datasketches 0.2.3 → 0.2.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +12 -0
- data/README.md +8 -8
- data/ext/datasketches/kll_wrapper.cpp +7 -3
- data/ext/datasketches/theta_wrapper.cpp +20 -4
- data/lib/datasketches/version.rb +1 -1
- data/vendor/datasketches-cpp/CMakeLists.txt +25 -5
- data/vendor/datasketches-cpp/MANIFEST.in +3 -0
- data/vendor/datasketches-cpp/NOTICE +6 -5
- data/vendor/datasketches-cpp/README.md +76 -9
- data/vendor/datasketches-cpp/cmake/DataSketchesConfig.cmake.in +10 -0
- data/vendor/datasketches-cpp/common/CMakeLists.txt +18 -13
- data/vendor/datasketches-cpp/common/include/binomial_bounds.hpp +1 -0
- data/vendor/datasketches-cpp/common/include/common_defs.hpp +14 -0
- data/vendor/datasketches-cpp/{kll → common}/include/kolmogorov_smirnov.hpp +5 -3
- data/vendor/datasketches-cpp/{kll → common}/include/kolmogorov_smirnov_impl.hpp +13 -16
- data/vendor/datasketches-cpp/common/include/quantile_sketch_sorted_view.hpp +121 -0
- data/vendor/datasketches-cpp/common/include/quantile_sketch_sorted_view_impl.hpp +91 -0
- data/vendor/datasketches-cpp/common/test/test_type.hpp +2 -0
- data/vendor/datasketches-cpp/cpc/CMakeLists.txt +15 -35
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor_impl.hpp +3 -1
- data/vendor/datasketches-cpp/cpc/include/cpc_confidence.hpp +1 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch_impl.hpp +5 -3
- data/vendor/datasketches-cpp/cpc/include/cpc_union_impl.hpp +10 -6
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_allocation_test.cpp +17 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_test.cpp +1 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_union_test.cpp +2 -0
- data/vendor/datasketches-cpp/fi/CMakeLists.txt +5 -15
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch.hpp +37 -5
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch_impl.hpp +29 -11
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_custom_type_test.cpp +2 -1
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_test.cpp +1 -0
- data/vendor/datasketches-cpp/hll/CMakeLists.txt +33 -56
- data/vendor/datasketches-cpp/hll/include/AuxHashMap-internal.hpp +2 -0
- data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable-internal.hpp +1 -0
- data/vendor/datasketches-cpp/hll/include/CouponHashSet-internal.hpp +2 -2
- data/vendor/datasketches-cpp/hll/include/CouponList-internal.hpp +1 -0
- data/vendor/datasketches-cpp/hll/include/HllSketch-internal.hpp +6 -4
- data/vendor/datasketches-cpp/hll/include/HllSketchImpl-internal.hpp +2 -0
- data/vendor/datasketches-cpp/hll/include/HllSketchImplFactory.hpp +2 -0
- data/vendor/datasketches-cpp/hll/test/AuxHashMapTest.cpp +1 -0
- data/vendor/datasketches-cpp/hll/test/CouponHashSetTest.cpp +1 -0
- data/vendor/datasketches-cpp/hll/test/CouponListTest.cpp +2 -0
- data/vendor/datasketches-cpp/hll/test/HllArrayTest.cpp +1 -0
- data/vendor/datasketches-cpp/hll/test/HllSketchTest.cpp +59 -0
- data/vendor/datasketches-cpp/hll/test/HllUnionTest.cpp +2 -0
- data/vendor/datasketches-cpp/hll/test/TablesTest.cpp +1 -0
- data/vendor/datasketches-cpp/kll/CMakeLists.txt +5 -19
- data/vendor/datasketches-cpp/kll/include/kll_helper.hpp +0 -4
- data/vendor/datasketches-cpp/kll/include/kll_helper_impl.hpp +5 -2
- data/vendor/datasketches-cpp/kll/include/kll_sketch.hpp +108 -41
- data/vendor/datasketches-cpp/kll/include/kll_sketch_impl.hpp +150 -132
- data/vendor/datasketches-cpp/kll/test/kll_sketch_test.cpp +165 -31
- data/vendor/datasketches-cpp/kll/test/kolmogorov_smirnov_test.cpp +1 -1
- data/vendor/datasketches-cpp/pyproject.toml +1 -1
- data/vendor/datasketches-cpp/python/CMakeLists.txt +8 -1
