akkoma/lib/pleroma/reverse_proxy.ex

344 lines
10 KiB
Elixir
Raw Normal View History

2018-11-23 16:40:45 +00:00
defmodule Pleroma.ReverseProxy do
2018-11-23 22:59:24 +00:00
@keep_req_headers ~w(accept user-agent accept-encoding cache-control if-modified-since if-unmodified-since if-none-match if-range range)
2018-11-23 16:40:45 +00:00
@resp_cache_headers ~w(etag date last-modified cache-control)
@keep_resp_headers @resp_cache_headers ++
2018-11-23 22:59:24 +00:00
~w(content-type content-disposition content-encoding content-range accept-ranges vary)
2018-11-23 16:40:45 +00:00
@default_cache_control_header "public, max-age=1209600"
@valid_resp_codes [200, 206, 304]
@max_read_duration :timer.minutes(2)
@max_body_length :infinity
@methods ~w(GET HEAD)
@moduledoc """
A reverse proxy.
Pleroma.ReverseProxy.call(conn, url, options)
It is not meant to be added into a plug pipeline, but to be called from another plug or controller.
Supports `#{inspect(@methods)}` HTTP methods, and only allows `#{inspect(@valid_resp_codes)}` status codes.
Responses are chunked to the client while downloading from the upstream.
Some request / responses headers are preserved:
* 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
errors. Any error during body processing will not be redirected as the response is chunked. This may expose
remote URL, clients IPs, .
* `max_body_length` (default `#{inspect(@max_body_length)}`): limits the content length to be approximately the
specified length. It is validated with the `content-length` header and also verified when proxying.
* `max_read_duration` (default `#{inspect(@max_read_duration)}` ms): the total time the connection is allowed to
read from the remote upstream.
* `inline_content_types`:
* `true` will not alter `content-disposition` (up to the upstream),
* `false` will add `content-disposition: attachment` to any request,
* a list of whitelisted content types
* `keep_user_agent` will forward the client's user-agent to the upstream. This may be useful if the upstream is
doing content transformation (encoding, ) depending on the request.
* `req_headers`, `resp_headers` additional headers.
* `http`: options for [hackney](https://github.com/benoitc/hackney).
"""
@hackney Application.get_env(:pleroma, :hackney, :hackney)
@httpoison Application.get_env(:pleroma, :httpoison, HTTPoison)
@default_hackney_options [{:follow_redirect, true}]
@inline_content_types [
"image/gif",
"image/jpeg",
"image/jpg",
"image/png",
"image/svg+xml",
"audio/mpeg",
"audio/mp3",
"video/webm",
"video/mp4",
"video/quicktime"
]
require Logger
import Plug.Conn
@type option() ::
{:keep_user_agent, boolean}
| {:max_read_duration, :timer.time() | :infinity}
| {:max_body_length, non_neg_integer() | :infinity}
| {:http, []}
| {:req_headers, [{String.t(), String.t()}]}
| {:resp_headers, [{String.t(), String.t()}]}
| {:inline_content_types, boolean() | [String.t()]}
| {:redirect_on_failure, boolean()}
@spec call(Plug.Conn.t(), url :: String.t(), [option()]) :: Plug.Conn.t()
def call(conn = %{method: method}, url, opts \\ []) when method in @methods do
hackney_opts =
@default_hackney_options
|> Keyword.merge(Keyword.get(opts, :http, []))
|> @httpoison.process_request_options()
req_headers = build_req_headers(conn.req_headers, opts)
opts =
if filename = Pleroma.Web.MediaProxy.filename(url) do
Keyword.put_new(opts, :attachment_name, filename)
else
opts
end
with {:ok, code, headers, client} <- request(method, url, req_headers, hackney_opts),
2018-11-29 20:11:45 +00:00
:ok <- header_length_constraint(headers, Keyword.get(opts, :max_body_length)) do
2018-11-23 16:40:45 +00:00
response(conn, client, url, code, headers, opts)
else
{:ok, code, headers} ->
head_response(conn, url, code, headers, opts)
|> halt()
{:error, {:invalid_http_response, code}} ->
Logger.error("#{__MODULE__}: request to #{inspect(url)} failed with HTTP status #{code}")
conn
|> error_or_redirect(
url,
code,
"Request failed: " <> Plug.Conn.Status.reason_phrase(code),
opts
)
|> halt()
{:error, error} ->
Logger.error("#{__MODULE__}: request to #{inspect(url)} failed: #{inspect(error)}")
conn
|> error_or_redirect(url, 500, "Request failed", opts)
|> halt()
end
end
def call(conn, _, _) do
conn
|> send_resp(400, Plug.Conn.Status.reason_phrase(400))
|> halt()
end
defp request(method, url, headers, hackney_opts) do
Logger.debug("#{__MODULE__} #{method} #{url} #{inspect(headers)}")
method = method |> String.downcase() |> String.to_existing_atom()
case @hackney.request(method, url, headers, "", hackney_opts) do
