CC @anton-bobukh, I'm not sure what the intention was for adding the empty Python files for generated types, but it causes a pretty major regression. See #8490 for more info.

@dbaileychess would you mind taking a look at this PR? This Python regression has meant that I have been unable to update to recent versions of flatbuffers, and I'd imagine it is the same for others as well. I am open to a different approach if you have another idea, but looking through the code I can't think of a situation where outputting a file for an already generated type would be desirable.

@dbaileychess Pinging you again as there has been a new release yesterday. This regression has been present since the v24.12.23 release, and has rendered projects that generate for Python largely unusable when flatbuffers with include statements are used. I'm open to suggestions if you feel that this fix isn't appropriate, but I would like to get the regression fixed so it's possible to use newer releases.

Is there a reason why you removed many files from the tests folder?

@EmixamPP these were added in #8292 which introduced the problem. The cause of the regression is that commit chose to output empty files (apart from comments) for any "already generated" types, which occur when you include another .fbs file. This can lead to both:

1. Overwriting previously generated files for those types with empty ones.
2. Adding extra files on the filesystem you don't want, such as including a .fbs file for another library (in a different namespace) where you will import them from a different location.

This commit reverts to the previous behavior of ignoring "already generated" types, and as a result removes these empty files that did not exist previously.

Thanks for the clarification!