Merge branch 'feature/reverse-proxy' into 'develop'

reverse proxy / uploads

See merge request pleroma/pleroma!470

.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
+/.vscode

config/config.exs

@@ -10,18 +10,18 @@
 
 config :pleroma, Pleroma.Repo, types: Pleroma.PostgresTypes
 
+# Upload configuration
 config :pleroma, Pleroma.Upload,
   uploader: Pleroma.Uploaders.Local,
-  strip_exif: false
+  filters: [],
+  proxy_remote: false,
+  proxy_opts: []
 
-config :pleroma, Pleroma.Uploaders.Local,
+config :pleroma, Pleroma.Uploaders.Local, uploads: "uploads"
-  uploads: "uploads",
-  uploads_url: "{{base_url}}/media/{{file}}"
 
 config :pleroma, Pleroma.Uploaders.S3,
   bucket: nil,
-  public_endpoint: "https://s3.amazonaws.com",
+  public_endpoint: "https://s3.amazonaws.com"
-  force_media_proxy: false
 
 config :pleroma, Pleroma.Uploaders.MDII,
   cgi: "https://mdii.sakura.ne.jp/mdii-post.cgi",
@@ -150,9 +150,11 @@
 
 config :pleroma, :media_proxy,
   enabled: false,
-  redirect_on_failure: true
+  # base_url: "https://cache.pleroma.social",
-
+  proxy_opts: [
-# base_url: "https://cache.pleroma.social"
+    # inline_content_types: [] | false | true,
+    # http: [:insecure]
+  ]
 
 config :pleroma, :chat, enabled: true
 

config/config.md

@@ -5,11 +5,19 @@ If you run Pleroma with ``MIX_ENV=prod`` the file is ``prod.secret.exs``, otherw
 
 ## Pleroma.Upload
 * `uploader`: Select which `Pleroma.Uploaders` to use
-* `strip_exif`: boolean, uses ImageMagick(!) to strip exif.
+* `filters`: List of `Pleroma.Upload.Filter` to use.
+* `base_url`: The base URL to access a user-uploaded file. Useful when you want to proxy the media files via another host.
+* `proxy_remote`: If you're using a remote uploader, Pleroma will proxy media requests instead of redirecting to it.
+* `proxy_opts`: Proxy options, see `Pleroma.ReverseProxy` documentation.
+
+Note: `strip_exif` has been replaced by `Pleroma.Upload.Filter.Mogrify`.
 
 ## Pleroma.Uploaders.Local
 * `uploads`: Which directory to store the user-uploads in, relative to pleroma’s working directory
-* `uploads_url`: The URL to access a user-uploaded file, ``{{base_url}}`` is replaced to the instance URL and ``{{file}}`` to the filename. Useful when you want to proxy the media files via another host.
+
+## Pleroma.Upload.Filter.Mogrify
+
+* `args`: List of actions for the `mogrify` command like `"strip"` or `["strip", {"impode", "1"}]`.
 
 ## :uri_schemes
 * `valid_schemes`: List of the scheme part that is considered valid to be an URL
@@ -68,7 +76,8 @@ This section is used to configure Pleroma-FE, unless ``:managed_config`` in ``:i
 
 ## :media_proxy
 * `enabled`: Enables proxying of remote media to the instance’s proxy
-* `redirect_on_failure`: Use the original URL when Media Proxy fails to get it
+* `base_url`: The base URL to access a user-uploaded file. Useful when you want to proxy the media files via another host/CDN fronts.
+* `proxy_opts`: All options defined in `Pleroma.ReverseProxy` documentation, defaults to `[max_body_length: (25*1_048_576)]`.
 
 ## :gopher
 * `enabled`: Enables the gopher interface

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,

installation/pleroma.nginx

@@ -70,10 +70,12 @@ server {
         client_max_body_size 16m;
     }
 
-    location /proxy {
+    location ~ ^/(media|proxy) {
         proxy_cache pleroma_media_cache;
         proxy_cache_lock on;
         proxy_ignore_client_abort on;
+        proxy_buffering off;
+        chunked_transfer_encoding on;
         proxy_pass http://localhost:4000;
     }
 }

lib/mix/tasks/migrate_local_uploads.ex

@@ -0,0 +1,97 @@
+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)
+      |> Enum.map(fn id ->
+        root_path = Path.join(local_path, id)
+
+        cond do
+          File.dir?(root_path) ->
+            files = for file <- File.ls!(root_path), do: {id, file, Path.join([root_path, file])}
+
+            case List.first(files) do
+              {id, file, path} ->
+                {%Pleroma.Upload{id: id, name: file, path: id <> "/" <> file, tempfile: path},
+                 root_path}
+
+              _ ->
+                nil
+            end
+
+          File.exists?(root_path) ->
+            file = Path.basename(id)
+            [hash, ext] = String.split(id, ".")
+            {%Pleroma.Upload{id: hash, name: file, path: file, tempfile: root_path}, root_path}
+
+          true ->
+            nil
+        end
+      end)
+      |> Enum.filter(& &1)
+
+    total_count = length(uploads)
+    Logger.info("Found #{total_count} uploads")
+
+    uploads
+    |> Task.async_stream(
+      fn {upload, root_path} ->
+        case Upload.store(upload, uploader: uploader, filters: [], size_limit: nil) do
+          {:ok, _} ->
+            if delete?, do: File.rm_rf!(root_path)
+            Logger.debug("uploaded: #{inspect(upload.path)} #{inspect(upload)}")
+            :ok
+
+          error ->
+            Logger.error("failed to upload #{inspect(upload.path)}: #{inspect(error)}")
+        end
+      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 S3|Swift [--delete]")
+  end
+end

