Merge pull request #254 from eseiler/feature/reverse_ibf_map

[FEATURE] Add prev_ibf_id

include/hibf/hierarchical_interleaved_bloom_filter.hpp

@@ -149,6 +149,8 @@ class hierarchical_interleaved_bloom_filter
     //!\brief Manages counting ranges of values for the seqan::hibf::hierarchical_interleaved_bloom_filter.
     template <std::integral value_t>
     class counting_agent_type;
+    //!\brief Contains information about the parent IBF and bin index.
+    struct previous_ibf_id_pair;
 
     /*!\name Constructors, destructor and assignment
      * \{
@@ -212,6 +214,17 @@ class hierarchical_interleaved_bloom_filter
      */
     std::vector<std::vector<uint64_t>> next_ibf_id;
 
+    /*!\brief Stores for each IBF of the HIBF the ID of the parent IBF and the bin index of the parent IBF.
+     * \details
+     * Assume we look up an IBF `i`, i.e. `prev_ibf_id[i]`.
+     * The reurned value contains two members:
+     *   * `ibf_idx` is the index of the parent IBF in `ibf_vector`.
+     *   * `bin_idx` is the index of the bin in the parent IBF of which IBF `i` is the child of.
+     *
+     * The root/top-level IBF has no parent and `ibf_idx == bin_idx == 0`.
+     */
+    std::vector<previous_ibf_id_pair> prev_ibf_id;
+
     /*!\brief Stores for each bin in each IBF of the HIBF the user bin ID.
     * \details
     * Assume we look up a bin `b` in IBF `i`, i.e. `ibf_bin_to_user_bin_id[i][b]`.
@@ -233,11 +246,15 @@ class hierarchical_interleaved_bloom_filter
     /*!\name Comparison operators
      * \{
      */
-    HIBF_CONSTEXPR_VECTOR bool operator==(hierarchical_interleaved_bloom_filter const &) const = default;
+    HIBF_CONSTEXPR_VECTOR bool operator==(hierarchical_interleaved_bloom_filter const &) const;
     //!\}
 
     /*!\cond DEV
-     * \brief Serialisation support function.
+     * \brief The version of the HIBF.
+     */
+    static constexpr uint32_t version{1};
+
+    /*!\brief Serialisation support function.
      * \tparam archive_t Type of `archive`; must satisfy seqan::hibf::cereal_archive.
      * \param[in] archive The archive being serialised from/to.
      *
@@ -247,9 +264,12 @@ class hierarchical_interleaved_bloom_filter
     template <seqan::hibf::cereal_archive archive_t>
     void CEREAL_SERIALIZE_FUNCTION_NAME(archive_t & archive)
     {
+        uint32_t parsed_version{version};
+        archive(cereal::make_nvp("version", parsed_version));
         archive(number_of_user_bins);
         archive(ibf_vector);
         archive(next_ibf_id);
+        archive(prev_ibf_id);
         archive(ibf_bin_to_user_bin_id);
     }
 
@@ -269,6 +289,25 @@ class hierarchical_interleaved_bloom_filter
     //!\endcond
 };
 
+//!\brief Contains information about the parent IBF and bin index.
+struct hierarchical_interleaved_bloom_filter::previous_ibf_id_pair
+{
+    size_t ibf_idx{}; //!< The index of the parent IBF in `ibf_vector`.
+    size_t bin_idx{}; //!< The index of the bin in the parent IBF of which an IBF is the child of.
+
+    friend constexpr auto operator<=>(previous_ibf_id_pair const &, previous_ibf_id_pair const &) = default;
+
+    template <seqan::hibf::cereal_archive archive_t>
+    void CEREAL_SERIALIZE_FUNCTION_NAME(archive_t & archive)
+    {
+        archive(ibf_idx);
+        archive(bin_idx);
+    }
+};
+
+HIBF_CONSTEXPR_VECTOR bool
+hierarchical_interleaved_bloom_filter::operator==(hierarchical_interleaved_bloom_filter const &) const = default;
+
 class hierarchical_interleaved_bloom_filter::membership_agent_type
 {
 private:

include/hibf/interleaved_bloom_filter.hpp

@@ -415,7 +415,11 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
     //!\}
 
     /*!\cond DEV
-     * \brief Serialisation support function.
+     * \brief The version of the HIBF.
+     */
+    static constexpr uint32_t version{1};
+
+    /*!\brief Serialisation support function.
      * \tparam archive_t Type of `archive`; must satisfy seqan::hibf::cereal_archive.
      * \param[in] archive The archive being serialised from/to.
      *
@@ -424,6 +428,8 @@ class interleaved_bloom_filter : private seqan::hibf::bit_vector
     template <cereal_archive archive_t>
     void CEREAL_SERIALIZE_FUNCTION_NAME(archive_t & archive)
     {
+        uint32_t parsed_version{version};
+        archive(cereal::make_nvp("version", parsed_version));
         archive(bins);
         archive(technical_bins);
         archive(bin_size_);

src/hierarchical_interleaved_bloom_filter.cpp

@@ -42,7 +42,8 @@ size_t hierarchical_build(hierarchical_interleaved_bloom_filter & hibf,
                           robin_hood::unordered_flat_set<uint64_t> & parent_kmers,
                           layout::graph::node const & current_node,
                           build::build_data & data,
-                          bool is_root)
+                          bool is_root,
+                          size_t const parent_ibf_idx = 0u)
 {
     size_t const ibf_pos{data.request_ibf_idx()};
 
@@ -68,7 +69,8 @@ size_t hierarchical_build(hierarchical_interleaved_bloom_filter & hibf,
                                    kmers,
                                    current_node.children[current_node.favourite_child_idx.value()],
                                    data,
-                                   false);
+                                   false,
+                                   ibf_pos);
             return 1;
         }
         else // max bin is not a merged bin
@@ -124,8 +126,9 @@ size_t hierarchical_build(hierarchical_interleaved_bloom_filter & hibf,
             auto & child = children[index];
 
             robin_hood::unordered_flat_set<uint64_t> local_kmers{};
-            size_t const local_ibf_pos = hierarchical_build(hibf, local_kmers, child, data, false);
+            size_t const local_ibf_pos = hierarchical_build(hibf, local_kmers, child, data, false, ibf_pos);
             auto parent_bin_index = child.parent_bin_index;
+            hibf.prev_ibf_id[local_ibf_pos] = {.ibf_idx = parent_ibf_idx, .bin_idx = parent_bin_index};
             {
                 size_t const mutex_id{parent_bin_index / 64};
                 std::lock_guard<std::mutex> guard{local_ibf_mutex[mutex_id]};
@@ -184,6 +187,7 @@ void build_index(hierarchical_interleaved_bloom_filter & hibf,
 
     hibf.ibf_vector.resize(number_of_ibfs);
     hibf.ibf_bin_to_user_bin_id.resize(number_of_ibfs);
+    hibf.prev_ibf_id.resize(number_of_ibfs);
     hibf.next_ibf_id.resize(number_of_ibfs);
 
     build::build_data data{.config = config, .ibf_graph = {hibf_layout}};