Problem with call prxnext if the first match is zero-length

js5 — Tue, 14 Jun 2022 06:28:21 GMT

Hello,

I am having issues parsing strings with leading blanks with call prxnext if the first match is zero-length. This is the string to parse:

 foo        foo foo foo foo

and here is the code snippet:

data read;
	length var $50;
	input var char50.;
	datalines;
 foo        foo foo foo foo
 ;
run;

data regex;
	set read;
	regex = prxparse("/(?:^| )([a-z\-]*)(?= |$)/");
	start = 1;
	stop = length(var);
	call prxnext(regex, start, stop, trim(var), pos, len);

	do while (pos GT 0);
		var_tall = prxposn(regex, 1, trim(var));
		output;
		call prxnext(regex, start, stop, trim(var), pos, len);
	end;
run;

The problem is that:

- only twelve matches are found while there are thirteen present

- the first zero-length match seems to overwrite the first non-zero one

regex101.com finds 13 matches as expected. Is this a bug in SAS? Or in my code? Thank you for your help in advance.

Re: Problem with call prxnext if the first match is zero-length

js5 — Tue, 14 Jun 2022 06:41:04 GMT

This works for some reason:

data read;
	length var $50;
	input var char50.;
	datalines;
 foo        foo foo foo foo
 ;
run;

data regex;
	set read;
	regex = prxparse("/(?:(?<=^)|(?<= ))([a-z\-]*)(?= |$)/");
	start = 1;
	stop = length(var);
	call prxnext(regex, start, stop, trim(var), pos, len);

	do while (pos GT 0);
		var_tall = prxposn(regex, 1, trim(var));
		output;
		call prxnext(regex, start, stop, trim(var), pos, len);
	end;
run;

topic Problem with call prxnext if the first match is zero-length in SAS Programming

Problem with call prxnext if the first match is zero-length

Re: Problem with call prxnext if the first match is zero-length