Add fuzzy matching to ParadeDB search

Make ParadeDB search typo-tolerant via configurable fuzzy distance and document the new setting. Add coverage for fuzzy query SQL and id parsing, and keep integration test stability under connection load.

config/config.exs

@@ -940,7 +940,8 @@ config :pleroma, Pleroma.Search.Meilisearch,
 
 config :pleroma, Pleroma.Search.ParadeDB,
   url: nil,
-  table: "pleroma_search_documents"
+  table: "pleroma_search_documents",
+  fuzzy_distance: 0
 
 config :pleroma, Pleroma.Search.ParadeDB.Repo,
   pool_size: 10,

config/description.exs

@@ -3560,6 +3560,12 @@ config :pleroma, :config_description, [
         type: :string,
         description: "Table name used for search documents.",
         suggestion: ["pleroma_search_documents"]
+      },
+      %{
+        key: :fuzzy_distance,
+        type: :integer,
+        description: "Edit-distance used for fuzzy token matching (0 disables fuzziness, max 2).",
+        suggestion: [0, 1, 2]
       }
     ]
   },

docs/configuration/search.md

@@ -160,7 +160,8 @@ To use it, set the search module to `Pleroma.Search.ParadeDB` and configure the
 >
 > config :pleroma, Pleroma.Search.ParadeDB,
 >   url: System.get_env("PARADEDB_DATABASE_URL") || "postgres://postgres:postgres@127.0.0.1:5432/paradedb",
->   table: "pleroma_search_documents"
+>   table: "pleroma_search_documents",
+>   fuzzy_distance: 0
 
 Then, create the table and BM25 index once:
 
@@ -178,4 +179,6 @@ After that, run an initial indexing pass to backfill existing posts:
 
 Note: Like the other external search backends, only public/unlisted Notes are indexed.
 
+Set `fuzzy_distance` to `1` or `2` to allow typo-tolerant matching (`2` is the maximum ParadeDB supports).
+
 Indexing runs via the Oban `search_indexing` queue, so ensure that queue is enabled (not paused) in your Oban configuration.

lib/pleroma/search/paradedb.ex

@@ -35,6 +35,22 @@ defmodule Pleroma.Search.ParadeDB do
     Config.get([Pleroma.Search.ParadeDB, :table], @default_table)
   end
 
+  defp fuzzy_distance do
+    case Config.get([Pleroma.Search.ParadeDB, :fuzzy_distance], 0) do
+      value when is_integer(value) and value in 0..2 ->
+        value
+
+      value when is_binary(value) ->
+        case Integer.parse(value) do
+          {parsed, ""} when parsed in 0..2 -> parsed
+          _ -> 0
+        end
+
+      _ ->
+        0
+    end
+  end
+
   @impl true
   def create_index do
     with {:ok, _} <- client_impl().query("CREATE EXTENSION IF NOT EXISTS pg_search", []),
@@ -190,7 +206,9 @@ defmodule Pleroma.Search.ParadeDB do
   end
 
   defp search_ids(query, limit, offset, author) do
-    base_sql = "SELECT id FROM #{table()} WHERE content &&& $1"
+    base_sql =
+      "SELECT id FROM #{table()} WHERE content &&& #{search_query_param_sql(fuzzy_distance())}"
+
     params = [query]
 
     {base_sql, params} =
@@ -216,6 +234,9 @@ defmodule Pleroma.Search.ParadeDB do
     end
   end
 
+  defp search_query_param_sql(0), do: "$1"
+  defp search_query_param_sql(distance), do: "$1::pdb.fuzzy(#{distance})"
+
   defp create_table_sql(table) do
     """
     CREATE TABLE IF NOT EXISTS #{table} (

test/pleroma/search/paradedb_integration_test.exs

@@ -73,7 +73,7 @@ defmodule Pleroma.Search.ParadeDBIntegrationTest do
         end
       end)
 
-      {:ok, _pid} = start_supervised({Pleroma.Search.ParadeDB.Repo, pool_size: 1})
+      {:ok, _pid} = start_supervised({Pleroma.Search.ParadeDB.Repo, pool_size: 5})
 
       previous_table = Pleroma.Config.get([Pleroma.Search.ParadeDB, :table], nil)
       table = "pleroma_search_documents_it_#{System.unique_integer([:positive])}"

test/pleroma/search/paradedb_test.exs

@@ -161,12 +161,15 @@ defmodule Pleroma.Search.ParadeDBTest do
       end)
 
       Config
-      |> expect(:get, 2, fn
+      |> expect(:get, 3, fn
         [Pleroma.Search.ParadeDB, :client_impl], nil ->
           ClientMock
 
         [Pleroma.Search.ParadeDB, :table], "pleroma_search_documents" ->
           "pleroma_search_documents"
+
+        [Pleroma.Search.ParadeDB, :fuzzy_distance], 0 ->
+          0
       end)
 
       assert [%{id: ^activity2_id}, %{id: ^activity1_id}] =
@@ -196,12 +199,15 @@ defmodule Pleroma.Search.ParadeDBTest do
       end)
 
       Config
-      |> expect(:get, 2, fn
+      |> expect(:get, 3, fn
         [Pleroma.Search.ParadeDB, :client_impl], nil ->
           ClientMock
 
         [Pleroma.Search.ParadeDB, :table], "pleroma_search_documents" ->
           "pleroma_search_documents"
+
+        [Pleroma.Search.ParadeDB, :fuzzy_distance], 0 ->
+          0
       end)
 
       assert [%{id: ^public_activity_id}] = ParadeDB.search(nil, "swamp", limit: 40, offset: 0)
@@ -225,15 +231,50 @@ defmodule Pleroma.Search.ParadeDBTest do
       end)
 
       Config
-      |> expect(:get, 2, fn
+      |> expect(:get, 3, fn
         [Pleroma.Search.ParadeDB, :client_impl], nil ->
           ClientMock
 
         [Pleroma.Search.ParadeDB, :table], "pleroma_search_documents" ->
           "pleroma_search_documents"
+
+        [Pleroma.Search.ParadeDB, :fuzzy_distance], 0 ->
+          0
       end)
 
       assert [%{id: ^activity_id}] = ParadeDB.search(nil, "swamp", limit: 40, offset: 0)
     end
+
+    test "search uses configured fuzzy distance" do
+      user = insert(:user)
+
+      {:ok, activity} = CommonAPI.post(user, %{status: "runing shoes", visibility: "public"})
+
+      activity_id = activity.id
+      {:ok, dumped_id} = FlakeId.Ecto.CompatType.dump(activity_id)
+
+      ClientMock
+      |> expect(:query, fn sql, params ->
+        assert sql =~ "SELECT id FROM pleroma_search_documents"
+        assert sql =~ "$1::pdb.fuzzy(1)"
+        assert ["running shoes", _limit, _offset] = params
+
+        {:ok, %{rows: [[dumped_id]]}}
+      end)
+
+      Config
+      |> expect(:get, 3, fn
+        [Pleroma.Search.ParadeDB, :client_impl], nil ->
+          ClientMock
+
+        [Pleroma.Search.ParadeDB, :table], "pleroma_search_documents" ->
+          "pleroma_search_documents"
+
+        [Pleroma.Search.ParadeDB, :fuzzy_distance], 0 ->
+          1
+      end)
+
+      assert [%{id: ^activity_id}] = ParadeDB.search(nil, "running shoes", limit: 40, offset: 0)
+    end
   end
 end