[FIX] PIBF not working

include/raptor/search/run_program_multiple.hpp

@@ -74,6 +74,8 @@ void run_program_multiple(search_arguments const & arguments)
         auto end = std::chrono::high_resolution_clock::now();
         reads_io_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
 
+        cereal_handle.wait();
+
         std::vector<seqan3::counting_vector<uint16_t>> counts(records.size(),
                                                               seqan3::counting_vector<uint16_t>(ibf.bin_count(), 0));
 
@@ -94,7 +96,6 @@ void run_program_multiple(search_arguments const & arguments)
             }
         };
 
-        cereal_handle.wait();
         do_parallel(count_task, records.size(), arguments.threads, compute_time);
 
         for (size_t const part : std::views::iota(1u, static_cast<unsigned int>(arguments.parts - 1)))

src/argument_parsing/search.cpp

@@ -15,8 +15,7 @@ void init_search_parser(seqan3::argument_parser & parser, search_arguments & arg
                       "index",
                       arguments.is_socks ? "Provide a valid path to an IBF." :
                                            "Provide a valid path to an IBF. Parts: Without suffix _0",
-                      seqan3::option_spec::required,
-                      seqan3::input_file_validator{});
+                      seqan3::option_spec::required);
     parser.add_option(arguments.query_file,
                       '\0',
                       "query",
@@ -81,6 +80,19 @@ void run_search(seqan3::argument_parser & parser, bool const is_socks)
 
     arguments.treshold_was_set = parser.is_option_set("threshold");
 
+    if (arguments.parts == 1)
+    {
+        seqan3::input_file_validator{}(arguments.ibf_file);
+    }
+    else
+    {
+        seqan3::input_file_validator validator{};
+        for (size_t part{0}; part < arguments.parts; ++part)
+        {
+            validator(arguments.ibf_file.string() + std::string{"_"} + std::to_string(part));
+        }
+    }
+
     // ==========================================
     // Process --pattern.
     // ==========================================
@@ -100,7 +112,9 @@ void run_search(seqan3::argument_parser & parser, bool const is_socks)
     // Read window and kmer size, and the bin paths.
     // ==========================================
     {
-        std::ifstream is{arguments.ibf_file, std::ios::binary};
+        std::ifstream is{arguments.parts == 1 ? arguments.ibf_file.string() :
+                                                arguments.ibf_file.string() + std::string{"_0"},
+                         std::ios::binary};
         cereal::BinaryInputArchive iarchive{is};
         iarchive(arguments.kmer_size);
         iarchive(arguments.window_size);

test/cli/cli_test.hpp

@@ -156,3 +156,4 @@ struct raptor : public cli_test
 
 struct raptor_build : public raptor, public testing::WithParamInterface<std::tuple<size_t, size_t, bool>> {};
 struct raptor_search : public raptor, public testing::WithParamInterface<std::tuple<size_t, size_t, size_t>> {};
+struct raptor_parts : public raptor, public testing::WithParamInterface<std::tuple<size_t, size_t, bool, size_t>> {};

test/cli/raptor_test.cpp

@@ -289,3 +289,92 @@ INSTANTIATE_TEST_SUITE_P(search_suite,
                                                 std::to_string(std::get<2>(info.param)) + "_error";
                              return name;
                          });
+
+////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+////////////////////////////////////////////////// raptor parts tests //////////////////////////////////////////////////
+////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+TEST_P(raptor_parts, pipeline)
+{
+    auto const [number_of_repeated_bins, window_size, run_parallel_tmp, number_of_errors] = GetParam();
+    bool const run_parallel = run_parallel_tmp && number_of_repeated_bins >= 32;
+
+    if (window_size == 23 && number_of_errors == 0)
+        GTEST_SKIP() << "Needs dynamic threshold correction";
+
+    std::stringstream header{};
+    {
+        std::string const expanded_bins = repeat_bins(number_of_repeated_bins);
+        std::ofstream file{"raptor_cli_test.txt"};
+        auto split_bins = expanded_bins
+                        | std::views::split(' ')
+                        | std::views::transform([](auto &&rng) {
+                            return std::string_view(&*rng.begin(), std::ranges::distance(rng));});
+        size_t usr_bin_id{0};
+        for (auto && file_path : split_bins)
+        {
+            header << '#' << usr_bin_id++ << '\t' << file_path << '\n';
+            file << file_path << '\n';
+        }
+        header << "#QUERY_NAME\tUSER_BINS\n";
+        file << '\n';
+    }
+
+    cli_test_result const result1 = execute_app("raptor", "build",
+                                                          "--kmer 19",
+                                                          "--window ", std::to_string(window_size),
+                                                          "--size 64k",
+                                                          "--threads ", run_parallel ? "2" : "1",
+                                                          "--output index.ibf",
+                                                          "--parts 4",
+                                                          "raptor_cli_test.txt");
+    EXPECT_EQ(result1.out, std::string{});
+    EXPECT_EQ(result1.err, std::string{});
+    ASSERT_EQ(result1.exit_code, 0);
+
+    cli_test_result const result2 = execute_app("raptor", "search",
+                                                          "--output search.out",
+                                                          "--error ", std::to_string(number_of_errors),
+                                                          "--index ", "index.ibf",
+                                                          "--parts 4",
+                                                          "--query ", data("query.fq"));
+    EXPECT_EQ(result2.out, std::string{});
+    EXPECT_EQ(result2.err, std::string{});
+    ASSERT_EQ(result2.exit_code, 0);
+
+    std::string const expected = [&] ()
+    {
+        std::string result{header.str()};
+        std::string line{};
+        std::ifstream search_result{search_result_path(number_of_repeated_bins,
+                                                       window_size,
+                                                       number_of_errors)};
+        while (std::getline(search_result, line) && line.substr(0, 6) != "query1")
+        {}
+        result += line;
+        result += '\n';
+        while (std::getline(search_result, line))
+        {
+            result += line;
+            result += '\n';
+        }
+
+        return result;
+    }();
+
+    std::string const actual = string_from_file("search.out");
+
+    EXPECT_EQ(expected, actual);
+}
+
+INSTANTIATE_TEST_SUITE_P(parts_suite,
+                         raptor_parts,
+                         testing::Combine(testing::Values(0), testing::Values(19, 23), testing::Values(true, false), testing::Values(0, 1)),
+                         [] (testing::TestParamInfo<raptor_parts::ParamType> const & info)
+                         {
+                             std::string name = std::to_string(std::max<int>(1, std::get<0>(info.param) * 4)) + "_bins_" +
+                                                std::to_string(std::get<1>(info.param)) + "_window_" +
+                                                (std::get<2>(info.param) ? "parallel" : "serial") +
+                                                std::to_string(std::get<3>(info.param)) + "_error";
+                             return name;
+                         });