Coverage Report

Created: 2026-04-01 03:37

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
be/src/exprs/function/function_hash.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
// This file is copied from
18
// https://github.com/ClickHouse/ClickHouse/blob/master/src/Functions/FunctionHash.cpp
19
// and modified by Doris
20
21
#include "exprs/function/function_hash.h"
22
23
#include "common/status.h"
24
#include "core/assert_cast.h"
25
#include "core/column/column.h"
26
#include "core/column/column_const.h"
27
#include "core/column/column_string.h"
28
#include "core/column/column_varbinary.h"
29
#include "core/column/column_vector.h"
30
#include "core/data_type/data_type.h"
31
#include "core/data_type/data_type_number.h"
32
#include "core/field.h"
33
#include "exec/common/template_helpers.hpp"
34
#include "exprs/function/function_helpers.h"
35
#include "exprs/function/function_variadic_arguments.h"
36
#include "exprs/function/simple_function_factory.h"
37
#include "util/hash/murmur_hash3.h"
38
#include "util/hash_util.hpp"
39
40
namespace doris {
41
#include "common/compile_check_begin.h"
42
constexpr uint64_t emtpy_value = 0xe28dbde7fe22e41c;
43
44
template <PrimitiveType ReturnType, bool is_mmh64_v2 = false>
45
struct MurmurHash3Impl {
46
3
    static constexpr auto get_name() {
47
3
        if constexpr (ReturnType == TYPE_INT) {
48
1
            return "murmur_hash3_32";
49
        } else if constexpr (ReturnType == TYPE_LARGEINT) {
50
            return "murmur_hash3_u64_v2";
51
1
        } else if constexpr (is_mmh64_v2) {
52
1
            return "murmur_hash3_64_v2";
53
1
        } else {
54
1
            return "murmur_hash3_64";
55
1
        }
56
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE5ELb0EE8get_nameEv
Line
Count
Source
46
1
    static constexpr auto get_name() {
47
1
        if constexpr (ReturnType == TYPE_INT) {
48
1
            return "murmur_hash3_32";
49
        } else if constexpr (ReturnType == TYPE_LARGEINT) {
50
            return "murmur_hash3_u64_v2";
51
        } else if constexpr (is_mmh64_v2) {
52
            return "murmur_hash3_64_v2";
53
        } else {
54
            return "murmur_hash3_64";
55
        }
56
1
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb0EE8get_nameEv
Line
Count
Source
46
1
    static constexpr auto get_name() {
47
        if constexpr (ReturnType == TYPE_INT) {
48
            return "murmur_hash3_32";
49
        } else if constexpr (ReturnType == TYPE_LARGEINT) {
50
            return "murmur_hash3_u64_v2";
51
        } else if constexpr (is_mmh64_v2) {
52
            return "murmur_hash3_64_v2";
53
1
        } else {
54
1
            return "murmur_hash3_64";
55
1
        }
56
1
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb1EE8get_nameEv
Line
Count
Source
46
1
    static constexpr auto get_name() {
47
        if constexpr (ReturnType == TYPE_INT) {
48
            return "murmur_hash3_32";
49
        } else if constexpr (ReturnType == TYPE_LARGEINT) {
50
            return "murmur_hash3_u64_v2";
51
1
        } else if constexpr (is_mmh64_v2) {
52
1
            return "murmur_hash3_64_v2";
53
        } else {
54
            return "murmur_hash3_64";
55
        }
56
1
    }
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE7ELb1EE8get_nameEv
57
    static constexpr auto name = get_name();
58
59
0
    static Status empty_apply(IColumn& icolumn, size_t input_rows_count) {
60
0
        ColumnVector<ReturnType>& vec_to = assert_cast<ColumnVector<ReturnType>&>(icolumn);
61
0
        vec_to.get_data().assign(
62
0
                input_rows_count,
63
0
                static_cast<typename PrimitiveTypeTraits<ReturnType>::CppType>(emtpy_value));
64
0
        return Status::OK();
65
0
    }
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE5ELb0EE11empty_applyERNS_7IColumnEm
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb0EE11empty_applyERNS_7IColumnEm
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb1EE11empty_applyERNS_7IColumnEm
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE7ELb1EE11empty_applyERNS_7IColumnEm
66
67
    static Status first_apply(const IDataType* type, const IColumn* column, size_t input_rows_count,
68
7
                              IColumn& icolumn) {
69
7
        return execute<true>(type, column, input_rows_count, icolumn);
70
7
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE5ELb0EE11first_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
68
3
                              IColumn& icolumn) {
69
3
        return execute<true>(type, column, input_rows_count, icolumn);
70
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb0EE11first_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
68
3
                              IColumn& icolumn) {
69
3
        return execute<true>(type, column, input_rows_count, icolumn);
70
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb1EE11first_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
68
1
                              IColumn& icolumn) {
69
1
        return execute<true>(type, column, input_rows_count, icolumn);
70
1
    }
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE7ELb1EE11first_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
71
72
    static Status combine_apply(const IDataType* type, const IColumn* column,
