From a77b2987bc90879af30a39ba274df9061cc7fbae Mon Sep 17 00:00:00 2001
From: Nathan Whitaker <17734409+nathanwhit@users.noreply.github.com>
Date: Mon, 11 Mar 2024 15:49:43 -0700
Subject: fix(ext/node): Match punycode module behavior to node (#22847)

Fixes #19214.

We were using the `idna` crate to implement our polyfill for
`punycode.toASCII` and `punycode.toUnicode`. The `idna` crate is
correct, and adheres to the IDNA2003/2008 spec, but it turns out
`node`'s implementations don't really follow any spec! Instead, node
splits the domain by `'.'` and punycode encodes/decodes each part. This
means that node's implementations will happily work on codepoints that
are disallowed by the IDNA specs, causing the error in #19214.

While fixing this, I went ahead and matched the node behavior on all of
the punycode functions and enabled node's punycode test in our
`node_compat` suite.
---
 tests/unit_node/punycode_test.ts | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
 create mode 100644 tests/unit_node/punycode_test.ts

(limited to 'tests/unit_node/punycode_test.ts')

diff --git a/tests/unit_node/punycode_test.ts b/tests/unit_node/punycode_test.ts
new file mode 100644
index 000000000..f5f8c4f97
--- /dev/null
+++ b/tests/unit_node/punycode_test.ts
@@ -0,0 +1,16 @@
+// Copyright 2018-2024 the Deno authors. All rights reserved. MIT license.
+
+import * as punycode from "node:punycode";
+import { assertEquals } from "@std/assert/mod.ts";
+
+Deno.test("regression #19214", () => {
+  const input = "个��.hk";
+
+  assertEquals(punycode.toASCII(input), "xn--ciq6844ba.hk");
+
+  assertEquals(punycode.toUnicode("xn--ciq6844ba.hk"), input);
+});
+
+Deno.test("Decode empty input", () => {
+  assertEquals(punycode.decode(""), "");
+});
-- 
cgit v1.2.3