X-Git-Url: https://git.squeep.com/?a=blobdiff_plain;f=lib%2Fpleroma%2Fweb%2Frich_media%2Fhelpers.ex;h=bd7f03cbeba8c3bf55bf86341e47440ae17464d9;hb=f18178cb096b9a00ed12ff0fe36893f118ec6649;hp=9bc8f2559b5d5d6c78174283bee33f156ffd7d4c;hpb=f4e2595592ccca6cedd64669baef7bdd2a6547d0;p=akkoma
diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex
index 9bc8f2559..bd7f03cbe 100644
--- a/lib/pleroma/web/rich_media/helpers.ex
+++ b/lib/pleroma/web/rich_media/helpers.ex
@@ -1,31 +1,65 @@
# Pleroma: A lightweight social networking server
-# Copyright _ 2017-2019 Pleroma Authors
+# Copyright _ 2017-2020 Pleroma Authors
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Helpers do
alias Pleroma.Activity
+ alias Pleroma.Config
alias Pleroma.HTML
alias Pleroma.Object
alias Pleroma.Web.RichMedia.Parser
+ @options [
+ pool: :media,
+ max_body: 2_000_000,
+ 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
- if AutoLinker.Parser.is_url?(page_url, true) do
- URI.parse(page_url) |> validate_page_url
- else
- :error
+ validate_tld = Config.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", authority: authority})
+ when is_binary(authority) do
+ cond do
+ host in Config.get([:rich_media, :ignore_hosts], []) ->
+ :error
+
+ get_tld(host) in Config.get([:rich_media, :ignore_tld], []) ->
+ :error
+
+ true ->
+ :ok
end
end
- defp validate_page_url(%URI{authority: nil}), do: :error
- defp validate_page_url(%URI{scheme: nil}), do: :error
- defp validate_page_url(%URI{}), do: :ok
defp validate_page_url(_), do: :error
- def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
- with true <- Pleroma.Config.get([:rich_media, :enabled]),
- %Object{} = object <- Object.normalize(activity),
+ 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.get([:rich_media, :enabled]),
false <- object.data["sensitive"] || false,
- {:ok, page_url} <- HTML.extract_first_external_url(object, object.data["content"]),
+ {: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}
@@ -34,7 +68,25 @@ defmodule Pleroma.Web.RichMedia.Helpers do
end
end
+ def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
+ with true <- Config.get([:rich_media, :enabled]),
+ %Object{} = object <- Object.normalize(activity) do
+ fetch_data_for_object(object)
+ else
+ _ -> %{}
+ end
+ end
+
def fetch_data_for_activity(_), do: %{}
- def perform(:fetch, %Activity{} = activity), do: fetch_data_for_activity(activity)
+ def perform(:fetch, %Activity{} = activity) do
+ fetch_data_for_activity(activity)
+ :ok
+ end
+
+ def rich_media_get(url) do
+ headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
+
+ Pleroma.HTTP.get(url, headers, @options)
+ end
end