instances: Store some metadata based on NodeInfo

lib/pleroma/instances/instance.ex

@@ -7,6 +7,7 @@ defmodule Pleroma.Instances.Instance do
 
   alias Pleroma.Instances
   alias Pleroma.Instances.Instance
+  alias Pleroma.Maps
   alias Pleroma.Repo
   alias Pleroma.User
   alias Pleroma.Workers.BackgroundWorker
@@ -24,6 +25,14 @@ defmodule Pleroma.Instances.Instance do
     field(:favicon, :string)
     field(:favicon_updated_at, :naive_datetime)
 
+    embeds_one :metadata, Pleroma.Instances.Metadata, primary_key: false do
+      field(:software_name, :string)
+      field(:software_version, :string)
+      field(:software_repository, :string)
+    end
+
+    field(:metadata_updated_at, :utc_datetime)
+
     timestamps()
   end
 
@@ -31,11 +40,17 @@ defmodule Pleroma.Instances.Instance do
 
   def changeset(struct, params \\ %{}) do
     struct
-    |> cast(params, [:host, :unreachable_since, :favicon, :favicon_updated_at])
+    |> cast(params, __schema__(:fields) -- [:metadata])
+    |> cast_embed(:metadata, with: &metadata_changeset/2)
     |> validate_required([:host])
     |> unique_constraint(:host)
   end
 
+  def metadata_changeset(struct, params \\ %{}) do
+    struct
+    |> cast(params, [:software_name, :software_version, :software_repository])
+  end
+
   def filter_reachable([]), do: %{}
 
   def filter_reachable(urls_or_hosts) when is_list(urls_or_hosts) do
@@ -198,6 +213,89 @@ defp scrape_favicon(%URI{} = instance_uri) do
     end
   end
 
+  def get_or_update_metadata(%URI{host: host} = instance_uri) do
+    existing_record = Repo.get_by(Instance, %{host: host})
+    now = NaiveDateTime.utc_now()
+
+    if existing_record && existing_record.metadata_updated_at &&
+         NaiveDateTime.diff(now, existing_record.metadata_updated_at) < 86_400 do
+      existing_record.metadata
+    else
+      metadata = scrape_metadata(instance_uri)
+
+      if existing_record do
+        existing_record
+        |> changeset(%{metadata: metadata, metadata_updated_at: now})
+        |> Repo.update()
+      else
+        %Instance{}
+        |> changeset(%{host: host, metadata: metadata, metadata_updated_at: now})
+        |> Repo.insert()
+      end
+
+      metadata
+    end
+  end
+
+  defp get_nodeinfo_uri(well_known) do
+    links = Map.get(well_known, "links", [])
+
+    nodeinfo21 =
+      Enum.find(links, &(&1["rel"] == "http://nodeinfo.diaspora.software/ns/schema/2.1"))["href"]
+
+    nodeinfo20 =
+      Enum.find(links, &(&1["rel"] == "http://nodeinfo.diaspora.software/ns/schema/2.0"))["href"]
+
+    cond do
+      is_binary(nodeinfo21) -> {:ok, nodeinfo21}
+      is_binary(nodeinfo20) -> {:ok, nodeinfo20}
+      true -> {:error, :no_links}
+    end
+  end
+
+  defp scrape_metadata(%URI{} = instance_uri) do
+    try do
+      with {_, true} <- {:reachable, reachable?(instance_uri.host)},
+           {:ok, %Tesla.Env{body: well_known_body}} <-
+             instance_uri
+             |> URI.merge("/.well-known/nodeinfo")
+             |> to_string()
+             |> Pleroma.HTTP.get([{"accept", "application/json"}]),
+           {:ok, well_known_json} <- Jason.decode(well_known_body),
+           {:ok, nodeinfo_uri} <- get_nodeinfo_uri(well_known_json),
+           {:ok, %Tesla.Env{body: nodeinfo_body}} <-
+             Pleroma.HTTP.get(nodeinfo_uri, [{"accept", "application/json"}]),
+           {:ok, nodeinfo} <- Jason.decode(nodeinfo_body) do
+        # Can extract more metadata from NodeInfo but need to be careful about it's size,
+        # can't just dump the entire thing
+        software = Map.get(nodeinfo, "software", %{})
+
+        %{
+          software_name: software["name"],
+          software_version: software["version"]
+        }
+        |> Maps.put_if_present(:software_repository, software["repository"])
+      else
+        {:reachable, false} ->
+          Logger.debug(
+            "Instance.scrape_metadata(\"#{to_string(instance_uri)}\") ignored unreachable host"
+          )
+
+          nil
+
+        _ ->
+          nil
+      end
+    rescue
+      e ->
+        Logger.warn(
+          "Instance.scrape_metadata(\"#{to_string(instance_uri)}\") error: #{inspect(e)}"
+        )
+
+        nil
+    end
+  end
+
   @doc """
   Deletes all users from an instance in a background task, thus also deleting
   all of those users' activities and notifications.

