Merge branch 'i1t/pleroma-477_user_search_improvements' into 'develop'

I1t/pleroma 477 user search improvements

See merge request pleroma/pleroma!685

lib/pleroma/user.ex

@@ -35,7 +35,7 @@ defmodule Pleroma.User do
     field(:avatar, :map)
     field(:local, :boolean, default: true)
     field(:follower_address, :string)
-    field(:search_distance, :float, virtual: true)
+    field(:search_rank, :float, virtual: true)
     field(:tags, {:array, :string}, default: [])
     field(:last_refreshed_at, :naive_datetime)
     has_many(:notifications, Notification)
@@ -510,6 +510,12 @@ def get_followers(user, page \\ nil) do
     {:ok, Repo.all(q)}
   end
 
+  def get_followers_ids(user, page \\ nil) do
+    q = get_followers_query(user, page)
+
+    Repo.all(from(u in q, select: u.id))
+  end
+
   def get_friends_query(%User{id: id, following: following}, nil) do
     from(
       u in User,
@@ -534,6 +540,12 @@ def get_friends(user, page \\ nil) do
     {:ok, Repo.all(q)}
   end
 
+  def get_friends_ids(user, page \\ nil) do
+    q = get_friends_query(user, page)
+
+    Repo.all(from(u in q, select: u.id))
+  end
+
   def get_follow_requests_query(%User{} = user) do
     from(
       a in Activity,
@@ -665,37 +677,120 @@ def get_recipients_from_activity(%Activity{recipients: to}) do
     Repo.all(query)
   end
 
-  def search(query, resolve \\ false) do
+  def search(query, resolve \\ false, for_user \\ nil) do
-    # strip the beginning @ off if there is a query
+    # Strip the beginning @ off if there is a query
     query = String.trim_leading(query, "@")
 
-    if resolve do
+    if resolve, do: User.get_or_fetch_by_nickname(query)
-      User.get_or_fetch_by_nickname(query)
-    end
 
-    inner =
+    fts_results = do_search(fts_search_subquery(query), for_user)
-      from(
-        u in User,
-        select_merge: %{
-          search_distance:
-            fragment(
-              "? <-> (? || coalesce(?, ''))",
-              ^query,
-              u.nickname,
-              u.name
-            )
-        },
-        where: not is_nil(u.nickname)
-      )
 
+    {:ok, trigram_results} =
+      Repo.transaction(fn ->
+        Ecto.Adapters.SQL.query(Repo, "select set_limit(0.25)", [])
+        do_search(trigram_search_subquery(query), for_user)
+      end)
+
+    Enum.uniq_by(fts_results ++ trigram_results, & &1.id)
+  end
+
+  defp do_search(subquery, for_user, options \\ []) do
     q =
       from(
-        s in subquery(inner),
+        s in subquery(subquery),
-        order_by: s.search_distance,
+        order_by: [desc: s.search_rank],
-        limit: 20
+        limit: ^(options[:limit] || 20)
       )
 
-    Repo.all(q)
+    results =
+      q
+      |> Repo.all()
+      |> Enum.filter(&(&1.search_rank > 0))
+
+    boost_search_results(results, for_user)
+  end
+
+  defp fts_search_subquery(query) do
+    processed_query =
+      query
+      |> String.replace(~r/\W+/, " ")
+      |> String.trim()
+      |> String.split()
+      |> Enum.map(&(&1 <> ":*"))
+      |> Enum.join(" | ")
+
+    from(
+      u in User,
+      select_merge: %{
+        search_rank:
+          fragment(
+            """
+            ts_rank_cd(
+              setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
+              setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B'),
+              to_tsquery('simple', ?),
+              32
+            )
+            """,
+            u.nickname,
+            u.name,
+            ^processed_query
+          )
+      },
+      where:
+        fragment(
+          """
+            (setweight(to_tsvector('simple', regexp_replace(?, '\\W', ' ', 'g')), 'A') ||
+            setweight(to_tsvector('simple', regexp_replace(coalesce(?, ''), '\\W', ' ', 'g')), 'B')) @@ to_tsquery('simple', ?)
+          """,
+          u.nickname,
+          u.name,
+          ^processed_query
+        )
+    )
+  end
+
+  defp trigram_search_subquery(query) do
+    from(
+      u in User,
+      select_merge: %{
+        search_rank:
+          fragment(
+            "similarity(?, trim(? || ' ' || coalesce(?, '')))",
+            ^query,
+            u.nickname,
+            u.name
+          )
+      },
+      where: fragment("trim(? || ' ' || coalesce(?, '')) % ?", u.nickname, u.name, ^query)
+    )
+  end
+
+  defp boost_search_results(results, nil), do: results
+
+  defp boost_search_results(results, for_user) do
+    friends_ids = get_friends_ids(for_user)
+    followers_ids = get_followers_ids(for_user)
+
+    Enum.map(
+      results,
+      fn u ->
+        search_rank_coef =
+          cond do
+            u.id in friends_ids ->
+              1.2
+
+            u.id in followers_ids ->
+              1.1
+
+            true ->
+              1
+          end
+
+        Map.put(u, :search_rank, u.search_rank * search_rank_coef)
+      end
+    )
+    |> Enum.sort_by(&(-&1.search_rank))
   end
 
