GH-117586: Speed up pathlib.Path.glob() by working with strings

[barneygale]

Move pathlib globbing implementation into a new private class: glob._Globber. This class implements fast string-based globbing. It's called by pathlib.Path.glob(), which then converts strings back to path objects.

In the private pathlib ABCs, add a pathlib._abc.Globber subclass that works with PathBase objects rather than strings, and calls user-defined path methods like PathBase.stat() rather than os.stat().

This sets the stage for two more improvements:

• Speed up pathlib.Path.glob() by removing redundant regex matching #115060
• Speed up glob.glob() by reducing number of system calls made #116380

Timings:

$ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()""list(p.glob('Lib/*'))" 1000 loops, best of 5: 392 usec per loop 1000 loops, best of 5: 365 usec per loop # --> 1.07x faster $ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()""list(p.glob('Lib/*.py'))" 1000 loops, best of 5: 393 usec per loop 1000 loops, best of 5: 371 usec per loop # --> 1.06x faster $ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()""list(p.glob('Lib/**'))" 50 loops, best of 5: 9.46 msec per loop 50 loops, best of 5: 9.06 msec per loop # --> 1.04x faster $ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()""list(p.glob('Lib/**/'))" 50 loops, best of 5: 4.98 msec per loop 50 loops, best of 5: 5.15 msec per loop # --> 1.03x slower (!) $ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()""list(p.glob('Lib/**/*'))" 20 loops, best of 5: 14 msec per loop 20 loops, best of 5: 12.9 msec per loop # --> 1.09x faster $ ./python -m timeit -s "from pathlib import Path; p = Path.cwd()""list(p.glob('Lib/**/*.py'))" 20 loops, best of 5: 12.2 msec per loop 20 loops, best of 5: 11.4 msec per loop # --> 1.07x faster

---

• Issue: Speed up pathlib.Path.glob() by working with strings internally #117586

[barneygale]

This is the first PR in a series that will hopefully unify the globbing implementations in the pathlib and glob modules, and speed both up in the process.

[barneygale]

Hey @serhiy-storchaka, does this PR look alright to you? Not requesting a detailed review, more of a sanity check, given you've looked after the glob module for the last few years.

This PR doesn't affect glob.[i]glob(), but it does move pathlib's globbing implementation into glob.py.

Thank you.

[barneygale]

I'll merge this now as it's important for #115060, which I'm hoping to get done in time for 3.13 beta 1.

But I'll leave glob.glob() and glob.iglob() unchanged in 3.13; any PRs I make will target 3.14.