datasketches 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/CHANGELOG.md +3 -0
- data/LICENSE +310 -0
- data/NOTICE +11 -0
- data/README.md +126 -0
- data/ext/datasketches/cpc_wrapper.cpp +50 -0
- data/ext/datasketches/ext.cpp +12 -0
- data/ext/datasketches/extconf.rb +11 -0
- data/ext/datasketches/hll_wrapper.cpp +69 -0
- data/lib/datasketches.rb +9 -0
- data/lib/datasketches/version.rb +3 -0
- data/vendor/datasketches-cpp/CMakeLists.txt +126 -0
- data/vendor/datasketches-cpp/LICENSE +311 -0
- data/vendor/datasketches-cpp/MANIFEST.in +19 -0
- data/vendor/datasketches-cpp/NOTICE +11 -0
- data/vendor/datasketches-cpp/README.md +42 -0
- data/vendor/datasketches-cpp/common/CMakeLists.txt +45 -0
- data/vendor/datasketches-cpp/common/include/MurmurHash3.h +173 -0
- data/vendor/datasketches-cpp/common/include/binomial_bounds.hpp +458 -0
- data/vendor/datasketches-cpp/common/include/bounds_binomial_proportions.hpp +291 -0
- data/vendor/datasketches-cpp/common/include/ceiling_power_of_2.hpp +41 -0
- data/vendor/datasketches-cpp/common/include/common_defs.hpp +51 -0
- data/vendor/datasketches-cpp/common/include/conditional_back_inserter.hpp +68 -0
- data/vendor/datasketches-cpp/common/include/conditional_forward.hpp +70 -0
- data/vendor/datasketches-cpp/common/include/count_zeros.hpp +114 -0
- data/vendor/datasketches-cpp/common/include/inv_pow2_table.hpp +107 -0
- data/vendor/datasketches-cpp/common/include/memory_operations.hpp +57 -0
- data/vendor/datasketches-cpp/common/include/serde.hpp +196 -0
- data/vendor/datasketches-cpp/common/test/CMakeLists.txt +38 -0
- data/vendor/datasketches-cpp/common/test/catch.hpp +17618 -0
- data/vendor/datasketches-cpp/common/test/catch_runner.cpp +7 -0
- data/vendor/datasketches-cpp/common/test/test_allocator.cpp +31 -0
- data/vendor/datasketches-cpp/common/test/test_allocator.hpp +108 -0
- data/vendor/datasketches-cpp/common/test/test_runner.cpp +29 -0
- data/vendor/datasketches-cpp/common/test/test_type.hpp +137 -0
- data/vendor/datasketches-cpp/cpc/CMakeLists.txt +74 -0
- data/vendor/datasketches-cpp/cpc/include/compression_data.hpp +6022 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_common.hpp +62 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor.hpp +147 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor_impl.hpp +742 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_confidence.hpp +167 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch.hpp +311 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch_impl.hpp +810 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_union.hpp +102 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_union_impl.hpp +346 -0
- data/vendor/datasketches-cpp/cpc/include/cpc_util.hpp +137 -0
- data/vendor/datasketches-cpp/cpc/include/icon_estimator.hpp +274 -0
- data/vendor/datasketches-cpp/cpc/include/kxp_byte_lookup.hpp +81 -0
- data/vendor/datasketches-cpp/cpc/include/u32_table.hpp +84 -0
- data/vendor/datasketches-cpp/cpc/include/u32_table_impl.hpp +266 -0
- data/vendor/datasketches-cpp/cpc/test/CMakeLists.txt +44 -0
- data/vendor/datasketches-cpp/cpc/test/compression_test.cpp +67 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_test.cpp +381 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_union_test.cpp +149 -0
- data/vendor/datasketches-cpp/fi/CMakeLists.txt +54 -0
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch.hpp +319 -0
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch_impl.hpp +484 -0
- data/vendor/datasketches-cpp/fi/include/reverse_purge_hash_map.hpp +114 -0
- data/vendor/datasketches-cpp/fi/include/reverse_purge_hash_map_impl.hpp +345 -0
- data/vendor/datasketches-cpp/fi/test/CMakeLists.txt +44 -0
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_custom_type_test.cpp +84 -0
- data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_test.cpp +360 -0
- 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/fi/test/reverse_purge_hash_map_test.cpp +47 -0
- data/vendor/datasketches-cpp/hll/CMakeLists.txt +92 -0
- data/vendor/datasketches-cpp/hll/include/AuxHashMap-internal.hpp +303 -0
- data/vendor/datasketches-cpp/hll/include/AuxHashMap.hpp +83 -0
- data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable-internal.hpp +811 -0
- data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable.hpp +40 -0
- data/vendor/datasketches-cpp/hll/include/CouponHashSet-internal.hpp +291 -0
- data/vendor/datasketches-cpp/hll/include/CouponHashSet.hpp +59 -0
- data/vendor/datasketches-cpp/hll/include/CouponList-internal.hpp +417 -0
- data/vendor/datasketches-cpp/hll/include/CouponList.hpp +91 -0
- data/vendor/datasketches-cpp/hll/include/CubicInterpolation-internal.hpp +233 -0
- data/vendor/datasketches-cpp/hll/include/CubicInterpolation.hpp +43 -0
- data/vendor/datasketches-cpp/hll/include/HarmonicNumbers-internal.hpp +90 -0
- data/vendor/datasketches-cpp/hll/include/HarmonicNumbers.hpp +48 -0
- data/vendor/datasketches-cpp/hll/include/Hll4Array-internal.hpp +335 -0
- data/vendor/datasketches-cpp/hll/include/Hll4Array.hpp +69 -0
- data/vendor/datasketches-cpp/hll/include/Hll6Array-internal.hpp +124 -0
- data/vendor/datasketches-cpp/hll/include/Hll6Array.hpp +55 -0
- data/vendor/datasketches-cpp/hll/include/Hll8Array-internal.hpp +158 -0
- data/vendor/datasketches-cpp/hll/include/Hll8Array.hpp +56 -0
- data/vendor/datasketches-cpp/hll/include/HllArray-internal.hpp +706 -0
- data/vendor/datasketches-cpp/hll/include/HllArray.hpp +136 -0
- data/vendor/datasketches-cpp/hll/include/HllSketch-internal.hpp +462 -0
- data/vendor/datasketches-cpp/hll/include/HllSketchImpl-internal.hpp +149 -0
- data/vendor/datasketches-cpp/hll/include/HllSketchImpl.hpp +85 -0
- data/vendor/datasketches-cpp/hll/include/HllSketchImplFactory.hpp +170 -0
- data/vendor/datasketches-cpp/hll/include/HllUnion-internal.hpp +287 -0
- data/vendor/datasketches-cpp/hll/include/HllUtil.hpp +239 -0
- data/vendor/datasketches-cpp/hll/include/RelativeErrorTables-internal.hpp +112 -0
- data/vendor/datasketches-cpp/hll/include/RelativeErrorTables.hpp +46 -0
- data/vendor/datasketches-cpp/hll/include/coupon_iterator-internal.hpp +56 -0
- data/vendor/datasketches-cpp/hll/include/coupon_iterator.hpp +43 -0
- data/vendor/datasketches-cpp/hll/include/hll.hpp +669 -0
- data/vendor/datasketches-cpp/hll/include/hll.private.hpp +32 -0
- data/vendor/datasketches-cpp/hll/test/AuxHashMapTest.cpp +79 -0
- data/vendor/datasketches-cpp/hll/test/CMakeLists.txt +51 -0
- data/vendor/datasketches-cpp/hll/test/CouponHashSetTest.cpp +130 -0
- data/vendor/datasketches-cpp/hll/test/CouponListTest.cpp +181 -0
- data/vendor/datasketches-cpp/hll/test/CrossCountingTest.cpp +93 -0
- data/vendor/datasketches-cpp/hll/test/HllArrayTest.cpp +191 -0
- data/vendor/datasketches-cpp/hll/test/HllSketchTest.cpp +389 -0
