From 380d0cce6b802baf4d13031a4a39f169dd65bffd Mon Sep 17 00:00:00 2001
From: Ivan Tashkinov <ivantashkinov@gmail.com>
Date: Fri, 29 Jan 2021 00:17:33 +0300
Subject: [PATCH] [#3213] Reinstated DISTINCT clause for hashtag "any"
 filtering with 2+ terms. Added test.

---
 lib/pleroma/web/activity_pub/activity_pub.ex    | 17 ++++++++++++-----
 .../controllers/timeline_controller.ex          |  2 +-
 .../web/activity_pub/activity_pub_test.exs      | 17 +++++++++++++++--
 3 files changed, 28 insertions(+), 8 deletions(-)

diff --git a/lib/pleroma/web/activity_pub/activity_pub.ex b/lib/pleroma/web/activity_pub/activity_pub.ex
index be81e0833..0a21ac2f2 100644
--- a/lib/pleroma/web/activity_pub/activity_pub.ex
+++ b/lib/pleroma/web/activity_pub/activity_pub.ex
@@ -846,11 +846,18 @@ defp restrict_hashtag_any(_query, %{tag: _tag, skip_preload: true}) do
   end
 
   defp restrict_hashtag_any(query, %{tag: tags}) when is_list(tags) do
-    from(
-      [_activity, object] in query,
-      join: hashtag in assoc(object, :hashtags),
-      where: hashtag.name in ^tags
-    )
+    query =
+      from(
+        [_activity, object] in query,
+        join: hashtag in assoc(object, :hashtags),
+        where: hashtag.name in ^tags
+      )
+
+    if length(tags) > 1 do
+      distinct(query, [activity], true)
+    else
+      query
+    end
   end
 
   defp restrict_hashtag_any(query, %{tag: tag}) when is_binary(tag) do
diff --git a/lib/pleroma/web/mastodon_api/controllers/timeline_controller.ex b/lib/pleroma/web/mastodon_api/controllers/timeline_controller.ex
index 08e6f23b9..1fb954a9b 100644
--- a/lib/pleroma/web/mastodon_api/controllers/timeline_controller.ex
+++ b/lib/pleroma/web/mastodon_api/controllers/timeline_controller.ex
@@ -134,9 +134,9 @@ defp hashtag_fetching(params, user, local_only) do
     tags =
       [params[:tag], params[:any]]
       |> List.flatten()
-      |> Enum.uniq()
       |> Enum.reject(&is_nil/1)
       |> Enum.map(&String.downcase/1)
+      |> Enum.uniq()
 
     tag_all =
       params
diff --git a/test/pleroma/web/activity_pub/activity_pub_test.exs b/test/pleroma/web/activity_pub/activity_pub_test.exs
index 1fcaf74d3..0b18269cd 100644
--- a/test/pleroma/web/activity_pub/activity_pub_test.exs
+++ b/test/pleroma/web/activity_pub/activity_pub_test.exs
@@ -217,6 +217,9 @@ test "it fetches the appropriate tag-restricted posts" do
     {:ok, status_two} = CommonAPI.post(user, %{status: ". #essais"})
     {:ok, status_three} = CommonAPI.post(user, %{status: ". #test #reject"})
 
+    {:ok, status_four} = CommonAPI.post(user, %{status: ". #any1 #any2"})
+    {:ok, status_five} = CommonAPI.post(user, %{status: ". #any2 #any1"})
+
     for hashtag_timeline_strategy <- [true, :prefer_aggregation, false] do
       clear_config([:instance, :improved_hashtag_timeline], hashtag_timeline_strategy)
 
@@ -238,8 +241,17 @@ test "it fetches the appropriate tag-restricted posts" do
           tag_all: ["test", "reject"]
         })
 
-      [fetch_one, fetch_two, fetch_three, fetch_four] =
-        Enum.map([fetch_one, fetch_two, fetch_three, fetch_four], fn statuses ->
+      # This test would fail if JOIN with 2+ terms in "any" clause is done without DISTINCT.
+      # The :limit is important (w/o DISTINCT 2 records are deduped by Ecto to 1 b/c of preload).
+      fetch_five =
+        ActivityPub.fetch_activities([], %{
+          type: "Create",
+          tag: ["any1", "any2"],
+          limit: 2
+        })
+
+      [fetch_one, fetch_two, fetch_three, fetch_four, fetch_five] =
+        Enum.map([fetch_one, fetch_two, fetch_three, fetch_four, fetch_five], fn statuses ->
           Enum.map(statuses, fn s -> Repo.preload(s, object: :hashtags) end)
         end)
 
@@ -247,6 +259,7 @@ test "it fetches the appropriate tag-restricted posts" do
       assert fetch_two == [status_one, status_two, status_three]
       assert fetch_three == [status_one, status_two]
       assert fetch_four == [status_three]
+      assert fetch_five == [status_four, status_five]
     end
   end