[FEATURE] Make hibf constructible from a layout file.

seqan · Aug 23, 2023 · 96456d7 · 96456d7
1 parent 991c413
commit 96456d7
Show file tree

Hide file tree

Showing 3 changed files with 72 additions and 0 deletions.
diff --git a/include/hibf/hierarchical_interleaved_bloom_filter.hpp b/include/hibf/hierarchical_interleaved_bloom_filter.hpp
@@ -105,6 +105,8 @@ class hierarchical_interleaved_bloom_filter
     ~hierarchical_interleaved_bloom_filter() = default;            //!< Defaulted.
 
     hierarchical_interleaved_bloom_filter(config const & configuration);
+    hierarchical_interleaved_bloom_filter(std::function<void(size_t const, insert_iterator &&)> input_fn,
+                                          std::filesystem::path const & layout_filename);
     //!\}
 
     //!\brief The individual interleaved Bloom filters.

diff --git a/src/hierarchical_interleaved_bloom_filter.cpp b/src/hierarchical_interleaved_bloom_filter.cpp
@@ -199,4 +199,20 @@ hierarchical_interleaved_bloom_filter::hierarchical_interleaved_bloom_filter(con
     build_index(*this, configuration, layout);
 }
 
+hierarchical_interleaved_bloom_filter::hierarchical_interleaved_bloom_filter(
+    std::function<void(size_t const, insert_iterator &&)> input_fn,
+    std::filesystem::path const & layout_filename)
+{
+    // read config and layout from file
+    std::ifstream layout_file{layout_filename};
+    config configuration;
+    layout::layout hibf_layout;
+    configuration.read_from(layout_file);
+    hibf_layout.read_from(layout_file);
+
+    configuration.input_fn = input_fn; // set input as it cannot be serialized.
+
+    build_index(*this, configuration, hibf_layout);
+}
+
 } // namespace hibf
diff --git a/test/unit/hibf/hierarchical_interleaved_bloom_filter_test.cpp b/test/unit/hibf/hierarchical_interleaved_bloom_filter_test.cpp
@@ -9,11 +9,13 @@
 
 #include <cstddef>    // for size_t
 #include <functional> // for function
+#include <iostream>
 #include <vector>     // for vector, allocator
 
 #include <hibf/config.hpp>                                // for insert_iterator, config
 #include <hibf/hierarchical_interleaved_bloom_filter.hpp> // for hierarchical_interleaved_bloom_filter
 #include <hibf/test/expect_range_eq.hpp>                  // for expect_range_eq, EXPECT_RANGE_EQ
+#include <hibf/test/tmp_directory.hpp>
 
 TEST(hibf_test, test_specific_hash_values)
 {
@@ -41,6 +43,58 @@ TEST(hibf_test, test_specific_hash_values)
     }
 }
 
+TEST(hibf_test, build_from_layout)
+{
+    // range of range of sequences
+    std::vector<std::vector<size_t>> hashes{{1u, 2u, 3u, 4u, 5u, 6u, 7u, 8u, 9u, 10u}, {1u, 2u, 3u, 4u, 5u}};
+
+    auto input_fn = [&](size_t const num, hibf::insert_iterator it)
+                        {
+                            for (auto const hash : hashes[num])
+                                it = hash;
+                        };
+
+    hibf::test::tmp_directory tmp{};
+    std::filesystem::path layout_filename{tmp.path()/"dummy.layout"};
+
+    {
+        std::ofstream stream{layout_filename};
+        stream << "##CONFIG:\n";
+        stream << "##{\n";
+        stream << "##    \"config\": {\n";
+        stream << "##        \"version\": 1,\n";
+        stream << "##        \"number_of_user_bins\": 2,\n";
+        stream << "##        \"number_of_hash_functions\": 2,\n";
+        stream << "##        \"maximum_false_positive_rate\": 0.05,\n";
+        stream << "##        \"threads\": 1,\n";
+        stream << "##        \"sketch_bits\": 12,\n";
+        stream << "##        \"tmax\": 64,\n";
+        stream << "##        \"alpha\": 1.2,\n";
+        stream << "##        \"max_rearrangement_ratio\": 0.5,\n";
+        stream << "##        \"disable_estimate_union\": false,\n";
+        stream << "##        \"disable_rearrangement\": true,\n";
+        stream << "##        \"disable_cutoffs\": false\n";
+        stream << "##    }\n";
+        stream << "##}\n";
+        stream << "##ENDCONFIG\n";
+        stream << "#HIGH_LEVEL_IBF max_bin_id:0\n";
+        stream << "#USER_BIN_IDX\tTECHNICAL_BIN_INDICES\tNUMBER_OF_TECHNICAL_BINS\n";
+        stream << "1\t0\t34\n";
+        stream << "0\t34\t30\n";
+    }
+
+    hibf::hierarchical_interleaved_bloom_filter hibf{input_fn, layout_filename};
+
+    {
+        std::vector<size_t> query{1, 2, 3, 4, 5};
+
+        auto agent = hibf.membership_agent();
+        auto result = agent.bulk_contains(query, 2);
+
+        EXPECT_RANGE_EQ(result, (std::vector<size_t>{0u, 1u}));
+    }
+}
+
 // #ifdef HIBF_HAS_SEQAN3
 
 // #include <seqan3/alphabet/nucleotide/dna4.hpp>