git.squeep.com Git - akkoma/blob - lib/pleroma/html.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.HTML do
   6   # Scrubbers are compiled on boot so they can be configured in OTP releases
   7   #  @on_load :compile_scrubbers
   8
   9   def compile_scrubbers do
  10     dir = Path.join(:code.priv_dir(:pleroma), "scrubbers")
  11
  12     dir
  13     |> Pleroma.Utils.compile_dir()
  14     |> case do
  15       {:error, _errors, _warnings} ->
  16         raise "Compiling scrubbers failed"
  17
  18       {:ok, _modules, _warnings} ->
  19         :ok
  20     end
  21   end
  22
  23   defp get_scrubbers(scrubber) when is_atom(scrubber), do: [scrubber]
  24   defp get_scrubbers(scrubbers) when is_list(scrubbers), do: scrubbers
  25   defp get_scrubbers(_), do: [Pleroma.HTML.Scrubber.Default]
  26
  27   def get_scrubbers do
  28     Pleroma.Config.get([:markup, :scrub_policy])
  29     |> get_scrubbers
  30   end
  31
  32   def filter_tags(html, nil) do
  33     filter_tags(html, get_scrubbers())
  34   end
  35
  36   def filter_tags(html, scrubbers) when is_list(scrubbers) do
  37     Enum.reduce(scrubbers, html, fn scrubber, html ->
  38       filter_tags(html, scrubber)
  39     end)
  40   end
  41
  42   def filter_tags(html, scrubber) do
  43     {:ok, content} = FastSanitize.Sanitizer.scrub(html, scrubber)
  44     content
  45   end
  46
  47   def filter_tags(html), do: filter_tags(html, nil)
  48   def strip_tags(html), do: filter_tags(html, FastSanitize.Sanitizer.StripTags)
  49
  50   def get_cached_scrubbed_html_for_activity(
  51         content,
  52         scrubbers,
  53         activity,
  54         key \\ "",
  55         callback \\ fn x -> x end
  56       ) do
  57     key = "#{key}#{generate_scrubber_signature(scrubbers)}|#{activity.id}"
  58
  59     Cachex.fetch!(:scrubber_cache, key, fn _key ->
  60       object = Pleroma.Object.normalize(activity)
  61       ensure_scrubbed_html(content, scrubbers, object.data["fake"] || false, callback)
  62     end)
  63   end
  64
  65   def get_cached_stripped_html_for_activity(content, activity, key) do
  66     get_cached_scrubbed_html_for_activity(
  67       content,
  68       FastSanitize.Sanitizer.StripTags,
  69       activity,
  70       key,
  71       &HtmlEntities.decode/1
  72     )
  73   end
  74
  75   def ensure_scrubbed_html(
  76         content,
  77         scrubbers,
  78         fake,
  79         callback
  80       ) do
  81     content =
  82       content
  83       |> filter_tags(scrubbers)
  84       |> callback.()
  85
  86     if fake do
  87       {:ignore, content}
  88     else
  89       {:commit, content}
  90     end
  91   end
  92
  93   defp generate_scrubber_signature(scrubber) when is_atom(scrubber) do
  94     generate_scrubber_signature([scrubber])
  95   end
  96
  97   defp generate_scrubber_signature(scrubbers) do
  98     Enum.reduce(scrubbers, "", fn scrubber, signature ->
  99       "#{signature}#{to_string(scrubber)}"
 100     end)
 101   end
 102
 103   def extract_first_external_url(_, nil), do: {:error, "No content"}
 104
 105   def extract_first_external_url(object, content) do
 106     key = "URL|#{object.id}"
 107
 108     Cachex.fetch!(:scrubber_cache, key, fn _key ->
 109       result =
 110         content
 111         |> Floki.parse_fragment!()
 112         |> Floki.filter_out("a.mention,a.hashtag,a.attachment,a[rel~=\"tag\"]")
 113         |> Floki.attribute("a", "href")
 114         |> Enum.at(0)
 115
 116       {:commit, {:ok, result}}
 117     end)
 118   end
 119 end