datasketches 0.2.0 → 0.2.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +16 -0
- data/LICENSE +40 -3
- data/NOTICE +1 -1
- data/README.md +7 -7
- data/ext/datasketches/extconf.rb +1 -1
- data/ext/datasketches/theta_wrapper.cpp +20 -4
- data/lib/datasketches/version.rb +1 -1
- data/vendor/datasketches-cpp/CMakeLists.txt +31 -3
- data/vendor/datasketches-cpp/LICENSE +40 -3
- data/vendor/datasketches-cpp/MANIFEST.in +3 -0
- data/vendor/datasketches-cpp/NOTICE +1 -1
- data/vendor/datasketches-cpp/README.md +76 -9
- data/vendor/datasketches-cpp/cmake/DataSketchesConfig.cmake.in +10 -0
- data/vendor/datasketches-cpp/common/CMakeLists.txt +14 -13
- data/vendor/datasketches-cpp/common/include/MurmurHash3.h +11 -7
- data/vendor/datasketches-cpp/common/include/binomial_bounds.hpp +8 -8
- data/vendor/datasketches-cpp/common/include/bounds_binomial_proportions.hpp +12 -15
- data/vendor/datasketches-cpp/common/include/common_defs.hpp +26 -0
- data/vendor/datasketches-cpp/common/include/conditional_forward.hpp +20 -8
- data/vendor/datasketches-cpp/common/include/count_zeros.hpp +2 -2
- data/vendor/datasketches-cpp/common/include/serde.hpp +7 -7
- data/vendor/datasketches-cpp/cpc/CMakeLists.txt +15 -35
- data/vendor/datasketches-cpp/cpc/include/cpc_common.hpp +10 -3
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor.hpp +19 -19
- data/vendor/datasketches-cpp/cpc/include/cpc_compressor_impl.hpp +91 -89
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch.hpp +15 -2
- data/vendor/datasketches-cpp/cpc/include/cpc_sketch_impl.hpp +126 -90
- data/vendor/datasketches-cpp/cpc/include/cpc_union.hpp +1 -1
- data/vendor/datasketches-cpp/cpc/include/cpc_union_impl.hpp +22 -20
- data/vendor/datasketches-cpp/cpc/include/cpc_util.hpp +10 -10
- data/vendor/datasketches-cpp/cpc/include/icon_estimator.hpp +4 -4
- data/vendor/datasketches-cpp/cpc/include/u32_table.hpp +8 -8
- data/vendor/datasketches-cpp/cpc/include/u32_table_impl.hpp +14 -14
- data/vendor/datasketches-cpp/cpc/test/compression_test.cpp +10 -10
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_allocation_test.cpp +17 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_sketch_test.cpp +25 -0
- data/vendor/datasketches-cpp/cpc/test/cpc_union_test.cpp +1 -1
- data/vendor/datasketches-cpp/fi/CMakeLists.txt +5 -15
- data/vendor/datasketches-cpp/fi/include/frequent_items_sketch_impl.hpp +69 -82
- data/vendor/datasketches-cpp/fi/include/reverse_purge_hash_map_impl.hpp +10 -10
- data/vendor/datasketches-cpp/fi/test/reverse_purge_hash_map_test.cpp +2 -2
- data/vendor/datasketches-cpp/hll/CMakeLists.txt +33 -56
- data/vendor/datasketches-cpp/hll/include/AuxHashMap-internal.hpp +60 -63
- data/vendor/datasketches-cpp/hll/include/AuxHashMap.hpp +19 -19
- data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable-internal.hpp +15 -15
- data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable.hpp +3 -3
- data/vendor/datasketches-cpp/hll/include/CouponHashSet-internal.hpp +74 -76
- data/vendor/datasketches-cpp/hll/include/CouponHashSet.hpp +6 -6
- data/vendor/datasketches-cpp/hll/include/CouponList-internal.hpp +110 -113
- data/vendor/datasketches-cpp/hll/include/CouponList.hpp +13 -13
- data/vendor/datasketches-cpp/hll/include/CubicInterpolation-internal.hpp +2 -4
- data/vendor/datasketches-cpp/hll/include/HarmonicNumbers-internal.hpp +1 -1
- data/vendor/datasketches-cpp/hll/include/Hll4Array-internal.hpp +80 -76
- data/vendor/datasketches-cpp/hll/include/Hll4Array.hpp +9 -9
- data/vendor/datasketches-cpp/hll/include/Hll6Array-internal.hpp +26 -26
- data/vendor/datasketches-cpp/hll/include/Hll6Array.hpp +6 -6
- data/vendor/datasketches-cpp/hll/include/Hll8Array-internal.hpp +33 -33
- data/vendor/datasketches-cpp/hll/include/Hll8Array.hpp +6 -6
- data/vendor/datasketches-cpp/hll/include/HllArray-internal.hpp +205 -209
- data/vendor/datasketches-cpp/hll/include/HllArray.hpp +36 -36
- data/vendor/datasketches-cpp/hll/include/HllSketch-internal.hpp +34 -32
- data/vendor/datasketches-cpp/hll/include/HllSketchImpl-internal.hpp +22 -22
- data/vendor/datasketches-cpp/hll/include/HllSketchImpl.hpp +13 -13
- data/vendor/datasketches-cpp/hll/include/HllSketchImplFactory.hpp +15 -15