- data/vendor/datasketches-cpp/python/README.md +13 -9
- data/vendor/datasketches-cpp/python/src/datasketches.cpp +4 -0
- data/vendor/datasketches-cpp/python/src/fi_wrapper.cpp +6 -1
- data/vendor/datasketches-cpp/python/src/kll_wrapper.cpp +48 -13
- data/vendor/datasketches-cpp/python/src/ks_wrapper.cpp +68 -0
- data/vendor/datasketches-cpp/python/src/quantiles_wrapper.cpp +240 -0
- data/vendor/datasketches-cpp/python/src/req_wrapper.cpp +9 -2
- data/vendor/datasketches-cpp/python/src/vector_of_kll.cpp +1 -0
- data/vendor/datasketches-cpp/python/tests/kll_test.py +10 -4
- data/vendor/datasketches-cpp/python/tests/quantiles_test.py +126 -0
- data/vendor/datasketches-cpp/quantiles/CMakeLists.txt +42 -0
- data/vendor/datasketches-cpp/quantiles/include/quantiles_sketch.hpp +656 -0
- data/vendor/datasketches-cpp/quantiles/include/quantiles_sketch_impl.hpp +1373 -0
- data/vendor/datasketches-cpp/quantiles/test/CMakeLists.txt +44 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n1000_v0.3.0.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n1000_v0.6.0.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n1000_v0.8.0.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n1000_v0.8.3.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n50_v0.3.0.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n50_v0.6.0.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n50_v0.8.0.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/Qk128_n50_v0.8.3.sk +0 -0
- data/vendor/datasketches-cpp/quantiles/test/kolmogorov_smirnov_test.cpp +110 -0
- data/vendor/datasketches-cpp/quantiles/test/quantiles_compatibility_test.cpp +129 -0
- data/vendor/datasketches-cpp/quantiles/test/quantiles_sketch_test.cpp +975 -0
- data/vendor/datasketches-cpp/req/CMakeLists.txt +6 -21
- data/vendor/datasketches-cpp/req/include/req_common.hpp +0 -5
- data/vendor/datasketches-cpp/req/include/req_compactor.hpp +6 -0
- data/vendor/datasketches-cpp/req/include/req_compactor_impl.hpp +30 -2
- data/vendor/datasketches-cpp/req/include/req_sketch.hpp +73 -23
- data/vendor/datasketches-cpp/req/include/req_sketch_impl.hpp +95 -63
- data/vendor/datasketches-cpp/req/test/req_sketch_test.cpp +74 -3
- data/vendor/datasketches-cpp/sampling/CMakeLists.txt +5 -9
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch.hpp +44 -7
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch_impl.hpp +44 -33
- data/vendor/datasketches-cpp/sampling/include/var_opt_union.hpp +41 -6
- data/vendor/datasketches-cpp/sampling/include/var_opt_union_impl.hpp +33 -15
- data/vendor/datasketches-cpp/sampling/test/var_opt_allocation_test.cpp +2 -2
- data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_test.cpp +1 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_union_test.cpp +1 -0
- data/vendor/datasketches-cpp/setup.py +1 -1
- data/vendor/datasketches-cpp/theta/CMakeLists.txt +26 -45
- data/vendor/datasketches-cpp/theta/include/bounds_on_ratios_in_sampled_sets.hpp +1 -0
- data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser_impl.hpp +92 -23
- data/vendor/datasketches-cpp/theta/include/theta_helpers.hpp +15 -0
- data/vendor/datasketches-cpp/theta/include/theta_intersection_base_impl.hpp +7 -6
- data/vendor/datasketches-cpp/theta/include/theta_set_difference_base_impl.hpp +3 -2
- data/vendor/datasketches-cpp/theta/include/theta_sketch.hpp +32 -15
- data/vendor/datasketches-cpp/theta/include/theta_sketch_impl.hpp +150 -93
- data/vendor/datasketches-cpp/theta/include/theta_union.hpp +6 -1
- data/vendor/datasketches-cpp/theta/include/theta_union_base.hpp +3 -1
- data/vendor/datasketches-cpp/theta/include/theta_union_base_impl.hpp +9 -2
- data/vendor/datasketches-cpp/theta/include/theta_union_impl.hpp +8 -5
