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
@@ -25,6 +25,7 @@
|
|
25
25
|
#include <cmath>
|
26
26
|
#include <random>
|
27
27
|
#include <algorithm>
|
28
|
+
#include <stdexcept>
|
28
29
|
|
29
30
|
#include "var_opt_sketch.hpp"
|
30
31
|
#include "serde.hpp"
|
@@ -311,8 +312,8 @@ var_opt_sketch<T,S,A>& var_opt_sketch<T,S,A>::operator=(var_opt_sketch&& other)
|
|
311
312
|
|
312
313
|
// implementation for fixed-size arithmetic types (integral and floating point)
|
313
314
|
template<typename T, typename S, typename A>
|
314
|
-
template<typename TT, typename std::enable_if<std::is_arithmetic<TT>::value, int>::type>
|
315
|
-
size_t var_opt_sketch<T,S,A>::get_serialized_size_bytes() const {
|
315
|
+
template<typename TT, typename SerDe, typename std::enable_if<std::is_arithmetic<TT>::value, int>::type>
|
316
|
+
size_t var_opt_sketch<T,S,A>::get_serialized_size_bytes(const SerDe&) const {
|
316
317
|
if (is_empty()) { return PREAMBLE_LONGS_EMPTY << 3; }
|
317
318
|
size_t num_bytes = (r_ == 0 ? PREAMBLE_LONGS_WARMUP : PREAMBLE_LONGS_FULL) << 3;
|
318
319
|
num_bytes += h_ * sizeof(double); // weights
|
@@ -325,8 +326,8 @@ size_t var_opt_sketch<T,S,A>::get_serialized_size_bytes() const {
|
|
325
326
|
|
326
327
|
// implementation for all other types
|
327
328
|
template<typename T, typename S, typename A>
|
328
|
-
template<typename TT, typename std::enable_if<!std::is_arithmetic<TT>::value, int>::type>
|
329
|
-
size_t var_opt_sketch<T,S,A>::get_serialized_size_bytes() const {
|
329
|
+
template<typename TT, typename SerDe, typename std::enable_if<!std::is_arithmetic<TT>::value, int>::type>
|
330
|
+
size_t var_opt_sketch<T,S,A>::get_serialized_size_bytes(const SerDe& sd) const {
|
330
331
|
if (is_empty()) { return PREAMBLE_LONGS_EMPTY << 3; }
|
331
332
|
size_t num_bytes = (r_ == 0 ? PREAMBLE_LONGS_WARMUP : PREAMBLE_LONGS_FULL) << 3;
|
332
333
|
num_bytes += h_ * sizeof(double); // weights
|
@@ -335,13 +336,14 @@ size_t var_opt_sketch<T,S,A>::get_serialized_size_bytes() const {
|
|
335
336
|
}
|
336
337
|
// must iterate over the items
|
337
338
|
for (auto it: *this)
|
338
|
-
num_bytes +=
|
339
|
+
num_bytes += sd.size_of_item(it.first);
|
339
340
|
return num_bytes;
|
340
341
|
}
|
341
342
|
|
342
343
|
template<typename T, typename S, typename A>
|
343
|
-
|
344
|
-
|
344
|
+
template<typename SerDe>
|
345
|
+
std::vector<uint8_t, AllocU8<A>> var_opt_sketch<T,S,A>::serialize(unsigned header_size_bytes, const SerDe& sd) const {
|
346
|
+
const size_t size = header_size_bytes + get_serialized_size_bytes(sd);
|
345
347
|
std::vector<uint8_t, AllocU8<A>> bytes(size, 0, allocator_);
|
346
348
|
uint8_t* ptr = bytes.data() + header_size_bytes;
|
347
349
|
uint8_t* end_ptr = ptr + size;
|
@@ -400,8 +402,8 @@ std::vector<uint8_t, AllocU8<A>> var_opt_sketch<T,S,A>::serialize(unsigned heade
|
|
400
402
|
}
|
401
403
|
|
402
404
|
// write the sample items, skipping the gap. Either h_ or r_ may be 0
|
403
|
-
ptr +=
|
404
|
-
ptr +=
|
405
|
+
ptr += sd.serialize(ptr, end_ptr - ptr, data_, h_);
|
406
|
+
ptr += sd.serialize(ptr, end_ptr - ptr, &data_[h_ + 1], r_);
|
405
407
|
}
|
406
408
|
|
407
409
|
size_t bytes_written = ptr - bytes.data();
|
@@ -413,7 +415,8 @@ std::vector<uint8_t, AllocU8<A>> var_opt_sketch<T,S,A>::serialize(unsigned heade
|
|
413
415
|
}
|
414
416
|
|
415
417
|
template<typename T, typename S, typename A>
|
416
|
-
|
418
|
+
template<typename SerDe>
|
419
|
+
void var_opt_sketch<T,S,A>::serialize(std::ostream& os, const SerDe& sd) const {
|
417
420
|
const bool empty = (h_ == 0) && (r_ == 0);
|
418
