Combining two Arrays which use regular expressions

[Mgarret]

Hi all. I have two arrays I’m trying to combine.  They both use regular expressions. One is looking for all the digits after categoryId= and one is looking for all the digits after productId=. I want to combine them into one so the array looks for either  categoryId= or productId=  so I don't have to use two data sets. I’m having trouble getting the regular expression syntax correct. 

data Camp1;
set Campaigns;
array url_array(*) $ url_:;
do i=1 to dim (url_array) ;
category_Id=input(prxchange('s/.+categoryId=(\d+).+/$1/o',-1,URL_array(i)), 12.);
output;
end;
drop i;run;


data Camp2;
set Campaigns;
array url_array(*) $ url_:;
do i=1 to dim (url_array) ;
category_Id=input(prxchange('s/.+productId=(\d+).+/$1/o',-1,URL_array(i)), 12.);
output;
end;
drop i;run; 

[Jagadishkatam]

Please try this untested code

data Camp1;
set Campaigns;
array url_array(*) $ url_:;
do i=1 to dim (url_array) ;
if prxmatch('m/.+categoryId=(\d+).+/',URL_array(i)) >0 then category_Id=input(prxchange('s/.+categoryId=(\d+).+/$1/o',-1,URL_array(i)), 12.);
else if prxmatch('m/.+productId=(\d+).+/',URL_array(i)) >0 then category_Id=input(prxchange('s/.+productId=(\d+).+/$1/o',-1,URL_array(i)), 12.);
output;
end;
drop i;
run;

[PGStats]

I like to use CALL PRXNEXT() for this kind of extraction. Here I also use look behind buffers (?<=...)

data test;

url_id = 1;
url = "The productId=123 and the categoryId=5555 or categoryId=6666";

if not catID then 
    catID + prxParse("/(?<=categoryId=)(\d+)/");
if not proID then 
    proID + prxParse("/(?<=productId=)(\d+)/");

type = "category";
start = 1; stop = length(url);
call prxnext(catID, start, stop, url, pos, len);
do while (pos> 0);
    category_id = input(substr(url, pos, len), best32.);
    output;
    call prxnext(catID, start, stop, url, pos, len);
    end;

type = "product";
start = 1; stop = length(url);
call prxnext(proID, start, stop, url, pos, len);
do while (pos> 0);
    category_id = input(substr(url, pos, len), best32.);
    output;
    call prxnext(proID, start, stop, url, pos, len);
    end;

keep url_Id type category_id;
run;

proc print data=test noobs; run;

[Patrick · Accepted solution]

Assuming "productid" and "categoryid" are mutually exclusive in your strings then the following could work:

data Camp1;
  set Campaigns;
  array url_array(*) $ url_:;
  do i=1 to dim (url_array);
    category_Id=input(prxchange('s/.+[[:^alnum:]](categoryId=|productId=)(\d+).+/$2/oi',1,URL_array(i)), 12.);
    output;
  end;
  drop i;
run;