5 files changed, 58 insertions, 5 deletions
diff --git a/lib/pleroma/web/activity_pub/activity_pub.ex b/lib/pleroma/web/activity_pub/activity_pub.ex
index 66a9f78a3..5aac3f53b 100644
--- a/lib/pleroma/web/activity_pub/activity_pub.ex
+++ b/lib/pleroma/web/activity_pub/activity_pub.ex
@@ -767,7 +767,7 @@ defmodule Pleroma.Web.ActivityPub.ActivityPub do
   end
 
   defp restrict_replies(query, %{
-         reply_filtering_user: user,
+         reply_filtering_user: %User{} = user,
          reply_visibility: "self"
        }) do
     from(
@@ -783,7 +783,7 @@ defmodule Pleroma.Web.ActivityPub.ActivityPub do
   end
 
   defp restrict_replies(query, %{
-         reply_filtering_user: user,
+         reply_filtering_user: %User{} = user,
          reply_visibility: "following"
        }) do
     from(
diff --git a/lib/pleroma/web/activity_pub/mrf/simple_policy.ex b/lib/pleroma/web/activity_pub/mrf/simple_policy.ex
index bb193475a..161177727 100644
--- a/lib/pleroma/web/activity_pub/mrf/simple_policy.ex
+++ b/lib/pleroma/web/activity_pub/mrf/simple_policy.ex
@@ -66,7 +66,8 @@ defmodule Pleroma.Web.ActivityPub.MRF.SimplePolicy do
            "type" => "Create",
            "object" => child_object
          } = object
-       ) do
+       )
+       when is_map(child_object) do
     media_nsfw =
       Config.get([:mrf_simple, :media_nsfw])
       |> MRF.subdomains_regex()
diff --git a/lib/pleroma/web/activity_pub/transmogrifier.ex b/lib/pleroma/web/activity_pub/transmogrifier.ex
index af4384213..8fe430644 100644
--- a/lib/pleroma/web/activity_pub/transmogrifier.ex
+++ b/lib/pleroma/web/activity_pub/transmogrifier.ex
@@ -309,7 +309,7 @@ defmodule Pleroma.Web.ActivityPub.Transmogrifier do
   def fix_emoji(%{"tag" => tags} = object) when is_list(tags) do
     emoji =
       tags
-      |> Enum.filter(fn data -> data["type"] == "Emoji" and data["icon"] end)
+      |> Enum.filter(fn data -> is_map(data) and data["type"] == "Emoji" and data["icon"] end)
       |> Enum.reduce(%{}, fn data, mapping ->
         name = String.trim(data["name"], ":")
 
diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex
index bd7f03cbe..d7a19df4a 100644
--- a/lib/pleroma/web/rich_media/helpers.ex
+++ b/lib/pleroma/web/rich_media/helpers.ex
@@ -87,6 +87,50 @@ defmodule Pleroma.Web.RichMedia.Helpers do
   def rich_media_get(url) do
     headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
 
-    Pleroma.HTTP.get(url, headers, @options)
+    head_check =
+      case Pleroma.HTTP.head(url, headers, @options) do
+        # If the HEAD request didn't reach the server for whatever reason,
+        # we assume the GET that comes right after won't either
+        {:error, _} = e ->
+          e
+
+        {:ok, %Tesla.Env{status: 200, headers: headers}} ->
+          with :ok <- check_content_type(headers),
+               :ok <- check_content_length(headers),
+               do: :ok
+
+        _ ->
+          :ok
+      end
+
+    with :ok <- head_check, do: Pleroma.HTTP.get(url, headers, @options)
+  end
+
+  defp check_content_type(headers) do
+    case List.keyfind(headers, "content-type", 0) do
+      {_, content_type} ->
+        case Plug.Conn.Utils.media_type(content_type) do
+          {:ok, "text", "html", _} -> :ok
+          _ -> {:error, {:content_type, content_type}}
+        end
+
+      _ ->
+        :ok
+    end
+  end
+
+  @max_body @options[:max_body]
+  defp check_content_length(headers) do
+    case List.keyfind(headers, "content-length", 0) do
+      {_, maybe_content_length} ->
+        case Integer.parse(maybe_content_length) do
+          {content_length, ""} when content_length <= @max_body -> :ok
+          {_, ""} -> {:error, :body_too_large}
+          _ -> :ok
+        end
+
+      _ ->
+        :ok
+    end
   end
 end
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index 5727fda18..33f6f1fa1 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -36,6 +36,14 @@ defmodule Pleroma.Web.RichMedia.Parser do
         {:ok, _data} = res ->
           res
 
+        {:error, :body_too_large} = e ->
+          e
+
+        {:error, {:content_type, _}} = e ->
+          e
+
+        # The TTL is not set for the errors above, since they are unlikely to change
+        # with time
         {:error, _} = e ->
           ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
           Cachex.expire(:rich_media_cache, url, ttl)