- data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base.hpp +5 -4
- data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base_impl.hpp +34 -9
- data/vendor/datasketches-cpp/theta/test/CMakeLists.txt +1 -0
- data/vendor/datasketches-cpp/theta/test/theta_a_not_b_test.cpp +2 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java_v1.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java_v2.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java_v1.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java_v2.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_exact_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_intersection_test.cpp +2 -0
- data/vendor/datasketches-cpp/theta/test/theta_setop_test.cpp +446 -0
- data/vendor/datasketches-cpp/theta/test/theta_sketch_test.cpp +429 -1
- data/vendor/datasketches-cpp/theta/test/theta_union_test.cpp +25 -11
- data/vendor/datasketches-cpp/tuple/CMakeLists.txt +18 -33
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch.hpp +1 -1
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch_impl.hpp +3 -3
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union.hpp +1 -1
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union_impl.hpp +3 -3
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch.hpp +29 -9
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch_impl.hpp +34 -14
- data/vendor/datasketches-cpp/tuple/include/tuple_union.hpp +6 -1
- data/vendor/datasketches-cpp/tuple/include/tuple_union_impl.hpp +8 -3
- data/vendor/datasketches-cpp/tuple/test/array_of_doubles_sketch_test.cpp +16 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_a_not_b_test.cpp +1 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_intersection_test.cpp +1 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_test.cpp +46 -8
- data/vendor/datasketches-cpp/tuple/test/tuple_union_test.cpp +8 -0
- metadata +33 -12
- data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator.hpp +0 -75
- data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator_impl.hpp +0 -184
- data/vendor/datasketches-cpp/req/include/req_quantile_calculator.hpp +0 -69
- data/vendor/datasketches-cpp/req/include/req_quantile_calculator_impl.hpp +0 -60
- data/vendor/datasketches-cpp/theta/test/theta_update_empty_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_update_estimation_from_java.sk +0 -0
@@ -32,29 +32,14 @@ target_include_directories(req
|
|
32
32
|
target_link_libraries(req INTERFACE common)
|
33
33
|
target_compile_features(req INTERFACE cxx_std_11)
|
34
34
|
|
35
|
-
set(req_HEADERS "")
|
36
|
-
list(APPEND req_HEADERS "include/req_common.hpp")
|
37
|
-
list(APPEND req_HEADERS "include/req_sketch.hpp")
|
38
|
-
list(APPEND req_HEADERS "include/req_sketch_impl.hpp")
|
39
|
-
list(APPEND req_HEADERS "include/req_compactor.hpp")
|
40
|
-
list(APPEND req_HEADERS "include/req_compactor_impl.hpp")
|
41
|
-
list(APPEND req_HEADERS "include/req_quantile_calculator.hpp")
|
42
|
-
list(APPEND req_HEADERS "include/req_quantile_calculator_impl.hpp")
|
43
|
-
|
44
35
|
install(TARGETS req
|
45
36
|
EXPORT ${PROJECT_NAME}
|
46
37
|
)
|
47
38
|
|
48
|
-
install(FILES
|
39
|
+
install(FILES
|
40
|
+
include/req_common.hpp
|
41
|
+
include/req_sketch.hpp
|
42
|
+
include/req_sketch_impl.hpp
|
43
|
+
include/req_compactor.hpp
|
44
|
+
include/req_compactor_impl.hpp
|
49
45
|
DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}/DataSketches")
|
50
|
-
|
51
|
-
target_sources(req
|
52
|
-
INTERFACE
|
53
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_common.hpp
|
54
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_sketch.hpp
|
55
