Merge branch 'feature/otp-configurable-scrubbers' into 'develop'

HTML: Compile Scrubbers on boot

See merge request pleroma/pleroma!2043

.formatter.exs

@@ -1,3 +1,3 @@
 [
-  inputs: ["mix.exs", "{config,lib,test}/**/*.{ex,exs}", "priv/repo/migrations/*.exs"]
+  inputs: ["mix.exs", "{config,lib,test}/**/*.{ex,exs}", "priv/repo/migrations/*.exs", "priv/scrubbers/*.ex"]
 ]

CHANGELOG.md

@@ -82,6 +82,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 - Report emails now include functional links to profiles of remote user accounts
 - Not being able to log in to some third-party apps when logged in to MastoFE
 - MRF: `Delete` activities being exempt from MRF policies
+- OTP releases: Not being able to configure HTML sanitization policy
 <details>
   <summary>API Changes</summary>
 

lib/pleroma/application.ex

@@ -30,6 +30,7 @@ def user_agent do
   # See http://elixir-lang.org/docs/stable/elixir/Application.html
   # for more information on OTP Applications
   def start(_type, _args) do
+    Pleroma.HTML.compile_scrubbers()
     Pleroma.Config.DeprecationWarnings.warn()
     setup_instrumenters()
 

lib/pleroma/html.ex

@@ -3,6 +3,25 @@
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.HTML do
+  # Scrubbers are compiled on boot so they can be configured in OTP releases
+  #  @on_load :compile_scrubbers
+
+  def compile_scrubbers do
+    dir = Path.join(:code.priv_dir(:pleroma), "scrubbers")
+
+    dir
+    |> File.ls!()
+    |> Enum.map(&Path.join(dir, &1))
+    |> Kernel.ParallelCompiler.compile()
+    |> case do
+      {:error, _errors, _warnings} ->
+        raise "Compiling scrubbers failed"
+
+      {:ok, _modules, _warnings} ->
+        :ok
+    end
+  end
+
   defp get_scrubbers(scrubber) when is_atom(scrubber), do: [scrubber]
   defp get_scrubbers(scrubbers) when is_list(scrubbers), do: scrubbers
   defp get_scrubbers(_), do: [Pleroma.HTML.Scrubber.Default]
@@ -99,216 +118,3 @@ def extract_first_external_url(object, content) do
     end)
   end
 end
-
-defmodule Pleroma.HTML.Scrubber.TwitterText do
-  @moduledoc """
-  An HTML scrubbing policy which limits to twitter-style text.  Only
-  paragraphs, breaks and links are allowed through the filter.
-  """
-
-  @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
-
-  require FastSanitize.Sanitizer.Meta
-  alias FastSanitize.Sanitizer.Meta
-
-  Meta.strip_comments()
-
-  # links
-  Meta.allow_tag_with_uri_attributes(:a, ["href", "data-user", "data-tag"], @valid_schemes)
-
-  Meta.allow_tag_with_this_attribute_values(:a, "class", [
-    "hashtag",
-    "u-url",
-    "mention",
-    "u-url mention",
-    "mention u-url"
-  ])
-
-  Meta.allow_tag_with_this_attribute_values(:a, "rel", [
-    "tag",
-    "nofollow",
-    "noopener",
-    "noreferrer"
-  ])
-
-  Meta.allow_tag_with_these_attributes(:a, ["name", "title"])
-
-  # paragraphs and linebreaks
-  Meta.allow_tag_with_these_attributes(:br, [])
-  Meta.allow_tag_with_these_attributes(:p, [])
-
-  # microformats
-  Meta.allow_tag_with_this_attribute_values(:span, "class", ["h-card"])
-  Meta.allow_tag_with_these_attributes(:span, [])
-
-  # allow inline images for custom emoji
-  if Pleroma.Config.get([:markup, :allow_inline_images]) do
-    # restrict img tags to http/https only, because of MediaProxy.
-    Meta.allow_tag_with_uri_attributes(:img, ["src"], ["http", "https"])
-
-    Meta.allow_tag_with_these_attributes(:img, [
-      "width",
-      "height",
-      "class",
-      "title",
-      "alt"
-    ])
-  end
-
-  Meta.strip_everything_not_covered()
-end
-
-defmodule Pleroma.HTML.Scrubber.Default do
-  @doc "The default HTML scrubbing policy: no "
-
-  require FastSanitize.Sanitizer.Meta
-  alias FastSanitize.Sanitizer.Meta
-
-  # credo:disable-for-previous-line
-  # No idea how to fix this one…
-
-  @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
-
-  Meta.strip_comments()
-
-  Meta.allow_tag_with_uri_attributes(:a, ["href", "data-user", "data-tag"], @valid_schemes)
-
-  Meta.allow_tag_with_this_attribute_values(:a, "class", [
