As mentioned in the comments.
$PATH in Linux is used to find a executable. In this case that's Python.
After this executable is found, it will be executed "in" the current working directory and work from there.
Meaning when Python gets the parameter one/toto.py it will start looking from ./ after your folder and file.
You could create a wrapper script, place that under /usr/bin/mywrap.py, mark it as a executable and utilize for path in sys.path: and try to find your module and dynamically import it and do a "hand over" to it.
There is no magic in the Python binary that will travers $PATH since this is a shell variable used to find binaries in the operating system. Much like PATH in windows is used for the same purpose.
~/python/one/toto.py. the$PATHvariable is parsed by the shell itself when trying to find executables. You could create a python script that you mark as executable that tries to do this for you. by usingos.path.abspath(...)to and usesys.pathto find modules that it can import and "hand over" to.PATHvariable will only affect where the shell finds the executable,pythonin your case. It doesn't affect wherepythonwill look for files to run