be/src/exec/common/join_utils.h
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 | | #pragma once |
19 | | |
20 | | #include <algorithm> |
21 | | #include <variant> |
22 | | |
23 | | #include "exec/common/hash_table/hash_crc32_return32.h" |
24 | | #include "exec/common/hash_table/hash_key_type.h" |
25 | | #include "exec/common/hash_table/hash_map_context.h" |
26 | | #include "exec/common/hash_table/join_hash_table.h" |
27 | | #include "exec/common/join_op_utils.h" // IWYU pragma: export (JoinOpVariants/AsofIndexGroup moved there) |
28 | | |
29 | | namespace doris { |
30 | | |
31 | | // Devirtualize compare_at for ASOF JOIN supported column types. |
32 | | // ASOF JOIN only supports DateV2, DateTimeV2, and TimestampTZ. |
33 | | // Dispatches to the concrete ColumnVector<T> once so that all compare_at |
34 | | // calls inside `func` are direct (non-virtual) calls. |
35 | | // `func` receives a single argument: a const pointer to the concrete column |
36 | | // (or const IColumn* as fallback for unexpected types). |
37 | | template <typename Func> |
38 | 4 | decltype(auto) asof_column_dispatch(const IColumn* col, Func&& func) { |
39 | 4 | if (const auto* c_dv2 = check_and_get_column<ColumnDateV2>(col)) { |
40 | 1 | return std::forward<Func>(func)(c_dv2); |
41 | 3 | } else if (const auto* c_dtv2 = check_and_get_column<ColumnDateTimeV2>(col)) { |
42 | 1 | return std::forward<Func>(func)(c_dtv2); |
43 | 2 | } else if (const auto* c_tstz = check_and_get_column<ColumnTimeStampTz>(col)) { |
44 | 1 | return std::forward<Func>(func)(c_tstz); |
45 | 1 | } else { |
46 | 1 | return std::forward<Func>(func)(col); |
47 | 1 | } |
48 | 4 | } asof_join_test.cpp:_ZN5doris20asof_column_dispatchIZNS_42AsofColumnDispatchTest_DateV2Dispatch_Test8TestBodyEvE3$_0EEDcPKNS_7IColumnEOT_ Line | Count | Source | 38 | 1 | decltype(auto) asof_column_dispatch(const IColumn* col, Func&& func) { | 39 | 1 | if (const auto* c_dv2 = check_and_get_column<ColumnDateV2>(col)) { | 40 | 1 | return std::forward<Func>(func)(c_dv2); | 41 | 1 | } else if (const auto* c_dtv2 = check_and_get_column<ColumnDateTimeV2>(col)) { | 42 | 0 | return std::forward<Func>(func)(c_dtv2); | 43 | 0 | } else if (const auto* c_tstz = check_and_get_column<ColumnTimeStampTz>(col)) { | 44 | 0 | return std::forward<Func>(func)(c_tstz); | 45 | 0 | } else { | 46 | 0 | return std::forward<Func>(func)(col); | 47 | 0 | } | 48 | 1 | } |
asof_join_test.cpp:_ZN5doris20asof_column_dispatchIZNS_46AsofColumnDispatchTest_DateTimeV2Dispatch_Test8TestBodyEvE3$_0EEDcPKNS_7IColumnEOT_ Line | Count | Source | 38 | 1 | decltype(auto) asof_column_dispatch(const IColumn* col, Func&& func) { | 39 | 1 | if (const auto* c_dv2 = check_and_get_column<ColumnDateV2>(col)) { | 40 | 0 | return std::forward<Func>(func)(c_dv2); | 41 | 1 | } else if (const auto* c_dtv2 = check_and_get_column<ColumnDateTimeV2>(col)) { | 42 | 1 | return std::forward<Func>(func)(c_dtv2); | 43 | 1 | } else if (const auto* c_tstz = check_and_get_column<ColumnTimeStampTz>(col)) { | 44 | 0 | return std::forward<Func>(func)(c_tstz); | 45 | 0 | } else { | 46 | 0 | return std::forward<Func>(func)(col); | 47 | 0 | } | 48 | 1 | } |