-    "hashtag",
-    "u-url",
-    "mention",
-    "u-url mention",
-    "mention u-url"
-  ])
-
-  Meta.allow_tag_with_this_attribute_values(:a, "rel", [
-    "tag",
-    "nofollow",
-    "noopener",
-    "noreferrer",
-    "ugc"
-  ])
-
-  Meta.allow_tag_with_these_attributes(:a, ["name", "title"])
-
-  Meta.allow_tag_with_these_attributes(:abbr, ["title"])
-
-  Meta.allow_tag_with_these_attributes(:b, [])
-  Meta.allow_tag_with_these_attributes(:blockquote, [])
-  Meta.allow_tag_with_these_attributes(:br, [])
-  Meta.allow_tag_with_these_attributes(:code, [])
-  Meta.allow_tag_with_these_attributes(:del, [])
-  Meta.allow_tag_with_these_attributes(:em, [])
-  Meta.allow_tag_with_these_attributes(:i, [])
-  Meta.allow_tag_with_these_attributes(:li, [])
-  Meta.allow_tag_with_these_attributes(:ol, [])
-  Meta.allow_tag_with_these_attributes(:p, [])
-  Meta.allow_tag_with_these_attributes(:pre, [])
-  Meta.allow_tag_with_these_attributes(:strong, [])
-  Meta.allow_tag_with_these_attributes(:sub, [])
-  Meta.allow_tag_with_these_attributes(:sup, [])
-  Meta.allow_tag_with_these_attributes(:u, [])
-  Meta.allow_tag_with_these_attributes(:ul, [])
-
-  Meta.allow_tag_with_this_attribute_values(:span, "class", ["h-card"])
-  Meta.allow_tag_with_these_attributes(:span, [])
-
-  @allow_inline_images Pleroma.Config.get([:markup, :allow_inline_images])
-
-  if @allow_inline_images do
-    # restrict img tags to http/https only, because of MediaProxy.
-    Meta.allow_tag_with_uri_attributes(:img, ["src"], ["http", "https"])
-
-    Meta.allow_tag_with_these_attributes(:img, [
-      "width",
-      "height",
-      "class",
-      "title",
-      "alt"
-    ])
-  end
-
-  if Pleroma.Config.get([:markup, :allow_tables]) do
-    Meta.allow_tag_with_these_attributes(:table, [])
-    Meta.allow_tag_with_these_attributes(:tbody, [])
-    Meta.allow_tag_with_these_attributes(:td, [])
-    Meta.allow_tag_with_these_attributes(:th, [])
-    Meta.allow_tag_with_these_attributes(:thead, [])
-    Meta.allow_tag_with_these_attributes(:tr, [])
-  end
-
-  if Pleroma.Config.get([:markup, :allow_headings]) do
-    Meta.allow_tag_with_these_attributes(:h1, [])
-    Meta.allow_tag_with_these_attributes(:h2, [])
-    Meta.allow_tag_with_these_attributes(:h3, [])
-    Meta.allow_tag_with_these_attributes(:h4, [])
-    Meta.allow_tag_with_these_attributes(:h5, [])
-  end
-
-  if Pleroma.Config.get([:markup, :allow_fonts]) do
-    Meta.allow_tag_with_these_attributes(:font, ["face"])
-  end
-
-  Meta.strip_everything_not_covered()
-end
-
-defmodule Pleroma.HTML.Transform.MediaProxy do
-  @moduledoc "Transforms inline image URIs to use MediaProxy."
-
-  alias Pleroma.Web.MediaProxy
-
-  def before_scrub(html), do: html
-
-  def scrub_attribute(:img, {"src", "http" <> target}) do
-    media_url =
-      ("http" <> target)
-      |> MediaProxy.url()
-
-    {"src", media_url}
-  end
-
-  def scrub_attribute(_tag, attribute), do: attribute
-
-  def scrub({:img, attributes, children}) do
-    attributes =
-      attributes
-      |> Enum.map(fn attr -> scrub_attribute(:img, attr) end)
-      |> Enum.reject(&is_nil(&1))
-
-    {:img, attributes, children}
-  end
-
-  def scrub({:comment, _text, _children}), do: ""
-
-  def scrub({tag, attributes, children}), do: {tag, attributes, children}
-  def scrub({_tag, children}), do: children
-  def scrub(text), do: text
-end
-
-defmodule Pleroma.HTML.Scrubber.LinksOnly do
-  @moduledoc """
-  An HTML scrubbing policy which limits to links only.
-  """
-
-  @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
-
-  require FastSanitize.Sanitizer.Meta
-  alias FastSanitize.Sanitizer.Meta
-
-  Meta.strip_comments()
-
-  # links
-  Meta.allow_tag_with_uri_attributes(:a, ["href"], @valid_schemes)
-
-  Meta.allow_tag_with_this_attribute_values(:a, "rel", [
-    "tag",
-    "nofollow",
-    "noopener",
-    "noreferrer",
-    "me",
-    "ugc"
-  ])
-
-  Meta.allow_tag_with_these_attributes(:a, ["name", "title"])
-  Meta.strip_everything_not_covered()
-end