73
6
                                size_t input_rows_count, IColumn& icolumn) {
74
6
        return execute<false>(type, column, input_rows_count, icolumn);
75
6
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE5ELb0EE13combine_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
73
3
                                size_t input_rows_count, IColumn& icolumn) {
74
3
        return execute<false>(type, column, input_rows_count, icolumn);
75
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb0EE13combine_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
73
3
                                size_t input_rows_count, IColumn& icolumn) {
74
3
        return execute<false>(type, column, input_rows_count, icolumn);
75
3
    }
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb1EE13combine_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE7ELb1EE13combine_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
76
77
    template <bool first>
78
    static Status execute(const IDataType* type, const IColumn* column, size_t input_rows_count,
79
13
                          IColumn& col_to) {
80
13
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
81
13
        if constexpr (first) {
82
7
            if constexpr (ReturnType == TYPE_INT) {
83
3
                to_column.insert_many_vals(static_cast<Int32>(HashUtil::MURMUR3_32_SEED),
84
3
                                           input_rows_count);
85
4
            } else {
86
4
                to_column.insert_many_defaults(input_rows_count);
87
4
            }
88
7
        }
89
13
        auto& col_to_data = to_column.get_data();
90
13
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
91
13
            const typename ColumnString::Chars& data = col_from->get_chars();
92
13
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
93
13
            size_t size = offsets.size();
94
13
            ColumnString::Offset current_offset = 0;
95
40
            for (size_t i = 0; i < size; ++i) {
96
27
                if constexpr (ReturnType == TYPE_INT) {
97
12
                    col_to_data[i] = HashUtil::murmur_hash3_32(
98
12
                            reinterpret_cast<const char*>(&data[current_offset]),
99
12
                            offsets[i] - current_offset, col_to_data[i]);
100
15
                } else {
101
15
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
102
15
                            reinterpret_cast<const char*>(&data[current_offset]),
103
15
                            offsets[i] - current_offset, static_cast<uint64_t>(col_to_data[i]));
104
15
                }
105
27
                current_offset = offsets[i];
106
27
            }
107
13
        } else if (const ColumnConst* col_from_const =
108
0
                           check_and_get_column_const_string_or_fixedstring(column)) {
109
0
            auto value = col_from_const->get_value<TYPE_STRING>();
110
0
            for (size_t i = 0; i < input_rows_count; ++i) {
111
0
                if constexpr (ReturnType == TYPE_INT) {
112
0
                    col_to_data[i] =
113
0
                            HashUtil::murmur_hash3_32(value.data(), value.size(), col_to_data[i]);
114
0
                } else {
115
0
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
116
0
                            value.data(), value.size(), static_cast<uint64_t>(col_to_data[i]));
117
0
                }
118
0
            }
119
0
        } else {
120
0
            DCHECK(false);
121
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
122
0
                                        column->get_name(), name);
123
0
        }
124
13
        return Status::OK();
125
13
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE5ELb0EE7executeILb1EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
79
3
                          IColumn& col_to) {
80
3
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
81
3
        if constexpr (first) {
82
3
            if constexpr (ReturnType == TYPE_INT) {
83
3
                to_column.insert_many_vals(static_cast<Int32>(HashUtil::MURMUR3_32_SEED),
84
3
                                           input_rows_count);
85
            } else {
86
                to_column.insert_many_defaults(input_rows_count);
87
            }
88
3
        }
89
3
        auto& col_to_data = to_column.get_data();
90
3
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
91
3
            const typename ColumnString::Chars& data = col_from->get_chars();
92
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
93
3
            size_t size = offsets.size();
94
3
            ColumnString::Offset current_offset = 0;
95
9
            for (size_t i = 0; i < size; ++i) {
96
6
                if constexpr (ReturnType == TYPE_INT) {
97
6
                    col_to_data[i] = HashUtil::murmur_hash3_32(
98
6
                            reinterpret_cast<const char*>(&data[current_offset]),
99
6
                            offsets[i] - current_offset, col_to_data[i]);
100
                } else {
101
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
102
                            reinterpret_cast<const char*>(&data[current_offset]),
103
                            offsets[i] - current_offset, static_cast<uint64_t>(col_to_data[i]));
104
                }
105
6
