LIEF legacy-install-failure issue

[Shoot-to-root]

Hi,

I'm trying to install via pip but this error came up (as shown in pic below).

I tried installing lief separately but another subprocess-exited-with-error and metadata-generation-failed error came up when I run pip install famework. I'm using Python 3.8.9 and the latest pip version. Please help, thanks!

[zRapha]

Hi, thanks for using FAME! Have you tried to run it with an earlier version of pip (such as pip install pip~=23.0)?

[Shoot-to-root]

Yes, I tried it with earlier version, it didn't work so I updated the pip version, but still doesn't work.

[zRapha]

It seems to be a Python issue with more recent versions, have you tried using Python3.7 as pointed out in the requirements? I cannot reproduce the issue when using Python3.7 and pip==24.0.

[Shoot-to-root]

Hi, sorry for the late reply. I managed to test it out today with Python 3.7.17 and pip==24.0, but there's still error. The error seems to be caused by gym:

Unrelated to the issue, I'm also curious to how you build back the generated malware sample from extracted features? Or do I misunderstand it, you didn't extract the feature, just inject perturbations into the files? If you can point me to the code that will be awesome, thanks!

[zRapha]

I am not able to test all versions now but I have an env with Python 3.7.15 and pip==22.0.4 and it seems to be installing fine. Have you checked if your pip is calling the proper Python version (in case you have more than one) as in:

python3.7 -m pip install famework

Re the question, the binary features are parsed and then built using the lief library. You can check the building function here.

[Shoot-to-root]

I managed to solve the problem with the help pf stack overflow! Turns out the gym 0.21 installation is broken with more recent versions. Now I can run it, thanks for your help!

However, when I used my malware samples to run, the program dismissed them: Object dismissed: samples/malware_set/. The files are PE32 executables. Are there any specific requirements for the input data? Or could you be so kind and send me the link to the datasets you used?

[zRapha]

Good to hear, WinPEs should work, however that message is raised if the input provided is not a file os.path.isfile(sample)==False

[Shoot-to-root]

I checked the file type, but the error is still raised:

[zRapha]

I see, however without the binary sample I cannot reproduce any behavior. This seems to be a problem outside of FAME since the operation to check the file type is a regular built-in function. I would proceed to close this request now since the original problem has been solved but feel free to re-open if there are issues related with the installation.