Remove preliminary output

include/utilities/consolidate/consolidate_matches.hpp

@@ -1,5 +1,7 @@
 #pragma once
 
+#include <filesystem>
+
 #include <valik/shared.hpp>
 #include <valik/split/database_metadata.hpp>
 
@@ -12,6 +14,7 @@ namespace valik
  * @brief Function that removes duplicates from split Stellar search results and writes the final output file.
  *
  * @param arguments Command line arguments.
+ * @return false If could not clean up intermediate results.
  */
 void consolidate_matches(search_arguments const & arguments);
 

src/consolidate/consolidate_matches.cpp

@@ -8,6 +8,8 @@ void consolidate_matches(search_arguments const & arguments)
     auto ref_meta = database_metadata(arguments.ref_meta_path, false);
     auto matches = read_stellar_output(arguments.all_matches, ref_meta);
 
+    seqan3::debug_stream << std::string(arguments.all_matches) << '\t' << matches.size() << '\n';
+
     std::sort( matches.begin(), matches.end(), std::greater<stellar_match>());
     matches.erase( std::unique( matches.begin(), matches.end() ), matches.end() );
 

src/valik_search.cpp

@@ -46,7 +46,13 @@ void valik_search(search_arguments const & arguments)
 
     // Consolidate matches (not necessary when searching a metagenomic database)
     if (!arguments.ref_meta_path.empty())
+    {
         consolidate_matches(arguments);
+        const bool error_in_delete = !std::filesystem::remove(arguments.all_matches);
+        if (error_in_delete)
+            std::cerr << "Could not clean up intermediate file: \t" << std::string(arguments.all_matches) << '\n';
+        failed = failed || error_in_delete;
+    }
 
     if (arguments.write_time)
         write_time_statistics(time_statistics, arguments.out_file.string() + ".time");

test/api/utilities/consolidate/consolidate_matches_test.cpp

@@ -3,7 +3,7 @@
 #include <utilities/consolidate/consolidate_matches.hpp>
 
 // Generate the full path of a test input file that is provided in the data directory.
-std::filesystem::path data(std::string const & filename)
+std::filesystem::path data_path(std::string const & filename)
 {
     return std::filesystem::path{std::string{DATADIR}}.concat(filename);
 }
@@ -15,7 +15,7 @@ std::filesystem::path consolidation_input_path(size_t const number_of_bins, size
     name += "bins";
     name += std::to_string(overlap);
     name += "overlap_dream_all.gff";
-    return data(name);
+    return data_path(name);
 }
 
 std::filesystem::path consolidation_meta_path(size_t const number_of_bins, size_t const overlap) noexcept
@@ -25,7 +25,7 @@ std::filesystem::path consolidation_meta_path(size_t const number_of_bins, size_
     name += "bins";
     name += std::to_string(overlap);
     name += "overlap_reference_metadata.tsv";
-    return data(name);
+    return data_path(name);
 }
 
 void compare_gff_out(std::vector<valik::stellar_match> const & expected,
@@ -62,8 +62,8 @@ TEST(consolidate_matches, bins8overlap50)
     valik::consolidate_matches(arguments);
 
     valik::database_metadata reference(arguments.ref_meta_path, false);
-    auto expected = valik::read_stellar_output(data("stellar_truth.gff"), reference, std::ios::binary);
-    auto actual = valik::read_stellar_output("consolidated.gff", reference);
+    auto expected = valik::read_stellar_output(data_path("stellar_truth.gff"), reference, std::ios::binary);
+    auto actual = valik::read_stellar_output("consolidated.gff", reference, std::ios::binary);
 
     compare_gff_out(expected, actual);
 }
@@ -81,8 +81,8 @@ TEST(consolidate_matches, bins16overlap50)
     valik::consolidate_matches(arguments);
 
     valik::database_metadata reference(arguments.ref_meta_path, false);
-    auto expected = valik::read_stellar_output(data("stellar_truth.gff"), reference, std::ios::binary);
-    auto actual = valik::read_stellar_output("consolidated.gff", reference);
+    auto expected = valik::read_stellar_output(data_path("stellar_truth.gff"), reference, std::ios::binary);
+    auto actual = valik::read_stellar_output("consolidated.gff", reference, std::ios::binary);
 
