Error when directory name contains non-ASCII characters? about b2_command_line_tool

[ppolewicz]

This looks like a problem with your terminal. Does your terminal allow you to list a file with that special character?

[Rolf-Smit]

With terminal are you referring to the the terminal being able to print a special character? Because based on the warning shown in the trace, that does not seem to be supported, however if you look at the first line it already breaks down earlier. So I don't think it is related to the terminal.

I found another example and got a better trace:

 'ordinal not in range(128)') 'ascii' codec can't encode character '\xe9' in position 54: ordinal not in range(128)
backblaze-download_1  | ERROR:b2sdk.sync.action:an exception occurred in a sync action
backblaze-download_1  | Traceback (most recent call last):
backblaze-download_1  |   File "b2sdk/sync/action.py", line 48, in run
backblaze-download_1  |   File "b2sdk/sync/action.py", line 252, in do_action
backblaze-download_1  |   File "b2sdk/sync/action.py", line 239, in _ensure_directory_existence
backblaze-download_1  |   File "os.py", line 225, in makedirs
backblaze-download_1  | UnicodeEncodeError: 'ascii' codec can't encode character '\xe9' in position 54: ordinal not in range(128)

_ensure_directory_existence seems to be a big clue.

[Rolf-Smit]

Did some further research and I found that the folder (in my previous example) that was uploaded to B2 contains:

U+0065 | e | LATIN SMALL LETTER E
U+0301 | ́  | COMBINING ACUTE ACCENT

[ppolewicz]

B2 cli can handle it if your terminal can - but it's set to 'ascii' which doesn't handle much

[ppolewicz]

or maybe it's your operating system that is not capable of writing those down - can you please try to create a file by yourself on that filesystem with the unicode char in the name and list it to make sure it works?

[Rolf-Smit]

So I was able to solve this by forcing my docker container to use UTF-8:

Dockerfile:

FROM ubuntu:22.04
# Set the locale to an UTF-8 one (instead of the default ASCII?)
RUN apt-get update && apt-get -y install locales && locale-gen en_US.UTF-8
ENV LANG en_US.UTF-8
ENV LANGUAGE en_US:en
ENV LC_ALL en_US.UTF-8
COPY b2-linux /backblaze/b2-linux
RUN chmod +x /backblaze/b2-linux
ENTRYPOINT ["/backblaze/b2-linux"]

[Rolf-Smit]

Would it make sense to document this behavior somewhere in the README.md for example? Specifically stating that the terminal (or even OS locale settings) may influence which files/folders can be downloaded/uploaded?