instance: Do not fetch unreachable instances

[akkoma] / lib / pleroma / instances / instance.ex

diff --git a/lib/pleroma/instances/instance.ex b/lib/pleroma/instances/instance.ex
index b97e229e5a04ff497cd4359f3b03fde1a1c2cefa..c9ca3aac717ec63546d450216109d36a904ffbdf 100644 (file)
--- a/lib/pleroma/instances/instance.ex
+++ b/lib/pleroma/instances/instance.ex
@@ -14,6 +14,8 @@ defmodule Pleroma.Instances.Instance do
   import Ecto.Query
   import Ecto.Changeset
 
+  require Logger
+
   schema "instances" do
     field(:host, :string)
     field(:unreachable_since, :naive_datetime_usec)
@@ -117,6 +119,17 @@ 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
@@ -127,50 +140,59 @@ defmodule Pleroma.Instances.Instance do
     existing_record = Repo.get_by(Instance, %{host: host})
     now = NaiveDateTime.utc_now()
 
-    if existing_record && existing_record.favicon &&
+    if existing_record && existing_record.favicon_updated_at &&
          NaiveDateTime.diff(now, existing_record.favicon_updated_at) < 86_400 do
       existing_record.favicon
     else
       favicon = scrape_favicon(instance_uri)
 
-      cond do
-        is_binary(favicon) && existing_record ->
-          existing_record
-          |> changeset(%{favicon: favicon, favicon_updated_at: now})
-          |> Repo.update()
-
-          favicon
-
-        is_binary(favicon) ->
-          %Instance{}
-          |> changeset(%{host: host, favicon: favicon, favicon_updated_at: now})
-          |> Repo.insert()
-
-          favicon
-
-        true ->
-          nil
+      if existing_record do
+        existing_record
+        |> changeset(%{favicon: favicon, favicon_updated_at: now})
+        |> Repo.update()
+      else
+        %Instance{}
+        |> changeset(%{host: host, favicon: favicon, favicon_updated_at: now})
+        |> Repo.insert()
       end
+
+      favicon
     end
+  rescue
+    e ->
+      Logger.warn("Instance.get_or_update_favicon(\"#{host}\") error: #{inspect(e)}")
+      nil
   end
 
   defp scrape_favicon(%URI{} = instance_uri) do
     try do
-      with {:ok, %Tesla.Env{body: html}} <-
-             Pleroma.HTTP.get(to_string(instance_uri), [{:Accept, "text/html"}]),
-           favicon_rel <-
-             html
-             |> Floki.parse_document!()
-             |> Floki.attribute("link[rel=icon]", "href")
-             |> List.first(),
-           favicon <- URI.merge(instance_uri, favicon_rel) |> to_string(),
-           true <- is_binary(favicon) do
+      with {_, true} <- {:reachable, reachable?(instance_uri.host)},
+           {:ok, %Tesla.Env{body: html}} <-
+             Pleroma.HTTP.get(to_string(instance_uri), [{"accept", "text/html"}], pool: :media),
+           {_, [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()} do
         favicon
       else
-        _ -> nil
+        {:reachable, false} ->
+          Logger.debug(
+            "Instance.scrape_favicon(\"#{to_string(instance_uri)}\") ignored unreachable host"
+          )
+
+          nil
+
+        _ ->
+          nil
       end
     rescue
-      _ -> nil
+      e ->
+        Logger.warn(
+          "Instance.scrape_favicon(\"#{to_string(instance_uri)}\") error: #{inspect(e)}"
+        )
+
+        nil
     end
   end
 end