Make URI encoding query quirks host-aware

This commit is contained in:
Phantasm 2025-10-21 00:50:08 +02:00
commit 73b337245b
No known key found for this signature in database
GPG key ID: 2669E588BCC634C8
2 changed files with 17 additions and 12 deletions

View file

@ -41,7 +41,7 @@ defmodule Pleroma.Utils.URIEncoding do
|> then(fn parsed -> |> then(fn parsed ->
path = encode_path(parsed.path, bypass_decode) path = encode_path(parsed.path, bypass_decode)
query = encode_query(parsed.query) query = encode_query(parsed.query, parsed.host)
%{parsed | path: path, query: query} %{parsed | path: path, query: query}
end) end)
@ -81,25 +81,27 @@ defmodule Pleroma.Utils.URIEncoding do
end) end)
end end
defp encode_query(nil), do: nil
# Order of kv pairs in query is not preserved when using URI.decode_query. # Order of kv pairs in query is not preserved when using URI.decode_query.
# URI.query_decoder/2 returns a stream which so far appears to not change order. # URI.query_decoder/2 returns a stream which so far appears to not change order.
# Immediately switch to a list to prevent breakage for sites that expect # Immediately switch to a list to prevent breakage for sites that expect
# the order of query keys to be always the same. # the order of query keys to be always the same.
defp encode_query(query) when is_binary(query) do defp encode_query(query, host) when is_binary(query) do
query query
|> URI.query_decoder() |> URI.query_decoder()
|> Enum.to_list() |> Enum.to_list()
|> do_encode_query() |> do_encode_query(host)
end end
# Always uses www_form encoding defp encode_query(nil, _), do: nil
defp do_encode_query(enumerable) do
Enum.map_join(enumerable, "&", &maybe_apply_query_quirk(&1)) # Always uses www_form encoding.
# Taken from Elixir's URI module.
defp do_encode_query(enumerable, host) do
Enum.map_join(enumerable, "&", &maybe_apply_query_quirk(&1, host))
end end
defp maybe_apply_query_quirk({key, value}) do # https://git.pleroma.social/pleroma/pleroma/-/issues/1055
defp maybe_apply_query_quirk({key, value}, "i.guim.co.uk" = _host) do
case key do case key do
"precrop" -> "precrop" ->
query_encode_kv_pair({key, value}, ~c":,") query_encode_kv_pair({key, value}, ~c":,")
@ -109,6 +111,9 @@ defmodule Pleroma.Utils.URIEncoding do
end end
end end
defp maybe_apply_query_quirk({key, value}, _), do: query_encode_kv_pair({key, value})
# Taken from Elixir's URI module and modified to support quirks.
defp query_encode_kv_pair({key, value}, rules \\ []) when is_list(rules) do defp query_encode_kv_pair({key, value}, rules \\ []) when is_list(rules) do
cond do cond do
length(rules) > 0 -> length(rules) > 0 ->

View file

@ -51,7 +51,7 @@ defmodule Pleroma.HTTPTest do
%{ %{
method: :get, method: :get,
url: url:
"https://example.com/emoji/Pack%201/koronebless.png?precrop=40:21,overlay-x0,overlay-y0&foo=bar+baz" "https://i.guim.co.uk/emoji/Pack%201/koronebless.png?precrop=40:21,overlay-x0,overlay-y0&foo=bar+baz"
} -> } ->
%Tesla.Env{status: 200, body: "Space in query with Guardian quirk"} %Tesla.Env{status: 200, body: "Space in query with Guardian quirk"}
@ -171,10 +171,10 @@ defmodule Pleroma.HTTPTest do
clear_config(:test_url_encoding, true) clear_config(:test_url_encoding, true)
url = url =
"https://example.com/emoji/Pack 1/koronebless.png?precrop=40:21,overlay-x0,overlay-y0&foo=bar baz" "https://i.guim.co.uk/emoji/Pack 1/koronebless.png?precrop=40:21,overlay-x0,overlay-y0&foo=bar baz"
properly_encoded_url = properly_encoded_url =
"https://example.com/emoji/Pack%201/koronebless.png?precrop=40:21,overlay-x0,overlay-y0&foo=bar+baz" "https://i.guim.co.uk/emoji/Pack%201/koronebless.png?precrop=40:21,overlay-x0,overlay-y0&foo=bar+baz"
result = URIEncoding.encode_url(url) result = URIEncoding.encode_url(url)