Fix atom leak in Rich Media Parser

lib/pleroma/web/mastodon_api/views/status_view.ex

@@ -377,8 +377,8 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
     page_url_data = URI.parse(page_url)
 
     page_url_data =
-      if rich_media[:url] != nil do
-        URI.merge(page_url_data, URI.parse(rich_media[:url]))
+      if is_binary(rich_media["url"]) do
+        URI.merge(page_url_data, URI.parse(rich_media["url"]))
       else
         page_url_data
       end
@@ -386,11 +386,9 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
     page_url = page_url_data |> to_string
 
     image_url =
-      if rich_media[:image] != nil do
-        URI.merge(page_url_data, URI.parse(rich_media[:image]))
+      if is_binary(rich_media["image"]) do
+        URI.merge(page_url_data, URI.parse(rich_media["image"]))
         |> to_string
-      else
-        nil
       end
 
     %{
@@ -399,8 +397,8 @@ defmodule Pleroma.Web.MastodonAPI.StatusView do
       provider_url: page_url_data.scheme <> "://" <> page_url_data.host,
       url: page_url,
       image: image_url |> MediaProxy.url(),
-      title: rich_media[:title] || "",
-      description: rich_media[:description] || "",
+      title: rich_media["title"] || "",
+      description: rich_media["description"] || "",
       pleroma: %{
         opengraph: rich_media
       }

lib/pleroma/web/rich_media/helpers.ex

@@ -9,7 +9,7 @@ defmodule Pleroma.Web.RichMedia.Helpers do
   alias Pleroma.Object
   alias Pleroma.Web.RichMedia.Parser
 
-  @spec validate_page_url(any()) :: :ok | :error
+  @spec validate_page_url(URI.t() | binary()) :: :ok | :error
   defp validate_page_url(page_url) when is_binary(page_url) do
     validate_tld = Application.get_env(:auto_linker, :opts)[:validate_tld]
 
@@ -18,8 +18,8 @@ defmodule Pleroma.Web.RichMedia.Helpers do
     |> parse_uri(page_url)
   end
 
-  defp validate_page_url(%URI{host: host, scheme: scheme, authority: authority})
-       when scheme == "https" and not is_nil(authority) do
+  defp validate_page_url(%URI{host: host, scheme: "https", authority: authority})
+       when is_binary(authority) do
     cond do
       host in Config.get([:rich_media, :ignore_hosts], []) ->
         :error

lib/pleroma/web/rich_media/parser.ex

@@ -91,7 +91,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
       html
       |> parse_html()
       |> maybe_parse()
-      |> Map.put(:url, url)
+      |> Map.put("url", url)
       |> clean_parsed_data()
       |> check_parsed_data()
     rescue
@@ -111,8 +111,8 @@ defmodule Pleroma.Web.RichMedia.Parser do
     end)
   end
 
-  defp check_parsed_data(%{title: title} = data)
-       when is_binary(title) and byte_size(title) > 0 do
+  defp check_parsed_data(%{"title" => title} = data)
+       when is_binary(title) and title != "" do
     {:ok, data}
   end
 
@@ -123,11 +123,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
   defp clean_parsed_data(data) do
     data
     |> Enum.reject(fn {key, val} ->
-      with {:ok, _} <- Jason.encode(%{key => val}) do
-        false
-      else
-        _ -> true
-      end
+      not match?({:ok, _}, Jason.encode(%{key => val}))
     end)
     |> Map.new()
   end

lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex

@@ -29,19 +29,19 @@ defmodule Pleroma.Web.RichMedia.Parsers.MetaTagsParser do
     {_tag, attributes, _children} = html_node
 
     data =
-      Enum.into(attributes, %{}, fn {name, value} ->
+      Map.new(attributes, fn {name, value} ->
         {name, String.trim_leading(value, "#{prefix}:")}
       end)
 
-    %{String.to_atom(data[key_name]) => data[value_name]}
+    %{data[key_name] => data[value_name]}
   end
 
-  defp maybe_put_title(%{title: _} = meta, _), do: meta
+  defp maybe_put_title(%{"title" => _} = meta, _), do: meta
 
   defp maybe_put_title(meta, html) when meta != %{} do
     case get_page_title(html) do
       "" -> meta
-      title -> Map.put_new(meta, :title, title)
+      title -> Map.put_new(meta, "title", title)
     end
   end
 

lib/pleroma/web/rich_media/parsers/oembed_parser.ex

@@ -5,7 +5,7 @@
 defmodule Pleroma.Web.RichMedia.Parsers.OEmbed do
   def parse(html, _data) do
     with elements = [_ | _] <- get_discovery_data(html),
-         {:ok, oembed_url} <- get_oembed_url(elements),
+         oembed_url when is_binary(oembed_url) <- get_oembed_url(elements),
          {:ok, oembed_data} <- get_oembed_data(oembed_url) do
       {:ok, oembed_data}
     else
@@ -17,19 +17,13 @@ defmodule Pleroma.Web.RichMedia.Parsers.OEmbed do
     html |> Floki.find("link[type='application/json+oembed']")
   end
 
-  defp get_oembed_url(nodes) do
-    {"link", attributes, _children} = nodes |> hd()
-
-    {:ok, Enum.into(attributes, %{})["href"]}
+  defp get_oembed_url([{"link", attributes, _children} | _]) do
+    Enum.find_value(attributes, fn {k, v} -> if k == "href", do: v end)
   end
 
   defp get_oembed_data(url) do
-    {:ok, %Tesla.Env{body: json}} = Pleroma.HTTP.get(url, [], adapter: [pool: :media])
-
-    {:ok, data} = Jason.decode(json)
-
-    data = data |> Map.new(fn {k, v} -> {String.to_atom(k), v} end)
-
-    {:ok, data}
+    with {:ok, %Tesla.Env{body: json}} <- Pleroma.HTTP.get(url, [], adapter: [pool: :media]) do
+      Jason.decode(json)
+    end
   end
 end