lib/pleroma/application.ex

@@ -8,6 +8,11 @@ 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/mime.ex

@@ -0,0 +1,108 @@
+defmodule Pleroma.MIME do
+  @moduledoc """
+  Returns the mime-type of a binary and optionally a normalized file-name.
+  """
+  @default "application/octet-stream"
+  @read_bytes 31
+
+  @spec file_mime_type(String.t()) ::
+          {:ok, content_type :: String.t(), filename :: String.t()} | {:error, any()} | :error
+  def file_mime_type(path, filename) do
+    with {:ok, content_type} <- file_mime_type(path),
+         filename <- fix_extension(filename, content_type) do
+      {:ok, content_type, filename}
+    end
+  end
+
+  @spec file_mime_type(String.t()) :: {:ok, String.t()} | {:error, any()} | :error
+  def file_mime_type(filename) do
+    File.open(filename, [:read], fn f ->
+      check_mime_type(IO.binread(f, @read_bytes))
+    end)
+  end
+
+  def bin_mime_type(binary, filename) do
+    with {:ok, content_type} <- bin_mime_type(binary),
+         filename <- fix_extension(filename, content_type) do
+      {:ok, content_type, filename}
+    end
+  end
+
+  @spec bin_mime_type(binary()) :: {:ok, String.t()} | :error
+  def bin_mime_type(<<head::binary-size(@read_bytes), _::binary>>) do
+    {:ok, check_mime_type(head)}
+  end
+
+  def mime_type(<<_::binary>>), do: {:ok, @default}
+
+  def bin_mime_type(_), do: :error
+
+  defp fix_extension(filename, content_type) do
+    parts = String.split(filename, ".")
+
+    new_filename =
+      if length(parts) > 1 do
+        Enum.drop(parts, -1) |> Enum.join(".")
+      else
+        Enum.join(parts)
+      end
+
+    cond do
+      content_type == "application/octet-stream" ->
+        filename
+
+      ext = List.first(MIME.extensions(content_type)) ->
+        new_filename <> "." <> ext
+
+      true ->
+        Enum.join([new_filename, String.split(content_type, "/") |> List.last()], ".")
+    end
+  end
+
+  defp check_mime_type(<<0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A, _::binary>>) do
+    "image/png"
+  end
+
+  defp check_mime_type(<<0x47, 0x49, 0x46, 0x38, _, 0x61, _::binary>>) do
+    "image/gif"
+  end
+
+  defp check_mime_type(<<0xFF, 0xD8, 0xFF, _::binary>>) do
+    "image/jpeg"
+  end
+
+  defp check_mime_type(<<0x1A, 0x45, 0xDF, 0xA3, _::binary>>) do
+    "video/webm"
+  end
+
+  defp check_mime_type(<<0x00, 0x00, 0x00, _, 0x66, 0x74, 0x79, 0x70, _::binary>>) do
+    "video/mp4"
+  end
+
+  defp check_mime_type(<<0x49, 0x44, 0x33, _::binary>>) do
+    "audio/mpeg"
+  end
+
+  defp check_mime_type(<<255, 251, _, 68, 0, 0, 0, 0, _::binary>>) do
+    "audio/mpeg"
+  end
+
+  defp check_mime_type(
+         <<0x4F, 0x67, 0x67, 0x53, 0x00, 0x02, 0x00, 0x00, _::size(160), 0x80, 0x74, 0x68, 0x65,
+           0x6F, 0x72, 0x61, _::binary>>
+       ) do
+    "video/ogg"
+  end
+
+  defp check_mime_type(<<0x4F, 0x67, 0x67, 0x53, 0x00, 0x02, 0x00, 0x00, _::binary>>) do
+    "audio/ogg"
+  end
+
+  defp check_mime_type(<<0x52, 0x49, 0x46, 0x46, _::binary>>) do
+    "audio/wav"
+  end
+
+  defp check_mime_type(_) do
+    @default
+  end
+end

lib/pleroma/plugs/uploaded_media.ex

@@ -0,0 +1,78 @@
+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

@@ -0,0 +1,343 @@
+defmodule Pleroma.ReverseProxy do
+  @keep_req_headers ~w(accept user-agent accept-encoding cache-control if-modified-since if-unmodified-since if-none-match if-range range)
+  @resp_cache_headers ~w(etag date last-modified cache-control)
+  @keep_resp_headers @resp_cache_headers ++
+                       ~w(content-type content-disposition content-encoding 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)
+  @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_length_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 get_content_type(headers) do
+    {_, content_type} =
+      List.keyfind(headers, "content-type", 0, {"content-type", "application/octet-stream"})
+
+    [content_type | _] = String.split(content_type, ";")
+    content_type
+  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
+    |> 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 = get_content_type(headers)
+
+    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_length_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_length_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
+      {: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

@@ -1,81 +1,209 @@
 defmodule Pleroma.Upload do
+  @moduledoc """
+  # Upload
+
+  Options:
+  * `:type`: presets for activity type (defaults to Document) and size limits from app configuration
+  * `:description`: upload alternative text
+  * `:base_url`: override base url
+  * `:uploader`: override uploader
+  * `:filters`: override filters
+  * `:size_limit`: override size limit
+  * `:activity_type`: override activity type
+
+  The `%Pleroma.Upload{}` struct: all documented fields are meant to be overwritten in filters:
+
+  * `:id` - the upload id.
+  * `:name` - the upload file name.
+  * `:path` - the upload path: set at first to `id/name` but can be changed. Keep in mind that the path
+    is once created permanent and changing it (especially in uploaders) is probably a bad idea!
+  * `:tempfile` - path to the temporary file. Prefer in-place changes on the file rather than changing the
+  path as the temporary file is also tracked by `Plug.Upload{}` and automatically deleted once the request is over.
+
+  Related behaviors:
+
+  * `Pleroma.Uploaders.Uploader`
+  * `Pleroma.Upload.Filter`
+
+  """
   alias Ecto.UUID
+  require Logger
 
-  def check_file_size(path, nil), do: true
+  @type source ::
+          Plug.Upload.t() | data_uri_string ::
+          String.t() | {:from_local, name :: String.t(), id :: String.t(), path :: String.t()}
 
-  def check_file_size(path, size_limit) do
+  @type option ::
-    {:ok, %{size: size}} = File.stat(path)
+          {:type, :avatar | :banner | :background}
-    size <= size_limit
+          | {:description, String.t()}
-  end
+          | {:activity_type, String.t()}
+          | {:size_limit, nil | non_neg_integer()}
+          | {:uploader, module()}
+          | {:filters, [module()]}
 
-  def store(file, should_dedupe, size_limit \\ nil)
+  @type t :: %__MODULE__{
+          id: String.t(),
+          name: String.t(),
+          tempfile: String.t(),
+          content_type: String.t(),
+          path: String.t()
+        }
+  defstruct [:id, :name, :tempfile, :content_type, :path]
 
-  def store(%Plug.Upload{} = file, should_dedupe, size_limit) do
+  @spec store(source, options :: [option()]) :: {:ok, Map.t()} | {:error, any()}
-    content_type = get_content_type(file.path)
+  def store(upload, opts \\ []) do
+    opts = get_opts(opts)
 
-    with uuid <- get_uuid(file, should_dedupe),
+    with {:ok, upload} <- prepare_upload(upload, opts),
-         name <- get_name(file, uuid, content_type, should_dedupe),
+         upload = %__MODULE__{upload | path: upload.path || "#{upload.id}/#{upload.name}"},
-         true <- check_file_size(file.path, size_limit) do
+         {:ok, upload} <- Pleroma.Upload.Filter.filter(opts.filters, upload),
-      strip_exif_data(content_type, file.path)
+         {:ok, url_spec} <- Pleroma.Uploaders.Uploader.put_file(opts.uploader, upload) do
-
+      {:ok,
-      {:ok, url_path} = uploader().put_file(name, uuid, file.path, content_type, should_dedupe)
+       %{
-
+         "type" => opts.activity_type,
-      %{
+         "url" => [
-        "type" => "Document",
+           %{
-        "url" => [
+             "type" => "Link",
-          %{
+             "mediaType" => upload.content_type,
-            "type" => "Link",
+             "href" => url_from_spec(opts.base_url, url_spec)
-            "mediaType" => content_type,
+           }
-            "href" => url_path
+         ],
-          }
+         "name" => Map.get(opts, :description) || upload.name
-        ],
+       }}
-        "name" => name
-      }
     else
-      _e -> nil
+      {:error, error} ->
-    end
+        Logger.error(
-  end
+          "#{__MODULE__} store (using #{inspect(opts.uploader)}) failed: #{inspect(error)}"
-
-  def store(%{"img" => "data:image/" <> image_data}, should_dedupe, size_limit) do
-    parsed = Regex.named_captures(~r/(?<filetype>jpeg|png|gif);base64,(?<data>.*)/, image_data)
-    data = Base.decode64!(parsed["data"], ignore: :whitespace)
-
-    with tmp_path <- tempfile_for_image(data),
-         uuid <- UUID.generate(),
-         true <- check_file_size(tmp_path, size_limit) do
-      content_type = get_content_type(tmp_path)
-      strip_exif_data(content_type, tmp_path)
-
-      name =
-        create_name(
-          String.downcase(Base.encode16(:crypto.hash(:sha256, data))),
-          parsed["filetype"],
-          content_type
         )
 
