topic Re: Conditionally Deleting First Digit in a Variable in SAS Programming

Conditionally Deleting First Digit in a Variable

ginak — Thu, 20 Apr 2017 05:47:24 GMT

Hello!

I have a variable called CellPhoneNo (cell phone numbers) that I want to look like this: 555-222-3333. This is what most look like:

5552223333

but some people included a 1 so now I also have:

15552223333

My goal is to add a format to change this to

555-222-3333.

How can I do this if some numbers start with a 1? I was thinking of eliminating the "1" and then applying the format. But is there a way to just add a dash in starting from the right, 4 digits in, and then keep going after 3, and then after 1 place in case there is a "1" at the beginning?

Thanks!

Best,

Gina

Re: Conditionally Deleting First Digit in a Variable

PeterClemmensen — Thu, 20 Apr 2017 06:32:25 GMT

Is CellPhoneNo a numeric or character variable?

Re: Conditionally Deleting First Digit in a Variable

RW9 — Thu, 20 Apr 2017 08:30:43 GMT

Here we are again. Post test data in the form of a datastep.

Personally I use formats very little, so I would just do:

phone_no=catx('-',substr(var,lengthn(var)-9,3),substr(var,lengthn(var)-6,3),substr(var,lengthn(var)-3));

Not seeing the point of keeping something which doesn't have to be numeric as numeric. (I.e. area codes in brackets)

Re: Conditionally Deleting First Digit in a Variable

Peter_C — Thu, 20 Apr 2017 08:48:26 GMT

Formatting numeric in phone number style is/was a standard demo example for user-defined formats
Check the on-line doc

Re: Conditionally Deleting First Digit in a Variable

Ksharp — Thu, 20 Apr 2017 10:49:45 GMT

Assuming it is numeric variable.




proc format;
picture fmt
 other='000-000-0000';
run;
data have;
input x;
p=int(log10(x));
if p=10 then x=mod(x,10**p) ;
format x fmt.;
drop p;
cards;
5552223333
15552223333
;
run;

Re: Conditionally Deleting First Digit in a Variable

Peter_C — Thu, 20 Apr 2017 10:52:46 GMT

Hi Gina @ginak

as I couldn't find it in current doc, here (thank you @Cynthia_sas ) is a link to an old posting about creating and using a user-defined format for phone numbers communities.sas.com/t5/SAS-Enterprise-Guide/format-phone-number/m-p/6001

hope it works for you

peterC

Re: Conditionally Deleting First Digit in a Variable

Peter_C — Thu, 20 Apr 2017 11:01:19 GMT

Hi @Ksharp

as the format is defined to present just those 10 digits, it is any removing leading digits when wider than 10, so I don't think it needs selective processing

peterC

Re: Conditionally Deleting First Digit in a Variable

Peter_C — Sat, 22 Apr 2017 14:37:59 GMT

RW9
surprising !
you miss so much value

Re: Conditionally Deleting First Digit in a Variable

RW9 — Mon, 24 Apr 2017 08:17:37 GMT

No, not really. Formats can be useful multi level formats in some procedures and formats on graphs. However I am really against any proprietary file format, compiled macros, functions, formats, etc. If I could use CSV like datasets I would drop datasets also. The move from 32bit to 64bit has really highlighted how restrictive and future falible such formats are.

Re: Conditionally Deleting First Digit in a Variable

Peter_C — Mon, 24 Apr 2017 16:23:15 GMT

while I won't argue SAS's case for the lack of migration support into 64bit from 32, I can only compare the scenario (incompatability of format catalogs between 32 and 64bit) to almost all other applications, where equally there is the equivalent of a transport file in .csv or .xml format -

here we should (easily) keep "permanent" formats converted to .csv (via cntlout ds) - providing no version dependency.

I would not recommend keeping macros in compiled catalogs - perhaps we should be saying the same about compiled formats.

Re: Conditionally Deleting First Digit in a Variable

Tom — Mon, 24 Apr 2017 16:55:16 GMT

Interesting case of seeing how PICTURE formats work.

By using '0' instead of '9' for the digit locations you can allow SAS to basically trim on the left.

So this format definition will work for numberic values that are either with or without the leading 1 .

picture phonefmt other='0-000-000-0000';

So a little test.

1755  data have;
1756    input x;
1757    put x phonefmt. x comma15.;
1758  cards;

  201-555-1234  2,015,551,234
1-212-555-9876 12,125,559,876

It doesn't work so well if you want to use () around the area code. Since then without the extra digit in front you do not get the opening paren.

  201)555-1234  2,015,551,234
1(212)555-9876 12,125,559,876

If you want to add the extra one.

if . < phone < 10**10 then phone=phone+10**10;