Update array of values based on subsequent elements

[Sir_Lancelot]

Hi All

I have a requirement to look at three numeric fields. if field one is not populated and field two and three are, then field two and three must be their values minus one. in summary, if the values on the left are not populated, the the values on the right must be reduced by either one or two.

Source Data 

v1	v2	v3
1	2	3
	2	3
		3
1		3

Expected Reslts

v1	v2	v3
1	2	3
	1	2
		1
1		2

[RW9]

Please clarify your test data, provide full test data in the form of a datastep:
https://communities.sas.com/t5/SAS-Communities-Library/How-to-create-a-data-step-version-of-your-dat...

Its a bit of a guess at the moment, for instance, why are they in that order, are the other variables?  Why is row 4 var3=2, it should be 3 following the logic?

[Sir_Lancelot]

Yes row 4 var3 must be 3. My mistake. there are wil be more variables in the future. So if I get get this as dynamic it will be great

[RW9 · Accepted solution]

With other variables, I meant variables which are not the vX variables, ones which indicate which position the items present should be in, as if you sorted the test data you presented then row 4 could appear after row 1, there is no indication in the data that that row should appear in position 4.  This is key as not be able to move data will limit your options in how to process it.  I mean you could do:

data have;
 infile datalines dsd;
 input v1 v2 v3;
datalines;
1,2,3
,2,3
,,3
1,,3
;
run;

data want;
 set have;
 retain lstv2 lstv3;
 if _n_=1 then do;
   lstv2=v2;
   lstv3=v3;
 end;
 else do;
   if v1=. then do;
     v2=lstv2-1;
     v3=lstv3-1;
     lstv2=v2;
     lstv3=v3;
   end;
   else if v1 ne . then do;
     lstv2=v2;
     lstv3=v3;
   end;
 end;
run;

I would also guess this is a bigger issue in that you are getting this type of structure in the first place.  Go back and look at previous steps and see if, by changing the output, you can simplify this whole matter earlier on in the process.

[Ksharp]

data have;
 infile datalines dsd;
 input v1 v2 v3;
datalines;
1,2,3
,2,3
,,3
1,,3
;
run;
data want;
 set have;
 array x{*} v1-v3;
 n=nmiss(of x{*});
 if n ne 0 then do;
 do k=1 to dim(x);
  if missing(x{k}) then i=k;
 end;
 do j=i to dim(x);
  x{j}=x{j}-n;
 end;
 end;
 drop n i j k;
 run;
 
 proc print;run;