Re: [Boost-users] boost.regex doesn't implement character set complements?

From: Eric Niebler (eric_at_[hidden])
Date: 2005-06-29 00:17:53

Robert Mathews wrote:
> I was trying to write "Match everything *except* Bar".
> 
> I wrote this:
> 
> bool aResult = boost::regex_match(L"Foo", boost::wregex(L"[^(Bar)]"));
> 
> This doesn't match. But if I wrote the same thing in Perl, it does work, ie:
> 
> print "match\n" if( "Foo" =~ "[^(Bar)]");
> 
> It looks a lot like sets can't contain subexpressions in boost::regex ... is
> that really true?
> Is there some other way to write the same thing?

I guarantee that your perl code isn't doing what you think it's doing. This:

    "Foo" =~ /[^(Bar)]/

will succeed because 'F' is not one of '(', 'B', 'a', 'r', or ')'. The 
same pattern means the same thing to Boost.Regex, but you're getting a 
different result because the regex_match algorithm only reports success 
if the pattern matches all of the input, which it doesn't in this case. 
Had you used regex_search, you'd have gotten the same results as perl. 
But it still wouldn't be doing what you want.

What you're looking for is a negative lookahead assertion. If you want 
to match three characters that are not "Bar", you can do it like:

     "(?!Bar)..."

(?!Bar) asserts that the next three characters are not "Bar", but 
doesn't consume them. The ... matches the next three characters (which 
are not "Bar") and consumes them.

HTH,

-- 
Eric Niebler
Boost Consulting
www.boost-consulting.com