-      {:ok, url_path} = uploader().put_file(name, uuid, tmp_path, content_type, should_dedupe)
+        {:error, error}
-
-      %{
-        "type" => "Image",
-        "url" => [
-          %{
-            "type" => "Link",
-            "mediaType" => content_type,
-            "href" => url_path
-          }
-        ],
-        "name" => name
-      }
-    else
-      _e -> nil
     end
   end
 
-  @doc """
+  defp get_opts(opts) do
-  Creates a tempfile using the Plug.Upload Genserver which cleans them up 
+    {size_limit, activity_type} =
-  automatically.
+      case Keyword.get(opts, :type) do
-  """
+        :banner ->
-  def tempfile_for_image(data) do
+          {Pleroma.Config.get!([:instance, :banner_upload_limit]), "Image"}
+
+        :avatar ->
+          {Pleroma.Config.get!([:instance, :avatar_upload_limit]), "Image"}
+
+        :background ->
+          {Pleroma.Config.get!([:instance, :background_upload_limit]), "Image"}
+
+        _ ->
+          {Pleroma.Config.get!([:instance, :upload_limit]), "Document"}
+      end
+
+    opts = %{
+      activity_type: Keyword.get(opts, :activity_type, activity_type),
+      size_limit: Keyword.get(opts, :size_limit, size_limit),
+      uploader: Keyword.get(opts, :uploader, Pleroma.Config.get([__MODULE__, :uploader])),
+      filters: Keyword.get(opts, :filters, Pleroma.Config.get([__MODULE__, :filters])),
+      description: Keyword.get(opts, :description),
+      base_url:
+        Keyword.get(
+          opts,
+          :base_url,
+          Pleroma.Config.get([__MODULE__, :base_url], Pleroma.Web.base_url())
+        )
+    }
+
+    # TODO: 1.0+ : remove old config compatibility
+    opts =
+      if Pleroma.Config.get([__MODULE__, :strip_exif]) == true &&
+           !Enum.member?(opts.filters, Pleroma.Upload.Filter.Mogrify) do
+        Logger.warn("""
+        Pleroma: configuration `:instance, :strip_exif` is deprecated, please instead set:
+
+          :pleroma, Pleroma.Upload, [filters: [Pleroma.Upload.Filter.Mogrify]]
+
+          :pleroma, Pleroma.Upload.Filter.Mogrify, args: "strip"
+        """)
+
+        Pleroma.Config.put([Pleroma.Upload.Filter.Mogrify], args: "strip")
+        Map.put(opts, :filters, opts.filters ++ [Pleroma.Upload.Filter.Mogrify])
+      else
+        opts
+      end
+
+    opts =
+      if Pleroma.Config.get([:instance, :dedupe_media]) == true &&
+           !Enum.member?(opts.filters, Pleroma.Upload.Filter.Dedupe) do
+        Logger.warn("""
+        Pleroma: configuration `:instance, :dedupe_media` is deprecated, please instead set:
+
+          :pleroma, Pleroma.Upload, [filters: [Pleroma.Upload.Filter.Dedupe]]
+        """)
+
+        Map.put(opts, :filters, opts.filters ++ [Pleroma.Upload.Filter.Dedupe])
+      else
+        opts
+      end
+  end
+
+  defp prepare_upload(%Plug.Upload{} = file, opts) do
+    with :ok <- check_file_size(file.path, opts.size_limit),
+         {:ok, content_type, name} <- Pleroma.MIME.file_mime_type(file.path, file.filename) do
+      {:ok,
+       %__MODULE__{
+         id: UUID.generate(),
+         name: name,
+         tempfile: file.path,
+         content_type: content_type
+       }}
+    end
+  end
+
+  defp prepare_upload(%{"img" => "data:image/" <> image_data}, opts) do
+    parsed = Regex.named_captures(~r/(?<filetype>jpeg|png|gif);base64,(?<data>.*)/, image_data)
+    data = Base.decode64!(parsed["data"], ignore: :whitespace)
+    hash = String.downcase(Base.encode16(:crypto.hash(:sha256, data)))
+
+    with :ok <- check_binary_size(data, opts.size_limit),
+         tmp_path <- tempfile_for_image(data),
+         {:ok, content_type, name} <-
+           Pleroma.MIME.bin_mime_type(data, hash <> "." <> parsed["filetype"]) do
+      {:ok,
+       %__MODULE__{
+         id: UUID.generate(),
+         name: name,
+         tempfile: tmp_path,
+         content_type: content_type
+       }}
+    end
+  end
+
+  # For Mix.Tasks.MigrateLocalUploads
+  defp prepare_upload(upload = %__MODULE__{tempfile: path}, _opts) do
+    with {:ok, content_type} <- Pleroma.MIME.file_mime_type(path) do
+      {:ok, %__MODULE__{upload | content_type: content_type}}
+    end
+  end
+
+  defp check_binary_size(binary, size_limit)
+       when is_integer(size_limit) and size_limit > 0 and byte_size(binary) >= size_limit do
+    {:error, :file_too_large}
+  end
+
+  defp check_binary_size(_, _), do: :ok
+
+  defp check_file_size(path, size_limit) when is_integer(size_limit) and size_limit > 0 do
+    with {:ok, %{size: size}} <- File.stat(path),
+         true <- size <= size_limit do
+      :ok
+    else
+      false -> {:error, :file_too_large}
+      error -> error
+    end
+  end
+
+  defp check_file_size(_, _), do: :ok
+
+  # Creates a tempfile using the Plug.Upload Genserver which cleans them up
+  # automatically.
+  defp tempfile_for_image(data) do
     {:ok, tmp_path} = Plug.Upload.random_file("profile_pics")
     {:ok, tmp_file} = File.open(tmp_path, [:write, :raw, :binary])
     IO.binwrite(tmp_file, data)
@@ -83,108 +211,12 @@ def tempfile_for_image(data) do
     tmp_path
   end
 
-  def strip_exif_data(content_type, file) do
+  defp url_from_spec(base_url, {:file, path}) do
-    settings = Application.get_env(:pleroma, Pleroma.Upload)
+    [base_url, "media", path]
-    do_strip = Keyword.fetch!(settings, :strip_exif)
+    |> Path.join()
-    [filetype, _ext] = String.split(content_type, "/")
-
-    if filetype == "image" and do_strip == true do
-      Mogrify.open(file) |> Mogrify.custom("strip") |> Mogrify.save(in_place: true)
-    end
   end
 
