akkoma/lib/pleroma/web/rich_media/helpers.ex

93 lines
2.3 KiB
Elixir
Raw Normal View History

# Pleroma: A lightweight social networking server
2020-03-03 22:44:49 +00:00
# Copyright _ 2017-2020 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Helpers do
2019-02-09 15:16:26 +00:00
alias Pleroma.Activity
alias Pleroma.Config
2019-02-09 15:16:26 +00:00
alias Pleroma.HTML
alias Pleroma.Object
alias Pleroma.Web.RichMedia.Parser
@options [
pool: :media,
max_body: 2_000_000,
recv_timeout: 2_000
]
2020-06-09 17:49:24 +00:00
@spec validate_page_url(URI.t() | binary()) :: :ok | :error
defp validate_page_url(page_url) when is_binary(page_url) do
validate_tld = Config.get([Pleroma.Formatter, :validate_tld])
2019-06-26 03:24:12 +00:00
page_url
|> Linkify.Parser.url?(validate_tld: validate_tld)
|> parse_uri(page_url)
end
2019-06-18 13:08:18 +00:00
2020-06-09 17:49:24 +00:00
defp validate_page_url(%URI{host: host, scheme: "https", authority: authority})
when is_binary(authority) do
2019-06-25 12:52:53 +00:00
cond do
host in Config.get([:rich_media, :ignore_hosts], []) ->
2019-06-25 12:52:53 +00:00
:error
get_tld(host) in Config.get([:rich_media, :ignore_tld], []) ->
:error
2019-06-25 12:52:53 +00:00
true ->
:ok
end
end
defp validate_page_url(_), do: :error
2019-02-24 19:13:46 +00:00
defp parse_uri(true, url) do
url
|> URI.parse()
|> validate_page_url
end
defp parse_uri(_, _), do: :error
defp get_tld(host) do
host
|> String.split(".")
|> Enum.reverse()
|> hd
end
def fetch_data_for_object(object) do
with true <- Config.get([:rich_media, :enabled]),
false <- object.data["sensitive"] || false,
{:ok, page_url} <-
HTML.extract_first_external_url_from_object(object),
2019-02-24 19:13:46 +00:00
:ok <- validate_page_url(page_url),
{:ok, rich_media} <- Parser.parse(page_url) do
%{page_url: page_url, rich_media: rich_media}
else
_ -> %{}
end
end
def fetch_data_for_activity(%Activity{data: %{"type" => "Create"}} = activity) do
with true <- Config.get([:rich_media, :enabled]),
%Object{} = object <- Object.normalize(activity) do
fetch_data_for_object(object)
else
_ -> %{}
end
end
def fetch_data_for_activity(_), do: %{}
def perform(:fetch, %Activity{} = activity) do
fetch_data_for_activity(activity)
:ok
end
def rich_media_get(url) do
headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
Pleroma.HTTP.get(url, headers, adapter: @options)
end
end