- data/vendor/datasketches-cpp/hll/test/HllUnionTest.cpp +313 -0
- data/vendor/datasketches-cpp/hll/test/IsomorphicTest.cpp +141 -0
- data/vendor/datasketches-cpp/hll/test/TablesTest.cpp +44 -0
- data/vendor/datasketches-cpp/hll/test/ToFromByteArrayTest.cpp +168 -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/CMakeLists.txt +58 -0
- data/vendor/datasketches-cpp/kll/include/kll_helper.hpp +150 -0
- data/vendor/datasketches-cpp/kll/include/kll_helper_impl.hpp +319 -0
- data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator.hpp +67 -0
- data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator_impl.hpp +169 -0
- data/vendor/datasketches-cpp/kll/include/kll_sketch.hpp +559 -0
- data/vendor/datasketches-cpp/kll/include/kll_sketch_impl.hpp +1131 -0
- data/vendor/datasketches-cpp/kll/test/CMakeLists.txt +44 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_custom_type_test.cpp +154 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_float_one_item_v1.sk +0 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_from_java.sk +0 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_test.cpp +685 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_validation.cpp +229 -0
- data/vendor/datasketches-cpp/pyproject.toml +17 -0
- data/vendor/datasketches-cpp/python/CMakeLists.txt +61 -0
- data/vendor/datasketches-cpp/python/README.md +78 -0
- data/vendor/datasketches-cpp/python/jupyter/CPCSketch.ipynb +345 -0
- data/vendor/datasketches-cpp/python/jupyter/FrequentItemsSketch.ipynb +354 -0
- data/vendor/datasketches-cpp/python/jupyter/HLLSketch.ipynb +346 -0
- data/vendor/datasketches-cpp/python/jupyter/KLLSketch.ipynb +463 -0
- data/vendor/datasketches-cpp/python/jupyter/ThetaSketchNotebook.ipynb +396 -0
- data/vendor/datasketches-cpp/python/src/__init__.py +2 -0
- data/vendor/datasketches-cpp/python/src/cpc_wrapper.cpp +90 -0
- data/vendor/datasketches-cpp/python/src/datasketches.cpp +40 -0
- data/vendor/datasketches-cpp/python/src/fi_wrapper.cpp +123 -0
- data/vendor/datasketches-cpp/python/src/hll_wrapper.cpp +136 -0
- data/vendor/datasketches-cpp/python/src/kll_wrapper.cpp +209 -0
- data/vendor/datasketches-cpp/python/src/theta_wrapper.cpp +162 -0
- data/vendor/datasketches-cpp/python/src/vector_of_kll.cpp +488 -0
- data/vendor/datasketches-cpp/python/src/vo_wrapper.cpp +140 -0
- data/vendor/datasketches-cpp/python/tests/__init__.py +0 -0
- data/vendor/datasketches-cpp/python/tests/cpc_test.py +64 -0
- data/vendor/datasketches-cpp/python/tests/fi_test.py +110 -0
- data/vendor/datasketches-cpp/python/tests/hll_test.py +131 -0
- data/vendor/datasketches-cpp/python/tests/kll_test.py +119 -0
- data/vendor/datasketches-cpp/python/tests/theta_test.py +121 -0
- data/vendor/datasketches-cpp/python/tests/vector_of_kll_test.py +148 -0
- data/vendor/datasketches-cpp/python/tests/vo_test.py +101 -0
- data/vendor/datasketches-cpp/sampling/CMakeLists.txt +48 -0
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch.hpp +392 -0
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch_impl.hpp +1752 -0
- data/vendor/datasketches-cpp/sampling/include/var_opt_union.hpp +239 -0
- data/vendor/datasketches-cpp/sampling/include/var_opt_union_impl.hpp +645 -0
- data/vendor/datasketches-cpp/sampling/test/CMakeLists.txt +43 -0
- data/vendor/datasketches-cpp/sampling/test/binaries_from_java.txt +67 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_test.cpp +509 -0
- data/vendor/datasketches-cpp/sampling/test/var_opt_union_test.cpp +358 -0
- 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 +94 -0
- data/vendor/datasketches-cpp/theta/CMakeLists.txt +57 -0
- data/vendor/datasketches-cpp/theta/include/theta_a_not_b.hpp +73 -0
- data/vendor/datasketches-cpp/theta/include/theta_a_not_b_impl.hpp +83 -0
- data/vendor/datasketches-cpp/theta/include/theta_intersection.hpp +88 -0
- data/vendor/datasketches-cpp/theta/include/theta_intersection_impl.hpp +130 -0
- data/vendor/datasketches-cpp/theta/include/theta_sketch.hpp +533 -0
- data/vendor/datasketches-cpp/theta/include/theta_sketch_impl.hpp +939 -0
- data/vendor/datasketches-cpp/theta/include/theta_union.hpp +122 -0
- data/vendor/datasketches-cpp/theta/include/theta_union_impl.hpp +109 -0
- data/vendor/datasketches-cpp/theta/test/CMakeLists.txt +45 -0
- data/vendor/datasketches-cpp/theta/test/theta_a_not_b_test.cpp +244 -0
- 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_single_item_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_intersection_test.cpp +218 -0
- data/vendor/datasketches-cpp/theta/test/theta_sketch_test.cpp +438 -0
- data/vendor/datasketches-cpp/theta/test/theta_union_test.cpp +97 -0
- 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
- data/vendor/datasketches-cpp/tuple/CMakeLists.txt +104 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_a_not_b.hpp +52 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_a_not_b_impl.hpp +32 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_intersection.hpp +52 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_intersection_impl.hpp +31 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch.hpp +179 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch_impl.hpp +238 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union.hpp +81 -0
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union_impl.hpp +43 -0
- data/vendor/datasketches-cpp/tuple/include/bounds_on_ratios_in_sampled_sets.hpp +135 -0
- data/vendor/datasketches-cpp/tuple/include/bounds_on_ratios_in_theta_sketched_sets.hpp +135 -0
- data/vendor/datasketches-cpp/tuple/include/jaccard_similarity.hpp +172 -0
- data/vendor/datasketches-cpp/tuple/include/theta_a_not_b_experimental.hpp +53 -0
- data/vendor/datasketches-cpp/tuple/include/theta_a_not_b_experimental_impl.hpp +33 -0
- data/vendor/datasketches-cpp/tuple/include/theta_comparators.hpp +48 -0
- data/vendor/datasketches-cpp/tuple/include/theta_constants.hpp +34 -0
- data/vendor/datasketches-cpp/tuple/include/theta_helpers.hpp +54 -0
- data/vendor/datasketches-cpp/tuple/include/theta_intersection_base.hpp +59 -0
- data/vendor/datasketches-cpp/tuple/include/theta_intersection_base_impl.hpp +121 -0
- data/vendor/datasketches-cpp/tuple/include/theta_intersection_experimental.hpp +78 -0
- data/vendor/datasketches-cpp/tuple/include/theta_intersection_experimental_impl.hpp +43 -0
- data/vendor/datasketches-cpp/tuple/include/theta_set_difference_base.hpp +54 -0
- data/vendor/datasketches-cpp/tuple/include/theta_set_difference_base_impl.hpp +80 -0
- data/vendor/datasketches-cpp/tuple/include/theta_sketch_experimental.hpp +393 -0
- data/vendor/datasketches-cpp/tuple/include/theta_sketch_experimental_impl.hpp +481 -0
- data/vendor/datasketches-cpp/tuple/include/theta_union_base.hpp +60 -0
- data/vendor/datasketches-cpp/tuple/include/theta_union_base_impl.hpp +84 -0
