add vrchat world filter based on content tags

Author: lun-4

.gitignore

@@ -29,3 +29,5 @@ vrhose-*.tar
 .Elixir*.zig
 priv/lib
 *.db*
+
+.envrc

lib/vrhose/hydrator.ex

@@ -35,7 +35,47 @@ defmodule VRHose.Hydrator do
     )
   end
 
-  defp process({did, post_data, subscribers} = event) do
+  defp process({_did, post_data, _subscribers} = event) do
+    if post_data.world_id == nil do
+      process_post(event)
+    else
+      is_good? = is_good_world?(post_data.world_id)
+      Logger.warning("is wrld #{post_data.world_id} good? #{inspect(is_good?)}")
+
+      if is_good? do
+        process_post(event)
+      end
+    end
+  end
+
+  @unwanted_tags [
+    "content_sex",
+    "content_adult"
+  ]
+
+  defp is_good_world?(wrld_id) do
+    case VRHose.VRChatWorld.fetch(wrld_id) do
+      {:ok, nil} ->
+        true
+
+      {:ok, world} ->
+        matches_unwanted_tags? =
+          world.tags
+          |> Jason.decode!()
+          |> Enum.map(fn tag ->
+            tag in @unwanted_tags
+          end)
+          |> Enum.any?()
+
+        # wanted when unwanted tags arent in the world
+        not matches_unwanted_tags?
+
+      {:error, _} ->
+        true
+    end
+  end
+
+  defp process_post({did, post_data, subscribers} = event) do
     identity = VRHose.Identity.one(did)
 
     post_data =

lib/vrhose/vrchat_world.ex

@@ -0,0 +1,80 @@
+defmodule VRHose.VRChatWorld do
+  use Ecto.Schema
+  import Ecto.Query
+  import Ecto.Changeset
+  alias VRHose.Repo
+
+  @type t :: %__MODULE__{}
+
+  schema "vrchat_worlds" do
+    field(:wrld_id, :string)
+    field(:name, :string)
+    field(:author_id, :string)
+    field(:tags, :string)
+    field(:capacity, :integer)
+    field(:description, :string)
+    timestamps(autogenerate: {VRHose.Data, :generate_unix_timestamp, []})
+  end
+
+  def one(wrld_id) do
+    query = from(s in __MODULE__, where: s.wrld_id == ^wrld_id, select: s)
+    Repo.replica(wrld_id).one(query, log: false)
+  end
+
+  def changeset(%__MODULE__{} = world, params) do
+    world
+    |> cast(params, [:wrld_id, :name, :author_id, :tags, :capacity, :description])
+    |> validate_required([:wrld_id, :name, :author_id, :tags, :capacity, :description])
+  end
+
+  def insert(wrld_id, name, author_id, tags, capacity, description) do
+    %__MODULE__{}
+    |> changeset(%{
+      wrld_id: wrld_id,
+      name: name,
+      author_id: author_id,
+      tags: tags,
+      capacity: capacity,
+      description: description
+    })
+    |> Repo.insert()
+  end
+
+  def fetch(wrld_id) do
+    maybe_world = one(wrld_id)
+
+    if maybe_world == nil do
+      {:ok, resp} = fetch_upstream(wrld_id)
+
+      tags =
+        resp.body["tags"]
+        |> Enum.filter(fn tag ->
+          String.starts_with?(tag, "content_")
+        end)
+        |> Jason.encode!()
+
+      insert(
+        resp.body["id"],
+        resp.body["name"],
+        resp.body["authorId"],
+        tags,
+        resp.body["capacity"],
+        resp.body["description"]
+      )
+    else
+      {:ok, maybe_world}
+    end
+  end
+
+  defp fetch_upstream(wrld_id) do
+    operator_email = System.get_env("VRHOSE_OPERATOR_EMAIL")
+
+    if operator_email == nil do
+      {:error, :no_operator_email}
+    else
+      Req.get("https://api.vrchat.cloud/api/1/worlds/#{wrld_id}",
+        headers: [{"User-Agent", "BlueskyFirehoseVR/0.0.0 #{operator_email}"}]
+      )
+    end
+  end
+end

priv/repo/migrations/20241214185646_add_world_labels.exs

@@ -0,0 +1,17 @@
+defmodule VRHose.Repo.Migrations.AddWorldLabels do
+  use Ecto.Migration
+
+  def change do
+    create table(:vrchat_worlds) do
+      add(:wrld_id, :text, null: false)
+      add(:name, :text, null: false)
+      add(:author_id, :text, null: false)
+      add(:tags, :text, null: false)
+      add(:capacity, :integer, null: false)
+      add(:description, :text, null: false)
+      timestamps()
+    end
+
+    create(unique_index(:vrchat_worlds, [:wrld_id]))
+  end
+end