X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Finstances%2Finstance.ex;h=8bf53c090a8cc07a7df4ddb775af4cb8121535a3;hb=a0b8e3c84280c11ab43b7660414b64b55c1b7259;hp=544c4b687c8d5e0a9c13a3247fa8e731609b0188;hpb=a0f101ee806af06bcd4271cd8d57d11ff85ea11a;p=akkoma diff --git a/lib/pleroma/instances/instance.ex b/lib/pleroma/instances/instance.ex index 544c4b687..5ee408f21 100644 --- a/lib/pleroma/instances/instance.ex +++ b/lib/pleroma/instances/instance.ex @@ -1,22 +1,31 @@ # Pleroma: A lightweight social networking server -# Copyright © 2017-2019 Pleroma Authors +# Copyright © 2017-2021 Pleroma Authors # SPDX-License-Identifier: AGPL-3.0-only defmodule Pleroma.Instances.Instance do @moduledoc "Instance." + @cachex Pleroma.Config.get([:cachex, :provider], Cachex) + alias Pleroma.Instances alias Pleroma.Instances.Instance alias Pleroma.Repo + alias Pleroma.User + alias Pleroma.Workers.BackgroundWorker use Ecto.Schema import Ecto.Query import Ecto.Changeset + require Logger + schema "instances" do field(:host, :string) field(:unreachable_since, :naive_datetime_usec) + field(:favicon, :string) + field(:metadata_updated_at, :naive_datetime) + field(:nodeinfo, :map, default: %{}) timestamps() end @@ -25,7 +34,7 @@ defmodule Pleroma.Instances.Instance do def changeset(struct, params \\ %{}) do struct - |> cast(params, [:host, :unreachable_since]) + |> cast(params, [:host, :unreachable_since, :favicon, :nodeinfo, :metadata_updated_at]) |> validate_required([:host]) |> unique_constraint(:host) end @@ -73,7 +82,7 @@ defmodule Pleroma.Instances.Instance do ) end - def reachable?(_), do: true + def reachable?(url_or_host) when is_binary(url_or_host), do: true def set_reachable(url_or_host) when is_binary(url_or_host) do with host <- host(url_or_host), @@ -115,9 +124,197 @@ defmodule Pleroma.Instances.Instance do def set_unreachable(_, _), do: {:error, nil} + def get_consistently_unreachable do + reachability_datetime_threshold = Instances.reachability_datetime_threshold() + + from(i in Instance, + where: ^reachability_datetime_threshold > i.unreachable_since, + order_by: i.unreachable_since, + select: {i.host, i.unreachable_since} + ) + |> Repo.all() + end + defp parse_datetime(datetime) when is_binary(datetime) do NaiveDateTime.from_iso8601(datetime) end defp parse_datetime(datetime), do: datetime + + def needs_update(nil), do: true + + def needs_update(%Instance{metadata_updated_at: nil}), do: true + + def needs_update(%Instance{metadata_updated_at: metadata_updated_at}) do + now = NaiveDateTime.utc_now() + NaiveDateTime.diff(now, metadata_updated_at) > 86_400 + end + + def local do + %Instance{ + host: Pleroma.Web.Endpoint.host(), + favicon: Pleroma.Web.Endpoint.url() <> "/favicon.png", + nodeinfo: Pleroma.Web.Nodeinfo.NodeinfoController.raw_nodeinfo() + } + end + + def update_metadata(%URI{host: host} = uri) do + Logger.info("Checking metadata for #{host}") + existing_record = Repo.get_by(Instance, %{host: host}) + + if reachable?