-  defp create_name(uuid, ext, type) do
+  defp url_from_spec({:url, url}) do
-    case type do
+    url
-      "application/octet-stream" ->
-        String.downcase(Enum.join([uuid, ext], "."))
-
-      "audio/mpeg" ->
-        String.downcase(Enum.join([uuid, "mp3"], "."))
-
-      _ ->
-        String.downcase(Enum.join([uuid, List.last(String.split(type, "/"))], "."))
-    end
-  end
-
-  defp get_uuid(file, should_dedupe) do
-    if should_dedupe do
-      Base.encode16(:crypto.hash(:sha256, File.read!(file.path)))
-    else
-      UUID.generate()
-    end
-  end
-
-  defp get_name(file, uuid, type, should_dedupe) do
-    if should_dedupe do
-      create_name(uuid, List.last(String.split(file.filename, ".")), type)
-    else
-      parts = String.split(file.filename, ".")
-
-      new_filename =
-        if length(parts) > 1 do
-          Enum.drop(parts, -1) |> Enum.join(".")
-        else
-          Enum.join(parts)
-        end
-
-      case type do
-        "application/octet-stream" -> file.filename
-        "audio/mpeg" -> new_filename <> ".mp3"
-        "image/jpeg" -> new_filename <> ".jpg"
-        _ -> Enum.join([new_filename, String.split(type, "/") |> List.last()], ".")
-      end
-    end
-  end
-
-  def get_content_type(file) do
-    match =
-      File.open(file, [:read], fn f ->
-        case IO.binread(f, 8) do
-          <<0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A>> ->
-            "image/png"
-
-          <<0x47, 0x49, 0x46, 0x38, _, 0x61, _, _>> ->
-            "image/gif"
-
-          <<0xFF, 0xD8, 0xFF, _, _, _, _, _>> ->
-            "image/jpeg"
-
-          <<0x1A, 0x45, 0xDF, 0xA3, _, _, _, _>> ->
-            "video/webm"
-
-          <<0x00, 0x00, 0x00, _, 0x66, 0x74, 0x79, 0x70>> ->
-            "video/mp4"
-
-          <<0x49, 0x44, 0x33, _, _, _, _, _>> ->
-            "audio/mpeg"
-
-          <<255, 251, _, 68, 0, 0, 0, 0>> ->
-            "audio/mpeg"
-
-          <<0x4F, 0x67, 0x67, 0x53, 0x00, 0x02, 0x00, 0x00>> ->
-            case IO.binread(f, 27) do
-              <<_::size(160), 0x80, 0x74, 0x68, 0x65, 0x6F, 0x72, 0x61>> ->
-                "video/ogg"
-
-              _ ->
-                "audio/ogg"
-            end
-
-          <<0x52, 0x49, 0x46, 0x46, _, _, _, _>> ->
-            "audio/wav"
-
-          _ ->
-            "application/octet-stream"
-        end
-      end)
-
-    case match do
-      {:ok, type} -> type
-      _e -> "application/octet-stream"
-    end
-  end
-
-  defp uploader() do
-    Pleroma.Config.get!([Pleroma.Upload, :uploader])
   end
 end

lib/pleroma/upload/filter.ex

@@ -0,0 +1,35 @@
+defmodule Pleroma.Upload.Filter do
+  @moduledoc """
+  Upload Filter behaviour
+
+  This behaviour allows to run filtering actions just before a file is uploaded. This allows to:
+
+  * morph in place the temporary file
+  * change any field of a `Pleroma.Upload` struct
+  * cancel/stop the upload
+  """
+
+  require Logger
+
+  @callback filter(Pleroma.Upload.t()) :: :ok | {:ok, Pleroma.Upload.t()} | {:error, any()}
+
+  @spec filter([module()], Pleroma.Upload.t()) :: {:ok, Pleroma.Upload.t()} | {:error, any()}
+
+  def filter([], upload) do
+    {:ok, upload}
+  end
+
+  def filter([filter | rest], upload) do
+    case filter.filter(upload) do
+      :ok ->
+        filter(rest, upload)
+
+      {:ok, upload} ->
+        filter(rest, upload)
+
+      error ->
+        Logger.error("#{__MODULE__}: Filter #{filter} failed: #{inspect(error)}")
+        error
+    end
+  end
+end

lib/pleroma/upload/filter/dedupe.ex

@@ -0,0 +1,10 @@
+defmodule Pleroma.Upload.Filter.Dedupe do
+  @behaviour Pleroma.Upload.Filter
+
+  def filter(upload = %Pleroma.Upload{name: name, tempfile: path}) do
+    extension = String.split(name, ".") |> List.last()
+    shasum = :crypto.hash(:sha256, File.read!(upload.tempfile)) |> Base.encode16(case: :lower)
+    filename = shasum <> "." <> extension
+    {:ok, %Pleroma.Upload{upload | id: shasum, path: filename}}
+  end
+end

lib/pleroma/upload/filter/mogrifun.ex

@@ -0,0 +1,60 @@
+defmodule Pleroma.Upload.Filter.Mogrifun do
+  @behaviour Pleroma.Upload.Filter
+
+  @filters [
+    {"implode", "1"},
+    {"-raise", "20"},
+    {"+raise", "20"},
+    [{"-interpolate", "nearest"}, {"-virtual-pixel", "mirror"}, {"-spread", "5"}],
+    "+polaroid",
+    {"-statistic", "Mode 10"},
+    {"-emboss", "0x1.1"},
+    {"-emboss", "0x2"},
+    {"-colorspace", "Gray"},
+    "-negate",
+    [{"-channel", "green"}, "-negate"],
+    [{"-channel", "red"}, "-negate"],
+    [{"-channel", "blue"}, "-negate"],
+    {"+level-colors", "green,gold"},
+    {"+level-colors", ",DodgerBlue"},
+    {"+level-colors", ",Gold"},
+    {"+level-colors", ",Lime"},
+    {"+level-colors", ",Red"},
+    {"+level-colors", ",DarkGreen"},
+    {"+level-colors", "firebrick,yellow"},
+    {"+level-colors", "'rgb(102,75,25)',lemonchiffon"},
+    [{"fill", "red"}, {"tint", "40"}],
+    [{"fill", "green"}, {"tint", "40"}],
+    [{"fill", "blue"}, {"tint", "40"}],
+    [{"fill", "yellow"}, {"tint", "40"}]
+  ]
+
+  def filter(%Pleroma.Upload{tempfile: file, content_type: "image" <> _}) do
+    filter = Enum.random(@filters)
+
+    file
+    |> Mogrify.open()
+    |> mogrify_filter(filter)
+    |> Mogrify.save(in_place: true)
+
+    :ok
+  end
+
+  def filter(_), do: :ok
+
+  defp mogrify_filter(mogrify, [filter | rest]) do
+    mogrify
+    |> mogrify_filter(filter)
+    |> mogrify_filter(rest)
+  end
+
+  defp mogrify_filter(mogrify, []), do: mogrify
+
+  defp mogrify_filter(mogrify, {action, options}) do
+    Mogrify.custom(mogrify, action, options)
+  end
+
+  defp mogrify_filter(mogrify, string) when is_binary(string) do
+    Mogrify.custom(mogrify, string)
+  end
+end

lib/pleroma/upload/filter/mogrify.ex

@@ -0,0 +1,37 @@
+defmodule Pleroma.Upload.Filter.Mogrify do
+  @behaviour Pleroma.Uploader.Filter
+
+  @type conversion :: action :: String.t() | {action :: String.t(), opts :: String.t()}
+  @type conversions :: conversion() | [conversion()]
+
+  def filter(%Pleroma.Upload{tempfile: file, content_type: "image" <> _}) do
+    filters = Pleroma.Config.get!([__MODULE__, :args])
+
+    file
+    |> Mogrify.open()
+    |> mogrify_filter(filters)
+    |> Mogrify.save(in_place: true)
+
+    :ok
+  end
+
+  def filter(_), do: :ok
+
+  defp mogrify_filter(mogrify, nil), do: mogrify
+
+  defp mogrify_filter(mogrify, [filter | rest]) do
+    mogrify
+    |> mogrify_filter(filter)
+    |> mogrify_filter(rest)
+  end
+
+  defp mogrify_filter(mogrify, []), do: mogrify
+
+  defp mogrify_filter(mogrify, {action, options}) do
+    Mogrify.custom(mogrify, action, options)
+  end
+
+  defp mogrify_filter(mogrify, action) when is_binary(action) do
+    Mogrify.custom(mogrify, action)
+  end
+end

lib/pleroma/uploaders/local.ex

@@ -3,49 +3,32 @@ defmodule Pleroma.Uploaders.Local do
 
   alias Pleroma.Web
 
-  def put_file(name, uuid, tmpfile, _content_type, should_dedupe) do
+  def get_file(_) do
-    upload_folder = get_upload_path(uuid, should_dedupe)
+    {:ok, {:static_dir, upload_path()}}
-    url_path = get_url(name, uuid, should_dedupe)
+  end
 