   def blocks_import(%User{} = blocker, blocked_identifiers) when is_list(blocked_identifiers) do

lib/pleroma/web/mastodon_api/mastodon_api_controller.ex

@@ -771,7 +771,7 @@ def status_search(user, query) do
   end
 
   def search2(%{assigns: %{user: user}} = conn, %{"q" => query} = params) do
-    accounts = User.search(query, params["resolve"] == "true")
+    accounts = User.search(query, params["resolve"] == "true", user)
 
     statuses = status_search(user, query)
 
@@ -795,7 +795,7 @@ def search2(%{assigns: %{user: user}} = conn, %{"q" => query} = params) do
   end
 
   def search(%{assigns: %{user: user}} = conn, %{"q" => query} = params) do
-    accounts = User.search(query, params["resolve"] == "true")
+    accounts = User.search(query, params["resolve"] == "true", user)
 
     statuses = status_search(user, query)
 
@@ -816,7 +816,7 @@ def search(%{assigns: %{user: user}} = conn, %{"q" => query} = params) do
   end
 
   def account_search(%{assigns: %{user: user}} = conn, %{"q" => query} = params) do
-    accounts = User.search(query, params["resolve"] == "true")
+    accounts = User.search(query, params["resolve"] == "true", user)
 
     res = AccountView.render("accounts.json", users: accounts, for: user, as: :user)
 

lib/pleroma/web/twitter_api/twitter_api_controller.ex

@@ -675,7 +675,7 @@ def search(%{assigns: %{user: user}} = conn, %{"q" => _query} = params) do
   end
 
   def search_user(%{assigns: %{user: user}} = conn, %{"query" => query}) do
-    users = User.search(query, true)
+    users = User.search(query, true, user)
 
     conn
     |> put_view(UserView)

priv/repo/migrations/20190115085500_create_user_fts_index.exs

@@ -0,0 +1,17 @@
+defmodule Pleroma.Repo.Migrations.CreateUserFtsIndex do
+  use Ecto.Migration
+
+  def change do
+    create index(
+             :users,
+             [
+               """
+               (setweight(to_tsvector('simple', regexp_replace(nickname, '\\W', ' ', 'g')), 'A') ||
+               setweight(to_tsvector('simple', regexp_replace(coalesce(name, ''), '\\W', ' ', 'g')), 'B'))
+               """
+             ],
+             name: :users_fts_index,
+             using: :gin
+           )
+  end
+end

priv/repo/migrations/20190118074940_fix_user_trigram_index.exs

@@ -0,0 +1,22 @@
+defmodule Pleroma.Repo.Migrations.FixUserTrigramIndex do
+  use Ecto.Migration
+
+  def up do
+    drop_if_exists(index(:users, [], name: :users_trigram_index))
+
+    create(
+      index(:users, ["(trim(nickname || ' ' || coalesce(name, ''))) gist_trgm_ops"],
+        name: :users_trigram_index,
+        using: :gist
+      )
+    )
+  end
+
+  def down do
+    drop_if_exists(index(:users, [], name: :users_trigram_index))
+
+    create(
+      index(:users, ["(nickname || name) gist_trgm_ops"], name: :users_trigram_index, using: :gist)
+    )
+  end
+end

test/user_test.exs

@@ -775,14 +775,61 @@ test "User.delete() plugs any possible zombie objects" do
   end
 
   describe "User.search" do
-    test "finds a user, ranking by similarity" do
+    test "finds a user by full or partial nickname" do
-      _user = insert(:user, %{name: "lain"})
+      user = insert(:user, %{nickname: "john"})
-      _user_two = insert(:user, %{name: "ean"})
-      _user_three = insert(:user, %{name: "ebn", nickname: "lain@mastodon.social"})
-      user_four = insert(:user, %{nickname: "lain@pleroma.soykaf.com"})
 
