datasketches 0.2.0 → 0.2.4

Sign up to get free protection for your applications and to get access to all the features.
Files changed (170) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +16 -0
  3. data/LICENSE +40 -3
  4. data/NOTICE +1 -1
  5. data/README.md +7 -7
  6. data/ext/datasketches/extconf.rb +1 -1
  7. data/ext/datasketches/theta_wrapper.cpp +20 -4
  8. data/lib/datasketches/version.rb +1 -1
  9. data/vendor/datasketches-cpp/CMakeLists.txt +31 -3
  10. data/vendor/datasketches-cpp/LICENSE +40 -3
  11. data/vendor/datasketches-cpp/MANIFEST.in +3 -0
  12. data/vendor/datasketches-cpp/NOTICE +1 -1
  13. data/vendor/datasketches-cpp/README.md +76 -9
  14. data/vendor/datasketches-cpp/cmake/DataSketchesConfig.cmake.in +10 -0
  15. data/vendor/datasketches-cpp/common/CMakeLists.txt +14 -13
  16. data/vendor/datasketches-cpp/common/include/MurmurHash3.h +11 -7
  17. data/vendor/datasketches-cpp/common/include/binomial_bounds.hpp +8 -8
  18. data/vendor/datasketches-cpp/common/include/bounds_binomial_proportions.hpp +12 -15
  19. data/vendor/datasketches-cpp/common/include/common_defs.hpp +26 -0
  20. data/vendor/datasketches-cpp/common/include/conditional_forward.hpp +20 -8
  21. data/vendor/datasketches-cpp/common/include/count_zeros.hpp +2 -2
  22. data/vendor/datasketches-cpp/common/include/serde.hpp +7 -7
  23. data/vendor/datasketches-cpp/cpc/CMakeLists.txt +15 -35
  24. data/vendor/datasketches-cpp/cpc/include/cpc_common.hpp +10 -3
  25. data/vendor/datasketches-cpp/cpc/include/cpc_compressor.hpp +19 -19
  26. data/vendor/datasketches-cpp/cpc/include/cpc_compressor_impl.hpp +91 -89
  27. data/vendor/datasketches-cpp/cpc/include/cpc_sketch.hpp +15 -2
  28. data/vendor/datasketches-cpp/cpc/include/cpc_sketch_impl.hpp +126 -90
  29. data/vendor/datasketches-cpp/cpc/include/cpc_union.hpp +1 -1
  30. data/vendor/datasketches-cpp/cpc/include/cpc_union_impl.hpp +22 -20
  31. data/vendor/datasketches-cpp/cpc/include/cpc_util.hpp +10 -10
  32. data/vendor/datasketches-cpp/cpc/include/icon_estimator.hpp +4 -4
  33. data/vendor/datasketches-cpp/cpc/include/u32_table.hpp +8 -8
  34. data/vendor/datasketches-cpp/cpc/include/u32_table_impl.hpp +14 -14
  35. data/vendor/datasketches-cpp/cpc/test/compression_test.cpp +10 -10
  36. data/vendor/datasketches-cpp/cpc/test/cpc_sketch_allocation_test.cpp +17 -0
  37. data/vendor/datasketches-cpp/cpc/test/cpc_sketch_test.cpp +25 -0
  38. data/vendor/datasketches-cpp/cpc/test/cpc_union_test.cpp +1 -1
  39. data/vendor/datasketches-cpp/fi/CMakeLists.txt +5 -15
  40. data/vendor/datasketches-cpp/fi/include/frequent_items_sketch_impl.hpp +69 -82
  41. data/vendor/datasketches-cpp/fi/include/reverse_purge_hash_map_impl.hpp +10 -10
  42. data/vendor/datasketches-cpp/fi/test/reverse_purge_hash_map_test.cpp +2 -2
  43. data/vendor/datasketches-cpp/hll/CMakeLists.txt +33 -56
  44. data/vendor/datasketches-cpp/hll/include/AuxHashMap-internal.hpp +60 -63
  45. data/vendor/datasketches-cpp/hll/include/AuxHashMap.hpp +19 -19
  46. data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable-internal.hpp +15 -15
  47. data/vendor/datasketches-cpp/hll/include/CompositeInterpolationXTable.hpp +3 -3
  48. data/vendor/datasketches-cpp/hll/include/CouponHashSet-internal.hpp +74 -76
  49. data/vendor/datasketches-cpp/hll/include/CouponHashSet.hpp +6 -6
  50. data/vendor/datasketches-cpp/hll/include/CouponList-internal.hpp +110 -113
