From ac6c4d98a8de540105d9906a8bb57c7780953509 Mon Sep 17 00:00:00 2001
From: Lewis Gaul <lewis.gaul@gmail.com>
Date: Wed, 10 Jul 2024 19:44:42 +0100
Subject: [PATCH] Replace unrecognised unicode characters rather than raising
 an exception

---
 python_on_whales/utils.py                           |  4 ++--
 tests/python_on_whales/components/test_container.py | 13 +++++++++++++
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/python_on_whales/utils.py b/python_on_whales/utils.py
index e74585e6..50372526 100644
--- a/python_on_whales/utils.py
+++ b/python_on_whales/utils.py
@@ -237,10 +237,10 @@ def run(
         return post_process_stream(completed_process.stdout)
 
 
-def post_process_stream(stream: Optional[bytes]):
+def post_process_stream(stream: Optional[bytes]) -> str:
     if stream is None:
         return ""
-    stream = stream.decode()
+    stream = stream.decode(errors="replace")
     if len(stream) != 0 and stream[-1] == "\n":
         stream = stream[:-1]
     return stream
diff --git a/tests/python_on_whales/components/test_container.py b/tests/python_on_whales/components/test_container.py
index ef7978cb..1f16f4fb 100644
--- a/tests/python_on_whales/components/test_container.py
+++ b/tests/python_on_whales/components/test_container.py
@@ -1437,3 +1437,16 @@ def test_run_always_pull_existent(
     assert remote_id != local_id
     docker_client.container.run(test_image_name, pull="always")
     assert docker_client.image.inspect(test_image_name).id == remote_id
+
+
+@pytest.mark.parametrize("ctr_client", ["docker", "podman"], indirect=True)
+def test_non_unicode_output(ctr_client: DockerClient):
+    """Non-unicode characters in container output should not lead to an exception."""
+    latin_char = "þ".encode("latin")
+    with pytest.raises(UnicodeDecodeError):
+        latin_char.decode(encoding="utf-8")
+    byte_repr = r"\x{:x}".format(ord(latin_char))
+    output = ctr_client.container.run(
+        "ubuntu", ["bash", "-c", f"echo -n $'{byte_repr}'"], remove=True
+    )
+    assert output == "�"