Do word boundaries work on symbol characters?

问题

I'm trying to implement word boundaries in my emoticons feature for a chat. But for some reason I can't seem to get the word boundaries to work. I am new to regex.

So when I do:

var reg = /\b\Hi\b/gi;
var str = 'HiHiHi Hi HiHiHi Hi';
alert(str.replace(reg, ''));

This happens: Jsfiddle

It actually works fine, and does remove those 2 Hi's that are standing alone.

But when I change the reg to an escaped smiley and then change the string:

var reg = /\b\:\)\b/gi;
var str = 'HiHi:) :) HiHiHi :)';
alert(str.replace(reg, ''));

This happens: Jsfiddle

It just doesn't work. The string stays the same. Is it that word boundaries can't be used on symbols? If so, how does Facebook do it on their chats?

回答1:

As \b will not work in this case, you could use:

var re = /(\s|^):\)(?!\S)/g;
var str = 'HiHi:) :) HiHiHi :)';
alert(str.replace(re, '$1'));

Which works like a space boundary.

You can add several smileys to it like so:

/(\s|^)(?::\)|:\(|:\||:\]|:\[)(?!\S)/g;

回答2:

Word boundaries \b represent a zero-width boundary between word characters \w (in javascript, [A-Za-z_]) and non-word characters \W (everything else).

Because of this, there will not be a boundary between two emoticons or when the emoticon is surrounded by spaces, punctuation, etc.

The simplest regex would be /[:;]-?[()]/gi, which supports smileys ) and frownies ( with optional dashes and eyes : or winks ;.

Edit:

This will require either a space or the beginning of the string (as a capturing group since Javascript doesn't support look-behinds), then it uses the above regex, then it must be followed by the end of string or whitespace.

var reg = /(\s|^)[:;]-?[()](?:\s|$)/gi;
var str = 'HiHi:) :) HiHiHi :)';
alert(str.replace(reg, '$1'));

Should replace in these situations: :-), cool :( not!

Should not replace in these situations: Digits:(0,1,2,3), hi :(.

来源：https://stackoverflow.com/questions/10983932/do-word-boundaries-work-on-symbol-characters