Code cleanup: Replace regex-strings by native regex

Throughout the code of highlight.js, there are many places where strings containing regular expressions are used instead of natvie regex-statements. For example:

  // Common regexps
  hljs.IDENT_RE = '[a-zA-Z]\\w*';
  hljs.UNDERSCORE_IDENT_RE = '[a-zA-Z_]\\w*';
  hljs.NUMBER_RE = '\\b\\d+(\\.\\d+)?';

I think the code would be much more readable by using native expressions. The primary gain would be to get rid of the confusing double backslashes (or quadruple backslashes in other cases).

  // Common regexps
  hljs.IDENT_RE = /[a-zA-Z]\w*/;
  hljs.UNDERSCORE_IDENT_RE = /[a-zA-Z_]\w*/;
  hljs.NUMBER_RE = /\b\d+(\.\d+)?/;

I actually tried this change and a lot of tests fail, so other code would have to be changed as well.

The question is: Do you think this is a valuable change? I’m not sure if I have the time to work on it, but I might if you want it.

[joshgoebel]

I think if we’re making “deep dives” into fixing lots of issues with a certain grammar (like the recent Handlebars work, etc.) and while we’re there (in a separate commit, to make review easy) it makes sense to turn strings into regexes (where no fancy string addition is necessary), I’d be ok with doing it as it comes up in the natural course of things.

But there are also 3rd party languages… so the “API” we offer now of serving up strings for the main library might stay the same for a long time, since that would be a breaking change (didn’t think of that earlier).

Or just do it organically… there are so many strings that a few more aren’t making my day significantly better or worse. 😃

[egor-rogov]

And it’s worth asking if there might be some reason we don’t know about that strings were preferred in the first place. (other than the need for concatenation)

If such a reason exists, I’m not aware of it.

I think there are a lot higher priorities Usually I’m against sweeping changes to a codebase like this kind of thing

Agree.