asof_join_test.cpp:_ZN5doris20asof_column_dispatchIZNS_47AsofColumnDispatchTest_TimestampTZDispatch_Test8TestBodyEvE3$_0EEDcPKNS_7IColumnEOT_ Line | Count | Source | 38 | 1 | decltype(auto) asof_column_dispatch(const IColumn* col, Func&& func) { | 39 | 1 | if (const auto* c_dv2 = check_and_get_column<ColumnDateV2>(col)) { | 40 | 0 | return std::forward<Func>(func)(c_dv2); | 41 | 1 | } else if (const auto* c_dtv2 = check_and_get_column<ColumnDateTimeV2>(col)) { | 42 | 0 | return std::forward<Func>(func)(c_dtv2); | 43 | 1 | } else if (const auto* c_tstz = check_and_get_column<ColumnTimeStampTz>(col)) { | 44 | 1 | return std::forward<Func>(func)(c_tstz); | 45 | 1 | } else { | 46 | 0 | return std::forward<Func>(func)(col); | 47 | 0 | } | 48 | 1 | } |
asof_join_test.cpp:_ZN5doris20asof_column_dispatchIZNS_44AsofColumnDispatchTest_FallbackDispatch_Test8TestBodyEvE3$_0EEDcPKNS_7IColumnEOT_ Line | Count | Source | 38 | 1 | decltype(auto) asof_column_dispatch(const IColumn* col, Func&& func) { | 39 | 1 | if (const auto* c_dv2 = check_and_get_column<ColumnDateV2>(col)) { | 40 | 0 | return std::forward<Func>(func)(c_dv2); | 41 | 1 | } else if (const auto* c_dtv2 = check_and_get_column<ColumnDateTimeV2>(col)) { | 42 | 0 | return std::forward<Func>(func)(c_dtv2); | 43 | 1 | } else if (const auto* c_tstz = check_and_get_column<ColumnTimeStampTz>(col)) { | 44 | 0 | return std::forward<Func>(func)(c_tstz); | 45 | 1 | } else { | 46 | 1 | return std::forward<Func>(func)(col); | 47 | 1 | } | 48 | 1 | } |
Unexecuted instantiation: hashjoin_build_sink.cpp:_ZN5doris20asof_column_dispatchIZNS_27HashJoinBuildSinkLocalState16build_asof_indexERNS_5BlockEE3$_2EEDcPKNS_7IColumnEOT_ |
49 | | template <class T> |
50 | | using PrimaryTypeHashTableContext = MethodOneNumber<T, JoinHashMap<T, HashCRC32Return32<T>, false>>; |
51 | | |
52 | | template <class T> |
53 | | using DirectPrimaryTypeHashTableContext = |
54 | | MethodOneNumberDirect<T, JoinHashMap<T, HashCRC32Return32<T>, true>>; |
55 | | |
56 | | template <class Key> |
57 | | using FixedKeyHashTableContext = MethodKeysFixed<JoinHashMap<Key, HashCRC32Return32<Key>, false>>; |
58 | | |
59 | | using SerializedHashTableContext = |
60 | | MethodSerialized<JoinHashMap<StringRef, HashCRC32Return32<StringRef>, false>>; |
61 | | using MethodOneString = |
62 | | MethodStringNoCache<JoinHashMap<StringRef, HashCRC32Return32<StringRef>, false>>; |
63 | | |
64 | | using HashTableVariants = std::variant< |
65 | | std::monostate, SerializedHashTableContext, PrimaryTypeHashTableContext<UInt8>, |
66 | | PrimaryTypeHashTableContext<UInt16>, PrimaryTypeHashTableContext<UInt32>, |
67 | | PrimaryTypeHashTableContext<UInt64>, PrimaryTypeHashTableContext<UInt128>, |
68 | | PrimaryTypeHashTableContext<UInt256>, DirectPrimaryTypeHashTableContext<UInt8>, |
69 | | DirectPrimaryTypeHashTableContext<UInt16>, DirectPrimaryTypeHashTableContext<UInt32>, |
70 | | DirectPrimaryTypeHashTableContext<UInt64>, DirectPrimaryTypeHashTableContext<UInt128>, |
71 | | FixedKeyHashTableContext<UInt64>, FixedKeyHashTableContext<UInt72>, |
72 | | FixedKeyHashTableContext<UInt96>, FixedKeyHashTableContext<UInt104>, |
73 | | FixedKeyHashTableContext<UInt128>, FixedKeyHashTableContext<UInt136>, |
74 | | FixedKeyHashTableContext<UInt256>, MethodOneString>; |
75 | | |
76 | | struct JoinDataVariants { |
77 | | HashTableVariants method_variant; |
78 | | |
79 | 48.0k | void init(const std::vector<DataTypePtr>& data_types, HashKeyType type) { |
80 | 48.0k | switch (type) { |