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_sketch_impl.hpp
|
56
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_compactor.hpp
|
57
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_compactor_impl.hpp
|
58
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_quantile_calculator.hpp
|
59
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/req_quantile_calculator_impl.hpp
|
60
|
-
)
|
@@ -21,17 +21,12 @@
|
|
21
21
|
#define REQ_COMMON_HPP_
|
22
22
|
|
23
23
|
#include <random>
|
24
|
-
#include <chrono>
|
25
24
|
|
26
25
|
#include "serde.hpp"
|
27
26
|
#include "common_defs.hpp"
|
28
27
|
|
29
28
|
namespace datasketches {
|
30
29
|
|
31
|
-
// TODO: have a common random bit with KLL
|
32
|
-
static std::independent_bits_engine<std::mt19937, 1, unsigned>
|
33
|
-
req_random_bit(static_cast<unsigned>(std::chrono::system_clock::now().time_since_epoch().count()));
|
34
|
-
|
35
30
|
namespace req_constants {
|
36
31
|
static const uint16_t MIN_K = 4;
|
37
32
|
static const uint8_t INIT_NUM_SECTIONS = 3;
|
@@ -38,6 +38,9 @@ public:
|
|
38
38
|
req_compactor& operator=(const req_compactor& other);
|
39
39
|
req_compactor& operator=(req_compactor&& other);
|
40
40
|
|
41
|
+
template<typename TT, typename CC, typename AA>
|
42
|
+
req_compactor(const req_compactor<TT, CC, AA>& other, const Allocator& allocator);
|
43
|
+
|
41
44
|
bool is_sorted() const;
|
42
45
|
uint32_t get_num_items() const;
|
43
46
|
uint32_t get_nom_capacity() const;
|
@@ -128,6 +131,9 @@ private:
|
|
128
131
|
template<typename S>
|
129
132
|
static std::pair<std::unique_ptr<T, items_deleter>, size_t> deserialize_items(const void* bytes, size_t size, const S& serde, const Allocator& allocator, uint32_t num);
|
130
133
|
|
134
|
+
// for type converting constructor
|
135
|
+
template<typename TT, typename CC, typename AA>
|
136
|
+
friend class req_compactor;
|
131
137
|
};
|
132
138
|
|
133
139
|
} /* namespace datasketches */
|
@@ -26,6 +26,7 @@
|
|
26
26
|
|
27
27
|
#include "count_zeros.hpp"
|
28
28
|
#include "conditional_forward.hpp"
|
29
|
+
#include "common_defs.hpp"
|
29
30
|
|
30
31
|
#include <iomanip>
|
31
32
|
|
@@ -131,6 +132,33 @@ req_compactor<T, C, A>& req_compactor<T, C, A>::operator=(req_compactor&& other)
|
|
131
132
|
return *this;
|
132
133
|
}
|
133
134
|
|
135
|
+
template<typename T, typename C, typename A>
|
136
|
+
template<typename TT, typename CC, typename AA>
|
137
|
+
req_compactor<T, C, A>::req_compactor(const req_compactor<TT, CC, AA>& other, const A& allocator):
|
138
|
+
allocator_(allocator),
|
139
|
+
lg_weight_(other.lg_weight_),
|
140
|
+
hra_(other.hra_),
|
141
|
+
coin_(other.coin_),
|
142
|
+
sorted_(other.sorted_),
|
143
|
+
section_size_raw_(other.section_size_raw_),
|
144
|
+
section_size_(other.section_size_),
|
145
|
+
num_sections_(other.num_sections_),
|
146
|
+
state_(other.state_),
|
147
|
+
num_items_(other.num_items_),
|
148
|
+
capacity_(other.capacity_),
|
149
|
+
items_(nullptr)
|
150
|
+
{
|
151
|
+
if (other.items_ != nullptr) {
|
152
|
+
items_ = allocator_.allocate(capacity_);
|
153
|
+
const uint32_t from = hra_ ? capacity_ - num_items_ : 0;
|
154
|
+
const uint32_t to = hra_ ? capacity_ : num_items_;
|
155
|
+
for (uint32_t i = from; i < to; ++i) new (items_ + i) T(other.items_[i]);
|
156
|
+
if (sorted_ && !std::is_sorted(items_ + from, items_ + to, C())) {
|
157
|
+
throw std::logic_error("items must be sorted");
|
158
|
+
}
|
159
|
+
}
|
160
|
+
}
|
161
|
+
|
134
162
|
template<typename T, typename C, typename A>
|
135
163
|
bool req_compactor<T, C, A>::is_sorted() const {
|
136
164
|
return sorted_;
|
@@ -245,7 +273,7 @@ std::pair<uint32_t, uint32_t> req_compactor<T, C, A>::compact(req_compactor& nex
