Merge pull request #530 from KhiopsML/528-remote-home

Fix a buggy samples_dir path construction when HOME contains a remote path

Author: tramora

CHANGELOG.md

@@ -6,6 +6,11 @@
   - Example: 10.2.1.4 is the 5th version that supports khiops 10.2.1.
 - Internals: Changes in *Internals* sections are unlikely to be of interest for data scientists.
 
+## Unreleased
+
+## Fixed
+- (`core`) Samples dir path construction when HOME is a remote path
+
 ## 11.0.0.1 - 2026-01-14
 
 ### Fixed

doc/requirements.txt

@@ -4,6 +4,6 @@ ipykernel>=6.9.1
 nbconvert==6.4.4
 nbformat==5.3.0
 numpydoc>=1.5.0
-pandas>=0.25.3
+pandas>=0.25.3,<=2.3.3
 scikit-learn>=0.22.2,<=1.7.2
 sphinx-copybutton>=0.5.0

khiops/core/internals/runner.py

@@ -14,7 +14,6 @@
 
 import io
 import os
-import pathlib
 import platform
 import shlex
 import shutil
@@ -65,7 +64,11 @@ def get_default_samples_dir():
     elif platform.system() == "Windows" and "PUBLIC" in os.environ:
         samples_dir = os.path.join(os.environ["PUBLIC"], "khiops_data", "samples")
     else:
-        samples_dir = str(pathlib.Path.home() / "khiops_data" / "samples")
+        # The filesystem abstract layer is used here
+        # as the path can be either local or remote
+        samples_dir = fs.get_child_path(
+            fs.get_child_path(os.environ["HOME"], "khiops_data"), "samples"
+        )
     return samples_dir
 
 

packaging/conda/meta.yaml

@@ -24,7 +24,7 @@ requirements:
   run:
     - python
     - khiops-core =11.0.0
-    - pandas >=0.25.3
+    - pandas >=0.25.3,<=2.3.3
     - scikit-learn >=0.22.2
   run_constrained:
     # do not necessary use the latest version

pyproject.toml

@@ -104,8 +104,8 @@ classifiers = [
 ]
 requires-python = ">=3.8"
 dependencies = [
-    "pandas>=0.25.3",
-    # do not use the latest version, to avoid undesired breaking changes
+    # do not use the latest versions, to avoid undesired breaking changes
+    "pandas>=0.25.3,<=2.3.3",
     "scikit-learn>=0.22.2,<=1.7.2",
 ]
 

tests/test_remote_access.py

@@ -227,6 +227,33 @@ def test_train_predictor_fail_and_log_with_remote_access(self):
             self.assertTrue(fs.exists(log_file_path), f"Path: {log_file_path}")
             fs.remove(log_file_path)
 
+        def test_samples_dir_inferred_from_remote_home(self):
+            """Test samples_dir is correctly inferred using a remote path in HOME"""
+
+            # Save initial state
+            # This runner has remote paths (for root_temp_dir for example)
+            initial_runner = kh.get_runner()
+            initial_home = os.environ.get("HOME")
+
+            # Set a remote path to HOME
+            os.environ["HOME"] = initial_runner.root_temp_dir
+            test_runner = KhiopsLocalRunner()
+            kh.set_runner(test_runner)
+
+            # Test the home path is indeed remote
+            self.assertFalse(fs.is_local_resource(os.environ["HOME"]))
+
+            # Test that samples_dir is built according to the expectations
+            expected_samples_dir = fs.get_child_path(
+                fs.get_child_path(os.environ["HOME"], "khiops_data"), "samples"
+            )
+            self.assertEqual(test_runner.samples_dir, expected_samples_dir)
+
+            # Restore initial state
+            if initial_home is not None:
+                os.environ["HOME"] = initial_home
+            kh.set_runner(initial_runner)
+
 
 class KhiopsS3RemoteFileTests(KhiopsRemoteAccessTestsContainer.KhiopsRemoteAccessTests):
     """Integration tests with Amazon S3 filesystems"""