extract numbers from variable with character and numeric features

[michokwu]

Hello Experts,

I have a variable that includes both characters and numbers and does not follow any specific coding standard as it has been obtained from multiple sources. I would just like to collect the numbers. I can't use substr because the codes are different. Is there a way I can do this?

Thank you

have	want
BIOL10128DCWC	10128
ANT102	102
PSY125	125
SOC101	101
SPAN2330	2330
HIST1500	1500
BCS-205	205
BCS-132	132

[ghosh]

want=compress(have, '', 'A');

[smantha]

data _new;
informat have $32.;
input have $ var2;
want=compress(compress(have,' +/*-%','A'));
cards;
BIOL10128DCWc	10128
ANT102	102
PSY125	125
SOC101	101
SPAN2330	2330
HIST1500	1500
BCS-205	205
BCS-132	132
;;;
run;

[michokwu]

Thanks! This works but it includes special characters. How do I get only the numbers?

[ghosh]

Did not not see my response, that is what 'A' modifier does.
want=compress(have, '', 'A');
It says, replace alpha with nothing

[michokwu]

Thank you! Some of observations still include special characters.

[Astounding · Accepted solution]

Try it this way:

want = compress(have, , 'kd');

As the third parameter, kd = keep digits

[michokwu]

Thanks you