- data/vendor/datasketches-cpp/hll/include/HllUnion-internal.hpp +61 -61
- data/vendor/datasketches-cpp/hll/include/HllUtil.hpp +120 -127
- data/vendor/datasketches-cpp/hll/include/coupon_iterator-internal.hpp +9 -9
- data/vendor/datasketches-cpp/hll/include/coupon_iterator.hpp +5 -5
- data/vendor/datasketches-cpp/hll/include/hll.hpp +21 -21
- data/vendor/datasketches-cpp/hll/test/AuxHashMapTest.cpp +1 -1
- data/vendor/datasketches-cpp/hll/test/CouponHashSetTest.cpp +34 -34
- data/vendor/datasketches-cpp/hll/test/CouponListTest.cpp +25 -25
- data/vendor/datasketches-cpp/hll/test/CrossCountingTest.cpp +2 -2
- data/vendor/datasketches-cpp/hll/test/HllArrayTest.cpp +35 -35
- data/vendor/datasketches-cpp/hll/test/HllSketchTest.cpp +15 -15
- data/vendor/datasketches-cpp/hll/test/HllUnionTest.cpp +10 -14
- data/vendor/datasketches-cpp/hll/test/IsomorphicTest.cpp +3 -3
- data/vendor/datasketches-cpp/hll/test/ToFromByteArrayTest.cpp +4 -4
- data/vendor/datasketches-cpp/kll/CMakeLists.txt +9 -19
- data/vendor/datasketches-cpp/kll/include/kll_helper.hpp +5 -4
- data/vendor/datasketches-cpp/kll/include/kll_helper_impl.hpp +6 -6
- data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator.hpp +14 -6
- data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator_impl.hpp +39 -24
- data/vendor/datasketches-cpp/kll/include/kll_sketch.hpp +41 -4
- data/vendor/datasketches-cpp/kll/include/kll_sketch_impl.hpp +76 -64
- data/vendor/datasketches-cpp/kll/include/kolmogorov_smirnov.hpp +67 -0
- data/vendor/datasketches-cpp/kll/include/kolmogorov_smirnov_impl.hpp +78 -0
- data/vendor/datasketches-cpp/kll/test/CMakeLists.txt +1 -0
- data/vendor/datasketches-cpp/kll/test/kll_sketch_test.cpp +133 -46
- data/vendor/datasketches-cpp/kll/test/kolmogorov_smirnov_test.cpp +111 -0
- data/vendor/datasketches-cpp/pyproject.toml +4 -2
- data/vendor/datasketches-cpp/python/CMakeLists.txt +10 -6
- data/vendor/datasketches-cpp/python/README.md +50 -50
- data/vendor/datasketches-cpp/python/pybind11Path.cmd +3 -0
- data/vendor/datasketches-cpp/python/src/cpc_wrapper.cpp +1 -1
- data/vendor/datasketches-cpp/python/src/hll_wrapper.cpp +4 -4
- data/vendor/datasketches-cpp/python/src/kll_wrapper.cpp +1 -1
- data/vendor/datasketches-cpp/python/src/theta_wrapper.cpp +8 -8
- data/vendor/datasketches-cpp/python/src/vector_of_kll.cpp +11 -5
- data/vendor/datasketches-cpp/python/src/vo_wrapper.cpp +2 -2
- data/vendor/datasketches-cpp/python/tests/hll_test.py +1 -1
- data/vendor/datasketches-cpp/python/tests/kll_test.py +2 -2
- data/vendor/datasketches-cpp/python/tests/req_test.py +2 -2
- data/vendor/datasketches-cpp/python/tests/vector_of_kll_test.py +4 -4
- data/vendor/datasketches-cpp/python/tests/vo_test.py +3 -3
- data/vendor/datasketches-cpp/req/CMakeLists.txt +8 -21
- data/vendor/datasketches-cpp/req/include/req_common.hpp +2 -1
- data/vendor/datasketches-cpp/req/include/req_compactor.hpp +4 -4
- data/vendor/datasketches-cpp/req/include/req_compactor_impl.hpp +26 -39
- data/vendor/datasketches-cpp/req/include/req_sketch.hpp +1 -1
- data/vendor/datasketches-cpp/req/include/req_sketch_impl.hpp +13 -11
- data/vendor/datasketches-cpp/req/test/req_sketch_test.cpp +52 -52
- data/vendor/datasketches-cpp/sampling/CMakeLists.txt +5 -9
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch.hpp +10 -5
- data/vendor/datasketches-cpp/sampling/include/var_opt_sketch_impl.hpp +61 -64
- data/vendor/datasketches-cpp/sampling/include/var_opt_union_impl.hpp +42 -48
- data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_test.cpp +6 -6
- data/vendor/datasketches-cpp/sampling/test/var_opt_union_test.cpp +13 -13
- data/vendor/datasketches-cpp/setup.py +10 -7
- data/vendor/datasketches-cpp/theta/CMakeLists.txt +26 -45
- data/vendor/datasketches-cpp/theta/include/bounds_on_ratios_in_sampled_sets.hpp +1 -1
- data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser.hpp +67 -0
- data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser_impl.hpp +137 -0
- data/vendor/datasketches-cpp/theta/include/theta_constants.hpp +9 -4
- data/vendor/datasketches-cpp/theta/include/theta_helpers.hpp +15 -0
- data/vendor/datasketches-cpp/theta/include/theta_intersection.hpp +9 -4
