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.
---
 ext/node/polyfills/internal/idna.ts | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

(limited to 'ext/node/polyfills/internal')

diff --git a/ext/node/polyfills/internal/idna.ts b/ext/node/polyfills/internal/idna.ts
index 6484fe951..93ed065cc 100644
--- a/ext/node/polyfills/internal/idna.ts
+++ b/ext/node/polyfills/internal/idna.ts
@@ -51,6 +51,11 @@
 
 "use strict";
 
+import {
+  op_node_idna_domain_to_ascii,
+  op_node_idna_domain_to_unicode,
+} from "ext:core/ops";
+
 /**
  * Creates an array containing the numeric code points of each Unicode
  * character in the string. While JavaScript uses UCS-2 internally,
@@ -105,3 +110,17 @@ export const ucs2 = {
   decode: ucs2decode,
   encode: ucs2encode,
 };
+
+/**
+ *  Converts a domain to ASCII as per the IDNA spec
+ */
+export function domainToASCII(domain: string) {
+  return op_node_idna_domain_to_ascii(domain);
+}
+
+/**
+ *  Converts a domain to Unicode as per the IDNA spec
+ */
+export function domainToUnicode(domain: string) {
+  return op_node_idna_domain_to_unicode(domain);
+}
-- 
cgit v1.2.3