Fails when LC_CTYPE is not explicitly UTF-8

[infinity0]

$ LC_CTYPE="C.UTF-8" python2.7 -m unittest discover
......
----------------------------------------------------------------------
Ran 6 tests in 0.020s

OK

$ LC_CTYPE="POSIX" python2.7 -m unittest discover
.E....
======================================================================
ERROR: test_aleksandra_slapik_44 (test.test_sagenb_writer.ReadSageNB)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/infinity0/var/lib/sage/sagenb-export/test/test_sagenb_writer.py", line 39, in test_aleksandra_slapik_44
    ipynb.write(self.tmp_filename(u'WDI projekt - R\xf3\u017cankowski, Kie\u0142pi\u0144ski, Kozok.ipynb'))
  File "/home/infinity0/var/lib/sage/sagenb-export/sagenb_export/ipynb_writer.py", line 58, in write
    write(ipynb, filename)
  File "/usr/lib/python2.7/dist-packages/nbformat/__init__.py", line 163, in write
    with io.open(fp, 'w', encoding='utf-8') as f:
UnicodeEncodeError: 'ascii' codec can't encode characters in position 41-42: ordinal not in range(128)

----------------------------------------------------------------------
Ran 6 tests in 0.020s

FAILED (errors=1)
1

This is because you pass a unicode string type as a filename. Python will implicitly try to encode this using LC_CTYPE, which doesn't support that particular value. You can fix this in several ways, I'm not sure which way you prefer:

1. Say that IpynbWriter.write can only accept byte filenames, and do the encode('utf-8') in the test. This is probably the simplest.
2. Say that IpynbWriter.write can accept unicode string filenames too, and do encode('utf-8') in there, if it's a string type.
3. Add a filename_encoding param to IpynbWriter.write.

[infinity0]

On further investigation it looks like this is more a problem with nbformat.write - even if you pass a byte path here, that function will not accept it - it only accepts string filenames or file objects. I'll go file a bug there too instead.

[infinity0]

(Note that there is still an issue in sagenb-export; but your solution will depend on how nbformat fix things on their end.)

[slel]

AFter nbformat issue jupyter/nbformat#76
was closed by pull request jupyter/nbformat#83
what remains to be done here?