Classification in laspy doesn't match with lasinfo about laspy HOT 6 CLOSED

It seems that its a problem with the laszip backend.

I can reproduce with laszip, but with lazrs I can't

from laspy.

Ok, I found the problem, basically, the file problem comes from the fact that the file is quite big

• 405MB of LAZ: 81_719_107 points of 34 bytes each so 2_778_449_638 bytes (2.147483647 GB)

The internal buffer used to get the uncompressed bytes given by laszip uses an 32bit signed int to count the
number of bytes however (81_719_107 * 34) >= int32_max, which made the buffer 'stop' and only output 0s.

When using chunking, since after reading each chunk, that buffer gets emptied you don't reach the int32 limit
and the file is read correctly. (unless you try to read 63_161_284 which makes you go over the limit)

I have a fix locally for this, which allows to laspy.read the whole file and
get the expected result.

It will require a new version of the laszip-python backend

from laspy.

a new release of laszip-python is available which fixes the problem

from laspy.

It happens only with this file ?

from laspy.

Hi @tmontaigu !
Thanks a lot for your feedback and glad you could identify the issue.

If I understand well, a quick fix on the user side would be to switch to the lazrs backend.

Let me know if I can help for the new version of laszip-python.

from laspy.

Thank you very much!
I can report that it works on my side with the new release (v0.2.3)

from laspy.