datasketches 0.3.1 → 0.4.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +10 -0
- data/NOTICE +1 -1
- data/README.md +0 -2
- data/ext/datasketches/cpc_wrapper.cpp +3 -3
- data/ext/datasketches/kll_wrapper.cpp +0 -10
- data/lib/datasketches/version.rb +1 -1
- data/lib/datasketches.rb +1 -1
- data/vendor/datasketches-cpp/CMakeLists.txt +23 -20
- data/vendor/datasketches-cpp/CODE_OF_CONDUCT.md +3 -0
- data/vendor/datasketches-cpp/CONTRIBUTING.md +50 -0
- data/vendor/datasketches-cpp/Doxyfile +2827 -0
- data/vendor/datasketches-cpp/LICENSE +0 -76
- data/vendor/datasketches-cpp/README.md +1 -3
- data/vendor/datasketches-cpp/common/CMakeLists.txt +12 -11
- data/vendor/datasketches-cpp/common/include/MurmurHash3.h +25 -27
- data/vendor/datasketches-cpp/common/include/common_defs.hpp +15 -10
- data/vendor/datasketches-cpp/common/include/count_zeros.hpp +11 -2
- data/vendor/datasketches-cpp/common/include/kolmogorov_smirnov.hpp +9 -6
- data/vendor/datasketches-cpp/common/include/memory_operations.hpp +5 -4
- data/vendor/datasketches-cpp/common/include/optional.hpp +148 -0
- data/vendor/datasketches-cpp/common/include/quantiles_sorted_view.hpp +95 -2
- data/vendor/datasketches-cpp/common/include/quantiles_sorted_view_impl.hpp +1 -1
- data/vendor/datasketches-cpp/common/include/serde.hpp +69 -20
- data/vendor/datasketches-cpp/common/test/CMakeLists.txt +2 -2
- data/vendor/datasketches-cpp/common/test/integration_test.cpp +6 -0
- data/vendor/datasketches-cpp/common/test/optional_test.cpp +85 -0
- data/vendor/datasketches-cpp/common/test/test_allocator.hpp +14 -14
- data/vendor/datasketches-cpp/{python/src/__init__.py → count/CMakeLists.txt} +25 -1
- data/vendor/datasketches-cpp/count/include/count_min.hpp +405 -0
- data/vendor/datasketches-cpp/count/include/count_min_impl.hpp +497 -0
- data/vendor/datasketches-cpp/{MANIFEST.in → count/test/CMakeLists.txt} +23 -20
- data/vendor/datasketches-cpp/count/test/count_min_allocation_test.cpp +155 -0
- data/vendor/datasketches-cpp/count/test/count_min_test.cpp +303 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_common.hpp +14 -20
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor.hpp +7 -4
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor_impl.hpp +17 -17
- data/vendor/datasketches-cpp/cpc/include/cpc_confidence.hpp +3 -3
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch.hpp +40 -40
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch_impl.hpp +14 -11
- data/vendor/datasketches-cpp/cpc/include/cpc_union.hpp +35 -11
- data/vendor/datasketches-cpp/cpc/include/cpc_union_impl.hpp +8 -8
- data/vendor/datasketches-cpp/cpc/include/cpc_util.hpp +16 -8
- data/vendor/datasketches-cpp/cpc/include/u32_table.hpp +3 -2
- data/vendor/datasketches-cpp/cpc/include/u32_table_impl.hpp +5 -5
- data/vendor/datasketches-cpp/cpc/test/CMakeLists.txt +20 -7
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_deserialize_from_java_test.cpp +60 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_serialize_for_java.cpp +38 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_test.cpp +4 -29
- data/vendor/datasketches-cpp/cpc/test/cpc_union_test.cpp +4 -4
- data/vendor/datasketches-cpp/{tox.ini → density/CMakeLists.txt} +24 -8
- data/vendor/datasketches-cpp/density/include/density_sketch.hpp +256 -0
- data/vendor/datasketches-cpp/density/include/density_sketch_impl.hpp +543 -0
- data/vendor/datasketches-cpp/{python/datasketches/__init__.py → density/test/CMakeLists.txt} +15 -3
- data/vendor/datasketches-cpp/density/test/density_sketch_test.cpp +244 -0
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch.hpp +21 -9
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch_impl.hpp +6 -4
- data/vendor/datasketches-cpp/fi/include/reverse_purge_hash_map.hpp +9 -3
- data/vendor/datasketches-cpp/fi/test/CMakeLists.txt +14 -1
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_deserialize_from_java_test.cpp +95 -0
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_serialize_for_java.cpp +83 -0
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_test.cpp +3 -42
- data/vendor/datasketches-cpp/hll/include/CouponList-internal.hpp +2 -2
- data/vendor/datasketches-cpp/hll/include/CouponList.hpp +3 -1
- data/vendor/datasketches-cpp/hll/include/Hll4Array-internal.hpp +19 -11
- data/vendor/datasketches-cpp/hll/include/Hll4Array.hpp +2 -5
- data/vendor/datasketches-cpp/hll/include/Hll6Array-internal.hpp +19 -7
- data/vendor/datasketches-cpp/hll/include/Hll6Array.hpp +1 -1
- data/vendor/datasketches-cpp/hll/include/Hll8Array-internal.hpp +98 -42
- data/vendor/datasketches-cpp/hll/include/Hll8Array.hpp +2 -0
- data/vendor/datasketches-cpp/hll/include/HllArray-internal.hpp +94 -61
- data/vendor/datasketches-cpp/hll/include/HllArray.hpp +20 -8
- data/vendor/datasketches-cpp/hll/include/HllSketch-internal.hpp +4 -4
- data/vendor/datasketches-cpp/hll/include/HllSketchImpl.hpp +3 -1
- data/vendor/datasketches-cpp/hll/include/HllSketchImplFactory.hpp +3 -21
- data/vendor/datasketches-cpp/hll/include/HllUnion-internal.hpp +8 -0
- data/vendor/datasketches-cpp/hll/include/HllUtil.hpp +14 -18
- data/vendor/datasketches-cpp/hll/include/coupon_iterator-internal.hpp +1 -1
