From 81657a288a4c286873fa84724e7e0e1b5b4e7923 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Thu, 5 Apr 2018 17:58:19 +0200
Subject: [PATCH] Modify script to accept compressed genome files

Former-commit-id: f816e3ce8b10e2ca3f1aa9ae969c24e699368e25
Former-commit-id: 16fb412552debdfd2172926e8a8b63be05257bdf
---
 detectors/normalize/tools/selectViridiplantae.sh | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/detectors/normalize/tools/selectViridiplantae.sh b/detectors/normalize/tools/selectViridiplantae.sh
index 895dd06..a75a6f3 100755
--- a/detectors/normalize/tools/selectViridiplantae.sh
+++ b/detectors/normalize/tools/selectViridiplantae.sh
@@ -1,8 +1,18 @@
 #!/bin/bash
 
-grep -A 1 '  ORGANISM' $* | \
+
+(                                          \
+	for f in $* ; do                       \
+		if [[ "$f" =~ \.gz$ ]] ; then      \
+			GREP=zgrep;                    \
+	    else                               \
+			GREP=grep;                     \
+	    fi;                                \
+		${GREP} -H -A 1 '  ORGANISM' $f;   \
+    done                                   \
+) | \
   grep -B 1 Viridiplantae | \
   gawk '{print $1}' | \
   grep '\.gbk' | \
-  sed -E 's/(^.*\.gbk).$/\1/' | \
+  sed -E 's/(^.*\.gbk(.gz)?).$/\1/' | \
   uniq
\ No newline at end of file