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 +1,44 @@
|
|
1
|
+
# Licensed to the Apache Software Foundation (ASF) under one
|
2
|
+
# or more contributor license agreements. See the NOTICE file
|
3
|
+
# distributed with this work for additional information
|
4
|
+
# regarding copyright ownership. The ASF licenses this file
|
5
|
+
# to you under the Apache License, Version 2.0 (the
|
6
|
+
# "License"); you may not use this file except in compliance
|
7
|
+
# with the License. You may obtain a copy of the License at
|
8
|
+
#
|
9
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
10
|
+
#
|
11
|
+
# Unless required by applicable law or agreed to in writing,
|
12
|
+
# software distributed under the License is distributed on an
|
13
|
+
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
14
|
+
# KIND, either express or implied. See the License for the
|
15
|
+
# specific language governing permissions and limitations
|
16
|
+
# under the License.
|
17
|
+
|
18
|
+
add_executable(fi_test)
|
19
|
+
|
20
|
+
target_link_libraries(fi_test fi common_test)
|
21
|
+
|
22
|
+
set_target_properties(fi_test PROPERTIES
|
23
|
+
CXX_STANDARD 11
|
24
|
+
CXX_STANDARD_REQUIRED YES
|
25
|
+
)
|
26
|
+
|
27
|
+
file(TO_CMAKE_PATH "${CMAKE_CURRENT_SOURCE_DIR}" FI_TEST_BINARY_PATH)
|
28
|
+
string(APPEND FI_TEST_BINARY_PATH "/")
|
29
|
+
target_compile_definitions(fi_test
|
30
|
+
PRIVATE
|
31
|
+
TEST_BINARY_INPUT_PATH="${FI_TEST_BINARY_PATH}"
|
32
|
+
)
|
33
|
+
|
34
|
+
add_test(
|
35
|
+
NAME fi_test
|
36
|
+
COMMAND fi_test
|
37
|
+
)
|
38
|
+
|
39
|
+
target_sources(fi_test
|
40
|
+
PRIVATE
|
41
|
+
reverse_purge_hash_map_test.cpp
|
42
|
+
frequent_items_sketch_test.cpp
|
43
|
+
frequent_items_sketch_custom_type_test.cpp
|
44
|
+
)
|
@@ -0,0 +1,84 @@
|
|
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 <catch.hpp>
|
21
|
+
#include <sstream>
|
22
|
+
|
23
|
+
#include "frequent_items_sketch.hpp"
|
24
|
+
#include "test_type.hpp"
|
25
|
+
|
26
|
+
namespace datasketches {
|
27
|
+
|
28
|
+
typedef frequent_items_sketch<test_type, float, test_type_hash, test_type_equal, test_type_serde> frequent_test_type_sketch;
|
29
|
+
|
30
|
+
TEST_CASE("frequent items: custom type", "[frequent_items_sketch]") {
|
31
|
+
frequent_test_type_sketch sketch(3);
|
32
|
+
sketch.update(1, 10); // should survive the purge
|
33
|
+
sketch.update(2);
|
34
|
+
sketch.update(3);
|
35
|
+
sketch.update(4);
|
36
|
+
sketch.update(5);
|
37
|
+
sketch.update(6);
|
38
|
+
sketch.update(7);
|
39
|
+
test_type a8(8);
|
40
|
+
sketch.update(a8);
|
41
|
+
REQUIRE_FALSE(sketch.is_empty());
|
42
|
+
REQUIRE(sketch.get_total_weight() == 17);
|
43
|
+
REQUIRE(sketch.get_estimate(1) == 10);
|
44
|
+
//std::cerr << "num active: " << sketch.get_num_active_items() << std::endl;
|
45
|
+
|
46
|
+