- data/vendor/datasketches-cpp/tuple/include/theta_union_experimental.hpp +88 -0
- data/vendor/datasketches-cpp/tuple/include/theta_union_experimental_impl.hpp +47 -0
- data/vendor/datasketches-cpp/tuple/include/theta_update_sketch_base.hpp +259 -0
- data/vendor/datasketches-cpp/tuple/include/theta_update_sketch_base_impl.hpp +389 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_a_not_b.hpp +57 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_a_not_b_impl.hpp +33 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_intersection.hpp +104 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_intersection_impl.hpp +43 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch.hpp +496 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch_impl.hpp +587 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_union.hpp +109 -0
- data/vendor/datasketches-cpp/tuple/include/tuple_union_impl.hpp +47 -0
- data/vendor/datasketches-cpp/tuple/test/CMakeLists.txt +53 -0
- data/vendor/datasketches-cpp/tuple/test/aod_1_compact_empty_from_java.sk +1 -0
- 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 +1 -0
- data/vendor/datasketches-cpp/tuple/test/array_of_doubles_sketch_test.cpp +298 -0
- data/vendor/datasketches-cpp/tuple/test/theta_a_not_b_experimental_test.cpp +250 -0
- data/vendor/datasketches-cpp/tuple/test/theta_compact_empty_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/test/theta_compact_estimation_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/test/theta_compact_single_item_from_java.sk +0 -0
- data/vendor/datasketches-cpp/tuple/test/theta_intersection_experimental_test.cpp +224 -0
- data/vendor/datasketches-cpp/tuple/test/theta_jaccard_similarity_test.cpp +144 -0
- data/vendor/datasketches-cpp/tuple/test/theta_sketch_experimental_test.cpp +247 -0
- data/vendor/datasketches-cpp/tuple/test/theta_union_experimental_test.cpp +44 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_a_not_b_test.cpp +289 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_intersection_test.cpp +235 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_jaccard_similarity_test.cpp +98 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_allocation_test.cpp +102 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_test.cpp +249 -0
- data/vendor/datasketches-cpp/tuple/test/tuple_union_test.cpp +187 -0
- metadata +302 -0
@@ -0,0 +1,144 @@
|
|
1
|
+
/*
|
2
|
+
* Licensed to the Apache Software Foundation (ASF) under one
|
3
|
+
* or more contributor license agreements. See the NOTICE file
|
4
|
+
* distributed with this work for additional information
|
5
|
+
* regarding copyright ownership. The ASF licenses this file
|
6
|
+
* to you under the Apache License, Version 2.0 (the
|
7
|
+
* "License"); you may not use this file except in compliance
|
8
|
+
* with the License. You may obtain a copy of the License at
|
9
|
+
*
|
10
|
+
* http://www.apache.org/licenses/LICENSE-2.0
|
11
|
+
*
|
12
|
+
* Unless required by applicable law or agreed to in writing,
|
13
|
+
* software distributed under the License is distributed on an
|
14
|
+
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
15
|
+
* KIND, either express or implied. See the License for the
|
16
|
+
* specific language governing permissions and limitations
|
17
|
+
* under the License.
|
18
|
+
*/
|
19
|
+
|
20
|
+
#include <iostream>
|
21
|
+
|
22
|
+
#include <catch.hpp>
|
23
|
+
#include <jaccard_similarity.hpp>
|
24
|
+
|
25
|
+
namespace datasketches {
|
26
|
+
|
27
|
+
using update_theta_sketch = update_theta_sketch_experimental<>;
|
28
|
+
|
29
|
+
TEST_CASE("theta jaccard: empty", "[theta_sketch]") {
|
30
|
+
auto sk_a = update_theta_sketch::builder().build();
|
31
|
+
auto sk_b = update_theta_sketch::builder().build();
|
32
|
+
auto jc = theta_jaccard_similarity::jaccard(sk_a, sk_b);
|
33
|
+
REQUIRE(jc == std::array<double, 3>{1, 1, 1});
|
34
|
+
}
|
35
|
+
|
36
|
+
TEST_CASE("theta jaccard: same sketch exact mode", "[theta_sketch]") {
|
37
|
+
auto sk = update_theta_sketch::builder().build();
|
38
|
+
for (int i = 0; i < 1000; ++i) sk.update(i);
|
39
|
+
|
40
|
+
// update sketch
|
41
|
+
auto jc = theta_jaccard_similarity::jaccard(sk, sk);
|
42
|
+
REQUIRE(jc == std::array<double, 3>{1, 1, 1});
|
43
|
+
|
44
|
+
// compact sketch
|
45
|
+
jc = theta_jaccard_similarity::jaccard(sk.compact(), sk.compact());
|
46
|
+
REQUIRE(jc == std::array<double, 3>{1, 1, 1});
|
47
|
+
}
|
48
|
+
|
49
|
+
TEST_CASE("theta jaccard: full overlap exact mode", "[theta_sketch]") {
|
50
|
+
auto sk_a = update_theta_sketch::builder().build();
|
51
|
+
auto sk_b = update_theta_sketch::builder().build();
|
52
|
+
for (int i = 0; i < 1000; ++i) {
|
53
|
+
sk_a.update(i);
|
54
|
+
sk_b.update(i);
|
55
|
+
}
|
56
|
+
|
57
|
+
// update sketches
|
58
|
+
auto jc = theta_jaccard_similarity::jaccard(sk_a, sk_b);
|
59
|
+
REQUIRE(jc == std::array<double, 3>{1, 1, 1});
|
60
|
+
|
61
|
+
// compact sketches
|
62
|
+
jc = theta_jaccard_similarity::jaccard(sk_a.compact(), sk_b.compact());
|
63
|
+
REQUIRE(jc == std::array<double, 3>{1, 1, 1});
|
64
|
+
}
|
65
|
+
|
66
|
+
TEST_CASE("theta jaccard: disjoint exact mode", "[theta_sketch]") {
|
67
|
+
auto sk_a = update_theta_sketch::builder().build();
|
68
|
+
auto sk_b = update_theta_sketch::builder().build();
|
69
|
+
for (int i = 0; i < 1000; ++i) {
|
70
|
+
sk_a.update(i);
|
71
|
+
sk_b.update(i + 1000);
|
72
|
+
}
|
73
|
+
|
74
|
+
// update sketches
|
75
|
+
auto jc = theta_jaccard_similarity::jaccard(sk_a, sk_b);
|
76
|
+
REQUIRE(jc == std::array<double, 3>{0, 0, 0});
|
77
|
+
|
78
|
+
// compact sketches
|
79
|
+
jc = theta_jaccard_similarity::jaccard(sk_a.compact(), sk_b.compact());
|
80
|
+
REQUIRE(jc == std::array<double, 3>{0, 0, 0});
|
81
|
+
}
|
82
|
+
|
83
|
+
TEST_CASE("theta jaccard: half overlap estimation mode", "[theta_sketch]") {
|
84
|
+
auto sk_a = update_theta_sketch::builder().build();
|
85
|
+
auto sk_b = update_theta_sketch::builder().build();
|
86
|
+
for (int i = 0; i < 10000; ++i) {
|
87
|
+
sk_a.update(i);
|
88
|
+
sk_b.update(i + 5000);
|
89
|
+
}
|
90
|
+
|
91
|
+
// update sketches
|
92
|
+
auto jc = theta_jaccard_similarity::jaccard(sk_a, sk_b);
|
93
|
+
REQUIRE(jc[0] == Approx(0.33).margin(0.01));
|
94
|
+
REQUIRE(jc[1] == Approx(0.33).margin(0.01));
|
95
|
+
REQUIRE(jc[2] == Approx(0.33).margin(0.01));
|
96
|
+
|
97
|
+
// compact sketches
|
98
|
+
jc = theta_jaccard_similarity::jaccard(sk_a.compact(), sk_b.compact());
|
99
|
+
REQUIRE(jc[0] == Approx(0.33).margin(0.01));
|
100
|
+
REQUIRE(jc[1] == Approx(0.33).margin(0.01));
|
101
|
+
REQUIRE(jc[2] == Approx(0.33).margin(0.01));
|
102
|
+
}
|
103
|
+
|
104
|
+
/**
|
105
|
+
* The distribution is quite tight, about +/- 0.7%, which is pretty good since the accuracy of the
|
106
|
+
* underlying sketch is about +/- 1.56%.