  51. data/vendor/datasketches-cpp/hll/include/CouponList.hpp +13 -13
  52. data/vendor/datasketches-cpp/hll/include/CubicInterpolation-internal.hpp +2 -4
  53. data/vendor/datasketches-cpp/hll/include/HarmonicNumbers-internal.hpp +1 -1
  54. data/vendor/datasketches-cpp/hll/include/Hll4Array-internal.hpp +80 -76
  55. data/vendor/datasketches-cpp/hll/include/Hll4Array.hpp +9 -9
  56. data/vendor/datasketches-cpp/hll/include/Hll6Array-internal.hpp +26 -26
  57. data/vendor/datasketches-cpp/hll/include/Hll6Array.hpp +6 -6
  58. data/vendor/datasketches-cpp/hll/include/Hll8Array-internal.hpp +33 -33
  59. data/vendor/datasketches-cpp/hll/include/Hll8Array.hpp +6 -6
  60. data/vendor/datasketches-cpp/hll/include/HllArray-internal.hpp +205 -209
  61. data/vendor/datasketches-cpp/hll/include/HllArray.hpp +36 -36
  62. data/vendor/datasketches-cpp/hll/include/HllSketch-internal.hpp +34 -32
  63. data/vendor/datasketches-cpp/hll/include/HllSketchImpl-internal.hpp +22 -22
  64. data/vendor/datasketches-cpp/hll/include/HllSketchImpl.hpp +13 -13
  65. data/vendor/datasketches-cpp/hll/include/HllSketchImplFactory.hpp +15 -15
  66. data/vendor/datasketches-cpp/hll/include/HllUnion-internal.hpp +61 -61
  67. data/vendor/datasketches-cpp/hll/include/HllUtil.hpp +120 -127
  68. data/vendor/datasketches-cpp/hll/include/coupon_iterator-internal.hpp +9 -9
  69. data/vendor/datasketches-cpp/hll/include/coupon_iterator.hpp +5 -5
  70. data/vendor/datasketches-cpp/hll/include/hll.hpp +21 -21
  71. data/vendor/datasketches-cpp/hll/test/AuxHashMapTest.cpp +1 -1
  72. data/vendor/datasketches-cpp/hll/test/CouponHashSetTest.cpp +34 -34
  73. data/vendor/datasketches-cpp/hll/test/CouponListTest.cpp +25 -25
  74. data/vendor/datasketches-cpp/hll/test/CrossCountingTest.cpp +2 -2
  75. data/vendor/datasketches-cpp/hll/test/HllArrayTest.cpp +35 -35
  76. data/vendor/datasketches-cpp/hll/test/HllSketchTest.cpp +15 -15
  77. data/vendor/datasketches-cpp/hll/test/HllUnionTest.cpp +10 -14
  78. data/vendor/datasketches-cpp/hll/test/IsomorphicTest.cpp +3 -3
  79. data/vendor/datasketches-cpp/hll/test/ToFromByteArrayTest.cpp +4 -4
  80. data/vendor/datasketches-cpp/kll/CMakeLists.txt +9 -19
  81. data/vendor/datasketches-cpp/kll/include/kll_helper.hpp +5 -4
  82. data/vendor/datasketches-cpp/kll/include/kll_helper_impl.hpp +6 -6
  83. data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator.hpp +14 -6
  84. data/vendor/datasketches-cpp/kll/include/kll_quantile_calculator_impl.hpp +39 -24
  85. data/vendor/datasketches-cpp/kll/include/kll_sketch.hpp +41 -4
  86. data/vendor/datasketches-cpp/kll/include/kll_sketch_impl.hpp +76 -64
  87. data/vendor/datasketches-cpp/kll/include/kolmogorov_smirnov.hpp +67 -0
  88. data/vendor/datasketches-cpp/kll/include/kolmogorov_smirnov_impl.hpp +78 -0
  89. data/vendor/datasketches-cpp/kll/test/CMakeLists.txt +1 -0
  90. data/vendor/datasketches-cpp/kll/test/kll_sketch_test.cpp +133 -46
  91. data/vendor/datasketches-cpp/kll/test/kolmogorov_smirnov_test.cpp +111 -0
  92. data/vendor/datasketches-cpp/pyproject.toml +4 -2
  93. data/vendor/datasketches-cpp/python/CMakeLists.txt +10 -6
  94. data/vendor/datasketches-cpp/python/README.md +50 -50
  95. data/vendor/datasketches-cpp/python/pybind11Path.cmd +3 -0
  96. data/vendor/datasketches-cpp/python/src/cpc_wrapper.cpp +1 -1
  97. data/vendor/datasketches-cpp/python/src/hll_wrapper.cpp +4 -4
  98. data/vendor/datasketches-cpp/python/src/kll_wrapper.cpp +1 -1