                current_offset = offsets[i];
106
6
            }
107
3
        } else if (const ColumnConst* col_from_const =
108
0
                           check_and_get_column_const_string_or_fixedstring(column)) {
109
0
            auto value = col_from_const->get_value<TYPE_STRING>();
110
0
            for (size_t i = 0; i < input_rows_count; ++i) {
111
0
                if constexpr (ReturnType == TYPE_INT) {
112
0
                    col_to_data[i] =
113
0
                            HashUtil::murmur_hash3_32(value.data(), value.size(), col_to_data[i]);
114
                } else {
115
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
116
                            value.data(), value.size(), static_cast<uint64_t>(col_to_data[i]));
117
                }
118
0
            }
119
0
        } else {
120
0
            DCHECK(false);
121
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
122
0
                                        column->get_name(), name);
123
0
        }
124
3
        return Status::OK();
125
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE5ELb0EE7executeILb0EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
79
3
                          IColumn& col_to) {
80
3
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
81
        if constexpr (first) {
82
            if constexpr (ReturnType == TYPE_INT) {
83
                to_column.insert_many_vals(static_cast<Int32>(HashUtil::MURMUR3_32_SEED),
84
                                           input_rows_count);
85
            } else {
86
                to_column.insert_many_defaults(input_rows_count);
87
            }
88
        }
89
3
        auto& col_to_data = to_column.get_data();
90
3
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
91
3
            const typename ColumnString::Chars& data = col_from->get_chars();
92
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
93
3
            size_t size = offsets.size();
94
3
            ColumnString::Offset current_offset = 0;
95
9
            for (size_t i = 0; i < size; ++i) {
96
6
                if constexpr (ReturnType == TYPE_INT) {
97
6
                    col_to_data[i] = HashUtil::murmur_hash3_32(
98
6
                            reinterpret_cast<const char*>(&data[current_offset]),
99
6
                            offsets[i] - current_offset, col_to_data[i]);
100
                } else {
101
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
102
                            reinterpret_cast<const char*>(&data[current_offset]),
103
                            offsets[i] - current_offset, static_cast<uint64_t>(col_to_data[i]));
104
                }
105
6
                current_offset = offsets[i];
106
6
            }
107
3
        } else if (const ColumnConst* col_from_const =
108
0
                           check_and_get_column_const_string_or_fixedstring(column)) {
109
0
            auto value = col_from_const->get_value<TYPE_STRING>();
110
0
            for (size_t i = 0; i < input_rows_count; ++i) {
111
0
                if constexpr (ReturnType == TYPE_INT) {
112
0
                    col_to_data[i] =
113
0
                            HashUtil::murmur_hash3_32(value.data(), value.size(), col_to_data[i]);
114
                } else {
115
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
116
                            value.data(), value.size(), static_cast<uint64_t>(col_to_data[i]));
117
                }
118
0
            }
119
0
        } else {
120
0
            DCHECK(false);
121
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
122
0
                                        column->get_name(), name);
123
0
        }
124
3
        return Status::OK();
125
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb0EE7executeILb1EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
79
3
                          IColumn& col_to) {
80
3
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
81
3
        if constexpr (first) {
82
            if constexpr (ReturnType == TYPE_INT) {
83
                to_column.insert_many_vals(static_cast<Int32>(HashUtil::MURMUR3_32_SEED),
84
                                           input_rows_count);
85
3
            } else {
86
3
                to_column.insert_many_defaults(input_rows_count);
87
3
            }
88
3
        }
89
3
        auto& col_to_data = to_column.get_data();
90
3
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
91
3
            const typename ColumnString::Chars& data = col_from->get_chars();
92
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
93
3
            size_t size = offsets.size();
94
3
            ColumnString::Offset current_offset = 0;
95
9
            for (size_t i = 0; i < size; ++i) {
96
                if constexpr (ReturnType == TYPE_INT) {
97
                    col_to_data[i] = HashUtil::murmur_hash3_32(
98
                            reinterpret_cast<const char*>(&data[current_offset]),
99
                            offsets[i] - current_offset, col_to_data[i]);
100
6
                } else {
101
6
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
102
6
                            reinterpret_cast<const char*>(&data[current_offset]),
103
6
                            offsets[i] - current_offset, static_cast<uint64_t>(col_to_data[i]));
104
6
                }
105
6
