Filesize detection doesn't work for compressed files

[tstenner]

https://github.com/xdf-modules/xdf-python/blob/a6c5f84860c41de9d63fc3f66ac7cac0f7779bc7/pyxdf/pyxdf.py#L235-L245

For compressed files, reading the file is aborted as soon as an invalid chunk size is read and the position in the uncompressed stream is after the compressed size. So, the new code checks if more than 1024 bytes are actually readable.

[cbrnr]

LGTM, but what's the special meaning of 1024 bytes here? Where does this number come from? A comment in the code would be helpful.

[tstenner]

I have no idea where the 1024 bytes came from. I suspect that it wasn't considered too useful to find a boundary chunk if the file is too close to the end, but how the 1024 came from that I don't know.

[cbrnr]

Can you add this explanation as a comment (it makes sense to not look for boundary chunks very close to the end of a file)?

[tstenner]

It doesn't matter how much data is left, so I changed the check to read only one byte. If there's not enough left for a boundary chunk, _scan_forward will read until the end of the file and then quit. It's a tiny bit slower, but we've gotten rid of one magic number.

[cbrnr]

And this happens only for corrupt files anyway, right?

[tstenner]

Yes with a tiny bit of fineprint. For regular files it shouldn't happen.

[cbrnr]

I guess you could get rid of this else: break block because you have the same thing right in the next else block. I.e. I think you can remove the first else block completely (the one belonging to the if _scan_forward(f)), and also remove the second else and dedent the two subsequent lines.

[tstenner]

Right, I've changed it.

[cbrnr]

Thanks @tstenner!