81 | 14.1k | case HashKeyType::serialized: |
82 | 14.1k | method_variant.emplace<SerializedHashTableContext>(); |
83 | 14.1k | break; |
84 | 160 | case HashKeyType::int8_key: |
85 | 160 | method_variant.emplace<PrimaryTypeHashTableContext<UInt8>>(); |
86 | 160 | break; |
87 | 80 | case HashKeyType::int16_key: |
88 | 80 | method_variant.emplace<PrimaryTypeHashTableContext<UInt16>>(); |
89 | 80 | break; |
90 | 324 | case HashKeyType::int32_key: |
91 | 324 | method_variant.emplace<PrimaryTypeHashTableContext<UInt32>>(); |
92 | 324 | break; |
93 | 800 | case HashKeyType::int64_key: |
94 | 800 | method_variant.emplace<PrimaryTypeHashTableContext<UInt64>>(); |
95 | 800 | break; |
96 | 240 | case HashKeyType::int128_key: |
97 | 240 | method_variant.emplace<PrimaryTypeHashTableContext<UInt128>>(); |
98 | 240 | break; |
99 | 0 | case HashKeyType::int256_key: |
100 | 0 | method_variant.emplace<PrimaryTypeHashTableContext<UInt256>>(); |
101 | 0 | break; |
102 | 324 | case HashKeyType::string_key: |
103 | 324 | method_variant.emplace<MethodOneString>(); |
104 | 324 | break; |
105 | 3.92k | case HashKeyType::fixed64: |
106 | 3.92k | method_variant.emplace<FixedKeyHashTableContext<UInt64>>(get_key_sizes(data_types)); |
107 | 3.92k | break; |
108 | 3.20k | case HashKeyType::fixed72: |
109 | 3.20k | method_variant.emplace<FixedKeyHashTableContext<UInt72>>(get_key_sizes(data_types)); |
110 | 3.20k | break; |
111 | 8.00k | case HashKeyType::fixed96: |
112 | 8.00k | method_variant.emplace<FixedKeyHashTableContext<UInt96>>(get_key_sizes(data_types)); |
113 | 8.00k | break; |
114 | 0 | case HashKeyType::fixed104: |
115 | 0 | method_variant.emplace<FixedKeyHashTableContext<UInt104>>(get_key_sizes(data_types)); |
116 | 0 | break; |
117 | 8.00k | case HashKeyType::fixed128: |
118 | 8.00k | method_variant.emplace<FixedKeyHashTableContext<UInt128>>(get_key_sizes(data_types)); |
119 | 8.00k | break; |
120 | 960 | case HashKeyType::fixed136: |
121 | 960 | method_variant.emplace<FixedKeyHashTableContext<UInt136>>(get_key_sizes(data_types)); |
122 | 960 | break; |
123 | 7.92k | case HashKeyType::fixed256: |
124 | 7.92k | method_variant.emplace<FixedKeyHashTableContext<UInt256>>(get_key_sizes(data_types)); |
125 | 7.92k | break; |
126 | 0 | default: |
127 | 0 | throw Exception(ErrorCode::INTERNAL_ERROR, |
128 | 0 | "JoinDataVariants meet invalid key type, type={}", type); |
129 | 48.0k | } |
130 | 48.0k | } |
131 | | }; |
132 | | |
133 | | template <typename Method> |
134 | | void primary_to_direct_mapping(Method* context, const ColumnRawPtrs& key_columns, |
135 | 1.60k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { |
136 | 1.60k | using FieldType = typename Method::Base::Key; |
137 | 1.60k | FieldType max_key = std::numeric_limits<FieldType>::min(); |
138 | 1.60k | FieldType min_key = std::numeric_limits<FieldType>::max(); |
139 | | |
140 | 1.60k | size_t num_rows = key_columns[0]->size(); |
141 | 1.60k | if (is_column_nullable(*key_columns[0])) { |
142 | 0 | const FieldType* input_keys = (FieldType*)assert_cast<const ColumnNullable*>(key_columns[0]) |
143 | 0 | ->get_nested_column_ptr() |
144 | 0 | ->get_raw_data() |
145 | 0 | .data; |
146 | 0 | const NullMap& null_map = |
147 | 0 | assert_cast<const ColumnNullable*>(key_columns[0])->get_null_map_data(); |
148 | | // skip first mocked row |
149 | 0 | for (size_t i = 1; i < num_rows; i++) { |