421
|
|
419
422
|
const uint8_t preLongs = (empty ? PREAMBLE_LONGS_EMPTY
|
@@ -469,13 +472,19 @@ void var_opt_sketch<T,S,A>::serialize(std::ostream& os) const {
|
|
469
472
|
}
|
470
473
|
|
471
474
|
// write the sample items, skipping the gap. Either h_ or r_ may be 0
|
472
|
-
|
473
|
-
|
475
|
+
sd.serialize(os, data_, h_);
|
476
|
+
sd.serialize(os, &data_[h_ + 1], r_);
|
474
477
|
}
|
475
478
|
}
|
476
479
|
|
477
480
|
template<typename T, typename S, typename A>
|
478
481
|
var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(const void* bytes, size_t size, const A& allocator) {
|
482
|
+
return deserialize(bytes, size, S(), allocator);
|
483
|
+
}
|
484
|
+
|
485
|
+
template<typename T, typename S, typename A>
|
486
|
+
template<typename SerDe>
|
487
|
+
var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(const void* bytes, size_t size, const SerDe& sd, const A& allocator) {
|
479
488
|
ensure_minimum_memory(size, 8);
|
480
489
|
const char* ptr = static_cast<const char*>(bytes);
|
481
490
|
const char* base = ptr;
|
@@ -559,10 +568,10 @@ var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(const void* bytes, size
|
|
559
568
|
items_deleter deleter(array_size, allocator);
|
560
569
|
std::unique_ptr<T, items_deleter> items(A(allocator).allocate(array_size), deleter);
|
561
570
|
|
562
|
-
ptr +=
|
571
|
+
ptr += sd.deserialize(ptr, end_ptr - ptr, items.get(), h);
|
563
572
|
items.get_deleter().set_h(h); // serde didn't throw, so the items are now valid
|
564
573
|
|
565
|
-
ptr +=
|
574
|
+
ptr += sd.deserialize(ptr, end_ptr - ptr, &(items.get()[h + 1]), r);
|
566
575
|
items.get_deleter().set_r(r); // serde didn't throw, so the items are now valid
|
567
576
|
|
568
577
|
return var_opt_sketch(k, h, (r > 0 ? 1 : 0), r, n, total_wt_r, rf, array_size, false,
|
@@ -571,6 +580,12 @@ var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(const void* bytes, size
|
|
571
580
|
|
572
581
|
template<typename T, typename S, typename A>
|
573
582
|
var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(std::istream& is, const A& allocator) {
|
583
|
+
return deserialize(is, S(), allocator);
|
584
|
+
}
|
585
|
+
|
586
|
+
template<typename T, typename S, typename A>
|
587
|
+
template<typename SerDe>
|
588
|
+
var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(std::istream& is, const SerDe& sd, const A& allocator) {
|
574
589
|
const auto first_byte = read<uint8_t>(is);
|
575
590
|
uint8_t preamble_longs = first_byte & 0x3f;
|
576
591
|
const resize_factor rf = static_cast<resize_factor>((first_byte >> 6) & 0x03);
|
@@ -640,10 +655,10 @@ var_opt_sketch<T,S,A> var_opt_sketch<T,S,A>::deserialize(std::istream& is, const
|
|
640
655
|
items_deleter deleter(array_size, allocator);
|
641
656
|
std::unique_ptr<T, items_deleter> items(A(allocator).allocate(array_size), deleter);
|
642
657
|
|
643
|
-
|
658
|
+
sd.deserialize(is, items.get(), h); // aka &data_[0]
|
644
659
|
items.get_deleter().set_h(h); // serde didn't throw, so the items are now valid
|
645
660
|
|
646
|
-
|
661
|
+
sd.deserialize(is, &(items.get()[h + 1]), r);
|
647
662
|
items.get_deleter().set_r(r); // serde didn't throw, so the items are now valid
|
648
663
|
|
649
664
|
if (!is.good())
|
@@ -731,8 +746,10 @@ void var_opt_sketch<T,S,A>::update(T&& item, double weight) {
|
|
731
746
|
|
732
747
|
template<typename T, typename S, typename A>
|
733
748
|
string<A> var_opt_sketch<T,S,A>::to_string() const {
|
734
|
-
|
735
|
-
|
749
|
+
// Using a temporary stream for implementation here does not comply with AllocatorAwareContainer requirements.