|
|
245
273
|
if (compaction_range.second - compaction_range.first < 2) throw std::logic_error("compaction range error");
|
246
274
|
|
247
275
|
if ((state_ & 1) == 1) { coin_ = !coin_; } // for odd flip coin;
|
248
|
-
else { coin_ =
|
276
|
+
else { coin_ = random_bit(); } // random coin flip
|
249
277
|
|
250
278
|
const auto num = (compaction_range.second - compaction_range.first) / 2;
|
251
279
|
next.ensure_space(num);
|
@@ -451,7 +479,7 @@ req_compactor<T, C, A>::req_compactor(bool hra, uint8_t lg_weight, bool sorted,
|
|
451
479
|
allocator_(allocator),
|
452
480
|
lg_weight_(lg_weight),
|
453
481
|
hra_(hra),
|
454
|
-
coin_(
|
482
|
+
coin_(random_bit()),
|
455
483
|
sorted_(sorted),
|
456
484
|
section_size_raw_(section_size_raw),
|
457
485
|
section_size_(nearest_even(section_size_raw)),
|
@@ -22,22 +22,25 @@
|
|
22
22
|
|
23
23
|
#include "req_common.hpp"
|
24
24
|
#include "req_compactor.hpp"
|
25
|
-
#include "
|
25
|
+
#include "quantile_sketch_sorted_view.hpp"
|
26
|
+
|
27
|
+
#include <stdexcept>
|
26
28
|
|
27
29
|
namespace datasketches {
|
28
30
|
|
29
31
|
template<
|
30
32
|
typename T,
|
31
|
-
typename Comparator = std::less<T>,
|
32
|
-
typename
|
33
|
+
typename Comparator = std::less<T>, // strict weak ordering function (see C++ named requirements: Compare)
|
34
|
+
typename S = serde<T>, // deprecated, to be removed in the next major version
|
33
35
|
typename Allocator = std::allocator<T>
|
34
36
|
>
|
35
37
|
class req_sketch {
|
36
38
|
public:
|
39
|
+
using value_type = T;
|
40
|
+
using comparator = Comparator;
|
37
41
|
using Compactor = req_compactor<T, Comparator, Allocator>;
|
38
42
|
using AllocCompactor = typename std::allocator_traits<Allocator>::template rebind_alloc<Compactor>;
|
39
|
-
using
|
40
|
-
using vector_double = std::vector<double, AllocDouble>;
|
43
|
+
using vector_double = std::vector<double, typename std::allocator_traits<Allocator>::template rebind_alloc<double>>;
|
41
44
|
|
42
45
|
/**
|
43
46
|
* Constructor
|
@@ -55,6 +58,14 @@ public:
|
|
55
58
|
req_sketch& operator=(const req_sketch& other);
|
56
59
|
req_sketch& operator=(req_sketch&& other);
|
57
60
|
|
61
|
+
/*
|
62
|
+
* Type converting constructor.
|
63
|
+
* @param other sketch of a different type
|
64
|
+
* @param allocator instance of an Allocator
|
65
|
+
*/
|
66
|
+
template<typename TT, typename CC, typename SS, typename AA>
|
67
|
+
explicit req_sketch(const req_sketch<TT, CC, SS, AA>& other, const Allocator& allocator = Allocator());
|
68
|
+
|
58
69
|
/**
|
59
70
|
* Returns configured parameter K
|
60
71
|
* @return parameter K
|
@@ -113,6 +124,12 @@ public:
|
|
113
124
|
*/
|
114
125
|
const T& get_max_value() const;
|
115
126
|
|
127
|
+
/**
|
128
|
+
* Returns an instance of the comparator for this sketch.
|
129
|
+
* @return comparator
|
130
|
+
*/
|
131
|
+
Comparator get_comparator() const;
|
132
|
+
|
116
133
|
/**
|
117
134
|
* Returns an approximation to the normalized (fractional) rank of the given item from 0 to 1 inclusive.
|
118
135
|
* With the template parameter inclusive=true the weight of the given item is included into the rank.
|
@@ -123,7 +140,6 @@ public:
|
|
123
140
|
* @param item to be ranked
|
124
141
|
* @return an approximate rank of the given item
|
125
142
|
*/
|
126
|
-
|
127
143
|
template<bool inclusive = false>
|
128
144
|
double get_rank(const T& item) const;
|
129
145
|
|
@@ -135,9 +151,10 @@ public:
|
|
135
151
|
*
|
136
152
|
* @param split_points an array of <i>m</i> unique, monotonically increasing values
|
137
153
|
* that divide the input domain into <i>m+1</i> consecutive disjoint intervals.
