Comment 2 for bug 1717935

IMHO this bug is important, because we're not in the 8+3 names' era anymore.

I've been playing around a bit with the code and it would seem this is an encoding issue. The Backblaze instructions on the subject [1] do not seem correct, as experimentally, at least the X-Bz-File-Name header should *not* be encoded (probably others as well). That leaves mainly the _get function to be encoded. A poor man's encoding (replacing spaces with "%20" or "+") seems to work, but a full solution should properly encode all URLs sent to B2.

[1] https://www.backblaze.com/b2/docs/string_encoding.html