  99. data/vendor/datasketches-cpp/python/src/theta_wrapper.cpp +8 -8
  100. data/vendor/datasketches-cpp/python/src/vector_of_kll.cpp +11 -5
  101. data/vendor/datasketches-cpp/python/src/vo_wrapper.cpp +2 -2
  102. data/vendor/datasketches-cpp/python/tests/hll_test.py +1 -1
  103. data/vendor/datasketches-cpp/python/tests/kll_test.py +2 -2
  104. data/vendor/datasketches-cpp/python/tests/req_test.py +2 -2
  105. data/vendor/datasketches-cpp/python/tests/vector_of_kll_test.py +4 -4
  106. data/vendor/datasketches-cpp/python/tests/vo_test.py +3 -3
  107. data/vendor/datasketches-cpp/req/CMakeLists.txt +8 -21
  108. data/vendor/datasketches-cpp/req/include/req_common.hpp +2 -1
  109. data/vendor/datasketches-cpp/req/include/req_compactor.hpp +4 -4
  110. data/vendor/datasketches-cpp/req/include/req_compactor_impl.hpp +26 -39
  111. data/vendor/datasketches-cpp/req/include/req_sketch.hpp +1 -1
  112. data/vendor/datasketches-cpp/req/include/req_sketch_impl.hpp +13 -11
  113. data/vendor/datasketches-cpp/req/test/req_sketch_test.cpp +52 -52
  114. data/vendor/datasketches-cpp/sampling/CMakeLists.txt +5 -9
  115. data/vendor/datasketches-cpp/sampling/include/var_opt_sketch.hpp +10 -5
  116. data/vendor/datasketches-cpp/sampling/include/var_opt_sketch_impl.hpp +61 -64
  117. data/vendor/datasketches-cpp/sampling/include/var_opt_union_impl.hpp +42 -48
  118. data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_test.cpp +6 -6
  119. data/vendor/datasketches-cpp/sampling/test/var_opt_union_test.cpp +13 -13
  120. data/vendor/datasketches-cpp/setup.py +10 -7
  121. data/vendor/datasketches-cpp/theta/CMakeLists.txt +26 -45
  122. data/vendor/datasketches-cpp/theta/include/bounds_on_ratios_in_sampled_sets.hpp +1 -1
  123. data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser.hpp +67 -0
  124. data/vendor/datasketches-cpp/theta/include/compact_theta_sketch_parser_impl.hpp +137 -0
  125. data/vendor/datasketches-cpp/theta/include/theta_constants.hpp +9 -4
  126. data/vendor/datasketches-cpp/theta/include/theta_helpers.hpp +15 -0
  127. data/vendor/datasketches-cpp/theta/include/theta_intersection.hpp +9 -4
  128. data/vendor/datasketches-cpp/theta/include/theta_intersection_base_impl.hpp +6 -6
  129. data/vendor/datasketches-cpp/theta/include/theta_intersection_impl.hpp +1 -1
  130. data/vendor/datasketches-cpp/theta/include/theta_jaccard_similarity_base.hpp +18 -14
  131. data/vendor/datasketches-cpp/theta/include/theta_set_difference_base_impl.hpp +2 -2
  132. data/vendor/datasketches-cpp/theta/include/theta_sketch.hpp +73 -15
  133. data/vendor/datasketches-cpp/theta/include/theta_sketch_impl.hpp +247 -103
  134. data/vendor/datasketches-cpp/theta/include/theta_union.hpp +10 -5
  135. data/vendor/datasketches-cpp/theta/include/theta_union_base.hpp +3 -1
  136. data/vendor/datasketches-cpp/theta/include/theta_union_base_impl.hpp +9 -3
  137. data/vendor/datasketches-cpp/theta/include/theta_union_impl.hpp +8 -5
  138. data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base.hpp +11 -5
  139. data/vendor/datasketches-cpp/theta/include/theta_update_sketch_base_impl.hpp +70 -37
  140. data/vendor/datasketches-cpp/theta/test/CMakeLists.txt +1 -0
  141. data/vendor/datasketches-cpp/theta/test/theta_a_not_b_test.cpp +23 -1
  142. data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java_v1.sk +0 -0