|
138
|
-
*
|
139
|
-
*
|
140
|
-
* the
|
154
|
+
* If the template parameter inclusive=false, the definition of an "interval" is inclusive of the left split point and exclusive of the right
|
155
|
+
* split point, with the exception that the last interval will include the maximum value.
|
156
|
+
* If the template parameter inclusive=true, the definition of an "interval" is exclusive of the left split point and inclusive of the right
|
157
|
+
* split point.
|
141
158
|
* It is not necessary to include either the min or max values in these split points.
|
142
159
|
*
|
143
160
|
* @return an array of m+1 doubles each of which is an approximation
|
@@ -178,8 +195,9 @@ public:
|
|
178
195
|
* @param rank the given normalized rank
|
179
196
|
* @return approximate quantile given the normalized rank
|
180
197
|
*/
|
198
|
+
using quantile_return_type = typename quantile_sketch_sorted_view<T, Comparator, Allocator>::quantile_return_type;
|
181
199
|
template<bool inclusive = false>
|
182
|
-
|
200
|
+
quantile_return_type get_quantile(double rank) const;
|
183
201
|
|
184
202
|
/**
|
185
203
|
* Returns an array of quantiles that correspond to the given array of normalized ranks.
|
@@ -221,24 +239,28 @@ public:
|
|
221
239
|
/**
|
222
240
|
* Computes size needed to serialize the current state of the sketch.
|
223
241
|
* This version is for fixed-size arithmetic types (integral and floating point).
|
242
|
+
* @param instance of a SerDe
|
224
243
|
* @return size in bytes needed to serialize this sketch
|
225
244
|
*/
|
226
|
-
template<typename TT = T, typename std::enable_if<std::is_arithmetic<TT>::value, int>::type = 0>
|
227
|
-
size_t get_serialized_size_bytes() const;
|
245
|
+
template<typename TT = T, typename SerDe = S, typename std::enable_if<std::is_arithmetic<TT>::value, int>::type = 0>
|
246
|
+
size_t get_serialized_size_bytes(const SerDe& sd = SerDe()) const;
|
228
247
|
|
229
248
|
/**
|
230
249
|
* Computes size needed to serialize the current state of the sketch.
|
231
250
|
* This version is for all other types and can be expensive since every item needs to be looked at.
|
251
|
+
* @param instance of a SerDe
|
232
252
|
* @return size in bytes needed to serialize this sketch
|
233
253
|
*/
|
234
|
-
template<typename TT = T, typename std::enable_if<!std::is_arithmetic<TT>::value, int>::type = 0>
|
235
|
-
size_t get_serialized_size_bytes() const;
|
254
|
+
template<typename TT = T, typename SerDe = S, typename std::enable_if<!std::is_arithmetic<TT>::value, int>::type = 0>
|
255
|
+
size_t get_serialized_size_bytes(const SerDe& sd = SerDe()) const;
|
236
256
|
|
237
257
|
/**
|
238
258
|
* This method serializes the sketch into a given stream in a binary form
|
239
259
|
* @param os output stream
|
260
|
+
* @param instance of a SerDe
|
240
261
|
*/
|
241
|
-
|
262
|
+
template<typename SerDe = S>
|
263
|
+
void serialize(std::ostream& os, const SerDe& sd = SerDe()) const;
|
242
264
|
|
243
265
|
// This is a convenience alias for users
|
244
266
|
// The type returned by the following serialize method
|
@@ -250,24 +272,53 @@ public:
|
|
250
272
|
* It is a blank space of a given size.
|
251
273
|
* This header is used in Datasketches PostgreSQL extension.
|
252
274
|
* @param header_size_bytes space to reserve in front of the sketch
|
275
|
+
* @param instance of a SerDe
|
253
276
|
*/
|
254
|
-
|
277
|
+
template<typename SerDe = S>
|
278
|
+
vector_bytes serialize(unsigned header_size_bytes = 0, const SerDe& sd = SerDe()) const;
|
255
279
|
|
256
280
|
/**
|
257
281
|
* This method deserializes a sketch from a given stream.