//std::cerr << "get frequent items" << std::endl;
|
47
|
+
auto items = sketch.get_frequent_items(frequent_items_error_type::NO_FALSE_POSITIVES);
|
48
|
+
REQUIRE(items.size() == 1); // only 1 item should be above threshold
|
49
|
+
REQUIRE(items[0].get_item().get_value() == 1);
|
50
|
+
REQUIRE(items[0].get_estimate() == 10);
|
51
|
+
|
52
|
+
std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
|
53
|
+
//std::cerr << "serialize" << std::endl;
|
54
|
+
sketch.serialize(s);
|
55
|
+
//std::cerr << "deserialize" << std::endl;
|
56
|
+
auto sketch2 = frequent_test_type_sketch::deserialize(s);
|
57
|
+
REQUIRE_FALSE(sketch2.is_empty());
|
58
|
+
REQUIRE(sketch2.get_total_weight() == 17);
|
59
|
+
REQUIRE(sketch2.get_estimate(1) == 10);
|
60
|
+
REQUIRE(sketch.get_num_active_items() == sketch2.get_num_active_items());
|
61
|
+
REQUIRE(sketch.get_maximum_error() == sketch2.get_maximum_error());
|
62
|
+
//std::cerr << "end" << std::endl;
|
63
|
+
|
64
|
+
std::cout << sketch2.to_string(true);
|
65
|
+
}
|
66
|
+
|
67
|
+
// this is to see the debug print from test_type if enabled there to make sure items are moved
|
68
|
+
TEST_CASE("frequent items: moving merge", "[frequent_items_sketch]") {
|
69
|
+
frequent_test_type_sketch sketch1(3);
|
70
|
+
sketch1.update(1);
|
71
|
+
|
72
|
+
frequent_test_type_sketch sketch2(3);
|
73
|
+
sketch2.update(2);
|
74
|
+
|
75
|
+
sketch2.merge(std::move(sketch1));
|
76
|
+
REQUIRE(sketch2.get_total_weight() == 2);
|
77
|
+
}
|
78
|
+
|
79
|
+
TEST_CASE("frequent items: negative weight", "[frequent_items_sketch]") {
|
80
|
+
frequent_test_type_sketch sketch(3);
|
81
|
+
REQUIRE_THROWS_AS(sketch.update(1, -1), std::invalid_argument);
|
82
|
+
}
|
83
|
+
|
84
|
+
} /* namespace datasketches */
|
@@ -0,0 +1,360 @@
|
|
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 <catch.hpp>
|
21
|
+
#include <sstream>
|
22
|
+
#include <fstream>
|
23
|
+
|
24
|
+
#include "frequent_items_sketch.hpp"
|
25
|
+
|
26
|
+
#ifdef TEST_BINARY_INPUT_PATH
|
27
|
+
static std::string testBinaryInputPath = TEST_BINARY_INPUT_PATH;
|
28
|
+
#else
|
29
|
+
static std::string testBinaryInputPath = "test/";
|
30
|
+
#endif
|
31
|
+
|
32
|
+
namespace datasketches {
|
33
|
+
|
34
|
+
TEST_CASE("frequent items: invalid k", "[frequent_items_sketch]") {
|
35
|
+
REQUIRE_THROWS_AS(frequent_items_sketch<int>(2), std::invalid_argument);
|
36
|
+
}
|
37
|
+
|
38
|
+
TEST_CASE("frequent items: empty", "[frequent_items_sketch]") {
|
39
|
+
frequent_items_sketch<int> sketch(3);
|
40
|
+
REQUIRE(sketch.is_empty());
|
41
|
+
REQUIRE(sketch.get_num_active_items() == 0);
|
42
|
+
REQUIRE(sketch.get_total_weight() == 0);
|
43
|
+
}
|
44
|
+
|
45
|
+
TEST_CASE("frequent items: one item", "[frequent_items_sketch]") {
|
46
|
+
frequent_items_sketch<std::string> sketch(3);
|
47
|
+
sketch.update("a");
