I'm trying to use regex.match() for a tokenizer.  So I want to compare my input to a bunch of different regex'es, expecting one to hit and the rest not to.

Problem is, I can't seem to figure out the "right" way to ask if a hit occurred.  My code is roughly this:

BEGIN CODE
while(input.length > 0)
{
	auto r1 = regex("^...something nasty...");
	auto r2 = regex("^...something worse...");

	auto tmp = match(input, r1);
	if(<check r1 here>)
	{
		ReportTokenType1(tmp.hit())
		input = tmp.post();
		continue;
	}

	... and so on ...
}
END CODE

What do I put inside the if()?

P.S. Does anybody know why dmd complains "cannot evaluate at compile time" when I set those regex objects to "static invariant" so I'm not rebuilding them with every pass?

On Tue, Jun 2, 2009 at 12:16 AM, Russell Lewis <webmaster@villagersonline.com> wrote:

>
> P.S. Does anybody know why dmd complains "cannot evaluate at compile time" when I set those regex objects to "static invariant" so I'm not rebuilding them with every pass?
>

Because it wants you to have a constant expression to initialize a static variable, and so it tries to evaluate regex(...) at compile time.  You know, there's a much easier way to prevent them from being reinitialized every iteration - move their declaration above the loop.

Jarrett Billingsley wrote:
> On Tue, Jun 2, 2009 at 12:16 AM, Russell Lewis
> <webmaster@villagersonline.com> wrote:
> 
>> P.S. Does anybody know why dmd complains "cannot evaluate at compile time"
>> when I set those regex objects to "static invariant" so I'm not rebuilding
>> them with every pass?
>>
> 
> Because it wants you to have a constant expression to initialize a
> static variable, and so it tries to evaluate regex(...) at compile
> time.  You know, there's a much easier way to prevent them from being
> reinitialized every iteration - move their declaration above the loop.

Of course!  But then, the code is harder to read.  I'll bite the bullet and move it above the loop, but it's ugly.

And yeah, I know about CTFI and the issues there...I just thought that regex was designed to allow CTFI.  I guess I was wrong?

BTW: I found out how to check if there was a match or not: the empty() function.

On Tue, Jun 2, 2009 at 11:47 AM, Russell Lewis <webmaster@villagersonline.com> wrote:

>
> Of course!  But then, the code is harder to read.  I'll bite the bullet and move it above the loop, but it's ugly.
>
> And yeah, I know about CTFI and the issues there...I just thought that regex was designed to allow CTFI.  I guess I was wrong?

For future reference, it's "CTFE", and I wasn't aware that regex was designed for it.

Jarrett Billingsley wrote:
> On Tue, Jun 2, 2009 at 11:47 AM, Russell Lewis
> <webmaster@villagersonline.com> wrote:
> 
>> Of course!  But then, the code is harder to read.  I'll bite the bullet and
>> move it above the loop, but it's ugly.
>>
>> And yeah, I know about CTFI and the issues there...I just thought that regex
>> was designed to allow CTFI.  I guess I was wrong?
> 
> For future reference, it's "CTFE", and I wasn't aware that regex was
> designed for it.

I kept looking at that acronym, and it just didn't look right...thanks for the correction. :)