Add py.typed marker for PEP 561 support (#151)

[qiinori]

• Add PEP 561 py.typed marker file so type checkers (pyright/pylance) recognize the package as typed

• Configure pyproject.toml to include py.typed in the distribution

Closes #151

[codecov]

Codecov Report

❌ Patch coverage is 70.00000% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 97.73%. Comparing base (dd4a357) to head (5915c22).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
fsrs/scheduler.py	70.00%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #153      +/-   ##
==========================================
- Coverage   98.62%   97.73%   -0.90%     
==========================================
  Files           7        7              
  Lines         653      661       +8     
==========================================
+ Hits          644      646       +2     
- Misses          9       15       +6     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[L-M-Sherlock]

It hasn't passed the ci.

[asukaminato0721]

https://pytorch.org/blog/pyrefly-now-type-checks-pytorch/

so actually I recommend pyrefly (

#149

[qiinori]

It hasn't passed the ci.

Hi! Fixed — ty 0.0.17 reclassified union attribute access from possibly-missing-attribute, so the ignore directives needed updating. Pushed the fix, CI needs approval to run.

[qiinori]

https://pytorch.org/blog/pyrefly-now-type-checks-pytorch/

so actually I recommend pyrefly (

#149

I can give pyrefly a try. Happy to fix them and switch to pyrefly in this PR if you'd like, or we can leave that for #149.

[asukaminato0721]

try switch, let's see the results.

Since for now too many ignore

[qiinori]

try switch, let's see the results.

Since for now too many ignore

Hi! I've pushed the transition from ty to pyrefly now. Eliminates all 4 ignore comments.

Code changes in fsrs/scheduler.py:

• Added @overload on _clamp_difficulty and _clamp_stability to type the dual float/torch.Tensor paths, letting pyrefly validate torch. Tensor methods like .clamp() in the else branch — eliminates all 4 ignore comments
• Replaced isinstance(x, Real) with isinstance(x, (int, float)) so pyrefly can narrow the else branch to torch. Tensor
• Added TYPE_CHECKING-gated from torch import Tensor (torch is optional; no runtime cost)
• Added wildcard case _: branches — IntEnum inherits int, so pyrefly can't prove exhaustiveness
• Added float() cast in _get_fuzz_range for mixed-type min() overload resolution

Code changes in pyproject.toml:

• switch ty with pyrefly in dev deps and tool config
• Added ignore-missing-imports = ["torch"] so pyrefly doesn't error on the optional torch dependency

Code changes in .github/workflows/type-check.yml:

• switch ty with pyrefly in install and run steps

[asukaminato0721]

Added wildcard case _: branches — IntEnum inherits int, so pyrefly can't prove exhaustiveness

let me try to reproduce it.

https://pyrefly.org/sandbox/?project=N4IgZglgNgpgziAXKOBDAdgEwEYHsAeAdAA4CeSIAxAAQDqMUAxrgLYzUAuunAFuwAqkATjDBRS1AMoYcBADS8IcakuqpqxVAHMYAHXQ0A7nxHVSuAK7VGGaoaEQOMBXB6pTM6rHfo1eCxycpMQQ6FoqvoIcPLjohPqUCdQAEjAeHtaWxLDUYDCoHBYiyrhgispoWHj4iEkAtNQA4jCBIqhQdRwQbEHE7Ix8jADWoeF5MDiow-XUAKpw7CIAbvlQAPocwTAAFACUnNyhcH2MgaF5QiKYvfAzMPh9DmzogYaOPLzsAJIAIgCiuXyhWKvAKdnYcAsxGIuCEgW2MEIWkIagCuGYLGyLWc1C0uE6%2BMwolCjggsQUMRWQgULUYuxmtAcTjUWGoQgskVI0VimSJ1EwEBEp3EETMliE1GwQlwhgWQn0%2BjA0pY1Bg6AsKu6MLh1C%2BLz%2B6pYCvQ%2BkYUFQcGUACUCqNtnqOAaNbtar5qO7dCBPd63e6nSqRMRimqumFPrlxRpcJaINgckJbWFlMY1WyYEsIDA3mH1DYhJh4r7qN6vV6TUWAIJaVChagAXmoAEZ9O73cl3NcGwAmFutxq4XCd6gAZl7fotEgbABZjUSymAB3s6gA%2BG2hrSu1vUQg7434esR3B7fQsAoDag1MfWC3sNejQhVmvoTdb1s7wtbmwLah3sKEP4Ti%2Br7bruRZfreibIv2g5Aa%2B75XuBP6QYQ7b5rBW7wegIByCAATQHAJDkIgVBzF0UCOBIYAcqcZLoHAs6iBGQinhwayGtgaTbDUEQcPsK7UHAHBCLBIjAr4YCegAchqHHCdQwD4AAvj62EgGQIhiKQhBcCwUAUDQggaSKlSyPuzDoJAWhFLasSFjQkgwOwPAcBwxBwIgAD0HnqaI4iELCWgeWqHmYOicAeeZlnWV0sQeUxahLDW5pxv0sRRQmMW%2BLgxCZQR%2BhkNy6B1FScC0QenrDoQjY9mWICKThUxdCsABi0AwBQJnVIRdVAA

so pyrefly can prove exhaustiveness.
I wrote that part in pyrefly :)

Edit: (oh I got this, it's another issue).

[asukaminato0721]

ERROR `next_interval` may be uninitialized [unbound-name]
   --> fsrs/scheduler.py:497:64
    |
497 |             next_interval = self._get_fuzzed_interval(interval=next_interval)
    |                                                                ^^^^^^^^^^^^^
    |
ERROR `next_interval` may be uninitialized [unbound-name]
   --> fsrs/scheduler.py:499:38
    |
499 |         card.due = review_datetime + next_interval
    |                                      ^^^^^^^^^^^^^
    |
 INFO 2 errors

sorry I found the problem, could you drop my two commits.

And we will merge this.

[qiinori]

ERROR ``next_interval`` may be uninitialized [unbound-name]
   --> fsrs/scheduler.py:497:64
    |
497 |             next_interval = self._get_fuzzed_interval(interval=next_interval)
    |                                                                ^^^^^^^^^^^^^
    |
ERROR ``next_interval`` may be uninitialized [unbound-name]
   --> fsrs/scheduler.py:499:38
    |
499 |         card.due = review_datetime + next_interval
    |                                      ^^^^^^^^^^^^^
    |
 INFO 2 errors

sorry I found the problem, could you drop my two commits.

And we will merge this.

Thanks for the approval!

[ishiko732]

Could you update the version here?

py-fsrs/pyproject.toml

Line 7 in a3804a6

version = "6.3.0"

The new release depends on this version number.