|
750
|
+
// The stream does not support passing an allocator instance, and alternatives are complicated.
|
751
|
+
std::ostringstream os;
|
752
|
+
os << "### VarOpt SUMMARY:" << std::endl;
|
736
753
|
os << " k : " << k_ << std::endl;
|
737
754
|
os << " h : " << h_ << std::endl;
|
738
755
|
os << " r : " << r_ << std::endl;
|
@@ -740,24 +757,28 @@ string<A> var_opt_sketch<T,S,A>::to_string() const {
|
|
740
757
|
os << " Current size : " << curr_items_alloc_ << std::endl;
|
741
758
|
os << " Resize factor: " << (1 << rf_) << std::endl;
|
742
759
|
os << "### END SKETCH SUMMARY" << std::endl;
|
743
|
-
return os.str();
|
760
|
+
return string<A>(os.str().c_str(), allocator_);
|
744
761
|
}
|
745
762
|
|
746
763
|
template<typename T, typename S, typename A>
|
747
764
|
string<A> var_opt_sketch<T,S,A>::items_to_string() const {
|
748
|
-
|
765
|
+
// Using a temporary stream for implementation here does not comply with AllocatorAwareContainer requirements.
|
766
|
+
// The stream does not support passing an allocator instance, and alternatives are complicated.
|
767
|
+
std::ostringstream os;
|
749
768
|
os << "### Sketch Items" << std::endl;
|
750
769
|
int idx = 0;
|
751
770
|
for (auto record : *this) {
|
752
771
|
os << idx << ": " << record.first << "\twt = " << record.second << std::endl;
|
753
772
|
++idx;
|
754
773
|
}
|
755
|
-
return os.str();
|
774
|
+
return string<A>(os.str().c_str(), allocator_);
|
756
775
|
}
|
757
776
|
|
758
777
|
template<typename T, typename S, typename A>
|
759
778
|
string<A> var_opt_sketch<T,S,A>::items_to_string(bool print_gap) const {
|
760
|
-
|
779
|
+
// Using a temporary stream for implementation here does not comply with AllocatorAwareContainer requirements.
|
780
|
+
// The stream does not support passing an allocator instance, and alternatives are complicated.