|
107
|
+
*/
|
108
|
+
TEST_CASE("theta jaccard: similarity test", "[theta_sketch]") {
|
109
|
+
const int8_t min_lg_k = 12;
|
110
|
+
const int u1 = 1 << 20;
|
111
|
+
const int u2 = u1 * 0.95;
|
112
|
+
const double threshold = 0.943;
|
113
|
+
|
114
|
+
auto expected = update_theta_sketch::builder().set_lg_k(min_lg_k).build();
|
115
|
+
for (int i = 0; i < u1; ++i) expected.update(i);
|
116
|
+
|
117
|
+
auto actual = update_theta_sketch::builder().set_lg_k(min_lg_k).build();
|
118
|
+
for (int i = 0; i < u2; ++i) actual.update(i);
|
119
|
+
|
120
|
+
REQUIRE(theta_jaccard_similarity::similarity_test(actual, expected, threshold));
|
121
|
+
REQUIRE(theta_jaccard_similarity::similarity_test(actual, actual, threshold));
|
122
|
+
}
|
123
|
+
|
124
|
+
/**
|
125
|
+
* The distribution is much looser here, about +/- 14%. This is due to the fact that intersections loose accuracy
|
126
|
+
* as the ratio of intersection to the union becomes a small number.
|
127
|
+
*/
|
128
|
+
TEST_CASE("theta jaccard: dissimilarity test", "[theta_sketch]") {
|
129
|
+
const int8_t min_lg_k = 12;
|
130
|
+
const int u1 = 1 << 20;
|
131
|
+
const int u2 = u1 * 0.05;
|
132
|
+
const double threshold = 0.061;
|
133
|
+
|
134
|
+
auto expected = update_theta_sketch::builder().set_lg_k(min_lg_k).build();
|
135
|
+
for (int i = 0; i < u1; ++i) expected.update(i);
|
136
|
+
|
137
|
+
auto actual = update_theta_sketch::builder().set_lg_k(min_lg_k).build();
|
138
|
+
for (int i = 0; i < u2; ++i) actual.update(i);
|
139
|
+
|
140
|
+
REQUIRE(theta_jaccard_similarity::dissimilarity_test(actual, expected, threshold));
|
141
|
+
REQUIRE_FALSE(theta_jaccard_similarity::dissimilarity_test(actual, actual, threshold));
|
142
|
+
}
|
143
|
+
|
144
|
+
} /* namespace datasketches */
|
@@ -0,0 +1,247 @@
|
|
1
|
+
/*
|
2
|
+
* Licensed to the Apache Software Foundation (ASF) under one
|
3
|
+
* or more contributor license agreements. See the NOTICE file
|
4
|
+
* distributed with this work for additional information
|
5
|
+
* regarding copyright ownership. The ASF licenses this file
|
6
|
+
* to you under the Apache License, Version 2.0 (the
|
7
|
+
* "License"); you may not use this file except in compliance
|
8
|
+
* with the License. You may obtain a copy of the License at
|
9
|
+
*
|
10
|
+
* http://www.apache.org/licenses/LICENSE-2.0
|
11
|
+
*
|
12
|
+
* Unless required by applicable law or agreed to in writing,
|
13
|
+
* software distributed under the License is distributed on an
|
14
|
+
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
15
|
+
* KIND, either express or implied. See the License for the
|
16
|
+
* specific language governing permissions and limitations
|
17
|
+
* under the License.
|
18
|
+
*/
|
19
|
+
|
20
|
+
#include <fstream>
|
21
|
+
#include <sstream>
|
22
|
+
|
23
|
+
#include <catch.hpp>
|
24
|
+
#include <theta_sketch_experimental.hpp>
|
25
|
+
|
26
|
+
namespace datasketches {
|
27
|
+
|
28
|
+
#ifdef TEST_BINARY_INPUT_PATH
|
29
|
+
const std::string inputPath = TEST_BINARY_INPUT_PATH;
|
30
|
+
#else
|
31
|
+
const std::string inputPath = "test/";
|
32
|
+
#endif
|
33
|
+
|
34
|
+
// These tests have been copied from the existing theta sketch implementation.
|
35
|
+
// Serialization as base class and serialization of update sketch have been removed.
|
36
|
+
|
37
|
+
using update_theta_sketch = update_theta_sketch_experimental<>;
|
38
|
+
using compact_theta_sketch = compact_theta_sketch_experimental<>;
|
39
|
+
|
40
|
+
TEST_CASE("theta sketch: empty", "[theta_sketch]") {
|
41
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().build();
|
42
|
+
REQUIRE(update_sketch.is_empty());
|
43
|
+
REQUIRE_FALSE(update_sketch.is_estimation_mode());
|
44
|
+
REQUIRE(update_sketch.get_theta() == 1.0);
|
45
|
+
REQUIRE(update_sketch.get_estimate() == 0.0);
|
46
|
+
REQUIRE(update_sketch.get_lower_bound(1) == 0.0);
|
47
|
+
REQUIRE(update_sketch.get_upper_bound(1) == 0.0);
|
48
|
+
|
49
|
+
compact_theta_sketch compact_sketch = update_sketch.compact();
|
50
|
+
REQUIRE(compact_sketch.is_empty());
|
51
|
+
REQUIRE_FALSE(compact_sketch.is_estimation_mode());
|
52
|
+
REQUIRE(compact_sketch.get_theta() == 1.0);
|
53
|
+
REQUIRE(compact_sketch.get_estimate() == 0.0);
|
54
|
+
REQUIRE(compact_sketch.get_lower_bound(1) == 0.0);
|
55
|
+
REQUIRE(compact_sketch.get_upper_bound(1) == 0.0);
|
56
|
+
}
|
57
|
+
|
58
|
+
TEST_CASE("theta sketch: non empty no retained keys", "[theta_sketch]") {
|
59
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().set_p(0.001).build();
|
60
|
+
update_sketch.update(1);
|
61
|
+
//std::cerr << update_sketch.to_string();
|
62
|
+
REQUIRE(update_sketch.get_num_retained() == 0);
|
63
|
+
REQUIRE_FALSE(update_sketch.is_empty());
|
64
|
+
REQUIRE(update_sketch.is_estimation_mode());
|
65
|
+
REQUIRE(update_sketch.get_estimate() == 0.0);
|
66
|
+
REQUIRE(update_sketch.get_lower_bound(1) == 0.0);
|
67
|
+
REQUIRE(update_sketch.get_upper_bound(1) > 0);
|
68
|
+
|
69
|
+
compact_theta_sketch compact_sketch = update_sketch.compact();
|
70
|
+
REQUIRE(compact_sketch.get_num_retained() == 0);
|
71
|
+
REQUIRE_FALSE(compact_sketch.is_empty());
|
72
|
+
REQUIRE(compact_sketch.is_estimation_mode());
|
73
|
+
REQUIRE(compact_sketch.get_estimate() == 0.0);
|
74
|
+
REQUIRE(compact_sketch.get_lower_bound(1) == 0.0);
|
75
|
+
REQUIRE(compact_sketch.get_upper_bound(1) > 0);
|
76
|
+
}
|
77
|
+
|
78
|
+
TEST_CASE("theta sketch: single item", "[theta_sketch]") {
|
79
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().build();
|
80
|
+
update_sketch.update(1);
|
81
|
+
REQUIRE_FALSE(update_sketch.is_empty());
|
82
|
+
REQUIRE_FALSE(update_sketch.is_estimation_mode());
|
83
|
+
REQUIRE(update_sketch.get_theta() == 1.0);
|
84
|
+
REQUIRE(update_sketch.get_estimate() == 1.0);
|
85
|
+
REQUIRE(update_sketch.get_lower_bound(1) == 1.0);
|
86
|
+
REQUIRE(update_sketch.get_upper_bound(1) == 1.0);
|
87
|
+
|
88
|
+
