Merge branch 'feature/rich-media' into 'develop'

URL previews in posts

Closes #402

See merge request pleroma/pleroma!617

4 files changed, 79 insertions, 0 deletions

diff --git a/lib/pleroma/web/rich_media/controllers/rich_media_controller.ex b/lib/pleroma/web/rich_media/controllers/rich_media_controller.ex
new file mode 100644
index 000000000..91019961d
--- /dev/null
+++ b/lib/pleroma/web/rich_media/controllers/rich_media_controller.ex
@@ -0,0 +1,17 @@
+defmodule Pleroma.Web.RichMedia.RichMediaController do
+  use Pleroma.Web, :controller
+
+  import Pleroma.Web.ControllerHelper, only: [json_response: 3]
+
+  def parse(conn, %{"url" => url}) do
+    case Pleroma.Web.RichMedia.Parser.parse(url) do
+      {:ok, data} ->
+        conn
+        |> json_response(200, data)
+
+      {:error, msg} ->
+        conn
+        |> json_response(404, msg)
+    end
+  end
+end
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
new file mode 100644
index 000000000..477a38196
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -0,0 +1,26 @@
+defmodule Pleroma.Web.RichMedia.Parser do
+  @parsers [Pleroma.Web.RichMedia.Parsers.OGP]
+
+  def parse(url) do
+    {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url)
+
+    html |> maybe_parse() |> get_parsed_data()
+  end
+
+  defp maybe_parse(html) do
+    Enum.reduce_while(@parsers, %{}, fn parser, acc ->
+      case parser.parse(html, acc) do
+        {:ok, data} -> {:halt, data}
+        {:error, _msg} -> {:cont, acc}
+      end
+    end)
+  end
+
+  defp get_parsed_data(data) when data == %{} do
+    {:error, "No metadata found"}
+  end
+
+  defp get_parsed_data(data) do
+    {:ok, data}
+  end
+end
diff --git a/lib/pleroma/web/rich_media/parsers/ogp.ex b/lib/pleroma/web/rich_media/parsers/ogp.ex
new file mode 100644
index 000000000..5773a5263
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parsers/ogp.ex
@@ -0,0 +1,30 @@
+defmodule Pleroma.Web.RichMedia.Parsers.OGP do
+  def parse(html, data) do
+    with elements = [_ | _] <- get_elements(html),
+         ogp_data =
+           Enum.reduce(elements, data, fn el, acc ->
+             attributes = normalize_attributes(el)
+
+             Map.merge(acc, attributes)
+           end) do
+      {:ok, ogp_data}
+    else
+      _e -> {:error, "No OGP metadata found"}
+    end
+  end
+
+  defp get_elements(html) do
+    html |> Floki.find("meta[property^='og:']")
+  end
+
+  defp normalize_attributes(html_node) do
+    {_tag, attributes, _children} = html_node
+
+    data =
+      Enum.into(attributes, %{}, fn {name, value} ->
+        {name, String.trim_leading(value, "og:")}
+      end)
+
+    %{String.to_atom(data["property"]) => data["content"]}
+  end
+end
diff --git a/lib/pleroma/web/router.ex b/lib/pleroma/web/router.ex
index 1f929ee21..8df45bf4d 100644
--- a/lib/pleroma/web/router.ex
+++ b/lib/pleroma/web/router.ex
@@ -232,6 +232,12 @@ defmodule Pleroma.Web.Router do
     put("/settings", MastodonAPIController, :put_settings)
   end
 
+  scope "/api", Pleroma.Web.RichMedia do
+    pipe_through(:authenticated_api)
+
+    get("/rich_media/parse", RichMediaController, :parse)
+  end
+
   scope "/api/v1", Pleroma.Web.MastodonAPI do
     pipe_through(:api)
     get("/instance", MastodonAPIController, :masto_instance)