Version 0.12.0? used 12 because 11 was last used.

suresttexas00 commented 7 years ago

Based off of 0.8.0, this version successfully implements the Linux special keys including the ESC.

added the TAB character, which was missing.

For both OS types, the new readkey() now wraps a `get_char` method, which is separate from the readchar methods (which only return a character byte). The get_char handles the reading need to get the full keystroke codes.

for Linux:

At the heart of this is a new reachar_linux.readchar() that successfully implements a real blocking bahavior, allowing us to get all queued up characters after an ESC.
This means if Escape was the only character, we can detect that now.
Users can also use the Linux (and Windows) versions to call readchar() in a non-blocking manner (the default is to block, i.e. wait for a keystroke to be typed).

for Windows:

I re-implemented Guiweber's windows methods, with blocking, however I don't have a windows testing platform to test or improve this section.
It differs from his version because the readkey() returns the standard constants from key.py that are used by Linux versus the originally proposed string values (so separate comparisons are not required between OSes).
for convenience, key.py contains a conversion function convertchar() to convert these back to string names like "ctrl_a", if desired.
The constants for windows will need to be updated and improved due to being incomplete. New items should only be added if we can create a valid byte-string constant for them.

suresttexas00 commented 7 years ago

@magmax .. Can we discuss?? I found the problem. The test suite is substituting its get_char function for the actual function in our code... which works when YOUR readkey was processing the keystrokes (not real keystrokes, but just a stream of values). As a result, when the test suite passes "\x1b ....", the stream returns x1b and the factory function returns that to the test code. The internal functions for get_char for linux or windows never get run.

This test suite does not properly test our code. I have just tweaked it to allow passing the factory function test stream...

suresttexas00 commented 7 years ago

If you don't like this whole thing, feel free to close it. I admit it is a bit of a mess and maybe you want to start from scratch with what you were doing. I don't like the hacks I implemented around the Travis testing, which created "passing" build, but horrible code...

However, on my own project, I did do a lot of windows testing and mapped most of the windows keys. I would suggest you take advantage of the mappings I made for windows in my other repo: https://github.com/suresttexas00/readkey

magmax commented 7 years ago

I see you already realized about the mess we have to do here. But I think it is better to do the mess here than in a lot of projects.

I started this project thinking it was much easier, but... well, now I am even more happy to have started it, despite maybe it is not the best solution in the world.

Thank you very much for your contributions. I will have a second look as soon as I can.

suresttexas00 commented 7 years ago

Not sure where we are with this now... Did you want me to make these changes first, or did you want to approve the pull request and then make the changes yourself?

magmax commented 3 years ago

I've changed the way to use versions. Do not care about them any more. This PR contains a lot of changes. Please, use smaller ones.

suresttexas00 commented 3 years ago

Wow, I had forgotten about this. I don't use it anymore and I have no idea where I left off, code-wise. I won't be issuing any more PR's at this time.

magmax / python-readchar