|
258
282
|
* @param is input stream
|
283
|
+
* @param instance of an Allocator
|
259
284
|
* @return an instance of a sketch
|
285
|
+
*
|
286
|
+
* Deprecated, to be removed in the next major version
|
260
287
|
*/
|
261
288
|
static req_sketch deserialize(std::istream& is, const Allocator& allocator = Allocator());
|
262
289
|
|
290
|
+
/**
|
291
|
+
* This method deserializes a sketch from a given stream.
|
292
|
+
* @param is input stream
|
293
|
+
* @param instance of a SerDe
|
294
|
+
* @param instance of an Allocator
|
295
|
+
* @return an instance of a sketch
|
296
|
+
*/
|
297
|
+
template<typename SerDe = S>
|
298
|
+
static req_sketch deserialize(std::istream& is, const SerDe& sd = SerDe(), const Allocator& allocator = Allocator());
|
299
|
+
|
263
300
|
/**
|
264
301
|
* This method deserializes a sketch from a given array of bytes.
|
265
302
|
* @param bytes pointer to the array of bytes
|
266
303
|
* @param size the size of the array
|
304
|
+
* @param instance of an Allocator
|
267
305
|
* @return an instance of a sketch
|
306
|
+
*
|
307
|
+
* Deprecated, to be removed in the next major version
|
268
308
|
*/
|
269
309
|
static req_sketch deserialize(const void* bytes, size_t size, const Allocator& allocator = Allocator());
|
270
310
|
|
311
|
+
/**
|
312
|
+
* This method deserializes a sketch from a given array of bytes.
|
313
|
+
* @param bytes pointer to the array of bytes
|
314
|
+
* @param size the size of the array
|
315
|
+
* @param instance of a SerDe
|
316
|
+
* @param instance of an Allocator
|
317
|
+
* @return an instance of a sketch
|
318
|
+
*/
|
319
|
+
template<typename SerDe = S>
|
320
|
+
static req_sketch deserialize(const void* bytes, size_t size, const SerDe& sd = SerDe(), const Allocator& allocator = Allocator());
|
321
|
+
|
271
322
|
/**
|
272
323
|
* Prints a summary of the sketch.
|
273
324
|
* @param print_levels if true include information about levels
|
@@ -279,6 +330,9 @@ public:
|
|
279
330
|
const_iterator begin() const;
|
280
331
|
const_iterator end() const;
|
281
332
|
|
333
|
+
template<bool inclusive = false>
|
334
|
+
quantile_sketch_sorted_view<T, Comparator, Allocator> get_sorted_view(bool cumulative) const;
|
335
|
+
|
282
336
|
private:
|
283
337
|
Allocator allocator_;
|
284
338
|
uint16_t k_;
|
@@ -310,13 +364,6 @@ private:
|
|
310
364
|
static double get_rank_ub(uint16_t k, uint8_t num_levels, double rank, uint8_t num_std_dev, uint64_t n, bool hra);
|
311
365
|
static bool is_exact_rank(uint16_t k, uint8_t num_levels, double rank, uint64_t n, bool hra);
|
312
366
|
|
313
|
-
using QuantileCalculator = req_quantile_calculator<T, Comparator, Allocator>;
|
314
|
-
using AllocCalc = typename std::allocator_traits<Allocator>::template rebind_alloc<QuantileCalculator>;
|
315
|
-
class calculator_deleter;
|
316
|
-
using QuantileCalculatorPtr = typename std::unique_ptr<QuantileCalculator, calculator_deleter>;
|
317
|
-
template<bool inclusive>
|
318
|
-
QuantileCalculatorPtr get_quantile_calculator() const;
|
319
|
-
|
320
367
|
// for deserialization
|
321
368
|
class item_deleter;
|
322
369
|
req_sketch(uint16_t k, bool hra, uint64_t n, std::unique_ptr<T, item_deleter> min_value, std::unique_ptr<T, item_deleter> max_value, std::vector<Compactor, AllocCompactor>&& compactors);
|
@@ -369,6 +416,9 @@ private:
|
|
369
416
|
}
|
370
417
|
}
|
371
418
|
|
419
|
+
// for type converting constructor
|
420
|
+
template<typename TT, typename CC, typename SS, typename AA>
|
421
|
+
friend class req_sketch;
|
372
422
|
};
|
373
423
|
|
374
424
|
template<typename T, typename C, typename S, typename A>
|