How to read from end of string?

[gobejo]

Hi I have a variable with different record lengths. I am trying to find out if i can read from the end of the string. Also, If any of the "want" is alphanumeric, then DELETE. I want only US based Zip, not canadian postal code

Have	Want
WARREN MI 48091-3241	48091
BLOOMFIELD MI 48304-2916	48304
SAINT CLAIR SHORES MI 48082-1158	48082
HUNTINGTON WOODS MI 48070	48070
MACOMB MI 48042-2398	48042
HUNTINGTN WDS     , MI 480700000	48070

[Haris]

You can try INDEX() function to find the location of the first instance of a numeric character preceded by a blank (including the blank will prevent you from reading Canadian postal codes that begin with a letter) in your string and then nest it in SUBSTRN() function to read 5 characters to the right of the first number.  Finally, sounds like you'll want to nest both of the first two in the INPUT() function to output a number rather than a character string.

[Haikuo]

data test;

     infile cards truncover;

     input str $100.;

     num=prxchange('s/(^\D+)(\b\d{5})(.+)/$2/io',-1, str);

     cards;

WARREN MI 48091-3241

BLOOMFIELD MI 48304-2916  

SAINT CLAIR SHORES MI 48082-1158    

HUNTINGTON WOODS MI 48070 

MACOMB MI 48042-2398

HUNTINGTN WDS     , MI 480700000

;

[Ksharp]

data test;
     infile cards truncover;
     input str $100.;
      length num $ 5;
     num=scan(str,-1,' ');
     cards;
WARREN MI 48091-3241
BLOOMFIELD MI 48304-2916  
SAINT CLAIR SHORES MI 48082-1158    
HUNTINGTON WOODS MI 48070 
MACOMB MI 48042-2398
HUNTINGTN WDS     , MI 480700000
;
run;

Xia Keshan