git.squeep.com Git - akkoma/blob - lib/pleroma/html.ex

   1 # Pleroma: A lightweight social networking server
   2 # Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
   3 # SPDX-License-Identifier: AGPL-3.0-only
   4
   5 defmodule Pleroma.HTML do
   6   alias HtmlSanitizeEx.Scrubber
   7
   8   defp get_scrubbers(scrubber) when is_atom(scrubber), do: [scrubber]
   9   defp get_scrubbers(scrubbers) when is_list(scrubbers), do: scrubbers
  10   defp get_scrubbers(_), do: [Pleroma.HTML.Scrubber.Default]
  11
  12   def get_scrubbers do
  13     Pleroma.Config.get([:markup, :scrub_policy])
  14     |> get_scrubbers
  15   end
  16
  17   def filter_tags(html, nil) do
  18     filter_tags(html, get_scrubbers())
  19   end
  20
  21   def filter_tags(html, scrubbers) when is_list(scrubbers) do
  22     Enum.reduce(scrubbers, html, fn scrubber, html ->
  23       filter_tags(html, scrubber)
  24     end)
  25   end
  26
  27   def filter_tags(html, scrubber), do: Scrubber.scrub(html, scrubber)
  28   def filter_tags(html), do: filter_tags(html, nil)
  29   def strip_tags(html), do: Scrubber.scrub(html, Scrubber.StripTags)
  30
  31   def get_cached_scrubbed_html_for_activity(content, scrubbers, activity, key \\ "") do
  32     key = "#{key}#{generate_scrubber_signature(scrubbers)}|#{activity.id}"
  33
  34     Cachex.fetch!(:scrubber_cache, key, fn _key ->
  35       object = Pleroma.Object.normalize(activity)
  36       ensure_scrubbed_html(content, scrubbers, object.data["fake"] || false)
  37     end)
  38   end
  39
  40   def get_cached_stripped_html_for_activity(content, activity, key) do
  41     get_cached_scrubbed_html_for_activity(
  42       content,
  43       HtmlSanitizeEx.Scrubber.StripTags,
  44       activity,
  45       key
  46     )
  47   end
  48
  49   def ensure_scrubbed_html(
  50         content,
  51         scrubbers,
  52         false = _fake
  53       ) do
  54     {:commit, filter_tags(content, scrubbers)}
  55   end
  56
  57   def ensure_scrubbed_html(
  58         content,
  59         scrubbers,
  60         true = _fake
  61       ) do
  62     {:ignore, filter_tags(content, scrubbers)}
  63   end
  64
  65   defp generate_scrubber_signature(scrubber) when is_atom(scrubber) do
  66     generate_scrubber_signature([scrubber])
  67   end
  68
  69   defp generate_scrubber_signature(scrubbers) do
  70     Enum.reduce(scrubbers, "", fn scrubber, signature ->
  71       "#{signature}#{to_string(scrubber)}"
  72     end)
  73   end
  74
  75   def extract_first_external_url(_, nil), do: {:error, "No content"}
  76
  77   def extract_first_external_url(object, content) do
  78     key = "URL|#{object.id}"
  79
  80     Cachex.fetch!(:scrubber_cache, key, fn _key ->
  81       result =
  82         content
  83         |> Floki.filter_out("a.mention")
  84         |> Floki.attribute("a", "href")
  85         |> Enum.at(0)
  86
  87       {:commit, {:ok, result}}
  88     end)
  89   end
  90 end
  91
  92 defmodule Pleroma.HTML.Scrubber.TwitterText do
  93   @moduledoc """
  94   An HTML scrubbing policy which limits to twitter-style text.  Only
  95   paragraphs, breaks and links are allowed through the filter.
  96   """
  97
  98   @markup Application.get_env(:pleroma, :markup)
  99   @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
 100
 101   require HtmlSanitizeEx.Scrubber.Meta
 102   alias HtmlSanitizeEx.Scrubber.Meta
 103
 104   Meta.remove_cdata_sections_before_scrub()
 105   Meta.strip_comments()
 106
 107   # links
 108   Meta.allow_tag_with_uri_attributes("a", ["href", "data-user", "data-tag"], @valid_schemes)
 109   Meta.allow_tag_with_these_attributes("a", ["name", "title", "class"])
 110
 111   Meta.allow_tag_with_this_attribute_values("a", "rel", [
 112     "tag",
 113     "nofollow",
 114     "noopener",
 115     "noreferrer"
 116   ])
 117
 118   # paragraphs and linebreaks
 119   Meta.allow_tag_with_these_attributes("br", [])
 120   Meta.allow_tag_with_these_attributes("p", [])
 121
 122   # microformats
 123   Meta.allow_tag_with_these_attributes("span", ["class"])
 124
 125   # allow inline images for custom emoji
 126   @allow_inline_images Keyword.get(@markup, :allow_inline_images)
 127
 128   if @allow_inline_images do
 129     # restrict img tags to http/https only, because of MediaProxy.