|
48
|
+
REQUIRE_FALSE(sketch.is_empty());
|
49
|
+
REQUIRE(sketch.get_num_active_items() == 1);
|
50
|
+
REQUIRE(sketch.get_total_weight() == 1);
|
51
|
+
REQUIRE(sketch.get_estimate("a") == 1);
|
52
|
+
REQUIRE(sketch.get_lower_bound("a") == 1);
|
53
|
+
REQUIRE(sketch.get_upper_bound("a") == 1);
|
54
|
+
}
|
55
|
+
|
56
|
+
TEST_CASE("frequent items: several items, no resize, no purge", "[frequent_items_sketch]") {
|
57
|
+
frequent_items_sketch<std::string> sketch(3);
|
58
|
+
sketch.update("a");
|
59
|
+
sketch.update("b");
|
60
|
+
sketch.update("c");
|
61
|
+
sketch.update("d");
|
62
|
+
sketch.update("b");
|
63
|
+
sketch.update("c");
|
64
|
+
sketch.update("b");
|
65
|
+
REQUIRE_FALSE(sketch.is_empty());
|
66
|
+
REQUIRE(sketch.get_total_weight() == 7);
|
67
|
+
REQUIRE(sketch.get_num_active_items() == 4);
|
68
|
+
REQUIRE(sketch.get_estimate("a") == 1);
|
69
|
+
REQUIRE(sketch.get_estimate("b") == 3);
|
70
|
+
REQUIRE(sketch.get_estimate("c") == 2);
|
71
|
+
REQUIRE(sketch.get_estimate("d") == 1);
|
72
|
+
}
|
73
|
+
|
74
|
+
TEST_CASE("frequent items: several items, with resize, no purge", "[frequent_items_sketch]") {
|
75
|
+
frequent_items_sketch<std::string> sketch(4);
|
76
|
+
sketch.update("a");
|
77
|
+
sketch.update("b");
|
78
|
+
sketch.update("c");
|
79
|
+
sketch.update("d");
|
80
|
+
sketch.update("b");
|
81
|
+
sketch.update("c");
|
82
|
+
sketch.update("b");
|
83
|
+
sketch.update("e");
|
84
|
+
sketch.update("f");
|
85
|
+
sketch.update("g");
|
86
|
+
sketch.update("h");
|
87
|
+
sketch.update("i");
|
88
|
+
sketch.update("j");
|
89
|
+
sketch.update("k");
|
90
|
+
sketch.update("l");
|
91
|
+
REQUIRE_FALSE(sketch.is_empty());
|
92
|
+
REQUIRE(sketch.get_total_weight() == 15);
|
93
|
+
REQUIRE(sketch.get_num_active_items() == 12);
|
94
|
+
REQUIRE(sketch.get_estimate("a") == 1);
|
95
|
+
REQUIRE(sketch.get_estimate("b") == 3);
|
96
|
+
REQUIRE(sketch.get_estimate("c") == 2);
|
97
|
+
REQUIRE(sketch.get_estimate("d") == 1);
|
98
|
+
}
|
99
|
+
|
100
|
+
TEST_CASE("frequent items: estimation mode", "[frequent_items_sketch]") {
|
101
|
+
frequent_items_sketch<int> sketch(3);
|
102
|
+
sketch.update(1, 10);
|
103
|
+
sketch.update(2);
|
104
|
+
sketch.update(3);
|
105
|
+
sketch.update(4);
|
106
|
+
sketch.update(5);
|
107
|
+
sketch.update(6);
|
108
|
+
sketch.update(7, 15);
|
109
|
+
sketch.update(8);
|
110
|
+
sketch.update(9);
|
111
|
+
sketch.update(10);
|
112
|
+
sketch.update(11);
|
113
|
+
sketch.update(12);
|
114
|
+
REQUIRE(sketch.get_maximum_error() > 0); // estimation mode
|
115
|
+
|
116
|
+
REQUIRE_FALSE(sketch.is_empty());
|
117
|
+
REQUIRE(sketch.get_total_weight() == 35);
|
118
|
+
|
119
|
+
auto items = sketch.get_frequent_items(frequent_items_error_type::NO_FALSE_POSITIVES);
|
120
|
+
REQUIRE(items.size() == 2); // only 2 items (1 and 7) should have counts more than 1
