From 24dc1b7b484bc0f00f5543e4b2ff7c15a3f5deb1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sebasti=C3=A1n=20Lucas?= <sebalucas@gmail.com>
Date: Sat, 8 Aug 2026 17:03:09 -0300
Subject: [PATCH] Anchor the utf-8-variants regexes at the true end of the
 buffer

Python's $ also matches before a trailing newline, so a \n at the end of
the input was consumed as part of a CESU-8 sequence or a Java null. Use \Z,
which only matches at the very end, keeping the intent for truncated input
in the streaming decoder.

Fixes #236
---
 ftfy/bad_codecs/utf8_variants.py | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/ftfy/bad_codecs/utf8_variants.py b/ftfy/bad_codecs/utf8_variants.py
index eaac3c1..f1c0f3f 100644
--- a/ftfy/bad_codecs/utf8_variants.py
+++ b/ftfy/bad_codecs/utf8_variants.py
@@ -58,11 +58,11 @@
 CESU8_EXPR = (
     b"("
     b"\xed"
-    b"([\xa0-\xaf]|$)"
-    b"([\x80-\xbf]|$)"
-    b"(\xed|$)"
-    b"([\xb0-\xbf]|$)"
-    b"([\x80-\xbf]|$)"
+    b"([\xa0-\xaf]|\\Z)"
+    b"([\x80-\xbf]|\\Z)"
+    b"(\xed|\\Z)"
+    b"([\xb0-\xbf]|\\Z)"
+    b"([\x80-\xbf]|\\Z)"
     b")"
 )
 
@@ -70,11 +70,11 @@
 
 # This expression matches isolated surrogate characters that aren't
 # CESU-8, which have to be handled carefully on Python 2.
-SURROGATE_EXPR = b"(\xed([\xa0-\xbf]|$)([\x80-\xbf]|$))"
+SURROGATE_EXPR = b"(\xed([\xa0-\xbf]|\\Z)([\x80-\xbf]|\\Z))"
 
 # This expression matches the Java encoding of U+0, including if it's
 # truncated and we need more bytes.
-NULL_EXPR = b"(\xc0(\x80|$))"
+NULL_EXPR = b"(\xc0(\x80|\\Z))"
 
 # This regex matches cases that we need to decode differently from
 # standard UTF-8.