Skip tests using PyTorch JIT on Python 3.14+

Author: thomas-chauchefoin-tob

test/test_pytorch.py

@@ -1,4 +1,5 @@
 import os
+import sys
 import unittest
 
 import torch
@@ -7,19 +8,24 @@
 from fickling.fickle import Pickled
 from fickling.pytorch import PyTorchModelWrapper
 
+_lacks_torch_jit_support = sys.version_info >= (3, 14)
 
 class TestPyTorchModule(unittest.TestCase):
     def setUp(self):
         model = models.mobilenet_v2()
         self.filename_v1_3 = "test_model.pth"
         torch.save(model, self.filename_v1_3)
         self.zip_filename = "test_random_data.zip"
-        m = torch.jit.script(model)
-        self.torchscript_filename = "test_model_torchscript.pth"
-        torch.jit.save(m, self.torchscript_filename)
+        if not _lacks_torch_jit_support:
+            m = torch.jit.script(model)
+            self.torchscript_filename = "test_model_torchscript.pth"
+            torch.jit.save(m, self.torchscript_filename)
 
     def tearDown(self):
-        for filename in [self.filename_v1_3, self.zip_filename, self.torchscript_filename]:
+        files = [self.filename_v1_3, self.zip_filename]
+        if not _lacks_torch_jit_support:
+            files.append(self.torchscript_filename)
+        for filename in files:
             if os.path.exists(filename):
                 os.remove(filename)
 
@@ -29,6 +35,7 @@ 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)
@@ -40,6 +47,7 @@ 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