priv/scrubbers/default.ex

@@ -0,0 +1,93 @@
+defmodule Pleroma.HTML.Scrubber.Default do
+  @doc "The default HTML scrubbing policy: no "
+
+  require FastSanitize.Sanitizer.Meta
+  alias FastSanitize.Sanitizer.Meta
+
+  # credo:disable-for-previous-line
+  # No idea how to fix this one…
+
+  @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
+
+  Meta.strip_comments()
+
+  Meta.allow_tag_with_uri_attributes(:a, ["href", "data-user", "data-tag"], @valid_schemes)
+
+  Meta.allow_tag_with_this_attribute_values(:a, "class", [
+    "hashtag",
+    "u-url",
+    "mention",
+    "u-url mention",
+    "mention u-url"
+  ])
+
+  Meta.allow_tag_with_this_attribute_values(:a, "rel", [
+    "tag",
+    "nofollow",
+    "noopener",
+    "noreferrer",
+    "ugc"
+  ])
+
+  Meta.allow_tag_with_these_attributes(:a, ["name", "title"])
+
+  Meta.allow_tag_with_these_attributes(:abbr, ["title"])
+
+  Meta.allow_tag_with_these_attributes(:b, [])
+  Meta.allow_tag_with_these_attributes(:blockquote, [])
+  Meta.allow_tag_with_these_attributes(:br, [])
+  Meta.allow_tag_with_these_attributes(:code, [])
+  Meta.allow_tag_with_these_attributes(:del, [])
+  Meta.allow_tag_with_these_attributes(:em, [])
+  Meta.allow_tag_with_these_attributes(:i, [])
+  Meta.allow_tag_with_these_attributes(:li, [])
+  Meta.allow_tag_with_these_attributes(:ol, [])
+  Meta.allow_tag_with_these_attributes(:p, [])
+  Meta.allow_tag_with_these_attributes(:pre, [])
+  Meta.allow_tag_with_these_attributes(:strong, [])
+  Meta.allow_tag_with_these_attributes(:sub, [])
+  Meta.allow_tag_with_these_attributes(:sup, [])
+  Meta.allow_tag_with_these_attributes(:u, [])
+  Meta.allow_tag_with_these_attributes(:ul, [])
+
+  Meta.allow_tag_with_this_attribute_values(:span, "class", ["h-card"])
+  Meta.allow_tag_with_these_attributes(:span, [])
+
+  @allow_inline_images Pleroma.Config.get([:markup, :allow_inline_images])
+
+  if @allow_inline_images do
+    # restrict img tags to http/https only, because of MediaProxy.
+    Meta.allow_tag_with_uri_attributes(:img, ["src"], ["http", "https"])
+
+    Meta.allow_tag_with_these_attributes(:img, [
+      "width",
+      "height",
+      "class",
+      "title",
+      "alt"
+    ])
+  end
+
+  if Pleroma.Config.get([:markup, :allow_tables]) do
+    Meta.allow_tag_with_these_attributes(:table, [])
+    Meta.allow_tag_with_these_attributes(:tbody, [])
+    Meta.allow_tag_with_these_attributes(:td, [])
+    Meta.allow_tag_with_these_attributes(:th, [])
+    Meta.allow_tag_with_these_attributes(:thead, [])
+    Meta.allow_tag_with_these_attributes(:tr, [])
+  end
+
+  if Pleroma.Config.get([:markup, :allow_headings]) do
+    Meta.allow_tag_with_these_attributes(:h1, [])
+    Meta.allow_tag_with_these_attributes(:h2, [])
+    Meta.allow_tag_with_these_attributes(:h3, [])
+    Meta.allow_tag_with_these_attributes(:h4, [])
+    Meta.allow_tag_with_these_attributes(:h5, [])
+  end
+
+  if Pleroma.Config.get([:markup, :allow_fonts]) do
+    Meta.allow_tag_with_these_attributes(:font, ["face"])
+  end
+
+  Meta.strip_everything_not_covered()
+end