                current_offset = offsets[i];
106
6
            }
107
3
        } else if (const ColumnConst* col_from_const =
108
0
                           check_and_get_column_const_string_or_fixedstring(column)) {
109
0
            auto value = col_from_const->get_value<TYPE_STRING>();
110
0
            for (size_t i = 0; i < input_rows_count; ++i) {
111
                if constexpr (ReturnType == TYPE_INT) {
112
                    col_to_data[i] =
113
                            HashUtil::murmur_hash3_32(value.data(), value.size(), col_to_data[i]);
114
0
                } else {
115
0
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
116
0
                            value.data(), value.size(), static_cast<uint64_t>(col_to_data[i]));
117
0
                }
118
0
            }
119
0
        } else {
120
0
            DCHECK(false);
121
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
122
0
                                        column->get_name(), name);
123
0
        }
124
3
        return Status::OK();
125
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb0EE7executeILb0EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
79
3
                          IColumn& col_to) {
80
3
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
81
        if constexpr (first) {
82
            if constexpr (ReturnType == TYPE_INT) {
83
                to_column.insert_many_vals(static_cast<Int32>(HashUtil::MURMUR3_32_SEED),
84
                                           input_rows_count);
85
            } else {
86
                to_column.insert_many_defaults(input_rows_count);
87
            }
88
        }
89
3
        auto& col_to_data = to_column.get_data();
90
3
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
91
3
            const typename ColumnString::Chars& data = col_from->get_chars();
92
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
93
3
            size_t size = offsets.size();
94
3
            ColumnString::Offset current_offset = 0;
95
9
            for (size_t i = 0; i < size; ++i) {
96
                if constexpr (ReturnType == TYPE_INT) {
97
                    col_to_data[i] = HashUtil::murmur_hash3_32(
98
                            reinterpret_cast<const char*>(&data[current_offset]),
99
                            offsets[i] - current_offset, col_to_data[i]);
100
6
                } else {
101
6
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
102
6
                            reinterpret_cast<const char*>(&data[current_offset]),
103
6
                            offsets[i] - current_offset, static_cast<uint64_t>(col_to_data[i]));
104
6
                }
105
6
                current_offset = offsets[i];
106
6
            }
107
3
        } else if (const ColumnConst* col_from_const =
108
0
                           check_and_get_column_const_string_or_fixedstring(column)) {
109
0
            auto value = col_from_const->get_value<TYPE_STRING>();
110
0
            for (size_t i = 0; i < input_rows_count; ++i) {
111
                if constexpr (ReturnType == TYPE_INT) {
112
                    col_to_data[i] =
113
                            HashUtil::murmur_hash3_32(value.data(), value.size(), col_to_data[i]);
114
0
                } else {
115
0
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
116
0
                            value.data(), value.size(), static_cast<uint64_t>(col_to_data[i]));
117
0
                }
118
0
            }
119
0
        } else {
120
0
            DCHECK(false);
121
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
122
0
                                        column->get_name(), name);
123
0
        }
124
3
        return Status::OK();
125
3
    }
_ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb1EE7executeILb1EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
79
1
                          IColumn& col_to) {
80
1
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
81
1
        if constexpr (first) {
82
            if constexpr (ReturnType == TYPE_INT) {
83
                to_column.insert_many_vals(static_cast<Int32>(HashUtil::MURMUR3_32_SEED),
84
                                           input_rows_count);
85
1
            } else {
86
1
                to_column.insert_many_defaults(input_rows_count);
87
1
            }
88
1
        }
89
1
        auto& col_to_data = to_column.get_data();
90
1
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
91
1
            const typename ColumnString::Chars& data = col_from->get_chars();
92
1
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
93
1
            size_t size = offsets.size();
94
1
            ColumnString::Offset current_offset = 0;
95
4
            for (size_t i = 0; i < size; ++i) {
96
                if constexpr (ReturnType == TYPE_INT) {
97
                    col_to_data[i] = HashUtil::murmur_hash3_32(
98
                            reinterpret_cast<const char*>(&data[current_offset]),
99
                            offsets[i] - current_offset, col_to_data[i]);
100
3
                } else {
101
3
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
102
3
                            reinterpret_cast<const char*>(&data[current_offset]),
103
3
                            offsets[i] - current_offset, static_cast<uint64_t>(col_to_data[i]));
104
3
                }
