Merge branch 'fix/scrubber-cache' into 'develop'

[#481] Store scrubbed posts in Cachex

Closes #481

See merge request pleroma/pleroma!610

5 files changed, 68 insertions, 7 deletions

diff --git a/lib/pleroma/application.ex b/lib/pleroma/application.ex
index 36a3694f2..4b997c048 100644
--- a/lib/pleroma/application.ex
+++ b/lib/pleroma/application.ex
@@ -56,6 +56,16 @@ defmodule Pleroma.Application do
         worker(
           Cachex,
           [
+            :scrubber_cache,
+            [
+              limit: 2500
+            ]
+          ],
+          id: :cachex_scrubber
+        ),
+        worker(
+          Cachex,
+          [
             :idempotency_cache,
             [
               expiration:
diff --git a/lib/pleroma/html.ex b/lib/pleroma/html.ex
index a0473676b..eb31f131e 100644
--- a/lib/pleroma/html.ex
+++ b/lib/pleroma/html.ex
@@ -15,8 +15,11 @@ defmodule Pleroma.HTML do
   end
 
   def filter_tags(html, nil) do
-    get_scrubbers()
-    |> Enum.reduce(html, fn scrubber, html ->
+    filter_tags(html, get_scrubbers())
+  end
+
+  def filter_tags(html, scrubbers) when is_list(scrubbers) do
+    Enum.reduce(scrubbers, html, fn scrubber, html ->
       filter_tags(html, scrubber)
     end)
   end
@@ -24,6 +27,40 @@ defmodule Pleroma.HTML do
   def filter_tags(html, scrubber), do: Scrubber.scrub(html, scrubber)
   def filter_tags(html), do: filter_tags(html, nil)
   def strip_tags(html), do: Scrubber.scrub(html, Scrubber.StripTags)
+
+  def get_cached_scrubbed_html_for_object(content, scrubbers, object) do
+    key = "#{generate_scrubber_signature(scrubbers)}|#{object.id}"
+    Cachex.fetch!(:scrubber_cache, key, fn _key -> ensure_scrubbed_html(content, scrubbers) end)
+  end
+
+  def get_cached_stripped_html_for_object(content, object) do
+    get_cached_scrubbed_html_for_object(content, HtmlSanitizeEx.Scrubber.StripTags, object)
+  end
+
+  def ensure_scrubbed_html(
+        content,
+        scrubbers
+      ) do
+    {:commit, filter_tags(content, scrubbers)}
+  end
+
+  defp generate_scrubber_signature(scrubber) when is_atom(scrubber) do
+    generate_scrubber_signature([scrubber])
+  end
+
+  defp generate_scrubber_signature(scrubbers) do
+    Enum.reduce(scrubbers, "", fn scrubber, signature ->
+      # If a scrubber does not have a version(e.g HtmlSanitizeEx.Scrubber.StripTags) it is assumed it is always 0)
+      version =
+        if Kernel.function_exported?(scrubber, :version, 0) do
+          scrubber.version
+        else
+          0
+        end
+
+      "#{signature}#{to_string(scrubber)}#{version}"
+    end)
+  end
 end
 
 defmodule Pleroma.HTML.Scrubber.TwitterText do
@@ -39,6 +76,10 @@ defmodule Pleroma.HTML.Scrubber.TwitterText do
   require HtmlSanitizeEx.Scrubber.Meta
   alias HtmlSanitizeEx.Scrubber.Meta
 
+  def version do
+    0
+  end
+
   Meta.remove_cdata_sections_before_scrub()
   Meta.strip_comments()
 
@@ -77,6 +118,10 @@ defmodule Pleroma.HTML.Scrubber.Default do
   require HtmlSanitizeEx.Scrubber.Meta
   alias HtmlSanitizeEx.Scrubber.Meta
 
+  def version do
+    0
+  end
+
   @markup Application.get_env(:pleroma, :markup)
   @uri_schemes Application.get_env(:pleroma, :uri_schemes, [])
   @valid_schemes Keyword.get(@uri_schemes, :valid_schemes, [])
@@ -154,6 +199,10 @@ defmodule Pleroma.HTML.Transform.MediaProxy do
 
   alias Pleroma.Web.MediaProxy
 
+  def version do
+    0
+  end
+
   def before_scrub(html), do: html
 
   def scrub_attribute("img", {"src", "http" <> target}) do
diff --git a/lib/pleroma/user.ex b/lib/pleroma/user.ex
index 5705098ea..b90a3efae 100644
--- a/lib/pleroma/user.ex
+++ b/lib/pleroma/user.ex
@@ -786,7 +786,9 @@ defmodule Pleroma.User do
     Pleroma.HTML.Scrubber.TwitterText
   end
 
-  def html_filter_policy(_), do: nil
+  @default_scrubbers Pleroma.Config.get([:markup, :scrub_policy])
+
+  def html_filter_policy(_), do: @default_scrubbers
 
   def get_or_fetch_by_ap_id(ap_id) do
     user = get_by_ap_id(ap_id)
diff --git a/lib/pleroma/web/mastodon_api/views/status_view.ex b/lib/pleroma/web/mastodon_api/views/status_view.ex
index 4d4681da8..da61bbd86 100644
--- a/lib/pleroma/web/mastodon_api/views/status_view.ex
+++ b/lib/pleroma/web/mastodon_api/views/status_view.ex
@@ -120,7 +120,7 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
     content =
       object
       |> render_content()
-      |> HTML.filter_tags(User.html_filter_policy(opts[:for]))
+      |> HTML.get_cached_scrubbed_html_for_object(User.html_filter_policy(opts[:for]), activity)
 
     %{
       id: to_string(activity.id),
diff --git a/lib/pleroma/web/twitter_api/views/activity_view.ex b/lib/pleroma/web/twitter_api/views/activity_view.ex
index 592cf622f..469f780c7 100644
--- a/lib/pleroma/web/twitter_api/views/activity_view.ex
+++ b/lib/pleroma/web/twitter_api/views/activity_view.ex
@@ -11,11 +11,11 @@ defmodule Pleroma.Web.TwitterAPI.ActivityView do
   alias Pleroma.Web.TwitterAPI.TwitterAPI
   alias Pleroma.Web.TwitterAPI.Representers.ObjectRepresenter
   alias Pleroma.Activity
+  alias Pleroma.HTML
   alias Pleroma.Object
   alias Pleroma.User
   alias Pleroma.Repo
   alias Pleroma.Formatter
-  alias Pleroma.HTML
 
   import Ecto.Query
   require Logger
@@ -245,14 +245,14 @@ defmodule Pleroma.Web.TwitterAPI.ActivityView do
 
     html =
       content
-      |> HTML.filter_tags(User.html_filter_policy(opts[:for]))
+      |> HTML.get_cached_scrubbed_html_for_object(User.html_filter_policy(opts[:for]), activity)
       |> Formatter.emojify(object["emoji"])
 
     text =
       if content do
         content
         |> String.replace(~r/<br\s?\/?>/, "\n")
-        |> HTML.strip_tags()
+        |> HTML.get_cached_stripped_html_for_object(activity)
       end
 
     reply_parent = Activity.get_in_reply_to_activity(activity)