Change default dataset download path to user home directory

Author: satwiksps

aeon/datasets/_data_loaders.py

@@ -41,7 +41,7 @@
 from aeon.utils.conversion import convert_collection
 
 DIRNAME = "data"
-MODULE = Path(aeon.__file__).parent / "datasets"
+MODULE = os.path.join(os.path.dirname(aeon.__file__), "datasets")
 
 CONNECTION_ERRORS = (
     HTTPError,
@@ -466,7 +466,9 @@ def _download_and_extract(url, extract_path=None):
         with open(zip_file_name, "wb") as out_file:
             out_file.write(response.read())
     if extract_path is None:
-        extract_path = os.path.join(MODULE, "local_data/%s/" % file_name.split(".")[0])
+        extract_path = os.path.join(
+            str(Path.home() / ".aeon"), "local_data/%s/" % file_name.split(".")[0]
+        )
     else:
         extract_path = os.path.join(extract_path, "%s/" % file_name.split(".")[0])
 
@@ -525,8 +527,14 @@ def _load_tsc_dataset(
         local_module = extract_path
         local_dirname = ""
     else:
-        local_module = MODULE
-        local_dirname = "data"
+        bundled_path = os.path.join(MODULE, "data", name)
+        if os.path.exists(bundled_path):
+            local_module = MODULE
+            local_dirname = "data"
+        else:
+            aeon_home = Path.home() / ".aeon"
+            local_module = str(aeon_home)
+            local_dirname = "data"
 
     if not os.path.exists(os.path.join(local_module, local_dirname)):
         os.makedirs(os.path.join(local_module, local_dirname))
@@ -546,7 +554,11 @@ def _load_tsc_dataset(
             try:
                 _download_and_extract(
                     url,
-                    extract_path=extract_path,
+                    extract_path=(
+                        extract_path
+                        if extract_path is not None
+                        else os.path.join(local_module, local_dirname)
+                    ),
                 )
             except zipfile.BadZipFile as e:
                 raise ValueError(
@@ -988,8 +1000,13 @@ def load_forecasting(name, extract_path=None, return_metadata=False):
         local_module = extract_path
         local_dirname = ""
     else:
-        local_module = MODULE
-        local_dirname = "data"
+        bundled_path = os.path.join(MODULE, "data", name)
+        if os.path.exists(bundled_path):
+            local_module = MODULE
+            local_dirname = "data"
+        else:
+            local_module = str(Path.home() / ".aeon")
+            local_dirname = "data"
 
     if not os.path.exists(os.path.join(local_module, local_dirname)):
         os.makedirs(os.path.join(local_module, local_dirname))
@@ -1029,7 +1046,11 @@ def load_forecasting(name, extract_path=None, return_metadata=False):
                 try:
                     _download_and_extract(
                         url,
-                        extract_path=extract_path,
+                        extract_path=(
+                            extract_path
+                            if extract_path is not None
+                            else os.path.join(local_module, local_dirname)
+                        ),
                     )
                 except zipfile.BadZipFile:
                     raise ValueError(
@@ -1142,8 +1163,13 @@ def load_regression(
         local_module = extract_path
         local_dirname = ""
     else:
-        local_module = MODULE
-        local_dirname = "data"
+        bundled_path = os.path.join(MODULE, "data", name)
+        if os.path.exists(bundled_path):
+            local_module = MODULE
+            local_dirname = "data"
+        else:
+            local_module = str(Path.home() / ".aeon")
+            local_dirname = "data"
     error_str = (
         f"File name {name} is not in the list of valid files to download,"
         f"see aeon.datasets.tser_datasetss.tser_soton for the list. "
@@ -1183,7 +1209,11 @@ def load_regression(
                 try:
                     _download_and_extract(
                         url,
-                        extract_path=extract_path,
+                        extract_path=(
+                            extract_path
+                            if extract_path is not None
+                            else os.path.join(local_module, local_dirname)
+                        ),
                     )
                 except zipfile.BadZipFile:
                     try_monash = True
@@ -1323,8 +1353,13 @@ def load_classification(
         local_module = extract_path
         local_dirname = None
     else:
-        local_module = MODULE
-        local_dirname = "data"
+        bundled_path = os.path.join(MODULE, "data", name)
+        if os.path.exists(bundled_path):
+            local_module = MODULE
+            local_dirname = "data"
+        else:
+            local_module = str(Path.home() / ".aeon")
+            local_dirname = "data"
     if local_dirname is None:
         path = local_module
     else:
@@ -1363,7 +1398,11 @@ def load_classification(
                 try:
                     _download_and_extract(
                         url,
-                        extract_path=extract_path,
+                        extract_path=(
+                            extract_path
+                            if extract_path is not None
+                            else os.path.join(local_module, local_dirname)
+                        ),
                     )
                 except zipfile.BadZipFile:
                     try_zenodo = True
@@ -1444,7 +1483,7 @@ def download_all_regression(extract_path=None):
         local_module = extract_path
         local_dirname = ""
     else:
-        local_module = MODULE
+        local_module = str(Path.home() / ".aeon")
         local_dirname = "data"
 
     if not os.path.exists(os.path.join(local_module, local_dirname)):

aeon/datasets/_single_problem_loaders.py

@@ -24,7 +24,6 @@
 ]
 
 import os
-from pathlib import Path
 
 import numpy as np
 import pandas as pd
@@ -33,7 +32,7 @@
 from aeon.datasets._data_loaders import _load_saved_dataset, _load_tsc_dataset
 
 DIRNAME = "data"
-MODULE = Path(__file__).parent
+MODULE = os.path.dirname(__file__)
 
 
 def load_gunpoint(split=None, return_type="numpy3d"):
@@ -990,4 +989,4 @@ def load_longley(return_array=True):
     data = data.astype(float)
     if return_array:
         return data.to_numpy().T
-    return data.T
+    return data.T

aeon/datasets/dataset_collections.py

@@ -34,14 +34,13 @@
     "get_available_tsf_datasets",
 ]
 import os
-from pathlib import Path
 
 import aeon
 from aeon.datasets.tsc_datasets import multivariate, univariate
 from aeon.datasets.tser_datasets import tser_monash, tser_soton
 from aeon.datasets.tsf_datasets import tsf_all
 
-MODULE = Path(aeon.__file__).parent / "datasets"
+MODULE = os.path.join(os.path.dirname(aeon.__file__), "datasets")
 
 
 def get_available_tser_datasets(name="tser_soton", return_list=True):
@@ -160,4 +159,4 @@ def get_downloaded_tsf_datasets(extract_path=None):
             all_files = os.listdir(sub_dir)
             if name + ".tsf" in all_files:
                 datasets.append(name)
-    return datasets
+    return datasets

aeon/datasets/tests/test_data_loaders.py

@@ -57,7 +57,7 @@ def test_load_forecasting_from_repo():
     assert not meta["contain_missing_values"]
     assert not meta["contain_equal_length"]
 
-    shutil.rmtree(os.path.dirname(__file__) + "/../local_data")
+    shutil.rmtree(os.path.dirname(__file__) + "/../local_data", ignore_errors=True)
 
 
 @pytest.mark.skipif(
@@ -84,7 +84,7 @@ def test_load_classification_from_repo():
     assert meta["classlabel"]
     assert not meta["targetlabel"]
     assert meta["class_values"] == ["1", "2"]
-    shutil.rmtree(os.path.dirname(__file__) + "/../local_data")
+    shutil.rmtree(os.path.dirname(__file__) + "/../local_data", ignore_errors=True)
 
 
 @pytest.mark.skipif(