Page MenuHomePhorge
Files F85712908

No OneTemporary
Actions

Size
5 KB
Referenced Files
None
Subscribers
None

View Options

diff --git a/lib/pleroma/web/rich_media/helpers.ex b/lib/pleroma/web/rich_media/helpers.ex
index 4d1c2dc9a..566fc8c8a 100644
--- a/lib/pleroma/web/rich_media/helpers.ex
+++ b/lib/pleroma/web/rich_media/helpers.ex
@@ -1,137 +1,130 @@
# Pleroma: A lightweight social networking server
# Copyright _ 2017-2020 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Helpers do
alias Pleroma.Activity
alias Pleroma.Config
alias Pleroma.HTML
alias Pleroma.Object
alias Pleroma.Web.RichMedia.Parser
@options [
pool: :media,
max_body: 2_000_000,
recv_timeout: 2_000
]
@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])
page_url
|> Linkify.Parser.url?(validate_tld: validate_tld)
|> parse_uri(page_url)
end
defp validate_page_url(%URI{host: host, scheme: "https", authority: authority})
when is_binary(authority) do
cond do
host in Config.get([:rich_media, :ignore_hosts], []) ->
:error
get_tld(host) in Config.get([:rich_media, :ignore_tld], []) ->
:error
true ->
:ok
end
end
defp validate_page_url(_), do: :error
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]),
{:ok, page_url} <-
HTML.extract_first_external_url_from_object(object),
: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, fetch: false) do
fetch_data_for_object(object)
else
_ -> %{}
end
end
def fetch_data_for_activity(_), do: %{}
- # YouTube's oEmbed implementation is broken, requiring this hack.
- # https://github.com/oscarotero/Embed/issues/417#issuecomment-746673027
- def rich_media_get("http://www.youtube.com/oembed?" <> params) do
- # Use HTTPS explicitly, even though YouTube returns HTTP
- rich_media_get("https://www.youtube.com/oembed?#{params}")
- end
-
def rich_media_get(url) do
headers = [{"user-agent", Pleroma.Application.user_agent() <> "; Bot"}]
head_check =
case Pleroma.HTTP.head(url, headers, @options) do
# If the HEAD request didn't reach the server for whatever reason,
# we assume the GET that comes right after won't either
{:error, _} = e ->
e
{:ok, %Tesla.Env{status: 200, headers: headers}} ->
with :ok <- check_content_type(headers),
:ok <- check_content_length(headers),
do: :ok
_ ->
:ok
end
with :ok <- head_check, do: Pleroma.HTTP.get(url, headers, @options)
end
defp check_content_type(headers) do
case List.keyfind(headers, "content-type", 0) do
{_, content_type} ->
case Plug.Conn.Utils.media_type(content_type) do
{:ok, "text", "html", _} -> :ok
_ -> {:error, {:content_type, content_type}}
end
_ ->
:ok
end
end
@max_body @options[:max_body]
defp check_content_length(headers) do
case List.keyfind(headers, "content-length", 0) do
{_, maybe_content_length} ->
case Integer.parse(maybe_content_length) do
{content_length, ""} when content_length <= @max_body -> :ok
{_, ""} -> {:error, :body_too_large}
_ -> :ok
end
_ ->
:ok
end
end
end
diff --git a/lib/pleroma/web/rich_media/parsers/o_embed.ex b/lib/pleroma/web/rich_media/parsers/o_embed.ex
index 09eabec56..24fca56bb 100644
--- a/lib/pleroma/web/rich_media/parsers/o_embed.ex
+++ b/lib/pleroma/web/rich_media/parsers/o_embed.ex
@@ -1,29 +1,36 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2021 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parsers.OEmbed do
def parse(html, _data) do
with elements = [_ | _] <- get_discovery_data(html),
oembed_url when is_binary(oembed_url) <- get_oembed_url(elements),
{:ok, oembed_data} <- get_oembed_data(oembed_url) do
oembed_data
else
_e -> %{}
end
end
defp get_discovery_data(html) do
html |> Floki.find("link[type='application/json+oembed']")
end
defp get_oembed_url([{"link", attributes, _children} | _]) do
Enum.find_value(attributes, fn {k, v} -> if k == "href", do: v end)
end
+ # YouTube's oEmbed implementation is broken, requiring this hack.
+ # https://github.com/oscarotero/Embed/issues/417#issuecomment-746673027
+ defp get_oembed_data("http://www.youtube.com/oembed?" <> params) do
+ # Use HTTPS explicitly, even though YouTube returns HTTP
+ get_oembed_data("https://www.youtube.com/oembed?#{params}")
+ end
+
defp get_oembed_data(url) do
with {:ok, %Tesla.Env{body: json}} <- Pleroma.Web.RichMedia.Helpers.rich_media_get(url) do
Jason.decode(json)
end
end
end

File Metadata

Mime Type
text/x-diff
Expires
Sat, Sep 19, 1:31 PM (20 h, 40 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1769522
Default Alt Text
(5 KB)

Event Timeline