datasketches 0.2.5 → 0.2.7

data/vendor/datasketches-cpp/req/include/req_compactor_impl.hpp

@@ -132,6 +132,33 @@ req_compactor<T, C, A>& req_compactor<T, C, A>::operator=(req_compactor&& other)
   return *this;
 }
 
+template<typename T, typename C, typename A>
+template<typename TT, typename CC, typename AA>
+req_compactor<T, C, A>::req_compactor(const req_compactor<TT, CC, AA>& other, const A& allocator):
+allocator_(allocator),
+lg_weight_(other.lg_weight_),
+hra_(other.hra_),
+coin_(other.coin_),
+sorted_(other.sorted_),
+section_size_raw_(other.section_size_raw_),
+section_size_(other.section_size_),
+num_sections_(other.num_sections_),
+state_(other.state_),
+num_items_(other.num_items_),
+capacity_(other.capacity_),
+items_(nullptr)
+{
+  if (other.items_ != nullptr) {
+    items_ = allocator_.allocate(capacity_);
+    const uint32_t from = hra_ ? capacity_ - num_items_ : 0;
+    const uint32_t to = hra_ ? capacity_ : num_items_;
+    for (uint32_t i = from; i < to; ++i) new (items_ + i) T(other.items_[i]);
+    if (sorted_ && !std::is_sorted(items_ + from, items_ + to, C())) {
+      throw std::logic_error("items must be sorted");
+    }
+  }
+}
+
 template<typename T, typename C, typename A>
 bool req_compactor<T, C, A>::is_sorted() const {
   return sorted_;

data/vendor/datasketches-cpp/req/include/req_sketch.hpp

@@ -58,6 +58,14 @@ public:
   req_sketch& operator=(const req_sketch& other);
   req_sketch& operator=(req_sketch&& other);
 
+  /*
+   * Type converting constructor.
+   * @param other sketch of a different type
+   * @param allocator instance of an Allocator
+   */
+  template<typename TT, typename CC, typename SS, typename AA>
+  explicit req_sketch(const req_sketch<TT, CC, SS, AA>& other, const Allocator& allocator = Allocator());
+
   /**
    * Returns configured parameter K
    * @return parameter K
@@ -408,6 +416,9 @@ private:
     }
   }
 
+  // for type converting constructor
+  template<typename TT, typename CC, typename SS, typename AA>
+  friend class req_sketch;
 };
 
 template<typename T, typename C, typename S, typename A>

data/vendor/datasketches-cpp/req/include/req_sketch_impl.hpp

@@ -64,8 +64,8 @@ compactors_(other.compactors_),
 min_value_(nullptr),
 max_value_(nullptr)
 {
-  if (other.min_value_ != nullptr) min_value_ = new (A().allocate(1)) T(*other.min_value_);
-  if (other.max_value_ != nullptr) max_value_ = new (A().allocate(1)) T(*other.max_value_);
+  if (other.min_value_ != nullptr) min_value_ = new (allocator_.allocate(1)) T(*other.min_value_);
+  if (other.max_value_ != nullptr) max_value_ = new (allocator_.allocate(1)) T(*other.max_value_);
 }
 
 template<typename T, typename C, typename S, typename A>
@@ -113,6 +113,33 @@ req_sketch<T, C, S, A>& req_sketch<T, C, S, A>::operator=(req_sketch&& other) {
   return *this;
 }
 
+template<typename T, typename C, typename S, typename A>
+template<typename TT, typename CC, typename SS, typename AA>
+req_sketch<T, C, S, A>::req_sketch(const req_sketch<TT, CC, SS, AA>& other, const A& allocator):
+allocator_(allocator),
+k_(other.k_),
+hra_(other.hra_),
+max_nom_size_(other.max_nom_size_),
+num_retained_(other.num_retained_),
+n_(other.n_),
+compactors_(allocator),
+min_value_(nullptr),
+max_value_(nullptr)
+{
+  static_assert(
+    std::is_constructible<T, TT>::value,
+    "Type converting constructor requires new type to be constructible from existing type"
+  );
+  compactors_.reserve(other.compactors_.size());
+  for (const auto& compactor: other.compactors_) {
+    compactors_.push_back(req_compactor<T, C, A>(compactor, allocator_));
+  }
+  if (!other.is_empty()) {
+    min_value_ = new (allocator_.allocate(1)) T(other.get_min_value());
+    max_value_ = new (allocator_.allocate(1)) T(other.get_max_value());
+  }
+}
+
 template<typename T, typename C, typename S, typename A>
 uint16_t req_sketch<T, C, S, A>::get_k() const {
   return k_;

