From 91b54e4a01515416465ce63fce0dc222f4e6d9e5 Mon Sep 17 00:00:00 2001
From: Voyager <voyagerxyx@arrmail.net>
Date: Fri, 21 Aug 2026 09:23:28 -0400
Subject: [PATCH] Fix Unicode chat sanitization

---
 network/network_chat_protocol.gd | 18 ++++++++++++---
 tests/chat_validation.gd         | 39 +++++++++++++++++++++++---------
 2 files changed, 43 insertions(+), 14 deletions(-)

diff --git a/network/network_chat_protocol.gd b/network/network_chat_protocol.gd
index 389260d..2e91432 100644
--- a/network/network_chat_protocol.gd
+++ b/network/network_chat_protocol.gd
@@ -9,15 +9,27 @@ const MAX_HISTORY: int = 100
 const LATE_JOIN_HISTORY: int = 50
 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 }
 
 
 static func sanitize_body(value: Variant) -> String:
 	if typeof(value) != TYPE_STRING:
 		return ""
-	var strippable_whitespace = RegEx.create_from_string("^[^\\w]+|[^\\w]+$")
-	var result := strippable_whitespace.sub(value, "")
-	if result.is_empty() or result.length() > MAX_VISIBLE_CHARACTERS:
+	var result: String = str(value).replace("\r", " ").replace("\n", " ")
+	result = result.replace("\t", " ")
+	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 ""
 	if result.to_utf8_buffer().size() > MAX_UTF8_BYTES:
 		return ""
diff --git a/tests/chat_validation.gd b/tests/chat_validation.gd
index aed8f17..fa0cd8e 100644
--- a/tests/chat_validation.gd
+++ b/tests/chat_validation.gd
@@ -1,21 +1,38 @@
 extends SceneTree
 
-const MainScene: PackedScene = preload("res://main/main.tscn")
-
 func _initialize() -> void:
 	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
-	assert(NetworkChatProtocol.sanitize_body(em_space + "a") == "a")
-	const braille_space := "⠀" # U+2800 BRAILLE PATTERN BLANK
-	assert(NetworkChatProtocol.sanitize_body(braille_space + "a") == "a")
+	const zero_width_space := "​" # U+200B ZERO WIDTH SPACE
+	const braille_blank := "⠀" # U+2800 BRAILLE PATTERN BLANK
+	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")
 	quit(0)