PrxChange: avoid spaces in wildcard replacement

[Oligolas]

Hi,

I hope you are doing fine, I am looking for a trick in a regular expression statement.

Question:Is there a direct way to avoid placing spaces in a 'prxchange replacement part' between a wildcard and a zero?

Example:/$1 0.$4/ (use group1, followed by a zero, followed by a dot, followed by the fourth group)?

Problem: I do not want any space between '$1' and my '0' but if I write /$10.$4/ SAS wants to replace with the 10th group, followed by a dot, followed by the fourth group...

Background: I am using a regex to ensure that my lab values have a leading zero if my string looks like a number with decimals (even when preceeded by a logical operator).

data test;
length a expect $40;

a='.064';      expect='0.064';      output;
a='.064   ';   expect='0.064';      output;
a='   .064   ';expect='0.064';      output;
a='<.01';      expect='<0.01';      output;
a='1.1';       expect='1.1';        output;
a='.064zzz';   expect='.064zzz';    output;
a='0';         expect='0';          output;
a='99.6';      expect='99.6';       output;
a='';          expect='';           output;
a='<3.17';     expect='<3.17';      output;
a='<3.17zzz';  expect='<3.17zzz';   output;
a='<=.17';     expect='<=0.17';     output;
a='>=.17';     expect='>=0.17';     output;

a='   <=.17';  expect='<=0.17';     output;
a='   >=.17';  expect='>=0.17';     output;

a='<>.17';     expect='<>0.17';     output;
a='<=>.17';    expect='<=>.17';     output;
a='<>.17zzz';  expect='<>.17zzz';   output;
a='.064 .064'; expect='.064 .064';  output;
a='>60';       expect='>60';        output;
run;

data test1;
   set test;
   length b $40;
   b=a;
   if prxmatch('/^\s*((<|>|=){0,2})\s*(\.)(\d+)\s*$/',a) then do;
      b=strip(tranwrd(a,'.','0.'));
   end;  
   c=prxchange('s/^\s*((<|>|=){0,2})\s*(\.)(\d+)\s*$/$1 0.$4/',1,a);

   c1=prxchange('s/^\s*((<|>|=){0,2})\s*(\.)(\d+)\s*$/$1/',1,a);
   c2=prxchange('s/^\s*((<|>|=){0,2})\s*(\.)(\d+)\s*$/$2/',1,a);
   c3=prxchange('s/^\s*((<|>|=){0,2})\s*(\.)(\d+)\s*$/$3/',1,a);
   c4=prxchange('s/^\s*((<|>|=){0,2})\s*(\.)(\d+)\s*$/$4/',1,a);

if b ne trim(expect) then FAILb='YES';
if c ne trim(expect) then FAILc='YES';
run;

/*
'^             Start of the string
\s*            any number of spaces
((<|>|=){0,2}) a combination of maximal 2 operators with <,> or = (First group)
\s*            any number of spaces
(\.)           a dot (Third group)
(\d+)          at least one digit (Fourth group)
\s*            any number of spaces
$              End of the string
*/

I would like c to get the same result as b

Thanks!

[gamotte · Accepted solution]

Hello,

Does the following replacement string answer your problem ?

/${1}0.$4/

[Oligolas]

Excellent, I didn't think of placing the group number in curlys. Thanks!