Fix save/load functions to accept pathlib.Path objects

[Copilot]

Fix save() and load() functions to accept pathlib.Path objects

The issue is that save() cannot take a Path object as it expects a string. The error occurs because Path objects don't support string operations like slicing (s[-5:]) or methods like .endswith() and .lower().

Solution:

Added os.fspath() calls to convert path-like objects (including Path objects) to strings at entry points before any string operations are performed. This is the standard Python way to handle path-like objects and is compatible with both strings and Path objects.

Changes Made:

1. In save(): Added os.fspath(filename) at the beginning to handle the .endswith('.sobj') call
2. In load(): Added os.fspath(filename) after extracting single filename to handle the .lower() call, with logic to skip URL fetching for Path objects
3. In _base_save(): Added os.fspath(filename) at the beginning to handle Path objects from SageObject.save()
4. Added comprehensive doctests in both save() and load() demonstrating Path object support using context manager

The conversion is now done at all entry points (save, load, _base_save) rather than in the internal helper function (_normalize_filename), making the code cleaner and more maintainable. Additionally, Path objects are explicitly prevented from triggering URL fetching logic.

Testing:

• Understand the issue: Path objects are not subscriptable but need to be converted to strings
• Add os.fspath() at entry points (save, load, _base_save)
• Keep _normalize_filename as internal string-only function
• Add doctests to save() and load() functions to verify Path objects work
• Update doctests to use context manager for TemporaryDirectory
• Prevent URL fetching when Path objects are provided
• Verify logic correctness with standalone Python tests
• Verify syntax and indentation of changes
• Confirm changes are minimal

The fix is minimal, follows Python best practices, and maintains backward compatibility with existing string-based code.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>save cannot take a Path object</issue_title>
<issue_description>save cannot take a Path object, as it expects a string.

sage: one = 1
sage: from pathlib import Path
sage: save(one, Path("foobar"))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[3], line 1
----> 1 save(one, Path("foobar"))

File /Applications/sage-dev/src/sage/misc/persist.pyx:283, in sage.misc.persist.save()
    281
    282     if not os.path.splitext(filename)[1] or not hasattr(obj, 'save'):
--> 283         filename = _normalize_filename(filename)
    284
    285     if filename.endswith('.sobj'):

File /Applications/sage-dev/src/sage/misc/persist.pyx:83, in sage.misc.persist._normalize_filename()
     81 Append the .sobj extension to a filename if it doesn't already have it.
     82 """
---> 83 if s[-5:] != '.sobj':
     84     return s + '.sobj'
     85

TypeError: 'PosixPath' object is unsliceable
```</issue_description>

## Comments on the Issue (you are @copilot in this section)

<comments>
</comments>

Fixes #1

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[edgarcosta]

@copilot, follow the advice from sagemath#41037 (comment) to use context manager ("with ..") will both (a) automatically clean up the directory when you're done, and (b) give you the name to begin with. For example, this should work:

sage: with tempfile.TemporaryDirectory() as d:
....:     p = Path(d) / "test_path"
....:     save(1, p)
....:     load(p)

[Copilot]

@copilot, follow the advice from sagemath#41037 (comment) to use context manager ("with ..") will both (a) automatically clean up the directory when you're do...

Updated the doctest to use a context manager as suggested. The new format is more concise and automatically handles cleanup. Changes in commit 2a8e6f1.

[edgarcosta]

@copilot This line is useless, remove it

[Copilot]

Removed the os.fspath() call from _normalize_filename and moved it to _base_save() instead. This keeps the conversion at entry points only, making _normalize_filename a simpler internal helper that works with strings. Changes in commit 90ca892.

[edgarcosta]

@copilot remove comment, add doctest, and make sure if filename is originally a Path then the branch below with fetching URL is never executed.

[Copilot]

Removed the comment, added a doctest for Path objects in load(), and ensured Path objects skip the URL fetching branch by checking for __fspath__ before conversion. Changes in commit 6b2b9e3.