-    File.mkdir_p!(upload_folder)
+  def put_file(upload) do
+    {local_path, file} =
+      case Enum.reverse(String.split(upload.path, "/", trim: true)) do
+        [file] ->
+          {upload_path(), file}
 
-    result_file = Path.join(upload_folder, name)
+        [file | folders] ->
+          path = Path.join([upload_path()] ++ Enum.reverse(folders))
+          File.mkdir_p!(path)
+          {path, file}
+      end
 
-    if File.exists?(result_file) do
+    result_file = Path.join(local_path, file)
-      File.rm!(tmpfile)
+
-    else
+    unless File.exists?(result_file) do
-      File.cp!(tmpfile, result_file)
+      File.cp!(upload.tempfile, result_file)
     end
 
-    {:ok, url_path}
+    :ok
   end
 
   def upload_path do
-    settings = Application.get_env(:pleroma, Pleroma.Uploaders.Local)
+    Pleroma.Config.get!([__MODULE__, :uploads])
-    Keyword.fetch!(settings, :uploads)
-  end
-
-  defp get_upload_path(uuid, should_dedupe) do
-    if should_dedupe do
-      upload_path()
-    else
-      Path.join(upload_path(), uuid)
-    end
-  end
-
-  defp get_url(name, uuid, should_dedupe) do
-    if should_dedupe do
-      url_for(:cow_uri.urlencode(name))
-    else
-      url_for(Path.join(uuid, :cow_uri.urlencode(name)))
-    end
-  end
-
-  defp url_for(file) do
-    settings = Application.get_env(:pleroma, Pleroma.Uploaders.Local)
-
-    Keyword.get(settings, :uploads_url)
-    |> String.replace("{{file}}", file)
-    |> String.replace("{{base_url}}", Web.base_url())
   end
 end

lib/pleroma/uploaders/mdii.ex

@@ -5,22 +5,27 @@ defmodule Pleroma.Uploaders.MDII do
 
   @httpoison Application.get_env(:pleroma, :httpoison)
 
-  def put_file(name, uuid, path, content_type, should_dedupe) do
+  # MDII-hosted images are never passed through the MediaPlug; only local media.
+  # Delegate to Pleroma.Uploaders.Local
+  def get_file(file) do
+    Pleroma.Uploaders.Local.get_file(file)
+  end
+
+  def put_file(upload) do
     cgi = Pleroma.Config.get([Pleroma.Uploaders.MDII, :cgi])
     files = Pleroma.Config.get([Pleroma.Uploaders.MDII, :files])
 
-    {:ok, file_data} = File.read(path)
+    {:ok, file_data} = File.read(upload.tempfile)
 
-    extension = String.split(name, ".") |> List.last()
+    extension = String.split(upload.name, ".") |> List.last()
     query = "#{cgi}?#{extension}"
 
     with {:ok, %{status_code: 200, body: body}} <- @httpoison.post(query, file_data) do
-      File.rm!(path)
       remote_file_name = String.split(body) |> List.first()
       public_url = "#{files}/#{remote_file_name}.#{extension}"
-      {:ok, public_url}
+      {:ok, {:url, public_url}}
     else
-      _ -> Pleroma.Uploaders.Local.put_file(name, uuid, path, content_type, should_dedupe)
+      _ -> Pleroma.Uploaders.Local.put_file(upload)
     end
   end
 end

lib/pleroma/uploaders/s3.ex

@@ -1,40 +1,46 @@
 defmodule Pleroma.Uploaders.S3 do
-  alias Pleroma.Web.MediaProxy
-
   @behaviour Pleroma.Uploaders.Uploader
+  require Logger
 
-  def put_file(name, uuid, path, content_type, _should_dedupe) do
+  # The file name is re-encoded with S3's constraints here to comply with previous links with less strict filenames
-    settings = Application.get_env(:pleroma, Pleroma.Uploaders.S3)
+  def get_file(file) do
-    bucket = Keyword.fetch!(settings, :bucket)
+    config = Pleroma.Config.get([__MODULE__])
-    public_endpoint = Keyword.fetch!(settings, :public_endpoint)
-    force_media_proxy = Keyword.fetch!(settings, :force_media_proxy)
 
-    {:ok, file_data} = File.read(path)
+    {:ok,
+     {:url,
+      Path.join([
+        Keyword.fetch!(config, :public_endpoint),
+        Keyword.fetch!(config, :bucket),
+        strict_encode(URI.decode(file))
+      ])}}
+  end
 
-    File.rm!(path)
+  def put_file(upload = %Pleroma.Upload{}) do
+    config = Pleroma.Config.get([__MODULE__])
+    bucket = Keyword.get(config, :bucket)
 
-    s3_name = "#{uuid}/#{encode(name)}"
+    {:ok, file_data} = File.read(upload.tempfile)
 
-    {:ok, _} =
+    s3_name = strict_encode(upload.path)
+
+    op =
       ExAws.S3.put_object(bucket, s3_name, file_data, [
         {:acl, :public_read},
-        {:content_type, content_type}
+        {:content_type, upload.content_type}
       ])
-      |> ExAws.request()
 
-    url_base = "#{public_endpoint}/#{bucket}/#{s3_name}"
+    case ExAws.request(op) do
+      {:ok, _} ->
+        {:ok, {:file, s3_name}}
 
-    public_url =
+      error ->
-      if force_media_proxy do
+        Logger.error("#{__MODULE__}: #{inspect(error)}")
-        MediaProxy.url(url_base)
+        {:error, "S3 Upload failed"}
-      else
+    end
-        url_base
-      end
-
-    {:ok, public_url}
   end
 
-  defp encode(name) do
+  @regex Regex.compile!("[^0-9a-zA-Z!.*/'()_-]")
-    String.replace(name, ~r/[^0-9a-zA-Z!.*'()_-]/, "-")
+  def strict_encode(name) do
+    String.replace(name, @regex, "-")
   end
 end

lib/pleroma/uploaders/swift/swift.ex

@@ -14,7 +14,7 @@ def upload_file(filename, body, content_type) do
 
     case put("#{filename}", body, "X-Auth-Token": token, "Content-Type": content_type) do
       {:ok, %HTTPoison.Response{status_code: 201}} ->
-        {:ok, "#{object_url}/#{filename}"}
+        {:ok, {:file, filename}}
 
