Skip to content

利用 AI 大模型和自动化工作流,根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.

fix(skill): decode piped uv output as UTF-8 in run_checked

c02062778641474c0eb
diff --git a/docs/skill/mpt_agent.py b/docs/skill/mpt_agent.py
index 4f65fd84..ac05ae57 100644
--- a/docs/skill/mpt_agent.py
+++ b/docs/skill/mpt_agent.py
@@ -585,6 +585,7 @@ def run_checked(command: list[str], *, cwd: Path) -> None:
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
text=True,
+ encoding="utf-8",
errors="replace",
check=False,
)
diff --git a/test/services/test_mpt_agent_skill.py b/test/services/test_mpt_agent_skill.py
index a4be796f..5df548bb 100644
--- a/test/services/test_mpt_agent_skill.py
+++ b/test/services/test_mpt_agent_skill.py
@@ -699,6 +699,28 @@ class TestMptAgentSkill(unittest.TestCase):
self.assertNotIn("package-a", stdout.getvalue())
self.assertNotIn("package-a", stderr.getvalue())
+ def test_run_checked_decodes_piped_output_as_utf8(self):
+ # run_checked pipes uv's output and re-decodes it with text=True. Without
+ # an explicit encoding, that decode follows the host locale (cp936/cp1252),
+ # so non-ASCII dependency errors reach the tail as mojibake. PR #1365 pinned
+ # every other PIPE-decoding subprocess.run to UTF-8; derive the requirement
+ # from the call so a future regression fails here instead of in the field.
+ with patch.object(
+ mpt_agent.subprocess,
+ "run",
+ return_value=SimpleNamespace(returncode=0, stdout=""),
+ ) as run_mock:
+ mpt_agent.run_checked(["uv", "sync", "--frozen"], cwd=Path.cwd())
+
+ kwargs = run_mock.call_args.kwargs
+ self.assertEqual(kwargs.get("stdout"), mpt_agent.subprocess.PIPE)
+ self.assertTrue(kwargs.get("text"))
+ self.assertEqual(
+ kwargs.get("encoding"),
+ "utf-8",
+ "piped output must be decoded as UTF-8, not the host locale",
+ )
+
def test_explicit_voice_is_not_overridden(self):
self.assertTrue(
mpt_agent.has_cli_option(