compact_theta_sketch compact_sketch = update_sketch.compact();
|
89
|
+
REQUIRE_FALSE(compact_sketch.is_empty());
|
90
|
+
REQUIRE_FALSE(compact_sketch.is_estimation_mode());
|
91
|
+
REQUIRE(compact_sketch.get_theta() == 1.0);
|
92
|
+
REQUIRE(compact_sketch.get_estimate() == 1.0);
|
93
|
+
REQUIRE(compact_sketch.get_lower_bound(1) == 1.0);
|
94
|
+
REQUIRE(compact_sketch.get_upper_bound(1) == 1.0);
|
95
|
+
}
|
96
|
+
|
97
|
+
TEST_CASE("theta sketch: resize exact", "[theta_sketch]") {
|
98
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().build();
|
99
|
+
for (int i = 0; i < 2000; i++) update_sketch.update(i);
|
100
|
+
REQUIRE_FALSE(update_sketch.is_empty());
|
101
|
+
REQUIRE_FALSE(update_sketch.is_estimation_mode());
|
102
|
+
REQUIRE(update_sketch.get_theta() == 1.0);
|
103
|
+
REQUIRE(update_sketch.get_estimate() == 2000.0);
|
104
|
+
REQUIRE(update_sketch.get_lower_bound(1) == 2000.0);
|
105
|
+
REQUIRE(update_sketch.get_upper_bound(1) == 2000.0);
|
106
|
+
|
107
|
+
compact_theta_sketch compact_sketch = update_sketch.compact();
|
108
|
+
REQUIRE_FALSE(compact_sketch.is_empty());
|
109
|
+
REQUIRE_FALSE(compact_sketch.is_estimation_mode());
|
110
|
+
REQUIRE(compact_sketch.get_theta() == 1.0);
|
111
|
+
REQUIRE(compact_sketch.get_estimate() == 2000.0);
|
112
|
+
REQUIRE(compact_sketch.get_lower_bound(1) == 2000.0);
|
113
|
+
REQUIRE(compact_sketch.get_upper_bound(1) == 2000.0);
|
114
|
+
}
|
115
|
+
|
116
|
+
TEST_CASE("theta sketch: estimation", "[theta_sketch]") {
|
117
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().set_resize_factor(update_theta_sketch::resize_factor::X1).build();
|
118
|
+
const int n = 8000;
|
119
|
+
for (int i = 0; i < n; i++) update_sketch.update(i);
|
120
|
+
//std::cerr << update_sketch.to_string();
|
121
|
+
REQUIRE_FALSE(update_sketch.is_empty());
|
122
|
+
REQUIRE(update_sketch.is_estimation_mode());
|
123
|
+
REQUIRE(update_sketch.get_theta() < 1.0);
|
124
|
+
REQUIRE(update_sketch.get_estimate() == Approx((double) n).margin(n * 0.01));
|
125
|
+
REQUIRE(update_sketch.get_lower_bound(1) < n);
|
126
|
+
REQUIRE(update_sketch.get_upper_bound(1) > n);
|
127
|
+
|
128
|
+
const uint32_t k = 1 << update_theta_sketch::builder::DEFAULT_LG_K;
|
129
|
+
REQUIRE(update_sketch.get_num_retained() >= k);
|
130
|
+
update_sketch.trim();
|
131
|
+
REQUIRE(update_sketch.get_num_retained() == k);
|
132
|
+
|
133
|
+
compact_theta_sketch compact_sketch = update_sketch.compact();
|
134
|
+
REQUIRE_FALSE(compact_sketch.is_empty());
|
135
|
+
REQUIRE(compact_sketch.is_ordered());
|
136
|
+
REQUIRE(compact_sketch.is_estimation_mode());
|
137
|
+
REQUIRE(compact_sketch.get_theta() < 1.0);
|
138
|
+
REQUIRE(compact_sketch.get_estimate() == Approx((double) n).margin(n * 0.01));
|
139
|
+
REQUIRE(compact_sketch.get_lower_bound(1) < n);
|
140
|
+
REQUIRE(compact_sketch.get_upper_bound(1) > n);
|
141
|
+
}
|
142
|
+
|
143
|
+
TEST_CASE("theta sketch: deserialize compact empty from java", "[theta_sketch]") {
|
144
|
+
std::ifstream is;
|
145
|
+
is.exceptions(std::ios::failbit | std::ios::badbit);
|
146
|
+
is.open(inputPath + "theta_compact_empty_from_java.sk", std::ios::binary);
|
147
|
+
auto sketch = compact_theta_sketch::deserialize(is);
|
148
|
+
REQUIRE(sketch.is_empty());
|
149
|
+
REQUIRE_FALSE(sketch.is_estimation_mode());
|
150
|
+
REQUIRE(sketch.get_num_retained() == 0);
|
151
|
+
REQUIRE(sketch.get_theta() == 1.0);
|
152
|
+
REQUIRE(sketch.get_estimate() == 0.0);
|
153
|
+
REQUIRE(sketch.get_lower_bound(1) == 0.0);
|
154
|
+
REQUIRE(sketch.get_upper_bound(1) == 0.0);
|
155
|
+
}
|
156
|
+
|
157
|
+
TEST_CASE("theta sketch: deserialize single item from java", "[theta_sketch]") {
|
158
|
+
std::ifstream is;
|
159
|
+
is.exceptions(std::ios::failbit | std::ios::badbit);
|
160
|
+
is.open(inputPath + "theta_compact_single_item_from_java.sk", std::ios::binary);
|
161
|
+
auto sketch = compact_theta_sketch::deserialize(is);
|
162
|
+
REQUIRE_FALSE(sketch.is_empty());
|
163
|
+
REQUIRE_FALSE(sketch.is_estimation_mode());
|
164
|
+
REQUIRE(sketch.get_num_retained() == 1);
|
165
|
+
REQUIRE(sketch.get_theta() == 1.0);
|
166
|
+
REQUIRE(sketch.get_estimate() == 1.0);
|
167
|
+
REQUIRE(sketch.get_lower_bound(1) == 1.0);
|
168
|
+
REQUIRE(sketch.get_upper_bound(1) == 1.0);
|
169
|
+
}
|
170
|
+
|
171
|
+
TEST_CASE("theta sketch: deserialize compact estimation from java", "[theta_sketch]") {
|
172
|
+
std::ifstream is;
|
173
|
+
is.exceptions(std::ios::failbit | std::ios::badbit);
|
174
|
+
is.open(inputPath + "theta_compact_estimation_from_java.sk", std::ios::binary);
|
175
|
+
auto sketch = compact_theta_sketch::deserialize(is);
|
176
|
+
REQUIRE_FALSE(sketch.is_empty());
|
177
|
+
REQUIRE(sketch.is_estimation_mode());
|
178
|
+
REQUIRE(sketch.is_ordered());
|
179
|
+
REQUIRE(sketch.get_num_retained() == 4342);
|
180
|
+
REQUIRE(sketch.get_theta() == Approx(0.531700444213199).margin(1e-10));
|
181
|
+
REQUIRE(sketch.get_estimate() == Approx(8166.25234614053).margin(1e-10));
|
182
|
+
REQUIRE(sketch.get_lower_bound(2) == Approx(7996.956955317471).margin(1e-10));
|
183
|
+
REQUIRE(sketch.get_upper_bound(2) == Approx(8339.090301078124).margin(1e-10));
|
184
|
+
|
185
|
+
// the same construction process in Java must have produced exactly the same sketch
|
186
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().build();
|
187
|
+
const int n = 8192;
|
188
|
+
for (int i = 0; i < n; i++) update_sketch.update(i);
|
189
|
+
REQUIRE(sketch.get_num_retained() == update_sketch.get_num_retained());
|
190
|
+
REQUIRE(sketch.get_theta() == Approx(update_sketch.get_theta()).margin(1e-10));
|
191
|
+
REQUIRE(sketch.get_estimate() == Approx(update_sketch.get_estimate()).margin(1e-10));
|
192
|
+
REQUIRE(sketch.get_lower_bound(1) == Approx(update_sketch.get_lower_bound(1)).margin(1e-10));
|
193
|
+
REQUIRE(sketch.get_upper_bound(1) == Approx(update_sketch.get_upper_bound(1)).margin(1e-10));
|
194
|
+