150 | 0 | if (null_map[i]) { |
151 | 0 | continue; |
152 | 0 | } |
153 | 0 | max_key = std::max(max_key, input_keys[i]); |
154 | 0 | min_key = std::min(min_key, input_keys[i]); |
155 | 0 | } |
156 | 1.60k | } else { |
157 | 1.60k | const FieldType* input_keys = (FieldType*)key_columns[0]->get_raw_data().data; |
158 | | // skip first mocked row |
159 | 4.83k | for (size_t i = 1; i < num_rows; i++) { |
160 | 3.22k | max_key = std::max(max_key, input_keys[i]); |
161 | 3.22k | min_key = std::min(min_key, input_keys[i]); |
162 | 3.22k | } |
163 | 1.60k | } |
164 | | |
165 | 1.60k | constexpr auto MAX_MAPPING_RANGE = 1 << 23; |
166 | 1.60k | bool allow_direct_mapping = (max_key >= min_key && max_key - min_key < MAX_MAPPING_RANGE - 1); |
167 | 1.60k | if (allow_direct_mapping) { |
168 | 1.60k | for (const auto& variant_ptr : variant_ptrs) { |
169 | 1.60k | variant_ptr->method_variant.emplace<DirectPrimaryTypeHashTableContext<FieldType>>( |
170 | 1.60k | max_key, min_key); |
171 | 1.60k | } |
172 | 1.60k | } |
173 | 1.60k | } _ZN5doris25primary_to_direct_mappingINS_15MethodOneNumberIhNS_13JoinHashTableIhNS_17HashCRC32Return32IhEELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Line | Count | Source | 135 | 160 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { | 136 | 160 | using FieldType = typename Method::Base::Key; | 137 | 160 | FieldType max_key = std::numeric_limits<FieldType>::min(); | 138 | 160 | FieldType min_key = std::numeric_limits<FieldType>::max(); | 139 | | | 140 | 160 | size_t num_rows = key_columns[0]->size(); | 141 | 160 | if (is_column_nullable(*key_columns[0])) { | 142 | 0 | const FieldType* input_keys = (FieldType*)assert_cast<const ColumnNullable*>(key_columns[0]) | 143 | 0 | ->get_nested_column_ptr() | 144 | 0 | ->get_raw_data() | 145 | 0 | .data; | 146 | 0 | const NullMap& null_map = | 147 | 0 | assert_cast<const ColumnNullable*>(key_columns[0])->get_null_map_data(); | 148 | | // skip first mocked row | 149 | 0 | for (size_t i = 1; i < num_rows; i++) { | 150 | 0 | if (null_map[i]) { | 151 | 0 | continue; | 152 | 0 | } | 153 | 0 | max_key = std::max(max_key, input_keys[i]); | 154 | 0 | min_key = std::min(min_key, input_keys[i]); | 155 | 0 | } | 156 | 160 | } else { | 157 | 160 | const FieldType* input_keys = (FieldType*)key_columns[0]->get_raw_data().data; | 158 | | // skip first mocked row | 159 | 480 | for (size_t i = 1; i < num_rows; i++) { | 160 | 320 | max_key = std::max(max_key, input_keys[i]); | 161 | 320 | min_key = std::min(min_key, input_keys[i]); | 162 | 320 | } | 163 | 160 | } | 164 | | | 165 | 160 | constexpr auto MAX_MAPPING_RANGE = 1 << 23; | 166 | 160 | bool allow_direct_mapping = (max_key >= min_key && max_key - min_key < MAX_MAPPING_RANGE - 1); | 167 | 160 | if (allow_direct_mapping) { | 168 | 160 | for (const auto& variant_ptr : variant_ptrs) { | 169 | 160 | variant_ptr->method_variant.emplace<DirectPrimaryTypeHashTableContext<FieldType>>( | 170 | 160 | max_key, min_key); | 171 | 160 | } | 172 | 160 | } | 173 | 160 | } |