- data/vendor/datasketches-cpp/theta/include/theta_intersection_base_impl.hpp +6 -6
- data/vendor/datasketches-cpp/theta/include/theta_intersection_impl.hpp +1 -1
- data/vendor/datasketches-cpp/theta/include/theta_jaccard_similarity_base.hpp +18 -14
- data/vendor/datasketches-cpp/theta/include/theta_set_difference_base_impl.hpp +2 -2
- data/vendor/datasketches-cpp/theta/include/theta_sketch.hpp +73 -15
- data/vendor/datasketches-cpp/theta/include/theta_sketch_impl.hpp +247 -103
- data/vendor/datasketches-cpp/theta/include/theta_union.hpp +10 -5
- data/vendor/datasketches-cpp/theta/include/theta_union_base.hpp +3 -1
- data/vendor/datasketches-cpp/theta/include/theta_union_base_impl.hpp +9 -3
- data/vendor/datasketches-cpp/theta/include/theta_union_impl.hpp +8 -5
- data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base.hpp +11 -5
- data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base_impl.hpp +70 -37
- data/vendor/datasketches-cpp/theta/test/CMakeLists.txt +1 -0
- data/vendor/datasketches-cpp/theta/test/theta_a_not_b_test.cpp +23 -1
- data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java_v1.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java_v2.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java_v1.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java_v2.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_compact_exact_from_java.sk +0 -0
- data/vendor/datasketches-cpp/theta/test/theta_intersection_test.cpp +21 -1
- data/vendor/datasketches-cpp/theta/test/theta_jaccard_similarity_test.cpp +58 -2
- data/vendor/datasketches-cpp/theta/test/theta_setop_test.cpp +445 -0
- data/vendor/datasketches-cpp/theta/test/theta_sketch_test.cpp +437 -1
- data/vendor/datasketches-cpp/theta/test/theta_union_test.cpp +41 -9
- data/vendor/datasketches-cpp/tuple/CMakeLists.txt +18 -33
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch.hpp +1 -1
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch_impl.hpp +50 -63
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union.hpp +1 -1
- data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union_impl.hpp +3 -3
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch.hpp +13 -9
- data/vendor/datasketches-cpp/tuple/include/tuple_sketch_impl.hpp +84 -78
- data/vendor/datasketches-cpp/tuple/include/tuple_union.hpp +6 -1
- data/vendor/datasketches-cpp/tuple/include/tuple_union_impl.hpp +8 -3
- data/vendor/datasketches-cpp/tuple/test/array_of_doubles_sketch_test.cpp +17 -1
- data/vendor/datasketches-cpp/tuple/test/tuple_a_not_b_test.cpp +17 -17
- data/vendor/datasketches-cpp/tuple/test/tuple_intersection_test.cpp +12 -12
- data/vendor/datasketches-cpp/tuple/test/tuple_jaccard_similarity_test.cpp +5 -5
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_allocation_test.cpp +1 -1
- data/vendor/datasketches-cpp/tuple/test/tuple_sketch_test.cpp +66 -28
- data/vendor/datasketches-cpp/tuple/test/tuple_union_test.cpp +19 -12
- metadata +18 -7
- 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
@@ -110,14 +110,14 @@ public:
|
|
110
110
|
* @return the lower bound of the approximate Clopper-Pearson confidence interval for the
|
111
111
|
* unknown success probability.
|
112
112
|
*/
|
113
|
-
static inline double approximate_lower_bound_on_p(
|
113
|
+
static inline double approximate_lower_bound_on_p(uint64_t n, uint64_t k, double num_std_devs) {
|
114
114
|
check_inputs(n, k);
|
115
115
|
if (n == 0) { return 0.0; } // the coin was never flipped, so we know nothing
|
116
116
|
else if (k == 0) { return 0.0; }
|
117
117
|
else if (k == 1) { return (exact_lower_bound_on_p_k_eq_1(n, delta_of_num_stdevs(num_std_devs))); }
|
118
118
|
else if (k == n) { return (exact_lower_bound_on_p_k_eq_n(n, delta_of_num_stdevs(num_std_devs))); }
|
119
119
|
else {
|
120
|
-
double x = abramowitz_stegun_formula_26p5p22((n - k) + 1, k, (-1.0 * num_std_devs));
|
120
|
+
double x = abramowitz_stegun_formula_26p5p22((n - k) + 1.0, static_cast<double>(k), (-1.0 * num_std_devs));
|
121
121
|
return (1.0 - x); // which is p
|
122
122
|
}
|
123
123
|
}
|
@@ -145,18 +145,18 @@ public:
|
|
145
145
|
* @return the upper bound of the approximate Clopper-Pearson confidence interval for the
|
146
146
|
* unknown success probability.
