fix: harden qdrant restore safety

scripts/vector-backup.sh

@@ -1,6 +1,10 @@
 #!/bin/sh
 set -eu
 
+helper_image='qdrant/qdrant:v1.18.2@sha256:75eab8c4ba42096724fdcfde8b4de0b5713d529dde32f285a1f86fdcb2c9e50c'
+archive_format='thothii-qdrant-backup-v1'
+volume_role='qdrant-data'
+
 usage() {
   echo "usage: $0 --project-name NAME --output FILE" >&2
   exit 2
@@ -23,37 +27,43 @@ output_dir=$(dirname "$output")
 output_name=$(basename "$output")
 [ -d "$output_dir" ] || { echo "backup destination directory does not exist" >&2; exit 2; }
 
+expected_volume_name="${project_name}_${volume_role}"
+
 resolve_volume() {
   names=$(docker volume ls \
     --filter "label=com.docker.compose.project=$project_name" \
-    --filter "label=com.docker.compose.volume=qdrant-data" \
+    --filter "label=com.docker.compose.volume=$volume_role" \
     --format '{{.Name}}')
   count=$(printf '%s\n' "$names" | sed '/^$/d' | wc -l | tr -d ' ')
   [ "$count" -eq 1 ] || {
     echo "expected exactly one qdrant-data volume for compose project $project_name" >&2
     exit 2
   }
-  printf '%s\n' "$names" | sed -n '/./{p;q;}'
+  resolved=$(printf '%s\n' "$names" | sed -n '/./{p;q;}')
+  [ "$resolved" = "$expected_volume_name" ] || {
+    echo "unexpected qdrant-data volume name: $resolved" >&2
+    exit 2
+  }
+  printf '%s\n' "$resolved"
 }
 
-resolve_mountpoint() {
-  mountpoint=$(docker volume inspect --format '{{ .Mountpoint }}' "$1")
-  [ -n "$mountpoint" ] || { echo "docker did not return a qdrant-data mountpoint" >&2; exit 2; }
-  case "$mountpoint" in
-    /*) ;;
-    *) echo "qdrant-data mountpoint is not absolute: $mountpoint" >&2; exit 2 ;;
-  esac
-  [ -d "$mountpoint" ] || { echo "qdrant-data mountpoint is not a directory: $mountpoint" >&2; exit 2; }
-  printf '%s\n' "$mountpoint"
+validate_volume_metadata() {
+  metadata=$(docker volume inspect --format '{{ .Name }} {{ index .Labels "com.docker.compose.project" }} {{ index .Labels "com.docker.compose.volume" }}' "$1")
+  set -- $metadata
+  [ "${1-}" = "$expected_volume_name" ] || { echo "volume metadata name mismatch" >&2; exit 2; }
+  [ "${2-}" = "$project_name" ] || { echo "volume metadata project label mismatch" >&2; exit 2; }
+  [ "${3-}" = "$volume_role" ] || { echo "volume metadata role label mismatch" >&2; exit 2; }
 }
 
 volume_name=$(resolve_volume)
-mountpoint=$(resolve_mountpoint "$volume_name")
+validate_volume_metadata "$volume_name"
+
 running_container=$(docker compose --project-name "$project_name" ps --status running -q qdrant)
 restart_qdrant=0
+temporary_output=
 cleanup() {
   status=$?
-  if [ "${temporary_output:-}" ] && [ -e "${temporary_output:-}" ]; then
+  if [ -n "${temporary_output:-}" ] && [ -e "${temporary_output:-}" ]; then
     rm -f "$temporary_output"
   fi
   if [ "$restart_qdrant" -eq 1 ]; then
@@ -70,7 +80,35 @@ fi
 
 umask 077
 temporary_output=$(mktemp "$output_dir/.${output_name}.tmp.XXXXXX")
-tar -C "$mountpoint" -cf "$temporary_output" .
+created_utc=$(date -u +"%Y-%m-%dT%H:%M:%SZ")
+
+docker run --rm \
+  --mount "type=volume,src=$volume_name,dst=/qdrant-data,readonly" \
+  --mount "type=bind,src=$output_dir,dst=/backup" \
+  "$helper_image" \
+  /bin/sh -eu -c '
+    archive_format=$1
+    project_name=$2
+    volume_name=$3
+    volume_role=$4
+    helper_image=$5
+    created_utc=$6
+    output_name=$7
+    staging=$(mktemp -d /tmp/qdrant-backup.XXXXXX)
+    trap "rm -rf \"$staging\"" EXIT HUP INT TERM
+    mkdir -p "$staging/payload"
+    cat >"$staging/manifest.env" <<EOF
+format=$archive_format
+project_name=$project_name
+volume_name=$volume_name
+volume_role=$volume_role
+helper_image=$helper_image
+created_utc=$created_utc
+    EOF
+    cp -R /qdrant-data/. "$staging/payload"/
+    tar -C "$staging" -cf "/backup/$output_name" manifest.env payload
+  ' sh "$archive_format" "$project_name" "$volume_name" "$volume_role" "$helper_image" "$created_utc" "$(basename "$temporary_output")"
+
 if ! ln "$temporary_output" "$output"; then
   echo "refusing to replace backup destination created concurrently: $output" >&2
   exit 2