- data/vendor/datasketches-cpp/hll/include/coupon_iterator.hpp +8 -2
- data/vendor/datasketches-cpp/hll/include/hll.hpp +79 -65
- data/vendor/datasketches-cpp/hll/test/CMakeLists.txt +14 -1
- data/vendor/datasketches-cpp/hll/test/HllUnionTest.cpp +7 -1
- data/vendor/datasketches-cpp/hll/test/ToFromByteArrayTest.cpp +0 -68
- data/vendor/datasketches-cpp/hll/test/hll_sketch_deserialize_from_java_test.cpp +69 -0
- data/vendor/datasketches-cpp/hll/test/hll_sketch_serialize_for_java.cpp +52 -0
- data/vendor/datasketches-cpp/kll/include/kll_helper.hpp +0 -1
- data/vendor/datasketches-cpp/kll/include/kll_helper_impl.hpp +2 -2
- data/vendor/datasketches-cpp/kll/include/kll_sketch.hpp +79 -53
- data/vendor/datasketches-cpp/kll/include/kll_sketch_impl.hpp +61 -132
- data/vendor/datasketches-cpp/kll/test/CMakeLists.txt +14 -1
- data/vendor/datasketches-cpp/kll/test/kll_sketch_deserialize_from_java_test.cpp +103 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_serialize_for_java.cpp +62 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_test.cpp +5 -40
- data/vendor/datasketches-cpp/quantiles/include/quantiles_sketch.hpp +76 -54
- data/vendor/datasketches-cpp/quantiles/include/quantiles_sketch_impl.hpp +66 -136
- data/vendor/datasketches-cpp/quantiles/test/CMakeLists.txt +14 -1
- data/vendor/datasketches-cpp/quantiles/test/quantiles_sketch_deserialize_from_java_test.cpp +84 -0
- data/vendor/datasketches-cpp/quantiles/test/quantiles_sketch_serialize_for_java.cpp +52 -0
- data/vendor/datasketches-cpp/quantiles/test/quantiles_sketch_test.cpp +15 -39
- data/vendor/datasketches-cpp/req/include/req_common.hpp +7 -3
- data/vendor/datasketches-cpp/req/include/req_compactor_impl.hpp +2 -4
- data/vendor/datasketches-cpp/req/include/req_sketch.hpp +105 -26
- data/vendor/datasketches-cpp/req/include/req_sketch_impl.hpp +50 -111
- data/vendor/datasketches-cpp/req/test/CMakeLists.txt +14 -1
- data/vendor/datasketches-cpp/req/test/req_sketch_deserialize_from_java_test.cpp +55 -0
- data/vendor/datasketches-cpp/{tuple/include/array_of_doubles_intersection_impl.hpp → req/test/req_sketch_serialize_for_java.cpp} +12 -7
- data/vendor/datasketches-cpp/req/test/req_sketch_test.cpp +3 -89
- data/vendor/datasketches-cpp/sampling/CMakeLists.txt +4 -0
- data/vendor/datasketches-cpp/sampling/include/ebpps_sample.hpp +210 -0
- data/vendor/datasketches-cpp/sampling/include/ebpps_sample_impl.hpp +535 -0
- data/vendor/datasketches-cpp/sampling/include/ebpps_sketch.hpp +281 -0
- data/vendor/datasketches-cpp/sampling/include/ebpps_sketch_impl.hpp +531 -0
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch.hpp +89 -32
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch_impl.hpp +33 -19
- data/vendor/datasketches-cpp/sampling/include/var_opt_union.hpp +13 -10
- data/vendor/datasketches-cpp/sampling/include/var_opt_union_impl.hpp +23 -19
- data/vendor/datasketches-cpp/sampling/test/CMakeLists.txt +55 -8
- data/vendor/datasketches-cpp/sampling/test/ebpps_allocation_test.cpp +96 -0
- data/vendor/datasketches-cpp/sampling/test/ebpps_sample_test.cpp +137 -0
- data/vendor/datasketches-cpp/sampling/test/ebpps_sketch_test.cpp +266 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_deserialize_from_java_test.cpp +81 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_serialize_for_java.cpp +54 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_test.cpp +33 -51
- data/vendor/datasketches-cpp/sampling/test/var_opt_union_deserialize_from_java_test.cpp +50 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_union_serialize_for_java.cpp +56 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_union_test.cpp +0 -20
- data/vendor/datasketches-cpp/theta/CMakeLists.txt +1 -0
- data/vendor/datasketches-cpp/theta/include/bit_packing.hpp +6279 -0
- data/vendor/datasketches-cpp/theta/include/bounds_on_ratios_in_sampled_sets.hpp +1 -0
- data/vendor/datasketches-cpp/theta/include/bounds_on_ratios_in_theta_sketched_sets.hpp +7 -6
- data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser.hpp +14 -8
- data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser_impl.hpp +60 -46
- data/vendor/datasketches-cpp/theta/include/theta_a_not_b.hpp +20 -5
- data/vendor/datasketches-cpp/theta/include/theta_constants.hpp +10 -4
- data/vendor/datasketches-cpp/theta/include/theta_helpers.hpp +3 -1
- data/vendor/datasketches-cpp/theta/include/theta_intersection.hpp +13 -5
- data/vendor/datasketches-cpp/theta/include/theta_intersection_base_impl.hpp +5 -5
- data/vendor/datasketches-cpp/theta/include/theta_intersection_impl.hpp +3 -3
- data/vendor/datasketches-cpp/theta/include/theta_jaccard_similarity.hpp +2 -1
- data/vendor/datasketches-cpp/theta/include/theta_jaccard_similarity_base.hpp +1 -0
- data/vendor/datasketches-cpp/theta/include/theta_set_difference_base_impl.hpp +1 -1
- data/vendor/datasketches-cpp/theta/include/theta_sketch.hpp +180 -33
- data/vendor/datasketches-cpp/theta/include/theta_sketch_impl.hpp +430 -130
- data/vendor/datasketches-cpp/theta/include/theta_union.hpp +17 -10
- data/vendor/datasketches-cpp/theta/include/theta_union_base_impl.hpp +10 -10
- data/vendor/datasketches-cpp/theta/include/theta_union_impl.hpp +3 -3
- data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base.hpp +21 -6
