Merge remote-tracking branch 'remotes/origin/develop' into 2168-media-preview-proxy

[akkoma] / lib / pleroma / reverse_proxy / reverse_proxy.ex

diff --git a/lib/pleroma/reverse_proxy/reverse_proxy.ex b/lib/pleroma/reverse_proxy/reverse_proxy.ex
index 9f5710c92ec60d5ff8e34ff1b381a941ad1d252a..613edf5658933c33688ac3c81b703f2f805245b3 100644 (file)
--- a/lib/pleroma/reverse_proxy/reverse_proxy.ex
+++ b/lib/pleroma/reverse_proxy/reverse_proxy.ex
@@ -1,14 +1,15 @@
 # Pleroma: A lightweight social networking server
-# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.ReverseProxy do
+  @range_headers ~w(range if-range)
   @keep_req_headers ~w(accept user-agent accept-encoding cache-control if-modified-since) ++
-                      ~w(if-unmodified-since if-none-match if-range range)
-  @resp_cache_headers ~w(etag date last-modified cache-control)
+                      ~w(if-unmodified-since if-none-match) ++ @range_headers
+  @resp_cache_headers ~w(etag date last-modified)
   @keep_resp_headers @resp_cache_headers ++
-                       ~w(content-type content-disposition content-encoding content-range) ++
-                       ~w(accept-ranges vary)
+                       ~w(content-length content-type content-disposition content-encoding) ++
+                       ~w(content-range accept-ranges vary)
   @default_cache_control_header "public, max-age=1209600"
   @valid_resp_codes [200, 206, 304]
   @max_read_duration :timer.seconds(30)
@@ -16,6 +17,8 @@ defmodule Pleroma.ReverseProxy do
   @failed_request_ttl :timer.seconds(60)
   @methods ~w(GET HEAD)
 
+  def max_read_duration_default, do: @max_read_duration
+
   @moduledoc """
   A reverse proxy.
 
@@ -32,9 +35,6 @@ defmodule Pleroma.ReverseProxy do
   * request: `#{inspect(@keep_req_headers)}`
   * response: `#{inspect(@keep_resp_headers)}`
 
-  If no caching headers (`#{inspect(@resp_cache_headers)}`) are returned by upstream, `cache-control` will be
-  set to `#{inspect(@default_cache_control_header)}`.
-
   Options:
 
   * `redirect_on_failure` (default `false`). Redirects the client to the real remote URL if there's any HTTP
@@ -59,7 +59,7 @@ defmodule Pleroma.ReverseProxy do
 
   * `req_headers`, `resp_headers` additional headers.
 
-  * `http`: options for [gun](https://github.com/ninenines/gun).
+  * `http`: options for [hackney](https://github.com/benoitc/hackney) or [gun](https://github.com/ninenines/gun).
 
   """
   @default_options [pool: :media]
@@ -173,6 +173,8 @@ defmodule Pleroma.ReverseProxy do
   end
 
   defp response(conn, client, url, status, headers, opts) do
+    Logger.debug("#{__MODULE__} #{status} #{url} #{inspect(headers)}")
+
     result =
       conn
       |> put_resp_headers(build_resp_headers(headers, opts))
@@ -223,7 +225,9 @@ defmodule Pleroma.ReverseProxy do
     end
   end
 
-  defp head_response(conn, _url, code, headers, opts) do
+  defp head_response(conn, url, code, headers, opts) do
+    Logger.debug("#{__MODULE__} #{code} #{url} #{inspect(headers)}")
+
     conn
     |> put_resp_headers(build_resp_headers(headers, opts))
     |> send_resp(code, "")
@@ -265,20 +269,33 @@ defmodule Pleroma.ReverseProxy do
     headers
     |> downcase_headers()
     |> Enum.filter(fn {k, _} -> k in @keep_req_headers end)
-    |> (fn headers ->
-          headers = headers ++ Keyword.get(opts, :req_headers, [])
-
-          if Keyword.get(opts, :keep_user_agent, false) do
-            List.keystore(
-              headers,
-              "user-agent",
-              0,
-              {"user-agent", Pleroma.Application.user_agent()}
-            )
-          else
-            headers
-          end
-        end).()
+    |> build_req_range_or_encoding_header(opts)
+    |> build_req_user_agent_header(opts)
+    |> Keyword.merge(Keyword.get(opts, :req_headers, []))
+  end
+
+  # Disable content-encoding if any @range_headers are requested (see #1823).
+  defp build_req_range_or_encoding_header(headers, _opts) do
+    range? = Enum.any?(headers, fn {header, _} -> Enum.member?(@range_headers, header) end)
+
+    if range? && List.keymember?(headers, "accept-encoding", 0) do
+      List.keydelete(headers, "accept-encoding", 0)
+    else
+      headers
+    end
+  end
+
+  defp build_req_user_agent_header(headers, opts) do
+    if Keyword.get(opts, :keep_user_agent, false) do
+      List.keystore(
+        headers,
+        "user-agent",
+        0,
+        {"user-agent", Pleroma.Application.user_agent()}
+      )
+    else
+      headers
+    end
   end
 
   defp build_resp_headers(headers, opts) do
@@ -286,21 +303,22 @@ defmodule Pleroma.ReverseProxy do
     |> Enum.filter(fn {k, _} -> k in @keep_resp_headers end)
     |> build_resp_cache_headers(opts)
     |> build_resp_content_disposition_header(opts)
-    |> (fn headers -> headers ++ Keyword.get(opts, :resp_headers, []) end).()
+    |> Keyword.merge(Keyword.get(opts, :resp_headers, []))
   end
 
   defp build_resp_cache_headers(headers, _opts) do
     has_cache? = Enum.any?(headers, fn {k, _} -> k in @resp_cache_headers end)
-    has_cache_control? = List.keymember?(headers, "cache-control", 0)
 
     cond do
-      has_cache? && has_cache_control? ->
-        headers
-
       has_cache? ->
-        # There's caching header present but no cache-control -- we need to explicitely override it
-        # to public as Plug defaults to "max-age=0, private, must-revalidate"
-        List.keystore(headers, "cache-control", 0, {"cache-control", "public"})
+        # There's caching header present but no cache-control -- we need to set our own
+        # as Plug defaults to "max-age=0, private, must-revalidate"
+        List.keystore(
+          headers,
+          "cache-control",
+          0,
+          {"cache-control", @default_cache_control_header}
+        )
 
       true ->
         List.keystore(
@@ -372,6 +390,8 @@ defmodule Pleroma.ReverseProxy do
 
   defp body_size_constraint(_, _), do: :ok
 
+  defp check_read_duration(nil = _duration, max), do: check_read_duration(@max_read_duration, max)
+
   defp check_read_duration(duration, max)
        when is_integer(duration) and is_integer(max) and max > 0 do
     if duration > max do