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

   1 # Pleroma: A lightweight social networking server
   2 # Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
   3 # SPDX-License-Identifier: AGPL-3.0-only
   4
   5 defmodule Pleroma.Web.RichMedia.Parser do
   6   require Logger
   7
   8   @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
   9
  10   defp parsers do
  11     Pleroma.Config.get([:rich_media, :parsers])
  12   end
  13
  14   def parse(nil), do: {:error, "No URL provided"}
  15
  16   if Pleroma.Config.get(:env) == :test do
  17     @spec parse(String.t()) :: {:ok, map()} | {:error, any()}
  18     def parse(url), do: parse_url(url)
  19   else
  20     @spec parse(String.t()) :: {:ok, map()} | {:error, any()}
  21     def parse(url) do
  22       with {:ok, data} <- get_cached_or_parse(url),
  23            {:ok, _} <- set_ttl_based_on_image(data, url) do
  24         {:ok, data}
  25       end
  26     end
  27
  28     defp get_cached_or_parse(url) do
  29       case @cachex.fetch(:rich_media_cache, url, fn ->
  30              case parse_url(url) do
  31                {:ok, _} = res ->
  32                  {:commit, res}
  33
  34                {:error, reason} = e ->
  35                  # Unfortunately we have to log errors here, instead of doing that
  36                  # along with ttl setting at the bottom. Otherwise we can get log spam
  37                  # if more than one process was waiting for the rich media card
  38                  # while it was generated. Ideally we would set ttl here as well,
  39                  # so we don't override it number_of_waiters_on_generation
  40                  # times, but one, obviously, can't set ttl for not-yet-created entry
  41                  # and Cachex doesn't support returning ttl from the fetch callback.
  42                  log_error(url, reason)
  43                  {:commit, e}
  44              end
  45            end) do
  46         {action, res} when action in [:commit, :ok] ->
  47           case res do
  48             {:ok, _data} = res ->
  49               res
  50
  51             {:error, reason} = e ->
  52               if action == :commit, do: set_error_ttl(url, reason)
  53               e
  54           end
  55
  56         {:error, e} ->
  57           {:error, {:cachex_error, e}}
  58       end
  59     end
  60
  61     defp set_error_ttl(_url, :body_too_large), do: :ok
  62     defp set_error_ttl(_url, {:content_type, _}), do: :ok
  63
  64     # The TTL is not set for the errors above, since they are unlikely to change
  65     # with time
  66
  67     defp set_error_ttl(url, _reason) do
  68       ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
  69       @cachex.expire(:rich_media_cache, url, ttl)
  70       :ok
  71     end
  72
  73     defp log_error(url, {:invalid_metadata, data}) do
  74       Logger.debug(fn -> "Incomplete or invalid metadata for #{url}: #{inspect(data)}" end)
  75     end
  76
  77     defp log_error(url, reason) do
  78       Logger.warn(fn -> "Rich media error for #{url}: #{inspect(reason)}" end)
  79     end
  80   end
  81
  82   @doc """
  83   Set the rich media cache based on the expiration time of image.
  84
  85   Adopt behaviour `Pleroma.Web.RichMedia.Parser.TTL`
  86
  87   ## Example
  88
  89       defmodule MyModule do
  90         @behaviour Pleroma.Web.RichMedia.Parser.TTL
  91         def ttl(data, url) do
  92           image_url = Map.get(data, :image)
  93           # do some parsing in the url and get the ttl of the image
  94           # and return ttl is unix time
  95           parse_ttl_from_url(image_url)
  96         end
  97       end
  98
  99   Define the module in the config
 100
 101       config :pleroma, :rich_media,
 102         ttl_setters: [MyModule]
 103   """
 104   @spec set_ttl_based_on_image(map(), String.t()) ::
 105           {:ok, Integer.t() | :noop} | {:error, :no_key}
 106   def set_ttl_based_on_image(data, url) do
 107     case get_ttl_from_image(data, url) do
 108       {:ok, ttl} when is_number(ttl) ->
 109         ttl = ttl * 1000
 110
 111         case @cachex.expire_at(:rich_media_cache, url, ttl) do
 112           {:ok, true} -> {:ok, ttl}
 113           {:ok, false} -> {:error, :no_key}
 114         end
 115
 116       _ ->
 117         {:ok, :noop}
 118     end
 119   end
 120
 121   defp get_ttl_from_image(data, url) do
 122     [:rich_media, :ttl_setters]
 123     |> Pleroma.Config.get()
 124     |> Enum.reduce({:ok, nil}, fn
 125       module, {:ok, _ttl} ->
 126         module.ttl(data, url)
 127
 128       _, error ->
 129         error
 130     end)
 131   end
 132
 133   def parse_url(url) do
 134     with {:ok, %Tesla.Env{body: html}} <- Pleroma.Web.RichMedia.Helpers.rich_media_get(url),
 135          {:ok, html} <- Floki.parse_document(html) do
 136       html
 137       |> maybe_parse()
 138       |> Map.put("url", url)
 139       |> clean_parsed_data()
 140       |> check_parsed_data()
 141     end
 142   end
 143
 144   defp maybe_parse(html) do
 145     Enum.reduce_while(parsers(), %{}, fn parser, acc ->
 146       case parser.parse(html, acc) do
 147         data when data != %{} -> {:halt, data}
 148         _ -> {:cont, acc}
 149       end
 150     end)
 151   end
 152
 153   defp check_parsed_data(%{"title" => title} = data)
 154        when is_binary(title) and title != "" do
 155     {:ok, data}
 156   end
 157
 158   defp check_parsed_data(data) do
 159     {:error, {:invalid_metadata, data}}
 160   end
 161
 162   defp clean_parsed_data(data) do
 163     data
 164     |> Enum.reject(fn {key, val} ->
 165       not match?({:ok, _}, Jason.encode(%{key => val}))
 166     end)
 167     |> Map.new()
 168   end
 169 end