Checking for proper ordering of numeric, time, etc - sas

Question

My data looks something like this:
data tmp ;
input id var1 - var5 ;
datalines ;
1 1 2 3 4 5
2 1 2 . . .
3 1 . . . 4
4 . 3 . . .
5 . . . . 5
6 1 3 2 2 3
7 5 3 7 8 9
8 1 . . . 2
9 1 . 2 3 4
;
run ;
I'm trying to determine if n variables are properly 'ordered.' By ordered, I mean numerically or sequential in time (or even alphabetic). So in this example, my desired output would be:
dummy = 1 1 1 1 1 0 0 1 1 since the ones where dummy = 1 are in correct order.
It would be trivial if I had complete data:
if var1 <= var2 <= ... <= varn then dummy = 1; else dummy = 0;
I do not have complete data unfortunately. So the problem may be that sas treats . as a very small number(?) and also that I cannot perform operations on . since this also failed:
if 0 * (var1 = .) + var1 <=
var1 * (var2 = .) + var2 <=
var2 * (var3 = .) + var3 <= ... <=
var_n-1 * (varn = .) + varn
then dummy = 1;
else dummy = 0;
Basically this would check to see if a variable is . and if it is, then use the previous value in the inequality, but if it is not missing, proceed as normal. This works sometimes, but still requires most of the info to be nonmissing.
I have also tried something like:
if var2 = max(var1, var2) & var1 <= var2 &
var3 = max(var1 -- var3) & var2 <= var3 & ...
but this approach also needs complete data. And I have tried transposing the data into a long format so that I can just delete the missing columns (and only keep variables I am interested in knowing the order of) but a transposed data set of thousands of variables isn't useful to me (if you would convert back to wide, there would still be missing columns).
Clearly, I am not the best SASer, but I would ideally like to write a macro or something since this issue comes up for me a lot (basically just a data check to see if dates are in order and occur when they should be regarding their relative timeline).
Here is all the code:
data tmp ;
input id var1 - var5 ;
datalines ;
1 1 2 3 4 5
2 1 2 . . .
3 1 . . . 4
4 . 3 . . .
5 . . . . 5
6 1 3 2 2 3
7 5 3 7 8 9
8 1 . . . 2
9 1 . 2 3 4
;
run ;
data tmp1 ;
set tmp ;
if var1 <= var2 <= var3 <= var4 <= var5 then dummy1 = 1 ; else dummy1 = 0 ;
if 0 * (var1 = .) + var1 <=
var1 * (var2 = .) + var2 <=
var2 * (var3 = .) + var3 <=
var3 * (var4 = .) + var4 <=
var4 * (var5 = .) + var5
then dummy2 = 1 ;
else dummy2 = 0 ;
if var2 = max(var1,var2) & var1 ~= var2 &
var3 = max(var1, var2, var3) & var2 ~= var3 &
var4 = max(var1, var2, var3, var4) & var3 ~= var4 &
var5 = max(var1, var2, var3, var4, var5) & var4 ~= var5
then dummy3 = 1 ;
else dummy3 = 0 ;
* none of dummy1 - 3 pick up the observations that are in proper order ;
run ;
data tmp1_varsIwant ;
set tmp1 ;
keep id var1 -- var5 ;
run ;
proc transpose data = tmp1_varsIwant out = tmp1_long ;
by id ;
run ;
data tmp1_long ;
set tmp1_long ;
if col1 = . then delete ;
if _name_ in('var6', 'var999') then delete ;
run ;
proc sort data = tmp1_long ;
by id col1 ;
run ;

Answer 1

Maybe you could force all the logic into one conditional, but it's probably simpler to use a loop like this:
data tmp1 ;
set tmp ;
array vars (*) var1-var5;
last_highest = .;
dummy = 1;
do i = 1 to 5;
if vars(i) > . and vars(i) < last_highest then do;
dummy = 0;
leave;
end;
last_highest = coalesce(vars(i),last_highest);
end;
run ;