# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parser do
- @hackney_options [
- pool: :media,
- recv_timeout: 2_000,
- max_body: 2_000_000,
- with_body: true
- ]
-
defp parsers do
Pleroma.Config.get([:rich_media, :parsers])
end
defp parse_url(url) do
try do
- {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options)
+ {:ok, %Tesla.Env{body: html}} = Pleroma.Web.RichMedia.Helpers.rich_media_get(url)
html
|> parse_html()
|> maybe_parse()
- |> Map.put(:url, url)
+ |> Map.put("url", url)
|> clean_parsed_data()
|> check_parsed_data()
rescue
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}
+ data when data != %{} -> {:halt, data}
+ _ -> {:cont, acc}
end
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
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