_ZN5doris25primary_to_direct_mappingINS_15MethodOneNumberItNS_13JoinHashTableItNS_17HashCRC32Return32ItEELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Line | Count | Source | 135 | 80 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { | 136 | 80 | using FieldType = typename Method::Base::Key; | 137 | 80 | FieldType max_key = std::numeric_limits<FieldType>::min(); | 138 | 80 | FieldType min_key = std::numeric_limits<FieldType>::max(); | 139 | | | 140 | 80 | size_t num_rows = key_columns[0]->size(); | 141 | 80 | if (is_column_nullable(*key_columns[0])) { | 142 | 0 | const FieldType* input_keys = (FieldType*)assert_cast<const ColumnNullable*>(key_columns[0]) | 143 | 0 | ->get_nested_column_ptr() | 144 | 0 | ->get_raw_data() | 145 | 0 | .data; | 146 | 0 | const NullMap& null_map = | 147 | 0 | assert_cast<const ColumnNullable*>(key_columns[0])->get_null_map_data(); | 148 | | // skip first mocked row | 149 | 0 | for (size_t i = 1; i < num_rows; i++) { | 150 | 0 | if (null_map[i]) { | 151 | 0 | continue; | 152 | 0 | } | 153 | 0 | max_key = std::max(max_key, input_keys[i]); | 154 | 0 | min_key = std::min(min_key, input_keys[i]); | 155 | 0 | } | 156 | 80 | } else { | 157 | 80 | const FieldType* input_keys = (FieldType*)key_columns[0]->get_raw_data().data; | 158 | | // skip first mocked row | 159 | 240 | for (size_t i = 1; i < num_rows; i++) { | 160 | 160 | max_key = std::max(max_key, input_keys[i]); | 161 | 160 | min_key = std::min(min_key, input_keys[i]); | 162 | 160 | } | 163 | 80 | } | 164 | | | 165 | 80 | constexpr auto MAX_MAPPING_RANGE = 1 << 23; | 166 | 80 | bool allow_direct_mapping = (max_key >= min_key && max_key - min_key < MAX_MAPPING_RANGE - 1); | 167 | 80 | if (allow_direct_mapping) { | 168 | 80 | for (const auto& variant_ptr : variant_ptrs) { | 169 | 80 | variant_ptr->method_variant.emplace<DirectPrimaryTypeHashTableContext<FieldType>>( | 170 | 80 | max_key, min_key); | 171 | 80 | } | 172 | 80 | } | 173 | 80 | } |
_ZN5doris25primary_to_direct_mappingINS_15MethodOneNumberIjNS_13JoinHashTableIjNS_17HashCRC32Return32IjEELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Line | Count | Source | 135 | 323 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { | 136 | 323 | using FieldType = typename Method::Base::Key; | 137 | 323 | FieldType max_key = std::numeric_limits<FieldType>::min(); | 138 | 323 | FieldType min_key = std::numeric_limits<FieldType>::max(); | 139 | | | 140 | 323 | size_t num_rows = key_columns[0]->size(); | 141 | 323 | if (is_column_nullable(*key_columns[0])) { | 142 | 0 | const FieldType* input_keys = (FieldType*)assert_cast<const ColumnNullable*>(key_columns[0]) | 143 | 0 | ->get_nested_column_ptr() | 144 | 0 | ->get_raw_data() | 145 | 0 | .data; | 146 | 0 | const NullMap& null_map = | 147 | 0 | assert_cast<const ColumnNullable*>(key_columns[0])->get_null_map_data(); | 148 | | // skip first mocked row | 149 | 0 | for (size_t i = 1; i < num_rows; i++) { | 150 | 0 | if (null_map[i]) { | 151 | 0 | continue; | 152 | 0 | } | 153 | 0 | max_key = std::max(max_key, input_keys[i]); | 154 | 0 | min_key = std::min(min_key, input_keys[i]); | 155 | 0 | } | 156 | 323 | } else { | 157 | 323 | const FieldType* input_keys = (FieldType*)key_columns[0]->get_raw_data().data; | 158 | | // skip first mocked row | 159 | 992 | for (size_t i = 1; i < num_rows; i++) { | 160 | 669 | max_key = std::max(max_key, input_keys[i]); | 161 | 669 | min_key = std::min(min_key, input_keys[i]); | 162 | 669 | } | 163 | 323 | } | 164 | | | 165 | 323 | constexpr auto MAX_MAPPING_RANGE = 1 << 23; | 166 | 323 | bool allow_direct_mapping = (max_key >= min_key && max_key - min_key < MAX_MAPPING_RANGE - 1); | 167 | 323 | if (allow_direct_mapping) { | 168 | 324 | for (const auto& variant_ptr : variant_ptrs) { | 169 | 324 | variant_ptr->method_variant.emplace<DirectPrimaryTypeHashTableContext<FieldType>>( | 170 | 324 | max_key, min_key); | 171 | 324 | } | 172 | 323 | } | 173 | 323 | } |
