Using regex to "sort" its matches

[Dmitry Olshansky]

16-Apr-2013 22:59, Linden Krouse пишет:
> Is there a way to use the regex library to put matches of different
> regexs or classes into different slices? For instance, if I had the
> regular expressions "(?<=#)\w+\b" and "(?<=%)\w+\b", could I use them to
> match a string at the same time and stop if the first one is found and
> keep their results separate (as if the regex was
> "(?<=#)\w+\b|(?<=%)\w+\b)" and gave a slice for "(?<=#)\w+\b" and a
> different one for "(?<=%)\w+\b" ). I'm a little new to regexs for the
> record.

For that particular case you can as well use this pattern:
"(?<=#|%)\w+\b"

and then check what character is the last one before your match by hand:

foreach(m; match(<your string>, "(?<=#|%)\w+\b")){
	if(m.pre.back == '#')
		//put in one place
	else if(m.pre.back == '%')
		// put somewhere else
}

More general interface to do it is being discussed recently, in essence 
it gives you ability to switch over a bunch of regular expressions.

See:
https://github.com/D-Programming-Language/phobos/pull/1241

The machinery to efficiently combine regular expressions like that will 
come sometime later but the interface should stay the same.

-- 
Dmitry Olshansky