issue with dropping leading zeros in a string

[buddha_d]

Dear community,

             I want to drop the leading zeros in a string which looks like the below dataset. I am trying to match these numbers in another table (which has the same numbers but no leading zeros).  Could you please help me do this. I tried this way, but all zero numbers are throwing error in my sas. Could you please help me. Thanks !!

data leadingzero;
input number $;
new_num1=substr(number, indexc(number,'123456789'));
cards;
0123
123
124-487
007_0k
000033y
00000
;
run;

The log for error is

1 data leadingzero;
2 input number $;
3 new_num1=substr(number, indexc(number,'123456789'));
4 cards;

NOTE: Invalid second argument to function SUBSTR at line 3 column 10.
RULE: ----+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8----+
10 00000
number=00000 new_num1= _ERROR_=1 _N_=6
NOTE: The data set WORK.LEADINGZERO has 6 observations and 2 variables.
NOTE: DATA statement used (Total process time):
real time 0.32 seconds
cpu time 0.04 seconds

[PGStats]

Depends what you want for those cases. One solution is:

new_num1=substr(number, max(1, indexc(number,'123456789')));

[PGStats · Accepted solution]

Another solution would be:

new_num2=prxChange("s/^0+//", 1, number);

[buddha_d]

Thanks PGStats. This worked well in my query.

[PGStats]

Or maybe:

new_num3=prxChange("s/^0+(\S)/\1/", 1, number);

[RichardDeVen]

Use VERIFY to determine position of first leading character that is not a 0.

string = substr(string, VERIFY(string,'0'));

[buddha_d]

My SAS EG gave me error when I tried this. But, Thanks for your help

[s_lassen]

I assume that what you want to do when the number part is all zeros is not to return an empty string, but to have a single zero, as that is the normal numeric representation of the number zero.

In which case I assume you want '00x' to become '0x' and '000000' to become '0'.

PRX expressions seem a good way of accomplishing that: What you want to do is to delete any zeros that come before a digit (including zero):

new_num1=prxChange('s/^0*(?=\d)//', 1, number);

The PRX expression: 

"^" means the beginning of the input string.

"0*" is any number of zeros.

"(?=\d)" is a look-ahead assertion, saying that after the string found there must be a single digit (including a zero).

[LeonidBatkhan]

Yes, all zeros throw an error because indexc(number,'123456789') returns 0, but 0 cannot be a value for the second argument of SUBSTR. Here is a solution that should work:

data leadingzero;
   input number $;
   new_num1 = substr(number,findc(number,'0','k'));
   cards;
0123
123
124-487
007_0k
000033y
00000
;

It creates the following data table:

Hope, this helps.

Also, I would recommend not to rely on the default column length and assign it explicitly:

data leadingzero;
   length number $10;
   input number $;
   new_num1 = substr(number,findc(number,'0','k'));
   cards;
0123
123
124-487
007_0k
000033y
00000
;

Otherwise, you may run in a situation of truncated variable values. The length must be large enough to accommodate all zeros plus one.

But even more robust solution would be this:

data leadingzero;
   length number $10;
   input number $;
   cards;
0123
123
124-487
007_0k
000033y
0000000000 
;

data noleadingzero (drop=pos);
   set leadingzero;
   length new_num1 $10;
   pos = findc(number,'0','k');
   if pos=0 then new_num1='';
            else new_num1=substr(number,pos);
run;

This way, your stripping leading zeros will not depend on possible errors in your input data set: you create your input data set first, make sure it is correct, and then use it in the next data step which takes care of stripping the leading zeros.

[buddha_d]

thanks for the explanation and solution

[buddha_d]

Thanks S_lassen for helping me