- data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base_impl.hpp +13 -3
- data/vendor/datasketches-cpp/theta/test/CMakeLists.txt +15 -1
- data/vendor/datasketches-cpp/theta/test/bit_packing_test.cpp +80 -0
- data/vendor/datasketches-cpp/theta/test/theta_sketch_deserialize_from_java_test.cpp +57 -0
- data/vendor/datasketches-cpp/theta/test/theta_sketch_serialize_for_java.cpp +61 -0
- data/vendor/datasketches-cpp/theta/test/theta_sketch_test.cpp +39 -188
- data/vendor/datasketches-cpp/theta/test/theta_union_test.cpp +25 -0
- data/vendor/datasketches-cpp/tuple/CMakeLists.txt +8 -7
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch.hpp +19 -144
- data/vendor/datasketches-cpp/tuple/include/{array_of_doubles_a_not_b.hpp → array_tuple_a_not_b.hpp} +24 -16
- data/vendor/datasketches-cpp/tuple/include/{array_of_doubles_a_not_b_impl.hpp → array_tuple_a_not_b_impl.hpp} +4 -4
- data/vendor/datasketches-cpp/tuple/include/array_tuple_intersection.hpp +65 -0
- data/vendor/datasketches-cpp/tuple/include/array_tuple_intersection_impl.hpp +31 -0
- data/vendor/datasketches-cpp/tuple/include/array_tuple_sketch.hpp +237 -0
- data/vendor/datasketches-cpp/tuple/include/{array_of_doubles_sketch_impl.hpp → array_tuple_sketch_impl.hpp} +40 -41
- data/vendor/datasketches-cpp/tuple/include/array_tuple_union.hpp +81 -0
- data/vendor/datasketches-cpp/tuple/include/array_tuple_union_impl.hpp +43 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_a_not_b.hpp +11 -2
- data/vendor/datasketches-cpp/tuple/include/tuple_intersection.hpp +17 -10
- data/vendor/datasketches-cpp/tuple/include/tuple_jaccard_similarity.hpp +2 -1
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch.hpp +95 -32
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch_impl.hpp +2 -1
- data/vendor/datasketches-cpp/tuple/include/tuple_union.hpp +19 -11
- data/vendor/datasketches-cpp/tuple/test/CMakeLists.txt +16 -1
- data/vendor/datasketches-cpp/tuple/test/aod_sketch_deserialize_from_java_test.cpp +76 -0
- data/vendor/datasketches-cpp/tuple/test/aod_sketch_serialize_for_java.cpp +62 -0
- data/vendor/datasketches-cpp/tuple/test/array_of_doubles_sketch_test.cpp +5 -129
- data/vendor/datasketches-cpp/tuple/test/engagement_test.cpp +85 -89
- data/vendor/datasketches-cpp/tuple/test/tuple_jaccard_similarity_test.cpp +3 -1
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_deserialize_from_java_test.cpp +47 -0
- data/vendor/datasketches-cpp/{python/src/datasketches.cpp → tuple/test/tuple_sketch_serialize_for_java.cpp} +16 -30
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_test.cpp +1 -1
- data/vendor/datasketches-cpp/version.cfg.in +1 -1
- metadata +61 -79
- data/vendor/datasketches-cpp/fi/test/items_sketch_string_from_java.sk +0 -0
- data/vendor/datasketches-cpp/fi/test/items_sketch_string_utf8_from_java.sk +0 -0
- data/vendor/datasketches-cpp/fi/test/longs_sketch_from_java.sk +0 -0
- data/vendor/datasketches-cpp/hll/test/array6_from_java.sk +0 -0
- data/vendor/datasketches-cpp/hll/test/compact_array4_from_java.sk +0 -0
- data/vendor/datasketches-cpp/hll/test/compact_set_from_java.sk +0 -0
- data/vendor/datasketches-cpp/hll/test/list_from_java.sk +0 -0
- data/vendor/datasketches-cpp/hll/test/updatable_array4_from_java.sk +0 -0
- data/vendor/datasketches-cpp/hll/test/updatable_set_from_java.sk +0 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_from_java.sk +0 -0
- data/vendor/datasketches-cpp/pyproject.toml +0 -23
- data/vendor/datasketches-cpp/python/CMakeLists.txt +0 -81
- data/vendor/datasketches-cpp/python/README.md +0 -85
- data/vendor/datasketches-cpp/python/datasketches/PySerDe.py +0 -104
- data/vendor/datasketches-cpp/python/include/py_serde.hpp +0 -113
- data/vendor/datasketches-cpp/python/jupyter/CPCSketch.ipynb +0 -345
- data/vendor/datasketches-cpp/python/jupyter/FrequentItemsSketch.ipynb +0 -354
- data/vendor/datasketches-cpp/python/jupyter/HLLSketch.ipynb +0 -346
- data/vendor/datasketches-cpp/python/jupyter/KLLSketch.ipynb +0 -463
- data/vendor/datasketches-cpp/python/jupyter/ThetaSketchNotebook.ipynb +0 -403
- data/vendor/datasketches-cpp/python/pybind11Path.cmd +0 -21
- data/vendor/datasketches-cpp/python/src/cpc_wrapper.cpp +0 -90
- data/vendor/datasketches-cpp/python/src/fi_wrapper.cpp +0 -128
- data/vendor/datasketches-cpp/python/src/hll_wrapper.cpp +0 -134
- data/vendor/datasketches-cpp/python/src/kll_wrapper.cpp +0 -210
- data/vendor/datasketches-cpp/python/src/ks_wrapper.cpp +0 -68
- data/vendor/datasketches-cpp/python/src/py_serde.cpp +0 -111
- data/vendor/datasketches-cpp/python/src/quantiles_wrapper.cpp +0 -204
- data/vendor/datasketches-cpp/python/src/req_wrapper.cpp +0 -215
- data/vendor/datasketches-cpp/python/src/theta_wrapper.cpp +0 -172
- data/vendor/datasketches-cpp/python/src/vector_of_kll.cpp +0 -490
- data/vendor/datasketches-cpp/python/src/vo_wrapper.cpp +0 -173
- data/vendor/datasketches-cpp/python/tests/__init__.py +0 -16
- data/vendor/datasketches-cpp/python/tests/cpc_test.py +0 -64
- data/vendor/datasketches-cpp/python/tests/fi_test.py +0 -110
- data/vendor/datasketches-cpp/python/tests/hll_test.py +0 -130
- data/vendor/datasketches-cpp/python/tests/kll_test.py +0 -125
- data/vendor/datasketches-cpp/python/tests/quantiles_test.py +0 -126