data/vendor/datasketches-cpp/req/test/req_sketch_custom_type_test.cpp

@@ -17,7 +17,7 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <sstream>
 
 #include <req_sketch.hpp>

data/vendor/datasketches-cpp/req/test/req_sketch_test.cpp

@@ -17,7 +17,7 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <req_sketch.hpp>
 
@@ -35,7 +35,7 @@ const std::string input_path = "test/";
 #endif
 
 TEST_CASE("req sketch: empty", "[req_sketch]") {
-  std::cout << "sizeof(req_float_sketch)=" << sizeof(req_sketch<float>) << "\n";
+  //std::cout << "sizeof(req_float_sketch)=" << sizeof(req_sketch<float>) << "\n";
   req_sketch<float> sketch(12);
   REQUIRE(sketch.get_k() == 12);
   REQUIRE(sketch.is_HRA());
@@ -245,7 +245,7 @@ TEST_CASE("req sketch: byte serialize-deserialize single item", "[req_sketch]")
   auto bytes = sketch.serialize();
   REQUIRE(bytes.size() == sketch.get_serialized_size_bytes());
   auto sketch2 = req_sketch<float>::deserialize(bytes.data(), bytes.size());
-  std::cout << sketch2.to_string(true);
+  //std::cout << sketch2.to_string(true);
   REQUIRE(bytes.size() == sketch2.get_serialized_size_bytes());
   REQUIRE(sketch2.is_empty() == sketch.is_empty());
   REQUIRE(sketch2.is_estimation_mode() == sketch.is_estimation_mode());
@@ -282,7 +282,7 @@ TEST_CASE("req sketch: byte serialize-deserialize exact mode", "[req_sketch]") {
   auto bytes = sketch.serialize();
   REQUIRE(bytes.size() == sketch.get_serialized_size_bytes());
   auto sketch2 = req_sketch<float>::deserialize(bytes.data(), bytes.size());
-  std::cout << sketch2.to_string(true);
+  //std::cout << sketch2.to_string(true);
   REQUIRE(bytes.size() == sketch2.get_serialized_size_bytes());
   REQUIRE(sketch2.is_empty() == sketch.is_empty());
   REQUIRE(sketch2.is_estimation_mode() == sketch.is_estimation_mode());
@@ -485,6 +485,72 @@ TEST_CASE("req sketch: merge incompatible HRA and LRA", "[req_sketch]") {
   REQUIRE_THROWS_AS(sketch1.merge(sketch2), std::invalid_argument);
 }
 