105
3
                current_offset = offsets[i];
106
3
            }
107
1
        } else if (const ColumnConst* col_from_const =
108
0
                           check_and_get_column_const_string_or_fixedstring(column)) {
109
0
            auto value = col_from_const->get_value<TYPE_STRING>();
110
0
            for (size_t i = 0; i < input_rows_count; ++i) {
111
                if constexpr (ReturnType == TYPE_INT) {
112
                    col_to_data[i] =
113
                            HashUtil::murmur_hash3_32(value.data(), value.size(), col_to_data[i]);
114
0
                } else {
115
0
                    col_to_data[i] = HashUtil::murmur_hash3_64<is_mmh64_v2>(
116
0
                            value.data(), value.size(), static_cast<uint64_t>(col_to_data[i]));
117
0
                }
118
0
            }
119
0
        } else {
120
0
            DCHECK(false);
121
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
122
0
                                        column->get_name(), name);
123
0
        }
124
1
        return Status::OK();
125
1
    }
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE6ELb1EE7executeILb0EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE7ELb1EE7executeILb1EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Unexecuted instantiation: _ZN5doris15MurmurHash3ImplILNS_13PrimitiveTypeE7ELb1EE7executeILb0EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
126
};
127
128
using FunctionMurmurHash3_32 =
129
        FunctionVariadicArgumentsBase<DataTypeInt32, MurmurHash3Impl<TYPE_INT>>;
130
using FunctionMurmurHash3_64 =
131
        FunctionVariadicArgumentsBase<DataTypeInt64, MurmurHash3Impl<TYPE_BIGINT>>;
132
using FunctionMurmurHash3_64_V2 =
133
        FunctionVariadicArgumentsBase<DataTypeInt64, MurmurHash3Impl<TYPE_BIGINT, true>>;
134
using FunctionMurmurHash3U64V2 =
135
        FunctionVariadicArgumentsBase<DataTypeInt128, MurmurHash3Impl<TYPE_LARGEINT, true>>;
136
137
#ifdef BE_TEST
138
1
const char* murmur_hash3_get_name_type_int_for_test() {
139
1
    return MurmurHash3Impl<TYPE_INT>::get_name();
140
1
}
141
142
1
const char* murmur_hash3_get_name_type_bigint_for_test() {
143
1
    return MurmurHash3Impl<TYPE_BIGINT>::get_name();
144
1
}
145
146
1
const char* murmur_hash3_get_name_type_bigint_v2_for_test() {
147
1
    return MurmurHash3Impl<TYPE_BIGINT, true>::get_name();
148
1
}
149
#endif
150
151
template <PrimitiveType ReturnType>
152
struct XxHashImpl {
153
    static constexpr auto name = ReturnType == TYPE_INT ? "xxhash_32" : "xxhash_64";
154
155
0
    static Status empty_apply(IColumn& icolumn, size_t input_rows_count) {
156
0
        ColumnVector<ReturnType>& vec_to = assert_cast<ColumnVector<ReturnType>&>(icolumn);
157
0
        vec_to.get_data().assign(
158
0
                input_rows_count,
159
0
                static_cast<typename PrimitiveTypeTraits<ReturnType>::CppType>(emtpy_value));
160
0
        return Status::OK();
161
0
    }
Unexecuted instantiation: _ZN5doris10XxHashImplILNS_13PrimitiveTypeE5EE11empty_applyERNS_7IColumnEm
Unexecuted instantiation: _ZN5doris10XxHashImplILNS_13PrimitiveTypeE6EE11empty_applyERNS_7IColumnEm
162
163
    static Status first_apply(const IDataType* type, const IColumn* column, size_t input_rows_count,
164
12
                              IColumn& icolumn) {
165
12
        return execute<true>(type, column, input_rows_count, icolumn);
166
12
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE5EE11first_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
164
6
                              IColumn& icolumn) {
165
6
        return execute<true>(type, column, input_rows_count, icolumn);
166
6
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE6EE11first_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
164
6
                              IColumn& icolumn) {
165
6
        return execute<true>(type, column, input_rows_count, icolumn);
166
6
    }
167
168
    static Status combine_apply(const IDataType* type, const IColumn* column,
169
12
                                size_t input_rows_count, IColumn& icolumn) {
170
12
        return execute<false>(type, column, input_rows_count, icolumn);
171
12
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE5EE13combine_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
169
6
                                size_t input_rows_count, IColumn& icolumn) {
170
6
        return execute<false>(type, column, input_rows_count, icolumn);
171
6
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE6EE13combine_applyEPKNS_9IDataTypeEPKNS_7IColumnEmRS6_
Line
Count
Source
169
6
                                size_t input_rows_count, IColumn& icolumn) {
170
6
        return execute<false>(type, column, input_rows_count, icolumn);
171
6
    }
172
173
    template <bool first>
174
    static Status execute(const IDataType* type, const IColumn* column, size_t input_rows_count,
175
24
                          IColumn& col_to) {
176
24
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
177
24
        if constexpr (first) {
178
12
            to_column.insert_many_defaults(input_rows_count);
179
12
        }
180
24
        auto& col_to_data = to_column.get_data();
181
24
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
182
12
