ThothII/scripts/vector-restore.sh

#!/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 --input FILE --confirm-project NAME" >&2
  exit 2
}

project_name=
input=
confirm_project=
while [ "$#" -gt 0 ]; do
  case "$1" in
    --project-name) project_name=${2-}; shift 2 ;;
    --input) input=${2-}; shift 2 ;;
    --confirm-project) confirm_project=${2-}; shift 2 ;;
    *) usage ;;
  esac
done

[ -n "$project_name" ] && [ -n "$input" ] && [ -n "$confirm_project" ] || usage
[ "$confirm_project" = "$project_name" ] || {
  echo "restore confirmation must match --project-name exactly" >&2
  exit 2
}
[ -r "$input" ] || { echo "backup input is not readable" >&2; exit 2; }

input_dir=$(dirname "$input")
input_name=$(basename "$input")
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=$volume_role" \
    --format '{{.Name}}')
  count=$(printf '%s\n' "$names" | sed '/^$/d' | wc -l | tr -d ' ')
  [ "$count" -eq 1 ] || {
    echo "expected exactly one qdrant-data volume for compose project $project_name" >&2
    exit 2
  }
  resolved=$(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"
}

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; }
}

validate_archive_paths() {
  paths=$(tar -tf "$input") || {
    echo "archive listing failed" >&2
    exit 2
  }
  saw_manifest=0
  saw_payload=0
  while IFS= read -r path; do
    [ -n "$path" ] || continue
    case "$path" in
      /*)
        echo "archive contains absolute paths" >&2
        exit 2
        ;;
    esac
    case "/$path/" in
      */../*|*/..//*)
        echo "archive contains parent traversal" >&2
        exit 2
        ;;
    esac
    case "$path" in
      manifest.env)
        saw_manifest=$((saw_manifest + 1))
        ;;
      payload|payload/*)
        case "$path" in
          payload/*) saw_payload=1 ;;
        esac
        ;;
      *)
        echo "archive contains unexpected top-level layout" >&2
        exit 2
        ;;
    esac
  done <<EOF
$paths
EOF
  [ "$saw_manifest" -eq 1 ] || { echo "archive manifest is missing or duplicated" >&2; exit 2; }
  [ "$saw_payload" -eq 1 ] || { echo "archive payload is missing" >&2; exit 2; }
}

validate_archive_types() {
  tar -tvf "$input" | while IFS= read -r entry; do
    [ -n "$entry" ] || continue
    type=$(printf '%.1s' "$entry")
    case "$type" in
      -|d) ;;
      l|h)
        echo "archive contains symlink or hardlink entries" >&2
        exit 2
        ;;
      b|c|p|s)
        echo "archive contains device or special-file entries" >&2
        exit 2
        ;;
      *)
        echo "archive contains unsupported entry types" >&2
        exit 2
        ;;
    esac
  done
}

validate_archive_manifest() {
  manifest=$(tar -xOf "$input" manifest.env 2>/dev/null) || {
    echo "archive manifest could not be read" >&2
    exit 2
  }
  format=
  manifest_project=
  manifest_volume=
  manifest_role=
  manifest_helper=
  created_utc=
  seen=
  while IFS='=' read -r key value; do
    [ -n "$key" ] || continue
    case "$key" in
      format) format=$value ;;
      project_name) manifest_project=$value ;;
      volume_name) manifest_volume=$value ;;
      volume_role) manifest_role=$value ;;
      helper_image) manifest_helper=$value ;;
      created_utc) created_utc=$value ;;
      *)
        echo "archive manifest contains unexpected fields" >&2
        exit 2
        ;;
    esac
    seen="$seen $key"
  done <<EOF
$manifest
EOF
  [ "$format" = "$archive_format" ] || { echo "archive format is unsupported" >&2; exit 2; }
  [ "$manifest_project" = "$project_name" ] || { echo "archive project does not match restore target" >&2; exit 2; }
  [ "$manifest_volume" = "$expected_volume_name" ] || { echo "archive volume does not match restore target" >&2; exit 2; }
  [ "$manifest_role" = "$volume_role" ] || { echo "archive volume role is invalid" >&2; exit 2; }
  [ "$manifest_helper" = "$helper_image" ] || { echo "archive helper image is invalid" >&2; exit 2; }
  [ -n "$created_utc" ] || { echo "archive manifest is incomplete" >&2; exit 2; }
}

validate_archive_paths
validate_archive_types
validate_archive_manifest

volume_name=$(resolve_volume)
validate_volume_metadata "$volume_name"

running_container=$(docker compose --project-name "$project_name" ps --status running -q qdrant)
restart_qdrant=0
cleanup() {
  status=$?
  if [ "$restart_qdrant" -eq 1 ]; then
    docker compose --project-name "$project_name" start qdrant >/dev/null
  fi
  exit "$status"
}
trap cleanup EXIT HUP INT TERM

if [ -n "$running_container" ]; then
  docker compose --project-name "$project_name" stop qdrant >/dev/null
  restart_qdrant=1
fi

docker run --rm \
  --mount "type=volume,src=$volume_name,dst=/qdrant-data" \
  --mount "type=bind,src=$input_dir,dst=/restore-backup,readonly" \
  "$helper_image" \
  /bin/sh -eu -c '
    input_name=$1
    rollback=$(mktemp -d /tmp/qdrant-rollback.XXXXXX)
    extracted=$(mktemp -d /tmp/qdrant-extract.XXXXXX)
    restore_original() {
      find /qdrant-data -mindepth 1 -maxdepth 1 -exec rm -rf -- {} +
      cp -R "$rollback"/. /qdrant-data/
    }
    cleanup() {
      status=$?
      if [ "$status" -ne 0 ] && [ -d "$rollback" ]; then
        restore_original
      fi
      rm -rf "$rollback" "$extracted"
      exit "$status"
    }
    trap cleanup EXIT HUP INT TERM
    cp -R /qdrant-data/. "$rollback"/
    tar -C "$extracted" -xf "/restore-backup/$input_name"
    find /qdrant-data -mindepth 1 -maxdepth 1 -exec rm -rf -- {} +
    cp -R "$extracted/payload"/. /qdrant-data/
  ' sh "$input_name"

echo "Qdrant restore completed into $volume_name for compose project $project_name"