- data/vendor/datasketches-cpp/python/tests/req_test.py +0 -126
- data/vendor/datasketches-cpp/python/tests/theta_test.py +0 -146
- data/vendor/datasketches-cpp/python/tests/vector_of_kll_test.py +0 -148
- data/vendor/datasketches-cpp/python/tests/vo_test.py +0 -125
- data/vendor/datasketches-cpp/req/test/req_float_empty_from_java.sk +0 -0
- data/vendor/datasketches-cpp/req/test/req_float_estimation_from_java.sk +0 -0
- data/vendor/datasketches-cpp/req/test/req_float_exact_from_java.sk +0 -0
- data/vendor/datasketches-cpp/req/test/req_float_raw_items_from_java.sk +0 -0
- data/vendor/datasketches-cpp/req/test/req_float_single_item_from_java.sk +0 -0
- data/vendor/datasketches-cpp/sampling/test/binaries_from_java.txt +0 -67
- data/vendor/datasketches-cpp/sampling/test/varopt_sketch_long_sampling.sk +0 -0
- data/vendor/datasketches-cpp/sampling/test/varopt_sketch_string_exact.sk +0 -0
- data/vendor/datasketches-cpp/sampling/test/varopt_union_double_sampling.sk +0 -0
- data/vendor/datasketches-cpp/setup.py +0 -110
- data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_exact_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_single_item_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_intersection.hpp +0 -52
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union.hpp +0 -81
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union_impl.hpp +0 -43
- data/vendor/datasketches-cpp/tuple/test/aod_1_compact_empty_from_java.sk +0 -1
- data/vendor/datasketches-cpp/tuple/test/aod_1_compact_estimation_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/test/aod_1_compact_non_empty_no_entries_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/test/aod_2_compact_exact_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/test/aod_3_compact_empty_from_java.sk +0 -1
@@ -29,6 +29,7 @@
|
|
29
29
|
namespace datasketches {
|
30
30
|
|
31
31
|
/**
|
32
|
+
* Bounds on ratios in sampled sets.
|
32
33
|
* This class is used to compute the bounds on the estimate of the ratio <i>|B| / |A|</i>, where:
|
33
34
|
* <ul>
|
34
35
|
* <li><i>|A|</i> is the unknown size of a set <i>A</i> of unique identifiers.</li>
|
@@ -28,6 +28,7 @@
|
|
28
28
|
namespace datasketches {
|
29
29
|
|
30
30
|
/**
|
31
|
+
* Bounds on ratios in Theta sketched sets.
|
31
32
|
* This is to compute the bounds on the estimate of the ratio <i>B / A</i>, where:
|
32
33
|
* <ul>
|
33
34
|
* <li><i>A</i> is a Theta Sketch of population <i>PopA</i>.</li>
|
@@ -50,8 +51,8 @@ class bounds_on_ratios_in_theta_sketched_sets {
|
|
50
51
|
public:
|
51
52
|
/**
|
52
53
|
* Gets the approximate lower bound for B over A based on a 95% confidence interval
|
53
|
-
* @param
|
54
|
-
* @param
|
54
|
+
* @param sketch_a the sketch A
|
55
|
+
* @param sketch_b the sketch B
|
55
56
|
* @return the approximate lower bound for B over A
|
56
57
|
*/
|
57
58
|
template<typename SketchA, typename SketchB>
|
@@ -72,8 +73,8 @@ public:
|
|
72
73
|
|
73
74
|
/**
|
74
75
|
* Gets the approximate upper bound for B over A based on a 95% confidence interval
|
75
|
-
* @param
|
76
|
-
* @param
|
76
|
+
* @param sketch_a the sketch A
|
77
|
+
* @param sketch_b the sketch B
|
77
78
|
* @return the approximate upper bound for B over A
|
78
79
|
*/
|
79
80
|
template<typename SketchA, typename SketchB>
|
@@ -94,8 +95,8 @@ public:
|
|
94
95
|
|
95
96
|
/**
|
96
97
|
* Gets the estimate for B over A
|
97
|
-
* @param
|
98
|
-
* @param
|
98
|
+
* @param sketch_a the sketch A
|
99
|
+
* @param sketch_b the sketch B
|
99
100
|
* @return the estimate for B over A
|
100
101
|
*/
|
101
102
|
template<typename SketchA, typename SketchB>
|
@@ -20,7 +20,7 @@
|
|
20
20
|
#ifndef COMPACT_THETA_SKETCH_PARSER_HPP_
|
21
21
|
#define COMPACT_THETA_SKETCH_PARSER_HPP_
|
22
22
|
|
23
|
-
#include <
|
23
|
+
#include <cstdint>
|
24
24
|
|
25
25
|
namespace datasketches {
|
26
26
|
|
@@ -33,7 +33,8 @@ public:
|
|
33
33
|
uint16_t seed_hash;
|
34
34
|
uint32_t num_entries;
|
35
35
|
uint64_t theta;
|
36
|
-
const
|
36
|
+
const void* entries_start_ptr;
|
37
|
+
uint8_t entry_bits;
|
37
38
|
};
|
38
39
|
|
39
40
|
static compact_theta_sketch_data parse(const void* ptr, size_t size, uint64_t seed, bool dump_on_error = false);
|
@@ -45,18 +46,23 @@ private:
|
|
45
46
|
static const size_t COMPACT_SKETCH_TYPE_BYTE = 2;
|
46
47
|
static const size_t COMPACT_SKETCH_FLAGS_BYTE = 5;
|
47
48
|
static const size_t COMPACT_SKETCH_SEED_HASH_U16 = 3;
|
48
|
-
static const size_t
|
49
|
-
static const size_t
|
50
|
-
static const size_t COMPACT_SKETCH_ENTRIES_EXACT_U64 = 2;
|
51
|
-
static const size_t
|
52
|
-
static const size_t
|
49
|
+
static const size_t COMPACT_SKETCH_SINGLE_ENTRY_U64 = 1; // ver 3
|
50
|
+
static const size_t COMPACT_SKETCH_NUM_ENTRIES_U32 = 2; // ver 1-3
|
51
|
+
static const size_t COMPACT_SKETCH_ENTRIES_EXACT_U64 = 2; // ver 1-3
|
52
|
+
static const size_t COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 = 3; // ver 1-3
|
53
|
+
static const size_t COMPACT_SKETCH_THETA_U64 = 2; // ver 1-3
|
54
|
+
static const size_t COMPACT_SKETCH_V4_ENTRY_BITS_BYTE = 3;
|
55
|
+
static const size_t COMPACT_SKETCH_V4_NUM_ENTRIES_BYTES_BYTE = 4;
|
56
|
+