            const typename ColumnString::Chars& data = col_from->get_chars();
183
12
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
184
12
            size_t size = offsets.size();
185
12
            ColumnString::Offset current_offset = 0;
186
36
            for (size_t i = 0; i < size; ++i) {
187
24
                if constexpr (ReturnType == TYPE_INT) {
188
12
                    col_to_data[i] = HashUtil::xxHash32WithSeed(
189
12
                            reinterpret_cast<const char*>(&data[current_offset]),
190
12
                            offsets[i] - current_offset, col_to_data[i]);
191
12
                } else {
192
12
                    col_to_data[i] = HashUtil::xxHash64WithSeed(
193
12
                            reinterpret_cast<const char*>(&data[current_offset]),
194
12
                            offsets[i] - current_offset, col_to_data[i]);
195
12
                }
196
24
                current_offset = offsets[i];
197
24
            }
198
12
        } else if (const ColumnConst* col_from_const =
199
12
                           check_and_get_column_const_string_or_fixedstring(column)) {
200
0
            auto value = col_from_const->get_value<TYPE_STRING>();
201
0
            for (size_t i = 0; i < input_rows_count; ++i) {
202
0
                if constexpr (ReturnType == TYPE_INT) {
203
0
                    col_to_data[i] =
204
0
                            HashUtil::xxHash32WithSeed(value.data(), value.size(), col_to_data[i]);
205
0
                } else {
206
0
                    col_to_data[i] =
207
0
                            HashUtil::xxHash64WithSeed(value.data(), value.size(), col_to_data[i]);
208
0
                }
209
0
            }
210
12
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(column)) {
211
36
            for (size_t i = 0; i < input_rows_count; ++i) {
212
24
                auto data_ref = vb_col->get_data_at(i);
213
24
                if constexpr (ReturnType == TYPE_INT) {
214
12
                    col_to_data[i] = HashUtil::xxHash32WithSeed(data_ref.data, data_ref.size,
215
12
                                                                col_to_data[i]);
216
12
                } else {
217
12
                    col_to_data[i] = HashUtil::xxHash64WithSeed(data_ref.data, data_ref.size,
218
12
                                                                col_to_data[i]);
219
12
                }
220
24
            }
221
12
        } else {
222
0
            DCHECK(false);
223
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
224
0
                                        column->get_name(), name);
225
0
        }
226
24
        return Status::OK();
227
24
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE5EE7executeILb1EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
175
6
                          IColumn& col_to) {
176
6
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
177
6
        if constexpr (first) {
178
6
            to_column.insert_many_defaults(input_rows_count);
179
6
        }
180
6
        auto& col_to_data = to_column.get_data();
181
6
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
182
3
            const typename ColumnString::Chars& data = col_from->get_chars();
183
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
184
3
            size_t size = offsets.size();
185
3
            ColumnString::Offset current_offset = 0;
186
9
            for (size_t i = 0; i < size; ++i) {
187
6
                if constexpr (ReturnType == TYPE_INT) {
188
6
                    col_to_data[i] = HashUtil::xxHash32WithSeed(
189
6
                            reinterpret_cast<const char*>(&data[current_offset]),
190
6
                            offsets[i] - current_offset, col_to_data[i]);
191
                } else {
192
                    col_to_data[i] = HashUtil::xxHash64WithSeed(
193
                            reinterpret_cast<const char*>(&data[current_offset]),
194
                            offsets[i] - current_offset, col_to_data[i]);
195
                }
196
6
                current_offset = offsets[i];
197
6
            }
198
3
        } else if (const ColumnConst* col_from_const =
199
3
                           check_and_get_column_const_string_or_fixedstring(column)) {
200
0
            auto value = col_from_const->get_value<TYPE_STRING>();
201
0
            for (size_t i = 0; i < input_rows_count; ++i) {
202
0
                if constexpr (ReturnType == TYPE_INT) {
203
0
                    col_to_data[i] =
204
0
                            HashUtil::xxHash32WithSeed(value.data(), value.size(), col_to_data[i]);
205
                } else {
206
                    col_to_data[i] =
207
                            HashUtil::xxHash64WithSeed(value.data(), value.size(), col_to_data[i]);
208
                }
209
0
            }
210
3
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(column)) {
211
9
            for (size_t i = 0; i < input_rows_count; ++i) {
212
6
                auto data_ref = vb_col->get_data_at(i);
213
6
                if constexpr (ReturnType == TYPE_INT) {
214
6
                    col_to_data[i] = HashUtil::xxHash32WithSeed(data_ref.data, data_ref.size,
215
6
                                                                col_to_data[i]);
216
                } else {
217