|
147
147
|
*/
|
148
|
-
static inline double approximate_upper_bound_on_p(
|
148
|
+
static inline double approximate_upper_bound_on_p(uint64_t n, uint64_t k, double num_std_devs) {
|
149
149
|
check_inputs(n, k);
|
150
150
|
if (n == 0) { return 1.0; } // the coin was never flipped, so we know nothing
|
151
151
|
else if (k == n) { return 1.0; }
|
152
152
|
else if (k == (n - 1)) {
|
153
|
-
return (
|
153
|
+
return (exact_upper_bound_on_p_k_eq_minusone(n, delta_of_num_stdevs(num_std_devs)));
|
154
154
|
}
|
155
155
|
else if (k == 0) {
|
156
156
|
return (exact_upper_bound_on_p_k_eq_zero(n, delta_of_num_stdevs(num_std_devs)));
|
157
157
|
}
|
158
158
|
else {
|
159
|
-
double x = abramowitz_stegun_formula_26p5p22(n - k, k + 1, num_std_devs);
|
159
|
+
double x = abramowitz_stegun_formula_26p5p22(static_cast<double>(n - k), k + 1.0, num_std_devs);
|
160
160
|
return (1.0 - x); // which is p
|
161
161
|
}
|
162
162
|
}
|
@@ -167,7 +167,7 @@ public:
|
|
167
167
|
* @param k is the number of successes. Must be non-negative, and cannot exceed n.
|
168
168
|
* @return the estimate of the unknown binomial proportion.
|
169
169
|
*/
|
170
|
-
static inline double estimate_unknown_p(
|
170
|
+
static inline double estimate_unknown_p(uint64_t n, uint64_t k) {
|
171
171
|
check_inputs(n, k);
|
172
172
|
if (n == 0) { return 0.5; } // the coin was never flipped, so we know nothing
|
173
173
|
else { return ((double) k / (double) n); }
|
@@ -193,9 +193,7 @@ public:
|
|
193
193
|
}
|
194
194
|
|
195
195
|
private:
|
196
|
-
static inline void check_inputs(
|
197
|
-
if (n < 0) { throw std::invalid_argument("N must be non-negative"); }
|
198
|
-
if (k < 0) { throw std::invalid_argument("K must be non-negative"); }
|
196
|
+
static inline void check_inputs(uint64_t n, uint64_t k) {
|
199
197
|
if (k > n) { throw std::invalid_argument("K cannot exceed N"); }
|
200
198
|
}
|
201
199
|
|
@@ -251,8 +249,7 @@ private:
|
|
251
249
|
// and it is worth keeping it that way so that it will always be easy to verify
|
252
250
|
// that the formula was typed in correctly.
|
253
251
|
|
254
|
-
static inline double abramowitz_stegun_formula_26p5p22(double a, double b,
|
255
|
-
double yp) {
|
252
|
+
static inline double abramowitz_stegun_formula_26p5p22(double a, double b, double yp) {
|
256
253
|
const double b2m1 = (2.0 * b) - 1.0;
|
257
254
|
const double a2m1 = (2.0 * a) - 1.0;
|
258
255
|
const double lambda = ((yp * yp) - 3.0) / 6.0;
|
@@ -268,19 +265,19 @@ private:
|
|
268
265
|
|
269
266
|
// Formulas for some special cases.