static const size_t COMPACT_SKETCH_V4_THETA_U64 = 1;
|
57
|
+
static const size_t COMPACT_SKETCH_V4_PACKED_DATA_EXACT_BYTE = 8;
|
58
|
+
static const size_t COMPACT_SKETCH_V4_PACKED_DATA_ESTIMATION_BYTE = 16;
|
53
59
|
|
54
60
|
static const uint8_t COMPACT_SKETCH_IS_EMPTY_FLAG = 2;
|
55
61
|
static const uint8_t COMPACT_SKETCH_IS_ORDERED_FLAG = 4;
|
56
62
|
|
57
|
-
static const uint8_t COMPACT_SKETCH_SERIAL_VERSION = 3;
|
58
63
|
static const uint8_t COMPACT_SKETCH_TYPE = 3;
|
59
64
|
|
65
|
+
static void check_memory_size(const void* ptr, size_t actual_bytes, size_t expected_bytes, bool dump_on_error);
|
60
66
|
static std::string hex_dump(const uint8_t* ptr, size_t size);
|
61
67
|
};
|
62
68
|
|
@@ -26,106 +26,120 @@
|
|
26
26
|
|
27
27
|
namespace datasketches {
|
28
28
|
|
29
|
+
template<typename T>
|
30
|
+
T whole_bytes_to_hold_bits(T bits) {
|
31
|
+
static_assert(std::is_integral<T>::value, "integral type expected");
|
32
|
+
return (bits >> 3) + ((bits & 7) > 0);
|
33
|
+
}
|
34
|
+
|
29
35
|
template<bool dummy>
|
30
36
|
auto compact_theta_sketch_parser<dummy>::parse(const void* ptr, size_t size, uint64_t seed, bool dump_on_error) -> compact_theta_sketch_data {
|
31
|
-
|
32
|
-
|
33
|
-
|
37
|
+
check_memory_size(ptr, size, 8, dump_on_error);
|
38
|
+
checker<true>::check_sketch_type(reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_TYPE_BYTE], COMPACT_SKETCH_TYPE);
|
34
39
|
uint8_t serial_version = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_SERIAL_VERSION_BYTE];
|
35
|
-
|
36
40
|
switch(serial_version) {
|
37
|
-
case
|
38
|
-
|
41
|
+
case 4: {
|
42
|
+
// version 4 sketches are ordered and always have entries (single item in exact mode is v3)
|
43
|
+
const uint16_t seed_hash = reinterpret_cast<const uint16_t*>(ptr)[COMPACT_SKETCH_SEED_HASH_U16];
|
44
|
+
checker<true>::check_seed_hash(seed_hash, compute_seed_hash(seed));
|
45
|
+
const bool has_theta = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE] > 1;
|
46
|
+
uint64_t theta = theta_constants::MAX_THETA;
|
47
|
+
if (has_theta) {
|
48
|
+
check_memory_size(ptr, size, 16, dump_on_error);
|
49
|
+
theta = reinterpret_cast<const uint64_t*>(ptr)[COMPACT_SKETCH_V4_THETA_U64];
|
50
|
+
}
|
51
|
+
const uint8_t num_entries_bytes = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_V4_NUM_ENTRIES_BYTES_BYTE];
|
52
|
+
size_t data_offset_bytes = has_theta ? COMPACT_SKETCH_V4_PACKED_DATA_ESTIMATION_BYTE : COMPACT_SKETCH_V4_PACKED_DATA_EXACT_BYTE;
|
53
|
+
check_memory_size(ptr, size, data_offset_bytes + num_entries_bytes, dump_on_error);
|
54
|
+
uint32_t num_entries = 0;
|
55
|
+
const uint8_t* num_entries_ptr = reinterpret_cast<const uint8_t*>(ptr) + data_offset_bytes;
|
56
|
+
for (unsigned i = 0; i < num_entries_bytes; ++i) {
|
57
|
+
num_entries |= (*num_entries_ptr++) << (i << 3);
|
58
|
+
}
|
59
|
+
data_offset_bytes += num_entries_bytes;
|
60
|
+
const uint8_t entry_bits = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_V4_ENTRY_BITS_BYTE];
|
61
|
+
const size_t expected_bits = entry_bits * num_entries;
|
62
|
+
const size_t expected_size_bytes = data_offset_bytes + whole_bytes_to_hold_bits(expected_bits);
|
63
|
+
check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
|
64
|
+
return {false, true, seed_hash, num_entries, theta,
|
65
|
+
reinterpret_cast<const uint8_t*>(ptr) + data_offset_bytes, entry_bits};
|
66
|
+
}
|
67
|
+
case 3: {
|
39
68
|
uint64_t theta = theta_constants::MAX_THETA;
|
40
69
|
const uint16_t seed_hash = reinterpret_cast<const uint16_t*>(ptr)[COMPACT_SKETCH_SEED_HASH_U16];
|
41
70
|
if (reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_FLAGS_BYTE] & (1 << COMPACT_SKETCH_IS_EMPTY_FLAG)) {
|
42
|
-
return {true, true, seed_hash, 0, theta, nullptr};
|
71
|
+
return {true, true, seed_hash, 0, theta, nullptr, 64};
|
43
72
|
}
|
44
73
|
checker<true>::check_seed_hash(seed_hash, compute_seed_hash(seed));
|
45
74
|
const bool has_theta = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE] > 2;
|
46
75
|
if (has_theta) {
|
47
|
-
|
76
|
+
check_memory_size(ptr, size, (COMPACT_SKETCH_THETA_U64 + 1) * sizeof(uint64_t), dump_on_error);
|
48
77
|
theta = reinterpret_cast<const uint64_t*>(ptr)[COMPACT_SKETCH_THETA_U64];
|
49
78
|
}
|
50
79
|
if (reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE] == 1) {
|
51
|
-
|
52
|
-
return {false, true, seed_hash, 1, theta, reinterpret_cast<const uint64_t*>(ptr) + COMPACT_SKETCH_SINGLE_ENTRY_U64};
|
80
|
+
check_memory_size(ptr, size, 16, dump_on_error);
|
81
|
+
return {false, true, seed_hash, 1, theta, reinterpret_cast<const uint64_t*>(ptr) + COMPACT_SKETCH_SINGLE_ENTRY_U64, 64};
|
53
82
|
}
|
54
83
|
const uint32_t num_entries = reinterpret_cast<const uint32_t*>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
|
55
84
|
const size_t entries_start_u64 = has_theta ? COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 : COMPACT_SKETCH_ENTRIES_EXACT_U64;
|
56
85
|
const uint64_t* entries = reinterpret_cast<const uint64_t*>(ptr) + entries_start_u64;
|
57
86
|
const size_t expected_size_bytes = (entries_start_u64 + num_entries) * sizeof(uint64_t);
|
58
|
-
|
59
|
-
throw std::out_of_range(std::to_string(expected_size_bytes) + " bytes expected, actual " + std::to_string(size)