|
781
|
+
std::ostringstream os;
|
761
782
|
os << "### Sketch Items" << std::endl;
|
762
783
|
const uint32_t array_length = (n_ < k_ ? n_ : k_ + 1);
|
763
784
|
for (uint32_t i = 0, display_idx = 0; i < array_length; ++i) {
|
@@ -774,7 +795,7 @@ string<A> var_opt_sketch<T,S,A>::items_to_string(bool print_gap) const {
|
|
774
795
|
++display_idx;
|
775
796
|
}
|
776
797
|
}
|
777
|
-
return os.str();
|
798
|
+
return string<A>(os.str().c_str(), allocator_);
|
778
799
|
}
|
779
800
|
|
780
801
|
template<typename T, typename S, typename A>
|
@@ -1677,16 +1698,6 @@ bool var_opt_sketch<T, S, A>::iterator::get_mark() const {
|
|
1677
1698
|
return sk_->marks_ == nullptr ? false : sk_->marks_[idx_];
|
1678
1699
|
}
|
1679
1700
|
|
1680
|
-
|
1681
|
-
|
1682
|
-
// ******************** MOVE TO COMMON UTILS AREA EVENTUALLY *********************
|
1683
|
-
|
1684
|
-
namespace random_utils {
|
1685
|
-
static std::random_device rd; // possibly unsafe in MinGW with GCC < 9.2
|
1686
|
-
static std::mt19937_64 rand(rd());
|
1687
|
-
static std::uniform_real_distribution<> next_double(0.0, 1.0);
|
1688
|
-
}
|
1689
|
-
|
1690
1701
|
/**
|
1691
1702
|
* Checks if target sampling allocation is more than 50% of max sampling size.
|
1692
1703
|
* If so, returns max sampling size, otherwise passes through target size.
|
@@ -45,7 +45,11 @@ template<typename A> using AllocU8 = typename std::allocator_traits<A>::template
|
|
45
45
|
* author Kevin Lang
|
46
46
|
* author Jon Malkin
|
47
47
|
*/
|
48
|
-
template
|
48
|
+
template<
|
49
|
+
typename T,
|
50
|
+
typename S = serde<T>, // deprecated, to be removed in the next major version
|
51
|
+
typename A = std::allocator<T>
|
52
|
+
>
|
49
53
|
class var_opt_union {
|
50
54
|
|
51
55
|
public:
|
@@ -88,14 +92,16 @@ public:
|
|
88
92
|
/**
|
89
93
|
* Computes size needed to serialize the current state of the union.
|
90
94
|
* This version is for all other types and can be expensive since every item needs to be looked at.
|
95
|
+
* @param instance of a SerDe
|
91
96
|
* @return size in bytes needed to serialize this sketch
|
92
97
|
*/
|
93
|
-
|
94
|
-
|
98
|
+
template<typename SerDe = S>
|
99
|
+
size_t get_serialized_size_bytes(const SerDe& sd = SerDe()) const;
|
100
|
+
|
95
101
|
// This is a convenience alias for users
|
96
102
|
// The type returned by the following serialize method
|
97
103
|
typedef vector_u8<A> vector_bytes;
|
98
|
-
|
104
|
+
|
99
105
|
/**
|
100
106
|
* NOTE: This method may be deprecated in a future version.
|
101
107
|
* This method serializes the sketch as a vector of bytes.
|
@@ -103,33 +109,62 @@ public:
|
|
103
109
|
* It is a blank space of a given size.
|
104
110
|
* This header is used in Datasketches PostgreSQL extension.
|
105
111
|
* @param header_size_bytes space to reserve in front of the sketch
|
112
|
+
* @param instance of a SerDe
|
106
113
|
*/
|
107
|
-
|
114
|
+
template<typename SerDe = S>
|
115
|
+
vector_bytes serialize(unsigned header_size_bytes = 0, const SerDe& sd = SerDe()) const;
|
108
116
|
|
109
117
|
/**
|
110
118
|
* NOTE: This method may be deprecated in a future version.
|
111
119
|
* This method serializes the sketch into a given stream in a binary form
|
112
120
|
* @param os output stream
|
121
|
+
* @param instance of a SerDe
|
113
122
|
*/
|
114
|
-
|
123
|
+
template<typename SerDe = S>
|
124
|
+
void serialize(std::ostream& os, const SerDe& sd = SerDe()) const;
|
115
125
|
|
116
126
|
/**
|
117
127
|
* NOTE: This method may be deprecated in a future version.
|
118
128
|
* This method deserializes a union from a given stream.
|
119
129
|
* @param is input stream
|
130
|
+
* @param instance of an Allocator
|
120
131
|
* @return an instance of a union
|
121
132
|
*/
|
122
133
|
static var_opt_union deserialize(std::istream& is, const A& allocator = A());
|
123
134
|
|
135
|
+
/**
|
136
|
+
* NOTE: This method may be deprecated in a future version.