_ZN5doris25primary_to_direct_mappingINS_15MethodOneNumberImNS_13JoinHashTableImNS_17HashCRC32Return32ImEELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Line | Count | Source | 135 | 800 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { | 136 | 800 | using FieldType = typename Method::Base::Key; | 137 | 800 | FieldType max_key = std::numeric_limits<FieldType>::min(); | 138 | 800 | FieldType min_key = std::numeric_limits<FieldType>::max(); | 139 | | | 140 | 800 | size_t num_rows = key_columns[0]->size(); | 141 | 800 | if (is_column_nullable(*key_columns[0])) { | 142 | 0 | const FieldType* input_keys = (FieldType*)assert_cast<const ColumnNullable*>(key_columns[0]) | 143 | 0 | ->get_nested_column_ptr() | 144 | 0 | ->get_raw_data() | 145 | 0 | .data; | 146 | 0 | const NullMap& null_map = | 147 | 0 | assert_cast<const ColumnNullable*>(key_columns[0])->get_null_map_data(); | 148 | | // skip first mocked row | 149 | 0 | for (size_t i = 1; i < num_rows; i++) { | 150 | 0 | if (null_map[i]) { | 151 | 0 | continue; | 152 | 0 | } | 153 | 0 | max_key = std::max(max_key, input_keys[i]); | 154 | 0 | min_key = std::min(min_key, input_keys[i]); | 155 | 0 | } | 156 | 800 | } else { | 157 | 800 | const FieldType* input_keys = (FieldType*)key_columns[0]->get_raw_data().data; | 158 | | // skip first mocked row | 159 | 2.40k | for (size_t i = 1; i < num_rows; i++) { | 160 | 1.60k | max_key = std::max(max_key, input_keys[i]); | 161 | 1.60k | min_key = std::min(min_key, input_keys[i]); | 162 | 1.60k | } | 163 | 800 | } | 164 | | | 165 | 800 | constexpr auto MAX_MAPPING_RANGE = 1 << 23; | 166 | 800 | bool allow_direct_mapping = (max_key >= min_key && max_key - min_key < MAX_MAPPING_RANGE - 1); | 167 | 800 | if (allow_direct_mapping) { | 168 | 800 | for (const auto& variant_ptr : variant_ptrs) { | 169 | 800 | variant_ptr->method_variant.emplace<DirectPrimaryTypeHashTableContext<FieldType>>( | 170 | 800 | max_key, min_key); | 171 | 800 | } | 172 | 800 | } | 173 | 800 | } |
_ZN5doris25primary_to_direct_mappingINS_15MethodOneNumberIN4wide7integerILm128EjEENS_13JoinHashTableIS4_NS_17HashCRC32Return32IS4_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISF_EERKSC_ISt10shared_ptrINS_16JoinDataVariantsEESaISM_EE Line | Count | Source | 135 | 240 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { | 136 | 240 | using FieldType = typename Method::Base::Key; | 137 | 240 | FieldType max_key = std::numeric_limits<FieldType>::min(); | 138 | 240 | FieldType min_key = std::numeric_limits<FieldType>::max(); | 139 | | | 140 | 240 | size_t num_rows = key_columns[0]->size(); | 141 | 240 | if (is_column_nullable(*key_columns[0])) { | 142 | 0 | const FieldType* input_keys = (FieldType*)assert_cast<const ColumnNullable*>(key_columns[0]) | 143 | 0 | ->get_nested_column_ptr() | 144 | 0 | ->get_raw_data() | 145 | 0 | .data; | 146 | 0 | const NullMap& null_map = | 147 | 0 | assert_cast<const ColumnNullable*>(key_columns[0])->get_null_map_data(); | 148 | | // skip first mocked row | 149 | 0 | for (size_t i = 1; i < num_rows; i++) { | 150 | 0 | if (null_map[i]) { | 151 | 0 | continue; | 152 | 0 | } | 153 | 0 | max_key = std::max(max_key, input_keys[i]); | 154 | 0 | min_key = std::min(min_key, input_keys[i]); | 155 | 0 | } | 156 | 240 | } else { | 157 | 240 | const FieldType* input_keys = (FieldType*)key_columns[0]->get_raw_data().data; | 158 | | // skip first mocked row | 159 | 720 | for (size_t i = 1; i < num_rows; i++) { | 160 | 480 | max_key = std::max(max_key, input_keys[i]); | 161 | 480 | min_key = std::min(min_key, input_keys[i]); | 162 | 480 | } | 163 | 240 | } | 164 | | | 165 | 240 | constexpr auto MAX_MAPPING_RANGE = 1 << 23; | 166 | 240 | bool allow_direct_mapping = (max_key >= min_key && max_key - min_key < MAX_MAPPING_RANGE - 1); | 167 | 240 | if (allow_direct_mapping) { | 168 | 240 | for (const auto& variant_ptr : variant_ptrs) { | 169 | 240 | variant_ptr->method_variant.emplace<DirectPrimaryTypeHashTableContext<FieldType>>( | 170 | 240 | max_key, min_key); | 171 | 240 | } | 172 | 240 | } | 173 | 240 | } |
|
174 | | |
175 | | template <typename Method> |
176 | | void try_convert_to_direct_mapping( |
177 | | Method* method, const ColumnRawPtrs& key_columns, |
178 | 46.4k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {}Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingISt9monostateEEvPT_RKSt6vectorIPKNS_7IColumnESaIS7_EERKS4_ISt10shared_ptrINS_16JoinDataVariantsEESaISE_EE _ZN5doris29try_convert_to_direct_mappingINS_16MethodSerializedINS_13JoinHashTableINS_9StringRefENS_17HashCRC32Return32IS3_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISD_EERKSA_ISt10shared_ptrINS_16JoinDataVariantsEESaISK_EE Line | Count | Source | 178 | 14.1k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_15MethodOneNumberIN4wide7integerILm256EjEENS_13JoinHashTableIS4_NS_17HashCRC32Return32IS4_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISF_EERKSC_ISt10shared_ptrINS_16JoinDataVariantsEESaISM_EE Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_21MethodOneNumberDirectIhNS_13JoinHashTableIhNS_17HashCRC32Return32IhEELb1EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_21MethodOneNumberDirectItNS_13JoinHashTableItNS_17HashCRC32Return32ItEELb1EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_21MethodOneNumberDirectIjNS_13JoinHashTableIjNS_17HashCRC32Return32IjEELb1EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_21MethodOneNumberDirectImNS_13JoinHashTableImNS_17HashCRC32Return32ImEELb1EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_21MethodOneNumberDirectIN4wide7integerILm128EjEENS_13JoinHashTableIS4_NS_17HashCRC32Return32IS4_EELb1EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISF_EERKSC_ISt10shared_ptrINS_16JoinDataVariantsEESaISM_EE _ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableImNS_17HashCRC32Return32ImEELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISC_EERKS9_ISt10shared_ptrINS_16JoinDataVariantsEESaISJ_EE Line | Count | Source | 178 | 3.92k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
_ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableINS_6UInt72ENS_17HashCRC32Return32IS3_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISD_EERKSA_ISt10shared_ptrINS_16JoinDataVariantsEESaISK_EE Line | Count | Source | 178 | 3.20k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
_ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableINS_6UInt96ENS_17HashCRC32Return32IS3_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISD_EERKSA_ISt10shared_ptrINS_16JoinDataVariantsEESaISK_EE Line | Count | Source | 178 | 8.00k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
Unexecuted instantiation: _ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableINS_7UInt104ENS_17HashCRC32Return32IS3_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISD_EERKSA_ISt10shared_ptrINS_16JoinDataVariantsEESaISK_EE _ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableIN4wide7integerILm128EjEENS_17HashCRC32Return32IS5_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISF_EERKSC_ISt10shared_ptrINS_16JoinDataVariantsEESaISM_EE Line | Count | Source | 178 | 8.00k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
_ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableINS_7UInt136ENS_17HashCRC32Return32IS3_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISD_EERKSA_ISt10shared_ptrINS_16JoinDataVariantsEESaISK_EE Line | Count | Source | 178 | 960 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
_ZN5doris29try_convert_to_direct_mappingINS_15MethodKeysFixedINS_13JoinHashTableIN4wide7integerILm256EjEENS_17HashCRC32Return32IS5_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISF_EERKSC_ISt10shared_ptrINS_16JoinDataVariantsEESaISM_EE Line | Count | Source | 178 | 7.92k | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
_ZN5doris29try_convert_to_direct_mappingINS_19MethodStringNoCacheINS_13JoinHashTableINS_9StringRefENS_17HashCRC32Return32IS3_EELb0EEEEEEEvPT_RKSt6vectorIPKNS_7IColumnESaISD_EERKSA_ISt10shared_ptrINS_16JoinDataVariantsEESaISK_EE Line | Count | Source | 178 | 324 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) {} |
|
179 | | |
180 | | inline void try_convert_to_direct_mapping( |
181 | | PrimaryTypeHashTableContext<UInt8>* context, const ColumnRawPtrs& key_columns, |
182 | 160 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { |
183 | 160 | primary_to_direct_mapping(context, key_columns, variant_ptrs); |
184 | 160 | } |
185 | | |
186 | | inline void try_convert_to_direct_mapping( |
187 | | PrimaryTypeHashTableContext<UInt16>* context, const ColumnRawPtrs& key_columns, |
188 | 80 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { |
189 | 80 | primary_to_direct_mapping(context, key_columns, variant_ptrs); |
190 | 80 | } |
191 | | |
192 | | inline void try_convert_to_direct_mapping( |
193 | | PrimaryTypeHashTableContext<UInt32>* context, const ColumnRawPtrs& key_columns, |
194 | 323 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { |
195 | 323 | primary_to_direct_mapping(context, key_columns, variant_ptrs); |
196 | 323 | } |
197 | | |
198 | | inline void try_convert_to_direct_mapping( |
199 | | PrimaryTypeHashTableContext<UInt64>* context, const ColumnRawPtrs& key_columns, |
200 | 800 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { |
201 | 800 | primary_to_direct_mapping(context, key_columns, variant_ptrs); |
202 | 800 | } |
203 | | |
204 | | inline void try_convert_to_direct_mapping( |
205 | | PrimaryTypeHashTableContext<UInt128>* context, const ColumnRawPtrs& key_columns, |
206 | 240 | const std::vector<std::shared_ptr<JoinDataVariants>>& variant_ptrs) { |
207 | 240 | primary_to_direct_mapping(context, key_columns, variant_ptrs); |
208 | 240 | } |
209 | | |
210 | | } // namespace doris |