From 9f2319e50dc0516bde4bfa3b117ec4792e553bd2 Mon Sep 17 00:00:00 2001
From: Mark Felder <feld@feld.me>
Date: Tue, 6 Feb 2024 16:54:52 -0500
Subject: [PATCH] RichMedia.Helpers: move the validate_page_url/1 function to
 the Parser module

This will ensure that the page validation happens in Parser.parse/1 so it can be called from anywhere and still filter invalid URLs.
---
 changelog.d/rich_media_tests.skip     |  0
 lib/pleroma/web/rich_media/helpers.ex | 43 -------------------------
 lib/pleroma/web/rich_media/parser.ex  | 46 ++++++++++++++++++++++++++-
 3 files changed, 45 insertions(+), 44 deletions(-)
 create mode 100644 changelog.d/rich_media_tests.skip

diff --git a/changelog.d/rich_media_tests.skip b/changelog.d/rich_media_tests.skip
new file mode 100644
index 000000000..e69de29bb
diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex
index 1501776d9..a711dc436 100644
--- a/lib/pleroma/web/rich_media/helpers.ex
+++ b/lib/pleroma/web/rich_media/helpers.ex
@@ -18,53 +18,10 @@ defmodule Pleroma.Web.RichMedia.Helpers do
     recv_timeout: 2_000
   ]
 
-  @spec validate_page_url(URI.t() | binary()) :: :ok | :error
-  defp validate_page_url(page_url) when is_binary(page_url) do
-    validate_tld = @config_impl.get([Pleroma.Formatter, :validate_tld])
-
-    page_url
-    |> Linkify.Parser.url?(validate_tld: validate_tld)
-    |> parse_uri(page_url)
-  end
-
-  defp validate_page_url(%URI{host: host, scheme: "https"}) do
-    cond do
-      Linkify.Parser.ip?(host) ->
-        :error
-
-      host in @config_impl.get([:rich_media, :ignore_hosts], []) ->
-        :error
-
-      get_tld(host) in @config_impl.get([:rich_media, :ignore_tld], []) ->
-        :error
-
-      true ->
-        :ok
-    end
-  end
-
-  defp validate_page_url(_), do: :error
-
-  defp parse_uri(true, url) do
-    url
-    |> URI.parse()
-    |> validate_page_url
-  end
-
-  defp parse_uri(_, _), do: :error
-
-  defp get_tld(host) do
-    host
-    |> String.split(".")
-    |> Enum.reverse()
-    |> hd
-  end
-
   def fetch_data_for_object(object) do
     with true <- @config_impl.get([:rich_media, :enabled]),
          {:ok, page_url} <-
            HTML.extract_first_external_url_from_object(object),
-         :ok <- validate_page_url(page_url),
          {:ok, rich_media} <- Parser.parse(page_url) do
       %{page_url: page_url, rich_media: rich_media}
     else
diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index a791b2bab..a73fbc4b9 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -6,6 +6,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
   require Logger
 
   @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
+  @config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
 
   defp parsers do
     Pleroma.Config.get([:rich_media, :parsers])
@@ -15,7 +16,8 @@ defmodule Pleroma.Web.RichMedia.Parser do
 
   @spec parse(String.t()) :: {:ok, map()} | {:error, any()}
   def parse(url) do
-    with {:ok, data} <- get_cached_or_parse(url),
+    with :ok <- validate_page_url(url),
+         {:ok, data} <- get_cached_or_parse(url),
          {:ok, _} <- set_ttl_based_on_image(data, url) do
       {:ok, data}
     end
@@ -161,4 +163,46 @@ defmodule Pleroma.Web.RichMedia.Parser do
     end)
     |> Map.new()
   end
+
+  @spec validate_page_url(URI.t() | binary()) :: :ok | :error
+  defp validate_page_url(page_url) when is_binary(page_url) do
+    validate_tld = @config_impl.get([Pleroma.Formatter, :validate_tld])
+
+    page_url
+    |> Linkify.Parser.url?(validate_tld: validate_tld)
+    |> parse_uri(page_url)
+  end
+
+  defp validate_page_url(%URI{host: host, scheme: "https"}) do
+    cond do
+      Linkify.Parser.ip?(host) ->
+        :error
+
+      host in @config_impl.get([:rich_media, :ignore_hosts], []) ->
+        :error
+
+      get_tld(host) in @config_impl.get([:rich_media, :ignore_tld], []) ->
+        :error
+
+      true ->
+        :ok
+    end
+  end
+
+  defp validate_page_url(_), do: :error
+
+  defp parse_uri(true, url) do
+    url
+    |> URI.parse()
+    |> validate_page_url
+  end
+
+  defp parse_uri(_, _), do: :error
+
+  defp get_tld(host) do
+    host
+    |> String.split(".")
+    |> Enum.reverse()
+    |> hd
+  end
 end