  143. data/vendor/datasketches-cpp/theta/test/theta_compact_empty_from_java_v2.sk +0 -0
  144. data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java_v1.sk +0 -0
  145. data/vendor/datasketches-cpp/theta/test/theta_compact_estimation_from_java_v2.sk +0 -0
  146. data/vendor/datasketches-cpp/theta/test/theta_compact_exact_from_java.sk +0 -0
  147. data/vendor/datasketches-cpp/theta/test/theta_intersection_test.cpp +21 -1
  148. data/vendor/datasketches-cpp/theta/test/theta_jaccard_similarity_test.cpp +58 -2
  149. data/vendor/datasketches-cpp/theta/test/theta_setop_test.cpp +445 -0
  150. data/vendor/datasketches-cpp/theta/test/theta_sketch_test.cpp +437 -1
  151. data/vendor/datasketches-cpp/theta/test/theta_union_test.cpp +41 -9
  152. data/vendor/datasketches-cpp/tuple/CMakeLists.txt +18 -33
  153. data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch.hpp +1 -1
  154. data/vendor/datasketches-cpp/tuple/include/array_of_doubles_sketch_impl.hpp +50 -63
  155. data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union.hpp +1 -1
  156. data/vendor/datasketches-cpp/tuple/include/array_of_doubles_union_impl.hpp +3 -3
  157. data/vendor/datasketches-cpp/tuple/include/tuple_sketch.hpp +13 -9
  158. data/vendor/datasketches-cpp/tuple/include/tuple_sketch_impl.hpp +84 -78
  159. data/vendor/datasketches-cpp/tuple/include/tuple_union.hpp +6 -1
  160. data/vendor/datasketches-cpp/tuple/include/tuple_union_impl.hpp +8 -3
  161. data/vendor/datasketches-cpp/tuple/test/array_of_doubles_sketch_test.cpp +17 -1
  162. data/vendor/datasketches-cpp/tuple/test/tuple_a_not_b_test.cpp +17 -17
  163. data/vendor/datasketches-cpp/tuple/test/tuple_intersection_test.cpp +12 -12
  164. data/vendor/datasketches-cpp/tuple/test/tuple_jaccard_similarity_test.cpp +5 -5
  165. data/vendor/datasketches-cpp/tuple/test/tuple_sketch_allocation_test.cpp +1 -1
  166. data/vendor/datasketches-cpp/tuple/test/tuple_sketch_test.cpp +66 -28
  167. data/vendor/datasketches-cpp/tuple/test/tuple_union_test.cpp +19 -12
  168. metadata +18 -7
  169. data/vendor/datasketches-cpp/theta/test/theta_update_empty_from_java.sk +0 -0
  170. data/vendor/datasketches-cpp/theta/test/theta_update_estimation_from_java.sk +0 -0
data/vendor/datasketches-cpp/hll/include/Hll8Array.hpp CHANGED
@@ -30,24 +30,24 @@ class Hll8Iterator;
30
30
  template<typename A>
31
31
  class Hll8Array final : public HllArray<A> {
32
32
  public:
33
- Hll8Array(int lgConfigK, bool startFullSize, const A& allocator);
33
+ Hll8Array(uint8_t lgConfigK, bool startFullSize, const A& allocator);
34
34
 
35
35
  virtual ~Hll8Array() = default;
36
36
  virtual std::function<void(HllSketchImpl<A>*)> get_deleter() const;
37
37
 
38
38
  virtual Hll8Array<A>* copy() const;
39
39
 
40
- inline uint8_t getSlot(int slotNo) const;
41
- inline void putSlot(int slotNo, uint8_t value);
40
+ inline uint8_t getSlot(uint32_t slotNo) const;
41
+ inline void putSlot(uint32_t slotNo, uint8_t value);
42
42
 
43
- virtual HllSketchImpl<A>* couponUpdate(int coupon) final;
43
+ virtual HllSketchImpl<A>* couponUpdate(uint32_t coupon) final;
44
44
  void mergeList(const CouponList<A>& src);
45
45
  void mergeHll(const HllArray<A>& src);
46
46
 
47
- virtual int getHllByteArrBytes() const;
47
+ virtual uint32_t getHllByteArrBytes() const;
48
48
 
49
49
  private:
50
- inline void internalCouponUpdate(int coupon);
50
+ inline void internalCouponUpdate(uint32_t coupon);
51
51
  };
52
52
 
53
53
  }