If you have a non-ASCII .py file without encoding declaration, then you can't normally import it:

$ python --version
Python 2.7.6

$ python -c 'import test'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "test.py", line 1
SyntaxError: Non-ASCII character '\xc2' in file test.py on line 1, but no encoding declared; see http://www.python.org/peps/pep-0263.html for details


However, "python -m" happily imports such files:

$ python -m test
¡Hello world!

Well, we aren't going to change 2.7 to have this case start throwing an error, since someone may be depending on it.  So I'm not sure there's anything to do here.

There are several bugs in processing encoding declaration (issue18961, issue18873) and yet several were fixed last time. Perhaps this issue or issue19942 relate to them.

Quite plausibly a bug in the pkgutil import system emulation. Does the same inconsistency appear in 3.3 with non-UTF8 bytes?

3.x is afflicted.

On 3.11 both are working (on a Mac):

cpython-1 % python -m tt  
¡Hello world!
cpython-1 % ./python.exe -c 'import tt'
¡Hello world!

test.py is a UTF-8 file, which is the default source encoding in Python 3. It fails as expected if the test script is encoded differently, such as Latin-1, unless the source encoding is declared.

I confirm: Python 3.10 works as expected.

Python 3.10 fails with the same SyntaxError using "python script.py" or "python -m script" if the script contains non-ASCII characters but is not encoded to UTF-8.

vstinner@apu$ python3 test.py
  File "/home/vstinner/test.py", line 1
    print('�Hello world!')
                         ^
SyntaxError: (unicode error) 'utf-8' codec can't decode byte 0xa1 in position 0: invalid start byte

vstinner@apu$ python3 -m test
Traceback (most recent call last):
  (...)
  File "/home/vstinner/test.py", line 1
    print('�Hello world!')
                         ^
SyntaxError: (unicode error) 'utf-8' codec can't decode byte 0xa1 in position 0: invalid start byte