find words in string

[emaneman]

Hello all,

respondents to a survey indicated a list of names (fiction authors), and they were recorded in a single variable, separated by a comma.

Thus, 

OBS	string
1	Toni Morrison
2	Kurt Vonnegut   ,Harper Lee   ,T.S. Eliot,Toni Morrison, Danielle Steel
3	Robert Ludlum   ,James Patterson   ,Tom Clancy   ,Danielle Steel   ,John Grisham

I need to create a new variable (NEW) that has the frequency of CERTAIN authors. That is, the NEW should not simply count the number of words in STRING. For instance, I want NEW to count Morrison and Steel. So to obtain the following:

OBS	string	NEW
1	Toni Morrison	1
2	Kurt Vonnegut   ,Harper Lee   ,T.S. Eliot,Toni Morrison, Danielle Steel	2
3	Robert Ludlum   ,James Patterson   ,Tom Clancy   ,Danielle Steel   ,John Grisham	1

I am having troubles finding a solution.

Any suggestion would be greatly appreciated

Eman

[PeterClemmensen]

One way

data have;
input OBS string : $ 3-203;
infile datalines4 dlm='|';
datalines;
1|Toni Morrison   
2|Kurt Vonnegut   ,Harper Lee   ,T.S. Eliot,Toni Morrison, Danielle Steel
3|Robert Ludlum   ,James Patterson   ,Tom Clancy   ,Danielle Steel   ,John Grisham
;

data want(drop=i);
    set have;
    new=0;
    do i=1 to countw(string);
        if propcase(scan(string, i)) in ('Morrison', 'Steel') then new+1;
    end;
run;

[emaneman · Accepted solution]

Thank you! Elegant, and fast.

I love this community. 

[PeterClemmensen]

Anytime 🙂 please mark my reply as the solution. This helps future users navigate the community.

[emaneman]

done, thank you again.