+TEST_CASE("req sketch: type conversion - empty", "[req_sketch]") {
+  req_sketch<double> req_double(12);
+  req_sketch<float> req_float(req_double);
+  REQUIRE(req_float.is_empty());
+  REQUIRE(req_float.get_k() == req_double.get_k());
+  REQUIRE(req_float.get_n() == 0);
+  REQUIRE(req_float.get_num_retained() == 0);
+}
+
+TEST_CASE("req sketch: type conversion - several levels", "[req_sketch]") {
+  req_sketch<double> req_double(12);
+  for (int i = 0; i < 1000; ++i) req_double.update(static_cast<double>(i));
+  req_sketch<float> req_float(req_double);
+  REQUIRE(!req_float.is_empty());
+  REQUIRE(req_float.get_k() == req_double.get_k());
+  REQUIRE(req_float.get_n() == req_double.get_n());
+  REQUIRE(req_float.get_num_retained() == req_double.get_num_retained());
+
+  auto sv_float = req_float.get_sorted_view(false);
+  auto sv_double = req_double.get_sorted_view(false);
+  auto sv_float_it = sv_float.begin();
+  auto sv_double_it = sv_double.begin();
+  while (sv_float_it != sv_float.end()) {
+    REQUIRE(sv_double_it != sv_double.end());
+    auto float_pair = *sv_float_it;
+    auto double_pair = *sv_double_it;
+    REQUIRE(float_pair.first == Approx(double_pair.first).margin(0.01));
+    REQUIRE(float_pair.second == double_pair.second);
+    ++sv_float_it;
+    ++sv_double_it;
+  }
+  REQUIRE(sv_double_it == sv_double.end());
+}
+
+class A {
+    int val;
+  public:
+    A(int val): val(val) {}
+    int get_val() const { return val; }
+  };
+
+  struct less_A {
+    bool operator()(const A& a1, const A& a2) const { return a1.get_val() < a2.get_val(); }
+  };
+
+  class B {
+    int val;
+  public:
+    explicit B(const A& a): val(a.get_val()) {}
+    int get_val() const { return val; }
+  };
+
+  struct less_B {
+    bool operator()(const B& b1, const B& b2) const { return b1.get_val() < b2.get_val(); }
+  };
+
+TEST_CASE("req sketch: type conversion - custom types") {
+  req_sketch<A, less_A> sa(4);
+  sa.update(1);
+  sa.update(2);
+  sa.update(3);
+
+  req_sketch<B, less_B> sb(sa);
+  REQUIRE(sb.get_n() == 3);
+}
+
 //TEST_CASE("for manual comparison with Java") {
 //  req_sketch<float> sketch(12, false);
 //  for (size_t i = 0; i < 100000; ++i) sketch.update(i);

data/vendor/datasketches-cpp/sampling/test/var_opt_allocation_test.cpp

@@ -22,7 +22,7 @@
 #include <test_type.hpp>
 #include <test_allocator.hpp>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <sstream>
 

data/vendor/datasketches-cpp/sampling/test/var_opt_sketch_test.cpp

@@ -19,7 +19,7 @@
 
 #include <var_opt_sketch.hpp>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <vector>
 #include <string>

data/vendor/datasketches-cpp/sampling/test/var_opt_union_test.cpp

@@ -19,7 +19,7 @@
 
 #include <var_opt_union.hpp>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <vector>
 #include <string>

data/vendor/datasketches-cpp/setup.py

@@ -20,7 +20,6 @@
 
 import os
 import sys
-import sysconfig
 import platform
 import subprocess
 
@@ -81,7 +80,7 @@ class CMakeBuild(build_ext):
 
 setup(
     name='datasketches',
-    version='3.4.0',
+    version='3.5.1',
     author='Apache Software Foundation',
     author_email='dev@datasketches.apache.org',
     description='The Apache DataSketches Library for Python',
@@ -89,7 +88,7 @@ setup(
     url='http://datasketches.apache.org',
     long_description=open('python/README.md').read(),
     long_description_content_type='text/markdown',
-    packages=find_packages('python'), # python pacakges only in this dir
+    packages=find_packages(where='python',exclude=['src','*tests*']), # src not needed if only the.so
     package_dir={'':'python'},
     # may need to add all source paths for sdist packages w/o MANIFEST.in
     ext_modules=[CMakeExtension('datasketches')],

data/vendor/datasketches-cpp/theta/test/theta_a_not_b_test.cpp

@@ -17,7 +17,7 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <theta_a_not_b.hpp>
 

data/vendor/datasketches-cpp/theta/test/theta_intersection_test.cpp

@@ -17,7 +17,7 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <theta_intersection.hpp>
 

data/vendor/datasketches-cpp/theta/test/theta_jaccard_similarity_test.cpp

@@ -19,7 +19,7 @@
 
 #include <iostream>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include "theta_jaccard_similarity.hpp"
 

