From 91c925d86a215a742b93ffaa7663c1457f44cbb5 Mon Sep 17 00:00:00 2001
From: Nick Frasser <1693461+nfrasser@users.noreply.github.com>
Date: Tue, 3 Dec 2024 22:49:17 -0500
Subject: [PATCH] Treat object replacement character as whitespace

Fixes #495
---
 packages/linkifyjs/src/scanner.mjs  | 3 +++
 test/spec/linkifyjs/parser.test.mjs | 5 +++++
 2 files changed, 8 insertions(+)

diff --git a/packages/linkifyjs/src/scanner.mjs b/packages/linkifyjs/src/scanner.mjs
index b54fb18..4ebc285 100644
--- a/packages/linkifyjs/src/scanner.mjs
+++ b/packages/linkifyjs/src/scanner.mjs
@@ -13,6 +13,7 @@ import assign from './assign.mjs';
 const NL = '\n'; // New line character
 const EMOJI_VARIATION = '\ufe0f'; // Variation selector, follows heart and others
 const EMOJI_JOINER = '\u200d'; // zero-width joiner
+const OBJECT_REPLACEMENT = '\ufffc'; // whitespace placeholder that sometimes appears in rich text editors
 
 let tlds = null,
 	utlds = null; // don't change so only have to be computed once
@@ -112,9 +113,11 @@ export function init(customSchemes = []) {
 	// Tokens of only non-newline whitespace are arbitrarily long
 	// If any whitespace except newline, more whitespace!
 	const Ws = tr(Start, re.SPACE, tk.WS, { [fsm.whitespace]: true });
+	tt(Start, OBJECT_REPLACEMENT, Ws);
 	tt(Start, NL, tk.NL, { [fsm.whitespace]: true });
 	tt(Ws, NL); // non-accepting state to avoid mixing whitespaces
 	tr(Ws, re.SPACE, Ws);
+	tt(Ws, OBJECT_REPLACEMENT, Ws);
 
 	// Emoji tokens. They are not grouped by the scanner except in cases where a
 	// zero-width joiner is present
diff --git a/test/spec/linkifyjs/parser.test.mjs b/test/spec/linkifyjs/parser.test.mjs
index ce6f118..6fefb43 100644
--- a/test/spec/linkifyjs/parser.test.mjs
+++ b/test/spec/linkifyjs/parser.test.mjs
@@ -313,6 +313,11 @@ const tests = [
 		[Text, Url, Text, Url, Text],
 		['Link 1『', 'http://foo.com/blah_blah', '』 Link 2『', 'http://foo.com/blah_blah_(wikipedia)_(again)', '』'],
 	],
+	[
+		'https://google.com\ufffcthis', // object replacement character
+		[Url, Text],
+		['https://google.com', '\ufffcthis'],
+	],
 ];
 
 describe('linkifyjs/parser#run()', () => {