pleroma/lib/pleroma/web/rich_media/parser.ex

57 lines
1.3 KiB
Elixir
Raw Normal View History

# Pleroma: A lightweight social networking server
# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
2019-01-01 21:26:40 +01:00
defmodule Pleroma.Web.RichMedia.Parser do
2019-01-13 01:06:50 +01:00
@parsers [
Pleroma.Web.RichMedia.Parsers.OGP,
Pleroma.Web.RichMedia.Parsers.TwitterCard,
Pleroma.Web.RichMedia.Parsers.OEmbed
]
2019-01-01 21:26:40 +01:00
def parse(nil), do: {:error, "No URL provided"}
if Mix.env() == :test do
def parse(url), do: parse_url(url)
else
def parse(url) do
2019-01-28 21:19:07 +01:00
try do
Cachex.fetch!(:rich_media_cache, url, fn _ ->
{:commit, parse_url(url)}
end)
rescue
e ->
{:error, "Cachex error: #{inspect(e)}"}
end
end
end
2019-01-01 21:26:40 +01:00
defp parse_url(url) do
try do
{:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url)
html |> maybe_parse() |> get_parsed_data()
rescue
2019-01-28 21:19:07 +01:00
e ->
{:error, "Parsing error: #{inspect(e)}"}
end
2019-01-02 15:02:50 +01:00
end
defp maybe_parse(html) do
Enum.reduce_while(@parsers, %{}, fn parser, acc ->
2019-01-01 21:26:40 +01:00
case parser.parse(html, acc) do
{:ok, data} -> {:halt, data}
{:error, _msg} -> {:cont, acc}
end
end)
end
2019-01-02 15:02:50 +01:00
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 21:26:40 +01:00
end