(host) do + do_update_metadata(uri, existing_record) + else + {:discard, :unreachable} + end + end + + defp do_update_metadata(%URI{host: host} = uri, existing_record) do + if existing_record do + if needs_update(existing_record) do + Logger.info("Updating metadata for #{host}") + favicon = scrape_favicon(uri) + nodeinfo = scrape_nodeinfo(uri) + + existing_record + |> changeset(%{ + host: host, + favicon: favicon, + nodeinfo: nodeinfo, + metadata_updated_at: NaiveDateTime.utc_now() + }) + |> Repo.update() + else + {:discard, "Does not require update"} + end + else + favicon = scrape_favicon(uri) + nodeinfo = scrape_nodeinfo(uri) + + Logger.info("Creating metadata for #{host}") + + {:ok, instance} = + %Instance{} + |> changeset(%{ + host: host, + favicon: favicon, + nodeinfo: nodeinfo, + metadata_updated_at: NaiveDateTime.utc_now() + }) + |> Repo.insert() + end + end + + def get_favicon(%URI{host: host}) do + existing_record = Repo.get_by(Instance, %{host: host}) + + if existing_record do + existing_record.favicon + else + nil + end + end + + defp scrape_nodeinfo(%URI{} = instance_uri) do + with true <- Pleroma.Config.get([:instances_nodeinfo, :enabled]), + {_, true} <- {:reachable, reachable?(instance_uri.host)}, + {:ok, %Tesla.Env{status: 200, body: body}} <- + Tesla.get( + "https://#{instance_uri.host}/.well-known/nodeinfo", + headers: [{"Accept", "application/json"}] + ), + {:ok, json} <- Jason.decode(body), + {:ok, %{"links" => links}} <- {:ok, json}, + {:ok, %{"href" => href}} <- + {:ok, + Enum.find(links, &(&1["rel"] == "http://nodeinfo.diaspora.software/ns/schema/2.0"))}, + {:ok, %Tesla.Env{body: data}} <- + Pleroma.HTTP.get(href, [{"accept", "application/json"}], []), + {:length, true} <- {:length, String.length(data) < 50_000}, + {:ok, nodeinfo} <- Jason.decode(data) do + nodeinfo + else + {:reachable, false} -> + Logger.debug( + "Instance.scrape_nodeinfo(\"#{to_string(instance_uri)}\") ignored unreachable host" + ) + + nil + + {:length, false} -> + Logger.debug( + "Instance.scrape_nodeinfo(\"#{to_string(instance_uri)}\") ignored too long body" + ) + + nil + + _ -> + nil + end + end + + defp scrape_favicon(%URI{} = instance_uri) do + with true <- Pleroma.Config.get([:instances_favicons, :enabled]), + {_, true} <- {:reachable, reachable?(instance_uri.host)}, + {:ok, %Tesla.Env{body: html}} <- + Pleroma.HTTP.get(to_string(instance_uri), [{"accept", "text/html"}], []), + {_, [favicon_rel | _]} when is_binary(favicon_rel) <- + {:parse, html |> Floki.parse_document!() |> Floki.attribute("link[rel=icon]", "href")}, + {_, favicon} when is_binary(favicon) <- + {:merge, URI.merge(instance_uri, favicon_rel) |> to_string()}, + {:length, true} <- {:length, String.length(favicon) < 255} do + favicon + else + {:reachable, false} -> + Logger.debug( + "Instance.scrape_favicon(\"#{to_string(instance_uri)}\") ignored unreachable host" + ) + + nil + + _ -> + nil + end + end + + @doc """ + Deletes all users from an instance in a background task, thus also deleting + all of those users' activities and notifications. + """ + def delete_users_and_activities(host) when is_binary(host) do + BackgroundWorker.enqueue("delete_instance", %{"host" => host}) + end + + def perform(:delete_instance, host) when is_binary(host) do + User.Query.build(%{nickname: "@#{host}"}) + |> Repo.chunk_stream(100, :batches) + |> Stream.each(fn users -> + users + |> Enum.each(fn user -> + User.perform(:delete, user) + end) + end) + |> Stream.run() + end + + def get_by_url(url_or_host) do + url = host(url_or_host) + Repo.get_by(Instance, host: url) + end + + def get_cached_by_url(url_or_host) do + url = host(url_or_host) + + if url == Pleroma.Web.Endpoint.host() do + {:ok, local()} + else + @cachex.fetch!(:instances_cache, "instances:#{url}", fn _ -> + with %Instance{} = instance <- get_by_url(url) do + {:commit, {:ok, instance}} + else + _ -> {:ignore, nil} + end + end) + end + end end