Merge branch 'fix/tag-feed-crashes' into 'develop'

fix: atom/rss feed issues

Closes #3045

See merge request pleroma/pleroma!3851

lib/pleroma/web/feed/feed_view.ex

@@ -6,7 +6,6 @@ defmodule Pleroma.Web.Feed.FeedView do
   use Phoenix.HTML
   use Pleroma.Web, :view
 
-  alias Pleroma.Formatter
   alias Pleroma.Object
   alias Pleroma.User
   alias Pleroma.Web.Gettext
@@ -72,7 +71,9 @@ def logo(user) do
 
   def last_activity(activities), do: List.last(activities)
 
-  def activity_title(%{"content" => content, "summary" => summary} = data, opts \\ %{}) do
+  def activity_title(%{"content" => content} = data, opts \\ %{}) do
+    summary = Map.get(data, "summary", "")
+
     title =
       cond do
         summary != "" -> summary
@@ -81,9 +82,8 @@ def activity_title(%{"content" => content, "summary" => summary} = data, opts \\
       end
 
     title
-    |> Pleroma.Web.Metadata.Utils.scrub_html()
+    |> Pleroma.Web.Metadata.Utils.scrub_html_and_truncate(opts[:max_length], opts[:omission])
-    |> Pleroma.Emoji.Formatter.demojify()
+    |> HtmlEntities.encode()
-    |> Formatter.truncate(opts[:max_length], opts[:omission])
   end
 
   def activity_description(data) do

lib/pleroma/web/metadata/utils.ex

@@ -30,12 +30,13 @@ def scrub_html_and_truncate(%{data: %{"content" => content}} = object) do
     |> scrub_html_and_truncate_object_field(object)
   end
 
-  def scrub_html_and_truncate(content, max_length \\ 200) when is_binary(content) do
+  def scrub_html_and_truncate(content, max_length \\ 200, omission \\ "...")
+      when is_binary(content) do
     content
     |> scrub_html
     |> Emoji.Formatter.demojify()
     |> HtmlEntities.decode()
-    |> Formatter.truncate(max_length)
+    |> Formatter.truncate(max_length, omission)
   end
 
   def scrub_html(content) when is_binary(content) do

lib/pleroma/web/templates/feed/feed/_activity.atom.eex

@@ -4,8 +4,8 @@
   <id><%= @data["id"] %></id>
   <title><%= activity_title(@data, Keyword.get(@feed_config, :post_title, %{})) %></title>
   <content type="html"><%= activity_description(@data) %></content>
-  <published><%= to_rfc3339(@activity.data["published"]) %></published>
+  <published><%= to_rfc3339(@data["published"]) %></published>
-  <updated><%= to_rfc3339(@activity.data["published"]) %></updated>
+  <updated><%= to_rfc3339(@data["published"]) %></updated>
   <ostatus:conversation ref="<%= activity_context(@activity) %>">
     <%= activity_context(@activity) %>
   </ostatus:conversation>

lib/pleroma/web/templates/feed/feed/_activity.rss.eex

@@ -4,7 +4,7 @@
   <guid><%= @data["id"] %></guid>
   <title><%= activity_title(@data, Keyword.get(@feed_config, :post_title, %{})) %></title>
   <description><%= activity_description(@data) %></description>
-  <pubDate><%= to_rfc2822(@activity.data["published"]) %></pubDate>
+  <pubDate><%= to_rfc2822(@data["published"]) %></pubDate>
   <ostatus:conversation ref="<%= activity_context(@activity) %>">
     <%= activity_context(@activity) %>
   </ostatus:conversation>

lib/pleroma/web/templates/feed/feed/_tag_activity.atom.eex

@@ -7,8 +7,8 @@
   <id><%= @data["id"] %></id>
   <title><%= activity_title(@data, Keyword.get(@feed_config, :post_title, %{})) %></title>
   <content type="html"><%= activity_description(@data) %></content>
-  <published><%= to_rfc3339(@activity.data["published"]) %></published>
+  <published><%= to_rfc3339(@data["published"]) %></published>
-  <updated><%= to_rfc3339(@activity.data["published"]) %></updated>
+  <updated><%= to_rfc3339(@data["published"]) %></updated>
   <ostatus:conversation ref="<%= activity_context(@activity) %>">
     <%= activity_context(@activity) %>
   </ostatus:conversation>

lib/pleroma/web/templates/feed/feed/_tag_activity.xml.eex

@@ -4,7 +4,7 @@
 
   <guid isPermalink="true"><%= activity_context(@activity) %></guid>
   <link><%= activity_context(@activity) %></link>
-  <pubDate><%= to_rfc2822(@activity.data["published"]) %></pubDate>
+  <pubDate><%= to_rfc2822(@data["published"]) %></pubDate>
 