                    col_to_data[i] = HashUtil::xxHash64WithSeed(data_ref.data, data_ref.size,
218
                                                                col_to_data[i]);
219
                }
220
6
            }
221
3
        } else {
222
0
            DCHECK(false);
223
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
224
0
                                        column->get_name(), name);
225
0
        }
226
6
        return Status::OK();
227
6
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE5EE7executeILb0EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
175
6
                          IColumn& col_to) {
176
6
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
177
        if constexpr (first) {
178
            to_column.insert_many_defaults(input_rows_count);
179
        }
180
6
        auto& col_to_data = to_column.get_data();
181
6
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
182
3
            const typename ColumnString::Chars& data = col_from->get_chars();
183
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
184
3
            size_t size = offsets.size();
185
3
            ColumnString::Offset current_offset = 0;
186
9
            for (size_t i = 0; i < size; ++i) {
187
6
                if constexpr (ReturnType == TYPE_INT) {
188
6
                    col_to_data[i] = HashUtil::xxHash32WithSeed(
189
6
                            reinterpret_cast<const char*>(&data[current_offset]),
190
6
                            offsets[i] - current_offset, col_to_data[i]);
191
                } else {
192
                    col_to_data[i] = HashUtil::xxHash64WithSeed(
193
                            reinterpret_cast<const char*>(&data[current_offset]),
194
                            offsets[i] - current_offset, col_to_data[i]);
195
                }
196
6
                current_offset = offsets[i];
197
6
            }
198
3
        } else if (const ColumnConst* col_from_const =
199
3
                           check_and_get_column_const_string_or_fixedstring(column)) {
200
0
            auto value = col_from_const->get_value<TYPE_STRING>();
201
0
            for (size_t i = 0; i < input_rows_count; ++i) {
202
0
                if constexpr (ReturnType == TYPE_INT) {
203
0
                    col_to_data[i] =
204
0
                            HashUtil::xxHash32WithSeed(value.data(), value.size(), col_to_data[i]);
205
                } else {
206
                    col_to_data[i] =
207
                            HashUtil::xxHash64WithSeed(value.data(), value.size(), col_to_data[i]);
208
                }
209
0
            }
210
3
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(column)) {
211
9
            for (size_t i = 0; i < input_rows_count; ++i) {
212
6
                auto data_ref = vb_col->get_data_at(i);
213
6
                if constexpr (ReturnType == TYPE_INT) {
214
6
                    col_to_data[i] = HashUtil::xxHash32WithSeed(data_ref.data, data_ref.size,
215
6
                                                                col_to_data[i]);
216
                } else {
217
                    col_to_data[i] = HashUtil::xxHash64WithSeed(data_ref.data, data_ref.size,
218
                                                                col_to_data[i]);
219
                }
220
6
            }
221
3
        } else {
222
0
            DCHECK(false);
223
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
224
0
                                        column->get_name(), name);
225
0
        }
226
6
        return Status::OK();
227
6
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE6EE7executeILb1EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
175
6
                          IColumn& col_to) {
176
6
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
177
6
        if constexpr (first) {
178
6
            to_column.insert_many_defaults(input_rows_count);
179
6
        }
180
6
        auto& col_to_data = to_column.get_data();
181
6
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
182
3
            const typename ColumnString::Chars& data = col_from->get_chars();
183
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
184
3
            size_t size = offsets.size();
185
3
            ColumnString::Offset current_offset = 0;
186
9
            for (size_t i = 0; i < size; ++i) {
187
                if constexpr (ReturnType == TYPE_INT) {
188
                    col_to_data[i] = HashUtil::xxHash32WithSeed(
189
                            reinterpret_cast<const char*>(&data[current_offset]),
190
                            offsets[i] - current_offset, col_to_data[i]);
191
6
                } else {
192
6
                    col_to_data[i] = HashUtil::xxHash64WithSeed(
193
6
                            reinterpret_cast<const char*>(&data[current_offset]),
194
6
                            offsets[i] - current_offset, col_to_data[i]);
195
6
                }
196
6
                current_offset = offsets[i];
197
6
            }
198
3
        } else if (const ColumnConst* col_from_const =
199
3
                           check_and_get_column_const_string_or_fixedstring(column)) {
200
0
            auto value = col_from_const->get_value<TYPE_STRING>();
201
0
            for (size_t i = 0; i < input_rows_count; ++i) {
202
                if constexpr (ReturnType == TYPE_INT) {
203
