topic Re: Regular expressions: match dot and comma as characters in SAS Programming

Regular expressions: match dot and comma as characters

Ga1ath — Wed, 17 Nov 2021 08:24:42 GMT

data test;
  input name $25.;
  cards;
x., test
.x test
test x.
;
run;

data want;
   set test;
   a=prxchange('s/\b(x\.,|x\.|\x)\b//i', -1, name);
run;

Hello there!
I'm trying to extract from string literal all parts which looks like x or x. or x.,
So I've wrtitten the code above
A result dataset you can see in the table below
It seems that SAS had deleted only x character, not x. or x.,
despite fact of existence x\. and x\., in regular expression before x
Does anyone know how it could be fixed?

Resulted dataset:

	name	a
1	x., test	., test
2	.x test	. test
3	test x.	test .

Desired dataset:

	name	a
1	x., test	test
2	.x test	. test
3	test x.	test

Re: Regular expressions: match dot and comma as characters

FreelanceReinh — Tue, 16 Nov 2021 19:12:08 GMT

Hello @Ga1ath and welcome to the SAS Support Communities!

I see two issues:

A word boundary (PRX metacharacter \b) cannot occur directly after a comma or a period because by definition it is "the position between a word and a space" -- but neither the comma nor the period is a word character (cf. PRX metacharacter \w).
The "\x" in your regular expression should read "|x".

Try this regex instead:

a=prxchange('s/\b(x\.,|x\.|x\b)//i', -1, name);

Re: Regular expressions: match dot and comma as characters

Tom — Tue, 16 Nov 2021 19:20:35 GMT

Is that last table supposed to be what you want to produce? Or is it the wrong output your current code is producing? If the latter then please provide the desired results for your example input.

Re: Regular expressions: match dot and comma as characters

Ga1ath — Tue, 16 Nov 2021 19:33:07 GMT

It is the wrong output. I'll edit the post.

Re: Regular expressions: match dot and comma as characters

Ga1ath — Tue, 16 Nov 2021 19:40:40 GMT

Thank you for reply!
All clear except second issue. You'h wrote The "\x" in your regular expression should read "|x".
But I don't see \x in my primordial regular expression. Please, can you explain what you have meant?

Re: Regular expressions: match dot and comma as characters

FreelanceReinh — Tue, 16 Nov 2021 21:39:56 GMT

@Ga1ath wrote:

All clear except second issue. You'h wrote The "\x" in your regular expression should read "|x".
But I don't see \x in my primordial regular expression. Please, can you explain what you have meant?

I meant this:

@Ga1ath wrote:

data want;
   set test;
   a=prxchange('s/\b(x\.,|x\.\x)\b//i', -1, name);
run;

whereas

@FreelanceReinh wrote:
a=prxchange('s/\b(x\.,|x\.|x\b)//i', -1, name);

Re: Regular expressions: match dot and comma as characters

Ga1ath — Wed, 17 Nov 2021 08:25:59 GMT

Of course, yeah. I'm sorry for my inattention.