π Geometric Computer Vision Library for Spatial AI
Python artificial-intelligencecomputer-visiondeep-learninghacktoberfestimage-processingmachine-learningneural-networkpythonpytorchroboticsspatial-ai
# LICENSE HEADER MANAGED BY add-license-header
#
# Copyright 2018 Kornia Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
import subprocess
import sys
import textwrap
def test_import_without_jit_script_deprecation():
"""Importing kornia must not eagerly call ``torch.jit.script``.
``torch.jit.script`` is deprecated (and unsupported on Python 3.14+). Decorating
functions with ``@torch.jit.script`` runs the deprecated path at import time, which
makes ``python -W error -c "import kornia"`` fail. This guards against reintroducing
an import-time scripting call. See https://github.com/kornia/kornia/issues/3727.
A fresh interpreter is used so the result is independent of kornia already being
imported by the test session. Only the ``torch.jit.script`` deprecation is promoted
to an error to avoid coupling the test to unrelated third-party warnings.
"""
script = textwrap.dedent(
"""
import warnings
warnings.filterwarnings("error", message=r".*torch\\.jit\\.script.*")
warnings.filterwarnings("error", category=DeprecationWarning, module=r"kornia.*")
warnings.filterwarnings("error", category=FutureWarning, module=r"kornia.*")
import kornia # noqa: F401
"""
)
# Trusted, fixed command (the current interpreter running a literal script); no external input.
result = subprocess.run( # noqa: S603
[sys.executable, "-c", script],
capture_output=True,
text=True,
check=False,
timeout=300,
)
assert result.returncode == 0, f"importing kornia raised a deprecation warning:\n{result.stderr}"
def test_import_emits_no_warnings():
"""Importing kornia must not emit ANY warning, whatever module raises it.
Stricter companion to the jit-script guard above: torch β and onnxruntime, when
installed β are imported first so their own import-time warnings (outside kornia's
control, and platform/version-dependent) are excluded, then every warning is promoted
to an error for the kornia import itself. This is the honest form of the
``python -W error -c "import kornia"`` gate from
https://github.com/kornia/kornia/issues/3727 β it fails on any warning kornia's own
import path emits or triggers, protecting downstream users who run with ``-W error``.
``resetwarnings`` first, so filters inherited from the environment (``PYTHONWARNINGS``,
a ``-W`` flag propagated by the test runner) cannot make the torch import fail and
cannot pre-arm filters that fight the ``simplefilter("error")`` below.
"""
script = textwrap.dedent(
"""
import contextlib
import warnings
warnings.resetwarnings()
import torch # noqa: F401 # torch's own import warnings are not kornia's to fix
# Same for onnxruntime, if installed: its import-time platform warnings (e.g.
# "Unsupported Windows version (2025server)" on windows-2025 runners) are not
# kornia's to fix either. `import kornia` no longer imports onnxruntime eagerly
# (kornia#4260), but pre-importing it here keeps the guard robust regardless.
with contextlib.suppress(ImportError):
import onnxruntime # noqa: F401
warnings.simplefilter("error")
import kornia # noqa: F401
"""
)
# Same trusted, fixed command shape as above.
result = subprocess.run( # noqa: S603
[sys.executable, "-c", script],
capture_output=True,
text=True,
check=False,
timeout=300,
)
assert result.returncode == 0, f"importing kornia emitted a warning:\n{result.stderr}"