X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fhelpers.ex;h=566fc8c8a91877a52448dba857e3abf49570064e;hb=ba512cbea42fc0a628d74d5680f0b34c3b1f1b5f;hp=bd7f03cbeba8c3bf55bf86341e47440ae17464d9;hpb=c4061f06ba26656a6a0e61f68c5f097930a4b054;p=akkoma

diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex
index bd7f03cbe..566fc8c8a 100644
--- a/lib/pleroma/web/rich_media/helpers.ex
+++ b/lib/pleroma/web/rich_media/helpers.ex
@@ -57,7 +57,6 @@ defmodule Pleroma.Web.RichMedia.Helpers do
 
   def fetch_data_for_object(object) do
     with true <- Config.get([:rich_media, :enabled]),
-         false <- object.data["sensitive"] || false,
          {:ok, page_url} <-
            HTML.extract_first_external_url_from_object(object),
          :ok <- validate_page_url(page_url),
@@ -70,7 +69,7 @@ defmodule Pleroma.Web.RichMedia.Helpers do
 
   def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
     with true <- Config.get([:rich_media, :enabled]),
-         %Object{} = object <- Object.normalize(activity) do
+         %Object{} = object <- Object.normalize(activity, fetch: false) do
       fetch_data_for_object(object)
     else
       _ -> %{}
@@ -79,14 +78,53 @@ defmodule Pleroma.Web.RichMedia.Helpers do
 
   def fetch_data_for_activity(_), do: %{}
 
-  def perform(:fetch, %Activity{} = activity) do
-    fetch_data_for_activity(activity)
-    :ok
-  end
-
   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