From 42fce4714f1c3e0e01e505ae8c3f3bd5282472c6 Mon Sep 17 00:00:00 2001
From: 0xacee <0xacee@users.noreply.github.com>
Date: Mon, 21 Sep 2026 16:03:24 +0000
Subject: [PATCH] check-yaml: report invalid UTF-8 instead of raising a
 traceback

---
 pre_commit_hooks/check_yaml.py | 3 +++
 tests/check_yaml_test.py       | 6 ++++++
 2 files changed, 9 insertions(+)

diff --git a/pre_commit_hooks/check_yaml.py b/pre_commit_hooks/check_yaml.py
index c94ea716..6524971f 100644
--- a/pre_commit_hooks/check_yaml.py
+++ b/pre_commit_hooks/check_yaml.py
@@ -65,6 +65,9 @@ def main(argv: Sequence[str] | None = None) -> int:
         except ruamel.yaml.YAMLError as exc:
             print(exc)
             retval = 1
+        except UnicodeDecodeError as exc:
+            print(f'{filename}: invalid UTF-8: {exc}')
+            retval = 1
     return retval
 
 
diff --git a/tests/check_yaml_test.py b/tests/check_yaml_test.py
index 54eb16e8..e59627b5 100644
--- a/tests/check_yaml_test.py
+++ b/tests/check_yaml_test.py
@@ -17,6 +17,12 @@ def test_main(filename, expected_retval):
     assert ret == expected_retval
 
 
+def test_main_non_utf8_file(tmpdir):
+    f = tmpdir.join('latin1.yaml')
+    f.write_binary(b'key: caf\xe9\n')
+    assert main((str(f),))
+
+
 def test_main_allow_multiple_documents(tmpdir):
     f = tmpdir.join('test.yaml')
     f.write('---\nfoo\n---\nbar\n')