|
60
|
-
+ (dump_on_error ? (", sketch dump: " + hex_dump(reinterpret_cast<const uint8_t*>(ptr), size)) : ""));
|
61
|
-
}
|
87
|
+
check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
|
62
88
|
const bool is_ordered = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_FLAGS_BYTE] & (1 << COMPACT_SKETCH_IS_ORDERED_FLAG);
|
63
|
-
return {false, is_ordered, seed_hash, num_entries, theta, entries};
|
89
|
+
return {false, is_ordered, seed_hash, num_entries, theta, entries, 64};
|
64
90
|
}
|
65
91
|
case 1: {
|
66
92
|
uint16_t seed_hash = compute_seed_hash(seed);
|
67
|
-
checker<true>::check_sketch_type(reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_TYPE_BYTE], COMPACT_SKETCH_TYPE);
|
68
93
|
const uint32_t num_entries = reinterpret_cast<const uint32_t*>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
|
69
94
|
uint64_t theta = reinterpret_cast<const uint64_t*>(ptr)[COMPACT_SKETCH_THETA_U64];
|
70
95
|
bool is_empty = (num_entries == 0) && (theta == theta_constants::MAX_THETA);
|
71
|
-
if (is_empty) {
|
72
|
-
return {true, true, seed_hash, 0, theta, nullptr};
|
73
|
-
}
|
96
|
+
if (is_empty) return {true, true, seed_hash, 0, theta, nullptr, 64};
|
74
97
|
const uint64_t* entries = reinterpret_cast<const uint64_t*>(ptr) + COMPACT_SKETCH_ENTRIES_ESTIMATION_U64;
|
75
98
|
const size_t expected_size_bytes = (COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 + num_entries) * sizeof(uint64_t);
|
76
|
-
|
77
|
-
|
78
|
-
+ (dump_on_error ? (", sketch dump: " + hex_dump(reinterpret_cast<const uint8_t*>(ptr), size)) : ""));
|
79
|
-
}
|
80
|
-
return {false, true, seed_hash, num_entries, theta, entries};
|
99
|
+
check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
|
100
|
+
return {false, true, seed_hash, num_entries, theta, entries, 64};
|
81
101
|
}
|
82
102
|
case 2: {
|
83
|
-
uint8_t preamble_size =
|
84
|
-
checker<true>::check_sketch_type(reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_TYPE_BYTE], COMPACT_SKETCH_TYPE);
|
103
|
+
uint8_t preamble_size = reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE];
|
85
104
|
const uint16_t seed_hash = reinterpret_cast<const uint16_t*>(ptr)[COMPACT_SKETCH_SEED_HASH_U16];
|
86
105
|
checker<true>::check_seed_hash(seed_hash, compute_seed_hash(seed));
|
87
106
|
if (preamble_size == 1) {
|
88
|
-
return {true, true, seed_hash, 0, theta_constants::MAX_THETA, nullptr};
|
107
|
+
return {true, true, seed_hash, 0, theta_constants::MAX_THETA, nullptr, 64};
|
89
108
|
} else if (preamble_size == 2) {
|
90
109
|
const uint32_t num_entries = reinterpret_cast<const uint32_t*>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
|
91
110
|
if (num_entries == 0) {
|
92
|
-
return {true, true, seed_hash, 0, theta_constants::MAX_THETA, nullptr};
|
111
|
+
return {true, true, seed_hash, 0, theta_constants::MAX_THETA, nullptr, 64};
|
93
112
|
} else {
|
94
113
|
const size_t expected_size_bytes = (preamble_size + num_entries) << 3;
|
95
|
-
|
96
|
-
throw std::out_of_range(std::to_string(expected_size_bytes) + " bytes expected, actual " + std::to_string(size)
|
97
|
-
+ (dump_on_error ? (", sketch dump: " + hex_dump(reinterpret_cast<const uint8_t*>(ptr), size)) : ""));
|
98
|
-
}
|
114
|
+
check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
|
99
115
|
const uint64_t* entries = reinterpret_cast<const uint64_t*>(ptr) + COMPACT_SKETCH_ENTRIES_EXACT_U64;
|
100
|
-
return {false, true, seed_hash, num_entries, theta_constants::MAX_THETA, entries};
|
116
|
+
return {false, true, seed_hash, num_entries, theta_constants::MAX_THETA, entries, 64};
|
101
117
|
}
|
102
118
|
} else if (preamble_size == 3) {
|
103
119
|
const uint32_t num_entries = reinterpret_cast<const uint32_t*>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
|
104
120
|
uint64_t theta = reinterpret_cast<const uint64_t*>(ptr)[COMPACT_SKETCH_THETA_U64];
|
105
121
|
bool is_empty = (num_entries == 0) && (theta == theta_constants::MAX_THETA);
|
106
|
-
if (is_empty) {
|
107
|
-
return {true, true, seed_hash, 0, theta, nullptr};
|
108
|
-
}
|
122
|
+
if (is_empty) return {true, true, seed_hash, 0, theta, nullptr, 64};
|
109
123
|
const uint64_t* entries = reinterpret_cast<const uint64_t*>(ptr) + COMPACT_SKETCH_ENTRIES_ESTIMATION_U64;
|
110
124
|
const size_t expected_size_bytes = (COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 + num_entries) * sizeof(uint64_t);
|
111
|
-
|
112
|
-
|
113
|
-
+ (dump_on_error ? (", sketch dump: " + hex_dump(reinterpret_cast<const uint8_t*>(ptr), size)) : ""));
|
114
|
-
}
|
115
|
-
return {false, true, seed_hash, num_entries, theta, entries};
|
125
|
+
check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
|
126
|
+
return {false, true, seed_hash, num_entries, theta, entries, 64};
|
116
127
|
} else {
|
117
128
|
throw std::invalid_argument(std::to_string(preamble_size) + " longs of premable, but expected 1, 2, or 3");
|
118
129
|
}
|
119
130
|
}
|
120
131
|
default:
|
121
|
-
|
122
|
-
checker<true>::check_serial_version(reinterpret_cast<const uint8_t*>(ptr)[COMPACT_SKETCH_SERIAL_VERSION_BYTE], COMPACT_SKETCH_SERIAL_VERSION);
|
123
|
-
// this throw is never reached, because check_serial_version will throw an informative exception.