|
270
267
|
|
271
|
-
static inline double exact_upper_bound_on_p_k_eq_zero(
|
268
|
+
static inline double exact_upper_bound_on_p_k_eq_zero(uint64_t n, double delta) {
|
272
269
|
return (1.0 - pow(delta, (1.0 / n)));
|
273
270
|
}
|
274
271
|
|
275
|
-
static inline double exact_lower_bound_on_p_k_eq_n(
|
272
|
+
static inline double exact_lower_bound_on_p_k_eq_n(uint64_t n, double delta) {
|
276
273
|
return (pow(delta, (1.0 / n)));
|
277
274
|
}
|
278
275
|
|
279
|
-
static inline double exact_lower_bound_on_p_k_eq_1(
|
276
|
+
static inline double exact_lower_bound_on_p_k_eq_1(uint64_t n, double delta) {
|
280
277
|
return (1.0 - pow((1.0 - delta), (1.0 / n)));
|
281
278
|
}
|
282
279
|
|
283
|
-
static inline double
|
280
|
+
static inline double exact_upper_bound_on_p_k_eq_minusone(uint64_t n, double delta) {
|
284
281
|
return (pow((1.0 - delta), (1.0 / n)));
|
285
282
|
}
|
286
283
|
|
@@ -23,11 +23,14 @@
|
|
23
23
|
#include <cstdint>
|
24
24
|
#include <string>
|
25
25
|
#include <memory>
|
26
|
+
#include <iostream>
|
26
27
|
|
27
28
|
namespace datasketches {
|
28
29
|
|
29
30
|
static const uint64_t DEFAULT_SEED = 9001;
|
30
31
|
|
32
|
+
enum resize_factor { X1 = 0, X2, X4, X8 };
|
33
|
+
|
31
34
|
template<typename A> using AllocChar = typename std::allocator_traits<A>::template rebind_alloc<char>;
|
32
35
|
template<typename A> using string = std::basic_string<char, std::char_traits<char>, AllocChar<A>>;
|
33
36
|
|
@@ -46,6 +49,29 @@ constexpr uint8_t lg_size_from_count(uint32_t n, double load_factor) {
|
|
46
49
|
return log2(n) + ((n > static_cast<uint32_t>((1 << (log2(n) + 1)) * load_factor)) ? 2 : 1);
|
47
50
|
}
|
48
51
|
|
52
|
+
// stream helpers to hide casts
|
53
|
+
template<typename T>
|
54
|
+
static inline T read(std::istream& is) {
|
55
|
+
T value;
|
56
|
+
is.read(reinterpret_cast<char*>(&value), sizeof(T));
|
57
|
+
return value;
|
58
|
+
}
|
59
|
+
|
60
|
+
template<typename T>
|
61
|
+
static inline void read(std::istream& is, T* ptr, size_t size_bytes) {
|
62
|
+
is.read(reinterpret_cast<char*>(ptr), size_bytes);
|
63
|
+
}
|
64
|
+
|
65
|
+
template<typename T>
|
66
|
+
static inline void write(std::ostream& os, T& value) {
|
67
|
+
os.write(reinterpret_cast<const char*>(&value), sizeof(T));
|
68
|
+
}
|
69
|
+
|
70
|
+
template<typename T>
|
71
|
+
static inline void write(std::ostream& os, const T* ptr, size_t size_bytes) {
|
72
|
+
os.write(reinterpret_cast<const char*>(ptr), size_bytes);
|
73
|
+
}
|
74
|
+
|
49
75
|
} // namespace
|
50
76
|
|
51
77
|
#endif // _COMMON_DEFS_HPP_
|
@@ -38,29 +38,41 @@ fwd_type<T1, T2> conditional_forward(T2&& value) {
|
|
38
38
|
// Forward container as iterators
|
39
39
|
|
40
40
|
template<typename Container>
|
41
|
-
auto forward_begin(Container&& c) ->
|
42
|
-
|
41
|
+
auto forward_begin(Container&& c) -> typename std::enable_if<
|
42
|
+
std::is_lvalue_reference<Container>::value ||
|
43
|
+
std::is_same<typename std::remove_reference<Container>::type::const_iterator, decltype(c.begin())>::value,
|
44
|
+
decltype(c.begin())
|
45
|
+
>::type
|
43
46
|
{
|
44
47
|
return c.begin();
|
45
48
|
}
|
46
49
|
|
47
50
|
template<typename Container>
|
48
|
-
auto forward_begin(Container&& c) ->
|
49
|
-
|
51
|
+
auto forward_begin(Container&& c) -> typename std::enable_if<
|
52
|
+
!std::is_lvalue_reference<Container>::value &&
|
53
|
+
!std::is_same<typename std::remove_reference<Container>::type::const_iterator, decltype(c.begin())>::value,
|
54
|
+
decltype(std::make_move_iterator(c.begin()))
|
55
|
+
>::type
|
50
56
|
{
|
51
57
|
return std::make_move_iterator(c.begin());
|
52
58
|
}
|
53
59
|
|
54
60
|
template<typename Container>
|
55
|
-
auto forward_end(Container&& c) ->
|
56
|
-
|
61
|
+
auto forward_end(Container&& c) -> typename std::enable_if<
|
62
|
+
std::is_lvalue_reference<Container>::value ||
|
63
|
+
std::is_same<typename std::remove_reference<Container>::type::const_iterator, decltype(c.begin())>::value,
|
64
|
+
decltype(c.end())
|
65
|
+
>::type
|
57
66
|
{
|
58
67
|
return c.end();
|
59
68
|
}
|
60
69
|
|
61
70
|
template<typename Container>
