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

76 lines
1.8 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 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
@hackney_options [
pool: :media,
recv_timeout: 2_000,
max_body: 2_000_000,
with_body: true
]
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 20:19:07 +00:00
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 20:26:40 +00:00
defp parse_url(url) do
try do
{:ok, %Tesla.Env{body: html}} = Pleroma.HTTP.get(url, [], adapter: @hackney_options)
html |> maybe_parse() |> clean_parsed_data() |> check_parsed_data()
rescue
2019-01-28 20:19:07 +00:00
e ->
{:error, "Parsing error: #{inspect(e)}"}
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 check_parsed_data(%{title: title} = data) when is_binary(title) and byte_size(title) > 0 do
{:ok, data}
2019-01-02 14:02:50 +00:00
end
defp check_parsed_data(data) do
{:error, "Found metadata was invalid or incomplete: #{inspect(data)}"}
2019-01-02 14:02:50 +00:00
end
defp clean_parsed_data(data) do
data
|> Enum.reject(fn {key, val} ->
with {:ok, _} <- Jason.encode(%{key => val}) do
false
else
_ -> true
end
end)
|> Map.new()
end
2019-01-01 20:26:40 +00:00
end