datasketches 0.2.6 → 0.3.0

data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_custom_type_test.cpp

@@ -17,7 +17,7 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <sstream>
 #include <stdexcept>
 
@@ -27,62 +27,75 @@
 
 namespace datasketches {
 
-using frequent_test_type_sketch = frequent_items_sketch<test_type, float, test_type_hash, test_type_equal, test_type_serde, test_allocator<test_type>>;
+using frequent_test_type_sketch = frequent_items_sketch<test_type, float, test_type_hash, test_type_equal, test_allocator<test_type>>;
 using alloc = test_allocator<test_type>;
 
 TEST_CASE("frequent items: custom type", "[frequent_items_sketch]") {
-  frequent_test_type_sketch sketch(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, 0);
-  sketch.update(1, 10); // should survive the purge
-  sketch.update(2);
-  sketch.update(3);
-  sketch.update(4);
-  sketch.update(5);
-  sketch.update(6);
-  sketch.update(7);
-  test_type a8(8);
-  sketch.update(a8);
-  REQUIRE_FALSE(sketch.is_empty());
-  REQUIRE(sketch.get_total_weight() == 17);
-  REQUIRE(sketch.get_estimate(1) == 10);
+  test_allocator_total_bytes = 0;
+  {
+    frequent_test_type_sketch sketch(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, test_type_equal(), 0);
+    sketch.update(1, 10); // should survive the purge
+    sketch.update(2);
+    sketch.update(3);
+    sketch.update(4);
+    sketch.update(5);
+    sketch.update(6);
+    sketch.update(7);
+    test_type a8(8);
+    sketch.update(a8);
+    REQUIRE_FALSE(sketch.is_empty());
+    REQUIRE(sketch.get_total_weight() == 17);
+    REQUIRE(sketch.get_estimate(1) == 10);
 
-  auto items = sketch.get_frequent_items(frequent_items_error_type::NO_FALSE_POSITIVES);
-  REQUIRE(items.size() == 1); // only 1 item should be above threshold
-  REQUIRE(items[0].get_item().get_value() == 1);
-  REQUIRE(items[0].get_estimate() == 10);
+    auto items = sketch.get_frequent_items(frequent_items_error_type::NO_FALSE_POSITIVES);
+    REQUIRE(items.size() == 1); // only 1 item should be above threshold
+    REQUIRE(items[0].get_item().get_value() == 1);
+    REQUIRE(items[0].get_estimate() == 10);
 
-  std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
-  sketch.serialize(s);
-  auto sketch2 = frequent_test_type_sketch::deserialize(s, alloc(0));
-  REQUIRE_FALSE(sketch2.is_empty());
-  REQUIRE(sketch2.get_total_weight() == 17);
-  REQUIRE(sketch2.get_estimate(1) == 10);
-  REQUIRE(sketch.get_num_active_items() == sketch2.get_num_active_items());
-  REQUIRE(sketch.get_maximum_error() == sketch2.get_maximum_error());
+    std::stringstream s(std::ios::in | std::ios::out | std::ios::binary);
+    sketch.serialize(s, test_type_serde());
+    auto sketch2 = frequent_test_type_sketch::deserialize(s, test_type_serde(), test_type_equal(), 0);
+    REQUIRE_FALSE(sketch2.is_empty());
+    REQUIRE(sketch2.get_total_weight() == 17);
+    REQUIRE(sketch2.get_estimate(1) == 10);
+    REQUIRE(sketch.get_num_active_items() == sketch2.get_num_active_items());
+    REQUIRE(sketch.get_maximum_error() == sketch2.get_maximum_error());
 
