git.squeep.com Git - akkoma/blob - lib/pleroma/web/rich_media/helpers.ex

   1 # Pleroma: A lightweight social networking server
   2 # Copyright _ 2017-2020 Pleroma Authors <https://pleroma.social/>
   3 # SPDX-License-Identifier: AGPL-3.0-only
   4
   5 defmodule Pleroma.Web.RichMedia.Helpers do
   6   alias Pleroma.Activity
   7   alias Pleroma.Config
   8   alias Pleroma.HTML
   9   alias Pleroma.Object
  10   alias Pleroma.Web.RichMedia.Parser
  11
  12   @options [
  13     pool: :media,
  14     max_body: 2_000_000,
  15     recv_timeout: 2_000
  16   ]
  17
  18   @spec validate_page_url(URI.t() | binary()) :: :ok | :error
  19   defp validate_page_url(page_url) when is_binary(page_url) do
  20     validate_tld = Config.get([Pleroma.Formatter, :validate_tld])
  21
  22     page_url
  23     |> Linkify.Parser.url?(validate_tld: validate_tld)
  24     |> parse_uri(page_url)
  25   end
  26
  27   defp validate_page_url(%URI{host: host, scheme: "https", authority: authority})
  28        when is_binary(authority) do
  29     cond do
  30       host in Config.get([:rich_media, :ignore_hosts], []) ->
  31         :error
  32
  33       get_tld(host) in Config.get([:rich_media, :ignore_tld], []) ->
  34         :error
  35
  36       true ->
  37         :ok
  38     end
  39   end
  40
  41   defp validate_page_url(_), do: :error
  42
  43   defp parse_uri(true, url) do
  44     url
  45     |> URI.parse()
  46     |> validate_page_url
  47   end
  48
  49   defp parse_uri(_, _), do: :error
  50
  51   defp get_tld(host) do
  52     host
  53     |> String.split(".")
  54     |> Enum.reverse()
  55     |> hd
  56   end
  57
  58   def fetch_data_for_object(object) do
  59     with true <- Config.get([:rich_media, :enabled]),
  60          {:ok, page_url} <-
  61            HTML.extract_first_external_url_from_object(object),
  62          :ok <- validate_page_url(page_url),
  63          {:ok, rich_media} <- Parser.parse(page_url) do
  64       %{page_url: page_url, rich_media: rich_media}
  65     else
  66       _ -> %{}
  67     end
  68   end
  69
  70   def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
  71     with true <- Config.get([:rich_media, :enabled]),
  72          %Object{} = object <- Object.normalize(activity, fetch: false) do
  73       fetch_data_for_object(object)
  74     else
  75       _ -> %{}
  76     end
  77   end
  78
  79   def fetch_data_for_activity(_), do: %{}
  80
  81   def rich_media_get(url) do
  82     headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
  83
  84     head_check =
  85       case Pleroma.HTTP.head(url, headers, @options) do
  86         # If the HEAD request didn't reach the server for whatever reason,
  87         # we assume the GET that comes right after won't either
  88         {:error, _} = e ->
  89           e
  90
  91         {:ok, %Tesla.Env{status: 200, headers: headers}} ->
  92           with :ok <- check_content_type(headers),
  93                :ok <- check_content_length(headers),
  94                do: :ok
  95
  96         _ ->
  97           :ok
  98       end
  99
 100     with :ok <- head_check, do: Pleroma.HTTP.get(url, headers, @options)
 101   end
 102
 103   defp check_content_type(headers) do
 104     case List.keyfind(headers, "content-type", 0) do
 105       {_, content_type} ->
 106         case Plug.Conn.Utils.media_type(content_type) do
 107           {:ok, "text", "html", _} -> :ok
 108           _ -> {:error, {:content_type, content_type}}
 109         end
 110
 111       _ ->
 112         :ok
 113     end
 114   end
 115
 116   @max_body @options[:max_body]
 117   defp check_content_length(headers) do
 118     case List.keyfind(headers, "content-length", 0) do
 119       {_, maybe_content_length} ->
 120         case Integer.parse(maybe_content_length) do
 121           {content_length, ""} when content_length <= @max_body -> :ok
 122           {_, ""} -> {:error, :body_too_large}
 123           _ -> :ok
 124         end
 125
 126       _ ->
 127         :ok
 128     end
 129   end
 130 end