download_databases_from_s3bucket.sh tests now if folders are misssing in...

download_databases_from_s3bucket.sh tests now if folders are misssing in datadir and download only these

psot/helpers/download_databases_from_s3bucket.sh

@@ -8,32 +8,56 @@ S3BUCKET='psot-databases'
 list='ghostx uniprot pfam-31'
 
 function test_datadir () {
-  
+
   if ! [[ -e $DATADIR && -d $DATADIR ]]; then
     mkdir $DATADIR
-    sync_data
-  elif [ -z "$(ls -A $DATADIR)" ]; then  
-    sync_data    
+    sync_data "$list"
+  elif ! all_folders_in_datadir "test_true"; then
+    sync_data "$(all_folders_in_datadir 'return_missing')"
   fi
 
-}  
+}
+
+function all_folders_in_datadir () {
+
+  new_list=""
+
+  for ele in $list; do
+    if ! $(ls $DATADIR | grep -qx "$ele"); then
+      if [ $1 == "test_true" ]; then
+        return 1
+      else
+        new_list="$new_list $ele"
+      fi
+    fi
+  done
+
+  if [ $1 == "test_true" ]; then
+    return 0
+  else
+    echo "$new_list"
+  fi
+
+}
 
 function sync_data () {
-  
+
   echo "Could not find databases in config given path"
   echo "Download databases... "
 
   cd $DATADIR
 
-  for ele in $list; do
+  download_list=$1
+  for ele in $download_list; do
     wget -q "https://${S3BUCKET}.s3.computational.bio.uni-giessen.de/${ele}.tar.gz"
     tar -zxf "${ele}.tar.gz"
     rm "${ele}.tar.gz"
   done
-  
+
   ln -s 'pfam-31/' 'pfam'
 
   echo "[$(date)] psot (PID $$) downloaded databases to $DATADIR" >> $LOGFILE
+
 }
 
 if [[ -e $LOGFILE ]]; then