-  auto bytes = sketch.serialize();
-  auto sketch3 = frequent_test_type_sketch::deserialize(bytes.data(), bytes.size(), alloc(0));
-  REQUIRE_FALSE(sketch3.is_empty());
-  REQUIRE(sketch3.get_total_weight() == 17);
-  REQUIRE(sketch3.get_estimate(1) == 10);
-  REQUIRE(sketch.get_num_active_items() == sketch3.get_num_active_items());
-  REQUIRE(sketch.get_maximum_error() == sketch3.get_maximum_error());
+    auto bytes = sketch.serialize(0, test_type_serde());
+    auto sketch3 = frequent_test_type_sketch::deserialize(bytes.data(), bytes.size(), test_type_serde(),
+        test_type_equal(), 0);
+    REQUIRE_FALSE(sketch3.is_empty());
+    REQUIRE(sketch3.get_total_weight() == 17);
+    REQUIRE(sketch3.get_estimate(1) == 10);
+    REQUIRE(sketch.get_num_active_items() == sketch3.get_num_active_items());
+    REQUIRE(sketch.get_maximum_error() == sketch3.get_maximum_error());
+  }
+  REQUIRE(test_allocator_total_bytes == 0);
 }
 
 // this is to see the debug print from test_type if enabled there to make sure items are moved
 TEST_CASE("frequent items: moving merge", "[frequent_items_sketch]") {
-  frequent_test_type_sketch sketch1(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, 0);
-  sketch1.update(1);
+  test_allocator_total_bytes = 0;
+  {
+    frequent_test_type_sketch sketch1(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, test_type_equal(), 0);
+    sketch1.update(1);
 
-  frequent_test_type_sketch sketch2(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, 0);
-  sketch2.update(2);
+    frequent_test_type_sketch sketch2(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, test_type_equal(), 0);
+    sketch2.update(2);
 
-  sketch2.merge(std::move(sketch1));
-  REQUIRE(sketch2.get_total_weight() == 2);
+    sketch2.merge(std::move(sketch1));
+    REQUIRE(sketch2.get_total_weight() == 2);
+  }
+  REQUIRE(test_allocator_total_bytes == 0);
 }
 
 TEST_CASE("frequent items: negative weight", "[frequent_items_sketch]") {
-  frequent_test_type_sketch sketch(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, 0);
-  REQUIRE_THROWS_AS(sketch.update(1, -1), std::invalid_argument);
+  test_allocator_total_bytes = 0;
+  {
+    frequent_test_type_sketch sketch(3, frequent_test_type_sketch::LG_MIN_MAP_SIZE, test_type_equal(), 0);
+    REQUIRE_THROWS_AS(sketch.update(1, -1), std::invalid_argument);
+  }
+  REQUIRE(test_allocator_total_bytes == 0);
 }
 
 } /* namespace datasketches */

data/vendor/datasketches-cpp/fi/test/frequent_items_sketch_test.cpp

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

data/vendor/datasketches-cpp/fi/test/reverse_purge_hash_map_test.cpp

@@ -17,27 +17,27 @@
  * under the License.
  */
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 #include <reverse_purge_hash_map.hpp>
 
 namespace datasketches {
 
 TEST_CASE("reverse purge hash map: empty", "[frequent_items_sketch]") {
-  reverse_purge_hash_map<int> map(3, 3, std::allocator<int>());
+  reverse_purge_hash_map<int> map(3, 3, std::equal_to<int>(), std::allocator<int>());
   REQUIRE(map.get_num_active() == 0);
   REQUIRE(map.get_lg_cur_size() == 3);  // static_cast<uint8_t>(3)
 }
 
 TEST_CASE("reverse purge hash map: one item", "[frequent_items_sketch]") {
-  reverse_purge_hash_map<int> map(3, 3, std::allocator<int>());
+  reverse_purge_hash_map<int> map(3, 3, std::equal_to<int>(), std::allocator<int>());
   map.adjust_or_insert(1, 1);
   REQUIRE(map.get_num_active() == 1);
   REQUIRE(map.get(1) == 1);
 }
 
 TEST_CASE("reverse purge hash map: iterator", "[frequent_items_sketch]") {
-  reverse_purge_hash_map<int> map(3, 4, std::allocator<int>());
+  reverse_purge_hash_map<int> map(3, 4, std::equal_to<int>(), std::allocator<int>());
   for (int i = 0; i < 11; i++) map.adjust_or_insert(i, 1); // this should fit with no purge
   uint64_t sum = 0;
   for (auto it: map) sum += it.second;

