Bloom filter code review changes

include/data_structures/bloom/bloom_filter.hpp

@@ -2,8 +2,17 @@
 #include <iostream>
 #include <vector>
 
+/*
+  Implementation of a generic Bloom Filter.
+
+  Read more about bloom filters here:
+    http://en.wikipedia.org/wiki/Bloom_filter
+    http://www.jasondavies.com/bloomfilter/
+*/
+
+// Type specifies the type of data stored
 template <class Type, int BucketSize = 8>
-class BasicBloomFilter {
+class BloomFilter {
  private:
   using HashFunction = std::function<uint64_t(const Type&)>;
   using CompresionFunction = std::function<int(uint64_t)>;
@@ -28,13 +37,13 @@ class BasicBloomFilter {
   }
 
  public:
-  BasicBloomFilter(std::vector<HashFunction> funcs,
-                   CompresionFunction compression = {})
+  BloomFilter(std::vector<HashFunction> funcs,
+              CompresionFunction compression = {})
       : hashes_(funcs) {
     if (!compression)
-      compression_ = std::bind(&BasicBloomFilter::default_compression, this,
+      compression_ = std::bind(&BloomFilter::default_compression, this,
                                std::placeholders::_1);
-    else 
+    else
       compression_ = compression;
 
     buckets.resize(hashes_.size());

tests/benchmark/data_structures/bloom/basic_bloom_filter.cpp

@@ -1,7 +1,7 @@
 #include <random>
 #include <thread>
 
-#include "data_structures/bloom/basic_bloom_filter.hpp"
+#include "data_structures/bloom/bloom_filter.hpp"
 #include "logging/default.hpp"
 #include "logging/streams/stdout.hpp"
 #include "utils/command_line/arguments.hpp"
@@ -14,7 +14,7 @@ using utils::random::StringGenerator;
 using StringHashFunction = std::function<uint64_t(const std::string&)>;
 
 template <class Type, int Size>
-static void TestBloom(benchmark::State& state, BasicBloomFilter<Type, Size>*
+static void TestBloom(benchmark::State& state, BloomFilter<Type, Size>*
 bloom, const std::vector<Type>& elements) {
   while(state.KeepRunning()) {
     for (int start = 0; start < state.range(0); start++)
@@ -46,7 +46,7 @@ int main(int argc, char** argv) {
     hash1, hash2
   };
 
-  BasicBloomFilter<std::string, 128> bloom(funcs);
+  BloomFilter<std::string, 128> bloom(funcs);
 
   benchmark::RegisterBenchmark("SimpleBloomFilter Benchmark Test", BM_Bloom,
                                &bloom, elements)

tests/benchmark/data_structures/concurrent/concurrent_bloom_map.cpp

@@ -1,7 +1,7 @@
 #include <random>
 #include <thread>
 
-#include "data_structures/bloom/basic_bloom_filter.hpp"
+#include "data_structures/bloom/bloom_filter.hpp"
 #include "data_structures/concurrent/concurrent_bloom_map.hpp"
 #include "logging/default.hpp"
 #include "logging/streams/stdout.hpp"
@@ -50,21 +50,6 @@ static void InsertValue(benchmark::State& state, ConcurrentBloomMap<K, V, F>* ma
   state.SetComplexityN(state.range(0));
 }
 
-/*
-  ConcurrentMap Deletion Benchmark Test
-template <class K, class V>
-static void DeleteValue(benchmark::State& state, ConcurrentBloomMap<K, V, F>* map,
-                        const std::vector<std::pair<K, V>> elements) {
-  while (state.KeepRunning()) {
-    auto accessor = map->access();
-    for (int start = 0; start < state.range(0); start++) {
-      accessor.remove(elements[start].first);
-    }
-  }
-  state.SetComplexityN(state.range(0));
-}
-*/
-
 /*
   ConcurrentMap Contains Benchmark Test
 */
@@ -83,12 +68,6 @@ auto BM_InsertValue = [](benchmark::State& state, auto* map, auto& elements) {
   InsertValue(state, map, elements);
 };
 
-/*
-auto BM_DeleteValue = [](benchmark::State& state, auto* map, auto elements) {
-  DeleteValue(state, map, elements);
-};
-*/
-
 auto BM_ContainsValue = [](benchmark::State& state, auto* map, auto elements) {
   ContainsValue(state, map, elements);
 };
@@ -154,12 +133,12 @@ int main(int argc, char** argv) {
     hash1, hash2
   };
 
