reverse proxy / uploads

.gitignore

@@ -6,6 +6,9 @@
 /uploads
 /test/uploads
 /.elixir_ls
+/test/fixtures/test_tmp.txt
+/test/fixtures/image_tmp.jpg
+/doc
 
 # Prevent committing custom emojis
 /priv/static/emoji/custom/*
@@ -28,4 +31,4 @@ erl_crash.dump
 .env
 
 # Editor config
-/.vscode
\ No newline at end of file
+/.vscode

config/config.exs

@@ -12,16 +12,15 @@
 
 config :pleroma, Pleroma.Upload,
   uploader: Pleroma.Uploaders.Local,
-  strip_exif: false
+  strip_exif: false,
+  proxy_remote: false,
+  proxy_opts: [inline_content_types: true, keep_user_agent: true]
 
-config :pleroma, Pleroma.Uploaders.Local,
-  uploads: "uploads",
-  uploads_url: "{{base_url}}/media/{{file}}"
+config :pleroma, Pleroma.Uploaders.Local, uploads: "uploads"
 
 config :pleroma, Pleroma.Uploaders.S3,
   bucket: nil,
-  public_endpoint: "https://s3.amazonaws.com",
-  force_media_proxy: false
+  public_endpoint: "https://s3.amazonaws.com"
 
 config :pleroma, Pleroma.Uploaders.MDII,
   cgi: "https://mdii.sakura.ne.jp/mdii-post.cgi",
@@ -150,9 +149,11 @@
 
 config :pleroma, :media_proxy,
   enabled: false,
-  redirect_on_failure: true
-
-# base_url: "https://cache.pleroma.social"
+  # base_url: "https://cache.pleroma.social",
+  proxy_opts: [
+    # inline_content_types: [] | false | true,
+    # http: [:insecure]
+  ]
 
 config :pleroma, :chat, enabled: true
 

config/test.exs

@@ -9,7 +9,7 @@
 # Print only warnings and errors during test
 config :logger, level: :warn
 
-config :pleroma, Pleroma.Upload, uploads: "test/uploads"
+config :pleroma, Pleroma.Uploaders.Local, uploads: "test/uploads"
 
 # Configure your database
 config :pleroma, Pleroma.Repo,

lib/mix/tasks/migrate_local_uploads.ex

+defmodule Mix.Tasks.MigrateLocalUploads do
+  use Mix.Task
+  import Mix.Ecto
+  alias Pleroma.{Upload, Uploaders.Local, Uploaders.S3}
+  require Logger
+
+  @log_every 50
+  @shortdoc "Migrate uploads from local to remote storage"
+
+  def run([target_uploader | args]) do
+    delete? = Enum.member?(args, "--delete")
+    Application.ensure_all_started(:pleroma)
+
+    local_path = Pleroma.Config.get!([Local, :uploads])
+    uploader = Module.concat(Pleroma.Uploaders, target_uploader)
+
+    unless Code.ensure_loaded?(uploader) do
+      raise("The uploader #{inspect(uploader)} is not an existing/loaded module.")
+    end
+
+    target_enabled? = Pleroma.Config.get([Upload, :uploader]) == uploader
+
+    unless target_enabled? do
+      Pleroma.Config.put([Upload, :uploader], uploader)
+    end
+
+    Logger.info("Migrating files from local #{local_path} to #{to_string(uploader)}")
+
+    if delete? do
+      Logger.warn(
+        "Attention: uploaded files will be deleted, hope you have backups! (--delete ; cancel with ^C)"
+      )
+
+      :timer.sleep(:timer.seconds(5))
+    end
+
+    uploads = File.ls!(local_path)
+    total_count = length(uploads)
+
+    uploads
+    |> Task.async_stream(
+      fn uuid ->
+        u_path = Path.join(local_path, uuid)
+
+        {name, path} =
+          cond do
+            File.dir?(u_path) ->
+              files = for file <- File.ls!(u_path), do: {{file, uuid}, Path.join([u_path, file])}
+              List.first(files)
+
+            File.exists?(u_path) ->
+              # {uuid, u_path}
+              raise "should_dedupe local storage not supported yet sorry"
+          end
+
+        {:ok, _} =
+          Upload.store({:from_local, name, path}, should_dedupe: false, uploader: uploader)
+
+        if delete? do
+          File.rm_rf!(u_path)
+        end
+
+        Logger.debug("uploaded: #{inspect(name)}")
+      end,
+      timeout: 150_000
+    )
+    |> Stream.chunk_every(@log_every)
+    |> Enum.reduce(0, fn done, count ->
+      count = count + length(done)
+      Logger.info("Uploaded #{count}/#{total_count} files")
+      count
+    end)
+
+    Logger.info("Done!")
+  end
+
+  def run(_) do
+    Logger.error("Usage: migrate_local_uploads UploaderName [--delete]")
+  end
+end

lib/pleroma/application.ex

@@ -7,6 +7,11 @@ def name, do: @name
   def version, do: @version
   def named_version(), do: @name <> " " <> @version
 