|
137
|
+
* This method deserializes a union from a given stream.
|
138
|
+
* @param is input stream
|
139
|
+
* @param instance of a SerDe
|
140
|
+
* @param instance of an Allocator
|
141
|
+
* @return an instance of a union
|
142
|
+
*/
|
143
|
+
template<typename SerDe = S>
|
144
|
+
static var_opt_union deserialize(std::istream& is, const SerDe& sd = SerDe(), const A& allocator = A());
|
145
|
+
|
124
146
|
/**
|
125
147
|
* NOTE: This method may be deprecated in a future version.
|
126
148
|
* This method deserializes a union from a given array of bytes.
|
127
149
|
* @param bytes pointer to the array of bytes
|
128
150
|
* @param size the size of the array
|
151
|
+
* @param instance of an Allocator
|
129
152
|
* @return an instance of a union
|
130
153
|
*/
|
131
154
|
static var_opt_union deserialize(const void* bytes, size_t size, const A& allocator = A());
|
132
155
|
|
156
|
+
/**
|
157
|
+
* NOTE: This method may be deprecated in a future version.
|
158
|
+
* This method deserializes a union from a given array of bytes.
|
159
|
+
* @param bytes pointer to the array of bytes
|
160
|
+
* @param size the size of the array
|
161
|
+
* @param instance of a SerDe
|
162
|
+
* @param instance of an Allocator
|
163
|
+
* @return an instance of a union
|
164
|
+
*/
|
165
|
+
template<typename SerDe = S>
|
166
|
+
static var_opt_union deserialize(const void* bytes, size_t size, const SerDe& sd = SerDe(), const A& allocator = A());
|
167
|
+
|
133
168
|
/**
|
134
169
|
* Prints a summary of the union as a string.
|
135
170
|
* @return the summary as a string
|
@@ -24,6 +24,7 @@
|
|
24
24
|
|
25
25
|
#include <cmath>
|
26
26
|
#include <sstream>
|
27
|
+
#include <stdexcept>
|
27
28
|
|
28
29
|
namespace datasketches {
|
29
30
|
|
@@ -129,6 +130,12 @@ var_opt_union<T,S,A>& var_opt_union<T,S,A>::operator=(var_opt_union&& other) {
|
|
129
130
|
|
130
131
|
template<typename T, typename S, typename A>
|
131
132
|
var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(std::istream& is, const A& allocator) {
|
133
|
+
return deserialize(is, S(), allocator);
|
134
|
+
}
|
135
|
+
|
136
|
+
template<typename T, typename S, typename A>
|
137
|
+
template<typename SerDe>
|
138
|
+
var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(std::istream& is, const SerDe& sd, const A& allocator) {
|
132
139
|
const auto preamble_longs = read<uint8_t>(is);
|
133
140
|
const auto serial_version = read<uint8_t>(is);
|
134
141
|
const auto family_id = read<uint8_t>(is);
|
@@ -155,7 +162,7 @@ var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(std::istream& is, const A
|
|
155
162
|
const auto outer_tau_numer = read<double>(is);
|
156
163
|
const auto outer_tau_denom = read<uint64_t>(is);
|
157
164
|
|
158
|
-
var_opt_sketch<T,S,A> gadget = var_opt_sketch<T,S,A>::deserialize(is, allocator);
|
165
|
+
var_opt_sketch<T,S,A> gadget = var_opt_sketch<T,S,A>::deserialize(is, sd, allocator);
|
159
166
|
|
160
167
|
if (!is.good())
|
161
168
|
throw std::runtime_error("error reading from std::istream");
|
@@ -165,6 +172,12 @@ var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(std::istream& is, const A
|
|
165
172
|
|
166
173
|
template<typename T, typename S, typename A>
|
167
174
|
var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(const void* bytes, size_t size, const A& allocator) {
|
175
|
+
return deserialize(bytes, size, S(), allocator);
|