data/vendor/datasketches-cpp/hll/include/Hll4Array-internal.hpp

@@ -267,10 +267,10 @@ void Hll4Array<A>::shiftToBiggerCurMin() {
     for (const auto coupon: *auxHashMap_) {
       slotNum = HllUtil<A>::getLow26(coupon) & configKmask;
       oldActualVal = HllUtil<A>::getValue(coupon);
-      newShiftedVal = oldActualVal - newCurMin;
-      if (newShiftedVal < 0) {
+      if (oldActualVal < newCurMin) {
         throw std::logic_error("oldActualVal < newCurMin when incrementing curMin");
       }
+      newShiftedVal = oldActualVal - newCurMin;
 
       if (getSlot(slotNum) != hll_constants::AUX_TOKEN) {
         throw std::logic_error("getSlot(slotNum) != AUX_TOKEN for item in auxiliary hash map");

data/vendor/datasketches-cpp/hll/test/AuxHashMapTest.cpp

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

data/vendor/datasketches-cpp/hll/test/CMakeLists.txt

@@ -17,7 +17,7 @@
 
 add_executable(hll_test)
 
-target_link_libraries(hll_test hll common_test)
+target_link_libraries(hll_test hll common_test_lib)
 
 set_target_properties(hll_test PROPERTIES
   CXX_STANDARD 11

data/vendor/datasketches-cpp/hll/test/CouponHashSetTest.cpp

@@ -21,7 +21,7 @@
 #include "CouponHashSet.hpp"
 #include "HllUtil.hpp"
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <ostream>
 #include <cmath>
 #include <string>

data/vendor/datasketches-cpp/hll/test/CouponListTest.cpp

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

data/vendor/datasketches-cpp/hll/test/CrossCountingTest.cpp

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

data/vendor/datasketches-cpp/hll/test/HllArrayTest.cpp

@@ -22,7 +22,7 @@
 #include <exception>
 #include <sstream>
 #include <stdexcept>
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 
 namespace datasketches {
 

data/vendor/datasketches-cpp/hll/test/HllSketchTest.cpp

@@ -21,7 +21,7 @@
 
 #include "hll.hpp"
 
-#include <catch.hpp>
+#include <catch2/catch.hpp>
 #include <test_allocator.hpp>
 
 namespace datasketches {

data/vendor/datasketches-cpp/hll/test/HllUnionTest.cpp

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

data/vendor/datasketches-cpp/hll/test/IsomorphicTest.cpp

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

data/vendor/datasketches-cpp/hll/test/TablesTest.cpp

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

data/vendor/datasketches-cpp/hll/test/ToFromByteArrayTest.cpp

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

data/vendor/datasketches-cpp/kll/include/kll_helper.hpp

@@ -48,38 +48,6 @@ class kll_helper {
     static inline uint16_t int_cap_aux_aux(uint16_t k, uint8_t depth);
     static inline uint64_t sum_the_sample_weights(uint8_t num_levels, const uint32_t* levels);
 
-    /*
-     * This version is for floating point types
-     * Checks the sequential validity of the given array of values.
-     * They must be unique, monotonically increasing and not NaN.
-     */
-    template <typename T, typename C>
-    static typename std::enable_if<std::is_floating_point<T>::value, void>::type
-    validate_values(const T* values, uint32_t size) {
-      for (uint32_t i = 0; i < size ; i++) {
-        if (std::isnan(values[i])) {
-          throw std::invalid_argument("Values must not be NaN");
-        }
-        if ((i < (size - 1)) && !(C()(values[i], values[i + 1]))) {
-          throw std::invalid_argument("Values must be unique and monotonically increasing");
-        }
-      }
-    }
-    /*
-     * This version is for non-floating point types
-     * Checks the sequential validity of the given array of values.
-     * They must be unique and monotonically increasing.
-     */
-    template <typename T, typename C>
-    static typename std::enable_if<!std::is_floating_point<T>::value, void>::type
-    validate_values(const T* values, uint32_t size) {
-      for (uint32_t i = 0; i < size ; i++) {
-        if ((i < (size - 1)) && !(C()(values[i], values[i + 1]))) {
-          throw std::invalid_argument("Values must be unique and monotonically increasing");
-        }
-      }
-    }
-
     template <typename T>
     static void randomly_halve_down(T* buf, uint32_t start, uint32_t length);