pleroma

commit: 94e4f215896dc7976a54fd146daf3e286602925a
parent f726e5fbbdab89dd2a8cb498112ed6866047d3d1
Author: Lain Soykaf <lain@lain.com>
Date:   Thu, 23 May 2024 14:38:30 +0400

QdrantSearch: Deal with actor restrictions

Diffstat:
M
lib/pleroma/search/qdrant_search.ex
22
++++++++++++++++++++--
M
test/pleroma/search/qdrant_search_test.exs
95
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-

2 files changed, 114 insertions(+), 3 deletions(-)
diff --git a/lib/pleroma/search/qdrant_search.ex b/lib/pleroma/search/qdrant_search.ex
@@ -43,23 +43,41 @@ defmodule Pleroma.Search.QdrantSearch do
     end
   end
 
+  defp actor_from_activity(%{data: %{"actor" => actor}}) do
+    actor
+  end
+
+  defp actor_from_activity(_), do: nil
+
   defp build_index_payload(activity, embedding) do
+    actor = actor_from_activity(activity)
+    published_at = activity.data["published"]
+
     %{
       points: [
         %{
           id: activity.id |> FlakeId.from_string() |> Ecto.UUID.cast!(),
-          vector: embedding
+          vector: embedding,
+          payload: %{actor: actor, published_at: published_at}
         }
       ]
     }
   end
 
   defp build_search_payload(embedding, options) do
-    %{
+    base = %{
       vector: embedding,
       limit: options[:limit] || 20,
       offset: options[:offset] || 0
     }
+
+    if options[:actor] do
+      Map.put(base, :filter, %{
+        must: [%{key: "actor", match: %{value: options[:actor].ap_id}}]
+      })
+    else
+      base
+    end
   end
 
   @impl true
diff --git a/test/pleroma/search/qdrant_search_test.exs b/test/pleroma/search/qdrant_search_test.exs
@@ -15,6 +15,94 @@ defmodule Pleroma.Search.QdrantSearchTest do
   alias Pleroma.Workers.SearchIndexingWorker
 
   describe "Qdrant search" do
+    test "searches for a term by encoding it and sending it to qdrant" do
+      user = insert(:user)
+
+      {:ok, activity} =
+        CommonAPI.post(user, %{
+          status: "guys i just don't wanna leave the swamp",
+          visibility: "public"
+        })
+
+      Config
+      |> expect(:get, 3, fn
+        [Pleroma.Search, :module], nil ->
+          QdrantSearch
+
+        [Pleroma.Search.QdrantSearch, key], nil ->
+          %{
+            openai_model: "a_model",
+            openai_url: "https://openai.url",
+            qdrant_url: "https://qdrant.url"
+          }[key]
+      end)
+
+      Tesla.Mock.mock(fn
+        %{url: "https://openai.url/v1/embeddings", method: :post} ->
+          Tesla.Mock.json(%{
+            data: [%{embedding: [1, 2, 3]}]
+          })
+
+        %{url: "https://qdrant.url/collections/posts/points/search", method: :post, body: body} ->
+          data = Jason.decode!(body)
+          refute data["filter"]
+
+          Tesla.Mock.json(%{
+            result: [%{"id" => activity.id |> FlakeId.from_string() |> Ecto.UUID.cast!()}]
+          })
+      end)
+
+      results = QdrantSearch.search(nil, "guys i just don't wanna leave the swamp", %{})
+
+      assert results == [activity]
+    end
+
+    test "for a given actor, ask for only relevant matches" do
+      user = insert(:user)
+
+      {:ok, activity} =
+        CommonAPI.post(user, %{
+          status: "guys i just don't wanna leave the swamp",
+          visibility: "public"
+        })
+
+      Config
+      |> expect(:get, 3, fn
+        [Pleroma.Search, :module], nil ->
+          QdrantSearch
+
+        [Pleroma.Search.QdrantSearch, key], nil ->
+          %{
+            openai_model: "a_model",
+            openai_url: "https://openai.url",
+            qdrant_url: "https://qdrant.url"
+          }[key]
+      end)
+
+      Tesla.Mock.mock(fn
+        %{url: "https://openai.url/v1/embeddings", method: :post} ->
+          Tesla.Mock.json(%{
+            data: [%{embedding: [1, 2, 3]}]
+          })
+
+        %{url: "https://qdrant.url/collections/posts/points/search", method: :post, body: body} ->
+          data = Jason.decode!(body)
+
+          assert data["filter"] == %{
+                   "must" => [%{"key" => "actor", "match" => %{"value" => user.ap_id}}]
+                 }
+
+          Tesla.Mock.json(%{
+            result: [%{"id" => activity.id |> FlakeId.from_string() |> Ecto.UUID.cast!()}]
+          })
+      end)
+
+      results =
+        QdrantSearch.search(nil, "guys i just don't wanna leave the swamp", %{actor: user})
+
+      assert results == [activity]
+    end
+
     test "indexes a public post on creation, deletes from the index on deletion" do
       user = insert(:user)
 
@@ -29,7 +117,12 @@ defmodule Pleroma.Search.QdrantSearchTest do
         %{method: :put, url: "https://qdrant.url/collections/posts/points", body: body} ->
           send(self(), "posted_to_qdrant")
 
-          assert match?(%{"points" => [%{"vector" => [1, 2, 3]}]}, Jason.decode!(body))
+          data = Jason.decode!(body)
+          %{"points" => [%{"vector" => vector, "payload" => payload}]} = data
+
+          assert vector == [1, 2, 3]
+          assert payload["actor"]
+          assert payload["published_at"]
 
           Tesla.Mock.json("ok")