176
|
+
}
|
177
|
+
|
178
|
+
template<typename T, typename S, typename A>
|
179
|
+
template<typename SerDe>
|
180
|
+
var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(const void* bytes, size_t size, const SerDe& sd, const A& allocator) {
|
168
181
|
ensure_minimum_memory(size, 8);
|
169
182
|
const char* ptr = static_cast<const char*>(bytes);
|
170
183
|
uint8_t preamble_longs;
|
@@ -199,22 +212,24 @@ var_opt_union<T,S,A> var_opt_union<T,S,A>::deserialize(const void* bytes, size_t
|
|
199
212
|
ptr += copy_from_mem(ptr, outer_tau_denom);
|
200
213
|
|
201
214
|
const size_t gadget_size = size - (PREAMBLE_LONGS_NON_EMPTY << 3);
|
202
|
-
var_opt_sketch<T,S,A> gadget = var_opt_sketch<T,S,A>::deserialize(ptr, gadget_size, allocator);
|
215
|
+
var_opt_sketch<T,S,A> gadget = var_opt_sketch<T,S,A>::deserialize(ptr, gadget_size, sd, allocator);
|
203
216
|
|
204
217
|
return var_opt_union<T,S,A>(items_seen, outer_tau_numer, outer_tau_denom, max_k, std::move(gadget));
|
205
218
|
}
|
206
219
|
|
207
220
|
template<typename T, typename S, typename A>
|
208
|
-
|
221
|
+
template<typename SerDe>
|
222
|
+
size_t var_opt_union<T,S,A>::get_serialized_size_bytes(const SerDe& sd) const {
|
209
223
|
if (n_ == 0) {
|
210
224
|
return PREAMBLE_LONGS_EMPTY << 3;
|
211
225
|
} else {
|
212
|
-
return (PREAMBLE_LONGS_NON_EMPTY << 3) + gadget_.get_serialized_size_bytes();
|
226
|
+
return (PREAMBLE_LONGS_NON_EMPTY << 3) + gadget_.get_serialized_size_bytes(sd);
|
213
227
|
}
|
214
228
|
}
|
215
229
|
|
216
230
|
template<typename T, typename S, typename A>
|
217
|
-
|
231
|
+
template<typename SerDe>
|
232
|
+
void var_opt_union<T,S,A>::serialize(std::ostream& os, const SerDe& sd) const {
|
218
233
|
bool empty = (n_ == 0);
|
219
234
|
|
220
235
|
const uint8_t serialization_version(SER_VER);
|
@@ -240,13 +255,14 @@ void var_opt_union<T,S,A>::serialize(std::ostream& os) const {
|
|
240
255
|
write(os, n_);
|
241
256
|
write(os, outer_tau_numer_);
|
242
257
|
write(os, outer_tau_denom_);
|
243
|
-
gadget_.serialize(os);
|
258
|
+
gadget_.serialize(os, sd);
|
244
259
|
}
|
245
260
|
}
|
246
261
|
|
247
262
|
template<typename T, typename S, typename A>
|
248
|
-
|
249
|
-
|
263
|
+
template<typename SerDe>
|
264
|
+
std::vector<uint8_t, AllocU8<A>> var_opt_union<T,S,A>::serialize(unsigned header_size_bytes, const SerDe& sd) const {
|
265
|
+
const size_t size = header_size_bytes + get_serialized_size_bytes(sd);
|
250
266
|
std::vector<uint8_t, AllocU8<A>> bytes(size, 0, gadget_.allocator_);
|
251
267
|
uint8_t* ptr = bytes.data() + header_size_bytes;
|
252
268
|
|
@@ -278,7 +294,7 @@ std::vector<uint8_t, AllocU8<A>> var_opt_union<T,S,A>::serialize(unsigned header
|
|
278
294
|
ptr += copy_to_mem(outer_tau_numer_, ptr);
|
279
295
|
ptr += copy_to_mem(outer_tau_denom_, ptr);
|
280
296
|
|
281
|
-
auto gadget_bytes = gadget_.serialize();
|
297
|
+
auto gadget_bytes = gadget_.serialize(0, sd);
|
282
298
|
ptr += copy_to_mem(gadget_bytes.data(), ptr, gadget_bytes.size() * sizeof(uint8_t));
|
283
299
|
}
|
284
300
|
|
@@ -295,14 +311,16 @@ void var_opt_union<T,S,A>::reset() {
|
|
295
311
|
|
296
312
|
template<typename T, typename S, typename A>
|
297
313
|
string<A> var_opt_union<T,S,A>::to_string() const {
|
298
|
-
|
299
|
-
|
300
|
-
|
314
|
+
// Using a temporary stream for implementation here does not comply with AllocatorAwareContainer requirements.
