Use pre-generated TorchScript fixture to avoid torch.jit deprecation warnings (#254)

* Use pre-generated TorchScript fixture to avoid torch.jit deprecation warnings

torch.jit.script/torch.jit.save emit DeprecationWarnings on Python 3.14+
and may be removed in a future PyTorch release. Replace runtime JIT calls
in test setUp with a pre-generated (smaller) fixture file, and include a
standalone script (test/fixtures/generate_fixtures.py) to regenerate it
with pinned versions of torch and torchvision.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Potential fix for pull request finding 'Unused import'

Co-authored-by: Copilot Autofix powered by AI <223894421+github-code-quality[bot]@users.noreply.github.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Copilot Autofix powered by AI <223894421+github-code-quality[bot]@users.noreply.github.com>

Author: 3 people

test/fixtures/generate_fixtures.py

@@ -0,0 +1,35 @@
+#!/usr/bin/env python3
+# /// script
+# requires-python = ">=3.10,<3.14"
+# dependencies = ["torch>=2.10,<3", "torchvision>=0.25,<1"]
+# ///
+"""Generate TorchScript fixture files for tests.
+
+These fixtures are checked into the repository so that tests don't need
+to call torch.jit.script/torch.jit.save at runtime (which emits
+deprecation warnings on Python 3.14+ and may be removed in future
+PyTorch versions).
+
+Usage:
+    uv run test/fixtures/generate_fixtures.py
+"""
+
+from pathlib import Path
+
+import torch
+import torchvision.models as models
+
+FIXTURES_DIR = Path(__file__).parent
+
+
+def main():
+    model = models.squeezenet1_0()
+    scripted = torch.jit.script(model)
+
+    out = FIXTURES_DIR / "squeezenet1_0_torchscript_v1_4.pt"
+    torch.jit.save(scripted, out)
+    print(f"Generated {out} ({out.stat().st_size} bytes)")
+
+
+if __name__ == "__main__":
+    main()

test/fixtures/squeezenet1_0_torchscript_v1_4.pt

@@ -1,6 +1,5 @@
 import random
 import string
-import sys
 import tarfile
 import tempfile
 import unittest
@@ -15,7 +14,7 @@
 import fickling.polyglot as polyglot
 from fickling.polyglot import FileProperties
 
-_lacks_torch_jit_support = sys.version_info >= (3, 14)
+FIXTURES_DIR = Path(__file__).parent / "fixtures"
 
 
 def _make_properties(**overrides):
@@ -79,17 +78,9 @@ def setUp(self):
         self.filename_legacy_pickle = tmppath / "model_legacy_pickle.pth"
         torch.save(model, self.filename_legacy_pickle, _use_new_zipfile_serialization=False)
 
-        if not _lacks_torch_jit_support:
-            # TorchScript v1.4
-            m = torch.jit.script(model)
-            self.filename_torchscript = tmppath / "model_torchscript.pt"
-            torch.jit.save(m, self.filename_torchscript)
-
-            # TorchScript v1.4 Dup
-            self.filename_torchscript_dup = tmppath / "model_torchscript_dup.pt"
-            torch.jit.save(m, self.filename_torchscript_dup)
-
-            self.standard_torchscript_polyglot_name = tmppath / "test_polyglot.pt"
+        # TorchScript v1.4 (pre-generated fixtures to avoid torch.jit deprecation warnings)
+        self.filename_torchscript = FIXTURES_DIR / "squeezenet1_0_torchscript_v1_4.pt"
+        self.standard_torchscript_polyglot_name = tmppath / "test_polyglot.pt"
 
         # PyTorch v0.1.1
         self.filename_legacy_tar = tmppath / "model_legacy_tar.pth"
@@ -132,7 +123,6 @@ def test_legacy_pickle(self):
         formats = polyglot.identify_pytorch_file_format(self.filename_legacy_pickle)
         self.assertEqual(formats, ["PyTorch v0.1.10"])
 
-    @unittest.skipIf(_lacks_torch_jit_support, "PyTorch 2.9.1 JIT broken with Python 3.14+")
     def test_torchscript(self):
         formats = polyglot.identify_pytorch_file_format(self.filename_torchscript)
         self.assertEqual(formats, ["TorchScript v1.4", "TorchScript v1.3", "PyTorch v1.3"])
@@ -200,7 +190,6 @@ def test_legacy_pickle_properties(self):
         proper_result = _make_properties(is_valid_pickle=True)
         self.assertEqual(properties, proper_result)
 
-    @unittest.skipIf(_lacks_torch_jit_support, "PyTorch 2.9.1 JIT broken with Python 3.14+")
     def test_torchscript_properties(self):
         properties = polyglot.find_file_properties(self.filename_torchscript)
         proper_result = _make_properties(
@@ -219,11 +208,10 @@ def test_zip_properties(self):
         proper_result = _make_properties(is_standard_zip=True, is_standard_not_torch=True)
         self.assertEqual(properties, proper_result)
 
-    @unittest.skipIf(_lacks_torch_jit_support, "PyTorch 2.9.1 JIT broken with Python 3.14+")
     def test_create_standard_torchscript_polyglot(self):
         polyglot.create_polyglot(
             self.filename_v1_3_dup,
-            self.filename_torchscript_dup,
+            self.filename_torchscript,
             self.standard_torchscript_polyglot_name,
             print_results=False,
         )

test/test_polyglot.py

@@ -1,4 +1,3 @@
-import sys
 import tempfile
 import unittest
 from pathlib import Path
@@ -9,7 +8,7 @@
 from fickling.fickle import Pickled
 from fickling.pytorch import PyTorchModelWrapper
 
-_lacks_torch_jit_support = sys.version_info >= (3, 14)
+FIXTURES_DIR = Path(__file__).parent / "fixtures"
 
 
 class TestPyTorchModule(unittest.TestCase):
@@ -21,10 +20,8 @@ def setUp(self):
         self.filename_v1_3 = tmppath / "test_model.pth"
         torch.save(model, self.filename_v1_3)
 
-        if not _lacks_torch_jit_support:
-            m = torch.jit.script(model)
-            self.torchscript_filename = tmppath / "test_model_torchscript.pth"
-            torch.jit.save(m, self.torchscript_filename)
+        # Pre-generated fixture to avoid torch.jit deprecation warnings
+        self.torchscript_filename = FIXTURES_DIR / "squeezenet1_0_torchscript_v1_4.pt"
 
     def tearDown(self):
         self.tmpdir.cleanup()
@@ -35,7 +32,6 @@ def test_wrapper(self):
         except Exception as e:  # noqa
             self.fail(f"PyTorchModelWrapper was not able to load a PyTorch v1.3 file: {e}")
 
-    @unittest.skipIf(_lacks_torch_jit_support, "PyTorch 2.9.1 JIT broken with Python 3.14+")
     def test_torchscript_wrapper(self):
         try:
             PyTorchModelWrapper(self.torchscript_filename)
@@ -47,7 +43,6 @@ def test_pickled(self):
         pickled_portion = result.pickled
         self.assertIsInstance(pickled_portion, Pickled)
 
-    @unittest.skipIf(_lacks_torch_jit_support, "PyTorch 2.9.1 JIT broken with Python 3.14+")
     def test_torchscript_pickled(self):
         result = PyTorchModelWrapper(self.torchscript_filename)
         pickled_portion = result.pickled