How to parse the first consecutive numeric substring from a character?

[xyxu]

I want to extract the first numeric substring that shows up in a character variable. For example, 

data have;
 	infile datalines delimiter = "," missover;
	input string $;
	datalines;
	abc 12345 &*ag21
	bld#5432 
	8642 12 asd
	kmdsf 657
run;

data want;
	length string $30. substring $30.;
	infile datalines dsd dlm='|' truncover ;
	input string $ substring $;
	datalines;
	abc 12345 &*ag21|12345
	bld#5432|5432
	8642 12 asd|8642
	kmdsf 657|657
run;

Is there a good way to do this in a data step? 

[AMSAS · Accepted solution]

You want to look at PRXPOSN Function - Extracting First and Last Names 

data have;
 	infile datalines delimiter = "," missover;
	input string $;
	datalines;
	abc 12345 &*ag21
	bld#5432 
	8642 12 asd
	kmdsf 657
run;

data results ;
	length string $30. substring $30.;
	infile datalines dsd dlm='|' truncover ;
	input string $ substring $;
	datalines;
	abc 12345 &*ag21|12345
	bld#5432|5432
	8642 12 asd|8642
	kmdsf 657|657
run;

data wamt ;
	retain patternID ;
	set have ;
	if _n_=1 then do ;
		patternID=prxparse('/([0-9]+)/') ;
	end ;
	found=prxmatch(patternID,string) ;
	want=prxposn(patternID,1,string) ;
	put string= found= want= ;

run ;