-      assert user_four ==
+      Enum.each(["john", "jo", "j"], fn query ->
-               User.search("lain@ple") |> List.first() |> Map.put(:search_distance, nil)
+        assert user == User.search(query) |> List.first() |> Map.put(:search_rank, nil)
+      end)
+    end
+
+    test "finds a user by full or partial name" 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)
+      end)
+    end
+
+    test "finds users, preferring nickname matches over name matches" do
+      u1 = insert(:user, %{name: "lain", nickname: "nick1"})
+      u2 = insert(:user, %{nickname: "lain", name: "nick1"})
+
+      assert [u2.id, u1.id] == Enum.map(User.search("lain"), & &1.id)
+    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"})
+
+      assert [u2.id, u1.id] == Enum.map(User.search("bar word"), & &1.id)
+    end
+
+    test "finds users, ranking by similarity" do
+      u1 = insert(:user, %{name: "lain"})
+      _u2 = insert(:user, %{name: "ean"})
+      u3 = insert(:user, %{name: "ebn", nickname: "lain@mastodon.social"})
+      u4 = insert(:user, %{nickname: "lain@pleroma.soykaf.com"})
+
+      assert [u4.id, u3.id, u1.id] == Enum.map(User.search("lain@ple"), & &1.id)
+    end
+
+    test "finds users, handling misspelled requests" do
+      u1 = insert(:user, %{name: "lain"})
+
+      assert [u1.id] == Enum.map(User.search("laiin"), & &1.id)
+    end
+
+    test "finds users, boosting ranks of friends and followers" do
+      u1 = insert(:user)
+      u2 = insert(:user, %{name: "Doe"})
+      follower = insert(:user, %{name: "Doe"})
+      friend = insert(:user, %{name: "Doe"})
+
+      {:ok, follower} = User.follow(follower, u1)
+      {:ok, u1} = User.follow(u1, friend)
+
+      assert [friend.id, follower.id, u2.id] == Enum.map(User.search("doe", false, u1), & &1.id)
     end
 
     test "finds a user whose name is nil" do
@@ -792,7 +839,15 @@ test "finds a user whose name is nil" do
       assert user_two ==
                User.search("lain@pleroma.soykaf.com")
                |> List.first()
-               |> Map.put(:search_distance, nil)
+               |> Map.put(:search_rank, nil)
+    end
+
+    test "does not yield false-positive matches" do
+      insert(:user, %{name: "John Doe"})
+
+      Enum.each(["mary", "a", ""], fn query ->
+        assert [] == User.search(query)
+      end)
     end
   end
 

test/web/twitter_api/twitter_api_controller_test.exs

@@ -1655,16 +1655,16 @@ test "it denies a friend request" do
   describe "GET /api/pleroma/search_user" do
     test "it returns users, ordered by similarity", %{conn: conn} do
       user = insert(:user, %{name: "eal"})
-      user_two = insert(:user, %{name: "ean"})
+      user_two = insert(:user, %{name: "eal me"})
-      user_three = insert(:user, %{name: "ebn"})
+      _user_three = insert(:user, %{name: "zzz"})
 
       resp =
         conn
-        |> get(twitter_api_search__path(conn, :search_user), query: "eal")
+        |> get(twitter_api_search__path(conn, :search_user), query: "eal me")
         |> json_response(200)
 
-      assert length(resp) == 3
+      assert length(resp) == 2
-      assert [user.id, user_two.id, user_three.id] == Enum.map(resp, fn %{"id" => id} -> id end)
+      assert [user_two.id, user.id] == Enum.map(resp, fn %{"id" => id} -> id end)
     end
   end