From b3a9a6f9513f2db10aaf5cd558fbc2ea5561edf8 Mon Sep 17 00:00:00 2001
From: Tuukka Norri <tsnorri@iki.fi>
Date: Thu, 20 Apr 2023 16:01:54 +0300
Subject: [PATCH] [MISC] Make seqan3::sam_file_header::program_info_t easier to
 copy

See #3137.
---
 include/seqan3/io/sam_file/header.hpp         | 32 +++++++++-------
 .../unit/io/sam_file/sam_file_output_test.cpp | 37 +++++++++++++++++++
 2 files changed, 56 insertions(+), 13 deletions(-)

diff --git a/include/seqan3/io/sam_file/header.hpp b/include/seqan3/io/sam_file/header.hpp
index b98af2ea8d..1e670abdaf 100644
--- a/include/seqan3/io/sam_file/header.hpp
+++ b/include/seqan3/io/sam_file/header.hpp
@@ -24,6 +24,21 @@
 namespace seqan3
 {
 
+/*!\brief Stores information of the program/tool that was used to create a SAM/BAM file.
+ * \ingroup io_sam_file
+ *
+ * \remark For a complete overview, take a look at \ref io_sam_file
+ */
+struct sam_file_program_info_t
+{
+    std::string id;                //!< A unique (file scope) id.
+    std::string name;              //!< The official name.
+    std::string command_line_call; //!< The command line call that produces the file.
+    std::string previous;          //!< The id of the previous program if program calls were chained.
+    std::string description;       //!< A description of the program and/or program call.
+    std::string version;           //!< The program/tool version.
+};
+
 /*!\brief Stores the header information of SAM/BAM files.
  * \ingroup io_sam_file
  *
@@ -63,19 +78,10 @@ class sam_file_header
     {}
     //!\}
 
-    //!\brief Stores information of the program/tool that was used to create the file.
-    struct program_info_t
-    {
-        std::string id;                //!< A unique (file scope) id.
-        std::string name;              //!< The official name.
-        std::string command_line_call; //!< The command line call that produces the file.
-        std::string previous;          //!< The id of the previous program if program calls were chained.
-        std::string description;       //!< A description of the program and/or program call.
-        std::string version;           //!< The program/tool version.
-    };
-
-    std::string format_version; //!< The file format version. Note: this is overwritten by our formats on output.
-    std::string sorting;        //!< The sorting of the file. SAM: [unknown, unsorted, queryname, coordinate].
+    using program_info_t =
+        sam_file_program_info_t; //!< Stores information of the program/tool that was used to create the file.
+    std::string format_version;  //!< The file format version. Note: this is overwritten by our formats on output.
+    std::string sorting;         //!< The sorting of the file. SAM: [unknown, unsorted, queryname, coordinate].
     std::string
         subsorting; //!< The sub-sorting of the file. SAM: [unknown, unsorted, queryname, coordinate](:[A-Za-z0-9_-]+)+.
     std::string grouping; //!< The grouping of the file. SAM: [none, query, reference].
diff --git a/test/unit/io/sam_file/sam_file_output_test.cpp b/test/unit/io/sam_file/sam_file_output_test.cpp
index 0d4fc35cb1..b8be9cbe7c 100644
--- a/test/unit/io/sam_file/sam_file_output_test.cpp
+++ b/test/unit/io/sam_file/sam_file_output_test.cpp
@@ -8,7 +8,9 @@
 #include <gtest/gtest.h>
 
 #include <iterator>
+#include <ranges>
 #include <sstream>
+#include <type_traits>
 
 #include <seqan3/io/sam_file/input.hpp>
 #include <seqan3/io/sam_file/output.hpp>
@@ -197,6 +199,41 @@ void assign_impl(source_t && source)
     EXPECT_EQ(reinterpret_cast<std::ostringstream &>(fout.get_stream()).str(), output_comp);
 }
 
+// ----------------------------------------------------------------------------
+// header
+// ----------------------------------------------------------------------------
+TEST(header, copy_program_info_t)
+{
+    std::string comp =
+        R"(@HD	VN:1.6	SO:unknown	GO:none
+@SQ	SN:ref	LN:26
+@PG	ID:prog1	PN:cool_program
+@CO	This is a comment.
+read1	41	ref	1	61	1S1M1D1M1I	ref	10	300	ACGT	!##$	AS:i:2	NM:i:7
+)";
+    seqan3::sam_file_input fin{std::istringstream{comp}, seqan3::format_sam{}};
+
+    auto & input_header(fin.header());
+
+    // We would like to test that (some of) the headers can be copied even if
+    // ref_ids_types do not match.
+    typedef std::remove_cvref_t<decltype(input_header.ref_ids())>
+        input_ref_ids_type; // std::deque <std::string> by default.
+    typedef std::array<std::string, 0> output_ref_ids_type;
+    static_assert(!std::is_same_v<input_ref_ids_type, output_ref_ids_type>);
+
+    seqan3::sam_file_output fout{std::ostringstream{},
+                                 output_ref_ids_type{},
+                                 std::ranges::empty_view<std::size_t>{},
+                                 seqan3::format_sam{}};
+    fout.header().program_infos = fin.header().program_infos;
+
+    // Verify that the above was sufficient to copy the headers.
+    auto const & pg_infos(fout.header().program_infos);
+    EXPECT_FALSE(pg_infos.empty());
+    EXPECT_EQ("cool_program", pg_infos.front().name);
+}
+
 // ----------------------------------------------------------------------------
 // row
 // ----------------------------------------------------------------------------