Replace Linkifier RegExp to RFC3986 based one

[tmyt]

Current Linkifier has some problems.
The PR replace Linkifier RegExp to RFC3986 based one, and aim to resolve issues #599 and #601.

These URLs are now Linkify.

• http://example.enterprises/
• http://[::1]/
• ftp://example.com/
• chrome-devtools://devtools/
• mailto:[email protected]
• about:blank

Could you discuss "should" or "should not" linkify the URI without "//"?

[coveralls]

Coverage increased (+0.1%) to 68.038% when pulling b2a4e9e on tmyt:propose/rfc3986 into 6d25706 on sourcelair:master.

[Tyriar]

Neat 😃

Did you write this up yourself or did you borrow it from somewhere as we might need to include the license.

Also does this have unicode support or is that still unresolved?

[tmyt]

The code wrote myself with refer the Appendix A of IETF RFC 3986 documentation.

Strictly, Unicode characters is not allowed in URI and should be encode with Percent or Punycode. Then this implementation does not support Unicode chars.

[Tyriar]

@tmyt strictly, same with brackets which get encoded. Browsers and such will try to help the user out by allowing unencoding characters in URLs, there are also legal unicode domain names now. I wonder if doing this is even worth it as the real fix imo is to implement a proper parser to get the ideal behavior as in #583

[tmyt]

@Tyriar Ah, I did not check #583 that implement in mean time, the PR is not needed. If not needed, please reject this.

[Tyriar]

@tmyt while this would be an improvement, it's probably a good idea to not switch to this and then a parser. Thanks for understanding 😃 something like this is what I'm thinking for the parser https://github.com/Microsoft/vscode/blob/master/src/vs/editor/common/modes/linkComputer.ts