numeric to character using the input function

[K_S]

SAS documentation states that the source variable for input() function must be character.

I just tried the input function with a numeric source and it worked - i got what i wanted to get.

I had a numeric variable that i wanted to convert into a character variable that had a length of 40 so i did the following:

new_char_var=input(original_num_var, $char40.)

It worked but i wonder if this is bad practice and if yes, then why? What are the downsides of doing what I did.

[Tom · Accepted solution]

Formats convert values to text. Informats convert text to values.

So if you want to convert your number to a character value use the PUT() function, not the INPUT() function.

When you used a numeric value as the first argument to the INPUT() function and a character informat as the second argument the SAS data step compiler noticed the conflict so it automatically converted the number into a character string (you will see messages about that in the SAS log).  It will use the BEST12. format to do that.   So you converted your number into a 12 character string (automatic conversion) and use the INPUT() function to convert that 12 character string into a 40 character string.  You then assigned the value to the variable new_char.  If NEW_CHAR was already defined then the value was either truncated or padded with spaces to fill its defined length.  If NEW_CHAR had not yet been defined then the data step compiler would make a guess that you wanted to define it as length $40 to match the width of the informat you used in the INPUT() function call.

So you effective did:

length new_var $40;
new_char_var=put(original_num_var, best12.);

[data_null__]

Did you notice the log message?  The number is converted to character then read with $CHAR which preserves leading spaces.  In general it makes you look like you don't know what you're doing.  There may be a scenario where it is exactly what you want to do but I can't think of one. 

[K_S]

the proper way of doing this is define the length of 40 at the beginning and then do

data out;

set in;

length new_char_var $40.;

new_char_var=put(original_num_var, 8. -L);

run;

?

by doing new_char_var=compress(input(original_num_var, $char40.)); i get the exact same output

[Reeza]

Because SAS is fixing it for you behind the scenes, which is why the log is different if you do it correctly versus doing it the original way.

[K_S]

is the following more palatable?

new_char_var=INPUT(PUT(original_num_var, 8. -L), $CHAR40.);

[Reeza]

What does the log show?

[K_S]

the log is clean.. no error messages or warnings

[Reeza]

It's confusing though, just use the single PUT to get what you need, why the INPUT() again?
The INPUT doesn't do anything.

[K_S]

the length needs to be 40 and i want to avoid specifying the length in a separate line. So converting numeric with lenght 8 into character with length 8 and then applying $char40. to make the length 40.

is there a way to do that with a single put?

[Reeza]

---

@K_S wrote:

the length needs to be 40 and i want to avoid specifying the length in a separate line. So converting numeric with lenght 8 into character with length 8 and then applying $char40. to make the length 40.

is there a way to do that with a single put?

---

That's an artificial requirement that really makes no sense. It means you start breaking the typical coding structure and your code is unclear but if that's what you want go for it.