|
62
|
-
auto forward_end(Container&& c) ->
|
63
|
-
|
71
|
+
auto forward_end(Container&& c) -> typename std::enable_if<
|
72
|
+
!std::is_lvalue_reference<Container>::value &&
|
73
|
+
!std::is_same<typename std::remove_reference<Container>::type::const_iterator, decltype(c.begin())>::value,
|
74
|
+
decltype(std::make_move_iterator(c.end()))
|
75
|
+
>::type
|
64
76
|
{
|
65
77
|
return std::make_move_iterator(c.end());
|
66
78
|
}
|
@@ -94,7 +94,7 @@ static inline uint8_t count_leading_zeros_in_u64(uint64_t input) {
|
|
94
94
|
static inline uint8_t count_trailing_zeros_in_u32(uint32_t input) {
|
95
95
|
for (int i = 0; i < 4; i++) {
|
96
96
|
const int byte = input & 0xff;
|
97
|
-
if (byte != 0) return (i << 3) + byte_trailing_zeros_table[byte];
|
97
|
+
if (byte != 0) return static_cast<uint8_t>((i << 3) + byte_trailing_zeros_table[byte]);
|
98
98
|
input >>= 8;
|
99
99
|
}
|
100
100
|
return 32;
|
@@ -103,7 +103,7 @@ static inline uint8_t count_trailing_zeros_in_u32(uint32_t input) {
|
|
103
103
|
static inline uint8_t count_trailing_zeros_in_u64(uint64_t input) {
|
104
104
|
for (int i = 0; i < 8; i++) {
|
105
105
|
const int byte = input & 0xff;
|
106
|
-
if (byte != 0) return (i << 3) + byte_trailing_zeros_table[byte];
|
106
|
+
if (byte != 0) return static_cast<uint8_t>((i << 3) + byte_trailing_zeros_table[byte]);
|
107
107
|
input >>= 8;
|
108
108
|
}
|
109
109
|
return 64;
|
@@ -51,7 +51,7 @@ struct serde<T, typename std::enable_if<std::is_arithmetic<T>::value>::type> {
|
|
51
51
|
bool failure = false;
|
52
52
|
try {
|
53
53
|
os.write(reinterpret_cast<const char*>(items), sizeof(T) * num);
|
54
|
-
} catch (std::ostream::failure&
|
54
|
+
} catch (std::ostream::failure&) {
|
55
55
|
failure = true;
|
56
56
|
}
|
57
57
|
if (failure || !os.good()) {
|
@@ -62,7 +62,7 @@ struct serde<T, typename std::enable_if<std::is_arithmetic<T>::value>::type> {
|
|
62
62
|
bool failure = false;
|
63
63
|
try {
|
64
64
|
is.read((char*)items, sizeof(T) * num);
|
65
|
-
} catch (std::istream::failure&
|
65
|
+
} catch (std::istream::failure&) {
|
66
66
|
failure = true;
|
67
67
|
}
|
68
68
|
if (failure || !is.good()) {
|
@@ -99,11 +99,11 @@ struct serde<std::string> {
|
|
99
99
|
bool failure = false;
|
100
100
|
try {
|
101
101
|
for (; i < num && os.good(); i++) {
|
102
|
-
uint32_t length = items[i].size();
|
102
|
+
uint32_t length = static_cast<uint32_t>(items[i].size());
|
103
103
|
os.write((char*)&length, sizeof(length));
|
104
104
|
os.write(items[i].c_str(), length);
|
105
105
|
}
|
106
|
-
} catch (std::ostream::failure&
|
106
|
+
} catch (std::ostream::failure&) {
|
107
107
|
failure = true;
|
108
108
|
}
|
109
109
|
if (failure || !os.good()) {
|
@@ -121,12 +121,12 @@ struct serde<std::string> {
|
|
121
121
|
std::string str;
|
122
122
|
str.reserve(length);
|
123
123
|
for (uint32_t j = 0; j < length; j++) {
|
124
|
-
str.push_back(is.get());
|
124
|
+
str.push_back(static_cast<char>(is.get()));
|
125
125
|
}
|
126
126
|
if (!is.good()) { break; }
|
127
127
|
new (&items[i]) std::string(std::move(str));
|
128
128
|
}
|
129
|
-
} catch (std::istream::failure&
|
129
|
+
} catch (std::istream::failure&) {
|
130
130
|
failure = true;
|
131
131
|
}
|
132
132
|
if (failure || !is.good()) {
|
@@ -143,7 +143,7 @@ struct serde<std::string> {
|
|
143
143
|
size_t serialize(void* ptr, size_t capacity, const std::string* items, unsigned num) const {
|
144
144
|
size_t bytes_written = 0;
|
145
145
|
for (unsigned i = 0; i < num; ++i) {
|
146
|
-
const uint32_t length = items[i].size();
|
146
|
+
const uint32_t length = static_cast<uint32_t>(items[i].size());
|
147
147
|
const size_t new_bytes = length + sizeof(length);
|
148
148
|
check_memory_size(bytes_written + new_bytes, capacity);
|
149
149
|
memcpy(ptr, &length, sizeof(length));
|
@@ -32,43 +32,23 @@ target_include_directories(cpc
|
|
32
32
|
target_link_libraries(cpc INTERFACE common)
|
33
33
|
target_compile_features(cpc INTERFACE cxx_std_11)
|
34
34
|
|
35
|
-
set(cpc_HEADERS "")
|
36
|
-
list(APPEND cpc_HEADERS "include/compression_data.hpp")
|
37
|
-