       {:ok, %HTTPoison.Response{status_code: 401}} ->
         {:error, "Unauthorized, Bad Token"}

lib/pleroma/uploaders/swift/uploader.ex

@@ -1,10 +1,15 @@
 defmodule Pleroma.Uploaders.Swift do
   @behaviour Pleroma.Uploaders.Uploader
 
-  def put_file(name, uuid, tmp_path, content_type, _should_dedupe) do
+  def get_file(name) do
-    {:ok, file_data} = File.read(tmp_path)
+    {:ok, {:url, Path.join([Pleroma.Config.get!([__MODULE__, :object_url]), name])}}
-    remote_name = "#{uuid}/#{name}"
+  end
 
-    Pleroma.Uploaders.Swift.Client.upload_file(remote_name, file_data, content_type)
+  def put_file(upload) do
+    Pleroma.Uploaders.Swift.Client.upload_file(
+      upload.path,
+      File.read!(upload.tmpfile),
+      upload.content_type
+    )
   end
 end

lib/pleroma/uploaders/uploader.ex

@@ -1,20 +1,40 @@
 defmodule Pleroma.Uploaders.Uploader do
   @moduledoc """
-  Defines the contract to put an uploaded file to any backend.
+  Defines the contract to put and get an uploaded file to any backend.
   """
 
+  @doc """
+  Instructs how to get the file from the backend.
+
+  Used by `Pleroma.Plugs.UploadedMedia`.
+  """
+  @type get_method :: {:static_dir, directory :: String.t()} | {:url, url :: String.t()}
+  @callback get_file(file :: String.t()) :: {:ok, get_method()}
+
   @doc """
   Put a file to the backend.
 
-  Returns `{:ok, String.t } | {:error, String.t} containing the path of the 
+  Returns:
-  uploaded file, or error information if the file failed to be saved to the 
+
-  respective backend.
+  * `:ok` which assumes `{:ok, upload.path}`
+  * `{:ok, spec}` where spec is:
+    * `{:file, filename :: String.t}` to handle reads with `get_file/1` (recommended)
+
+    This allows to correctly proxy or redirect requests to the backend, while allowing to migrate backends without breaking any URL.
+  * `{url, url :: String.t}` to bypass `get_file/2` and use the `url` directly in the activity.
+  * `{:error, String.t}` error information if the file failed to be saved to the backend.
+
+
   """
-  @callback put_file(
+  @callback put_file(Pleroma.Upload.t()) ::
-              name :: String.t(),
+              :ok | {:ok, {:file | :url, String.t()}} | {:error, String.t()}
-              uuid :: String.t(),
+
-              file :: File.t(),
+  @spec put_file(module(), Pleroma.Upload.t()) ::
-              content_type :: String.t(),
+          {:ok, {:file | :url, String.t()}} | {:error, String.t()}
-              should_dedupe :: Boolean.t()
+  def put_file(uploader, upload) do
-            ) :: {:ok, String.t()} | {:error, String.t()}
+    case uploader.put_file(upload) do
+      :ok -> {:ok, {:file, upload.path}}
+      other -> other
+    end
+  end
 end

lib/pleroma/web/activity_pub/activity_pub.ex

@@ -572,10 +572,8 @@ def fetch_activities_bounded(recipients_to, recipients_cc, opts \\ %{}) do
     |> Enum.reverse()
   end
 
-  def upload(file, size_limit \\ nil) do
+  def upload(file, opts \\ []) do
-    with data <-
+    with {:ok, data} <- Upload.store(file, opts) do
-           Upload.store(file, Application.get_env(:pleroma, :instance)[:dedupe_media], size_limit),
-         false <- is_nil(data) do
       Repo.insert(%Object{data: data})
     end
   end

lib/pleroma/web/endpoint.ex

@@ -12,7 +12,7 @@ defmodule Pleroma.Web.Endpoint do
   plug(CORSPlug)
   plug(Pleroma.Plugs.HTTPSecurityPlug)
 
-  plug(Plug.Static, at: "/media", from: Pleroma.Uploaders.Local.upload_path(), gzip: false)
+  plug(Pleroma.Plugs.UploadedMedia)
 
   plug(
     Plug.Static,

lib/pleroma/web/mastodon_api/mastodon_api_controller.ex

@@ -35,14 +35,6 @@ def create_app(conn, params) do
   def update_credentials(%{assigns: %{user: user}} = conn, params) do
     original_user = user
 
-    avatar_upload_limit =
-      Application.get_env(:pleroma, :instance)
-      |> Keyword.fetch(:avatar_upload_limit)
-
-    banner_upload_limit =
-      Application.get_env(:pleroma, :instance)
-      |> Keyword.fetch(:banner_upload_limit)
-
     params =
       if bio = params["note"] do
         Map.put(params, "bio", bio)
@@ -60,7 +52,7 @@ def update_credentials(%{assigns: %{user: user}} = conn, params) do
     user =
       if avatar = params["avatar"] do
         with %Plug.Upload{} <- avatar,
-             {:ok, object} <- ActivityPub.upload(avatar, avatar_upload_limit),
+             {:ok, object} <- ActivityPub.upload(avatar, type: :avatar),
              change = Ecto.Changeset.change(user, %{avatar: object.data}),
              {:ok, user} = User.update_and_set_cache(change) do
           user
@@ -74,7 +66,7 @@ def update_credentials(%{assigns: %{user: user}} = conn, params) do
     user =
       if banner = params["header"] do
         with %Plug.Upload{} <- banner,
-             {:ok, object} <- ActivityPub.upload(banner, banner_upload_limit),
+             {:ok, object} <- ActivityPub.upload(banner, type: :banner),
              new_info <- Map.put(user.info, "banner", object.data),
              change <- User.info_changeset(user, %{info: new_info}),
              {:ok, user} <- User.update_and_set_cache(change) do
@@ -471,19 +463,12 @@ def update_media(%{assigns: %{user: _}} = conn, data) do
   end
 
   def upload(%{assigns: %{user: _}} = conn, %{"file" => file} = data) do
-    with {:ok, object} <- ActivityPub.upload(file) do
+    with {:ok, object} <- ActivityPub.upload(file, description: Map.get(data, "description")) do
-      objdata =
+      change = Object.change(object, %{data: object.data})
-        if Map.has_key?(data, "description") do
-          Map.put(object.data, "name", data["description"])
-        else
-          object.data
-        end
-
-      change = Object.change(object, %{data: objdata})
       {:ok, object} = Repo.update(change)
 
       objdata =
-        objdata
+        object.data
         |> Map.put("id", object.id)
 
       render(conn, StatusView, "attachment.json", %{attachment: objdata})

lib/pleroma/web/media_proxy/controller.ex

@@ -1,135 +1,34 @@
 defmodule Pleroma.Web.MediaProxy.MediaProxyController do
   use Pleroma.Web, :controller
-  require Logger
+  alias Pleroma.{Web.MediaProxy, ReverseProxy}
 
-  @httpoison Application.get_env(:pleroma, :httpoison)
+  @default_proxy_opts [max_body_length: 25 * 1_048_576]
 
