Import of csv succesfull but not xlsx

[Gorzyne]

Hi all

I have this first piece of code that works fine

proc import datafile="~/NAS/A/myfolder/myfile.csv"
   out=work.myfile
   dbms=csv
   replace;
   delimiter=";";
   getnames=no;
run;

and a roughly similar piec of code, just trying to import a xlsx instead of csv, in the same folder

proc import datafile="~/NAS/A/myfolder/anotherfile.xlsx"
   out=work.myfile
   dbms=xlsx
   replace;
   getnames=yes;
run;

that tells me stuff like

712       libname _XLSXIMP xlsx "/home/SXXXXXX/NAS/A/Myfolder
712     ! /anotherfile.xlsx
ERROR: Could not load excel spreadsheet "/home/SXXXXXX/NAS/A/Myfolder/anotherfile.xlsx": No such file o
       directory.
713       header=YES
714       ;
715       data work.anotherfile;
716       set _XLSXIMP.'_firstsheet_'n;
              ^
ERROR: Library "_XLSXIMP" is not known
717       ;
718       run;

NOTE: DATA step was not executed because of errors detected
NOTE: The data step took :
      real time : 0.000
      cpu time  : 0.001


719       libname _XLSXIMP clear;
WARNING: The library "_XLSXIMP" is not defined, so cannot be cleared
720       quit; run;
721       ODS _ALL_ CLOSE;

Anyhow I can stick with csv if I'm stuck but I thought you guys might get me out of that one

Best regards,

Gorz

[RW9 · Accepted solution]

No such file or directory.

I am not sure what else we can say on the matter, it is telling you there is not file named that, in the location you have provided.  Only oyu have access to your systems to confirm this.

In terms of data import, CSV is far and away the better medium, being portable across systems, plain text, so readable without proprietary software etc.  I would hihgly recommend to use CSV, or XML/Json/Html or one of those types.  I would also highly recommend to avoid using proc import - this is a guessing procedure, which will guess what it thinks your data should look like.  As you are the best person to know what your data should look like (and of course you will have an import agreement stating formats/lengths etc. wont you), you should input this information into a datastep (you can take the one generated by the proc import from the log and use that as a basis).  That way every import of the data will conform to the same structure (or fail) and will be clean and repeatable.