If you are going to use PROC IMPORT then you will have to live with the fact that the lengths of character variables will depend on the actual subset of the data that exists in the one file that PROC IMPORT is currently reading.
Are these files supposed to always have the same columns in the same order every time you use them? If so the fix can be very simple. Just add code to fix the variables. Perhaps something like:
proc import 'todays_excel_file.xlsx' out=step1 replace;
run;
data actual_dataset;
attrib pid length=$7 format=$7;
attrib pid_ini length=$3 format=$3.;
....
set step1;
run;
If the files change a lot then you will need to make something smarter that actually checks the metadata about the files and make decisions about how to fix (or if it can be fixed without human help).
Note you got lucky. You won't see truncation if the first version of the variable that SAS sees is the longest. but just trying to put the longest one first in the MERGE or SET statement cannot work when there are multiple character variables with varying lengths. The longest version of VAR1 might be in DS1 while the longest version of VAR2 might be DS2. So either order will truncate one of the variables.
