[component] Restrict character set for component.ID name

[mwear]

Description

While working on #10495 we discovered that there are not any restrictions on the name field of a component.ID. There are restrictions on the type field introduced in #9208. This PR adds similar restrictions to name.

A type must

• have at least one character,
• start with an ASCII alphabetic character and
• can only contain ASCII alphanumeric characters and '_'.

I found that we need a slightly different set of rules for name as some tests use a digit and others use a uuid as a name. A name is still optional, but if it's provided it must:

• have at least one character,
• start with an ASCII alphanumeric character and
• can only contain ASCII alphanumeric characters, '_', and '-'.

I'd be willing to adjust these restrictions if anyone has any opinions on what should or should not be allowed.

Link to tracking issue

Fixes #10673

Testing

Unit tests

Documentation

Code comments

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.24%. Comparing base (7d5b1ba) to head (f1d7584).
Report is 22 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #10674      +/-   ##
==========================================
- Coverage   92.38%   92.24%   -0.15%     
==========================================
  Files         403      403              
  Lines       18729    18726       -3     
==========================================
- Hits        17303    17274      -29     
- Misses       1066     1097      +31     
+ Partials      360      355       -5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[axw]

Looks great, thank you! Just one thought about the allowed characters.

[axw]

What do you think about opening this up to all Unicode letters and numbers? i.e. swapping [0-9] for \pN (numbers) and [a-zA-Z] for \pL (letters). Perhaps also replacing the hyphen and dash with \pP (punctuation).

I think ASCII is reasonable for type, since there's a specific set of them. On the other hand, the name component is arbitrary and user-defined. So, for example, allowing Chinese characters may be more inclusive of native Chinese speakers/readers/writers.

[mx-psi]

Can we start with introducing this restriction only on UnmarshalText?

We can later think about adding Name or not, but that is the part that impacts end-users of the binary and I think we can start there

[mwear]

Updated with both suggestions @mx-psi and @axw.

[axw]

Thank you for the changes, looks great!