Generating distinct list of variables of which can have a specific string value

[stayhydrated]

Summary: I have a very large table (more than 1000 columns) where I need to find (create a list that I can copy from) the variable names that have values that contain the string "XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX". 

I have sample code I wrote and some sample data to help:

data have;
input cust_id lorem $ ipsum $ dolor $ sit $ amet $ consectetur $ adipiscing $ elit $ sed $ do $ eiusmod $ tempor $
;
datalines;
1 XXXX KW23 RW16 BA62 KR39 HO32 FX55 VT44 RG81 XXXX SI10 ZK27
1 WW10 ST89 RT44 YQ20 FD70 ZB99 MF40 TV38 HE86 XXXX MU18 OC98
1 IQ59 VM72 XXXX FI95 AT96 GU59 JO89 AW21 FV34 XXXX TU46 TK69
2 IX95 KV92 NU30 BZ39 YS62 VU51 RW99 YX60 AG35 XXXX CM70 BV53
2 XXXX FS77 QI34 NT73 CW86 ED83 HU82 PY22 WS60 XXXX QM63 AN29
3 UP55 RU88 DO14 YW45 OR14 KH75 RA76 MZ65 FD65 XXXX BR51 TE67
4 VX45 NP14 II57 VR57 WY25 FH58 EZ58 XXXX BJ11 XXXX TH83 FY17
4 UJ73 HE29 GV27 FT69 RS17 MK32 CE47 ZJ30 KK48 XXXX KD69 XXXX
4 ZE25 UM46 XXXX UC55 XX99 YS61 YJ23 IM52 ZH76 XXXX MO29 XXXX
5 EA96 AY41 DO99 US92 VC61 HV97 ZR44 CF46 YN99 HC55 HC72 XXXX
;
run; 
quit; 

data have_char; 
	set have (KEEP=_CHARACTER_); 
run; 

proc contents data = have_char NOPRINT OUT=VBLS (KEEP=NAME VARNUM); run; 
proc sort data = VBLS; by VARNUM; run; 

PROC SQL NOPRINT;
SELECT DISTINCT NAME
INTO :var_list separated by " "
FROM VBLS;
QUIT; 

data want_if_it_worked (keep=VARS_WITH_XXXX);
	set have;  
	array val{12} &var_list.; 
	VARS_WITH_XXXX=vname(val(whichc("XXXX",of val(*))));
run; 

This is the table I am getting with the above code that I do not want: 

Below is the table I want to output from the "have" data:

Thanks for the help! 

[ChrisNZ]

Like this?

data WANT; 
  set HAVE;
  array C[*] _CHARACTER_; 
  if whichc('XXXX',of C[*]) then
    FIRSTVAR_WITH_XXX=vname(C[whichc('XXXX',of C[*])]);
run; 

FIRSTVAR_WITH_XXX
lorem
do
dolor
do
lorem
do
elit
do
dolor
tempor

[stayhydrated]

This could work in a pinch - I would have to select the distinct values (which is easy enough). Nevertheless this is still very helpful and I can use this in other times - thank you!

[Ksharp · Accepted solution]

data have;
input cust_id lorem $ ipsum $ dolor $ sit $ amet $ consectetur $ adipiscing $ elit $ sed $ do $ eiusmod $ tempor $
;
datalines;
1 XXXX KW23 RW16 BA62 KR39 HO32 FX55 VT44 RG81 XXXX SI10 ZK27
1 WW10 ST89 RT44 YQ20 FD70 ZB99 MF40 TV38 HE86 XXXX MU18 OC98
1 IQ59 VM72 XXXX FI95 AT96 GU59 JO89 AW21 FV34 XXXX TU46 TK69
2 IX95 KV92 NU30 BZ39 YS62 VU51 RW99 YX60 AG35 XXXX CM70 BV53
2 XXXX FS77 QI34 NT73 CW86 ED83 HU82 PY22 WS60 XXXX QM63 AN29
3 UP55 RU88 DO14 YW45 OR14 KH75 RA76 MZ65 FD65 XXXX BR51 TE67
4 VX45 NP14 II57 VR57 WY25 FH58 EZ58 XXXX BJ11 XXXX TH83 FY17
4 UJ73 HE29 GV27 FT69 RS17 MK32 CE47 ZJ30 KK48 XXXX KD69 XXXX
4 ZE25 UM46 XXXX UC55 XX99 YS61 YJ23 IM52 ZH76 XXXX MO29 XXXX
5 EA96 AY41 DO99 US92 VC61 HV97 ZR44 CF46 YN99 HC55 HC72 XXXX
;
run;

proc transpose data=have(obs=0) out=vname;
var _character_;
run;

data _null_;
 set vname end=last;
 if _n_=1 then call execute('proc sql;create table temp as select ');
 call execute(catx(' ','max(upcase(',_name_,')="XXXX") as ',_name_));
 if not last then call execute(',');
  else call execute('from have ;quit;');
run;

proc transpose data=temp out=want(where=(col1=1)) name=vars_with_xxxx;
var _all_;
run;

[stayhydrated]

This is excellent, thank you very much!

[Astounding]

A slightly different twist with additional information.  Once you have read in the data:

data want;
   set have;
   array xx {*} _character_;
   do _n_ = 1 to dim(xx);
      if xx{_n_} = 'XXXX' then do;
         xxxx_name = vname(xx{_n_});
         output;
      end;
   end;
   keep xxxx_name;
run;
proc freq data=want;
   tables xxxx_name;
run;

This way, you not only get the list of variables names, you also get a count of how many times each name contains XXXX.

[stayhydrated]

I really like this slightly different twist! It will be helpful building my skills. Thank you so much!