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       ensure_scrubbed_html(content, scrubbers, activity.data["object"]["fake"] || false)
  36     end)
  37   end
  38
  39   def get_cached_stripped_html_for_activity(content, activity, key) do
  40     get_cached_scrubbed_html_for_activity(
  41       content,
  42       HtmlSanitizeEx.Scrubber.StripTags,
  43       activity,
  44       key
  45     )
  46   end
  47
  48   def ensure_scrubbed_html(
  49         content,
  50         scrubbers,
  51         false = _fake
  52       ) do
  53     {:commit, filter_tags(content, scrubbers)}
  54   end
  55
  56   def ensure_scrubbed_html(
  57         content,
  58         scrubbers,
  59         true = _fake
  60       ) do
  61     {:ignore, filter_tags(content, scrubbers)}
  62   end
  63
  64   defp generate_scrubber_signature(scrubber) when is_atom(scrubber) do
  65     generate_scrubber_signature([scrubber])
  66   end
  67
  68   defp generate_scrubber_signature(scrubbers) do
  69     Enum.reduce(scrubbers, "", fn scrubber, signature ->
  70       "#{signature}#{to_string(scrubber)}"
  71     end)
  72   end
  73
  74   def extract_first_external_url(_, nil), do: {:error, "No content"}
  75
  76   def extract_first_external_url(object, content) do
  77     key = "URL|#{object.id}"
  78
  79     Cachex.fetch!(:scrubber_cache, key, fn _key ->
  80       result =
  81         content
  82         |> Floki.filter_out("a.mention")
  83         |> Floki.attribute("a", "href")
  84         |> Enum.at(0)
  85
  86       {:commit, {:ok, result}}
  87     end)
  88   end
  89 end
  90
  91 defmodule Pleroma.HTML.Scrubber.TwitterText do
  92   @moduledoc """
  93   An HTML scrubbing policy which limits to twitter-style text.  Only
  94   paragraphs, breaks and links are allowed through the filter.
  95   """
  96
  97   @markup Application.get_env(:pleroma, :markup)
  98   @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
  99
 100   require HtmlSanitizeEx.Scrubber.Meta
 101   alias HtmlSanitizeEx.Scrubber.Meta
 102
 103   Meta.remove_cdata_sections_before_scrub()
 104   Meta.strip_comments()
 105
 106   # links
 107   Meta.allow_tag_with_uri_attributes("a", ["href", "data-user", "data-tag"], @valid_schemes)
 108
 109   Meta.allow_tag_with_this_attribute_values("a", "class", [
 110     "hashtag",
 111     "u-url",
 112     "mention",
 113     "u-url mention",
 114     "mention u-url"
 115   ])
 116
 117   Meta.allow_tag_with_this_attribute_values("a", "rel", [
 118     "tag",
 119     "nofollow",
 120     "noopener",
 121     "noreferrer"
 122   ])
 123
 124   Meta.allow_tag_with_these_attributes("a", ["name", "title"])
 125
 126   # paragraphs and linebreaks
 127   Meta.allow_tag_with_these_attributes("br", [])
 128   Meta.allow_tag_with_these_attributes("p", [])
 129
 130   # microformats
 131   Meta.allow_tag_with_this_attribute_values("span", "class", ["h-card"])
 132   Meta.allow_tag_with_these_attributes("span", [])
 133
 134   # allow inline images for custom emoji
 135   @allow_inline_images Keyword.get(@markup, :allow_inline_images)
 136
 137   if @allow_inline_images do
 138     # restrict img tags to http/https only, because of MediaProxy.