list(APPEND cpc_HEADERS "include/cpc_common.hpp")
|
38
|
-
list(APPEND cpc_HEADERS "include/cpc_compressor.hpp")
|
39
|
-
list(APPEND cpc_HEADERS "include/cpc_compressor_impl.hpp")
|
40
|
-
list(APPEND cpc_HEADERS "include/cpc_confidence.hpp")
|
41
|
-
list(APPEND cpc_HEADERS "include/cpc_sketch.hpp")
|
42
|
-
list(APPEND cpc_HEADERS "include/cpc_sketch_impl.hpp")
|
43
|
-
list(APPEND cpc_HEADERS "include/cpc_union.hpp")
|
44
|
-
list(APPEND cpc_HEADERS "include/cpc_union_impl.hpp")
|
45
|
-
list(APPEND cpc_HEADERS "include/cpc_util.hpp")
|
46
|
-
list(APPEND cpc_HEADERS "include/icon_estimator.hpp")
|
47
|
-
list(APPEND cpc_HEADERS "include/kxp_byte_lookup.hpp")
|
48
|
-
list(APPEND cpc_HEADERS "include/u32_table.hpp")
|
49
|
-
list(APPEND cpc_HEADERS "include/u32_table_impl.hpp")
|
50
|
-
|
51
35
|
install(TARGETS cpc
|
52
36
|
EXPORT ${PROJECT_NAME}
|
53
37
|
)
|
54
38
|
|
55
|
-
install(FILES
|
39
|
+
install(FILES
|
40
|
+
include/compression_data.hpp
|
41
|
+
include/cpc_common.hpp
|
42
|
+
include/cpc_compressor.hpp
|
43
|
+
include/cpc_compressor_impl.hpp
|
44
|
+
include/cpc_confidence.hpp
|
45
|
+
include/cpc_sketch.hpp
|
46
|
+
include/cpc_sketch_impl.hpp
|
47
|
+
include/cpc_union.hpp
|
48
|
+
include/cpc_union_impl.hpp
|
49
|
+
include/cpc_util.hpp
|
50
|
+
include/icon_estimator.hpp
|
51
|
+
include/kxp_byte_lookup.hpp
|
52
|
+
include/u32_table.hpp
|
53
|
+
include/u32_table_impl.hpp
|
56
54
|
DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}/DataSketches")
|
57
|
-
|
58
|
-
target_sources(cpc
|
59
|
-
INTERFACE
|
60
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/compression_data.hpp
|
61
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_common.hpp
|
62
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_compressor.hpp
|
63
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_compressor_impl.hpp
|
64
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_confidence.hpp
|
65
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_sketch.hpp
|
66
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_sketch_impl.hpp
|
67
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_union.hpp
|
68
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_union_impl.hpp
|
69
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/cpc_util.hpp
|
70
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/icon_estimator.hpp
|
71
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/kxp_byte_lookup.hpp
|
72
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/u32_table.hpp
|
73
|
-
${CMAKE_CURRENT_SOURCE_DIR}/include/u32_table_impl.hpp
|
74
|
-
)
|
@@ -26,9 +26,16 @@
|
|
26
26
|
|
27
27
|
namespace datasketches {
|
28
28
|
|
29
|
-
|
30
|
-
|
31
|
-
|
29
|
+
namespace cpc_constants {
|
30
|
+
const uint8_t MIN_LG_K = 4;
|
31
|
+
const uint8_t MAX_LG_K = 26;
|
32
|
+
const uint8_t DEFAULT_LG_K = 11;
|
33
|
+
}
|
34
|
+
|
35
|
+
// TODO: Redundant and deprecated. Will be removed in next major version release.
|
36
|
+
static const uint8_t CPC_MIN_LG_K = cpc_constants::MIN_LG_K;
|
37
|
+
static const uint8_t CPC_MAX_LG_K = cpc_constants::MAX_LG_K;
|
38
|
+
static const uint8_t CPC_DEFAULT_LG_K = cpc_constants::DEFAULT_LG_K;
|
32
39
|
|
33
40
|
template<typename A> using AllocU8 = typename std::allocator_traits<A>::template rebind_alloc<uint8_t>;
|
34
41
|
template<typename A> using AllocU16 = typename std::allocator_traits<A>::template rebind_alloc<uint16_t>;
|
@@ -48,44 +48,44 @@ template<typename A>
|
|
48
48
|
class cpc_compressor {
|
49
49
|
public:
|
50
50
|
void compress(const cpc_sketch_alloc<A>& source, compressed_state<A>& target) const;
|
51
|
-
void uncompress(const compressed_state<A>& source, uncompressed_state<A>& target, uint8_t lg_k,
|
51
|
+
void uncompress(const compressed_state<A>& source, uncompressed_state<A>& target, uint8_t lg_k, uint32_t num_coupons) const;
|
52
52
|
|
53
53
|
// methods below are public for testing
|
54
54
|
|
55
55
|
// This returns the number of compressed words that were actually used. It is the caller's
|
56
56
|
// responsibility to ensure that the compressed_words array is long enough to prevent over-run.
