RichMedia.Helpers: move the validate_page_url/1 function to the Parser module

This will ensure that the page validation happens in Parser.parse/1 so it can be called from anywhere and still filter invalid URLs.
This commit is contained in:
Mark Felder 2024-02-06 16:54:52 -05:00
parent 6b7b443ff9
commit 9f2319e50d
3 changed files with 45 additions and 44 deletions

View file

View file

@ -18,53 +18,10 @@ defmodule Pleroma.Web.RichMedia.Helpers do
recv_timeout: 2_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_impl.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"}) do
cond do
Linkify.Parser.ip?(host) ->
:error
host in @config_impl.get([:rich_media, :ignore_hosts], []) ->
:error
get_tld(host) in @config_impl.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 def fetch_data_for_object(object) do
with true <- @config_impl.get([:rich_media, :enabled]), with true <- @config_impl.get([:rich_media, :enabled]),
{:ok, page_url} <- {:ok, page_url} <-
HTML.extract_first_external_url_from_object(object), HTML.extract_first_external_url_from_object(object),
:ok <- validate_page_url(page_url),
{:ok, rich_media} <- Parser.parse(page_url) do {:ok, rich_media} <- Parser.parse(page_url) do
%{page_url: page_url, rich_media: rich_media} %{page_url: page_url, rich_media: rich_media}
else else

View file

@ -6,6 +6,7 @@ defmodule Pleroma.Web.RichMedia.Parser do
require Logger require Logger
@cachex Pleroma.Config.get([:cachex, :provider], Cachex) @cachex Pleroma.Config.get([:cachex, :provider], Cachex)
@config_impl Application.compile_env(:pleroma, [__MODULE__, :config_impl], Pleroma.Config)
defp parsers do defp parsers do
Pleroma.Config.get([:rich_media, :parsers]) Pleroma.Config.get([:rich_media, :parsers])
@ -15,7 +16,8 @@ defmodule Pleroma.Web.RichMedia.Parser do
@spec parse(String.t()) :: {:ok, map()} | {:error, any()} @spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url) do def parse(url) do
with {:ok, data} <- get_cached_or_parse(url), with :ok <- validate_page_url(url),
{:ok, data} <- get_cached_or_parse(url),
{:ok, _} <- set_ttl_based_on_image(data, url) do {:ok, _} <- set_ttl_based_on_image(data, url) do
{:ok, data} {:ok, data}
end end
@ -161,4 +163,46 @@ defmodule Pleroma.Web.RichMedia.Parser do
end) end)
|> Map.new() |> Map.new()
end end
@spec validate_page_url(URI.t() | binary()) :: :ok | :error
defp validate_page_url(page_url) when is_binary(page_url) do
validate_tld = @config_impl.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"}) do
cond do
Linkify.Parser.ip?(host) ->
:error
host in @config_impl.get([:rich_media, :ignore_hosts], []) ->
:error
get_tld(host) in @config_impl.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
end end