-  @max_body_length 25 * 1_048_576
+  def remote(conn, params = %{"sig" => sig64, "url" => url64}) do
-
+    with config <- Pleroma.Config.get([:media_proxy]),
-  @cache_control %{
+         true <- Keyword.get(config, :enabled, false),
-    default: "public, max-age=1209600",
+         {:ok, url} <- MediaProxy.decode_url(sig64, url64),
-    error: "public, must-revalidate, max-age=160"
-  }
-
-  # Content-types that will not be returned as content-disposition attachments
-  # Override with :media_proxy, :safe_content_types in the configuration
-  @safe_content_types [
-    "image/gif",
-    "image/jpeg",
-    "image/jpg",
-    "image/png",
-    "image/svg+xml",
-    "audio/mpeg",
-    "audio/mp3",
-    "video/webm",
-    "video/mp4"
-  ]
-
-  def remote(conn, params = %{"sig" => sig, "url" => url}) do
-    config = Application.get_env(:pleroma, :media_proxy, [])
-
-    with true <- Keyword.get(config, :enabled, false),
-         {:ok, url} <- Pleroma.Web.MediaProxy.decode_url(sig, url),
          filename <- Path.basename(URI.parse(url).path),
-         true <-
+         :ok <- filename_matches(Map.has_key?(params, "filename"), conn.request_path, url) do
-           if(Map.get(params, "filename"),
+      ReverseProxy.call(conn, url, Keyword.get(config, :proxy_opts, @default_proxy_length))
-             do: filename == Path.basename(conn.request_path),
-             else: true
-           ),
-         {:ok, content_type, body} <- proxy_request(url),
-         safe_content_type <-
-           Enum.member?(
-             Keyword.get(config, :safe_content_types, @safe_content_types),
-             content_type
-           ) do
-      conn
-      |> put_resp_content_type(content_type)
-      |> set_cache_header(:default)
-      |> put_resp_header(
-        "content-security-policy",
-        "default-src 'none'; style-src 'unsafe-inline'; media-src data:; img-src 'self' data:"
-      )
-      |> put_resp_header("x-xss-protection", "1; mode=block")
-      |> put_resp_header("x-content-type-options", "nosniff")
-      |> put_attachement_header(safe_content_type, filename)
-      |> send_resp(200, body)
     else
       false ->
-        send_error(conn, 404)
+        send_resp(conn, 404, Plug.Conn.Status.reason_phrase(404))
 
       {:error, :invalid_signature} ->
-        send_error(conn, 403)
+        send_resp(conn, 403, Plug.Conn.Status.reason_phrase(403))
 
-      {:error, {:http, _, url}} ->
+      {:wrong_filename, filename} ->
-        redirect_or_error(conn, url, Keyword.get(config, :redirect_on_failure, true))
+        redirect(conn, external: MediaProxy.build_url(sig64, url64, filename))
     end
   end
 
-  defp proxy_request(link) do
+  def filename_matches(has_filename, path, url) do
-    headers = [
+    filename = MediaProxy.filename(url)
-      {"user-agent",
-       "Pleroma/MediaProxy; #{Pleroma.Web.base_url()} <#{
-         Application.get_env(:pleroma, :instance)[:email]
-       }>"}
-    ]
 
-    options =
+    cond do
-      @httpoison.process_request_options([:insecure, {:follow_redirect, true}]) ++
+      has_filename && filename && Path.basename(path) != filename -> {:wrong_filename, filename}
-        [{:pool, :default}]
+      true -> :ok
-
-    with {:ok, 200, headers, client} <- :hackney.request(:get, link, headers, "", options),
-         headers = Enum.into(headers, Map.new()),
-         {:ok, body} <- proxy_request_body(client),
-         content_type <- proxy_request_content_type(headers, body) do
-      {:ok, content_type, body}
-    else
-      {:ok, status, _, _} ->
-        Logger.warn("MediaProxy: request failed, status #{status}, link: #{link}")
-        {:error, {:http, :bad_status, link}}
-
-      {:error, error} ->
-        Logger.warn("MediaProxy: request failed, error #{inspect(error)}, link: #{link}")
-        {:error, {:http, error, link}}
     end
   end
-
-  defp set_cache_header(conn, key) do
-    Plug.Conn.put_resp_header(conn, "cache-control", @cache_control[key])
-  end
-
-  defp redirect_or_error(conn, url, true), do: redirect(conn, external: url)
-  defp redirect_or_error(conn, url, _), do: send_error(conn, 502, "Media proxy error: " <> url)
-
-  defp send_error(conn, code, body \\ "") do
-    conn
-    |> set_cache_header(:error)
-    |> send_resp(code, body)
-  end
-
-  defp proxy_request_body(client), do: proxy_request_body(client, <<>>)
-
-  defp proxy_request_body(client, body) when byte_size(body) < @max_body_length do
-    case :hackney.stream_body(client) do
-      {:ok, data} -> proxy_request_body(client, <<body::binary, data::binary>>)
-      :done -> {:ok, body}
-      {:error, reason} -> {:error, reason}
-    end
-  end
-
-  defp proxy_request_body(client, _) do
-    :hackney.close(client)
-    {:error, :body_too_large}
-  end
-
-  # TODO: the body is passed here as well because some hosts do not provide a content-type.
-  # At some point we may want to use magic numbers to discover the content-type and reply a proper one.
-  defp proxy_request_content_type(headers, _body) do
-    headers["Content-Type"] || headers["content-type"] || "application/octet-stream"
-  end
-
-  defp put_attachement_header(conn, true, _), do: conn
-
-  defp put_attachement_header(conn, false, filename) do
-    put_resp_header(conn, "content-disposition", "attachment; filename='#{filename}'")
-  end
 end

lib/pleroma/web/media_proxy/media_proxy.ex

@@ -17,10 +17,8 @@ def url(url) do
       base64 = Base.url_encode64(url, @base64_opts)
       sig = :crypto.hmac(:sha, secret, base64)
       sig64 = sig |> Base.url_encode64(@base64_opts)
-      filename = if path = URI.parse(url).path, do: "/" <> Path.basename(path), else: ""
 
-      Keyword.get(config, :base_url, Pleroma.Web.base_url()) <>
+      build_url(sig64, base64, filename(url))
-        "/proxy/#{sig64}/#{base64}#{filename}"
     end
   end
 
@@ -35,4 +33,20 @@ def decode_url(sig, url) do
       {:error, :invalid_signature}
     end
   end
+
+  def filename(url_or_path) do
+    if path = URI.parse(url_or_path).path, do: Path.basename(path)
+  end
+
+  def build_url(sig_base64, url_base64, filename \\ nil) do
+    [
+      Pleroma.Config.get([:media_proxy, :base_url], Pleroma.Web.base_url()),
+      "proxy",
+      sig_base64,
+      url_base64,
+      filename
+    ]
+    |> Enum.filter(fn value -> value end)
+    |> Path.join()
+  end
 end

lib/pleroma/web/twitter_api/twitter_api.ex

@@ -97,7 +97,7 @@ def upload(%Plug.Upload{} = file, format \\ "xml") do
     {:ok, object} = ActivityPub.upload(file)
 
     url = List.first(object.data["url"])
-    href = url["href"] |> MediaProxy.url()
+    href = url["href"]
     type = url["mediaType"]
 
