Extracting a phrase using Regex (PRXNEXT)

[BrianB4233]

Hi all,

I've written what I thought was some solid code to extract multiple dates, and the preceding 18 words/non-words, from a free text field. I'm using PRXNEXT b/c there are often multiple dates within the text field and I'd like to extract all of them. However, testing this in https://regex101.com/ and then viewing the results doesn't result in a match. It is correctly identifying, and outputting, the date using PRXPOSN but it's not including all of the words/non-words preceding the date.

What is being output in the temp dataset is this:

year: 2.9 %..........[average
woman <1.67%]
NCI Lifetime: 15.1 %..........[average
woman <10%]

A

Whereas in regex101 it's showing this: https://regex101.com/r/LWRcqN/1

data data_chk1;

length dt_1-dt_12 $150 
dt_out1-dt_out12 $30
imp_rep_concat $11000
;

set work.birad_score_0_3(drop=cht_in impressiontext reporttext obs=max);

/* Combine impression & report text together to search as one */
imp_rep_concat = catx(' REPORT_TEXT ',impression_copy,report_copy);

*** Identifies ddOctdd or dOctdddd or ddOctdddd as well if there is a
space/hyphen/whatever between the day & month or month & year;
if _n_ = 1 then do;
retain dt_pattern;
 dt_pattern = prxparse("/(?:\w+\W+){0,18}(\d{1,2}(\.|\/|-)\d{1,2}(\.|\/|-)\d{2,4})/i");
end;

/*if prxmatch(dt_pattern,impression_copy) then do;*/
/*match = 1;*/
/* date_out = prxposn(dt_pattern,1,impression_copy);*/
/*end;*/

start = 1;
stop = length(imp_rep_concat);

call prxnext(dt_pattern,start,stop,imp_rep_concat,pos,len);
	array comm[12] $dt_1-dt_12;
	array comm1[12] $dt_out1-dt_out12;
	do i = 1 to 12 while (pos > 0);
		comm(i) = substr(imp_rep_concat,pos,len);
		comm1(i) = prxPosn(dt_pattern, 1, imp_rep_concat);
 call prxnext(dt_pattern,start,stop,imp_rep_concat,pos,len);
end;


*drop dt_1-dt_12 dt_pattern: start: stop: pos len i;
run;

Any ideas what is causing the inconsistency? Thank you.

[ChrisNZ · Accepted solution]

1. The link you sent uses extensions gmi. SAS does not support option g, and you only use i in your code.

2. In any case only using i gives the same result on regex101.com

3. The result is the same in SAS and regex101 except that the SAS result is truncated at length 200. Try lengthening the variable. 

[BrianB4233]

Thank you ChrisNZ - not sure how I missed this but it was indeed #3.