Fix HTML Entities regex

[JonathanWolfe]

Caused JonathanWolfe/language-polymer#1. Just need a simple lookahead assertion.

[hediyi]

Could you also update the specs?

[hediyi]

What do you think of rewriting the grammar to

        'match': '(&)([a-zA-Z0-9]+|#[0-9]+|#x[0-9a-fA-F]+)(;)'
        'captures':
          '1':
            'name': 'punctuation.definition.entity.begin.html'
          '2':
            'name': 'entity.name.entity.other.html'
          '3':
            'name': 'punctuation.definition.entity.end.html'
        'name': 'constant.character.entity.html'

[JonathanWolfe]

I think that's the better way forward since entities aren't able to wrap around anything.

[JonathanWolfe]

Swapped to use match instead of the begin/end semantic. Also corrected spec changes.

[winstliu]

/cc @DavidePastore I may have to revert your change in order to fix the recent regressions regarding ampersands (such as #141). Any suggestions?

[DavidePastore]

@50Wliu thanks for the cc. The fact is that &a has a particular meaning in my implementation: you're in the middle of an entity and you could use this information to trigger autocomplete. Maybe we could try to preserve the first pattern and edit the second.

[JonathanWolfe]

@DavidePastore - I'm not a regex guru but I can't think of any way to have that functionality and not break the rest of the spec. It will constantly try to eat the rest of the document (or line at a minimum with a change) thinking you're still writing an entity. It is possible to preserve autocomplete but it would require you to write &; then go back and fill in the entity.

[winstliu]

I'm reverting this for now and will investigate alternative options later.

[DavidePastore]

@JonathanWolfe - Yes, you're right. I'm not a regex guru either.
@50Wliu You did the right thing 👍