Add nodeinfo tasker

config/config.exs

@@ -567,7 +567,8 @@
     attachments_cleanup: 1,
     new_users_digest: 1,
     mute_expire: 5,
-    search_indexing: 10
+    search_indexing: 10,
+    nodeinfo_fetcher: 1
   ],
   plugins: [
     Oban.Plugins.Pruner,

lib/pleroma/instances/instance.ex

@@ -22,7 +22,8 @@ defmodule Pleroma.Instances.Instance do
     field(:host, :string)
     field(:unreachable_since, :naive_datetime_usec)
     field(:favicon, :string)
-    field(:favicon_updated_at, :naive_datetime)
+    field(:metadata_updated_at, :naive_datetime)
+    field(:nodeinfo, :map, default: %{})
 
     timestamps()
   end
@@ -138,42 +139,80 @@ defp parse_datetime(datetime) when is_binary(datetime) do
 
   defp parse_datetime(datetime), do: datetime
 
-  def get_or_update_favicon(%URI{host: host} = instance_uri) do
-    existing_record = Repo.get_by(Instance, %{host: host})
-    now = NaiveDateTime.utc_now()
+  def needs_update(nil), do: true
 
-    if existing_record && existing_record.favicon_updated_at &&
-         NaiveDateTime.diff(now, existing_record.favicon_updated_at) < 86_400 do
-      existing_record.favicon
-    else
-      favicon = scrape_favicon(instance_uri)
+  def needs_update(%Instance{metadata_updated_at: metadata_updated_at}) do
+    now = NaiveDateTime.utc_now()
+    NaiveDateTime.diff(now, metadata_updated_at) > 86_400
+  end
+
+  def update_metadata(%URI{host: host} = uri) do
+    existing_record = Repo.get_by(Instance, %{host: host})
 
     if existing_record do
-        existing_record
-        |> changeset(%{favicon: favicon, favicon_updated_at: now})
-        |> Repo.update()
-      else
+      if needs_update(existing_record) do
+        favicon = scrape_favicon(uri)
+        nodeinfo = scrape_nodeinfo(uri)
+
         %Instance{}
-        |> changeset(%{host: host, favicon: favicon, favicon_updated_at: now})
+        |> changeset(%{host: host, favicon: favicon, nodeinfo: nodeinfo})
+        |> Repo.update()
+      end
+    else
+      favicon = scrape_favicon(uri)
+      nodeinfo = scrape_nodeinfo(uri)
+
+      %Instance{}
+      |> changeset(%{host: host, favicon: favicon, nodeinfo: nodeinfo})
       |> Repo.insert()
     end
-
-      favicon
   end
-  rescue
-    e ->
-      Logger.warn("Instance.get_or_update_favicon(\"#{host}\") error: #{inspect(e)}")
+
+  def get_favicon(%URI{host: host} = instance_uri) do
+    existing_record = Repo.get_by(Instance, %{host: host})
+
+    if existing_record do
+      existing_record.favicon
+    else
       nil
     end
+  end
+
+  defp scrape_nodeinfo(%URI{} = instance_uri) do
+    with {_, true} <- {:reachable, reachable?(instance_uri.host)},
+         {:ok, %Tesla.Env{status: 200, body: body}} <-
+           Tesla.get(
+             "https://#{instance_uri.host}/.well-known/nodeinfo",
+             headers: [{"Accept", "application/json"}]
+           ),
+         {:ok, json} <- Jason.decode(body),
+         {:ok, %{"links" => links}} <- {:ok, json},
+         {:ok, %{"href" => href}} <-
+           {:ok,
+            Enum.find(links, &(&1["rel"] == "http://nodeinfo.diaspora.software/ns/schema/2.0"))},
+         {:ok, %Tesla.Env{body: data}} <-
+           Pleroma.HTTP.get(ref, [{"accept", "application/json"}], []),
+         {:ok, nodeinfo} <- Jason.decode(data) do
+      nodeinfo
+    else
+      {:reachable, false} ->
+        Logger.debug(
+          "Instance.scrape_favicon(\"#{to_string(instance_uri)}\") ignored unreachable host"
+        )
+
+        nil
+
+      _ ->
+        nil
+    end
+  end
 
   defp scrape_favicon(%URI{} = instance_uri) do
-    try do
     with {_, true} <- {:reachable, reachable?(instance_uri.host)},
          {:ok, %Tesla.Env{body: html}} <-
            Pleroma.HTTP.get(to_string(instance_uri), [{"accept", "text/html"}], []),
          {_, [favicon_rel | _]} when is_binary(favicon_rel) <-
-             {:parse,
-              html |> Floki.parse_document!() |> Floki.attribute("link[rel=icon]", "href")},
+           {:parse, html |> Floki.parse_document!() |> Floki.attribute("link[rel=icon]", "href")},
          {_, favicon} when is_binary(favicon) <-
            {:merge, URI.merge(instance_uri, favicon_rel) |> to_string()} do
       favicon
@@ -186,14 +225,6 @@ defp scrape_favicon(%URI{} = instance_uri) do
         nil
 
       _ ->
-          nil
-      end
-    rescue
-      e ->
-        Logger.warn(
-          "Instance.scrape_favicon(\"#{to_string(instance_uri)}\") error: #{inspect(e)}"
-        )
-
         nil
     end
   end

lib/pleroma/workers/nodeinfo_fetcher_worker.ex

@@ -0,0 +1,19 @@
+defmodule Pleroma.Workers.NodeInfoFetcherWorker do
+  use Oban.Worker, queue: :backup, max_attempts: 1
+
+  alias Oban.Job
+  alias Pleroma.Instance
+
+  def process(domain) do
+    %{"op" => "process", "domain" => domain}
+    |> new()
+    |> Oban.insert()
+  end
+
+  def perform(%Job{
+        args: %{"op" => "process", "domain" => domain}
+      }) do
+    uri = URI.parse(domain)
+    Instance.get_or_update_favicon(uri)
+  end
+end

priv/repo/migrations/20221020135943_add_nodeinfo.exs

@@ -0,0 +1,15 @@
+defmodule Pleroma.Repo.Migrations.AddNodeinfo do
+  use Ecto.Migration
+
+  def up do
+    alter table(:instances) do
+      add_if_not_exists(:nodeinfo, :map, default: %{})
+    end
+  end
+
+  def down do
+    alter table(:instances) do
+      remove_if_exists(:nodeinfo, :map)
+    end
+  end
+end