|
315
|
+
// The stream does not support passing an allocator instance, and alternatives are complicated.
|
316
|
+
std::ostringstream os;
|
317
|
+
os << "### VarOpt Union SUMMARY:" << std::endl;
|
318
|
+
os << " n : " << n_ << std::endl;
|
301
319
|
os << " Max k : " << max_k_ << std::endl;
|
302
|
-
os << " Gadget Summary:
|
320
|
+
os << " Gadget Summary:" << std::endl;
|
303
321
|
os << gadget_.to_string();
|
304
|
-
os << "### END VarOpt Union SUMMARY
|
305
|
-
return os.str();
|
322
|
+
os << "### END VarOpt Union SUMMARY" << std::endl;
|
323
|
+
return string<A>(os.str().c_str(), gadget_.allocator_);
|
306
324
|
}
|
307
325
|
|
308
326
|
template<typename T, typename S, typename A>
|
@@ -39,7 +39,7 @@ TEST_CASE("varopt allocation test", "[var_opt_sketch]") {
|
|
39
39
|
var_opt_test_sketch sk1(10, var_opt_test_sketch::DEFAULT_RESIZE_FACTOR, 0);
|
40
40
|
for (int i = 0; i < 100; ++i) sk1.update(i);
|
41
41
|
auto bytes1 = sk1.serialize();
|
42
|
-
auto sk2 = var_opt_test_sketch::deserialize(bytes1.data(), bytes1.size(), 0);
|
42
|
+
auto sk2 = var_opt_test_sketch::deserialize(bytes1.data(), bytes1.size(), test_type_serde(), 0);
|
43
43
|
|
44
44
|
std::stringstream ss;
|
45
45
|
sk1.serialize(ss);
|
@@ -51,7 +51,7 @@ TEST_CASE("varopt allocation test", "[var_opt_sketch]") {
|
|
51
51
|
u1.update(sk3);
|
52
52
|
|
53
53
|
auto bytes2 = u1.serialize();
|
54
|
-
auto u2 = var_opt_test_union::deserialize(bytes2.data(), bytes2.size(), 0);
|
54
|
+
auto u2 = var_opt_test_union::deserialize(bytes2.data(), bytes2.size(), test_type_serde(), 0);
|
55
55
|
}
|
56
56
|
REQUIRE(test_allocator_total_bytes == 0);
|
57
57
|
REQUIRE(test_allocator_net_allocations == 0);
|
@@ -32,53 +32,34 @@ target_include_directories(theta
|
|
32
32
|
target_link_libraries(theta INTERFACE common)
|
33
33
|
target_compile_features(theta INTERFACE cxx_std_11)
|
34
34
|
|
35
|
-
set(theta_HEADERS "")
|
36
|
-
list(APPEND theta_HEADERS "include/theta_sketch.hpp;include/theta_sketch_impl.hpp")
|
37
|
-
list(APPEND theta_HEADERS "include/theta_union.hpp;include/theta_union_impl.hpp")
|
38
|
-
list(APPEND theta_HEADERS "include/theta_intersection.hpp;include/theta_intersection_impl.hpp")
|
39
|
-
list(APPEND theta_HEADERS "include/theta_a_not_b.hpp;include/theta_a_not_b_impl.hpp")
|
40
|
-
list(APPEND theta_HEADERS "include/theta_jaccard_similarity.hpp")
|
41
|
-
list(APPEND theta_HEADERS "include/theta_comparators.hpp")
|
42
|
-
list(APPEND theta_HEADERS "include/theta_constants.hpp")
|
43
|
-
list(APPEND theta_HEADERS "include/theta_helpers.hpp")
|
44
|
-
list(APPEND theta_HEADERS "include/theta_update_sketch_base.hpp;include/theta_update_sketch_base_impl.hpp")
|
45
|
-
list(APPEND theta_HEADERS "include/theta_union_base.hpp;include/theta_union_base_impl.hpp")
|
46
|
-
list(APPEND theta_HEADERS "include/theta_intersection_base.hpp;include/theta_intersection_base_impl.hpp")
|
47
|
-
list(APPEND theta_HEADERS "include/theta_set_difference_base.hpp;include/theta_set_difference_base_impl.hpp")
|
48
|
-
list(APPEND theta_HEADERS "include/theta_jaccard_similarity_base.hpp")
|
49
|
-
list(APPEND theta_HEADERS "include/bounds_on_ratios_in_sampled_sets.hpp")
|
50
|
-
list(APPEND theta_HEADERS "include/bounds_on_ratios_in_theta_sketched_sets.hpp")