{:ok, code, headers, client} when code in @valid_resp_codes ->
{:ok, code, downcase_headers(headers), client}
{:ok, code, headers} when code in @valid_resp_codes ->
{:ok, code, downcase_headers(headers)}
{:ok, code, _, _} ->
{:error, {:invalid_http_response, code}}
{:error, error} ->
{:error, error}
end
end
defp response(conn, client, url, status, headers, opts) do
result =
conn
|> put_resp_headers(build_resp_headers(headers, opts))
|> send_chunked(status)
|> chunk_reply(client, opts)
case result do
{:ok, conn} ->
halt(conn)
{:error, :closed, conn} ->
:hackney.close(client)
halt(conn)
{:error, error, conn} ->
Logger.warn(
"#{__MODULE__} request to #{url} failed while reading/chunking: #{inspect(error)}"
)
:hackney.close(client)
halt(conn)
end
end
defp chunk_reply(conn, client, opts) do
chunk_reply(conn, client, opts, 0, 0)
end
defp chunk_reply(conn, client, opts, sent_so_far, duration) do
with {:ok, duration} <-
check_read_duration(
duration,
Keyword.get(opts, :max_read_duration, @max_read_duration)
),
{:ok, data} <- @hackney.stream_body(client),
{:ok, duration} <- increase_read_duration(duration),
sent_so_far = sent_so_far + byte_size(data),
:ok <- body_size_constraint(sent_so_far, Keyword.get(opts, :max_body_size)),
{:ok, conn} <- chunk(conn, data) do
chunk_reply(conn, client, opts, sent_so_far, duration)
else
:done -> {:ok, conn}
{:error, error} -> {:error, error, conn}
end
end
defp head_response(conn, _url, code, headers, opts) do
conn
|> put_resp_headers(build_resp_headers(headers, opts))
|> send_resp(code, "")
end
defp error_or_redirect(conn, url, code, body, opts) do
if Keyword.get(opts, :redirect_on_failure, false) do
conn
|> Phoenix.Controller.redirect(external: url)
|> halt()
else
conn
|> send_resp(code, body)
|> halt
end
end
defp downcase_headers(headers) do
Enum.map(headers, fn {k, v} ->
{String.downcase(k), v}
end)
end
defp get_content_type(headers) do
2018-11-23 22:59:24 +00:00
{_, content_type} =
List.keyfind(headers, "content-type", 0, {"content-type", "application/octet-stream"})
[content_type | _] = String.split(content_type, ";")
content_type
end
2018-11-23 16:40:45 +00:00
defp put_resp_headers(conn, headers) do
Enum.reduce(headers, conn, fn {k, v}, conn ->
put_resp_header(conn, k, v)
end)
end
defp build_req_headers(headers, opts) do
headers =
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).()
end
defp build_resp_headers(headers, opts) do
2018-11-23 22:59:24 +00:00
headers
|> 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).()
2018-11-23 16:40:45 +00:00
end
defp build_resp_cache_headers(headers, opts) do
has_cache? = Enum.any?(headers, fn {k, _} -> k in @resp_cache_headers end)
if has_cache? do
headers
else
List.keystore(headers, "cache-control", 0, {"cache-control", @default_cache_control_header})
end
end
defp build_resp_content_disposition_header(headers, opts) do
opt = Keyword.get(opts, :inline_content_types, @inline_content_types)
content_type = get_content_type(headers)
2018-11-23 16:40:45 +00:00
attachment? =
cond do
is_list(opt) && !Enum.member?(opt, content_type) -> true
opt == false -> true
true -> false
end
if attachment? do
disposition = "attachment; filename=" <> Keyword.get(opts, :attachment_name, "attachment")
List.keystore(headers, "content-disposition", 0, {"content-disposition", disposition})
else
headers
end
end
2018-11-29 20:11:45 +00:00
defp header_length_constraint(headers, limit) when is_integer(limit) and limit > 0 do
2018-11-23 16:40:45 +00:00
with {_, size} <- List.keyfind(headers, "content-length", 0),
{size, _} <- Integer.parse(size),
true <- size <= limit do
:ok
else
false ->
{:error, :body_too_large}
_ ->
:ok
end
end
2018-11-29 20:11:45 +00:00
defp header_length_constraint(_, _), do: :ok
2018-11-23 16:40:45 +00:00
defp body_size_constraint(size, limit) when is_integer(limit) and limit > 0 and size >= limit do
{:error, :body_too_large}
end
defp body_size_constraint(_, _), do: :ok
defp check_read_duration(duration, max)
when is_integer(duration) and is_integer(max) and max > 0 do
if duration > max do
{:error, :read_duration_exceeded}
else
{:ok, {duration, :erlang.system_time(:millisecond)}}
end
end
defp check_read_duration(_, _), do: {:ok, :no_duration_limit, :no_duration_limit}
defp increase_read_duration({previous_duration, started})
when is_integer(previous_duration) and is_integer(started) do
duration = :erlang.system_time(:millisecond) - started
{:ok, previous_duration + duration}
end
defp increase_read_duration(_) do
{:ok, :no_duration_limit, :no_duration_limit}
end
end