GH-115060: Speed up pathlib.Path.glob() by omitting initial stat()

[barneygale]

Since 6258844, paths that might not exist can be fed into pathlib's globbing implementation, which will call os.scandir() / os.lstat() only when strictly necessary. This allows us to drop an initial self.is_dir() call, which saves a stat().

$ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()" "list(p.glob('Lib'))"
20000 loops, best of 5: 13.6 usec per loop
20000 loops, best of 5: 10.4 usec per loop
# --> 1.31x faster

$ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()" "list(p.glob('*.py'))"
5000 loops, best of 5: 88.4 usec per loop
5000 loops, best of 5: 83.8 usec per loop
# --> 1.05x faster

$ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()" "list(p.glob('*'))"
2000 loops, best of 5: 145 usec per loop
2000 loops, best of 5: 139 usec per loop
# --> 1.04x faster

• Issue: Speed up pathlib.Path.glob() by removing redundant regex matching #115060

---

📚 Documentation preview 📚: https://cpython-previews--117831.org.readthedocs.build/

[hauntsaninja]

Since this was previously explicitly documented, should this have a versionchanged in the docs? Oh hmm, I guess this was just documented recently in #114036 by you, so it's probably fine... :-)

I also wonder if we can improve tests, e.g. it looks like the if not self.is_dir(): branch was not covered by tests

[barneygale]

Thanks! I think it's probably not important enough for .. versionchanged::, particularly as we don't document the sorts of OSError that are raised or suppressed from is_dir().

[barneygale]

On reflection, I think this works best as a .. versionchanged:: directive. Thank you for the pointer :)