data/vendor/datasketches-cpp/theta/test/theta_setop_test.cpp

@@ -20,7 +20,7 @@
 #include <sstream>
 #include <stdexcept>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <theta_union.hpp>
 #include <theta_intersection.hpp>
 #include <theta_a_not_b.hpp>

data/vendor/datasketches-cpp/theta/test/theta_sketch_test.cpp

@@ -23,7 +23,7 @@
 #include <vector>
 #include <stdexcept>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <theta_sketch.hpp>
 
 namespace datasketches {

data/vendor/datasketches-cpp/theta/test/theta_union_test.cpp

@@ -17,7 +17,7 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <theta_union.hpp>
 

data/vendor/datasketches-cpp/tuple/include/tuple_sketch_impl.hpp

@@ -353,8 +353,7 @@ size_t compact_tuple_sketch<S, A>::get_serialized_size_summaries_bytes(const SD&
 template<typename S, typename A>
 template<typename SerDe>
 void compact_tuple_sketch<S, A>::serialize(std::ostream& os, const SerDe& sd) const {
-  const bool is_single_item = entries_.size() == 1 && !this->is_estimation_mode();
-  const uint8_t preamble_longs = this->is_empty() || is_single_item ? 1 : this->is_estimation_mode() ? 3 : 2;
+  const uint8_t preamble_longs = this->is_estimation_mode() ? 3 : this->is_empty() || entries_.size() == 1 ? 1 : 2;
   write(os, preamble_longs);
   const uint8_t serial_version = SERIAL_VERSION;
   write(os, serial_version);
@@ -373,28 +372,25 @@ void compact_tuple_sketch<S, A>::serialize(std::ostream& os, const SerDe& sd) co
   write(os, flags_byte);
   const uint16_t seed_hash = get_seed_hash();
   write(os, seed_hash);
-  if (!this->is_empty()) {
-    if (!is_single_item) {
-      const uint32_t num_entries = static_cast<uint32_t>(entries_.size());
-      write(os, num_entries);
-      const uint32_t unused32 = 0;
-      write(os, unused32);
-      if (this->is_estimation_mode()) {
-        write(os, this->theta_);
-      }
-    }
-    for (const auto& it: entries_) {
-      write(os, it.first);
-      sd.serialize(os, &it.second, 1);
-    }
+  if (preamble_longs > 1) {
+    const uint32_t num_entries = static_cast<uint32_t>(entries_.size());
+    write(os, num_entries);
+    const uint32_t unused32 = 0;
+    write(os, unused32);
+  }
+  if (this->is_estimation_mode()) {
+    write(os, this->theta_);
+  }
+  for (const auto& it: entries_) {
+    write(os, it.first);
+    sd.serialize(os, &it.second, 1);
   }
 }
 
 template<typename S, typename A>
 template<typename SerDe>
 auto compact_tuple_sketch<S, A>::serialize(unsigned header_size_bytes, const SerDe& sd) const -> vector_bytes {
-  const bool is_single_item = entries_.size() == 1 && !this->is_estimation_mode();
-  const uint8_t preamble_longs = this->is_empty() || is_single_item ? 1 : this->is_estimation_mode() ? 3 : 2;
+  const uint8_t preamble_longs = this->is_estimation_mode() ? 3 : this->is_empty() || entries_.size() == 1 ? 1 : 2;
   const size_t size = header_size_bytes + sizeof(uint64_t) * preamble_longs
       + sizeof(uint64_t) * entries_.size() + get_serialized_size_summaries_bytes(sd);
   vector_bytes bytes(size, 0, entries_.get_allocator());
@@ -418,19 +414,17 @@ auto compact_tuple_sketch<S, A>::serialize(unsigned header_size_bytes, const Ser
   ptr += copy_to_mem(flags_byte, ptr);
   const uint16_t seed_hash = get_seed_hash();
   ptr += copy_to_mem(seed_hash, ptr);
-  if (!this->is_empty()) {
-    if (!is_single_item) {
-      const uint32_t num_entries = static_cast<uint32_t>(entries_.size());
-      ptr += copy_to_mem(num_entries, ptr);
-      ptr += sizeof(uint32_t); // unused
-      if (this->is_estimation_mode()) {
-        ptr += copy_to_mem(theta_, ptr);
-      }
-    }
-    for (const auto& it: entries_) {
-      ptr += copy_to_mem(it.first, ptr);
-      ptr += sd.serialize(ptr, end_ptr - ptr, &it.second, 1);
-    }
+  if (preamble_longs > 1) {
+    const uint32_t num_entries = static_cast<uint32_t>(entries_.size());
+    ptr += copy_to_mem(num_entries, ptr);
+    ptr += sizeof(uint32_t); // unused
+  }
+  if (this->is_estimation_mode()) {
+    ptr += copy_to_mem(theta_, ptr);
+  }
+  for (const auto& it: entries_) {
+    ptr += copy_to_mem(it.first, ptr);
+    ptr += sd.serialize(ptr, end_ptr - ptr, &it.second, 1);
   }
   return bytes;
 }

