If you take advantage of the features of CATX and use VVALUE this can be written slightly more concisely.

proc format lib=work; value xx 1='ABC' 2='XYZL' 3='FSFGS'; quit;

data test; input x $; length newx $500; do _n_ = 1 to length(x); xN = input(substr(x,_n_,1),F1.); newx = catx(',',newx,vvalue(xN)); end; format xN xx.; drop xN; datalines; 123 112 12 222 1 11 ; run; proc print; run;

On 10/27/09, Joe Matise <snoopy369@gmail.com> wrote: > proc format lib=work; > value xx > 1='ABC' > 2='XYZL' > 3='FSFGS' > ; > quit; > > data test; > input x $; > format newx $500.; > newx = put(input(substr(x,1,1),BEST.),xx.); > if length(x) ge 2 then do _t = 2 to length(x); > newx=cats(newx,',',put(input(substr(x,_t,1),BEST.),xx.)); > end; > put x= newx=; > datalines; > 123 > 112 > 12 > 1 > 11 > ; > run; > > Probably could also have some fun with TRANWRD, or hash solution if you > don't need to use a format. > > -Joe > > On Tue, Oct 27, 2009 at 2:04 PM, Ya Huang <ya.huang@amylin.com> wrote: > > > Hi there, > > > > Assuming I have a format, something like this. > > > > proc format; > > value xx > > 1='ABC' > > 2='XYZL' > > 3='FSFGS' > > ; > > > > Now I have a string '213', and I want a decoded string 'XYZL,ABC,FSFGS'. > > The string can be '123','312','223','1','21' etc., which is dynamically > > generated. The decoded string should follow the order of the original > > number in the string if more than one numbers are in the original string. > > > > Looking for some elegant solutions. > > > > Thanks > > > > Ya > > >