priv/scrubbers/links_only.ex

@@ -0,0 +1,27 @@
+defmodule Pleroma.HTML.Scrubber.LinksOnly do
+  @moduledoc """
+  An HTML scrubbing policy which limits to links only.
+  """
+
+  @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
+
+  require FastSanitize.Sanitizer.Meta
+  alias FastSanitize.Sanitizer.Meta
+
+  Meta.strip_comments()
+
+  # links
+  Meta.allow_tag_with_uri_attributes(:a, ["href"], @valid_schemes)
+
+  Meta.allow_tag_with_this_attribute_values(:a, "rel", [
+    "tag",
+    "nofollow",
+    "noopener",
+    "noreferrer",
+    "me",
+    "ugc"
+  ])
+
+  Meta.allow_tag_with_these_attributes(:a, ["name", "title"])
+  Meta.strip_everything_not_covered()
+end

priv/scrubbers/media_proxy.ex

@@ -0,0 +1,32 @@
+defmodule Pleroma.HTML.Transform.MediaProxy do
+  @moduledoc "Transforms inline image URIs to use MediaProxy."
+
+  alias Pleroma.Web.MediaProxy
+
+  def before_scrub(html), do: html
+
+  def scrub_attribute(:img, {"src", "http" <> target}) do
+    media_url =
+      ("http" <> target)
+      |> MediaProxy.url()
+
+    {"src", media_url}
+  end
+
+  def scrub_attribute(_tag, attribute), do: attribute
+
+  def scrub({:img, attributes, children}) do
+    attributes =
+      attributes
+      |> Enum.map(fn attr -> scrub_attribute(:img, attr) end)
+      |> Enum.reject(&is_nil(&1))
+
+    {:img, attributes, children}
+  end
+
+  def scrub({:comment, _text, _children}), do: ""
+
+  def scrub({tag, attributes, children}), do: {tag, attributes, children}
+  def scrub({_tag, children}), do: children
+  def scrub(text), do: text
+end

priv/scrubbers/twitter_text.ex

@@ -0,0 +1,57 @@
+defmodule Pleroma.HTML.Scrubber.TwitterText do
+  @moduledoc """
+  An HTML scrubbing policy which limits to twitter-style text.  Only
+  paragraphs, breaks and links are allowed through the filter.
+  """
+
+  @valid_schemes Pleroma.Config.get([:uri_schemes, :valid_schemes], [])
+
+  require FastSanitize.Sanitizer.Meta
+  alias FastSanitize.Sanitizer.Meta
+
+  Meta.strip_comments()
+
+  # links
+  Meta.allow_tag_with_uri_attributes(:a, ["href", "data-user", "data-tag"], @valid_schemes)
+
+  Meta.allow_tag_with_this_attribute_values(:a, "class", [
+    "hashtag",
+    "u-url",
+    "mention",
+    "u-url mention",
+    "mention u-url"
+  ])
+
+  Meta.allow_tag_with_this_attribute_values(:a, "rel", [
+    "tag",
+    "nofollow",
+    "noopener",
+    "noreferrer"
+  ])
+
+  Meta.allow_tag_with_these_attributes(:a, ["name", "title"])
+
+  # paragraphs and linebreaks
+  Meta.allow_tag_with_these_attributes(:br, [])
+  Meta.allow_tag_with_these_attributes(:p, [])
+
+  # microformats
+  Meta.allow_tag_with_this_attribute_values(:span, "class", ["h-card"])
+  Meta.allow_tag_with_these_attributes(:span, [])
+
+  # allow inline images for custom emoji
+  if Pleroma.Config.get([:markup, :allow_inline_images]) do
+    # restrict img tags to http/https only, because of MediaProxy.
+    Meta.allow_tag_with_uri_attributes(:img, ["src"], ["http", "https"])
+
+    Meta.allow_tag_with_these_attributes(:img, [
+      "width",
+      "height",
+      "class",
+      "title",
+      "alt"
+    ])
+  end
+
+  Meta.strip_everything_not_covered()
+end