|
57
|
-
|
57
|
+
uint32_t low_level_compress_bytes(
|
58
58
|
const uint8_t* byte_array, // input
|
59
|
-
|
59
|
+
uint32_t num_bytes_to_encode,
|
60
60
|
const uint16_t* encoding_table,
|
61
61
|
uint32_t* compressed_words // output
|
62
62
|
) const;
|
63
63
|
|
64
64
|
void low_level_uncompress_bytes(
|
65
65
|
uint8_t* byte_array, // output
|
66
|
-
|
66
|
+
uint32_t num_bytes_to_decode,
|
67
67
|
const uint16_t* decoding_table,
|
68
68
|
const uint32_t* compressed_words,
|
69
|
-
|
69
|
+
uint32_t num_compressed_words // input
|
70
70
|
) const;
|
71
71
|
|
72
72
|
// Here "pairs" refers to row-column pairs that specify
|
73
73
|
// the positions of surprising values in the bit matrix.
|
74
74
|
|
75
75
|
// returns the number of compressedWords actually used
|
76
|
-
|
76
|
+
uint32_t low_level_compress_pairs(
|
77
77
|
const uint32_t* pair_array, // input
|
78
|
-
|
79
|
-
|
78
|
+
uint32_t num_pairs_to_encode,
|
79
|
+
uint8_t num_base_bits,
|
80
80
|
uint32_t* compressed_words // output
|
81
81
|
) const;
|
82
82
|
|
83
83
|
void low_level_uncompress_pairs(
|
84
84
|
uint32_t* pair_array, // output
|
85
|
-
|
86
|
-
|
85
|
+
uint32_t num_pairs_to_decode,
|
86
|
+
uint8_t num_base_bits,
|
87
87
|
const uint32_t* compressed_words, // input
|
88
|
-
|
88
|
+
uint32_t num_compressed_words // input
|
89
89
|
) const;
|
90
90
|
|
91
91
|
private:
|
@@ -122,22 +122,22 @@ private:
|
|
122
122
|
void uncompress_pinned_flavor(const compressed_state<A>& source, uncompressed_state<A>& target, uint8_t lg_k, uint32_t num_coupons) const;
|
123
123
|
void uncompress_sliding_flavor(const compressed_state<A>& source, uncompressed_state<A>& target, uint8_t lg_k, uint32_t num_coupons) const;
|
124
124
|
|
125
|
-
uint8_t* make_inverse_permutation(const uint8_t* permu,
|
126
|
-
uint16_t* make_decoding_table(const uint16_t* encoding_table,
|
125
|
+
uint8_t* make_inverse_permutation(const uint8_t* permu, unsigned length);
|
126
|
+
uint16_t* make_decoding_table(const uint16_t* encoding_table, unsigned num_byte_values);
|
127
127
|
void validate_decoding_table(const uint16_t* decoding_table, const uint16_t* encoding_table) const;
|
128
128
|
|
129
129
|
void compress_surprising_values(const vector_u32<A>& pairs, uint8_t lg_k, compressed_state<A>& result) const;
|
130
130
|
void compress_sliding_window(const uint8_t* window, uint8_t lg_k, uint32_t num_coupons, compressed_state<A>& target) const;
|
131
131
|
|
132
|
-
vector_u32<A> uncompress_surprising_values(const uint32_t* data,
|
133
|
-
void uncompress_sliding_window(const uint32_t* data,
|
132
|
+
vector_u32<A> uncompress_surprising_values(const uint32_t* data, uint32_t data_words, uint32_t num_pairs, uint8_t lg_k, const A& allocator) const;
|
133
|
+
void uncompress_sliding_window(const uint32_t* data, uint32_t data_words, vector_u8<A>& window, uint8_t lg_k, uint32_t num_coupons) const;
|
134
134
|
|
135
|
-
static size_t safe_length_for_compressed_pair_buf(
|
136
|
-
static size_t safe_length_for_compressed_window_buf(
|
137
|
-
static uint8_t determine_pseudo_phase(uint8_t lg_k,
|
135
|
+
static size_t safe_length_for_compressed_pair_buf(uint32_t k, uint32_t num_pairs, uint8_t num_base_bits);
|
136
|
+
static size_t safe_length_for_compressed_window_buf(uint32_t k);
|
137
|
+
static uint8_t determine_pseudo_phase(uint8_t lg_k, uint32_t c);
|
138
138
|
|
139
139
|
static inline vector_u32<A> tricky_get_pairs_from_window(const uint8_t* window, uint32_t k, uint32_t num_pairs_to_get, uint32_t empty_space, const A& allocator);
|
140
|
-
static inline
|
140
|
+
static inline uint8_t golomb_choose_number_of_base_bits(uint32_t k, uint64_t count);
|
141
141
|
};
|
142
142
|
|
143
143
|
} /* namespace datasketches */
|