Using PCRXFIND with DS2

Question

I want to replicate this using PCRXFIND instead of prxmatch for speed and because prxmatch is not working correctly.

I am trying to find text in a data file. Here is a reproducible example. There is one file with the data to search and another with the search terms. I want to use regular expressions. Normally this is done with prxmatch. I want to do it with PCRXFIND because I am working inside DS2. I use a hash iterator to iterate through the search terms or each observation in the data set.

The search result do not follow any pattern I can recognize. I would be happy for any suggestions. I posted this on sas communities with not results.

    * ds2 PCRXFIND example;

data person;
    input name $ dept $;
    datalines;
John Sales
Mary Acctng
Joe Findme
Sue Hereiam
;
run;


data searchterms;
    infile datalines missover;
    input s_index $ term $;
datalines;
1   Hereiam
2   Findme
3   Acc
;
run;

proc contents data=searchterms; run;
proc print data=searchterms; run;

proc ds2;
    data search_results (overwrite=yes);

    dcl double rc c ;
    declare char(8) s_index;
    declare char(8) term;
    declare char(11) name dept;
    declare char(1) c_options;
    declare char(20) search_term search_text;
    dcl package hash h(1, '{select s_index, term from searchterms}');
    dcl package hiter hi('h');
        method init();
            c_options = 'i';
            rc = h.defineKey('s_index');
            rc = h.defineData('term');
            rc = h.defineDone();
        end;
        method run();
            dcl double rc;
            set {select  name, dept from person};
            rc = hi.first();
            do while(rc=0);
                c = prxmatch('/'||compress(term)||'/i',name||'   '||dept);
                search_term = '/'||compress(term)||'/i';
                search_text = name||'   '||dept;
                rc = hi.next();
                output;

            end;
        end;
    enddata;
run;
quit;

Answer 1

In SAS 9.4M5 a new package was added for regular expression matching, DS2 PCRXFIND Package. The below is an example of using that package along with a custom list object package to complete your task.

proc ds2;
package prxElement/overwrite=yes;
  declare package prxElement next;
  declare int                id;
  declare package pcrxfind   prx;

  method prxElement( int                               id,
                     varchar(32767) character set utf8 regex);
    this.id = id;
    prx = _new_ pcrxfind(regex);
  end;
endpackage;
run;

package prxList/overwrite=yes;
  declare package prxElement head;
  declare int count;

  method prxList();
    head  = null; *null is an empty list;
    count = 0;
  end;

  method last() returns package prxElement;
    declare package prxElement item;
    item = head;
    do while ( ^null(item.next) );
      item = item.next;
    end;
    return item;
  end;

  method add(package prxElement element);
    if ( ^null(head) ) then do;
      declare package prxElement joinTo;
      joinTo=last();
      joinTo.next = element;
      count = count+1;
    end;
    else do;
      head = element;
      count = 1;
    end;
  end;
endpackage;
run;

data results(overwrite=yes);
declare double rc;
drop term;
declare package pcrxfind   prx;
declare package prxElement item;
declare package prxList    prxList;
method init();
  declare int i;
  prxList=_new_ prxList();
  do i=1 to &termdim;
    set searchterms;
    item=_new_ prxElement(s_index,cats('/',term,'/i'));
    prxList.add(item);
  end;
end;
method run();
  declare int i;
  set person;
  item = prxList.head;
  do i=1 to prxList.count;
    prx = item.prx;
    s_index = item.id;
    rc = (prx.Match(catx(' ',name,dept)))>0 ;
    output;
    item=item.next;
  end;
end;
enddata;
run;
quit;

Answer 2

This looks like it works fine. I'm not 100% sure why yours didn't work, but usually when you're not being sufficiently careful with surrounding spaces things happen that don't always make sense.

I used CATS and CATX for concatenation, which seems the most useful way to handle the string combinations. I also shifted around the output, and used c_options.

data person;
    input name $ dept $;
    datalines;
John Sales
Mary Acctng
Joe Findme
Sue Hereiam
;
run;


data searchterms;
    infile datalines missover;
    input s_index $ term $;
datalines;
1   Hereiam
2   Findme
3   Acc
;
run;

proc contents data=searchterms; run;
proc print data=searchterms; run;

proc ds2;
    data search_results (overwrite=yes);

    dcl double rc c ;
    declare char(8) s_index;
    declare char(8) term;
    declare char(11) name dept;
    declare char(1) c_options;
    declare char(20) search_term search_text;
    dcl package hash h(1, 'searchterms');
    dcl package hiter hi('h');
        method init();
            c_options = 'i';
            rc = h.defineKey('s_index');
            rc = h.defineData('term');
            rc = h.defineDone();
        end;
        method run();
            dcl double rc;
            set person;
            rc = hi.first();
            do while(rc=0);
                search_term = cats('/',term,'/',c_options);
                search_text = catx(' ',name,dept);
                c = prxmatch(search_term, search_text);
                output;
                rc = hi.next();
            end;
        end;
    enddata;
run;
quit;

proc print data=search_results;
run;

Answer 3

I have not used DS2, so donot know much about the code ther but You can simply do it on join condition in proc sql. something like below

 proc sql;
create table want as 
select  distinct t1.* 
 from person T1
 INNER JOIN searchterms t2 
ON find(t1.dept,trim(t2.term) ) > 0;