|
121
|
+
REQUIRE(items[0].get_item() == 7);
|
122
|
+
REQUIRE(items[0].get_estimate() == 15);
|
123
|
+
REQUIRE(items[1].get_item() == 1);
|
124
|
+
REQUIRE(items[1].get_estimate() == 10);
|
125
|
+
|
126
|
+
items = sketch.get_frequent_items(frequent_items_error_type::NO_FALSE_NEGATIVES);
|
127
|
+
REQUIRE(2 <= items.size()); // at least 2 items
|
128
|
+
REQUIRE(12 >= items.size()); // but not more than 12 items
|
129
|
+
}
|
130
|
+
|
131
|
+
TEST_CASE("frequent items: merge exact mode", "[frequent_items_sketch]") {
|
132
|
+
frequent_items_sketch<int> sketch1(3);
|
133
|
+
sketch1.update(1);
|
134
|
+
sketch1.update(2);
|
135
|
+
sketch1.update(3);
|
136
|
+
sketch1.update(4);
|
137
|
+
|
138
|
+
frequent_items_sketch<int> sketch2(3);
|
139
|
+
sketch1.update(2);
|
140
|
+
sketch1.update(3);
|
141
|
+
sketch1.update(2);
|
142
|
+
|
143
|
+
sketch1.merge(sketch2);
|
144
|
+
REQUIRE_FALSE(sketch1.is_empty());
|
145
|
+
REQUIRE(sketch1.get_total_weight() == 7);
|
146
|
+
REQUIRE(sketch1.get_num_active_items() == 4);
|
147
|
+
REQUIRE(sketch1.get_estimate(1) == 1);
|
148
|
+
REQUIRE(sketch1.get_estimate(2) == 3);
|
149
|
+
REQUIRE(sketch1.get_estimate(3) == 2);
|
150
|
+
REQUIRE(sketch1.get_estimate(4) == 1);
|
151
|
+
}
|
152
|
+
|
153
|
+
TEST_CASE("frequent items: merge estimation mode", "[frequent_items_sketch]") {
|
154
|
+
frequent_items_sketch<int> sketch1(4);
|
155
|
+
sketch1.update(1, 9); // to make sure it survives the purge
|
156
|
+
sketch1.update(2);
|
157
|
+
sketch1.update(3);
|
158
|
+
sketch1.update(4);
|
159
|
+
sketch1.update(5);
|
160
|
+
sketch1.update(6);
|
161
|
+
sketch1.update(7);
|
162
|
+
sketch1.update(8);
|
163
|
+
sketch1.update(9);
|
164
|
+
sketch1.update(10);
|
165
|
+
sketch1.update(11);
|
166
|
+
sketch1.update(12);
|
167
|
+
sketch1.update(13);
|
168
|
+
sketch1.update(14);
|
169
|
+
REQUIRE(sketch1.get_maximum_error() > 0); // estimation mode
|
170
|
+
|
171
|
+
frequent_items_sketch<int> sketch2(4);
|
172
|
+
sketch2.update(8);
|
173
|
+
sketch2.update(9);
|
174
|
+
sketch2.update(10);
|
175
|
+
sketch2.update(11);
|
176
|
+
sketch2.update(12);
|
177
|
+
sketch2.update(13);
|
178
|
+
sketch2.update(14);
|
179
|
+
sketch2.update(15);
|
180
|
+
sketch2.update(16);
|
181
|
+
sketch2.update(17);
|
182
|
+
sketch2.update(18);
|
183
|
+
sketch2.update(19);
|
184
|
+
sketch2.update(20);
|
185
|
+
sketch2.update(21, 11); // to make sure it survives the purge
|
186
|
+
REQUIRE(sketch2.get_maximum_error() > 0); // estimation mode
|
187
|
+
|
188
|
+
sketch1.merge(sketch2);
|
189
|
+
REQUIRE_FALSE(sketch1.is_empty());
|
190
|
+
REQUIRE(sketch1.get_total_weight() == 46);
|
191
|
+
REQUIRE(2 <= sketch1.get_num_active_items());
|
192
|
+
|
193
|
+
auto items = sketch1.get_frequent_items(frequent_items_error_type::NO_FALSE_POSITIVES, 2);
