Increase regex readability

[Vampire]

No description provided.

[Vampire]

This PR makes all regex patterns that contain escapes raw String literals, so that you do not have to mentally double unescape to understand the pattern.

It also removes unnecessary start- and end-of-input anchors for regex that are anyway only used to match entire inputs.
If the anchors should stay, they should at least be changed to always be used, currently in most cases it was like ^foo|bar|baz$ instead of ^(?:foo|bar|baz)$, so only matching start-of-input for foo and end-of-input for baz.

[OptimumCode]

Hi, @Vampire. Thank you very much for the contribution

If the anchors should stay, they should at least be changed to always be used, currently in most cases it was like ^foo|bar|baz$ instead of ^(?:foo|bar|baz)$

I think it would be better to correct the patterns so the anchors are used correctly - just a precaution in case an incorrect method on Regex is used (that does not match the entire string against the pattern).

[Vampire]

I think it would be better to correct the patterns so the anchors are used correctly - just a precaution in case an incorrect method on Regex is used (that does not match the entire string against the pattern).

Sure, I hope I didn't break it now. :-D

[OptimumCode]

All looks good to me, thank you @Vampire. @krzema12 have you had a chance to take a look at the PR? Would you have any suggestions/objections?

[krzema12]

Generally looks sane, but I'm worried a bit about compatibility of the regexes across all targets. AFAIK, there are some minor discrepancies, and our tests aren't fully multiplatform yet (even if they are, not sure if coverage is good enough).

What I propose to do is to go to the modified file where we use something suspicious, like the non-capturing groups, and checking if the tests exercising the regexes are KMP already. If not, I'd recommend porting them first.

[Vampire]

Did you mistype and mean you are worried?
Because you said you are not.
Besides that, there were already non-capturing groups used in other regexes.
They are supported in JVM and JS.
On Native and WASM I'm not sure which flavor is used as the docs just say "this can be altered in the future to match the JVM behavior" so assuming it should be fine there too.

[Vampire]

From a quick look I'd say Native and WASM is an own Regex implementation also supporting non-capturing groups.
And I guess the intention is to be consistent with JVM implementation and if not fix that, hence the hint that it might change into that direction.

[krzema12]

Did you mistype and mean you are worried? Because you said you are not.

Yes,sorry!

Yeah, things should probably work fine, although as a rule of thumb, if we touch some code, I'd prefer to have coverage for all targets. I can migrate the two tests tomorrow (they're pretty simple), and this way we'll increase confidence regarding correctness of this change.