Merge branch 'feature-a/4269-tsv-gz' into 'dev'

--gz makes also output AIRR .tsv.gz

Closes #4249

See merge request !675

algo/core/tools.cpp

@@ -473,3 +473,19 @@ string extractGeneName(string label){
 	}
 	return result;
 }
+
+
+/*
+   Opens a ostream, possibly gz-compressed
+*/
+std::ostream* new_ofgzstream(const char *f, bool gz)
+{
+  if (gz)
+  {
+    return new ogzstream(f);
+  }
+  else
+  {
+    return new ofstream(f);
+  }
+}
\ No newline at end of file

algo/core/tools.h

@@ -43,6 +43,7 @@ typedef string junction ;
 #include <cassert>
 #include <vector>
 #include "bioreader.hpp"
+#include "../lib/gzstream.h"
 #include "kmeraffect.h"
 #include "../lib/json_fwd.hpp"
 using json = nlohmann::json;
@@ -281,6 +282,11 @@ void output_label_average(ostream &out, string label, long long int nb, double a
 
 void json_add_warning(json &clone, string code, string msg, string level=LEVEL_WARN);
 
+/*
+   Opens a ostream, possibly gz-compressed
+*/
+std::ostream* new_ofgzstream(const char *f, bool gz);
+
 
 //////////////////////////////////////////////////
 // Template code

algo/tests/should-get-tests/gzip-out.should

@@ -2,15 +2,20 @@
 
 $ Output
 1: out-a.vidjil
+1: out-a.tsv
 
 !LAUNCH: $VIDJIL_DIR/$EXEC -g $VIDJIL_DIR/germline/homo-sapiens.g:IGH --gz -b out-b $VIDJIL_DATA/clones_simul.fa
 
-$ Output
+$ Compressed output
 1: out-b.vidjil.gz
+1: out-b.tsv.gz
 
 !NO_LAUNCHER:
 !NO_EXTRA:
-gunzip -f out/out-b.vidjil.gz && diff -s -I '"....-..-..' -I "gz" -I "out-" out/out-a.vidjil out/out-b.vidjil ; echo "Diff: $?"
+gunzip -f out/out-b.vidjil.gz && diff -s -I '"....-..-..' -I "gz" -I "out-" out/out-a.vidjil out/out-b.vidjil ; echo "Diff .vidjil.gz: $?"
+gunzip -f out/out-b.tsv.gz    && diff -s out/out-a.tsv out/out-b.tsv ; echo "Diff .tsv.gz: $?"
+
 
 $ Same output
-1:Diff: 0
+1:Diff .vidjil.gz: 0
+1:Diff .tsv.gz: 0

algo/vidjil.cpp

@@ -58,7 +58,6 @@
 #include "lib/CLI11.hpp"
 #include "lib/json.hpp"
 #include "lib/CLI11_json.hpp"
-#include "lib/gzstream.h"
 
 #include "vidjil.h"
 
@@ -585,7 +584,7 @@ int main (int argc, char **argv)
   app.add_option("--base,-b", f_basename, "output basename (by default basename of the input file)") -> group(group) -> type_name("STRING");
 
   bool out_gz = false;
-  app.add_flag("--gz", out_gz, "output compressed .vidjil.gz file") -> group(group) -> level();
+  app.add_flag("--gz", out_gz, "output compressed .tsv.gz and .vidjil.gz files") -> group(group) -> level();
 
   bool no_airr = false;
   bool no_vidjil = false;
@@ -826,7 +825,10 @@ int main (int argc, char **argv)
   string f_json = out_dir + f_basename + JSON_SUFFIX ;
 
   if (out_gz)
+  {
+    f_airr += GZ_SUFFIX;
     f_json += GZ_SUFFIX;
+  }
 
   ostringstream stream_cmdline;
   for (int i=0; i < argc; i++) stream_cmdline << argv[i] << " ";
@@ -1299,7 +1301,7 @@ int main (int argc, char **argv)
         << endl ;
 
     string f_clones = out_dir + f_basename + CLONES_FILENAME ;
-    cout << "  ==> " << f_clones << "   \t(main result file)" << endl ;
+    cout << "  ==> " << f_clones << "   \t(for post-processing with other software)" << endl ;
     ofstream out_clones(f_clones.c_str()) ;
 
     cout << "  ==> " << out_seqdir + CLONE_FILENAME + "*" << "\t(detail, by clone)" << endl ; 
@@ -1745,36 +1747,27 @@ int main (int argc, char **argv)
     cout << endl;
   }
 
-  //$ Output AIRR .tsv
+  //$ Output AIRR .tsv(.gz)
   if (!no_airr)
   {
     cout << "  ==> " << f_airr << "   \t(AIRR output)" << endl;
-    ofstream out_airr(f_airr.c_str());
-    static_cast<SampleOutputAIRR *>(&output) -> out(out_airr);
+    std::ostream *out_airr = new_ofgzstream(f_airr.c_str(), out_gz);
+    static_cast<SampleOutputAIRR *>(&output) -> out(*out_airr);
+    delete out_airr;
   }
 
   //$ Output .vidjil(.gz) json
   cout << "  ==> " << f_json ;
   if (!no_vidjil)
   {
-    cout << "\t(data file for the Vidjil web application)" << endl;
+    cout << "\t(main output file, may be opened by the Vidjil web application)" << endl;
   }
   else
   {
     cout << "\t(only metadata, no clone output)" << endl;
   }
 
-  std::ostream *out_json;
-
-  if (out_gz)
-  {
-    out_json = new ogzstream(f_json.c_str());
-  }
-  else
-  {
-    out_json = new ofstream(f_json.c_str());
-  }
-
+  std::ostream *out_json = new_ofgzstream(f_json.c_str(), out_gz);
   SampleOutputVidjil *outputVidjil = static_cast<SampleOutputVidjil *>(&output);
 
   outputVidjil -> out(*out_json, !no_vidjil);

doc/vidjil-algo.md

@@ -549,6 +549,8 @@ The main output of Vidjil-algo (with the default `-c clones` command) are the th
   - The `.tsv` file is the AIRR output, for compatibility with other software
     using the same format. See [below](#airr-tsv-output) for details.
 
+    With the `--gz` option, this file is output as a compressed `.tsv.gz` file.
+
   - The `.vdj.fa` file is *a FASTA file for further processing by other bioinformatics tools*.
     Even if it is advised to rather use the full information in the `.vijdil` file,
     the `.vdj.fa` is a convenient way to have sequences of clones for further processing.