Merge pull request #90 from eseiler/misc/membership_for

[MISC] membership_for

include/hibf/hierarchical_interleaved_bloom_filter.hpp

@@ -198,7 +198,7 @@ class hierarchical_interleaved_bloom_filter::membership_agent_type
 
     //!\brief Helper for recursive membership querying.
     template <std::ranges::forward_range value_range_t>
-    void membership_for_impl(value_range_t && values, int64_t const ibf_idx, size_t const threshold)
+    void membership_for_impl(value_range_t && values, int64_t const ibf_idx, uint16_t const threshold)
     {
         auto agent = hibf_ptr->ibf_vector[ibf_idx].template counting_agent<uint16_t>();
         auto & result = agent.bulk_count(values);
@@ -229,8 +229,8 @@ class hierarchical_interleaved_bloom_filter::membership_agent_type
 
 public:
     /*!\name Constructors, destructor and assignment
-        * \{
-        */
+     * \{
+     */
     membership_agent_type() = default;                                         //!< Defaulted.
     membership_agent_type(membership_agent_type const &) = default;            //!< Defaulted.
     membership_agent_type & operator=(membership_agent_type const &) = delete; //!< Deleted. hibf_ptr is const.
@@ -249,6 +249,12 @@ class hierarchical_interleaved_bloom_filter::membership_agent_type
     //!\brief Stores the result of membership_for().
     std::vector<int64_t> result_buffer;
 
+    //!\brief Sorts the results.
+    void sort_results()
+    {
+        std::ranges::sort(result_buffer);
+    }
+
     /*!\name Lookup
      * \{
      */
@@ -290,7 +296,7 @@ class hierarchical_interleaved_bloom_filter::membership_agent_type
      */
     template <std::ranges::forward_range value_range_t>
     [[nodiscard]] std::vector<int64_t> const & membership_for(value_range_t && values,
-                                                              size_t const threshold) & noexcept
+                                                              uint16_t const threshold) & noexcept
     {
         assert(hibf_ptr != nullptr);
 
@@ -302,16 +308,14 @@ class hierarchical_interleaved_bloom_filter::membership_agent_type
 
         membership_for_impl(values, 0, threshold);
 
-        std::ranges::sort(result_buffer); // TODO: necessary?
-
         return result_buffer;
     }
 
     // `membership_for` cannot be called on a temporary, since the object the returned reference points to
     // is immediately destroyed.
     template <std::ranges::range value_range_t>
     [[nodiscard]] std::vector<int64_t> const & membership_for(value_range_t && values,
-                                                              size_t const threshold) && noexcept = delete;
+                                                              uint16_t const threshold) && noexcept = delete;
     //!\}
 };
 

test/snippet/readme.cpp

@@ -72,15 +72,16 @@ int main()
 
     // query1 hits in user_bin_1 and user_bin_3, which have the IDs 0 and 2, respectively.
     for (int64_t hit_user_bin : result1)
-        std::cout << hit_user_bin << ' '; // 0 2
+        std::cout << hit_user_bin << ' '; // The results are not sorted: 2 0
     std::cout << '\n';
 
     // Another query. A query is simply a range of unsigned integer values, e.g., it does not have to be a vector.
     auto query2 = std::views::iota(0u, 15u); // 0,1,2,...,14
     auto & result2 = agent.membership_for(query2, 5u);
+    agent.sort_results(); // Sort the results.
 
     // query2 hits in user_bin_1 and user_bin_2, which have the IDs 0 and 1, respectively.
     for (int64_t hit_user_bin : result2)
-        std::cout << hit_user_bin << ' '; // 0 1
+        std::cout << hit_user_bin << ' '; // The results are sorted: 0 1
     std::cout << '\n';
 }

test/snippet/readme.out

@@ -1,2 +1,2 @@
-0 2 
+2 0 
 0 1 

test/unit/hibf/hierarchical_interleaved_bloom_filter_test.cpp

@@ -38,6 +38,7 @@ TEST(hibf_test, small_example_with_direct_hashes)
     std::vector<size_t> query{1u, 2u, 3u, 4u, 5u};
 
     auto & result = agent.membership_for(query, 2u);
+    agent.sort_results();
     EXPECT_RANGE_EQ(result, (std::vector<size_t>{0u, 1u}));
 }
 
@@ -87,6 +88,7 @@ TEST(hibf_test, build_from_layout)
     std::vector<size_t> query{1u, 2u, 3u, 4u, 5u};
 
     auto & result = agent.membership_for(query, 2u);
+    agent.sort_results();
     EXPECT_RANGE_EQ(result, (std::vector<size_t>{0u, 1u}));
 }
 
@@ -122,6 +124,7 @@ TEST(hibf_test, three_level_hibf)
         auto unique_query = std::views::iota(start + 5u, start + 9u);
 
         auto & overlap_result = agent.membership_for(overlap_query, 4u); // one overlapping user bin
+        agent.sort_results();
         EXPECT_EQ(overlap_result.size(), 2u);
         EXPECT_EQ(overlap_result[0], ub_id);
         EXPECT_EQ(overlap_result[1], ub_id + 1u);
@@ -202,6 +205,7 @@ TEST(hibf_test, evenly_sized_and_highly_similar_user_bins)
         auto similar_query = std::views::iota(start, start + 5u);
 
         auto & similar_result = agent.membership_for(similar_query, 5u); // t = 5 results in 6 similar user bins
+        agent.sort_results();
         EXPECT_RANGE_EQ(similar_result, (std::views::iota(ub_id - 6u, ub_id + 1u)));
     }
 }