From 78d0a69febd70d7247a9efea030be90a14509b13 Mon Sep 17 00:00:00 2001
From: "eugen.betke" <betke@dkrz.de>
Date: Tue, 18 Aug 2020 15:27:39 +0200
Subject: [PATCH] Compress script

---
 datasets/compress.sh | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
 create mode 100755 datasets/compress.sh

diff --git a/datasets/compress.sh b/datasets/compress.sh
new file mode 100755
index 0000000..bb2abdb
--- /dev/null
+++ b/datasets/compress.sh
@@ -0,0 +1,19 @@
+#!/bin/bash
+
+filenames=( "job_codings_v3.csv" )
+filenames=( ${filenames[@]} "job_metadata.csv" )
+filenames=( ${filenames[@]} $( ls job_similarities_*.csv ) )
+
+echo "${filenames[*]}"
+
+for in_fn in ${filenames[@]}; do
+    out_fn="${in_fn}.tar.xz"
+    if [ -f "${in_fn}" ]; then
+        if [ ! -f "${out_fn}" ]; then
+            echo "Compressing ${in_fn}"
+            tar -cf - "${in_fn}" | xz -9 > "${out_fn}"
+        else
+            echo "Skipping ${out_fn}. File exists."
+        fi
+    fi
+done