     case format do

lib/pleroma/web/twitter_api/twitter_api_controller.ex

@@ -290,11 +290,7 @@ def register(conn, params) do
   end
 
   def update_avatar(%{assigns: %{user: user}} = conn, params) do
-    upload_limit =
+    {:ok, object} = ActivityPub.upload(params, type: :avatar)
-      Application.get_env(:pleroma, :instance)
-      |> Keyword.fetch(:avatar_upload_limit)
-
-    {:ok, object} = ActivityPub.upload(params, upload_limit)
     change = Changeset.change(user, %{avatar: object.data})
     {:ok, user} = User.update_and_set_cache(change)
     CommonAPI.update(user)
@@ -303,11 +299,7 @@ def update_avatar(%{assigns: %{user: user}} = conn, params) do
   end
 
   def update_banner(%{assigns: %{user: user}} = conn, params) do
-    upload_limit =
+    with {:ok, object} <- ActivityPub.upload(%{"img" => params["banner"]}, type: :banner),
-      Application.get_env(:pleroma, :instance)
-      |> Keyword.fetch(:banner_upload_limit)
-
-    with {:ok, object} <- ActivityPub.upload(%{"img" => params["banner"]}, upload_limit),
          new_info <- Map.put(user.info, "banner", object.data),
          change <- User.info_changeset(user, %{info: new_info}),
          {:ok, user} <- User.update_and_set_cache(change) do
@@ -321,11 +313,7 @@ def update_banner(%{assigns: %{user: user}} = conn, params) do
   end
 
   def update_background(%{assigns: %{user: user}} = conn, params) do
-    upload_limit =
+    with {:ok, object} <- ActivityPub.upload(params, type: :background),
-      Application.get_env(:pleroma, :instance)
-      |> Keyword.fetch(:background_upload_limit)
-
-    with {:ok, object} <- ActivityPub.upload(params, upload_limit),
          new_info <- Map.put(user.info, "background", object.data),
          change <- User.info_changeset(user, %{info: new_info}),
          {:ok, _user} <- User.update_and_set_cache(change) do

test/media_proxy_test.exs

@@ -82,6 +82,23 @@ test "validates signature" do
       [_, "proxy", sig, base64 | _] = URI.parse(encoded).path |> String.split("/")
       assert decode_url(sig, base64) == {:error, :invalid_signature}
     end
+
+    test "uses the configured base_url" do
+      base_url = Pleroma.Config.get([:media_proxy, :base_url])
+
+      if base_url do
+        on_exit(fn ->
+          Pleroma.Config.put([:media_proxy, :base_url], base_url)
+        end)
+      end
+
+      Pleroma.Config.put([:media_proxy, :base_url], "https://cache.pleroma.social")
+
+      url = "https://pleroma.soykaf.com/static/logo.png"
+      encoded = url(url)
+
+      assert String.starts_with?(encoded, Pleroma.Config.get([:media_proxy, :base_url]))
+    end
   end
 
   describe "when disabled" do

test/support/httpoison_mock.ex

@@ -1,6 +1,8 @@
 defmodule HTTPoisonMock do
   alias HTTPoison.Response
 
+  def process_request_options(options), do: options
+
   def get(url, body \\ [], headers \\ [])
 
   def get("https://prismo.news/@mxb", _, _) do

test/upload_test.exs

@@ -2,7 +2,58 @@ defmodule Pleroma.UploadTest do
   alias Pleroma.Upload
   use Pleroma.DataCase
 
-  describe "Storing a file" do
+  describe "Storing a file with the Local uploader" do
+    setup do
+      uploader = Pleroma.Config.get([Pleroma.Upload, :uploader])
+      filters = Pleroma.Config.get([Pleroma.Upload, :filters])
+
+      unless uploader == Pleroma.Uploaders.Local || filters != [] do
+        Pleroma.Config.put([Pleroma.Upload, :uploader], Pleroma.Uploaders.Local)
+        Pleroma.Config.put([Pleroma.Upload, :filters], [])
+
+        on_exit(fn ->
+          Pleroma.Config.put([Pleroma.Upload, :uploader], uploader)
+          Pleroma.Config.put([Pleroma.Upload, :filters], filters)
+        end)
+      end
+
+      :ok
+    end
+
+    test "returns a media url" do
+      File.cp!("test/fixtures/image.jpg", "test/fixtures/image_tmp.jpg")
+
+      file = %Plug.Upload{
+        content_type: "image/jpg",
+        path: Path.absname("test/fixtures/image_tmp.jpg"),
+        filename: "image.jpg"
+      }
+
+      {:ok, data} = Upload.store(file)
+
+      assert %{"url" => [%{"href" => url}]} = data
+
+      assert String.starts_with?(url, Pleroma.Web.base_url() <> "/media/")
+    end
+
+    test "returns a media url with configured base_url" do
+      base_url = "https://cache.pleroma.social"
+
+      File.cp!("test/fixtures/image.jpg", "test/fixtures/image_tmp.jpg")
+
+      file = %Plug.Upload{
+        content_type: "image/jpg",
+        path: Path.absname("test/fixtures/image_tmp.jpg"),
+        filename: "image.jpg"
+      }
+
+      {:ok, data} = Upload.store(file, base_url: base_url)
+
+      assert %{"url" => [%{"href" => url}]} = data
+
+      assert String.starts_with?(url, base_url <> "/media/")
+    end
+
     test "copies the file to the configured folder with deduping" do
       File.cp!("test/fixtures/image.jpg", "test/fixtures/image_tmp.jpg")
 
@@ -12,10 +63,11 @@ test "copies the file to the configured folder with deduping" do
         filename: "an [image.jpg"
       }
 
-      data = Upload.store(file, true)
+      {:ok, data} = Upload.store(file, filters: [Pleroma.Upload.Filter.Dedupe])
 
-      assert data["name"] ==
+      assert List.first(data["url"])["href"] ==
-               "e7a6d0cf595bff76f14c9a98b6c199539559e8b844e02e51e5efcfd1f614a2df.jpeg"
+               Pleroma.Web.base_url() <>
+                 "/media/e7a6d0cf595bff76f14c9a98b6c199539559e8b844e02e51e5efcfd1f614a2df.jpg"
     end
 
     test "copies the file to the configured folder without deduping" do
@@ -27,7 +79,7 @@ test "copies the file to the configured folder without deduping" do
         filename: "an [image.jpg"
       }
 
-      data = Upload.store(file, false)
+      {:ok, data} = Upload.store(file)
       assert data["name"] == "an [image.jpg"
     end
 
@@ -40,7 +92,7 @@ test "fixes incorrect content type" do
         filename: "an [image.jpg"
       }
 
-      data = Upload.store(file, true)
+      {:ok, data} = Upload.store(file, filters: [Pleroma.Upload.Filter.Dedupe])
       assert hd(data["url"])["mediaType"] == "image/jpeg"
     end
 
@@ -53,7 +105,7 @@ test "adds missing extension" do
         filename: "an [image"
       }
 
-      data = Upload.store(file, false)
+      {:ok, data} = Upload.store(file)
       assert data["name"] == "an [image.jpg"
     end
 
@@ -66,7 +118,7 @@ test "fixes incorrect file extension" do
         filename: "an [image.blah"
       }
 
-      data = Upload.store(file, false)
+      {:ok, data} = Upload.store(file)
       assert data["name"] == "an [image.jpg"
     end
 
@@ -79,7 +131,7 @@ test "don't modify filename of an unknown type" do
         filename: "test.txt"
       }
 
-      data = Upload.store(file, false)
+      {:ok, data} = Upload.store(file)
       assert data["name"] == "test.txt"
     end
   end