->get_data_at(i);
280
0
            digest.reset(data_ref.data, data_ref.size);
281
0
            std::string_view ans = digest.digest();
282
283
0
            StringOP::push_value_string(ans, i, res_data, res_offset);
284
0
        }
285
0
    }
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA224DigestENS_9ColumnStrIjEEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA224DigestENS_15ColumnVarbinaryEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS7_IjLm4096ESA_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA256DigestENS_9ColumnStrIjEEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA256DigestENS_15ColumnVarbinaryEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS7_IjLm4096ESA_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA384DigestENS_9ColumnStrIjEEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA384DigestENS_15ColumnVarbinaryEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS7_IjLm4096ESA_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA512DigestENS_9ColumnStrIjEEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS8_IjLm4096ESB_Lm16ELm15EEE
Unexecuted instantiation: _ZNK5doris24FunctionStringDigestSHA214vector_executeINS_12SHA512DigestENS_15ColumnVarbinaryEEEvPKT0_mRNS_8PODArrayIhLm4096ENS_9AllocatorILb0ELb0ELb0ENS_22DefaultMemoryAllocatorELb1EEELm16ELm15EEERNS7_IjLm4096ESA_Lm16ELm15EEE
286
};
287
288
1
void register_function_string_digest(SimpleFunctionFactory& factory) {
289
1
    factory.register_function<FunctionStringDigestMulti<SM3Sum>>();
290
1
    factory.register_function<FunctionStringDigestMulti<MD5Sum>>();
291
1
    factory.register_function<FunctionStringDigestSHA1>();
292
1
    factory.register_function<FunctionStringDigestSHA2>();
293
294
1
    factory.register_alias(FunctionStringDigestMulti<MD5Sum>::name, "md5");
295
1
    factory.register_alias(FunctionStringDigestMulti<SM3Sum>::name, "sm3");
296
1
    factory.register_alias(FunctionStringDigestSHA1::name, "sha");
297
1
}
298
299
#include "common/compile_check_avoid_end.h"
300
} // namespace doris