REQUIRE(sketch.get_lower_bound(2) == Approx(update_sketch.get_lower_bound(2)).margin(1e-10));
|
195
|
+
REQUIRE(sketch.get_upper_bound(2) == Approx(update_sketch.get_upper_bound(2)).margin(1e-10));
|
196
|
+
REQUIRE(sketch.get_lower_bound(3) == Approx(update_sketch.get_lower_bound(3)).margin(1e-10));
|
197
|
+
REQUIRE(sketch.get_upper_bound(3) == Approx(update_sketch.get_upper_bound(3)).margin(1e-10));
|
198
|
+
compact_theta_sketch compact_sketch = update_sketch.compact();
|
199
|
+
// the sketches are ordered, so the iteration sequence must match exactly
|
200
|
+
auto iter = sketch.begin();
|
201
|
+
for (const auto& key: compact_sketch) {
|
202
|
+
REQUIRE(*iter == key);
|
203
|
+
++iter;
|
204
|
+
}
|
205
|
+
}
|
206
|
+
|
207
|
+
TEST_CASE("theta sketch: serialize deserialize stream and bytes equivalence", "[theta_sketch]") {
|
208
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().build();
|
209
|
+
const int n = 8192;
|
210
|
+
for (int i = 0; i < n; i++) update_sketch.update(i);
|
211
|
+
|
212
|
+
std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
|
213
|
+
update_sketch.compact().serialize(s);
|
214
|
+
auto bytes = update_sketch.compact().serialize();
|
215
|
+
REQUIRE(bytes.size() == static_cast<size_t>(s.tellp()));
|
216
|
+
for (size_t i = 0; i < bytes.size(); ++i) {
|
217
|
+
REQUIRE(((char*)bytes.data())[i] == (char)s.get());
|
218
|
+
}
|
219
|
+
|
220
|
+
s.seekg(0); // rewind
|
221
|
+
compact_theta_sketch deserialized_sketch1 = compact_theta_sketch::deserialize(s);
|
222
|
+
compact_theta_sketch deserialized_sketch2 = compact_theta_sketch::deserialize(bytes.data(), bytes.size());
|
223
|
+
REQUIRE(bytes.size() == static_cast<size_t>(s.tellg()));
|
224
|
+
REQUIRE(deserialized_sketch2.is_empty() == deserialized_sketch1.is_empty());
|
225
|
+
REQUIRE(deserialized_sketch2.is_ordered() == deserialized_sketch1.is_ordered());
|
226
|
+
REQUIRE(deserialized_sketch2.get_num_retained() == deserialized_sketch1.get_num_retained());
|
227
|
+
REQUIRE(deserialized_sketch2.get_theta() == deserialized_sketch1.get_theta());
|
228
|
+
REQUIRE(deserialized_sketch2.get_estimate() == deserialized_sketch1.get_estimate());
|
229
|
+
REQUIRE(deserialized_sketch2.get_lower_bound(1) == deserialized_sketch1.get_lower_bound(1));
|
230
|
+
REQUIRE(deserialized_sketch2.get_upper_bound(1) == deserialized_sketch1.get_upper_bound(1));
|
231
|
+
// the sketches are ordered, so the iteration sequence must match exactly
|
232
|
+
auto iter = deserialized_sketch1.begin();
|
233
|
+
for (auto key: deserialized_sketch2) {
|
234
|
+
REQUIRE(*iter == key);
|
235
|
+
++iter;
|
236
|
+
}
|
237
|
+
}
|
238
|
+
|
239
|
+
TEST_CASE("theta sketch: deserialize compact single item buffer overrun", "[theta_sketch]") {
|
240
|
+
update_theta_sketch update_sketch = update_theta_sketch::builder().build();
|
241
|
+
update_sketch.update(1);
|
242
|
+
auto bytes = update_sketch.compact().serialize();
|
243
|
+
REQUIRE_THROWS_AS(compact_theta_sketch::deserialize(bytes.data(), 7), std::out_of_range);
|
244
|
+
REQUIRE_THROWS_AS(compact_theta_sketch::deserialize(bytes.data(), bytes.size() - 1), std::out_of_range);
|
245
|
+
}
|
246
|
+
|
247
|
+
} /* namespace datasketches */
|
@@ -0,0 +1,44 @@
|
|
1
|
+
/*
|
2
|
+
* Licensed to the Apache Software Foundation (ASF) under one
|
3
|
+
* or more contributor license agreements. See the NOTICE file
|
4
|
+
* distributed with this work for additional information
|
5
|
+
* regarding copyright ownership. The ASF licenses this file
|
6
|
+
* to you under the Apache License, Version 2.0 (the
|
7
|
+
* "License"); you may not use this file except in compliance
|
8
|
+
* with the License. You may obtain a copy of the License at
|
9
|
+
*
|
10
|
+
* http://www.apache.org/licenses/LICENSE-2.0
|
11
|
+
*
|
12
|
+
* Unless required by applicable law or agreed to in writing,
|
13
|
+
* software distributed under the License is distributed on an
|
14
|
+
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
15
|
+
* KIND, either express or implied. See the License for the
|
16
|
+
* specific language governing permissions and limitations
|
17
|
+
* under the License.
|
18
|
+
*/
|
19
|
+
|
20
|
+
#include <iostream>
|
21
|
+
|
22
|
+
#include <catch.hpp>
|
23
|
+
#include <tuple_union.hpp>
|
24
|
+
|
25
|
+
#include <theta_union_experimental.hpp>
|
26
|
+
|
27
|
+
namespace datasketches {
|
28
|
+
|
29
|
+
TEST_CASE("theta_union_exeperimental") {
|
30
|
+
auto update_sketch1 = update_theta_sketch_experimental<>::builder().build();
|
31
|
+
update_sketch1.update(1);
|
32
|
+
update_sketch1.update(2);
|
33
|
+
|
34
|
+
auto update_sketch2 = update_theta_sketch_experimental<>::builder().build();
|
35
|
+
update_sketch2.update(1);
|
36
|
+
update_sketch2.update(3);
|
37
|
+
|
38
|
+
auto u = theta_union_experimental<>::builder().build();
|
39
|
+
u.update(update_sketch1);
|
40
|
+
u.update(update_sketch2);
|
41
|
+
auto r = u.get_result();
|
42
|
+
}
|
43
|
+
|
44
|
+
} /* namespace datasketches */
|
@@ -0,0 +1,289 @@
|
|
1
|
+
/*
|
2
|
+
* Licensed to the Apache Software Foundation (ASF) under one
|
3
|
+
* or more contributor license agreements. See the NOTICE file
|
4
|
+
* distributed with this work for additional information
|
5
|
+
* regarding copyright ownership. The ASF licenses this file
|
6
|
+
* to you under the Apache License, Version 2.0 (the
|
7
|
+
* "License"); you may not use this file except in compliance
|
8
|
+
* with the License. You may obtain a copy of the License at
|
9
|
+
*
|
10
|
+
* http://www.apache.org/licenses/LICENSE-2.0
|
11
|
+
*
|
12
|
+
* Unless required by applicable law or agreed to in writing,
|
13
|
+
* software distributed under the License is distributed on an
|
14
|
+
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
15
|
+
* KIND, either express or implied. See the License for the
|
16
|
+
* specific language governing permissions and limitations
|
17
|
+
* under the License.
