Reading Data with INPUT

The INPUT statement describes how to read raw text into variables. So far you've used list input: values separated by blanks.

The 8-character default

With list input, a character variable gets a length of 8 unless you say otherwise — longer values are cut off:

data people;
   input name $ city $;
   datalines;
Christopher Johannesburg
Ada London
;
run;

proc print data=people;
run;

Obs    name        city

  1    Christop    Johannes
  2    Ada         London

Fix it with a LENGTH statement before the INPUT:

data people;
   length name $ 12 city $ 15;
   input name $ city $;
   datalines;
Christopher Johannesburg
Ada London
;
run;

proc print data=people;
run;

Obs    name           city

  1    Christopher    Johannesburg
  2    Ada            London

Missing values and delimiters

In list input, a period . means "missing". For comma-separated values, use INFILE DATALINES with DSD, which also treats two commas in a row as a missing value:

data scores;
   infile datalines dsd;
   input name $ math english;
   datalines;
Ada,90,85
Alan,,70
Grace,78,
;
run;

proc print data=scores;
run;

Obs    name     math    english

  1    Ada        90         85
  2    Alan        .         70
  3    Grace      78          .

Column and formatted input

For fixed-width data, give column positions — spaces inside values are then allowed:

data staff;
   input name $ 1-12 dept $ 14-18 salary 20-24;
   datalines;
Mary Jones   Sales 52000
Li Wei       IT    61000
;
run;

proc print data=staff;
run;

Obs    name          dept     salary

  1    Mary Jones    Sales     52000
  2    Li Wei        IT        61000

Formatted input uses a pointer (@column) and an informat — a rule for reading a value, such as date9. for dates or comma8. for numbers with commas. The next lesson but one covers informats.

Several rows per line

A trailing @@ keeps reading the same line for the next observation:

data points;
   input x y @@;
   datalines;
1 2 3 4 5 6
;
run;

NOTE: SAS went to a new line when INPUT statement reached past the end of a line.
NOTE: The data set WORK.POINTS has 3 observations and 2 variables.

The first note is expected with @@ and harmless here.

Common mistakes

• Forgetting $ — SAS then tries to read text as a number, sets it to missing, and writes NOTE: Invalid data for name in line ... to the log.
• Putting LENGTH after INPUT — by then the length is already set to 8, and SAS warns that the length can't be changed.