|
194
|
+
REQUIRE(items.size() == 2); // only 2 items (1 and 21) should be above threshold
|
195
|
+
REQUIRE(items[0].get_item() == 21);
|
196
|
+
REQUIRE(11 <= items[0].get_estimate()); // always overestimated
|
197
|
+
REQUIRE(items[1].get_item() == 1);
|
198
|
+
REQUIRE(9 <= items[1].get_estimate()); // always overestimated
|
199
|
+
}
|
200
|
+
|
201
|
+
TEST_CASE("frequent items: deserialize from java long", "[frequent_items_sketch]") {
|
202
|
+
std::ifstream is;
|
203
|
+
is.exceptions(std::ios::failbit | std::ios::badbit);
|
204
|
+
is.open(testBinaryInputPath + "longs_sketch_from_java.sk", std::ios::binary);
|
205
|
+
auto sketch = frequent_items_sketch<long long>::deserialize(is);
|
206
|
+
REQUIRE_FALSE(sketch.is_empty());
|
207
|
+
REQUIRE(sketch.get_total_weight() == 4);
|
208
|
+
REQUIRE(sketch.get_num_active_items() == 4);
|
209
|
+
REQUIRE(sketch.get_estimate(1) == 1);
|
210
|
+
REQUIRE(sketch.get_estimate(2) == 1);
|
211
|
+
REQUIRE(sketch.get_estimate(3) == 1);
|
212
|
+
REQUIRE(sketch.get_estimate(4) == 1);
|
213
|
+
}
|
214
|
+
|
215
|
+
TEST_CASE("frequent items: deserialize from java string", "[frequent_items_sketch]") {
|
216
|
+
std::ifstream is;
|
217
|
+
is.exceptions(std::ios::failbit | std::ios::badbit);
|
218
|
+
is.open(testBinaryInputPath + "items_sketch_string_from_java.sk", std::ios::binary);
|
219
|
+
auto sketch = frequent_items_sketch<std::string>::deserialize(is);
|
220
|
+
REQUIRE_FALSE(sketch.is_empty());
|
221
|
+
REQUIRE(sketch.get_total_weight() == 4);
|
222
|
+
REQUIRE(sketch.get_num_active_items() == 4);
|
223
|
+
REQUIRE(sketch.get_estimate("aaaaaaaaaaaaaaaaaaaaaaaaaaaaa") == 1);
|
224
|
+
REQUIRE(sketch.get_estimate("bbbbbbbbbbbbbbbbbbbbbbbbbbbbb") == 1);
|
225
|
+
REQUIRE(sketch.get_estimate("ccccccccccccccccccccccccccccc") == 1);
|
226
|
+
REQUIRE(sketch.get_estimate("ddddddddddddddddddddddddddddd") == 1);
|
227
|
+
}
|
228
|
+
|
229
|
+
TEST_CASE("frequent items: deserialize from java string, utf-8", "[frequent_items_sketch]") {
|
230
|
+
std::ifstream is;
|
231
|
+
is.exceptions(std::ios::failbit | std::ios::badbit);
|
232
|
+
is.open(testBinaryInputPath + "items_sketch_string_utf8_from_java.sk", std::ios::binary);
|
233
|
+
auto sketch = frequent_items_sketch<std::string>::deserialize(is);
|
234
|
+
REQUIRE_FALSE(sketch.is_empty());
|
235
|
+
REQUIRE(sketch.get_total_weight() == 10);
|
236
|
+
REQUIRE(sketch.get_num_active_items() == 4);
|
237
|
+
REQUIRE(sketch.get_estimate("абвгд") == 1);
|
238
|
+
REQUIRE(sketch.get_estimate("еёжзи") == 2);
|
239
|
+
REQUIRE(sketch.get_estimate("йклмн") == 3);
|
240
|
+
REQUIRE(sketch.get_estimate("опрст") == 4);
|
241
|
+
}
|
242
|
+
|
243
|
+
TEST_CASE("frequent items: deserialize long64 stream", "[frequent_items_sketch]") {
|
244
|
+
frequent_items_sketch<long long> sketch1(3);