|
18
|
+
*/
|
19
|
+
|
20
|
+
#include <iostream>
|
21
|
+
|
22
|
+
#include <catch.hpp>
|
23
|
+
#include <tuple_a_not_b.hpp>
|
24
|
+
#include <theta_sketch_experimental.hpp>
|
25
|
+
|
26
|
+
namespace datasketches {
|
27
|
+
|
28
|
+
TEST_CASE("tuple a-not-b: empty", "[tuple_a_not_b]") {
|
29
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
30
|
+
auto b = update_tuple_sketch<float>::builder().build();
|
31
|
+
tuple_a_not_b<float> a_not_b;
|
32
|
+
auto result = a_not_b.compute(a, b);
|
33
|
+
REQUIRE(result.get_num_retained() == 0);
|
34
|
+
REQUIRE(result.is_empty());
|
35
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
36
|
+
REQUIRE(result.get_estimate() == 0.0);
|
37
|
+
}
|
38
|
+
|
39
|
+
TEST_CASE("tuple a-not-b: non empty no retained keys", "[tuple_a_not_b]") {
|
40
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
41
|
+
a.update(1, 1);
|
42
|
+
auto b = update_tuple_sketch<float>::builder().set_p(0.001).build();
|
43
|
+
tuple_a_not_b<float> a_not_b;
|
44
|
+
|
45
|
+
// B is still empty
|
46
|
+
auto result = a_not_b.compute(a, b);
|
47
|
+
REQUIRE_FALSE(result.is_empty());
|
48
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
49
|
+
REQUIRE(result.get_num_retained() == 1);
|
50
|
+
REQUIRE(result.get_theta() == Approx(1).margin(1e-10));
|
51
|
+
REQUIRE(result.get_estimate() == 1.0);
|
52
|
+
|
53
|
+
// B is not empty in estimation mode and no entries
|
54
|
+
b.update(1, 1);
|
55
|
+
REQUIRE(b.get_num_retained() == 0);
|
56
|
+
|
57
|
+
result = a_not_b.compute(a, b);
|
58
|
+
REQUIRE_FALSE(result.is_empty());
|
59
|
+
REQUIRE(result.is_estimation_mode());
|
60
|
+
REQUIRE(result.get_num_retained() == 0);
|
61
|
+
REQUIRE(result.get_theta() == Approx(0.001).margin(1e-10));
|
62
|
+
REQUIRE(result.get_estimate() == 0.0);
|
63
|
+
}
|
64
|
+
|
65
|
+
TEST_CASE("tuple a-not-b: exact mode half overlap", "[tuple_a_not_b]") {
|
66
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
67
|
+
int value = 0;
|
68
|
+
for (int i = 0; i < 1000; i++) a.update(value++, 1);
|
69
|
+
|
70
|
+
auto b = update_tuple_sketch<float>::builder().build();
|
71
|
+
value = 500;
|
72
|
+
for (int i = 0; i < 1000; i++) b.update(value++, 1);
|
73
|
+
|
74
|
+
tuple_a_not_b<float> a_not_b;
|
75
|
+
|
76
|
+
// unordered inputs, ordered result
|
77
|
+
auto result = a_not_b.compute(a, b);
|
78
|
+
REQUIRE_FALSE(result.is_empty());
|
79
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
80
|
+
REQUIRE(result.is_ordered());
|
81
|
+
REQUIRE(result.get_estimate() == 500.0);
|
82
|
+
|
83
|
+
// unordered inputs, unordered result
|
84
|
+
result = a_not_b.compute(a, b, false);
|
85
|
+
REQUIRE_FALSE(result.is_empty());
|
86
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
87
|
+
REQUIRE_FALSE(result.is_ordered());
|
88
|
+
REQUIRE(result.get_estimate() == 500.0);
|
89
|
+
|
90
|
+
// ordered inputs
|
91
|
+
result = a_not_b.compute(a.compact(), b.compact());
|
92
|
+
REQUIRE_FALSE(result.is_empty());
|
93
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
94
|
+
REQUIRE(result.is_ordered());
|
95
|
+
REQUIRE(result.get_estimate() == 500.0);
|
96
|
+
|
97
|
+
// A is ordered, so the result is ordered regardless
|
98
|
+
result = a_not_b.compute(a.compact(), b, false);
|
99
|
+
REQUIRE_FALSE(result.is_empty());
|
100
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
101
|
+
REQUIRE(result.is_ordered());
|
102
|
+
REQUIRE(result.get_estimate() == 500.0);
|
103
|
+
}
|
104
|
+
|
105
|
+
// needed until promotion of experimental to replace existing theta sketch
|
106
|
+
using update_theta_sketch = update_theta_sketch_experimental<>;
|
107
|
+
|
108
|
+
TEST_CASE("mixed a-not-b: exact mode half overlap", "[tuple_a_not_b]") {
|
109
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
110
|
+
int value = 0;
|
111
|
+
for (int i = 0; i < 1000; i++) a.update(value++, 1);
|
112
|
+
|
113
|
+
auto b = update_theta_sketch::builder().build();
|
114
|
+
value = 500;
|
115
|
+
for (int i = 0; i < 1000; i++) b.update(value++);
|
116
|
+
|
117
|
+
tuple_a_not_b<float> a_not_b;
|
118
|
+
|
119
|
+
// unordered inputs, ordered result
|
120
|
+
auto result = a_not_b.compute(a, compact_tuple_sketch<float>(b, 1, false));
|
121
|
+
REQUIRE_FALSE(result.is_empty());
|
122
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
123
|
+
REQUIRE(result.is_ordered());
|
124
|
+
REQUIRE(result.get_estimate() == 500.0);
|
125
|
+
|
126
|
+
// unordered inputs, unordered result
|
127
|
+
result = a_not_b.compute(a, compact_tuple_sketch<float>(b, 1, false), false);
|
128
|
+
REQUIRE_FALSE(result.is_empty());
|
129
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
130
|
+
REQUIRE_FALSE(result.is_ordered());
|
131
|
+
REQUIRE(result.get_estimate() == 500.0);
|
132
|
+
|
133
|
+
// ordered inputs
|
134
|
+
result = a_not_b.compute(a.compact(), compact_tuple_sketch<float>(b.compact(), 1));
|
135
|
+
REQUIRE_FALSE(result.is_empty());
|
136
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
137
|
+
REQUIRE(result.is_ordered());
|
138
|
+
REQUIRE(result.get_estimate() == 500.0);
|
139
|
+
|
140
|
+
// A is ordered, so the result is ordered regardless
|
141
|
+
result = a_not_b.compute(a.compact(), compact_tuple_sketch<float>(b, 1, false), false);
|
142
|
+
REQUIRE_FALSE(result.is_empty());
|
143
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
144
|
+
REQUIRE(result.is_ordered());
|
145
|
+
REQUIRE(result.get_estimate() == 500.0);
|
146
|
+
}
|
147
|
+
|
148
|
+
TEST_CASE("tuple a-not-b: exact mode disjoint", "[tuple_a_not_b]") {
|
149
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
150
|
+
int value = 0;
|
151
|
+
for (int i = 0; i < 1000; i++) a.update(value++, 1);
|
152
|
+
