improved username generation for new users

[edchapman88]

✅ Checklist

• I have searched open and closed issues for duplicates.
• This is a request for a new feature in the Data Safe Haven or an upgrade to an existing feature.
• The feature is still missing in the latest version.
• I have read through the documentation.
• This isn't an open-ended question (open a discussion if it is).

🍓 Suggested change

1. Validate that the autogenerated usernames for new users meet the requirements specified by Entra ID (e.g. not all characters are allowed, such as "Ö").
2. Return the autogenerated usernames for the newly created users from the dsh users add command in a structured/concise format.

• Handle characters with accents
• Handle double surnames
• Handle double-barrelled surnames

🚂 How could this be done?

1. Extend the username generation function to sanitise inputs such that they meet the username criteria specified by Entra ID.
   • Optionally a different mechanism for generating usernames could be implemented (rather than <firstname>.<surname>), e.g. random words / numbers.
2. Implement a function that writes to stdout at the end of the command (after all users have been created) so it appears last in the calling shell. Write e.g. an ascii table similar to that of the dsh users list command. This is more user friendly where the new usernames must be passed on to another process (e.g. saved in a spreadsheet in the case of DSGs).

[jemrobinson]

This isn't a problem with a programmatic solution. I suggest that the DSH code should add a field for "username" and let the TRE operator decide.

(Note that we still need to come up with a solution for TRESA - one option could be to ask the users to decide for themselves).

[helendduncan]

Agree with issues of accents - also names with spaces fail (didn't try hyphens)

When processing a number of users, it is useful that if one fails, the process continues rather than stopping and needing re-starting

Current time to generate one user and verify they exist on Entra is approximately 90 seconds - is this something that can be sped up? Parallelised?

Is the country code needed in the input? It's a small barrier when generating users - external programs can fairly readily be made to extract the country code from the phone number, but then is that something that should or could be done within the DSH code?