Add initial implementation of resampy library

[hunterhogan]

Introduce core functionality and metadata for the resampy library, including initial stubs for resampling methods and filter functions. Update metadata to specify dependencies.

See also #15338.

[donbarbos]

Thank you! I’ve found a few more points; sorry for such a fragmentary review.

[hunterhogan]

Thank you! I’ve found a few more points; sorry for such a fragmentary review.

No, thank YOU! About the "fragmentary review": I certainly understand that if you were reviewing the work of a peer, they would likely prefer everything at once. But I'm not your peer, and my PR needs a lot more reviewing than a PR from an experienced programmer.

And to be frank, I'm very happy you did not put all of this in the first review. I think that would have been disheartening for me. These "fragments" are just the right size for me: I'm learning, but not overwhelmed.

Thank you very much for your patience with my inexperience and for essentially mentoring me through this process.

I'll study the comments you left and make changes.

[hunterhogan]

unrecognized setting "$schema".

Running: /opt/hostedtoolcache/Python/3.13.11/x64/bin/pyright --pythonplatform Windows --pythonversion 3.9
Config contains unrecognized setting "$schema".

https://github.com/python/typeshed/actions/runs/21557033413/job/62115164783?pr=15341#step:11:24

On my computer, I get the same output.

>py -m tests.runtests stubs/resampy --run-stubtest
...

Running: C:\Program Files\nodejs\npx.CMD pyright@1.1.408 stubs\resampy --pythonversion 3.9 -p pyrightconfig.stricter.json
0 errors, 0 warnings, 0 informations
Config contains unrecognized setting "$schema".

All tests.runtests Details

(.venv) C:\clones\typeshed>py -m tests.runtests stubs/resampy --run-stubtest

Running pre-commit...
trim trailing whitespace.................................................Passed
fix end of files.........................................................Passed
check yaml...........................................(no files to check)Skipped
check toml...............................................................Passed
check for merge conflicts................................................Passed
mixed line ending........................................................Passed
check for case conflicts.................................................Passed
Run ruff on stubs, tests and scripts.....................................Passed
Run ruff on the test cases...........................(no files to check)Skipped
black....................................................................Passed
flake8...................................................................Passed
Check hooks apply to the repository..................(no files to check)Skipped

Running check_typeshed_structure.py...

Running Pyright (stricter configs) for Python 3.9...
11.4.2
Running: C:\Program Files\nodejs\npx.CMD pyright@1.1.408 stubs\resampy --pythonversion 3.9 -p pyrightconfig.stricter.json
0 errors, 0 warnings, 0 informations
Config contains unrecognized setting "$schema".


Running mypy for Python 3.9...
*** Testing Python 3.9 on win32
Testing third-party packages...
testing resampy (5 files)... success

--- success, 5 files checked ---

Running stubtest...
resampy... (65.07 s) success
Note: resampy is not currently tested on win32 in typeshed's CI

Regression tests: No test_cases folder for 'resampy'!


--- TEST SUMMARY: All tests passed! ---

pre-commit Success
Check structure: Success
Pyright: Success
mypy: Success
stubtest: Success
Pyright regression tests: Success
mypy regression test: Success

Is that important?

np.floating

The failures on Python 3.9 are of the following type:

error: Expected type arguments for generic class "floating"

But maybe that is because of the schema, above.

kwargs and ndarray

I wrote this stub 11 months ago, and I guess I have learned a lot more than I realized since then. I fixed kwargs: Any. We actually know exactly what the keywords and their types are. I mimicked Unpack from ast.pyi, but I'm certainly not claiming that is the "best" or the "right" way to annotate this.

I also realized that I was letting the type checker boss me around. The np.ndarray used as filters in this package are very specific: 1-axis, float64. Always. I was annotating based on what Pylance was telling me instead of using my brain to read the freaking code. Therefore, I tightened up those annotations.

"internal" functions

_resample_loop is clearly internal, so I removed it.

I think resample_f_s and resample_f_p are best categorized as internal also. I cannot imagine why I would bypass core.resample or core.resample_nu and directly call resample_f_p or resample_f_s. The two "core" functions have defensive coding and logic that I would need to duplicate. The authors didn't add either of those functions to any __all__ statement or to __init__.py, either. But that would create an empty stub file, and the idea of an intentionally empty stub file seems bizarre to me. It's probably normal, but I compulsively annotate everything.

I tried removing them and the allowlist, but the tests on my computer were unhappy.

error: resampy.interpn.resample_f_p is not present in stub
Stub: in file C:\clones\typeshed\stubs\resampy\resampy\interpn.pyi
MISSING
Runtime: at line 73
def (x, t_out, interp_win, interp_delta, num_table, scale, y)

error: resampy.interpn.resample_f_s is not present in stub
Stub: in file C:\clones\typeshed\stubs\resampy\resampy\interpn.pyi
MISSING
Runtime: at line 81
def (x, t_out, interp_win, interp_delta, num_table, scale, y)

I restored the stubtest_allowlist, and wrote a comment.

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