Pleroma/lib/pleroma/workers/attachments_cleanup_worker.ex

110 lines
3.1 KiB
Elixir
Raw Normal View History

2020-01-19 17:45:20 +01:00
# Pleroma: A lightweight social networking server
# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
2020-01-19 17:45:20 +01:00
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Workers.AttachmentsCleanupWorker do
import Ecto.Query
alias Pleroma.Object
alias Pleroma.Repo
use Pleroma.Workers.WorkerHelper, queue: "attachments_cleanup"
@impl Oban.Worker
2020-06-23 14:09:01 +02:00
def perform(%Job{
args: %{
"op" => "cleanup_attachments",
"object" => %{"data" => %{"attachment" => [_ | _] = attachments, "actor" => actor}}
2020-06-23 14:09:01 +02:00
}
}) 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
2020-06-14 20:02:57 +02:00
{:ok, :success}
end
2020-01-19 17:45:20 +01:00
2020-06-23 14:09:01 +02:00
def perform(%Job{args: %{"op" => "cleanup_attachments", "object" => _object}}), do: {:ok, :skip}
2020-06-14 20:02:57 +02:00
defp do_clean({object_ids, attachment_urls}) do
2020-01-19 17:45:20 +01:00
uploader = Pleroma.Config.get([Pleroma.Upload, :uploader])
2020-05-15 20:34:46 +02:00
base_url =
String.trim_trailing(
Pleroma.Upload.base_url(),
2020-05-15 20:34:46 +02:00
"/"
)
2020-06-14 20:02:57 +02:00
Enum.each(attachment_urls, fn href ->
href
|> String.trim_leading("#{base_url}")
2020-06-14 20:02:57 +02:00
|> uploader.delete_file()
end)
2020-05-15 20:34:46 +02:00
2020-06-14 20:02:57 +02:00
delete_objects(object_ids)
end
2020-05-15 20:34:46 +02:00
2020-06-14 20:02:57 +02:00
defp delete_objects([_ | _] = object_ids) do
Repo.delete_all(from(o in Object, where: o.id in ^object_ids))
end
2020-05-15 20:34:46 +02:00
2020-06-14 20:02:57 +02:00
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
2020-06-15 19:47:02 +02:00
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
2020-06-14 20:02:57 +02:00
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)
2020-01-19 17:45:20 +01:00
end
2020-06-15 19:47:02 +02:00
defp fetch_objects(hrefs) do
2020-06-14 20:02:57 +02:00
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
2020-01-19 17:45:20 +01:00
end