|
124
|
-
// This is only here to avoid a compiler warning about a path without a return value.
|
125
|
-
throw std::invalid_argument("unexpected sketch serialization version");
|
132
|
+
throw std::invalid_argument("unsupported serial version " + std::to_string(serial_version));
|
126
133
|
}
|
127
134
|
}
|
128
135
|
|
136
|
+
template<bool dummy>
|
137
|
+
void compact_theta_sketch_parser<dummy>::check_memory_size(const void* ptr, size_t actual_bytes, size_t expected_bytes, bool dump_on_error) {
|
138
|
+
if (actual_bytes < expected_bytes) throw std::out_of_range("at least " + std::to_string(expected_bytes)
|
139
|
+
+ " bytes expected, actual " + std::to_string(actual_bytes)
|
140
|
+
+ (dump_on_error ? (", sketch dump: " + hex_dump(reinterpret_cast<const uint8_t*>(ptr), actual_bytes)) : ""));
|
141
|
+
}
|
142
|
+
|
129
143
|
template<bool dummy>
|
130
144
|
std::string compact_theta_sketch_parser<dummy>::hex_dump(const uint8_t* ptr, size_t size) {
|
131
145
|
std::stringstream s;
|
@@ -25,6 +25,16 @@
|
|
25
25
|
|
26
26
|
namespace datasketches {
|
27
27
|
|
28
|
+
// forward declaration
|
29
|
+
template<typename A> class theta_a_not_b_alloc;
|
30
|
+
|
31
|
+
// alias with default allocator for convenience
|
32
|
+
using theta_a_not_b = theta_a_not_b_alloc<std::allocator<uint64_t>>;
|
33
|
+
|
34
|
+
/**
|
35
|
+
* Theta A-not-B (set difference).
|
36
|
+
* Computes set difference of Theta sketches.
|
37
|
+
*/
|
28
38
|
template<typename Allocator = std::allocator<uint64_t>>
|
29
39
|
class theta_a_not_b_alloc {
|
30
40
|
public:
|
@@ -33,11 +43,19 @@ public:
|
|
33
43
|
using CompactSketch = compact_theta_sketch_alloc<Allocator>;
|
34
44
|
using State = theta_set_difference_base<Entry, ExtractKey, CompactSketch, Allocator>;
|
35
45
|
|
46
|
+
/**
|
47
|
+
* Constructor
|
48
|
+
* @param seed for the hash function that was used to create the sketch
|
49
|
+
* @param allocator to use for allocating and deallocating memory
|
50
|
+
*/
|
36
51
|
explicit theta_a_not_b_alloc(uint64_t seed = DEFAULT_SEED, const Allocator& allocator = Allocator());
|
37
52
|
|
38
53
|
/**
|
39
|
-
* Computes the
|
40
|
-
* @
|
54
|
+
* Computes the A-not-B set operation given two sketches.
|
55
|
+
* @param a sketch A
|
56
|
+
* @param b sketch B
|
57
|
+
* @param ordered optional flag to specify if an ordered sketch should be produced
|
58
|
+
* @return the result of A-not-B as a compact sketch
|
41
59
|
*/
|
42
60
|
template<typename FwdSketch, typename Sketch>
|
43
61
|
CompactSketch compute(FwdSketch&& a, const Sketch& b, bool ordered = true) const;
|
@@ -46,9 +64,6 @@ private:
|
|
46
64
|
State state_;
|
47
65
|
};
|
48
66
|
|
49
|
-
// alias with default allocator for convenience
|
50
|
-
using theta_a_not_b = theta_a_not_b_alloc<std::allocator<uint64_t>>;
|
51
|
-
|
52
67
|
} /* namespace datasketches */
|
53
68
|
|
54
69
|
#include "theta_a_not_b_impl.hpp"
|
@@ -25,15 +25,21 @@
|
|
25
25
|
|
26
26
|
namespace datasketches {
|
27
27
|
|
28
|
+
/// Theta constants
|
28
29
|
namespace theta_constants {
|
30
|
+
/// hash table resize factor
|
29
31
|
using resize_factor = datasketches::resize_factor;
|
30
|
-
|
31
|
-
const
|
32
|
+
/// default resize factor
|
33
|
+
const resize_factor DEFAULT_RESIZE_FACTOR = resize_factor::X8;
|
34
|
+
|
35
|
+
/// max theta - signed max for compatibility with Java
|
36
|
+
const uint64_t MAX_THETA = LLONG_MAX;
|
37
|
+
/// min log2 of K
|
32
38
|
const uint8_t MIN_LG_K = 5;
|
39
|
+
/// max log2 of K
|
33
40
|
const uint8_t MAX_LG_K = 26;
|
34
|
-
|
41
|
+
/// default log2 of K
|
35
42
|
const uint8_t DEFAULT_LG_K = 12;
|
36
|
-
const resize_factor DEFAULT_RESIZE_FACTOR = resize_factor::X8;
|
37
43
|
}
|
38
44
|
|
39
45
|
} /* namespace datasketches */
|
@@ -25,6 +25,16 @@
|
|
25
25
|
|
26
26
|
namespace datasketches {
|
27
27
|
|
28
|
+
// forward declaration
|
29
|
+
template<typename A> class theta_intersection_alloc;
|
30
|
+
|
31
|
+
// alias with default allocator for convenience
|
32
|
+
using theta_intersection = theta_intersection_alloc<std::allocator<uint64_t>>;
|
33
|
+
|
34
|
+
/**
|
35
|
+
* Theta intersection.
|
36
|
+
* Computes intersection of Theta sketches.