                    col_to_data[i] =
204
                            HashUtil::xxHash32WithSeed(value.data(), value.size(), col_to_data[i]);
205
0
                } else {
206
0
                    col_to_data[i] =
207
0
                            HashUtil::xxHash64WithSeed(value.data(), value.size(), col_to_data[i]);
208
0
                }
209
0
            }
210
3
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(column)) {
211
9
            for (size_t i = 0; i < input_rows_count; ++i) {
212
6
                auto data_ref = vb_col->get_data_at(i);
213
                if constexpr (ReturnType == TYPE_INT) {
214
                    col_to_data[i] = HashUtil::xxHash32WithSeed(data_ref.data, data_ref.size,
215
                                                                col_to_data[i]);
216
6
                } else {
217
6
                    col_to_data[i] = HashUtil::xxHash64WithSeed(data_ref.data, data_ref.size,
218
6
                                                                col_to_data[i]);
219
6
                }
220
6
            }
221
3
        } else {
222
0
            DCHECK(false);
223
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
224
0
                                        column->get_name(), name);
225
0
        }
226
6
        return Status::OK();
227
6
    }
_ZN5doris10XxHashImplILNS_13PrimitiveTypeE6EE7executeILb0EEENS_6StatusEPKNS_9IDataTypeEPKNS_7IColumnEmRS8_
Line
Count
Source
175
6
                          IColumn& col_to) {
176
6
        auto& to_column = assert_cast<ColumnVector<ReturnType>&>(col_to);
177
        if constexpr (first) {
178
            to_column.insert_many_defaults(input_rows_count);
179
        }
180
6
        auto& col_to_data = to_column.get_data();
181
6
        if (const auto* col_from = check_and_get_column<ColumnString>(column)) {
182
3
            const typename ColumnString::Chars& data = col_from->get_chars();
183
3
            const typename ColumnString::Offsets& offsets = col_from->get_offsets();
184
3
            size_t size = offsets.size();
185
3
            ColumnString::Offset current_offset = 0;
186
9
            for (size_t i = 0; i < size; ++i) {
187
                if constexpr (ReturnType == TYPE_INT) {
188
                    col_to_data[i] = HashUtil::xxHash32WithSeed(
189
                            reinterpret_cast<const char*>(&data[current_offset]),
190
                            offsets[i] - current_offset, col_to_data[i]);
191
6
                } else {
192
6
                    col_to_data[i] = HashUtil::xxHash64WithSeed(
193
6
                            reinterpret_cast<const char*>(&data[current_offset]),
194
6
                            offsets[i] - current_offset, col_to_data[i]);
195
6
                }
196
6
                current_offset = offsets[i];
197
6
            }
198
3
        } else if (const ColumnConst* col_from_const =
199
3
                           check_and_get_column_const_string_or_fixedstring(column)) {
200
0
            auto value = col_from_const->get_value<TYPE_STRING>();
201
0
            for (size_t i = 0; i < input_rows_count; ++i) {
202
                if constexpr (ReturnType == TYPE_INT) {
203
                    col_to_data[i] =
204
                            HashUtil::xxHash32WithSeed(value.data(), value.size(), col_to_data[i]);
205
0
                } else {
206
0
                    col_to_data[i] =
207
0
                            HashUtil::xxHash64WithSeed(value.data(), value.size(), col_to_data[i]);
208
0
                }
209
0
            }
210
3
        } else if (const auto* vb_col = check_and_get_column<ColumnVarbinary>(column)) {
211
9
            for (size_t i = 0; i < input_rows_count; ++i) {
212
6
                auto data_ref = vb_col->get_data_at(i);
213
                if constexpr (ReturnType == TYPE_INT) {
214
                    col_to_data[i] = HashUtil::xxHash32WithSeed(data_ref.data, data_ref.size,
215
                                                                col_to_data[i]);
216
6
                } else {
217
6
                    col_to_data[i] = HashUtil::xxHash64WithSeed(data_ref.data, data_ref.size,
218
6
                                                                col_to_data[i]);
219
6
                }
220
6
            }
221
3
        } else {
222
0
            DCHECK(false);
223
0
            return Status::NotSupported("Illegal column {} of argument of function {}",
224
0
                                        column->get_name(), name);
225
0
        }
226
6
        return Status::OK();
227
6
    }
228
};
229
230
using FunctionXxHash_32 = FunctionVariadicArgumentsBase<DataTypeInt32, XxHashImpl<TYPE_INT>>;
231
using FunctionXxHash_64 = FunctionVariadicArgumentsBase<DataTypeInt64, XxHashImpl<TYPE_BIGINT>>;
232
233
1
void register_function_hash(SimpleFunctionFactory& factory) {
234
1
    factory.register_function<FunctionMurmurHash3_32>();
235
1
    factory.register_function<FunctionMurmurHash3_64>();
236
1
    factory.register_function<FunctionMurmurHash3_64_V2>();
237
1
    factory.register_function<FunctionMurmurHash3U64V2>();
238
1
    factory.register_function<FunctionXxHash_32>();
239
1
    factory.register_function<FunctionXxHash_64>();
240
1
    factory.register_alias("xxhash_64", "xxhash3_64");
241
1
}
242
} // namespace doris