feat: regex sampler

[llllvvuu]

Vendors a tiny dependency (total dependency tree <10 KB
uncompressed).

Other than removing unused code from the dependency, the only change
made is to change the following line:

https://github.com/fent/randexp.js/blob/2b35ea607883fa8cafa63bd3bdb5c12d695f873a/lib/randexp.js#L109

to a separately customizable callback, in order to adjust the length to the minLength and maxLength constraints.

What/Why/How?

Fixes #152 and Redocly/redoc#2312

Reference

#152

Testing

Added unit tests and fuzz tests for both basic and advanced cases.

Screenshots (optional)

N/A

Check yourself

• Code is linted
• Tested
• All new/updated code is covered with tests

Security

• Security impact of change has been considered
  Regex denial-of-service applies as usual, ensure code is not running on untrusted input in a place where a hanging process would be harmful.
• Code follows company security practices and guidelines

[RomanHotsiy]

This looks more or less good but I'm not sure if the complexity involved is worth it.
People can always provide example manually which will be better quality.

[RomanHotsiy]

This is strange to include this as dependency but not use it directly. I think it's better to add 2 other dependencies here.

But I'm also concerned by those dependencies. They are not very active. Last update 5 years ago.

[llllvvuu]

That's a valid concern.

One way to reduce the complexity would be to simply ignore the min_length and max_length for regexes. Another way could be to copy faker.js fromRegExp, which is less complete but has no dependencies:

https://github.com/faker-js/faker/blob/fd6557e4b3716ebe41aa38a58b4ae112e6baf0d2/src/modules/helpers/index.ts#L401-L580

They also had a pretty long discussion of what to do about this: faker-js/faker#1569

[RomanHotsiy]

I would cope the faker-js one. This looks good.

[llllvvuu]

Done! It's definitely less capable than randexp but I have added to the README a link to their docs which note the limitations.

[RomanHotsiy]

Oh, so this tool is not deterministic 😢

Can we adjust it to produce same result always.

One of the features of this tools is "deterministic" (check the readme, it's first on the list).

There are multiple reasons to want determinism in such tool.

[llllvvuu]

Yes, adjusted

[RomanHotsiy]

Let's make it a bit more consistent.

let's move it from vendor folder into samplers/string-regex.js and rename the internal input from fromRegExp to regexSample.

[llllvvuu]

OK, adjusted. I could not import as just regexSample due to naming conflict so I did a namespace import

[RomanHotsiy]

we don't need to adjust it. Let's leave it is.
This is a string with format: regex which means it has to be a valid regex. I doubt someone will provide regex for regex.

[llllvvuu]

Ah, I misunderstood what format: regex meant. Thank you, fixed

[RomanHotsiy]

@Redocly/keyboard-warriors please, merge it when you're ready 🙏

[kanvik175]

we really need this feature 😢. when can you merge it? 🙏

[lornajane]

Thanks @llllvvuu for adding this feature!