|
153
|
+
auto b = update_tuple_sketch<float>::builder().build();
|
154
|
+
for (int i = 0; i < 1000; i++) b.update(value++, 1);
|
155
|
+
|
156
|
+
tuple_a_not_b<float> a_not_b;
|
157
|
+
|
158
|
+
// unordered inputs
|
159
|
+
auto result = a_not_b.compute(a, b);
|
160
|
+
REQUIRE_FALSE(result.is_empty());
|
161
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
162
|
+
REQUIRE(result.get_estimate() == 1000.0);
|
163
|
+
|
164
|
+
// ordered inputs
|
165
|
+
result = a_not_b.compute(a.compact(), b.compact());
|
166
|
+
REQUIRE_FALSE(result.is_empty());
|
167
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
168
|
+
REQUIRE(result.get_estimate() == 1000.0);
|
169
|
+
}
|
170
|
+
|
171
|
+
TEST_CASE("tuple a-not-b: exact mode full overlap", "[tuple_a_not_b]") {
|
172
|
+
auto sketch = update_tuple_sketch<float>::builder().build();
|
173
|
+
int value = 0;
|
174
|
+
for (int i = 0; i < 1000; i++) sketch.update(value++, 1);
|
175
|
+
|
176
|
+
tuple_a_not_b<float> a_not_b;
|
177
|
+
|
178
|
+
// unordered inputs
|
179
|
+
auto result = a_not_b.compute(sketch, sketch);
|
180
|
+
REQUIRE(result.is_empty());
|
181
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
182
|
+
REQUIRE(result.get_estimate() == 0.0);
|
183
|
+
|
184
|
+
// ordered inputs
|
185
|
+
result = a_not_b.compute(sketch.compact(), sketch.compact());
|
186
|
+
REQUIRE(result.is_empty());
|
187
|
+
REQUIRE_FALSE(result.is_estimation_mode());
|
188
|
+
REQUIRE(result.get_estimate() == 0.0);
|
189
|
+
}
|
190
|
+
|
191
|
+
TEST_CASE("tuple a-not-b: estimation mode half overlap", "[tuple_a_not_b]") {
|
192
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
193
|
+
int value = 0;
|
194
|
+
for (int i = 0; i < 10000; i++) a.update(value++, 1);
|
195
|
+
|
196
|
+
auto b = update_tuple_sketch<float>::builder().build();
|
197
|
+
value = 5000;
|
198
|
+
for (int i = 0; i < 10000; i++) b.update(value++, 1);
|
199
|
+
|
200
|
+
tuple_a_not_b<float> a_not_b;
|
201
|
+
|
202
|
+
// unordered inputs
|
203
|
+
auto result = a_not_b.compute(a, b);
|
204
|
+
REQUIRE_FALSE(result.is_empty());
|
205
|
+
REQUIRE(result.is_estimation_mode());
|
206
|
+
REQUIRE(result.get_estimate() == Approx(5000).margin(5000 * 0.02));
|
207
|
+
|
208
|
+
// ordered inputs
|
209
|
+
result = a_not_b.compute(a.compact(), b.compact());
|
210
|
+
REQUIRE_FALSE(result.is_empty());
|
211
|
+
REQUIRE(result.is_estimation_mode());
|
212
|
+
REQUIRE(result.get_estimate() == Approx(5000).margin(5000 * 0.02));
|
213
|
+
}
|
214
|
+
|
215
|
+
TEST_CASE("tuple a-not-b: estimation mode disjoint", "[tuple_a_not_b]") {
|
216
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
217
|
+
int value = 0;
|
218
|
+
for (int i = 0; i < 10000; i++) a.update(value++, 1);
|
219
|
+
|
220
|
+
auto b = update_tuple_sketch<float>::builder().build();
|
221
|
+
for (int i = 0; i < 10000; i++) b.update(value++, 1);
|
222
|
+
|
223
|
+
tuple_a_not_b<float> a_not_b;
|
224
|
+
|
225
|
+
// unordered inputs
|
226
|
+
auto result = a_not_b.compute(a, b);
|
227
|
+
REQUIRE_FALSE(result.is_empty());
|
228
|
+
REQUIRE(result.is_estimation_mode());
|
229
|
+
REQUIRE(result.get_estimate() == Approx(10000).margin(10000 * 0.02));
|
230
|
+
|
231
|
+
// ordered inputs
|
232
|
+
result = a_not_b.compute(a.compact(), b.compact());
|
233
|
+
REQUIRE_FALSE(result.is_empty());
|
234
|
+
REQUIRE(result.is_estimation_mode());
|
235
|
+
REQUIRE(result.get_estimate() == Approx(10000).margin(10000 * 0.02));
|
236
|
+
}
|
237
|
+
|
238
|
+
TEST_CASE("tuple a-not-b: estimation mode full overlap", "[tuple_a_not_b]") {
|
239
|
+
auto sketch = update_tuple_sketch<float>::builder().build();
|
240
|
+
int value = 0;
|
241
|
+
for (int i = 0; i < 10000; i++) sketch.update(value++, 1);
|
242
|
+
|
243
|
+
tuple_a_not_b<float> a_not_b;
|
244
|
+
|
245
|
+
// unordered inputs
|
246
|
+
auto result = a_not_b.compute(sketch, sketch);
|
247
|
+
REQUIRE_FALSE(result.is_empty());
|
248
|
+
REQUIRE(result.is_estimation_mode());
|
249
|
+
REQUIRE(result.get_estimate() == 0.0);
|
250
|
+
|
251
|
+
// ordered inputs
|
252
|
+
result = a_not_b.compute(sketch.compact(), sketch.compact());
|
253
|
+
REQUIRE_FALSE(result.is_empty());
|
254
|
+
REQUIRE(result.is_estimation_mode());
|
255
|
+
REQUIRE(result.get_estimate() == 0.0);
|
256
|
+
}
|
257
|
+
|
258
|
+
TEST_CASE("tuple a-not-b: seed mismatch", "[tuple_a_not_b]") {
|
259
|
+
auto sketch = update_tuple_sketch<float>::builder().build();
|
260
|
+
sketch.update(1, 1); // non-empty should not be ignored
|
261
|
+
tuple_a_not_b<float> a_not_b(123);
|
262
|
+
REQUIRE_THROWS_AS(a_not_b.compute(sketch, sketch), std::invalid_argument);
|
263
|
+
}
|
264
|
+
|
265
|
+
TEST_CASE("tuple a-not-b: issue #152", "[tuple_a_not_b]") {
|
266
|
+
auto a = update_tuple_sketch<float>::builder().build();
|
267
|
+
int value = 0;
|
268
|
+
for (int i = 0; i < 10000; i++) a.update(value++, 1);
|
269
|
+
|
270
|
+
auto b = update_tuple_sketch<float>::builder().build();
|
271
|
+
value = 5000;
|
272
|
+
for (int i = 0; i < 25000; i++) b.update(value++, 1);
|
273
|
+
|
274
|
+
tuple_a_not_b<float> a_not_b;
|
275
|
+
|
276
|
+
// unordered inputs
|
277
|
+
auto result = a_not_b.compute(a, b);
|
278
|
+
REQUIRE_FALSE(result.is_empty());
|
279
|
+
REQUIRE(result.is_estimation_mode());
|
280
|
+
REQUIRE(result.get_estimate() == Approx(5000).margin(5000 * 0.03));
|
281
|
+
|
282
|
+
// ordered inputs
|
283
|
+
result = a_not_b.compute(a.compact(), b.compact());
|
284
|
+
REQUIRE_FALSE(result.is_empty());
|
285
|
+
REQUIRE(result.is_estimation_mode());
|
286
|
+
REQUIRE(result.get_estimate() == Approx(5000).margin(5000 * 0.03));
|
287
|
+
}
|
288
|
+
|
289
|
+
} /* namespace datasketches */
|