|
245
|
+
sketch1.update(1, 1);
|
246
|
+
sketch1.update(2, 2);
|
247
|
+
sketch1.update(3, 3);
|
248
|
+
sketch1.update(4, 4);
|
249
|
+
sketch1.update(5, 5);
|
250
|
+
|
251
|
+
std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
|
252
|
+
sketch1.serialize(s);
|
253
|
+
auto sketch2 = frequent_items_sketch<long long>::deserialize(s);
|
254
|
+
REQUIRE_FALSE(sketch2.is_empty());
|
255
|
+
REQUIRE(sketch2.get_total_weight() == 15);
|
256
|
+
REQUIRE(sketch2.get_num_active_items() == 5);
|
257
|
+
REQUIRE(sketch2.get_estimate(1) == 1);
|
258
|
+
REQUIRE(sketch2.get_estimate(2) == 2);
|
259
|
+
REQUIRE(sketch2.get_estimate(3) == 3);
|
260
|
+
REQUIRE(sketch2.get_estimate(4) == 4);
|
261
|
+
REQUIRE(sketch2.get_estimate(5) == 5);
|
262
|
+
}
|
263
|
+
|
264
|
+
TEST_CASE("frequent items: serialize deserialiation long64 bytes", "[frequent_items_sketch]") {
|
265
|
+
frequent_items_sketch<long long> sketch1(3);
|
266
|
+
sketch1.update(1, 1);
|
267
|
+
sketch1.update(2, 2);
|
268
|
+
sketch1.update(3, 3);
|
269
|
+
sketch1.update(4, 4);
|
270
|
+
sketch1.update(5, 5);
|
271
|
+
|
272
|
+
auto bytes = sketch1.serialize();
|
273
|
+
auto sketch2 = frequent_items_sketch<long long>::deserialize(bytes.data(), bytes.size());
|
274
|
+
REQUIRE_FALSE(sketch2.is_empty());
|
275
|
+
REQUIRE(sketch2.get_total_weight() == 15);
|
276
|
+
REQUIRE(sketch2.get_num_active_items() == 5);
|
277
|
+
REQUIRE(sketch2.get_estimate(1) == 1);
|
278
|
+
REQUIRE(sketch2.get_estimate(2) == 2);
|
279
|
+
REQUIRE(sketch2.get_estimate(3) == 3);
|
280
|
+
REQUIRE(sketch2.get_estimate(4) == 4);
|
281
|
+
REQUIRE(sketch2.get_estimate(5) == 5);
|
282
|
+
}
|
283
|
+
|
284
|
+
TEST_CASE("frequent items: serialize deserialize string stream", "[frequent_items_sketch]") {
|
285
|
+
frequent_items_sketch<std::string> sketch1(3);
|
286
|
+
sketch1.update("aaaaaaaaaaaaaaaa", 1);
|
287
|
+
sketch1.update("bbbbbbbbbbbbbbbb", 2);
|
288
|
+
sketch1.update("cccccccccccccccc", 3);
|
289
|
+
sketch1.update("dddddddddddddddd", 4);
|
290
|
+
sketch1.update("eeeeeeeeeeeeeeee", 5);
|
291
|
+
|
292
|
+
std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
|
293
|
+
sketch1.serialize(s);
|
294
|
+
auto sketch2 = frequent_items_sketch<std::string>::deserialize(s);
|
295
|
+
REQUIRE_FALSE(sketch2.is_empty());
|
296
|
+
REQUIRE(sketch2.get_total_weight() == 15);
|
297
|
+
REQUIRE(sketch2.get_num_active_items() == 5);
|
298
|
+
REQUIRE(sketch2.get_estimate("aaaaaaaaaaaaaaaa") == 1);
|
299
|
+
REQUIRE(sketch2.get_estimate("bbbbbbbbbbbbbbbb") == 2);
|
300
|
+
REQUIRE(sketch2.get_estimate("cccccccccccccccc") == 3);
|
301
|
+
REQUIRE(sketch2.get_estimate("dddddddddddddddd") == 4);
|
302
|
+
REQUIRE(sketch2.get_estimate("eeeeeeeeeeeeeeee") == 5);
|
303
|
+
}
|
304
|
+
|
305
|
+
