⬆ Update qurator_data_lib.sh (Fixes GH-5)

qurator_data_lib.sh

@@ -1,3 +1,14 @@
+# ______________________________________
+#/ always copy the file from            \
+#| mono-repo/qurator_data_lib.sh, never |
+#\ edit the copy in the project         /
+# --------------------------------------
+#        \   ^__^
+#         \  (oo)\_______
+#            (__)\       )\/\
+#                ||----w |
+#                ||     ||
+
 if [ -z "$BASH" ]; then
   echo "qurator_data_lib.sh uses bash features, please make sure to run $0 in bash"
   exit 1
@@ -23,14 +34,31 @@ check_data_subdir() {
 }
 
 annex_get() {
+  if [[ "$1" = '--allow_symlinks' ]]; then
+    allow_symlinks=1
+    shift
+  else
+    allow_symlinks=0
+  fi
   file_pattern="$1"
 
   (
     cd $DATA_SUBDIR
     git annex get $file_pattern
 
-    # fsck seems to be necessary to fix the files if we're in a submodule
+    # fsck seems to be necessary to fix the files if we are in a submodule
     git annex fsck $file_pattern
+
+    # Check that there are no symlinks = only unlocked files. This is needed for
+    # Docker builds, as we cannot dereference symlinks in a Dockerfile COPY.
+    if [[ $allow_symlinks = 0 ]]; then
+      git ls-files $file_pattern | while read f; do
+        if ! [[ -f "$f" ]]; then
+           echo "$DATA_SUBDIR/$f is not a regular file – Is an unlock needed?"
+           exit
+        fi
+      done
+    fi
   )
 }
 
@@ -58,9 +86,24 @@ suggest_commands() {
 }
 
 handle_data() {
+  if [[ "$1" = '--no-download' ]]; then
+    no_download=1
+    shift
+  else
+    no_download=0
+  fi
+
   if [ -n "$FORCE_DOWNLOAD" ]; then
     get_from_web
   elif ! check_data_subdir; then
+    if [[ $no_download = 1 ]]; then
+      select choice in "Abort to manually fix $DATA_SUBDIR submodule"; do
+        if [ $REPLY = 1 ]; then
+          suggest_commands
+          exit
+        fi
+      done
+    else
       select choice in "Abort to manually fix $DATA_SUBDIR submodule" "Download data files from the web"; do
         if [ $REPLY = 1 ]; then
           suggest_commands
@@ -70,6 +113,7 @@ handle_data() {
           break
         fi
       done
+    fi
   else
     get_from_annex
   fi