2019-01-01 20:26:40 +00:00
|
|
|
defmodule Pleroma.Web.RichMedia.Parser do
|
|
|
|
@parsers [Pleroma.Web.RichMedia.Parsers.OGP]
|
|
|
|
|
|
|
|
def parse(url) do
|
2019-01-04 23:23:47 +00:00
|
|
|
Cachex.fetch!(:rich_media_cache, url, fn _ ->
|
|
|
|
{:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url)
|
2019-01-01 20:26:40 +00:00
|
|
|
|
2019-01-04 23:23:47 +00:00
|
|
|
result = html |> maybe_parse() |> get_parsed_data()
|
|
|
|
|
|
|
|
{:commit, result}
|
|
|
|
end)
|
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
|