|
51
|
-
|
52
35
|
install(TARGETS theta
|
53
36
|
EXPORT ${PROJECT_NAME}
|
54
37
|
)
|
55
38
|
|
56
|
-
install(FILES
|
39
|
+
install(FILES
|
40
|
+
include/theta_sketch.hpp
|
41
|
+
include/theta_sketch_impl.hpp
|
42
|
+
include/theta_union.hpp
|
43
|
+
include/theta_union_impl.hpp
|
44
|
+
include/theta_intersection.hpp
|
45
|
+
include/theta_intersection_impl.hpp
|
46
|
+
include/theta_a_not_b.hpp
|
47
|
+
include/theta_a_not_b_impl.hpp
|
48
|
+
include/theta_jaccard_similarity.hpp
|
49
|
+
include/theta_comparators.hpp
|
50
|
+
include/theta_constants.hpp
|
51
|
+
include/theta_helpers.hpp
|
52
|
+
include/theta_update_sketch_base.hpp
|
53
|
+
include/theta_update_sketch_base_impl.hpp
|
54
|
+
include/theta_union_base.hpp
|
55
|
+
include/theta_union_base_impl.hpp
|
56
|
+
include/theta_intersection_base.hpp
|
57
|
+
include/theta_intersection_base_impl.hpp
|
58
|
+
include/theta_set_difference_base.hpp
|
59
|
+
include/theta_set_difference_base_impl.hpp
|
60
|
+
include/theta_jaccard_similarity_base.hpp
|
61
|
+
include/bounds_on_ratios_in_sampled_sets.hpp
|
62
|
+
include/bounds_on_ratios_in_theta_sketched_sets.hpp
|
63
|
+
include/compact_theta_sketch_parser.hpp
|
64
|
+
include/compact_theta_sketch_parser_impl.hpp
|
57
65
|
DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}/DataSketches")
|
58
|
-
|
59
|
-
target_sources(theta
|
60
|
-
INTERFACE
|
61
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_sketch.hpp
|
62
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_union.hpp
|
63
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_intersection.hpp
|
64
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_a_not_b.hpp
|
65
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_sketch_impl.hpp
|
66
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_union_impl.hpp
|
67
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_intersection_impl.hpp
|
68
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_a_not_b_impl.hpp
|
69
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_jaccard_similarity.hpp
|
70
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_comparators.hpp
|
71
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_constants.hpp
|
72
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_helpers.hpp
|
73
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_update_sketch_base.hpp
|
74
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_update_sketch_base_impl.hpp
|
75
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_union_base.hpp
|
76
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_union_base_impl.hpp
|
77
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_intersection_base.hpp
|
78
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_intersection_base_impl.hpp
|
79
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_set_difference_base.hpp
|
80
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_set_difference_base_impl.hpp
|
81
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/theta_jaccard_similarity_base.hpp
|
82
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/bounds_on_ratios_in_sampled_sets.hpp
|
83
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/bounds_on_ratios_in_theta_sketched_sets.hpp
|
84
|
-
)
|