Switch from pywin32 to ctype

[samepaul]

A bit less fancy looking code, but doesn't depend on pywin32 which is evil :)

[tikuma-lsuhsc]

@samepaul - thanks for the PR. I'll review it more in depth later, but I think you left out the most important change to exorcise the evil pywin32: it must be removed from pyproject.tml's dependencies. Could you do that, too?

[tikuma-lsuhsc]

Addresses Issue #4

[tikuma-lsuhsc]

It's not passing the CI, failing both read & write tests with FFmpeg. If you could run FFmpeg on your system, please test with pytest.

[samepaul]

It's not passing the CI, failing both read & write tests with FFmpeg. If you could run FFmpeg on your system, please test with pytest.

I'm unfortunately not really familiar with ffmpeg and what are expectations of it with regard to pipes. I notices that ERROR_BROKEN_PIPE should not generate exception and perhaps non-zero buffers affect behavior as well. I will commit that fix, but other than that is above me, sorry :(

[samepaul]

It's not passing the CI, failing both read & write tests with FFmpeg. If you could run FFmpeg on your system, please test with pytest.

I would actually appreciate if you could review readinto because I am not entirely sure what exactly it is expected to do.

[tikuma-lsuhsc]

It's not passing the CI, failing both read & write tests with FFmpeg. If you could run FFmpeg on your system, please test with pytest.

I'm unfortunately not really familiar with ffmpeg and what are expectations of it with regard to pipes. I notices that ERROR_BROKEN_PIPE should not generate exception and perhaps non-zero buffers affect behavior as well. I will commit that fix, but other than that is above me, sorry :(

no biggie. I'll review and edit the code later when I find a chunk of time.

[tikuma-lsuhsc]

The test is passing on my PC, so it's good so far.

I would actually appreciate if you could review readinto because I am not entirely sure what exactly it is expected to do.

I think you're confused because the original version was overly cautious... My (unfounded) fear was that the read operation somehow times out and does not return the number of bytes requested without throwing an error. The ReadFile documentation does not suggest such possibility (I think) so we can stick to how you have it, and if a problem arises in the future I'll address it.

I'll let it sit for another day and will try merge it tomorrow night and publish the updated release.

[samepaul]

The test is passing on my PC, so it's good so far.

I would actually appreciate if you could review readinto because I am not entirely sure what exactly it is expected to do.

I think you're confused because the original version was overly cautious... My (unfounded) fear was that the read operation somehow times out and does not return the number of bytes requested without throwing an error. The ReadFile documentation does not suggest such possibility (I think) so we can stick to how you have it, and if a problem arises in the future I'll address it.

I'll let it sit for another day and will try merge it tomorrow night and publish the updated release.

Sure. I’m glad you liked it. That was an interesting first time journey for me. 😊