X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fworkers%2Fattachments_cleanup_worker.ex;h=f5090dae737b2131d8318ce23e0a58718fe13704;hb=ce517ff4e5a9cc144853ec8112bfac2a38a3a68a;hp=3c5820a866ab93fc83cc5abd0d39d53199cd2564;hpb=99a6c660a909d8c74289015b3f69357196256112;p=akkoma diff --git a/lib/pleroma/workers/attachments_cleanup_worker.ex b/lib/pleroma/workers/attachments_cleanup_worker.ex index 3c5820a86..f5090dae7 100644 --- a/lib/pleroma/workers/attachments_cleanup_worker.ex +++ b/lib/pleroma/workers/attachments_cleanup_worker.ex @@ -1,5 +1,5 @@ # Pleroma: A lightweight social networking server -# Copyright © 2017-2020 Pleroma Authors +# Copyright © 2017-2021 Pleroma Authors # SPDX-License-Identifier: AGPL-3.0-only defmodule Pleroma.Workers.AttachmentsCleanupWorker do @@ -11,85 +11,99 @@ defmodule Pleroma.Workers.AttachmentsCleanupWorker do use Pleroma.Workers.WorkerHelper, queue: "attachments_cleanup" @impl Oban.Worker - def perform( - %{ + def perform(%Job{ + args: %{ "op" => "cleanup_attachments", "object" => %{"data" => %{"attachment" => [_ | _] = attachments, "actor" => actor}} - }, - _job - ) do - hrefs = - Enum.flat_map(attachments, fn attachment -> - Enum.map(attachment["url"], & &1["href"]) - end) + } + }) do + if Pleroma.Config.get([:instance, :cleanup_attachments], false) do + attachments + |> Enum.flat_map(fn item -> Enum.map(item["url"], & &1["href"]) end) + |> fetch_objects + |> prepare_objects(actor, Enum.map(attachments, & &1["name"])) + |> filter_objects + |> do_clean + end + + {:ok, :success} + end - names = Enum.map(attachments, & &1["name"]) + def perform(%Job{args: %{"op" => "cleanup_attachments", "object" => _object}}), do: {:ok, :skip} + defp do_clean({object_ids, attachment_urls}) do uploader = Pleroma.Config.get([Pleroma.Upload, :uploader]) - # find all objects for copies of the attachments, name and actor doesn't matter here - delete_ids = - from(o in Object, - where: - fragment( - "to_jsonb(array(select jsonb_array_elements((?)#>'{url}') ->> 'href' where jsonb_typeof((?)#>'{url}') = 'array'))::jsonb \\?| (?)", - o.data, - o.data, - ^hrefs - ) + base_url = + String.trim_trailing( + Pleroma.Upload.base_url(), + "/" ) - # The query above can be time consumptive on large instances until we - # refactor how uploads are stored - |> Repo.all(timeout: :infinity) - # we should delete 1 object for any given attachment, but don't delete - # files if there are more than 1 object for it - |> Enum.reduce(%{}, fn %{ - id: id, - data: %{ - "url" => [%{"href" => href}], - "actor" => obj_actor, - "name" => name - } - }, - acc -> - Map.update(acc, href, %{id: id, count: 1}, fn val -> - case obj_actor == actor and name in names do - true -> - # set id of the actor's object that will be deleted - %{val | id: id, count: val.count + 1} - - false -> - # another actor's object, just increase count to not delete file - %{val | count: val.count + 1} - end - end) - end) - |> Enum.map(fn {href, %{id: id, count: count}} -> - # only delete files that have single instance - with 1 <- count do - prefix = - case Pleroma.Config.get([Pleroma.Upload, :base_url]) do - nil -> "media" - _ -> "" - end - - base_url = - String.trim_trailing( - Pleroma.Config.get([Pleroma.Upload, :base_url], Pleroma.Web.base_url()), - "/" - ) - - file_path = String.trim_leading(href, "#{base_url}/#{prefix}") - - uploader.delete_file(file_path) - end - id - end) + Enum.each(attachment_urls, fn href -> + href + |> String.trim_leading("#{base_url}") + |> uploader.delete_file() + end) + + delete_objects(object_ids) + end - from(o in Object, where: o.id in ^delete_ids) - |> Repo.delete_all() + defp delete_objects([_ | _] = object_ids) do + Repo.delete_all(from(o in Object, where: o.id in ^object_ids)) end - def perform(%{"op" => "cleanup_attachments", "object" => _object}, _job), do: :ok + defp delete_objects(_), do: :ok + + # we should delete 1 object for any given attachment, but don't delete + # files if there are more than 1 object for it + defp filter_objects(objects) do + Enum.reduce(objects, {[], []}, fn {href, %{id: id, count: count}}, {ids, hrefs} -> + with 1 <- count do + {ids ++ [id], hrefs ++ [href]} + else + _ -> {ids ++ [id], hrefs} + end + end) + end + + defp prepare_objects(objects, actor, names) do + objects + |> Enum.reduce(%{}, fn %{ + id: id, + data: %{ + "url" => [%{"href" => href}], + "actor" => obj_actor, + "name" => name + } + }, + acc -> + Map.update(acc, href, %{id: id, count: 1}, fn val -> + case obj_actor == actor and name in names do + true -> + # set id of the actor's object that will be deleted + %{val | id: id, count: val.count + 1} + + false -> + # another actor's object, just increase count to not delete file + %{val | count: val.count + 1} + end + end) + end) + end + + defp fetch_objects(hrefs) do + from(o in Object, + where: + fragment( + "to_jsonb(array(select jsonb_array_elements((?)#>'{url}') ->> 'href' where jsonb_typeof((?)#>'{url}') = 'array'))::jsonb \\?| (?)", + o.data, + o.data, + ^hrefs + ) + ) + # The query above can be time consumptive on large instances until we + # refactor how uploads are stored + |> Repo.all(timeout: :infinity) + end end