Cache all datasets at the beginning of tests takes too long (#573)

charavelg · web-flow · commit fd27a6f1e916 · 2025-07-28T13:53:30.000+02:00
Check on list files rather than cache all wich takes too long
diff --git a/conftest.py b/conftest.py
@@ -40,13 +40,12 @@ def pytest_collection_modifyitems(config, items):
                 item.add_marker(skip_marker)
 
     # Skip related doctests if UCR UEA datasets cannot be fetched
-    ucr_uea_datasets = True
     try:
         datasets = UCR_UEA_datasets()
-        datasets.cache_all()
         ucr_uea_datasets = bool(datasets.list_datasets())
     except Exception as exc:
-        warnings.warn("Error caching UCR UEA datasets: {}".format(exc))
+        ucr_uea_datasets = False
+        warnings.warn("Error listing UCR UEA datasets: {}".format(exc))
 
     if not ucr_uea_datasets:
         warnings.warn("Skipping doctests requiring UCR UEA dataset download")
@@ -57,5 +56,7 @@ def pytest_collection_modifyitems(config, items):
                 "tslearn.datasets.ucr_uea.UCR_UEA_datasets.list_multivariate_datasets",
                 "tslearn.datasets.ucr_uea.UCR_UEA_datasets.list_univariate_datasets",
                 "tslearn.datasets.ucr_uea.UCR_UEA_datasets.load_dataset",
+                "tslearn.datasets.ucr_uea.UCR_UEA_datasets.baseline_accuracy",
+                "tslearn.datasets.ucr_uea.UCR_UEA_datasets.list_cached_datasets"
             ]:
                 item.add_marker(skip_marker)
diff --git a/tslearn/datasets/ucr_uea.py b/tslearn/datasets/ucr_uea.py
@@ -198,7 +198,7 @@ def list_datasets(self):
         Returns
         -------
         list of str:
-            A list of names of all (univariate and multivariate) dataset namas.
+            A list of names of all (univariate and multivariate) dataset names.
         """
         return (self.list_univariate_datasets()
                 + self.list_multivariate_datasets())
