Is there better way than this
%macro optimizing_var_len(__dataset=,label_refine=0);
%if "&__dataset"^="" and %sysfunc(exist(&__dataset))=1 %then %do;
libname temp111 "/SAS_Shared_Data/DATA_DUMPS/optimizing_var_len";
data _null_;
if index("&__dataset",'.') then do;
call symputx('__lib',scan("&__dataset",1,'.'),'l');
call symputx('__dataset',scan("&__dataset",2,'.'),'l');
end;
else
call symputx('__lib','work','l');
run;
%put &__lib &__dataset;
proc contents noprint data=&__lib..&__dataset out=__col;
run;
proc sql noprint;
select cat("max(length(trim(",name,"))) as ",name) into : __char SEPARATED by ","
from __col where type=2;
select count(name) into : __char_cnt SEPARATED by ","
from __col where type=2;
select count(name) into : __num_cnt from __col where type=1;
quit;
%if &__num_cnt>0 %then %do;
proc means data=&__lib..&__dataset min max sum stackods ;
ods output summary=__mean ;
run;
%end;
%else %do;
data __mean;
length variable $8 min max sum 3;
delete;
run;
%end;
data mean_slect3 mean_slect4;
set __mean;
if max(abs(min),abs(max))<=100 and int(sum)=sum then output mean_slect3;
else if max(abs(min),abs(max))<=100000 and int(sum)=sum then output mean_slect4;
run;
%if &__char_cnt>0 %then %do;
proc sql;
create table o as
select &__char from &__lib..&__dataset;
quit;
proc transpose data=o out=o_t ;
run;
%end;
%else %do;
data o_t;
length _name_ $10 col1 3;
delete;
run;
%end;
%DEL_VAR1:
%let __char_len=;
%let _len_3=;
%let _len_4=;
proc sql noprint;
select cat(_name_," $",strip(put(col1,best10.))) into: __char_len SEPARATED by " " from o_t;
select cat(Variable," 3") into : _len_3 SEPARATED by " " from mean_slect3;
select cat(Variable," 4") into : _len_4 SEPARATED by " " from mean_slect4;
quit;
options varlenchk=nowarn;
data temp111.&__dataset ;
%if "&__char_len"^="" %then length &__char_len;;
set &__lib..&__dataset;
%if "&_len_3"^="" or "&_len_4"^="" %then length &_len_3 &_len_4 ;;
run;
options varlenchk=warn;
proc compare base=&__lib..&__dataset compare=temp111.&__dataset
outstat=outcompstat(where=(_TYPE_='NDIF' )) nosummary novalues ;
ods output CompareSummary(nowarn)=compare_summ ;
run;
proc sql noprint;
select count(1) into: __unequal_var from outcompstat where _BASE_^=0 or _comp_^=0;
select count(1) into: __equal_var from outcompstat where _BASE_=0 and _comp_=0;
quit;
%if &__unequal_var>0 %then %do;
proc sql noprint;
delete * from mean_slect3 where variable in (select _var_ from outcompstat where _BASE_^=0 or _comp_^=0);
delete * from mean_slect4 where variable in (select _var_ from outcompstat _BASE_^=0 or _comp_^=0);
quit;
%goto del_var1;
%end;
%else %if &__equal_var=&__num_cnt and %sysfunc(exist(compare_summ))=0 %then %do;
proc copy in=temp111 out=&__lib. move ;
select &__dataset;
run;
%end;
%if &label_refine=1 %then %do;
data _null_;
set __col end=last;
if _n_=1 then call execute('proc datasets library=&__lib nolist noprint; modify &__dataset ;');
call execute(cat('attrib ',strip(name),' label="',upcase(strip(name)),'";'));
if last then call execute(';quit;run;');
run;
%end;
proc sql;
drop table __col, o, o_t, mean_slect3 ,mean_slect4,outcompstat,__mean
%if %sysfunc(exist(compare_summ))=1 %then ,compare_summ;
%if %sysfunc(exist(temp111.&__dataset))=1 %then ,temp111.&__dataset;
;quit;
libname temp111 clear;
%end;
%else
%put WARNING: Invalide dataset &__dataset;
%mend optimizing_var_len;
%optimizing_var_len(__dataset=j);
