Merge branch 'user-search-nickname-rank-boost' into 'develop'

Improved search results for localized nickname match

See merge request pleroma/pleroma!2733

lib/pleroma/user/search.ex

@@ -69,11 +69,15 @@ defp fts_search(query, query_string) do
       u in query,
       where:
         fragment(
+          # The fragment must _exactly_ match `users_fts_index`, otherwise the index won't work
           """
-          (to_tsvector('simple', ?) || to_tsvector('simple', ?)) @@ to_tsquery('simple', ?)
+          (
+            setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
+            setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')
+          ) @@ to_tsquery('simple', ?)
           """,
-          u.name,
           u.nickname,
+          u.name,
           ^query_string
         )
     )
@@ -88,15 +92,23 @@ defp to_tsquery(query_string) do
     |> Enum.join(" | ")
   end
 
+  # Considers nickname match, localized nickname match, name match; preferences nickname match
   defp trigram_rank(query, query_string) do
     from(
       u in query,
       select_merge: %{
         search_rank:
           fragment(
-            "similarity(?, trim(? || ' ' || coalesce(?, '')))",
+            """
+            similarity(?, ?) +
+            similarity(?, regexp_replace(?, '@.+', '')) +
+            similarity(?, trim(coalesce(?, '')))
+            """,
             ^query_string,
             u.nickname,
+            ^query_string,
+            u.nickname,
+            ^query_string,
             u.name
           )
       }

priv/repo/migrations/20200708193702_drop_user_trigram_index.exs

@@ -0,0 +1,18 @@
+defmodule Pleroma.Repo.Migrations.DropUserTrigramIndex do
+  @moduledoc "Drops unused trigram index on `users` (FTS index is being used instead)"
+
+  use Ecto.Migration
+
+  def up do
+    drop_if_exists(index(:users, [], name: :users_trigram_index))
+  end
+
+  def down do
+    create_if_not_exists(
+      index(:users, ["(trim(nickname || ' ' || coalesce(name, ''))) gist_trgm_ops"],
+        name: :users_trigram_index,
+        using: :gist
+      )
+    )
+  end
+end

test/tasks/user_test.exs

@@ -481,17 +481,17 @@ test "it returns users matching" do
       moot = insert(:user, nickname: "moot")
       kawen = insert(:user, nickname: "kawen", name: "fediverse expert moon")
 
-      {:ok, user} = User.follow(user, kawen)
+      {:ok, user} = User.follow(user, moon)
 
       assert [moon.id, kawen.id] == User.Search.search("moon") |> Enum.map(& &1.id)
-      res = User.search("moo") |> Enum.map(& &1.id)
-      assert moon.id in res
-      assert moot.id in res
-      assert kawen.id in res
-      assert [moon.id, kawen.id] == User.Search.search("moon fediverse") |> Enum.map(& &1.id)
 
-      assert [kawen.id, moon.id] ==
+      res = User.search("moo") |> Enum.map(& &1.id)
-               User.Search.search("moon fediverse", for_user: user) |> Enum.map(& &1.id)
+      assert Enum.sort([moon.id, moot.id, kawen.id]) == Enum.sort(res)
+
+      assert [kawen.id, moon.id] == User.Search.search("expert fediverse") |> Enum.map(& &1.id)
+
+      assert [moon.id, kawen.id] ==
+               User.Search.search("expert fediverse", for_user: user) |> Enum.map(& &1.id)
     end
   end
 

test/user_search_test.exs

@@ -46,30 +46,49 @@ test "accepts offset parameter" do
       assert length(User.search("john", limit: 3, offset: 3)) == 2
     end
 
-    test "finds a user by full or partial nickname" do
+    defp clear_virtual_fields(user) do
+      Map.merge(user, %{search_rank: nil, search_type: nil})
+    end
+
+    test "finds a user by full nickname or its leading fragment" do
       user = insert(:user, %{nickname: "john"})
 
       Enum.each(["john", "jo", "j"], fn query ->
         assert user ==
                  User.search(query)
                  |> List.first()
-                 |> Map.put(:search_rank, nil)
+                 |> clear_virtual_fields()
-                 |> Map.put(:search_type, nil)
       end)
     end
 
-    test "finds a user by full or partial name" do
+    test "finds a user by full name or leading fragment(s) of its words" do
       user = insert(:user, %{name: "John Doe"})
 
       Enum.each(["John Doe", "JOHN", "doe", "j d", "j", "d"], fn query ->
         assert user ==
                  User.search(query)
                  |> List.first()
-                 |> Map.put(:search_rank, nil)
+                 |> clear_virtual_fields()
-                 |> Map.put(:search_type, nil)
       end)
     end
 
+    test "matches by leading fragment of user domain" do
+      user = insert(:user, %{nickname: "arandom@dude.com"})
+      insert(:user, %{nickname: "iamthedude"})
+
+      assert [user.id] == User.search("dud") |> Enum.map(& &1.id)
+    end
+
+    test "ranks full nickname match higher than full name match" do
+      nicknamed_user = insert(:user, %{nickname: "hj@shigusegubu.club"})
+      named_user = insert(:user, %{nickname: "xyz@sample.com", name: "HJ"})
+
+      results = User.search("hj")
+
+      assert [nicknamed_user.id, named_user.id] == Enum.map(results, & &1.id)
+      assert Enum.at(results, 0).search_rank > Enum.at(results, 1).search_rank
+    end
+
     test "finds users, considering density of matched tokens" do
       u1 = insert(:user, %{name: "Bar Bar plus Word Word"})
       u2 = insert(:user, %{name: "Word Word Bar Bar Bar"})