From 6c5ba9f86a8d3a4cb0c40314cb2536bdcb2d6d1f Mon Sep 17 00:00:00 2001
From: fgvieira <1151762+fgvieira@users.noreply.github.com>
Date: Mon, 30 Jun 2025 10:20:25 +0200
Subject: [PATCH] Output gzip stats file

---
 metaDMG.cpp     | 15 +++++++--------
 test/testAll.sh |  8 ++++----
 2 files changed, 11 insertions(+), 12 deletions(-)

diff --git a/metaDMG.cpp b/metaDMG.cpp
index b2f4f05..0678f51 100644
--- a/metaDMG.cpp
+++ b/metaDMG.cpp
@@ -311,24 +311,23 @@ int main_getdamage(int argc, char **argv) {
 
     // write stat
     char buf[1024];
-    snprintf(buf, 1024, "%s.stat", onam);
+    snprintf(buf, 1024, "%s.stat.gz", onam);
     fprintf(stderr, "\t-> Outputting overall statistic in file: \"%s\"\n", buf);
 
-    FILE *fpstat = NULL;
-    assert(((fpstat = fopen(buf, "wb"))) != NULL);
-    fprintf(fpstat,"taxid\tnreads\tmea_len\tvar_len\tmean_gc\tvar_gc\tlca\trank\n");
+    gzFile fpstat = NULL;
+    assert((fpstat = gzopen(buf, "wb")) != NULL);
+    gzprintf(fpstat,"taxid\tnreads\tmea_len\tvar_len\tmean_gc\tvar_gc\tlca\trank\n");
     for (std::map<int, std::vector<float> >::iterator it = gcconts.begin(); it != gcconts.end(); it++) {
         std::map<int, triple>::iterator it2 = dmg->assoc.find(it->first);
         assert(it2 != dmg->assoc.end());
         std::map<int, std::vector<float> >::iterator it3 = seqlens.find(it->first);
         assert(it3 != seqlens.end());
         if (0)
-            fprintf(fpstat, "%d\t%lu\t%f\t%f\t%f\t%f\tNA\tNA\n", it->first, it2->second.nreads, mean(it3->second), var(it3->second), mean(it->second), var(it->second));
+            gzprintf(fpstat, "%d\t%lu\t%f\t%f\t%f\t%f\tNA\tNA\n", it->first, it2->second.nreads, mean(it3->second), var(it3->second), mean(it->second), var(it->second));
         else
-            fprintf(fpstat, "%s\t%lu\t%f\t%f\t%f\t%f\tNA\tNA\n", sam_hdr_tid2name(hdr, it->first), it2->second.nreads, mean(it3->second), var(it3->second), mean(it->second), var(it->second));
+            gzprintf(fpstat, "%s\t%lu\t%f\t%f\t%f\t%f\tNA\tNA\n", sam_hdr_tid2name(hdr, it->first), it2->second.nreads, mean(it3->second), var(it3->second), mean(it->second), var(it->second));
     }
-    if (fpstat != NULL)
-        fclose(fpstat);
+    gzclose(fpstat);
 
     sam_hdr_destroy(hdr);
     bam_destroy1(b);
diff --git a/test/testAll.sh b/test/testAll.sh
index 8afc006..a93b754 100755
--- a/test/testAll.sh
+++ b/test/testAll.sh
@@ -66,7 +66,7 @@ if [[ $? -ne 0 ]]; then
     echo "Problem running command: ${CMD}"
     RVAL=$((32+RVAL))
 fi
-gunzip -c  output/test_lca.lca.gz |cut -f1,3-100 >output/test_lca.lca.sub
+gunzip -c  output/test_lca.lca.gz | cut -f1,3-100 >output/test_lca.lca.sub
 
 echo "Running aggregate"
 CMD="${PRG} aggregate output/test_lca.bdamage.gz --nodes data/nodes.dmp.gz --names data/names.dmp.gz --lcastat output/test_lca.stat.gz --out_prefix output/test_aggregate"
@@ -87,13 +87,13 @@ fi
 echo "Running test of output/test_dfit_local.dfit.gz by gzip -t  output/test_dfit_local.dfit.gz "
 gzip -t  output/test_dfit_local.dfit.gz
 echo -ne "Return value of test $?\nWill now run gunzip -c output/test_dfit_local.dfit.gz|wc -c\n"
- gunzip -c output/test_dfit_local.dfit.gz|wc -c
+gunzip -c output/test_dfit_local.dfit.gz | wc -c
 
 
 gunzip -c output/test_dfit_local.dfit.gz > tmp.txt
 head -n 10 tmp.txt > tmp2.txt
 #rm tmp.txt
-cut -f 1-6,8- tmp2.txt| numfmt -d $'\t' --header --format='%.2f' --field=2- --invalid=ignore > output/test_dfit_local.dfit.fix
+cut -f 1-6,8- tmp2.txt | numfmt -d $'\t' --header --format='%.2f' --field=2- --invalid=ignore > output/test_dfit_local.dfit.fix
 #rm tmp2.txt
 #zcat output/test_dfit_local.dfit.gz | head -n 10|cut -f 1-6,8- | numfmt -d $'\t' --header --format='%.2f' --field=2- --invalid=ignore > output/test_dfit_local.dfit.fix
 #exit 0
@@ -112,7 +112,7 @@ fi
 gunzip -c output/test_dfit_local_10threads.dfit.gz > tmp.txt
 head -n 10 tmp.txt > tmp2.txt
 rm tmp.txt
-cut -f 1-6,8- tmp2.txt| numfmt -d $'\t' --header --format='%.2f' --field=2- --invalid=ignore | sort -r > output/test_dfit_local_10threads.dfit.fix
+cut -f 1-6,8- tmp2.txt | numfmt -d $'\t' --header --format='%.2f' --field=2- --invalid=ignore | sort -r > output/test_dfit_local_10threads.dfit.fix
 rm tmp2.txt
 
 echo "Running dfit global"