|
37
|
+
*/
|
28
38
|
template<typename Allocator = std::allocator<uint64_t>>
|
29
39
|
class theta_intersection_alloc {
|
30
40
|
public:
|
@@ -33,6 +43,7 @@ public:
|
|
33
43
|
using Sketch = theta_sketch_alloc<Allocator>;
|
34
44
|
using CompactSketch = compact_theta_sketch_alloc<Allocator>;
|
35
45
|
|
46
|
+
// there is no payload in Theta sketch entry
|
36
47
|
struct nop_policy {
|
37
48
|
void operator()(uint64_t internal_entry, uint64_t incoming_entry) const {
|
38
49
|
unused(incoming_entry);
|
@@ -41,7 +52,7 @@ public:
|
|
41
52
|
};
|
42
53
|
using State = theta_intersection_base<Entry, ExtractKey, nop_policy, Sketch, CompactSketch, Allocator>;
|
43
54
|
|
44
|
-
|
55
|
+
/**
|
45
56
|
* Constructor
|
46
57
|
* @param seed for the hash function that was used to create the sketch
|
47
58
|
* @param allocator to use for allocating and deallocating memory
|
@@ -61,7 +72,7 @@ public:
|
|
61
72
|
* Produces a copy of the current state of the intersection.
|
62
73
|
* If update() was not called, the state is the infinite "universe",
|
63
74
|
* which is considered an undefined state, and throws an exception.
|
64
|
-
* @param ordered optional flag to specify if ordered sketch should be produced
|
75
|
+
* @param ordered optional flag to specify if an ordered sketch should be produced
|
65
76
|
* @return the result of the intersection
|
66
77
|
*/
|
67
78
|
CompactSketch get_result(bool ordered = true) const;
|
@@ -76,9 +87,6 @@ private:
|
|
76
87
|
State state_;
|
77
88
|
};
|
78
89
|
|
79
|
-
// alias with default allocator for convenience
|
80
|
-
using theta_intersection = theta_intersection_alloc<std::allocator<uint64_t>>;
|
81
|
-
|
82
90
|
} /* namespace datasketches */
|
83
91
|
|
84
92
|
#include "theta_intersection_impl.hpp"
|
@@ -49,8 +49,8 @@ void theta_intersection_base<EN, EK, P, S, CS, A>::update(SS&& sketch) {
|
|
49
49
|
if (!is_valid_) { // first update, copy or move incoming sketch
|
50
50
|
is_valid_ = true;
|
51
51
|
const uint8_t lg_size = lg_size_from_count(sketch.get_num_retained(), theta_update_sketch_base<EN, EK, A>::REBUILD_THRESHOLD);
|
52
|
-
table_ = hash_table(lg_size, lg_size, resize_factor::X1, 1, table_.theta_, table_.seed_, table_.allocator_, table_.is_empty_);
|
53
|
-
for (auto
|
52
|
+
table_ = hash_table(lg_size, lg_size - 1, resize_factor::X1, 1, table_.theta_, table_.seed_, table_.allocator_, table_.is_empty_);
|
53
|
+
for (auto&& entry: sketch) {
|
54
54
|
auto result = table_.find(EK()(entry));
|
55
55
|
if (result.second) {
|
56
56
|
throw std::invalid_argument("duplicate key, possibly corrupted input sketch");
|
@@ -64,7 +64,7 @@ void theta_intersection_base<EN, EK, P, S, CS, A>::update(SS&& sketch) {
|
|
64
64
|
matched_entries.reserve(max_matches);
|
65
65
|
uint32_t match_count = 0;
|
66
66
|
uint32_t count = 0;
|
67
|
-
for (auto
|
67
|
+
for (auto&& entry: sketch) {
|
68
68
|
if (EK()(entry) < table_.theta_) {
|
69
69
|
auto result = table_.find(EK()(entry));
|
70
70
|
if (result.second) {
|
@@ -88,8 +88,8 @@ void theta_intersection_base<EN, EK, P, S, CS, A>::update(SS&& sketch) {
|
|
88
88
|
if (table_.theta_ == theta_constants::MAX_THETA) table_.is_empty_ = true;
|
89
89
|
} else {
|
90
90
|
const uint8_t lg_size = lg_size_from_count(match_count, theta_update_sketch_base<EN, EK, A>::REBUILD_THRESHOLD);
|
91
|
-
table_ = hash_table(lg_size, lg_size, resize_factor::X1, 1, table_.theta_, table_.seed_, table_.allocator_, table_.is_empty_);
|
92
|
-
for (uint32_t i = 0; i < match_count; i
|
91
|
+
table_ = hash_table(lg_size, lg_size - 1, resize_factor::X1, 1, table_.theta_, table_.seed_, table_.allocator_, table_.is_empty_);
|
92
|
+
for (uint32_t i = 0; i < match_count; ++i) {
|
93
93
|
auto result = table_.find(EK()(matched_entries[i]));
|
94
94
|
table_.insert(result.first, std::move(matched_entries[i]));
|
95
95
|
}
|
@@ -28,9 +28,9 @@ state_(seed, nop_policy(), allocator)
|
|
28
28
|
{}
|
29
29
|
|
30
30
|
template<typename A>
|
31
|
-
template<typename
|
32
|
-
void theta_intersection_alloc<A>::update(
|
33
|
-
state_.update(std::forward<
|
31
|
+
template<typename FwdSketch>
|
32
|
+
void theta_intersection_alloc<A>::update(FwdSketch&& sketch) {
|
33
|
+
state_.update(std::forward<FwdSketch>(sketch));
|
34
34
|
}
|
35
35
|
|
36
36
|
template<typename A>
|
@@ -26,10 +26,11 @@
|
|
26
26
|
|
27
27
|
namespace datasketches {
|
28
28
|
|
29
|
+
/// Theta Jaccard similarity alias
|
29
30
|
template<typename Allocator = std::allocator<uint64_t>>
|
30
31
|
using theta_jaccard_similarity_alloc = jaccard_similarity_base<theta_union_alloc<Allocator>, theta_intersection_alloc<Allocator>, trivial_extract_key>;
|
31
32
|
|
32
|
-
|
33
|
+
/// Theta Jaccard similarity alias with default allocator
|
33
34
|
using theta_jaccard_similarity = theta_jaccard_similarity_alloc<std::allocator<uint64_t>>;
|
34
35
|
|
35
36
|
} /* namespace datasketches */
|