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

   1 # Pleroma: A lightweight social networking server
   2 # Copyright © 2017-2018 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     get_scrubbers()
  19     |> Enum.reduce(html, fn scrubber, html ->
  20       filter_tags(html, scrubber)
  21     end)
  22   end
  23
  24   def filter_tags(html, scrubber), do: Scrubber.scrub(html, scrubber)
  25   def filter_tags(html), do: filter_tags(html, nil)
  26   def strip_tags(html), do: Scrubber.scrub(html, Scrubber.StripTags)
  27 end
  28
  29 defmodule Pleroma.HTML.Scrubber.TwitterText do
  30   @moduledoc """
  31   An HTML scrubbing policy which limits to twitter-style text.  Only
  32   paragraphs, breaks and links are allowed through the filter.
  33   """
  34
  35   @markup Application.get_env(:pleroma, :markup)
  36   @uri_schemes Application.get_env(:pleroma, :uri_schemes, [])
  37   @valid_schemes Keyword.get(@uri_schemes, :valid_schemes, [])
  38
  39   require HtmlSanitizeEx.Scrubber.Meta
  40   alias HtmlSanitizeEx.Scrubber.Meta
  41
  42   def version do
  43     0
  44   end
  45
  46   Meta.remove_cdata_sections_before_scrub()
  47   Meta.strip_comments()
  48
  49   # links
  50   Meta.allow_tag_with_uri_attributes("a", ["href", "data-user", "data-tag"], @valid_schemes)
  51   Meta.allow_tag_with_these_attributes("a", ["name", "title"])
  52
  53   # paragraphs and linebreaks
  54   Meta.allow_tag_with_these_attributes("br", [])
  55   Meta.allow_tag_with_these_attributes("p", [])
  56
  57   # microformats
  58   Meta.allow_tag_with_these_attributes("span", [])
  59
  60   # allow inline images for custom emoji
  61   @allow_inline_images Keyword.get(@markup, :allow_inline_images)
  62
  63   if @allow_inline_images do
  64     # restrict img tags to http/https only, because of MediaProxy.
  65     Meta.allow_tag_with_uri_attributes("img", ["src"], ["http", "https"])
  66
  67     Meta.allow_tag_with_these_attributes("img", [
  68       "width",
  69       "height",
  70       "title",
  71       "alt"
  72     ])
  73   end
  74
  75   Meta.strip_everything_not_covered()
  76 end
  77
  78 defmodule Pleroma.HTML.Scrubber.Default do
  79   @doc "The default HTML scrubbing policy: no "
  80
  81   require HtmlSanitizeEx.Scrubber.Meta
  82   alias HtmlSanitizeEx.Scrubber.Meta
  83
  84   def version do
  85     0
  86   end
  87
  88   @markup Application.get_env(:pleroma, :markup)
  89   @uri_schemes Application.get_env(:pleroma, :uri_schemes, [])
  90   @valid_schemes Keyword.get(@uri_schemes, :valid_schemes, [])
  91
  92   Meta.remove_cdata_sections_before_scrub()
  93   Meta.strip_comments()
  94
  95   Meta.allow_tag_with_uri_attributes("a", ["href", "data-user", "data-tag"], @valid_schemes)
  96   Meta.allow_tag_with_these_attributes("a", ["name", "title"])
  97
  98   Meta.allow_tag_with_these_attributes("abbr", ["title"])
  99
 100   Meta.allow_tag_with_these_attributes("b", [])
 101   Meta.allow_tag_with_these_attributes("blockquote", [])
 102   Meta.allow_tag_with_these_attributes("br", [])
 103   Meta.allow_tag_with_these_attributes("code", [])
 104   Meta.allow_tag_with_these_attributes("del", [])
 105   Meta.allow_tag_with_these_attributes("em", [])
 106   Meta.allow_tag_with_these_attributes("i", [])
 107   Meta.allow_tag_with_these_attributes("li", [])
 108   Meta.allow_tag_with_these_attributes("ol", [])
 109   Meta.allow_tag_with_these_attributes("p", [])
 110   Meta.allow_tag_with_these_attributes("pre", [])
 111   Meta.allow_tag_with_these_attributes("span", [])
 112   Meta.allow_tag_with_these_attributes("strong", [])
 113   Meta.allow_tag_with_these_attributes("u", [])
 114   Meta.allow_tag_with_these_attributes("ul", [])
 115
 116   @allow_inline_images Keyword.get(@markup, :allow_inline_images)
 117
 118   if @allow_inline_images do
 119     # restrict img tags to http/https only, because of MediaProxy.
 120     Meta.allow_tag_with_uri_attributes("img", ["src"], ["http", "https"])
 121
 122     Meta.allow_tag_with_these_attributes("img", [
 123       "width",
 124       "height",
 125       "title",
 126       "alt"
 127     ])
 128   end
 129
 130   @allow_tables Keyword.get(@markup, :allow_tables)
 131
 132   if @allow_tables do
 133     Meta.allow_tag_with_these_attributes("table", [])
 134     Meta.allow_tag_with_these_attributes("tbody", [])
 135     Meta.allow_tag_with_these_attributes("td", [])
 136     Meta.allow_tag_with_these_attributes("th", [])
 137     Meta.allow_tag_with_these_attributes("thead", [])
 138     Meta.allow_tag_with_these_attributes("tr", [])
 139   end
 140
 141   @allow_headings Keyword.get(@markup, :allow_headings)
 142
 143   if @allow_headings do
 144     Meta.allow_tag_with_these_attributes("h1", [])
 145     Meta.allow_tag_with_these_attributes("h2", [])
 146     Meta.allow_tag_with_these_attributes("h3", [])
 147     Meta.allow_tag_with_these_attributes("h4", [])
 148     Meta.allow_tag_with_these_attributes("h5", [])
 149   end
 150
 151   @allow_fonts Keyword.get(@markup, :allow_fonts)
 152
 153   if @allow_fonts do
 154     Meta.allow_tag_with_these_attributes("font", ["face"])
 155   end
 156
 157   Meta.strip_everything_not_covered()
 158 end
 159
 160 defmodule Pleroma.HTML.Transform.MediaProxy do
 161   @moduledoc "Transforms inline image URIs to use MediaProxy."
 162
 163   alias Pleroma.Web.MediaProxy
 164
 165   def version do
 166     0
 167   end
 168
 169   def before_scrub(html), do: html
 170
 171   def scrub_attribute("img", {"src", "http" <> target}) do
 172     media_url =
 173       ("http" <> target)
 174       |> MediaProxy.url()
 175
 176     {"src", media_url}
 177   end
 178
 179   def scrub_attribute(_tag, attribute), do: attribute
 180
 181   def scrub({"img", attributes, children}) do
 182     attributes =
 183       attributes
 184       |> Enum.map(fn attr -> scrub_attribute("img", attr) end)
 185       |> Enum.reject(&is_nil(&1))
 186
 187     {"img", attributes, children}
 188   end
 189
 190   def scrub({:comment, _children}), do: ""
 191
 192   def scrub({tag, attributes, children}), do: {tag, attributes, children}
 193   def scrub({_tag, children}), do: children
 194   def scrub(text), do: text
 195 end