     compare_gff_out(expected, actual);
 }

test/api/valik/split/write_seg_sequences_test.cpp

@@ -6,7 +6,7 @@
 #include <seqan3/test/expect_range_eq.hpp>
 
 // Generate the full path of a test input file that is provided in the data directory.
-static std::filesystem::path data(std::string const & filename)
+static std::filesystem::path data_path(std::string const & filename)
 {
     return std::filesystem::path{std::string{DATADIR}}.concat(filename);
 }
@@ -35,17 +35,17 @@ static void const test_reference_out(size_t overlap, size_t bins)
 {
     std::string path_prefix = "write_out_" + std::to_string(overlap) + "_" + std::to_string(bins);
 
-    valik::database_metadata reference(data(path_prefix + "_reference_metadata.txt"), false);
-    valik::database_segments segments(data(path_prefix + "_reference_segments.txt"));
-    valik::write_reference_segments(reference, segments, data("database.fasta"));
+    valik::database_metadata reference(data_path(path_prefix + "_reference_metadata.txt"), false);
+    valik::database_segments segments(data_path(path_prefix + "_reference_segments.txt"));
+    valik::write_reference_segments(reference, segments, data_path("database.fasta"));
 
     for (size_t i = 0; i < bins - 1; i++)
     {
         valik::database_segments::segment current_seg = segments.members[i];
         valik::database_segments::segment next_seg = segments.members[i + 1];
 
-        std::string current_seg_seq = string_from_file(data("database_" + std::to_string(i) + ".fasta"), std::ios::binary);
-        std::string next_seg_seq = string_from_file(data("database_" + std::to_string(i + 1) + ".fasta"), std::ios::binary);
+        std::string current_seg_seq = string_from_file(data_path("database_" + std::to_string(i) + ".fasta"), std::ios::binary);
+        std::string next_seg_seq = string_from_file(data_path("database_" + std::to_string(i + 1) + ".fasta"), std::ios::binary);
         trim_fasta_id(current_seg_seq);
         trim_fasta_id(next_seg_seq);
 
@@ -95,15 +95,15 @@ static void const test_query_out(size_t overlap, size_t bins)
 {
     std::string path_prefix = "write_out_" + std::to_string(overlap) + "_" + std::to_string(bins);
 
-    valik::database_metadata reference(data(path_prefix + "_reference_metadata.txt"), false);
-    valik::database_segments segments(data(path_prefix + "_reference_segments.txt"));
-    valik::write_query_segments(reference, segments, data("database.fasta"));
+    valik::database_metadata reference(data_path(path_prefix + "_reference_metadata.txt"), false);
+    valik::database_segments segments(data_path(path_prefix + "_reference_segments.txt"));
+    valik::write_query_segments(reference, segments, data_path("database.fasta"));
 
     using sequence_file_t = seqan3::sequence_file_input<valik::dna4_traits, seqan3::fields<seqan3::field::seq>>;
 
     size_t i{0};
     seqan3::dna4_vector previous_seg_seq;
-    for (auto && [current_seg_seq] : sequence_file_t{data("database.segments.fasta")})
+    for (auto && [current_seg_seq] : sequence_file_t{data_path("database.segments.fasta")})
     {
         if (i > 1)
         {

test/cli/CMakeLists.txt

@@ -70,11 +70,6 @@ target_use_datasources (valik_test FILES 8bins23window1error100pattern1overlap.g
 target_use_datasources (valik_test FILES 8bins23window1error100pattern40overlap.gff.out)
 target_use_datasources (valik_test FILES query.fq)
 
