bug: Unable to install awscli plugin with recent Swedish locale

[jmidgren]

Describe the Bug

I run Kubuntu 20.04 in Swedish and I'm unable to install the awscli plugin:

$ asdf plugin-add awscli
awscli is invalid. Name must match regex ^[a-zA-Z0-9_-]+$

After some digging I found that there may be a problem with the swedish locale not including the letter "w" properly. (We swedes have always had a complicated relationship with "w", e.g. in collation it should according to some be sorted under "v".) After reading more about it, it struck me that expressions like a-z very much depends on the language implied. If languages have additional characters within the a-z range or if alphabets are ordered differently, expressions like a-z do not make sense.

One way to solve this is to specify the language every time such expressions are used. E.g. in plugins.bash, grep could be preceded with LANG=C.

Relevant posts for more background:

• Stack Overflow: Regular expression [A-Za-z] seems to not include letter W and w
• Glibc bug report: Handle [a-z] and [A-Z] in consistent portable fashion regardless of locale

Steps to Reproduce

LANG=sv_SE.UTF-8 asdf plugin-add awscli

Expected Behaviour

Plugin is being added or the message Plugin named awscli already added shows if it's already installed.

Actual Behaviour

asdf fails with return code 1 and the message

awscli is invalid. Name must match regex ^[a-zA-Z0-9_-]+$

Environment

OS:
Linux lap5cg1111r2p 5.13.0-44-generic #49~20.04.1-Ubuntu SMP Wed May 18 18:44:28 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux

SHELL:
zsh 5.8 (x86_64-ubuntu-linux-gnu)

ASDF VERSION:
v0.10.1-f8109e0

ASDF ENVIRONMENT VARIABLES:
ASDF_DIRENV_BIN=/home/johannem/.asdf/installs/direnv/2.31.0/bin/direnv
ASDF_DIR=/home/johannem/.asdf

ASDF INSTALLED PLUGINS:
ansible-base                 https://github.com/amrox/asdf-pyapp.git master 45a4efd
awscli                       https://github.com/MetricMike/asdf-awscli.git main ec4ccdc
direnv                       https://github.com/asdf-community/asdf-direnv.git master 7dcd571
golang                       https://github.com/kennyp/asdf-golang.git master cc8bc47
golangci-lint                https://github.com/hypnoglow/asdf-golangci-lint.git master 5b537fa
nodejs                       https://github.com/asdf-vm/asdf-nodejs.git master f9957f3
yadm                         https://github.com/particledecay/asdf-yadm.git master 7504791
yq                           https://github.com/sudermanjr/asdf-yq.git master 772992f

Also:
glibc (libc-bin, locales etc): 2.31-0ubuntu9.9
grep: 3.4-1

asdf plugins affected (if relevant)

No response

[jmidgren]

Another way to solve the problem would be to use regex character classes instead, e.g. [[:alpha:][:digit:]_-]. This is probably not what we want unless plugins with names in other alphabets should be considered. However, languages that don't include the letters a-z will then not work with existing plugin names...

[Stratus3D]

Thanks for the bug report @jmidgren ! This is new to me.

I need to check past issues before fixing this, but I think our reasoning for adding the ^[a-zA-Z0-9_-]+$ regex check for plugin names was to prevent whitespace from being used in a plugin name, as that caused issues in the .tool-versions files. If that is the only reason we added the regex check, expanding to something like ^[[:alpha:][:digit:]_-]$ should be fine.

[Stratus3D]

I've got a PR out to fix this issue. Feel free to give it a try to confirm it works @jmidgren

[Stratus3D]

I marked this as fixed by my PR. Sometime this week I will release a new patch version of asdf.