woofmeow/straywild
woofmeow/straywild
3
5
Fork 2
Code Issues 37 Projects 1 Releases 43 Packages Wiki Activity Actions

Fix Unicode chat sanitization

Browse source
This commit is contained in:
Alexander Sellite 2026-08-21 09:23:28 -04:00
parent c20bb35929
commit 91b54e4a01
2 changed files with 43 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

18
network/network_chat_protocol.gd
View file

@ -9,15 +9,27 @@ const MAX_HISTORY: int = 100
const LATE_JOIN_HISTORY: int = 50 const LATE_JOIN_HISTORY: int = 50
const MAX_ID_LENGTH: int = 96 const MAX_ID_LENGTH: int = 96
static var _invisible_edge_pattern: RegEx = RegEx.create_from_string(
"^[\\s\\p{Z}\\p{Cf}\\x{2800}]+|[\\s\\p{Z}\\p{Cf}\\x{2800}]+$"
)
static var _visible_content_pattern: RegEx = RegEx.create_from_string(
"[^\\s\\p{Z}\\p{Cf}\\x{2800}]"
)
enum Kind { PLAYER, SYSTEM } enum Kind { PLAYER, SYSTEM }
static func sanitize_body(value: Variant) -> String: static func sanitize_body(value: Variant) -> String:
if typeof(value) != TYPE_STRING: if typeof(value) != TYPE_STRING:
return "" return ""
var strippable_whitespace = RegEx.create_from_string("^[^\\w]+|[^\\w]+$") var result: String = str(value).replace("\r", " ").replace("\n", " ")
var result := strippable_whitespace.sub(value, "") result = result.replace("\t", " ")
if result.is_empty() or result.length() > MAX_VISIBLE_CHARACTERS: result = _invisible_edge_pattern.sub(result, "", true)
if (
result.is_empty()
or _visible_content_pattern.search(result) == null
or result.length() > MAX_VISIBLE_CHARACTERS
):
return "" return ""
if result.to_utf8_buffer().size() > MAX_UTF8_BYTES: if result.to_utf8_buffer().size() > MAX_UTF8_BYTES:
return "" return ""

39
tests/chat_validation.gd
View file

@ -1,21 +1,38 @@
extends SceneTree extends SceneTree
const MainScene: PackedScene = preload("res://main/main.tscn")
func _initialize() -> void: func _initialize() -> void:
call_deferred("_run") call_deferred("_run")
func _run() -> void:
## Regular whitespace should be stripped
assert(NetworkChatProtocol.sanitize_body(" a") == "a", "Leading space not trimmed")
assert(NetworkChatProtocol.sanitize_body("b ") == "b", "Trailing space not trimmed")
assert(NetworkChatProtocol.sanitize_body("a b") == "a b", "Innocent whitespace culled")
## Some examples of Unicode whitespace that should be stripped func _run() -> void:
assert(NetworkChatProtocol.sanitize_body(42).is_empty())
assert(NetworkChatProtocol.sanitize_body(" a ") == "a")
assert(NetworkChatProtocol.sanitize_body("a b") == "a b")
assert(NetworkChatProtocol.sanitize_body("a\tb\nc") == "a b c")
assert(NetworkChatProtocol.sanitize_body(" !hello!? ") == "!hello!?")
assert(NetworkChatProtocol.sanitize_body("!!!") == "!!!")
const em_space := " " # U+2003 EM SPACE const em_space := " " # U+2003 EM SPACE
assert(NetworkChatProtocol.sanitize_body(em_space + "a") == "a") const zero_width_space := "​" # U+200B ZERO WIDTH SPACE
const braille_space := "⠀" # U+2800 BRAILLE PATTERN BLANK const braille_blank := "⠀" # U+2800 BRAILLE PATTERN BLANK
assert(NetworkChatProtocol.sanitize_body(braille_space + "a") == "a") const invisible_padding := em_space + zero_width_space + braille_blank
assert(
NetworkChatProtocol.sanitize_body(invisible_padding + "a" + invisible_padding)
== "a"
)
assert(NetworkChatProtocol.sanitize_body(invisible_padding).is_empty())
assert(NetworkChatProtocol.sanitize_body(" \t\n" + invisible_padding).is_empty())
assert(
NetworkChatProtocol.sanitize_body("a".repeat(
NetworkChatProtocol.MAX_VISIBLE_CHARACTERS
)).length() == NetworkChatProtocol.MAX_VISIBLE_CHARACTERS
)
assert(
NetworkChatProtocol.sanitize_body("a".repeat(
NetworkChatProtocol.MAX_VISIBLE_CHARACTERS + 1
)).is_empty()
)
print("Chat validation: PASS") print("Chat validation: PASS")
quit(0) quit(0)