I've been trying to achieve this: I want to wrap words into tag and spaces (which may be multiple) in tag, assuming original text can contain html tags that should not be toched
This is <b>very bold</b> word.
convert to -->
<w>This</w><s> </s><w>is</w><s> </s><b><w>very</w><s> </s><w>bold</w></b><s> </s><w>word</w>
What is the right regEx to achieve that?
You should use two replacements >>
s.replace(/([^\s<>]+)(?:(?=\s)|$)/g, '<w>$1</w>').replace(/(\s+)/g, '<s>$1</s>')
Check this demo.
EDIT:
For more complex inputs (based on your comment below), go with >>
s.replace(/([^\s<>]+)(?![^<>]*>)(?:(?=[<\s])|$)/g, '<w>$1</w>').replace(/(\s+)(?![^<>]*>)/g, '<s>$1</s>');
Check this demo.
(?=...) part?
(?=\s) means "followed by whitespace"
data-foo="<bar>" tag parameter, which is possible, but very unlikely. As I don't know what kind of HTML source code OP needs to parse, it is hard to say how far with regex complexity we have to go. I believe my most recent code should work for OP.Regular expressions are not suited for every task. If your string can contain arbitrary HTML, than it's not possible to handle all cases using regular expressions, because HTML is a context-free language and regular expressions covers only a subset of them. Now before messing around with loops and a load of code to handle this, let me suggest the following:
If you are in a browser environment or have access to a DOM library, you could put this string inside a temporary DOM element, then work on the text nodes and then read the string back.
Here's an example using a lib I wrote some month and updated now which is called Linguigi
var element = document.createElement('div');
element.innerHTML = 'This is <b>very bold</b> word.';
var ling = new Linguigi(element);
ling.eachWord(true, function(text) {
return '<w>' + text + '</w>';
});
ling.eachToken(/ +/g, true, function(text) {
return '<s>' + text + '</s>';
});
alert(element.innerHTML);
Example: http://prinzhorn.github.com/Linguigi/ (hit the Stackoverflow 12758422 button)