1 # Pleroma: A lightweight social networking server
2 # Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
3 # SPDX-License-Identifier: AGPL-3.0-only
5 defmodule Pleroma.Web.RichMedia.Parser do
7 Pleroma.Web.RichMedia.Parsers.OGP,
8 Pleroma.Web.RichMedia.Parsers.TwitterCard,
9 Pleroma.Web.RichMedia.Parsers.OEmbed
18 def parse(nil), do: {:error, "No URL provided"}
20 if Mix.env() == :test do
21 def parse(url), do: parse_url(url)
25 Cachex.fetch!(:rich_media_cache, url, fn _ ->
26 {:commit, parse_url(url)}
30 {:error, "Cachex error: #{inspect(e)}"}
35 defp parse_url(url) do
37 {:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options)
39 html |> maybe_parse() |> clean_parsed_data() |> check_parsed_data()
42 {:error, "Parsing error: #{inspect(e)}"}
46 defp maybe_parse(html) do
47 Enum.reduce_while(@parsers, %{}, fn parser, acc ->
48 case parser.parse(html, acc) do
49 {:ok, data} -> {:halt, data}
50 {:error, _msg} -> {:cont, acc}
55 defp check_parsed_data(%{title: title} = data) when is_binary(title) and byte_size(title) > 0 do
59 defp check_parsed_data(data) do
60 {:error, "Found metadata was invalid or incomplete: #{inspect(data)}"}
63 defp clean_parsed_data(data) do
65 |> Enum.reject(fn {key, val} ->
66 with {:ok, _} <- Jason.encode(%{key => val}) do