   <description><%= activity_description(@data) %></description>
   <%= for attachment <- @data["attachment"] || [] do %>

test/pleroma/web/feed/user_controller_test.exs

@@ -57,9 +57,23 @@ defmodule Pleroma.Web.Feed.UserControllerTest do
         )
 
       note_activity2 = insert(:note_activity, note: note2)
+
+      note3 =
+        insert(:note,
+          user: user,
+          data: %{
+            "content" => "This note tests whether HTML entities are truncated properly",
+            "summary" => "Won't, didn't fail",
+            "inReplyTo" => note_activity2.id
+          }
+        )
+
+      _note_activity3 = insert(:note_activity, note: note3)
       object = Object.normalize(note_activity, fetch: false)
 
-      [user: user, object: object, max_id: note_activity2.id]
+      encoded_title = FeedView.activity_title(note3.data)
+
+      [user: user, object: object, max_id: note_activity2.id, encoded_title: encoded_title]
     end
 
     test "gets an atom feed", %{conn: conn, user: user, object: object, max_id: max_id} do
@@ -74,7 +88,7 @@ test "gets an atom feed", %{conn: conn, user: user, object: object, max_id: max_
         |> SweetXml.parse()
         |> SweetXml.xpath(~x"//entry/title/text()"l)
 
-      assert activity_titles == ['2hu', '2hu & as']
+      assert activity_titles == ['Won\'t, didn\'...', '2hu', '2hu & as']
       assert resp =~ FeedView.escape(object.data["content"])
       assert resp =~ FeedView.escape(object.data["summary"])
       assert resp =~ FeedView.escape(object.data["context"])
@@ -105,7 +119,7 @@ test "gets a rss feed", %{conn: conn, user: user, object: object, max_id: max_id
         |> SweetXml.parse()
         |> SweetXml.xpath(~x"//item/title/text()"l)
 
-      assert activity_titles == ['2hu', '2hu & as']
+      assert activity_titles == ['Won\'t, didn\'...', '2hu', '2hu & as']
       assert resp =~ FeedView.escape(object.data["content"])
       assert resp =~ FeedView.escape(object.data["summary"])
       assert resp =~ FeedView.escape(object.data["context"])
@@ -176,6 +190,30 @@ test "does not require authentication on non-federating instances", %{conn: conn
       |> get("/users/#{user.nickname}/feed.rss")
       |> response(200)
     end
+
+    test "does not mangle HTML entities midway", %{
+      conn: conn,
+      user: user,
+      object: object,
+      encoded_title: encoded_title
+    } do
+      resp =
+        conn
+        |> put_req_header("accept", "application/atom+xml")
+        |> get(user_feed_path(conn, :feed, user.nickname))
+        |> response(200)
+
+      activity_titles =
+        resp
+        |> SweetXml.parse()
+        |> SweetXml.xpath(~x"//entry/title/text()"l)
+
+      assert activity_titles == ['Won\'t, didn\'...', '2hu', '2hu & as']
+      assert resp =~ FeedView.escape(object.data["content"])
+      assert resp =~ FeedView.escape(object.data["summary"])
+      assert resp =~ FeedView.escape(object.data["context"])
+      assert resp =~ encoded_title
+    end
   end
 
   # Note: see ActivityPubControllerTest for JSON format tests

test/pleroma/web/metadata/utils_test.exs

@@ -72,7 +72,7 @@ test "it does not return old content after editing" do
     end
   end
 
-  describe "scrub_html_and_truncate/2" do
+  describe "scrub_html_and_truncate/3" do
     test "it returns text without encode HTML" do
       assert Utils.scrub_html_and_truncate("Pleroma's really cool!") == "Pleroma's really cool!"
     end