1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
|
#pragma once
#include "yql_cost_function.h"
#include <yql/essentials/core/minsketch/count_min_sketch.h>
#include <yql/essentials/core/histogram/eq_width_histogram.h>
#include <yql/essentials/core/cbo/cbo_interesting_orderings.h>
#include <library/cpp/json/json_reader.h>
#include <util/generic/vector.h>
#include <util/generic/hash.h>
#include <util/generic/hash_set.h>
#include <util/generic/maybe.h>
#include <util/string/builder.h>
#include <util/generic/string.h>
#include <optional>
#include <iostream>
namespace NYql {
enum EStatisticsType : ui32 {
BaseTable,
FilteredFactTable,
ManyManyJoin
};
enum EStorageType : ui32 {
NA,
RowStorage,
ColumnStorage
};
class TShufflingOrderingsByJoinLabels {
public:
void Add(TVector<TString> joinLabels, NDq::TOrderingsStateMachine::TLogicalOrderings shufflings) {
std::sort(joinLabels.begin(), joinLabels.end());
ShufflingOrderingsByJoinLabels_.emplace_back(joinLabels, shufflings);
}
TMaybe<NDq::TOrderingsStateMachine::TLogicalOrderings> GetShufflingOrderigsByJoinLabels(
TVector<TString> searchingLabels
) {
std::sort(searchingLabels.begin(), searchingLabels.end());
for (const auto& [joinLabels, shufflings]: ShufflingOrderingsByJoinLabels_) {
if (searchingLabels == joinLabels) {
return shufflings;
}
}
return Nothing();
}
TString ToString() const;
private:
TVector<std::pair<TVector<TString>, NDq::TOrderingsStateMachine::TLogicalOrderings>> ShufflingOrderingsByJoinLabels_;
};
// Providers may subclass this struct to associate specific statistics, useful to
// derive stats for higher-level operators in the plan.
struct IProviderStatistics {
virtual ~IProviderStatistics() {}
};
struct TColumnStatistics {
std::optional<double> NumUniqueVals;
std::optional<double> HyperLogLog;
std::shared_ptr<NKikimr::TCountMinSketch> CountMinSketch;
std::shared_ptr<NKikimr::TEqWidthHistogramEstimator> EqWidthHistogramEstimator;
TString Type;
TColumnStatistics() {}
};
/**
* Optimizer Statistics struct records per-table and per-column statistics
* for the current operator in the plan. Currently, only Nrows and Ncols are
* recorded.
* Cost is also included in statistics, as its updated concurrently with statistics
* all of the time.
*/
struct TOptimizerStatistics {
struct TKeyColumns : public TSimpleRefCount<TKeyColumns> {
TVector<TString> Data;
TKeyColumns(TVector<TString> data) : Data(std::move(data)) {}
TVector<NDq::TJoinColumn> ToJoinColumns(const TString& alias) {
TVector<NDq::TJoinColumn> columns;
columns.reserve(Data.size());
for (std::size_t i = 0; i < Data.size(); ++i) {
columns.push_back(NDq::TJoinColumn(alias, Data[i]));
}
return columns;
}
};
struct TSortColumns : public TSimpleRefCount<TSortColumns> {
TVector<TString> Columns;
TVector<TString> Aliases;
TSortColumns(const TVector<TString>& cols, const TVector<TString>& aliases)
: Columns(cols)
, Aliases(aliases)
{}
};
struct TColumnStatMap : public TSimpleRefCount<TColumnStatMap> {
THashMap<TString,TColumnStatistics> Data;
TColumnStatMap() {}
TColumnStatMap(THashMap<TString,TColumnStatistics> data) : Data(std::move(data)) {}
};
struct TShuffledByColumns : public TSimpleRefCount<TShuffledByColumns> {
TVector<NDq::TJoinColumn> Data;
TShuffledByColumns(TVector<NDq::TJoinColumn> data) : Data(std::move(data)) {}
TString ToString() {
TString result;
for (const auto& column: Data) {
result.append(column.RelName).append(".").append(column.AttributeName).append(", ");
}
if (!result.empty()) {
result.pop_back();
result.pop_back();
}
return result;
}
};
EStatisticsType Type = BaseTable;
double Nrows = 0;
int Ncols = 0;
double ByteSize = 0;
double Cost = 0;
double Selectivity = 1.0;
TIntrusivePtr<TKeyColumns> KeyColumns;
TIntrusivePtr<TColumnStatMap> ColumnStatistics;
TIntrusivePtr<TShuffledByColumns> ShuffledByColumns;
TIntrusivePtr<TSortColumns> SortColumns;
EStorageType StorageType = EStorageType::NA;
std::shared_ptr<IProviderStatistics> Specific;
std::shared_ptr<TVector<TString>> Labels = {};
TString SourceTableName;
TSimpleSharedPtr<THashSet<TString>> Aliases;
TIntrusivePtr<NDq::TTableAliasMap> TableAliases;
NDq::TOrderingsStateMachine::TLogicalOrderings LogicalOrderings;
NDq::TOrderingsStateMachine::TLogicalOrderings SortingOrderings;
NDq::TOrderingsStateMachine::TLogicalOrderings ReversedSortingOrderings;
std::optional<std::size_t> ShuffleOrderingIdx;
std::int64_t SortingOrderingIdx = -1;
std::int64_t ShufflingOrderingIdx = -1;
// special flag for equijoin
bool CBOFired = false;
TOptimizerStatistics(TOptimizerStatistics&&) = default;
TOptimizerStatistics& operator=(TOptimizerStatistics&&) = default;
TOptimizerStatistics(const TOptimizerStatistics&) = default;
TOptimizerStatistics& operator=(const TOptimizerStatistics&) = default;
TOptimizerStatistics() = default;
TOptimizerStatistics(
EStatisticsType type,
double nrows = 0.0,
int ncols = 0,
double byteSize = 0.0,
double cost = 0.0,
TIntrusivePtr<TKeyColumns> keyColumns = {},
TIntrusivePtr<TColumnStatMap> columnMap = {},
EStorageType storageType = EStorageType::NA,
std::shared_ptr<IProviderStatistics> specific = nullptr
);
TOptimizerStatistics& operator+=(const TOptimizerStatistics& other);
bool Empty() const;
friend std::ostream& operator<<(std::ostream& os, const TOptimizerStatistics& s);
TString ToString() const;
};
std::shared_ptr<TOptimizerStatistics> OverrideStatistics(const TOptimizerStatistics& s, const TStringBuf& tablePath, const std::shared_ptr<NJson::TJsonValue>& stats);
}
|