data/vendor/datasketches-cpp/tuple/test/array_of_doubles_sketch_test.cpp

@@ -22,7 +22,7 @@
 #include <sstream>
 #include <array>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <array_of_doubles_sketch.hpp>
 #include <array_of_doubles_union.hpp>
 #include <array_of_doubles_intersection.hpp>

data/vendor/datasketches-cpp/tuple/test/tuple_a_not_b_test.cpp

@@ -19,7 +19,7 @@
 
 #include <iostream>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <tuple_a_not_b.hpp>
 #include <theta_sketch.hpp>
 #include <stdexcept>

data/vendor/datasketches-cpp/tuple/test/tuple_intersection_test.cpp

@@ -19,7 +19,7 @@
 
 #include <iostream>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <tuple_intersection.hpp>
 #include <theta_sketch.hpp>
 #include <stdexcept>

data/vendor/datasketches-cpp/tuple/test/tuple_jaccard_similarity_test.cpp

@@ -19,7 +19,7 @@
 
 #include <iostream>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include "tuple_jaccard_similarity.hpp"
 

data/vendor/datasketches-cpp/tuple/test/tuple_sketch_allocation_test.cpp

@@ -19,7 +19,7 @@
 
 #include <iostream>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <tuple_sketch.hpp>
 #include <test_allocator.hpp>
 #include <test_type.hpp>

data/vendor/datasketches-cpp/tuple/test/tuple_sketch_test.cpp

@@ -32,7 +32,7 @@ std::ostream& operator<<(std::ostream& os, const three_doubles& tuple) {
 
 }
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <tuple_sketch.hpp>
 
 namespace datasketches {

data/vendor/datasketches-cpp/tuple/test/tuple_union_test.cpp

@@ -20,7 +20,7 @@
 #include <iostream>
 #include <stdexcept>
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <tuple_union.hpp>
 #include <theta_sketch.hpp>
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: datasketches
 version: !ruby/object:Gem::Version
-  version: 0.2.5
+  version: 0.2.7
 platform: ruby
 authors:
 - Andrew Kane
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-05-21 00:00:00.000000000 Z
+date: 2022-11-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rice
@@ -69,12 +69,10 @@ files:
 - vendor/datasketches-cpp/common/include/quantile_sketch_sorted_view_impl.hpp
 - vendor/datasketches-cpp/common/include/serde.hpp
 - vendor/datasketches-cpp/common/test/CMakeLists.txt
-- vendor/datasketches-cpp/common/test/catch.hpp
 - vendor/datasketches-cpp/common/test/catch_runner.cpp
 - vendor/datasketches-cpp/common/test/integration_test.cpp
 - vendor/datasketches-cpp/common/test/test_allocator.cpp
 - vendor/datasketches-cpp/common/test/test_allocator.hpp
-- vendor/datasketches-cpp/common/test/test_runner.cpp
 - vendor/datasketches-cpp/common/test/test_type.hpp
 - vendor/datasketches-cpp/cpc/CMakeLists.txt
 - vendor/datasketches-cpp/cpc/include/compression_data.hpp