Error with UnicodeDecode

[namepunk]

I have used the drive as a SAMBA file share for document folders. Folders are used by windows customers. If i try to backup this folder receive an error
_root@test:~# attic create -s -v /mnt/backup/attic::test2 /test/docs/test
/test/docs/test
Traceback (most recent call last):
File "/usr/local/bin/attic", line 3, in
main()
File "/usr/local/lib/python3.4/dist-packages/attic/archiver.py", line 730, in main
exit_code = archiver.run(sys.argv[1:])
File "/usr/local/lib/python3.4/dist-packages/attic/archiver.py", line 720, in run
return args.func(args)
File "/usr/local/lib/python3.4/dist-packages/attic/archiver.py", line 129, in do_create
self._process(archive, cache, args.excludes, args.exclude_caches, skip_inode s, path, restrict_dev)
File "/usr/local/lib/python3.4/dist-packages/attic/archiver.py", line 170, in process
archive.process_item(path, st)
File "/usr/local/lib/python3.4/dist-packages/attic/archive.py", line 375, in p rocess_item
item.update(self.stat_attrs(st, path))
File "/usr/local/lib/python3.4/dist-packages/attic/archive.py", line 370, in s tat_attrs
acl_get(path, item, st, self.numeric_owner)
File "attic/platform_linux.pyx", line 101, in attic.platform_linux.acl_get (at tic/platform_linux.c:2480)
File "attic/platform_linux.pyx", line 65, in attic.platform_linux.acl_numeric ids (attic/platform_linux.c:1698)
UnicodeDecodeError: 'ascii' codec can't decode byte 0xd0 in position 97: ordinal not in range(128)

[ThomasWaldmann]

the problem is caused by a non-ascii access control list (maybe a user or group name with non-ascii chars).

[namepunk]

I know it. I'm ready to backup without ACL but how? In documentation is not written about ACL.

[ThomasWaldmann]

There is no attic option to switch that off. You'ld have to edit attic/platform.py so it runs into the "else" case there.

[namepunk]

Thank you. It works.

[anarcat]

@namepunk can you provide the filename that triggers the crash? you should be able to see where it fails using -vv...

i wonder if it's a good idea to assume the ACLs are encoded in utf-8...

[ThomasWaldmann]

@anarcat this is not the filename, but likely some non-ascii user or group name inside the acl.

[anarcat]

from irc://irc.freenodenet/#borg:

15:37:33 <anarcat> maybe the quick fix is to add a --acl flag that'S enabled by default
15:37:39 <anarcat> a bit like what i did with --progress
15:38:09 <anarcat> 0xd0 is        320   208   D0     Ð     LATIN CAPITAL LETTER ETH
15:38:15 <anarcat> in latin1
15:38:17 <anarcat> aka wtf
15:44:05 <ThomasWaldmann> >>> b'\xd0\x80'.decode('utf-8')
15:44:05 <ThomasWaldmann> 'Ѐ'
15:44:16 <anarcat> aaah
15:44:24 <ThomasWaldmann> or anything else following it
15:45:16 <anarcat> weird
15:46:27 <anarcat> >>> 'È'.encode('utf-8')
15:46:27 <anarcat> b'\xc3\x88'
15:46:56 <anarcat> >>> 'Ѐ'.encode('utf-8')
15:46:56 <anarcat> b'\xd0\x80'
15:47:16 <anarcat> even more clearly: >>> 'Ѐ' == 'È'
15:47:16 <anarcat> False
15:47:19 <anarcat> wtf
15:47:28 <ThomasWaldmann> hehe
15:47:40 <ThomasWaldmann> scammers paradise
15:54:21 <ThomasWaldmann> >>> b'\xd0\x00'.decode('utf-8', 'backslashreplace')
15:54:21 <ThomasWaldmann> Traceback (most recent call last): File "<stdin>", line 1, in <module>
15:54:21 <ThomasWaldmann> TypeError: don't know how to handle UnicodeDecodeError in error callback
15:54:30 <ThomasWaldmann> wat?
15:58:39 <ThomasWaldmann> omg, a bug in 3.4.3
16:03:27 <ThomasWaldmann> Changed in version 3.5: The 'backslashreplace' error handlers now works with decoding and translating.
16:03:34 <ThomasWaldmann> no, brand new
16:04:23 <ThomasWaldmann> guess i want surrogatescape
16:16:15 <ThomasWaldmann> how did that code ever work
16:17:52 <ThomasWaldmann> anarcat: can you have a look at acl_use_local_uid_gid