+  def user_agent() do
+    info = "#{Pleroma.Web.base_url()} <#{Pleroma.Config.get([:instance, :email], "")}>"
+    named_version() <> "; " <> info
+  end
+
   # See http://elixir-lang.org/docs/stable/elixir/Application.html
   # for more information on OTP Applications
   @env Mix.env()

lib/pleroma/plugs/uploaded_media.ex

+defmodule Pleroma.Plugs.UploadedMedia do
+  @moduledoc """
+  """
+
+  import Plug.Conn
+  require Logger
+
+  @behaviour Plug
+  # no slashes
+  @path "media"
+  @cache_control %{
+    default: "public, max-age=1209600",
+    error: "public, must-revalidate, max-age=160"
+  }
+
+  def init(_opts) do
+    static_plug_opts =
+      []
+      |> Keyword.put(:from, "__unconfigured_media_plug")
+      |> Keyword.put(:at, "/__unconfigured_media_plug")
+      |> Plug.Static.init()
+
+    %{static_plug_opts: static_plug_opts}
+  end
+
+  def call(conn = %{request_path: <<"/", @path, "/", file::binary>>}, opts) do
+    config = Pleroma.Config.get([Pleroma.Upload])
+
+    with uploader <- Keyword.fetch!(config, :uploader),
+         proxy_remote = Keyword.get(config, :proxy_remote, false),
+         {:ok, get_method} <- uploader.get_file(file) do
+      get_media(conn, get_method, proxy_remote, opts)
+    else
+      _ ->
+        conn
+        |> send_resp(500, "Failed")
+        |> halt()
+    end
+  end
+
+  def call(conn, _opts), do: conn
+
+  defp get_media(conn, {:static_dir, directory}, _, opts) do
+    static_opts =
+      Map.get(opts, :static_plug_opts)
+      |> Map.put(:at, [@path])
+      |> Map.put(:from, directory)
+
+    conn = Plug.Static.call(conn, static_opts)
+
+    if conn.halted do
+      conn
+    else
+      conn
+      |> send_resp(404, "Not found")
+      |> halt()
+    end
+  end
+
+  defp get_media(conn, {:url, url}, true, _) do
+    conn
+    |> Pleroma.ReverseProxy.call(url, Pleroma.Config.get([Pleroma.Upload, :proxy_opts], []))
+  end
+
+  defp get_media(conn, {:url, url}, _, _) do
+    conn
+    |> Phoenix.Controller.redirect(external: url)
+    |> halt()
+  end
+
+  defp get_media(conn, unknown, _, _) do
+    Logger.error("#{__MODULE__}: Unknown get startegy: #{inspect(unknown)}")
+
+    conn
+    |> send_resp(500, "Internal Error")
+    |> halt()
+  end
+end

lib/pleroma/reverse_proxy.ex

+defmodule Pleroma.ReverseProxy do
+  @keep_req_headers ~w(accept user-agent accept-encoding cache-control if-modified-since if-none-match range)
+  @resp_cache_headers ~w(etag date last-modified cache-control)
+  @keep_resp_headers @resp_cache_headers ++
+                       ~w(content-type content-disposition content-length accept-ranges vary)
+  @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),
+         :ok <- header_lenght_constraint(headers, Keyword.get(opts, :max_body_length)) do
+      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 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
+    headers =
+      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).()
+  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} =
+      List.keyfind(headers, "content-type", 0, {"content-type", "application/octect-stream"})
+
+    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
+
+  defp header_lenght_constraint(headers, limit) when is_integer(limit) and limit > 0 do
+    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
+
+  defp header_lenght_constraint(_, _), do: :ok
+
+  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
+      Logger.debug("Duration #{inspect(duration)}")
+      {: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

lib/pleroma/upload.ex

 defmodule Pleroma.Upload do
   alias Ecto.UUID
+  require Logger
+
+  @type upload_option ::
+          {:dedupe, boolean()} | {:size_limit, non_neg_integer()} | {:uploader, module()}
+  @type upload_source ::
+          Plug.Upload.t() | data_uri_string() ::
+          String.t() | {:from_local, name :: String.t(), uuid :: String.t(), path :: String.t()}
+
+  @spec store(upload_source, options :: [upload_option()]) :: {:ok, Map.t()} | {:error, any()}
+  def store(upload, opts \\ []) do
+    opts = get_opts(opts)
+
+    with {:ok, name, uuid, path, content_type} <- process_upload(upload, opts),
+         _ <- strip_exif_data(content_type, path),
+         {:ok, url_spec} <- opts.uploader.put_file(name, uuid, path, content_type, opts) do
+      {:ok,
+       %{
+         "type" => "Image",
+         "url" => [
+           %{
+             "type" => "Link",
+             "mediaType" => content_type,
+             "href" => url_from_spec(url_spec)
+           }
+         ],
+         "name" => name
+       }}
+    else