TEST_CASE("frequent items: serialize deserialize string bytes", "[frequent_items_sketch]") {
|
306
|
+
frequent_items_sketch<std::string> sketch1(3);
|
307
|
+
sketch1.update("aaaaaaaaaaaaaaaa", 1);
|
308
|
+
sketch1.update("bbbbbbbbbbbbbbbb", 2);
|
309
|
+
sketch1.update("cccccccccccccccc", 3);
|
310
|
+
sketch1.update("dddddddddddddddd", 4);
|
311
|
+
sketch1.update("eeeeeeeeeeeeeeee", 5);
|
312
|
+
|
313
|
+
auto bytes = sketch1.serialize();
|
314
|
+
auto sketch2 = frequent_items_sketch<std::string>::deserialize(bytes.data(), bytes.size());
|
315
|
+
REQUIRE_FALSE(sketch2.is_empty());
|
316
|
+
REQUIRE(sketch2.get_total_weight() == 15);
|
317
|
+
REQUIRE(sketch2.get_num_active_items() == 5);
|
318
|
+
REQUIRE(sketch2.get_estimate("aaaaaaaaaaaaaaaa") == 1);
|
319
|
+
REQUIRE(sketch2.get_estimate("bbbbbbbbbbbbbbbb") == 2);
|
320
|
+
REQUIRE(sketch2.get_estimate("cccccccccccccccc") == 3);
|
321
|
+
REQUIRE(sketch2.get_estimate("dddddddddddddddd") == 4);
|
322
|
+
REQUIRE(sketch2.get_estimate("eeeeeeeeeeeeeeee") == 5);
|
323
|
+
}
|
324
|
+
|
325
|
+
TEST_CASE("frequent items: serialize deserialize string, utf-8 stream", "[frequent_items_sketch]") {
|
326
|
+
frequent_items_sketch<std::string> sketch1(3);
|
327
|
+
sketch1.update("абвгд", 1);
|
328
|
+
sketch1.update("еёжзи", 2);
|
329
|
+
sketch1.update("йклмн", 3);
|
330
|
+
sketch1.update("опрст", 4);
|
331
|
+
sketch1.update("уфхцч", 5);
|
332
|
+
|
333
|
+
std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
|
334
|
+
sketch1.serialize(s);
|
335
|
+
auto sketch2 = frequent_items_sketch<std::string>::deserialize(s);
|
336
|
+
REQUIRE_FALSE(sketch2.is_empty());
|
337
|
+
REQUIRE(sketch2.get_total_weight() == 15);
|
338
|
+
REQUIRE(sketch2.get_num_active_items() == 5);
|
339
|
+
REQUIRE(sketch2.get_estimate("абвгд") == 1);
|
340
|
+
REQUIRE(sketch2.get_estimate("еёжзи") == 2);
|
341
|
+
REQUIRE(sketch2.get_estimate("йклмн") == 3);
|
342
|
+
REQUIRE(sketch2.get_estimate("опрст") == 4);
|
343
|
+
REQUIRE(sketch2.get_estimate("уфхцч") == 5);
|
344
|
+
}
|
345
|
+
|
346
|
+
TEST_CASE("frequent items: int64 deserialize single item buffer overrun", "[frequent_items_sketch]") {
|
347
|
+
frequent_items_sketch<int64_t> sketch(3);
|
348
|
+
sketch.update(1);
|
349
|
+
auto bytes = sketch.serialize();
|
350
|
+
REQUIRE_THROWS_AS(frequent_items_sketch<int64_t>::deserialize(bytes.data(), bytes.size() - 1), std::out_of_range);
|
351
|
+
}
|
352
|
+
|
353
|
+
TEST_CASE("frequent items: string deserialize single item buffer overrun", "[frequent_items_sketch]") {
|
354
|
+
frequent_items_sketch<std::string> sketch(3);
|
355
|
+
sketch.update("a");
|
356
|
+
auto bytes = sketch.serialize();
|
357
|
+
REQUIRE_THROWS_AS(frequent_items_sketch<std::string>::deserialize(bytes.data(), bytes.size() - 1), std::out_of_range);
|
358
|
+
}
|
359
|
+
|
360
|
+
} /* namespace datasketches */
|