 130     Meta.allow_tag_with_uri_attributes("img", ["src"], ["http", "https"])
 131
 132     Meta.allow_tag_with_these_attributes("img", [
 133       "width",
 134       "height",
 135       "title",
 136       "alt"
 137     ])
 138   end
 139
 140   Meta.strip_everything_not_covered()
 141 end
 142
 143 defmodule Pleroma.HTML.Scrubber.Default do
 144   @doc "The default HTML scrubbing policy: no "
 145
 146   require HtmlSanitizeEx.Scrubber.Meta
 147   alias HtmlSanitizeEx.Scrubber.Meta
 148   # credo:disable-for-previous-line
 149   # No idea how to fix this one…
 150
 151   @markup Application.get_env(:pleroma, :markup)
 152   @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
 153
 154   Meta.remove_cdata_sections_before_scrub()
 155   Meta.strip_comments()
 156
 157   Meta.allow_tag_with_uri_attributes("a", ["href", "data-user", "data-tag"], @valid_schemes)
 158   Meta.allow_tag_with_these_attributes("a", ["name", "title", "class"])
 159
 160   Meta.allow_tag_with_this_attribute_values("a", "rel", [
 161     "tag",
 162     "nofollow",
 163     "noopener",
 164     "noreferrer"
 165   ])
 166
 167   Meta.allow_tag_with_these_attributes("abbr", ["title"])
 168
 169   Meta.allow_tag_with_these_attributes("b", [])
 170   Meta.allow_tag_with_these_attributes("blockquote", [])
 171   Meta.allow_tag_with_these_attributes("br", [])
 172   Meta.allow_tag_with_these_attributes("code", [])
 173   Meta.allow_tag_with_these_attributes("del", [])
 174   Meta.allow_tag_with_these_attributes("em", [])
 175   Meta.allow_tag_with_these_attributes("i", [])
 176   Meta.allow_tag_with_these_attributes("li", [])
 177   Meta.allow_tag_with_these_attributes("ol", [])
 178   Meta.allow_tag_with_these_attributes("p", [])
 179   Meta.allow_tag_with_these_attributes("pre", [])
 180   Meta.allow_tag_with_these_attributes("span", ["class"])
 181   Meta.allow_tag_with_these_attributes("strong", [])
 182   Meta.allow_tag_with_these_attributes("u", [])
 183   Meta.allow_tag_with_these_attributes("ul", [])
 184
 185   @allow_inline_images Keyword.get(@markup, :allow_inline_images)
 186
 187   if @allow_inline_images do
 188     # restrict img tags to http/https only, because of MediaProxy.
 189     Meta.allow_tag_with_uri_attributes("img", ["src"], ["http", "https"])
 190
 191     Meta.allow_tag_with_these_attributes("img", [
 192       "width",
 193       "height",
 194       "title",
 195       "alt"
 196     ])
 197   end
 198
 199   @allow_tables Keyword.get(@markup, :allow_tables)
 200
 201   if @allow_tables do
 202     Meta.allow_tag_with_these_attributes("table", [])
 203     Meta.allow_tag_with_these_attributes("tbody", [])
 204     Meta.allow_tag_with_these_attributes("td", [])
 205     Meta.allow_tag_with_these_attributes("th", [])
 206     Meta.allow_tag_with_these_attributes("thead", [])
 207     Meta.allow_tag_with_these_attributes("tr", [])
 208   end
 209
 210   @allow_headings Keyword.get(@markup, :allow_headings)
 211
 212   if @allow_headings do
 213     Meta.allow_tag_with_these_attributes("h1", [])
 214     Meta.allow_tag_with_these_attributes("h2", [])
 215     Meta.allow_tag_with_these_attributes("h3", [])
 216     Meta.allow_tag_with_these_attributes("h4", [])
 217     Meta.allow_tag_with_these_attributes("h5", [])
 218   end
 219
 220   @allow_fonts Keyword.get(@markup, :allow_fonts)
 221
 222   if @allow_fonts do
 223     Meta.allow_tag_with_these_attributes("font", ["face"])
 224   end
 225
 226   Meta.strip_everything_not_covered()
 227 end
 228
 229 defmodule Pleroma.HTML.Transform.MediaProxy do
 230   @moduledoc "Transforms inline image URIs to use MediaProxy."
 231
 232   alias Pleroma.Web.MediaProxy
 233
 234   def before_scrub(html), do: html
 235
 236   def scrub_attribute("img", {"src", "http" <> target}) do
 237     media_url =
 238       ("http" <> target)
 239       |> MediaProxy.url()
 240
 241     {"src", media_url}
 242   end
 243
 244   def scrub_attribute(_tag, attribute), do: attribute
 245
 246   def scrub({"img", attributes, children}) do
 247     attributes =
 248       attributes
 249       |> Enum.map(fn attr -> scrub_attribute("img", attr) end)
 250       |> Enum.reject(&is_nil(&1))
 251
 252     {"img", attributes, children}
 253   end
 254
 255   def scrub({:comment, _children}), do: ""
 256
 257   def scrub({tag, attributes, children}), do: {tag, attributes, children}
 258   def scrub({_tag, children}), do: children
 259   def scrub(text), do: text
 260 end