2019-01-01 20:26:40 +00:00
|
|
|
defmodule Pleroma.Web.RichMedia.Parser do
|
2019-01-13 00:06:50 +00:00
|
|
|
@parsers [
|
|
|
|
Pleroma.Web.RichMedia.Parsers.OGP,
|
|
|
|
Pleroma.Web.RichMedia.Parsers.TwitterCard,
|
|
|
|
Pleroma.Web.RichMedia.Parsers.OEmbed
|
|
|
|
]
|
2019-01-01 20:26:40 +00:00
|
|
|
|
2019-01-04 23:50:54 +00:00
|
|
|
if Mix.env() == :test do
|
|
|
|
def parse(url), do: parse_url(url)
|
|
|
|
else
|
|
|
|
def parse(url),
|
2019-01-09 18:35:01 +00:00
|
|
|
do: Cachex.fetch!(:rich_media_cache, url, fn _ -> parse_url(url) end)
|
2019-01-04 23:50:54 +00:00
|
|
|
end
|
2019-01-01 20:26:40 +00:00
|
|
|
|
2019-01-04 23:50:54 +00:00
|
|
|
defp parse_url(url) do
|
|
|
|
{:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url)
|
2019-01-04 23:23:47 +00:00
|
|
|
|
2019-01-04 23:50:54 +00:00
|
|
|
html |> maybe_parse() |> get_parsed_data()
|
2019-01-02 14:02:50 +00:00
|
|
|
end
|
|
|
|
|
|
|
|
defp maybe_parse(html) do
|
|
|
|
Enum.reduce_while(@parsers, %{}, fn parser, acc ->
|
2019-01-01 20:26:40 +00:00
|
|
|
case parser.parse(html, acc) do
|
|
|
|
{:ok, data} -> {:halt, data}
|
|
|
|
{:error, _msg} -> {:cont, acc}
|
|
|
|
end
|
|
|
|
end)
|
|
|
|
end
|
2019-01-02 14:02:50 +00:00
|
|
|
|
|
|
|
defp get_parsed_data(data) when data == %{} do
|
|
|
|
{:error, "No metadata found"}
|
|
|
|
end
|
|
|
|
|
|
|
|
defp get_parsed_data(data) do
|
|
|
|
{:ok, data}
|
|
|
|
end
|
2019-01-01 20:26:40 +00:00
|
|
|
end
|