-  BasicBloomFilter<std::string, 128> bloom_filter_(funcs);
+  BloomFilter<std::string, 128> bloom_filter_(funcs);
 
   // maps used for testing
   //ConcurrentBloomMap<int, int> ii_map;
   //ConcurrentBloomMap<int, std::string> is_map;
-  using Filter = BasicBloomFilter<std::string, 128>;
+  using Filter = BloomFilter<std::string, 128>;
   ConcurrentBloomMap<std::string, int, Filter > si_map(bloom_filter_);
   ConcurrentBloomMap<std::string, std::string, Filter>
 ss_map(bloom_filter_);
@@ -171,21 +150,6 @@ ss_map(bloom_filter_);
   auto ss_elems = utils::random::generate_vector(pssg, MAX_ELEMENTS);
 
   /* insertion Tests */
-  /*
-  benchmark::RegisterBenchmark("InsertValue[Int, Int]", BM_InsertValue, &ii_map,
-                               ii_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-
-  benchmark::RegisterBenchmark("InsertValue[Int, String]", BM_InsertValue,
-                               &is_map, is_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-  */
   benchmark::RegisterBenchmark("InsertValue[String, Int]", BM_InsertValue,
                                &si_map, si_elems)
       ->RangeMultiplier(MULTIPLIER)
@@ -201,22 +165,6 @@ ss_map(bloom_filter_);
       ->Threads(THREADS);
 
   // Contains Benchmark Tests
-
-  /*
-  benchmark::RegisterBenchmark("ContainsValue[Int, Int]", BM_ContainsValue,
-                               &ii_map, ii_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-
-  benchmark::RegisterBenchmark("ContainsValue[Int, String]", BM_ContainsValue,
-                               &is_map, is_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-  */
   benchmark::RegisterBenchmark("ContainsValue[String, Int]", BM_ContainsValue,
                                &si_map, si_elems)
       ->RangeMultiplier(MULTIPLIER)
@@ -231,38 +179,6 @@ ss_map(bloom_filter_);
       ->Complexity(benchmark::oN)
       ->Threads(THREADS);
 
-  // Deletion Banchamark Tests
-  /*
-
-  benchmark::RegisterBenchmark("DeleteValue[Int, Int]", BM_DeleteValue, &ii_map,
-                               ii_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-
-  benchmark::RegisterBenchmark("DeleteValue[Int, String]", BM_DeleteValue,
-                               &is_map, is_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-
-  benchmark::RegisterBenchmark("DeleteValue[String, Int]", BM_DeleteValue,
-                               &si_map, si_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-
-  benchmark::RegisterBenchmark("DeleteValue[String, String]", BM_DeleteValue,
-                               &ss_map, ss_elems)
-      ->RangeMultiplier(MULTIPLIER)
-      ->Range(1, MAX_ELEMENTS)
-      ->Complexity(benchmark::oN)
-      ->Threads(THREADS);
-  */
-
   benchmark::Initialize(&argc, argv);
   benchmark::RunSpecifiedBenchmarks();
 

tests/unit/basic_bloom_filter.cpp

@@ -4,14 +4,14 @@
 #include "utils/command_line/arguments.hpp"
 #include "utils/hashing/fnv64.hpp"
 
-#include "data_structures/bloom/basic_bloom_filter.hpp"
+#include "data_structures/bloom/bloom_filter.hpp"
 
 #pragma clang diagnostic push
 #pragma clang diagnostic ignored "-Wwritable-strings"
 
 using StringHashFunction = std::function<uint64_t(const std::string&)>;
  
-TEST_CASE("BasicBloomFilter Test") {
+TEST_CASE("BloomFilter Test") {
   StringHashFunction hash1 = fnv64<std::string>;
   StringHashFunction hash2 = fnv1a64<std::string>;
 
@@ -22,10 +22,10 @@ TEST_CASE("BasicBloomFilter Test") {
     hash1, hash2
   };
 
-  BasicBloomFilter<std::string, 64> bloom(funcs);
+  BloomFilter<std::string, 64> bloom(funcs);
 
   std::string test = "test";
-  std::string kifla = "pizda";
+  std::string kifla = "kifla";
 
   std::cout << hash1(test) << std::endl;
   std::cout << hash2(test) << std::endl;