From e1a2e8b17cca0d9f50b72fcea0ec5ffb8e613db1 Mon Sep 17 00:00:00 2001
From: "Haelwenn (lanodan) Monnier" <contact@hacktivis.me>
Date: Mon, 7 Dec 2020 20:09:34 +0100
Subject: [PATCH 1/2] instance: Do not fetch unreachable instances

Closes: https://git.pleroma.social/pleroma/pleroma/-/issues/2346
---
 lib/pleroma/instances/instance.ex        | 13 +++++++++++--
 test/pleroma/instances/instance_test.exs |  9 +++++++++
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/lib/pleroma/instances/instance.ex b/lib/pleroma/instances/instance.ex
index df471a39d..c9ca3aac7 100644
--- a/lib/pleroma/instances/instance.ex
+++ b/lib/pleroma/instances/instance.ex
@@ -166,7 +166,8 @@ defmodule Pleroma.Instances.Instance do
 
   defp scrape_favicon(%URI{} = instance_uri) do
     try do
-      with {:ok, %Tesla.Env{body: html}} <-
+      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,
@@ -175,7 +176,15 @@ defmodule Pleroma.Instances.Instance do
              {: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
       e ->
diff --git a/test/pleroma/instances/instance_test.exs b/test/pleroma/instances/instance_test.exs
index 4f0805100..2c6389e4f 100644
--- a/test/pleroma/instances/instance_test.exs
+++ b/test/pleroma/instances/instance_test.exs
@@ -3,6 +3,7 @@
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Instances.InstanceTest do
+  alias Pleroma.Instances
   alias Pleroma.Instances.Instance
   alias Pleroma.Repo
 
@@ -148,5 +149,13 @@ defmodule Pleroma.Instances.InstanceTest do
                         )
              end) =~ "Instance.scrape_favicon(\"https://no-favicon.example.org/\") error: "
     end
+
+    test "Doesn't scrapes unreachable instances" do
+      instance = insert(:instance, unreachable_since: Instances.reachability_datetime_threshold())
+      url = "https://" <> instance.host
+
+      assert capture_log(fn -> assert nil == Instance.get_or_update_favicon(URI.parse(url)) end) =~
+               "Instance.scrape_favicon(\"#{url}\") ignored unreachable host"
+    end
   end
 end

From 1403798820da21660fb8787ffaf9f54817597636 Mon Sep 17 00:00:00 2001
From: "Haelwenn (lanodan) Monnier" <contact@hacktivis.me>
Date: Mon, 7 Dec 2020 21:18:51 +0100
Subject: [PATCH 2/2] instance.reachable?: Limit to binary input

---
 lib/pleroma/instances/instance.ex | 2 +-
 test/pleroma/instances_test.exs   | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/lib/pleroma/instances/instance.ex b/lib/pleroma/instances/instance.ex
index c9ca3aac7..2e1696fe2 100644
--- a/lib/pleroma/instances/instance.ex
+++ b/lib/pleroma/instances/instance.ex
@@ -77,7 +77,7 @@ 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),
diff --git a/test/pleroma/instances_test.exs b/test/pleroma/instances_test.exs
index d2618025c..5d0ce6237 100644
--- a/test/pleroma/instances_test.exs
+++ b/test/pleroma/instances_test.exs
@@ -32,9 +32,9 @@ defmodule Pleroma.InstancesTest do
       assert Instances.reachable?(URI.parse(url).host)
     end
 
-    test "returns true on non-binary input" do
-      assert Instances.reachable?(nil)
-      assert Instances.reachable?(1)
+    test "raises FunctionClauseError exception on non-binary input" do
+      assert_raise FunctionClauseError, fn -> Instances.reachable?(nil) end
+      assert_raise FunctionClauseError, fn -> Instances.reachable?(1) end
     end
   end