-target_use_datasources (valik_test FILES 16bins50overlap_dream_all.gff)
-target_use_datasources (valik_test FILES 16bins50overlap_reference_metadata.tsv)
-target_use_datasources (valik_test FILES 8bins50overlap_dream_all.gff)
-target_use_datasources (valik_test FILES 8bins50overlap_reference_metadata.tsv)
-target_use_datasources (valik_test FILES multi_seq_ref.fasta)
 target_use_datasources (valik_test FILES query_e0.06.fasta)
 
 add_cli_test (dream_test.cpp)

test/data/datasources.cmake

@@ -209,9 +209,6 @@ declare_datasource (FILE 16bins13window1error.gff
 declare_datasource (FILE 16bins13window1error.gff.out
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/16bins13window1error.gff.out
                 URL_HASH SHA256=89e61426004aa15bff39de90b4cd2d7e9080dc19cc3596a797f00d7c8f0c95c0)
-declare_datasource (FILE 16bins13window1error.gff.preliminary
-                URL ${CMAKE_SOURCE_DIR}/test/data/dream/16bins13window1error.gff.preliminary
-                URL_HASH SHA256=1b554d97c35e6bc5cfbda12a8fb817a980da5aeeb8d0d62a827d4996a18ab245)
 declare_datasource (FILE 16bins13window.ibf
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/16bins13window.ibf
                 URL_HASH SHA256=0875cd2d90d1320f93c575f3197e7879a5b8e51c5c11de76975561d30e4cf324)
@@ -221,9 +218,6 @@ declare_datasource (FILE 16bins15window1error.gff
 declare_datasource (FILE 16bins15window1error.gff.out
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/16bins15window1error.gff.out
                 URL_HASH SHA256=66d219613d6a715bbf6cd282868ee31612147287471b7e1820ca77bc370dec9d)
-declare_datasource (FILE 16bins15window1error.gff.preliminary
-                URL ${CMAKE_SOURCE_DIR}/test/data/dream/16bins15window1error.gff.preliminary
-                URL_HASH SHA256=d728395b3760ef0dc97e4ca37aa324b7ac4aa25601f6e62143a7b189e792b24b)
 declare_datasource (FILE 16bins15window.ibf
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/16bins15window.ibf
                 URL_HASH SHA256=7f1ce2bbdf8d657da29d39879ab23c68cb19dbe0b58c69a9c5a576f6528ad24c)
@@ -233,9 +227,6 @@ declare_datasource (FILE 4bins13window1error.gff
 declare_datasource (FILE 4bins13window1error.gff.out
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/4bins13window1error.gff.out
                 URL_HASH SHA256=5d285b543887e94829f15112b8d3c9659d48ff61a74cb216f2fa9f4bb54b4c90)
-declare_datasource (FILE 4bins13window1error.gff.preliminary
-                URL ${CMAKE_SOURCE_DIR}/test/data/dream/4bins13window1error.gff.preliminary
-                URL_HASH SHA256=a095ac60b3077cbb3a34a97886b19bb2cfcc0936cdbc8938694d7c829a06547d)
 declare_datasource (FILE 4bins13window.ibf
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/4bins13window.ibf
                 URL_HASH SHA256=b08ec3c196dc45faf73c24b86113e2c89adaf3d1844799d646e25dc0e77ac6bb)
@@ -245,9 +236,6 @@ declare_datasource (FILE 4bins15window1error.gff
 declare_datasource (FILE 4bins15window1error.gff.out
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/4bins15window1error.gff.out
                 URL_HASH SHA256=d7c0691b24eb8c949eb522b9844d647b62b0a621953692b15cad5c022128c29e)
-declare_datasource (FILE 4bins15window1error.gff.preliminary
-                URL ${CMAKE_SOURCE_DIR}/test/data/dream/4bins15window1error.gff.preliminary
-                URL_HASH SHA256=163e226fba9cca687ea233209d291bba4558cee6ef0a3d00e7b61631c712cd9e)
 declare_datasource (FILE 4bins15window.ibf
                 URL ${CMAKE_SOURCE_DIR}/test/data/dream/4bins15window.ibf
                 URL_HASH SHA256=8b584d0e55043b3cc0835674dc83f5a7db6143645071f49973a1d085ac4fb919)