priv/repo/migrations/20230306112859_instances_add_metadata.exs

@@ -0,0 +1,14 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2023 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Repo.Migrations.InstancesAddMetadata do
+  use Ecto.Migration
+
+  def change do
+    alter table(:instances) do
+      add(:metadata, :map)
+      add(:metadata_updated_at, :utc_datetime)
+    end
+  end
+end

test/fixtures/mastodon-nodeinfo20.json

@@ -0,0 +1 @@
+{"version":"2.0","software":{"name":"mastodon","version":"4.1.0"},"protocols":["activitypub"],"services":{"outbound":[],"inbound":[]},"usage":{"users":{"total":971090,"activeMonth":167218,"activeHalfyear":384808},"localPosts":52071541},"openRegistrations":true,"metadata":{}}

test/fixtures/mastodon-well-known-nodeinfo.json

@@ -0,0 +1 @@
+{"links":[{"rel":"http://nodeinfo.diaspora.software/ns/schema/2.0","href":"https://mastodon.example.org/nodeinfo/2.0"}]}

test/fixtures/wildebeest-nodeinfo21.json

@@ -0,0 +1 @@
+{"version":"2.1","software":{"name":"wildebeest","version":"0.0.1","repository":"https://github.com/cloudflare/wildebeest"},"protocols":["activitypub"],"usage":{"users":{"total":1,"activeMonth":1,"activeHalfyear":1}},"openRegistrations":false,"metadata":{"upstream":{"name":"mastodon","version":"3.5.1"}}}

test/fixtures/wildebeest-well-known-nodeinfo.json

@@ -0,0 +1 @@
+{"links":[{"rel":"http://nodeinfo.diaspora.software/ns/schema/2.0","href":"https://wildebeest.example.org/nodeinfo/2.0"},{"rel":"http://nodeinfo.diaspora.software/ns/schema/2.1","href":"https://wildebeest.example.org/nodeinfo/2.1"}]}

test/pleroma/instances/instance_test.exs

@@ -161,6 +161,66 @@ test "Doesn't scrapes unreachable instances" do
     end
   end
 
+  describe "get_or_update_metadata/1" do
+    test "Scrapes Wildebeest NodeInfo" do
+      Tesla.Mock.mock(fn
+        %{url: "https://wildebeest.example.org/.well-known/nodeinfo"} ->
+          %Tesla.Env{
+            status: 200,
+            body: File.read!("test/fixtures/wildebeest-well-known-nodeinfo.json")
+          }
+
+        %{url: "https://wildebeest.example.org/nodeinfo/2.1"} ->
+          %Tesla.Env{
+            status: 200,
+            body: File.read!("test/fixtures/wildebeest-nodeinfo21.json")
+          }
+      end)
+
+      expected = %{
+        software_name: "wildebeest",
+        software_repository: "https://github.com/cloudflare/wildebeest",
+        software_version: "0.0.1"
+      }
+
+      assert expected ==
+               Instance.get_or_update_metadata(URI.parse("https://wildebeest.example.org/"))
+
+      expected = %Pleroma.Instances.Instance.Pleroma.Instances.Metadata{
+        software_name: "wildebeest",
+        software_repository: "https://github.com/cloudflare/wildebeest",
+        software_version: "0.0.1"
+      }
+
+      assert expected ==
+               Repo.get_by(Pleroma.Instances.Instance, %{host: "wildebeest.example.org"}).metadata
+    end
+
+    test "Scrapes Mastodon NodeInfo" do
+      Tesla.Mock.mock(fn
+        %{url: "https://mastodon.example.org/.well-known/nodeinfo"} ->
+          %Tesla.Env{
+            status: 200,
+            body: File.read!("test/fixtures/mastodon-well-known-nodeinfo.json")
+          }
+
+        %{url: "https://mastodon.example.org/nodeinfo/2.0"} ->
+          %Tesla.Env{
+            status: 200,
+            body: File.read!("test/fixtures/mastodon-nodeinfo20.json")
+          }
+      end)
+
+      expected = %{
+        software_name: "mastodon",
+        software_version: "4.1.0"
+      }
+
+      assert expected ==
+               Instance.get_or_update_metadata(URI.parse("https://mastodon.example.org/"))
+    end
+  end
+
   test "delete_users_and_activities/1 deletes remote instance users and activities" do
     [mario, luigi, _peach, wario] =
       users = [