 139     Meta.allow_tag_with_uri_attributes("img", ["src"], ["http", "https"])
 140
 141     Meta.allow_tag_with_these_attributes("img", [
 142       "width",
 143       "height",
 144       "title",
 145       "alt"
 146     ])
 147   end
 148
 149   Meta.strip_everything_not_covered()
 150 end
 151
 152 defmodule Pleroma.HTML.Scrubber.Default do
 153   @doc "The default HTML scrubbing policy: no "
 154
 155   require HtmlSanitizeEx.Scrubber.Meta
 156   alias HtmlSanitizeEx.Scrubber.Meta
 157   # credo:disable-for-previous-line
 158   # No idea how to fix this one…
 159
 160   @markup Application.get_env(:pleroma, :markup)
 161   @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
 162
 163   Meta.remove_cdata_sections_before_scrub()
 164   Meta.strip_comments()
 165
 166   Meta.allow_tag_with_uri_attributes("a", ["href", "data-user", "data-tag"], @valid_schemes)
 167
 168   Meta.allow_tag_with_this_attribute_values("a", "class", [
 169     "hashtag",
 170     "u-url",
 171     "mention",
 172     "u-url mention",
 173     "mention u-url"
 174   ])
 175
 176   Meta.allow_tag_with_this_attribute_values("a", "rel", [
 177     "tag",
 178     "nofollow",
 179     "noopener",
 180     "noreferrer"
 181   ])
 182
 183   Meta.allow_tag_with_these_attributes("a", ["name", "title"])
 184
 185   Meta.allow_tag_with_these_attributes("abbr", ["title"])
 186
 187   Meta.allow_tag_with_these_attributes("b", [])
 188   Meta.allow_tag_with_these_attributes("blockquote", [])
 189   Meta.allow_tag_with_these_attributes("br", [])
 190   Meta.allow_tag_with_these_attributes("code", [])
 191   Meta.allow_tag_with_these_attributes("del", [])
 192   Meta.allow_tag_with_these_attributes("em", [])
 193   Meta.allow_tag_with_these_attributes("i", [])
 194   Meta.allow_tag_with_these_attributes("li", [])
 195   Meta.allow_tag_with_these_attributes("ol", [])
 196   Meta.allow_tag_with_these_attributes("p", [])
 197   Meta.allow_tag_with_these_attributes("pre", [])
 198   Meta.allow_tag_with_these_attributes("strong", [])
 199   Meta.allow_tag_with_these_attributes("u", [])
 200   Meta.allow_tag_with_these_attributes("ul", [])
 201
 202   Meta.allow_tag_with_this_attribute_values("span", "class", ["h-card"])
 203   Meta.allow_tag_with_these_attributes("span", [])
 204
 205   @allow_inline_images Keyword.get(@markup, :allow_inline_images)
 206
 207   if @allow_inline_images do
 208     # restrict img tags to http/https only, because of MediaProxy.
 209     Meta.allow_tag_with_uri_attributes("img", ["src"], ["http", "https"])
 210
 211     Meta.allow_tag_with_these_attributes("img", [
 212       "width",
 213       "height",
 214       "title",
 215       "alt"
 216     ])
 217   end
 218
 219   @allow_tables Keyword.get(@markup, :allow_tables)
 220
 221   if @allow_tables do
 222     Meta.allow_tag_with_these_attributes("table", [])
 223     Meta.allow_tag_with_these_attributes("tbody", [])
 224     Meta.allow_tag_with_these_attributes("td", [])
 225     Meta.allow_tag_with_these_attributes("th", [])
 226     Meta.allow_tag_with_these_attributes("thead", [])
 227     Meta.allow_tag_with_these_attributes("tr", [])
 228   end
 229
 230   @allow_headings Keyword.get(@markup, :allow_headings)
 231
 232   if @allow_headings do
 233     Meta.allow_tag_with_these_attributes("h1", [])
 234     Meta.allow_tag_with_these_attributes("h2", [])
 235     Meta.allow_tag_with_these_attributes("h3", [])
 236     Meta.allow_tag_with_these_attributes("h4", [])
 237     Meta.allow_tag_with_these_attributes("h5", [])
 238   end
 239
 240   @allow_fonts Keyword.get(@markup, :allow_fonts)
 241
 242   if @allow_fonts do
 243     Meta.allow_tag_with_these_attributes("font", ["face"])
 244   end
 245
 246   Meta.strip_everything_not_covered()
 247 end
 248
 249 defmodule Pleroma.HTML.Transform.MediaProxy do
 250   @moduledoc "Transforms inline image URIs to use MediaProxy."
 251
 252   alias Pleroma.Web.MediaProxy
 253
 254   def before_scrub(html), do: html
 255
 256   def scrub_attribute("img", {"src", "http" <> target}) do
 257     media_url =
 258       ("http" <> target)
 259       |> MediaProxy.url()
 260
 261     {"src", media_url}
 262   end
 263
 264   def scrub_attribute(_tag, attribute), do: attribute
 265
 266   def scrub({"img", attributes, children}) do
 267     attributes =
 268       attributes
 269       |> Enum.map(fn attr -> scrub_attribute("img", attr) end)
 270       |> Enum.reject(&is_nil(&1))
 271
 272     {"img", attributes, children}
 273   end
 274
 275   def scrub({:comment, _children}), do: ""
 276
 277   def scrub({tag, attributes, children}), do: {tag, attributes, children}
 278   def scrub({_tag, children}), do: children
 279   def scrub(text), do: text
 280 end