Replace urllib.request.url2pathname with local shim for Python 3.14 (#7240)

In Python 3.14, `url2pathname` was rewritten to be a full file-URL
parser per RFC 8089 (python/cpython#126838).
The new implementation prepends `file:` to the input and re-parses it
with `urlsplit`, then validates that the URL authority resolves to
localhost.

This breaks our usage because we pass `urlparse(uri).path` — a pre-parsed
path component — not a full URL. The new `url2pathname` misinterprets
this input, and raises:

    URLError: file:// scheme is supported only on localhost

Since we already validate that the scheme is `file://` before extracting
the path, we replace the stdlib import with a local `url2pathname` shim
that preserves the simple Python <= 3.13 behavior.

Author: agoscinski

src/aiida/common/__init__.py

@@ -22,6 +22,7 @@
 from .links import *
 from .log import *
 from .progress_reporter import *
+from .utils import *
 
 __all__ = (
     'AIIDA_LOGGER',
@@ -87,6 +88,7 @@
     'override_log_level',
     'set_progress_bar_tqdm',
     'set_progress_reporter',
+    'url2pathname',
     'validate_link_label',
 )
 

src/aiida/common/utils.py

@@ -10,6 +10,8 @@
 
 from __future__ import annotations
 
+__all__ = ('url2pathname',)
+
 import filecmp
 import inspect
 import io
@@ -659,3 +661,25 @@ def batch_iter(
 # balance between memory efficiency and database round-trip overhead. Setting it too low increases
 # the number of database queries needed, while setting it too high increases memory consumption.
 DEFAULT_BATCH_SIZE: int = 1000
+
+# To obtain pre py3.14 behavior of url2pathname we copied the code to here
+# https://github.com/python/cpython/blob/1a2b0fb3e5eac4e767e6bbb0b2c3cedaedafc07b/Lib/urllib/request.py#L1664-L1679
+# Only minor changes were applied to conform with mypy
+if os.name == 'nt':
+    from nturl2path import url2pathname
+else:
+    from urllib.parse import unquote
+
+    def url2pathname(url: str) -> str:
+        """OS-specific conversion from a relative URL of the 'file' scheme
+        to a file system path; not recommended for general use."""
+        if url[:3] == '///':
+            # URL has an empty authority section, so the path begins on the
+            # third character.
+            url = url[2:]
+        elif url[:12] == '//localhost/':
+            # Skip past 'localhost' authority.
+            url = url[11:]
+        encoding = sys.getfilesystemencoding()
+        errors = sys.getfilesystemencodeerrors()
+        return unquote(url, encoding=encoding, errors=errors)

src/aiida/manage/configuration/profile.py

@@ -214,8 +214,8 @@ def repository_path(self) -> pathlib.Path:
         :return: absolute filepath of the profile's file repository
         """
         from urllib.parse import urlparse
-        from urllib.request import url2pathname
 
+        from aiida.common.utils import url2pathname
         from aiida.common.warnings import warn_deprecation
 
         warn_deprecation('This method has been deprecated', version=3)

src/aiida/storage/psql_dos/backend.py

@@ -49,7 +49,8 @@
 def get_filepath_container(profile: Profile) -> pathlib.Path:
     """Return the filepath of the disk-object store container."""
     from urllib.parse import urlparse
-    from urllib.request import url2pathname
+
+    from aiida.common.utils import url2pathname
 
     try:
         parts = urlparse(profile.storage_config['repository_uri'])