regex date capture - greed, reluctance, and precedence problem

[Chris Treglio]

I've got a long list of strings with dates in the format dd-mmm-yyyy which I'm trying to capture. I'd like to be able to handle missing leading zeroes in the day part (i.e. properly capture 01-Jan-2011 and 1-Jan-2011).

My current code doesn't handle leading zeroes.

		String reg = ".*((?:[12][0-9]|3[01]|0[1-9])-" + // day part  
		 "(?i:jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)-" +  // Month part
		 "20\\d\\d).*"; // year part

		p = Pattern.compile(reg);

		Matcher m = p.matcher(source);

		m.matches() ;

		try {
			return m.group(1);
		} 
		catch (IllegalStateException e)
		{
			return "";
		}

I thought by changing the day part to ".*((?:[12][0-9]|3[01]|0?[1-9])-", I would make the leading zero "greedy" optional, and capture it if it's there. It does not. And furthermore, it turns dates like "12-Mar-2011" into "2-Mar-2011". Obviously, I'd want matches in the teens, twenties, or thirties to get captured too.

what am I doing?

[Wouter Oet]

You're trying to reinvent the wheel. Why use regex to parse dates when you can use DateFormat/SimpleDateFormat?

[Chris Treglio]

I'm not really trying to turn a String into a Date object, I'm trying to pull out String dates from a longer String filled with other stuff. My String sources are like "bla bla blah blah 02-Apr-2011 blah bla blah".

Can you do that with the DateFormat/SimpleDateFormat?

[Wouter Oet]

Aha. I'm not sure if that is possible.

[Mike Simmons]

I'm guessing the issue is that the initial .* is greedy, and that wins out over your other intentions here. It will probably overlook a leading 1, 2, or 3 as well, not just leading 0. As long as there's at least one digit after, to match the rest of the expression.

I suggest either:

(a) replace .* with .*?, which is reluctant

or

(b) drop the .* entirely, and replace matches() with find().